diff --git a/.bumpversion.toml b/.bumpversion.toml index e32f518c9..a6912bed5 100644 --- a/.bumpversion.toml +++ b/.bumpversion.toml @@ -1,5 +1,5 @@ [tool.bumpversion] -current_version = "0.40.0-beta.11" +current_version = "0.40.0-beta.12" parse = """(?x) (?P0|[1-9]\\d*)\\. (?P0|[1-9]\\d*)\\. diff --git a/.github/workflows/nodejs.yml b/.github/workflows/nodejs.yml index 98acccf02..2e290166e 100644 --- a/.github/workflows/nodejs.yml +++ b/.github/workflows/nodejs.yml @@ -164,6 +164,71 @@ jobs: python ci/mock_openai.py & cd nodejs/examples node_modules/.bin/jest --testEnvironment jest-environment-node-single-context --verbose + arrow-compat: + # The main jobs resolve `apache-arrow` to the version pinned in + # devDependencies (the newest supported release). This job re-runs the + # suite with the library itself running on each older Arrow release the + # peer range still accepts, which is what a consumer on that version gets. + name: Linux (Arrow ${{ matrix.arrow-version }}) + timeout-minutes: 30 + strategy: + matrix: + arrow-version: [ "19.0.1", "20.0.0" ] + runs-on: "ubuntu-22.04" + defaults: + run: + shell: bash + working-directory: nodejs + steps: + - uses: actions/checkout@v6 + with: + fetch-depth: 0 + lfs: true + - uses: pnpm/action-setup@v6 + with: + version: 11.1.1 + - uses: actions/setup-node@v6 + with: + node-version: 24 + cache: 'pnpm' + cache-dependency-path: nodejs/pnpm-lock.yaml + - uses: Swatinem/rust-cache@v2 + with: + # Same crate, profile and target as the `linux` job above, so reuse + # its cache: rust-cache keys on the job id by default, and this job + # would otherwise start cold on every run until it had saved its own. + shared-key: linux + save-if: ${{ github.ref == 'refs/heads/main' }} + - name: Install dependencies + run: | + sudo apt update + sudo apt install -y protobuf-compiler libssl-dev + - name: Build + run: | + pnpm install --frozen-lockfile + # No `--` separator: pnpm forwards it literally, which would + # make napi-rs treat `--profile ci` as a cargo passthrough arg. + pnpm build:debug --profile ci + - name: Resolve apache-arrow to ${{ matrix.arrow-version }} + run: | + # The older releases are already installed as the `apache-arrow-NN` + # test aliases. With pnpm's hoisted linker each alias is a complete, + # self-contained directory, so swap it in for the library's own + # `apache-arrow` instead of running a second install, which has been + # seen to hang after printing "Done" in CI. A copy, not a symlink: the + # cross-version tests rely on the alias being a separate module + # instance from the one the library loads. + major="$(echo '${{ matrix.arrow-version }}' | cut -d. -f1)" + mv node_modules/apache-arrow node_modules/apache-arrow.pinned + cp -R "node_modules/apache-arrow-${major}" node_modules/apache-arrow + # `require("apache-arrow/package.json")` is blocked by the package's exports map. + node -p "'apache-arrow resolved to ' + JSON.parse(require('fs').readFileSync('node_modules/apache-arrow/package.json')).version" + pnpm tsc + - name: Test + env: + # Newer @smithy/core uses dynamic ESM imports. + NODE_OPTIONS: "--experimental-vm-modules" + run: node_modules/.bin/jest --verbose macos: timeout-minutes: 30 # macos-15 ships a newer linker; the older macos-14 linker fails to insert diff --git a/Cargo.lock b/Cargo.lock index 87bef3a25..77c7412f4 100644 --- a/Cargo.lock +++ b/Cargo.lock @@ -141,7 +141,7 @@ version = "1.1.5" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "40c48f72fd53cd289104fc64099abca73db4166ad86ea0b4341abe65af83dadc" dependencies = [ - "windows-sys 0.60.2", + "windows-sys 0.61.2", ] [[package]] @@ -152,7 +152,7 @@ checksum = "291e6a250ff86cd4a820112fb8898808a366d8f9f58ce16d1f538353ad55747d" dependencies = [ "anstyle", "once_cell_polyfill", - "windows-sys 0.60.2", + "windows-sys 0.61.2", ] [[package]] @@ -181,9 +181,9 @@ dependencies = [ [[package]] name = "arc-swap" -version = "1.9.1" +version = "1.9.2" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "6a3a1fd6f75306b68087b831f025c712524bcb19aad54e557b1129cfa0a2b207" +checksum = "c049c0be4daef0b145cb3555416b3b8ef5b7888a38aea1a3a155801fe7b0810b" dependencies = [ "rustversion", ] @@ -1110,9 +1110,9 @@ dependencies = [ [[package]] name = "aws-smithy-types" -version = "1.6.3" +version = "1.6.4" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "8f94d16e797ec62cd999fc9d5942b48fa7050c3093ddadff48e4d7528d16fcb9" +checksum = "0b791f3ac597193fe1d08b82366986eb1f5bc31f2ac6c194c0855276116c76cd" dependencies = [ "base64-simd", "bytes", @@ -1819,7 +1819,7 @@ version = "3.1.1" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "faf9468729b8cbcea668e36183cb69d317348c2e08e994829fb56ebfdfbaac34" dependencies = [ - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -3087,7 +3087,7 @@ dependencies = [ "libc", "option-ext", "redox_users", - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -3315,7 +3315,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb" dependencies = [ "libc", - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -3588,8 +3588,8 @@ checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c" [[package]] name = "fsst" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "rand 0.9.5", @@ -4538,7 +4538,7 @@ dependencies = [ "libc", "percent-encoding", "pin-project-lite", - "socket2 0.5.10", + "socket2 0.6.3", "system-configuration", "tokio", "tower-service", @@ -4880,7 +4880,7 @@ checksum = "3640c1c38b8e4e43584d8df18be5fc6b0aa314ce6ebf51b53313d4306cca8e46" dependencies = [ "hermit-abi", "libc", - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -5143,8 +5143,8 @@ checksum = "e037a2e1d8d5fdbd49b16a4ea09d5d6401c1f29eca5ff29d03d3824dba16256a" [[package]] name = "lance" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arc-swap", "arrow", @@ -5193,7 +5193,7 @@ dependencies = [ "lance-tokenizer", "log", "moka", - "object_store 0.14.1", + "object_store 0.14.2", "permutation", "pin-project", "prost", @@ -5217,8 +5217,8 @@ dependencies = [ [[package]] name = "lance-arrow" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-buffer", @@ -5240,7 +5240,7 @@ dependencies = [ [[package]] name = "lance-arrow-scalar" version = "58.0.0" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-buffer", @@ -5254,7 +5254,7 @@ dependencies = [ [[package]] name = "lance-arrow-stats" version = "58.0.0" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-schema", @@ -5263,8 +5263,8 @@ dependencies = [ [[package]] name = "lance-bitpacking" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrayref", "crunchy", @@ -5274,8 +5274,8 @@ dependencies = [ [[package]] name = "lance-core" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-buffer", @@ -5294,7 +5294,7 @@ dependencies = [ "log", "moka", "num_cpus", - "object_store 0.14.1", + "object_store 0.14.2", "pin-project", "prost", "quick_cache", @@ -5312,8 +5312,8 @@ dependencies = [ [[package]] name = "lance-datafusion" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow", "arrow-array", @@ -5344,8 +5344,8 @@ dependencies = [ [[package]] name = "lance-datagen" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow", "arrow-array", @@ -5362,8 +5362,8 @@ dependencies = [ [[package]] name = "lance-derive" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "proc-macro2", "quote", @@ -5372,8 +5372,8 @@ dependencies = [ [[package]] name = "lance-encoding" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-arith", "arrow-array", @@ -5406,8 +5406,8 @@ dependencies = [ [[package]] name = "lance-file" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-arith", "arrow-array", @@ -5428,7 +5428,7 @@ dependencies = [ "lance-io", "log", "num-traits", - "object_store 0.14.1", + "object_store 0.14.2", "prost", "prost-build", "prost-types", @@ -5439,8 +5439,8 @@ dependencies = [ [[package]] name = "lance-geo" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "datafusion", "geo-traits", @@ -5454,8 +5454,8 @@ dependencies = [ [[package]] name = "lance-index" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arc-swap", "arrow", @@ -5482,6 +5482,7 @@ dependencies = [ "geoarrow-array", "geoarrow-schema", "half", + "hashbrown 0.16.1", "hex", "itertools 0.14.0", "jieba-rs", @@ -5504,7 +5505,7 @@ dependencies = [ "log", "ndarray", "num-traits", - "object_store 0.14.1", + "object_store 0.14.2", "prost", "prost-build", "prost-types", @@ -5525,8 +5526,8 @@ dependencies = [ [[package]] name = "lance-index-core" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-schema", @@ -5549,8 +5550,8 @@ dependencies = [ [[package]] name = "lance-io" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow", "arrow-array", @@ -5570,7 +5571,7 @@ dependencies = [ "log", "metrics", "moka", - "object_store 0.14.1", + "object_store 0.14.2", "object_store_opendal", "opendal", "opendal-service-goosefs", @@ -5592,8 +5593,8 @@ dependencies = [ [[package]] name = "lance-linalg" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-schema", @@ -5607,8 +5608,8 @@ dependencies = [ [[package]] name = "lance-namespace" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow", "async-trait", @@ -5622,9 +5623,10 @@ dependencies = [ [[package]] name = "lance-namespace-impls" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ + "arc-swap", "arrow", "arrow-ipc", "arrow-schema", @@ -5645,7 +5647,7 @@ dependencies = [ "lance-namespace", "lance-table", "log", - "object_store 0.14.1", + "object_store 0.14.2", "quick-xml 0.40.1", "rand 0.9.5", "reqwest 0.12.28", @@ -5662,9 +5664,9 @@ dependencies = [ [[package]] name = "lance-namespace-reqwest-client" -version = "0.12.0" +version = "0.13.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "d8d23e54b1634d5bbb434f8dd33dc3c05f6e58d876a9a27b3b4aef58ddbe11af" +checksum = "828f291764feda98898452bea4ce3756b73e5d5c7c67b62edda27456b7cc72c1" dependencies = [ "reqwest 0.12.28", "serde", @@ -5676,8 +5678,8 @@ dependencies = [ [[package]] name = "lance-select" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-buffer", @@ -5691,8 +5693,8 @@ dependencies = [ [[package]] name = "lance-table" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow", "arrow-array", @@ -5713,7 +5715,7 @@ dependencies = [ "lance-io", "lance-select", "log", - "object_store 0.14.1", + "object_store 0.14.2", "prost", "prost-build", "prost-types", @@ -5732,8 +5734,8 @@ dependencies = [ [[package]] name = "lance-testing" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "arrow-array", "arrow-schema", @@ -5746,8 +5748,8 @@ dependencies = [ [[package]] name = "lance-tokenizer" -version = "13.0.0-beta.15" -source = "git+https://github.com/lance-format/lance.git?tag=v13.0.0-beta.15#db53dcf367088b32ae3a7461407d43593759ae3e" +version = "13.1.0-beta.1" +source = "git+https://github.com/lance-format/lance.git?tag=v13.1.0-beta.1#e5a3553b1699a7062ff7afe4ee268d596bbf2e0d" dependencies = [ "frostem", "icu_segmenter", @@ -5760,7 +5762,7 @@ dependencies = [ [[package]] name = "lancedb" -version = "0.40.0-beta.11" +version = "0.40.0-beta.12" dependencies = [ "ahash", "anyhow", @@ -5824,7 +5826,7 @@ dependencies = [ "moka", "num-traits", "oauth2", - "object_store 0.14.1", + "object_store 0.14.2", "pin-project", "polars", "polars-arrow", @@ -5857,7 +5859,7 @@ dependencies = [ [[package]] name = "lancedb-nodejs" -version = "0.40.0-beta.11" +version = "0.40.0-beta.12" dependencies = [ "arrow-array", "arrow-buffer", @@ -5882,7 +5884,7 @@ dependencies = [ [[package]] name = "lancedb-python" -version = "0.40.0-beta.11" +version = "0.40.0-beta.12" dependencies = [ "arc-swap", "arrow", @@ -6473,9 +6475,9 @@ dependencies = [ [[package]] name = "napi" -version = "3.12.2" +version = "3.12.5" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "58c5f4d5375213fdb7be2655e152386e82f026f9a5ba36a75556e11359aafe09" +checksum = "f0c007d4a8ead952a81887661d41fd56b7e7a70d3d4d921f445fb37a8f6efa1e" dependencies = [ "bitflags 2.11.1", "chrono", @@ -6493,15 +6495,15 @@ dependencies = [ [[package]] name = "napi-build" -version = "2.4.2" +version = "2.5.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "860e7c40864f95cfb83cde99f9ebadd88ef3d9bdccd7dd2cee0cc96a2dd4ffa7" +checksum = "0941983559cd62bfa7f04646a16428967a847365adeef27711e4e3927046bbed" [[package]] name = "napi-derive" -version = "3.6.5" +version = "3.6.9" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "350057056a30368aa76c11a0d406b0aa61321710be2c36656ab4f6efe0b785a2" +checksum = "6b1d232c24dc6754979b27efdf2047a5849efc7c92c1245de87e6bedbb9f5c3c" dependencies = [ "convert_case", "ctor", @@ -6513,9 +6515,9 @@ dependencies = [ [[package]] name = "napi-derive-backend" -version = "6.1.3" +version = "6.1.4" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "4c1c87a71568f3fe5c736b10878ff55064b250bb4ea4b48c8055713e07b9e463" +checksum = "fd6856fe1686867da90ed0dc292d7c0c87a695eb7c578b913360f33619ee01e3" dependencies = [ "convert_case", "proc-macro2", @@ -6526,9 +6528,9 @@ dependencies = [ [[package]] name = "napi-sys" -version = "3.3.0" +version = "3.4.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "85fbf1fa9f1babfe396d74bbbf52b3643770243e8f5b0b46715d4caf7f0dfc9a" +checksum = "e22a4f25c16a5c5411d987cd6fbd48313522ae1789ade2d6dd3efdc6d40a0fc8" dependencies = [ "libloading", ] @@ -6635,7 +6637,7 @@ version = "0.50.3" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "7957b9740744892f114936ab4a57b3f487491bbeafaf8083688b16841a4240e5" dependencies = [ - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -6764,7 +6766,7 @@ version = "5.0.0" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "51e219e79014df21a225b1860a479e2dcd7cbd9130f4defd4bd0e191ea31d67d" dependencies = [ - "base64 0.21.7", + "base64 0.22.1", "chrono", "getrandom 0.2.17", "http 1.5.0", @@ -6881,13 +6883,13 @@ dependencies = [ [[package]] name = "object_store" -version = "0.14.1" +version = "0.14.2" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "d354792e39fa5f0009e47623cf8b15b099bf9a652fa55c6f817fe28ac84fea50" +checksum = "f1796bc93603f78c5760a69f2d58badc9618d22adade0a95385bb2adbae4eb94" dependencies = [ "async-trait", "aws-lc-rs", - "base64 0.22.1", + "base64 0.23.1", "bytes", "chrono", "crc-fast", @@ -6933,7 +6935,7 @@ dependencies = [ "bytes", "chrono", "futures", - "object_store 0.14.1", + "object_store 0.14.2", "opendal", "pin-project", "tokio", @@ -8138,8 +8140,8 @@ version = "0.14.3" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "343d3bd7056eda839b03204e68deff7d1b13aba7af2b2fd16890697274262ee7" dependencies = [ - "heck 0.4.1", - "itertools 0.11.0", + "heck 0.5.0", + "itertools 0.14.0", "log", "multimap", "petgraph", @@ -8158,7 +8160,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "b570b25f7617e43d59005d0990ccb79e950a423952cea19671b7a876da390adf" dependencies = [ "anyhow", - "itertools 0.11.0", + "itertools 0.14.0", "proc-macro2", "quote", "syn 2.0.117", @@ -8380,7 +8382,7 @@ dependencies = [ "quinn-udp", "rustc-hash", "rustls 0.23.45", - "socket2 0.5.10", + "socket2 0.6.3", "thiserror 2.0.18", "tokio", "tracing", @@ -8418,7 +8420,7 @@ dependencies = [ "cfg_aliases", "libc", "once_cell", - "socket2 0.5.10", + "socket2 0.6.3", "tracing", "windows-sys 0.60.2", ] @@ -9218,7 +9220,7 @@ dependencies = [ "errno", "libc", "linux-raw-sys", - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -9289,7 +9291,7 @@ dependencies = [ "security-framework", "security-framework-sys", "webpki-root-certs", - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -9580,9 +9582,9 @@ dependencies = [ [[package]] name = "serde_with" -version = "3.23.0" +version = "3.24.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "935177bb8c0cd8ca1a4e6d1a2ac8988bea69cab4f9d3a31311e012ad27868ea4" +checksum = "df9adc193c780ef8f159aee8b61e2d5801aaa555e6eb0947fe45530ec506296f" dependencies = [ "base64 0.23.1", "bs58", @@ -9601,9 +9603,9 @@ dependencies = [ [[package]] name = "serde_with_macros" -version = "3.23.0" +version = "3.24.0" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "1d607aa01a3cb0ad757d6fd216136910db3c97b102fe686585689615a02dbcdc" +checksum = "3e17bbc68e28663bbbb90df47e058aa7eda4fb445b89fe70457bb94fbccf6e49" dependencies = [ "darling 0.24.1", "proc-macro2", @@ -9836,7 +9838,7 @@ version = "0.8.9" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "c1c97747dbf44bb1ca44a561ece23508e99cb592e862f22222dcf42f51d1e451" dependencies = [ - "heck 0.4.1", + "heck 0.5.0", "proc-macro2", "quote", "syn 2.0.117", @@ -9848,7 +9850,7 @@ version = "0.9.0" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "54254b8531cafa275c5e096f62d48c81435d1015405a91198ddb11e967301d40" dependencies = [ - "heck 0.4.1", + "heck 0.5.0", "proc-macro2", "quote", "syn 2.0.117", @@ -9871,7 +9873,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "3a766e1110788c36f4fa1c2b71b387a7815aa65f88ce0229841826633d93723e" dependencies = [ "libc", - "windows-sys 0.60.2", + "windows-sys 0.61.2", ] [[package]] @@ -9990,7 +9992,7 @@ dependencies = [ "cfg-if 1.0.4", "libc", "psm", - "windows-sys 0.60.2", + "windows-sys 0.61.2", ] [[package]] @@ -10282,7 +10284,7 @@ dependencies = [ "getrandom 0.4.2", "once_cell", "rustix", - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] @@ -11303,7 +11305,7 @@ version = "0.1.11" source = "registry+https://github.com/rust-lang/crates.io-index" checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22" dependencies = [ - "windows-sys 0.59.0", + "windows-sys 0.61.2", ] [[package]] diff --git a/Cargo.toml b/Cargo.toml index 1bf7ce86a..183e5f4f7 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -13,20 +13,20 @@ categories = ["database-implementations"] rust-version = "1.91.0" [workspace.dependencies] -lance = { "version" = "=13.0.0-beta.15", default-features = false, "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-core = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-datagen = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-file = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-io = { "version" = "=13.0.0-beta.15", default-features = false, "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-index = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-linalg = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-namespace = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-namespace-impls = { "version" = "=13.0.0-beta.15", default-features = false, "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-table = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-testing = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-datafusion = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-encoding = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } -lance-arrow = { "version" = "=13.0.0-beta.15", "tag" = "v13.0.0-beta.15", "git" = "https://github.com/lance-format/lance.git" } +lance = { "version" = "=13.1.0-beta.1", default-features = false, "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-core = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-datagen = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-file = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-io = { "version" = "=13.1.0-beta.1", default-features = false, "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-index = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-linalg = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-namespace = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-namespace-impls = { "version" = "=13.1.0-beta.1", default-features = false, "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-table = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-testing = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-datafusion = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-encoding = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } +lance-arrow = { "version" = "=13.1.0-beta.1", "tag" = "v13.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" } lancedb = { path = "rust/lancedb", default-features = false } ahash = "0.8" # Note that this one does not include pyarrow diff --git a/docs/src/java/java.md b/docs/src/java/java.md index 5946e5862..0565156dc 100644 --- a/docs/src/java/java.md +++ b/docs/src/java/java.md @@ -14,7 +14,7 @@ Add the following dependency to your `pom.xml`: com.lancedb lancedb-core - 0.40.0-beta.11 + 0.40.0-beta.12 ``` diff --git a/docs/src/js/classes/AutoQuery.md b/docs/src/js/classes/AutoQuery.md index 1d7ea6952..79d666124 100644 --- a/docs/src/js/classes/AutoQuery.md +++ b/docs/src/js/classes/AutoQuery.md @@ -350,7 +350,7 @@ input to this method would be: #### Parameters -* **columns**: `string` \| `string`[] \| `Record`<`string`, `string`> \| `Map`<`string`, `string`> +* **columns**: `string` \| `string`[] \| `Map`<`string`, `string`> \| `Record`<`string`, `string`> #### Returns diff --git a/docs/src/js/classes/BlobFile.md b/docs/src/js/classes/BlobFile.md index 84a596d1a..ee012be25 100644 --- a/docs/src/js/classes/BlobFile.md +++ b/docs/src/js/classes/BlobFile.md @@ -10,16 +10,52 @@ A lazy handle to blob bytes. Create one with [Table.fetchBlobFiles](Table.md#fet ## Methods +### close() + +```ts +close(): Promise +``` + +Releases the handle. Reads after `close()` fail. Calling it again does +nothing. + +#### Returns + +`Promise`<`void`> + +*** + +### isClosed() + +```ts +isClosed(): boolean +``` + +Returns true after [BlobFile.close](BlobFile.md#close). + +#### Returns + +`boolean` + +*** + ### read() ```ts -read(): Promise +read(maxBytes?, options?): Promise ``` -Reads from the cursor to the end and advances the cursor. +Reads from the cursor and advances the cursor. -A second call returns an empty buffer. [BlobFile.readRange](BlobFile.md#readrange) does -not move the cursor. +Reads to the end when `maxBytes` is omitted, or at most `maxBytes` bytes +otherwise. Returns an empty buffer at the end of the blob. +[BlobFile.readRange](BlobFile.md#readrange) does not move the cursor. + +#### Parameters + +* **maxBytes?**: `bigint` + +* **options?**: [`BlobReadOptions`](../type-aliases/BlobReadOptions.md) #### Returns @@ -30,7 +66,10 @@ not move the cursor. ### readRange() ```ts -readRange(start, end): Promise +readRange( + start, + end, + options?): Promise ``` Reads the half-open byte range `[start, end)`. @@ -43,12 +82,55 @@ Fails when `end` is past the blob size. Does not move the cursor. * **end**: `bigint` +* **options?**: [`BlobReadOptions`](../type-aliases/BlobReadOptions.md) + #### Returns `Promise`<`Buffer`> *** +### readRanges() + +```ts +readRanges(ranges, options?): Promise +``` + +Reads several half-open byte ranges. Returns one buffer per range, in +the order given. + +Fails when any `end` is past the blob size. Does not move the cursor. + +#### Parameters + +* **ranges**: [`BlobRange`](../type-aliases/BlobRange.md)[] + +* **options?**: [`BlobReadOptions`](../type-aliases/BlobReadOptions.md) + +#### Returns + +`Promise`<`Buffer`[]> + +*** + +### seek() + +```ts +seek(position): Promise +``` + +Moves the cursor to `position`, in bytes from the start of the blob. + +#### Parameters + +* **position**: `bigint` + +#### Returns + +`Promise`<`void`> + +*** + ### size() ```ts @@ -60,3 +142,17 @@ Returns the blob size in bytes. #### Returns `bigint` + +*** + +### tell() + +```ts +tell(): Promise +``` + +Returns the cursor position in bytes. + +#### Returns + +`Promise`<`bigint`> diff --git a/docs/src/js/classes/Query.md b/docs/src/js/classes/Query.md index 6ebaebd75..0abf49755 100644 --- a/docs/src/js/classes/Query.md +++ b/docs/src/js/classes/Query.md @@ -424,7 +424,7 @@ input to this method would be: #### Parameters -* **columns**: `string` \| `string`[] \| `Record`<`string`, `string`> \| `Map`<`string`, `string`> +* **columns**: `string` \| `string`[] \| `Map`<`string`, `string`> \| `Record`<`string`, `string`> #### Returns diff --git a/docs/src/js/classes/QueryBase.md b/docs/src/js/classes/QueryBase.md index 35c071525..664337bad 100644 --- a/docs/src/js/classes/QueryBase.md +++ b/docs/src/js/classes/QueryBase.md @@ -193,7 +193,7 @@ input to this method would be: #### Parameters -* **columns**: `string` \| `string`[] \| `Record`<`string`, `string`> \| `Map`<`string`, `string`> +* **columns**: `string` \| `string`[] \| `Map`<`string`, `string`> \| `Record`<`string`, `string`> #### Returns diff --git a/docs/src/js/classes/Table.md b/docs/src/js/classes/Table.md index 17e2d4f34..ac16b8a81 100644 --- a/docs/src/js/classes/Table.md +++ b/docs/src/js/classes/Table.md @@ -516,7 +516,10 @@ Drop an index from the table. ### fetchBlobFiles() ```ts -abstract fetchBlobFiles(column, rowIds): Promise<(null | BlobFile)[]> +abstract fetchBlobFiles( + column, + rowIds, + options?): Promise<(null | BlobFile)[]> ``` Opens lazy blob handles for `column` at the given row IDs using the @@ -531,6 +534,8 @@ See [Table.fetchBlobs](Table.md#fetchblobs) for row-ID validity across versions. * **rowIds**: readonly (`number` \| `bigint`)[] +* **options?**: [`BlobReadOptions`](../type-aliases/BlobReadOptions.md) + #### Returns `Promise`<(`null` \| [`BlobFile`](BlobFile.md))[]> @@ -540,7 +545,10 @@ See [Table.fetchBlobs](Table.md#fetchblobs) for row-ID validity across versions. ### fetchBlobs() ```ts -abstract fetchBlobs(column, rowIds): Promise<(null | Buffer)[]> +abstract fetchBlobs( + column, + rowIds, + options?): Promise<(null | Buffer)[]> ``` Bytes for `column` at row IDs from [Query.withRowId](Query.md#withrowid). @@ -559,6 +567,8 @@ memory; use [Table.fetchBlobFiles](Table.md#fetchblobfiles) for large values. * **rowIds**: readonly (`number` \| `bigint`)[] +* **options?**: [`BlobReadOptions`](../type-aliases/BlobReadOptions.md) + #### Returns `Promise`<(`null` \| `Buffer`)[]> @@ -1371,7 +1381,7 @@ repeatedly calling this method. ##### Parameters -* **updates**: `Record`<`string`, `string`> \| `Map`<`string`, `string`> +* **updates**: `Map`<`string`, `string`> \| `Record`<`string`, `string`> the columns to update diff --git a/docs/src/js/classes/TakeQuery.md b/docs/src/js/classes/TakeQuery.md index 6ae2f9c8c..79a53f753 100644 --- a/docs/src/js/classes/TakeQuery.md +++ b/docs/src/js/classes/TakeQuery.md @@ -200,7 +200,7 @@ input to this method would be: #### Parameters -* **columns**: `string` \| `string`[] \| `Record`<`string`, `string`> \| `Map`<`string`, `string`> +* **columns**: `string` \| `string`[] \| `Map`<`string`, `string`> \| `Record`<`string`, `string`> #### Returns diff --git a/docs/src/js/classes/VectorQuery.md b/docs/src/js/classes/VectorQuery.md index f9412c76d..29749c28f 100644 --- a/docs/src/js/classes/VectorQuery.md +++ b/docs/src/js/classes/VectorQuery.md @@ -673,7 +673,7 @@ input to this method would be: #### Parameters -* **columns**: `string` \| `string`[] \| `Record`<`string`, `string`> \| `Map`<`string`, `string`> +* **columns**: `string` \| `string`[] \| `Map`<`string`, `string`> \| `Record`<`string`, `string`> #### Returns diff --git a/docs/src/js/globals.md b/docs/src/js/globals.md index 2f49578b1..edb0bacc3 100644 --- a/docs/src/js/globals.md +++ b/docs/src/js/globals.md @@ -158,8 +158,11 @@ - [AnalyzePlanDistributedMetrics](type-aliases/AnalyzePlanDistributedMetrics.md) - [BaseTokenizer](type-aliases/BaseTokenizer.md) - [BlobOptions](type-aliases/BlobOptions.md) +- [BlobRange](type-aliases/BlobRange.md) +- [BlobReadOptions](type-aliases/BlobReadOptions.md) - [Data](type-aliases/Data.md) - [DataLike](type-aliases/DataLike.md) +- [DataTypeLike](type-aliases/DataTypeLike.md) - [FieldLike](type-aliases/FieldLike.md) - [IntoSql](type-aliases/IntoSql.md) - [IntoVector](type-aliases/IntoVector.md) diff --git a/docs/src/js/interfaces/CherryPickResult.md b/docs/src/js/interfaces/CherryPickResult.md index c837ab143..fdec8eedd 100644 --- a/docs/src/js/interfaces/CherryPickResult.md +++ b/docs/src/js/interfaces/CherryPickResult.md @@ -38,9 +38,9 @@ preview: CherryPickPreview; ```ts status: - | "failed" - | "unknown" | "ready" + | "failed" | "notImplemented" - | "cherryPicked"; + | "cherryPicked" + | "unknown"; ``` diff --git a/docs/src/js/interfaces/DropNamespaceOptions.md b/docs/src/js/interfaces/DropNamespaceOptions.md index 1863bb0cd..6e5f01d34 100644 --- a/docs/src/js/interfaces/DropNamespaceOptions.md +++ b/docs/src/js/interfaces/DropNamespaceOptions.md @@ -21,7 +21,7 @@ Refuse to drop if non-empty (restrict) or drop recursively (cascade). ### mode? ```ts -optional mode: "fail" | "skip"; +optional mode: "skip" | "fail"; ``` Whether to skip if the namespace doesn't exist, or fail. diff --git a/docs/src/js/namespaces/embedding/classes/EmbeddingFunction.md b/docs/src/js/namespaces/embedding/classes/EmbeddingFunction.md index 574a0d71f..b6291e0be 100644 --- a/docs/src/js/namespaces/embedding/classes/EmbeddingFunction.md +++ b/docs/src/js/namespaces/embedding/classes/EmbeddingFunction.md @@ -52,7 +52,7 @@ new EmbeddingFunction(): EmbeddingFunction ### computeQueryEmbeddings() ```ts -computeQueryEmbeddings(data): Promise +computeQueryEmbeddings(data): Promise ``` Compute the embeddings for a single query @@ -63,7 +63,7 @@ Compute the embeddings for a single query #### Returns -`Promise`<`number`[] \| `Uint8Array` \| `Float32Array` \| `Float64Array`> +`Promise`<`Float32Array` \| `Float64Array` \| `Uint8Array` \| `number`[]> *** diff --git a/docs/src/js/namespaces/embedding/classes/TextEmbeddingFunction.md b/docs/src/js/namespaces/embedding/classes/TextEmbeddingFunction.md index 444c4c3f0..0a0cc619d 100644 --- a/docs/src/js/namespaces/embedding/classes/TextEmbeddingFunction.md +++ b/docs/src/js/namespaces/embedding/classes/TextEmbeddingFunction.md @@ -37,7 +37,7 @@ new TextEmbeddingFunction(): TextEmbeddingFunction ### computeQueryEmbeddings() ```ts -computeQueryEmbeddings(data): Promise +computeQueryEmbeddings(data): Promise ``` Compute the embeddings for a single query @@ -48,7 +48,7 @@ Compute the embeddings for a single query #### Returns -`Promise`<`number`[] \| `Uint8Array` \| `Float32Array` \| `Float64Array`> +`Promise`<`Float32Array` \| `Float64Array` \| `Uint8Array` \| `number`[]> #### Overrides diff --git a/docs/src/js/type-aliases/BlobRange.md b/docs/src/js/type-aliases/BlobRange.md new file mode 100644 index 000000000..85cd48242 --- /dev/null +++ b/docs/src/js/type-aliases/BlobRange.md @@ -0,0 +1,27 @@ +[**@lancedb/lancedb**](../README.md) • **Docs** + +*** + +[@lancedb/lancedb](../globals.md) / BlobRange + +# Type Alias: BlobRange + +```ts +type BlobRange: object; +``` + +A half-open byte range `[start, end)` for [BlobFile.readRanges](../classes/BlobFile.md#readranges). + +## Type declaration + +### end + +```ts +end: bigint; +``` + +### start + +```ts +start: bigint; +``` diff --git a/docs/src/js/type-aliases/BlobReadOptions.md b/docs/src/js/type-aliases/BlobReadOptions.md new file mode 100644 index 000000000..17264b0e4 --- /dev/null +++ b/docs/src/js/type-aliases/BlobReadOptions.md @@ -0,0 +1,24 @@ +[**@lancedb/lancedb**](../README.md) • **Docs** + +*** + +[@lancedb/lancedb](../globals.md) / BlobReadOptions + +# Type Alias: BlobReadOptions + +```ts +type BlobReadOptions: object; +``` + +Options for blob reads. + +## Type declaration + +### signal? + +```ts +optional signal: AbortSignal; +``` + +Cancels the read. The call rejects with `signal.reason`, and the native +read stops, including in-flight requests to a remote table. diff --git a/docs/src/js/type-aliases/DataTypeLike.md b/docs/src/js/type-aliases/DataTypeLike.md new file mode 100644 index 000000000..1f0ee6b81 --- /dev/null +++ b/docs/src/js/type-aliases/DataTypeLike.md @@ -0,0 +1,18 @@ +[**@lancedb/lancedb**](../README.md) • **Docs** + +*** + +[@lancedb/lancedb](../globals.md) / DataTypeLike + +# Type Alias: DataTypeLike + +```ts +type DataTypeLike: DataType | object; +``` + +A `DataType` from any copy or version of apache-arrow. + +Arrow 21 brands its classes with `unique symbol` properties, so a type +object from a second copy of the library no longer satisfies the `DataType` +type of this one even though it is structurally identical. Inputs that only +need to be sanitized accept this looser shape instead. diff --git a/java/lancedb-core/pom.xml b/java/lancedb-core/pom.xml index b7cdb9e41..77635c74e 100644 --- a/java/lancedb-core/pom.xml +++ b/java/lancedb-core/pom.xml @@ -8,7 +8,7 @@ com.lancedb lancedb-parent - 0.40.0-beta.11 + 0.40.0-beta.12 ../pom.xml diff --git a/java/pom.xml b/java/pom.xml index 752dacefd..9d7249f92 100644 --- a/java/pom.xml +++ b/java/pom.xml @@ -6,7 +6,7 @@ com.lancedb lancedb-parent - 0.40.0-beta.11 + 0.40.0-beta.12 pom ${project.artifactId} LanceDB Java SDK Parent POM @@ -28,7 +28,7 @@ UTF-8 15.0.0 - 13.0.0-beta.14 + 13.1.0-beta.1 false 2.30.0 1.7 diff --git a/nodejs/Cargo.toml b/nodejs/Cargo.toml index 87d4973d2..051e16474 100644 --- a/nodejs/Cargo.toml +++ b/nodejs/Cargo.toml @@ -1,7 +1,7 @@ [package] name = "lancedb-nodejs" edition.workspace = true -version = "0.40.0-beta.11" +version = "0.40.0-beta.12" publish = false license.workspace = true description.workspace = true diff --git a/nodejs/NODEJS_THIRD_PARTY_LICENSES.md b/nodejs/NODEJS_THIRD_PARTY_LICENSES.md index c0881f61e..550d2888c 100644 --- a/nodejs/NODEJS_THIRD_PARTY_LICENSES.md +++ b/nodejs/NODEJS_THIRD_PARTY_LICENSES.md @@ -275,6 +275,7 @@ [@types/node@20.16.10](https://github.com/DefinitelyTyped/DefinitelyTyped) - MIT [@types/node@20.17.9](https://github.com/DefinitelyTyped/DefinitelyTyped) - MIT [@types/node@22.7.4](https://github.com/DefinitelyTyped/DefinitelyTyped) - MIT +[@types/node@25.9.8](https://github.com/DefinitelyTyped/DefinitelyTyped) - MIT [@types/semver@7.5.6](https://github.com/DefinitelyTyped/DefinitelyTyped) - MIT [@types/stack-utils@2.0.3](https://github.com/DefinitelyTyped/DefinitelyTyped) - MIT [@types/tmp@0.2.6](https://github.com/DefinitelyTyped/DefinitelyTyped) - MIT @@ -306,6 +307,9 @@ [apache-arrow@16.0.0](https://github.com/apache/arrow) - Apache-2.0 [apache-arrow@17.0.0](https://github.com/apache/arrow) - Apache-2.0 [apache-arrow@18.0.0](https://github.com/apache/arrow) - Apache-2.0 +[apache-arrow@19.0.1](https://github.com/apache/arrow) - Apache-2.0 +[apache-arrow@20.0.0](https://github.com/apache/arrow) - Apache-2.0 +[apache-arrow@21.2.0](https://github.com/apache/arrow-js) - Apache-2.0 [argparse@1.0.10](https://github.com/nodeca/argparse) - MIT [argparse@2.0.1](https://github.com/nodeca/argparse) - Python-2.0 [array-back@3.1.0](https://github.com/75lb/array-back) - MIT @@ -348,6 +352,7 @@ [color@4.2.3](https://github.com/Qix-/color) - MIT [combined-stream@1.0.8](https://github.com/felixge/node-combined-stream) - MIT [command-line-args@5.2.1](https://github.com/75lb/command-line-args) - MIT +[command-line-args@6.0.2](https://github.com/75lb/command-line-args) - MIT [command-line-usage@7.0.1](https://github.com/75lb/command-line-usage) - MIT [concat-map@0.0.1](https://github.com/substack/node-concat-map) - MIT [convert-source-map@2.0.0](https://github.com/thlorenz/convert-source-map) - MIT @@ -398,12 +403,14 @@ [file-entry-cache@6.0.1](https://github.com/royriojas/file-entry-cache) - MIT [fill-range@7.1.1](https://github.com/jonschlinkert/fill-range) - MIT [find-replace@3.0.0](https://github.com/75lb/find-replace) - MIT +[find-replace@5.0.2](https://github.com/75lb/find-replace) - MIT [find-up@4.1.0](https://github.com/sindresorhus/find-up) - MIT [find-up@5.0.0](https://github.com/sindresorhus/find-up) - MIT [flat-cache@3.2.0](https://github.com/jaredwray/flat-cache) - MIT [flatbuffers@1.12.0](https://github.com/google/flatbuffers) - Apache* [flatbuffers@23.5.26](https://github.com/google/flatbuffers) - Apache* [flatbuffers@24.3.25](https://github.com/google/flatbuffers) - Apache-2.0 +[flatbuffers@25.9.23](https://github.com/google/flatbuffers) - Apache-2.0 [flatted@3.2.9](https://github.com/WebReflection/flatted) - ISC [follow-redirects@1.15.6](https://github.com/follow-redirects/follow-redirects) - MIT [foreground-child@3.3.0](https://github.com/tapjs/foreground-child) - ISC @@ -644,6 +651,7 @@ [uc.micro@2.1.0](https://github.com/markdown-it/uc.micro) - MIT [undici-types@5.26.5](https://github.com/nodejs/undici) - MIT [undici-types@6.19.8](https://github.com/nodejs/undici) - MIT +[undici-types@7.24.6](https://github.com/nodejs/undici) - MIT [update-browserslist-db@1.0.13](https://github.com/browserslist/update-db) - MIT [uri-js@4.4.1](https://github.com/garycourt/uri-js) - BSD-2-Clause [uuid@9.0.1](https://github.com/uuidjs/uuid) - MIT diff --git a/nodejs/__test__/arrow.test.ts b/nodejs/__test__/arrow.test.ts index df02b4936..e054a5495 100644 --- a/nodejs/__test__/arrow.test.ts +++ b/nodejs/__test__/arrow.test.ts @@ -2,10 +2,8 @@ // SPDX-FileCopyrightText: Copyright The LanceDB Authors import * as fs from "node:fs"; import * as vm from "node:vm"; -import * as arrow15 from "apache-arrow-15"; -import * as arrow16 from "apache-arrow-16"; -import * as arrow17 from "apache-arrow-17"; -import * as arrow18 from "apache-arrow-18"; + +import { type ApacheArrow, arrow15, arrowVersions } from "./arrow_versions"; import { Field as CurrentField, @@ -110,1541 +108,1502 @@ it("preserves field metadata from a provided schema", async function () { expect(roundTripped.schema.fields[0].metadata).toEqual(jsonMetadata); }); -describe.each([arrow15, arrow16, arrow17, arrow18])( - "Arrow", - ( - arrow: typeof arrow15 | typeof arrow16 | typeof arrow17 | typeof arrow18, - ) => { - type ApacheArrow = - | typeof arrow15 - | typeof arrow16 - | typeof arrow17 - | typeof arrow18; - const { - Schema, - Field, - Binary, - Bool, - Utf8, - Float64, - Struct, - List, - Map_, - Int16, - Int32, - Int64, - Float, - Float16, - Float32, - FixedSizeList, - Precision, - tableFromIPC, - DataType, - Dictionary, - RecordBatch: ArrowRecordBatch, - Table: ArrowTable, - Uint8: ArrowUint8, - makeData: arrowMakeData, - vectorFromArray, +describe.each(arrowVersions)("Arrow %#", (arrow: ApacheArrow) => { + const { + Schema, + Field, + Binary, + Bool, + Utf8, + Float64, + Struct, + List, + Map_, + Int16, + Int32, + Int64, + Float, + Float16, + Float32, + FixedSizeList, + Precision, + tableFromIPC, + DataType, + Dictionary, + RecordBatch: ArrowRecordBatch, + Table: ArrowTable, + Uint8: ArrowUint8, + makeData: arrowMakeData, + vectorFromArray, + // biome-ignore lint/suspicious/noExplicitAny: + } = arrow; + type Schema = ApacheArrow["Schema"]; + type Table = ApacheArrow["Table"]; + + function expectValidMapField( + // biome-ignore lint/suspicious/noExplicitAny: Arrow Field types vary across supported versions + field: any, + ): void { + expect(DataType.isMap(field.type)).toBe(true); + expect(field.type.keysSorted).toBe(true); + expect(field.type.children).toHaveLength(1); + + const entries = field.type.children[0]; + expect(entries.name).toBe("entries"); + expect(entries.nullable).toBe(false); + expect(DataType.isStruct(entries.type)).toBe(true); + expect(entries.type.children).toHaveLength(2); + + const [key, value] = entries.type.children; + expect([key.name, value.name]).toEqual(["key", "value"]); + expect(key.nullable).toBe(false); + expect(DataType.isUtf8(key.type)).toBe(true); + expect(value.nullable).toBe(true); + expect(DataType.isInt(value.type)).toBe(true); + expect(value.type.bitWidth).toBe(32); + expect(value.type.isSigned).toBe(true); + } + + // Helper method to verify various ways to create a table + async function checkTableCreation( + tableCreationMethod: ( + records: Record[], + recordsReversed: Record[], + schema: Schema, + ) => Promise, + infersTypes: boolean, + ): Promise { + const records = sampleRecords(); + const recordsReversed = [ + { + list: ["anime", "action", "comedy"], + struct: { x: 0, y: 0 }, + string: "hello", + number: 7, + boolean: false, + binary: Buffer.alloc(5), + }, + ]; + const schema = new Schema([ + new Field("binary", new Binary(), false), + new Field("boolean", new Bool(), false), + new Field("number", new Float64(), false), + new Field("string", new Utf8(), false), + new Field( + "struct", + new Struct([ + new Field("x", new Float64(), false), + new Field("y", new Float64(), false), + ]), + ), + new Field("list", new List(new Field("item", new Utf8(), false)), false), + ]); + const table = (await tableCreationMethod( + records, + recordsReversed, + schema, // biome-ignore lint/suspicious/noExplicitAny: - } = arrow; - type Schema = ApacheArrow["Schema"]; - type Table = ApacheArrow["Table"]; + )) as any; - function expectValidMapField( - // biome-ignore lint/suspicious/noExplicitAny: Arrow Field types vary across supported versions - field: any, - ): void { - expect(DataType.isMap(field.type)).toBe(true); - expect(field.type.keysSorted).toBe(true); - expect(field.type.children).toHaveLength(1); + // We expect deterministic ordering of the fields + expect(table.schema.names).toEqual(schema.names); - const entries = field.type.children[0]; - expect(entries.name).toBe("entries"); - expect(entries.nullable).toBe(false); - expect(DataType.isStruct(entries.type)).toBe(true); - expect(entries.type.children).toHaveLength(2); + schema.fields.forEach( + ( + // biome-ignore lint/suspicious/noExplicitAny: + field: { name: any; type: { toString: () => any } }, + idx: string | number, + ) => { + const actualField = table.schema.fields[idx]; + // Type inference always assumes nullable=true + if (infersTypes) { + expect(actualField.nullable).toBe(true); + } else { + expect(actualField.nullable).toBe(false); + } + expect(table.getChild(field.name)?.type.toString()).toEqual( + field.type.toString(), + ); + expect(table.getChildAt(idx)?.type.toString()).toEqual( + field.type.toString(), + ); + }, + ); + } - const [key, value] = entries.type.children; - expect([key.name, value.name]).toEqual(["key", "value"]); - expect(key.nullable).toBe(false); - expect(DataType.isUtf8(key.type)).toBe(true); - expect(value.nullable).toBe(true); - expect(DataType.isInt(value.type)).toBe(true); - expect(value.type.bitWidth).toBe(32); - expect(value.type.isSigned).toBe(true); - } - - // Helper method to verify various ways to create a table - async function checkTableCreation( - tableCreationMethod: ( - records: Record[], - recordsReversed: Record[], - schema: Schema, - ) => Promise
, - infersTypes: boolean, - ): Promise { - const records = sampleRecords(); - const recordsReversed = [ - { - list: ["anime", "action", "comedy"], - struct: { x: 0, y: 0 }, - string: "hello", - number: 7, - boolean: false, - binary: Buffer.alloc(5), - }, + describe("The function makeArrowTable", function () { + it("accepts snake_case embedding metadata like camelCase", function () { + const spellings = [ + // biome-ignore lint/style/useNamingConvention: the Python wire spelling + { source_column: "text", vector_column: "vector" }, + { sourceColumn: "text", vectorColumn: "vector" }, ]; - const schema = new Schema([ - new Field("binary", new Binary(), false), - new Field("boolean", new Bool(), false), - new Field("number", new Float64(), false), - new Field("string", new Utf8(), false), - new Field( - "struct", - new Struct([ - new Field("x", new Float64(), false), - new Field("y", new Float64(), false), + for (const columns of spellings) { + const schema = new Schema( + [ + new Field("text", new Utf8(), false), + new Field( + "vector", + new FixedSizeList(3, new Field("item", new Float32(), true)), + false, + ), + ], + new Map([ + [ + "embedding_functions", + JSON.stringify([{ name: "mock", model: {}, ...columns }]), + ], ]), - ), + ); + // The vector field is non-nullable and absent from the data; only a + // recognized embedding config makes that acceptable. + const table = makeArrowTable([{ text: "hello" }], { schema }); + expect(table.numRows).toBe(1); + } + }); + + it("will use data types from a provided schema instead of inference", async function () { + const schema = new Schema([ + new Field("a", new Int32(), false), + new Field("b", new Float32(), true), + new Field("c", new FixedSizeList(3, new Field("item", new Float16()))), + new Field("d", new Int64(), true), + ]); + const table = makeArrowTable( + [ + { a: 1, b: 2, c: [1, 2, 3], d: 9 }, + { a: 4, b: 5, c: [4, 5, 6], d: 10 }, + { a: 7, b: 8, c: [7, 8, 9], d: null }, + ], + { schema }, + ); + + const buf = await fromTableToBuffer(table); + expect(buf.byteLength).toBeGreaterThan(0); + + const actual = tableFromIPC(buf); + expect(actual.numRows).toBe(3); + const actualSchema = actual.schema; + expect(actualSchema).toEqual(schema); + expect(table.getChild("a")?.toJSON()).toEqual([1, 4, 7]); + expect(table.getChild("b")?.toJSON()).toEqual([2, 5, 8]); + expect(table.getChild("d")?.toJSON()).toEqual([9n, 10n, null]); + }); + + it("will use a provided FixedSizeList schema with typed array values", function () { + const schema = new Schema([ + new Field("text", new Utf8(), false), new Field( - "list", - new List(new Field("item", new Utf8(), false)), + "vector", + new FixedSizeList(3, new Field("item", new Float32(), false)), false, ), ]); - const table = (await tableCreationMethod( - records, - recordsReversed, - schema, - // biome-ignore lint/suspicious/noExplicitAny: - )) as any; - // We expect deterministic ordering of the fields - expect(table.schema.names).toEqual(schema.names); + const table = makeArrowTable( + [ + { + text: "foo", + vector: new Float32Array([1, 2, 3]), + }, + ], + { schema }, + ); - schema.fields.forEach( - ( - // biome-ignore lint/suspicious/noExplicitAny: - field: { name: any; type: { toString: () => any } }, - idx: string | number, - ) => { - const actualField = table.schema.fields[idx]; - // Type inference always assumes nullable=true - if (infersTypes) { - expect(actualField.nullable).toBe(true); - } else { - expect(actualField.nullable).toBe(false); - } - expect(table.getChild(field.name)?.type.toString()).toEqual( - field.type.toString(), - ); - expect(table.getChildAt(idx)?.type.toString()).toEqual( - field.type.toString(), - ); + expect(table.getChild("text")?.toJSON()).toEqual(["foo"]); + expect( + table + .getChild("vector") + ?.toJSON() + .map((value) => value.toJSON()), + ).toEqual([[1, 2, 3]]); + }); + + it("will assume the column `vector` is FixedSizeList by default", async function () { + const schema = new Schema([ + new Field("a", new Float(Precision.DOUBLE), true), + new Field("b", new Int64(), true), + new Field( + "vector", + new FixedSizeList( + 3, + new Field("item", new Float(Precision.SINGLE), true), + ), + true, + ), + ]); + const table = makeArrowTable([ + { a: 1, b: 2n, vector: [1, 2, 3] }, + { a: 4, b: 5n, vector: [4, 5, 6] }, + { a: 7, b: 8n, vector: [7, 8, 9] }, + ]); + + const buf = await fromTableToBuffer(table); + expect(buf.byteLength).toBeGreaterThan(0); + + const actual = tableFromIPC(buf); + expect(actual.numRows).toBe(3); + const actualSchema = actual.schema; + expect(actualSchema).toEqual(schema); + + expect(table.getChild("a")?.toJSON()).toEqual([1, 4, 7]); + expect(table.getChild("b")?.toJSON()).toEqual([2n, 5n, 8n]); + expect( + table + .getChild("vector") + ?.toJSON() + .map((v) => v.toJSON()), + ).toEqual([ + [1, 2, 3], + [4, 5, 6], + [7, 8, 9], + ]); + }); + + it("can support multiple vector columns", async function () { + const schema = new Schema([ + new Field("a", new Float(Precision.DOUBLE), true), + new Field("b", new Float(Precision.DOUBLE), true), + new Field( + "vec1", + new FixedSizeList(3, new Field("item", new Float16(), true)), + true, + ), + new Field( + "vec2", + new FixedSizeList(3, new Field("item", new Float64(), true)), + true, + ), + ]); + const table = makeArrowTable( + [ + { a: 1, b: 2, vec1: [1, 2, 3], vec2: [2, 4, 6] }, + { a: 4, b: 5, vec1: [4, 5, 6], vec2: [8, 10, 12] }, + { a: 7, b: 8, vec1: [7, 8, 9], vec2: [14, 16, 18] }, + ], + { + vectorColumns: { + vec1: { type: new Float16() }, + vec2: { type: new Float64() }, + }, }, ); - } - describe("The function makeArrowTable", function () { - it("accepts snake_case embedding metadata like camelCase", function () { - const spellings = [ - // biome-ignore lint/style/useNamingConvention: the Python wire spelling - { source_column: "text", vector_column: "vector" }, - { sourceColumn: "text", vectorColumn: "vector" }, - ]; - for (const columns of spellings) { - const schema = new Schema( - [ - new Field("text", new Utf8(), false), - new Field( - "vector", - new FixedSizeList(3, new Field("item", new Float32(), true)), - false, - ), - ], - new Map([ - [ - "embedding_functions", - JSON.stringify([{ name: "mock", model: {}, ...columns }]), - ], - ]), - ); - // The vector field is non-nullable and absent from the data; only a - // recognized embedding config makes that acceptable. - const table = makeArrowTable([{ text: "hello" }], { schema }); - expect(table.numRows).toBe(1); - } - }); + const buf = await fromTableToBuffer(table); + expect(buf.byteLength).toBeGreaterThan(0); - it("will use data types from a provided schema instead of inference", async function () { - const schema = new Schema([ - new Field("a", new Int32(), false), - new Field("b", new Float32(), true), - new Field( - "c", - new FixedSizeList(3, new Field("item", new Float16())), - ), - new Field("d", new Int64(), true), - ]); - const table = makeArrowTable( - [ - { a: 1, b: 2, c: [1, 2, 3], d: 9 }, - { a: 4, b: 5, c: [4, 5, 6], d: 10 }, - { a: 7, b: 8, c: [7, 8, 9], d: null }, - ], - { schema }, - ); + const actual = tableFromIPC(buf); + expect(actual.numRows).toBe(3); + const actualSchema = actual.schema; + expect(actualSchema).toEqual(schema); + }); - const buf = await fromTableToBuffer(table); - expect(buf.byteLength).toBeGreaterThan(0); + it("will detect vector columns when name contains 'vector' or 'embedding'", async function () { + // Test various naming patterns that should be detected as vector columns + const floatVectorTable = makeArrowTable([ + { + // Float vectors (use decimal values to ensure they're treated as floats) + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + user_vector: [1.1, 2.2], + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + text_embedding: [3.3, 4.4], + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + doc_embeddings: [5.5, 6.6], + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + my_vector_field: [7.7, 8.8], + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + embedding_model: [9.9, 10.1], + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + VECTOR_COL: [11.1, 12.2], // uppercase + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + Vector_Mixed: [13.3, 14.4], // mixed case + }, + ]); - const actual = tableFromIPC(buf); - expect(actual.numRows).toBe(3); - const actualSchema = actual.schema; - expect(actualSchema).toEqual(schema); - expect(table.getChild("a")?.toJSON()).toEqual([1, 4, 7]); - expect(table.getChild("b")?.toJSON()).toEqual([2, 5, 8]); - expect(table.getChild("d")?.toJSON()).toEqual([9n, 10n, null]); - }); + // Check that columns with 'vector' or 'embedding' in name are converted to FixedSizeList + const floatVectorColumns = [ + "user_vector", + "text_embedding", + "doc_embeddings", + "my_vector_field", + "embedding_model", + "VECTOR_COL", + "Vector_Mixed", + ]; - it("will use a provided FixedSizeList schema with typed array values", function () { - const schema = new Schema([ - new Field("text", new Utf8(), false), - new Field( - "vector", - new FixedSizeList(3, new Field("item", new Float32(), false)), - false, - ), - ]); - - const table = makeArrowTable( - [ - { - text: "foo", - vector: new Float32Array([1, 2, 3]), - }, - ], - { schema }, - ); - - expect(table.getChild("text")?.toJSON()).toEqual(["foo"]); + for (const columnName of floatVectorColumns) { expect( - table - .getChild("vector") - ?.toJSON() - .map((value) => value.toJSON()), - ).toEqual([[1, 2, 3]]); - }); - - it("will assume the column `vector` is FixedSizeList by default", async function () { - const schema = new Schema([ - new Field("a", new Float(Precision.DOUBLE), true), - new Field("b", new Int64(), true), - new Field( - "vector", - new FixedSizeList( - 3, - new Field("item", new Float(Precision.SINGLE), true), - ), - true, - ), - ]); - const table = makeArrowTable([ - { a: 1, b: 2n, vector: [1, 2, 3] }, - { a: 4, b: 5n, vector: [4, 5, 6] }, - { a: 7, b: 8n, vector: [7, 8, 9] }, - ]); - - const buf = await fromTableToBuffer(table); - expect(buf.byteLength).toBeGreaterThan(0); - - const actual = tableFromIPC(buf); - expect(actual.numRows).toBe(3); - const actualSchema = actual.schema; - expect(actualSchema).toEqual(schema); - - expect(table.getChild("a")?.toJSON()).toEqual([1, 4, 7]); - expect(table.getChild("b")?.toJSON()).toEqual([2n, 5n, 8n]); + DataType.isFixedSizeList(floatVectorTable.getChild(columnName)?.type), + ).toBe(true); + // Check that float vectors use Float32 by default expect( - table - .getChild("vector") - ?.toJSON() - .map((v) => v.toJSON()), - ).toEqual([ - [1, 2, 3], - [4, 5, 6], - [7, 8, 9], - ]); - }); + floatVectorTable + .getChild(columnName) + ?.type.children[0].type.toString(), + ).toEqual(new Float32().toString()); + } - it("can support multiple vector columns", async function () { - const schema = new Schema([ - new Field("a", new Float(Precision.DOUBLE), true), - new Field("b", new Float(Precision.DOUBLE), true), - new Field( - "vec1", - new FixedSizeList(3, new Field("item", new Float16(), true)), - true, - ), - new Field( - "vec2", - new FixedSizeList(3, new Field("item", new Float64(), true)), - true, - ), - ]); - const table = makeArrowTable( - [ - { a: 1, b: 2, vec1: [1, 2, 3], vec2: [2, 4, 6] }, - { a: 4, b: 5, vec1: [4, 5, 6], vec2: [8, 10, 12] }, - { a: 7, b: 8, vec1: [7, 8, 9], vec2: [14, 16, 18] }, - ], - { - vectorColumns: { - vec1: { type: new Float16() }, - vec2: { type: new Float64() }, - }, - }, - ); + // Test that regular integer arrays still get treated as float vectors + // (since JavaScript doesn't distinguish integers from floats at runtime) + const integerArrayTable = makeArrowTable([ + { + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + vector_int: [1, 2], // Regular array with integers - should be Float32 + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + embedding_int: [3, 4], // Regular array with integers - should be Float32 + }, + ]); - const buf = await fromTableToBuffer(table); - expect(buf.byteLength).toBeGreaterThan(0); - - const actual = tableFromIPC(buf); - expect(actual.numRows).toBe(3); - const actualSchema = actual.schema; - expect(actualSchema).toEqual(schema); - }); - - it("will detect vector columns when name contains 'vector' or 'embedding'", async function () { - // Test various naming patterns that should be detected as vector columns - const floatVectorTable = makeArrowTable([ - { - // Float vectors (use decimal values to ensure they're treated as floats) - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - user_vector: [1.1, 2.2], - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - text_embedding: [3.3, 4.4], - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - doc_embeddings: [5.5, 6.6], - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - my_vector_field: [7.7, 8.8], - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - embedding_model: [9.9, 10.1], - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - VECTOR_COL: [11.1, 12.2], // uppercase - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - Vector_Mixed: [13.3, 14.4], // mixed case - }, - ]); - - // Check that columns with 'vector' or 'embedding' in name are converted to FixedSizeList - const floatVectorColumns = [ - "user_vector", - "text_embedding", - "doc_embeddings", - "my_vector_field", - "embedding_model", - "VECTOR_COL", - "Vector_Mixed", - ]; - - for (const columnName of floatVectorColumns) { - expect( - DataType.isFixedSizeList( - floatVectorTable.getChild(columnName)?.type, - ), - ).toBe(true); - // Check that float vectors use Float32 by default - expect( - floatVectorTable - .getChild(columnName) - ?.type.children[0].type.toString(), - ).toEqual(new Float32().toString()); - } - - // Test that regular integer arrays still get treated as float vectors - // (since JavaScript doesn't distinguish integers from floats at runtime) - const integerArrayTable = makeArrowTable([ - { - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - vector_int: [1, 2], // Regular array with integers - should be Float32 - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - embedding_int: [3, 4], // Regular array with integers - should be Float32 - }, - ]); - - const integerArrayColumns = ["vector_int", "embedding_int"]; - - for (const columnName of integerArrayColumns) { - expect( - DataType.isFixedSizeList( - integerArrayTable.getChild(columnName)?.type, - ), - ).toBe(true); - // Regular integer arrays should use Float32 (avoiding false positives) - expect( - integerArrayTable - .getChild(columnName) - ?.type.children[0].type.toString(), - ).toEqual(new Float32().toString()); - } - - // Test normal list should NOT be converted to FixedSizeList - const normalListTable = makeArrowTable([ - { - // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns - normal_list: [15.5, 16.6], // should NOT be detected as vector - }, - ]); + const integerArrayColumns = ["vector_int", "embedding_int"]; + for (const columnName of integerArrayColumns) { expect( DataType.isFixedSizeList( - normalListTable.getChild("normal_list")?.type, + integerArrayTable.getChild(columnName)?.type, ), - ).toBe(false); - expect( - DataType.isList(normalListTable.getChild("normal_list")?.type), ).toBe(true); - }); - - it("will allow different vector column types", async function () { - const table = makeArrowTable([{ fp16: [1], fp32: [1], fp64: [1] }], { - vectorColumns: { - fp16: { type: new Float16() }, - fp32: { type: new Float32() }, - fp64: { type: new Float64() }, - }, - }); - + // Regular integer arrays should use Float32 (avoiding false positives) expect( - table.getChild("fp16")?.type.children[0].type.toString(), - ).toEqual(new Float16().toString()); - expect( - table.getChild("fp32")?.type.children[0].type.toString(), + integerArrayTable + .getChild(columnName) + ?.type.children[0].type.toString(), ).toEqual(new Float32().toString()); - expect( - table.getChild("fp64")?.type.children[0].type.toString(), - ).toEqual(new Float64().toString()); + } + + // Test normal list should NOT be converted to FixedSizeList + const normalListTable = makeArrowTable([ + { + // biome-ignore lint/style/useNamingConvention: Testing vector column detection patterns + normal_list: [15.5, 16.6], // should NOT be detected as vector + }, + ]); + + expect( + DataType.isFixedSizeList(normalListTable.getChild("normal_list")?.type), + ).toBe(false); + expect( + DataType.isList(normalListTable.getChild("normal_list")?.type), + ).toBe(true); + }); + + it("will allow different vector column types", async function () { + const table = makeArrowTable([{ fp16: [1], fp32: [1], fp64: [1] }], { + vectorColumns: { + fp16: { type: new Float16() }, + fp32: { type: new Float32() }, + fp64: { type: new Float64() }, + }, }); - it("will infer FixedSizeList from Float32Array values", async function () { - const table = makeArrowTable([ - { id: "a", vector: new Float32Array([0.1, 0.2, 0.3]) }, - { id: "b", vector: new Float32Array([0.4, 0.5, 0.6]) }, - ]); + expect(table.getChild("fp16")?.type.children[0].type.toString()).toEqual( + new Float16().toString(), + ); + expect(table.getChild("fp32")?.type.children[0].type.toString()).toEqual( + new Float32().toString(), + ); + expect(table.getChild("fp64")?.type.children[0].type.toString()).toEqual( + new Float64().toString(), + ); + }); - expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( + it("will infer FixedSizeList from Float32Array values", async function () { + const table = makeArrowTable([ + { id: "a", vector: new Float32Array([0.1, 0.2, 0.3]) }, + { id: "b", vector: new Float32Array([0.4, 0.5, 0.6]) }, + ]); + + expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( + true, + ); + const vectorType = table.getChild("vector")?.type; + expect(vectorType.listSize).toBe(3); + expect(vectorType.children[0].type.toString()).toEqual( + new Float32().toString(), + ); + }); + + it("will infer FixedSizeList from Uint8Array values", async function () { + const table = makeArrowTable([ + { id: "a", vector: new Uint8Array([1, 2, 3]) }, + { id: "b", vector: new Uint8Array([4, 5, 6]) }, + ]); + + expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( + true, + ); + const vectorType = table.getChild("vector")?.type; + expect(vectorType.listSize).toBe(3); + expect(vectorType.children[0].type.toString()).toEqual( + new ArrowUint8().toString(), + ); + }); + + it("will use dictionary encoded strings if asked", async function () { + const table = makeArrowTable([{ str: "hello" }]); + expect(DataType.isUtf8(table.getChild("str")?.type)).toBe(true); + + const tableWithDict = makeArrowTable([{ str: "hello" }], { + dictionaryEncodeStrings: true, + }); + expect(DataType.isDictionary(tableWithDict.getChild("str")?.type)).toBe( + true, + ); + + const schema = new Schema([ + new Field("str", new Dictionary(new Utf8(), new Int32())), + ]); + + const tableWithDict2 = makeArrowTable([{ str: "hello" }], { schema }); + expect(DataType.isDictionary(tableWithDict2.getChild("str")?.type)).toBe( + true, + ); + }); + + it("will infer data types correctly", async function () { + await checkTableCreation( + // biome-ignore lint/suspicious/noExplicitAny: + async (records) => (makeArrowTable)(records), + true, + ); + }); + + it("will allow matching inferred types across records", function () { + expect(() => makeArrowTable([{ value: 1 }, { value: 2 }])).not.toThrow(); + }); + + it("will reject mismatched inferred types across records", function () { + expect(() => makeArrowTable([{ value: 1 }, { value: "two" }])).toThrow( + "Failed to infer schema for data. Previously inferred type Float64 but found Utf8 for field value at row 1. Consider providing an explicit schema.", + ); + }); + + it.each([ + ["ArrayBuffer", new Uint8Array([104]).buffer], + ["URL", new URL("https://example.com/")], + ["empty object", {}], + [ + "class instance with enumerable fields", + Object.assign(new Blob(["x"]), { position: 0 }), + ], + ])("rejects %s values without a schema", (_name, value) => { + expect(() => makeArrowTable([{ id: 1, value }])).toThrow( + /field value at row 0/, + ); + expect(() => makeArrowTable([{ value }])).toThrow(/field value at row 0/); + }); + + it.each([ + ["Binary", new Binary(), new Uint8Array([104]).buffer], + ["Utf8", new Utf8(), new URL("https://example.com/")], + ])("rejects %s values with a schema", (_name, type, value) => { + const schema = new Schema([new Field("value", type, true)]); + expect(() => makeArrowTable([{ value }], { schema })).toThrow( + /field value at row 0/, + ); + }); + + it("rejects unsupported values in nested struct fields", () => { + const schema = new Schema([ + new Field( + "metadata", + new Struct([new Field("bytes", new Binary(), true)]), true, - ); - const vectorType = table.getChild("vector")?.type; - expect(vectorType.listSize).toBe(3); - expect(vectorType.children[0].type.toString()).toEqual( - new Float32().toString(), - ); - }); + ), + ]); + expect(() => + makeArrowTable( + [{ metadata: { bytes: new Uint8Array([104]).buffer } }], + { schema }, + ), + ).toThrow(/field metadata\.bytes at row 0/); + }); - it("will infer FixedSizeList from Uint8Array values", async function () { - const table = makeArrowTable([ - { id: "a", vector: new Uint8Array([1, 2, 3]) }, - { id: "b", vector: new Uint8Array([4, 5, 6]) }, - ]); - - expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( + it("rejects unsupported objects in list elements", () => { + const schema = new Schema([ + new Field( + "images", + new List(new Field("item", new Binary(), true)), true, - ); - const vectorType = table.getChild("vector")?.type; - expect(vectorType.listSize).toBe(3); - expect(vectorType.children[0].type.toString()).toEqual( - new ArrowUint8().toString(), - ); - }); + ), + ]); + expect(() => + makeArrowTable([{ images: [new Blob(["abc"])] }], { schema }), + ).toThrow(/field images\[0\] at row 0/); + }); - it("will use dictionary encoded strings if asked", async function () { - const table = makeArrowTable([{ str: "hello" }]); - expect(DataType.isUtf8(table.getChild("str")?.type)).toBe(true); - - const tableWithDict = makeArrowTable([{ str: "hello" }], { - dictionaryEncodeStrings: true, - }); - expect(DataType.isDictionary(tableWithDict.getChild("str")?.type)).toBe( - true, - ); - - const schema = new Schema([ - new Field("str", new Dictionary(new Utf8(), new Int32())), - ]); - - const tableWithDict2 = makeArrowTable([{ str: "hello" }], { schema }); - expect( - DataType.isDictionary(tableWithDict2.getChild("str")?.type), - ).toBe(true); - }); - - it("will infer data types correctly", async function () { - await checkTableCreation( - // biome-ignore lint/suspicious/noExplicitAny: - async (records) => (makeArrowTable)(records), - true, - ); - }); - - it("will allow matching inferred types across records", function () { - expect(() => - makeArrowTable([{ value: 1 }, { value: 2 }]), - ).not.toThrow(); - }); - - it("will reject mismatched inferred types across records", function () { - expect(() => makeArrowTable([{ value: 1 }, { value: "two" }])).toThrow( - "Failed to infer schema for data. Previously inferred type Float64 but found Utf8 for field value at row 1. Consider providing an explicit schema.", - ); - }); - - it.each([ - ["ArrayBuffer", new Uint8Array([104]).buffer], - ["URL", new URL("https://example.com/")], - ["empty object", {}], - [ - "class instance with enumerable fields", - Object.assign(new Blob(["x"]), { position: 0 }), - ], - ])("rejects %s values without a schema", (_name, value) => { - expect(() => makeArrowTable([{ id: 1, value }])).toThrow( - /field value at row 0/, - ); - expect(() => makeArrowTable([{ value }])).toThrow( - /field value at row 0/, - ); - }); - - it.each([ - ["Binary", new Binary(), new Uint8Array([104]).buffer], - ["Utf8", new Utf8(), new URL("https://example.com/")], - ])("rejects %s values with a schema", (_name, type, value) => { - const schema = new Schema([new Field("value", type, true)]); - expect(() => makeArrowTable([{ value }], { schema })).toThrow( - /field value at row 0/, - ); - }); - - it("rejects unsupported values in nested struct fields", () => { - const schema = new Schema([ - new Field( - "metadata", - new Struct([new Field("bytes", new Binary(), true)]), - true, - ), - ]); - expect(() => - makeArrowTable( - [{ metadata: { bytes: new Uint8Array([104]).buffer } }], - { schema }, - ), - ).toThrow(/field metadata\.bytes at row 0/); - }); - - it("rejects unsupported objects in list elements", () => { - const schema = new Schema([ - new Field( - "images", - new List(new Field("item", new Binary(), true)), - true, - ), - ]); - expect(() => - makeArrowTable([{ images: [new Blob(["abc"])] }], { schema }), - ).toThrow(/field images\[0\] at row 0/); - }); - - it("rejects unsupported objects in nested lists", () => { - const schema = new Schema([ - new Field( - "images", - new List( - new Field( - "item", - new List(new Field("item", new Binary(), true)), - true, - ), + it("rejects unsupported objects in nested lists", () => { + const schema = new Schema([ + new Field( + "images", + new List( + new Field( + "item", + new List(new Field("item", new Binary(), true)), + true, ), - true, ), - ]); - expect(() => - makeArrowTable([{ images: [[new Blob(["abc"])]] }], { schema }), - ).toThrow(/field images\[0\]\[0\] at row 0/); - }); + true, + ), + ]); + expect(() => + makeArrowTable([{ images: [[new Blob(["abc"])]] }], { schema }), + ).toThrow(/field images\[0\]\[0\] at row 0/); + }); - it("rejects unsupported nested struct fields in lists", () => { - const schema = new Schema([ - new Field( - "items", - new List( - new Field( - "item", - new Struct([new Field("bytes", new Binary(), true)]), - true, - ), + it("rejects unsupported nested struct fields in lists", () => { + const schema = new Schema([ + new Field( + "items", + new List( + new Field( + "item", + new Struct([new Field("bytes", new Binary(), true)]), + true, ), - true, ), - ]); - expect(() => - makeArrowTable( - [{ items: [{ bytes: new Uint8Array([104]).buffer }] }], - { schema }, - ), - ).toThrow(/field items\[0\]\.bytes at row 0/); + true, + ), + ]); + expect(() => + makeArrowTable([{ items: [{ bytes: new Uint8Array([104]).buffer }] }], { + schema, + }), + ).toThrow(/field items\[0\]\.bytes at row 0/); + }); + + it("will ignore generated dictionary IDs when comparing inferred types", function () { + const table = makeArrowTable([{ str: "a" }, { str: "b" }], { + dictionaryEncodeStrings: true, }); - it("will ignore generated dictionary IDs when comparing inferred types", function () { - const table = makeArrowTable([{ str: "a" }, { str: "b" }], { - dictionaryEncodeStrings: true, - }); + expect(table.getChild("str")?.toJSON()).toEqual(["a", "b"]); + }); - expect(table.getChild("str")?.toJSON()).toEqual(["a", "b"]); - }); + it("will preserve null values without treating them as type mismatches", function () { + for (const records of [ + [{ vector: [1, 2, 3] }, { vector: null }], + [{ vector: null }, { vector: [1, 2, 3] }], + ]) { + const table = makeArrowTable(records); - it("will preserve null values without treating them as type mismatches", function () { - for (const records of [ - [{ vector: [1, 2, 3] }, { vector: null }], - [{ vector: null }, { vector: [1, 2, 3] }], - ]) { - const table = makeArrowTable(records); + expect(table.numRows).toBe(2); + expect(table.getChild("vector")?.nullCount).toBe(1); + } + }); - expect(table.numRows).toBe(2); - expect(table.getChild("vector")?.nullCount).toBe(1); - } - }); - - it("will preserve empty variable-size lists", function () { - for (const records of [ - [{ items: [1] }, { items: [] }], - [{ items: [] }, { items: [1] }], - ]) { - const table = makeArrowTable(records); - expect( - table - .getChild("items") - ?.toJSON() - .map((value) => value.toJSON()), - ).toEqual(records.map((record) => record.items)); - } - }); - - it("will propagate deferred evidence through nested lists", function () { - for (const records of [ - [{ items: [1] }, { items: [null] }], - [{ items: [null] }, { items: [1] }], - [{ items: [null, 1] }, { items: [2, null] }], - ]) { - const table = makeArrowTable(records); - expect( - table - .getChild("items") - ?.toJSON() - .map((value) => value.toJSON()), - ).toEqual(records.map((record) => record.items)); - } - - const nestedRecords = [{ items: [[1]] }, { items: [[null]] }]; - const nestedTable = makeArrowTable(nestedRecords); + it("will preserve empty variable-size lists", function () { + for (const records of [ + [{ items: [1] }, { items: [] }], + [{ items: [] }, { items: [1] }], + ]) { + const table = makeArrowTable(records); expect( - nestedTable + table .getChild("items") ?.toJSON() - .map((value) => - value - .toJSON() - .map((nestedValue: { toJSON: () => unknown[] }) => - nestedValue.toJSON(), - ), - ), - ).toEqual(nestedRecords.map((record) => record.items)); - }); + .map((value) => value.toJSON()), + ).toEqual(records.map((record) => record.items)); + } + }); - it("will reject incompatible deferred evidence within a list", function () { - for (const items of [ - [[], 1], - [1, []], - [[null], 1], - [1, [null]], - ]) { - expect(() => makeArrowTable([{ items }])).toThrow( - "Failed to infer data type for field items at row 0.", - ); - } - }); + it("will propagate deferred evidence through nested lists", function () { + for (const records of [ + [{ items: [1] }, { items: [null] }], + [{ items: [null] }, { items: [1] }], + [{ items: [null, 1] }, { items: [2, null] }], + ]) { + const table = makeArrowTable(records); + expect( + table + .getChild("items") + ?.toJSON() + .map((value) => value.toJSON()), + ).toEqual(records.map((record) => record.items)); + } - it("will reject empty fixed-size lists", function () { - expect(() => - makeArrowTable([{ vector: [1, 2, 3] }, { vector: [] }]), - ).toThrow( - "Failed to infer schema for data. Previously inferred type FixedSizeList[3] but found List[0] for field vector at row 1.", + const nestedRecords = [{ items: [[1]] }, { items: [[null]] }]; + const nestedTable = makeArrowTable(nestedRecords); + expect( + nestedTable + .getChild("items") + ?.toJSON() + .map((value) => + value + .toJSON() + .map((nestedValue: { toJSON: () => unknown[] }) => + nestedValue.toJSON(), + ), + ), + ).toEqual(nestedRecords.map((record) => record.items)); + }); + + it("will reject incompatible deferred evidence within a list", function () { + for (const items of [ + [[], 1], + [1, []], + [[null], 1], + [1, [null]], + ]) { + expect(() => makeArrowTable([{ items }])).toThrow( + "Failed to infer data type for field items at row 0.", ); - }); + } + }); - it("will reject inferred leaf and branch shape changes", function () { - expect(() => - makeArrowTable([{ value: 1 }, { value: { nested: 2 } }]), - ).toThrow( - "Failed to infer schema for data. Previously inferred type Float64 but found Struct for field value at row 1.", - ); - expect(() => - makeArrowTable([{ value: { nested: 1 } }, { value: 2 }]), - ).toThrow( - "Failed to infer schema for data. Previously inferred type Struct but found Float64 for field value at row 1.", - ); - }); + it("will reject empty fixed-size lists", function () { + expect(() => + makeArrowTable([{ vector: [1, 2, 3] }, { vector: [] }]), + ).toThrow( + "Failed to infer schema for data. Previously inferred type FixedSizeList[3] but found List[0] for field vector at row 1.", + ); + }); - it("will allow null values around inferred struct values", function () { - for (const { records, nullIndex } of [ - { - records: [{ value: null }, { value: { nested: 2 } }], - nullIndex: 0, - }, - { - records: [{ value: { nested: 1 } }, { value: null }], - nullIndex: 1, - }, - ]) { - const table = makeArrowTable(records); - const values = table.getChild("value"); + it("will reject inferred leaf and branch shape changes", function () { + expect(() => + makeArrowTable([{ value: 1 }, { value: { nested: 2 } }]), + ).toThrow( + "Failed to infer schema for data. Previously inferred type Float64 but found Struct for field value at row 1.", + ); + expect(() => + makeArrowTable([{ value: { nested: 1 } }, { value: 2 }]), + ).toThrow( + "Failed to infer schema for data. Previously inferred type Struct but found Float64 for field value at row 1.", + ); + }); - expect(values?.nullCount).toBe(1); - expect(values?.get(nullIndex)).toBeNull(); - } - }); + it("will allow null values around inferred struct values", function () { + for (const { records, nullIndex } of [ + { + records: [{ value: null }, { value: { nested: 2 } }], + nullIndex: 0, + }, + { + records: [{ value: { nested: 1 } }, { value: null }], + nullIndex: 1, + }, + ]) { + const table = makeArrowTable(records); + const values = table.getChild("value"); - it("will allow a schema to be provided", async function () { - await checkTableCreation( - async (records, _, schema) => - // biome-ignore lint/suspicious/noExplicitAny: - (makeArrowTable)(records, { schema }), - false, - ); - }); + expect(values?.nullCount).toBe(1); + expect(values?.get(nullIndex)).toBeNull(); + } + }); - it("will use the field order of any provided schema", async function () { - await checkTableCreation( - async (_, recordsReversed, schema) => - // biome-ignore lint/suspicious/noExplicitAny: - (makeArrowTable)(recordsReversed, { schema }), - false, - ); - }); - - it("will make an empty table", async function () { - await checkTableCreation( + it("will allow a schema to be provided", async function () { + await checkTableCreation( + async (records, _, schema) => // biome-ignore lint/suspicious/noExplicitAny: - async (_, __, schema) => (makeArrowTable)([], { schema }), - false, - ); - }); + (makeArrowTable)(records, { schema }), + false, + ); + }); - it("will allow subsets of columns if nullable", async function () { - const schema = new Schema([ - new Field("a", new Int64(), true), + it("will use the field order of any provided schema", async function () { + await checkTableCreation( + async (_, recordsReversed, schema) => + // biome-ignore lint/suspicious/noExplicitAny: + (makeArrowTable)(recordsReversed, { schema }), + false, + ); + }); + + it("will make an empty table", async function () { + await checkTableCreation( + // biome-ignore lint/suspicious/noExplicitAny: + async (_, __, schema) => (makeArrowTable)([], { schema }), + false, + ); + }); + + it("will allow subsets of columns if nullable", async function () { + const schema = new Schema([ + new Field("a", new Int64(), true), + new Field( + "s", + new Struct([ + new Field("x", new Int32(), true), + new Field("y", new Int32(), true), + ]), + true, + ), + new Field("d", new Int16(), true), + ]); + + const table = makeArrowTable([{ a: 1n }], { schema }); + expect(table.numCols).toBe(1); + expect(table.numRows).toBe(1); + + const table2 = makeArrowTable([{ a: 1n, d: 2 }], { schema }); + expect(table2.numCols).toBe(2); + + const table3 = makeArrowTable([{ s: { y: 3 } }], { schema }); + expect(table3.numCols).toBe(1); + const expectedSchema = new Schema([ + new Field("s", new Struct([new Field("y", new Int32(), true)]), true), + ]); + expect(table3.schema).toEqual(expectedSchema); + }); + + it("will work even if columns are sparsely provided", async function () { + const sparseRecords = [{ a: 1n }, { b: 2n }, { c: 3n }, { d: 4n }]; + const table = makeArrowTable(sparseRecords); + expect(table.numCols).toBe(4); + expect(table.numRows).toBe(4); + + const schema = new Schema([ + new Field("a", new Int64(), true), + new Field("b", new Int32(), true), + new Field("c", new Int64(), true), + new Field("d", new Int16(), true), + ]); + const table2 = makeArrowTable(sparseRecords, { schema }); + expect(table2.numCols).toBe(4); + expect(table2.numRows).toBe(4); + expect(table2.schema).toEqual(schema); + }); + + it("will handle missing columns in schema alignment when using embeddings", async function () { + const schema = new Schema( + [ + new Field("domain", new Utf8(), true), + new Field("name", new Utf8(), true), + new Field("description", new Utf8(), true), + ], + new Map([["embedding_functions", JSON.stringify([])]]), + ); + + const data = [ + { domain: "google.com", name: "Google" }, + { domain: "facebook.com", name: "Facebook" }, + ]; + + const table = await convertToTable(data, undefined, { schema }); + + expect(table.numCols).toBe(3); + expect(table.numRows).toBe(2); + + const descriptionColumn = table.getChild("description"); + expect(descriptionColumn).toBeDefined(); + expect(descriptionColumn?.nullCount).toBe(2); + expect(descriptionColumn?.toArray()).toEqual([null, null]); + + expect(table.getChild("domain")?.toArray()).toEqual([ + "google.com", + "facebook.com", + ]); + expect(table.getChild("name")?.toArray()).toEqual(["Google", "Facebook"]); + }); + + it("will handle completely missing nested struct columns", async function () { + const schema = new Schema( + [ + new Field("id", new Utf8(), true), + new Field("name", new Utf8(), true), new Field( - "s", + "metadata", new Struct([ - new Field("x", new Int32(), true), - new Field("y", new Int32(), true), + new Field("version", new Int32(), true), + new Field("author", new Utf8(), true), + new Field( + "tags", + new List(new Field("item", new Utf8(), true)), + true, + ), ]), true, ), - new Field("d", new Int16(), true), - ]); - - const table = makeArrowTable([{ a: 1n }], { schema }); - expect(table.numCols).toBe(1); - expect(table.numRows).toBe(1); - - const table2 = makeArrowTable([{ a: 1n, d: 2 }], { schema }); - expect(table2.numCols).toBe(2); - - const table3 = makeArrowTable([{ s: { y: 3 } }], { schema }); - expect(table3.numCols).toBe(1); - const expectedSchema = new Schema([ - new Field("s", new Struct([new Field("y", new Int32(), true)]), true), - ]); - expect(table3.schema).toEqual(expectedSchema); - }); - - it("will work even if columns are sparsely provided", async function () { - const sparseRecords = [{ a: 1n }, { b: 2n }, { c: 3n }, { d: 4n }]; - const table = makeArrowTable(sparseRecords); - expect(table.numCols).toBe(4); - expect(table.numRows).toBe(4); - - const schema = new Schema([ - new Field("a", new Int64(), true), - new Field("b", new Int32(), true), - new Field("c", new Int64(), true), - new Field("d", new Int16(), true), - ]); - const table2 = makeArrowTable(sparseRecords, { schema }); - expect(table2.numCols).toBe(4); - expect(table2.numRows).toBe(4); - expect(table2.schema).toEqual(schema); - }); - - it("will handle missing columns in schema alignment when using embeddings", async function () { - const schema = new Schema( - [ - new Field("domain", new Utf8(), true), - new Field("name", new Utf8(), true), - new Field("description", new Utf8(), true), - ], - new Map([["embedding_functions", JSON.stringify([])]]), - ); - - const data = [ - { domain: "google.com", name: "Google" }, - { domain: "facebook.com", name: "Facebook" }, - ]; - - const table = await convertToTable(data, undefined, { schema }); - - expect(table.numCols).toBe(3); - expect(table.numRows).toBe(2); - - const descriptionColumn = table.getChild("description"); - expect(descriptionColumn).toBeDefined(); - expect(descriptionColumn?.nullCount).toBe(2); - expect(descriptionColumn?.toArray()).toEqual([null, null]); - - expect(table.getChild("domain")?.toArray()).toEqual([ - "google.com", - "facebook.com", - ]); - expect(table.getChild("name")?.toArray()).toEqual([ - "Google", - "Facebook", - ]); - }); - - it("will handle completely missing nested struct columns", async function () { - const schema = new Schema( - [ - new Field("id", new Utf8(), true), - new Field("name", new Utf8(), true), - new Field( - "metadata", - new Struct([ - new Field("version", new Int32(), true), - new Field("author", new Utf8(), true), - new Field( - "tags", - new List(new Field("item", new Utf8(), true)), - true, - ), - ]), - true, - ), - ], - new Map([["embedding_functions", JSON.stringify([])]]), - ); - - const data = [ - { id: "doc1", name: "Document 1" }, - { id: "doc2", name: "Document 2" }, - ]; - - const table = await convertToTable(data, undefined, { schema }); - - expect(table.numCols).toBe(3); - expect(table.numRows).toBe(2); - - const buf = await fromTableToBuffer(table); - const retrievedTable = tableFromIPC(buf); - - const rows = []; - for (let i = 0; i < retrievedTable.numRows; i++) { - rows.push(retrievedTable.get(i)); - } - - expect(rows[0].metadata.version).toBe(null); - expect(rows[0].metadata.author).toBe(null); - expect(rows[0].metadata.tags).toBe(null); - expect(rows[0].id).toBe("doc1"); - expect(rows[0].name).toBe("Document 1"); - }); - - it("will handle partially missing nested struct fields", async function () { - const schema = new Schema( - [ - new Field("id", new Utf8(), true), - new Field( - "metadata", - new Struct([ - new Field("version", new Int32(), true), - new Field("author", new Utf8(), true), - new Field("created_at", new Utf8(), true), - ]), - true, - ), - ], - new Map([["embedding_functions", JSON.stringify([])]]), - ); - - const data = [ - { id: "doc1", metadata: { version: 1, author: "Alice" } }, - { id: "doc2", metadata: { version: 2 } }, - ]; - - const table = await convertToTable(data, undefined, { schema }); - - expect(table.numCols).toBe(2); - expect(table.numRows).toBe(2); - - const metadataColumn = table.getChild("metadata"); - expect(metadataColumn).toBeDefined(); - expect(metadataColumn?.type.toString()).toBe( - "Struct<{version:Int32, author:Utf8, created_at:Utf8}>", - ); - }); - - it("will handle multiple levels of nested structures", async function () { - const schema = new Schema( - [ - new Field("id", new Utf8(), true), - new Field( - "config", - new Struct([ - new Field("database", new Utf8(), true), - new Field( - "connection", - new Struct([ - new Field("host", new Utf8(), true), - new Field("port", new Int32(), true), - new Field( - "ssl", - new Struct([ - new Field("enabled", new Bool(), true), - new Field("cert_path", new Utf8(), true), - ]), - true, - ), - ]), - true, - ), - ]), - true, - ), - ], - new Map([["embedding_functions", JSON.stringify([])]]), - ); - - const data = [ - { - id: "config1", - config: { - database: "postgres", - connection: { host: "localhost" }, - }, - }, - { - id: "config2", - config: { database: "mysql" }, - }, - { - id: "config3", - }, - ]; - - const table = await convertToTable(data, undefined, { schema }); - - expect(table.numCols).toBe(2); - expect(table.numRows).toBe(3); - - const configColumn = table.getChild("config"); - expect(configColumn).toBeDefined(); - expect(configColumn?.type.toString()).toBe( - "Struct<{database:Utf8, connection:Struct<{host:Utf8, port:Int32, ssl:Struct<{enabled:Bool, cert_path:Utf8}>}>}>", - ); - }); - - it("will handle missing columns in Arrow table input when using embeddings", async function () { - const incompleteTable = makeArrowTable([ - { domain: "google.com", name: "Google" }, - { domain: "facebook.com", name: "Facebook" }, - ]); - - const schema = new Schema( - [ - new Field("domain", new Utf8(), true), - new Field("name", new Utf8(), true), - new Field("description", new Utf8(), true), - ], - new Map([["embedding_functions", JSON.stringify([])]]), - ); - - const buf = await fromDataToBuffer(incompleteTable, undefined, schema); - - expect(buf.byteLength).toBeGreaterThan(0); - - const retrievedTable = tableFromIPC(buf); - expect(retrievedTable.numCols).toBe(3); - expect(retrievedTable.numRows).toBe(2); - - const descriptionColumn = retrievedTable.getChild("description"); - expect(descriptionColumn).toBeDefined(); - expect(descriptionColumn?.nullCount).toBe(2); - expect(descriptionColumn?.toArray()).toEqual([null, null]); - - expect(retrievedTable.getChild("domain")?.toArray()).toEqual([ - "google.com", - "facebook.com", - ]); - expect(retrievedTable.getChild("name")?.toArray()).toEqual([ - "Google", - "Facebook", - ]); - }); - - it("should correctly retain values in nested struct fields", async function () { - const testData = [ - { - id: "doc1", - vector: [1, 2, 3], - metadata: { - filePath: "/path/to/file1.ts", - startLine: 10, - endLine: 20, - text: "function test() { return true; }", - }, - }, - { - id: "doc2", - vector: [4, 5, 6], - metadata: { - filePath: "/path/to/file2.ts", - startLine: 30, - endLine: 40, - text: "function test2() { return false; }", - }, - }, - ]; - - const table = makeArrowTable(testData); - - const metadataField = table.schema.fields.find( - (f) => f.name === "metadata", - ); - expect(metadataField).toBeDefined(); - // biome-ignore lint/suspicious/noExplicitAny: accessing fields in different Arrow versions - const childNames = metadataField?.type.children.map((c: any) => c.name); - expect(childNames).toEqual([ - "filePath", - "startLine", - "endLine", - "text", - ]); - - const buf = await fromTableToBuffer(table); - const retrievedTable = tableFromIPC(buf); - - const rows = []; - for (let i = 0; i < retrievedTable.numRows; i++) { - rows.push(retrievedTable.get(i)); - } - - const firstRow = rows[0]; - expect(firstRow.id).toBe("doc1"); - expect(firstRow.vector.toJSON()).toEqual([1, 2, 3]); - expect(firstRow.metadata.filePath).toBe("/path/to/file1.ts"); - expect(firstRow.metadata.startLine).toBe(10); - expect(firstRow.metadata.endLine).toBe(20); - expect(firstRow.metadata.text).toBe("function test() { return true; }"); - }); - }); - - class DummyEmbedding extends EmbeddingFunction { - toJSON(): Partial { - return {}; - } - - async computeSourceEmbeddings(data: string[]): Promise { - return data.map(() => [0.0, 0.0]); - } - - ndims(): number { - return 2; - } - - embeddingDataType() { - return new Float16(); - } - } - - class DummyEmbeddingWithNoDimension extends EmbeddingFunction { - toJSON(): Partial { - return {}; - } - - embeddingDataType() { - return new Float16(); - } - - async computeSourceEmbeddings(data: string[]): Promise { - return data.map(() => [0.0, 0.0]); - } - } - const dummyEmbeddingConfig: EmbeddingFunctionConfig = { - sourceColumn: "string", - function: new DummyEmbedding(), - }; - - const dummyEmbeddingConfigWithNoDimension: EmbeddingFunctionConfig = { - sourceColumn: "string", - function: new DummyEmbeddingWithNoDimension(), - }; - - describe("convertToTable", function () { - it("will infer data types correctly", async function () { - await checkTableCreation( - // biome-ignore lint/suspicious/noExplicitAny: - async (records) => await (convertToTable)(records), - true, - ); - }); - - it("will allow a schema to be provided", async function () { - await checkTableCreation( - async (records, _, schema) => - // biome-ignore lint/suspicious/noExplicitAny: - await (convertToTable)(records, undefined, { schema }), - false, - ); - }); - - it("will use the field order of any provided schema", async function () { - await checkTableCreation( - async (_, recordsReversed, schema) => - // biome-ignore lint/suspicious/noExplicitAny: - await (convertToTable)(recordsReversed, undefined, { schema }), - false, - ); - }); - - it("will make an empty table", async function () { - await checkTableCreation( - async (_, __, schema) => - // biome-ignore lint/suspicious/noExplicitAny: - await (convertToTable)([], undefined, { schema }), - false, - ); - }); - - it("will apply embeddings", async function () { - const records = sampleRecords(); - const table = await convertToTable(records, dummyEmbeddingConfig); - expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( - true, - ); - expect( - table.getChild("vector")?.type.children[0].type.toString(), - ).toEqual(new Float16().toString()); - }); - - it("will fail if missing the embedding source column", async function () { - await expect( - convertToTable([{ id: 1 }], dummyEmbeddingConfig), - ).rejects.toThrow("'string' was not present"); - }); - - it("use embeddingDimension if embedding missing from table", async function () { - const schema = new Schema([new Field("string", new Utf8(), false)]); - // Simulate getting an empty Arrow table (minus embedding) from some other source - // In other words, we aren't starting with records - const table = makeEmptyTable(schema); - - // If the embedding specifies the dimension we are fine - await fromTableToBuffer(table, dummyEmbeddingConfig); - - // We can also supply a schema and should be ok - const schemaWithEmbedding = new Schema([ - new Field("string", new Utf8(), false), - new Field( - "vector", - new FixedSizeList(2, new Field("item", new Float16(), false)), - false, - ), - ]); - await fromTableToBuffer( - table, - dummyEmbeddingConfigWithNoDimension, - schemaWithEmbedding, - ); - - // Otherwise we will get an error - await expect( - fromTableToBuffer(table, dummyEmbeddingConfigWithNoDimension), - ).rejects.toThrow("does not specify `embeddingDimension`"); - }); - - it("will apply embeddings to an empty table", async function () { - const schema = new Schema([ - new Field("string", new Utf8(), false), - new Field( - "vector", - new FixedSizeList(2, new Field("item", new Float16(), false)), - false, - ), - ]); - const table = await convertToTable([], dummyEmbeddingConfig, { - schema, - }); - expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( - true, - ); - expect( - table.getChild("vector")?.type.children[0].type.toString(), - ).toEqual(new Float16().toString()); - }); - - it("will complain if embeddings present but schema missing embedding column", async function () { - const schema = new Schema([new Field("string", new Utf8(), false)]); - await expect( - convertToTable([], dummyEmbeddingConfig, { schema }), - ).rejects.toThrow("column vector was missing"); - }); - - it("will skip embedding application if already applied", async function () { - const records = sampleRecords(); - const table = await convertToTable(records, dummyEmbeddingConfig); - - // fromTableToBuffer will try and apply the embeddings again - // but should skip since the column already has non-null values - const result = await fromTableToBuffer(table, dummyEmbeddingConfig); - expect(result.byteLength).toBeGreaterThan(0); - }); - }); - - describe("makeEmptyTable", function () { - it("will make an empty table", async function () { - await checkTableCreation( - // biome-ignore lint/suspicious/noExplicitAny: - async (_, __, schema) => (makeEmptyTable)(schema), - false, - ); - }); - - it("will make an empty table with a Map field", async function () { - const schema = new Schema([ - new Field( - "attributes", - new Map_( - new Field( - "entries", - new Struct([ - new Field("key", new Utf8(), false), - new Field("value", new Int32(), true), - ]), - false, - ), - true, - ), - ), - ]); - - const table = makeEmptyTable(schema); - - expectValidMapField(table.schema.fields[0]); - - const buffer = await fromTableToBuffer(table); - const roundTripped = tableFromIPC(buffer); - - expectValidMapField(roundTripped.schema.fields[0]); - }); - - it("preserves string schema metadata", function () { - const metadata = new Map([["source", "fixture"]]); - const schema = new Schema( - [new Field("value", new Int32(), true)], - metadata, - ); - - expect(makeEmptyTable(schema).schema.metadata.get("source")).toBe( - "fixture", - ); - }); - - it.each([ - ["non-string keys", new Map([[42, "fixture"]])], - ["non-string values", new Map([["source", 42]])], - [ - "non-string keys and values", - new Map([[42, false]]), ], - ])("rejects schema metadata with %s", function (_, metadataLike) { - const metadata = metadataLike as unknown as Map; - const schema = new Schema( - [new Field("value", new Int32(), true)], - metadata, - ); + new Map([["embedding_functions", JSON.stringify([])]]), + ); - expect(() => makeEmptyTable(schema)).toThrow( - "Expected metadata, if present, to be a Map but it had non-string keys or values", - ); - }); + const data = [ + { id: "doc1", name: "Document 1" }, + { id: "doc2", name: "Document 2" }, + ]; + + const table = await convertToTable(data, undefined, { schema }); + + expect(table.numCols).toBe(3); + expect(table.numRows).toBe(2); + + const buf = await fromTableToBuffer(table); + const retrievedTable = tableFromIPC(buf); + + const rows = []; + for (let i = 0; i < retrievedTable.numRows; i++) { + rows.push(retrievedTable.get(i)); + } + + expect(rows[0].metadata.version).toBe(null); + expect(rows[0].metadata.author).toBe(null); + expect(rows[0].metadata.tags).toBe(null); + expect(rows[0].id).toBe("doc1"); + expect(rows[0].name).toBe("Document 1"); }); - describe("when using two versions of arrow", function () { - it("preserves a dictionary shared by multiple fields", async function () { - const values = ["alpha", "beta", "alpha"]; - const dictionaryVector = vectorFromArray(values); - const batch = new ArrowRecordBatch({ - first: dictionaryVector.data[0], - second: dictionaryVector.data[0], - }); - const table = new ArrowTable([batch]); - - const sanitized = sanitizeTable(table); - expect([...sanitized.getChild("first")!]).toEqual(values); - expect([...sanitized.getChild("second")!]).toEqual(values); - const firstType = sanitized.schema.fields[0].type as { - dictionary: unknown; - }; - const secondType = sanitized.schema.fields[1].type as { - dictionary: unknown; - }; - expect(secondType.dictionary).toBe(firstType.dictionary); - expect(sanitized.batches[0].data.children[1].dictionary).toBe( - sanitized.batches[0].data.children[0].dictionary, - ); - - const buf = await fromDataToBuffer(table); - const actual = currentTableFromIPC(buf); - expect([...actual.getChild("first")!]).toEqual(values); - expect([...actual.getChild("second")!]).toEqual(values); - }); - - it("preserves shared dictionary data from another Arrow version", async function () { - const values = ["alpha", "beta", "alpha"]; - const dictionaryVector = vectorFromArray(values); - const firstBatch = new ArrowRecordBatch({ - label: dictionaryVector.slice(0, 2).data[0], - }); - const secondBatch = new ArrowRecordBatch({ - label: dictionaryVector.slice(2).data[0], - }); - const table = new ArrowTable([firstBatch, secondBatch]); - - const sanitized = sanitizeTable(table); - expect([...sanitized.getChild("label")!]).toEqual(values); - - const dictionaries = sanitized.batches.map( - (batch) => batch.data.children[0].dictionary, - ); - expect(dictionaries[0]).toBeInstanceOf(CurrentVector); - expect(dictionaries[1]).toBe(dictionaries[0]); - - const buf = await fromDataToBuffer(table); - const actual = currentTableFromIPC(buf); - expect([...actual.getChild("label")!]).toEqual(values); - }); - - it("preserves shared chunks in growing dictionaries", async function () { - const type = new Dictionary(new Utf8(), new Int32(), 42, false); - const firstDictionary = vectorFromArray(["alpha", "beta"], new Utf8()); - const secondDictionary = firstDictionary.concat( - vectorFromArray(["gamma"], new Utf8()), - ); - const firstData = arrowMakeData({ - type, - data: Int32Array.from([0, 1]), - dictionary: firstDictionary, - }); - const secondData = arrowMakeData({ - type, - data: Int32Array.from([2]), - dictionary: secondDictionary, - }); - const table = new ArrowTable([ - new ArrowRecordBatch({ label: firstData }), - new ArrowRecordBatch({ label: secondData }), - ]); - - const sanitized = sanitizeTable(table); - const expected = ["alpha", "beta", "gamma"]; - expect([...sanitized.getChild("label")!]).toEqual(expected); - const firstLocalDictionary = - sanitized.batches[0].data.children[0].dictionary!; - const secondLocalDictionary = - sanitized.batches[1].data.children[0].dictionary!; - expect(secondLocalDictionary.data[0]).toBe( - firstLocalDictionary.data[0], - ); - - const buf = await fromTableToBuffer(sanitized); - const actual = currentTableFromIPC(buf); - expect([...actual.getChild("label")!]).toEqual(expected); - }); - - it("can serialize list data from another Arrow version", async function () { - const values = [["anime", "action"], [], null]; - const vector = vectorFromArray( - values, - new List(new Field("item", new Utf8(), true)), - ); - const table = new ArrowTable({ tags: vector }); - - const buf = await fromDataToBuffer(table); - const actual = currentTableFromIPC(buf); - const actualTags = actual.getChild("tags"); - - expect(actualTags?.get(0)?.toJSON()).toEqual(values[0]); - expect(actualTags?.get(1)?.toJSON()).toEqual(values[1]); - expect(actualTags?.get(2)).toBeNull(); - }); - - it("can still import data", async function () { - const schema = new arrow15.Schema([ - new arrow15.Field("id", new arrow15.Int32()), - new arrow15.Field( - "vector", - new arrow15.FixedSizeList( - 1024, - new arrow15.Field("item", new arrow15.Float32(), true), - ), + it("will handle partially missing nested struct fields", async function () { + const schema = new Schema( + [ + new Field("id", new Utf8(), true), + new Field( + "metadata", + new Struct([ + new Field("version", new Int32(), true), + new Field("author", new Utf8(), true), + new Field("created_at", new Utf8(), true), + ]), + true, ), - new arrow15.Field( - "struct", - new arrow15.Struct([ - new arrow15.Field( - "nested", - new arrow15.Dictionary( - new arrow15.Utf8(), - new arrow15.Int32(), - 1, - true, - ), - ), - new arrow15.Field( - "ts_with_tz", - new arrow15.TimestampNanosecond("some_tz"), - ), - new arrow15.Field( - "ts_no_tz", - new arrow15.TimestampNanosecond(null), + ], + new Map([["embedding_functions", JSON.stringify([])]]), + ); + + const data = [ + { id: "doc1", metadata: { version: 1, author: "Alice" } }, + { id: "doc2", metadata: { version: 2 } }, + ]; + + const table = await convertToTable(data, undefined, { schema }); + + expect(table.numCols).toBe(2); + expect(table.numRows).toBe(2); + + const metadataColumn = table.getChild("metadata"); + expect(metadataColumn).toBeDefined(); + expect(metadataColumn?.type.toString()).toBe( + "Struct<{version:Int32, author:Utf8, created_at:Utf8}>", + ); + }); + + it("will handle multiple levels of nested structures", async function () { + const schema = new Schema( + [ + new Field("id", new Utf8(), true), + new Field( + "config", + new Struct([ + new Field("database", new Utf8(), true), + new Field( + "connection", + new Struct([ + new Field("host", new Utf8(), true), + new Field("port", new Int32(), true), + new Field( + "ssl", + new Struct([ + new Field("enabled", new Bool(), true), + new Field("cert_path", new Utf8(), true), + ]), + true, + ), + ]), + true, ), ]), + true, ), - // biome-ignore lint/suspicious/noExplicitAny: skip - ]) as any; - schema.metadataVersion = arrow15.MetadataVersion.V5; - const table = makeArrowTable([], { schema }); + ], + new Map([["embedding_functions", JSON.stringify([])]]), + ); - const buf = await fromTableToBuffer(table); - expect(buf.byteLength).toBeGreaterThan(0); - const actual = tableFromIPC(buf); - const actualSchema = actual.schema; - expect(actualSchema.fields.length).toBe(3); + const data = [ + { + id: "config1", + config: { + database: "postgres", + connection: { host: "localhost" }, + }, + }, + { + id: "config2", + config: { database: "mysql" }, + }, + { + id: "config3", + }, + ]; - // Deep equality gets hung up on some very minor unimportant differences - // between arrow version 13 and 15 which isn't really what we're testing for - // and so we do our own comparison that just checks name/type/nullability - function compareFields(lhs: arrow15.Field, rhs: arrow15.Field) { - expect(lhs.name).toEqual(rhs.name); - expect(lhs.nullable).toEqual(rhs.nullable); - expect(lhs.typeId).toEqual(rhs.typeId); - if ("children" in lhs.type && lhs.type.children !== null) { - const lhsChildren = lhs.type.children as arrow15.Field[]; - lhsChildren.forEach((child: arrow15.Field, idx) => { - compareFields(child, rhs.type.children[idx]); - }); - } - } + const table = await convertToTable(data, undefined, { schema }); + + expect(table.numCols).toBe(2); + expect(table.numRows).toBe(3); + + const configColumn = table.getChild("config"); + expect(configColumn).toBeDefined(); + expect(configColumn?.type.toString()).toBe( + "Struct<{database:Utf8, connection:Struct<{host:Utf8, port:Int32, ssl:Struct<{enabled:Bool, cert_path:Utf8}>}>}>", + ); + }); + + it("will handle missing columns in Arrow table input when using embeddings", async function () { + const incompleteTable = makeArrowTable([ + { domain: "google.com", name: "Google" }, + { domain: "facebook.com", name: "Facebook" }, + ]); + + const schema = new Schema( + [ + new Field("domain", new Utf8(), true), + new Field("name", new Utf8(), true), + new Field("description", new Utf8(), true), + ], + new Map([["embedding_functions", JSON.stringify([])]]), + ); + + const buf = await fromDataToBuffer(incompleteTable, undefined, schema); + + expect(buf.byteLength).toBeGreaterThan(0); + + const retrievedTable = tableFromIPC(buf); + expect(retrievedTable.numCols).toBe(3); + expect(retrievedTable.numRows).toBe(2); + + const descriptionColumn = retrievedTable.getChild("description"); + expect(descriptionColumn).toBeDefined(); + expect(descriptionColumn?.nullCount).toBe(2); + expect(descriptionColumn?.toArray()).toEqual([null, null]); + + expect(retrievedTable.getChild("domain")?.toArray()).toEqual([ + "google.com", + "facebook.com", + ]); + expect(retrievedTable.getChild("name")?.toArray()).toEqual([ + "Google", + "Facebook", + ]); + }); + + it("should correctly retain values in nested struct fields", async function () { + const testData = [ + { + id: "doc1", + vector: [1, 2, 3], + metadata: { + filePath: "/path/to/file1.ts", + startLine: 10, + endLine: 20, + text: "function test() { return true; }", + }, + }, + { + id: "doc2", + vector: [4, 5, 6], + metadata: { + filePath: "/path/to/file2.ts", + startLine: 30, + endLine: 40, + text: "function test2() { return false; }", + }, + }, + ]; + + const table = makeArrowTable(testData); + + const metadataField = table.schema.fields.find( + (f) => f.name === "metadata", + ); + expect(metadataField).toBeDefined(); + // biome-ignore lint/suspicious/noExplicitAny: accessing fields in different Arrow versions + const childNames = metadataField?.type.children.map((c: any) => c.name); + expect(childNames).toEqual(["filePath", "startLine", "endLine", "text"]); + + const buf = await fromTableToBuffer(table); + const retrievedTable = tableFromIPC(buf); + + const rows = []; + for (let i = 0; i < retrievedTable.numRows; i++) { + rows.push(retrievedTable.get(i)); + } + + const firstRow = rows[0]; + expect(firstRow.id).toBe("doc1"); + expect(firstRow.vector.toJSON()).toEqual([1, 2, 3]); + expect(firstRow.metadata.filePath).toBe("/path/to/file1.ts"); + expect(firstRow.metadata.startLine).toBe(10); + expect(firstRow.metadata.endLine).toBe(20); + expect(firstRow.metadata.text).toBe("function test() { return true; }"); + }); + }); + + class DummyEmbedding extends EmbeddingFunction { + toJSON(): Partial { + return {}; + } + + async computeSourceEmbeddings(data: string[]): Promise { + return data.map(() => [0.0, 0.0]); + } + + ndims(): number { + return 2; + } + + embeddingDataType() { + return new Float16(); + } + } + + class DummyEmbeddingWithNoDimension extends EmbeddingFunction { + toJSON(): Partial { + return {}; + } + + embeddingDataType() { + return new Float16(); + } + + async computeSourceEmbeddings(data: string[]): Promise { + return data.map(() => [0.0, 0.0]); + } + } + const dummyEmbeddingConfig: EmbeddingFunctionConfig = { + sourceColumn: "string", + function: new DummyEmbedding(), + }; + + const dummyEmbeddingConfigWithNoDimension: EmbeddingFunctionConfig = { + sourceColumn: "string", + function: new DummyEmbeddingWithNoDimension(), + }; + + describe("convertToTable", function () { + it("will infer data types correctly", async function () { + await checkTableCreation( // biome-ignore lint/suspicious/noExplicitAny: - actualSchema.fields.forEach((field: any, idx: string | number) => { - compareFields(field, actualSchema.fields[idx]); - }); - }); + async (records) => await (convertToTable)(records), + true, + ); }); - describe("converting record batches to buffers", function () { - it("can convert to buffered record batch and back again", async function () { - const records = [ - { text: "dog", vector: [0.1, 0.2] }, - { text: "cat", vector: [0.3, 0.4] }, - ]; - const table = await convertToTable(records); - const batch = table.batches[0]; - - const buffer = await fromRecordBatchToBuffer(batch); - const result = await fromBufferToRecordBatch(buffer); - - expect(JSON.stringify(batch.toArray())).toEqual( - JSON.stringify(result?.toArray()), - ); - }); - - it("converting from buffer returns null if buffer has no record batches", async function () { - const result = await fromBufferToRecordBatch(Buffer.from([0x01, 0x02])); // bad data - expect(result).toEqual(null); - }); + it("will allow a schema to be provided", async function () { + await checkTableCreation( + async (records, _, schema) => + // biome-ignore lint/suspicious/noExplicitAny: + await (convertToTable)(records, undefined, { schema }), + false, + ); }); - describe("boolean null handling", function () { - it("should handle null values in nullable boolean fields", () => { - const { makeArrowTable } = require("../lancedb/arrow"); - const schema = new Schema([new Field("test", new arrow.Bool(), true)]); - - // Test with all null values - const data = [{ test: null }]; - const table = makeArrowTable(data, { schema }); - - expect(table.numRows).toBe(1); - expect(table.schema.names).toEqual(["test"]); - expect(table.getChild("test")!.get(0)).toBeNull(); - }); - - it("should handle mixed null and non-null boolean values", () => { - const { makeArrowTable } = require("../lancedb/arrow"); - const schema = new Schema([new Field("test", new Bool(), true)]); - - // Test with mixed values - const data = [{ test: true }, { test: null }, { test: false }]; - const table = makeArrowTable(data, { schema }); - - expect(table.numRows).toBe(3); - expect(table.getChild("test")!.get(0)).toBe(true); - expect(table.getChild("test")!.get(1)).toBeNull(); - expect(table.getChild("test")!.get(2)).toBe(false); - }); + it("will use the field order of any provided schema", async function () { + await checkTableCreation( + async (_, recordsReversed, schema) => + // biome-ignore lint/suspicious/noExplicitAny: + await (convertToTable)(recordsReversed, undefined, { schema }), + false, + ); }); - // Test for the undefined values bug fix - describe("undefined values handling", () => { - it("should handle mixed undefined and actual values", () => { - const schema = new Schema([ - new Field("text", new Utf8(), true), // nullable - new Field("number", new Int32(), true), // nullable - new Field("bool", new Bool(), true), // nullable - ]); + it("will make an empty table", async function () { + await checkTableCreation( + async (_, __, schema) => + // biome-ignore lint/suspicious/noExplicitAny: + await (convertToTable)([], undefined, { schema }), + false, + ); + }); - const data = [ - { text: undefined, number: 42, bool: true }, - { text: "hello", number: undefined, bool: false }, - { text: "world", number: 123, bool: undefined }, - ]; - const table = makeArrowTable(data, { schema }); + it("will apply embeddings", async function () { + const records = sampleRecords(); + const table = await convertToTable(records, dummyEmbeddingConfig); + expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( + true, + ); + expect( + table.getChild("vector")?.type.children[0].type.toString(), + ).toEqual(new Float16().toString()); + }); - const result = table.toArray(); - expect(result).toHaveLength(3); - expect(result[0].text).toBe(null); - expect(result[0].number).toBe(42); - expect(result[0].bool).toBe(true); - expect(result[1].text).toBe("hello"); - expect(result[1].number).toBe(null); - expect(result[1].bool).toBe(false); - expect(result[2].text).toBe("world"); - expect(result[2].number).toBe(123); - expect(result[2].bool).toBe(null); + it("will fail if missing the embedding source column", async function () { + await expect( + convertToTable([{ id: 1 }], dummyEmbeddingConfig), + ).rejects.toThrow("'string' was not present"); + }); + + it("use embeddingDimension if embedding missing from table", async function () { + const schema = new Schema([new Field("string", new Utf8(), false)]); + // Simulate getting an empty Arrow table (minus embedding) from some other source + // In other words, we aren't starting with records + const table = makeEmptyTable(schema); + + // If the embedding specifies the dimension we are fine + await fromTableToBuffer(table, dummyEmbeddingConfig); + + // We can also supply a schema and should be ok + const schemaWithEmbedding = new Schema([ + new Field("string", new Utf8(), false), + new Field( + "vector", + new FixedSizeList(2, new Field("item", new Float16(), false)), + false, + ), + ]); + await fromTableToBuffer( + table, + dummyEmbeddingConfigWithNoDimension, + schemaWithEmbedding, + ); + + // Otherwise we will get an error + await expect( + fromTableToBuffer(table, dummyEmbeddingConfigWithNoDimension), + ).rejects.toThrow("does not specify `embeddingDimension`"); + }); + + it("will apply embeddings to an empty table", async function () { + const schema = new Schema([ + new Field("string", new Utf8(), false), + new Field( + "vector", + new FixedSizeList(2, new Field("item", new Float16(), false)), + false, + ), + ]); + const table = await convertToTable([], dummyEmbeddingConfig, { + schema, + }); + expect(DataType.isFixedSizeList(table.getChild("vector")?.type)).toBe( + true, + ); + expect( + table.getChild("vector")?.type.children[0].type.toString(), + ).toEqual(new Float16().toString()); + }); + + it("will complain if embeddings present but schema missing embedding column", async function () { + const schema = new Schema([new Field("string", new Utf8(), false)]); + await expect( + convertToTable([], dummyEmbeddingConfig, { schema }), + ).rejects.toThrow("column vector was missing"); + }); + + it("will skip embedding application if already applied", async function () { + const records = sampleRecords(); + const table = await convertToTable(records, dummyEmbeddingConfig); + + // fromTableToBuffer will try and apply the embeddings again + // but should skip since the column already has non-null values + const result = await fromTableToBuffer(table, dummyEmbeddingConfig); + expect(result.byteLength).toBeGreaterThan(0); + }); + }); + + describe("makeEmptyTable", function () { + it("will make an empty table", async function () { + await checkTableCreation( + // biome-ignore lint/suspicious/noExplicitAny: + async (_, __, schema) => (makeEmptyTable)(schema), + false, + ); + }); + + it("will make an empty table with a Map field", async function () { + const schema = new Schema([ + new Field( + "attributes", + new Map_( + new Field( + "entries", + new Struct([ + new Field("key", new Utf8(), false), + new Field("value", new Int32(), true), + ]), + false, + ), + true, + ), + ), + ]); + + const table = makeEmptyTable(schema); + + expectValidMapField(table.schema.fields[0]); + + const buffer = await fromTableToBuffer(table); + const roundTripped = tableFromIPC(buffer); + + expectValidMapField(roundTripped.schema.fields[0]); + }); + + it("preserves string schema metadata", function () { + const metadata = new Map([["source", "fixture"]]); + const schema = new Schema( + [new Field("value", new Int32(), true)], + metadata, + ); + + expect(makeEmptyTable(schema).schema.metadata.get("source")).toBe( + "fixture", + ); + }); + + it.each([ + ["non-string keys", new Map([[42, "fixture"]])], + ["non-string values", new Map([["source", 42]])], + ["non-string keys and values", new Map([[42, false]])], + ])("rejects schema metadata with %s", function (_, metadataLike) { + const metadata = metadataLike as unknown as Map; + const schema = new Schema( + [new Field("value", new Int32(), true)], + metadata, + ); + + expect(() => makeEmptyTable(schema)).toThrow( + "Expected metadata, if present, to be a Map but it had non-string keys or values", + ); + }); + }); + + describe("when using two versions of arrow", function () { + it("preserves a dictionary shared by multiple fields", async function () { + const values = ["alpha", "beta", "alpha"]; + const dictionaryVector = vectorFromArray(values); + const batch = new ArrowRecordBatch({ + first: dictionaryVector.data[0], + second: dictionaryVector.data[0], + }); + const table = new ArrowTable([batch]); + + const sanitized = sanitizeTable(table); + expect([...sanitized.getChild("first")!]).toEqual(values); + expect([...sanitized.getChild("second")!]).toEqual(values); + const firstType = sanitized.schema.fields[0].type as { + dictionary: unknown; + }; + const secondType = sanitized.schema.fields[1].type as { + dictionary: unknown; + }; + expect(secondType.dictionary).toBe(firstType.dictionary); + expect(sanitized.batches[0].data.children[1].dictionary).toBe( + sanitized.batches[0].data.children[0].dictionary, + ); + + const buf = await fromDataToBuffer(table); + const actual = currentTableFromIPC(buf); + expect([...actual.getChild("first")!]).toEqual(values); + expect([...actual.getChild("second")!]).toEqual(values); + }); + + it("preserves shared dictionary data from another Arrow version", async function () { + const values = ["alpha", "beta", "alpha"]; + const dictionaryVector = vectorFromArray(values); + const firstBatch = new ArrowRecordBatch({ + label: dictionaryVector.slice(0, 2).data[0], + }); + const secondBatch = new ArrowRecordBatch({ + label: dictionaryVector.slice(2).data[0], + }); + const table = new ArrowTable([firstBatch, secondBatch]); + + const sanitized = sanitizeTable(table); + expect([...sanitized.getChild("label")!]).toEqual(values); + + const dictionaries = sanitized.batches.map( + (batch) => batch.data.children[0].dictionary, + ); + expect(dictionaries[0]).toBeInstanceOf(CurrentVector); + expect(dictionaries[1]).toBe(dictionaries[0]); + + const buf = await fromDataToBuffer(table); + const actual = currentTableFromIPC(buf); + expect([...actual.getChild("label")!]).toEqual(values); + }); + + it("preserves shared chunks in growing dictionaries", async function () { + const type = new Dictionary(new Utf8(), new Int32(), 42, false); + const firstDictionary = vectorFromArray(["alpha", "beta"], new Utf8()); + const secondDictionary = firstDictionary.concat( + vectorFromArray(["gamma"], new Utf8()), + ); + const firstData = arrowMakeData({ + type, + data: Int32Array.from([0, 1]), + dictionary: firstDictionary, + }); + const secondData = arrowMakeData({ + type, + data: Int32Array.from([2]), + dictionary: secondDictionary, + }); + const table = new ArrowTable([ + new ArrowRecordBatch({ label: firstData }), + new ArrowRecordBatch({ label: secondData }), + ]); + + const sanitized = sanitizeTable(table); + const expected = ["alpha", "beta", "gamma"]; + expect([...sanitized.getChild("label")!]).toEqual(expected); + const firstLocalDictionary = + sanitized.batches[0].data.children[0].dictionary!; + const secondLocalDictionary = + sanitized.batches[1].data.children[0].dictionary!; + expect(secondLocalDictionary.data[0]).toBe(firstLocalDictionary.data[0]); + + const buf = await fromTableToBuffer(sanitized); + const actual = currentTableFromIPC(buf); + expect([...actual.getChild("label")!]).toEqual(expected); + }); + + it("can serialize list data from another Arrow version", async function () { + const values = [["anime", "action"], [], null]; + const vector = vectorFromArray( + values, + new List(new Field("item", new Utf8(), true)), + ); + const table = new ArrowTable({ tags: vector }); + + const buf = await fromDataToBuffer(table); + const actual = currentTableFromIPC(buf); + const actualTags = actual.getChild("tags"); + + expect(actualTags?.get(0)?.toJSON()).toEqual(values[0]); + expect(actualTags?.get(1)?.toJSON()).toEqual(values[1]); + expect(actualTags?.get(2)).toBeNull(); + }); + + it("can still import data", async function () { + const schema = new arrow15.Schema([ + new arrow15.Field("id", new arrow15.Int32()), + new arrow15.Field( + "vector", + new arrow15.FixedSizeList( + 1024, + new arrow15.Field("item", new arrow15.Float32(), true), + ), + ), + new arrow15.Field( + "struct", + new arrow15.Struct([ + new arrow15.Field( + "nested", + new arrow15.Dictionary( + new arrow15.Utf8(), + new arrow15.Int32(), + 1, + true, + ), + ), + new arrow15.Field( + "ts_with_tz", + new arrow15.TimestampNanosecond("some_tz"), + ), + new arrow15.Field( + "ts_no_tz", + new arrow15.TimestampNanosecond(null), + ), + ]), + ), + // biome-ignore lint/suspicious/noExplicitAny: skip + ]) as any; + schema.metadataVersion = arrow15.MetadataVersion.V5; + const table = makeArrowTable([], { schema }); + + const buf = await fromTableToBuffer(table); + expect(buf.byteLength).toBeGreaterThan(0); + const actual = tableFromIPC(buf); + const actualSchema = actual.schema; + expect(actualSchema.fields.length).toBe(3); + + // Deep equality gets hung up on some very minor unimportant differences + // between arrow version 13 and 15 which isn't really what we're testing for + // and so we do our own comparison that just checks name/type/nullability + function compareFields(lhs: arrow15.Field, rhs: arrow15.Field) { + expect(lhs.name).toEqual(rhs.name); + expect(lhs.nullable).toEqual(rhs.nullable); + expect(lhs.typeId).toEqual(rhs.typeId); + if ("children" in lhs.type && lhs.type.children !== null) { + const lhsChildren = lhs.type.children as arrow15.Field[]; + lhsChildren.forEach((child: arrow15.Field, idx) => { + compareFields(child, rhs.type.children[idx]); + }); + } + } + // biome-ignore lint/suspicious/noExplicitAny: + actualSchema.fields.forEach((field: any, idx: string | number) => { + compareFields(field, actualSchema.fields[idx]); }); }); - }, -); + }); + + describe("converting record batches to buffers", function () { + it("can convert to buffered record batch and back again", async function () { + const records = [ + { text: "dog", vector: [0.1, 0.2] }, + { text: "cat", vector: [0.3, 0.4] }, + ]; + const table = await convertToTable(records); + const batch = table.batches[0]; + + const buffer = await fromRecordBatchToBuffer(batch); + const result = await fromBufferToRecordBatch(buffer); + + expect(JSON.stringify(batch.toArray())).toEqual( + JSON.stringify(result?.toArray()), + ); + }); + + it("converting from buffer returns null if buffer has no record batches", async function () { + const result = await fromBufferToRecordBatch(Buffer.from([0x01, 0x02])); // bad data + expect(result).toEqual(null); + }); + }); + + describe("boolean null handling", function () { + it("should handle null values in nullable boolean fields", () => { + const { makeArrowTable } = require("../lancedb/arrow"); + const schema = new Schema([new Field("test", new arrow.Bool(), true)]); + + // Test with all null values + const data = [{ test: null }]; + const table = makeArrowTable(data, { schema }); + + expect(table.numRows).toBe(1); + expect(table.schema.names).toEqual(["test"]); + expect(table.getChild("test")!.get(0)).toBeNull(); + }); + + it("should handle mixed null and non-null boolean values", () => { + const { makeArrowTable } = require("../lancedb/arrow"); + const schema = new Schema([new Field("test", new Bool(), true)]); + + // Test with mixed values + const data = [{ test: true }, { test: null }, { test: false }]; + const table = makeArrowTable(data, { schema }); + + expect(table.numRows).toBe(3); + expect(table.getChild("test")!.get(0)).toBe(true); + expect(table.getChild("test")!.get(1)).toBeNull(); + expect(table.getChild("test")!.get(2)).toBe(false); + }); + }); + + // Test for the undefined values bug fix + describe("undefined values handling", () => { + it("should handle mixed undefined and actual values", () => { + const schema = new Schema([ + new Field("text", new Utf8(), true), // nullable + new Field("number", new Int32(), true), // nullable + new Field("bool", new Bool(), true), // nullable + ]); + + const data = [ + { text: undefined, number: 42, bool: true }, + { text: "hello", number: undefined, bool: false }, + { text: "world", number: 123, bool: undefined }, + ]; + const table = makeArrowTable(data, { schema }); + + const result = table.toArray(); + expect(result).toHaveLength(3); + expect(result[0].text).toBe(null); + expect(result[0].number).toBe(42); + expect(result[0].bool).toBe(true); + expect(result[1].text).toBe("hello"); + expect(result[1].number).toBe(null); + expect(result[1].bool).toBe(false); + expect(result[2].text).toBe("world"); + expect(result[2].number).toBe(123); + expect(result[2].bool).toBe(null); + }); + }); +}); diff --git a/nodejs/__test__/arrow_versions.ts b/nodejs/__test__/arrow_versions.ts new file mode 100644 index 000000000..bfbbfa5ef --- /dev/null +++ b/nodejs/__test__/arrow_versions.ts @@ -0,0 +1,39 @@ +// SPDX-License-Identifier: Apache-2.0 +// SPDX-FileCopyrightText: Copyright The LanceDB Authors + +// Every Apache Arrow release the package accepts as a peer dependency, as +// separately installed copies so the tests can feed objects built by one +// Arrow version into a LanceDB running on another. When a new Arrow major +// is released, add the alias to `package.json` and extend this list; the +// test files pick the change up from here. +import * as arrow15 from "apache-arrow-15"; +import * as arrow16 from "apache-arrow-16"; +import * as arrow17 from "apache-arrow-17"; +import * as arrow18 from "apache-arrow-18"; +import * as arrow19 from "apache-arrow-19"; +import * as arrow20 from "apache-arrow-20"; +import * as arrow21 from "apache-arrow-21"; + +export { arrow15, arrow16, arrow17, arrow18, arrow19, arrow20, arrow21 }; + +export type ApacheArrow = + | typeof arrow15 + | typeof arrow16 + | typeof arrow17 + | typeof arrow18 + | typeof arrow19 + | typeof arrow20 + | typeof arrow21; + +export const arrowVersions: ApacheArrow[] = [ + arrow15, + arrow16, + arrow17, + arrow18, + arrow19, + arrow20, + arrow21, +]; + +/** The newest supported Arrow release, for tests that only need one. */ +export const latestArrow = arrow21; diff --git a/nodejs/__test__/registry.test.ts b/nodejs/__test__/registry.test.ts index 973ad8a25..5c92941d6 100644 --- a/nodejs/__test__/registry.test.ts +++ b/nodejs/__test__/registry.test.ts @@ -3,13 +3,10 @@ import * as apiArrow from "apache-arrow"; -import * as arrow15 from "apache-arrow-15"; -import * as arrow16 from "apache-arrow-16"; -import * as arrow17 from "apache-arrow-17"; -import * as arrow18 from "apache-arrow-18"; - import * as tmp from "tmp"; +import { arrowVersions } from "./arrow_versions"; + import { connect } from "../lancedb"; import { EmbeddingFunction, @@ -18,7 +15,7 @@ import { } from "../lancedb/embedding"; import { getRegistry, register } from "../lancedb/embedding/registry"; -describe.each([arrow15, arrow16, arrow17, arrow18])("LanceSchema", (arrow) => { +describe.each(arrowVersions)("LanceSchema", (arrow) => { test("should preserve input order", async () => { const schema = LanceSchema({ id: new arrow.Int32(), @@ -29,7 +26,7 @@ describe.each([arrow15, arrow16, arrow17, arrow18])("LanceSchema", (arrow) => { }); }); -describe.each([arrow15, arrow16, arrow17, arrow18])("Registry", (arrow) => { +describe.each(arrowVersions)("Registry", (arrow) => { let tmpDir: tmp.DirResult; beforeEach(() => { tmpDir = tmp.dirSync({ unsafeCleanup: true }); @@ -240,7 +237,7 @@ describe("Registry.setVar", () => { return data.map(() => [1, 2, 3]); } embeddingDataType() { - return new arrow18.Float32() as apiArrow.Float; + return new apiArrow.Float32(); } protected getSensitiveKeys() { return ["secretKey"]; diff --git a/nodejs/__test__/sanitize.test.ts b/nodejs/__test__/sanitize.test.ts index 587805d53..97856a55b 100644 --- a/nodejs/__test__/sanitize.test.ts +++ b/nodejs/__test__/sanitize.test.ts @@ -2,7 +2,13 @@ // SPDX-FileCopyrightText: Copyright The LanceDB Authors import * as arrow from "../lancedb/arrow"; -import { sanitizeField, sanitizeMap, sanitizeType } from "../lancedb/sanitize"; +import { + sanitizeField, + sanitizeMap, + sanitizeSchema, + sanitizeType, +} from "../lancedb/sanitize"; +import { arrow21 } from "./arrow_versions"; describe("sanitize", function () { describe("sanitizeType function", function () { @@ -59,6 +65,8 @@ describe("sanitize", function () { ["durationmicrosecond", new arrow.DurationMicrosecond()], ["durationmillisecond", new arrow.DurationMillisecond()], ["durationsecond", new arrow.DurationSecond()], + ["largebinary", new arrow.LargeBinary()], + ["largeutf8", new arrow.LargeUtf8()], ] as const; it.each(allTypeNameTestCases)( @@ -89,6 +97,110 @@ describe("sanitize", function () { "Unrecognized type name in schema: invalid_type", ); }); + + // These types only exist in Arrow 20 (IntervalMonthDayNano) and 21 (the + // rest), so the objects come from a second copy of the newest release. The + // library itself may be running on an older Arrow (see the CI matrix), in + // which case sanitizing them has to fail with a clear message instead. + const hostArrow = arrow as unknown as Record; + const hostSupportsNewerTypes = typeof hostArrow.Utf8View === "function"; + const whenSupported = hostSupportsNewerTypes ? describe : describe.skip; + const whenUnsupported = hostSupportsNewerTypes ? describe.skip : describe; + + const newerTypeTestCases = [ + ["Utf8View", () => new arrow21.Utf8View()], + ["BinaryView", () => new arrow21.BinaryView()], + ["IntervalMonthDayNano", () => new arrow21.IntervalMonthDayNano()], + [ + "LargeList", + () => + new arrow21.LargeList(new arrow21.Field("item", new arrow21.Utf8())), + ], + ] as const; + + whenSupported("types added in Arrow 20 and 21", function () { + it.each(newerTypeTestCases)( + "should sanitize a %s type from another copy of Arrow", + function (name, makeForeign) { + const foreign = makeForeign(); + const result = sanitizeType(foreign); + // Interval subclasses share `Type.Interval`, so the sanitized type + // is the generic `Interval` carrying the same unit, as for the + // other units. + const expected = name === "IntervalMonthDayNano" ? "Interval" : name; + expect(result).toBeInstanceOf(hostArrow[expected] as () => unknown); + expect(result.typeId).toBe(foreign.typeId); + expect(result.toString()).toBe(foreign.toString()); + }, + ); + + it.each([ + ["binaryview", "BinaryView"], + ["utf8view", "Utf8View"], + ["intervalmonthdaynano", "Interval"], + ])('should map type name "%s" to %s', function (name, expected) { + expect(sanitizeType(name)).toBeInstanceOf( + hostArrow[expected] as () => unknown, + ); + }); + + it("should sanitize the child field of a LargeList", function () { + const foreign = new arrow21.LargeList( + new arrow21.Field("item", new arrow21.Int32(), false), + ); + const result = sanitizeType(foreign) as arrow.List; + expect(result.children[0]).toBeInstanceOf(arrow.Field); + expect(result.children[0].name).toBe("item"); + expect(result.children[0].nullable).toBe(false); + const child = result.children[0].type; + expect(arrow.DataType.isInt(child)).toBe(true); + expect((child as arrow.Int).bitWidth).toBe(32); + expect((child as arrow.Int).isSigned).toBe(true); + }); + }); + + whenUnsupported("types newer than the installed Arrow", function () { + it.each( + newerTypeTestCases.filter(([name]) => name !== "IntervalMonthDayNano"), + )( + "should explain that %s needs a newer apache-arrow", + function (name, makeForeign) { + expect(() => sanitizeType(makeForeign())).toThrow( + `The Arrow type ${name} requires apache-arrow`, + ); + }, + ); + + it("should sanitize IntervalMonthDayNano to a generic Interval", () => { + // The subclass shares `Type.Interval`, which every supported Arrow + // release has; only the unit is new. + const result = sanitizeType(new arrow21.IntervalMonthDayNano()); + expect(result).toBeInstanceOf(arrow.Interval); + expect((result as arrow.Interval).unit).toBe( + arrow21.IntervalUnit.MONTH_DAY_NANO, + ); + }); + }); + + it("should rebuild a schema from a second copy of the same Arrow version", function () { + // Arrow 21 makes `instanceof Schema` succeed across copies of the library + // but not `instanceof `; the schema must still be rebuilt. + const foreign = new arrow21.Schema([ + new arrow21.Field("x", new arrow21.Float32(), true), + ]); + // Only Arrow 21+ overrides `Symbol.hasInstance` this way. + const crossCopyInstanceof = Object.getOwnPropertySymbols( + arrow.Schema, + ).includes(Symbol.hasInstance); + expect(foreign instanceof arrow.Schema).toBe(crossCopyInstanceof); + const result = sanitizeSchema(foreign); + expect(result).not.toBe(foreign); + const type = result.fields[0].type; + expect(type).toBeInstanceOf(arrow.DataType); + expect(arrow.DataType.isFloat(type)).toBe(true); + expect((type as arrow.Float).precision).toBe(arrow.Precision.SINGLE); + expect(() => arrow.makeBuilder({ type })).not.toThrow(); + }); }); describe("sanitizeField function", function () { diff --git a/nodejs/__test__/table.test.ts b/nodejs/__test__/table.test.ts index ae257455d..eaf59211c 100644 --- a/nodejs/__test__/table.test.ts +++ b/nodejs/__test__/table.test.ts @@ -2,15 +2,11 @@ // SPDX-FileCopyrightText: Copyright The LanceDB Authors import * as fs from "fs"; +import { getEventListeners } from "node:events"; import { pathToFileURL } from "node:url"; import * as path from "path"; import * as tmp from "tmp"; -import * as arrow15 from "apache-arrow-15"; -import * as arrow16 from "apache-arrow-16"; -import * as arrow17 from "apache-arrow-17"; -import * as arrow18 from "apache-arrow-18"; - import { AutoQuery, Connection, @@ -55,19 +51,21 @@ import { instanceOfFullTextQuery, } from "../lancedb/query"; import { LocalTable } from "../lancedb/table"; +import { + type ApacheArrow, + arrow18, + arrowVersions, + latestArrow, +} from "./arrow_versions"; -describe.each([arrow15, arrow16, arrow17, arrow18])( +describe.each(arrowVersions)( "Given a table", // biome-ignore lint/suspicious/noExplicitAny: (arrow: any) => { let tmpDir: tmp.DirResult; let table: Table; - const schema: - | import("apache-arrow-15").Schema - | import("apache-arrow-16").Schema - | import("apache-arrow-17").Schema - | import("apache-arrow-18").Schema = new arrow.Schema([ + const schema: InstanceType = new arrow.Schema([ new arrow.Field("id", new arrow.Float64(), true), ]); @@ -92,6 +90,38 @@ describe.each([arrow15, arrow16, arrow17, arrow18])( await expect(table.countRows()).resolves.toBe(3); }); + it("creates and adds Arrow tables with large string and binary columns", async () => { + const data = new arrow.Table({ + text: arrow.vectorFromArray(["alpha", "beta"], new arrow.LargeUtf8()), + bytes: arrow.vectorFromArray( + [Buffer.from("one"), Buffer.from("two")], + new arrow.LargeBinary(), + ), + }); + const conn = await connect(tmpDir.name); + const largeTable = await conn.createTable("large_columns", data); + await largeTable.add(data); + + const rows = await largeTable.query().toArray(); + expect(rows.map((row) => row.text)).toEqual([ + "alpha", + "beta", + "alpha", + "beta", + ]); + expect(rows.map((row) => Buffer.from(row.bytes).toString())).toEqual([ + "one", + "two", + "one", + "two", + ]); + const fields = (await largeTable.schema()).fields; + expect(fields.map((field) => field.type.typeId)).toEqual([ + arrow.Type.LargeUtf8, + arrow.Type.LargeBinary, + ]); + }); + it("should support a foreign Float64 vector schema end to end", async () => { const conn = await connect(tmpDir.name); const schema = new arrow.Schema([ @@ -709,6 +739,114 @@ describe.each([arrow15, arrow16, arrow17, arrow18])( ); // https://github.com/lancedb/lancedb/issues/1963 +// These types were added in Arrow 21, so they are built with a second copy of +// that release rather than the per-version matrix above, and reading them back +// needs the library's own Arrow to know them too. +const hostArrowSupportsViewTypes = + typeof (arrow as unknown as Record).Utf8View === "function"; +(hostArrowSupportsViewTypes ? it : it.skip)( + "creates and adds tables with LargeList, Utf8View and BinaryView columns", + async () => { + const tmpDir = tmp.dirSync({ unsafeCleanup: true }); + const db = await connect(tmpDir.name); + const makeData = () => + new latestArrow.Table({ + id: latestArrow.vectorFromArray([1, 2], new latestArrow.Int32()), + tags: latestArrow.vectorFromArray( + [["a", "b"], ["c"]], + new latestArrow.LargeList( + new latestArrow.Field("item", new latestArrow.Utf8(), true), + ), + ), + text: latestArrow.vectorFromArray( + ["short", "a string that is longer than the twelve inline bytes"], + new latestArrow.Utf8View(), + ), + bytes: latestArrow.vectorFromArray( + [new Uint8Array([1, 2, 3]), new Uint8Array(20).fill(7)], + new latestArrow.BinaryView(), + ), + }); + const table = await db.createTable("views", makeData()); + await table.add(makeData()); + + const schema = await table.schema(); + // Take the enum from the Arrow 21 copy: the host Arrow may predate + // `Type.LargeList`, and this file still has to type-check there. + expect(schema.fields.map((f) => f.type.typeId)).toEqual([ + latestArrow.Type.Int, + latestArrow.Type.LargeList, + // Lance stores the view types as their offset-based equivalents. + latestArrow.Type.Utf8, + latestArrow.Type.Binary, + ]); + const rows = await table.query().toArrow(); + expect(rows.numRows).toBe(4); + const row = rows.get(1)!.toJSON(); + expect(row.id).toBe(2); + expect(row.tags.toJSON()).toEqual(["c"]); + expect(row.text).toBe( + "a string that is longer than the twelve inline bytes", + ); + expect(Array.from(row.bytes)).toEqual(Array(20).fill(7)); + + // Arrow JS 21 serializes a *sliced* view column with a truncated views + // buffer; the library rebuilds view columns as Utf8/Binary so this works. + const sliced = makeData().slice(1, 2); + expect(sliced.getChild("text")!.data[0].offset).toBe(1); + const slicedTable = await db.createTable("views_sliced", sliced); + await slicedTable.add(sliced); + const slicedRows = await slicedTable.query().toArrow(); + expect(slicedRows.numRows).toBe(2); + for (const i of [0, 1]) { + const r = slicedRows.get(i)!.toJSON(); + expect(r.id).toBe(2); + expect(r.text).toBe( + "a string that is longer than the twelve inline bytes", + ); + expect(Array.from(r.bytes)).toEqual(Array(20).fill(7)); + } + + // Regression: a field with an integer-like name. JavaScript enumerates + // such keys first, so a key-ordered rebuild would pair "1" with the text + // column's values. The batches are built explicitly to pin field order. + const numericSchema = new latestArrow.Schema([ + new latestArrow.Field("text", new latestArrow.Utf8View(), true), + new latestArrow.Field("1", new latestArrow.Utf8View(), true), + new latestArrow.Field("n", new latestArrow.Int32(), true), + ]); + const column = ( + values: unknown[], + type: InstanceType, + ) => latestArrow.vectorFromArray(values, type).data[0]; + const numericBatch = new latestArrow.RecordBatch( + numericSchema, + latestArrow.makeData({ + type: new latestArrow.Struct(numericSchema.fields), + length: 3, + nullCount: 0, + children: [ + column(["t0", "t1", "t2"], new latestArrow.Utf8View()), + column(["one0", "one1", "one2"], new latestArrow.Utf8View()), + column([0, 1, 2], new latestArrow.Int32()), + ], + }), + ); + const numericData = new latestArrow.Table(numericSchema, [numericBatch]); + const numericTable = await db.createTable("views_numeric", numericData); + await numericTable.add(numericData.slice(1)); + const numericRows = (await numericTable.query().toArrow()).toArray(); + expect(numericRows.map((r) => r.toJSON())).toEqual([ + { text: "t0", "1": "one0", n: 0 }, + { text: "t1", "1": "one1", n: 1 }, + { text: "t2", "1": "one2", n: 2 }, + { text: "t1", "1": "one1", n: 1 }, + { text: "t2", "1": "one2", n: 2 }, + ]); + tmpDir.removeCallback(); + }, +); + it("should query documents with LangChain PDF metadata", async () => { const tmpDir = tmp.dirSync({ unsafeCleanup: true }); try { @@ -2601,6 +2739,27 @@ describe("when dealing with blob columns", () => { expect(Buffer.from(await files[1]!.read()).toString()).toBe("beta"); }); + it("adds a LargeBinary Arrow column to a blob table", async () => { + const db = await connect(tmpDir.name); + const schema = new Schema([new Field("id", new Int64()), blob("image")]); + const table = await db.createTable( + "blobs", + [{ id: 1n, image: Buffer.from("alpha") }], + { schema }, + ); + const payload = Buffer.from("beta"); + const data = new arrow18.Table({ + id: arrow18.vectorFromArray([2n], new arrow18.Int64()), + image: arrow18.vectorFromArray([payload], new arrow18.LargeBinary()), + }); + await table.add(data); + + const rows = await table.query().withRowId().toArray(); + const rowId = rows.find((row) => row.id === 2n)!._rowid as bigint; + const [actual] = await table.fetchBlobs("image", [rowId]); + expect(actual).toEqual(payload); + }); + it("reads a half-open range", async () => { const { table, rowIds } = await openBlobTable(); const files = await table.fetchBlobFiles("image", rowIds); @@ -2625,6 +2784,148 @@ describe("when dealing with blob columns", () => { ).rejects.toThrow(/exceeds blob size/); }); + it("reads at most maxBytes and advances the cursor", async () => { + const { table, rowIds } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + expect((await handle!.read(2n)).toString()).toBe("al"); + expect(await handle!.tell()).toBe(2n); + expect((await handle!.read(2n)).toString()).toBe("ph"); + expect((await handle!.read(10n)).toString()).toBe("a"); + expect(await handle!.tell()).toBe(5n); + expect(await handle!.read(1n)).toEqual(Buffer.alloc(0)); + }); + + it("seeks the cursor", async () => { + const { table, rowIds, alpha } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + await handle!.seek(3n); + expect(await handle!.tell()).toBe(3n); + expect((await handle!.read()).toString()).toBe("ha"); + await handle!.seek(0n); + expect(await handle!.read()).toEqual(alpha); + }); + + it("rejects a negative seek position", async () => { + const { table, rowIds } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + await expect(handle!.seek(-1n)).rejects.toThrow(/cannot be negative/); + }); + + it("readRanges returns one buffer per range in order", async () => { + const { table, rowIds } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + const buffers = await handle!.readRanges([ + { start: 3n, end: 5n }, + { start: 0n, end: 2n }, + { start: 3n, end: 5n }, + ]); + expect(buffers.map((b) => b.toString())).toEqual(["ha", "al", "ha"]); + expect(await handle!.tell()).toBe(0n); + }); + + it("fails when a readRanges end is past the blob size", async () => { + const { table, rowIds, alpha } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + await expect( + handle!.readRanges([ + { start: 0n, end: 1n }, + { start: 0n, end: BigInt(alpha.length + 1) }, + ]), + ).rejects.toThrow(/exceeds blob size/); + }); + + it("fails reads after close and closes once", async () => { + const { table, rowIds } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + expect(handle!.isClosed()).toBe(false); + await handle!.close(); + expect(handle!.isClosed()).toBe(true); + await expect(handle!.close()).resolves.toBeUndefined(); + await expect(handle!.read()).rejects.toThrow(/already closed/); + await expect(handle!.read(1n)).rejects.toThrow(/already closed/); + await expect(handle!.readRange(0n, 1n)).rejects.toThrow(/already closed/); + await expect(handle!.readRanges([{ start: 0n, end: 1n }])).rejects.toThrow( + /already closed/, + ); + await expect(handle!.seek(0n)).rejects.toThrow(/already closed/); + await expect(handle!.tell()).rejects.toThrow(/already closed/); + expect(handle!.size()).toBe(5n); + }); + + it("rejects reads with an already aborted signal", async () => { + const { table, rowIds, alpha } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + const signal = AbortSignal.abort(); + const abortError = { name: "AbortError" }; + await expect(handle!.read(undefined, { signal })).rejects.toMatchObject( + abortError, + ); + await expect(handle!.read(2n, { signal })).rejects.toMatchObject( + abortError, + ); + await expect(handle!.readRange(0n, 1n, { signal })).rejects.toMatchObject( + abortError, + ); + await expect( + handle!.readRanges([{ start: 0n, end: 1n }], { signal }), + ).rejects.toMatchObject(abortError); + await expect( + table.fetchBlobs("image", rowIds, { signal }), + ).rejects.toMatchObject(abortError); + await expect( + table.fetchBlobFiles("image", rowIds, { signal }), + ).rejects.toMatchObject(abortError); + expect(await handle!.read()).toEqual(alpha); + }); + + it("rejects with the abort reason", async () => { + const { table, rowIds } = await openBlobTable(); + const [handle] = await table.fetchBlobFiles("image", rowIds); + const reason = new Error("caller gave up"); + await expect( + handle!.readRange(0n, 1n, { signal: AbortSignal.abort(reason) }), + ).rejects.toBe(reason); + }); + + it("reads with a signal and removes its abort listener", async () => { + const { table, rowIds, alpha, beta } = await openBlobTable(); + const controller = new AbortController(); + const { signal } = controller; + const [handle] = await table.fetchBlobFiles("image", rowIds, { signal }); + expect((await handle!.readRange(0n, 2n, { signal })).toString()).toBe("al"); + expect(await table.fetchBlobs("image", rowIds, { signal })).toEqual([ + alpha, + beta, + null, + ]); + expect(getEventListeners(signal, "abort")).toHaveLength(0); + controller.abort(); + expect(await handle!.read()).toEqual(alpha); + }); + + it("cancels an in-flight read", async () => { + const db = await connect(tmpDir.name); + const schema = new Schema([ + new Field("id", new Int64(), true), + blob("payload", { inlineSizeThreshold: 0, dedicatedSizeThreshold: 1 }), + ]); + const payload = Buffer.alloc(32 * 1024 * 1024, 7); + const table = await db.createTable("large_blobs", [{ id: 1n, payload }], { + schema, + }); + const [row] = await table.query().withRowId().toArray(); + const [handle] = await table.fetchBlobFiles("payload", [ + row._rowid as bigint, + ]); + const controller = new AbortController(); + const pending = handle!.readRange(0n, handle!.size(), { + signal: controller.signal, + }); + controller.abort(); + await expect(pending).rejects.toMatchObject({ name: "AbortError" }); + expect(await handle!.readRange(0n, 4n)).toEqual(payload.subarray(0, 4)); + }); + it("rejects fetchBlobs on a non-blob column", async () => { const { table, rowIds } = await openBlobTable(); await expect(table.fetchBlobs("id", rowIds)).rejects.toThrow(/blob/i); @@ -3007,7 +3308,7 @@ it("passes cleanupOlderThan to the native binding as an absolute timestamp", asy expect(optimize).toHaveBeenCalledWith(cutoff.getTime(), true); }); -describe.each([arrow15, arrow16, arrow17, arrow18])( +describe.each(arrowVersions)( "when optimizing a dataset", // biome-ignore lint/suspicious/noExplicitAny: (arrow: any) => { diff --git a/nodejs/lancedb/arrow.ts b/nodejs/lancedb/arrow.ts index df875a808..3daee7d65 100644 --- a/nodejs/lancedb/arrow.ts +++ b/nodejs/lancedb/arrow.ts @@ -66,11 +66,20 @@ export type SchemaLike = export type FieldLike = | Field | { - type: string; + type: string | DataTypeLike; name: string; nullable: boolean; metadata?: Map; }; +/** + * A `DataType` from any copy or version of apache-arrow. + * + * Arrow 21 brands its classes with `unique symbol` properties, so a type + * object from a second copy of the library no longer satisfies the `DataType` + * type of this one even though it is structurally identical. Inputs that only + * need to be sanitized accept this looser shape instead. + */ +export type DataTypeLike = DataType | { readonly typeId: number }; /** * Create an Arrow field backed by LanceDB's JSON extension type. @@ -1059,6 +1068,71 @@ export async function fromRecordsToStreamBuffer( return Buffer.from(await writer.toUint8Array()); } +// `Type.Utf8View` / `Type.BinaryView` as numbers: the enum members only exist +// in Arrow 21+, and this module compiles against every supported release. +const UTF8_VIEW_TYPE_ID = 24; +const BINARY_VIEW_TYPE_ID = 23; + +/** + * Copy a Utf8View / BinaryView `Data` into a single `Data` of `type`. + */ +function materializeViewData(child: ArrowData, type: DataType): ArrowData { + const builder = makeBuilder({ type, nullValues: [null] }); + for (const value of new Vector([child])) { + builder.append(value); + } + return builder.finish().flush(); +} + +/** + * Rebuild any top-level Utf8View / BinaryView column as Utf8 / Binary. + * + * Lance stores the view types as their offset-based equivalents anyway, so + * nothing is lost. Doing it here also sidesteps an Arrow JS 21 bug: its IPC + * writer emits a truncated views buffer for a *sliced* view array, which the + * Rust reader rejects with "Need at least N bytes in buffers[0]". + * + * The record batches are rebuilt positionally rather than through a + * `Record`: JavaScript enumerates integer-like keys first, so + * a field named e.g. `"1"` would otherwise be paired with the wrong column. + * + * Tables without view columns are returned as-is. + */ +function materializeViewColumns(table: ArrowTable): ArrowTable { + const replacements = new Map(); + table.schema.fields.forEach((field, i) => { + if (field.type.typeId === UTF8_VIEW_TYPE_ID) { + replacements.set(i, new Utf8()); + } else if (field.type.typeId === BINARY_VIEW_TYPE_ID) { + replacements.set(i, new Binary()); + } + }); + if (replacements.size === 0) { + return table; + } + const fields = table.schema.fields.map((field, i) => { + const type = replacements.get(i); + return type === undefined + ? field + : new Field(field.name, type, field.nullable, field.metadata); + }); + const schema = new Schema(fields, table.schema.metadata); + const batches = table.batches.map((batch) => { + const children = batch.data.children.map((child, i) => { + const type = replacements.get(i); + return type === undefined ? child : materializeViewData(child, type); + }); + const data = makeData({ + type: new Struct(fields), + length: batch.numRows, + nullCount: 0, + children, + }); + return new RecordBatch(schema, data); + }); + return new ArrowTable(schema, batches); +} + /** * Serialize an Arrow Table into a buffer using the Arrow IPC File serialization * @@ -1075,7 +1149,9 @@ export async function fromTableToBuffer( if (schema !== undefined && schema !== null) { schema = sanitizeSchema(schema); } - const tableWithEmbeddings = await applyEmbeddings(table, embeddings, schema); + const tableWithEmbeddings = materializeViewColumns( + await applyEmbeddings(table, embeddings, schema), + ); validateBlobSchema(tableWithEmbeddings.schema); const writer = RecordBatchFileWriter.writeAll(tableWithEmbeddings); return Buffer.from(await writer.toUint8Array()); @@ -1146,7 +1222,8 @@ export async function fromRecordBatchToBuffer( export async function fromRecordBatchToStreamBuffer( batch: RecordBatch, ): Promise { - const writer = RecordBatchStreamWriter.writeAll([batch]); + const table = materializeViewColumns(new ArrowTable([batch])); + const writer = RecordBatchStreamWriter.writeAll(table); return Buffer.from(await writer.toUint8Array()); } @@ -1163,7 +1240,9 @@ export async function fromTableToStreamBuffer( embeddings?: EmbeddingFunctionConfig, schema?: SchemaLike, ): Promise { - const tableWithEmbeddings = await applyEmbeddings(table, embeddings, schema); + const tableWithEmbeddings = materializeViewColumns( + await applyEmbeddings(table, embeddings, schema), + ); const writer = RecordBatchStreamWriter.writeAll(tableWithEmbeddings); return Buffer.from(await writer.toUint8Array()); } diff --git a/nodejs/lancedb/blob.ts b/nodejs/lancedb/blob.ts index 244faac09..e6c780ca9 100644 --- a/nodejs/lancedb/blob.ts +++ b/nodejs/lancedb/blob.ts @@ -141,13 +141,16 @@ export class BlobFile { } /** - * Reads from the cursor to the end and advances the cursor. + * Reads from the cursor and advances the cursor. * - * A second call returns an empty buffer. {@link BlobFile.readRange} does - * not move the cursor. + * Reads to the end when `maxBytes` is omitted, or at most `maxBytes` bytes + * otherwise. Returns an empty buffer at the end of the blob. + * {@link BlobFile.readRange} does not move the cursor. */ - read(): Promise { - return this.inner.read(); + read(maxBytes?: bigint, options?: BlobReadOptions): Promise { + return runWithSignal(options?.signal, (signal) => + this.inner.read(maxBytes, signal), + ); } /** @@ -155,11 +158,96 @@ export class BlobFile { * * Fails when `end` is past the blob size. Does not move the cursor. */ - readRange(start: bigint, end: bigint): Promise { - return this.inner.readRange(start, end); + readRange( + start: bigint, + end: bigint, + options?: BlobReadOptions, + ): Promise { + return runWithSignal(options?.signal, (signal) => + this.inner.readRange(start, end, signal), + ); + } + + /** + * Reads several half-open byte ranges. Returns one buffer per range, in + * the order given. + * + * Fails when any `end` is past the blob size. Does not move the cursor. + */ + readRanges( + ranges: BlobRange[], + options?: BlobReadOptions, + ): Promise { + return runWithSignal(options?.signal, (signal) => + this.inner.readRanges(ranges, signal), + ); + } + + /** Moves the cursor to `position`, in bytes from the start of the blob. */ + seek(position: bigint): Promise { + return this.inner.seek(position); + } + + /** Returns the cursor position in bytes. */ + tell(): Promise { + return this.inner.tell(); + } + + /** + * Releases the handle. Reads after `close()` fail. Calling it again does + * nothing. + */ + close(): Promise { + return this.inner.close(); + } + + /** Returns true after {@link BlobFile.close}. */ + isClosed(): boolean { + return this.inner.isClosed(); } } +/** Options for blob reads. */ +export type BlobReadOptions = { + /** + * Cancels the read. The call rejects with `signal.reason`, and the native + * read stops, including in-flight requests to a remote table. + */ + signal?: AbortSignal; +}; + +/** @ignore */ +export async function runWithSignal( + signal: AbortSignal | undefined, + run: (signal: AbortSignal | undefined) => Promise, +): Promise { + if (signal === undefined) { + return run(undefined); + } + signal.throwIfAborted(); + // napi-rs replaces `onabort` on the signal it receives, so pass a private + // signal and forward the caller's abort to it. + const controller = new AbortController(); + const onAbort = () => controller.abort(); + signal.addEventListener("abort", onAbort, { once: true }); + try { + return await run(controller.signal); + } catch (err) { + if (signal.aborted) { + throw signal.reason; + } + throw err; + } finally { + signal.removeEventListener("abort", onAbort); + } +} + +/** A half-open byte range `[start, end)` for {@link BlobFile.readRanges}. */ +export type BlobRange = { + start: bigint; + end: bigint; +}; + export function coerceBlobValue(value: unknown): BlobInput | null { if (value == null) { return null; diff --git a/nodejs/lancedb/index.ts b/nodejs/lancedb/index.ts index ed6889232..9d4e5608b 100644 --- a/nodejs/lancedb/index.ts +++ b/nodejs/lancedb/index.ts @@ -84,7 +84,7 @@ export { } from "./arrow"; export { blob, isBlobField, BlobFile } from "./blob"; -export type { BlobOptions } from "./blob"; +export type { BlobOptions, BlobRange, BlobReadOptions } from "./blob"; export { Connection, @@ -195,6 +195,7 @@ export * as rerankers from "./rerankers"; export { SchemaLike, TableLike, + DataTypeLike, FieldLike, RecordBatchLike, DataLike, diff --git a/nodejs/lancedb/sanitize.ts b/nodejs/lancedb/sanitize.ts index 3e5d583c4..6ba4cc881 100644 --- a/nodejs/lancedb/sanitize.ts +++ b/nodejs/lancedb/sanitize.ts @@ -9,6 +9,7 @@ // comes from the exact same library instance. This is not always the case // and so we must sanitize the input to ensure that it is compatible. +import * as arrowModule from "apache-arrow"; import { BufferType, Data, Vector } from "apache-arrow"; import type { IntBitWidth, TKeys, TimeBitWidth } from "apache-arrow/type"; import { @@ -43,6 +44,8 @@ import { Interval, IntervalDayTime, IntervalYearMonth, + LargeBinary, + LargeUtf8, List, Map_, Null, @@ -74,6 +77,66 @@ import { Utf8, } from "./arrow"; +/** + * Whether `value` was created by *this* copy of apache-arrow's `ctor`. + * + * Arrow 21 overrides `Symbol.hasInstance` on `Schema`, `Field`, `Table`, + * `RecordBatch`, `Vector` and `Data` so that `instanceof` also matches objects + * built by a different copy of the library. Those are exactly the objects + * this module has to rebuild: the `DataType` instances nested inside them are + * deliberately *not* given that treatment upstream, so Arrow's own builders and + * writers reject them. Walk the real prototype chain instead of trusting + * `instanceof`. + */ +function isOwnInstance( + value: unknown, + // biome-ignore lint/suspicious/noExplicitAny: matches any class constructor + ctor: abstract new (...args: any[]) => T, +): value is T { + return ( + typeof value === "object" && + value !== null && + Object.prototype.isPrototypeOf.call(ctor.prototype, value) + ); +} + +/** + * Types added to Arrow JS after release 18, keyed by the `Type` enum value + * they use. The library compiles and runs against every Arrow release in the + * peer range, so these are looked up by name at runtime instead of imported: + * an older Arrow simply does not have the class. + */ +type NewerTypeName = + | "IntervalMonthDayNano" + | "LargeList" + | "BinaryView" + | "Utf8View"; +const NEWER_TYPES: ReadonlyArray<{ + name: NewerTypeName; + typeId: number; + since: number; +}> = [ + { name: "IntervalMonthDayNano", typeId: -31, since: 20 }, + { name: "LargeList", typeId: 21, since: 21 }, + { name: "BinaryView", typeId: 23, since: 21 }, + { name: "Utf8View", typeId: 24, since: 21 }, +]; + +// biome-ignore lint/suspicious/noExplicitAny: constructor arity varies by type +type DataTypeConstructor = new (...args: any[]) => DataType; + +function newerTypeConstructor(name: NewerTypeName): DataTypeConstructor { + const ctor = (arrowModule as unknown as Record)[name]; + if (typeof ctor !== "function") { + const since = NEWER_TYPES.find((t) => t.name === name)?.since; + throw new Error( + `The Arrow type ${name} requires apache-arrow ${since} or newer, ` + + "but the installed apache-arrow does not provide it", + ); + } + return ctor as DataTypeConstructor; +} + type SanitizationContext = { types: WeakMap; vectors: WeakMap; @@ -210,19 +273,37 @@ export function sanitizeList(typeLike: object) { return sanitizeListWithContext(typeLike, createSanitizationContext()); } +function sanitizeListChildWithContext( + typeLike: object, + context: SanitizationContext, + typeName: string, +): Field { + if (!("children" in typeLike) || !Array.isArray(typeLike.children)) { + throw Error( + `Expected a ${typeName} type to have an array-like \`children\` property`, + ); + } + if (typeLike.children.length !== 1) { + throw Error(`Expected a ${typeName} type to have exactly one child`); + } + return sanitizeFieldWithContext(typeLike.children[0], context); +} + function sanitizeListWithContext( typeLike: object, context: SanitizationContext, ) { - if (!("children" in typeLike) || !Array.isArray(typeLike.children)) { - throw Error( - "Expected a List type to have an array-like `children` property", - ); - } - if (typeLike.children.length !== 1) { - throw Error("Expected a List type to have exactly one child"); - } - return new List(sanitizeFieldWithContext(typeLike.children[0], context)); + return new List(sanitizeListChildWithContext(typeLike, context, "List")); +} + +function sanitizeLargeListWithContext( + typeLike: object, + context: SanitizationContext, +) { + const LargeList = newerTypeConstructor("LargeList"); + return new LargeList( + sanitizeListChildWithContext(typeLike, context, "LargeList"), + ); } export function sanitizeStruct(typeLike: object) { @@ -449,11 +530,30 @@ function sanitizeTypeWithContext( return type; } +function sanitizeNewerTypeById( + typeLike: object, + typeId: number, + context: SanitizationContext, +): DataType | undefined { + const entry = NEWER_TYPES.find((t) => t.typeId === typeId); + if (entry === undefined) { + return undefined; + } + if (entry.name === "LargeList") { + return sanitizeLargeListWithContext(typeLike, context); + } + return new (newerTypeConstructor(entry.name))(); +} + function sanitizeTypeById( typeLike: object, typeId: Type, context: SanitizationContext, ): DataType { + const newer = sanitizeNewerTypeById(typeLike, typeId, context); + if (newer !== undefined) { + return newer; + } switch (typeId) { case Type.NONE: throw Error("Received a Type with a typeId of NONE"); @@ -465,8 +565,12 @@ function sanitizeTypeById( return sanitizeFloat(typeLike); case Type.Binary: return new Binary(); + case Type.LargeBinary: + return new LargeBinary(); case Type.Utf8: return new Utf8(); + case Type.LargeUtf8: + return new LargeUtf8(); case Type.Bool: return new Bool(); case Type.Decimal: @@ -566,7 +670,7 @@ function sanitizeFieldWithContext( fieldLike: unknown, context: SanitizationContext, ): Field { - if (fieldLike instanceof Field) { + if (isOwnInstance(fieldLike, Field)) { return fieldLike; } if (typeof fieldLike !== "object" || fieldLike === null) { @@ -620,7 +724,7 @@ function sanitizeSchemaWithContext( schemaLike: SchemaLike, context: SanitizationContext, ): Schema { - if (schemaLike instanceof Schema) { + if (isOwnInstance(schemaLike, Schema)) { return schemaLike; } if (typeof schemaLike !== "object" || schemaLike === null) { @@ -647,7 +751,7 @@ function sanitizeSchemaWithContext( } export function sanitizeTable(tableLike: TableLike): Table { - if (tableLike instanceof Table) { + if (isOwnInstance(tableLike, Table)) { return tableLike; } if (typeof tableLike !== "object" || tableLike === null) { @@ -675,7 +779,7 @@ function sanitizeRecordBatch( batchLike: RecordBatchLike, context: SanitizationContext, ): RecordBatch { - if (batchLike instanceof RecordBatch) { + if (isOwnInstance(batchLike, RecordBatch)) { return batchLike; } if (typeof batchLike !== "object" || batchLike === null) { @@ -704,11 +808,17 @@ type DictionaryDataLike = DataLike & { dictionary?: DictionaryVectorLike; }; +// Arrow 21 added `variadicBuffers` to `Data` for the Utf8View / BinaryView +// types. Older Arrow versions do not have the property, so it is optional. +type VariadicDataLike = DataLike & { + variadicBuffers?: ReadonlyArray; +}; + function sanitizeData( dataLike: DataLike, context: SanitizationContext, ): Data { - if (dataLike instanceof Data) { + if (isOwnInstance(dataLike, Data)) { return dataLike; } const cachedData = context.data.get(dataLike); @@ -726,7 +836,12 @@ function sanitizeData( context.vectors.set(dictionaryLike, dictionary); } } - const data = new Data( + // Arrow 21 added an eighth constructor parameter (`variadicBuffers`); older + // releases ignore the extra argument but their typings do not declare it. + const DataCtor = Data as unknown as new ( + ...args: unknown[] + ) => Data; + const data = new DataCtor( sanitizeTypeWithContext(dataLike.type, context), dataLike.offset, dataLike.length, @@ -739,6 +854,7 @@ function sanitizeData( }, dataLike.children.map((child) => sanitizeData(child, context)), dictionary, + (dataLike as VariadicDataLike).variadicBuffers, ); context.data.set(dataLike, data); return data; @@ -747,7 +863,11 @@ function sanitizeData( const constructorsByTypeName = { null: () => new Null(), binary: () => new Binary(), + largebinary: () => new LargeBinary(), + binaryview: () => new (newerTypeConstructor("BinaryView"))(), utf8: () => new Utf8(), + largeutf8: () => new LargeUtf8(), + utf8view: () => new (newerTypeConstructor("Utf8View"))(), bool: () => new Bool(), int8: () => new Int8(), int16: () => new Int16(), @@ -768,6 +888,8 @@ const constructorsByTypeName = { timesecond: () => new TimeSecond(), intervaldaytime: () => new IntervalDayTime(), intervalyearmonth: () => new IntervalYearMonth(), + intervalmonthdaynano: () => + new (newerTypeConstructor("IntervalMonthDayNano"))(), durationnanosecond: () => new DurationNanosecond(), durationmicrosecond: () => new DurationMicrosecond(), durationmillisecond: () => new DurationMillisecond(), diff --git a/nodejs/lancedb/table.ts b/nodejs/lancedb/table.ts index 132e8bc20..d4c5db509 100644 --- a/nodejs/lancedb/table.ts +++ b/nodejs/lancedb/table.ts @@ -17,7 +17,7 @@ import { tableFromIPC, } from "./arrow"; -import { BlobFile } from "./blob"; +import { BlobFile, BlobReadOptions, runWithSignal } from "./blob"; import { EmbeddingFunctionConfig, getRegistry } from "./embedding/registry"; import { IndexOptions } from "./indices"; import { Job } from "./job"; @@ -545,6 +545,7 @@ export abstract class Table { abstract fetchBlobs( column: string, rowIds: readonly (bigint | number)[], + options?: BlobReadOptions, ): Promise<(Buffer | null)[]>; /** @@ -557,6 +558,7 @@ export abstract class Table { abstract fetchBlobFiles( column: string, rowIds: readonly (bigint | number)[], + options?: BlobReadOptions, ): Promise<(BlobFile | null)[]>; /** @@ -1248,8 +1250,11 @@ export class LocalTable extends Table { async fetchBlobs( column: string, rowIds: readonly (bigint | number)[], + options?: BlobReadOptions, ): Promise<(Buffer | null)[]> { - const values = await this.inner.fetchBlobs(column, rowIdsToBigInts(rowIds)); + const values = await runWithSignal(options?.signal, (signal) => + this.inner.fetchBlobs(column, rowIdsToBigInts(rowIds), signal), + ); // N-API Option maps missing values to undefined. Collapse those to null. return values.map((value) => value ?? null); } @@ -1257,10 +1262,10 @@ export class LocalTable extends Table { async fetchBlobFiles( column: string, rowIds: readonly (bigint | number)[], + options?: BlobReadOptions, ): Promise<(BlobFile | null)[]> { - const files = await this.inner.fetchBlobFiles( - column, - rowIdsToBigInts(rowIds), + const files = await runWithSignal(options?.signal, (signal) => + this.inner.fetchBlobFiles(column, rowIdsToBigInts(rowIds), signal), ); // N-API Option maps missing values to undefined. Collapse those to null. return files.map((file) => diff --git a/nodejs/npm/darwin-arm64/package.json b/nodejs/npm/darwin-arm64/package.json index 0e9d131b0..051bdb98f 100644 --- a/nodejs/npm/darwin-arm64/package.json +++ b/nodejs/npm/darwin-arm64/package.json @@ -1,6 +1,6 @@ { "name": "@lancedb/lancedb-darwin-arm64", - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "os": ["darwin"], "cpu": ["arm64"], "main": "lancedb.darwin-arm64.node", diff --git a/nodejs/npm/linux-arm64-gnu/package.json b/nodejs/npm/linux-arm64-gnu/package.json index 39002b700..755035ee2 100644 --- a/nodejs/npm/linux-arm64-gnu/package.json +++ b/nodejs/npm/linux-arm64-gnu/package.json @@ -1,6 +1,6 @@ { "name": "@lancedb/lancedb-linux-arm64-gnu", - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "os": ["linux"], "cpu": ["arm64"], "main": "lancedb.linux-arm64-gnu.node", diff --git a/nodejs/npm/linux-arm64-musl/package.json b/nodejs/npm/linux-arm64-musl/package.json index 232796091..d31737229 100644 --- a/nodejs/npm/linux-arm64-musl/package.json +++ b/nodejs/npm/linux-arm64-musl/package.json @@ -1,6 +1,6 @@ { "name": "@lancedb/lancedb-linux-arm64-musl", - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "os": ["linux"], "cpu": ["arm64"], "main": "lancedb.linux-arm64-musl.node", diff --git a/nodejs/npm/linux-x64-gnu/package.json b/nodejs/npm/linux-x64-gnu/package.json index cc9c003e0..618df080d 100644 --- a/nodejs/npm/linux-x64-gnu/package.json +++ b/nodejs/npm/linux-x64-gnu/package.json @@ -1,6 +1,6 @@ { "name": "@lancedb/lancedb-linux-x64-gnu", - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "os": ["linux"], "cpu": ["x64"], "main": "lancedb.linux-x64-gnu.node", diff --git a/nodejs/npm/linux-x64-musl/package.json b/nodejs/npm/linux-x64-musl/package.json index 2f0356da2..87628c0c3 100644 --- a/nodejs/npm/linux-x64-musl/package.json +++ b/nodejs/npm/linux-x64-musl/package.json @@ -1,6 +1,6 @@ { "name": "@lancedb/lancedb-linux-x64-musl", - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "os": ["linux"], "cpu": ["x64"], "main": "lancedb.linux-x64-musl.node", diff --git a/nodejs/npm/win32-arm64-msvc/package.json b/nodejs/npm/win32-arm64-msvc/package.json index 2f96d1019..65c7b500c 100644 --- a/nodejs/npm/win32-arm64-msvc/package.json +++ b/nodejs/npm/win32-arm64-msvc/package.json @@ -1,6 +1,6 @@ { "name": "@lancedb/lancedb-win32-arm64-msvc", - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "os": [ "win32" ], diff --git a/nodejs/npm/win32-x64-msvc/package.json b/nodejs/npm/win32-x64-msvc/package.json index c941d597f..ad6a81acc 100644 --- a/nodejs/npm/win32-x64-msvc/package.json +++ b/nodejs/npm/win32-x64-msvc/package.json @@ -1,6 +1,6 @@ { "name": "@lancedb/lancedb-win32-x64-msvc", - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "os": ["win32"], "cpu": ["x64"], "main": "lancedb.win32-x64-msvc.node", diff --git a/nodejs/package.json b/nodejs/package.json index bd37d97e2..da5dc77b2 100644 --- a/nodejs/package.json +++ b/nodejs/package.json @@ -11,7 +11,7 @@ "ann" ], "private": false, - "version": "0.40.0-beta.11", + "version": "0.40.0-beta.12", "main": "dist/index.js", "exports": { ".": "./dist/index.js", @@ -49,10 +49,14 @@ "@types/jest": "^29.1.2", "@types/node": "22.7.4", "@types/tmp": "^0.2.6", + "apache-arrow": "21.2.0", "apache-arrow-15": "npm:apache-arrow@15.0.0", "apache-arrow-16": "npm:apache-arrow@16.0.0", "apache-arrow-17": "npm:apache-arrow@17.0.0", "apache-arrow-18": "npm:apache-arrow@18.0.0", + "apache-arrow-19": "npm:apache-arrow@19.0.1", + "apache-arrow-20": "npm:apache-arrow@20.0.0", + "apache-arrow-21": "npm:apache-arrow@21.2.0", "eslint": "^8.57.0", "jest": "^29.7.0", "shx": "^0.3.4", @@ -102,7 +106,7 @@ }, "peerDependencies": { "@types/node": ">=22", - "apache-arrow": ">=15.0.0 <=18.1.0" + "apache-arrow": ">=15.0.0 <22.0.0" }, "peerDependenciesMeta": { "@types/node": { diff --git a/nodejs/pnpm-lock.yaml b/nodejs/pnpm-lock.yaml index a03838999..4c19305c8 100644 --- a/nodejs/pnpm-lock.yaml +++ b/nodejs/pnpm-lock.yaml @@ -14,9 +14,6 @@ importers: '@opentelemetry/api': specifier: ^1.9.0 version: 1.9.1 - apache-arrow: - specifier: '>=15.0.0 <=18.1.0' - version: 18.1.0 reflect-metadata: specifier: ^0.2.2 version: 0.2.2 @@ -54,6 +51,9 @@ importers: '@types/tmp': specifier: ^0.2.6 version: 0.2.6 + apache-arrow: + specifier: 21.2.0 + version: 21.2.0 apache-arrow-15: specifier: npm:apache-arrow@15.0.0 version: apache-arrow@15.0.0 @@ -66,6 +66,15 @@ importers: apache-arrow-18: specifier: npm:apache-arrow@18.0.0 version: apache-arrow@18.0.0 + apache-arrow-19: + specifier: npm:apache-arrow@19.0.1 + version: apache-arrow@19.0.1 + apache-arrow-20: + specifier: npm:apache-arrow@20.0.0 + version: apache-arrow@20.0.0 + apache-arrow-21: + specifier: npm:apache-arrow@21.2.0 + version: apache-arrow@21.2.0 eslint: specifier: ^8.57.0 version: 8.57.1 @@ -1703,6 +1712,9 @@ packages: '@types/node@22.7.4': resolution: {integrity: sha512-y+NPi1rFzDs1NdQHHToqeiX2TIS79SWEAw9GYhkkx8bD0ChpfqC+n2j5OXOCpzfojBEBt6DnEnnG9MY0zk1XLg==} + '@types/node@25.9.8': + resolution: {integrity: sha512-VfMrScDmMhUJQmd5hArdQnFvK0OIeD36uN2Va1FcpYaLB/BgkgM8Ulc50XtYISPMz7APJ30+N0T5EM0jlcdfRw==} + '@types/stack-utils@2.0.3': resolution: {integrity: sha512-9aEbYZ3TbYMznPdcdr3SmIrLXwC/AKZXQeCf9Pgao5CKb8CyHuEX5jzWPTkvregvhRJHcpRO6BFoGW9ycaOkYw==} @@ -1840,8 +1852,16 @@ packages: resolution: {integrity: sha512-gFlPaqN9osetbB83zC29AbbZqGiCuFH1vyyPseJ+B7SIbfBtESV62mMT/CkiIt77W6ykC/nTWFzTXFs0Uldg4g==} hasBin: true - apache-arrow@18.1.0: - resolution: {integrity: sha512-v/ShMp57iBnBp4lDgV8Jx3d3Q5/Hac25FWmQ98eMahUiHPXcvwIMKJD0hBIgclm/FCG+LwPkAKtkRO1O/W0YGg==} + apache-arrow@19.0.1: + resolution: {integrity: sha512-APmMLzS4qbTivLrPdQXexGM4JRr+0g62QDaobzEvip/FdQIrv2qLy0mD5Qdmw4buydtVJgbFeKR8f59I6PPGDg==} + hasBin: true + + apache-arrow@20.0.0: + resolution: {integrity: sha512-JUeK0jFRUd7rbmrhhzR3O2KXjLaZ4YYYFOptyUfxOsMIoZCPi6bZR58gVi/xi3HTBMPseXm9PXyQ2V916930pA==} + hasBin: true + + apache-arrow@21.2.0: + resolution: {integrity: sha512-Hxe6Agq26gQOM954qpzYSllJBPJl+e16U5CkfuMUhLrNba+5nKkttIVlflaovN6oaTratqMGAO8H5u/aNhmHWQ==} hasBin: true argparse@1.0.10: @@ -2033,6 +2053,15 @@ packages: resolution: {integrity: sha512-H4UfQhZyakIjC74I9d34fGYDwk3XpSr17QhEd0Q3I9Xq1CETHo4Hcuo87WyWHpAF1aSLjLRf5lD9ZGX2qStUvg==} engines: {node: '>=4.0.0'} + command-line-args@6.0.2: + resolution: {integrity: sha512-AIjYVxrV9X752LmPDLbVYv8aMCuHPSLZJXEo2qo/xJfv+NYhaZ4sMSF01rM+gHPaMgvPM0l5D/F+Qx+i2WfSmQ==} + engines: {node: '>=12.20'} + peerDependencies: + '@75lb/nature': latest + peerDependenciesMeta: + '@75lb/nature': + optional: true + command-line-usage@7.0.4: resolution: {integrity: sha512-85UdvzTNx/+s5CkSgBm/0hzP80RFHAa7PsfeADE5ezZF3uHz3/Tqj9gIKGT9PTtpycc3Ua64T0oVulGfKxzfqg==} engines: {node: '>=12.20.0'} @@ -2281,6 +2310,15 @@ packages: resolution: {integrity: sha512-6Tb2myMioCAgv5kfvP5/PkZZ/ntTpVK39fHY7WkWBgvbeE+VHd/tZuZ4mrC+bxh4cfOZeYKVPaJIZtZXV7GNCQ==} engines: {node: '>=4.0.0'} + find-replace@5.0.2: + resolution: {integrity: sha512-Y45BAiE3mz2QsrN2fb5QEtO4qb44NcS7en/0y9PEVsg351HsLeVclP8QPMH79Le9sH3rs5RSwJu99W0WPZO43Q==} + engines: {node: '>=14'} + peerDependencies: + '@75lb/nature': latest + peerDependenciesMeta: + '@75lb/nature': + optional: true + find-up@4.1.0: resolution: {integrity: sha512-PpOwAdQ/YlXQ2vj8a3h8IipDuYRi3wceVQQGYWxNINccq40Anw7BlsEXCMbt1Zt+OLA6Fq9suIpIWD0OsnISlw==} engines: {node: '>=8'} @@ -2302,6 +2340,9 @@ packages: flatbuffers@24.12.23: resolution: {integrity: sha512-dLVCAISd5mhls514keQzmEG6QHmUUsNuWsb4tFafIUwvvgDjXhtfAYSKOzt5SWOy+qByV5pbsDZ+Vb7HUOBEdA==} + flatbuffers@25.9.23: + resolution: {integrity: sha512-MI1qs7Lo4Syw0EOzUl0xjs2lsoeqFku44KpngfIduHBYvzm8h2+7K8YMQh1JtVVVrUvhLpNwqVi4DERegUJhPQ==} + flatted@3.4.2: resolution: {integrity: sha512-PjDse7RzhcPkIJwy5t7KPWQSZ9cAbzQXcafsetQoD7sOJRQlGikNbx7yZp2OotDnJyrDcbyRq3Ttb18iYOqkxA==} @@ -3345,6 +3386,9 @@ packages: undici-types@6.21.0: resolution: {integrity: sha512-iwDZqg0QAGrg9Rav5H4n0M64c3mkR59cJ6wQp+7C4nI0gsmExaedaYLNO44eT4AtBBwjbTiGPMlt2Md0T9H9JQ==} + undici-types@7.24.6: + resolution: {integrity: sha512-WRNW+sJgj5OBN4/0JpHFqtqzhpbnV0GuB+OozA9gCL7a993SmU+1JBZCzLNxYsbMfIeDL+lTsphD5jN5N+n0zg==} + unist-util-is@6.0.1: resolution: {integrity: sha512-LsiILbtBETkDz8I9p1dQ0uyRUWuaQzd/cuEeS1hoRSyW5E5XGmTzlwY1OrNzzakGowI9Dr/I8HVaw4hTtnxy8g==} @@ -5505,6 +5549,10 @@ snapshots: dependencies: undici-types: 6.19.8 + '@types/node@25.9.8': + dependencies: + undici-types: 7.24.6 + '@types/stack-utils@2.0.3': {} '@types/tmp@0.2.6': {} @@ -5694,17 +5742,40 @@ snapshots: json-bignum: 0.0.3 tslib: 2.8.1 - apache-arrow@18.1.0: + apache-arrow@19.0.1: dependencies: '@swc/helpers': 0.5.21 '@types/command-line-args': 5.2.3 '@types/command-line-usage': 5.0.4 '@types/node': 20.19.41 - command-line-args: 5.2.1 + command-line-args: 6.0.2 command-line-usage: 7.0.4 flatbuffers: 24.12.23 json-bignum: 0.0.3 tslib: 2.8.1 + transitivePeerDependencies: + - '@75lb/nature' + + apache-arrow@20.0.0: + dependencies: + '@swc/helpers': 0.5.21 + '@types/command-line-args': 5.2.3 + '@types/command-line-usage': 5.0.4 + '@types/node': 20.19.41 + command-line-args: 6.0.2 + command-line-usage: 7.0.4 + flatbuffers: 25.9.23 + json-bignum: 0.0.3 + tslib: 2.8.1 + transitivePeerDependencies: + - '@75lb/nature' + + apache-arrow@21.2.0: + dependencies: + '@types/node': 25.9.8 + flatbuffers: 25.9.23 + json-with-bigint: 3.5.8 + tslib: 2.8.1 argparse@1.0.10: dependencies: @@ -5906,6 +5977,13 @@ snapshots: lodash.camelcase: 4.3.0 typical: 4.0.0 + command-line-args@6.0.2: + dependencies: + array-back: 6.2.3 + find-replace: 5.0.2 + lodash.camelcase: 4.3.0 + typical: 7.3.0 + command-line-usage@7.0.4: dependencies: array-back: 6.2.3 @@ -6178,6 +6256,8 @@ snapshots: dependencies: array-back: 3.1.0 + find-replace@5.0.2: {} + find-up@4.1.0: dependencies: locate-path: 5.0.0 @@ -6201,6 +6281,8 @@ snapshots: flatbuffers@24.12.23: {} + flatbuffers@25.9.23: {} + flatted@3.4.2: {} follow-redirects@1.16.0: {} @@ -7405,6 +7487,8 @@ snapshots: undici-types@6.21.0: {} + undici-types@7.24.6: {} + unist-util-is@6.0.1: dependencies: '@types/unist': 3.0.3 diff --git a/nodejs/src/blob.rs b/nodejs/src/blob.rs index 230a11bae..59cf61af6 100644 --- a/nodejs/src/blob.rs +++ b/nodejs/src/blob.rs @@ -1,25 +1,45 @@ // SPDX-License-Identifier: Apache-2.0 // SPDX-FileCopyrightText: Copyright The LanceDB Authors +use std::future::Future; use std::ops::Range; use std::sync::Arc; +use std::sync::atomic::{AtomicBool, Ordering}; use arrow_array::{Array, LargeBinaryArray}; +use futures::future::abortable; use lancedb::blob::BlobFile as LanceBlobFile; +use napi::Env; use napi::bindgen_prelude::*; use napi_derive::napi; use crate::error::convert_error; +#[napi(object)] +pub struct BlobRange { + pub start: BigInt, + pub end: BigInt, +} + #[napi] pub struct BlobFile { inner: Arc, + closed: AtomicBool, } impl BlobFile { pub(crate) fn new(inner: LanceBlobFile) -> Self { Self { inner: Arc::new(inner), + closed: AtomicBool::new(false), + } + } + + fn ensure_open(&self) -> napi::Result<()> { + if self.closed.load(Ordering::Acquire) { + Err(napi::Error::from_reason("blob file is already closed")) + } else { + Ok(()) } } } @@ -32,21 +52,124 @@ impl BlobFile { } #[napi] - pub async fn read(&self) -> napi::Result { - let bytes = self.inner.read().await.map_err(|err| convert_error(&err))?; - Ok(Buffer::from(bytes.as_ref())) + pub fn read<'env>( + &self, + env: &'env Env, + max_bytes: Option, + signal: Option, + ) -> napi::Result> { + self.ensure_open()?; + let max_bytes = max_bytes + .map(|max_bytes| { + usize::try_from(parse_u64(max_bytes, "maxBytes")?) + .map_err(|_| napi::Error::from_reason("maxBytes is too large")) + }) + .transpose()?; + let inner = self.inner.clone(); + spawn_abortable(env, signal, async move { + let bytes = match max_bytes { + None => inner.read().await, + Some(max_bytes) => inner.read_up_to(max_bytes).await, + } + .map_err(|err| convert_error(&err))?; + Ok(Buffer::from(bytes.as_ref())) + }) } #[napi] - pub async fn read_range(&self, start: BigInt, end: BigInt) -> napi::Result { + pub fn read_range<'env>( + &self, + env: &'env Env, + start: BigInt, + end: BigInt, + signal: Option, + ) -> napi::Result> { + self.ensure_open()?; let range = bigint_range(start, end)?; - let bytes = self - .inner - .read_range(range) - .await - .map_err(|err| convert_error(&err))?; - Ok(Buffer::from(bytes.as_ref())) + let inner = self.inner.clone(); + spawn_abortable(env, signal, async move { + let bytes = inner + .read_range(range) + .await + .map_err(|err| convert_error(&err))?; + Ok(Buffer::from(bytes.as_ref())) + }) } + + #[napi] + pub fn read_ranges<'env>( + &self, + env: &'env Env, + ranges: Vec, + signal: Option, + ) -> napi::Result>> { + self.ensure_open()?; + let ranges = ranges + .into_iter() + .map(|range| bigint_range(range.start, range.end)) + .collect::>>()?; + let inner = self.inner.clone(); + spawn_abortable(env, signal, async move { + let buffers = inner + .read_ranges(&ranges) + .await + .map_err(|err| convert_error(&err))?; + Ok(buffers + .iter() + .map(|bytes| Buffer::from(bytes.as_ref())) + .collect()) + }) + } + + #[napi] + pub async fn seek(&self, position: BigInt) -> napi::Result<()> { + self.ensure_open()?; + let position = parse_u64(position, "position")?; + self.inner + .seek(position) + .await + .map_err(|err| convert_error(&err)) + } + + #[napi] + pub async fn tell(&self) -> napi::Result { + self.ensure_open()?; + let position = self.inner.tell().await.map_err(|err| convert_error(&err))?; + Ok(BigInt::from(position)) + } + + #[napi] + pub async fn close(&self) -> napi::Result<()> { + if self.closed.swap(true, Ordering::AcqRel) { + return Ok(()); + } + self.inner.close().await.map_err(|err| convert_error(&err)) + } + + #[napi] + pub fn is_closed(&self) -> bool { + self.closed.load(Ordering::Acquire) + } +} + +/// Runs `fut` on the napi runtime. Aborting `signal` drops the future, which +/// cancels in-flight reads. +pub fn spawn_abortable<'env, T>( + env: &'env Env, + signal: Option, + fut: impl Future> + Send + 'static, +) -> napi::Result> +where + T: ToNapiValue + Send + 'static, +{ + let (fut, handle) = abortable(fut); + if let Some(signal) = signal { + signal.on_abort(move || handle.abort()); + } + env.spawn_future(async move { + fut.await + .unwrap_or_else(|_| Err(napi::Error::from_reason("blob read aborted"))) + }) } fn bigint_range(start: BigInt, end: BigInt) -> napi::Result> { diff --git a/nodejs/src/query.rs b/nodejs/src/query.rs index f9fe91d35..f542fe856 100644 --- a/nodejs/src/query.rs +++ b/nodejs/src/query.rs @@ -658,6 +658,7 @@ impl JsFullTextQuery { FtsQuery::Phrase(_) => "phrase".to_string(), FtsQuery::Boost(_) => "boost".to_string(), FtsQuery::MultiMatch(_) => "multi_match".to_string(), + FtsQuery::CombinedFields(_) => "combined_fields".to_string(), FtsQuery::Boolean(_) => "boolean".to_string(), } } diff --git a/nodejs/src/table.rs b/nodejs/src/table.rs index d381d98e5..f5c3b2987 100644 --- a/nodejs/src/table.rs +++ b/nodejs/src/table.rs @@ -11,11 +11,12 @@ use lancedb::table::{ FieldMetadataUpdate as LanceFieldMetadataUpdate, FtsToken as LanceDbFtsToken, NewColumnTransform, OptimizeAction, OptimizeOptions, Ref, Table as LanceDbTable, }; +use napi::Env; use napi::bindgen_prelude::*; use napi::threadsafe_function::{ThreadsafeFunction, ThreadsafeFunctionCallMode}; use napi_derive::napi; -use crate::blob::{BlobFile, copy_blob_buffers, parse_row_ids}; +use crate::blob::{BlobFile, copy_blob_buffers, parse_row_ids, spawn_abortable}; use crate::error::NapiErrorExt; use crate::index::Index; use crate::merge::NativeMergeInsertBuilder; @@ -340,36 +341,44 @@ impl Table { } #[napi(catch_unwind)] - pub async fn fetch_blobs( + pub fn fetch_blobs<'env>( &self, + env: &'env Env, column: String, row_ids: Vec, - ) -> napi::Result>> { + signal: Option, + ) -> napi::Result>>> { let row_ids = parse_row_ids(row_ids)?; - let array = self - .inner_ref()? - .fetch_blobs(column.as_str(), &row_ids) - .await - .default_error()?; - Ok(copy_blob_buffers(array)) + let table = self.inner_ref()?.clone(); + spawn_abortable(env, signal, async move { + let array = table + .fetch_blobs(column.as_str(), &row_ids) + .await + .default_error()?; + Ok(copy_blob_buffers(array)) + }) } #[napi(catch_unwind)] - pub async fn fetch_blob_files( + pub fn fetch_blob_files<'env>( &self, + env: &'env Env, column: String, row_ids: Vec, - ) -> napi::Result>> { + signal: Option, + ) -> napi::Result>>> { let row_ids = parse_row_ids(row_ids)?; - let files = self - .inner_ref()? - .fetch_blob_files(column.as_str(), &row_ids) - .await - .default_error()?; - Ok(files - .into_iter() - .map(|file| file.map(BlobFile::new)) - .collect()) + let table = self.inner_ref()?.clone(); + spawn_abortable(env, signal, async move { + let files = table + .fetch_blob_files(column.as_str(), &row_ids) + .await + .default_error()?; + Ok(files + .into_iter() + .map(|file| file.map(BlobFile::new)) + .collect()) + }) } #[napi(catch_unwind)] diff --git a/nodejs/tsconfig.json b/nodejs/tsconfig.json index 9b83da5ec..094fe5003 100644 --- a/nodejs/tsconfig.json +++ b/nodejs/tsconfig.json @@ -10,7 +10,10 @@ "resolveJsonModule": true, "emitDecoratorMetadata": true, "experimentalDecorators": true, - "moduleResolution": "Node" + "moduleResolution": "Node", + // flatbuffers 25 (via apache-arrow 21) ships declarations that need + // TypeScript 5.7+ generic typed arrays; skip checking third-party .d.ts. + "skipLibCheck": true }, "exclude": ["./dist/*", "./examples/*"], "typedocOptions": { diff --git a/python/Cargo.toml b/python/Cargo.toml index 17ff3f4fc..9ba96b55f 100644 --- a/python/Cargo.toml +++ b/python/Cargo.toml @@ -1,6 +1,6 @@ [package] name = "lancedb-python" -version = "0.40.0-beta.11" +version = "0.40.0-beta.12" publish = false edition.workspace = true description = "Python bindings for LanceDB" diff --git a/python/python/lancedb/_lancedb.pyi b/python/python/lancedb/_lancedb.pyi index cf63d7a9b..176e57ac3 100644 --- a/python/python/lancedb/_lancedb.pyi +++ b/python/python/lancedb/_lancedb.pyi @@ -149,12 +149,20 @@ class Connection(object): limit: Optional[int], ) -> list[str]: ... # Deprecated: Use list_tables instead async def open_job(self, job_id: str) -> Job: ... - async def create_function_async(self, request_json: str) -> Job: ... - async def get_function(self, name: str, version: str) -> str: ... - async def list_functions(self) -> List[str]: ... - async def drop_function(self, name: str, version: str) -> bool: ... + async def create_function_async( + self, request_json: str, namespace_path: Optional[List[str]] = None + ) -> Job: ... + async def get_function( + self, name: str, version: str, namespace_path: Optional[List[str]] = None + ) -> str: ... + async def list_functions( + self, namespace_path: Optional[List[str]] = None + ) -> List[str]: ... + async def drop_function( + self, name: str, version: str, namespace_path: Optional[List[str]] = None + ) -> bool: ... async def drop_function_async( - self, name: str, version: str + self, name: str, version: str, namespace_path: Optional[List[str]] = None ) -> Tuple[bool, Job]: ... async def create_secret( self, name: str, value: str, namespace_path: Optional[List[str]] = None diff --git a/python/python/lancedb/db.py b/python/python/lancedb/db.py index 490987c3c..60e1780f0 100644 --- a/python/python/lancedb/db.py +++ b/python/python/lancedb/db.py @@ -784,6 +784,7 @@ class DBConnection(EnforceOverrides): definition: UdfDefinition, *, secrets: Optional[Sequence[EnvVarSecret]] = None, + namespace_path: Optional[List[str]] = None, ) -> FunctionVersion: """Build and register a scalar Python UDF, then return its version. @@ -800,6 +801,9 @@ class DBConnection(EnforceOverrides): Function needs, each naming a Secret and the environment variable its value arrives in. The Function's source is unchanged by this; it reads the variable the way it already did. + namespace_path : list of str, optional + The namespace holding the Function. None or an empty list is the + root namespace. Examples -------- @@ -815,13 +819,16 @@ class DBConnection(EnforceOverrides): ) ``` """ - return self.create_function_async(definition, secrets=secrets).wait() + return self.create_function_async( + definition, secrets=secrets, namespace_path=namespace_path + ).wait() def create_function_async( self, definition: UdfDefinition, *, secrets: Optional[Sequence[EnvVarSecret]] = None, + namespace_path: Optional[List[str]] = None, ) -> Job[FunctionVersion]: """Submit a scalar Python UDF for building and registration. @@ -833,17 +840,26 @@ class DBConnection(EnforceOverrides): "Function catalog operations are not supported for this connection type" ) - def get_function(self, name: str, *, version: str) -> FunctionVersion: + def get_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> FunctionVersion: """Open one exact immutable Function version from the remote catalog.""" raise NotImplementedError( "Function catalog operations are not supported for this connection type" ) - def list_functions(self) -> List[FunctionVersion]: - """List every published immutable Function version. + def list_functions( + self, *, namespace_path: Optional[List[str]] = None + ) -> List[FunctionVersion]: + """List every published immutable Function version in a namespace. - Results are ordered by Function name then version. Local connections - raise ``NotImplementedError``. + Functions in child namespaces are not included. Results are ordered by + Function name then version. Local connections raise + ``NotImplementedError``. Examples -------- @@ -857,7 +873,13 @@ class DBConnection(EnforceOverrides): "Function catalog operations are not supported for this connection type" ) - def drop_function(self, name: str, *, version: str) -> bool: + def drop_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> bool: """Drop a Function name and the object it was bound to. The requested version must exist in the currently named object. Returns @@ -870,7 +892,13 @@ class DBConnection(EnforceOverrides): "Function catalog operations are not supported for this connection type" ) - def drop_function_async(self, name: str, *, version: str) -> Tuple[bool, Job]: + def drop_function_async( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> Tuple[bool, Job]: """Drop a Function name and return its cleanup job. The name is unbound before this returns; the object's content may still @@ -1808,25 +1836,62 @@ class LanceDBConnection(DBConnection): definition: UdfDefinition, *, secrets: Optional[Sequence[EnvVarSecret]] = None, + namespace_path: Optional[List[str]] = None, ) -> Job[FunctionVersion]: - job = LOOP.run(self._conn.create_function_async(definition, secrets=secrets)) + job = LOOP.run( + self._conn.create_function_async( + definition, secrets=secrets, namespace_path=namespace_path + ) + ) return Job(job) @override - def get_function(self, name: str, *, version: str) -> FunctionVersion: - return LOOP.run(self._conn.get_function(name, version=version)) + def get_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> FunctionVersion: + return LOOP.run( + self._conn.get_function( + name, version=version, namespace_path=namespace_path + ) + ) @override - def list_functions(self) -> List[FunctionVersion]: - return LOOP.run(self._conn.list_functions()) + def list_functions( + self, *, namespace_path: Optional[List[str]] = None + ) -> List[FunctionVersion]: + return LOOP.run(self._conn.list_functions(namespace_path=namespace_path)) @override - def drop_function(self, name: str, *, version: str) -> bool: - return LOOP.run(self._conn.drop_function(name, version=version)) + def drop_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> bool: + return LOOP.run( + self._conn.drop_function( + name, version=version, namespace_path=namespace_path + ) + ) @override - def drop_function_async(self, name: str, *, version: str) -> Tuple[bool, Job]: - dropped, job = LOOP.run(self._conn.drop_function_async(name, version=version)) + def drop_function_async( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> Tuple[bool, Job]: + dropped, job = LOOP.run( + self._conn.drop_function_async( + name, version=version, namespace_path=namespace_path + ) + ) return dropped, Job(job) @override @@ -2787,6 +2852,7 @@ class AsyncConnection(object): definition: UdfDefinition, *, secrets: Optional[Sequence[EnvVarSecret]] = None, + namespace_path: Optional[List[str]] = None, ) -> AsyncJob[FunctionVersion]: """Submit a scalar Python UDF for building and registration. @@ -2794,36 +2860,60 @@ class AsyncConnection(object): artifact. Waiting on the job returns the immutable Function version. ``secrets`` is a sequence of [EnvVarSecret][lancedb.secrets.EnvVarSecret], each naming a Secret and - the environment variable its value arrives in. Local connections raise - ``NotImplementedError``. + the environment variable its value arrives in. ``namespace_path`` is + the namespace the Function is registered in; None or an empty list is + the root. Local connections raise ``NotImplementedError``. """ if not isinstance(definition, UdfDefinition): raise TypeError("create_function_async requires a @udf definition") request = definition.bind_secrets(secrets) - inner = await self._inner.create_function_async(request.to_canonical_json()) + inner = await self._inner.create_function_async( + request.to_canonical_json(), namespace_path + ) return _typed_job(inner, FunctionVersion.from_json) - async def get_function(self, name: str, *, version: str) -> FunctionVersion: + async def get_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> FunctionVersion: """Open one exact immutable Function version from the remote catalog.""" - return FunctionVersion.from_json(await self._inner.get_function(name, version)) + return FunctionVersion.from_json( + await self._inner.get_function(name, version, namespace_path) + ) - async def list_functions(self) -> List[FunctionVersion]: - """List every published immutable Function version. + async def list_functions( + self, *, namespace_path: Optional[List[str]] = None + ) -> List[FunctionVersion]: + """List every published immutable Function version in a namespace. - Results are ordered by Function name then version. Local connections - raise ``NotImplementedError``. + Functions in child namespaces are not included. Results are ordered by + Function name then version. Local connections raise + ``NotImplementedError``. """ return [ FunctionVersion.from_json(value) - for value in await self._inner.list_functions() + for value in await self._inner.list_functions(namespace_path) ] - async def drop_function(self, name: str, *, version: str) -> bool: + async def drop_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> bool: """Drop a Function name and the object it was bound to.""" - return await self._inner.drop_function(name, version) + return await self._inner.drop_function(name, version, namespace_path) async def drop_function_async( - self, name: str, *, version: str + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, ) -> Tuple[bool, AsyncJob]: """Drop a Function name and return its cleanup job. @@ -2831,7 +2921,9 @@ class AsyncConnection(object): be being deleted. Await :meth:`AsyncJob.wait` to wait for that to finish. """ - dropped, job = await self._inner.drop_function_async(name, version) + dropped, job = await self._inner.drop_function_async( + name, version, namespace_path + ) return dropped, AsyncJob(job) async def create_secret( diff --git a/python/python/lancedb/embeddings/base.py b/python/python/lancedb/embeddings/base.py index 149b9c089..fbe43860d 100644 --- a/python/python/lancedb/embeddings/base.py +++ b/python/python/lancedb/embeddings/base.py @@ -182,7 +182,11 @@ class EmbeddingFunction(BaseModel, ABC): def VectorField(self, **kwargs): """ Creates a pydantic Field that can automatically annotate - the target vector column for this embedding function + the target vector column for this embedding function. + + The field can be annotated as ``list[float]`` for compatibility with + static type checkers. LanceDB will infer the fixed vector dimension from + this embedding function. """ return Field(json_schema_extra={"vector_column_for": self}, **kwargs) diff --git a/python/python/lancedb/functions.py b/python/python/lancedb/functions.py index 5f26b590e..c170c7375 100644 --- a/python/python/lancedb/functions.py +++ b/python/python/lancedb/functions.py @@ -356,9 +356,16 @@ _ObjectVersion = Annotated[ class FunctionVersion(_RemoteValue): - """A pinned object revision, independent of its executable image digest.""" + """A pinned object revision, independent of its executable image digest. + + ``name`` and ``namespace_path`` are the Function's parts, as a + [SecretReference][lancedb.functions.SecretReference] carries a Secret's; + they are what [get_function][lancedb.db.DBConnection.get_function] takes. + An empty ``namespace_path`` is the root namespace. + """ name: str + namespace_path: tuple[str, ...] = () object_id: str location: str version: _ObjectVersion @@ -444,6 +451,7 @@ class FunctionVersion(_RemoteValue): return FunctionApplication( function=FunctionVersionRef( name=self.name, + namespace_path=self.namespace_path, object_id=self.object_id, location=self.location, version=self.version, @@ -502,6 +510,7 @@ class FunctionRegistrationRequest(_RemoteValue): class FunctionVersionRef(_OpenRemoteValue): name: str + namespace_path: tuple[str, ...] = () object_id: str location: str version: _ObjectVersion diff --git a/python/python/lancedb/pydantic.py b/python/python/lancedb/pydantic.py index 528d865d7..313c514c3 100644 --- a/python/python/lancedb/pydantic.py +++ b/python/python/lancedb/pydantic.py @@ -88,6 +88,17 @@ def Vector( ... pa.field("url", pa.utf8(), False), ... pa.field("embeddings", pa.list_(pa.float32(), 768)) ... ]) + + Notes + ----- + ``Vector`` creates a type dynamically, so calls such as ``Vector(768)`` are + not valid static type annotations. For an embedding field, use the standard + ``list[float]`` annotation when running mypy; ``VectorField`` supplies the + fixed dimension to LanceDB:: + + class MyModel(LanceModel): + text: str = embeddings.SourceField() + vector: list[float] = embeddings.VectorField() """ # TODO: make a public parameterized type. @@ -313,6 +324,10 @@ def _unwrap_optional_annotation(annotation: Any) -> Any | None: def _pydantic_to_arrow_type(field: FieldInfo) -> pa.DataType: """Convert a Pydantic FieldInfo to Arrow DataType""" + embedding_vector_type = _embedding_vector_to_arrow_type(field) + if embedding_vector_type is not None: + return embedding_vector_type + unwrapped = _unwrap_optional_annotation(field.annotation) if unwrapped is not None: return _pydantic_type_to_arrow_type(unwrapped, field) @@ -326,8 +341,32 @@ def _pydantic_to_arrow_type(field: FieldInfo) -> pa.DataType: return _pydantic_type_to_arrow_type(field.annotation, field) +def _embedding_vector_to_arrow_type(field: FieldInfo) -> pa.DataType | None: + """Infer a fixed-size vector type from ``VectorField`` metadata.""" + if not _is_embedding_vector_annotation(field): + return None + + function = get_extras(field, "vector_column_for") + return pa.list_(pa.float32(), function.ndims()) + + +def _is_embedding_vector_annotation(field: FieldInfo) -> bool: + if get_extras(field, "vector_column_for") is None: + return False + + annotation = _unwrap_optional_annotation(field.annotation) + if annotation is None: + annotation = field.annotation + + origin = getattr(annotation, "__origin__", None) + args = getattr(annotation, "__args__", ()) + return origin is list and args == (float,) + + def is_nullable(field: FieldInfo) -> bool: """Check if a Pydantic FieldInfo is nullable.""" + if _is_embedding_vector_annotation(field): + return True if _unwrap_optional_annotation(field.annotation) is not None: return True if isinstance(field.annotation, (_GenericAlias, GenericAlias)): diff --git a/python/python/lancedb/remote/db.py b/python/python/lancedb/remote/db.py index 051a98095..de2390983 100644 --- a/python/python/lancedb/remote/db.py +++ b/python/python/lancedb/remote/db.py @@ -863,25 +863,62 @@ class RemoteDBConnection(DBConnection): definition: UdfDefinition, *, secrets: Optional[Sequence[EnvVarSecret]] = None, + namespace_path: Optional[List[str]] = None, ) -> Job[FunctionVersion]: - job = LOOP.run(self._conn.create_function_async(definition, secrets=secrets)) + job = LOOP.run( + self._conn.create_function_async( + definition, secrets=secrets, namespace_path=namespace_path + ) + ) return Job(job) @override - def get_function(self, name: str, *, version: str) -> FunctionVersion: - return LOOP.run(self._conn.get_function(name, version=version)) + def get_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> FunctionVersion: + return LOOP.run( + self._conn.get_function( + name, version=version, namespace_path=namespace_path + ) + ) @override - def list_functions(self) -> List[FunctionVersion]: - return LOOP.run(self._conn.list_functions()) + def list_functions( + self, *, namespace_path: Optional[List[str]] = None + ) -> List[FunctionVersion]: + return LOOP.run(self._conn.list_functions(namespace_path=namespace_path)) @override - def drop_function(self, name: str, *, version: str) -> bool: - return LOOP.run(self._conn.drop_function(name, version=version)) + def drop_function( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> bool: + return LOOP.run( + self._conn.drop_function( + name, version=version, namespace_path=namespace_path + ) + ) @override - def drop_function_async(self, name: str, *, version: str) -> Tuple[bool, Job]: - dropped, job = LOOP.run(self._conn.drop_function_async(name, version=version)) + def drop_function_async( + self, + name: str, + *, + version: str, + namespace_path: Optional[List[str]] = None, + ) -> Tuple[bool, Job]: + dropped, job = LOOP.run( + self._conn.drop_function_async( + name, version=version, namespace_path=namespace_path + ) + ) return dropped, Job(job) @override diff --git a/python/python/tests/test_blob.py b/python/python/tests/test_blob.py index e100597cf..72107364c 100644 --- a/python/python/tests/test_blob.py +++ b/python/python/tests/test_blob.py @@ -1506,7 +1506,7 @@ def test_merge_insert_mixed_python_blobs_validates_external_uri(tmp_path, as_pan {"id": 2, "image": blob_path.as_uri()}, ] - with pytest.raises(ValueError, match="allow_external_blob_outside_bases"): + with pytest.raises(ValueError, match="outside registered external bases"): ( table.merge_insert("id") .when_matched_update_all() diff --git a/python/python/tests/test_first_class_function_slice1.py b/python/python/tests/test_first_class_function_slice1.py index 1abd42e56..b08dadf36 100644 --- a/python/python/tests/test_first_class_function_slice1.py +++ b/python/python/tests/test_first_class_function_slice1.py @@ -162,6 +162,29 @@ def test_function_version_binds_named_columns_as_one_immutable_application(): ] == [("text", "column", "documents.body")] +def test_function_version_names_its_namespace_as_a_separate_field(): + """A Function is named by its parts, as a Secret is; a root one states no + namespace.""" + value = job_result("remote_function_job.json") + root = FunctionVersion.from_json(json.dumps(value)) + assert root.namespace_path == () + assert "namespace_path" not in json.loads(root.to_canonical_json()) + + nested = FunctionVersion(**{**value, "namespace_path": ["analytics", "features"]}) + assert nested.name == "embed" + assert nested.namespace_path == ("analytics", "features") + assert json.loads(nested.to_canonical_json())["namespace_path"] == [ + "analytics", + "features", + ] + application = nested(text=col("documents.body")) + assert application.function.name == "embed" + assert application.function.namespace_path == ("analytics", "features") + assert json.loads(application.to_canonical_json())["function"][ + "namespace_path" + ] == ["analytics", "features"] + + def test_function_version_binding_validates_names_and_direct_columns(): version = FunctionVersion.from_json( json.dumps(job_result("remote_function_job.json")) diff --git a/python/python/tests/test_first_class_function_slice2.py b/python/python/tests/test_first_class_function_slice2.py index 1cd8d4cf5..4610170f6 100644 --- a/python/python/tests/test_first_class_function_slice2.py +++ b/python/python/tests/test_first_class_function_slice2.py @@ -1539,8 +1539,10 @@ def _mock_remote_function_catalog(): else (None, None) ) if function_action[1] == "create": + *namespace_path, name = function_action[0].split("$") state["version"] = { - "name": "normalize_score", + "name": name, + **({"namespace_path": namespace_path} if namespace_path else {}), "version": FUNCTION_VERSION, "object_id": "fixture", "location": "memory:///fixture", @@ -1565,10 +1567,18 @@ def _mock_remote_function_catalog(): "job_state": "DONE", "result": state["version"], } - elif self.path == "/v1/function/normalize_score/describe": + # A namespaced Function's id is its namespace path and name joined + # by `$`, so the name is the last segment of it. + elif ( + function_action[1] == "describe" + and function_action[0].split("$")[-1] == "normalize_score" + ): assert body == {"version": FUNCTION_VERSION} response = state["version"] - elif self.path == "/v1/function/normalize_score/drop": + elif ( + function_action[1] == "drop" + and function_action[0].split("$")[-1] == "normalize_score" + ): assert body == {"version": FUNCTION_VERSION} response = {"dropped": True} elif secret_action[1] in ("create", "alter"): @@ -1602,7 +1612,8 @@ def _mock_remote_function_catalog(): assert query["page_token"] == "next" self._write_response(200, {"secrets": [{"name": "hf-prod"}]}) return - if url.path != "/v1/namespace/$/function/list": + parts = url.path.strip("/").split("/") + if parts[:2] != ["v1", "namespace"] or parts[3:] != ["function", "list"]: self._write_response(404, {"error": "not found"}) return assert query["include_definition"] == "true" @@ -1840,3 +1851,102 @@ async def test_async_remote_drop_function_sends_exact_version(): {"version": FUNCTION_VERSION}, ) ] + + +def test_remote_function_verbs_address_a_namespace_in_the_path(): + """A Function's namespace is part of the identifier each route addresses; + no body carries it, so a namespaced request differs only in its path.""" + namespace = ["analytics", "features"] + with _mock_remote_function_catalog() as (host, state): + db = lancedb.connect( + "db://dev", + api_key="fake", + host_override=host, + client_config={"retry_config": {"retries": 0}}, + ) + created = db.create_function(normalize_score, namespace_path=namespace) + # The returned parts are what the other verbs take, so a created + # Function reopens from its own fields. + assert created.name == "normalize_score" + assert created.namespace_path == tuple(namespace) + reopened = db.get_function( + created.name, + version=created.version, + namespace_path=list(created.namespace_path), + ) + assert reopened == created + assert reopened(value=lancedb.col("score")).function.namespace_path == tuple( + namespace + ) + db.list_functions(namespace_path=namespace) + assert ( + db.drop_function( + "normalize_score", version=FUNCTION_VERSION, namespace_path=namespace + ) + is True + ) + dropped, _ = db.drop_function_async( + "normalize_score", version=FUNCTION_VERSION, namespace_path=namespace + ) + assert dropped is True + + assert [path for path, _ in state["requests"]] == [ + "/v1/function/analytics$features$normalize_score/create", + "/v1/jobs/describe", + "/v1/function/analytics$features$normalize_score/describe", + "/v1/namespace/analytics$features/function/list", + "/v1/namespace/analytics$features/function/list", + "/v1/function/analytics$features$normalize_score/drop", + "/v1/function/analytics$features$normalize_score/drop", + ] + + +@pytest.mark.asyncio +async def test_async_remote_function_verbs_address_a_namespace_in_the_path(): + namespace = ["analytics", "features"] + with _mock_remote_function_catalog() as (host, state): + db = await lancedb.connect_async( + "db://dev", + api_key="fake", + host_override=host, + client_config={"retry_config": {"retries": 0}}, + ) + registration = await db.create_function_async( + normalize_score, namespace_path=namespace + ) + await registration.wait() + await db.get_function( + "normalize_score", version=FUNCTION_VERSION, namespace_path=namespace + ) + await db.list_functions(namespace_path=namespace) + assert ( + await db.drop_function( + "normalize_score", version=FUNCTION_VERSION, namespace_path=namespace + ) + is True + ) + + assert [path for path, _ in state["requests"]] == [ + "/v1/function/analytics$features$normalize_score/create", + "/v1/jobs/describe", + "/v1/function/analytics$features$normalize_score/describe", + "/v1/namespace/analytics$features/function/list", + "/v1/namespace/analytics$features/function/list", + "/v1/function/analytics$features$normalize_score/drop", + ] + + +def test_a_function_namespace_path_must_be_a_list_of_segments(): + """A bare string is refused rather than read as a sequence of one-character + segments, and nothing reaches the service.""" + with _mock_remote_function_catalog() as (host, state): + db = lancedb.connect( + "db://dev", + api_key="fake", + host_override=host, + client_config={"retry_config": {"retries": 0}}, + ) + with pytest.raises(TypeError): + db.list_functions(namespace_path="analytics") + + assert state["requests"] == [] diff --git a/python/python/tests/test_lsm_write_spec.py b/python/python/tests/test_lsm_write_spec.py index d43cb7532..ac767ea31 100644 --- a/python/python/tests/test_lsm_write_spec.py +++ b/python/python/tests/test_lsm_write_spec.py @@ -52,9 +52,9 @@ def test_set_lsm_write_spec_validates(tmp_path): with pytest.raises(Exception, match="num_buckets"): table.set_lsm_write_spec(LsmWriteSpec.bucket("id", 1025)) - # Happy path then mutation rejected. + # Happy path, then a set over the installed spec is refused. table.set_lsm_write_spec(LsmWriteSpec.bucket("id", 4)) - with pytest.raises(Exception, match="mutation"): + with pytest.raises(Exception, match="already set"): table.set_lsm_write_spec(LsmWriteSpec.bucket("id", 8)) @@ -176,18 +176,18 @@ def test_get_lsm_write_spec(tmp_path): table.unset_lsm_write_spec() assert table.get_lsm_write_spec() is None - # Identity round-trips (column recovered from the schema). Leaving the - # maintained set to be inferred picks up the index on the table, so the - # spec reads back naming it rather than as "infer". + # Identity round-trips (column recovered from the schema). An unnamed + # maintained set reads back as None: what is stored is the intent to + # maintain every index, not the set it resolves to now. table.set_lsm_write_spec(LsmWriteSpec.identity("id")) spec = table.get_lsm_write_spec() assert spec.spec_type == "identity" assert spec.column == "id" - assert spec.maintained_indexes == [idx_name] + assert spec.maintained_indexes is None table.unset_lsm_write_spec() # Unsharded round-trips (no routing column). Opting out is distinct from - # the inferred default. + # the unnamed default. table.set_lsm_write_spec(LsmWriteSpec.unsharded().with_maintained_indexes([])) spec = table.get_lsm_write_spec() assert spec.spec_type == "unsharded" diff --git a/python/python/tests/test_pydantic.py b/python/python/tests/test_pydantic.py index ca03940f2..81674d735 100644 --- a/python/python/tests/test_pydantic.py +++ b/python/python/tests/test_pydantic.py @@ -9,6 +9,7 @@ from typing import List, Optional, Tuple import pyarrow as pa import pydantic import pytest +from lancedb.conftest import MockTextEmbeddingFunction from lancedb.pydantic import ( LanceModel, MultiVector, @@ -425,6 +426,25 @@ def test_bare_vector_raises_clear_error(): exec("class TestModel(LanceModel):\n vector: Vector", namespace) +def test_embedding_vector_list_annotation(): + embedding = MockTextEmbeddingFunction.create() + + class StaticTypingModel(LanceModel): + text: str = embedding.SourceField() + vector: list[float] = embedding.VectorField() + + schema = pydantic_to_schema(StaticTypingModel) + assert schema == pa.schema( + [ + pa.field("text", pa.utf8(), False), + pa.field("vector", pa.list_(pa.float32(), embedding.ndims()), True), + ] + ) + + model = StaticTypingModel(text="hello", vector=[0.0] * embedding.ndims()) + assert model.vector == [0.0] * embedding.ndims() + + def test_fixed_size_list_field(): class TestModel(pydantic.BaseModel): vec: Vector(16) diff --git a/python/python/tests/test_remote_db.py b/python/python/tests/test_remote_db.py index 494fe73f1..def2daab6 100644 --- a/python/python/tests/test_remote_db.py +++ b/python/python/tests/test_remote_db.py @@ -2396,6 +2396,27 @@ def blob_query_response_table(): ) +def blob_descriptor_take_table(): + """What `fetch_blob_files` reads for rows 10, 20, 30, and 40.""" + queried = blob_query_response_table() + image_field = queried.schema.field("image") + images = pa.StructArray.from_arrays( + [ + pa.array([0, 0, 0, 0], type=pa.uint8()), + pa.array([0, 0, 0, 0], type=pa.uint64()), + pa.array([5, 0, 5, 0], type=pa.uint64()), + pa.array([0, 0, 0, 0], type=pa.uint32()), + pa.array(["", "", "", ""], type=pa.string()), + ], + fields=image_field.type, + mask=pa.array([False, True, False, False]), + ) + return pa.Table.from_arrays( + [images, pa.array([10, 20, 30, 40], type=pa.uint64())], + schema=pa.schema([image_field, pa.field("_rowid", pa.uint64())]), + ) + + @contextlib.contextmanager def blob_remote_table(*, server_version=Version("0.5.0")): def handler(request): @@ -2434,9 +2455,14 @@ def blob_remote_table(*, server_version=Version("0.5.0")): elif request.path == "/v1/table/test/query/": content_len = int(request.headers.get("Content-Length", 0)) body = json.loads(request.rfile.read(content_len)) - assert body["columns"] == ["id", "image"] assert body["with_row_id"] is True - response_table = blob_query_response_table() + if body["columns"] == ["image"]: + # fetch_blob_files sizes its handles from a descriptor take. + assert body["filter"].startswith("_rowid IN") + response_table = blob_descriptor_take_table() + else: + assert body["columns"] == ["id", "image"] + response_table = blob_query_response_table() request.send_response(200) request.send_header("Content-Type", "application/vnd.apache.arrow.file") request.end_headers() diff --git a/python/src/connection.rs b/python/src/connection.rs index e3754cc3d..480d55a91 100644 --- a/python/src/connection.rs +++ b/python/src/connection.rs @@ -754,31 +754,37 @@ impl Connection { }) } + #[pyo3(signature = (request_json, namespace_path=None))] pub fn create_function_async( self_: PyRef<'_, Self>, request_json: String, + namespace_path: Option>, ) -> PyResult> { let inner = self_.get_inner()?.clone(); let request = lancedb::function::FunctionRegistrationRequest::from_json(&request_json) .infer_error()?; + let namespace_path = namespace_path.unwrap_or_default(); future_into_py(self_.py(), async move { inner - .create_function_async(request) + .create_function_async(request, &namespace_path) .await .infer_error() .map(crate::job::Job::new_typed) }) } + #[pyo3(signature = (name, version, namespace_path=None))] pub fn get_function( self_: PyRef<'_, Self>, name: String, version: String, + namespace_path: Option>, ) -> PyResult> { let inner = self_.get_inner()?.clone(); + let namespace_path = namespace_path.unwrap_or_default(); future_into_py(self_.py(), async move { inner - .get_function(name, version) + .get_function(name, version, &namespace_path) .await .infer_error()? .to_canonical_json() @@ -786,11 +792,16 @@ impl Connection { }) } - pub fn list_functions(self_: PyRef<'_, Self>) -> PyResult> { + #[pyo3(signature = (namespace_path=None))] + pub fn list_functions( + self_: PyRef<'_, Self>, + namespace_path: Option>, + ) -> PyResult> { let inner = self_.get_inner()?.clone(); + let namespace_path = namespace_path.unwrap_or_default(); future_into_py(self_.py(), async move { inner - .list_functions() + .list_functions(&namespace_path) .await .infer_error()? .into_iter() @@ -799,26 +810,35 @@ impl Connection { }) } + #[pyo3(signature = (name, version, namespace_path=None))] pub fn drop_function( self_: PyRef<'_, Self>, name: String, version: String, + namespace_path: Option>, ) -> PyResult> { let inner = self_.get_inner()?.clone(); + let namespace_path = namespace_path.unwrap_or_default(); future_into_py(self_.py(), async move { - inner.drop_function(name, version).await.infer_error() + inner + .drop_function(name, version, &namespace_path) + .await + .infer_error() }) } + #[pyo3(signature = (name, version, namespace_path=None))] pub fn drop_function_async( self_: PyRef<'_, Self>, name: String, version: String, + namespace_path: Option>, ) -> PyResult> { let inner = self_.get_inner()?.clone(); + let namespace_path = namespace_path.unwrap_or_default(); future_into_py(self_.py(), async move { inner - .drop_function_async(name, version) + .drop_function_async(name, version, &namespace_path) .await .infer_error() .map(|(dropped, job)| (dropped, crate::job::Job::new(job))) diff --git a/python/src/query.rs b/python/src/query.rs index 2398376d5..3a4737ee0 100644 --- a/python/src/query.rs +++ b/python/src/query.rs @@ -180,6 +180,9 @@ impl<'py> IntoPyObject<'py> for PyLanceDB { .expect("Failed to import namespace"); match self.0 { + FtsQuery::CombinedFields(_) => Err(PyValueError::new_err( + "Combined-fields FTS queries are not yet supported in Python", + )), FtsQuery::Match(query) => { let kwargs = PyDict::new(py); kwargs.set_item("boost", query.boost)?; diff --git a/rust/lancedb/Cargo.toml b/rust/lancedb/Cargo.toml index fbc9ab5ee..cdb0609d3 100644 --- a/rust/lancedb/Cargo.toml +++ b/rust/lancedb/Cargo.toml @@ -1,6 +1,6 @@ [package] name = "lancedb" -version = "0.40.0-beta.11" +version = "0.40.0-beta.12" edition.workspace = true description = "LanceDB: A serverless, low-latency vector database for AI applications" license.workspace = true diff --git a/rust/lancedb/src/connection.rs b/rust/lancedb/src/connection.rs index 0e5bbf5da..985195c07 100644 --- a/rust/lancedb/src/connection.rs +++ b/rust/lancedb/src/connection.rs @@ -587,7 +587,8 @@ impl Connection { ) } - /// Build and register a Python callable as an immutable Function version. + /// Build and register a Python callable as an immutable Function version + /// in `namespace_path`, which is empty for the root namespace. /// /// The server-side job builds the OCI image, then registers the completed /// artifact. Waiting on the returned typed job yields the durable @@ -597,8 +598,11 @@ impl Connection { pub async fn create_function_async( &self, request: crate::function::FunctionRegistrationRequest, + namespace_path: &[String], ) -> Result> { - self.internal.create_function_async(request).await + self.internal + .create_function_async(request, namespace_path) + .await } /// Look up one exact immutable Function version in the remote catalog. @@ -610,13 +614,15 @@ impl Connection { &self, name: impl AsRef, version: impl AsRef, + namespace_path: &[String], ) -> Result { self.internal - .get_function(name.as_ref(), version.as_ref()) + .get_function(name.as_ref(), version.as_ref(), namespace_path) .await } - /// List every published immutable Function version in the remote catalog. + /// List every published immutable Function version in `namespace_path` of + /// the remote catalog. Functions in child namespaces are not included. /// /// Results are ordered by Function name then version. The client walks all /// server pages before returning. Local databases return @@ -628,14 +634,17 @@ impl Connection { /// # async fn list_functions( /// # connection: &lancedb::Connection, /// # ) -> Result<(), Box> { - /// for function in connection.list_functions().await? { + /// for function in connection.list_functions(&[]).await? { /// println!("{} {}", function.name(), function.version()); /// } /// # Ok(()) /// # } /// ``` - pub async fn list_functions(&self) -> Result> { - self.internal.list_functions().await + pub async fn list_functions( + &self, + namespace_path: &[String], + ) -> Result> { + self.internal.list_functions(namespace_path).await } /// Remove the current Function name binding, retaining the object history. @@ -647,9 +656,10 @@ impl Connection { &self, name: impl AsRef, version: impl AsRef, + namespace_path: &[String], ) -> Result { self.internal - .drop_function(name.as_ref(), version.as_ref()) + .drop_function(name.as_ref(), version.as_ref(), namespace_path) .await } @@ -663,9 +673,10 @@ impl Connection { &self, name: impl AsRef, version: impl AsRef, + namespace_path: &[String], ) -> Result<(bool, crate::job::Job)> { self.internal - .drop_function_async(name.as_ref(), version.as_ref()) + .drop_function_async(name.as_ref(), version.as_ref(), namespace_path) .await } diff --git a/rust/lancedb/src/database.rs b/rust/lancedb/src/database.rs index 224fd8770..2518edf13 100644 --- a/rust/lancedb/src/database.rs +++ b/rust/lancedb/src/database.rs @@ -333,10 +333,12 @@ pub trait Database: /// /// See [`CloneTableRequest`] for detailed documentation and examples. async fn clone_table(&self, request: CloneTableRequest) -> Result>; - /// Submit a Function creation job that builds an image and registers it. + /// Submit a Function creation job that builds an image and registers it + /// in `namespace_path`. async fn create_function_async( &self, _request: crate::function::FunctionRegistrationRequest, + _namespace_path: &[String], ) -> Result> { function_catalog_not_supported() } @@ -412,15 +414,25 @@ pub trait Database: &self, _name: &str, _version: &str, + _namespace_path: &[String], ) -> Result { function_catalog_not_supported() } - /// List every published immutable Function version in the remote catalog. - async fn list_functions(&self) -> Result> { + /// List every published immutable Function version in one namespace of the + /// remote catalog. + async fn list_functions( + &self, + _namespace_path: &[String], + ) -> Result> { function_catalog_not_supported() } /// Remove the current Function name binding, retaining the object history. - async fn drop_function(&self, _name: &str, _version: &str) -> Result { + async fn drop_function( + &self, + _name: &str, + _version: &str, + _namespace_path: &[String], + ) -> Result { function_catalog_not_supported() } /// Start dropping a Function and return a handle to the cleanup job. @@ -431,8 +443,9 @@ pub trait Database: &self, name: &str, version: &str, + namespace_path: &[String], ) -> Result<(bool, crate::job::Job)> { - let dropped = self.drop_function(name, version).await?; + let dropped = self.drop_function(name, version, namespace_path).await?; Ok((dropped, crate::job::Job::new_done())) } /// Create a named Secret in this database. Fails if the name is taken, so diff --git a/rust/lancedb/src/function.rs b/rust/lancedb/src/function.rs index 3b7209bd5..f05633281 100644 --- a/rust/lancedb/src/function.rs +++ b/rust/lancedb/src/function.rs @@ -85,22 +85,25 @@ fn has_unknown_keys(value: &Value, allowed: &[&str]) -> bool { .is_some_and(|object| object.keys().any(|key| !allowed.contains(&key.as_str()))) } +/// Every field a [`FunctionVersionRef`] document may carry. An application or +/// binding with any other key comes from a newer contract. +pub(crate) const FUNCTION_VERSION_REF_FIELDS: &[&str] = &[ + "name", + "namespace_path", + "object_id", + "location", + "version", + "manifest_digest", +]; + fn application_has_unknown_nested_fields(value: &Value) -> bool { let Some(application) = value.as_object() else { return false; }; - if application.get("function").is_some_and(|value| { - has_unknown_keys( - value, - &[ - "name", - "object_id", - "location", - "version", - "manifest_digest", - ], - ) - }) { + if application + .get("function") + .is_some_and(|value| has_unknown_keys(value, FUNCTION_VERSION_REF_FIELDS)) + { return true; } if application @@ -431,9 +434,17 @@ fn deserialize_object_version<'de, D: Deserializer<'de>>( } /// One immutable Function object revision and its executable artifact. +/// +/// The Function is named by its parts, `name` and `namespace_path`, the way a +/// [`SecretReference`](crate::secrets::SecretReference) names a Secret. #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)] pub struct FunctionVersion { name: String, + /// The namespace holding the Function. Empty is the root, and is absent + /// from the wire, as a [`SecretReference`](crate::secrets::SecretReference) + /// omits it. + #[serde(default, skip_serializing_if = "Vec::is_empty")] + namespace_path: Vec, object_id: String, location: String, #[serde(deserialize_with = "deserialize_object_version")] @@ -463,6 +474,7 @@ impl FunctionVersion { pub fn reference(&self) -> FunctionVersionRef { FunctionVersionRef { name: self.name.clone(), + namespace_path: self.namespace_path.clone(), object_id: self.object_id.clone(), location: self.location.clone(), version: self.version.clone(), @@ -472,6 +484,9 @@ impl FunctionVersion { pub fn name(&self) -> &str { &self.name } + pub fn namespace_path(&self) -> &[String] { + &self.namespace_path + } pub fn version(&self) -> &str { &self.version } @@ -556,6 +571,10 @@ impl_json!(FunctionRegistrationRequest); #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)] pub struct FunctionVersionRef { pub name: String, + /// The namespace holding the Function; empty is the root and is absent + /// from the wire. + #[serde(default, skip_serializing_if = "Vec::is_empty")] + pub namespace_path: Vec, pub object_id: String, pub location: String, #[serde(deserialize_with = "deserialize_object_version")] @@ -954,6 +973,46 @@ mod tests { ); } + /// A Function is named by its parts, and a root one states no namespace. + #[test] + fn function_version_carries_its_namespace_as_a_separate_field() { + const VERSION: &str = include_str!( + "../tests/fixtures/first_class_functions/v1/remote_function_version.canonical.json" + ); + let root = FunctionVersion::from_json(VERSION).unwrap(); + assert!(root.namespace_path().is_empty()); + assert_eq!(root.to_canonical_json().unwrap(), VERSION.trim_end()); + + let mut document: Value = serde_json::from_str(VERSION).unwrap(); + document["namespace_path"] = serde_json::json!(["analytics", "features"]); + let nested = FunctionVersion::from_json(&document.to_string()).unwrap(); + assert_eq!(nested.name(), "embed"); + assert_eq!(nested.namespace_path(), ["analytics", "features"]); + let reference = nested.reference(); + assert_eq!(reference.name, "embed"); + assert_eq!(reference.namespace_path, ["analytics", "features"]); + let reference: Value = serde_json::to_value(&reference).unwrap(); + assert_eq!( + reference["namespace_path"], + serde_json::json!(["analytics", "features"]) + ); + let root_reference = serde_json::to_value(root.reference()).unwrap(); + assert!(root_reference.get("namespace_path").is_none()); + + // An application carrying the namespace is not from a newer contract. + let application = serde_json::json!({ + "function": reference, + "inputs": [], + "output": serde_json::to_value(nested.signature().output.clone()).unwrap(), + }); + let application = FunctionApplication::from_json(&application.to_string()).unwrap(); + assert!(!application.has_unknown_fields()); + assert_eq!( + application.function().namespace_path, + ["analytics", "features"] + ); + } + /// Arrays are ordered by the caller, so canonicalization must leave them /// alone -- sorting them would change what a signature means. #[test] diff --git a/rust/lancedb/src/remote/db.rs b/rust/lancedb/src/remote/db.rs index 39bb5a528..27606cdfe 100644 --- a/rust/lancedb/src/remote/db.rs +++ b/rust/lancedb/src/remote/db.rs @@ -979,8 +979,9 @@ impl Database for RemoteDatabase { async fn create_function_async( &self, request: FunctionRegistrationRequest, + namespace_path: &[String], ) -> Result> { - let function_id = build_object_identifier("Function name", &request.name, &[])?; + let function_id = build_object_identifier("Function name", &request.name, namespace_path)?; let req = self .client .post(&format!("/v1/function/{function_id}/create")) @@ -1005,8 +1006,13 @@ impl Database for RemoteDatabase { )))) } - async fn get_function(&self, name: &str, version: &str) -> Result { - let function_id = build_object_identifier("Function name", name, &[])?; + async fn get_function( + &self, + name: &str, + version: &str, + namespace_path: &[String], + ) -> Result { + let function_id = build_object_identifier("Function name", name, namespace_path)?; let req = self .client .post(&format!("/v1/function/{function_id}/describe")) @@ -1018,8 +1024,8 @@ impl Database for RemoteDatabase { response.json().await.err_to_http(request_id) } - async fn list_functions(&self) -> Result> { - let namespace_id = build_namespace_identifier(&[])?; + async fn list_functions(&self, namespace_path: &[String]) -> Result> { + let namespace_id = build_namespace_identifier(namespace_path)?; let path = format!("/v1/namespace/{namespace_id}/function/list"); let mut functions = Vec::new(); let mut page_token: Option = None; @@ -1059,12 +1065,25 @@ impl Database for RemoteDatabase { Ok(functions) } - async fn drop_function(&self, name: &str, version: &str) -> Result { - Ok(self.drop_function_async(name, version).await?.0) + async fn drop_function( + &self, + name: &str, + version: &str, + namespace_path: &[String], + ) -> Result { + Ok(self + .drop_function_async(name, version, namespace_path) + .await? + .0) } - async fn drop_function_async(&self, name: &str, version: &str) -> Result<(bool, Job)> { - let function_id = build_object_identifier("Function name", name, &[])?; + async fn drop_function_async( + &self, + name: &str, + version: &str, + namespace_path: &[String], + ) -> Result<(bool, Job)> { + let function_id = build_object_identifier("Function name", name, namespace_path)?; let req = self .client .post(&format!("/v1/function/{function_id}/drop")) @@ -2004,7 +2023,7 @@ mod tests { .body(serde_json::json!({"dropped": true, "job_id": "j1-fn-drop"}).to_string()) .unwrap() }); - let (dropped, job) = db.drop_function_async("embed", "1").await.unwrap(); + let (dropped, job) = db.drop_function_async("embed", "1", &[]).await.unwrap(); assert!(dropped); assert_eq!(job.id(), Some("j1-fn-drop")); } @@ -2019,7 +2038,7 @@ mod tests { .body(serde_json::json!({"dropped": false}).to_string()) .unwrap() }); - let (dropped, job) = db.drop_function_async("embed", "1").await.unwrap(); + let (dropped, job) = db.drop_function_async("embed", "1", &[]).await.unwrap(); assert!(!dropped); assert_eq!(job.id(), None); assert_eq!(job.status().await.unwrap(), "finished"); @@ -2036,7 +2055,11 @@ mod tests { let db = super::RemoteDatabase::new_mock(move |_| { http::Response::builder().status(202).body(body).unwrap() }); - let error = db.drop_function_async("embed", "1").await.err().unwrap(); + let error = db + .drop_function_async("embed", "1", &[]) + .await + .err() + .unwrap(); assert!(error.to_string().contains("valid job_id")); } } @@ -3905,7 +3928,7 @@ mod tests { http::Response::builder().status(200).body("{}").unwrap() }); let error = conn - .drop_function(name, "fv_1") + .drop_function(name, "fv_1", &[]) .await .expect_err("a dot-only Function name must be refused"); assert!( @@ -4307,7 +4330,7 @@ mod tests { path => panic!("unexpected path: {path}"), }); let request = crate::function::FunctionRegistrationRequest::from_json(REQUEST).unwrap(); - let job = conn.create_function_async(request).await.unwrap(); + let job = conn.create_function_async(request, &[]).await.unwrap(); assert_eq!(job.id(), Some("job-function-1")); let version = job.wait().await.unwrap(); assert_eq!(version.name(), "embed"); @@ -4327,7 +4350,7 @@ mod tests { assert_eq!(body, serde_json::json!({"version": "1"})); http::Response::builder().status(200).body(VERSION).unwrap() }); - let version = conn.get_function("embed", "1").await.unwrap(); + let version = conn.get_function("embed", "1", &[]).await.unwrap(); assert_eq!(version.name(), "embed"); assert_eq!(version.version(), "1"); } @@ -4370,7 +4393,7 @@ mod tests { } } }); - let functions = conn.list_functions().await.unwrap(); + let functions = conn.list_functions(&[]).await.unwrap(); assert_eq!(functions.len(), 1); assert_eq!(functions[0].name(), "embed"); assert_eq!(functions[0].version(), "1"); @@ -4393,7 +4416,7 @@ mod tests { .unwrap() }); - let functions = conn.list_functions().await.unwrap(); + let functions = conn.list_functions(&[]).await.unwrap(); assert!(functions.is_empty()); assert_eq!(requests.load(Ordering::SeqCst), 1); } @@ -4434,7 +4457,7 @@ mod tests { .unwrap() }); - let error = conn.list_functions().await.unwrap_err(); + let error = conn.list_functions(&[]).await.unwrap_err(); assert!( matches!( &error, @@ -4461,7 +4484,82 @@ mod tests { .body(r#"{"dropped":false}"#) .unwrap() }); - assert!(!conn.drop_function("embed", "1").await.unwrap()); + assert!(!conn.drop_function("embed", "1", &[]).await.unwrap()); + } + + /// A Function's namespace is addressed in the path the way a Secret's is: + /// every route takes the joined identifier and no body carries the + /// namespace, so a namespaced request differs from a root one only in its + /// path. + #[tokio::test] + async fn test_a_function_namespace_path_is_addressed_in_the_path() { + const REQUEST: &str = include_str!( + "../../tests/fixtures/first_class_functions/v1/remote_function_registration_request.json" + ); + const VERSION: &str = include_str!( + "../../tests/fixtures/first_class_functions/v1/remote_function_version.canonical.json" + ); + let namespace = ["analytics".to_string(), "features".to_string()]; + let mut expected_create: serde_json::Value = serde_json::from_str(REQUEST).unwrap(); + expected_create.as_object_mut().unwrap().remove("name"); + let conn = Connection::new_with_handler(move |request| { + let body = request + .body() + .and_then(|body| body.as_bytes()) + .map(|bytes| serde_json::from_slice::(bytes).unwrap()); + match request.url().path() { + "/v1/function/analytics$features$normalize_score/create" => { + assert_eq!(body.unwrap(), expected_create); + http::Response::builder() + .status(202) + .body(r#"{"job_id":"job-function-1"}"#.to_string()) + .unwrap() + } + "/v1/function/analytics$features$embed/describe" => { + assert_eq!(body.unwrap(), serde_json::json!({"version": "1"})); + http::Response::builder() + .status(200) + .body(VERSION.to_string()) + .unwrap() + } + "/v1/function/analytics$features$embed/drop" => { + assert_eq!(body.unwrap(), serde_json::json!({"version": "1"})); + http::Response::builder() + .status(200) + .body(r#"{"dropped":true}"#.to_string()) + .unwrap() + } + // Listing is namespace-scoped, so the namespace is the whole + // identifier. + "/v1/namespace/analytics$features/function/list" => http::Response::builder() + .status(200) + .body(r#"{"functions":[]}"#.to_string()) + .unwrap(), + path => panic!("unexpected path: {path}"), + } + }); + let request = crate::function::FunctionRegistrationRequest::from_json(REQUEST).unwrap(); + let job = conn + .create_function_async(request, &namespace) + .await + .unwrap(); + assert_eq!(job.id(), Some("job-function-1")); + conn.get_function("embed", "1", &namespace).await.unwrap(); + assert!(conn.list_functions(&namespace).await.unwrap().is_empty()); + assert!(conn.drop_function("embed", "1", &namespace).await.unwrap()); + } + + /// Each namespace segment is checked before a request is built, so an + /// empty segment cannot collapse the identifier onto the parent namespace. + #[tokio::test] + async fn test_an_unaddressable_function_namespace_segment_is_refused() { + let conn = Connection::new_with_handler(|request| -> http::Response { + panic!("reached the transport: {}", request.url().path()) + }); + let namespace = ["analytics".to_string(), String::new()]; + assert!(conn.get_function("embed", "1", &namespace).await.is_err()); + assert!(conn.list_functions(&namespace).await.is_err()); + assert!(conn.drop_function("embed", "1", &namespace).await.is_err()); } #[tokio::test] diff --git a/rust/lancedb/src/remote/table.rs b/rust/lancedb/src/remote/table.rs index 65219cad5..f19958c22 100644 --- a/rust/lancedb/src/remote/table.rs +++ b/rust/lancedb/src/remote/table.rs @@ -96,6 +96,8 @@ const SCHEMA_SELECTOR_CHANGED: &str = "table selector changed while fetching sch fn fts_query_requires_document_granularity_support(query: &FtsQuery) -> bool { match query { + // Combined-fields queries do not expose a document granularity option. + FtsQuery::CombinedFields(_) => false, FtsQuery::Match(query) => query .document_granularity .is_some_and(|granularity| granularity.is_list_element()), @@ -6339,6 +6341,53 @@ mod tests { .unwrap(); } + #[tokio::test] + async fn test_query_combined_fields_uses_structured_fts() { + use lance_index::scalar::inverted::query::CombinedFieldsQuery; + + let table = + Table::new_with_handler_version("my_table", semver::Version::new(0, 3, 0), |request| { + let body = request.body().unwrap().as_bytes().unwrap(); + let body: serde_json::Value = serde_json::from_slice(body).unwrap(); + assert_eq!( + body["full_text_query"]["query"], + serde_json::json!({ + "combined_fields": { + "query": "hello world", + "columns": ["title", "text"], + "boost": [1.0, 1.0], + "operator": "Or" + } + }) + ); + + let data = RecordBatch::try_new( + Arc::new(Schema::new(vec![Field::new("a", DataType::Int32, false)])), + vec![Arc::new(Int32Array::from(vec![1]))], + ) + .unwrap(); + http::Response::builder() + .status(200) + .header(CONTENT_TYPE, ARROW_FILE_CONTENT_TYPE) + .body(write_ipc_file(&data)) + .unwrap() + }); + + table + .query() + .full_text_search(FullTextSearchQuery::new_query( + CombinedFieldsQuery::try_new( + "hello world".into(), + vec!["title".into(), "text".into()], + ) + .unwrap() + .into(), + )) + .execute() + .await + .unwrap(); + } + #[tokio::test] async fn test_query_row_document_granularity_uses_structured_fts() { let table = diff --git a/rust/lancedb/src/remote/table/blobs.rs b/rust/lancedb/src/remote/table/blobs.rs index f222ae31c..7ae74157f 100644 --- a/rust/lancedb/src/remote/table/blobs.rs +++ b/rust/lancedb/src/remote/table/blobs.rs @@ -3,23 +3,28 @@ //! Cloud blob column listing, whole-byte fetch, and seekable HTTP byte-range handles. +use std::collections::HashMap; use std::ops::Range; use std::sync::Arc; use std::sync::atomic::{AtomicBool, Ordering}; use std::time::Duration; -use arrow_array::{Array, LargeBinaryArray}; +use arrow_array::{Array, LargeBinaryArray, StructArray, UInt8Array, UInt64Array}; use arrow_schema::DataType; use bytes::Bytes; +use datafusion_expr::{col, lit}; use futures::{StreamExt, TryStreamExt}; +use lance::dataset::ROW_ID; +use lance_core::datatypes::BlobKind; use reqwest::{Response, StatusCode, header}; use tokio::sync::{Mutex, OwnedSemaphorePermit}; use crate::Error; use crate::blob::BlobFile; use crate::error::Result; +use crate::query::{QueryFilter, QueryRequest, Select}; use crate::remote::client::{HttpSend, RequestResultExt, RestfulLanceDbClient}; -use crate::table::BaseTable; +use crate::table::{AnyQuery, BaseTable}; use super::{ FreshnessHeaders, FreshnessState, ReadSnapshot, RemoteTable, VERSION_HEADER, @@ -142,7 +147,7 @@ struct RemoteBlobState { /// Seekable Cloud blob handle over HTTP Range. /// -/// Nonempty handles read the table version returned by their size probe. +/// Handles read the table version used to fetch their descriptor or probe their size. #[derive(Debug)] pub struct RemoteBlobFile { requester: Arc, @@ -503,7 +508,11 @@ impl RemoteTable { /// Open seekable handles for `row_ids`. /// - /// Each non-null row is probed once to determine its size. + /// Descriptors for the requested rows come from one `_rowid` take query per + /// [`MAX_FETCH_BLOBS_ROW_IDS`] unique row ids, all read at one exact table + /// version. Each handle is sized from its descriptor and reads that + /// version. Only external blobs whose descriptor records no size are + /// probed, because their length is the whole object's. pub(super) async fn fetch_blob_files_impl( &self, column: &str, @@ -513,46 +522,251 @@ impl RemoteTable { if row_ids.is_empty() { return Ok(Vec::new()); } - if !self.server_version.support_blobs() { - return Err(Error::NotSupported { - message: "fetch_blob_files requires LanceDB Cloud server 0.5.0 or newer".into(), - }); - } + self.ensure_blob_files_supported()?; - let read_snapshot = self.snapshot_read_state().await; - self.fetch_blob_files_with_snapshot(column, row_ids, read_snapshot) - .await + let mut read_snapshot = self.snapshot_read_state().await; + let sizes = self + .take_blob_descriptor_sizes(column, row_ids, &mut read_snapshot) + .await?; + let version = read_snapshot.version; + + let mut files = Vec::with_capacity(row_ids.len()); + let mut probe_indices = Vec::new(); + let mut probe_requesters = Vec::new(); + for (index, row_id) in row_ids.iter().enumerate() { + match sizes[row_id] { + DescriptorSize::Null => files.push(None), + DescriptorSize::Known(size) => { + let requester = self.blob_range_requester(column, *row_id, &read_snapshot); + files.push(Some(RemoteBlobFile::new(requester, size, version).into())); + } + DescriptorSize::Unresolved => { + files.push(None); + probe_indices.push(index); + probe_requesters.push(self.blob_range_requester( + column, + *row_id, + &read_snapshot, + )); + } + } + } + let probed = probe_blob_files(probe_requesters).await?; + for (index, file) in probe_indices.into_iter().zip(probed) { + files[index] = file; + } + Ok(files) } + /// Probe each row once for its size. + /// + /// Serves callers that must not issue a query, such as a single blob that + /// the whole-byte route could not return. async fn fetch_blob_files_with_snapshot( &self, column: &str, row_ids: &[u64], read_snapshot: ReadSnapshot, ) -> Result>> { - let encoded_column = urlencoding::encode(column); let requesters = row_ids .iter() - .map(|row_id| { - let path = format!( - "/v1/table/{}/blob/{encoded_column}/{row_id}/bytes", - self.identifier - ); - let requester: Arc = Arc::new(TableBlobRangeRequester { - client: self.client.clone(), - path, - version: read_snapshot.version, - branch: self.branch.clone(), - freshness: Arc::new(std::sync::Mutex::new(read_snapshot.freshness_state)), - parent_freshness: self.freshness.clone(), - parent_freshness_request: read_snapshot.freshness, - read_consistency_interval: self.client.read_consistency_interval, - }); - requester - }) + .map(|row_id| self.blob_range_requester(column, *row_id, &read_snapshot)) .collect(); probe_blob_files(requesters).await } + + /// Read the blob descriptor of every unique row in `row_ids`. + /// + /// Pins `read_snapshot` to the version the first take read, so later chunks + /// and the returned handles all see one table version. + async fn take_blob_descriptor_sizes( + &self, + column: &str, + row_ids: &[u64], + read_snapshot: &mut ReadSnapshot, + ) -> Result> { + let mut unique_row_ids = row_ids.to_vec(); + unique_row_ids.sort_unstable(); + unique_row_ids.dedup(); + + let mut sizes = HashMap::with_capacity(unique_row_ids.len()); + for chunk in unique_row_ids.chunks(MAX_FETCH_BLOBS_ROW_IDS) { + let (mut chunk_sizes, response_version) = self + .take_blob_descriptor_chunk(column, chunk, *read_snapshot) + .await?; + if read_snapshot.version.is_none() { + match response_version { + Some(version) => read_snapshot.version = Some(version), + None => { + // A server that does not report its read version may have read + // an older version than a describe now returns, so the take is + // repeated at the described version. + read_snapshot.version = + Some(self.describe_read_snapshot(*read_snapshot).await?.version); + (chunk_sizes, _) = self + .take_blob_descriptor_chunk(column, chunk, *read_snapshot) + .await?; + } + } + } + sizes.extend(chunk_sizes); + } + let found = unique_row_ids + .iter() + .filter(|row_id| sizes.contains_key(row_id)) + .count(); + if found < unique_row_ids.len() { + return Err(Error::InvalidInput { + message: format!( + "blob read for column '{column}' requested {} row ids but only {found} exist \ + in the table; pass row ids collected from this table", + unique_row_ids.len() + ), + }); + } + Ok(sizes) + } + + /// Take one chunk of descriptors, also returning the version the server + /// reported reading. + async fn take_blob_descriptor_chunk( + &self, + column: &str, + row_ids: &[u64], + read_snapshot: ReadSnapshot, + ) -> Result<(HashMap, Option)> { + let query = AnyQuery::Query(QueryRequest { + filter: Some(QueryFilter::Datafusion( + col(ROW_ID).in_list(row_ids.iter().map(|id| lit(*id)).collect(), false), + )), + select: Select::columns(&[column]), + with_row_id: true, + // Row ids name base-table rows, and the MemWAL scanner has no stable `_rowid`. + use_lsm: Some(false), + ..Default::default() + }); + let body = self + .prepare_query_bodies(&query, read_snapshot.version)? + .pop() + .expect("a plain query has one body"); + let request = self + .client + .post(&format!("/v1/table/{}/query/", self.identifier)) + .json(&body); + let (request_id, response) = self + .send_with_freshness(request, true, read_snapshot.freshness) + .await?; + let response_version = response + .headers() + .get(&VERSION_HEADER) + .and_then(|value| value.to_str().ok()) + .and_then(|value| value.parse().ok()); + let mut stream = self.read_arrow_response(&request_id, response).await?; + + let mut sizes = HashMap::with_capacity(row_ids.len()); + let invalid_response = |message: String| Error::Http { + source: message.into(), + request_id: request_id.clone(), + status_code: None, + }; + while let Some(batch) = stream.try_next().await? { + let taken_row_ids = batch + .column_by_name(ROW_ID) + .and_then(|array| array.as_any().downcast_ref::()) + .ok_or_else(|| { + invalid_response(format!("blob descriptor take is missing UInt64 '{ROW_ID}'")) + })?; + let values = batch.column_by_name(column).ok_or_else(|| { + invalid_response(format!("blob descriptor take is missing '{column}'")) + })?; + // Only blob v2 columns read back as descriptors. + let descriptors = values + .as_any() + .downcast_ref::() + .and_then(BlobDescriptorSizes::try_new) + .ok_or_else(|| Error::InvalidInput { + message: format!("column '{column}' is not a blob column"), + })?; + for (index, row_id) in taken_row_ids.values().iter().enumerate() { + sizes.insert(*row_id, descriptors.size(index)?); + } + } + Ok((sizes, response_version)) + } + + fn ensure_blob_files_supported(&self) -> Result<()> { + if self.server_version.support_blobs() { + Ok(()) + } else { + Err(Error::NotSupported { + message: "fetch_blob_files requires LanceDB Cloud server 0.5.0 or newer".into(), + }) + } + } + + fn blob_range_requester( + &self, + column: &str, + row_id: u64, + read_snapshot: &ReadSnapshot, + ) -> Arc { + let path = format!( + "/v1/table/{}/blob/{}/{row_id}/bytes", + self.identifier, + urlencoding::encode(column) + ); + Arc::new(TableBlobRangeRequester { + client: self.client.clone(), + path, + version: read_snapshot.version, + branch: self.branch.clone(), + freshness: Arc::new(std::sync::Mutex::new(read_snapshot.freshness_state)), + parent_freshness: self.freshness.clone(), + parent_freshness_request: read_snapshot.freshness, + read_consistency_interval: self.client.read_consistency_interval, + }) + } +} + +/// What a blob v2 descriptor says about a row's size. +#[derive(Debug, Clone, Copy)] +enum DescriptorSize { + Null, + Known(u64), + /// An external blob that recorded no size. Its length is the whole + /// object's, which only the server can resolve. + Unresolved, +} + +/// The `kind` and `size` children of a blob v2 descriptor array. +struct BlobDescriptorSizes<'a> { + descriptors: &'a StructArray, + kinds: &'a UInt8Array, + sizes: &'a UInt64Array, +} + +impl<'a> BlobDescriptorSizes<'a> { + fn try_new(descriptors: &'a StructArray) -> Option { + let child = |name: &str| descriptors.column_by_name(name); + Some(Self { + descriptors, + kinds: child("kind")?.as_any().downcast_ref::()?, + sizes: child("size")?.as_any().downcast_ref::()?, + }) + } + + fn size(&self, index: usize) -> Result { + if self.descriptors.is_null(index) || self.kinds.is_null(index) { + return Ok(DescriptorSize::Null); + } + let kind = BlobKind::try_from(self.kinds.value(index))?; + let size = self.sizes.value(index); + Ok(if kind == BlobKind::External && size == 0 { + DescriptorSize::Unresolved + } else { + DescriptorSize::Known(size) + }) + } } /// Probe blob sizes while preserving row order. @@ -676,6 +890,8 @@ mod tests { use reqwest::Request; use semver::Version; + use arrow_array::RecordBatch; + use super::*; const PAYLOAD: &[u8] = b"0123456789abcdefghijklmnopqrstuvwxyz"; @@ -759,10 +975,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); - let mut files = table - .fetch_blob_files_impl("image", &[10, 20]) - .await - .unwrap(); + let mut files = table.probe_files("image", &[10, 20]).await.unwrap(); assert_eq!(files.len(), 2); assert!(files[1].is_none()); @@ -774,12 +987,368 @@ mod tests { ); } + /// Descriptors as a remote query returns them, one `(kind, size)` per row. + fn descriptors(rows: &[Option<(BlobKind, u64)>]) -> StructArray { + let kinds = rows + .iter() + .map(|row| row.map_or(0, |(kind, _)| kind as u8)) + .collect::(); + let sizes = rows + .iter() + .map(|row| row.map_or(0, |(_, size)| size)) + .collect::(); + let len = rows.len(); + StructArray::new( + lance_core::datatypes::BLOB_V2_DESC_FIELDS.clone(), + vec![ + Arc::new(kinds), + Arc::new(UInt64Array::from(vec![0; len])), + Arc::new(sizes), + Arc::new(arrow_array::UInt32Array::from(vec![0; len])), + Arc::new(arrow_array::StringArray::from(vec![""; len])), + ], + Some(rows.iter().map(Option::is_some).collect()), + ) + } + + impl RemoteTable { + /// Open handles through the per-row size probe alone. + async fn probe_files( + &self, + column: &str, + row_ids: &[u64], + ) -> Result>> { + let read_snapshot = self.snapshot_read_state().await; + self.fetch_blob_files_with_snapshot(column, row_ids, read_snapshot) + .await + } + } + + /// A request the take mock received. + #[derive(Debug, Clone, PartialEq)] + enum TakeMockRequest { + Query { + row_ids: Vec, + version: Option, + }, + Describe, + Range { + row_id: u64, + range: String, + version: Option, + }, + } + + /// Row ids named by a `_rowid IN (...)` filter. + fn filter_row_ids(filter: &str) -> Vec { + let list = filter.split_once('(').unwrap().1.trim_end_matches(')'); + list.split(',') + .map(|id| id.trim().parse().unwrap()) + .collect() + } + + /// A table whose query route answers `_rowid` takes from `rows` and whose + /// byte routes serve [`PAYLOAD`]. Takes return rows in reverse order, and + /// report `query_version` in `x-lancedb-version` when it is set. + fn mock_take_blob_table( + rows: Vec<(u64, Option<(BlobKind, u64)>)>, + query_version: Option<&'static str>, + ) -> ( + RemoteTable, + Arc>>, + ) { + let requests = Arc::new(StdMutex::new(Vec::new())); + let captured = requests.clone(); + let table = RemoteTable::new_mock( + "my_table".to_string(), + move |request| { + let path = request.url().path().to_string(); + if path == "/v1/table/my_table/query/" { + let body: serde_json::Value = + serde_json::from_slice(request.body().unwrap().as_bytes().unwrap()) + .unwrap(); + assert_eq!(body["columns"], serde_json::json!(["image"])); + assert_eq!(body["with_row_id"], serde_json::json!(true)); + assert_eq!(body["use_lsm"], serde_json::json!(false)); + let requested = filter_row_ids(body["filter"].as_str().unwrap()); + captured.lock().unwrap().push(TakeMockRequest::Query { + row_ids: requested.clone(), + version: body["version"].as_u64(), + }); + let taken = rows + .iter() + .rev() + .filter(|(row_id, _)| requested.contains(row_id)) + .collect::>(); + let batch = RecordBatch::try_from_iter([ + ( + "image", + Arc::new(descriptors( + &taken.iter().map(|(_, row)| *row).collect::>(), + )) as Arc, + ), + ( + ROW_ID, + Arc::new(UInt64Array::from_iter_values( + taken.iter().map(|(row_id, _)| *row_id), + )) as Arc, + ), + ]) + .unwrap(); + let mut body = Vec::new(); + let mut writer = + arrow_ipc::writer::FileWriter::try_new(&mut body, &batch.schema()).unwrap(); + writer.write(&batch).unwrap(); + writer.finish().unwrap(); + drop(writer); + let mut response = http::Response::builder().status(200); + if let Some(version) = query_version { + response = response.header(VERSION_HEADER, version); + } + return response.body(body).unwrap(); + } + if path == "/v1/table/my_table/describe/" { + captured.lock().unwrap().push(TakeMockRequest::Describe); + return http::Response::builder() + .status(200) + .body(r#"{"version":9,"schema":{"fields":[]}}"#.as_bytes().to_vec()) + .unwrap(); + } + let row_id = path + .strip_prefix("/v1/table/my_table/blob/image/") + .and_then(|rest| rest.strip_suffix("/bytes")) + .unwrap_or_else(|| panic!("unexpected path: {path}")) + .parse() + .unwrap(); + captured.lock().unwrap().push(TakeMockRequest::Range { + row_id, + range: request.headers()[header::RANGE] + .to_str() + .unwrap() + .to_string(), + version: request + .url() + .query_pairs() + .find(|(name, _)| name == "version") + .map(|(_, value)| value.into_owned()), + }); + range_response(&request, PAYLOAD) + }, + Some(Version::new(0, 5, 0)), + ); + (table, requests) + } + + #[tokio::test] + async fn remote_blob_files_are_sized_from_one_descriptor_take() { + let payload_size = PAYLOAD.len() as u64; + let (table, requests) = mock_take_blob_table( + vec![ + (10, Some((BlobKind::Inline, payload_size))), + (20, None), + (30, Some((BlobKind::Inline, 0))), + (40, Some((BlobKind::Packed, payload_size))), + (50, Some((BlobKind::Dedicated, payload_size))), + (60, Some((BlobKind::External, 0))), + ], + Some("7"), + ); + + // Reordered and duplicated row ids come back in request order. + let files = table + .fetch_blob_files_impl("image", &[60, 10, 20, 30, 40, 50, 10]) + .await + .unwrap(); + + let sizes = files + .iter() + .map(|file| file.as_ref().map(BlobFile::size)) + .collect::>(); + assert_eq!( + sizes, + [ + Some(payload_size), + Some(payload_size), + None, + Some(0), + Some(payload_size), + Some(payload_size), + Some(payload_size), + ] + ); + // One take, then a probe for the external blob that recorded no size. + assert_eq!( + requests.lock().unwrap().as_slice(), + [ + TakeMockRequest::Query { + row_ids: vec![10, 20, 30, 40, 50, 60], + version: None, + }, + TakeMockRequest::Range { + row_id: 60, + range: "bytes=0-0".into(), + version: Some("7".into()), + }, + ] + ); + + // Sized handles read the version the take read. + assert_eq!( + files[1].as_ref().unwrap().read_range(5..12).await.unwrap(), + &PAYLOAD[5..12] + ); + assert!(files[3].as_ref().unwrap().read().await.unwrap().is_empty()); + assert_eq!( + requests.lock().unwrap().last().unwrap(), + &TakeMockRequest::Range { + row_id: 10, + range: "bytes=5-11".into(), + version: Some("7".into()), + } + ); + } + + #[tokio::test] + async fn remote_blob_files_take_every_chunk_at_the_first_version() { + let row_ids = (0..MAX_FETCH_BLOBS_ROW_IDS as u64 + 1).collect::>(); + let (table, requests) = mock_take_blob_table( + row_ids.iter().map(|row_id| (*row_id, None)).collect(), + Some("7"), + ); + + let files = table + .fetch_blob_files_impl("image", &row_ids) + .await + .unwrap(); + + assert_eq!(files.len(), row_ids.len()); + let versions = requests + .lock() + .unwrap() + .iter() + .map(|request| match request { + TakeMockRequest::Query { row_ids, version } => (row_ids.len(), *version), + other => panic!("unexpected request: {other:?}"), + }) + .collect::>(); + assert_eq!(versions, [(MAX_FETCH_BLOBS_ROW_IDS, None), (1, Some(7))]); + } + + #[tokio::test] + async fn remote_blob_files_retake_at_the_described_version_when_a_take_does_not_report_it() { + let payload_size = PAYLOAD.len() as u64; + let (table, requests) = + mock_take_blob_table(vec![(10, Some((BlobKind::Inline, payload_size)))], None); + + let file = table + .fetch_blob_files_impl("image", &[10]) + .await + .unwrap() + .pop() + .flatten() + .unwrap(); + file.read_range(0..4).await.unwrap(); + + // The unversioned take is repeated at the described version. + assert_eq!( + requests.lock().unwrap().as_slice(), + [ + TakeMockRequest::Query { + row_ids: vec![10], + version: None, + }, + TakeMockRequest::Describe, + TakeMockRequest::Query { + row_ids: vec![10], + version: Some(9), + }, + TakeMockRequest::Range { + row_id: 10, + range: "bytes=0-3".into(), + version: Some("9".into()), + }, + ] + ); + } + + #[tokio::test] + async fn remote_blob_files_take_at_the_checked_out_version() { + let payload_size = PAYLOAD.len() as u64; + let (table, requests) = mock_take_blob_table( + vec![(10, Some((BlobKind::Inline, payload_size)))], + Some("7"), + ); + table.checkout(9).await.unwrap(); + requests.lock().unwrap().clear(); + + table.fetch_blob_files_impl("image", &[10]).await.unwrap(); + + assert_eq!( + requests.lock().unwrap().as_slice(), + [TakeMockRequest::Query { + row_ids: vec![10], + version: Some(9), + }] + ); + } + + #[tokio::test] + async fn remote_blob_files_reject_row_ids_missing_from_the_take() { + let (table, requests) = mock_take_blob_table(vec![(10, None)], Some("7")); + + let error = table + .fetch_blob_files_impl("image", &[10, 99]) + .await + .unwrap_err(); + + assert!(matches!(error, Error::InvalidInput { .. }), "{error}"); + assert!(error.to_string().contains("only 1 exist"), "{error}"); + assert_eq!(requests.lock().unwrap().len(), 1); + } + + #[tokio::test] + async fn remote_blob_files_reject_a_non_blob_column() { + let table = RemoteTable::new_mock( + "my_table".to_string(), + |request| { + assert_eq!(request.url().path(), "/v1/table/my_table/query/"); + let batch = RecordBatch::try_from_iter([ + ( + "id", + Arc::new(arrow_array::Int64Array::from(vec![1])) as Arc, + ), + ( + ROW_ID, + Arc::new(UInt64Array::from(vec![10])) as Arc, + ), + ]) + .unwrap(); + let mut body = Vec::new(); + let mut writer = + arrow_ipc::writer::FileWriter::try_new(&mut body, &batch.schema()).unwrap(); + writer.write(&batch).unwrap(); + writer.finish().unwrap(); + drop(writer); + http::Response::builder().status(200).body(body).unwrap() + }, + Some(Version::new(0, 5, 0)), + ); + + let error = table.fetch_blob_files_impl("id", &[10]).await.unwrap_err(); + + assert!(matches!(error, Error::InvalidInput { .. }), "{error}"); + assert!( + error.to_string().contains("'id' is not a blob column"), + "{error}" + ); + } + #[tokio::test] async fn remote_blob_file_reads_the_requested_range() { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -817,7 +1386,7 @@ mod tests { }, ); let files = table - .fetch_blob_files_impl("image", &(0..16).collect::>()) + .probe_files("image", &(0..16).collect::>()) .await .unwrap(); let start = Arc::new(tokio::sync::Barrier::new(files.len() + 1)); @@ -852,10 +1421,7 @@ mod tests { ..Default::default() }, ); - let files = table - .fetch_blob_files_impl("image", &[10, 20]) - .await - .unwrap(); + let files = table.probe_files("image", &[10, 20]).await.unwrap(); let reads = files .iter() .map(|file| file.as_ref().unwrap().read_up_to(1)); @@ -891,7 +1457,7 @@ mod tests { ..Default::default() }, ); - let file = table.fetch_blob_files_impl("image", &[10]).await.unwrap(); + let file = table.probe_files("image", &[10]).await.unwrap(); let error = file[0] .as_ref() .unwrap() @@ -939,7 +1505,7 @@ mod tests { ); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -991,7 +1557,7 @@ mod tests { table.checkout(5).await.unwrap(); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -1011,7 +1577,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -1034,10 +1600,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); - let mut files = table - .fetch_blob_files_impl("image", &[10, 20, 30]) - .await - .unwrap(); + let mut files = table.probe_files("image", &[10, 20, 30]).await.unwrap(); assert_eq!(files.len(), 3); assert!(files[1].is_none()); @@ -1065,10 +1628,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); - let error = table - .fetch_blob_files_impl("image", &[10]) - .await - .unwrap_err(); + let error = table.probe_files("image", &[10]).await.unwrap_err(); assert!( error .to_string() @@ -1092,10 +1652,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); - let error = table - .fetch_blob_files_impl("image", &[10]) - .await - .unwrap_err(); + let error = table.probe_files("image", &[10]).await.unwrap_err(); assert!( error .to_string() @@ -1116,10 +1673,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); - let error = table - .fetch_blob_files_impl("image", &[10]) - .await - .unwrap_err(); + let error = table.probe_files("image", &[10]).await.unwrap_err(); assert!( error .to_string() @@ -1177,7 +1731,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -1217,7 +1771,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -1267,7 +1821,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -1292,7 +1846,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -1371,7 +1925,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); - let mut files = table.fetch_blob_files_impl("image", &[10]).await.unwrap(); + let mut files = table.probe_files("image", &[10]).await.unwrap(); let file = files.remove(0).unwrap(); assert_eq!(file.size(), PAYLOAD.len() as u64); @@ -1386,7 +1940,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); - let mut files = table.fetch_blob_files_impl("image", &[10]).await.unwrap(); + let mut files = table.probe_files("image", &[10]).await.unwrap(); let file = files.remove(0).unwrap(); let probe_requests = requests.lock().unwrap().len(); @@ -1411,7 +1965,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); - let mut files = table.fetch_blob_files_impl("image", &[10]).await.unwrap(); + let mut files = table.probe_files("image", &[10]).await.unwrap(); let file = files.remove(0).unwrap(); let probe_requests = requests.lock().unwrap().len(); @@ -1463,7 +2017,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); - let mut files = table.fetch_blob_files_impl("image", &[10]).await.unwrap(); + let mut files = table.probe_files("image", &[10]).await.unwrap(); let file = files.remove(0).unwrap(); let error = file.read_range(1..3).await.unwrap_err(); @@ -1554,7 +2108,7 @@ mod tests { let requests = Arc::new(StdMutex::new(Vec::new())); let table = mock_remote_blob_table(requests.clone()); let file = table - .fetch_blob_files_impl("image", &[10]) + .probe_files("image", &[10]) .await .unwrap() .pop() @@ -1600,7 +2154,7 @@ mod tests { Some(Version::new(0, 5, 0)), ); - let mut files = table.fetch_blob_files_impl("image", &[10]).await.unwrap(); + let mut files = table.probe_files("image", &[10]).await.unwrap(); let file = files.remove(0).unwrap(); let error = file.read_range(1..3).await.unwrap_err(); assert!(error.to_string().contains("206"), "got: {error}"); diff --git a/rust/lancedb/src/table.rs b/rust/lancedb/src/table.rs index 8ce8014b3..3e601f689 100644 --- a/rust/lancedb/src/table.rs +++ b/rust/lancedb/src/table.rs @@ -330,7 +330,8 @@ pub use self::merge::MergeResult; /// date) and [`LsmWriteSpec::with_writer_config_defaults`] (default /// `ShardWriter` configuration recorded in the MemWAL index). /// -/// A fresh spec maintains every index on the table, resolved on install. +/// A fresh spec maintains every index on the table, including ones created +/// after it is installed. /// /// Install a spec with [`Table::set_lsm_write_spec`] and remove it with /// [`Table::unset_lsm_write_spec`]. The actual `merge_insert` dispatch @@ -348,9 +349,8 @@ pub enum LsmWriteSpec { num_buckets: u32, /// Indexes the MemWAL maintains in-memory as rows are appended. /// - /// `None` means every index it can maintain, resolved on install — a - /// snapshot, so indexes created later need the spec unset and re-set. - /// `Some([])` maintains nothing. + /// `None` means every index the table has, including ones created + /// later. `Some([])` maintains nothing. maintained_indexes: Option>, /// Default `ShardWriter` configuration recorded in the MemWAL index. writer_config_defaults: HashMap, @@ -363,9 +363,8 @@ pub enum LsmWriteSpec { column: String, /// Indexes the MemWAL maintains in-memory as rows are appended. /// - /// `None` means every index it can maintain, resolved on install — a - /// snapshot, so indexes created later need the spec unset and re-set. - /// `Some([])` maintains nothing. + /// `None` means every index the table has, including ones created + /// later. `Some([])` maintains nothing. maintained_indexes: Option>, /// Default `ShardWriter` configuration recorded in the MemWAL index. writer_config_defaults: HashMap, @@ -374,9 +373,8 @@ pub enum LsmWriteSpec { Unsharded { /// Indexes the MemWAL maintains in-memory as rows are appended. /// - /// `None` means every index it can maintain, resolved on install — a - /// snapshot, so indexes created later need the spec unset and re-set. - /// `Some([])` maintains nothing. + /// `None` means every index the table has, including ones created + /// later. `Some([])` maintains nothing. maintained_indexes: Option>, /// Default `ShardWriter` configuration recorded in the MemWAL index. writer_config_defaults: HashMap, @@ -422,14 +420,15 @@ impl LsmWriteSpec { /// Set which indexes the MemWAL maintains. /// - /// `None` (the default) resolves to every index on the table at install, - /// failing if one cannot be maintained — name the set to install anyway. A - /// list is verbatim: each name must already exist and be maintainable, and - /// an empty list maintains nothing. + /// `None` (the default) is every index on the table, re-read as the table + /// changes, so an index created later is maintained too; one of a kind the + /// MemWAL cannot mirror is skipped rather than failing the table. A list is + /// verbatim: each name must already exist and be maintainable, and an empty + /// list maintains nothing. /// /// ``` /// # use lancedb::table::LsmWriteSpec; - /// // Every index the table has when the spec is installed: + /// // Every index the table has, now and later: /// LsmWriteSpec::unsharded().with_maintained_indexes(None); /// // Exactly these: /// LsmWriteSpec::unsharded().with_maintained_indexes(vec!["id_idx".to_string()]); @@ -1991,6 +1990,10 @@ impl Table { /// - [`LsmWriteSpec::identity`] — shard by the raw value of a scalar column. /// - [`LsmWriteSpec::unsharded`] — route every write to a single shard. /// + /// A table carries one spec: this fails while one is installed, since the + /// generations already written were homed under it. + /// [`Table::unset_lsm_write_spec`] removes one. + /// /// # Example /// /// ``` @@ -3794,7 +3797,20 @@ impl BaseTable for NativeTable { let Some(segment) = segments.first() else { continue; }; - let params = load_segment_params(&dataset, segment).await?; + // The listing itself only needs the manifest. Missing index files must + // not hide every other index, or callers cannot find the one to repair. + let params = match load_segment_params(&dataset, segment).await { + Ok(params) => params, + Err(err) => { + log::warn!( + "Failed to read full text search configuration for index '{}': {}", + index.name, + err + ); + index.index_details = None; + continue; + } + }; let details = serde_json::to_string(¶ms).map_err(|source| Error::Other { message: format!( "Failed to serialize full text search configuration for index '{}'", @@ -5654,29 +5670,22 @@ mod tests { assert_eq!(table.get_lsm_write_spec().await.unwrap(), None); // Identity sharding round-trips (column recovered from the schema). - // A spec left at its default maintains every index on the table, so it - // reads back naming the one on the table rather than as "infer". + // A spec left at its default round-trips as the default: what is stored + // is the intent to maintain everything, not the set it resolves to now. let spec = LsmWriteSpec::identity("region"); table.set_lsm_write_spec(spec.clone()).await.unwrap(); - assert_eq!( - table.get_lsm_write_spec().await.unwrap(), - Some(spec.with_maintained_indexes(vec![idx_name.clone()])) - ); + assert_eq!(table.get_lsm_write_spec().await.unwrap(), Some(spec)); table.unset_lsm_write_spec().await.unwrap(); // Unsharded round-trips (no routing column). let spec = LsmWriteSpec::unsharded(); table.set_lsm_write_spec(spec.clone()).await.unwrap(); - assert_eq!( - table.get_lsm_write_spec().await.unwrap(), - Some(spec.with_maintained_indexes(vec![idx_name])) - ); + assert_eq!(table.get_lsm_write_spec().await.unwrap(), Some(spec)); } - /// The maintained set defaults to every index on the table, resolved at - /// install. An index the memtable cannot build fails the install rather - /// than being dropped: maintaining it would take the table offline for - /// writes, dropping it would hide that from the caller. + /// The maintained set defaults to every index on the table. A named index + /// the memtable cannot build fails the install, because the caller asked + /// for it; an unnamed one is skipped, because it arrived by existing. #[tokio::test] async fn test_set_lsm_write_spec_infers_maintained_indexes() { let tmp_dir = tempdir().unwrap(); @@ -5729,19 +5738,27 @@ mod tests { ); assert_eq!(table.get_lsm_write_spec().await.unwrap(), None); - // The default covers every index, so the bitmap fails it too. - let err = table + // The default names nothing, so the bitmap is skipped rather than + // refusing the table: it is maintained by existing, not by being asked + // for. The intent is what is stored, not the set it resolves to today. + table .set_lsm_write_spec(LsmWriteSpec::unsharded()) .await - .unwrap_err(); - assert!( - matches!(err, Error::InvalidInput { ref message } - if message.contains("tag_bitmap") && message.contains("maintained_indexes")), - "expected the inferred set to be rejected, got {err:?}" + .unwrap(); + assert_eq!( + table + .get_lsm_write_spec() + .await + .unwrap() + .unwrap() + .maintained_indexes(), + None, + "an unnamed set is stored as the intent to maintain everything" ); - assert_eq!(table.get_lsm_write_spec().await.unwrap(), None); - // Naming the maintainable subset installs. + // Narrowing to a named subset: unset first, since an installed spec + // cannot be set over. + table.unset_lsm_write_spec().await.unwrap(); table .set_lsm_write_spec( LsmWriteSpec::unsharded().with_maintained_indexes(vec!["id_btree".to_string()]), @@ -5773,6 +5790,28 @@ mod tests { .maintained_indexes(), Some([].as_slice()) ); + + // An installed spec is never set over, whatever the new one asks for, + // and the refused call leaves it alone. + let err = table + .set_lsm_write_spec(LsmWriteSpec::bucket("id", 4)) + .await + .unwrap_err(); + assert!( + matches!(err, Error::InvalidInput { ref message } + if message.contains("already set")), + "expected the repeat to be refused, got {err:?}" + ); + assert_eq!( + table + .get_lsm_write_spec() + .await + .unwrap() + .unwrap() + .maintained_indexes(), + Some([].as_slice()), + "a refused call changes nothing" + ); } #[tokio::test] diff --git a/rust/lancedb/src/table/computed_columns.rs b/rust/lancedb/src/table/computed_columns.rs index 98b4e1627..d92599294 100644 --- a/rust/lancedb/src/table/computed_columns.rs +++ b/rust/lancedb/src/table/computed_columns.rs @@ -704,13 +704,7 @@ fn ensure_known_binding_shape(value: &Value) -> Result<()> { object .get("function") .ok_or_else(|| invalid_function("Function binding is missing its exact version"))?, - &[ - "name", - "object_id", - "location", - "version", - "manifest_digest", - ], + crate::function::FUNCTION_VERSION_REF_FIELDS, "version reference", )?; for input in object @@ -3760,6 +3754,20 @@ mod tests { ensure_known_binding_shape(&raw_binding).unwrap(); } + /// A binding names its Function by parts, so one declared against a + /// namespaced Function is part of the known contract too. + #[test] + fn test_namespaced_bindings_are_a_known_shape() { + let mut raw_binding: Value = serde_json::from_str(include_str!( + "../../tests/fixtures/first_class_functions/v1/remote_initialized_function_binding.json" + )) + .unwrap(); + raw_binding["function"]["namespace_path"] = serde_json::json!(["analytics", "features"]); + ensure_known_binding_shape(&raw_binding).unwrap(); + let binding = FunctionBinding::from_json(&raw_binding.to_string()).unwrap(); + assert_eq!(binding.function().namespace_path, ["analytics", "features"]); + } + #[test] fn test_named_struct_can_be_kept_as_one_nullable_physical_column() { let application = named_struct_application("{}"); diff --git a/rust/lancedb/src/table/create_index.rs b/rust/lancedb/src/table/create_index.rs index 0a0d00da1..916b28ed1 100644 --- a/rust/lancedb/src/table/create_index.rs +++ b/rust/lancedb/src/table/create_index.rs @@ -160,6 +160,12 @@ impl NativeTable { builder = builder.name(name); } builder.await?; + // A writer already open was built before this index existed, so it has + // to pick it up here or a read needing it is refused until it reopens. + self.dataset + .shard_writer() + .refresh_maintained_indexes(&dataset) + .await; self.dataset.update(dataset); Ok(()) } @@ -2078,6 +2084,50 @@ mod tests { assert_eq!(index.columns, vec!["tags".to_string()]); } + #[tokio::test] + async fn test_list_indices_with_missing_fts_index_files() { + let tmp_dir = tempdir().unwrap(); + let uri = tmp_dir.path().to_str().unwrap(); + let conn = connect(uri).execute().await.unwrap(); + let batch = record_batch!( + ("id", Int32, [1, 2, 3]), + ("text", Utf8, ["alpha", "beta", "gamma"]) + ) + .unwrap(); + let table = conn.create_table("t", batch).execute().await.unwrap(); + table + .create_index(&["text"], Index::FTS(FtsIndexBuilder::default())) + .execute() + .await + .unwrap(); + table + .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default())) + .execute() + .await + .unwrap(); + + let fts_uuid = table + .list_indices() + .await + .unwrap() + .into_iter() + .find(|index| index.name == "text_idx") + .and_then(|index| index.index_uuid) + .unwrap(); + std::fs::remove_dir_all(tmp_dir.path().join("t.lance/_indices").join(fts_uuid)).unwrap(); + + // A new connection, so the index cache cannot serve the deleted files. + let conn = connect(uri).execute().await.unwrap(); + let table = conn.open_table("t").execute().await.unwrap(); + let mut indices = table.list_indices().await.unwrap(); + indices.sort_by(|a, b| a.name.cmp(&b.name)); + let names = indices.iter().map(|i| i.name.as_str()).collect::>(); + assert_eq!(names, vec!["id_idx", "text_idx"]); + assert_eq!(indices[1].index_type, crate::index::IndexType::FTS); + assert_eq!(indices[1].index_details, None); + assert!(indices[0].index_details.is_some()); + } + #[tokio::test] async fn test_create_inverted_index() { let conn = connect("memory://").execute().await.unwrap(); diff --git a/rust/lancedb/src/table/merge.rs b/rust/lancedb/src/table/merge.rs index 982818ff6..354eef5a5 100644 --- a/rust/lancedb/src/table/merge.rs +++ b/rust/lancedb/src/table/merge.rs @@ -1505,6 +1505,30 @@ mod lsm_tests { ); } + /// A table with `id`/`text` and a primary key, ready for an LSM write spec. + async fn lsm_text_table(dir: &tempfile::TempDir) -> crate::Table { + let conn = connect(dir.path().to_str().unwrap()) + .execute() + .await + .unwrap(); + let table = conn + .create_table("t", id_text_reader(vec![(1, "alpha")])) + .execute() + .await + .unwrap(); + table.set_unenforced_primary_key(["id"]).await.unwrap(); + table + } + + /// Upsert `rows` through the LSM write path. + async fn upsert_text(table: &crate::Table, rows: Vec<(i64, &str)>) { + let mut builder = table.merge_insert(&[]); + builder + .when_matched_update_all(None) + .when_not_matched_insert_all(); + builder.execute(id_text_reader(rows)).await.unwrap(); + } + /// A reader of `[id: Int64, text: Utf8]` rows. fn id_text_reader(rows: Vec<(i64, &str)>) -> Box { let schema = Arc::new(Schema::new(vec![ @@ -1524,6 +1548,165 @@ mod lsm_tests { Box::new(RecordBatchIterator::new(vec![Ok(batch)], schema)) } + /// Under the default set, an index created while a writer is open reaches + /// that writer. + /// + /// The default maintains every index the table has, so creating one changes + /// what the table maintains without anyone touching the spec. A writer left + /// on its old configs would keep building MemTables the new index does not + /// cover. + #[tokio::test] + async fn lsm_default_set_picks_up_an_index_created_while_writing() { + use crate::index::Index; + use lance_index::scalar::FullTextSearchQuery; + + let dir = tempdir().unwrap(); + let table = lsm_text_table(&dir).await; + // The default set, on a table with no FTS index yet. + table + .set_lsm_write_spec(LsmWriteSpec::unsharded()) + .await + .unwrap(); + + // Opens the writer, whose MemTable carries no FTS index. + upsert_text(&table, vec![(99, "zebra")]).await; + + // Created with the writer still open and never reopened. + table + .create_index(&["text"], Index::FTS(Default::default())) + .execute() + .await + .unwrap(); + + upsert_text(&table, vec![(100, "zebra stripes")]).await; + + let query = FullTextSearchQuery::new("stripes".to_string()) + .with_column("text".to_string()) + .unwrap(); + let batches = table + .query() + .full_text_search(query) + .execute() + .await + .expect("the created index reached the open writer") + .try_collect::>() + .await + .unwrap(); + let found: usize = batches.iter().map(|b| b.num_rows()).sum(); + assert_eq!(found, 1, "the row written after create_index must be found"); + } + + /// A spec is replaced by unsetting it and setting the new one. + /// + /// `unset_lsm_write_spec` drains the open writer, so the rows it held + /// survive the replacement, and the writer the next write opens builds its + /// MemTables from the new set. + #[tokio::test] + async fn lsm_spec_is_replaced_by_unsetting_it_first() { + use crate::index::Index; + use lance_index::scalar::FullTextSearchQuery; + + let dir = tempdir().unwrap(); + let table = lsm_text_table(&dir).await; + table + .create_index(&["text"], Index::FTS(Default::default())) + .execute() + .await + .unwrap(); + let fts_index = table.list_indices().await.unwrap()[0].name.clone(); + + // Installed maintaining nothing, then written to: the writer opens with + // no FTS index. + table + .set_lsm_write_spec(LsmWriteSpec::unsharded().with_maintained_indexes(Vec::new())) + .await + .unwrap(); + upsert_text(&table, vec![(99, "zebra")]).await; + + // Setting over an installed spec is refused; unset, then set. + let err = table + .set_lsm_write_spec( + LsmWriteSpec::unsharded().with_maintained_indexes(vec![fts_index.clone()]), + ) + .await + .expect_err("an installed spec cannot be set over"); + assert!( + err.to_string().contains("already set"), + "unexpected error: {err}" + ); + table.unset_lsm_write_spec().await.unwrap(); + table + .set_lsm_write_spec(LsmWriteSpec::unsharded().with_maintained_indexes(vec![fts_index])) + .await + .unwrap(); + + // A row written after the replacement is answered by the index the new + // writer maintains. + upsert_text(&table, vec![(100, "zebra stripes")]).await; + + let query = FullTextSearchQuery::new("stripes".to_string()) + .with_column("text".to_string()) + .unwrap(); + let batches = table + .query() + .full_text_search(query) + .execute() + .await + .expect("the replaced spec maintains the index") + .try_collect::>() + .await + .unwrap(); + let found: usize = batches.iter().map(|b| b.num_rows()).sum(); + assert_eq!( + found, 1, + "the row written after the replacement must be found" + ); + } + + /// A row written before the index exists is searchable once it does. + /// + /// The MemTable holding it was built without the index. Creating the index + /// replaces the writer's configs, which seals that MemTable and waits for + /// its flush, so no resident MemTable is left that cannot answer the read. + #[tokio::test] + async fn lsm_a_row_written_before_the_index_is_searchable_after_it() { + use crate::index::Index; + use lance_index::scalar::FullTextSearchQuery; + + let dir = tempdir().unwrap(); + let table = lsm_text_table(&dir).await; + // No index yet, and the default set: maintain whatever the table has. + table + .set_lsm_write_spec(LsmWriteSpec::unsharded()) + .await + .unwrap(); + + // Opens the writer, whose MemTable therefore carries no FTS index. + upsert_text(&table, vec![(99, "zebra")]).await; + + // Built afterwards: the spec now resolves to it, the MemTable does not. + table + .create_index(&["text"], Index::FTS(Default::default())) + .execute() + .await + .unwrap(); + + let query = FullTextSearchQuery::new("zebra".to_string()) + .with_column("text".to_string()) + .unwrap(); + let batches = table + .query() + .full_text_search(query) + .execute() + .await + .expect("no resident MemTable is left that cannot answer") + .try_collect::>() + .await + .unwrap(); + let found: usize = batches.iter().map(|b| b.num_rows()).sum(); + assert_eq!(found, 1, "the row written before the index must be found"); + } + #[tokio::test] async fn lsm_read_full_text_search() { use crate::index::Index; diff --git a/rust/lancedb/src/table/merge/lsm.rs b/rust/lancedb/src/table/merge/lsm.rs index fc925c014..0f86fdcac 100644 --- a/rust/lancedb/src/table/merge/lsm.rs +++ b/rust/lancedb/src/table/merge/lsm.rs @@ -77,21 +77,36 @@ pub(crate) async fn set_lsm_write_spec(table: &NativeTable, spec: LsmWriteSpec) let dataset = table.dataset.get().await?; if dataset.mem_wal_index_details().await?.is_some() { return Err(Error::InvalidInput { - message: "set_lsm_write_spec: an LSM write spec is already set on this table; mutation is not supported".into(), + message: "set_lsm_write_spec: an LSM write spec is already set on this \ + table and cannot be changed" + .into(), }); } } - // Before the builder borrows the dataset clone. `list_indices` merges an - // index's segments into one entry, so the result needs no dedup. - let maintained_indexes = { - let dataset = table.dataset.get().await?; - resolve_maintained_indexes( - &dataset, - &table.list_indices().await?, - spec.maintained_indexes(), - ) - .await? + // A named set is checked against the table here, where the caller can still + // be told which names exist. An unnamed set is not resolved at all: it is an + // intent lance re-reads whenever it builds a MemTable, so an index created + // later is maintained without another call. + let maintained_indexes = match spec.maintained_indexes() { + Some(requested) => { + let indices = table.list_indices().await?; + for name in requested { + if !indices.iter().any(|index| &index.name == name) { + return Err(Error::InvalidInput { + message: format!( + "maintained index '{}' does not exist on this table; it has {}", + name, + index_name_list(&indices), + ), + }); + } + } + let dataset = table.dataset.get().await?; + validate_maintained_indexes(&dataset, requested).await?; + Some(requested.to_vec()) + } + None => None, }; table.checkout_latest().await?; @@ -138,7 +153,9 @@ pub(crate) async fn set_lsm_write_spec(table: &NativeTable, spec: LsmWriteSpec) writer_config_defaults } }; - builder = builder.maintained_indexes(maintained_indexes); + if let Some(maintained_indexes) = maintained_indexes { + builder = builder.maintained_indexes(maintained_indexes); + } for (key, value) in writer_config_defaults { builder = builder.add_writer_config_default(key, value); } @@ -147,48 +164,6 @@ pub(crate) async fn set_lsm_write_spec(table: &NativeTable, spec: LsmWriteSpec) Ok(()) } -/// Resolve a spec's maintained-index selection against `indices`, as reported -/// by [`Table::list_indices`](crate::Table::list_indices). -/// -/// `None` means every index on the table, snapshotted now. Lance validates -/// either selection against its shard-writer rules, so a spec that installs is -/// one the MemWAL can open. -/// -/// An unmaintainable index fails an inferred set rather than being dropped from -/// it — dropping would leave the caller believing it is maintained. -async fn resolve_maintained_indexes( - dataset: &Dataset, - indices: &[IndexConfig], - requested: Option<&[String]>, -) -> Result> { - let Some(requested) = requested else { - let all: Vec = indices.iter().map(|index| index.name.clone()).collect(); - validate_maintained_indexes(dataset, &all) - .await - .map_err(|source| Error::InvalidInput { - message: format!( - "cannot maintain every index on this table: {source}. Set \ - maintained_indexes explicitly to choose from {}", - index_name_list(indices), - ), - })?; - return Ok(all); - }; - for name in requested { - if !indices.iter().any(|index| &index.name == name) { - return Err(Error::InvalidInput { - message: format!( - "maintained index '{}' does not exist on this table; it has {}", - name, - index_name_list(indices), - ), - }); - } - } - validate_maintained_indexes(dataset, requested).await?; - Ok(requested.to_vec()) -} - /// Index names for an error message. fn index_name_list(indices: &[IndexConfig]) -> String { if indices.is_empty() { @@ -298,7 +273,9 @@ fn lsm_write_spec_from_details( }; Ok(base - .with_maintained_indexes(details.maintained_indexes.clone()) + .with_maintained_indexes( + (!details.maintain_all_indexes).then(|| details.maintained_indexes.clone()), + ) .with_writer_config_defaults(details.writer_config_defaults.clone())) } @@ -460,6 +437,55 @@ impl ShardWriterCache { Ok(entry) } + /// Install `dataset`'s maintained index set on the cached writer, if one is + /// open, so a change reaches a table that is already being written to. + /// + /// Reported rather than propagated: the index is committed durably by the + /// time this runs, and the writer this process happens to hold may be fenced + /// or poisoned -- which is not a reason to fail creating an index. Any + /// writer opened afterwards reads the committed set. + /// + /// Holds the entry's read lock across the refresh, so a concurrent close + /// waits rather than retiring the writer mid-seal. Writes take the same + /// lock first, so the order matches theirs. + pub(crate) async fn refresh_maintained_indexes(&self, dataset: &Dataset) { + let entry = { + let guard = self.slot.read().await; + guard.as_ref().map(|(_, entry)| entry.clone()) + }; + let Some(entry) = entry else { + return; + }; + let guard = entry.inner.read().await; + let Some(writer) = guard.as_ref() else { + return; + }; + // Replacing the configs seals the active memtable so the next one carries + // the new indexes. The outgoing one keeps the set it was built with, so + // its flush has to land before a read: while it is still frozen, an + // indexed read sees a resident memtable that cannot answer and is + // refused. + let sealed = match dataset.refresh_mem_wal_index_configs(writer).await { + Ok(sealed) => sealed, + Err(error) => { + log::warn!( + "the maintained index set changed but the open shard writer could \ + not be refreshed, so it keeps the set it opened with: {error}" + ); + return; + } + }; + if let Some(fence) = sealed + && let Err(error) = fence.wait().await + { + log::warn!( + "the shard writer took the new maintained index set, but the memtable \ + it sealed was not flushed, so an indexed read stays refused until \ + it is: {error}" + ); + } + } + /// Snapshot the cached writer's shard for the LSM read path: its shard id, /// authoritative in-memory manifest, and active + frozen memtable refs. /// Returns `None` when no writer is currently cached (e.g. nothing has been diff --git a/rust/lancedb/src/table/query/lsm.rs b/rust/lancedb/src/table/query/lsm.rs index 86c1fe5f2..f3e40fc7b 100644 --- a/rust/lancedb/src/table/query/lsm.rs +++ b/rust/lancedb/src/table/query/lsm.rs @@ -33,7 +33,7 @@ use datafusion_physical_plan::expressions::Column; use datafusion_physical_plan::projection::ProjectionExec; use datafusion_physical_plan::{ExecutionPlan, PhysicalExpr}; use lance::Dataset; -use lance::dataset::mem_wal::scanner::InMemoryMemTables; +use lance::dataset::mem_wal::scanner::{InMemoryMemTableRef, InMemoryMemTables}; use lance::dataset::mem_wal::{ DatasetMemWalExt, LsmScanner, ShardManifestStore, ShardSnapshot, ShardWriterConfig, }; @@ -468,20 +468,19 @@ async fn fts_plan( .to_string(), })?; - // Without a maintained in-memory FTS index for this column, the active memtable - // arm produces an empty plan (`active_source_can_execute_fts` returns false), so - // the search silently omits un-compacted documents. Reject rather than mislead. - if !index_maintained( - dataset, - column, - &details.maintained_indexes, - "InvertedIndexDetails", - ) - .await? + // Two ways the fresh tier cannot answer: the spec maintains no FTS index for + // this column, or a resident MemTable was built before the index joined the + // set. Either leaves the memtable arm an empty plan + // (`active_source_can_execute_fts` returns false), so the search silently + // omits un-compacted documents. Reject rather than mislead. + if !index_maintained(dataset, column, details, "InvertedIndexDetails").await? + || !resident_memtables_carry(&in_memory, |memtable| { + memtable.index_store.get_fts_by_column(column).is_some() + }) { return Err(Error::NotSupported { message: format!( - "the MemWAL LSM scanner full-text search requires the FTS index on '{column}' to be maintained by the write spec (LsmWriteSpec::with_maintained_indexes); otherwise un-compacted documents are omitted. set use_lsm(false) to read the base table only" + "the MemWAL LSM scanner full-text search requires the FTS index on '{column}' to be maintained by the write spec (LsmWriteSpec::with_maintained_indexes) and carried by every resident MemTable; otherwise un-compacted documents are omitted. set use_lsm(false) to read the base table only" ), }); } @@ -493,6 +492,27 @@ async fn fts_plan( Ok(scanner.create_plan().await?) } +/// Whether the spec maintains the index named `name`. +/// +/// An unnamed set names nothing and covers everything, so the stored list alone +/// cannot answer this. +fn spec_maintains(details: &MemWalIndexDetails, name: &str) -> bool { + details.maintain_all_indexes || details.maintained_indexes.iter().any(|n| n == name) +} + +/// Whether every resident MemTable carries the index an arm needs. +/// +/// The spec is intent; a MemTable built before the index joined the set +/// contributes nothing to an indexed read, so its rows would go missing. +fn resident_memtables_carry( + in_memory: &HashMap, + carries: impl Fn(&InMemoryMemTableRef) -> bool, +) -> bool { + in_memory + .values() + .all(|tables| carries(&tables.active) && tables.frozen.iter().all(&carries)) +} + /// Whether an index of `type_url_suffix` covering `column` is in the MemWAL spec's /// maintained set. Only a maintained index has its catch-up tracked (so exclusion is /// gated correctly) and its in-memory arm kept current; an unmaintained base index @@ -502,7 +522,7 @@ async fn fts_plan( async fn index_maintained( dataset: &Dataset, column: &str, - maintained: &[String], + details: &MemWalIndexDetails, type_url_suffix: &str, ) -> Result { use lance::index::DatasetIndexExt; @@ -512,7 +532,7 @@ async fn index_maintained( let indices = dataset.load_indices().await?; Ok(indices.iter().any(|idx| { idx.fields.contains(&field.id) - && maintained.iter().any(|m| m == &idx.name) + && spec_maintains(details, &idx.name) && idx .index_details .as_ref() @@ -583,9 +603,7 @@ async fn arm_maintained_index_names( }) .map(|idx| idx.name.clone()) .collect(); - if let Some(name) = - resolve_single_index(segment_names, &details.maintained_indexes, arm, &column)? - { + if let Some(name) = resolve_single_index(segment_names, details, arm, &column)? { names.push(name); } } @@ -604,7 +622,7 @@ async fn arm_maintained_index_names( /// compaction watermark). fn resolve_single_index( mut names: Vec, - maintained: &[String], + details: &MemWalIndexDetails, arm: &str, column: &str, ) -> Result> { @@ -620,7 +638,7 @@ fn resolve_single_index( Ok(names .into_iter() .next() - .filter(|name| maintained.contains(name))) + .filter(|name| spec_maintains(details, name))) } /// Drop the primary-key columns Lance appends internally for dedup when the user's @@ -688,21 +706,19 @@ async fn vector_plan( None => default_vector_column(&arrow_schema, Some(query_vector.len() as i32))?, }; - // The base arm relies on the column's vector index (`fast_search`). Unless it is - // maintained, its catch-up is untracked and exclusion falls back to the - // compaction watermark — dropping compacted SSTables the (lagging) base index has - // not re-indexed. Reject rather than silently omit rows, mirroring the FTS arm. - if !index_maintained( - dataset, - &column, - &details.maintained_indexes, - "VectorIndexDetails", - ) - .await? + // The base arm relies on the column's vector index (`fast_search`). Unmaintained, + // its catch-up is untracked and exclusion falls back to the compaction watermark, + // dropping compacted SSTables the lagging base index has not re-indexed; and a + // resident MemTable built before the index joined the set carries none either. + // Reject rather than silently omit rows, mirroring the FTS arm. + if !index_maintained(dataset, &column, details, "VectorIndexDetails").await? + || !resident_memtables_carry(&in_memory, |memtable| { + memtable.index_store.get_hnsw_by_column(&column).is_some() + }) { return Err(Error::NotSupported { message: format!( - "the MemWAL LSM scanner requires the vector index on '{column}' to be maintained by the write spec (LsmWriteSpec::with_maintained_indexes); otherwise compacted rows not yet re-indexed are omitted. set use_lsm(false) to read the base table only" + "the MemWAL LSM scanner requires the vector index on '{column}' to be maintained by the write spec (LsmWriteSpec::with_maintained_indexes) and carried by every resident MemTable; otherwise compacted rows not yet re-indexed are omitted. set use_lsm(false) to read the base table only" ), }); } @@ -888,7 +904,10 @@ mod tests { #[test] fn resolve_single_index_dedupes_segments() { - let maintained = vec!["fts_idx".to_string()]; + let maintained = MemWalIndexDetails { + maintained_indexes: vec!["fts_idx".to_string()], + ..Default::default() + }; // Two physical segments of ONE logical index must not count as "multiple". assert_eq!( resolve_single_index( @@ -917,4 +936,33 @@ mod tests { None ); } + + /// A spec that maintains every index names none of them, so the stored list + /// is empty while every index on the table is in fact maintained. Reading + /// the list alone would refuse the read path on the default spec. + #[test] + fn an_unnamed_set_maintains_every_index() { + let all = MemWalIndexDetails { + maintained_indexes: Vec::new(), + maintain_all_indexes: true, + ..Default::default() + }; + assert!(spec_maintains(&all, "fts_idx")); + assert_eq!( + resolve_single_index(vec!["fts_idx".to_string()], &all, "full-text", "text").unwrap(), + Some("fts_idx".to_string()), + ); + + // The same empty list without the flag maintains nothing. + let none = MemWalIndexDetails { + maintained_indexes: Vec::new(), + maintain_all_indexes: false, + ..Default::default() + }; + assert!(!spec_maintains(&none, "fts_idx")); + assert_eq!( + resolve_single_index(vec!["fts_idx".to_string()], &none, "full-text", "text").unwrap(), + None, + ); + } } diff --git a/rust/lancedb/tests/first_class_function_slice2.rs b/rust/lancedb/tests/first_class_function_slice2.rs index 14ceb511d..380766e03 100644 --- a/rust/lancedb/tests/first_class_function_slice2.rs +++ b/rust/lancedb/tests/first_class_function_slice2.rs @@ -97,13 +97,16 @@ async fn local_function_catalog_operations_return_stable_not_supported() { )) .unwrap(); - let create_error = connection.create_function_async(request).await.unwrap_err(); + let create_error = connection + .create_function_async(request, &[]) + .await + .unwrap_err(); let lookup_error = connection - .get_function("normalize_score", "1") + .get_function("normalize_score", "1", &[]) .await .unwrap_err(); let drop_error = connection - .drop_function("normalize_score", "1") + .drop_function("normalize_score", "1", &[]) .await .unwrap_err(); for error in [create_error, lookup_error, drop_error] {