Bump version: 0.1.7 → 0.1.8

* Refactors the Node module to load the shared library from a separate
package. When a user does `npm install vectordb`, the correct optional
dependency is automatically downloaded by npm.
* Add scripts and instructions to build Linux and MacOS node artifacts
locally.
* Add instructions for publishing the npm module and crates.

Co-authored-by: Will Jones <willjones127@gmail.com>

.bumpversion.cfg

@@ -0,0 +1,12 @@
+[bumpversion]
+current_version = 0.1.8
+commit = True
+message = Bump version: {current_version} → {new_version}
+tag = True
+tag_name = v{new_version}
+
+[bumpversion:file:node/package.json]
+
+[bumpversion:file:rust/ffi/node/Cargo.toml]
+
+[bumpversion:file:rust/vectordb/Cargo.toml]

.github/workflows/cargo-publish.yml

@@ -0,0 +1,29 @@
+name: Cargo Publish
+
+on:
+  release:
+    types: [ published ]
+
+env:
+  # This env var is used by Swatinem/rust-cache@v2 for the cache
+  # key, so we set it to make sure it is always consistent.
+  CARGO_TERM_COLOR: always
+
+jobs:
+  build:
+    runs-on: ubuntu-22.04
+    timeout-minutes: 30
+    # Only runs on tags that matches the make-release action
+    if: startsWith(github.ref, 'refs/tags/v')
+    steps:
+      - uses: actions/checkout@v3
+      - uses: Swatinem/rust-cache@v2
+        with:
+          workspaces: rust
+      - name: Install dependencies
+        run: |
+          sudo apt update
+          sudo apt install -y protobuf-compiler libssl-dev
+      - name: Publish the package
+        run: |
+          cargo publish -p vectordb --all-features --token ${{ secrets.CARGO_REGISTRY_TOKEN }}

.github/workflows/make-release-commit.yml

@@ -0,0 +1,55 @@
+name: Create release commit
+
+on:
+  workflow_dispatch:
+    inputs:
+      dry_run:
+        description: 'Dry run (create the local commit/tags but do not push it)'
+        required: true
+        default: "false"
+        type: choice
+        options:
+          - "true"
+          - "false"
+      part:
+        description: 'What kind of release is this?'
+        required: true
+        default: 'patch'
+        type: choice
+        options:
+          - patch
+          - minor
+          - major
+
+jobs:
+  bump-version:
+    runs-on: ubuntu-latest
+    steps:
+    - name: Check out main
+      uses: actions/checkout@v3
+      with:
+        ref: main
+        persist-credentials: false
+        fetch-depth: 0
+        lfs: true
+    - name: Set git configs for bumpversion
+      shell: bash
+      run: |
+        git config user.name 'Lance Release'
+        git config user.email 'lance-dev@lancedb.com'
+    - name: Set up Python 3.10
+      uses: actions/setup-python@v4
+      with:
+        python-version: "3.10"
+    - name: Bump version, create tag and commit
+      run: |
+        pip install bump2version
+        bumpversion --verbose ${{ inputs.part }}
+    - name: Push new version and tag
+      if: ${{ inputs.dry_run }} == "false"
+      uses: ad-m/github-push-action@master
+      with:
+        github_token: ${{ secrets.LANCEDB_RELEASE_TOKEN }}
+        branch: main
+        tags: true
+

.github/workflows/node.yml

@@ -67,8 +67,12 @@ jobs:
     - name: Build
       run: |
         npm ci
-        npm run build
         npm run tsc
+        npm run build
+        npm run pack-build
+        npm install --no-save ./dist/vectordb-*.tgz
+        # Remove index.node to test with dependency installed
+        rm index.node
     - name: Test
       run: npm run test
   macos:
@@ -94,8 +98,12 @@ jobs:
     - name: Build
       run: |
         npm ci
-        npm run build
         npm run tsc
+        npm run build
+        npm run pack-build
+        npm install --no-save ./dist/vectordb-*.tgz
+        # Remove index.node to test with dependency installed
+        rm index.node
     - name: Test
       run: |
         npm run test

.github/workflows/npm-publish.yml

@@ -0,0 +1,137 @@
+name: NPM Publish
+
+on:
+  release:
+    types: [ published ]
+
+jobs:
+  node:
+    runs-on: ubuntu-latest
+    # Only runs on tags that matches the make-release action
+    if: startsWith(github.ref, 'refs/tags/v')
+    defaults:
+      run:
+        shell: bash
+        working-directory: node
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+      - uses: actions/setup-node@v3
+        with:
+          node-version: 20
+          cache: 'npm'
+          cache-dependency-path: node/package-lock.json
+      - name: Install dependencies
+        run: |
+          sudo apt update
+          sudo apt install -y protobuf-compiler libssl-dev
+      - name: Build
+        run: |
+          npm ci
+          npm run tsc
+          npm pack
+      - name: Upload Linux Artifacts
+        uses: actions/upload-artifact@v3
+        with:
+          name: node-package
+          path: |
+            node/vectordb-*.tgz
+
+  node-macos:
+    runs-on: macos-12
+    # Only runs on tags that matches the make-release action
+    if: startsWith(github.ref, 'refs/tags/v')
+    strategy:
+      fail-fast: false
+      matrix:
+        target: [x86_64-apple-darwin, aarch64-apple-darwin]
+    steps:
+    - name: Checkout
+      uses: actions/checkout@v3
+    - name: Install system dependencies
+      run: brew install protobuf
+    - name: Install npm dependencies
+      run: |
+        cd node
+        npm ci
+    - name: Install rustup target
+      if: ${{ matrix.target == 'aarch64-apple-darwin' }}
+      run: rustup target add aarch64-apple-darwin
+    - name: Build MacOS native node modules
+      run: bash ci/build_macos_artifacts.sh ${{ matrix.target }}
+    - name: Upload Darwin Artifacts
+      uses: actions/upload-artifact@v3
+      with:
+        name: darwin-native
+        path: |
+          node/dist/vectordb-darwin*.tgz
+
+  node-linux:
+    name: node-linux (${{ matrix.arch}}-unknown-linux-${{ matrix.libc }})
+    runs-on: ubuntu-latest
+    # Only runs on tags that matches the make-release action
+    if: startsWith(github.ref, 'refs/tags/v')
+    strategy:
+      fail-fast: false
+      matrix:
+        libc:
+          - gnu
+          # TODO: re-enable musl once we have refactored to pre-built containers
+          # Right now we have to build node from source which is too expensive.
+          # - musl
+        arch:
+          - x86_64
+          # Building on aarch64 is too slow for now
+          # - aarch64
+    steps:
+    - name: Checkout
+      uses: actions/checkout@v3
+    - name: Change owner to root (for npm)
+      # The docker container is run as root, so we need the files to be owned by root
+      # Otherwise npm is a nightmare: https://github.com/npm/cli/issues/3773
+      run: sudo chown -R root:root .
+    - name: Set up QEMU
+      if: ${{ matrix.arch == 'aarch64' }}
+      uses: docker/setup-qemu-action@v2
+      with:
+        platforms: arm64
+    - name: Build Linux GNU native node modules
+      if: ${{ matrix.libc == 'gnu' }}
+      run: |
+        docker run \
+          -v $(pwd):/io -w /io \
+          quay.io/pypa/manylinux2014_${{ matrix.arch }} \
+          bash ci/build_linux_artifacts.sh ${{ matrix.arch }}-unknown-linux-gnu
+    - name: Build musl Linux native node modules
+      if: ${{ matrix.libc == 'musl' }}
+      run: |
+        docker run --platform linux/arm64/v8 \
+          -v $(pwd):/io -w /io \
+          quay.io/pypa/musllinux_1_1_${{ matrix.arch }} \
+          bash ci/build_linux_artifacts.sh ${{ matrix.arch }}-unknown-linux-musl
+    - name: Upload Linux Artifacts
+      uses: actions/upload-artifact@v3
+      with:
+        name: linux-native
+        path: |
+          node/dist/vectordb-linux*.tgz
+
+  release:
+    needs: [node, node-macos, node-linux]
+    runs-on: ubuntu-latest
+    # Only runs on tags that matches the make-release action
+    if: startsWith(github.ref, 'refs/tags/v')
+    steps:
+      - uses: actions/download-artifact@v3
+      - name: Display structure of downloaded files
+        run: ls -R
+      - uses: actions/setup-node@v3
+        with:
+          node-version: 20
+      - name: Publish to NPM
+        env:
+          NODE_AUTH_TOKEN: ${{ secrets.LANCEDB_NPM_REGISTRY_TOKEN }}
+        run: |
+          for filename in */*.tgz; do
+            npm publish $filename
+          done

.github/workflows/pypi-publish.yml

@@ -0,0 +1,31 @@
+name: PyPI Publish
+
+on:
+  release:
+    types: [ published ]
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+    # Only runs on tags that matches the python-make-release action
+    if: startsWith(github.ref, 'refs/tags/python-v')
+    defaults:
+      run:
+        shell: bash
+        working-directory: python
+    steps:
+      - uses: actions/checkout@v3
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.8"
+      - name: Build distribution
+        run: |
+          ls -la
+          pip install wheel setuptools --upgrade
+          python setup.py sdist bdist_wheel
+      - name: Publish
+        uses: pypa/gh-action-pypi-publish@v1.8.5
+        with:
+          password: ${{ secrets.LANCEDB_PYPI_API_TOKEN }}
+          packages-dir: python/dist

.github/workflows/python-make-release-commit.yml

@@ -0,0 +1,56 @@
+name: Python - Create release commit
+
+on:
+  workflow_dispatch:
+    inputs:
+      dry_run:
+        description: 'Dry run (create the local commit/tags but do not push it)'
+        required: true
+        default: "false"
+        type: choice
+        options:
+          - "true"
+          - "false"
+      part:
+        description: 'What kind of release is this?'
+        required: true
+        default: 'patch'
+        type: choice
+        options:
+          - patch
+          - minor
+          - major
+
+jobs:
+  bump-version:
+    runs-on: ubuntu-latest
+    steps:
+    - name: Check out main
+      uses: actions/checkout@v3
+      with:
+        ref: main
+        persist-credentials: false
+        fetch-depth: 0
+        lfs: true
+    - name: Set git configs for bumpversion
+      shell: bash
+      run: |
+        git config user.name 'Lance Release'
+        git config user.email 'lance-dev@lancedb.com'
+    - name: Set up Python 3.10
+      uses: actions/setup-python@v4
+      with:
+        python-version: "3.10"
+    - name: Bump version, create tag and commit
+      working-directory: python
+      run: |
+        pip install bump2version
+        bumpversion --verbose ${{ inputs.part }}
+    - name: Push new version and tag
+      if: ${{ inputs.dry_run }} == "false"
+      uses: ad-m/github-push-action@master
+      with:
+        github_token: ${{ secrets.LANCEDB_RELEASE_TOKEN }}
+        branch: main
+        tags: true
+

.github/workflows/python.yml

@@ -32,9 +32,15 @@ jobs:
       run: |
         pip install -e .
         pip install tantivy@git+https://github.com/quickwit-oss/tantivy-py#164adc87e1a033117001cf70e38c82a53014d985
-        pip install pytest pytest-mock
+        pip install pytest pytest-mock black isort
+    - name: Black
+      run: black --check --diff --no-color --quiet .
+    - name: isort
+      run: isort --check --diff --quiet .
     - name: Run tests
       run: pytest -x -v --durations=30 tests
+    - name: doctest
+      run: pytest --doctest-modules lancedb
   mac:
     timeout-minutes: 30
     runs-on: "macos-12"

.github/workflows/rust.yml

@@ -0,0 +1,67 @@
+name: Rust
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+    paths:
+      - rust/**
+      - .github/workflows/rust.yml
+
+env:
+  # This env var is used by Swatinem/rust-cache@v2 for the cache
+  # key, so we set it to make sure it is always consistent.
+  CARGO_TERM_COLOR: always
+  # Disable full debug symbol generation to speed up CI build and keep memory down
+  # "1" means line tables only, which is useful for panic tracebacks.
+  RUSTFLAGS: "-C debuginfo=1"
+  RUST_BACKTRACE: "1"
+
+jobs:
+  linux:
+    timeout-minutes: 30
+    runs-on: ubuntu-22.04
+    defaults:
+      run:
+        shell: bash
+        working-directory: rust
+    steps:
+    - uses: actions/checkout@v3
+      with:
+          fetch-depth: 0
+          lfs: true
+    - uses: Swatinem/rust-cache@v2
+      with:
+          workspaces: rust
+    - name: Install dependencies
+      run: |
+          sudo apt update
+          sudo apt install -y protobuf-compiler libssl-dev
+    - name: Build
+      run: cargo build --all-features
+    - name: Run tests
+      run: cargo test --all-features
+  macos:
+    runs-on: macos-12
+    timeout-minutes: 30
+    defaults:
+      run:
+        shell: bash
+        working-directory: rust
+    steps:
+      - uses: actions/checkout@v3
+        with:
+            fetch-depth: 0
+            lfs: true
+      - name: CPU features
+        run: sysctl -a | grep cpu
+      - uses: Swatinem/rust-cache@v2
+        with:
+          workspaces: rust
+      - name: Install dependencies
+        run: brew install protobuf
+      - name: Build
+        run: cargo build --all-features
+      - name: Run tests
+        run: cargo test --all-features

.gitignore

@@ -4,6 +4,8 @@
 **/__pycache__
 .DS_Store
 
+.vscode
+
 rust/target
 rust/Cargo.lock
 

Cargo.lock

@@ -190,6 +190,7 @@ dependencies = [
  "arrow-data",
  "arrow-schema",
  "flatbuffers",
+ "zstd",
 ]
 
 [[package]]
@@ -654,6 +655,12 @@ version = "3.12.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9b1ce199063694f33ffb7dd4e0ee620741495c32833cde5aa08f02a0bf96f0c8"
 
+[[package]]
+name = "bytemuck"
+version = "1.13.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "17febce684fd15d89027105661fec94afb475cb995fbc59d2865198446ba2eea"
+
 [[package]]
 name = "byteorder"
 version = "1.4.3"
@@ -1646,9 +1653,9 @@ dependencies = [
 
 [[package]]
 name = "lance"
-version = "0.4.17"
+version = "0.4.21"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "86dda8185bd1ffae7b910c1f68035af23be9b717c52e9cc4de176cd30b47f772"
+checksum = "3d6c2e7bcfc71c7167ec70cd06c6d55c644a148f6580218c5a0b66e13ac5b5cc"
 dependencies = [
  "accelerate-src",
  "arrow",
@@ -1657,7 +1664,9 @@ dependencies = [
  "arrow-buffer",
  "arrow-cast",
  "arrow-data",
+ "arrow-ipc",
  "arrow-ord",
+ "arrow-row",
  "arrow-schema",
  "arrow-select",
  "async-recursion",
@@ -1668,6 +1677,7 @@ dependencies = [
  "bytes",
  "cblas",
  "chrono",
+ "dashmap",
  "datafusion",
  "futures",
  "lapack",
@@ -1684,6 +1694,7 @@ dependencies = [
  "prost-types",
  "rand",
  "reqwest",
+ "roaring",
  "shellexpand",
  "snafu",
  "sqlparser-lance",
@@ -2598,6 +2609,12 @@ dependencies = [
  "winreg",
 ]
 
+[[package]]
+name = "retain_mut"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8c31b5c4033f8fdde8700e4657be2c497e7288f01515be52168c631e2e4d4086"
+
 [[package]]
 name = "ring"
 version = "0.16.20"
@@ -2613,6 +2630,17 @@ dependencies = [
  "winapi",
 ]
 
+[[package]]
+name = "roaring"
+version = "0.10.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ef0fb5e826a8bde011ecae6a8539dd333884335c57ff0f003fbe27c25bbe8f71"
+dependencies = [
+ "bytemuck",
+ "byteorder",
+ "retain_mut",
+]
+
 [[package]]
 name = "rustc_version"
 version = "0.4.0"
@@ -3358,7 +3386,7 @@ checksum = "accd4ea62f7bb7a82fe23066fb0957d48ef677f6eeb8215f372f52e48bb32426"
 
 [[package]]
 name = "vectordb"
-version = "0.0.1"
+version = "0.1.6"
 dependencies = [
  "arrow-array",
  "arrow-data",
@@ -3373,7 +3401,7 @@ dependencies = [
 
 [[package]]
 name = "vectordb-node"
-version = "0.1.0"
+version = "0.1.6"
 dependencies = [
  "arrow-array",
  "arrow-ipc",

README.md

@@ -75,4 +75,4 @@ result = table.search([100, 100]).limit(2).to_df()
 
 ## Blogs, Tutorials & Videos
 * 📈 <a href="https://blog.eto.ai/benchmarking-random-access-in-lance-ed690757a826">2000x better performance with Lance over Parquet</a>
-* 🤖 <a href="https://github.com/lancedb/lancedb/blob/main/notebooks/youtube_transcript_search.ipynb">Build a question and answer bot with LanceDB</a>
+* 🤖 <a href="https://github.com/lancedb/lancedb/blob/main/docs/src/notebooks/youtube_transcript_search.ipynb">Build a question and answer bot with LanceDB</a>

ci/build_linux_artifacts.sh

@@ -0,0 +1,91 @@
+#!/bin/bash
+# Builds the Linux artifacts (node binaries).
+# Usage: ./build_linux_artifacts.sh [target]
+# Targets supported: 
+# - x86_64-unknown-linux-gnu:centos
+# - aarch64-unknown-linux-gnu:centos
+# - aarch64-unknown-linux-musl
+# - x86_64-unknown-linux-musl
+
+# TODO: refactor this into a Docker container we can pull
+
+set -e
+
+setup_dependencies() {
+    echo "Installing system dependencies..."
+    if [[ $1 == *musl ]]; then
+        # musllinux
+        apk add openssl-dev
+    else
+        # manylinux2014
+        yum install -y openssl-devel unzip
+    fi
+
+    if [[ $1 == x86_64* ]]; then
+        ARCH=x86_64
+    else
+        # gnu target
+        ARCH=aarch_64
+    fi
+    
+    # Install new enough protobuf (yum-provided is old)
+    PB_REL=https://github.com/protocolbuffers/protobuf/releases
+    PB_VERSION=23.1
+    curl -LO $PB_REL/download/v$PB_VERSION/protoc-$PB_VERSION-linux-$ARCH.zip
+    unzip protoc-$PB_VERSION-linux-$ARCH.zip -d /usr/local
+}
+
+install_node() {
+    echo "Installing node..."
+    curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.34.0/install.sh | bash
+    source "$HOME"/.bashrc
+
+    if [[ $1 == *musl ]]; then
+        # This node version is 15, we need 16 or higher:
+        # apk add nodejs-current npm 
+        # So instead we install from source (nvm doesn't provide binaries for musl):
+        nvm install -s --no-progress 17
+    else
+        nvm install --no-progress 17 # latest that supports glibc 2.17
+    fi
+}
+
+install_rust() {
+    echo "Installing rust..."
+    curl https://sh.rustup.rs -sSf | bash -s -- -y
+    export PATH="$PATH:/root/.cargo/bin"
+}
+
+build_node_binary() {
+    echo "Building node library for $1..."
+    pushd node
+
+    npm ci
+    
+    if [[ $1 == *musl ]]; then
+        # This is needed for cargo to allow build cdylibs with musl
+        export RUSTFLAGS="-C target-feature=-crt-static"
+    fi
+
+    # Cargo can run out of memory while pulling dependencies, espcially when running
+    # in QEMU. This is a workaround for that.
+    export CARGO_NET_GIT_FETCH_WITH_CLI=true
+
+    # We don't pass in target, since the native target here already matches
+    # and openblas-src doesn't do well with cross-compilation.
+    npm run build-release
+    npm run pack-build
+
+    popd
+}
+
+TARGET=${1:-x86_64-unknown-linux-gnu}
+# Others:
+# aarch64-unknown-linux-gnu
+# x86_64-unknown-linux-musl
+# aarch64-unknown-linux-musl
+
+setup_dependencies $TARGET
+install_node $TARGET
+install_rust
+build_node_binary $TARGET

ci/build_macos_artifacts.sh

@@ -0,0 +1,33 @@
+# Builds the macOS artifacts (node binaries).
+# Usage: ./ci/build_macos_artifacts.sh [target]
+# Targets supported: x86_64-apple-darwin aarch64-apple-darwin
+
+prebuild_rust() {
+    # Building here for the sake of easier debugging.
+    pushd rust/ffi/node
+    echo "Building rust library for $1"
+    export RUST_BACKTRACE=1
+    cargo build --release --target $1
+    popd
+}
+
+build_node_binaries() {
+    pushd node
+    echo "Building node library for $1"
+    npm run build-release -- --target $1
+    npm run pack-build -- --target $1
+    popd
+}
+
+if [ -n "$1" ]; then
+    targets=$1
+else
+    targets="x86_64-apple-darwin aarch64-apple-darwin"
+fi
+
+echo "Building artifacts for targets: $targets"
+for target in $targets
+    do
+    prebuild_rust $target
+    build_node_binaries $target
+done

docs/mkdocs.yml

@@ -14,10 +14,24 @@ theme:
 
 plugins:
 - search
+- autorefs
 - mkdocstrings:
     handlers:
       python:
         paths: [../python]
+        selection:
+          docstring_style: numpy
+        rendering:
+          heading_level: 4
+          show_source: false
+          show_symbol_type_in_heading: true
+          show_signature_annotations: true
+          show_root_heading: true
+          members_order: source
+        import:
+          # for cross references
+          - https://arrow.apache.org/docs/objects.inv
+          - https://pandas.pydata.org/docs/objects.inv
 - mkdocs-jupyter
 
 markdown_extensions:
@@ -41,9 +55,13 @@ nav:
 - Python full-text search: fts.md
 - Python integrations: integrations.md
 - Python examples:
-  - YouTube Transcript Search using OpenAI: notebooks/youtube_transcript_search.ipynb
+  - YouTube Transcript Search: notebooks/youtube_transcript_search.ipynb
   - Documentation QA Bot using LangChain: notebooks/code_qa_bot.ipynb
   - Multimodal search using CLIP: notebooks/multimodal_search.ipynb
+  - Serverless QA Bot with S3 and Lambda: examples/serverless_lancedb_with_s3_and_lambda.md
+  - Serverless QA Bot with Modal: examples/serverless_qa_bot_with_modal_and_langchain.md
+- Javascript examples:
+  - YouTube Transcript Search: examples/youtube_transcript_bot_with_nodejs.md
 - References:
   - Vector Search: search.md
   - Indexing: ann_indexes.md

docs/src/ann_indexes.md

@@ -67,7 +67,7 @@ There are a couple of parameters that can be used to fine-tune the search:
   e.g., for 1M vectors divided up into 256 partitions, nprobes should be set to ~20-40.<br/>
   Note: nprobes is only applicable if an ANN index is present. If specified on a table without an ANN index, it is ignored.
 - **refine_factor** (default: None): Refine the results by reading extra elements and re-ranking them in memory.<br/>
-  A higher number makes search more accurate but also slower. If you find the recall is less than idea, try refine_factor=10 to start.<br/>
+  A higher number makes search more accurate but also slower. If you find the recall is less than ideal, try refine_factor=10 to start.<br/>
   e.g., for 1M vectors divided into 256 partitions, if you're looking for top 20, then refine_factor=200 reranks the whole partition.<br/>
   Note: refine_factor is only applicable if an ANN index is present. If specified on a table without an ANN index, it is ignored.
 

docs/src/embedding.md

@@ -45,7 +45,7 @@ You can also use an external API like OpenAI to generate embeddings
         assert len(openai.Model.list()["data"]) > 0
 
         def embed_func(c):
-        rs = openai.Embedding.create(input=c, engine="text-embedding-ada-002")
+            rs = openai.Embedding.create(input=c, engine="text-embedding-ada-002")
         return [record["embedding"] for record in rs["data"]]
       ```
 

docs/src/examples/modal_langchain.py

@@ -1,27 +1,22 @@
-import sys
-from modal import Secret, Stub, Image, web_endpoint
-import lancedb
-import re
 import pickle
-import requests
+import re
+import sys
 import zipfile
 from pathlib import Path
 
+import requests
+from langchain.chains import RetrievalQA
 from langchain.document_loaders import UnstructuredHTMLLoader
 from langchain.embeddings import OpenAIEmbeddings
+from langchain.llms import OpenAI
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain.vectorstores import LanceDB
-from langchain.llms import OpenAI
-from langchain.chains import RetrievalQA
+from modal import Image, Secret, Stub, web_endpoint
+
+import lancedb
 
 lancedb_image = Image.debian_slim().pip_install(
-    "lancedb",
-    "langchain",
-    "openai",
-    "pandas",
-    "tiktoken",
-    "unstructured",
-    "tabulate"
+    "lancedb", "langchain", "openai", "pandas", "tiktoken", "unstructured", "tabulate"
 )
 
 stub = Stub(
@@ -34,21 +29,26 @@ docsearch = None
 docs_path = Path("docs.pkl")
 db_path = Path("lancedb")
 
+
 def get_document_title(document):
     m = str(document.metadata["source"])
     title = re.findall("pandas.documentation(.*).html", m)
     if title[0] is not None:
-        return(title[0])
-    return ''
+        return title[0]
+    return ""
+
 
 def download_docs():
-    pandas_docs = requests.get("https://eto-public.s3.us-west-2.amazonaws.com/datasets/pandas_docs/pandas.documentation.zip")
+    pandas_docs = requests.get(
+        "https://eto-public.s3.us-west-2.amazonaws.com/datasets/pandas_docs/pandas.documentation.zip"
+    )
     with open(Path("pandas.documentation.zip"), "wb") as f:
         f.write(pandas_docs.content)
 
     file = zipfile.ZipFile(Path("pandas.documentation.zip"))
     file.extractall(path=Path("pandas_docs"))
 
+
 def store_docs():
     docs = []
 
@@ -74,6 +74,7 @@ def store_docs():
 
     return docs
 
+
 def qanda_langchain(query):
     download_docs()
     docs = store_docs()
@@ -85,14 +86,25 @@ def qanda_langchain(query):
     documents = text_splitter.split_documents(docs)
     embeddings = OpenAIEmbeddings()
 
-    db = lancedb.connect(db_path) 
-    table = db.create_table("pandas_docs", data=[
-        {"vector": embeddings.embed_query("Hello World"), "text": "Hello World", "id": "1"}
-    ], mode="overwrite")
+    db = lancedb.connect(db_path)
+    table = db.create_table(
+        "pandas_docs",
+        data=[
+            {
+                "vector": embeddings.embed_query("Hello World"),
+                "text": "Hello World",
+                "id": "1",
+            }
+        ],
+        mode="overwrite",
+    )
     docsearch = LanceDB.from_documents(documents, embeddings, connection=table)
-    qa = RetrievalQA.from_chain_type(llm=OpenAI(), chain_type="stuff", retriever=docsearch.as_retriever())
+    qa = RetrievalQA.from_chain_type(
+        llm=OpenAI(), chain_type="stuff", retriever=docsearch.as_retriever()
+    )
     return qa.run(query)
 
+
 @stub.function()
 @web_endpoint(method="GET")
 def web(query: str):
@@ -101,6 +113,7 @@ def web(query: str):
         "answer": answer,
     }
 
+
 @stub.function()
 def cli(query: str):
     answer = qanda_langchain(query)

docs/src/examples/nodejs.md

@@ -1,99 +0,0 @@
-# YouTube transcript QA bot with NodeJS
-
-## use LanceDB's Javascript API and OpenAI to build a QA bot for YouTube transcripts
-
-<img id="splash" width="400" alt="nodejs" src="https://github.com/lancedb/lancedb/assets/917119/3a140e75-bf8e-438a-a1e4-af14a72bcf98">
-
-This Q&A bot will allow you to search through youtube transcripts using natural language! We'll introduce how you can use LanceDB's Javascript API to store and manage your data easily.
-
-For this example we're using a HuggingFace dataset that contains YouTube transcriptions: `jamescalam/youtube-transcriptions`, to make it easier, we've converted it to a LanceDB `db` already, which you can download and put in a working directory:
-
-```wget -c https://eto-public.s3.us-west-2.amazonaws.com/lancedb_demo.tar.gz -O - | tar -xz -C .```
-
-Now, we'll create a simple app that can:
-1. Take a text based query and search for contexts in our corpus, using embeddings generated from the OpenAI Embedding API.
-2. Create a prompt with the contexts, and call the OpenAI Completion API to answer the text based query.
-
-Dependencies and setup of OpenAI API:
-
-```javascript
-const lancedb = require("vectordb");
-const { Configuration, OpenAIApi } = require("openai");
-
-const configuration = new Configuration({
-    apiKey: process.env.OPENAI_API_KEY,
-    });
-const openai = new OpenAIApi(configuration);
-```
-
-First, let's set our question and the context amount. The context amount will be used to query similar documents in our corpus.
-
-```javascript
-const QUESTION = "who was the 12th person on the moon and when did they land?";
-const CONTEXT_AMOUNT = 3;
-```
-
-Now, let's generate an embedding from this question:
-
-```javascript
-const embeddingResponse = await openai.createEmbedding({
-    model: "text-embedding-ada-002",
-    input: QUESTION,
-});
-
-const embedding = embeddingResponse.data["data"][0]["embedding"];
-```
-
-Once we have the embedding, we can connect to LanceDB (using the database we downloaded earlier), and search through the chatbot table.
-We'll extract 3 similar documents found.
-
-```javascript
-const db = await lancedb.connect('./lancedb');
-const tbl = await db.openTable('chatbot');
-const query = tbl.search(embedding);
-query.limit = CONTEXT_AMOUNT;
-const context = await query.execute();
-```
-
-Let's combine the context together so we can pass it into our prompt:
-
-```javascript
-for (let i = 1; i < context.length; i++) {
-    context[0]["text"] += " " + context[i]["text"];
-}
-```
-
-Lastly, let's construct the prompt. You could play around with this to create more accurate/better prompts to yield results.
-
-```javascript
-const prompt = "Answer the question based on the context below.\n\n" +
-    "Context:\n" +
-    `${context[0]["text"]}\n` +
-    `\n\nQuestion: ${QUESTION}\nAnswer:`;
-```
-
-We pass the prompt, along with the context, to the completion API.
-
-```javascript
-const completion = await openai.createCompletion({
-    model: "text-davinci-003",
-    prompt,
-    temperature: 0,
-    max_tokens: 400,
-    top_p: 1,
-    frequency_penalty: 0,
-    presence_penalty: 0,
-});
-```
-
-And that's it!
-
-```javascript
-console.log(completion.data.choices[0].text);
-```
-
-The response is (which is non deterministic):
-
-```
-The 12th person on the moon was Harrison Schmitt and he landed on December 11, 1972.
-```

docs/src/examples/youtube_transcript_bot_with_nodejs.md

@@ -0,0 +1,139 @@
+# YouTube transcript QA bot with NodeJS
+
+## use LanceDB's Javascript API and OpenAI to build a QA bot for YouTube transcripts
+
+<img id="splash" width="400" alt="nodejs" src="https://github.com/lancedb/lancedb/assets/917119/3a140e75-bf8e-438a-a1e4-af14a72bcf98">
+
+This Q&A bot will allow you to search through youtube transcripts using natural language! We'll introduce how to use LanceDB's Javascript API to store and manage your data easily.
+
+```bash
+npm install vectordb
+```
+
+## Download the data
+
+For this example, we're using a sample of a HuggingFace dataset that contains YouTube transcriptions: `jamescalam/youtube-transcriptions`. Download and extract this file under the `data` folder:
+
+```bash
+wget -c https://eto-public.s3.us-west-2.amazonaws.com/datasets/youtube_transcript/youtube-transcriptions_sample.jsonl
+```
+
+## Prepare Context
+
+Each item in the dataset contains just a short chunk of text. We'll need to merge a bunch of these chunks together on a rolling basis. For this demo, we'll look back 20 records to create a more complete context for each sentence.
+
+First, we need to read and parse the input file.
+
+```javascript
+const lines = (await fs.readFile(INPUT_FILE_NAME, 'utf-8'))
+  .toString()
+  .split('\n')
+  .filter(line => line.length > 0)
+  .map(line => JSON.parse(line))
+
+const data = contextualize(lines, 20, 'video_id')
+```
+
+The contextualize function groups the transcripts by video_id and then creates the expanded context for each item.
+
+```javascript
+function contextualize (rows, contextSize, groupColumn) {
+  const grouped = []
+  rows.forEach(row => {
+    if (!grouped[row[groupColumn]]) {
+      grouped[row[groupColumn]] = []
+    }
+    grouped[row[groupColumn]].push(row)
+  })
+
+  const data = []
+  Object.keys(grouped).forEach(key => {
+    for (let i = 0; i < grouped[key].length; i++) {
+      const start = i - contextSize > 0 ? i - contextSize : 0
+      grouped[key][i].context = grouped[key].slice(start, i + 1).map(r => r.text).join(' ')
+    }
+    data.push(...grouped[key])
+  })
+  return data
+}
+```
+
+## Create the LanceDB Table
+
+To load our data into LanceDB, we need to create embedding (vectors) for each item. For this example, we will use the OpenAI embedding functions, which have a native integration with LanceDB.
+
+```javascript
+// You need to provide an OpenAI API key, here we read it from the OPENAI_API_KEY environment variable
+const apiKey = process.env.OPENAI_API_KEY
+// The embedding function will create embeddings for the 'context' column
+const embedFunction = new lancedb.OpenAIEmbeddingFunction('context', apiKey)
+// Connects to LanceDB
+const db = await lancedb.connect('data/youtube-lancedb')
+const tbl = await db.createTable('vectors', data, embedFunction)
+```
+
+## Create and answer the prompt
+
+We will accept questions in natural language and use our corpus stored in LanceDB to answer them. First, we need to set up the OpenAI client:
+
+```javascript
+const configuration = new Configuration({ apiKey })
+const openai = new OpenAIApi(configuration)
+```
+
+Then we can prompt questions and use LanceDB to retrieve the three most relevant transcripts for this prompt.
+
+```javascript
+const query = await rl.question('Prompt: ')
+const results = await tbl
+  .search(query)
+  .select(['title', 'text', 'context'])
+  .limit(3)
+  .execute()
+```
+
+The query and the transcripts' context are appended together in a single prompt:
+
+```javascript
+function createPrompt (query, context) {
+    let prompt =
+        'Answer the question based on the context below.\n\n' +
+        'Context:\n'
+
+    // need to make sure our prompt is not larger than max size
+    prompt = prompt + context.map(c => c.context).join('\n\n---\n\n').substring(0, 3750)
+    prompt = prompt + `\n\nQuestion: ${query}\nAnswer:`
+    return prompt
+}
+```
+
+We can now use the OpenAI Completion API to process our custom prompt and give us an answer.
+
+```javascript
+const response = await openai.createCompletion({
+  model: 'text-davinci-003',
+  prompt: createPrompt(query, results),
+  max_tokens: 400,
+  temperature: 0,
+  top_p: 1,
+  frequency_penalty: 0,
+  presence_penalty: 0
+})
+console.log(response.data.choices[0].text)
+```
+
+## Let's put it all together now
+
+Now we can provide queries and have them answered based on your local LanceDB data.
+
+```bash
+Prompt: who was the 12th person on the moon and when did they land?
+ The 12th person on the moon was Harrison Schmitt and he landed on December 11, 1972.
+Prompt: Which training method should I use for sentence transformers when I only have pairs of related sentences?
+ NLI with multiple negative ranking loss.
+```
+
+## That's a wrap
+
+In this example, you learned how to use LanceDB to store and query embedding representations of your local data. The complete example code is on [GitHub](https://github.com/lancedb/lancedb/tree/main/node/examples), and you can also download the LanceDB dataset using [this link](https://eto-public.s3.us-west-2.amazonaws.com/datasets/youtube_transcript/youtube-lancedb.zip).
+

docs/src/index.md

@@ -8,11 +8,13 @@ The key features of LanceDB include:
 
 * Store, query and filter vectors, metadata and multi-modal data (text, images, videos, point clouds, and more).
 
+* Support for vector similarity search, full-text search and SQL.
+
 * Native Python and Javascript/Typescript support.
 
 * Zero-copy, automatic versioning, manage versions of your data without needing extra infrastructure.
 
-* Ecosystem integrations with [LangChain 🦜️🔗](https://python.langchain.com/en/latest/modules/indexes/vectorstores/examples/lanecdb.html), [LlamaIndex 🦙](https://gpt-index.readthedocs.io/en/latest/examples/vector_stores/LanceDBIndexDemo.html), Apache-Arrow, Pandas, Polars, DuckDB and more on the way.
+* Ecosystem integrations with [LangChain 🦜️🔗](https://python.langchain.com/en/latest/modules/indexes/vectorstores/examples/lancedb.html), [LlamaIndex 🦙](https://gpt-index.readthedocs.io/en/latest/examples/vector_stores/LanceDBIndexDemo.html), Apache-Arrow, Pandas, Polars, DuckDB and more on the way.
 
 LanceDB's core is written in Rust 🦀 and is built using <a href="https://github.com/lancedb/lance">Lance</a>, an open-source columnar format designed for performant ML workloads.
 
@@ -54,6 +56,11 @@ LanceDB's core is written in Rust 🦀 and is built using <a href="https://githu
 - [YouTube Transcript Search](notebooks/youtube_transcript_search.ipynb)
 - [Documentation QA Bot using LangChain](notebooks/code_qa_bot.ipynb)
 - [Multimodal search using CLIP](notebooks/multimodal_search.ipynb)
+- [Serverless QA Bot with S3 and Lambda](examples/serverless_lancedb_with_s3_and_lambda.md)
+- [Serverless QA Bot with Modal](examples/serverless_qa_bot_with_modal_and_langchain.md)
+
+## Complete Demos (JavaScript)
+- [YouTube Transcript Search](examples/youtube_transcript_bot_with_nodejs.md)
 
 ## Documentation Quick Links
 * [`Basic Operations`](basic.md) - basic functionality of LanceDB.
@@ -61,4 +68,5 @@ LanceDB's core is written in Rust 🦀 and is built using <a href="https://githu
 * [`Indexing`](ann_indexes.md) - create vector indexes to speed up queries.
 * [`Full text search`](fts.md) - [EXPERIMENTAL] full-text search API
 * [`Ecosystem Integrations`](integrations.md) - integrating LanceDB with python data tooling ecosystem.
-* [`API Reference`](python.md) - detailed documentation for the LanceDB Python SDK.
+* [`Python API Reference`](python/python.md) - detailed documentation for the LanceDB Python SDK.
+* [`Node API Reference`](javascript/modules.md) - detailed documentation for the LanceDB Python SDK.

docs/src/integrations.md

@@ -24,9 +24,6 @@ data = pd.DataFrame({
     "price": [10.0, 20.0]
 })
 table = db.create_table("pd_table", data=data)
-
-# Optionally, create a IVF_PQ index
-table.create_index(num_partitions=256, num_sub_vectors=96)
 ```
 
 You will find detailed instructions of creating dataset and index in [Basic Operations](basic.md) and [Indexing](indexing.md)

docs/src/notebooks/diffusiondb/datagen.py

@@ -21,12 +21,13 @@ from argparse import ArgumentParser
 from multiprocessing import Pool
 
 import lance
-import lancedb
 import pyarrow as pa
 from datasets import load_dataset
 from PIL import Image
 from transformers import CLIPModel, CLIPProcessor, CLIPTokenizerFast
 
+import lancedb
+
 MODEL_ID = "openai/clip-vit-base-patch32"
 
 device = "cuda"

docs/src/notebooks/youtube_transcript_search.ipynb

@@ -1,11 +1,12 @@
 {
  "cells": [
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "42bf01fb",
    "metadata": {},
    "source": [
-    "# We're going to build question and answer bot\n",
+    "# Youtube Transcript Search QA Bot\n",
     "\n",
     "This Q&A bot will allow you to search through youtube transcripts using natural language! By going through this notebook, we'll introduce how you can use LanceDB to store and manage your data easily."
    ]
@@ -35,6 +36,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "22e570f4",
    "metadata": {},
@@ -87,6 +89,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "5ac2b6a3",
    "metadata": {},
@@ -181,6 +184,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "3044e0b0",
    "metadata": {},
@@ -209,6 +213,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "db586267",
    "metadata": {},
@@ -229,6 +234,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "2106b5bb",
    "metadata": {},
@@ -338,6 +344,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "53e4bff1",
    "metadata": {},
@@ -371,6 +378,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "8ef34fca",
    "metadata": {},
@@ -459,6 +467,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "23afc2f9",
    "metadata": {},
@@ -541,6 +550,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "28705959",
    "metadata": {},
@@ -571,6 +581,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "559a095b",
    "metadata": {},

docs/src/python/python.md

@@ -6,9 +6,38 @@
 pip install lancedb
 ```
 
-## ::: lancedb
-## ::: lancedb.db
-## ::: lancedb.table
-## ::: lancedb.query
-## ::: lancedb.embeddings
-## ::: lancedb.context
+## Connection
+
+::: lancedb.connect
+
+::: lancedb.LanceDBConnection
+
+## Table
+
+::: lancedb.table.LanceTable
+
+## Querying
+
+::: lancedb.query.LanceQueryBuilder
+
+::: lancedb.query.LanceFtsQueryBuilder
+
+## Embeddings
+
+::: lancedb.embeddings.with_embeddings
+
+::: lancedb.embeddings.EmbeddingFunction
+
+## Context
+
+::: lancedb.context.contextualize
+
+::: lancedb.context.Contextualizer
+
+## Full text search
+
+::: lancedb.fts.create_index
+
+::: lancedb.fts.populate_index
+
+::: lancedb.fts.search_index

node/.npmignore

@@ -0,0 +1,4 @@
+gen_test_data.py
+index.node
+dist/lancedb*.tgz
+vectordb*.tgz

node/CHANGELOG.md

@@ -5,6 +5,12 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.1.5] - 2023-06-00
+
+### Added
+
+- Support for macOS X86
+
 ## [0.1.4] - 2023-06-03
 
 ### Added

node/README.md

@@ -8,6 +8,10 @@ A JavaScript / Node.js library for [LanceDB](https://github.com/lancedb/lancedb)
 npm install vectordb
 ```
 
+This will download the appropriate native library for your platform. We currently
+support x86_64 Linux, aarch64 Linux, Intel MacOS, and ARM (M1/M2) MacOS. We do not
+yet support Windows or musl-based Linux (such as Alpine Linux).
+
 ## Usage
 
 ### Basic Example
@@ -24,17 +28,33 @@ The [examples](./examples) folder contains complete examples.
 
 ## Development
 
-The LanceDB javascript is built with npm:
+To build everything fresh:
+
+```bash
+npm install
+npm run tsc
+npm run build
+```
+
+Then you should be able to run the tests with:
+
+```bash
+npm test
+```
+
+### Rebuilding Rust library
+
+```bash
+npm run build
+```
+
+### Rebuilding Typescript
 
 ```bash
 npm run tsc
 ```
 
-Run the tests with
-
-```bash
-npm test
-```
+### Fix lints
 
 To run the linter and have it automatically fix all errors
 

node/examples/js-youtube-transcripts/index.js

@@ -0,0 +1,122 @@
+// Copyright 2023 Lance Developers.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+'use strict'
+
+const lancedb = require('vectordb')
+const fs = require('fs/promises')
+const readline = require('readline/promises')
+const { stdin: input, stdout: output } = require('process')
+const { Configuration, OpenAIApi } = require('openai')
+
+// Download file from XYZ
+const INPUT_FILE_NAME = 'data/youtube-transcriptions_sample.jsonl';
+
+(async () => {
+  // You need to provide an OpenAI API key, here we read it from the OPENAI_API_KEY environment variable
+  const apiKey = process.env.OPENAI_API_KEY
+  // The embedding function will create embeddings for the 'context' column
+  const embedFunction = new lancedb.OpenAIEmbeddingFunction('context', apiKey)
+
+  // Connects to LanceDB
+  const db = await lancedb.connect('data/youtube-lancedb')
+
+  // Open the vectors table or create one if it does not exist
+  let tbl
+  if ((await db.tableNames()).includes('vectors')) {
+    tbl = await db.openTable('vectors', embedFunction)
+  } else {
+    tbl = await createEmbeddingsTable(db, embedFunction)
+  }
+
+  // Use OpenAI Completion API to generate and answer based on the context that LanceDB provides
+  const configuration = new Configuration({ apiKey })
+  const openai = new OpenAIApi(configuration)
+  const rl = readline.createInterface({ input, output })
+  try {
+    while (true) {
+      const query = await rl.question('Prompt: ')
+      const results = await tbl
+        .search(query)
+        .select(['title', 'text', 'context'])
+        .limit(3)
+        .execute()
+
+      // console.table(results)
+
+      const response = await openai.createCompletion({
+        model: 'text-davinci-003',
+        prompt: createPrompt(query, results),
+        max_tokens: 400,
+        temperature: 0,
+        top_p: 1,
+        frequency_penalty: 0,
+        presence_penalty: 0
+      })
+      console.log(response.data.choices[0].text)
+    }
+  } catch (err) {
+    console.log('Error: ', err)
+  } finally {
+    rl.close()
+  }
+  process.exit(1)
+})()
+
+async function createEmbeddingsTable (db, embedFunction) {
+  console.log(`Creating embeddings from ${INPUT_FILE_NAME}`)
+  // read the input file into a JSON array, skipping empty lines
+  const lines = (await fs.readFile(INPUT_FILE_NAME, 'utf-8'))
+    .toString()
+    .split('\n')
+    .filter(line => line.length > 0)
+    .map(line => JSON.parse(line))
+
+  const data = contextualize(lines, 20, 'video_id')
+  return await db.createTable('vectors', data, embedFunction)
+}
+
+// Each transcript has a small text column, we include previous transcripts in order to
+// have more context information when creating embeddings
+function contextualize (rows, contextSize, groupColumn) {
+  const grouped = []
+  rows.forEach(row => {
+    if (!grouped[row[groupColumn]]) {
+      grouped[row[groupColumn]] = []
+    }
+    grouped[row[groupColumn]].push(row)
+  })
+
+  const data = []
+  Object.keys(grouped).forEach(key => {
+    for (let i = 0; i < grouped[key].length; i++) {
+      const start = i - contextSize > 0 ? i - contextSize : 0
+      grouped[key][i].context = grouped[key].slice(start, i + 1).map(r => r.text).join(' ')
+    }
+    data.push(...grouped[key])
+  })
+  return data
+}
+
+// Creates a prompt by aggregating all relevant contexts
+function createPrompt (query, context) {
+  let prompt =
+      'Answer the question based on the context below.\n\n' +
+      'Context:\n'
+
+  // need to make sure our prompt is not larger than max size
+  prompt = prompt + context.map(c => c.context).join('\n\n---\n\n').substring(0, 3750)
+  prompt = prompt + `\n\nQuestion: ${query}\nAnswer:`
+  return prompt
+}

node/examples/js-youtube-transcripts/package.json

@@ -0,0 +1,15 @@
+{
+  "name": "vectordb-example-js-openai",
+  "version": "1.0.0",
+  "description": "",
+  "main": "index.js",
+  "scripts": {
+    "test": "echo \"Error: no test specified\" && exit 1"
+  },
+  "author": "Lance Devs",
+  "license": "Apache-2.0",
+  "dependencies": {
+    "vectordb": "file:../..",
+    "openai": "^3.2.1"
+  }
+}

node/native.js

@@ -12,29 +12,26 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+const { currentTarget } = require('@neon-rs/load');
+
 let nativeLib;
 
-function getPlatformLibrary() {
-    if (process.platform === "darwin" && process.arch == "arm64") {
-        return require('./aarch64-apple-darwin.node');
-    } else if (process.platform === "darwin" && process.arch == "x64") {
-        return require('./x86_64-apple-darwin.node');
-    } else if (process.platform === "linux" && process.arch == "x64") {
-        return require('./x86_64-unknown-linux-gnu.node');
-    } else {
-        throw new Error(`vectordb: unsupported platform ${process.platform}_${process.arch}. Please file a bug report at https://github.com/lancedb/lancedb/issues`)
-    }
-}
-
 try {
-    nativeLib = require('./index.node')
+    nativeLib = require(`vectordb-${currentTarget()}`);
 } catch (e) {
-    if (e.code === "MODULE_NOT_FOUND") {
-        nativeLib = getPlatformLibrary();
-    } else {
-        throw new Error('vectordb: failed to load native library. Please file a bug report at https://github.com/lancedb/lancedb/issues');
-    }
+    try {
+        // Might be developing locally, so try that. But don't expose that error
+        // to the user.
+        nativeLib = require("./index.node");
+    } catch {
+        throw new Error(`vectordb: failed to load native library.
+  You may need to run \`npm install vectordb-${currentTarget()}\`.
+      
+  If that does not work, please file a bug report at https://github.com/lancedb/lancedb/issues
+      
+  Source error: ${e}`);
+    }   
 }
 
-module.exports = nativeLib
-
+// Dynamic require for runtime.
+module.exports = nativeLib;

node/package-lock.json

@@ -1,18 +1,28 @@
 {
   "name": "vectordb",
-  "version": "0.1.4",
+  "version": "0.1.5",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "vectordb",
-      "version": "0.1.4",
+      "version": "0.1.5",
+      "cpu": [
+        "x64",
+        "arm64"
+      ],
       "license": "Apache-2.0",
+      "os": [
+        "darwin",
+        "linux"
+      ],
       "dependencies": {
         "@apache-arrow/ts": "^12.0.0",
+        "@neon-rs/load": "^0.0.74",
         "apache-arrow": "^12.0.0"
       },
       "devDependencies": {
+        "@neon-rs/cli": "^0.0.74",
         "@types/chai": "^4.3.4",
         "@types/mocha": "^10.0.1",
         "@types/node": "^18.16.2",
@@ -35,6 +45,12 @@
         "typedoc": "^0.24.7",
         "typedoc-plugin-markdown": "^3.15.3",
         "typescript": "*"
+      },
+      "optionalDependencies": {
+        "vectordb-darwin-arm64": "0.1.2",
+        "vectordb-darwin-x64": "0.1.2",
+        "vectordb-linux-arm64-gnu": "0.1.2",
+        "vectordb-linux-x64-gnu": "0.1.2"
       }
     },
     "node_modules/@apache-arrow/ts": {
@@ -202,6 +218,20 @@
         "@jridgewell/sourcemap-codec": "^1.4.10"
       }
     },
+    "node_modules/@neon-rs/cli": {
+      "version": "0.0.74",
+      "resolved": "https://registry.npmjs.org/@neon-rs/cli/-/cli-0.0.74.tgz",
+      "integrity": "sha512-9lPmNmjej5iKKOTMPryOMubwkgMRyTWRuaq1yokASvI5mPhr2kzPN7UVjdCOjQvpunNPngR9yAHoirpjiWhUHw==",
+      "dev": true,
+      "bin": {
+        "neon": "index.js"
+      }
+    },
+    "node_modules/@neon-rs/load": {
+      "version": "0.0.74",
+      "resolved": "https://registry.npmjs.org/@neon-rs/load/-/load-0.0.74.tgz",
+      "integrity": "sha512-/cPZD907UNz55yrc/ud4wDgQKtU1TvkD9jeqZWG6J4IMmZkp6zgjkQcKA8UvpkZlcpPHvc8J17sGzLFbP/LUYg=="
+    },
     "node_modules/@nodelib/fs.scandir": {
       "version": "2.1.5",
       "resolved": "https://registry.npmjs.org/@nodelib/fs.scandir/-/fs.scandir-2.1.5.tgz",
@@ -4601,6 +4631,17 @@
         "@jridgewell/sourcemap-codec": "^1.4.10"
       }
     },
+    "@neon-rs/cli": {
+      "version": "0.0.74",
+      "resolved": "https://registry.npmjs.org/@neon-rs/cli/-/cli-0.0.74.tgz",
+      "integrity": "sha512-9lPmNmjej5iKKOTMPryOMubwkgMRyTWRuaq1yokASvI5mPhr2kzPN7UVjdCOjQvpunNPngR9yAHoirpjiWhUHw==",
+      "dev": true
+    },
+    "@neon-rs/load": {
+      "version": "0.0.74",
+      "resolved": "https://registry.npmjs.org/@neon-rs/load/-/load-0.0.74.tgz",
+      "integrity": "sha512-/cPZD907UNz55yrc/ud4wDgQKtU1TvkD9jeqZWG6J4IMmZkp6zgjkQcKA8UvpkZlcpPHvc8J17sGzLFbP/LUYg=="
+    },
     "@nodelib/fs.scandir": {
       "version": "2.1.5",
       "resolved": "https://registry.npmjs.org/@nodelib/fs.scandir/-/fs.scandir-2.1.5.tgz",

node/package.json

@@ -1,16 +1,19 @@
 {
   "name": "vectordb",
-  "version": "0.1.4",
+  "version": "0.1.8",
   "description": " Serverless, low-latency vector database for AI applications",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
   "scripts": {
     "tsc": "tsc -b",
-    "build": "cargo-cp-artifact --artifact cdylib vectordb-node index.node -- cargo build --message-format=json-render-diagnostics",
+    "build": "cargo-cp-artifact --artifact cdylib vectordb-node index.node -- cargo build --message-format=json",
     "build-release": "npm run build -- --release",
+    "cross-release": "cargo-cp-artifact --artifact cdylib vectordb-node index.node -- cross build --message-format=json --release -p vectordb-node",
     "test": "mocha -recursive dist/test",
     "lint": "eslint src --ext .js,.ts",
-    "clean": "rm -rf node_modules *.node dist/"
+    "clean": "rm -rf node_modules *.node dist/",
+    "pack-build": "neon pack-build",
+    "check-npm": "printenv && which node && which npm && npm --version"
   },
   "repository": {
     "type": "git",
@@ -25,6 +28,7 @@
   "author": "Lance Devs",
   "license": "Apache-2.0",
   "devDependencies": {
+    "@neon-rs/cli": "^0.0.74",
     "@types/chai": "^4.3.4",
     "@types/mocha": "^10.0.1",
     "@types/node": "^18.16.2",
@@ -50,6 +54,29 @@
   },
   "dependencies": {
     "@apache-arrow/ts": "^12.0.0",
+    "@neon-rs/load": "^0.0.74",
     "apache-arrow": "^12.0.0"
+  },
+  "os": [
+    "darwin",
+    "linux"
+  ],
+  "cpu": [
+    "x64",
+    "arm64"
+  ],
+  "neon": {
+    "targets": {
+      "x86_64-apple-darwin": "vectordb-darwin-x64",
+      "aarch64-apple-darwin": "vectordb-darwin-arm64",
+      "x86_64-unknown-linux-gnu": "vectordb-linux-x64-gnu",
+      "aarch64-unknown-linux-gnu": "vectordb-linux-arm64-gnu"
+    }
+  },
+  "optionalDependencies": {
+    "vectordb-darwin-arm64": "0.1.8",
+    "vectordb-darwin-x64": "0.1.8",
+    "vectordb-linux-x64-gnu": "0.1.8",
+    "vectordb-linux-arm64-gnu": "0.1.8"
   }
 }

node/src/index.ts

@@ -22,7 +22,7 @@ import { fromRecordsToBuffer } from './arrow'
 import type { EmbeddingFunction } from './embedding/embedding_function'
 
 // eslint-disable-next-line @typescript-eslint/no-var-requires
-const { databaseNew, databaseTableNames, databaseOpenTable, tableCreate, tableSearch, tableAdd, tableCreateVectorIndex } = require('../native.js')
+const { databaseNew, databaseTableNames, databaseOpenTable, tableCreate, tableSearch, tableAdd, tableCreateVectorIndex, tableCountRows } = require('../native.js')
 
 export type { EmbeddingFunction }
 export { OpenAIEmbeddingFunction } from './embedding/openai'
@@ -178,6 +178,13 @@ export class Table<T = number[]> {
   async create_index (indexParams: VectorIndexParams): Promise<any> {
     return await this.createIndex(indexParams)
   }
+
+  /**
+   * Returns the number of rows in this table.
+   */
+  async countRows (): Promise<number> {
+    return tableCountRows.call(this._tbl)
+  }
 }
 
 interface IvfPQIndexConfig {
@@ -293,6 +300,8 @@ export class Query<T = number[]> {
     return this
   }
 
+  where = this.filter
+
   /** Return only the specified columns.
    *
    * @param value Only select the specified columns. If not specified, all columns will be returned.

node/src/test/test.ts

@@ -64,13 +64,20 @@ describe('LanceDB client', function () {
       assert.equal(results[0].id, 1)
     })
 
-    it('uses a filter', async function () {
+    it('uses a filter / where clause', async function () {
+      // eslint-disable-next-line @typescript-eslint/explicit-function-return-type
+      const assertResults = (results: Array<Record<string, unknown>>) => {
+        assert.equal(results.length, 1)
+        assert.equal(results[0].id, 2)
+      }
+
       const uri = await createTestDB()
       const con = await lancedb.connect(uri)
       const table = await con.openTable('vectors')
-      const results = await table.search([0.1, 0.1]).filter('id == 2').execute()
-      assert.equal(results.length, 1)
-      assert.equal(results[0].id, 2)
+      let results = await table.search([0.1, 0.1]).filter('id == 2').execute()
+      assertResults(results)
+      results = await table.search([0.1, 0.1]).where('id == 2').execute()
+      assertResults(results)
     })
 
     it('select only a subset of columns', async function () {
@@ -103,9 +110,7 @@ describe('LanceDB client', function () {
       const tableName = `vectors_${Math.floor(Math.random() * 100)}`
       const table = await con.createTable(tableName, data)
       assert.equal(table.name, tableName)
-
-      const results = await table.search([0.1, 0.3]).execute()
-      assert.equal(results.length, 2)
+      assert.equal(await table.countRows(), 2)
     })
 
     it('appends records to an existing table ', async function () {
@@ -118,16 +123,14 @@ describe('LanceDB client', function () {
       ]
 
       const table = await con.createTable('vectors', data)
-      const results = await table.search([0.1, 0.3]).execute()
-      assert.equal(results.length, 2)
+      assert.equal(await table.countRows(), 2)
 
       const dataAdd = [
         { id: 3, vector: [2.1, 2.2], price: 10, name: 'c' },
         { id: 4, vector: [3.1, 3.2], price: 50, name: 'd' }
       ]
       await table.add(dataAdd)
-      const resultsAdd = await table.search([0.1, 0.3]).execute()
-      assert.equal(resultsAdd.length, 4)
+      assert.equal(await table.countRows(), 4)
     })
 
     it('overwrite all records in a table', async function () {
@@ -135,16 +138,14 @@ describe('LanceDB client', function () {
       const con = await lancedb.connect(uri)
 
       const table = await con.openTable('vectors')
-      const results = await table.search([0.1, 0.3]).execute()
-      assert.equal(results.length, 2)
+      assert.equal(await table.countRows(), 2)
 
       const dataOver = [
         { vector: [2.1, 2.2], price: 10, name: 'foo' },
         { vector: [3.1, 3.2], price: 50, name: 'bar' }
       ]
       await table.overwrite(dataOver)
-      const resultsAdd = await table.search([0.1, 0.3]).execute()
-      assert.equal(resultsAdd.length, 2)
+      assert.equal(await table.countRows(), 2)
     })
   })
 

python/.bumpversion.cfg

@@ -0,0 +1,8 @@
+[bumpversion]
+current_version = 0.1.8
+commit = True
+message = [python] Bump version: {current_version} → {new_version}
+tag = True
+tag_name = python-v{new_version}
+
+[bumpversion:file:pyproject.toml]

python/lancedb/__init__.py

@@ -22,8 +22,21 @@ def connect(uri: URI) -> LanceDBConnection:
     uri: str or Path
         The uri of the database.
 
+    Examples
+    --------
+
+    For a local directory, provide a path for the database:
+
+    >>> import lancedb
+    >>> db = lancedb.connect("~/.lancedb")
+
+    For object storage, use a URI prefix:
+
+    >>> db = lancedb.connect("s3://my-bucket/lancedb")
+
     Returns
     -------
-    A connection to a LanceDB database.
+    conn : LanceDBConnection
+        A connection to a LanceDB database.
     """
     return LanceDBConnection(uri)

python/lancedb/conftest.py

@@ -0,0 +1,18 @@
+import builtins
+import os
+
+import pytest
+
+# import lancedb so we don't have to in every example
+import lancedb
+
+
+@pytest.fixture(autouse=True)
+def doctest_setup(monkeypatch, tmpdir):
+    # disable color for doctests so we don't have to include
+    # escape codes in docstrings
+    monkeypatch.setitem(os.environ, "NO_COLOR", "1")
+    # Explicitly set the column width
+    monkeypatch.setitem(os.environ, "COLUMNS", "80")
+    # Work in a temporary directory
+    monkeypatch.chdir(tmpdir)

python/lancedb/context.py

@@ -14,20 +14,109 @@ from __future__ import annotations
 
 import pandas as pd
 
+from .exceptions import MissingColumnError, MissingValueError
+
 
 def contextualize(raw_df: pd.DataFrame) -> Contextualizer:
     """Create a Contextualizer object for the given DataFrame.
-    Used to create context windows.
+
+    Used to create context windows. Context windows are rolling subsets of text
+    data.
+
+    The input text column should already be separated into rows that will be the
+    unit of the window. So to create a context window over tokens, start with
+    a DataFrame with one token per row. To create a context window over sentences,
+    start with a DataFrame with one sentence per row.
+
+    Examples
+    --------
+    >>> from lancedb.context import contextualize
+    >>> import pandas as pd
+    >>> data = pd.DataFrame({
+    ...    'token': ['The', 'quick', 'brown', 'fox', 'jumped', 'over',
+    ...              'the', 'lazy', 'dog', 'I', 'love', 'sandwiches'],
+    ...    'document_id': [1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2]
+    ... })
+
+    ``window`` determines how many rows to include in each window. In our case
+    this how many tokens, but depending on the input data, it could be sentences,
+    paragraphs, messages, etc.
+
+    >>> contextualize(data).window(3).stride(1).text_col('token').to_df()
+                    token  document_id
+    0     The quick brown            1
+    1     quick brown fox            1
+    2    brown fox jumped            1
+    3     fox jumped over            1
+    4     jumped over the            1
+    5       over the lazy            1
+    6        the lazy dog            1
+    7          lazy dog I            1
+    8          dog I love            1
+    9   I love sandwiches            2
+    10    love sandwiches            2
+    >>> contextualize(data).window(7).stride(1).min_window_size(7).text_col('token').to_df()
+                                      token  document_id
+    0   The quick brown fox jumped over the            1
+    1  quick brown fox jumped over the lazy            1
+    2    brown fox jumped over the lazy dog            1
+    3        fox jumped over the lazy dog I            1
+    4       jumped over the lazy dog I love            1
+    5   over the lazy dog I love sandwiches            1
+
+    ``stride`` determines how many rows to skip between each window start. This can
+    be used to reduce the total number of windows generated.
+
+    >>> contextualize(data).window(4).stride(2).text_col('token').to_df()
+                        token  document_id
+    0     The quick brown fox            1
+    2   brown fox jumped over            1
+    4    jumped over the lazy            1
+    6          the lazy dog I            1
+    8   dog I love sandwiches            1
+    10        love sandwiches            2
+
+    ``groupby`` determines how to group the rows. For example, we would like to have
+    context windows that don't cross document boundaries. In this case, we can
+    pass ``document_id`` as the group by.
+
+    >>> contextualize(data).window(4).stride(2).text_col('token').groupby('document_id').to_df()
+                       token  document_id
+    0    The quick brown fox            1
+    2  brown fox jumped over            1
+    4   jumped over the lazy            1
+    6           the lazy dog            1
+    9      I love sandwiches            2
+
+    ``min_window_size`` determines the minimum size of the  context windows that are generated
+    This can be used to trim the last few context windows which have size less than
+    ``min_window_size``. By default context windows of size 1 are skipped.
+
+    >>> contextualize(data).window(6).stride(3).text_col('token').groupby('document_id').to_df()
+                                 token  document_id
+    0  The quick brown fox jumped over            1
+    3     fox jumped over the lazy dog            1
+    6                     the lazy dog            1
+    9                I love sandwiches            2
+
+    >>> contextualize(data).window(6).stride(3).min_window_size(4).text_col('token').groupby('document_id').to_df()
+                                 token  document_id
+    0  The quick brown fox jumped over            1
+    3     fox jumped over the lazy dog            1
+
     """
     return Contextualizer(raw_df)
 
 
 class Contextualizer:
+    """Create context windows from a DataFrame. See [lancedb.context.contextualize][]."""
+
     def __init__(self, raw_df):
         self._text_col = None
         self._groupby = None
         self._stride = None
         self._window = None
+        self._min_window_size = 2
         self._raw_df = raw_df
 
     def window(self, window: int) -> Contextualizer:
@@ -75,17 +164,50 @@ class Contextualizer:
         self._text_col = text_col
         return self
 
+    def min_window_size(self, min_window_size: int) -> Contextualizer:
+        """Set the (optional) min_window_size size for the context window.
+
+        Parameters
+        ----------
+        min_window_size: int
+            The min_window_size.
+        """
+        self._min_window_size = min_window_size
+        return self
+
     def to_df(self) -> pd.DataFrame:
         """Create the context windows and return a DataFrame."""
 
+        if self._text_col not in self._raw_df.columns.tolist():
+            raise MissingColumnError(self._text_col)
+
+        if self._window is None or self._window < 1:
+            raise MissingValueError(
+                "The value of window is None or less than 1. Specify the "
+                "window size (number of rows to include in each window)"
+            )
+
+        if self._stride is None or self._stride < 1:
+            raise MissingValueError(
+                "The value of stride is None or less than 1. Specify the "
+                "stride (number of rows to skip between each window)"
+            )
+
         def process_group(grp):
             # For each group, create the text rolling window
+            # with values of size >= min_window_size
             text = grp[self._text_col].values
-            contexts = grp.iloc[: -self._window : self._stride, :].copy()
-            contexts[self._text_col] = [
-                " ".join(text[start_i : start_i + self._window])
-                for start_i in range(0, len(grp) - self._window, self._stride)
+            contexts = grp.iloc[:: self._stride, :].copy()
+            windows = [
+                " ".join(text[start_i : min(start_i + self._window, len(grp))])
+                for start_i in range(0, len(grp), self._stride)
+                if start_i + self._window <= len(grp)
+                or len(grp) - start_i >= self._min_window_size
             ]
+            # if last few rows dropped
+            if len(windows) < len(contexts):
+                contexts = contexts.iloc[: len(windows)]
+            contexts[self._text_col] = windows
             return contexts
 
         if self._groupby is None:

python/lancedb/db.py

@@ -13,25 +13,59 @@
 
 from __future__ import annotations
 
+import functools
 import os
 from pathlib import Path
-import os
 
 import pyarrow as pa
 from pyarrow import fs
 
 from .common import DATA, URI
 from .table import LanceTable
-from .util import get_uri_scheme, get_uri_location
+from .util import get_uri_location, get_uri_scheme
 
 
 class LanceDBConnection:
     """
     A connection to a LanceDB database.
+
+    Parameters
+    ----------
+    uri: str or Path
+        The root uri of the database.
+
+    Examples
+    --------
+    >>> import lancedb
+    >>> db = lancedb.connect("./.lancedb")
+    >>> db.create_table("my_table", data=[{"vector": [1.1, 1.2], "b": 2},
+    ...                                   {"vector": [0.5, 1.3], "b": 4}])
+    LanceTable(my_table)
+    >>> db.create_table("another_table", data=[{"vector": [0.4, 0.4], "b": 6}])
+    LanceTable(another_table)
+    >>> db.table_names()
+    ['another_table', 'my_table']
+    >>> len(db)
+    2
+    >>> db["my_table"]
+    LanceTable(my_table)
+    >>> "my_table" in db
+    True
+    >>> db.drop_table("my_table")
+    >>> db.drop_table("another_table")
     """
 
     def __init__(self, uri: URI):
-        is_local = isinstance(uri, Path) or get_uri_scheme(uri) == "file"
+        if not isinstance(uri, Path):
+            scheme = get_uri_scheme(uri)
+        is_local = isinstance(uri, Path) or scheme == "file"
+        # managed lancedb remote uses schema like lancedb+[http|grpc|...]://
+        self._is_managed_remote = not is_local and scheme.startswith("lancedb")
+        if self._is_managed_remote:
+            if len(scheme.split("+")) != 2:
+                raise ValueError(
+                    f"Invalid LanceDB URI: {uri}, expected uri to have scheme like lancedb+<flavor>://..."
+                )
         if is_local:
             if isinstance(uri, str):
                 uri = Path(uri)
@@ -39,30 +73,74 @@ class LanceDBConnection:
             Path(uri).mkdir(parents=True, exist_ok=True)
         self._uri = str(uri)
 
+        self._entered = False
+
     @property
     def uri(self) -> str:
         return self._uri
 
+    @functools.cached_property
+    def is_managed_remote(self) -> bool:
+        return self._is_managed_remote
+
+    @functools.cached_property
+    def remote_flavor(self) -> str:
+        if not self.is_managed_remote:
+            raise ValueError(
+                "Not a managed remote LanceDB, there should be no server flavor"
+            )
+        return get_uri_scheme(self.uri).split("+")[1]
+
+    @functools.cached_property
+    def _client(self) -> "lancedb.remote.LanceDBClient":
+        if not self.is_managed_remote:
+            raise ValueError("Not a managed remote LanceDB, there should be no client")
+
+        # don't import unless we are really using remote
+        from lancedb.remote.client import RestfulLanceDBClient
+
+        if self.remote_flavor == "http":
+            return RestfulLanceDBClient(self._uri)
+
+        raise ValueError("Unsupported remote flavor: " + self.remote_flavor)
+
+    async def close(self):
+        if self._entered:
+            raise ValueError("Cannot re-enter the same LanceDBConnection twice")
+        self._entered = True
+        await self._client.close()
+
+    async def __aenter__(self) -> LanceDBConnection:
+        return self
+
+    async def __aexit__(self, exc_type, exc_value, traceback):
+        await self.close()
+
     def table_names(self) -> list[str]:
         """Get the names of all tables in the database.
 
         Returns
         -------
-        A list of table names.
+        list of str
+            A list of table names.
         """
         try:
             filesystem, path = fs.FileSystem.from_uri(self.uri)
         except pa.ArrowInvalid:
-            raise NotImplementedError(
-                "Unsupported scheme: " + self.uri
-            )
+            raise NotImplementedError("Unsupported scheme: " + self.uri)
 
         try:
-            paths = filesystem.get_file_info(fs.FileSelector(get_uri_location(self.uri)))
+            paths = filesystem.get_file_info(
+                fs.FileSelector(get_uri_location(self.uri))
+            )
         except FileNotFoundError:
             # It is ok if the file does not exist since it will be created
             paths = []
-        tables = [os.path.splitext(file_info.base_name)[0] for file_info in paths if file_info.extension == 'lance']
+        tables = [
+            os.path.splitext(file_info.base_name)[0]
+            for file_info in paths
+            if file_info.extension == "lance"
+        ]
         return tables
 
     def __len__(self) -> int:
@@ -103,7 +181,73 @@ class LanceDBConnection:
 
         Returns
         -------
-        A LanceTable object representing the table.
+        LanceTable
+            A reference to the newly created table.
+
+        Examples
+        --------
+
+        Can create with list of tuples or dictionaries:
+
+        >>> import lancedb
+        >>> db = lancedb.connect("./.lancedb")
+        >>> data = [{"vector": [1.1, 1.2], "lat": 45.5, "long": -122.7},
+        ...         {"vector": [0.2, 1.8], "lat": 40.1, "long":  -74.1}]
+        >>> db.create_table("my_table", data)
+        LanceTable(my_table)
+        >>> db["my_table"].head()
+        pyarrow.Table
+        vector: fixed_size_list<item: float>[2]
+          child 0, item: float
+        lat: double
+        long: double
+        ----
+        vector: [[[1.1,1.2],[0.2,1.8]]]
+        lat: [[45.5,40.1]]
+        long: [[-122.7,-74.1]]
+
+        You can also pass a pandas DataFrame:
+
+        >>> import pandas as pd
+        >>> data = pd.DataFrame({
+        ...    "vector": [[1.1, 1.2], [0.2, 1.8]],
+        ...    "lat": [45.5, 40.1],
+        ...    "long": [-122.7, -74.1]
+        ... })
+        >>> db.create_table("table2", data)
+        LanceTable(table2)
+        >>> db["table2"].head()
+        pyarrow.Table
+        vector: fixed_size_list<item: float>[2]
+          child 0, item: float
+        lat: double
+        long: double
+        ----
+        vector: [[[1.1,1.2],[0.2,1.8]]]
+        lat: [[45.5,40.1]]
+        long: [[-122.7,-74.1]]
+
+        Data is converted to Arrow before being written to disk. For maximum
+        control over how data is saved, either provide the PyArrow schema to
+        convert to or else provide a PyArrow table directly.
+
+        >>> custom_schema = pa.schema([
+        ...   pa.field("vector", pa.list_(pa.float32(), 2)),
+        ...   pa.field("lat", pa.float32()),
+        ...   pa.field("long", pa.float32())
+        ... ])
+        >>> db.create_table("table3", data, schema = custom_schema)
+        LanceTable(table3)
+        >>> db["table3"].head()
+        pyarrow.Table
+        vector: fixed_size_list<item: float>[2]
+          child 0, item: float
+        lat: float
+        long: float
+        ----
+        vector: [[[1.1,1.2],[0.2,1.8]]]
+        lat: [[45.5,40.1]]
+        long: [[-122.7,-74.1]]
         """
         if data is not None:
             tbl = LanceTable.create(self, name, data, schema, mode=mode)

python/lancedb/embeddings.py

@@ -29,7 +29,31 @@ def with_embeddings(
     wrap_api: bool = True,
     show_progress: bool = False,
     batch_size: int = 1000,
-):
+) -> pa.Table:
+    """Add a vector column to a table using the given embedding function.
+
+    The new columns will be called "vector".
+
+    Parameters
+    ----------
+    func : Callable
+        A function that takes a list of strings and returns a list of vectors.
+    data : pa.Table or pd.DataFrame
+        The data to add an embedding column to.
+    column : str, default "text"
+        The name of the column to use as input to the embedding function.
+    wrap_api : bool, default True
+        Whether to wrap the embedding function in a retry and rate limiter.
+    show_progress : bool, default False
+        Whether to show a progress bar.
+    batch_size : int, default 1000
+        The number of row values to pass to each call of the embedding function.
+
+    Returns
+    -------
+    pa.Table
+        The input table with a new column called "vector" containing the embeddings.
+    """
     func = EmbeddingFunction(func)
     if wrap_api:
         func = func.retry().rate_limit()

python/lancedb/exceptions.py

@@ -0,0 +1,22 @@
+"""Custom exception handling"""
+
+
+class MissingValueError(ValueError):
+    """Exception raised when a required value is missing."""
+
+    pass
+
+
+class MissingColumnError(KeyError):
+    """
+    Exception raised when a column name specified is not in
+    the  DataFrame object
+    """
+
+    def __init__(self, column_name):
+        self.column_name = column_name
+
+    def __str__(self):
+        return (
+            f"Error: Column '{self.column_name}' does not exist in the DataFrame object"
+        )

python/lancedb/fts.py

@@ -68,6 +68,11 @@ def populate_index(index: tantivy.Index, table: LanceTable, fields: List[str]) -
         The table to index
     fields : List[str]
         List of fields to index
+
+    Returns
+    -------
+    int
+        The number of rows indexed
     """
     # first check the fields exist and are string or large string type
     for name in fields:

python/lancedb/query.py

@@ -12,6 +12,9 @@
 #  limitations under the License.
 from __future__ import annotations
 
+import asyncio
+from typing import Awaitable, Literal
+
 import numpy as np
 import pandas as pd
 import pyarrow as pa
@@ -22,6 +25,24 @@ from .common import VECTOR_COLUMN_NAME
 class LanceQueryBuilder:
     """
     A builder for nearest neighbor queries for LanceDB.
+
+    Examples
+    --------
+    >>> import lancedb
+    >>> data = [{"vector": [1.1, 1.2], "b": 2},
+    ...         {"vector": [0.5, 1.3], "b": 4},
+    ...         {"vector": [0.4, 0.4], "b": 6},
+    ...         {"vector": [0.4, 0.4], "b": 10}]
+    >>> db = lancedb.connect("./.lancedb")
+    >>> table = db.create_table("my_table", data=data)
+    >>> (table.search([0.4, 0.4])
+    ...       .metric("cosine")
+    ...       .where("b < 10")
+    ...       .select(["b"])
+    ...       .limit(2)
+    ...       .to_df())
+       b      vector  score
+    0  6  [0.4, 0.4]    0.0
     """
 
     def __init__(self, table: "lancedb.table.LanceTable", query: np.ndarray):
@@ -44,7 +65,8 @@ class LanceQueryBuilder:
 
         Returns
         -------
-        The LanceQueryBuilder object.
+        LanceQueryBuilder
+            The LanceQueryBuilder object.
         """
         self._limit = limit
         return self
@@ -59,7 +81,8 @@ class LanceQueryBuilder:
 
         Returns
         -------
-        The LanceQueryBuilder object.
+        LanceQueryBuilder
+            The LanceQueryBuilder object.
         """
         self._columns = columns
         return self
@@ -74,22 +97,24 @@ class LanceQueryBuilder:
 
         Returns
         -------
-        The LanceQueryBuilder object.
+        LanceQueryBuilder
+            The LanceQueryBuilder object.
         """
         self._where = where
         return self
 
-    def metric(self, metric: str) -> LanceQueryBuilder:
+    def metric(self, metric: Literal["L2", "cosine"]) -> LanceQueryBuilder:
         """Set the distance metric to use.
 
         Parameters
         ----------
-        metric: str
-            The distance metric to use. By default "l2" is used.
+        metric: "L2" or "cosine"
+            The distance metric to use. By default "L2" is used.
 
         Returns
         -------
-        The LanceQueryBuilder object.
+        LanceQueryBuilder
+            The LanceQueryBuilder object.
         """
         self._metric = metric
         return self
@@ -97,6 +122,12 @@ class LanceQueryBuilder:
     def nprobes(self, nprobes: int) -> LanceQueryBuilder:
         """Set the number of probes to use.
 
+        Higher values will yield better recall (more likely to find vectors if
+        they exist) at the expense of latency.
+
+        See discussion in [Querying an ANN Index][../querying-an-ann-index] for
+        tuning advice.
+
         Parameters
         ----------
         nprobes: int
@@ -104,13 +135,20 @@ class LanceQueryBuilder:
 
         Returns
         -------
-        The LanceQueryBuilder object.
+        LanceQueryBuilder
+            The LanceQueryBuilder object.
         """
         self._nprobes = nprobes
         return self
 
     def refine_factor(self, refine_factor: int) -> LanceQueryBuilder:
-        """Set the refine factor to use.
+        """Set the refine factor to use, increasing the number of vectors sampled.
+
+        As an example, a refine factor of 2 will sample 2x as many vectors as
+        requested, re-ranks them, and returns the top half most relevant results.
+
+        See discussion in [Querying an ANN Index][querying-an-ann-index] for
+        tuning advice.
 
         Parameters
         ----------
@@ -119,7 +157,8 @@ class LanceQueryBuilder:
 
         Returns
         -------
-        The LanceQueryBuilder object.
+        LanceQueryBuilder
+            The LanceQueryBuilder object.
         """
         self._refine_factor = refine_factor
         return self
@@ -131,8 +170,28 @@ class LanceQueryBuilder:
         and also the "score" column which is the distance between the query
         vector and the returned vector.
         """
+
+        return self.to_arrow().to_pandas()
+
+    def to_arrow(self) -> pa.Table:
+        """
+        Execute the query and return the results as a arrow Table.
+        In addition to the selected columns, LanceDB also returns a vector
+        and also the "score" column which is the distance between the query
+        vector and the returned vector.
+        """
+        if self._table._conn.is_managed_remote:
+            try:
+                loop = asyncio.get_running_loop()
+            except RuntimeError:
+                loop = asyncio.get_event_loop()
+            result = self._table._conn._client.query(
+                self._table.name, self.to_remote_query()
+            )
+            return loop.run_until_complete(result).to_arrow()
+
         ds = self._table.to_lance()
-        tbl = ds.to_table(
+        return ds.to_table(
             columns=self._columns,
             filter=self._where,
             nearest={
@@ -144,7 +203,20 @@ class LanceQueryBuilder:
                 "refine_factor": self._refine_factor,
             },
         )
-        return tbl.to_pandas()
+
+    def to_remote_query(self) -> "VectorQuery":
+        # don't import unless we are connecting to remote
+        from lancedb.remote.client import VectorQuery
+
+        return VectorQuery(
+            vector=self._query.tolist(),
+            filter=self._where,
+            k=self._limit,
+            _metric=self._metric,
+            columns=self._columns,
+            nprobes=self._nprobes,
+            refine_factor=self._refine_factor,
+        )
 
 
 class LanceFtsQueryBuilder(LanceQueryBuilder):

python/lancedb/remote/__init__.py

@@ -0,0 +1,61 @@
+#  Copyright 2023 LanceDB Developers
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+import abc
+from typing import List, Optional
+
+import attr
+import pandas as pd
+import pyarrow as pa
+from pydantic import BaseModel
+
+__all__ = ["LanceDBClient", "VectorQuery", "VectorQueryResult"]
+
+
+class VectorQuery(BaseModel):
+    # vector to search for
+    vector: List[float]
+
+    # sql filter to refine the query with
+    filter: Optional[str] = None
+
+    # top k results to return
+    k: int
+
+    # # metrics
+    _metric: str = "L2"
+
+    # which columns to return in the results
+    columns: Optional[List[str]] = None
+
+    # optional query parameters for tuning the results,
+    # e.g. `{"nprobes": "10", "refine_factor": "10"}`
+    nprobes: int = 10
+
+    refine_factor: Optional[int] = None
+
+
+@attr.define
+class VectorQueryResult:
+    # for now the response is directly seralized into a pandas dataframe
+    tbl: pa.Table
+
+    def to_arrow(self) -> pa.Table:
+        return self.tbl
+
+
+class LanceDBClient(abc.ABC):
+    @abc.abstractmethod
+    def query(self, table_name: str, query: VectorQuery) -> VectorQueryResult:
+        """Query the LanceDB server for the given table and query."""
+        pass

python/lancedb/remote/client.py

@@ -0,0 +1,79 @@
+#  Copyright 2023 LanceDB Developers
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+
+import functools
+import urllib.parse
+
+import aiohttp
+import attr
+import pyarrow as pa
+
+from lancedb.remote import VectorQuery, VectorQueryResult
+from lancedb.remote.errors import LanceDBClientError
+
+
+def _check_not_closed(f):
+    @functools.wraps(f)
+    def wrapped(self, *args, **kwargs):
+        if self.closed:
+            raise ValueError("Connection is closed")
+        return f(self, *args, **kwargs)
+
+    return wrapped
+
+
+@attr.define(slots=False)
+class RestfulLanceDBClient:
+    url: str
+    closed: bool = attr.field(default=False, init=False)
+
+    @functools.cached_property
+    def session(self) -> aiohttp.ClientSession:
+        parsed = urllib.parse.urlparse(self.url)
+        scheme = parsed.scheme
+        if not scheme.startswith("lancedb"):
+            raise ValueError(
+                f"Invalid scheme: {scheme}, must be like lancedb+<flavor>://"
+            )
+        flavor = scheme.split("+")[1]
+        url = f"{flavor}://{parsed.hostname}:{parsed.port}"
+        return aiohttp.ClientSession(url)
+
+    async def close(self):
+        await self.session.close()
+        self.closed = True
+
+    @_check_not_closed
+    async def query(self, table_name: str, query: VectorQuery) -> VectorQueryResult:
+        async with self.session.post(
+            f"/table/{table_name}/", json=query.dict(exclude_none=True)
+        ) as resp:
+            resp: aiohttp.ClientResponse = resp
+            if 400 <= resp.status < 500:
+                raise LanceDBClientError(
+                    f"Bad Request: {resp.status}, error: {await resp.text()}"
+                )
+            if 500 <= resp.status < 600:
+                raise LanceDBClientError(
+                    f"Internal Server Error: {resp.status}, error: {await resp.text()}"
+                )
+            if resp.status != 200:
+                raise LanceDBClientError(
+                    f"Unknown Error: {resp.status}, error: {await resp.text()}"
+                )
+
+            resp_body = await resp.read()
+            with pa.ipc.open_file(pa.BufferReader(resp_body)) as reader:
+                tbl = reader.read_all()
+        return VectorQueryResult(tbl)

python/lancedb/remote/errors.py

@@ -0,0 +1,16 @@
+#  Copyright 2023 LanceDB Developers
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+
+class LanceDBClientError(RuntimeError):
+    pass

python/lancedb/table.py

@@ -14,7 +14,6 @@
 from __future__ import annotations
 
 import os
-import shutil
 from functools import cached_property
 from typing import List, Union
 
@@ -27,7 +26,6 @@ from lance.vector import vec_to_table
 
 from .common import DATA, VEC, VECTOR_COLUMN_NAME
 from .query import LanceFtsQueryBuilder, LanceQueryBuilder
-from .util import get_uri_scheme
 
 
 def _sanitize_data(data, schema):
@@ -47,6 +45,40 @@ def _sanitize_data(data, schema):
 class LanceTable:
     """
     A table in a LanceDB database.
+
+    Examples
+    --------
+
+    Create using [LanceDBConnection.create_table][lancedb.LanceDBConnection.create_table]
+    (more examples in that method's documentation).
+
+    >>> import lancedb
+    >>> db = lancedb.connect("./.lancedb")
+    >>> table = db.create_table("my_table", data=[{"vector": [1.1, 1.2], "b": 2}])
+    >>> table.head()
+    pyarrow.Table
+    vector: fixed_size_list<item: float>[2]
+      child 0, item: float
+    b: int64
+    ----
+    vector: [[[1.1,1.2]]]
+    b: [[2]]
+
+    Can append new data with [LanceTable.add][lancedb.table.LanceTable.add].
+
+    >>> table.add([{"vector": [0.5, 1.3], "b": 4}])
+    2
+
+    Can query the table with [LanceTable.search][lancedb.table.LanceTable.search].
+
+    >>> table.search([0.4, 0.4]).select(["b"]).to_df()
+       b      vector  score
+    0  4  [0.5, 1.3]   0.82
+    1  2  [1.1, 1.2]   1.13
+
+    Search queries are much faster when an index is created. See
+    [LanceTable.create_index][lancedb.table.LanceTable.create_index].
+
     """
 
     def __init__(
@@ -64,7 +96,12 @@ class LanceTable:
 
     @property
     def schema(self) -> pa.Schema:
-        """Return the schema of the table."""
+        """Return the schema of the table.
+
+        Returns
+        -------
+        pa.Schema
+            A PyArrow schema object."""
         return self._dataset.schema
 
     def list_versions(self):
@@ -72,12 +109,39 @@ class LanceTable:
         return self._dataset.versions()
 
     @property
-    def version(self):
+    def version(self) -> int:
         """Get the current version of the table"""
         return self._dataset.version
 
     def checkout(self, version: int):
-        """Checkout a version of the table"""
+        """Checkout a version of the table. This is an in-place operation.
+
+        This allows viewing previous versions of the table.
+
+        Parameters
+        ----------
+        version : int
+            The version to checkout.
+
+        Examples
+        --------
+        >>> import lancedb
+        >>> db = lancedb.connect("./.lancedb")
+        >>> table = db.create_table("my_table", [{"vector": [1.1, 0.9], "type": "vector"}])
+        >>> table.version
+        1
+        >>> table.to_pandas()
+               vector    type
+        0  [1.1, 0.9]  vector
+        >>> table.add([{"vector": [0.5, 0.2], "type": "vector"}])
+        2
+        >>> table.version
+        2
+        >>> table.checkout(1)
+        >>> table.to_pandas()
+               vector    type
+        0  [1.1, 0.9]  vector
+        """
         max_ver = max([v["version"] for v in self._dataset.versions()])
         if version < 1 or version > max_ver:
             raise ValueError(f"Invalid version {version}")
@@ -98,11 +162,20 @@ class LanceTable:
         return self._dataset.head(n)
 
     def to_pandas(self) -> pd.DataFrame:
-        """Return the table as a pandas DataFrame."""
+        """Return the table as a pandas DataFrame.
+
+        Returns
+        -------
+        pd.DataFrame
+        """
         return self.to_arrow().to_pandas()
 
     def to_arrow(self) -> pa.Table:
-        """Return the table as a pyarrow Table."""
+        """Return the table as a pyarrow Table.
+
+        Returns
+        -------
+        pa.Table"""
         return self._dataset.to_table()
 
     @property
@@ -175,7 +248,8 @@ class LanceTable:
 
         Returns
         -------
-        The number of vectors added to the table.
+        int
+            The number of vectors in the table.
         """
         data = _sanitize_data(data, self.schema)
         lance.write_dataset(data, self._dataset_uri, mode=mode)
@@ -193,10 +267,11 @@ class LanceTable:
 
         Returns
         -------
-        A LanceQueryBuilder object representing the query.
-        Once executed, the query returns selected columns, the vector,
-        and also the "score" column which is the distance between the query
-        vector and the returned vector.
+        LanceQueryBuilder
+            A query builder object representing the query.
+            Once executed, the query returns selected columns, the vector,
+            and also the "score" column which is the distance between the query
+            vector and the returned vector.
         """
         if isinstance(query, str):
             # fts
@@ -265,4 +340,6 @@ def _sanitize_vector_column(data: pa.Table, vector_column_name: str) -> pa.Table
         values = values.cast(pa.float32())
     list_size = len(values) / len(data)
     vec_arr = pa.FixedSizeListArray.from_arrays(values, list_size)
-    return data.set_column(data.column_names.index(vector_column_name), vector_column_name, vec_arr)
+    return data.set_column(
+        data.column_names.index(vector_column_name), vector_column_name, vec_arr
+    )

python/pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "lancedb"
-version = "0.1.6"
-dependencies = ["pylance>=0.4.17", "ratelimiter", "retry", "tqdm"]
+version = "0.1.8"
+dependencies = ["pylance>=0.4.20", "ratelimiter", "retry", "tqdm", "aiohttp", "pydantic", "attr"]
 description = "lancedb"
 authors = [
     { name = "LanceDB Devs", email = "dev@lancedb.com" },
@@ -37,7 +37,7 @@ repository = "https://github.com/lancedb/lancedb"
 
 [project.optional-dependencies]
 tests = [
-    "pytest", "pytest-mock"
+    "pytest", "pytest-mock", "doctest", "pytest-asyncio"
 ]
 dev = [
     "ruff", "pre-commit", "black"

python/tests/test_context.py

@@ -0,0 +1,77 @@
+#  Copyright 2023 LanceDB Developers
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+import pandas as pd
+import pytest
+
+from lancedb.context import contextualize
+
+
+@pytest.fixture
+def raw_df() -> pd.DataFrame:
+    return pd.DataFrame(
+        {
+            "token": [
+                "The",
+                "quick",
+                "brown",
+                "fox",
+                "jumped",
+                "over",
+                "the",
+                "lazy",
+                "dog",
+                "I",
+                "love",
+                "sandwiches",
+            ],
+            "document_id": [1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2],
+        }
+    )
+
+
+def test_contextualizer(raw_df: pd.DataFrame):
+    result = (
+        contextualize(raw_df)
+        .window(6)
+        .stride(3)
+        .text_col("token")
+        .groupby("document_id")
+        .to_df()["token"]
+        .to_list()
+    )
+
+    assert result == [
+        "The quick brown fox jumped over",
+        "fox jumped over the lazy dog",
+        "the lazy dog",
+        "I love sandwiches",
+    ]
+
+
+def test_contextualizer_with_threshold(raw_df: pd.DataFrame):
+    result = (
+        contextualize(raw_df)
+        .window(6)
+        .stride(3)
+        .text_col("token")
+        .groupby("document_id")
+        .min_window_size(4)
+        .to_df()["token"]
+        .to_list()
+    )
+
+    assert result == [
+        "The quick brown fox jumped over",
+        "fox jumped over the lazy dog",
+    ]

python/tests/test_db.py

@@ -119,4 +119,4 @@ def test_delete_table(tmp_path):
     assert db.table_names() == []
 
     db.create_table("test", data=data)
-    assert db.table_names() == ["test"]
+    assert db.table_names() == ["test"]

python/tests/test_e2e_remote_db.py

@@ -0,0 +1,27 @@
+#  Copyright 2023 LanceDB Developers
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+import numpy as np
+import pytest
+
+from lancedb import LanceDBConnection
+
+# TODO: setup integ test mark and script
+
+
+@pytest.mark.skip(reason="Need to set up a local server")
+def test_against_local_server():
+    conn = LanceDBConnection("lancedb+http://localhost:10024")
+    table = conn.open_table("sift1m_ivf1024_pq16")
+    df = table.search(np.random.rand(128)).to_df()
+    assert len(df) == 10

python/tests/test_embeddings.py

@@ -14,6 +14,7 @@ import sys
 
 import numpy as np
 import pyarrow as pa
+
 from lancedb.embeddings import with_embeddings
 
 

python/tests/test_fts.py

@@ -13,13 +13,13 @@
 import os
 import random
 
-import lancedb.fts
 import numpy as np
 import pandas as pd
 import pytest
 import tantivy
 
 import lancedb as ldb
+import lancedb.fts
 
 
 @pytest.fixture

python/tests/test_io.py

@@ -12,6 +12,7 @@
 #  limitations under the License.
 
 import os
+
 import pytest
 
 import lancedb
@@ -19,6 +20,7 @@ import lancedb
 # You need to setup AWS credentials an a base path to run this test. Example
 #    AWS_PROFILE=default TEST_S3_BASE_URL=s3://my_bucket/dataset pytest tests/test_io.py
 
+
 @pytest.mark.skipif(
     (os.environ.get("TEST_S3_BASE_URL") is None),
     reason="please setup s3 base url",

python/tests/test_query.py

@@ -17,12 +17,15 @@ import pandas as pd
 import pandas.testing as tm
 import pyarrow as pa
 import pytest
+
+from lancedb.db import LanceDBConnection
 from lancedb.query import LanceQueryBuilder
 
 
 class MockTable:
     def __init__(self, tmp_path):
         self.uri = tmp_path
+        self._conn = LanceDBConnection("/tmp/lance/")
 
     def to_lance(self):
         return lance.dataset(self.uri)
@@ -30,23 +33,17 @@ class MockTable:
 
 @pytest.fixture
 def table(tmp_path) -> MockTable:
-    df = pd.DataFrame(
+    df = pa.table(
         {
-            "vector": [[1, 2], [3, 4]],
-            "id": [1, 2],
-            "str_field": ["a", "b"],
-            "float_field": [1.0, 2.0],
+            "vector": pa.array(
+                [[1, 2], [3, 4]], type=pa.list_(pa.float32(), list_size=2)
+            ),
+            "id": pa.array([1, 2]),
+            "str_field": pa.array(["a", "b"]),
+            "float_field": pa.array([1.0, 2.0]),
         }
     )
-    schema = pa.schema(
-        [
-            pa.field("vector", pa.list_(pa.float32(), list_size=2)),
-            pa.field("id", pa.int32()),
-            pa.field("str_field", pa.string()),
-            pa.field("float_field", pa.float64()),
-        ]
-    )
-    lance.write_dataset(df, tmp_path, schema)
+    lance.write_dataset(df, tmp_path)
     return MockTable(tmp_path)
 
 
@@ -65,7 +62,7 @@ def test_query_builder_with_filter(table):
 def test_query_builder_with_metric(table):
     query = [4, 8]
     df_default = LanceQueryBuilder(table, query).to_df()
-    df_l2 = LanceQueryBuilder(table, query).metric("l2").to_df()
+    df_l2 = LanceQueryBuilder(table, query).metric("L2").to_df()
     tm.assert_frame_equal(df_default, df_l2)
 
     df_cosine = LanceQueryBuilder(table, query).metric("cosine").limit(1).to_df()

python/tests/test_remote_client.py

@@ -0,0 +1,95 @@
+#  Copyright 2023 LanceDB Developers
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+import attr
+import numpy as np
+import pandas as pd
+import pyarrow as pa
+import pytest
+from aiohttp import web
+
+from lancedb.remote.client import RestfulLanceDBClient, VectorQuery
+
+
+@attr.define
+class MockLanceDBServer:
+    runner: web.AppRunner = attr.field(init=False)
+    site: web.TCPSite = attr.field(init=False)
+
+    async def query_handler(self, request: web.Request) -> web.Response:
+        table_name = request.match_info["table_name"]
+        assert table_name == "test_table"
+
+        request_json = await request.json()
+        # TODO: do some matching
+
+        vecs = pd.Series([np.random.rand(128) for x in range(10)], name="vector")
+        ids = pd.Series(range(10), name="id")
+        df = pd.DataFrame([vecs, ids]).T
+
+        batch = pa.RecordBatch.from_pandas(
+            df,
+            schema=pa.schema(
+                [
+                    pa.field("vector", pa.list_(pa.float32(), 128)),
+                    pa.field("id", pa.int64()),
+                ]
+            ),
+        )
+
+        sink = pa.BufferOutputStream()
+        with pa.ipc.new_file(sink, batch.schema) as writer:
+            writer.write_batch(batch)
+
+        return web.Response(body=sink.getvalue().to_pybytes())
+
+    async def setup(self):
+        app = web.Application()
+        app.add_routes([web.post("/table/{table_name}", self.query_handler)])
+        self.runner = web.AppRunner(app)
+        await self.runner.setup()
+        self.site = web.TCPSite(self.runner, "localhost", 8111)
+
+    async def start(self):
+        await self.site.start()
+
+    async def stop(self):
+        await self.runner.cleanup()
+
+
+@pytest.mark.skip(reason="flaky somehow, fix later")
+@pytest.mark.asyncio
+async def test_e2e_with_mock_server():
+    mock_server = MockLanceDBServer()
+    await mock_server.setup()
+    await mock_server.start()
+
+    try:
+        client = RestfulLanceDBClient("lancedb+http://localhost:8111")
+        df = (
+            await client.query(
+                "test_table",
+                VectorQuery(
+                    vector=np.random.rand(128).tolist(),
+                    k=10,
+                    _metric="L2",
+                    columns=["id", "vector"],
+                ),
+            )
+        ).to_df()
+
+        assert "vector" in df.columns
+        assert "id" in df.columns
+    finally:
+        # make sure we don't leak resources
+        await mock_server.stop()

python/tests/test_remote_db.py

@@ -0,0 +1,35 @@
+#  Copyright 2023 LanceDB Developers
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+import pyarrow as pa
+
+from lancedb.db import LanceDBConnection
+from lancedb.remote.client import VectorQuery, VectorQueryResult
+
+
+class FakeLanceDBClient:
+    async def close(self):
+        pass
+
+    async def query(self, table_name: str, query: VectorQuery) -> VectorQueryResult:
+        assert table_name == "test"
+        t = pa.schema([]).empty_table()
+        return VectorQueryResult(t)
+
+
+def test_remote_db():
+    conn = LanceDBConnection("lancedb+http://client-will-be-injected")
+    setattr(conn, "_client", FakeLanceDBClient())
+
+    table = conn["test"]
+    table.search([1.0, 2.0]).to_df()

python/tests/test_table.py

@@ -11,11 +11,13 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 
+import functools
 from pathlib import Path
 
 import pandas as pd
 import pyarrow as pa
 import pytest
+
 from lancedb.table import LanceTable
 
 
@@ -23,6 +25,10 @@ class MockDB:
     def __init__(self, uri: Path):
         self.uri = uri
 
+    @functools.cached_property
+    def is_managed_remote(self) -> bool:
+        return False
+
 
 @pytest.fixture
 def db(tmp_path) -> MockDB:

rust/ffi/node/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb-node"
-version = "0.1.0"
+version = "0.1.8"
 description = "Serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"
 edition = "2018"

rust/ffi/node/src/index/vector.rs

@@ -97,6 +97,7 @@ fn get_index_params_builder(
                 let ivf_params = IvfBuildParams {
                     num_partitions: np,
                     max_iters,
+                    centroids: None,
                 };
                 index_builder.ivf_params(ivf_params)
             });

rust/ffi/node/src/lib.rs

@@ -264,6 +264,25 @@ fn table_add(mut cx: FunctionContext) -> JsResult<JsPromise> {
     Ok(promise)
 }
 
+fn table_count_rows(mut cx: FunctionContext) -> JsResult<JsPromise> {
+    let js_table = cx.this().downcast_or_throw::<JsBox<JsTable>, _>(&mut cx)?;
+    let rt = runtime(&mut cx)?;
+    let channel = cx.channel();
+
+    let (deferred, promise) = cx.promise();
+    let table = js_table.table.clone();
+
+    rt.block_on(async move {
+        let num_rows_result = table.lock().unwrap().count_rows().await;
+
+        deferred.settle_with(&channel, move |mut cx| {
+            let num_rows = num_rows_result.or_else(|err| cx.throw_error(err.to_string()))?;
+            Ok(cx.number(num_rows as f64))
+        });
+    });
+    Ok(promise)
+}
+
 #[neon::main]
 fn main(mut cx: ModuleContext) -> NeonResult<()> {
     cx.export_function("databaseNew", database_new)?;
@@ -272,6 +291,7 @@ fn main(mut cx: ModuleContext) -> NeonResult<()> {
     cx.export_function("tableSearch", table_search)?;
     cx.export_function("tableCreate", table_create)?;
     cx.export_function("tableAdd", table_add)?;
+    cx.export_function("tableCountRows", table_count_rows)?;
     cx.export_function(
         "tableCreateVectorIndex",
         index::vector::table_create_vector_index,

rust/vectordb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb"
-version = "0.0.1"
+version = "0.1.8"
 edition = "2021"
 description = "Serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"
@@ -14,7 +14,7 @@ arrow-data = "37.0"
 arrow-schema = "37.0"
 object_store = "0.5.6"
 snafu = "0.7.4"
-lance = "0.4.17"
+lance = "0.4.21"
 tokio = { version = "1.23", features = ["rt-multi-thread"] }
 
 [dev-dependencies]

rust/vectordb/src/database.rs

@@ -42,7 +42,7 @@ impl Database {
     ///
     /// * A [Database] object.
     pub async fn connect(uri: &str) -> Result<Database> {
-        let object_store = ObjectStore::new(uri).await?;
+        let (object_store, _) = ObjectStore::from_uri(uri).await?;
         if object_store.is_local() {
             Self::try_create_dir(uri).context(CreateDirSnafu { path: uri })?;
         }
@@ -69,7 +69,7 @@ impl Database {
     pub async fn table_names(&self) -> Result<Vec<String>> {
         let f = self
             .object_store
-            .read_dir("/")
+            .read_dir(self.uri.as_str())
             .await?
             .iter()
             .map(|fname| Path::new(fname))

rust/vectordb/src/index/vector.rs

@@ -20,6 +20,8 @@ pub trait VectorIndexBuilder {
     fn get_column(&self) -> Option<String>;
     fn get_index_name(&self) -> Option<String>;
     fn build(&self) -> VectorIndexParams;
+
+    fn get_replace(&self) -> bool;
 }
 
 pub struct IvfPQIndexBuilder {
@@ -28,6 +30,7 @@ pub struct IvfPQIndexBuilder {
     metric_type: Option<MetricType>,
     ivf_params: Option<IvfBuildParams>,
     pq_params: Option<PQBuildParams>,
+    replace: bool,
 }
 
 impl IvfPQIndexBuilder {
@@ -38,6 +41,7 @@ impl IvfPQIndexBuilder {
             metric_type: None,
             ivf_params: None,
             pq_params: None,
+            replace: true,
         }
     }
 }
@@ -67,6 +71,11 @@ impl IvfPQIndexBuilder {
         self.pq_params = Some(pq_params);
         self
     }
+
+    pub fn replace(&mut self, replace: bool) -> &mut IvfPQIndexBuilder {
+        self.replace = replace;
+        self
+    }
 }
 
 impl VectorIndexBuilder for IvfPQIndexBuilder {
@@ -84,6 +93,10 @@ impl VectorIndexBuilder for IvfPQIndexBuilder {
 
         VectorIndexParams::with_ivf_pq_params(pq_params.metric_type, ivf_params, pq_params)
     }
+
+    fn get_replace(&self) -> bool {
+        self.replace
+    }
 }
 
 #[cfg(test)]

rust/vectordb/src/query.rs

@@ -177,7 +177,7 @@ mod tests {
     #[tokio::test]
     async fn test_setters_getters() {
         let mut batches: Box<dyn RecordBatchReader> = Box::new(make_test_batches());
-        let ds = Dataset::write(&mut batches, ":memory:", None)
+        let ds = Dataset::write(&mut batches, "memory://foo", None)
             .await
             .unwrap();
 
@@ -206,7 +206,7 @@ mod tests {
     #[tokio::test]
     async fn test_execute() {
         let mut batches: Box<dyn RecordBatchReader> = Box::new(make_test_batches());
-        let ds = Dataset::write(&mut batches, ":memory:", None)
+        let ds = Dataset::write(&mut batches, "memory://foo", None)
             .await
             .unwrap();
 

rust/vectordb/src/table.rs

@@ -130,6 +130,7 @@ impl Table {
                 IndexType::Vector,
                 index_builder.get_index_name(),
                 &index_builder.build(),
+                index_builder.get_replace(),
             )
             .await?;
         self.dataset = Arc::new(dataset);
@@ -233,7 +234,7 @@ mod tests {
         let uri = tmp_dir.path().to_str().unwrap();
 
         let batches: Box<dyn RecordBatchReader> = Box::new(make_test_batches());
-        let schema = batches.schema().clone();
+        let _ = batches.schema().clone();
         Table::create(&uri, "test", batches).await.unwrap();
 
         let batches: Box<dyn RecordBatchReader> = Box::new(make_test_batches());