[python] Bump version: 0.3.2 → 0.3.3

- posthog charges per event and docs events are registered very
frequently. We can keep tracking them on GA
- Reduced sentry trace factor

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.3.3
+current_version = 0.3.5
 commit = True
 message = Bump version: {current_version} → {new_version}
 tag = True

.github/workflows/node.yml

@@ -11,6 +11,10 @@ on:
       - .github/workflows/node.yml
       - docker-compose.yml
 
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
 env:
   # 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.

.github/workflows/npm-publish.yml

@@ -38,7 +38,7 @@ jobs:
             node/vectordb-*.tgz
 
   node-macos:
-    runs-on: macos-12
+    runs-on: macos-13
     # Only runs on tags that matches the make-release action
     if: startsWith(github.ref, 'refs/tags/v')
     strategy:

.github/workflows/python.yml

@@ -8,6 +8,11 @@ on:
     paths:
       - python/**
       - .github/workflows/python.yml
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
 jobs:
   linux:
     timeout-minutes: 30
@@ -43,7 +48,7 @@ jobs:
       run: pytest --doctest-modules lancedb
   mac:
     timeout-minutes: 30
-    runs-on: "macos-12"
+    runs-on: "macos-13"
     defaults:
       run:
         shell: bash

.github/workflows/rust.yml

@@ -10,6 +10,10 @@ on:
       - rust/**
       - .github/workflows/rust.yml
 
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
 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.
@@ -44,7 +48,7 @@ jobs:
     - name: Run tests
       run: cargo test --all-features
   macos:
-    runs-on: macos-12
+    runs-on: macos-13
     timeout-minutes: 30
     defaults:
       run:

Cargo.toml

@@ -5,9 +5,9 @@ exclude = ["python"]
 resolver = "2"
 
 [workspace.dependencies]
-lance = { "version" = "=0.8.7", "features" = ["dynamodb"] }
+lance = { "version" = "=0.8.10", "features" = ["dynamodb"] }
-lance-linalg = { "version" = "=0.8.7" }
+lance-linalg = { "version" = "=0.8.10" }
-lance-testing = { "version" = "=0.8.7" }
+lance-testing = { "version" = "=0.8.10" }
 # Note that this one does not include pyarrow
 arrow = { version = "47.0.0", optional = false }
 arrow-array = "47.0"
@@ -19,7 +19,7 @@ arrow-arith = "47.0"
 arrow-cast = "47.0"
 chrono = "0.4.23"
 half = { "version" = "=2.3.1", default-features = false, features = [
-    "num-traits"
+    "num-traits",
 ] }
 log = "0.4"
 object_store = "0.7.1"

docs/mkdocs.yml

@@ -150,8 +150,6 @@ nav:
 
 extra_css:
   - styles/global.css
-extra_javascript:
-  - scripts/posthog.js
 
 extra:
   analytics:

docs/src/ann_indexes.md

@@ -71,9 +71,41 @@ a single PQ code.
 ### Use GPU to build vector index
 
 Lance Python SDK has experimental GPU support for creating IVF index.
+Using GPU for index creation requires [PyTorch>2.0](https://pytorch.org/) being installed.
+
 You can specify the GPU device to train IVF partitions via
 
-- **accelerator**: Specify to `"cuda"`` to enable GPU training.
+- **accelerator**: Specify to ``cuda`` or ``mps`` (on Apple Silicon) to enable GPU training.
+
+=== "Linux"
+
+     <!-- skip-test -->
+     ``` { .python .copy }
+     # Create index using CUDA on Nvidia GPUs.
+     tbl.create_index(
+          num_partitions=256,
+          num_sub_vectors=96,
+          accelerator="cuda"
+     )
+     ```
+
+=== "Macos"
+
+     <!-- skip-test -->
+     ```python
+     # Create index using MPS on Apple Silicon.
+     tbl.create_index(
+          num_partitions=256,
+          num_sub_vectors=96,
+          accelerator="mps"
+     )
+     ```
+
+Trouble shootings:
+
+If you see ``AssertionError: Torch not compiled with CUDA enabled``, you need to [install
+PyTorch with CUDA support](https://pytorch.org/get-started/locally/).
+
 
 ## Querying an ANN Index
 

docs/src/python/python.md

@@ -22,8 +22,6 @@ pip install lancedb
 
 ::: lancedb.query.LanceQueryBuilder
 
-::: lancedb.query.LanceFtsQueryBuilder
-
 ## Embeddings
 
 ::: lancedb.embeddings.registry.EmbeddingFunctionRegistry
@@ -56,7 +54,7 @@ pip install lancedb
 
 ## Utilities
 
-::: lancedb.vector
+::: lancedb.schema.vector
 
 ## Integrations
 

docs/test/md_testing.py

@@ -18,29 +18,45 @@ python_file = ".py"
 python_folder = "python"
 
 files = glob.glob(glob_string, recursive=True)
-excluded_files = [f for excluded_glob in excluded_globs for f in glob.glob(excluded_glob, recursive=True)]
+excluded_files = [
+    f
+    for excluded_glob in excluded_globs
+    for f in glob.glob(excluded_glob, recursive=True)
+]
+
 
 def yield_lines(lines: Iterator[str], prefix: str, suffix: str):
     in_code_block = False
     # Python code has strict indentation
     strip_length = 0
+    skip_test = False
     for line in lines:
+        if "skip-test" in line:
+            skip_test = True
         if line.strip().startswith(prefix + python_prefix):
             in_code_block = True
             strip_length = len(line) - len(line.lstrip())
         elif in_code_block and line.strip().startswith(suffix):
             in_code_block = False
-            yield "\n"
+            if not skip_test:
+                yield "\n"
+            skip_test = False
         elif in_code_block:
-            yield line[strip_length:]
+            if not skip_test:
+                yield line[strip_length:]
 
 for file in filter(lambda file: file not in excluded_files, files):
     with open(file, "r") as f:
         lines = list(yield_lines(iter(f), "```", "```"))
 
     if len(lines) > 0:
-        out_path = Path(python_folder) / Path(file).name.strip(".md") / (Path(file).name.strip(".md") + python_file)
+        print(lines)
+        out_path = (
+            Path(python_folder)
+            / Path(file).name.strip(".md")
+            / (Path(file).name.strip(".md") + python_file)
+        )
         print(out_path)
         out_path.parent.mkdir(exist_ok=True, parents=True)
         with open(out_path, "w") as out:
-            out.writelines(lines)
+            out.writelines(lines)

node/README.md

@@ -10,7 +10,7 @@ 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).
+yet support musl-based Linux (such as Alpine Linux).
 
 ## Usage
 

node/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "vectordb",
-  "version": "0.3.3",
+  "version": "0.3.5",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "vectordb",
-      "version": "0.3.3",
+      "version": "0.3.5",
       "cpu": [
         "x64",
         "arm64"
@@ -53,11 +53,11 @@
         "uuid": "^9.0.0"
       },
       "optionalDependencies": {
-        "@lancedb/vectordb-darwin-arm64": "0.3.3",
+        "@lancedb/vectordb-darwin-arm64": "0.3.5",
-        "@lancedb/vectordb-darwin-x64": "0.3.3",
+        "@lancedb/vectordb-darwin-x64": "0.3.5",
-        "@lancedb/vectordb-linux-arm64-gnu": "0.3.3",
+        "@lancedb/vectordb-linux-arm64-gnu": "0.3.5",
-        "@lancedb/vectordb-linux-x64-gnu": "0.3.3",
+        "@lancedb/vectordb-linux-x64-gnu": "0.3.5",
-        "@lancedb/vectordb-win32-x64-msvc": "0.3.3"
+        "@lancedb/vectordb-win32-x64-msvc": "0.3.5"
       }
     },
     "node_modules/@apache-arrow/ts": {
@@ -317,9 +317,9 @@
       }
     },
     "node_modules/@lancedb/vectordb-darwin-arm64": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.5.tgz",
-      "integrity": "sha512-nvyj7xNX2/wb/PH5TjyhLR/NQ1jVuoBw2B5UaSg7qf8Tnm5SSXWQ7F25RVKcKwh72fz1qB+CWW24ftZnRzbT/Q==",
+      "integrity": "sha512-Nnso+WXMSTIUouddDgPDNt40K6d2fF7W5OsfgAMDXAhUrdSMOZbVP0bWklRz9J7JluseBL9/MfLSEYZDTvrACg==",
       "cpu": [
         "arm64"
       ],
@@ -329,9 +329,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-darwin-x64": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.5.tgz",
-      "integrity": "sha512-7CW+nILyPHp6cua0Rl0xaTDWw/vajEn/jCsEjFYgDmE+rtf5Z5Fum41FxR9C2TtIAvUK+nWb5mkYeOLqU6vRvg==",
+      "integrity": "sha512-gvg/iq13zAamLL7jueiIw7Q67dygm/NmILkFQ3WrAOUjr0IMxLBCv+XMxt62xajTrA+ObyfmU1uiuhrJL81PWw==",
       "cpu": [
         "x64"
       ],
@@ -341,9 +341,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-linux-arm64-gnu": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.5.tgz",
-      "integrity": "sha512-MmhwbacKxZPkLwwOqysVY8mUb8lFoyFIPlYhSLV4xS1C8X4HWALljIul1qMl1RYudp9Uc3PsOzRexl+OvCGfUw==",
+      "integrity": "sha512-6PvCBIXI9zPqF478TibZxxiAehFZ530g0FOFDT49xtp540HvhE9+XQk/yO0w96mvyoCfzB2lK4haDmdhCoehNw==",
       "cpu": [
         "arm64"
       ],
@@ -353,9 +353,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-linux-x64-gnu": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.5.tgz",
-      "integrity": "sha512-OrNlsKi/QPw59Po040oRKn8IuqFEk4upc/4FaFKqVkcmQjjZrMg5Kgy9ZfWIhHdAnWXXggZZIPArpt0X1B0ceA==",
+      "integrity": "sha512-e3nqurUeCow4QONeNf/QP50Z90mgrh9xoUfjRSHcCPQcP6WgmFEafbt0jeSVgZ7tbt7+03/MK0YexhHM/5sBjA==",
       "cpu": [
         "x64"
       ],
@@ -365,9 +365,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-win32-x64-msvc": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.5.tgz",
-      "integrity": "sha512-lIT0A7a6eqX51IfGyhECtpXXgsr//kgbd+HZbcCdPy2GMmNezSch/7V22zExDSpF32hX8WfgcTLYCVWVilggDQ==",
+      "integrity": "sha512-RC1FfgEr6Z9sADuvspT2PG1B2mpKRdckgeiHqTHkIXdq3Qp5V5TeQJAbVvMr2xd1q99W6zreub52QXf+AilLVQ==",
       "cpu": [
         "x64"
       ],
@@ -4869,33 +4869,33 @@
       }
     },
     "@lancedb/vectordb-darwin-arm64": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.5.tgz",
-      "integrity": "sha512-nvyj7xNX2/wb/PH5TjyhLR/NQ1jVuoBw2B5UaSg7qf8Tnm5SSXWQ7F25RVKcKwh72fz1qB+CWW24ftZnRzbT/Q==",
+      "integrity": "sha512-Nnso+WXMSTIUouddDgPDNt40K6d2fF7W5OsfgAMDXAhUrdSMOZbVP0bWklRz9J7JluseBL9/MfLSEYZDTvrACg==",
       "optional": true
     },
     "@lancedb/vectordb-darwin-x64": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.5.tgz",
-      "integrity": "sha512-7CW+nILyPHp6cua0Rl0xaTDWw/vajEn/jCsEjFYgDmE+rtf5Z5Fum41FxR9C2TtIAvUK+nWb5mkYeOLqU6vRvg==",
+      "integrity": "sha512-gvg/iq13zAamLL7jueiIw7Q67dygm/NmILkFQ3WrAOUjr0IMxLBCv+XMxt62xajTrA+ObyfmU1uiuhrJL81PWw==",
       "optional": true
     },
     "@lancedb/vectordb-linux-arm64-gnu": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.5.tgz",
-      "integrity": "sha512-MmhwbacKxZPkLwwOqysVY8mUb8lFoyFIPlYhSLV4xS1C8X4HWALljIul1qMl1RYudp9Uc3PsOzRexl+OvCGfUw==",
+      "integrity": "sha512-6PvCBIXI9zPqF478TibZxxiAehFZ530g0FOFDT49xtp540HvhE9+XQk/yO0w96mvyoCfzB2lK4haDmdhCoehNw==",
       "optional": true
     },
     "@lancedb/vectordb-linux-x64-gnu": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.5.tgz",
-      "integrity": "sha512-OrNlsKi/QPw59Po040oRKn8IuqFEk4upc/4FaFKqVkcmQjjZrMg5Kgy9ZfWIhHdAnWXXggZZIPArpt0X1B0ceA==",
+      "integrity": "sha512-e3nqurUeCow4QONeNf/QP50Z90mgrh9xoUfjRSHcCPQcP6WgmFEafbt0jeSVgZ7tbt7+03/MK0YexhHM/5sBjA==",
       "optional": true
     },
     "@lancedb/vectordb-win32-x64-msvc": {
-      "version": "0.3.3",
+      "version": "0.3.5",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.3.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.5.tgz",
-      "integrity": "sha512-lIT0A7a6eqX51IfGyhECtpXXgsr//kgbd+HZbcCdPy2GMmNezSch/7V22zExDSpF32hX8WfgcTLYCVWVilggDQ==",
+      "integrity": "sha512-RC1FfgEr6Z9sADuvspT2PG1B2mpKRdckgeiHqTHkIXdq3Qp5V5TeQJAbVvMr2xd1q99W6zreub52QXf+AilLVQ==",
       "optional": true
     },
     "@neon-rs/cli": {

node/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vectordb",
-  "version": "0.3.3",
+  "version": "0.3.5",
   "description": " Serverless, low-latency vector database for AI applications",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
@@ -81,10 +81,10 @@
     }
   },
   "optionalDependencies": {
-    "@lancedb/vectordb-darwin-arm64": "0.3.3",
+    "@lancedb/vectordb-darwin-arm64": "0.3.5",
-    "@lancedb/vectordb-darwin-x64": "0.3.3",
+    "@lancedb/vectordb-darwin-x64": "0.3.5",
-    "@lancedb/vectordb-linux-arm64-gnu": "0.3.3",
+    "@lancedb/vectordb-linux-arm64-gnu": "0.3.5",
-    "@lancedb/vectordb-linux-x64-gnu": "0.3.3",
+    "@lancedb/vectordb-linux-x64-gnu": "0.3.5",
-    "@lancedb/vectordb-win32-x64-msvc": "0.3.3"
+    "@lancedb/vectordb-win32-x64-msvc": "0.3.5"
   }
 }

node/src/index.ts

@@ -23,7 +23,7 @@ import { Query } from './query'
 import { isEmbeddingFunction } from './embedding/embedding_function'
 
 // eslint-disable-next-line @typescript-eslint/no-var-requires
-const { databaseNew, databaseTableNames, databaseOpenTable, databaseDropTable, tableCreate, tableAdd, tableCreateVectorIndex, tableCountRows, tableDelete, tableCleanupOldVersions, tableCompactFiles } = require('../native.js')
+const { databaseNew, databaseTableNames, databaseOpenTable, databaseDropTable, tableCreate, tableAdd, tableCreateVectorIndex, tableCountRows, tableDelete, tableCleanupOldVersions, tableCompactFiles, tableListIndices, tableIndexStats } = require('../native.js')
 
 export { Query }
 export type { EmbeddingFunction }
@@ -260,6 +260,27 @@ export interface Table<T = number[]> {
    * ```
    */
   delete: (filter: string) => Promise<void>
+
+  /**
+   * List the indicies on this table.
+   */
+  listIndices: () => Promise<VectorIndex[]>
+
+  /**
+   * Get statistics about an index.
+   */
+  indexStats: (indexUuid: string) => Promise<IndexStats>
+}
+
+export interface VectorIndex {
+  columns: string[]
+  name: string
+  uuid: string
+}
+
+export interface IndexStats {
+  numIndexedRows: number | null
+  numUnindexedRows: number | null
 }
 
 /**
@@ -502,6 +523,14 @@ export class LocalTable<T = number[]> implements Table<T> {
         return res.metrics
       })
   }
+
+  async listIndices (): Promise<VectorIndex[]> {
+    return tableListIndices.call(this._tbl)
+  }
+
+  async indexStats (indexUuid: string): Promise<IndexStats> {
+    return tableIndexStats.call(this._tbl, indexUuid)
+  }
 }
 
 export interface CleanupStats {

node/src/remote/index.ts

@@ -14,7 +14,9 @@
 
 import {
   type EmbeddingFunction, type Table, type VectorIndexParams, type Connection,
-  type ConnectionOptions, type CreateTableOptions, type WriteOptions
+  type ConnectionOptions, type CreateTableOptions, type VectorIndex,
+  type WriteOptions,
+  type IndexStats
 } from '../index'
 import { Query } from '../query'
 
@@ -241,4 +243,21 @@ export class RemoteTable<T = number[]> implements Table<T> {
   async delete (filter: string): Promise<void> {
     await this._client.post(`/v1/table/${this._name}/delete/`, { predicate: filter })
   }
+
+  async listIndices (): Promise<VectorIndex[]> {
+    const results = await this._client.post(`/v1/table/${this._name}/index/list/`)
+    return results.data.indexes?.map((index: any) => ({
+      columns: index.columns,
+      name: index.index_name,
+      uuid: index.index_uuid
+    }))
+  }
+
+  async indexStats (indexUuid: string): Promise<IndexStats> {
+    const results = await this._client.post(`/v1/table/${this._name}/index/${indexUuid}/stats/`)
+    return {
+      numIndexedRows: results.data.num_indexed_rows,
+      numUnindexedRows: results.data.num_unindexed_rows
+    }
+  }
 }

node/src/test/test.ts

@@ -328,6 +328,24 @@ describe('LanceDB client', function () {
       const createIndex = table.createIndex({ type: 'ivf_pq', column: 'name', num_partitions: -1, max_iters: 2, num_sub_vectors: 2 })
       await expect(createIndex).to.be.rejectedWith('num_partitions: must be > 0')
     })
+
+    it('should be able to list index and stats', async function () {
+      const uri = await createTestDB(32, 300)
+      const con = await lancedb.connect(uri)
+      const table = await con.openTable('vectors')
+      await table.createIndex({ type: 'ivf_pq', column: 'vector', num_partitions: 2, max_iters: 2, num_sub_vectors: 2 })
+
+      const indices = await table.listIndices()
+      expect(indices).to.have.lengthOf(1)
+      expect(indices[0].name).to.equal('vector_idx')
+      expect(indices[0].uuid).to.not.be.equal(undefined)
+      expect(indices[0].columns).to.have.lengthOf(1)
+      expect(indices[0].columns[0]).to.equal('vector')
+
+      const stats = await table.indexStats(indices[0].uuid)
+      expect(stats.numIndexedRows).to.equal(300)
+      expect(stats.numUnindexedRows).to.equal(0)
+    }).timeout(50_000)
   })
 
   describe('when using a custom embedding function', function () {

python/.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.3.2
+current_version = 0.3.3
 commit = True
 message = [python] Bump version: {current_version} → {new_version}
 tag = True

python/lancedb/context.py

@@ -84,7 +84,9 @@ def contextualize(raw_df: "pd.DataFrame") -> Contextualizer:
     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_pandas()
+    >>> (contextualize(data)
+    ...     .window(4).stride(2).text_col('token').groupby('document_id')
+    ...     .to_pandas())
                        token  document_id
     0    The quick brown fox            1
     2  brown fox jumped over            1
@@ -92,18 +94,24 @@ def contextualize(raw_df: "pd.DataFrame") -> Contextualizer:
     6           the lazy dog            1
     9      I love sandwiches            2
 
-    ``min_window_size`` determines the minimum size of the  context windows that are generated
+    ``min_window_size`` determines the minimum size of the context windows
-    This can be used to trim the last few context windows which have size less than
+    that are generated.This can be used to trim the last few context windows
-    ``min_window_size``. By default context windows of size 1 are skipped.
+    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_pandas()
+    >>> (contextualize(data)
+    ...     .window(6).stride(3).text_col('token').groupby('document_id')
+    ...     .to_pandas())
                                  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_pandas()
+    >>> (contextualize(data)
+    ...     .window(6).stride(3).min_window_size(4).text_col('token')
+    ...     .groupby('document_id')
+    ...     .to_pandas())
                                  token  document_id
     0  The quick brown fox jumped over            1
     3     fox jumped over the lazy dog            1
@@ -113,7 +121,9 @@ def contextualize(raw_df: "pd.DataFrame") -> Contextualizer:
 
 
 class Contextualizer:
-    """Create context windows from a DataFrame. See [lancedb.context.contextualize][]."""
+    """Create context windows from a DataFrame.
+    See [lancedb.context.contextualize][].
+    """
 
     def __init__(self, raw_df):
         self._text_col = None
@@ -183,7 +193,7 @@ class Contextualizer:
         deprecated_in="0.3.1",
         removed_in="0.4.0",
         current_version=__version__,
-        details="Use the bar function instead",
+        details="Use to_pandas() instead",
     )
     def to_df(self) -> "pd.DataFrame":
         return self.to_pandas()

python/lancedb/db.py

@@ -52,12 +52,24 @@ class DBConnection(ABC):
         ----------
         name: str
             The name of the table.
-        data: list, tuple, dict, pd.DataFrame; optional
+        data: The data to initialize the table, *optional*
-            The data to initialize the table. User must provide at least one of `data` or `schema`.
+            User must provide at least one of `data` or `schema`.
-        schema: pyarrow.Schema or LanceModel; optional
+            Acceptable types are:
-            The schema of the table.
+
+            - dict or list-of-dict
+
+            - pandas.DataFrame
+
+            - pyarrow.Table or pyarrow.RecordBatch
+        schema: The schema of the table, *optional*
+            Acceptable types are:
+
+            - pyarrow.Schema
+
+            - [LanceModel][lancedb.pydantic.LanceModel]
         mode: str; default "create"
-            The mode to use when creating the table. Can be either "create" or "overwrite".
+            The mode to use when creating the table.
+            Can be either "create" or "overwrite".
             By default, if the table already exists, an exception is raised.
             If you want to overwrite the table, use mode="overwrite".
         on_bad_vectors: str, default "error"
@@ -150,7 +162,8 @@ class DBConnection(ABC):
         ...     for i in range(5):
         ...         yield pa.RecordBatch.from_arrays(
         ...             [
-        ...                 pa.array([[3.1, 4.1], [5.9, 26.5]], pa.list_(pa.float32(), 2)),
+        ...                 pa.array([[3.1, 4.1], [5.9, 26.5]],
+        ...                     pa.list_(pa.float32(), 2)),
         ...                 pa.array(["foo", "bar"]),
         ...                 pa.array([10.0, 20.0]),
         ...             ],
@@ -250,7 +263,7 @@ class LanceDBConnection(DBConnection):
         return self._uri
 
     def table_names(self) -> list[str]:
-        """Get the names of all tables in the database.
+        """Get the names of all tables in the database. The names are sorted.
 
         Returns
         -------
@@ -274,6 +287,7 @@ class LanceDBConnection(DBConnection):
             for file_info in paths
             if file_info.extension == "lance"
         ]
+        tables.sort()
         return tables
 
     def __len__(self) -> int:

python/lancedb/query.py

@@ -30,7 +30,40 @@ pd = safe_import_pandas()
 
 
 class Query(pydantic.BaseModel):
-    """A Query"""
+    """The LanceDB Query
+
+    Attributes
+    ----------
+    vector : List[float]
+        the vector to search for
+    filter : Optional[str]
+        sql filter to refine the query with, optional
+    prefilter : bool
+        if True then apply the filter before vector search
+    k : int
+        top k results to return
+    metric : str
+        the distance metric between a pair of vectors,
+
+        can support L2 (default), Cosine and Dot.
+        [metric definitions][search]
+    columns : Optional[List[str]]
+        which columns to return in the results
+    nprobes : int
+        The number of probes used - optional
+
+        - A higher number makes search more accurate but also slower.
+
+        - See discussion in [Querying an ANN Index][querying-an-ann-index] for
+          tuning advice.
+    refine_factor : Optional[int]
+        Refine the results by reading extra elements and re-ranking them in memory - optional
+
+        - A higher number makes search more accurate but also slower.
+
+        - See discussion in [Querying an ANN Index][querying-an-ann-index] for
+          tuning advice.
+    """
 
     vector_column: str = VECTOR_COLUMN_NAME
 
@@ -61,6 +94,10 @@ class Query(pydantic.BaseModel):
 
 
 class LanceQueryBuilder(ABC):
+    """Build LanceDB query based on specific query type:
+    vector or full text search.
+    """
+
     @classmethod
     def create(
         cls,
@@ -133,11 +170,11 @@ class LanceQueryBuilder(ABC):
         deprecated_in="0.3.1",
         removed_in="0.4.0",
         current_version=__version__,
-        details="Use the bar function instead",
+        details="Use to_pandas() instead",
     )
     def to_df(self) -> "pd.DataFrame":
         """
-        Deprecated alias for `to_pandas()`. Please use `to_pandas()` instead.
+        *Deprecated alias for `to_pandas()`. Please use `to_pandas()` instead.*
 
         Execute the query and return the results as a pandas DataFrame.
         In addition to the selected columns, LanceDB also returns a vector
@@ -226,13 +263,20 @@ class LanceQueryBuilder(ABC):
         self._columns = columns
         return self
 
-    def where(self, where) -> LanceQueryBuilder:
+    def where(self, where: str, prefilter: bool = False) -> LanceQueryBuilder:
         """Set the where clause.
 
         Parameters
         ----------
         where: str
-            The where clause.
+            The where clause which is a valid SQL where clause. See
+            `Lance filter pushdown <https://lancedb.github.io/lance/read_and_write.html#filter-push-down>`_
+            for valid SQL expressions.
+        prefilter: bool, default False
+            If True, apply the filter before vector search, otherwise the
+            filter is applied on the result of vector search.
+            This feature is **EXPERIMENTAL** and may be removed and modified
+            without warning in the future.
 
         Returns
         -------
@@ -240,13 +284,12 @@ class LanceQueryBuilder(ABC):
             The LanceQueryBuilder object.
         """
         self._where = where
+        self._prefilter = prefilter
         return self
 
 
 class LanceVectorQueryBuilder(LanceQueryBuilder):
     """
-    A builder for nearest neighbor queries for LanceDB.
-
     Examples
     --------
     >>> import lancedb
@@ -302,7 +345,7 @@ class LanceVectorQueryBuilder(LanceQueryBuilder):
         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
+        See discussion in [Querying an ANN Index][querying-an-ann-index] for
         tuning advice.
 
         Parameters
@@ -369,14 +412,14 @@ class LanceVectorQueryBuilder(LanceQueryBuilder):
         Parameters
         ----------
         where: str
-            The where clause.
+            The where clause which is a valid SQL where clause. See
+            `Lance filter pushdown <https://lancedb.github.io/lance/read_and_write.html#filter-push-down>`_
+            for valid SQL expressions.
         prefilter: bool, default False
             If True, apply the filter before vector search, otherwise the
             filter is applied on the result of vector search.
             This feature is **EXPERIMENTAL** and may be removed and modified
-            without warning in the future. Currently this is only supported
+            without warning in the future.
-            in OSS and can only be used with a table that does not have an ANN
-            index.
 
         Returns
         -------
@@ -389,6 +432,8 @@ class LanceVectorQueryBuilder(LanceQueryBuilder):
 
 
 class LanceFtsQueryBuilder(LanceQueryBuilder):
+    """A builder for full text search for LanceDB."""
+
     def __init__(self, table: "lancedb.table.Table", query: str):
         super().__init__(table)
         self._query = query

python/lancedb/remote/db.py

@@ -104,7 +104,11 @@ class RemoteDBConnection(DBConnection):
             raise ValueError("Either data or schema must be provided.")
         if data is not None:
             data = _sanitize_data(
-                data, schema, on_bad_vectors=on_bad_vectors, fill_value=fill_value
+                data,
+                schema,
+                metadata=None,
+                on_bad_vectors=on_bad_vectors,
+                fill_value=fill_value,
             )
         else:
             if schema is None:

python/lancedb/table.py

@@ -149,13 +149,13 @@ class Table(ABC):
     @property
     @abstractmethod
     def schema(self) -> pa.Schema:
-        """The [Arrow Schema](https://arrow.apache.org/docs/python/api/datatypes.html#) of
+        """The [Arrow Schema](https://arrow.apache.org/docs/python/api/datatypes.html#)
-        this Table
+        of this Table
 
         """
         raise NotImplementedError
 
-    def to_pandas(self):
+    def to_pandas(self) -> "pd.DataFrame":
         """Return the table as a pandas DataFrame.
 
         Returns
@@ -191,17 +191,18 @@ class Table(ABC):
             The distance metric to use when creating the index.
             Valid values are "L2", "cosine", or "dot".
             L2 is euclidean distance.
-        num_partitions: int
+        num_partitions: int, default 256
             The number of IVF partitions to use when creating the index.
             Default is 256.
-        num_sub_vectors: int
+        num_sub_vectors: int, default 96
             The number of PQ sub-vectors to use when creating the index.
             Default is 96.
         vector_column_name: str, default "vector"
             The vector column name to create the index.
         replace: bool, default True
-            If True, replace the existing index if it exists.
+            - If True, replace the existing index if it exists.
-            If False, raise an error if duplicate index exists.
+
+            - If False, raise an error if duplicate index exists.
         accelerator: str, default None
             If set, use the given accelerator to create the index.
             Only support "cuda" for now.
@@ -220,8 +221,14 @@ class Table(ABC):
 
         Parameters
         ----------
-        data: list-of-dict, dict, pd.DataFrame
+        data: DATA
-            The data to insert into the table.
+            The data to insert into the table. Acceptable types are:
+
+            - dict or list-of-dict
+
+            - pandas.DataFrame
+
+            - pyarrow.Table or pyarrow.RecordBatch
         mode: str
             The mode to use when writing the data. Valid values are
             "append" and "overwrite".
@@ -242,31 +249,70 @@ class Table(ABC):
         query_type: str = "auto",
     ) -> LanceQueryBuilder:
         """Create a search query to find the nearest neighbors
-        of the given query vector.
+        of the given query vector. We currently support [vector search][search]
+        and [full-text search][experimental-full-text-search].
+
+        All query options are defined in [Query][lancedb.query.Query].
+
+        Examples
+        --------
+        >>> import lancedb
+        >>> db = lancedb.connect("./.lancedb")
+        >>> data = [
+        ...    {"original_width": 100, "caption": "bar", "vector": [0.1, 2.3, 4.5]},
+        ...    {"original_width": 2000, "caption": "foo",  "vector": [0.5, 3.4, 1.3]},
+        ...    {"original_width": 3000, "caption": "test", "vector": [0.3, 6.2, 2.6]}
+        ... ]
+        >>> table = db.create_table("my_table", data)
+        >>> query = [0.4, 1.4, 2.4]
+        >>> (table.search(query, vector_column_name="vector")
+        ...     .where("original_width > 1000", prefilter=True)
+        ...     .select(["caption", "original_width"])
+        ...     .limit(2)
+        ...     .to_pandas())
+          caption  original_width           vector  _distance
+        0     foo            2000  [0.5, 3.4, 1.3]   5.220000
+        1    test            3000  [0.3, 6.2, 2.6]  23.089996
 
         Parameters
         ----------
-        query: str, list, np.ndarray, PIL.Image.Image, default None
+        query: list/np.ndarray/str/PIL.Image.Image, default None
-            The query to search for. If None then
+            The targetted vector to search for.
-            the select/where/limit clauses are applied to filter
+
+            - *default None*.
+            Acceptable types are: list, np.ndarray, PIL.Image.Image
+
+            - If None then the select/where/limit clauses are applied to filter
             the table
-        vector_column_name: str, default "vector"
+        vector_column_name: str
             The name of the vector column to search.
-        query_type: str, default "auto"
+            *default "vector"*
-            "vector", "fts", or "auto"
+        query_type: str
-            If "auto" then the query type is inferred from the query;
+            *default "auto"*.
-            If `query` is a list/np.ndarray then the query type is "vector";
+            Acceptable types are: "vector", "fts", or "auto"
-            If `query` is a PIL.Image.Image then either do vector search
+
-            or raise an error if no corresponding embedding function is found.
+            - If "auto" then the query type is inferred from the query;
-            If `query` is a string, then the query type is "vector" if the
+
+                - If `query` is a list/np.ndarray then the query type is
+                "vector";
+
+                - If `query` is a PIL.Image.Image then either do vector search,
+                or raise an error if no corresponding embedding function is found.
+
+            - If `query` is a string, then the query type is "vector" if the
             table has embedding functions else the query type is "fts"
 
         Returns
         -------
         LanceQueryBuilder
             A query builder object representing the query.
-            Once executed, the query returns selected columns, the vector,
+            Once executed, the query returns
-            and also the "_distance" column which is the distance between the query
+
+            - selected columns
+
+            - the vector
+
+            - and also the "_distance" column which is the distance between the query
             vector and the returned vector.
         """
         raise NotImplementedError
@@ -285,14 +331,19 @@ class Table(ABC):
         Parameters
         ----------
         where: str
-            The SQL where clause to use when deleting rows. For example, 'x = 2'
+            The SQL where clause to use when deleting rows.
-            or 'x IN (1, 2, 3)'. The filter must not be empty, or it will error.
+
+            - For example, 'x = 2' or 'x IN (1, 2, 3)'.
+
+            The filter must not be empty, or it will error.
 
         Examples
         --------
         >>> import lancedb
         >>> data = [
-        ...   {"x": 1, "vector": [1, 2]}, {"x": 2, "vector": [3, 4]}, {"x": 3, "vector": [5, 6]}
+        ...    {"x": 1, "vector": [1, 2]},
+        ...    {"x": 2, "vector": [3, 4]},
+        ...    {"x": 3, "vector": [5, 6]}
         ... ]
         >>> db = lancedb.connect("./.lancedb")
         >>> table = db.create_table("my_table", data)
@@ -377,7 +428,8 @@ class LanceTable(Table):
         --------
         >>> import lancedb
         >>> db = lancedb.connect("./.lancedb")
-        >>> table = db.create_table("my_table", [{"vector": [1.1, 0.9], "type": "vector"}])
+        >>> table = db.create_table("my_table",
+        ...    [{"vector": [1.1, 0.9], "type": "vector"}])
         >>> table.version
         2
         >>> table.to_pandas()
@@ -424,7 +476,8 @@ class LanceTable(Table):
         --------
         >>> import lancedb
         >>> db = lancedb.connect("./.lancedb")
-        >>> table = db.create_table("my_table", [{"vector": [1.1, 0.9], "type": "vector"}])
+        >>> table = db.create_table("my_table", [
+        ...     {"vector": [1.1, 0.9], "type": "vector"}])
         >>> table.version
         2
         >>> table.to_pandas()
@@ -669,14 +722,39 @@ class LanceTable(Table):
         query_type: str = "auto",
     ) -> LanceQueryBuilder:
         """Create a search query to find the nearest neighbors
-        of the given query vector.
+        of the given query vector. We currently support [vector search][search]
+        and [full-text search][search].
+
+        Examples
+        --------
+        >>> import lancedb
+        >>> db = lancedb.connect("./.lancedb")
+        >>> data = [
+        ...    {"original_width": 100, "caption": "bar", "vector": [0.1, 2.3, 4.5]},
+        ...    {"original_width": 2000, "caption": "foo",  "vector": [0.5, 3.4, 1.3]},
+        ...    {"original_width": 3000, "caption": "test", "vector": [0.3, 6.2, 2.6]}
+        ... ]
+        >>> table = db.create_table("my_table", data)
+        >>> query = [0.4, 1.4, 2.4]
+        >>> (table.search(query, vector_column_name="vector")
+        ...     .where("original_width > 1000", prefilter=True)
+        ...     .select(["caption", "original_width"])
+        ...     .limit(2)
+        ...     .to_pandas())
+          caption  original_width           vector  _distance
+        0     foo            2000  [0.5, 3.4, 1.3]   5.220000
+        1    test            3000  [0.3, 6.2, 2.6]  23.089996
 
         Parameters
         ----------
-        query: str, list, np.ndarray, a PIL Image or None
+        query: list/np.ndarray/str/PIL.Image.Image, default None
-            The query to search for. If None then
+            The targetted vector to search for.
-            the select/where/limit clauses are applied to filter
+
-            the table
+            - *default None*.
+            Acceptable types are: list, np.ndarray, PIL.Image.Image
+
+            - If None then the select/[where][sql]/limit clauses are applied
+            to filter the table
         vector_column_name: str, default "vector"
             The name of the vector column to search.
         query_type: str, default "auto"
@@ -685,7 +763,7 @@ class LanceTable(Table):
             If `query` is a list/np.ndarray then the query type is "vector";
             If `query` is a PIL.Image.Image then either do vector search
             or raise an error if no corresponding embedding function is found.
-            If the query is a string, then the query type is "vector" if the
+            If the `query` is a string, then the query type is "vector" if the
             table has embedding functions, else the query type is "fts"
 
         Returns
@@ -720,7 +798,9 @@ class LanceTable(Table):
         --------
         >>> import lancedb
         >>> data = [
-        ...   {"x": 1, "vector": [1, 2]}, {"x": 2, "vector": [3, 4]}, {"x": 3, "vector": [5, 6]}
+        ...    {"x": 1, "vector": [1, 2]},
+        ...    {"x": 2, "vector": [3, 4]},
+        ...    {"x": 3, "vector": [5, 6]}
         ... ]
         >>> db = lancedb.connect("./.lancedb")
         >>> table = db.create_table("my_table", data)
@@ -740,7 +820,8 @@ class LanceTable(Table):
             The data to insert into the table.
             At least one of `data` or `schema` must be provided.
         schema: pa.Schema or LanceModel, optional
-            The schema of the table. If not provided, the schema is inferred from the data.
+            The schema of the table. If not provided,
+            the schema is inferred from the data.
             At least one of `data` or `schema` must be provided.
         mode: str, default "create"
             The mode to use when writing the data. Valid values are
@@ -811,7 +892,8 @@ class LanceTable(Table):
         file_info = fs.get_file_info(path)
         if file_info.type != pa.fs.FileType.Directory:
             raise FileNotFoundError(
-                f"Table {name} does not exist. Please first call db.create_table({name}, data)"
+                f"Table {name} does not exist."
+                f"Please first call db.create_table({name}, data)"
             )
         return tbl
 
@@ -838,7 +920,9 @@ class LanceTable(Table):
         --------
         >>> import lancedb
         >>> data = [
-        ...   {"x": 1, "vector": [1, 2]}, {"x": 2, "vector": [3, 4]}, {"x": 3, "vector": [5, 6]}
+        ...    {"x": 1, "vector": [1, 2]},
+        ...    {"x": 2, "vector": [3, 4]},
+        ...    {"x": 3, "vector": [5, 6]}
         ... ]
         >>> db = lancedb.connect("./.lancedb")
         >>> table = db.create_table("my_table", data)
@@ -872,12 +956,6 @@ class LanceTable(Table):
 
     def _execute_query(self, query: Query) -> pa.Table:
         ds = self.to_lance()
-        if query.prefilter:
-            for idx in ds.list_indices():
-                if query.vector_column in idx["fields"]:
-                    raise NotImplementedError(
-                        "Prefiltering for indexed vector column is coming soon."
-                    )
         return ds.to_table(
             columns=query.columns,
             filter=query.filter,
@@ -1019,7 +1097,8 @@ def _sanitize_vector_column(
     # ChunkedArray is annoying to work with, so we combine chunks here
     vec_arr = data[vector_column_name].combine_chunks()
     if pa.types.is_list(data[vector_column_name].type):
-        # if it's a variable size list array we make sure the dimensions are all the same
+        # if it's a variable size list array,
+        # we make sure the dimensions are all the same
         has_jagged_ndims = len(vec_arr.values) % len(data) != 0
         if has_jagged_ndims:
             data = _sanitize_jagged(

python/lancedb/utils/sentry_log.py

@@ -63,7 +63,8 @@ def set_sentry():
         """
         if "exc_info" in hint:
             exc_type, exc_value, tb = hint["exc_info"]
-            if "out of memory" in str(exc_value).lower():
+            ignored_errors = ["out of memory", "no space left on device", "testing"]
+            if any(error in str(exc_value).lower() for error in ignored_errors):
                 return None
 
         if is_git_dir():
@@ -97,7 +98,7 @@ def set_sentry():
             dsn="https://c63ef8c64e05d1aa1a96513361f3ca2f@o4505950840946688.ingest.sentry.io/4505950933614592",
             debug=False,
             include_local_variables=False,
-            traces_sample_rate=1.0,
+            traces_sample_rate=0.5,
             environment="production",  # 'dev' or 'production'
             before_send=before_send,
             ignore_errors=[KeyboardInterrupt, FileNotFoundError, bdb.BdbQuit],

python/pyproject.toml

@@ -1,9 +1,9 @@
 [project]
 name = "lancedb"
-version = "0.3.2"
+version = "0.3.3"
 dependencies = [
     "deprecation",
-    "pylance==0.8.7",
+    "pylance==0.8.10",
     "ratelimiter~=1.0",
     "retry>=0.9.2",
     "tqdm>=4.1.0",

python/tests/test_db.py

@@ -150,6 +150,21 @@ def test_ingest_iterator(tmp_path):
     run_tests(PydanticSchema)
 
 
+def test_table_names(tmp_path):
+    db = lancedb.connect(tmp_path)
+    data = pd.DataFrame(
+        {
+            "vector": [[3.1, 4.1], [5.9, 26.5]],
+            "item": ["foo", "bar"],
+            "price": [10.0, 20.0],
+        }
+    )
+    db.create_table("test2", data=data)
+    db.create_table("test1", data=data)
+    db.create_table("test3", data=data)
+    assert db.table_names() == ["test1", "test2", "test3"]
+
+
 def test_create_mode(tmp_path):
     db = lancedb.connect(tmp_path)
     data = pd.DataFrame(
@@ -287,3 +302,27 @@ def test_replace_index(tmp_path):
         num_sub_vectors=4,
         replace=True,
     )
+
+
+def test_prefilter_with_index(tmp_path):
+    db = lancedb.connect(uri=tmp_path)
+    data = [
+        {"vector": np.random.rand(128), "item": "foo", "price": float(i)}
+        for i in range(1000)
+    ]
+    sample_key = data[100]["vector"]
+    table = db.create_table(
+        "test",
+        data,
+    )
+    table.create_index(
+        num_partitions=2,
+        num_sub_vectors=4,
+    )
+    table = (
+        table.search(sample_key)
+        .where("price == 500", prefilter=True)
+        .limit(5)
+        .to_arrow()
+    )
+    assert table.num_rows == 1

rust/ffi/node/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb-node"
-version = "0.3.3"
+version = "0.3.5"
 description = "Serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"
 edition = "2018"

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

@@ -70,7 +70,6 @@ fn get_index_params_builder(
                 .map(|mt| {
                     let metric_type = mt.unwrap();
                     index_builder.metric_type(metric_type);
-                    pq_params.metric_type = metric_type;
                 });
 
             let num_partitions = obj.get_opt_usize(cx, "num_partitions")?;

rust/ffi/node/src/lib.rs

@@ -239,6 +239,8 @@ fn main(mut cx: ModuleContext) -> NeonResult<()> {
     cx.export_function("tableDelete", JsTable::js_delete)?;
     cx.export_function("tableCleanupOldVersions", JsTable::js_cleanup)?;
     cx.export_function("tableCompactFiles", JsTable::js_compact)?;
+    cx.export_function("tableListIndices", JsTable::js_list_indices)?;
+    cx.export_function("tableIndexStats", JsTable::js_index_stats)?;
     cx.export_function(
         "tableCreateVectorIndex",
         index::vector::table_create_vector_index,

rust/ffi/node/src/table.rs

@@ -247,7 +247,7 @@ impl JsTable {
         }
 
         rt.spawn(async move {
-            let stats = table.compact_files(options).await;
+            let stats = table.compact_files(options, None).await;
 
             deferred.settle_with(&channel, move |mut cx| {
                 let stats = stats.or_throw(&mut cx)?;
@@ -276,4 +276,91 @@ impl JsTable {
         });
         Ok(promise)
     }
+
+    pub(crate) fn js_list_indices(mut cx: FunctionContext) -> JsResult<JsPromise> {
+        let js_table = cx.this().downcast_or_throw::<JsBox<JsTable>, _>(&mut cx)?;
+        let rt = runtime(&mut cx)?;
+        let (deferred, promise) = cx.promise();
+        // let predicate = cx.argument::<JsString>(0)?.value(&mut cx);
+        let channel = cx.channel();
+        let table = js_table.table.clone();
+
+        rt.spawn(async move {
+            let indices = table.load_indices().await;
+
+            deferred.settle_with(&channel, move |mut cx| {
+                let indices = indices.or_throw(&mut cx)?;
+
+                let output = JsArray::new(&mut cx, indices.len() as u32);
+                for (i, index) in indices.iter().enumerate() {
+                    let js_index = JsObject::new(&mut cx);
+                    let index_name = cx.string(index.index_name.clone());
+                    js_index.set(&mut cx, "name", index_name)?;
+
+                    let index_uuid = cx.string(index.index_uuid.clone());
+                    js_index.set(&mut cx, "uuid", index_uuid)?;
+
+                    let js_index_columns = JsArray::new(&mut cx, index.columns.len() as u32);
+                    for (j, column) in index.columns.iter().enumerate() {
+                        let js_column = cx.string(column.clone());
+                        js_index_columns.set(&mut cx, j as u32, js_column)?;
+                    }
+                    js_index.set(&mut cx, "columns", js_index_columns)?;
+
+                    output.set(&mut cx, i as u32, js_index)?;
+                }
+
+                Ok(output)
+            })
+        });
+        Ok(promise)
+    }
+
+    pub(crate) fn js_index_stats(mut cx: FunctionContext) -> JsResult<JsPromise> {
+        let js_table = cx.this().downcast_or_throw::<JsBox<JsTable>, _>(&mut cx)?;
+        let rt = runtime(&mut cx)?;
+        let (deferred, promise) = cx.promise();
+        let index_uuid = cx.argument::<JsString>(0)?.value(&mut cx);
+        let channel = cx.channel();
+        let table = js_table.table.clone();
+
+        rt.spawn(async move {
+            let load_stats = futures::try_join!(
+                table.count_indexed_rows(&index_uuid),
+                table.count_unindexed_rows(&index_uuid)
+            );
+
+            deferred.settle_with(&channel, move |mut cx| {
+                let (indexed_rows, unindexed_rows) = load_stats.or_throw(&mut cx)?;
+
+                let output = JsObject::new(&mut cx);
+
+                match indexed_rows {
+                    Some(x) => {
+                        let i = cx.number(x as f64);
+                        output.set(&mut cx, "numIndexedRows", i)?;
+                    }
+                    None => {
+                        let null = cx.null();
+                        output.set(&mut cx, "numIndexedRows", null)?;
+                    }
+                };
+
+                match unindexed_rows {
+                    Some(x) => {
+                        let i = cx.number(x as f64);
+                        output.set(&mut cx, "numUnindexedRows", i)?;
+                    }
+                    None => {
+                        let null = cx.null();
+                        output.set(&mut cx, "numUnindexedRows", null)?;
+                    }
+                };
+
+                Ok(output)
+            })
+        });
+
+        Ok(promise)
+    }
 }

rust/vectordb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb"
-version = "0.3.3"
+version = "0.3.5"
 edition = "2021"
 description = "LanceDB: A serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"

rust/vectordb/src/database.rs

@@ -161,7 +161,7 @@ impl Database {
     ///
     /// * A [Vec<String>] with all table names.
     pub async fn table_names(&self) -> Result<Vec<String>> {
-        let f = self
+        let mut f = self
             .object_store
             .read_dir(self.base_path.clone())
             .await?
@@ -175,7 +175,8 @@ impl Database {
                 is_lance.unwrap_or(false)
             })
             .filter_map(|p| p.file_stem().and_then(|s| s.to_str().map(String::from)))
-            .collect();
+            .collect::<Vec<String>>();
+        f.sort();
         Ok(f)
     }
 
@@ -312,8 +313,8 @@ mod tests {
         let db = Database::connect(uri).await.unwrap();
         let tables = db.table_names().await.unwrap();
         assert_eq!(tables.len(), 2);
-        assert!(tables.contains(&String::from("table1")));
+        assert!(tables[0].eq(&String::from("table1")));
-        assert!(tables.contains(&String::from("table2")));
+        assert!(tables[1].eq(&String::from("table2")));
     }
 
     #[tokio::test]

rust/vectordb/src/index/vector.rs

@@ -12,6 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use lance::format::{Index, Manifest};
 use lance::index::vector::ivf::IvfBuildParams;
 use lance::index::vector::pq::PQBuildParams;
 use lance::index::vector::VectorIndexParams;
@@ -98,7 +99,11 @@ impl VectorIndexBuilder for IvfPQIndexBuilder {
         let ivf_params = self.ivf_params.clone().unwrap_or_default();
         let pq_params = self.pq_params.clone().unwrap_or_default();
 
-        VectorIndexParams::with_ivf_pq_params(pq_params.metric_type, ivf_params, pq_params)
+        VectorIndexParams::with_ivf_pq_params(
+            self.metric_type.unwrap_or(MetricType::L2),
+            ivf_params,
+            pq_params,
+        )
     }
 
     fn get_replace(&self) -> bool {
@@ -106,6 +111,27 @@ impl VectorIndexBuilder for IvfPQIndexBuilder {
     }
 }
 
+pub struct VectorIndex {
+    pub columns: Vec<String>,
+    pub index_name: String,
+    pub index_uuid: String,
+}
+
+impl VectorIndex {
+    pub fn new_from_format(manifest: &Manifest, index: &Index) -> VectorIndex {
+        let fields = index
+            .fields
+            .iter()
+            .map(|i| manifest.schema.fields[*i as usize].name.clone())
+            .collect();
+        VectorIndex {
+            columns: fields,
+            index_name: index.name.clone(),
+            index_uuid: index.uuid.to_string(),
+        }
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -158,7 +184,6 @@ mod tests {
         pq_params.max_iters = 1;
         pq_params.num_bits = 8;
         pq_params.num_sub_vectors = 50;
-        pq_params.metric_type = MetricType::Cosine;
         pq_params.max_opq_iters = 2;
         index_builder.ivf_params(ivf_params);
         index_builder.pq_params(pq_params);
@@ -176,7 +201,6 @@ mod tests {
             assert_eq!(pq_params.max_iters, 1);
             assert_eq!(pq_params.num_bits, 8);
             assert_eq!(pq_params.num_sub_vectors, 50);
-            assert_eq!(pq_params.metric_type, MetricType::Cosine);
             assert_eq!(pq_params.max_opq_iters, 2);
         } else {
             assert!(false, "Expected second stage to be pq")

rust/vectordb/src/table.rs

@@ -18,14 +18,16 @@ use std::sync::Arc;
 use arrow_array::{Float32Array, RecordBatchReader};
 use arrow_schema::SchemaRef;
 use lance::dataset::cleanup::RemovalStats;
-use lance::dataset::optimize::{compact_files, CompactionMetrics, CompactionOptions};
+use lance::dataset::optimize::{
+    compact_files, CompactionMetrics, CompactionOptions, IndexRemapperOptions,
+};
 use lance::dataset::{Dataset, WriteParams};
-use lance::index::IndexType;
+use lance::index::{DatasetIndexExt, IndexType};
 use lance::io::object_store::WrappingObjectStore;
 use std::path::Path;
 
 use crate::error::{Error, Result};
-use crate::index::vector::VectorIndexBuilder;
+use crate::index::vector::{VectorIndex, VectorIndexBuilder};
 use crate::query::Query;
 use crate::utils::{PatchReadParam, PatchWriteParam};
 use crate::WriteMode;
@@ -238,8 +240,6 @@ impl Table {
 
     /// Create index on the table.
     pub async fn create_index(&mut self, index_builder: &impl VectorIndexBuilder) -> Result<()> {
-        use lance::index::DatasetIndexExt;
-
         let mut dataset = self.dataset.as_ref().clone();
         dataset
             .create_index(
@@ -257,6 +257,14 @@ impl Table {
         Ok(())
     }
 
+    pub async fn optimize_indices(&mut self) -> Result<()> {
+        let mut dataset = self.dataset.as_ref().clone();
+
+        dataset.optimize_indices().await?;
+
+        Ok(())
+    }
+
     /// Insert records into this Table
     ///
     /// # Arguments
@@ -353,12 +361,45 @@ impl Table {
     /// for faster reads.
     ///
     /// This calls into [lance::dataset::optimize::compact_files].
-    pub async fn compact_files(&mut self, options: CompactionOptions) -> Result<CompactionMetrics> {
+    pub async fn compact_files(
+        &mut self,
+        options: CompactionOptions,
+        remap_options: Option<Arc<dyn IndexRemapperOptions>>,
+    ) -> Result<CompactionMetrics> {
         let mut dataset = self.dataset.as_ref().clone();
-        let metrics = compact_files(&mut dataset, options, None).await?;
+        let metrics = compact_files(&mut dataset, options, remap_options).await?;
         self.dataset = Arc::new(dataset);
         Ok(metrics)
     }
+
+    pub fn count_fragments(&self) -> usize {
+        self.dataset.count_fragments()
+    }
+
+    pub fn count_deleted_rows(&self) -> usize {
+        self.dataset.count_deleted_rows()
+    }
+
+    pub fn num_small_files(&self, max_rows_per_group: usize) -> usize {
+        self.dataset.num_small_files(max_rows_per_group)
+    }
+
+    pub async fn count_indexed_rows(&self, index_uuid: &str) -> Result<Option<usize>> {
+        Ok(self.dataset.count_indexed_rows(index_uuid).await?)
+    }
+
+    pub async fn count_unindexed_rows(&self, index_uuid: &str) -> Result<Option<usize>> {
+        Ok(self.dataset.count_unindexed_rows(index_uuid).await?)
+    }
+
+    pub async fn load_indices(&self) -> Result<Vec<VectorIndex>> {
+        let (indices, mf) =
+            futures::try_join!(self.dataset.load_indices(), self.dataset.latest_manifest())?;
+        Ok(indices
+            .iter()
+            .map(|i| VectorIndex::new_from_format(&mf, i))
+            .collect())
+    }
 }
 
 #[cfg(test)]