[python] Bump version: 0.1.15 → 0.1.16

- Fixes many numeric conversions that results in hard to reproduce issues
- JsObjectExt extends JsObject with safe methods to extract numericvalues

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.1.18
+current_version = 0.1.19
 commit = True
 message = Bump version: {current_version} → {new_version}
 tag = True

.github/workflows/python.yml

@@ -30,7 +30,7 @@ jobs:
         python-version: 3.${{ matrix.python-minor-version }}
     - name: Install lancedb
       run: |
-        pip install -e .
+        pip install -e .[tests]
         pip install tantivy@git+https://github.com/quickwit-oss/tantivy-py#164adc87e1a033117001cf70e38c82a53014d985
         pip install pytest pytest-mock black isort
     - name: Black
@@ -59,7 +59,7 @@ jobs:
         python-version: "3.11"
     - name: Install lancedb
       run: |
-        pip install -e .
+        pip install -e .[tests]
         pip install tantivy@git+https://github.com/quickwit-oss/tantivy-py#164adc87e1a033117001cf70e38c82a53014d985
         pip install pytest pytest-mock black
     - name: Black

Cargo.toml

@@ -13,4 +13,5 @@ arrow-schema = "42.0"
 arrow-ipc = "42.0"
 half = { "version" = "=2.2.1", default-features = false }
 object_store = "0.6.1"
+snafu = "0.7.4"
 

docs/mkdocs.yml

@@ -57,12 +57,14 @@ nav:
 - Basics: basic.md
 - Embeddings: embedding.md
 - Python full-text search: fts.md
-- Python integrations:
+- Integrations:
   - Pandas and PyArrow: python/arrow.md
   - DuckDB: python/duckdb.md
   - LangChain 🦜️🔗: https://python.langchain.com/en/latest/modules/indexes/vectorstores/examples/lancedb.html
+  - LangChain JS/TS 🦜️🔗: https://js.langchain.com/docs/modules/data_connection/vectorstores/integrations/lancedb
   - LlamaIndex 🦙: https://gpt-index.readthedocs.io/en/latest/examples/vector_stores/LanceDBIndexDemo.html
   - Pydantic: python/pydantic.md
+  - Voxel51: integrations/voxel51.md
 - Python examples:
   - YouTube Transcript Search: notebooks/youtube_transcript_search.ipynb
   - Documentation QA Bot using LangChain: notebooks/code_qa_bot.ipynb
@@ -72,6 +74,7 @@ nav:
 - Javascript examples:
   - YouTube Transcript Search: examples/youtube_transcript_bot_with_nodejs.md
   - TransformersJS Embedding Search: examples/transformerjs_embedding_search_nodejs.md
+
 - References:
   - Vector Search: search.md
   - SQL filters: sql.md

docs/src/examples/youtube_transcript_bot.md

@@ -4,4 +4,10 @@
 
 <img id="splash" width="400" alt="youtube transcript search" src="https://user-images.githubusercontent.com/917119/236965568-def7394d-171c-45f2-939d-8edfeaadd88c.png">
 
+
+<a href="https://colab.research.google.com/github/lancedb/vectordb-recipes/blob/main/examples/youtube_bot/main.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab">
+
+Scripts - [![Python](https://img.shields.io/badge/python-3670A0?style=for-the-badge&logo=python&logoColor=ffdd54)](./examples/youtube_bot/main.py)  [![JavaScript](https://img.shields.io/badge/javascript-%23323330.svg?style=for-the-badge&logo=javascript&logoColor=%23F7DF1E)](./examples/youtube_bot/index.js)
+
+
 This example is in a [notebook](https://github.com/lancedb/lancedb/blob/main/docs/src/notebooks/youtube_transcript_search.ipynb)

docs/src/integrations/voxel51.md

@@ -0,0 +1,71 @@
+![example](/assets/voxel.gif)
+
+Basic recipe
+____________
+
+The basic workflow to use LanceDB to create a similarity index on your FiftyOne
+datasets and use this to query your data is as follows:
+
+1)  Load a dataset into FiftyOne
+
+2)  Compute embedding vectors for samples or patches in your dataset, or select
+    a model to use to generate embeddings
+
+3)  Use the `compute_similarity()`
+    method to generate a LanceDB table for the samples or object
+    patches embeddings in a dataset by setting the parameter `backend="lancedb"` and
+    specifying a `brain_key` of your choice
+
+4)  Use this LanceDB table to query your data with
+    `sort_by_similarity()`
+
+5) If desired, delete the table
+
+The example below demonstrates this workflow.
+
+!!! Note
+
+    You must install the LanceDB Python client to run this
+    ```
+    pip install lancedb
+    ```
+
+```python
+
+import fiftyone as fo
+import fiftyone.brain as fob
+import fiftyone.zoo as foz
+
+# Step 1: Load your data into FiftyOne
+dataset = foz.load_zoo_dataset("quickstart")
+
+# Steps 2 and 3: Compute embeddings and create a similarity index
+lancedb_index = fob.compute_similarity(
+    dataset, 
+    model="clip-vit-base32-torch",
+    brain_key="lancedb_index",
+    backend="lancedb",
+)
+```
+Once the similarity index has been generated, we can query our data in FiftyOne
+by specifying the `brain_key`:
+
+```python
+# Step 4: Query your data
+query = dataset.first().id  # query by sample ID
+view = dataset.sort_by_similarity(
+    query, 
+    brain_key="lancedb_index",
+    k=10,  # limit to 10 most similar samples
+)
+
+# Step 5 (optional): Cleanup
+
+# Delete the LanceDB table
+lancedb_index.cleanup()
+
+# Delete run record from FiftyOne
+dataset.delete_brain_run("lancedb_index")
+```
+
+More in depth walkthrough of the integration, visit the LanceDB guide on Voxel51 - [LaceDB x Voxel51](https://docs.voxel51.com/integrations/lancedb.html)

docs/src/notebooks/code_qa_bot.ipynb

@@ -10,7 +10,11 @@
     "\n",
     "This Q&A bot will allow you to query your own documentation easily using questions. We'll also demonstrate the use of LangChain and LanceDB using the OpenAI API. \n",
     "\n",
-    "In this example we'll use Pandas 2.0 documentation, but, this could be replaced for your own docs as well"
+    "In this example we'll use Pandas 2.0 documentation, but, this could be replaced for your own docs as well\n",
+    "\n",
+    "<a href=\"https://colab.research.google.com/github/lancedb/vectordb-recipes/blob/main/examples/Code-Documentation-QA-Bot/main.ipynb\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"></a>\n",
+    "\n",
+    "Scripts -  [![Python](https://img.shields.io/badge/python-3670A0?style=for-the-badge&logo=python&logoColor=ffdd54)](./examples/Code-Documentation-QA-Bot/main.py)  [![JavaScript](https://img.shields.io/badge/javascript-%23323330.svg?style=for-the-badge&logo=javascript&logoColor=%23F7DF1E)](./examples/Code-Documentation-QA-Bot/index.js)"
    ]
   },
   {

docs/src/notebooks/multimodal_search.ipynb

@@ -1,5 +1,14 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "![example](https://github.com/lancedb/vectordb-recipes/assets/15766192/799f94a1-a01d-4a5b-a627-2a733bbb4227)\n",
+    "\n",
+    " <a href=\"https://colab.research.google.com/github/lancedb/vectordb-recipes/blob/main/examples/multimodal_clip/main.ipynb\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"></a>| [![Python](https://img.shields.io/badge/python-3670A0?style=for-the-badge&logo=python&logoColor=ffdd54)](./examples/multimodal_clip/main.py)  |"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 2,
@@ -42,6 +51,19 @@
     "## First run setup: Download data and pre-process"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "### Get dataset\n",
+    "\n",
+    "!wget https://eto-public.s3.us-west-2.amazonaws.com/datasets/diffusiondb_lance.tar.gz\n",
+    "!tar -xvf diffusiondb_lance.tar.gz\n",
+    "!mv diffusiondb_test rawdata.lance\n"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 30,
@@ -247,7 +269,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "Python 3.11.4 64-bit",
    "language": "python",
    "name": "python3"
   },
@@ -261,7 +283,12 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.11.3"
+   "version": "3.11.4"
+  },
+  "vscode": {
+   "interpreter": {
+    "hash": "b0fa6594d8f4cbf19f97940f81e996739fb7646882a419484c72d19e05852a7e"
+   }
   }
  },
  "nbformat": 4,

docs/src/notebooks/youtube_transcript_search.ipynb

@@ -8,7 +8,12 @@
    "source": [
     "# 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."
+    "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.\n",
+    "\n",
+    "\n",
+    "<a href=\"https://colab.research.google.com/github/lancedb/vectordb-recipes/blob/main/examples/youtube_bot/main.ipynb\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\">\n",
+    "\n",
+    "Scripts - [![Python](https://img.shields.io/badge/python-3670A0?style=for-the-badge&logo=python&logoColor=ffdd54)](./examples/youtube_bot/main.py)  [![JavaScript](https://img.shields.io/badge/javascript-%23323330.svg?style=for-the-badge&logo=javascript&logoColor=%23F7DF1E)](./examples/youtube_bot/index.js)\n"
    ]
   },
   {

docs/src/python/pydantic.md

@@ -1,6 +1,8 @@
 # Pydantic
 
 [Pydantic](https://docs.pydantic.dev/latest/) is a data validation library in Python.
+LanceDB integrates with Pydantic for schema inference, data ingestion, and query result casting.
+
 
 ## Schema
 

docs/test/md_testing.py

@@ -7,7 +7,8 @@ excluded_files = [
     "../src/embedding.md",
     "../src/examples/serverless_lancedb_with_s3_and_lambda.md",
     "../src/examples/serverless_qa_bot_with_modal_and_langchain.md",
-    "../src/examples/youtube_transcript_bot_with_nodejs.md"
+    "../src/examples/youtube_transcript_bot_with_nodejs.md",
+    "../src/integrations/voxel51.md",
 ]
 
 python_prefix = "py"

node/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "vectordb",
-  "version": "0.1.18",
+  "version": "0.1.19",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "vectordb",
-      "version": "0.1.18",
+      "version": "0.1.19",
       "cpu": [
         "x64",
         "arm64"
@@ -51,11 +51,11 @@
         "typescript": "*"
       },
       "optionalDependencies": {
-        "@lancedb/vectordb-darwin-arm64": "0.1.18",
+        "@lancedb/vectordb-darwin-arm64": "0.1.19",
-        "@lancedb/vectordb-darwin-x64": "0.1.18",
+        "@lancedb/vectordb-darwin-x64": "0.1.19",
-        "@lancedb/vectordb-linux-arm64-gnu": "0.1.18",
+        "@lancedb/vectordb-linux-arm64-gnu": "0.1.19",
-        "@lancedb/vectordb-linux-x64-gnu": "0.1.18",
+        "@lancedb/vectordb-linux-x64-gnu": "0.1.19",
-        "@lancedb/vectordb-win32-x64-msvc": "0.1.18"
+        "@lancedb/vectordb-win32-x64-msvc": "0.1.19"
       }
     },
     "node_modules/@apache-arrow/ts": {
@@ -315,9 +315,9 @@
       }
     },
     "node_modules/@lancedb/vectordb-darwin-arm64": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.1.19.tgz",
-      "integrity": "sha512-vu8MCFgaAAGmTJF+4RaoApROMpRVVgrCk+V9my4adAfWkkXbSmtxiDgiIwwL1VqdGb8UwzGn3kVbNW7idE1ojA==",
+      "integrity": "sha512-efQhJkBKvMNhjFq3Sw3/qHo9D9gb9UqiIr98n3STsbNxBQjMnWemXn91Ckl40siRG1O8qXcINW7Qs/EGmus+kg==",
       "cpu": [
         "arm64"
       ],
@@ -327,9 +327,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-darwin-x64": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.1.19.tgz",
-      "integrity": "sha512-ZU30bd6frRyKJ515ow972PlqO2wIiNT4Ohor9+KbUwl/VKDyAwKOKG8cWhRJXTxk0k1oqpiJ6+Q28TcYJ0sSAw==",
+      "integrity": "sha512-r6OZNVyemAssABz2w7CRhe7dyREwBEfTytn+ux1zzTnzsgMgDovCQ0rQ3WZcxWvcy7SFCxiemA9IP1b/lsb4tQ==",
       "cpu": [
         "x64"
       ],
@@ -339,9 +339,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-linux-arm64-gnu": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.1.19.tgz",
-      "integrity": "sha512-2UroC026bUYwyciSRonYlXei0SoYbKgfWpozxYOu7GgBAV2CQQtaAPgWJTEl6ZiCNeBmBTx+j0h3+ydUfZA73Q==",
+      "integrity": "sha512-mL/hRmZp6Kw7hmGJBdOZfp/tTYiCdlOcs8DA/+nr2eiXERv0gIhyiKvr2P5DwbBmut3qXEkDalMHTo95BSdL2A==",
       "cpu": [
         "arm64"
       ],
@@ -351,9 +351,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-linux-x64-gnu": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.1.19.tgz",
-      "integrity": "sha512-DoQBskl22JAJFZh219ZOJ6o+f1niTZp0qRYngHa/kTIpLKzHWQ0OTtMCz32VBAjAsKjSLNxHE8rrT/S6tvS7KQ==",
+      "integrity": "sha512-AG0FHksbbr+cHVKPi4B8cmBtqb6T9E0uaK4kyZkXrX52/xtv9RYVZcykaB/tSSm0XNFPWWRnx9R8UqNZV/hxMA==",
       "cpu": [
         "x64"
       ],
@@ -363,9 +363,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-win32-x64-msvc": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.1.19.tgz",
-      "integrity": "sha512-a/kUM3V6rWuXS80pPECYxKfCUAnq56Of/GPCvnAkpk9C9ldyX10iff4aA6DiPHjEk9V2ytqDfJKl9N3QcMLKLA==",
+      "integrity": "sha512-PDWZ2hvLVXH4Z4WIO1rsWY8ev3NpNm7aXlaey32P+l1Iz9Hia9+F2GBpp2UiEQKfvbk82ucAvBLRmpSsHY8Tlw==",
       "cpu": [
         "x64"
       ],
@@ -4852,33 +4852,33 @@
       }
     },
     "@lancedb/vectordb-darwin-arm64": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.1.19.tgz",
-      "integrity": "sha512-vu8MCFgaAAGmTJF+4RaoApROMpRVVgrCk+V9my4adAfWkkXbSmtxiDgiIwwL1VqdGb8UwzGn3kVbNW7idE1ojA==",
+      "integrity": "sha512-efQhJkBKvMNhjFq3Sw3/qHo9D9gb9UqiIr98n3STsbNxBQjMnWemXn91Ckl40siRG1O8qXcINW7Qs/EGmus+kg==",
       "optional": true
     },
     "@lancedb/vectordb-darwin-x64": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.1.19.tgz",
-      "integrity": "sha512-ZU30bd6frRyKJ515ow972PlqO2wIiNT4Ohor9+KbUwl/VKDyAwKOKG8cWhRJXTxk0k1oqpiJ6+Q28TcYJ0sSAw==",
+      "integrity": "sha512-r6OZNVyemAssABz2w7CRhe7dyREwBEfTytn+ux1zzTnzsgMgDovCQ0rQ3WZcxWvcy7SFCxiemA9IP1b/lsb4tQ==",
       "optional": true
     },
     "@lancedb/vectordb-linux-arm64-gnu": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.1.19.tgz",
-      "integrity": "sha512-2UroC026bUYwyciSRonYlXei0SoYbKgfWpozxYOu7GgBAV2CQQtaAPgWJTEl6ZiCNeBmBTx+j0h3+ydUfZA73Q==",
+      "integrity": "sha512-mL/hRmZp6Kw7hmGJBdOZfp/tTYiCdlOcs8DA/+nr2eiXERv0gIhyiKvr2P5DwbBmut3qXEkDalMHTo95BSdL2A==",
       "optional": true
     },
     "@lancedb/vectordb-linux-x64-gnu": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.1.19.tgz",
-      "integrity": "sha512-DoQBskl22JAJFZh219ZOJ6o+f1niTZp0qRYngHa/kTIpLKzHWQ0OTtMCz32VBAjAsKjSLNxHE8rrT/S6tvS7KQ==",
+      "integrity": "sha512-AG0FHksbbr+cHVKPi4B8cmBtqb6T9E0uaK4kyZkXrX52/xtv9RYVZcykaB/tSSm0XNFPWWRnx9R8UqNZV/hxMA==",
       "optional": true
     },
     "@lancedb/vectordb-win32-x64-msvc": {
-      "version": "0.1.18",
+      "version": "0.1.19",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.1.18.tgz",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.1.19.tgz",
-      "integrity": "sha512-a/kUM3V6rWuXS80pPECYxKfCUAnq56Of/GPCvnAkpk9C9ldyX10iff4aA6DiPHjEk9V2ytqDfJKl9N3QcMLKLA==",
+      "integrity": "sha512-PDWZ2hvLVXH4Z4WIO1rsWY8ev3NpNm7aXlaey32P+l1Iz9Hia9+F2GBpp2UiEQKfvbk82ucAvBLRmpSsHY8Tlw==",
       "optional": true
     },
     "@neon-rs/cli": {

node/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vectordb",
-  "version": "0.1.18",
+  "version": "0.1.19",
   "description": " Serverless, low-latency vector database for AI applications",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
@@ -78,10 +78,10 @@
     }
   },
   "optionalDependencies": {
-    "@lancedb/vectordb-darwin-arm64": "0.1.18",
+    "@lancedb/vectordb-darwin-arm64": "0.1.19",
-    "@lancedb/vectordb-darwin-x64": "0.1.18",
+    "@lancedb/vectordb-darwin-x64": "0.1.19",
-    "@lancedb/vectordb-linux-arm64-gnu": "0.1.18",
+    "@lancedb/vectordb-linux-arm64-gnu": "0.1.19",
-    "@lancedb/vectordb-linux-x64-gnu": "0.1.18",
+    "@lancedb/vectordb-linux-x64-gnu": "0.1.19",
-    "@lancedb/vectordb-win32-x64-msvc": "0.1.18"
+    "@lancedb/vectordb-win32-x64-msvc": "0.1.19"
   }
 }

node/src/embedding/embedding_function.ts

@@ -28,7 +28,6 @@ export interface EmbeddingFunction<T> {
 }
 
 export function isEmbeddingFunction<T> (value: any): value is EmbeddingFunction<T> {
-  return Object.keys(value).length === 2 &&
+  return typeof value.sourceColumn === 'string' &&
-      typeof value.sourceColumn === 'string' &&
       typeof value.embed === 'function'
 }

node/src/test/embedding/openai.ts

@@ -16,6 +16,7 @@ import { describe } from 'mocha'
 import { assert } from 'chai'
 
 import { OpenAIEmbeddingFunction } from '../../embedding/openai'
+import { isEmbeddingFunction } from '../../embedding/embedding_function'
 
 // eslint-disable-next-line @typescript-eslint/no-var-requires
 const { OpenAIApi } = require('openai')
@@ -47,4 +48,10 @@ describe('OpenAPIEmbeddings', function () {
       assert.deepEqual(vectors[1], stubValue.data.data[1].embedding)
     })
   })
+
+  describe('isEmbeddingFunction', function () {
+    it('should match the isEmbeddingFunction guard', function () {
+      assert.isTrue(isEmbeddingFunction(new OpenAIEmbeddingFunction('text', 'sk-key')))
+    })
+  })
 })

node/src/test/test.ts

@@ -134,6 +134,18 @@ describe('LanceDB client', function () {
       assert.equal(await table.countRows(), 2)
     })
 
+    it('fails to create a new table when the vector column is missing', async function () {
+      const dir = await track().mkdir('lancejs')
+      const con = await lancedb.connect(dir)
+
+      const data = [
+        { id: 1, price: 10 }
+      ]
+
+      const create = con.createTable('missing_vector', data)
+      await expect(create).to.be.rejectedWith(Error, 'column \'vector\' is missing')
+    })
+
     it('use overwrite flag to overwrite existing table', async function () {
       const dir = await track().mkdir('lancejs')
       const con = await lancedb.connect(dir)
@@ -230,6 +242,22 @@ describe('LanceDB client', function () {
       // Default replace = true
       await table.createIndex({ type: 'ivf_pq', column: 'vector', num_partitions: 2, max_iters: 2, num_sub_vectors: 2 })
     }).timeout(50_000)
+
+    it('it should fail when the column is not a vector', async function () {
+      const uri = await createTestDB(32, 300)
+      const con = await lancedb.connect(uri)
+      const table = await con.openTable('vectors')
+      const createIndex = table.createIndex({ type: 'ivf_pq', column: 'name', num_partitions: 2, max_iters: 2, num_sub_vectors: 2 })
+      await expect(createIndex).to.be.rejectedWith(/VectorIndex requires the column data type to be fixed size list of float32s/)
+    })
+
+    it('it should fail when the column is not a vector', async function () {
+      const uri = await createTestDB(32, 300)
+      const con = await lancedb.connect(uri)
+      const table = await con.openTable('vectors')
+      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')
+    })
   })
 
   describe('when using a custom embedding function', function () {

python/.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.1.15
+current_version = 0.1.16
 commit = True
 message = [python] Bump version: {current_version} → {new_version}
 tag = True

python/lancedb/common.py

@@ -11,17 +11,18 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 from pathlib import Path
-from typing import List, Union
+from typing import Iterable, List, Union
 
 import numpy as np
-import pandas as pd
 import pyarrow as pa
 
+from .util import safe_import_pandas
+
+pd = safe_import_pandas()
+
+DATA = Union[List[dict], dict, "pd.DataFrame", pa.Table, Iterable[pa.RecordBatch]]
 VEC = Union[list, np.ndarray, pa.Array, pa.ChunkedArray]
 URI = Union[str, Path]
-
-# TODO support generator
-DATA = Union[List[dict], dict, pd.DataFrame]
 VECTOR_COLUMN_NAME = "vector"
 
 

python/lancedb/context.py

@@ -12,12 +12,13 @@
 #  limitations under the License.
 from __future__ import annotations
 
-import pandas as pd
-
 from .exceptions import MissingColumnError, MissingValueError
+from .util import safe_import_pandas
+
+pd = safe_import_pandas()
 
 
-def contextualize(raw_df: pd.DataFrame) -> Contextualizer:
+def contextualize(raw_df: "pd.DataFrame") -> Contextualizer:
     """Create a Contextualizer object for the given DataFrame.
 
     Used to create context windows. Context windows are rolling subsets of text
@@ -175,8 +176,12 @@ class Contextualizer:
         self._min_window_size = min_window_size
         return self
 
-    def to_df(self) -> pd.DataFrame:
+    def to_df(self) -> "pd.DataFrame":
         """Create the context windows and return a DataFrame."""
+        if pd is None:
+            raise ImportError(
+                "pandas is required to create context windows using lancedb"
+            )
 
         if self._text_col not in self._raw_df.columns.tolist():
             raise MissingColumnError(self._text_col)

python/lancedb/db.py

@@ -16,9 +16,8 @@ from __future__ import annotations
 import os
 from abc import ABC, abstractmethod
 from pathlib import Path
-from typing import Dict, Iterable, List, Optional, Tuple, Union
+from typing import Optional
 
-import pandas as pd
 import pyarrow as pa
 from pyarrow import fs
 
@@ -39,9 +38,7 @@ class DBConnection(ABC):
     def create_table(
         self,
         name: str,
-        data: Optional[
+        data: Optional[DATA] = None,
-            Union[List[dict], dict, pd.DataFrame, pa.Table, Iterable[pa.RecordBatch]],
-        ] = None,
         schema: Optional[pa.Schema] = None,
         mode: str = "create",
         on_bad_vectors: str = "error",
@@ -279,7 +276,7 @@ class LanceDBConnection(DBConnection):
     def create_table(
         self,
         name: str,
-        data: Optional[Union[List[dict], dict, pd.DataFrame]] = None,
+        data: Optional[DATA] = None,
         schema: pa.Schema = None,
         mode: str = "create",
         on_bad_vectors: str = "error",
@@ -319,14 +316,20 @@ class LanceDBConnection(DBConnection):
         """
         return LanceTable.open(self, name)
 
-    def drop_table(self, name: str):
+    def drop_table(self, name: str, ignore_missing: bool = False):
         """Drop a table from the database.
 
         Parameters
         ----------
         name: str
             The name of the table.
+        ignore_missing: bool, default False
+            If True, ignore if the table does not exist.
         """
-        filesystem, path = fs_from_uri(self.uri)
+        try:
-        table_path = os.path.join(path, name + ".lance")
+            filesystem, path = fs_from_uri(self.uri)
-        filesystem.delete_dir(table_path)
+            table_path = os.path.join(path, name + ".lance")
+            filesystem.delete_dir(table_path)
+        except FileNotFoundError:
+            if not ignore_missing:
+                raise

python/lancedb/embeddings.py

@@ -16,15 +16,19 @@ import sys
 from typing import Callable, Union
 
 import numpy as np
-import pandas as pd
 import pyarrow as pa
 from lance.vector import vec_to_table
 from retry import retry
 
+from .util import safe_import_pandas
+
+pd = safe_import_pandas()
+DATA = Union[pa.Table, "pd.DataFrame"]
+
 
 def with_embeddings(
     func: Callable,
-    data: Union[pa.Table, pd.DataFrame],
+    data: DATA,
     column: str = "text",
     wrap_api: bool = True,
     show_progress: bool = False,
@@ -60,7 +64,7 @@ def with_embeddings(
     func = func.batch_size(batch_size)
     if show_progress:
         func = func.show_progress()
-    if isinstance(data, pd.DataFrame):
+    if pd is not None and isinstance(data, pd.DataFrame):
         data = pa.Table.from_pandas(data, preserve_index=False)
     embeddings = func(data[column].to_numpy())
     table = vec_to_table(np.array(embeddings))

python/lancedb/pydantic.py

@@ -249,3 +249,36 @@ def pydantic_to_schema(model: Type[pydantic.BaseModel]) -> pa.Schema:
     """
     fields = _pydantic_model_to_fields(model)
     return pa.schema(fields)
+
+
+class LanceModel(pydantic.BaseModel):
+    """
+    A Pydantic Model base class that can be converted to a LanceDB Table.
+
+    Examples
+    --------
+    >>> import lancedb
+    >>> from lancedb.pydantic import LanceModel, vector
+    >>>
+    >>> class TestModel(LanceModel):
+    ...     name: str
+    ...     vector: vector(2)
+    ...
+    >>> db = lancedb.connect("/tmp")
+    >>> table = db.create_table("test", schema=TestModel.to_arrow_schema())
+    >>> table.add([
+    ...     TestModel(name="test", vector=[1.0, 2.0])
+    ... ])
+    >>> table.search([0., 0.]).limit(1).to_pydantic(TestModel)
+    [TestModel(name='test', vector=FixedSizeList(dim=2))]
+    """
+
+    @classmethod
+    def to_arrow_schema(cls):
+        return pydantic_to_schema(cls)
+
+    @classmethod
+    def field_names(cls) -> List[str]:
+        if PYDANTIC_VERSION.major < 2:
+            return list(cls.__fields__.keys())
+        return list(cls.model_fields.keys())

python/lancedb/query.py

@@ -13,17 +13,20 @@
 
 from __future__ import annotations
 
-from typing import List, Literal, Optional, Union
+from typing import List, Literal, Optional, Type, Union
 
 import numpy as np
-import pandas as pd
 import pyarrow as pa
-from pydantic import BaseModel
+import pydantic
 
 from .common import VECTOR_COLUMN_NAME
+from .pydantic import LanceModel
+from .util import safe_import_pandas
+
+pd = safe_import_pandas()
 
 
-class Query(BaseModel):
+class Query(pydantic.BaseModel):
     """A Query"""
 
     vector_column: str = VECTOR_COLUMN_NAME
@@ -198,7 +201,7 @@ class LanceQueryBuilder:
         self._refine_factor = refine_factor
         return self
 
-    def to_df(self) -> pd.DataFrame:
+    def to_df(self) -> "pd.DataFrame":
         """
         Execute the query and return the results as a pandas DataFrame.
         In addition to the selected columns, LanceDB also returns a vector
@@ -230,9 +233,26 @@ class LanceQueryBuilder:
         )
         return self._table._execute_query(query)
 
+    def to_pydantic(self, model: Type[LanceModel]) -> List[LanceModel]:
+        """Return the table as a list of pydantic models.
+
+        Parameters
+        ----------
+        model: Type[LanceModel]
+            The pydantic model to use.
+
+        Returns
+        -------
+        List[LanceModel]
+        """
+        return [
+            model(**{k: v for k, v in row.items() if k in model.field_names()})
+            for row in self.to_arrow().to_pylist()
+        ]
+
 
 class LanceFtsQueryBuilder(LanceQueryBuilder):
-    def to_arrow(self) -> pd.Table:
+    def to_arrow(self) -> pa.Table:
         try:
             import tantivy
         except ImportError:

python/lancedb/remote/db.py

@@ -20,7 +20,6 @@ import pyarrow as pa
 
 from lancedb.common import DATA
 from lancedb.db import DBConnection
-from lancedb.schema import schema_to_json
 from lancedb.table import Table, _sanitize_data
 
 from .arrow import to_ipc_binary

python/lancedb/remote/table.py

@@ -16,11 +16,11 @@ from functools import cached_property
 from typing import Union
 
 import pyarrow as pa
+from lance import json_to_schema
 
 from lancedb.common import DATA, VEC, VECTOR_COLUMN_NAME
 
-from ..query import LanceQueryBuilder, Query
+from ..query import LanceQueryBuilder
-from ..schema import json_to_schema
 from ..table import Query, Table, _sanitize_data
 from .arrow import to_ipc_binary
 from .client import ARROW_STREAM_CONTENT_TYPE

python/lancedb/schema.py

@@ -12,11 +12,7 @@
 #  limitations under the License.
 
 """Schema related utilities."""
-
-from typing import Any, Dict, Type
-
 import pyarrow as pa
-from lance import json_to_schema, schema_to_json
 
 
 def vector(dimension: int, value_type: pa.DataType = pa.float32()) -> pa.DataType:

python/lancedb/table.py

@@ -20,26 +20,32 @@ from typing import Iterable, List, Union
 
 import lance
 import numpy as np
-import pandas as pd
 import pyarrow as pa
 import pyarrow.compute as pc
 from lance import LanceDataset
 from lance.vector import vec_to_table
 
 from .common import DATA, VEC, VECTOR_COLUMN_NAME
+from .pydantic import LanceModel
 from .query import LanceFtsQueryBuilder, LanceQueryBuilder, Query
-from .util import fs_from_uri
+from .util import fs_from_uri, safe_import_pandas
+
+pd = safe_import_pandas()
 
 
 def _sanitize_data(data, schema, on_bad_vectors, fill_value):
     if isinstance(data, list):
+        # convert to list of dict if data is a bunch of LanceModels
+        if isinstance(data[0], LanceModel):
+            schema = data[0].__class__.to_arrow_schema()
+            data = [dict(d) for d in data]
         data = pa.Table.from_pylist(data)
         data = _sanitize_schema(
             data, schema=schema, on_bad_vectors=on_bad_vectors, fill_value=fill_value
         )
     if isinstance(data, dict):
         data = vec_to_table(data)
-    if isinstance(data, pd.DataFrame):
+    if pd is not None and isinstance(data, pd.DataFrame):
         data = pa.Table.from_pandas(data)
         data = _sanitize_schema(
             data, schema=schema, on_bad_vectors=on_bad_vectors, fill_value=fill_value
@@ -94,7 +100,7 @@ class Table(ABC):
         """
         raise NotImplementedError
 
-    def to_pandas(self) -> pd.DataFrame:
+    def to_pandas(self):
         """Return the table as a pandas DataFrame.
 
         Returns
@@ -328,7 +334,7 @@ class LanceTable(Table):
         """Return the first n rows of the table."""
         return self._dataset.head(n)
 
-    def to_pandas(self) -> pd.DataFrame:
+    def to_pandas(self) -> "pd.DataFrame":
         """Return the table as a pandas DataFrame.
 
         Returns

python/lancedb/util.py

@@ -15,7 +15,6 @@ import os
 from typing import Tuple
 from urllib.parse import urlparse
 
-import pyarrow as pa
 import pyarrow.fs as pa_fs
 
 
@@ -76,3 +75,12 @@ def fs_from_uri(uri: str) -> Tuple[pa_fs.FileSystem, str]:
         return fs, path
 
     return pa_fs.FileSystem.from_uri(uri)
+
+
+def safe_import_pandas():
+    try:
+        import pandas as pd
+
+        return pd
+    except ImportError:
+        return None

python/pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "lancedb"
-version = "0.1.15"
+version = "0.1.16"
-dependencies = ["pylance~=0.5.8", "ratelimiter", "retry", "tqdm", "aiohttp", "pydantic", "attr", "semver"]
+dependencies = ["pylance==0.5.10", "ratelimiter", "retry", "tqdm", "aiohttp", "pydantic", "attr", "semver"]
 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-asyncio"
+    "pandas>=1.4", "pytest", "pytest-mock", "pytest-asyncio"
 ]
 dev = [
     "ruff", "pre-commit", "black"

python/tests/test_db.py

@@ -149,6 +149,10 @@ def test_delete_table(tmp_path):
     db.create_table("test", data=data)
     assert db.table_names() == ["test"]
 
+    # dropping a table that does not exist should pass
+    # if ignore_missing=True
+    db.drop_table("does_not_exist", ignore_missing=True)
+
 
 def test_empty_or_nonexistent_table(tmp_path):
     db = lancedb.connect(tmp_path)

python/tests/test_pydantic.py

@@ -20,7 +20,7 @@ import pyarrow as pa
 import pydantic
 import pytest
 
-from lancedb.pydantic import PYDANTIC_VERSION, pydantic_to_schema, vector
+from lancedb.pydantic import PYDANTIC_VERSION, LanceModel, pydantic_to_schema, vector
 
 
 @pytest.mark.skipif(
@@ -163,3 +163,13 @@ def test_fixed_size_list_validation():
         TestModel(vec=range(7))
 
     TestModel(vec=range(8))
+
+
+def test_lance_model():
+    class TestModel(LanceModel):
+        vec: vector(16)
+        li: List[int]
+
+    schema = pydantic_to_schema(TestModel)
+    assert schema == TestModel.to_arrow_schema()
+    assert TestModel.field_names() == ["vec", "li"]

python/tests/test_query.py

@@ -20,6 +20,7 @@ import pyarrow as pa
 import pytest
 
 from lancedb.db import LanceDBConnection
+from lancedb.pydantic import LanceModel, vector
 from lancedb.query import LanceQueryBuilder, Query
 from lancedb.table import LanceTable
 
@@ -64,6 +65,24 @@ def table(tmp_path) -> MockTable:
     return MockTable(tmp_path)
 
 
+def test_cast(table):
+    class TestModel(LanceModel):
+        vector: vector(2)
+        id: int
+        str_field: str
+        float_field: float
+
+    q = LanceQueryBuilder(table, [0, 0], "vector").limit(1)
+    results = q.to_pydantic(TestModel)
+    assert len(results) == 1
+    r0 = results[0]
+    assert isinstance(r0, TestModel)
+    assert r0.id == 1
+    assert r0.vector == [1, 2]
+    assert r0.str_field == "a"
+    assert r0.float_field == 1.0
+
+
 def test_query_builder(table):
     df = LanceQueryBuilder(table, [0, 0], "vector").limit(1).select(["id"]).to_df()
     assert df["id"].values[0] == 1

python/tests/test_table.py

@@ -13,15 +13,16 @@
 
 import functools
 from pathlib import Path
+from typing import List
 from unittest.mock import PropertyMock, patch
 
 import numpy as np
 import pandas as pd
 import pyarrow as pa
 import pytest
-from lance.vector import vec_to_table
 
 from lancedb.db import LanceDBConnection
+from lancedb.pydantic import LanceModel, vector
 from lancedb.table import LanceTable
 
 
@@ -135,6 +136,17 @@ def test_add(db):
     _add(table, schema)
 
 
+def test_add_pydantic_model(db):
+    class TestModel(LanceModel):
+        vector: vector(16)
+        li: List[int]
+
+    data = TestModel(vector=list(range(16)), li=[1, 2, 3])
+    table = LanceTable.create(db, "test", data=[data])
+    assert len(table) == 1
+    assert table.schema == TestModel.to_arrow_schema()
+
+
 def _add(table, schema):
     # table = LanceTable(db, "test")
     assert len(table) == 2

rust/ffi/node/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb-node"
-version = "0.1.18"
+version = "0.1.19"
 description = "Serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"
 edition = "2018"
@@ -13,6 +13,7 @@ crate-type = ["cdylib"]
 arrow-array = { workspace = true }
 arrow-ipc = { workspace = true }
 arrow-schema = { workspace = true }
+conv = "0.3.3"
 once_cell = "1"
 futures = "0.3"
 half = { workspace = true }
@@ -21,5 +22,6 @@ vectordb = { path = "../../vectordb" }
 tokio = { version = "1.23", features = ["rt-multi-thread"] }
 neon = {version = "0.10.1", default-features = false, features = ["channel-api", "napi-6", "promise-api", "task-api"] }
 object_store = { workspace = true, features = ["aws"] }
+snafu = { workspace = true }
 async-trait = "0"
 env_logger = "0"

rust/ffi/node/src/arrow.rs

@@ -13,27 +13,30 @@
 // limitations under the License.
 
 use std::io::Cursor;
+use std::ops::Deref;
 use std::sync::Arc;
 
 use arrow_array::cast::as_list_array;
-use arrow_array::{Array, FixedSizeListArray, RecordBatch};
+use arrow_array::{Array, ArrayRef, FixedSizeListArray, RecordBatch};
 use arrow_ipc::reader::FileReader;
+use arrow_ipc::writer::FileWriter;
 use arrow_schema::{DataType, Field, Schema};
 use lance::arrow::{FixedSizeListArrayExt, RecordBatchExt};
+use vectordb::table::VECTOR_COLUMN_NAME;
+
+use crate::error::{MissingColumnSnafu, Result};
+use snafu::prelude::*;
+
+pub(crate) fn convert_record_batch(record_batch: RecordBatch) -> Result<RecordBatch> {
+    let column = get_column(VECTOR_COLUMN_NAME, &record_batch)?;
 
-pub(crate) fn convert_record_batch(record_batch: RecordBatch) -> RecordBatch {
-    let column = record_batch
-        .column_by_name("vector")
-        .cloned()
-        .expect("vector column is missing");
     // TODO: we should just consume the underlying js buffer in the future instead of this arrow around a bunch of times
     let arr = as_list_array(column.as_ref());
     let list_size = arr.values().len() / record_batch.num_rows();
-    let r =
+    let r = FixedSizeListArray::try_new_from_values(arr.values().to_owned(), list_size as i32)?;
-        FixedSizeListArray::try_new_from_values(arr.values().to_owned(), list_size as i32).unwrap();
 
     let schema = Arc::new(Schema::new(vec![Field::new(
-        "vector",
+        VECTOR_COLUMN_NAME,
         DataType::FixedSizeList(
             Arc::new(Field::new("item", DataType::Float32, true)),
             list_size as i32,
@@ -41,22 +44,42 @@ pub(crate) fn convert_record_batch(record_batch: RecordBatch) -> RecordBatch {
         true,
     )]));
 
-    let mut new_batch = RecordBatch::try_new(schema.clone(), vec![Arc::new(r)]).unwrap();
+    let mut new_batch = RecordBatch::try_new(schema.clone(), vec![Arc::new(r)])?;
 
     if record_batch.num_columns() > 1 {
-        let rb = record_batch.drop_column("vector").unwrap();
+        let rb = record_batch.drop_column(VECTOR_COLUMN_NAME)?;
-        new_batch = new_batch.merge(&rb).unwrap();
+        new_batch = new_batch.merge(&rb)?;
     }
-    new_batch
+    Ok(new_batch)
 }
 
-pub(crate) fn arrow_buffer_to_record_batch(slice: &[u8]) -> Vec<RecordBatch> {
+fn get_column(column_name: &str, record_batch: &RecordBatch) -> Result<ArrayRef> {
+    record_batch
+        .column_by_name(column_name)
+        .cloned()
+        .context(MissingColumnSnafu { name: column_name })
+}
+
+pub(crate) fn arrow_buffer_to_record_batch(slice: &[u8]) -> Result<Vec<RecordBatch>> {
     let mut batches: Vec<RecordBatch> = Vec::new();
-    let fr = FileReader::try_new(Cursor::new(slice), None);
+    let file_reader = FileReader::try_new(Cursor::new(slice), None)?;
-    let file_reader = fr.unwrap();
     for b in file_reader {
-        let record_batch = convert_record_batch(b.unwrap());
+        let record_batch = convert_record_batch(b?)?;
         batches.push(record_batch);
     }
-    batches
+    Ok(batches)
+}
+
+pub(crate) fn record_batch_to_buffer(batches: Vec<RecordBatch>) -> Result<Vec<u8>> {
+    if batches.is_empty() {
+        return Ok(Vec::new());
+    }
+
+    let schema = batches.get(0).unwrap().schema();
+    let mut fr = FileWriter::try_new(Vec::new(), schema.deref())?;
+    for batch in batches.iter() {
+        fr.write(batch)?
+    }
+    fr.finish()?;
+    Ok(fr.into_inner()?)
 }

rust/ffi/node/src/error.rs

@@ -0,0 +1,88 @@
+// 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 arrow_schema::ArrowError;
+use neon::context::Context;
+use neon::prelude::NeonResult;
+use snafu::Snafu;
+
+#[derive(Debug, Snafu)]
+#[snafu(visibility(pub(crate)))]
+pub enum Error {
+    #[snafu(display("column '{name}' is missing"))]
+    MissingColumn { name: String },
+    #[snafu(display("{name}: {message}"))]
+    RangeError { name: String, message: String },
+    #[snafu(display("{index_type} is not a valid index type"))]
+    InvalidIndexType { index_type: String },
+
+    #[snafu(display("{message}"))]
+    LanceDB { message: String },
+    #[snafu(display("{message}"))]
+    Neon { message: String },
+}
+
+pub type Result<T> = std::result::Result<T, Error>;
+
+impl From<vectordb::error::Error> for Error {
+    fn from(e: vectordb::error::Error) -> Self {
+        Self::LanceDB {
+            message: e.to_string(),
+        }
+    }
+}
+
+impl From<lance::Error> for Error {
+    fn from(e: lance::Error) -> Self {
+        Self::LanceDB {
+            message: e.to_string(),
+        }
+    }
+}
+
+impl From<ArrowError> for Error {
+    fn from(value: ArrowError) -> Self {
+        Self::LanceDB {
+            message: value.to_string(),
+        }
+    }
+}
+
+impl From<neon::result::Throw> for Error {
+    fn from(value: neon::result::Throw) -> Self {
+        Self::Neon {
+            message: value.to_string(),
+        }
+    }
+}
+
+/// ResultExt is used to transform a [`Result`] into a [`NeonResult`],
+/// so it can be returned as a JavaScript error
+/// Copied from [Neon](https://github.com/neon-bindings/neon/blob/4c2e455a9e6814f1ba0178616d63caec7f4df317/crates/neon/src/result/mod.rs#L88)
+pub trait ResultExt<T> {
+    fn or_throw<'a, C: Context<'a>>(self, cx: &mut C) -> NeonResult<T>;
+}
+
+/// Implement ResultExt for the std Result so it can be used any Result type
+impl<T, E> ResultExt<T> for std::result::Result<T, E>
+where
+    E: std::fmt::Display,
+{
+    fn or_throw<'a, C: Context<'a>>(self, cx: &mut C) -> NeonResult<T> {
+        match self {
+            Ok(value) => Ok(value),
+            Err(error) => cx.throw_error(error.to_string()),
+        }
+    }
+}

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

@@ -22,12 +22,15 @@ use neon::prelude::*;
 
 use vectordb::index::vector::{IvfPQIndexBuilder, VectorIndexBuilder};
 
+use crate::error::Error::InvalidIndexType;
+use crate::error::ResultExt;
+use crate::neon_ext::js_object_ext::JsObjectExt;
 use crate::{runtime, JsTable};
 
 pub(crate) fn table_create_vector_index(mut cx: FunctionContext) -> JsResult<JsPromise> {
     let js_table = cx.this().downcast_or_throw::<JsBox<JsTable>, _>(&mut cx)?;
     let index_params = cx.argument::<JsObject>(0)?;
-    let index_params_builder = get_index_params_builder(&mut cx, index_params).unwrap();
+    let index_params_builder = get_index_params_builder(&mut cx, index_params).or_throw(&mut cx)?;
 
     let rt = runtime(&mut cx)?;
     let channel = cx.channel();
@@ -54,27 +57,21 @@ pub(crate) fn table_create_vector_index(mut cx: FunctionContext) -> JsResult<JsP
 fn get_index_params_builder(
     cx: &mut FunctionContext,
     obj: Handle<JsObject>,
-) -> Result<impl VectorIndexBuilder, String> {
+) -> crate::error::Result<impl VectorIndexBuilder> {
-    let idx_type = obj
+    let idx_type = obj.get::<JsString, _, _>(cx, "type")?.value(cx);
-        .get::<JsString, _, _>(cx, "type")
-        .map_err(|t| t.to_string())?
-        .value(cx);
 
     match idx_type.as_str() {
         "ivf_pq" => {
             let mut index_builder: IvfPQIndexBuilder = IvfPQIndexBuilder::new();
             let mut pq_params = PQBuildParams::default();
 
-            obj.get_opt::<JsString, _, _>(cx, "column")
+            obj.get_opt::<JsString, _, _>(cx, "column")?
-                .map_err(|t| t.to_string())?
                 .map(|s| index_builder.column(s.value(cx)));
 
-            obj.get_opt::<JsString, _, _>(cx, "index_name")
+            obj.get_opt::<JsString, _, _>(cx, "index_name")?
-                .map_err(|t| t.to_string())?
                 .map(|s| index_builder.index_name(s.value(cx)));
 
-            obj.get_opt::<JsString, _, _>(cx, "metric_type")
+            obj.get_opt::<JsString, _, _>(cx, "metric_type")?
-                .map_err(|t| t.to_string())?
                 .map(|s| MetricType::try_from(s.value(cx).as_str()))
                 .map(|mt| {
                     let metric_type = mt.unwrap();
@@ -82,15 +79,8 @@ fn get_index_params_builder(
                     pq_params.metric_type = metric_type;
                 });
 
-            let num_partitions = obj
+            let num_partitions = obj.get_opt_usize(cx, "num_partitions")?;
-                .get_opt::<JsNumber, _, _>(cx, "num_partitions")
+            let max_iters = obj.get_opt_usize(cx, "max_iters")?;
-                .map_err(|t| t.to_string())?
-                .map(|s| s.value(cx) as usize);
-
-            let max_iters = obj
-                .get_opt::<JsNumber, _, _>(cx, "max_iters")
-                .map_err(|t| t.to_string())?
-                .map(|s| s.value(cx) as usize);
 
             num_partitions.map(|np| {
                 let max_iters = max_iters.unwrap_or(50);
@@ -102,32 +92,28 @@ fn get_index_params_builder(
                 index_builder.ivf_params(ivf_params)
             });
 
-            obj.get_opt::<JsBoolean, _, _>(cx, "use_opq")
+            obj.get_opt::<JsBoolean, _, _>(cx, "use_opq")?
-                .map_err(|t| t.to_string())?
                 .map(|s| pq_params.use_opq = s.value(cx));
 
-            obj.get_opt::<JsNumber, _, _>(cx, "num_sub_vectors")
+            obj.get_opt_usize(cx, "num_sub_vectors")?
-                .map_err(|t| t.to_string())?
+                .map(|s| pq_params.num_sub_vectors = s);
-                .map(|s| pq_params.num_sub_vectors = s.value(cx) as usize);
 
-            obj.get_opt::<JsNumber, _, _>(cx, "num_bits")
+            obj.get_opt_usize(cx, "num_bits")?
-                .map_err(|t| t.to_string())?
+                .map(|s| pq_params.num_bits = s);
-                .map(|s| pq_params.num_bits = s.value(cx) as usize);
 
-            obj.get_opt::<JsNumber, _, _>(cx, "max_iters")
+            obj.get_opt_usize(cx, "max_iters")?
-                .map_err(|t| t.to_string())?
+                .map(|s| pq_params.max_iters = s);
-                .map(|s| pq_params.max_iters = s.value(cx) as usize);
 
-            obj.get_opt::<JsNumber, _, _>(cx, "max_opq_iters")
+            obj.get_opt_usize(cx, "max_opq_iters")?
-                .map_err(|t| t.to_string())?
+                .map(|s| pq_params.max_opq_iters = s);
-                .map(|s| pq_params.max_opq_iters = s.value(cx) as usize);
 
-            obj.get_opt::<JsBoolean, _, _>(cx, "replace")
+            obj.get_opt::<JsBoolean, _, _>(cx, "replace")?
-                .map_err(|t| t.to_string())?
                 .map(|s| index_builder.replace(s.value(cx)));
 
             Ok(index_builder)
         }
-        t => Err(format!("{} is not a valid index type", t).to_string()),
+        index_type => Err(InvalidIndexType {
+            index_type: index_type.into(),
+        }),
     }
 }

rust/ffi/node/src/lib.rs

@@ -18,7 +18,6 @@ use std::ops::Deref;
 use std::sync::{Arc, Mutex};
 
 use arrow_array::{Float32Array, RecordBatchIterator};
-use arrow_ipc::writer::FileWriter;
 use async_trait::async_trait;
 use futures::{TryFutureExt, TryStreamExt};
 use lance::dataset::{WriteMode, WriteParams};
@@ -32,14 +31,17 @@ use once_cell::sync::OnceCell;
 use tokio::runtime::Runtime;
 
 use vectordb::database::Database;
-use vectordb::error::Error;
 use vectordb::table::{ReadParams, Table};
 
-use crate::arrow::arrow_buffer_to_record_batch;
+use crate::arrow::{arrow_buffer_to_record_batch, record_batch_to_buffer};
+use crate::error::ResultExt;
+use crate::neon_ext::js_object_ext::JsObjectExt;
 
 mod arrow;
 mod convert;
+mod error;
 mod index;
+mod neon_ext;
 
 struct JsDatabase {
     database: Arc<Database>,
@@ -86,7 +88,7 @@ fn runtime<'a, C: Context<'a>>(cx: &mut C) -> NeonResult<&'static Runtime> {
 
     LOG.get_or_init(|| env_logger::init());
 
-    RUNTIME.get_or_try_init(|| Runtime::new().or_else(|err| cx.throw_error(err.to_string())))
+    RUNTIME.get_or_try_init(|| Runtime::new().or_throw(cx))
 }
 
 fn database_new(mut cx: FunctionContext) -> JsResult<JsPromise> {
@@ -101,7 +103,7 @@ fn database_new(mut cx: FunctionContext) -> JsResult<JsPromise> {
 
         deferred.settle_with(&channel, move |mut cx| {
             let db = JsDatabase {
-                database: Arc::new(database.or_else(|err| cx.throw_error(err.to_string()))?),
+                database: Arc::new(database.or_throw(&mut cx)?),
             };
             Ok(cx.boxed(db))
         });
@@ -123,7 +125,7 @@ fn database_table_names(mut cx: FunctionContext) -> JsResult<JsPromise> {
         let tables_rst = database.table_names().await;
 
         deferred.settle_with(&channel, move |mut cx| {
-            let tables = tables_rst.or_else(|err| cx.throw_error(err.to_string()))?;
+            let tables = tables_rst.or_throw(&mut cx)?;
             let table_names = convert::vec_str_to_array(&tables, &mut cx);
             table_names
         });
@@ -194,9 +196,7 @@ fn database_open_table(mut cx: FunctionContext) -> JsResult<JsPromise> {
         let table_rst = database.open_table_with_params(&table_name, &params).await;
 
         deferred.settle_with(&channel, move |mut cx| {
-            let table = Arc::new(Mutex::new(
+            let table = Arc::new(Mutex::new(table_rst.or_throw(&mut cx)?));
-                table_rst.or_else(|err| cx.throw_error(err.to_string()))?,
-            ));
             Ok(cx.boxed(JsTable { table }))
         });
     });
@@ -217,7 +217,7 @@ fn database_drop_table(mut cx: FunctionContext) -> JsResult<JsPromise> {
     rt.spawn(async move {
         let result = database.drop_table(&table_name).await;
         deferred.settle_with(&channel, move |mut cx| {
-            result.or_else(|err| cx.throw_error(err.to_string()))?;
+            result.or_throw(&mut cx)?;
             Ok(cx.null())
         });
     });
@@ -246,12 +246,9 @@ fn table_search(mut cx: FunctionContext) -> JsResult<JsPromise> {
         .get_opt::<JsString, _, _>(&mut cx, "_filter")?
         .map(|s| s.value(&mut cx));
     let refine_factor = query_obj
-        .get_opt::<JsNumber, _, _>(&mut cx, "_refineFactor")?
+        .get_opt_u32(&mut cx, "_refineFactor")
-        .map(|s| s.value(&mut cx))
+        .or_throw(&mut cx)?;
-        .map(|i| i as u32);
+    let nprobes = query_obj.get_usize(&mut cx, "_nprobes").or_throw(&mut cx)?;
-    let nprobes = query_obj
-        .get::<JsNumber, _, _>(&mut cx, "_nprobes")?
-        .value(&mut cx) as usize;
     let metric_type = query_obj
         .get_opt::<JsString, _, _>(&mut cx, "_metricType")?
         .map(|s| s.value(&mut cx))
@@ -278,30 +275,17 @@ fn table_search(mut cx: FunctionContext) -> JsResult<JsPromise> {
             .select(select);
         let record_batch_stream = builder.execute();
         let results = record_batch_stream
-            .and_then(|stream| stream.try_collect::<Vec<_>>().map_err(Error::from))
+            .and_then(|stream| {
+                stream
+                    .try_collect::<Vec<_>>()
+                    .map_err(vectordb::error::Error::from)
+            })
             .await;
 
         deferred.settle_with(&channel, move |mut cx| {
-            let results = results.or_else(|err| cx.throw_error(err.to_string()))?;
+            let results = results.or_throw(&mut cx)?;
-            let vector: Vec<u8> = Vec::new();
+            let buffer = record_batch_to_buffer(results).or_throw(&mut cx)?;
-
+            Ok(JsBuffer::external(&mut cx, buffer))
-            if results.is_empty() {
-                return cx.buffer(0);
-            }
-
-            let schema = results.get(0).unwrap().schema();
-            let mut fr = FileWriter::try_new(vector, schema.deref())
-                .or_else(|err| cx.throw_error(err.to_string()))?;
-
-            for batch in results.iter() {
-                fr.write(batch)
-                    .or_else(|err| cx.throw_error(err.to_string()))?;
-            }
-            fr.finish().or_else(|err| cx.throw_error(err.to_string()))?;
-            let buf = fr
-                .into_inner()
-                .or_else(|err| cx.throw_error(err.to_string()))?;
-            Ok(JsBuffer::external(&mut cx, buf))
         });
     });
     Ok(promise)
@@ -313,7 +297,7 @@ fn table_create(mut cx: FunctionContext) -> JsResult<JsPromise> {
         .downcast_or_throw::<JsBox<JsDatabase>, _>(&mut cx)?;
     let table_name = cx.argument::<JsString>(0)?.value(&mut cx);
     let buffer = cx.argument::<JsBuffer>(1)?;
-    let batches = arrow_buffer_to_record_batch(buffer.as_slice(&mut cx));
+    let batches = arrow_buffer_to_record_batch(buffer.as_slice(&mut cx)).or_throw(&mut cx)?;
     let schema = batches[0].schema();
 
     // Write mode
@@ -351,9 +335,7 @@ fn table_create(mut cx: FunctionContext) -> JsResult<JsPromise> {
             .await;
 
         deferred.settle_with(&channel, move |mut cx| {
-            let table = Arc::new(Mutex::new(
+            let table = Arc::new(Mutex::new(table_rst.or_throw(&mut cx)?));
-                table_rst.or_else(|err| cx.throw_error(err.to_string()))?,
-            ));
             Ok(cx.boxed(JsTable { table }))
         });
     });
@@ -370,7 +352,8 @@ fn table_add(mut cx: FunctionContext) -> JsResult<JsPromise> {
     let js_table = cx.this().downcast_or_throw::<JsBox<JsTable>, _>(&mut cx)?;
     let buffer = cx.argument::<JsBuffer>(0)?;
     let write_mode = cx.argument::<JsString>(1)?.value(&mut cx);
-    let batches = arrow_buffer_to_record_batch(buffer.as_slice(&mut cx));
+
+    let batches = arrow_buffer_to_record_batch(buffer.as_slice(&mut cx)).or_throw(&mut cx)?;
     let schema = batches[0].schema();
 
     let rt = runtime(&mut cx)?;
@@ -399,7 +382,7 @@ fn table_add(mut cx: FunctionContext) -> JsResult<JsPromise> {
         let add_result = table.lock().unwrap().add(batch_reader, Some(params)).await;
 
         deferred.settle_with(&channel, move |mut cx| {
-            let _added = add_result.or_else(|err| cx.throw_error(err.to_string()))?;
+            let _added = add_result.or_throw(&mut cx)?;
             Ok(cx.boolean(true))
         });
     });
@@ -418,7 +401,7 @@ fn table_count_rows(mut cx: FunctionContext) -> JsResult<JsPromise> {
         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()))?;
+            let num_rows = num_rows_result.or_throw(&mut cx)?;
             Ok(cx.number(num_rows as f64))
         });
     });
@@ -438,7 +421,7 @@ fn table_delete(mut cx: FunctionContext) -> JsResult<JsPromise> {
     let delete_result = rt.block_on(async move { table.lock().unwrap().delete(&predicate).await });
 
     deferred.settle_with(&channel, move |mut cx| {
-        delete_result.or_else(|err| cx.throw_error(err.to_string()))?;
+        delete_result.or_throw(&mut cx)?;
         Ok(cx.undefined())
     });
 

rust/ffi/node/src/neon_ext.rs

@@ -0,0 +1,15 @@
+// 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.
+
+pub mod js_object_ext;

rust/ffi/node/src/neon_ext/js_object_ext.rs

@@ -0,0 +1,82 @@
+// 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 crate::error::{Error, Result};
+use neon::prelude::*;
+
+// extends neon's [JsObject] with helper functions to extract properties
+pub trait JsObjectExt {
+    fn get_opt_u32(&self, cx: &mut FunctionContext, key: &str) -> Result<Option<u32>>;
+    fn get_usize(&self, cx: &mut FunctionContext, key: &str) -> Result<usize>;
+    fn get_opt_usize(&self, cx: &mut FunctionContext, key: &str) -> Result<Option<usize>>;
+}
+
+impl JsObjectExt for JsObject {
+    fn get_opt_u32(&self, cx: &mut FunctionContext, key: &str) -> Result<Option<u32>> {
+        let val_opt = self
+            .get_opt::<JsNumber, _, _>(cx, key)?
+            .map(|s| f64_to_u32_safe(s.value(cx), key));
+        val_opt.transpose()
+    }
+
+    fn get_usize(&self, cx: &mut FunctionContext, key: &str) -> Result<usize> {
+        let val = self.get::<JsNumber, _, _>(cx, key)?.value(cx);
+        f64_to_usize_safe(val, key)
+    }
+
+    fn get_opt_usize(&self, cx: &mut FunctionContext, key: &str) -> Result<Option<usize>> {
+        let val_opt = self
+            .get_opt::<JsNumber, _, _>(cx, key)?
+            .map(|s| f64_to_usize_safe(s.value(cx), key));
+        val_opt.transpose()
+    }
+}
+
+fn f64_to_u32_safe(n: f64, key: &str) -> Result<u32> {
+    use conv::*;
+
+    n.approx_as::<u32>().map_err(|e| match e {
+        FloatError::NegOverflow(_) => Error::RangeError {
+            name: key.into(),
+            message: "must be > 0".to_string(),
+        },
+        FloatError::PosOverflow(_) => Error::RangeError {
+            name: key.into(),
+            message: format!("must be < {}", u32::MAX),
+        },
+        FloatError::NotANumber(_) => Error::RangeError {
+            name: key.into(),
+            message: "not a valid number".to_string(),
+        },
+    })
+}
+
+fn f64_to_usize_safe(n: f64, key: &str) -> Result<usize> {
+    use conv::*;
+
+    n.approx_as::<usize>().map_err(|e| match e {
+        FloatError::NegOverflow(_) => Error::RangeError {
+            name: key.into(),
+            message: "must be > 0".to_string(),
+        },
+        FloatError::PosOverflow(_) => Error::RangeError {
+            name: key.into(),
+            message: format!("must be < {}", usize::MAX),
+        },
+        FloatError::NotANumber(_) => Error::RangeError {
+            name: key.into(),
+            message: "not a valid number".to_string(),
+        },
+    })
+}

rust/vectordb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb"
-version = "0.1.18"
+version = "0.1.19"
 edition = "2021"
 description = "Serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"
@@ -12,7 +12,7 @@ arrow-array = { workspace = true }
 arrow-data = { workspace = true }
 arrow-schema = { workspace = true }
 object_store = { workspace = true }
-snafu = "0.7.4"
+snafu = { workspace = true }
 half = { workspace = true }
 lance = { workspace = true }
 tokio = { version = "1.23", features = ["rt-multi-thread"] }