Bump version: 0.3.1 → 0.3.2

Add more APIs to remote table for Node SDK
* `add` rows
* `overwrite` table with rows
* `create` table

This has been tested against dev stack

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.3.0
+current_version = 0.3.2
 commit = True
 message = Bump version: {current_version} → {new_version}
 tag = True

Cargo.toml

@@ -5,23 +5,23 @@ exclude = ["python"]
 resolver = "2"
 
 [workspace.dependencies]
-lance = { "version" = "=0.8.3", "features" = ["dynamodb"] }
-lance-linalg = { "version" = "=0.8.3" }
-lance-testing = { "version" = "=0.8.3" }
+lance = { "version" = "=0.8.5", "features" = ["dynamodb"] }
+lance-linalg = { "version" = "=0.8.5" }
+lance-testing = { "version" = "=0.8.5" }
 # Note that this one does not include pyarrow
-arrow = { version = "43.0.0", optional = false }
-arrow-array = "43.0"
-arrow-data = "43.0"
-arrow-ipc = "43.0"
-arrow-ord = "43.0"
-arrow-schema = "43.0"
-arrow-arith = "43.0"
-arrow-cast = "43.0"
+arrow = { version = "47.0.0", optional = false }
+arrow-array = "47.0"
+arrow-data = "47.0"
+arrow-ipc = "47.0"
+arrow-ord = "47.0"
+arrow-schema = "47.0"
+arrow-arith = "47.0"
+arrow-cast = "47.0"
 chrono = "0.4.23"
 half = { "version" = "=2.2.1", default-features = false, features = [
     "num-traits"
 ] }
 log = "0.4"
-object_store = "0.6.1"
+object_store = "0.7.1"
 snafu = "0.7.4"
 url = "2"

docs/mkdocs.yml

@@ -37,7 +37,7 @@ plugins:
           docstring_style: numpy
         rendering:
           heading_level: 4
-          show_source: false
+          show_source: true
           show_symbol_type_in_heading: true
           show_signature_annotations: true
           show_root_heading: true
@@ -73,7 +73,12 @@ nav:
     - Vector Search: search.md
     - SQL filters: sql.md
     - Indexing: ann_indexes.md
-  - 🧬 Embeddings: embedding.md
+  - 🧬 Embeddings:
+    - embeddings/index.md
+    - Ingest Embedding Functions: embeddings/embedding_functions.md
+    - Available Functions: embeddings/default_embedding_functions.md
+    - Create Custom Embedding Functions: embeddings/api.md
+    - Example- MultiModal CLIP Embeddings: notebooks/DisappearingEmbeddingFunction.ipynb
   - 🔍 Python full-text search: fts.md
   - 🔌 Integrations:
     - integrations/index.md
@@ -105,7 +110,12 @@ nav:
   - Vector Search: search.md
   - SQL filters: sql.md
   - Indexing: ann_indexes.md
-- Embeddings: embedding.md
+- Embeddings:
+  - embeddings/index.md
+  - Ingest Embedding Functions: embeddings/embedding_functions.md
+  - Available Functions: embeddings/default_embedding_functions.md
+  - Create Custom Embedding Functions: embeddings/api.md
+  - Example- MultiModal CLIP Embeddings: notebooks/DisappearingEmbeddingFunction.ipynb
 - Python full-text search: fts.md
 - Integrations:
   - integrations/index.md

docs/src/embeddings/api.md

@@ -0,0 +1,213 @@
+To use your own custom embedding function, you need to follow these 2 simple steps.
+1. Create your embedding function by implementing the `EmbeddingFunction` interface
+2. Register your embedding function in the global `EmbeddingFunctionRegistry`.
+
+Let us see how this looks like in action.
+
+![](../assets/embeddings_api.png)
+
+
+`EmbeddingFunction` & `EmbeddingFunctionRegistry` handle low-level details for serializing schema and model information as metadata. To build a custom embdding function, you don't need to worry about those details and simply focus on setting up the model.
+
+## `TextEmbeddingFunction` Interface
+
+There is another optional layer of abstraction provided in form of `TextEmbeddingFunction`. You can use this if your model isn't multi-modal in nature and only operates on text. In such case both source and vector fields will have the same pathway for vectorization, so you simply just need to setup the model and rest is handled by `TextEmbeddingFunction`. You can read more about the class and its attributes in the class reference.
+
+
+Let's implement `SentenceTransformerEmbeddings` class. All you need to do is implement the `generate_embeddings()` and `ndims` function to handle the input types you expect and register the class in the global `EmbeddingFunctionRegistry`
+
+```python
+from lancedb.embeddings import register
+
+@register("sentence-transformers")
+class SentenceTransformerEmbeddings(TextEmbeddingFunction):
+    name: str = "all-MiniLM-L6-v2"
+    # set more default instance vars like device, etc.
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self._ndims = None
+    
+    def generate_embeddings(self, texts):
+        return self._embedding_model().encode(list(texts), ...).tolist()
+
+    def ndims(self):
+        if self._ndims is None:
+            self._ndims = len(self.generate_embeddings("foo")[0])
+        return self._ndims
+
+    @cached(cache={}) 
+    def _embedding_model(self):
+        return sentence_transformers.SentenceTransformer(name)
+
+```
+
+This is a stripped down version of our implementation of `SentenceTransformerEmbeddings` that removes certain optimizations and defaul settings.
+
+Now you can use this embedding function to create your table schema and that's it! you can then ingest data and run queries without manually vectorizing the inputs.
+
+```python
+from lancedb.pydantic import LanceModel, Vector
+
+registry = EmbeddingFunctionRegistry.get_instance()
+stransformer = registry.get("sentence-transformers").create()
+
+class TextModelSchema(LanceModel):
+    vector: Vector(stransformer.ndims) = stransformer.VectorField()
+    text: str = stransformer.SourceField()
+
+tbl = db.create_table("table", schema=TextModelSchema)
+
+tbl.add(pd.DataFrame({"text": ["halo", "world"]}))
+result = tbl.search("world").limit(5)
+```
+
+NOTE:
+
+You can always implement the `EmbeddingFunction` interface directly if you want or need to, `TextEmbeddingFunction` just makes it much simpler and faster for you to do so, by setting up the boiler plat for text-specific use case
+
+## Multi-modal embedding function example
+You can also use the `EmbeddingFunction` interface to implement more complex workflows such as multi-modal embedding function support. LanceDB implements `OpenClipEmeddingFunction` class that suppports multi-modal seach. Here's the implementation that you can use as a reference to build your own multi-modal embedding functions.
+
+```python
+@register("open-clip")
+class OpenClipEmbeddings(EmbeddingFunction):
+    name: str = "ViT-B-32"
+    pretrained: str = "laion2b_s34b_b79k"
+    device: str = "cpu"
+    batch_size: int = 64
+    normalize: bool = True
+    _model = PrivateAttr()
+    _preprocess = PrivateAttr()
+    _tokenizer = PrivateAttr()
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        open_clip = self.safe_import("open_clip", "open-clip") # EmbeddingFunction util to import external libs and raise if not found
+        model, _, preprocess = open_clip.create_model_and_transforms(
+            self.name, pretrained=self.pretrained
+        )
+        model.to(self.device)
+        self._model, self._preprocess = model, preprocess
+        self._tokenizer = open_clip.get_tokenizer(self.name)
+        self._ndims = None
+
+    def ndims(self):
+        if self._ndims is None:
+            self._ndims = self.generate_text_embeddings("foo").shape[0]
+        return self._ndims
+
+    def compute_query_embeddings(
+        self, query: Union[str, "PIL.Image.Image"], *args, **kwargs
+    ) -> List[np.ndarray]:
+        """
+        Compute the embeddings for a given user query
+
+        Parameters
+        ----------
+        query : Union[str, PIL.Image.Image]
+            The query to embed. A query can be either text or an image.
+        """
+        if isinstance(query, str):
+            return [self.generate_text_embeddings(query)]
+        else:
+            PIL = self.safe_import("PIL", "pillow")
+            if isinstance(query, PIL.Image.Image):
+                return [self.generate_image_embedding(query)]
+            else:
+                raise TypeError("OpenClip supports str or PIL Image as query")
+
+    def generate_text_embeddings(self, text: str) -> np.ndarray:
+        torch = self.safe_import("torch")
+        text = self.sanitize_input(text)
+        text = self._tokenizer(text)
+        text.to(self.device)
+        with torch.no_grad():
+            text_features = self._model.encode_text(text.to(self.device))
+            if self.normalize:
+                text_features /= text_features.norm(dim=-1, keepdim=True)
+            return text_features.cpu().numpy().squeeze()
+
+    def sanitize_input(self, images: IMAGES) -> Union[List[bytes], np.ndarray]:
+        """
+        Sanitize the input to the embedding function.
+        """
+        if isinstance(images, (str, bytes)):
+            images = [images]
+        elif isinstance(images, pa.Array):
+            images = images.to_pylist()
+        elif isinstance(images, pa.ChunkedArray):
+            images = images.combine_chunks().to_pylist()
+        return images
+
+    def compute_source_embeddings(
+        self, images: IMAGES, *args, **kwargs
+    ) -> List[np.array]:
+        """
+        Get the embeddings for the given images
+        """
+        images = self.sanitize_input(images)
+        embeddings = []
+        for i in range(0, len(images), self.batch_size):
+            j = min(i + self.batch_size, len(images))
+            batch = images[i:j]
+            embeddings.extend(self._parallel_get(batch))
+        return embeddings
+
+    def _parallel_get(self, images: Union[List[str], List[bytes]]) -> List[np.ndarray]:
+        """
+        Issue concurrent requests to retrieve the image data
+        """
+        with concurrent.futures.ThreadPoolExecutor() as executor:
+            futures = [
+                executor.submit(self.generate_image_embedding, image)
+                for image in images
+            ]
+            return [future.result() for future in futures]
+
+    def generate_image_embedding(
+        self, image: Union[str, bytes, "PIL.Image.Image"]
+    ) -> np.ndarray:
+        """
+        Generate the embedding for a single image
+
+        Parameters
+        ----------
+        image : Union[str, bytes, PIL.Image.Image]
+            The image to embed. If the image is a str, it is treated as a uri.
+            If the image is bytes, it is treated as the raw image bytes.
+        """
+        torch = self.safe_import("torch")
+        # TODO handle retry and errors for https
+        image = self._to_pil(image)
+        image = self._preprocess(image).unsqueeze(0)
+        with torch.no_grad():
+            return self._encode_and_normalize_image(image)
+
+    def _to_pil(self, image: Union[str, bytes]):
+        PIL = self.safe_import("PIL", "pillow")
+        if isinstance(image, bytes):
+            return PIL.Image.open(io.BytesIO(image))
+        if isinstance(image, PIL.Image.Image):
+            return image
+        elif isinstance(image, str):
+            parsed = urlparse.urlparse(image)
+            # TODO handle drive letter on windows.
+            if parsed.scheme == "file":
+                return PIL.Image.open(parsed.path)
+            elif parsed.scheme == "":
+                return PIL.Image.open(image if os.name == "nt" else parsed.path)
+            elif parsed.scheme.startswith("http"):
+                return PIL.Image.open(io.BytesIO(url_retrieve(image)))
+            else:
+                raise NotImplementedError("Only local and http(s) urls are supported")
+
+    def _encode_and_normalize_image(self, image_tensor: "torch.Tensor"):
+        """
+        encode a single image tensor and optionally normalize the output
+        """
+        image_features = self._model.encode_image(image_tensor)
+        if self.normalize:
+            image_features /= image_features.norm(dim=-1, keepdim=True)
+        return image_features.cpu().numpy().squeeze()
+```

docs/src/embeddings/default_embedding_functions.md

@@ -0,0 +1,156 @@
+There are various Embedding functions available out of the box with lancedb. We're working on supporting other popular embedding APIs.
+
+## Text Embedding Functions
+Here are the text embedding functions registered by default
+
+### Sentence Transformers
+Here are the parameters that you can set when registering a `sentence-transformers` object, and their default values:
+
+| Parameter | Type | Default Value | Description |
+|---|---|---|---|
+| `name` | `str` | `"all-MiniLM-L6-v2"` | The name of the model. |
+| `device` | `str` | `"cpu"` | The device to run the model on. Can be `"cpu"` or `"gpu"`. |
+| `normalize` | `bool` | `True` | Whether to normalize the input text before feeding it to the model. |
+
+
+```python
+db = lancedb.connect("/tmp/db")
+registry = EmbeddingFunctionRegistry.get_instance()
+func = registry.get("sentence-transformers").create(device="cpu")
+
+class Words(LanceModel):
+    text: str = func.SourceField()
+    vector: Vector(func.ndims()) = func.VectorField()
+
+table = db.create_table("words", schema=Words)
+table.add(
+    [
+        {"text": "hello world"}
+        {"text": "goodbye world"}
+    ]
+)
+
+query = "greetings"
+actual = table.search(query).limit(1).to_pydantic(Words)[0]
+print(actual.text)
+```
+
+### OpenAIEmbeddings
+LanceDB has OpenAI embeddings function in the registry by default. It is registered as `openai` and here are the parameters that you can customize when creating the instances
+
+| Parameter | Type | Default Value | Description |
+|---|---|---|---|
+| `name` | `str` | `"text-embedding-ada-002"` | The name of the model. |
+
+
+
+```python
+db = lancedb.connect("/tmp/db")
+registry = EmbeddingFunctionRegistry.get_instance()
+func = registry.get("openai").create()
+
+class Words(LanceModel):
+    text: str = func.SourceField()
+    vector: Vector(func.ndims()) = func.VectorField()
+
+table = db.create_table("words", schema=Words)
+table.add(
+    [
+        {"text": "hello world"}
+        {"text": "goodbye world"}
+    ]
+    )
+
+query = "greetings"
+actual = table.search(query).limit(1).to_pydantic(Words)[0]
+print(actual.text)
+```
+
+## Multi-modal embedding functions
+Multi-modal embedding functions allow you query your table using both images and text.
+
+### OpenClipEmbeddings
+We support CLIP model embeddings using the open souce alternbative, open-clip which support various customizations. It is registered as `open-clip` and supports following customizations.
+
+
+| Parameter | Type | Default Value | Description |
+|---|---|---|---|
+| `name` | `str` | `"ViT-B-32"` | The name of the model. |
+| `pretrained` | `str` | `"laion2b_s34b_b79k"` | The name of the pretrained model to load. |
+| `device` | `str` | `"cpu"` | The device to run the model on. Can be `"cpu"` or `"gpu"`. |
+| `batch_size` | `int` | `64` | The number of images to process in a batch. |
+| `normalize` | `bool` | `True` | Whether to normalize the input images before feeding them to the model. |
+
+
+This embedding function supports ingesting images as both bytes and urls. You can query them using both test and other images.
+
+NOTE:
+LanceDB supports ingesting images directly from accessible links.
+
+
+```python
+
+db = lancedb.connect(tmp_path)
+registry = EmbeddingFunctionRegistry.get_instance()
+func = registry.get("open-clip").create()
+
+class Images(LanceModel):
+    label: str
+    image_uri: str = func.SourceField() # image uri as the source
+    image_bytes: bytes = func.SourceField() # image bytes as the source
+    vector: Vector(func.ndims()) = func.VectorField() # vector column 
+    vec_from_bytes: Vector(func.ndims()) = func.VectorField() # Another vector column 
+
+table = db.create_table("images", schema=Images)
+labels = ["cat", "cat", "dog", "dog", "horse", "horse"]
+uris = [
+    "http://farm1.staticflickr.com/53/167798175_7c7845bbbd_z.jpg",
+    "http://farm1.staticflickr.com/134/332220238_da527d8140_z.jpg",
+    "http://farm9.staticflickr.com/8387/8602747737_2e5c2a45d4_z.jpg",
+    "http://farm5.staticflickr.com/4092/5017326486_1f46057f5f_z.jpg",
+    "http://farm9.staticflickr.com/8216/8434969557_d37882c42d_z.jpg",
+    "http://farm6.staticflickr.com/5142/5835678453_4f3a4edb45_z.jpg",
+]
+# get each uri as bytes
+image_bytes = [requests.get(uri).content for uri in uris]
+table.add(
+    [{"label": labels, "image_uri": uris, "image_bytes": image_bytes}]
+)
+```
+Now we can search using text from both the default vector column and the custom vector column
+```python
+
+# text search
+actual = table.search("man's best friend").limit(1).to_pydantic(Images)[0]
+print(actual.label) # prints "dog"
+
+frombytes = (
+    table.search("man's best friend", vector_column_name="vec_from_bytes")
+    .limit(1)
+    .to_pydantic(Images)[0]
+)
+print(frombytes.label)
+
+```
+
+Because we're using a multi-modal embedding function, we can also search using images
+
+```python
+# image search
+query_image_uri = "http://farm1.staticflickr.com/200/467715466_ed4a31801f_z.jpg"
+image_bytes = requests.get(query_image_uri).content
+query_image = Image.open(io.BytesIO(image_bytes))
+actual = table.search(query_image).limit(1).to_pydantic(Images)[0]
+print(actual.label == "dog")
+
+# image search using a custom vector column
+other = (
+    table.search(query_image, vector_column_name="vec_from_bytes")
+    .limit(1)
+    .to_pydantic(Images)[0]
+)
+print(actual.label)
+
+```
+
+If you have any questions about the embeddings API, supported models, or see a relevant model missing, please raise an issue.

docs/src/embeddings/embedding_functions.md

@@ -0,0 +1,82 @@
+Representing multi-modal data as vector embeddings is becoming a standard practice. Embedding functions themselves be thought of as a part of the processing pipeline that each request(input) has to be passed through. After initial setup these components are not expected to change for a particular project. 
+
+This is main motivation behind our new embedding functions API, that allow you simply set it up once and the table remembers it, effectively making the **embedding functions disappear in the background** so you don't have to worry about modelling and simply focus on the DB aspects of VectorDB.
+
+
+You can simply follow these steps and forget about the details of your embedding functions as long as you don't intend to change it.
+
+### Step 1 - Define the embedding function
+We have some pre-defined embedding functions in the global registry with more coming soon. Here's let's an implementation of CLIP as example.
+```
+registry = EmbeddingFunctionRegistry.get_instance()
+clip = registry.get("open-clip").create()
+
+```
+You can also define your own embedding function by implementing the `EmbeddingFunction` abstract base interface. It subclasses PyDantic Model which can be utilized to write complex schemas simply as we'll see next!
+
+### Step 2 - Define the Data Model or Schema
+Our embedding function from the previous section abstracts away all the details about the models and dimensions required to define the schema. You can simply set a feild as **source** or **vector** column. Here's how
+
+```python
+class Pets(LanceModel):
+    vector: Vector(clip.ndims) = clip.VectorField()
+    image_uri: str = clip.SourceField()
+
+```
+`VectorField` tells LanceDB to use the clip embedding function to generate query embeddings for `vector` column & `SourceField` tells that when adding data, automatically use the embedding function to encode `image_uri`.
+
+
+### Step 3 - Create LanceDB Table
+Now that we have chosen/defined our embedding function and the schema, we can create the table
+
+```python
+db = lancedb.connect("~/lancedb")
+table = db.create_table("pets", schema=Pets)
+
+```
+That's it! We have ingested all the information needed to embed source and query inputs. We can now forget about the model and dimension details and start to build or VectorDB
+
+### Step 4 - Ingest lots of data and run vector search!
+Now you can just add the data and it'll be vectorized automatically
+
+```python
+table.add([{"image_uri": u} for u in uris])
+```
+
+Our OpenCLIP query embedding function support querying via both text and images. 
+
+```python
+result = table.search("dog")
+```
+
+Let's query an image
+
+```python
+p = Path("path/to/images/samoyed_100.jpg")
+query_image = Image.open(p)
+table.search(query_image)
+
+```
+
+### A little fun with PyDantic
+LanceDB is integrated with PyDantic. Infact we've used the integration in the above example to define the schema. It is also being used behing the scene by the embdding function API to ingest useful information as table metadata.
+You can also use it for adding utility operations in the schema. For example, in our multi-modal example, you can search images using text or another image. Let us define a utility function to plot the image.
+```python
+class Pets(LanceModel):
+    vector: Vector(clip.ndims) = clip.VectorField()
+    image_uri: str = clip.SourceField()
+
+    @property
+    def image(self):
+        return Image.open(self.image_uri)
+```
+Now, you can covert your search results to pydantic model and use this property.
+
+```python
+rs = table.search(query_image).limit(3).to_pydantic(Pets)
+rs[2].image
+```
+
+![](../assets/dog_clip_output.png)
+
+Now that you've the basic idea about LanceDB embedding function, let us now dive deeper into the API that you can use to implement your own embedding functions!

docs/src/embeddings/index.md

@@ -1,13 +1,20 @@
-# Embedding Functions
+# Embedding
 
-Embeddings are high dimensional floating-point vector representations of your data or query.
-Anything can be embedded using some embedding model or function.
-For a given embedding function, the output will always have the same number of dimensions.
+Embeddings are high dimensional floating-point vector representations of your data or query. Anything can be embedded using some embedding model or function. Position of embedding in a high dimensional vector space has semantic significance to a degree that depends on the type of modal and training. These embeddings when projected in a 2-D space generally group similar entities close-by forming groups.
 
-## Creating an embedding function
+![](../assets/embedding_intro.png)
 
-Any function that takes as input a batch (list) of data and outputs a batch (list) of embeddings
-can be used by LanceDB as an embedding function. The input and output batch sizes should be the same.
+# Creating an embedding function
+
+LanceDB supports 2 major ways of vectorizing your data, explicit and implicit.
+
+1. By manually embedding the data before ingesting in the table
+2. By automatically embedding the data and query as they come, by ingesting embedding function information in the table itself! Covered in [Next Section](embedding_functions.md)
+
+Whatever workflow you prefer, we have the tools to support you.
+## Explicit Vectorization
+
+In this workflow, you can create your embedding function and vectorize your data using lancedb's `with_embedding` function. Let's look at some examples.
 
 ### HuggingFace example
 
@@ -134,9 +141,9 @@ belong in the same latent space and your results will be nonsensical.
      The above snippet returns an array of records with the 10 closest vectors to the query.
 
 
-## Roadmap
+## Implicit vectorization / Ingesting embedding functions
+Representing multi-modal data as vector embeddings is becoming a standard practice. Embedding functions themselves be thought of as a part of the processing pipeline that each request(input) has to be passed through. After initial setup these components are not expected to change for a particular project. 
 
-In the near future, we'll be integrating the embedding functions deeper into LanceDB<br/>.
-The goal is that you just have to configure the function once when you create the table,
-and then you'll never have to deal with embeddings / vectors after that unless you want to.
-We'll also integrate more popular models and APIs.
+This is main motivation behind our new embedding functions API, that allow you simply set it up once and the table remembers it, effectively making the **embedding functions disappear in the background** so you don't have to worry about modelling and simply focus on the DB aspects of VectorDB.
+
+Learn more in the Next Section

docs/src/guides/tables.md

@@ -251,8 +251,9 @@ After a table has been created, you can always add more data to it using
     ### Adding Pandas DataFrame
 
     ```python
-    df = pd.DataFrame([{"vector": [1.3, 1.4], "item": "fizz", "price": 100.0},
-                  {"vector": [9.5, 56.2], "item": "buzz", "price": 200.0}])
+    df = pd.DataFrame({
+        "vector": [[1.3, 1.4], [9.5, 56.2]], "item": ["fizz", "buzz"], "price": [100.0, 200.0]
+    })
     tbl.add(df)
     ```
 
@@ -261,17 +262,12 @@ After a table has been created, you can always add more data to it using
     ### Adding to table using Iterator
 
     ```python
-    import pandas as pd
-
     def make_batches():
         for i in range(5):
-            yield pd.DataFrame(
-                {
-                    "vector": [[3.1, 4.1], [1, 1]],
-                    "item": ["foo", "bar"],
-                    "price": [10.0, 20.0],
-                })
-
+            yield [
+                    {"vector": [3.1, 4.1], "item": "foo", "price": 10.0},
+                    {"vector": [5.9, 26.5], "item": "bar", "price": 20.0}
+                ]
     tbl.add(make_batches())
     ```
 
@@ -306,9 +302,10 @@ Use the `delete()` method on tables to delete rows from a table. To choose which
 
     ```python
     import lancedb
-    import pandas as pd
 
-    data = pd.DataFrame({"x": [1, 2, 3], "vector": [[1, 2], [3, 4], [5, 6]]})
+    data = [{"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)
     table.to_pandas()

docs/src/notebooks/tables_guide.ipynb

@@ -114,13 +114,10 @@
     }
    ],
    "source": [
-    "import pandas as pd\n",
-    "\n",
-    "data = pd.DataFrame({\n",
-    "    \"vector\": [[1.1, 1.2], [0.2, 1.8]],\n",
-    "    \"lat\": [45.5, 40.1],\n",
-    "    \"long\": [-122.7, -74.1]\n",
-    "})\n",
+    "data = [\n",
+    "    {\"vector\": [1.1, 1.2], \"lat\": 45.5, \"long\": -122.7},\n",
+    "    {\"vector\": [0.2, 1.8], \"lat\": 40.1, \"long\": -74.1},\n",
+    "]\n",
     "\n",
     "db.create_table(\"table2\", data)\n",
     "\n",
@@ -366,11 +363,11 @@
     "def make_batches():\n",
     "    for i in range(5):\n",
     "        yield pd.DataFrame(\n",
-    "                    {\n",
-    "                        \"vector\": [[3.1, 4.1], [1, 1]],\n",
-    "                        \"item\": [\"foo\", \"bar\"],\n",
-    "                        \"price\": [10.0, 20.0],\n",
-    "                    })\n",
+    "                {\n",
+    "                    \"vector\": [[3.1, 4.1], [1, 1]],\n",
+    "                    \"item\": [\"foo\", \"bar\"],\n",
+    "                    \"price\": [10.0, 20.0],\n",
+    "                })\n",
     "\n",
     "tbl = db.create_table(\"table5\", make_batches(), schema=PydanticSchema)\n",
     "tbl.schema"
@@ -572,9 +569,11 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "df = pd.DataFrame([{\"vector\": [1.3, 1.4], \"item\": \"fizz\", \"price\": 100.0},\n",
-    "              {\"vector\": [9.5, 56.2], \"item\": \"buzz\", \"price\": 200.0}])\n",
-    "tbl.add(df)"
+    "data = [\n",
+    "        {\"vector\": [1.3, 1.4], \"item\": \"fizz\", \"price\": 100.0},\n",
+    "        {\"vector\": [9.5, 56.2], \"item\": \"buzz\", \"price\": 200.0}\n",
+    "]\n",
+    "tbl.add(data)"
    ]
   },
   {
@@ -596,17 +595,12 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "\n",
-    "import pandas as pd\n",
-    "\n",
     "def make_batches():\n",
     "    for i in range(5):\n",
-    "        yield pd.DataFrame(\n",
-    "                {\n",
-    "                    \"vector\": [[3.1, 4.1], [1, 1]],\n",
-    "                    \"item\": [\"foo\", \"bar\"],\n",
-    "                    \"price\": [10.0, 20.0],\n",
-    "                })\n",
+    "        yield [\n",
+    "                  {\"vector\": [3.1, 4.1], \"item\": \"foo\", \"price\": 10.0},\n",
+    "                  {\"vector\": [1, 1], \"item\": \"bar\", \"price\": 20.0},\n",
+    "              ]\n",
     "tbl.add(make_batches())"
    ]
   },

docs/src/python/arrow.md

@@ -39,7 +39,6 @@ to lazily generate data:
 
 from typing import Iterable
 import pyarrow as pa
-import lancedb
 
 def make_batches() -> Iterable[pa.RecordBatch]:
     for i in range(5):

docs/src/python/duckdb.md

@@ -11,15 +11,13 @@ pip install duckdb lancedb
 We will re-use [the dataset created previously](./arrow.md):
 
 ```python
-import pandas as pd
 import lancedb
 
 db = lancedb.connect("data/sample-lancedb")
-data = pd.DataFrame({
-    "vector": [[3.1, 4.1], [5.9, 26.5]],
-    "item": ["foo", "bar"],
-    "price": [10.0, 20.0]
-})
+data = [
+    {"vector": [3.1, 4.1], "item": "foo", "price": 10.0},
+    {"vector": [5.9, 26.5], "item": "bar", "price": 20.0}
+]
 table = db.create_table("pd_table", data=data)
 arrow_table = table.to_arrow()
 ```

docs/test/md_testing.js

@@ -8,6 +8,7 @@ const excludedGlobs = [
   "../src/embedding.md",
   "../src/examples/*.md",
   "../src/guides/tables.md",
+  "../src/embeddings/*.md",
 ];
 
 const nodePrefix = "javascript";

docs/test/md_testing.py

@@ -10,6 +10,7 @@ excluded_globs = [
     "../src/integrations/voxel51.md",
     "../src/guides/tables.md",
     "../src/python/duckdb.md",
+    "../src/embeddings/*.md",
 ]
 
 python_prefix = "py"

node/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "vectordb",
-  "version": "0.3.0",
+  "version": "0.3.1",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "vectordb",
-      "version": "0.3.0",
+      "version": "0.3.1",
       "cpu": [
         "x64",
         "arm64"
@@ -53,11 +53,11 @@
         "uuid": "^9.0.0"
       },
       "optionalDependencies": {
-        "@lancedb/vectordb-darwin-arm64": "0.3.0",
-        "@lancedb/vectordb-darwin-x64": "0.3.0",
-        "@lancedb/vectordb-linux-arm64-gnu": "0.3.0",
-        "@lancedb/vectordb-linux-x64-gnu": "0.3.0",
-        "@lancedb/vectordb-win32-x64-msvc": "0.3.0"
+        "@lancedb/vectordb-darwin-arm64": "0.3.1",
+        "@lancedb/vectordb-darwin-x64": "0.3.1",
+        "@lancedb/vectordb-linux-arm64-gnu": "0.3.1",
+        "@lancedb/vectordb-linux-x64-gnu": "0.3.1",
+        "@lancedb/vectordb-win32-x64-msvc": "0.3.1"
       }
     },
     "node_modules/@apache-arrow/ts": {
@@ -317,9 +317,9 @@
       }
     },
     "node_modules/@lancedb/vectordb-darwin-arm64": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.0.tgz",
-      "integrity": "sha512-Fg+k/cSnqmNQlSWyDp0PpaAJ67kAISfZAD+zZ3mcE8/3ml2I/wM/GVjPy2zeiQX9aR93lG1mZXFSNTDUc74tWQ==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.1.tgz",
+      "integrity": "sha512-h3yUP249xaO3rrRuVC4oRxEm5/9T66CGKiI8OwYCJUOEFrfz/jj+6PK8geMn7IqbPnOY9YRPSEi/Cc3EdFd6Sg==",
       "cpu": [
         "arm64"
       ],
@@ -329,9 +329,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-darwin-x64": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.0.tgz",
-      "integrity": "sha512-CXp4b/brMbnBPZuGzKIOskd9uD90R73rWubaJ0du/Kt6fcyQX1dM1wEhWTLxI6eKf8IDL/R9QLL2cIahm1J86w==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.1.tgz",
+      "integrity": "sha512-SQ32iMMVfvjXgvFGSGdsXcSnVDypR6eE06d7VIXsuKAg6P9e1XUhB4YcsHGeAEEv3gEoUSgsljo92ZvXJcWouQ==",
       "cpu": [
         "x64"
       ],
@@ -341,9 +341,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-linux-arm64-gnu": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.0.tgz",
-      "integrity": "sha512-1bjaRzYcDsWIRUbO2K/f+ohNmNvCgKcrrOhmiXSHVlYY8kH1LUMFZj+BhqBC0Ea0Stt7/1rsRLMRXRtaeVOEHw==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.1.tgz",
+      "integrity": "sha512-+jk2nJnaIWTqcOAyix2y+ClLNM5ECIdwyHZp5KjDqOlP6Z7eb5V2Xsah0AFp8nX3BiRRvqj3zR3zi26D7OBnYw==",
       "cpu": [
         "arm64"
       ],
@@ -353,9 +353,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-linux-x64-gnu": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.0.tgz",
-      "integrity": "sha512-BEDIJ6ReGAi+tLTS/RzxIw621yo1UUUiVNTzPGV2didyiJCr1chIGbES+39d/wiFQM43Xs3CBZLNzp+jKkv0/w==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.1.tgz",
+      "integrity": "sha512-I42Zf2lH8SUZLLYDDG4kzZ8iPq2wf1cXMh9iKNiLwgl5BnRsZVQ5A5k0uCX7IV7FcnHL/febKOxixXQyoKNAzw==",
       "cpu": [
         "x64"
       ],
@@ -365,9 +365,9 @@
       ]
     },
     "node_modules/@lancedb/vectordb-win32-x64-msvc": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.0.tgz",
-      "integrity": "sha512-7K2kbWbShuifQF/6L/tWSz2DhKfIreHKlBdVOuBTYYOReQMHn5cJxgwuFgQHqMubZ9zcagtHpmo+Wtqd034OKQ==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.1.tgz",
+      "integrity": "sha512-3OBS+fc4kcwhkqIy5b2Nump/iYoAgQd6gmYIJux3LJbMCc4yDcPJdFGVQkWu43JfBh7YOWPfOng2NSCUDBGmoA==",
       "cpu": [
         "x64"
       ],
@@ -4869,33 +4869,33 @@
       }
     },
     "@lancedb/vectordb-darwin-arm64": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.0.tgz",
-      "integrity": "sha512-Fg+k/cSnqmNQlSWyDp0PpaAJ67kAISfZAD+zZ3mcE8/3ml2I/wM/GVjPy2zeiQX9aR93lG1mZXFSNTDUc74tWQ==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-arm64/-/vectordb-darwin-arm64-0.3.1.tgz",
+      "integrity": "sha512-h3yUP249xaO3rrRuVC4oRxEm5/9T66CGKiI8OwYCJUOEFrfz/jj+6PK8geMn7IqbPnOY9YRPSEi/Cc3EdFd6Sg==",
       "optional": true
     },
     "@lancedb/vectordb-darwin-x64": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.0.tgz",
-      "integrity": "sha512-CXp4b/brMbnBPZuGzKIOskd9uD90R73rWubaJ0du/Kt6fcyQX1dM1wEhWTLxI6eKf8IDL/R9QLL2cIahm1J86w==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-darwin-x64/-/vectordb-darwin-x64-0.3.1.tgz",
+      "integrity": "sha512-SQ32iMMVfvjXgvFGSGdsXcSnVDypR6eE06d7VIXsuKAg6P9e1XUhB4YcsHGeAEEv3gEoUSgsljo92ZvXJcWouQ==",
       "optional": true
     },
     "@lancedb/vectordb-linux-arm64-gnu": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.0.tgz",
-      "integrity": "sha512-1bjaRzYcDsWIRUbO2K/f+ohNmNvCgKcrrOhmiXSHVlYY8kH1LUMFZj+BhqBC0Ea0Stt7/1rsRLMRXRtaeVOEHw==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-arm64-gnu/-/vectordb-linux-arm64-gnu-0.3.1.tgz",
+      "integrity": "sha512-+jk2nJnaIWTqcOAyix2y+ClLNM5ECIdwyHZp5KjDqOlP6Z7eb5V2Xsah0AFp8nX3BiRRvqj3zR3zi26D7OBnYw==",
       "optional": true
     },
     "@lancedb/vectordb-linux-x64-gnu": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.0.tgz",
-      "integrity": "sha512-BEDIJ6ReGAi+tLTS/RzxIw621yo1UUUiVNTzPGV2didyiJCr1chIGbES+39d/wiFQM43Xs3CBZLNzp+jKkv0/w==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-linux-x64-gnu/-/vectordb-linux-x64-gnu-0.3.1.tgz",
+      "integrity": "sha512-I42Zf2lH8SUZLLYDDG4kzZ8iPq2wf1cXMh9iKNiLwgl5BnRsZVQ5A5k0uCX7IV7FcnHL/febKOxixXQyoKNAzw==",
       "optional": true
     },
     "@lancedb/vectordb-win32-x64-msvc": {
-      "version": "0.3.0",
-      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.0.tgz",
-      "integrity": "sha512-7K2kbWbShuifQF/6L/tWSz2DhKfIreHKlBdVOuBTYYOReQMHn5cJxgwuFgQHqMubZ9zcagtHpmo+Wtqd034OKQ==",
+      "version": "0.3.1",
+      "resolved": "https://registry.npmjs.org/@lancedb/vectordb-win32-x64-msvc/-/vectordb-win32-x64-msvc-0.3.1.tgz",
+      "integrity": "sha512-3OBS+fc4kcwhkqIy5b2Nump/iYoAgQd6gmYIJux3LJbMCc4yDcPJdFGVQkWu43JfBh7YOWPfOng2NSCUDBGmoA==",
       "optional": true
     },
     "@neon-rs/cli": {

node/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vectordb",
-  "version": "0.3.0",
+  "version": "0.3.2",
   "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": "0.3.0",
-    "@lancedb/vectordb-darwin-x64": "0.3.0",
-    "@lancedb/vectordb-linux-arm64-gnu": "0.3.0",
-    "@lancedb/vectordb-linux-x64-gnu": "0.3.0",
-    "@lancedb/vectordb-win32-x64-msvc": "0.3.0"
+    "@lancedb/vectordb-darwin-arm64": "0.3.2",
+    "@lancedb/vectordb-darwin-x64": "0.3.2",
+    "@lancedb/vectordb-linux-arm64-gnu": "0.3.2",
+    "@lancedb/vectordb-linux-x64-gnu": "0.3.2",
+    "@lancedb/vectordb-win32-x64-msvc": "0.3.2"
   }
 }

node/src/arrow.ts

@@ -20,7 +20,7 @@ import {
   Utf8,
   type Vector,
   FixedSizeList,
-  vectorFromArray, type Schema, Table as ArrowTable
+  vectorFromArray, type Schema, Table as ArrowTable, RecordBatchStreamWriter
 } from 'apache-arrow'
 import { type EmbeddingFunction } from './index'
 
@@ -77,7 +77,9 @@ function newVectorBuilder (dim: number): FixedSizeListBuilder<Float32> {
 
 // Creates the Arrow Type for a Vector column with dimension `dim`
 function newVectorType (dim: number): FixedSizeList<Float32> {
-  const children = new Field<Float32>('item', new Float32())
+  // Somewhere we always default to have the elements nullable, so we need to set it to true
+  // otherwise we often get schema mismatches because the stored data always has schema with nullable elements
+  const children = new Field<Float32>('item', new Float32(), true)
   return new FixedSizeList(dim, children)
 }
 
@@ -88,6 +90,13 @@ export async function fromRecordsToBuffer<T> (data: Array<Record<string, unknown
   return Buffer.from(await writer.toUint8Array())
 }
 
+// Converts an Array of records into Arrow IPC stream format
+export async function fromRecordsToStreamBuffer<T> (data: Array<Record<string, unknown>>, embeddings?: EmbeddingFunction<T>): Promise<Buffer> {
+  const table = await convertToTable(data, embeddings)
+  const writer = RecordBatchStreamWriter.writeAll(table)
+  return Buffer.from(await writer.toUint8Array())
+}
+
 // Converts an Arrow Table into Arrow IPC format
 export async function fromTableToBuffer<T> (table: ArrowTable, embeddings?: EmbeddingFunction<T>): Promise<Buffer> {
   if (embeddings !== undefined) {
@@ -105,6 +114,23 @@ export async function fromTableToBuffer<T> (table: ArrowTable, embeddings?: Embe
   return Buffer.from(await writer.toUint8Array())
 }
 
+// Converts an Arrow Table into Arrow IPC stream format
+export async function fromTableToStreamBuffer<T> (table: ArrowTable, embeddings?: EmbeddingFunction<T>): Promise<Buffer> {
+  if (embeddings !== undefined) {
+    const source = table.getChild(embeddings.sourceColumn)
+
+    if (source === null) {
+      throw new Error(`The embedding source column ${embeddings.sourceColumn} was not found in the Arrow Table`)
+    }
+
+    const vectors = await embeddings.embed(source.toArray() as T[])
+    const column = vectorFromArray(vectors, newVectorType(vectors[0].length))
+    table = table.assign(new ArrowTable({ vector: column }))
+  }
+  const writer = RecordBatchStreamWriter.writeAll(table)
+  return Buffer.from(await writer.toUint8Array())
+}
+
 // Creates an empty Arrow Table
 export function createEmptyTable (schema: Schema): ArrowTable {
   return new ArrowTable(schema)

node/src/remote/client.ts

@@ -108,13 +108,18 @@ export class HttpLancedbClient {
   /**
    * Sent POST request.
    */
-  public async post (path: string, data?: any, params?: Record<string, string | number>): Promise<AxiosResponse> {
+  public async post (
+    path: string,
+    data?: any,
+    params?: Record<string, string | number>,
+    content?: string | undefined
+  ): Promise<AxiosResponse> {
     const response = await axios.post(
         `${this._url}${path}`,
         data,
         {
           headers: {
-            'Content-Type': 'application/json',
+            'Content-Type': content ?? 'application/json',
             'x-api-key': this._apiKey(),
             ...(this._dbName !== undefined ? { 'x-lancedb-database': this._dbName } : {})
           },

node/src/remote/index.ts

@@ -18,8 +18,10 @@ import {
 } from '../index'
 import { Query } from '../query'
 
-import { Vector } from 'apache-arrow'
+import { Vector, Table as ArrowTable } from 'apache-arrow'
 import { HttpLancedbClient } from './client'
+import { isEmbeddingFunction } from '../embedding/embedding_function'
+import { createEmptyTable, fromRecordsToStreamBuffer, fromTableToStreamBuffer } from '../arrow'
 
 /**
  * Remote connection.
@@ -66,8 +68,60 @@ export class RemoteConnection implements Connection {
     }
   }
 
-  async createTable<T> (name: string | CreateTableOptions<T>, data?: Array<Record<string, unknown>>, optsOrEmbedding?: WriteOptions | EmbeddingFunction<T>, opt?: WriteOptions): Promise<Table<T>> {
-    throw new Error('Not implemented')
+  async createTable<T> (nameOrOpts: string | CreateTableOptions<T>, data?: Array<Record<string, unknown>>, optsOrEmbedding?: WriteOptions | EmbeddingFunction<T>, opt?: WriteOptions): Promise<Table<T>> {
+    // Logic copied from LocatlConnection, refactor these to a base class + connectionImpl pattern
+    let schema
+    let embeddings: undefined | EmbeddingFunction<T>
+    let tableName: string
+    if (typeof nameOrOpts === 'string') {
+      if (optsOrEmbedding !== undefined && isEmbeddingFunction(optsOrEmbedding)) {
+        embeddings = optsOrEmbedding
+      }
+      tableName = nameOrOpts
+    } else {
+      schema = nameOrOpts.schema
+      embeddings = nameOrOpts.embeddingFunction
+      tableName = nameOrOpts.name
+    }
+
+    let buffer: Buffer
+
+    function isEmpty (data: Array<Record<string, unknown>> | ArrowTable<any>): boolean {
+      if (data instanceof ArrowTable) {
+        return data.data.length === 0
+      }
+      return data.length === 0
+    }
+
+    if ((data === undefined) || isEmpty(data)) {
+      if (schema === undefined) {
+        throw new Error('Either data or schema needs to defined')
+      }
+      buffer = await fromTableToStreamBuffer(createEmptyTable(schema))
+    } else if (data instanceof ArrowTable) {
+      buffer = await fromTableToStreamBuffer(data, embeddings)
+    } else {
+      // data is Array<Record<...>>
+      buffer = await fromRecordsToStreamBuffer(data, embeddings)
+    }
+
+    const res = await this._client.post(
+      `/v1/table/${tableName}/create/`,
+      buffer,
+      undefined,
+      'application/vnd.apache.arrow.stream'
+    )
+    if (res.status !== 200) {
+      throw new Error(`Server Error, status: ${res.status}, ` +
+      // eslint-disable-next-line @typescript-eslint/restrict-template-expressions
+        `message: ${res.statusText}: ${res.data}`)
+    }
+
+    if (embeddings === undefined) {
+      return new RemoteTable(this._client, tableName)
+    } else {
+      return new RemoteTable(this._client, tableName, embeddings)
+    }
   }
 
   async dropTable (name: string): Promise<void> {
@@ -141,11 +195,39 @@ export class RemoteTable<T = number[]> implements Table<T> {
   }
 
   async add (data: Array<Record<string, unknown>>): Promise<number> {
-    throw new Error('Not implemented')
+    const buffer = await fromRecordsToStreamBuffer(data, this._embeddings)
+    const res = await this._client.post(
+      `/v1/table/${this._name}/insert/`,
+      buffer,
+      {
+        mode: 'append'
+      },
+      'application/vnd.apache.arrow.stream'
+    )
+    if (res.status !== 200) {
+      throw new Error(`Server Error, status: ${res.status}, ` +
+      // eslint-disable-next-line @typescript-eslint/restrict-template-expressions
+        `message: ${res.statusText}: ${res.data}`)
+    }
+    return data.length
   }
 
   async overwrite (data: Array<Record<string, unknown>>): Promise<number> {
-    throw new Error('Not implemented')
+    const buffer = await fromRecordsToStreamBuffer(data, this._embeddings)
+    const res = await this._client.post(
+      `/v1/table/${this._name}/insert/`,
+      buffer,
+      {
+        mode: 'overwrite'
+      },
+      'application/vnd.apache.arrow.stream'
+    )
+    if (res.status !== 200) {
+      throw new Error(`Server Error, status: ${res.status}, ` +
+      // eslint-disable-next-line @typescript-eslint/restrict-template-expressions
+        `message: ${res.statusText}: ${res.data}`)
+    }
+    return data.length
   }
 
   async createIndex (indexParams: VectorIndexParams): Promise<any> {
@@ -157,6 +239,6 @@ export class RemoteTable<T = number[]> implements Table<T> {
   }
 
   async delete (filter: string): Promise<void> {
-    throw new Error('Not implemented')
+    await this._client.post(`/v1/table/${this._name}/delete/`, { predicate: filter })
   }
 }

python/LICENSE

@@ -0,0 +1 @@
+../LICENSE

python/lancedb/embeddings/__init__.py

@@ -21,5 +21,6 @@ from .functions import (
     OpenClipEmbeddings,
     SentenceTransformerEmbeddings,
     TextEmbeddingFunction,
+    register,
 )
 from .utils import with_embeddings

python/lancedb/remote/table.py

@@ -105,4 +105,8 @@ class RemoteTable(Table):
         return self._conn._loop.run_until_complete(result).to_arrow()
 
     def delete(self, predicate: str):
-        raise NotImplementedError
+        """Delete rows from the table."""
+        payload = {"predicate": predicate}
+        self._conn._loop.run_until_complete(
+            self._conn._client.post(f"/v1/table/{self._name}/delete/", data=payload)
+        )

python/lancedb/table.py

@@ -151,7 +151,7 @@ class Table(ABC):
     @abstractmethod
     def schema(self) -> pa.Schema:
         """The [Arrow Schema](https://arrow.apache.org/docs/python/api/datatypes.html#) of
-        this [Table](Table)
+        this Table
 
         """
         raise NotImplementedError
@@ -292,8 +292,9 @@ class Table(ABC):
         Examples
         --------
         >>> import lancedb
-        >>> import pandas as pd
-        >>> data = pd.DataFrame({"x": [1, 2, 3], "vector": [[1, 2], [3, 4], [5, 6]]})
+        >>> data = [
+        ...   {"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)
         >>> table.to_pandas()
@@ -719,8 +720,9 @@ class LanceTable(Table):
         Examples
         --------
         >>> import lancedb
-        >>> import pandas as pd
-        >>> data = pd.DataFrame({"x": [1, 2, 3], "vector": [[1, 2], [3, 4], [5, 6]]})
+        >>> data = [
+        ...   {"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)
         >>> table.to_pandas()
@@ -836,8 +838,9 @@ class LanceTable(Table):
         Examples
         --------
         >>> import lancedb
-        >>> import pandas as pd
-        >>> data = pd.DataFrame({"x": [1, 2, 3], "vector": [[1, 2], [3, 4], [5, 6]]})
+        >>> data = [
+        ...   {"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)
         >>> table.to_pandas()

python/pyproject.toml

@@ -3,7 +3,7 @@ name = "lancedb"
 version = "0.3.1"
 dependencies = [
     "deprecation",
-    "pylance==0.8.3",
+    "pylance==0.8.5",
     "ratelimiter~=1.0",
     "retry>=0.9.2",
     "tqdm>=4.1.0",

python/tests/test_table.py

@@ -458,7 +458,8 @@ def test_compact_cleanup(db):
 
     stats = table.compact_files()
     assert len(table) == 3
-    assert table.version == 4
+    # Compact_files bump 2 versions.
+    assert table.version == 5
     assert stats.fragments_removed > 0
     assert stats.fragments_added == 1
 
@@ -467,7 +468,7 @@ def test_compact_cleanup(db):
 
     stats = table.cleanup_old_versions(older_than=timedelta(0), delete_unverified=True)
     assert stats.bytes_removed > 0
-    assert table.version == 4
+    assert table.version == 5
 
     with pytest.raises(Exception, match="Version 3 no longer exists"):
         table.checkout(3)

rust/ffi/node/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb-node"
-version = "0.3.0"
+version = "0.3.2"
 description = "Serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"
 edition = "2018"

rust/ffi/node/src/lib.rs

@@ -74,7 +74,7 @@ fn runtime<'a, C: Context<'a>>(cx: &mut C) -> NeonResult<&'static Runtime> {
     static RUNTIME: OnceCell<Runtime> = OnceCell::new();
     static LOG: OnceCell<()> = OnceCell::new();
 
-    LOG.get_or_init(|| env_logger::init());
+    LOG.get_or_init(env_logger::init);
 
     RUNTIME.get_or_try_init(|| Runtime::new().or_throw(cx))
 }
@@ -148,7 +148,7 @@ fn get_aws_creds(
     match (secret_key_id, secret_key, temp_token) {
         (Some(key_id), Some(key), optional_token) => Ok(Some(Arc::new(
             StaticCredentialProvider::new(AwsCredential {
-                key_id: key_id,
+                key_id,
                 secret_key: key,
                 token: optional_token,
             }),

rust/ffi/node/src/table.rs

@@ -70,7 +70,7 @@ impl JsTable {
             store_params: Some(ObjectStoreParams::with_aws_credentials(
                 aws_creds, aws_region,
             )),
-            mode: mode,
+            mode,
             ..WriteParams::default()
         };
 
@@ -121,7 +121,7 @@ impl JsTable {
             let add_result = table.add(batch_reader, Some(params)).await;
 
             deferred.settle_with(&channel, move |mut cx| {
-                let _added = add_result.or_throw(&mut cx)?;
+                add_result.or_throw(&mut cx)?;
                 Ok(cx.boxed(JsTable::from(table)))
             });
         });

rust/vectordb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vectordb"
-version = "0.3.0"
+version = "0.3.2"
 edition = "2021"
 description = "LanceDB: A serverless, low-latency vector database for AI applications"
 license = "Apache-2.0"

rust/vectordb/src/data/inspect.rs

@@ -18,9 +18,9 @@ use arrow::compute::kernels::{aggregate::bool_and, length::length};
 use arrow_array::{
     cast::AsArray,
     types::{ArrowPrimitiveType, Int32Type, Int64Type},
-    Array, GenericListArray, OffsetSizeTrait, RecordBatchReader,
+    Array, GenericListArray, OffsetSizeTrait, PrimitiveArray, RecordBatchReader,
 };
-use arrow_ord::comparison::eq_dyn_scalar;
+use arrow_ord::cmp::eq;
 use arrow_schema::DataType;
 use num_traits::{ToPrimitive, Zero};
 
@@ -38,7 +38,8 @@ where
     }
 
     let dim = len_arr.as_primitive::<T>().value(0);
-    if bool_and(&eq_dyn_scalar(len_arr.as_primitive::<T>(), dim)?) != Some(true) {
+    let datum = PrimitiveArray::<T>::new_scalar(dim);
+    if bool_and(&eq(len_arr.as_primitive::<T>(), &datum)?) != Some(true) {
         Ok(None)
     } else {
         Ok(Some(dim))

rust/vectordb/src/database.rs

@@ -135,7 +135,7 @@ impl Database {
     async fn open_path(path: &str) -> Result<Database> {
         let (object_store, base_path) = ObjectStore::from_uri(path).await?;
         if object_store.is_local() {
-            Self::try_create_dir(path).context(CreateDirSnafu { path: path })?;
+            Self::try_create_dir(path).context(CreateDirSnafu { path })?;
         }
         Ok(Self {
             uri: path.to_string(),

rust/vectordb/src/index/vector.rs

@@ -95,8 +95,8 @@ impl VectorIndexBuilder for IvfPQIndexBuilder {
     }
 
     fn build(&self) -> VectorIndexParams {
-        let ivf_params = self.ivf_params.clone().unwrap_or(IvfBuildParams::default());
-        let pq_params = self.pq_params.clone().unwrap_or(PQBuildParams::default());
+        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)
     }

rust/vectordb/src/table.rs

@@ -339,7 +339,7 @@ impl Table {
     /// This calls into [lance::dataset::optimize::compact_files].
     pub async fn compact_files(&mut self, options: CompactionOptions) -> Result<CompactionMetrics> {
         let mut dataset = self.dataset.as_ref().clone();
-        let metrics = compact_files(&mut dataset, options).await?;
+        let metrics = compact_files(&mut dataset, options, None).await?;
         self.dataset = Arc::new(dataset);
         Ok(metrics)
     }