fix(python): preserve embedding functions in sync remote table creation (#4408)

Sync remote `create_table(..., embedding_functions=...)` discarded the
supplied configuration and only logged an unsupported-feature warning.
Schema-only creation therefore lost the metadata needed by `add()` to
embed text, while text-only initial data produced no vector column.

Forward the configuration to the existing async creation path, remove
the stale warning, and document the parameter. The shared path stores
the embedding metadata and generates vectors for initial data and
subsequent additions.

Add a regression test against the mocked remote HTTP API for both
schema-only and text-only creation. It checks the outgoing Arrow schema
metadata, the metadata returned by the remote table, inferred vector
types, and generated vectors during creation and `add()`.

Validation:

- Both regression cases failed before the fix because the outgoing
embedding metadata was missing, then passed after the fix.
- `uv run --extra tests --extra dev pytest
python/tests/test_remote_db.py python/tests/test_embeddings.py -q -m
'not slow and not s3_test'`: 82 passed, 4 skipped, 8 deselected.
- Repository-wide `ruff check .` and `ruff format .` passed using the
uv-managed environment; `git diff --check` passed.

Fixes #4376

<!-- lance-gatekeeper-fix:v1 agent=a6816b1f3335438bd8ad192f36395213
generation=1 -->

Co-authored-by: Gatefixer <313497061+lancedb-gatefixer[bot]@users.noreply.github.com>

python/python/lancedb/remote/db.py

@@ -608,6 +608,9 @@ class RemoteDBConnection(DBConnection):
             One of "error", "drop", "fill".
         fill_value: float
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
+        embedding_functions: list of EmbeddingFunctionConfig, optional
+            The embedding functions to store in the table schema and use to
+            generate vectors when creating the table or adding data.
 
         Returns
         -------
@@ -682,12 +685,6 @@ class RemoteDBConnection(DBConnection):
         if namespace_path is None:
             namespace_path = []
         validate_table_name(name)
-        if embedding_functions is not None:
-            logging.warning(
-                "embedding_functions is not yet supported on LanceDB Cloud."
-                "Please vote https://github.com/lancedb/lancedb/issues/626 "
-                "for this feature."
-            )
 
         from .table import RemoteTable
 
@@ -700,6 +697,7 @@ class RemoteDBConnection(DBConnection):
                 schema=schema,
                 on_bad_vectors=on_bad_vectors,
                 fill_value=fill_value,
+                embedding_functions=embedding_functions,
             )
         )
         return RemoteTable(

python/python/tests/test_remote_db.py

@@ -17,7 +17,8 @@ from packaging.version import Version
 
 import lancedb
 import numpy as np
-from lancedb.conftest import MockTextEmbeddingFunction
+from lancedb.conftest import MockNonNormTextEmbeddingFunction, MockTextEmbeddingFunction
+from lancedb.embeddings import EmbeddingFunctionConfig, EmbeddingFunctionRegistry
 from lancedb.query import AsyncQuery, ColumnOrdering
 from lancedb.remote import ClientConfig
 from lancedb.remote.errors import HttpError, RetryError
@@ -596,6 +597,106 @@ def test_remote_permutation_is_picklable():
         ]
 
 
+@pytest.mark.parametrize("with_schema", [True, False])
+def test_create_table_embedding_functions(with_schema):
+    func = MockNonNormTextEmbeddingFunction.create()
+    config = EmbeddingFunctionConfig(
+        source_column="text", vector_column="vector", function=func
+    )
+    schema = pa.schema(
+        [pa.field("text", pa.string()), pa.field("vector", pa.list_(pa.float32(), 10))]
+    )
+    received = {}
+
+    def handler(request):
+        if request.path == "/v1/table/test/describe/":
+            # Echo the metadata actually sent by create_table, as the server does.
+            metadata = received["create"].schema.metadata or {}
+            send_json(
+                request,
+                {
+                    "version": 1,
+                    "schema": {
+                        "fields": [
+                            {
+                                "name": "text",
+                                "type": {"type": "string"},
+                                "nullable": True,
+                            },
+                            {
+                                "name": "vector",
+                                "type": {
+                                    "type": "fixed_size_list",
+                                    "fields": [
+                                        {
+                                            "name": "item",
+                                            "type": {"type": "float"},
+                                            "nullable": True,
+                                        }
+                                    ],
+                                    "length": 10,
+                                },
+                                "nullable": True,
+                            },
+                        ],
+                        "metadata": {
+                            k.decode(): v.decode() for k, v in metadata.items()
+                        },
+                    },
+                },
+            )
+        elif request.path in (
+            "/v1/table/test/create/?mode=create",
+            "/v1/table/test/insert/",
+        ):
+            if request.headers.get("Transfer-Encoding") == "chunked":
+                body = bytearray()
+                while True:
+                    size = int(request.rfile.readline(), 16)
+                    if size == 0:
+                        request.rfile.readline()
+                        break
+                    body.extend(request.rfile.read(size))
+                    request.rfile.read(2)
+            else:
+                body = request.rfile.read(int(request.headers["Content-Length"]))
+            operation = "create" if "/create/" in request.path else "insert"
+            received[operation] = pa.ipc.open_stream(body).read_all()
+            send_json(request, {})
+        else:
+            request.send_response(404)
+            request.end_headers()
+
+    with mock_lancedb_connection(handler) as db:
+        table = db.create_table(
+            "test",
+            schema=schema if with_schema else None,
+            data=None if with_schema else [{"text": "hello world"}],
+            embedding_functions=[config],
+        )
+        metadata = pa.schema(
+            [],
+            metadata=EmbeddingFunctionRegistry.get_instance().get_table_metadata(
+                [config]
+            ),
+        ).metadata
+        assert received["create"].schema.metadata == metadata
+        assert table.schema.metadata == metadata
+        assert table.schema.field("vector").type == pa.list_(pa.float32(), 10)
+        assert table.embedding_functions["vector"].source_column == "text"
+        if with_schema:
+            assert received["create"].num_rows == 0
+        else:
+            assert received["create"]["vector"].to_pylist() == [
+                func.compute_source_embeddings(["hello world"])[0].tolist()
+            ]
+
+        table.add([{"text": "goodbye world"}])
+        assert received["insert"]["vector"].to_pylist() == [
+            func.compute_source_embeddings(["goodbye world"])[0].tolist()
+        ]
+
+
 def test_create_table_exist_ok():
     def handler(request):
         if request.path == "/v1/table/test/create/?mode=exist_ok":