feat: add LiteLLM as embedding provider

pyproject.toml

@@ -47,6 +47,7 @@ dependencies = [
     "fastembed>=0.7.4",
     "sqlite-vec>=0.1.6",
     "openai>=1.100.2",
+    "litellm>=1.60.0,<2.0.0",
     "logfire>=4.19.0",
     "psutil>=5.9.0",
 ]

src/basic_memory/repository/embedding_provider_factory.py

@@ -151,6 +151,16 @@ def create_embedding_provider(app_config: BasicMemoryConfig) -> EmbeddingProvide
             request_concurrency=app_config.semantic_embedding_request_concurrency,
             **extra_kwargs,
         )
+    elif provider_name == "litellm":
+        from basic_memory.repository.litellm_provider import LiteLLMEmbeddingProvider
+
+        model_name = app_config.semantic_embedding_model or "openai/text-embedding-3-small"
+        provider = LiteLLMEmbeddingProvider(
+            model_name=model_name,
+            batch_size=app_config.semantic_embedding_batch_size,
+            request_concurrency=app_config.semantic_embedding_request_concurrency,
+            **extra_kwargs,
+        )
     else:
         raise ValueError(f"Unsupported semantic embedding provider: {provider_name}")
 

src/basic_memory/repository/litellm_provider.py

@@ -0,0 +1,116 @@
+"""LiteLLM-based embedding provider for semantic indexing.
+
+Routes embedding requests to 100+ providers (OpenAI, Anthropic, Google, Azure,
+Bedrock, Cohere, etc.) via the litellm SDK. No proxy server needed.
+
+Model strings use the ``provider/model`` format, e.g.
+``openai/text-embedding-3-small``, ``cohere/embed-english-v3.0``,
+``azure/my-embedding-deployment``.
+
+See https://docs.litellm.ai/docs/embedding/supported_embedding for all
+supported embedding models.
+"""
+
+from __future__ import annotations
+
+import asyncio
+from typing import Any
+
+from basic_memory.repository.embedding_provider import EmbeddingProvider
+from basic_memory.repository.semantic_errors import SemanticDependenciesMissingError
+
+
+class LiteLLMEmbeddingProvider(EmbeddingProvider):
+    """Embedding provider backed by the litellm SDK."""
+
+    def __init__(
+        self,
+        model_name: str = "openai/text-embedding-3-small",
+        *,
+        batch_size: int = 64,
+        request_concurrency: int = 4,
+        dimensions: int = 1536,
+        api_key: str | None = None,
+        timeout: float = 30.0,
+    ) -> None:
+        self.model_name = model_name
+        self.dimensions = dimensions
+        self.batch_size = batch_size
+        self.request_concurrency = request_concurrency
+        self._api_key = api_key
+        self._timeout = timeout
+
+    def runtime_log_attrs(self) -> dict[str, int]:
+        """Return provider-specific runtime settings suitable for startup logs."""
+        return {
+            "provider_batch_size": self.batch_size,
+            "request_concurrency": self.request_concurrency,
+        }
+
+    async def embed_documents(self, texts: list[str]) -> list[list[float]]:
+        if not texts:
+            return []
+
+        try:
+            import litellm
+        except ImportError as exc:
+            raise SemanticDependenciesMissingError(
+                "litellm dependency is missing. Install with: pip install litellm"
+            ) from exc
+
+        batches = [
+            texts[start : start + self.batch_size]
+            for start in range(0, len(texts), self.batch_size)
+        ]
+        batch_vectors: list[list[list[float]] | None] = [None] * len(batches)
+        semaphore = asyncio.Semaphore(self.request_concurrency)
+
+        async def embed_batch(batch_index: int, batch: list[str]) -> None:
+            async with semaphore:
+                params: dict[str, Any] = {
+                    "model": self.model_name,
+                    "input": batch,
+                    "drop_params": True,
+                    "timeout": self._timeout,
+                }
+                if self._api_key:
+                    params["api_key"] = self._api_key
+
+                response = await litellm.aembedding(**params)
+
+            vectors_by_index: dict[int, list[float]] = {}
+            for item in response.data:
+                response_index = int(item["index"])
+                vectors_by_index[response_index] = [float(v) for v in item["embedding"]]
+
+            ordered_vectors: list[list[float]] = []
+            for index in range(len(batch)):
+                vector = vectors_by_index.get(index)
+                if vector is None:
+                    raise RuntimeError(
+                        "LiteLLM embedding response is missing expected vector index."
+                    )
+                ordered_vectors.append(vector)
+
+            batch_vectors[batch_index] = ordered_vectors
+
+        await asyncio.gather(
+            *(embed_batch(batch_index, batch) for batch_index, batch in enumerate(batches))
+        )
+
+        all_vectors: list[list[float]] = []
+        for vectors in batch_vectors:
+            if vectors is None:
+                raise RuntimeError("LiteLLM embedding batch did not produce vectors.")
+            all_vectors.extend(vectors)
+
+        if all_vectors and len(all_vectors[0]) != self.dimensions:
+            raise RuntimeError(
+                f"Embedding model returned {len(all_vectors[0])}-dimensional vectors "
+                f"but provider was configured for {self.dimensions} dimensions."
+            )
+        return all_vectors
+
+    async def embed_query(self, text: str) -> list[float]:
+        vectors = await self.embed_documents([text])
+        return vectors[0] if vectors else [0.0] * self.dimensions

tests/repository/test_litellm_provider.py

@@ -0,0 +1,181 @@
+"""Tests for LiteLLMEmbeddingProvider.
+
+Uses AST parsing and direct SDK mocking to avoid importing the full
+basic_memory dependency chain (logfire, alembic, etc.).
+"""
+
+import ast
+import sys
+import types
+from pathlib import Path
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+
+PROVIDER_PATH = (
+    Path(__file__).resolve().parents[2]
+    / "src"
+    / "basic_memory"
+    / "repository"
+    / "litellm_provider.py"
+)
+FACTORY_PATH = (
+    Path(__file__).resolve().parents[2]
+    / "src"
+    / "basic_memory"
+    / "repository"
+    / "embedding_provider_factory.py"
+)
+
+
+class TestLiteLLMProviderStructure:
+    """Verify the provider file has the correct structure."""
+
+    def _parse(self):
+        return ast.parse(PROVIDER_PATH.read_text())
+
+    def test_file_exists(self):
+        assert PROVIDER_PATH.exists()
+
+    def test_has_litellm_embedding_provider_class(self):
+        tree = self._parse()
+        classes = [n.name for n in ast.walk(tree) if isinstance(n, ast.ClassDef)]
+        assert "LiteLLMEmbeddingProvider" in classes
+
+    def test_has_embed_documents_method(self):
+        tree = self._parse()
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ClassDef) and node.name == "LiteLLMEmbeddingProvider":
+                methods = [
+                    n.name
+                    for n in node.body
+                    if isinstance(n, (ast.FunctionDef, ast.AsyncFunctionDef))
+                ]
+                assert "embed_documents" in methods
+                assert "embed_query" in methods
+                return
+        pytest.fail("LiteLLMEmbeddingProvider class not found")
+
+    def test_embed_documents_is_async(self):
+        tree = self._parse()
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ClassDef) and node.name == "LiteLLMEmbeddingProvider":
+                for item in node.body:
+                    if isinstance(item, ast.AsyncFunctionDef) and item.name == "embed_documents":
+                        return
+        pytest.fail("embed_documents is not async")
+
+    def test_uses_drop_params_true(self):
+        src = PROVIDER_PATH.read_text()
+        assert "drop_params" in src
+
+    def test_uses_litellm_aembedding(self):
+        src = PROVIDER_PATH.read_text()
+        assert "aembedding" in src
+
+    def test_has_runtime_log_attrs(self):
+        tree = self._parse()
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ClassDef) and node.name == "LiteLLMEmbeddingProvider":
+                methods = [
+                    n.name
+                    for n in node.body
+                    if isinstance(n, (ast.FunctionDef, ast.AsyncFunctionDef))
+                ]
+                assert "runtime_log_attrs" in methods
+                return
+
+    def test_default_model_in_source(self):
+        src = PROVIDER_PATH.read_text()
+        assert "openai/text-embedding-3-small" in src
+
+
+class TestFactoryRegistration:
+    """Verify the factory recognizes litellm as a provider."""
+
+    def test_litellm_branch_in_factory(self):
+        src = FACTORY_PATH.read_text()
+        assert 'provider_name == "litellm"' in src
+
+    def test_imports_litellm_provider(self):
+        src = FACTORY_PATH.read_text()
+        assert "LiteLLMEmbeddingProvider" in src
+
+
+class TestLiteLLMSDKInteraction:
+    """Test litellm SDK calls directly (no basic_memory deps needed)."""
+
+    def test_aembedding_called_with_drop_params(self):
+        fake = types.ModuleType("litellm")
+        response = MagicMock()
+        response.data = [{"index": 0, "embedding": [0.1, 0.2]}]
+        fake.aembedding = AsyncMock(return_value=response)
+        sys.modules["litellm"] = fake
+
+        try:
+            import asyncio
+
+            async def run():
+                await fake.aembedding(
+                    model="openai/text-embedding-3-small",
+                    input=["hello"],
+                    drop_params=True,
+                )
+
+            asyncio.run(run())
+            kwargs = fake.aembedding.call_args.kwargs
+            assert kwargs["drop_params"] is True
+            assert kwargs["model"] == "openai/text-embedding-3-small"
+        finally:
+            del sys.modules["litellm"]
+
+    def test_aembedding_forwards_api_key(self):
+        fake = types.ModuleType("litellm")
+        response = MagicMock()
+        response.data = [{"index": 0, "embedding": [0.1]}]
+        fake.aembedding = AsyncMock(return_value=response)
+        sys.modules["litellm"] = fake
+
+        try:
+            import asyncio
+
+            async def run():
+                await fake.aembedding(
+                    model="openai/text-embedding-3-small",
+                    input=["hello"],
+                    api_key="sk-test",
+                    drop_params=True,
+                )
+
+            asyncio.run(run())
+            assert fake.aembedding.call_args.kwargs["api_key"] == "sk-test"
+        finally:
+            del sys.modules["litellm"]
+
+    def test_aembedding_response_has_vectors(self):
+        fake = types.ModuleType("litellm")
+        response = MagicMock()
+        response.data = [
+            {"index": 0, "embedding": [0.1, 0.2, 0.3]},
+            {"index": 1, "embedding": [0.4, 0.5, 0.6]},
+        ]
+        fake.aembedding = AsyncMock(return_value=response)
+        sys.modules["litellm"] = fake
+
+        try:
+            import asyncio
+
+            async def run():
+                resp = await fake.aembedding(
+                    model="openai/text-embedding-3-small",
+                    input=["hello", "world"],
+                    drop_params=True,
+                )
+                return resp
+
+            resp = asyncio.run(run())
+            assert len(resp.data) == 2
+            assert resp.data[0]["embedding"] == [0.1, 0.2, 0.3]
+            assert resp.data[1]["embedding"] == [0.4, 0.5, 0.6]
+        finally:
+            del sys.modules["litellm"]