feat(cli): add a note to get_embedding_function error message. Fix #46

Author: Davidyz

src/vectorcode/common.py

@@ -134,7 +134,7 @@ def get_collection_name(full_path: str) -> str:
     return collection_id
 
 
-def get_embedding_function(configs: Config) -> chromadb.EmbeddingFunction:
+def get_embedding_function(configs: Config) -> chromadb.EmbeddingFunction | None:
     try:
         return getattr(embedding_functions, configs.embedding_function)(
             **configs.embedding_params
@@ -145,6 +145,16 @@ def get_embedding_function(configs: Config) -> chromadb.EmbeddingFunction:
             file=sys.stderr,
         )
         return embedding_functions.SentenceTransformerEmbeddingFunction()
+    except Exception as e:
+        print(
+            f"Failed to use {configs.embedding_function} with the following error:",
+            file=sys.stderr,
+        )
+        e.add_note(
+            "\nFor errors caused by missing dependency, consult the documentation of pipx (or whatever package manager that you installed VectorCode with) for instructions to inject libraries into the virtual environment."
+        )
+
+        raise
 
 
 __COLLECTION_CACHE: dict[str, AsyncCollection] = {}

tests/test_common.py

@@ -9,6 +9,7 @@
 import pytest
 from chromadb.api import AsyncClientAPI
 from chromadb.api.models.AsyncCollection import AsyncCollection
+from chromadb.utils import embedding_functions
 
 from vectorcode.cli_utils import Config
 from vectorcode.common import (
@@ -68,6 +69,34 @@ def test_get_embedding_function():
     assert "SentenceTransformerEmbeddingFunction" in str(type(embedding_function))
 
 
+def test_get_embedding_function_init_exception():
+    # Test when the embedding function exists but raises an error during initialization
+    config = Config(
+        embedding_function="SentenceTransformerEmbeddingFunction",
+        embedding_params={"model_name": "non_existent_model_should_cause_error"},
+    )
+
+    # Mock SentenceTransformerEmbeddingFunction.__init__ to raise a generic exception
+    with patch.object(
+        embedding_functions, "SentenceTransformerEmbeddingFunction", autospec=True
+    ) as mock_stef:
+        # Simulate an error during the embedding function's __init__
+        mock_stef.side_effect = Exception("Simulated initialization error")
+
+        with pytest.raises(Exception) as excinfo:
+            get_embedding_function(config)
+
+        # Check if the raised exception is the one we simulated
+        assert "Simulated initialization error" in str(excinfo.value)
+        # Check if the additional note was added
+        assert "For errors caused by missing dependency" in excinfo.value.__notes__[0]
+
+        # Verify that the constructor was called with the correct parameters
+        mock_stef.assert_called_once_with(
+            model_name="non_existent_model_should_cause_error"
+        )
+
+
 @pytest.mark.asyncio
 async def test_try_server():
     # This test requires a server to be running, so it's difficult to make it truly isolated.