vector-inspector 0.3.4__tar.gz → 0.3.6__tar.gz

{vector_inspector-0.3.4 → vector_inspector-0.3.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vector-inspector
-Version: 0.3.4
+Version: 0.3.6
 Summary: A comprehensive desktop application for visualizing, querying, and managing vector database data
 Author-Email: Anthony Dawson <anthonypdawson+github@gmail.com>
 License: MIT
@@ -26,15 +26,16 @@ Requires-Dist: keyring>=25.7.0
 Requires-Dist: hf-xet>=1.2.0
 Requires-Dist: pymilvus>=2.6.6
 Requires-Dist: lancedb>=0.27.0
+Requires-Dist: psycopg2>=2.9.11
+Requires-Dist: pgvector>=0.4.2
 Description-Content-Type: text/markdown
 
-- Added and tested support for python 3.13
-- Miscellaneous documentation updates and fixes
-- Splash screen now shows after main window is visible
-
+# Latest updates
+- Added Postgres (pgvector extension) support as a new vector database connection option.
+- Fixed issue with embedding regeneration on add/edit
+- Added option for editing items without regenerating embeddings
 ---
 
-
 # Vector Inspector
 
 > **Disclaimer:** This tool is currently under active development and is **not production ready**. Not all features have been thoroughly tested and code is released frequently. Use with caution in critical or production environments.

{vector_inspector-0.3.4 → vector_inspector-0.3.6}/README.md

@@ -1,10 +1,9 @@
-- Added and tested support for python 3.13
-- Miscellaneous documentation updates and fixes
-- Splash screen now shows after main window is visible
-
+# Latest updates
+- Added Postgres (pgvector extension) support as a new vector database connection option.
+- Fixed issue with embedding regeneration on add/edit
+- Added option for editing items without regenerating embeddings
 ---
 
-
 # Vector Inspector
 
 > **Disclaimer:** This tool is currently under active development and is **not production ready**. Not all features have been thoroughly tested and code is released frequently. Use with caution in critical or production environments.

{vector_inspector-0.3.4 → vector_inspector-0.3.6}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "vector-inspector"
-version = "0.3.4"
+version = "0.3.6"
 description = "A comprehensive desktop application for visualizing, querying, and managing vector database data"
 authors = [
     { name = "Anthony Dawson", email = "anthonypdawson+github@gmail.com" },
@@ -23,6 +23,8 @@ dependencies = [
     "hf-xet>=1.2.0",
     "pymilvus>=2.6.6",
     "lancedb>=0.27.0",
+    "psycopg2>=2.9.11",
+    "pgvector>=0.4.2",
 ]
 requires-python = ">=3.10,<3.14"
 readme = "README.md"

{vector_inspector-0.3.4 → vector_inspector-0.3.6}/src/vector_inspector/core/connections/base_connection.py

@@ -261,7 +261,7 @@ class VectorDBConnection(ABC):
 
             # Finally, check user settings (for collections we can't modify)
             if connection_id:
-                from ...services.settings_service import SettingsService
+                from vector_inspector.services.settings_service import SettingsService
 
                 settings = SettingsService()
                 model_info = settings.get_embedding_model(connection_id, collection_name)
@@ -272,3 +272,88 @@ class VectorDBConnection(ABC):
         except Exception as e:
             log_error("Failed to get embedding model: %s", e)
             return None
+
+    def load_embedding_model_for_collection(
+        self, collection_name: str, connection_id: Optional[str] = None
+    ):
+        """
+        Resolve and load an embedding model for a collection.
+
+        Resolution order:
+        1. User settings (SettingsService)
+        2. Collection metadata (get_collection_info)
+        3. Dimension-based registry (embedding_utils.get_embedding_model_for_dimension)
+        4. DEFAULT_MODEL
+
+        Returns:
+            (loaded_model, model_name, model_type)
+        """
+        try:
+            from vector_inspector.services.settings_service import SettingsService
+            from vector_inspector.core.embedding_utils import (
+                load_embedding_model,
+                get_embedding_model_for_dimension,
+                DEFAULT_MODEL,
+            )
+
+            # 1) settings
+            if connection_id:
+                settings = SettingsService()
+                cfg = settings.get_embedding_model(connection_id, collection_name)
+                if cfg and cfg.get("model"):
+                    model_name = cfg.get("model")
+                    model_type = cfg.get("type", "sentence-transformer")
+                    model = load_embedding_model(model_name, model_type)
+                    return (model, model_name, model_type)
+
+            # 2) collection metadata
+            try:
+                info = self.get_collection_info(collection_name)
+            except Exception:
+                info = None
+
+            if info and info.get("embedding_model"):
+                model_name = info.get("embedding_model")
+                model_type = info.get("embedding_model_type", "sentence-transformer")
+                model = load_embedding_model(model_name, model_type)
+                return (model, model_name, model_type)
+
+            # 3) dimension based
+            if info and info.get("vector_dimension"):
+                try:
+                    dim = int(info.get("vector_dimension"))
+                    model, model_name, model_type = get_embedding_model_for_dimension(dim)
+                    return (model, model_name, model_type)
+                except Exception:
+                    pass
+
+            # 4) fallback
+            model_name, model_type = DEFAULT_MODEL
+            model = load_embedding_model(model_name, model_type)
+            return (model, model_name, model_type)
+        except Exception as e:
+            log_error("Failed to load embedding model for collection %s: %s", collection_name, e)
+            raise
+
+    def compute_embeddings_for_documents(
+        self, collection_name: str, documents: List[str], connection_id: Optional[str] = None
+    ) -> List[List[float]]:
+        """
+        Compute embeddings for a list of documents using the resolved model for the collection.
+
+        Returns a list of embedding vectors (one per document). If encoding fails,
+        raises an exception.
+        """
+        model, model_name, model_type = self.load_embedding_model_for_collection(
+            collection_name, connection_id
+        )
+
+        # Use batch encoding when available (sentence-transformer), otherwise per-doc
+        if model_type != "clip":
+            # sentence-transformer-like models support batch encode
+            return model.encode(documents, show_progress_bar=False).tolist()
+        else:
+            # CLIP - use encode_text helper for each document
+            from vector_inspector.core.embedding_utils import encode_text
+
+            return [encode_text(d, model, model_type) for d in documents]

{vector_inspector-0.3.4 → vector_inspector-0.3.6}/src/vector_inspector/core/connections/chroma_connection.py

@@ -8,7 +8,7 @@ from chromadb.api import ClientAPI
 from chromadb.api.models.Collection import Collection
 from chromadb import Documents, EmbeddingFunction, Embeddings
 
-from .base_connection import VectorDBConnection
+from vector_inspector.core.connections.base_connection import VectorDBConnection
 from vector_inspector.core.logging import log_info, log_error
 
 
@@ -28,7 +28,7 @@ class DimensionAwareEmbeddingFunction(EmbeddingFunction):
         if self._initialized:
             return
 
-        from ..embedding_utils import get_embedding_model_for_dimension
+        from vector_inspector.core.embedding_utils import get_embedding_model_for_dimension
 
         log_info("[ChromaDB] Loading embedding model for %dd vectors...", self.expected_dimension)
         self.model, self.model_name, self.model_type = get_embedding_model_for_dimension(
@@ -45,7 +45,7 @@ class DimensionAwareEmbeddingFunction(EmbeddingFunction):
     def __call__(self, input: Documents) -> Embeddings:
         """Embed documents using the dimension-appropriate model."""
         self._ensure_model_loaded()
-        from ..embedding_utils import encode_text
+        from vector_inspector.core.embedding_utils import encode_text
 
         embeddings = []
         for text in input:
@@ -385,6 +385,16 @@ class ChromaDBConnection(VectorDBConnection):
             return False
 
         try:
+            # If embeddings not provided, compute using collection model
+            if not embeddings and documents:
+                try:
+                    embeddings = self.compute_embeddings_for_documents(
+                        collection_name, documents, getattr(self, "connection_id", None)
+                    )
+                except Exception as e:
+                    log_error("Failed to compute embeddings for Chroma add_items: %s", e)
+                    return False
+
             collection.add(
                 documents=documents,
                 metadatas=metadatas,  # type: ignore
@@ -422,6 +432,16 @@ class ChromaDBConnection(VectorDBConnection):
             return False
 
         try:
+            # If embeddings not provided but documents changed, compute embeddings
+            if (not embeddings) and documents:
+                try:
+                    embeddings = self.compute_embeddings_for_documents(
+                        collection_name, documents, getattr(self, "connection_id", None)
+                    )
+                except Exception as e:
+                    log_error("Failed to compute embeddings for Chroma update_items: %s", e)
+                    return False
+
             collection.update(
                 ids=ids,
                 documents=documents,