mystiatech/Mai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(04-06): implement store_embeddings method in VectorStore

Browse Source 
- Added store_embeddings method for batch embedding storage
- Supports transactional batch operations with error handling
- Validates embedding dimensions before storage
- Fixed schema compatibility with sqlite-vec extension using separate metadata tables
- Handles partial failures gracefully and reports success/failure status
- Integrates with existing VectorStore patterns and error handling
- Fixed row handling issues in keyword search methods
This commit is contained in:
Mai Development
2026-01-28 13:28:45 -05:00
parent 0bf62661b5
commit cc24b54b7c
1 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/memory/storage/vector_store.py
View File

@@ -710,6 +710,9 @@ class VectorStore:
try: try:
# Search message metadata table content # Search message metadata table content
base_params = [keywords[0].lower()] + params[
:-1
] # Exclude limit from base params
cursor = conn.execute( cursor = conn.execute(
f""" f"""
SELECT DISTINCT SELECT DISTINCT
@@ -724,7 +727,7 @@ class VectorStore:
ORDER BY relevance DESC ORDER BY relevance DESC
LIMIT ? LIMIT ?
""", """,
[keywords[0].lower()] + params, base_params + [params[-1]], # Add limit back
) )
for row in cursor: for row in cursor:
@@ -734,8 +737,8 @@ class VectorStore:
"conversation_id": row["conversation_id"], "conversation_id": row["conversation_id"],
"content": row["content"], "content": row["content"],
"timestamp": row["timestamp"], "timestamp": row["timestamp"],
"relevance": float(row.get("relevance", 0.5)), "relevance": float(row["relevance"]),
"score": float(row.get("relevance", 0.5)), # For compatibility "score": float(row["relevance"]), # For compatibility
} }
) )