apunkt/dograh
1
0
Fork 0
mirror of https://github.com/dograh-hq/dograh.git synced 2026-06-13 08:15:21 +02:00
Code Issues Projects Releases Packages Wiki Activity

feat: add openai embedding service

Browse source
This commit is contained in:
Abhishek Kumar 2026-01-17 13:36:26 +05:30
parent eb41285204
commit 3f0e500fde
39 changed files with 1902 additions and 339 deletions
Download patch file Download diff file Expand all files Collapse all files

22
api/db/knowledge_base_client.py
View file

@ -332,6 +332,7 @@ class KnowledgeBaseClient(BaseDBClient):
limit: int = 5,
document_ids: Optional[List[int]] = None,
document_uuids: Optional[List[str]] = None,
embedding_model: Optional[str] = None,
) -> List[dict]:
"""Search for similar chunks using vector similarity.
@ -344,6 +345,7 @@ class KnowledgeBaseClient(BaseDBClient):
limit: Maximum number of results to return
document_ids: Optional list of document IDs to filter by
document_uuids: Optional list of document UUIDs to filter by
embedding_model: Optional embedding model to filter by (for dimension compatibility)
Returns:
List of dictionaries with chunk data and similarity scores, ordered by similarity (highest first)
@ -359,23 +361,37 @@ class KnowledgeBaseClient(BaseDBClient):
"c.organization_id = $2",
"d.is_active = true",
]
params = [None, organization_id, limit] # $1 will be embedding_str, $3 is limit
params = [
None,
organization_id,
limit,
] # $1 will be embedding_str, $3 is limit
param_index = 4 # Next available parameter index
# Add document_ids filter if provided
if document_ids:
placeholders = ", ".join(f"${param_index + i}" for i in range(len(document_ids)))
placeholders = ", ".join(
f"${param_index + i}" for i in range(len(document_ids))
)
where_conditions.append(f"c.document_id IN ({placeholders})")
params.extend(document_ids)
param_index += len(document_ids)
# Add document_uuids filter if provided
if document_uuids:
placeholders = ", ".join(f"${param_index + i}" for i in range(len(document_uuids)))
placeholders = ", ".join(
f"${param_index + i}" for i in range(len(document_uuids))
)
where_conditions.append(f"d.document_uuid IN ({placeholders})")
params.extend(document_uuids)
param_index += len(document_uuids)
# Add embedding_model filter if provided (for dimension compatibility)
if embedding_model:
where_conditions.append(f"c.embedding_model = ${param_index}")
params.append(embedding_model)
param_index += 1
# Build the complete SQL query
where_clause = " AND ".join(where_conditions)
query_sql = f"""