aljaz/enclava
1
0
Fork 0
mirror of https://github.com/aljazceru/enclava.git synced 2025-12-17 07:24:34 +01:00
Code Issues Packages Projects Releases Wiki Activity

ratelimiting and rag

Browse Source
This commit is contained in:
Aljaz Ceru
2025-09-21 06:49:55 +02:00
parent 0c20de4ca1
commit f58a76ac59
7 changed files with 410 additions and 130 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
backend/modules/rag/main.py
View File

@@ -60,6 +60,7 @@ import tiktoken
from app.core.config import settings
from app.core.logging import log_module_event
from app.services.base_module import BaseModule, Permission
from app.services.enhanced_embedding_service import enhanced_embedding_service
@dataclass
@@ -1125,9 +1126,17 @@ class RAGModule(BaseModule):
# Chunk the document
chunks = self._chunk_text(content)
# Generate embeddings for all chunks in batch (more efficient)
embeddings = await self._generate_embeddings(chunks)
# Generate embeddings with enhanced rate limiting handling
embeddings, success = await enhanced_embedding_service.get_embeddings_with_retry(chunks)
# Log if fallback embeddings were used
if not success:
logger.warning(f"Used fallback embeddings for document {doc_id} - search quality may be degraded")
log_module_event("rag", "fallback_embeddings_used", {
"document_id": doc_id,
"content_preview": content[:100] + "..." if len(content) > 100 else content
})
# Create document points
points = []
for i, (chunk, embedding) in enumerate(zip(chunks, embeddings)):
@@ -1188,9 +1197,17 @@ class RAGModule(BaseModule):
# Chunk the document
chunks = self._chunk_text(processed_doc.content)
# Generate embeddings for all chunks in batch (more efficient)
embeddings = await self._generate_embeddings(chunks)
# Generate embeddings with enhanced rate limiting handling
embeddings, success = await enhanced_embedding_service.get_embeddings_with_retry(chunks)
# Log if fallback embeddings were used
if not success:
logger.warning(f"Used fallback embeddings for document {processed_doc.id} - search quality may be degraded")
log_module_event("rag", "fallback_embeddings_used", {
"document_id": processed_doc.id,
"filename": processed_doc.original_filename
})
# Create document points with enhanced metadata
points = []
for i, (chunk, embedding) in enumerate(zip(chunks, embeddings)):