Chris Coutinho
b8010270c1
This commit addresses multiple issues with async operations, PDF metadata extraction, and type safety in document processing and search. ## Async/Await Fixes - processor.py:259 - Added await for chunker.chunk_text(content) - processor.py:270 - Added await for bm25_service.encode_batch(chunk_texts) - tests/unit/test_document_chunker.py - Converted all 12 test methods to async ## PDF Metadata Enhancement - pymupdf.py:143 - Added file_size metadata extraction - pymupdf.py:145-206 - Refactored to extract text page-by-page - Manually loop through pages instead of using page_chunks=True - Generate page_boundaries metadata for precise page tracking - Works around pymupdf.layout.activate() breaking page_chunks=True - processor.py:32-66 - Added assign_page_numbers() helper function - Assigns page numbers to chunks based on overlap with page boundaries - Handles chunks spanning multiple pages - processor.py:298-300 - Call assign_page_numbers() for PDF files ## Type Safety Fixes - bm25_hybrid.py:184 - Removed int() conversion of doc_id - semantic.py:131 - Removed int() conversion of doc_id - viz_routes.py:275 - Removed int() conversion of doc_id - Added comments documenting that doc_id can be int (notes) or str (file paths) ## Testing - All 18 tests passing (12 unit + 6 integration) - No type errors in modified files - Container logs show successful processing - Vector viz searches working correctly 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
80 lines
2.5 KiB
Python
80 lines
2.5 KiB
Python
"""BM25 sparse embedding provider using FastEmbed."""
|
|
|
|
import logging
|
|
from typing import Any
|
|
|
|
from fastembed import SparseTextEmbedding
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
class BM25SparseEmbeddingProvider:
|
|
"""
|
|
BM25 sparse embedding provider for hybrid search.
|
|
|
|
Uses FastEmbed's BM25 model to generate sparse vectors for keyword-based
|
|
retrieval. These sparse vectors are combined with dense semantic vectors
|
|
in Qdrant using Reciprocal Rank Fusion (RRF) for hybrid search.
|
|
|
|
Unlike dense embeddings which have fixed dimensions, sparse embeddings
|
|
have variable-length vectors with (index, value) pairs representing
|
|
term frequencies in the BM25 vocabulary.
|
|
"""
|
|
|
|
def __init__(self, model_name: str = "Qdrant/bm25"):
|
|
"""
|
|
Initialize BM25 sparse embedding provider.
|
|
|
|
Args:
|
|
model_name: FastEmbed BM25 model name (default: Qdrant/bm25)
|
|
"""
|
|
self.model_name = model_name
|
|
logger.info(f"Initializing BM25 sparse embedding provider: {model_name}")
|
|
|
|
# Initialize FastEmbed sparse embedding model
|
|
self.model = SparseTextEmbedding(model_name=model_name)
|
|
logger.info(f"BM25 sparse embedding model loaded: {model_name}")
|
|
|
|
def encode(self, text: str) -> dict[str, Any]:
|
|
"""
|
|
Generate BM25 sparse embedding for a single text.
|
|
|
|
Args:
|
|
text: Input text to encode
|
|
|
|
Returns:
|
|
Dictionary with 'indices' and 'values' keys for Qdrant sparse vector
|
|
"""
|
|
# FastEmbed returns a generator, take first result
|
|
sparse_embedding = next(iter(self.model.embed([text])))
|
|
|
|
return {
|
|
"indices": sparse_embedding.indices.tolist(),
|
|
"values": sparse_embedding.values.tolist(),
|
|
}
|
|
|
|
async def encode_batch(self, texts: list[str]) -> list[dict[str, Any]]:
|
|
"""
|
|
Generate BM25 sparse embeddings for multiple texts (batched).
|
|
|
|
Args:
|
|
texts: List of texts to encode
|
|
|
|
Returns:
|
|
List of dictionaries with 'indices' and 'values' for each text
|
|
"""
|
|
import anyio
|
|
|
|
# Run CPU-bound BM25 encoding in thread pool to avoid blocking event loop
|
|
sparse_embeddings = await anyio.to_thread.run_sync(
|
|
lambda: list(self.model.embed(texts))
|
|
)
|
|
|
|
return [
|
|
{
|
|
"indices": emb.indices.tolist(),
|
|
"values": emb.values.tolist(),
|
|
}
|
|
for emb in sparse_embeddings
|
|
]
|