#!/usr/bin/env python3
"""Test that progressive context doesn't use dummy vectors for search."""
import sys
import os
import json
import numpy as np
from pathlib import Path
# Add parent directory to path
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from src.utils.progressive_context import ProgressiveContextManager
from src.utils.embeddings import get_embedding_model
from qdrant_client import QdrantClient
from qdrant_client.models import VectorParams, Distance
def test_progressive_no_dummy_vectors():
"""Test that progressive context preserves original vector search scores."""
# Initialize test client (in-memory)
client = QdrantClient(":memory:")
# Create test collection
collection_name = "test_collection"
embedding_dim = 384 # For all-MiniLM-L6-v2
client.create_collection(
collection_name=collection_name,
vectors_config=VectorParams(size=embedding_dim, distance=Distance.COSINE)
)
# Get embedding model
embed_model = get_embedding_model()
# Add test documents
test_docs = [
{
"id": 1,
"content": "def authenticate_user(username, password): return check_credentials(username, password)",
"file_path": "auth.py",
"chunk_index": 0,
"chunk_type": "function"
},
{
"id": 2,
"content": "class UserAuthentication: def login(self, user, pwd): self.authenticate(user, pwd)",
"file_path": "auth.py",
"chunk_index": 1,
"chunk_type": "class"
},
{
"id": 3,
"content": "# Authentication module for user login and security",
"file_path": "auth.py",
"chunk_index": 2,
"chunk_type": "comment"
}
]
# Index documents
points = []
for doc in test_docs:
embedding = embed_model.embed_documents([doc["content"]])[0]
points.append({
"id": doc["id"],
"vector": embedding,
"payload": {
"content": doc["content"],
"file_path": doc["file_path"],
"chunk_index": doc["chunk_index"],
"chunk_type": doc["chunk_type"]
}
})
client.upsert(collection_name=collection_name, points=points)
# Initialize progressive context manager
context_manager = ProgressiveContextManager(
qdrant_client=client,
collections=[collection_name],
cache_dir="./test_cache"
)
# Test query
query = "authentication user login"
query_embedding = embed_model.embed_documents([query])[0]
# Test hybrid search (the problematic mode)
print("\n=== Testing Hybrid Search ===")
results = context_manager._search_at_level(
query=query,
query_embedding=query_embedding,
n_results=3,
level="snippet",
search_collections=[collection_name],
search_mode="hybrid"
)
print(f"Found {len(results)} results")
# Check results
for i, result in enumerate(results):
print(f"\nResult {i+1}:")
print(f" Score: {result['score']:.4f}")
print(f" Content: {result['content'][:80]}...")
# Key check: vector_score should be properly set (not 0 or dummy)
if 'vector_score' in result:
print(f" Vector score: {result['vector_score']:.4f}")
assert result['vector_score'] > 0, "Vector score should be positive, not from dummy vector"
# BM25 score might be 0 if keyword didn't match
if 'bm25_score' in result:
print(f" BM25 score: {result['bm25_score']:.4f}")
# Also test keyword-only mode
print("\n\n=== Testing Keyword Search ===")
results_keyword = context_manager._search_at_level(
query=query,
query_embedding=query_embedding,
n_results=3,
level="snippet",
search_collections=[collection_name],
search_mode="keyword"
)
print(f"Found {len(results_keyword)} keyword results")
# Verify scores are reasonable (not corrupted by dummy vectors)
for result in results_keyword:
# For keyword search, the score comes from BM25
assert result['score'] >= 0, "Score should be non-negative"
print(f" Score: {result['score']:.4f} - {result['content'][:60]}...")
print("\n✅ Test passed! No dummy vectors detected.")
# Clean up
if os.path.exists("./test_cache"):
import shutil
shutil.rmtree("./test_cache")
if __name__ == "__main__":
test_progressive_no_dummy_vectors()
