"""
Example: Using Vector Memory
This example demonstrates how to use the vector memory system
for semantic search over past experiences.
"""
from pathlib import Path
from boring.vector_memory import create_vector_memory
def main():
"""Demonstrate vector memory usage."""
# Create vector memory (requires chromadb)
memory = create_vector_memory(project_root=Path.cwd(), log_dir=Path("logs"))
if not memory.enabled:
print("Vector memory not available. Install with:")
print(" pip install boring-gemini[vector]")
return
# Add some experiences
memory.add_experience(
error_type="ImportError",
error_message="No module named 'requests'",
solution="pip install requests",
context="web_client.py",
)
memory.add_experience(
error_type="TypeError",
error_message="'NoneType' object is not subscriptable",
solution="Check if result is None before accessing: if result: value = result['key']",
context="data_parser.py",
)
# Query for similar errors
print("\n--- Searching for similar errors ---\n")
results = memory.retrieve_similar("module not found error", n_results=2)
for result in results:
print(f"Error: {result['error_type']}")
print(f"Solution: {result['solution']}")
print(f"Similarity: {result['similarity']:.2%}")
print()
# Get context injection for prompts
context = memory.generate_context_injection("TypeError: NoneType")
print("--- Context Injection ---")
print(context)
if __name__ == "__main__":
main()
