app/core/conversation_retrieval.py

import json
import numpy as np
from app.core.embedding import generate_embedding
from app.core.db_setup import conversations_collection
from app.core.logging_setup import logger

# Removed DB_PATH as it is no longer needed

# Update find_similar_conversations to use MongoDB
def find_similar_conversations(query_message):
    """Find similar conversations based on embedding similarity."""
    query_embedding = generate_embedding(query_message)
    if query_embedding is None:
        return []

    try:
        conversations = list(conversations_collection.find({}, {"message": 1, "embedding": 1, "_id": 0}))
    except Exception as e:
        logger.error(f"Unexpected error: {e}")
        return []

    similarities = []
    for conversation in conversations:
        message = conversation["message"]
        embedding = json.loads(conversation["embedding"])
        similarity = np.dot(query_embedding, embedding) / (np.linalg.norm(query_embedding) * np.linalg.norm(embedding))
        similarities.append((message, similarity))

    similarities.sort(key=lambda x: x[1], reverse=True)
    return [msg for msg, _ in similarities[:5]]

# Update get_similar_conversations to use MongoDB
def get_similar_conversations(user_input, top_n=3):
    """Retrieve past relevant messages using semantic similarity."""
    user_embedding = generate_embedding(user_input)
    if user_embedding is None:
        return []

    try:
        conversations = list(conversations_collection.find({}, {"message": 1, "embedding": 1, "_id": 0}))
    except Exception as e:
        logger.error(f"Unexpected error: {e}")
        return []

    similarities = []

    for conversation in conversations:
        message = conversation["message"]
        embedding = json.loads(conversation["embedding"])

        # Ensure user_embedding is a numpy array
        user_embedding = np.array(user_embedding)

        # Ensure embedding is a numpy array
        embedding = np.array(embedding)

        # Validate embedding dimensions
        if user_embedding.shape[0] != len(embedding):
            logger.warning(f"Skipping embedding due to dimension mismatch: user_embedding {user_embedding.shape[0]} vs embedding {len(embedding)}")
            continue

        # Adjust embedding dimensions to match user_embedding
        if len(embedding) < user_embedding.shape[0]:
            embedding = np.pad(embedding, (0, user_embedding.shape[0] - len(embedding)), mode='constant')
        elif len(embedding) > user_embedding.shape[0]:
            embedding = embedding[:user_embedding.shape[0]]

        similarity = np.dot(user_embedding, embedding) / (np.linalg.norm(user_embedding) * np.linalg.norm(embedding))
        similarities.append((message, similarity))

    similarities.sort(key=lambda x: x[1], reverse=True)
    return [msg for msg, _ in similarities[:top_n]]