mongo_tools.py

# mongo_tools.py
try:
    from .mongochain import MongoDBHandler
except ImportError:
    from mongochain import MongoDBHandler

from typing import Dict, List, Any

# Global handler instance
_mongo_handler = MongoDBHandler()

def mongo_configure(uri: str, db: str, coll: str):
    """Configures the global MongoDB handler."""
    global _mongo_handler
    _mongo_handler = MongoDBHandler(uri=uri, db_name=db, collection_name=coll)
    return _mongo_handler.connect()

def mongo_insert_doc(content: str, metadata: Dict[str, Any] = None) -> str:
    """
    Inserts a document into MongoDB.
    
    Args:
        content: The text content to store.
        metadata: Optional dictionary of metadata (source, author, etc.).
    """
    success = _mongo_handler.insert_chunk(content, metadata or {})
    return "Successfully inserted doc." if success else "Failed to insert doc."

def mongo_find_docs(query_json: Dict[str, Any], limit: int = 5) -> List[Dict[str, Any]]:
    """
    Finds documents in MongoDB matching a JSON query.
    
    Args:
        query_json: A MongoDB query dictionary (e.g., {"metadata.source": "tech_specs.pdf"}).
        limit: Max number of documents to return.
    """
    results = _mongo_handler.find_relevant(query_json, limit=limit)
    # Convert ObjectId to string for JSON compatibility
    for res in results:
        if "_id" in res:
            res["_id"] = str(res["_id"])
    return results

def mongo_get_collection_stats() -> Dict[str, Any]:
    """Returns statistics about the currently connected MongoDB collection."""
    return _mongo_handler.get_stats()

def mongo_clear_collection() -> str:
    """Deletes all documents in the current collection."""
    success = _mongo_handler.clear()
    return "Collection cleared." if success else "Failed to clear collection."

def mongo_keyword_search(keyword: str, limit: int = 5) -> List[Dict[str, Any]]:
    """
    Performs a simple regex keyword search in the document content.
    
    Args:
        keyword: The string to search for.
        limit: Max results.
    """
    query = {"content": {"$regex": keyword, "$options": "i"}}
    return mongo_find_docs(query, limit)