app/routes/keyword.py

"""Keyword routes for API endpoints."""
import logging
from fastapi import APIRouter
from fastapi.responses import JSONResponse

from collectors.finfast.keyword import collect as run_historical_backfill # pylint: disable=import-error
from models.database import keywords_collection, summary_collection # pylint: disable=import-error

# Configure logger
logger = logging.getLogger(__name__)

# Create FastAPI Router
router = APIRouter(prefix="/keyword", tags=["keyword"])

@router.post('/historical-backfill')
async def historical_backfill():
    """
    Manual trigger endpoint for 60-day historical keyword backfill.
    
    Returns:
        JSON response with status and details
    """
    try:
        logger.info("Received request for historical keyword backfill")

        # Run the historical backfill
        run_historical_backfill()

        logger.info("Historical keyword backfill completed successfully")
        return JSONResponse(content={
            "status": "success",
            "message": "Historical keyword backfill completed successfully",
            "details": {
                "operation": "60-day historical keyword backfill",
                "method": "standard approach (day-by-day processing)"
            }
        }, status_code=200)

    except Exception as e: # pylint: disable=broad-exception-caught
        logger.error("Error during historical keyword backfill: %s", e)
        return JSONResponse(content={
            "status": "error",
            "message": "Failed to complete historical keyword backfill",
            "error": str(e)
        }, status_code=500)

@router.get('/status')
async def keyword_status():
    """
    Get status of keyword processing.
    
    Returns:
        JSON response with keyword processing status
    """
    try:

        # Get basic statistics
        total_documents = keywords_collection.count_documents({})

        # Get date range
        oldest_doc = keywords_collection.find_one(sort=[("_id", 1)])
        newest_doc = keywords_collection.find_one(sort=[("_id", -1)])

        status_info = {
            "status": "operational",
            "total_keyword_documents": total_documents,
            "date_range": {
                "oldest": oldest_doc["_id"] if oldest_doc else None,
                "newest": newest_doc["_id"] if newest_doc else None
            }
        }

        return JSONResponse(content=status_info, status_code=200)

    except Exception as e: # pylint: disable=broad-exception-caught
        logger.error("Error getting keyword status: %s", e)
        return JSONResponse(content={
            "status": "error",
            "message": "Failed to get keyword status",
            "error": str(e)
        }, status_code=500)

@router.get('/summary')
async def get_all_summaries():
    """
    Get all summaries (today, week, month).
    
    Returns:
        JSON response with all summary data
    """
    try:
        logger.info("Received request for all summaries")

        summaries = {}
        periods = ["today", "week", "month"]

        for period in periods:
            summary_doc = summary_collection.find_one({"_id": period})
            if summary_doc:
                summaries[period] = {
                    "keywords": summary_doc.get("keywords", []),
                    "categories": summary_doc.get("categories", [])
                }
            else:
                summaries[period] = {
                    "keywords": [],
                    "categories": [],
                    "message": f"No summary available for {period}"
                }

        return JSONResponse(content={
            "status": "success",
            "summaries": summaries
        }, status_code=200)

    except Exception as e: # pylint: disable=broad-exception-caught
        logger.error("Error getting all summaries: %s", e)
        return JSONResponse(content={
            "status": "error",
            "message": "Failed to get summaries",
            "error": str(e)
        }, status_code=500)

@router.get('/summary/{period}')
async def get_summary_by_period(period: str):
    """
    Get specific summary by period (today/week/month).
    
    Args:
        period: Analysis period (today, week, or month)
    
    Returns:
        JSON response with specific summary data
    """
    try:
        # Validate period parameter
        valid_periods = ["today", "week", "month"]
        if period not in valid_periods:
            return JSONResponse(content={
                "status": "error",
                "message": "Invalid period. Must be one of: today, week, month",
                "valid_periods": valid_periods
            }, status_code=400)

        logger.info("Received request for %s summary", period)

        summary_doc = summary_collection.find_one({"_id": period})

        if summary_doc:
            return JSONResponse(content={
                "status": "success",
                "period": period,
                "summary": {
                    "keywords": summary_doc.get("keywords", []),
                    "categories": summary_doc.get("categories", [])
                }
            }, status_code=200)
        return JSONResponse(content={
            "status": "not_found",
            "message": f"No summary available for {period}",
            "period": period
        }, status_code=404)

    except Exception as e: # pylint: disable=broad-exception-caught
        logger.error("Error getting %s summary: %s", period, e)
        return JSONResponse(content={
            "status": "error",
            "message": f"Failed to get {period} summary",
            "error": str(e)
        }, status_code=500)