doc_processer/app/api/v1/endpoints/image.py

"""Image OCR endpoint."""

import time
import uuid

from fastapi import APIRouter, Depends, HTTPException, Request, Response

from app.core.dependencies import (
    get_image_processor,
    get_layout_detector,
    get_ocr_service,
    get_mineru_ocr_service,
    get_glmocr_service,
)
from app.core.logging_config import get_logger, RequestIDAdapter
from app.schemas.image import ImageOCRRequest, ImageOCRResponse
from app.services.image_processor import ImageProcessor
from app.services.layout_detector import LayoutDetector
from app.services.ocr_service import OCRService, MineruOCRService, GLMOCRService

router = APIRouter()
logger = get_logger()


@router.post("/ocr", response_model=ImageOCRResponse)
async def process_image_ocr(
    request: ImageOCRRequest,
    http_request: Request,
    response: Response,
    image_processor: ImageProcessor = Depends(get_image_processor),
    layout_detector: LayoutDetector = Depends(get_layout_detector),
    mineru_service: MineruOCRService = Depends(get_mineru_ocr_service),
    paddle_service: OCRService = Depends(get_ocr_service),
    glmocr_service: GLMOCRService = Depends(get_glmocr_service),
) -> ImageOCRResponse:
    """Process an image and extract content as LaTeX, Markdown, and MathML.

    The processing pipeline:
    1. Load and preprocess image (add 30% whitespace padding)
    2. Detect layout using DocLayout-YOLO
    3. Based on layout:
       - If plain text exists: use PP-DocLayoutV2 for mixed recognition
       - Otherwise: use PaddleOCR-VL with formula prompt
    4. Convert output to LaTeX, Markdown, and MathML formats

    Note: OMML conversion is not included due to performance overhead.
    Use the /convert/latex-to-omml endpoint to convert LaTeX to OMML separately.
    """
    # Get or generate request ID
    request_id = http_request.headers.get("x-request-id", str(uuid.uuid4()))
    response.headers["x-request-id"] = request_id

    # Create logger adapter with request_id
    log = RequestIDAdapter(logger, {"request_id": request_id})
    log.request_id = request_id

    try:
        log.info("Starting image OCR processing")

        # Preprocess image
        preprocess_start = time.time()
        image = image_processor.preprocess(
            image_url=request.image_url,
            image_base64=request.image_base64,
        )
        preprocess_time = time.time() - preprocess_start
        log.debug(f"Image preprocessing completed in {preprocess_time:.3f}s")

        # Layout detection
        layout_start = time.time()
        layout_info = layout_detector.detect(image)
        layout_time = time.time() - layout_start
        log.info(f"Layout detection completed in {layout_time:.3f}s")

        # OCR recognition
        ocr_start = time.time()
        if layout_info.MixedRecognition:
            recognition_method = "MixedRecognition (MinerU)"
            log.info(f"Using {recognition_method}")
            ocr_result = mineru_service.recognize(image)
        else:
            recognition_method = "FormulaOnly (GLMOCR)"
            log.info(f"Using {recognition_method}")
            ocr_result = glmocr_service.recognize(image)
        ocr_time = time.time() - ocr_start

        total_time = time.time() - preprocess_start
        log.info(f"OCR processing completed - Method: {recognition_method}, " f"Layout time: {layout_time:.3f}s, OCR time: {ocr_time:.3f}s, " f"Total time: {total_time:.3f}s")

    except RuntimeError as e:
        log.error(f"OCR processing failed: {str(e)}", exc_info=True)
        raise HTTPException(status_code=503, detail=str(e))
    except Exception as e:
        log.error(f"Unexpected error during OCR processing: {str(e)}", exc_info=True)
        raise HTTPException(status_code=500, detail="Internal server error")

    return ImageOCRResponse(
        latex=ocr_result.get("latex", ""),
        markdown=ocr_result.get("markdown", ""),
        mathml=ocr_result.get("mathml", ""),
        mml=ocr_result.get("mml", ""),
    )