# utils.py import os import uuid import json import cv2 import numpy as np import google.generativeai as genai from PIL import Image import fitz # Configure the Gemini API key try: genai.configure(api_key=os.environ.get("AIzaSyCJimpvYx_rUZo043u3iFwtKig9uqsbEHY")) except AttributeError as e: print(f"Error: GOOGLE_API_KEY not found. Please set it in your .env file. {e}") exit() # --- File and Image Processing --- def combine_images_to_pdf(image1_path, image2_path, output_folder): """Combines two images into a single PDF.""" try: img1 = Image.open(image1_path).convert("RGB") img2 = Image.open(image2_path).convert("RGB") pdf_filename = f"{uuid.uuid4()}.pdf" pdf_path = os.path.join(output_folder, pdf_filename) img1.save(pdf_path, save_all=True, append_images=[img2]) return pdf_path, pdf_filename except Exception as e: print(f"Error combining images to PDF: {e}") return None, None def convert_pdf_first_page_to_image(pdf_path, output_folder): """Converts the first page of a PDF to a JPG image and returns the path.""" try: doc = fitz.open(pdf_path) if doc.page_count == 0: doc.close(); return None page = doc.load_page(0) pix = page.get_pixmap() image_filename = f"pdf_page_{uuid.uuid4()}.jpg" image_filepath = os.path.join(output_folder, image_filename) pix.save(image_filepath) doc.close() return image_filepath except Exception as e: print(f"Error converting PDF to image: {e}") return None # --- STABLE OPENCV DNN FACE DETECTION --- def detect_and_extract_face(image_path, output_folder): """Detects a face using a stable OpenCV DNN model.""" try: # These model files must be in your root project folder prototxt_path = 'deploy.prototxt' model_path = 'res10_300x300_ssd_iter_140000.caffemodel' net = cv2.dnn.readNetFromCaffe(prototxt_path, model_path) img = cv2.imread(image_path) if img is None: print(f"!!! FACE DETECTION ERROR: Could not read image file at {image_path}") return None (h, w) = img.shape[:2] blob = cv2.dnn.blobFromImage(cv2.resize(img, (300, 300)), 1.0, (300, 300), (104.0, 177.0, 123.0)) net.setInput(blob) detections = net.forward() best_detection_index = np.argmax(detections[0, 0, :, 2]) confidence = detections[0, 0, best_detection_index, 2] if confidence > 0.3: # Using a lenient threshold for scanned documents print(f">>> FACE DETECTION: Found a face with confidence: {confidence:.2f}") box = detections[0, 0, best_detection_index, 3:7] * np.array([w, h, w, h]) (startX, startY, endX, endY) = box.astype("int") padding = 20 startX = max(0, startX - padding); startY = max(0, startY - padding) endX = min(w, endX + padding); endY = min(h, endY + padding) face_cropped = img[startY:endY, startX:endX] if face_cropped.size == 0: return None face_filename = f"face_{uuid.uuid4()}.jpg" face_filepath = os.path.join(output_folder, face_filename) cv2.imwrite(face_filepath, face_cropped) print(f">>> FACE DETECTION SUCCESS: Face extracted to {face_filepath}") return face_filepath else: print(f">>> FACE DETECTION: No faces found with sufficient confidence (Max Confidence: {confidence:.2f}).") return None except Exception as e: print(f"!!! FACE DETECTION CRITICAL ERROR: An exception occurred: {e}") return None # --- OCR with Google Gemini (UPDATED WITH STRICT PROMPT) --- def call_gemini_ocr(file_path): """Performs OCR on an image or PDF file using Gemini.""" try: model = genai.GenerativeModel('gemini-1.5-flash') uploaded_file = genai.upload_file(path=file_path) # --- NEW, STRICTER PROMPT TO FORCE JSON-ONLY OUTPUT --- prompt = """ Analyze the document. Extract all textual information into a structured JSON object. **CRITICAL INSTRUCTIONS:** 1. **JSON ONLY:** The entire output must be a single, valid JSON object. 2. **NO MARKDOWN:** Do not wrap the JSON in ```json ... ``` or any other markdown formatting. 3. **NO EXPLANATIONS:** Do not include any introductory text, summary, or any other explanatory sentences before or after the JSON object. Your entire response must start with `{` and end with `}`. 4. Use descriptive keys like "fullName", "dateOfBirth", "idNumber", "address". 5. If a field is not present, the value must be null. """ response = model.generate_content( [prompt, uploaded_file], safety_settings={'HARM_CATEGORY_HARASSMENT': 'BLOCK_NONE', 'HARM_CATEGORY_HATE_SPEECH': 'BLOCK_NONE', 'HARM_CATEGORY_SEXUALLY_EXPLICIT': 'BLOCK_NONE', 'HARM_CATEGORY_DANGEROUS_CONTENT': 'BLOCK_NONE'} ) print(f"--- RAW GEMINI RESPONSE ---\n{response.text}\n-------------------------") if not response.text.strip(): print("!!! GEMINI ERROR: Received an empty response. Likely due to safety filters.") if response.prompt_feedback: print(f"Prompt Feedback from API: {response.prompt_feedback}") return {"error": "API returned an empty response, likely blocked by safety filters."} # Clean the response just in case the model adds markdown despite instructions cleaned_text = response.text.strip() if cleaned_text.startswith("```json"): cleaned_text = cleaned_text[7:] if cleaned_text.endswith("```"): cleaned_text = cleaned_text[:-3] data = json.loads(cleaned_text) return data except json.JSONDecodeError as e: print(f"!!! JSON DECODE ERROR: Failed to parse Gemini response as JSON. Error: {e}") return {"error": f"Failed to parse non-JSON response from API.", "raw_response": response.text} except Exception as e: print(f"!!! An unexpected error occurred calling Gemini API: {e}") return {"error": str(e)}