}hddlZddlZddlZddlZddlZddlmZddl m Z ddl Z ejejjddZdZdZd Zy#e$rZedeeYdZ[)dZ[wwxYw) N)ImageGOOGLE_API_KEY)api_keyzBError: GOOGLE_API_KEY not found. Please set it in your .env file. c| tj|jd}tj|jd}tjd}t j j||}|j|d|g||fS#t$r}td|Yd}~yd}~wwxYw)z&Combines two images into a single PDF.RGBz.pdfT)save_all append_imageszError combining images to PDF: N)NN) ropenconvertuuiduuid4ospathjoinsave Exceptionprint) image1_path image2_path output_folderimg1img2 pdf_filenamepdf_pathes /root/ocr/utils.pycombine_images_to_pdfrs zz+&..u5zz+&..u5**,t, 77<< |< (T$ @%%  /s34sBB B;#B66B;c tj|}|jdk(r|jy|j d}|j }dt jd}tjj||}|j||j|S#t$r}td|Yd}~yd}~wwxYw)zEConverts the first page of a PDF to a JPG image and returns the path.rN pdf_page_.jpgzError converting PDF to image: )fitzr page_countclose load_page get_pixmapr r rrrrrr)rrdocpagepiximage_filenameimage_filepathrs rconvert_pdf_first_page_to_imager+!s ii! >>Q  IIK}}Qoo$TZZ\N$7m^D     /s34s4B3A;B33 C<CCc0 d}d}tjj||}tj|}|t d|y|j dd\}}tjj tj|dddd}|j||j} tj| d d dddf} | d d | df} | d kDrt d | d | d d | d dftj||||gz} | jd\} }}}d}td | |z } td ||z }t|||z}t|||z}|||| |f}|j d k(rydt#j$d}t&j(j+||}tj,||t d||St d| d dy#t.$r}t d|Yd}~yd}~wwxYw)z/Detects a face using a stable OpenCV DNN model.zdeploy.prototxtz(res10_300x300_ssd_iter_140000.caffemodelNz7!!! FACE DETECTION ERROR: Could not read image file at ),r.g?)gZ@g f@g^@rg333333?z2>>> FACE DETECTION: Found a face with confidence: z.2fintface_r z.>>> FACE DETECTION SUCCESS: Face extracted to zO>>> FACE DETECTION: No faces found with sufficient confidence (Max Confidence: z).z:!!! FACE DETECTION CRITICAL ERROR: An exception occurred: )cv2dnnreadNetFromCaffeimreadrshape blobFromImageresizesetInputforwardnpargmaxarrayastypemaxminsizer r rrrimwriter) image_pathr prototxt_path model_pathnetimghwblob detectionsbest_detection_index confidenceboxstartXstartYendXendYpadding face_cropped face_filename face_filepathrs rdetect_and_extract_facerY3s,,) ? gg&&}jAjj$ ; KJ<X Y2AAww$$SZZZ%@#%/1FH T[[] !yyAq!QJ)?@1&:A =>   FzRUFVW XQ#71<=!QPQST@VVC+.::e+< (VVT4GFW,-FAv?O8Pvq$.)D#a2H4vd{F4K78L  A%d#DJJL>6MGGLL FM KK | 4 B=/R S cdnorcssuv w  J1#NOs,A G4 D8G4AG4#G44 H=HHc tjd}tj|}d}|j||gddddd}t d|j d|j j s3t d |jrt d |jd d iS|j j }|jd r|dd}|jdr|dd}tj|}|S#tj$r'}t d|dj dcYd}~Sd}~wt$r%}t d|d t|icYd}~Sd}~wwxYw)z2Performs OCR on an image or PDF file using Gemini.zgemini-1.5-flash)ra 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. BLOCK_NONE)HARM_CATEGORY_HARASSMENTHARM_CATEGORY_HATE_SPEECHHARM_CATEGORY_SEXUALLY_EXPLICITHARM_CATEGORY_DANGEROUS_CONTENT)safety_settingsz--- RAW GEMINI RESPONSE --- z -------------------------zK!!! GEMINI ERROR: Received an empty response. Likely due to safety filters.zPrompt Feedback from API: errorzAAPI returned an empty response, likely blocked by safety filters.z```jsonr0Nz```zG!!! JSON DECODE ERROR: Failed to parse Gemini response as JSON. Error: z+Failed to parse non-JSON response from API.)ra raw_responsez5!!! An unexpected error occurred calling Gemini API: )genaiGenerativeModel upload_filegenerate_contentrtextstripprompt_feedback startswithendswithjsonloadsJSONDecodeErrorrstr) file_pathmodel uploaded_filepromptresponse cleaned_textdatars rcall_gemini_ocrrxds|/!%%&89))y9   )) ] #9E:F@L@LN*  -hmm_rs} #  EOOBJJNN+;<= $.b1!m  Nqc RSFF s,AB$A;;B