Explicite/api-test
0
1�
2 O�ii � �^ � d dl Z d dlmZmZ d dlZd� Zdedefd�Zdej defd�Z3y) � N)�Dict�Listc �. � dd l } | S # t $ r Y y w xY w)Nr )� pytesseract� Exception)r s �Z/Users/alexiseduardojuarezjaimes/Desktop/api-face-extractor/app/services/doc_classifier.py�_safe_import_pytesseractr s" � ������ � ���s � � ��s�returnc � � | j � } t j dd| � } t j dd| � j � } | S )u1 Normaliza el texto OCR para búsqueda de patronesz [^A-Z0-9<\s]� z\s+)�upper�re�sub�strip)r4 s r �_normalize_textr s? � � ��� �A�5�����Q�'�A�6���v�s�A� �$�$�&�A� �H� �np_rgbc � �� g }t � }|�dddgd�S |j | dd�� }t |� �t
�� d7k rdd d gd�S d
|v xr" t
t j d|� � dk\ }|r|j d� g d�}t �fd�|D � � }|r|j d|� �� g d�}t �fd�|D � � } | r|j d| � �� g d�}8t �fd�|9D � � } | r|j d| � �� g d�} t �fd�| D � � }
|
r|j d |
� �� g d �}t �fd �|D � � }|r|j d |� �� |rd d!|d�S |d"k\ rt d#d$d%|z z � }d&||d�S |d'k( rd&d(|d�S | d"k\ r d)�vrt d*d+d,| z z � }d-||d�S | d"k\ rt d!d.d%| z z � }d/||d�S |
d'k\ rt d0d1d,|
z z � }d2||d�S |d'k\ rd3d4|d�S dd5|r|d�S d6gd�S # t $ r'}ddd t |� j � �gd�cY d}~S d}~ww xY w)7uI 10 Clasifica el tipo de documento mediante OCR y reglas heurísticas.11 12 NO bloquea el flujo - solo añade información al JSON.13 14 Tipos soportados:15 - ine_mx: INE (Credencial para votar) de México16 - passport: Pasaporte (detecta MRZ)17 - driver_license_mx: Licencia de conducir México18 - us_driver_license: Licencia de conducir USA19 - us_id_card: ID estatal USA20 - id_card_generic: Credencial genérica21 - unknown: No se pudo determinar22 23 Returns:24 Dict con: type (str), confidence (float 0-1), signals (List[str])25 N�unknowng�������?z3ocr:not_available (install pytesseract + tesseract))�type�26confidence�signalszeng+spaz--psm 6)�lang�configz27ocr:error:�28 g333333�?zocr:too_little_textz<<�<� z mrz:detected (<< and many <))z INSTITUTO NACIONAL ELECTORALzCREDENCIAL PARA VOTAR�ELECTOR�SECCION�VIGENCIAzCLAVE DE ELECTOR�CURPc 3 �, �K � | ] }|�v s�d �� �
y�w�� N� ��.0�k�texts �r � <genexpr>z$classify_document.<locals>.<genexpr>Y s �� �� �8�l��a�4�i�1�l�� � �z ine:hits=)�LICENCIA�CONDUCIR� CONDUCTOR�29EXPEDICIONr! �FOLIOc 3 �, �K � | ] }|�v s�d �� �
y�wr$ r& r' s �r r+ z$classify_document.<locals>.<genexpr>f � �� �� �<��1�!�t�)�Q��r, z mx_dl:hits=)30�DRIVER LICENSEzDRIVER LICENCE�DL�CLASS�ISS�EXP�DOB�HEIGHT�EYES�SEXc 3 �, �K � | ] }|�v s�d �� �
y�wr$ r& r' s �r r+ z$classify_document.<locals>.<genexpr>w r3 r, z us_dl:hits=)zIDENTIFICATION CARDz
IDENTITY CARDzSTATE IDzID CARDc 3 �, �K � | ] }|�v s�d �� �
y�wr$ r& r' s �r r+ z$classify_document.<locals>.<genexpr>� r3 r, z us_id:hits=)�IDENTIFICACION�IDENTIFICATION�ID�31CREDENCIALc 3 �, �K � | ] }|�v s�d �� �
y�wr$ r& r' s �r r+ z$classify_document.<locals>.<genexpr>� s �� �� �C�"5�Q��d��q�"5�r, z
generic:hits=�passportg�������?� gffffff�?g333333�?g� ��Q��?�ine_mxr% g�������?r4 gq=32ףp�?g �?g{�G�z�?�driver_license_mxg�(\����?�us_driver_licenseg)\���(�?g33ףp=34�?�35us_id_card�id_card_genericg�������?g�������?�no_keywords_matched) r �image_to_stringr r �__name__r �lenr �findall�append�sum�min)r r r �raw_text�e� mrz_detected� ine_keywords�ine_hits�mx_dl_keywords�36mx_dl_hits�us_dl_keywords�37us_dl_hits�us_id_keywords�38us_id_hits�generic_id_keywords� generic_hitsr r* s @r �classify_documentr` s� �� �$ �G�*�,�K��� � �M�N�39� 40�41� �.�.�v�I�i�.�X�� �8� $�D�42�4�y�2�~� � �-�.�43� 44� �H�$�N��R�Z�Z��h�-G�)H�B�)N�L�����5�6��L� �8�l�8�8�H������8�*�-�.��N� �<��<�<�J������Z�L�1�2� �N� �<��<�<�J������Z�L�1�2��N� �<��<�<�J������Z�L�1�2��� �C�"5�C�C�L����� �|�n�5�6�45 �"�$�7�K�K� �1�}���t�d�X�o� 5�6�46� �47�w�O�O��1�}� ���I�I� �Q��+�4�7���t�d�Z�&7� 7�8�48�+�:�RY�Z�Z� �Q����t�d�Z�&7� 7�8�49�+�:�RY�Z�Z� �Q����t�d�Z�&7� 7�8�50�$�J�7�S�S� �q��)��'�R�R� �T��g� n�n�Wl�Vm� n�n��w � 51� � �$�T�!�W�%5�%5�$6�7�8�52� 53��54�s �H � I � H;�5I �;I ) r �typingr r �numpy�npr �strr �ndarrayr` r& r r �<module>rf sC �� � � ��
�s�
�s�
�Zo�b�j�j� Zo�T� Zor