Team Ai
Apppublic

documentExtractionag051/ExtractDocument

sourceHugging Faceupdated 9mo agoView on Hugging Face
0likes
custom_processor.cpython-313.pyc44 linesDownload Raw Back to __pycache__
1�

2��_i����SSKrSSKJr \R"\RS9 \R3"\5rS\/0rSSjr	Sr4g)	�N�)�extract_invoice_tables)�level�Invoicec��[U5n[/n[RS[	U5S35 UH2n	[RSU	R535 U	"UUUUUUUS9nM4 U$![aDn6[RSU	R7SU835 SSKnUR5 Sn9A10M�Sn11A12ff=f)	a�13Main entry point for layout-based post-processing.14 15Args:16    input_data (dict): Original input JSON data17    validation_data (dict): Validation data structure18    ocr_elems (list): List of TextRect objects19    column_headers (dict): Optional custom column headers20    anchor_column (str): Optional anchor column name21    anchor_pattern (str): Optional anchor pattern regex22    advanced_options (dict): Optional advanced options23 24Returns:25    dict: Updated validation data26zRunning z handler(s)zRunning handler: )�	ocr_elems�validation_data�page_dimensions�column_headers�
anchor_column�anchor_pattern�advanced_optionszHandler z	 failed: rN)27�_extract_page_dimensionsr�logger�info�len�__name__�	Exception�error�	traceback�	print_exc)�28input_datar	rrrr
rr29�handler_list�handler�ers            �nC:\Users\poona\OneDrive\Documents\nextlifyai\Document_Automation\Document_Extraction\utils\custom_processor.py�do_custom_post_processingr,s���2/�z�:�O�+�+�L�30�K�K�(�3�|�,�-�[�9�:������'��(8�(8�'9�:�;�
	"�%�#� /� /�-�+�-�!1��O�	 �$����	"��L�L�8�G�$4�$4�#5�Y�q�c�B�C�����!�!��	"�s�A/�/31B=�99B8�8B=c32�n�SU;a�[URS5[5(akUSnU(a_SUS;aV[U5VVs0sH=up#URSUS-5URSS5URSS54_M? snn$SU;axURS05RS05RS/5n[U5VVs0sH-up#US-URSS5URSS54_M/ snn$[RS	5 0$s snnfs snnf)33zn34Extract page dimensions from input data.35 36Supports both simplified schema and original engine output format.37�pages�pageNumrr�width�height�38engineData�imagePreprocessingResultz!Could not extract page dimensions)�39isinstance�get�list�	enumerater�warning)�datar�idx�ps    rrrbs0���$��:�d�h�h�w�&7��>�>��W�
���Y�%��(�*�(��.��.�F�C����i��q��)�A�E�E�'�1�,=�q�u�u�X�q�?Q�+R�R�.��
��t���H�H�\�2�&�
�S�+�R�
0�
�S��"�
�	�$�E�*�40�*���
�!�G�a�e�e�G�Q�'����x��);�<�<�*�41�	42�43�N�N�6�7�
�I��%��44s
�AD+�4D1)NNNN)�logging�invoice_post_processingr�basicConfig�INFO�	getLoggerrr�LAYOUT_PROCESSORSrr��r�<module>r5s^���;����'�,�,�'�	�	�	�8�	$���<�&�'�������3�lr4