Team Ai
Apppublic

documentExtractionag051/ExtractDocument

sourceHugging Faceupdated 9mo agoView on Hugging Face
0likes
validation_utils.cpython-313.pyc49 linesDownload Raw Back to __pycache__
1�

2u�_i�	���SrSSKJrJr SSKr\R3"\5rS\\\4S\\	\444SjrS\\\4S\4Sjr
S	\\	\54S\\	\	44S6jrg)z�7validation_utils.py8===================9 10Utility functions for validation data operations.11 12This module provides helper functions for working with validation data structures.13�)�Dict�AnyN�data�returnc14�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$s snnfs snnf)15z�16Extract page dimensions from input data.17 18Supports both simplified schema and original engine output format.19 20Args:21    data (dict): Input data containing page information.22 23Returns:24    dict: {page_num: (width, height)}25�pages�pageNumr��width�height�26engineData�imagePreprocessingResultz2Could not extract page dimensions from input data.)�27isinstance�get�list�	enumerate�logger�warning)rr�idx�ps    �nC:\Users\poona\OneDrive\Documents\nextlifyai\Document_Automation\Document_Extraction\utils\validation_utils.py�extract_page_dimensionsrs0���$��:�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��"�
�	�$�E�*�28�*���
�!�G�a�e�e�G�Q�'����x��);�<�<�*�29�	30�31�N�N�G�H�
�I��%��32s
�AD+�4D1�	data_dictc�B�[SUR555$)z�33Check if all 'value' fields in a dictionary are empty strings.34 35Args:36    data_dict (dict): Dictionary with nested value fields.37 38Returns:39    bool: True if all values are empty, False otherwise.40c3�z# �UH1n[U[5=(a URSS5S:Hv� M3 g7f)�value�N)r�dictr)�.0�vs  r�	<genexpr>�%check_empty_values.<locals>.<genexpr>@s6����#�A�	�1�d��8����g�r� 2�b� 8�8�#�s�9;)�all�values)rs r�check_empty_valuesr%6s'�����!�!�#�����page_dimensionsc�j�0nSn[UR55HnX!U'XupEX%-
nM U$)z�41Calculate cumulative page heights for multi-page documents.42 43Args:44    page_dimensions (dict): {page_num: (width, height)}45 46Returns:47    dict: {page_num: cumulative_height_offset}48r)�sorted�keys)r'�cumulative_heights�cumulative_height�page_num�_rs      r�calculate_cumulative_heightsr/FsK�������?�/�/�1�2��'8�8�$�#�-�	���#��3�49�r&)�__doc__�typingrr�logging�	getLogger�__name__r�str�int�tupler�boolr%r/�r&r�<module>r:s������	�	�	�8�	$��#�$�s�C�x�.�#�T�#�u�*�5E�#�L
�$�s�C�x�.�
�T�
� �$�s�E�z�2B��t�C�QT�H�~�r&