a
    ¡˜ëj '  ã                   @  s   d dl mZ d dlZd dlZd dlmZ d dlZd dlZd dl	Z
d dlZd dlmZ d dlmZ d dlmZ d dlmZ d dlmZmZmZmZ d d	lmZ d d
lmZ erÄd dlmZ d dl m!Z! dej"vrØdej"d< G dd„ deƒZ#dddddœdd„Z$dS )é    )ÚannotationsN)ÚTYPE_CHECKING)Úetree)ÚImage)Útrace_logger)Ú
env_config)ÚIMAGE_COLOR_DEPTHÚTESSERACT_MAX_SIZEÚTESSERACT_TEXT_HEIGHTÚSource)ÚOCRAgent)Úrequires_dependencies©ÚTextRegions)ÚLayoutElementsZOMP_THREAD_LIMITÚ1c                   @  sÄ   e Zd ZdZddiZd*ddœdd„Zd	d
„ Zdddœdd„Zdddœdd„Zd+ddddddœdd„Z	d,ddddœdd„Z
d-ddddœdd „Zed!ƒdd"dœd#d$„ƒZed!ƒd.dddd&œd'd(„ƒZd)S )/ÚOCRAgentTesseractz)OCR service implementation for Tesseract.Úhzhttp://www.w3.org/1999/xhtmlÚengÚstr©Úlanguagec                 C  s
   || _ d S ©Nr   )Úselfr   © r   út/mnt/bigstor/www/usecases/.venv/lib/python3.9/site-packages/unstructured/partition/utils/ocr_models/tesseract_ocr.pyÚ__init__'   s    zOCRAgentTesseract.__init__c                 C  s   dS )NTr   )r   r   r   r   Úis_text_sorted*   s    z OCRAgentTesseract.is_text_sortedúPILImage.Image)ÚimageÚreturnc                 C  s   t jt |¡| jd�S )N)Úlang)Úunstructured_pytesseractZimage_to_stringÚnpÚarrayr   )r   r   r   r   r   Úget_text_from_image-   s    z%OCRAgentTesseract.get_text_from_imager   c              
   C  sÎ   t  d¡ d}| jt |¡| jtjd�}| ¡ }|t	  
tj¡}|tjk sT|tjkr¼tdt t tt |j¡ t ¡d¡ƒ}tt tj| d¡|ƒ}| jt t||ƒ¡| jtjd�}| ¡ }| j||d�}|S )zHGet the OCR regions from image as a list of text regions with tesseract.z,Processing entire page OCR with tesseract...é   )r!   Úcharacter_confidence_thresholdr   )Úzoom)r   ZdetailÚ.image_to_data_with_character_confidence_filterr#   r$   r   r   Z(TESSERACT_CHARACTER_CONFIDENCE_THRESHOLDZdropnar
   ZquantileZTESSERACT_TEXT_HEIGHT_QUANTILEZTESSERACT_MIN_TEXT_HEIGHTZTESSERACT_MAX_TEXT_HEIGHTÚmaxÚroundÚsqrtr	   ÚprodÚsizer   ÚminZTESSERACT_OPTIMUM_TEXT_HEIGHTÚ
zoom_imageÚ
parse_data)r   r   r(   Úocr_dfZtext_heightZmax_zoomÚocr_regionsr   r   r   Úget_layout_from_image0   s>    
ýÿÿþ þþýz'OCRAgentTesseract.get_layout_from_imageÚ ç        z
np.ndarrayÚfloatzpd.DataFrame)r   r!   Úconfigr'   r    c                 C  s&   t j||d| dd�}|  ||¡}|S )Nz-c hocr_char_boxes=1 Úhocr)r!   r8   Ú	extension)r"   Zimage_to_pdf_or_hocrÚhocr_to_dataframe)r   r   r!   r8   r'   r9   r2   r   r   r   r)   \   s    üz@OCRAgentTesseract.image_to_data_with_character_confidence_filter)r9   r'   r    c              	   C  sì   g }|st j|g d¢d�S t |¡}| d| j¡}|D ]f}| dd¡}t d|¡}| j	||d�}	|	r6|r6t
tt| ¡ ƒƒ}
|
\}}}}| |||||	dœ¡ q6t j|g d¢d�}|d	 |d
  |d< |d |d  |d< |jd	dgd�}|S )N)ÚleftÚtopÚwidthÚheightÚtext)Úcolumnsz.//h:span[@class="ocrx_word"]Útitler5   zbbox (\d+) (\d+) (\d+) (\d+))Úwordr'   )r<   r=   ÚrightÚbottomr@   rD   r<   r>   rE   r=   r?   )ÚpdZ	DataFramer   Ú
fromstringÚfindallÚhocr_namespaceÚgetÚreÚsearchÚextract_word_from_hocrÚlistÚmapÚintÚgroupsÚappendZdrop)r   r9   r'   Z
df_entriesÚrootZ
word_spansZ	word_spanZ
word_titleZ
bbox_matchr@   Z	word_bboxr<   r=   rD   rE   r2   r   r   r   r;   l   s6    
ÿûÿ	z#OCRAgentTesseract.hocr_to_dataframezetree.Element)rC   r'   r    c           
      C  sz   |  d| j¡}t|ƒdkrdS d}|D ]N}|j}| dd¡}t d|¡}|r&|sRq&t| d¡ƒd }	|	|kr&||7 }q&|S )zTExtracts a word from an hOCR word tag, filtering out characters with low confidence.z.//h:span[@class="ocrx_cinfo"]r   r5   rB   zx_conf (\d+\.\d+)r&   éd   )	rH   rI   Úlenr@   rJ   rK   rL   r7   Úgroup)
r   rC   r'   Zcharacter_spansZ	word_textZcharacter_spanÚcharZ
char_titleZ
conf_matchZcharacter_probabilityr   r   r   rM   “   s    
z(OCRAgentTesseract.extract_word_from_hocrZunstructured_inferencer   c                 C  s.   ddl m} |  |¡}|  |¡}|||dd�S )Nr   )Ú&build_layout_elements_from_ocr_regionsT)r3   Úocr_textZgroup_by_ocr_text)Z0unstructured.partition.pdf_image.inference_utilsrX   r4   r%   )r   r   rX   r3   rY   r   r   r   Úget_layout_elements_from_image­   s    
	
ýz0OCRAgentTesseract.get_layout_elements_from_imager&   )Úocr_datar(   r    c                 C  s¾   ddl m} |dkrd}|j dd„ ¡j}|dk}|g d¢ j}|dd…d	f  |dd…df 7  < |dd…d
f  |dd…df 7  < | t¡| }||| || t t	j
g| ¡  ¡d�S )a  Parse the OCR result data to extract a list of TextRegion objects from tesseract.

        The function processes the OCR result data frame, looking for bounding
        box information and associated text to create instances of the TextRegion
        class, which are then appended to a list.

        Parameters:
        - ocr_data (pd.DataFrame):
            A Pandas DataFrame containing the OCR result data.
            It should have columns like 'text', 'left', 'top', 'width', and 'height'.

        - zoom (float, optional):
            A zoom factor to scale the coordinates of the bounding boxes from image scaling.
            Default is 1.

        Returns:
        - TextRegions:
            TextRegions object, containing data from all text regions in numpy arrays; each row
            represents a detected text region within the OCR-ed image.

        Note:
        - An empty string or a None value for the 'text' key in the input
          data frame will result in its associated bounding box being ignored.
        r   r   r&   c                 S  s   t | tƒst| ƒS |  ¡ S r   )Ú
isinstancer   Ústrip)r@   r   r   r   Ú<lambda>å   ó    z.OCRAgentTesseract.parse_data.<locals>.<lambda>r5   )r<   r=   r>   r?   Né   é   )Úelement_coordsÚtextsÚsources)Ú)unstructured_inference.inference.elementsr   r@   ÚapplyÚvaluesZastyper7   r#   r$   r   ZOCR_TESSERACTÚsum)r   r[   r(   r   rc   Úmaskrb   r   r   r   r1   Ä   s     ÿ$$ýzOCRAgentTesseract.parse_dataN)r   )r   r5   r6   )r6   )r6   )r&   )Ú__name__Ú
__module__Ú__qualname__Ú__doc__rI   r   r   r%   r4   r)   r;   rM   r   rZ   r1   r   r   r   r   r   "   s$   /   û ÿ( ÿr   r&   r   r7   )r   r(   r    c                 C  sj   |dkrd}t jt  t | ¡t j¡d||t jd�}t dtj¡}t j	||dd�}t j
||dd�}t |¡S )zšscale an image based on the zoom factor using cv2; the scaled image is post processed by
    dilation then erosion to improve edge sharpness for OCR tasksr   r&   N)ZfxZfyÚinterpolation)r&   r&   )Z
iterations)Úcv2ÚresizeZcvtColorr#   r$   ZCOLOR_RGB2BGRZINTER_CUBICZonesZuint8ZdilateZerodeÚPILImageZ	fromarray)r   r(   Z	new_imageÚkernelr   r   r   r0   ó   s    ûr0   )r&   )%Ú
__future__r   ÚosrK   Útypingr   ro   Únumpyr#   ZpandasrF   r"   Úlxmlr   ZPILr   rq   Zunstructured.loggerr   Z#unstructured.partition.utils.configr   Z&unstructured.partition.utils.constantsr   r	   r
   r   Z5unstructured.partition.utils.ocr_models.ocr_interfacer   Zunstructured.utilsr   re   r   Z.unstructured_inference.inference.layoutelementr   Úenvironr   r0   r   r   r   r   Ú<module>   s,   

 R