Ng|FTddlmZddlZddlZddlmZmZmZmZm Z m Z ddl Z ddl m Zddl mZddlmZddlmZddlmZdd lmZmZdd lmZdd lmZdd lmZdd lm Z erddl!m"Z"ddl#m$Z$m%Z%ddl&m'Z'ddl(m)Z)dddej*j+ddfdJd&Z,e d'dddej*j+ddfdKd)Z-e d'ddej*j+ddfdLd0Z.e d' dMdNd5Z/ dOdPd<Z0 dQdRdAZ1ej2fdSdGZ3e d'ej2fdTdIZ4dS)U) annotationsN)IO TYPE_CHECKINGAnyListOptionalcast)Image) ImageSequence) ElementType)SimpleTableCell)OCRLayoutDumper)pad_element_bboxes valid_text) env_config)OCRMode)OCRAgent)requires_dependencies) TextRegion)DocumentLayout PageLayout) LayoutElement)!UnstructuredTableTransformerModelFengdatabytes | IO[bytes] out_layout'DocumentLayout'extracted_layoutList[List['TextRegion']]is_imageboolinfer_table_structure ocr_languagesstrocr_mode pdf_image_dpiintocr_layout_dumperOptional[OCRLayoutDumper]returnc t|tr|n|} tj5} t j| d} t| d5} | | dddn #1swxYwYt| |||||||| } dddn #1swxYwY| S)a Process OCR data from a given data and supplement the output DocumentLayout from unstructured_inference with ocr. Parameters: - data (Union[bytes, BinaryIO]): The input file data, which can be either bytes or a BinaryIO object. - out_layout (DocumentLayout): The output layout from unstructured-inference. - is_image (bool, optional): Indicates if the input data is an image (True) or not (False). Defaults to False. - infer_table_structure (bool, optional): If true, extract the table content. - ocr_languages (str, optional): The languages for OCR processing. Defaults to "eng" (English). - ocr_mode (str, optional): The OCR processing mode, e.g., "entire_page" or "individual_blocks". Defaults to "entire_page". If choose "entire_page" OCR, OCR processes the entire image page and will be merged with the output layout. If choose "individual_blocks" OCR, OCR is performed on individual elements by cropping the image. - pdf_image_dpi (int, optional): DPI (dots per inch) for processing PDF images. Defaults to 200. - ocr_layout_dumper (OCRLayoutDumper, optional): The OCR layout dumper to save the OCR layout. Returns: DocumentLayout: The merged layout information obtained after OCR processing. tmp_filewbN) filenamerr r"r$r%r'r(r*) isinstancebytesreadtempfileTemporaryDirectoryospathjoinopenwriteprocess_file_with_ocr)rrr r"r$r%r'r(r* data_bytes tmp_dir_path tmp_file_pathr.merged_layoutss `/var/www/html/ai-engine/env/lib/python3.11/site-packages/unstructured/partition/pdf_image/ocr.pyprocess_data_with_ocrrAsHP$D%00AdiikkJ  $ & & , \:>> - & & '( NN: & & & ' ' ' ' ' ' ' ' ' ' ' ' ' ' '/"!-"7''/                    " s51C0B CB CB CCCunstructured_inferencer0c ddlm} g} |rtj|5} | j} t t j| D]t\} }|d}| |_| t|kr|| nd}t|j | ||||||}| |u| j | cdddS#1swxYwYdStj5}t!j|||d}t%t&t(|}t |D]\} }| t|kr|| nd}tj|5}t|j | ||||||}| |dddn #1swxYwY| j | cdddS#1swxYwYdS#t*$rY}t,j|st,j|r|t5d|d |d}~wwxYw) a Process OCR data from a given file and supplement the output DocumentLayout from unstructured-inference with ocr. Parameters: - filename (str): The path to the input file, which can be an image or a PDF. - out_layout (DocumentLayout): The output layout from unstructured-inference. - is_image (bool, optional): Indicates if the input data is an image (True) or not (False). Defaults to False. - infer_table_structure (bool, optional): If true, extract the table content. - ocr_languages (str, optional): The languages for OCR processing. Defaults to "eng" (English). - ocr_mode (str, optional): The OCR processing mode, e.g., "entire_page" or "individual_blocks". Defaults to "entire_page". If choose "entire_page" OCR, OCR processes the entire image page and will be merged with the output layout. If choose "individual_blocks" OCR, OCR is performed on individual elements by cropping the image. - pdf_image_dpi (int, optional): DPI (dots per inch) for processing PDF images. Defaults to 200. Returns: DocumentLayout: The merged layout information obtained after OCR processing. r)rRGBN) page_layoutimager$r%r'extracted_regionsr*T)dpi output_folder paths_onlyzFile "z " not found!)'unstructured_inference.inference.layoutrPILImager9format enumerater Iteratorconvertlensupplement_page_layout_with_ocrpagesappend from_pagesr4r5 pdf2imageconvert_from_pathr rr& Exceptionr6r7isdirisfileFileNotFoundError)r0rr r"r$r%r'r(r*rmerged_page_layoutsimages image_formatirFrGmerged_page_layouttemp_dir _image_paths image_paths image_pathes r@r;r;\sNGFFFFF,..L ( Fx(( FF%} )-*@*H*H I I C CHAu!MM%00E#/EL?@3GWCXCX?X?X(8(;(;^b%)H$.$4Q$7#.C&3!)*;*;***&(../ABBBB0~01DEE! F F F F F F F F F F F F F F F F F F$,.. F((:%"*#     #49l;; %.{%;%; G GMAz?@3GWCXCX?X?X(8(;(;^b%!z22 Ge-L(2(8(;"'2G*7%-.?.?...*,223EFFF G G G G G G G G G G G G G G G1~01DEE+ F F F F F F F F F F F F F F F F F F, LLL 7== " " LbgnnX&>&> LG#$CX$C$C$CDD! K LsG0B-C G0CG0!C"G0'G0:A>G#87F;/ G#;F? ?G#F? G# G0#G''G0*G'+G00 I:AIIrE 'PageLayout'rFPILImage.ImagerGOptional[List['TextRegion']]cltj|}|tjjkrf||}|r||tttd|j ||j dd<n|tj jkr|j D]} | j sztj} t| | } || jj| jj| jj| jjf} || } | | _ nt/d|rqddlm}||jt9d t;ttd|j ||j|| |j dd<|S) a6 Supplement an PageLayout with OCR results depending on OCR mode. If mode is "entire_page", we get the OCR layout for the entire image and merge it with PageLayout. If mode is "individual_blocks", we find the elements from PageLayout with no text and add text from OCR to each element. )languager)r ocr_layoutNpaddingz[Invalid OCR mode. Parameter `ocr_mode` must be set to `entire_page` or `individual_blocks`.r)tablesz&Unable to load table extraction agent.)elementsrF tables_agent ocr_agentrG)r get_agentr FULL_PAGEvalueget_layout_from_imageadd_ocred_page merge_out_layout_with_ocr_layoutr rroINDIVIDUAL_BLOCKStextrIMAGE_CROP_PADrcropbboxx1y1x2y2get_text_from_image ValueErrorunstructured_inference.modelsrn load_agentrp RuntimeError(supplement_element_with_table_extraction)rErFr$r%r'rGr*rqrkelementrmpadded_element cropped_image text_from_ocrrns r@rRrRs$"M:::I7$***44U;;  9  , ,Z 8 8 8"BD1;3GHH!# # # QQQ W.4 4 4"+ - -G< -$3!3GW!M!M!M % &+.&+.&+.&+. !! !* = =m L L ,  -" C     888888   &GHH H"J$/1EFF,/ # # # QQQ roList['LayoutElement']rp#'UnstructuredTableTransformerModel'cddlm}d|D}|D]}tj}t ||} || jj| jj| jj | jj f} t| ||| } | | | d} | dkrdn || } | |_ tjrd | D}||_|S) aHSupplement the existing layout with table extraction. Any Table elements that are extracted will have a metadata fields "text_as_html" where the table's text content is rendered into a html string and "table_as_cells" with the raw table cells output from table agent if env_config.EXTRACT_TABLE_AS_CELLS is True r) cells_to_htmlc<g|]}|jtjk|S)typer TABLE).0els r@ z. s'LLLRrw+:K/K/Kb/K/K/Krrl)table_element_imagerqrG table_elementcells) ocr_tokens result_formatcZg|](}tj|)Sr)r from_table_transformer_cellto_dict)rcells r@rz.&s<"""PT;DAAIIKK"""r)$unstructured_inference.models.tablesrrTABLE_IMAGE_CROP_PADrr{r|r}r~rrget_table_tokenspredict text_as_htmlEXTRACT_TABLE_AS_CELLStable_as_cells)rorFrprqrGrtable_elementsrrmrr table_tokens tatr_cellsrsimple_table_cellss r@rrs5CBBBBBLL8LLLN!881+GWEEE #&#&#&#&    ( -/(    ")) l'*  (2--rr==3L3L +  , 8""Xb""" &8G " OrrrqrrOptional['LayoutElement']List[dict[str, Any]]c2||}g}|D]K}||jj|jj|jj|jjg|jdLt|D] \}}d|vr||d<d|vrd|d<d|vrd|d<!|S)z1Get OCR tokens from either paddleocr or tesseract)rF)r|ryspan_numline_numr block_num) rurTr|r}r~rrryrN) rrqrGrrkr ocr_regionidxtokens r@rr.s007J0KKJL   O&O&O&O&  #    --## U U " " #E*  U " " !E*  e # #!"E+  rTrkList['TextRegion']supplement_with_ocr_elementsczd|D}|D]}t|||_|rt||n|}|S)a Merge the out layout with the OCR-detected text regions on page level. This function iterates over each out layout element and aggregates the associated text from the OCR layout using the specified threshold. The out layout's text attribute is then updated with this aggregated text. If `supplement_with_ocr_elements` is `True`, the out layout will be supplemented with the OCR layout. c:g|]}t|j|Sr)rry)rregions r@rz4merge_out_layout_with_ocr_layout..`s)___6zRXR]G^G^____r)aggregate_ocr_text_by_blockry#supplement_layout_with_ocr_elements)rrkrout_regions_without_text out_region final_layouts r@rwrwRsk `_Z___.  5     ( +J CCC  rr 'TextRegion'subregion_thresholdfloat Optional[str]cg}|D]E}|j|j|}|r!|jr||jF|rd|ndS)zdExtracts the text aggregated from the regions of the ocr layout that lie within the given block. r)r|is_almost_subregion_ofryrTr8)rkrrextracted_textsr'ocr_region_is_subregion_of_given_regions r@rrqs~O 44 2.s$aaaVVK`=`=`&=`=`=`r)0unstructured.partition.pdf_image.inference_utilsrr|rrT) rrkrrrr!ocr_region_is_subregion_of_out_elocr_regions_to_addocr_elements_to_addrrs @r@rrsB/1    B0:0V0V#11 -1 %,,Z888 baaazaaaDDEWXX 33  r)rrrrr r!r"r#r$r#r%r&r'r&r(r)r*r+r,r)r0r&rrr r!r"r#r$r#r%r&r'r&r(r)r*r+r,r)rErfrFrgr$r#r%r&r'r&rGrhr*r+r,rf)N) rorrFrgrprrGrhr,r)NN) rrgrqrrGrhrrr,r)T)rrrkrrr#r,r)rkrrrrrr,r)rrrkrrrr,r)5 __future__rr6r4typingrrrrrr rVPILr rLr unstructured.documents.elementsr (unstructured.metrics.table.table_formatsr 5unstructured.partition.pdf_image.analysis.layout_dumpr0unstructured.partition.pdf_image.pdf_image_utilsrr#unstructured.partition.utils.configr&unstructured.partition.utils.constantsr5unstructured.partition.utils.ocr_models.ocr_interfacerunstructured.utilsr)unstructured_inference.inference.elementsrrKrr.unstructured_inference.inference.layoutelementrrrrsrtrAr;rRrrrwOCR_LAYOUT_SUBREGION_THRESHOLDrrrrr@rs5"""""" ????????????????"!!!!!777777DDDDDDQQQQQQ[[[[[[[[::::::::::::JJJJJJ444444WDDDDDDRRRRRRRRLLLLLLVVVVVV"'%+37;;;;;|/00 "'%+37WLWLWLWL10WLt/00#(%+6:37AAAA10AH/00 7; ....10.h7;/3 !!!!!N*.D",!J@@@@@*/00",!J666610666r