NgDVddlZddlZddlmZddlmZddlmZddlm Z m Z m Z m Z m Z mZddlmZmZmZmZmZmZdedee fd Zd efd Zddd d d d dddded ededeedeedededededeedefdZ ddededeedededeedefdZdS)N)BytesIO)Path)Optional)AnalysisDrawerFinalLayoutDrawer LayoutDrawerOCRLayoutDrawerODModelLayoutDrawerPdfminerLayoutDrawer)ExtractedLayoutDumperFinalLayoutDumperJsonLayoutDumper LayoutDumperObjectDetectionLayoutDumperOCRLayoutDumperdumperreturnct|tr"t|St|tr"t |St|t r"t|St|tr"t|Std|)zFor a given layout dumper, return the corresponding layout drawer instance initialized with a dumped layout dict. Args: dumper: The layout dumper instance Returns: LayoutDrawer: The corresponding layout drawer instance  layout_dumpzUnknown dumper type: ) isinstancerr dumpr r rr r r ValueError)rs k/var/www/html/ai-engine/env/lib/python3.11/site-packages/unstructured/partition/pdf_image/analysis/tools.py_get_drawer_for_dumperrs&566 ;"v{{}}==== F1 2 2;# >>>> FO , ,;6;;==9999 F- . .; V[[]];;;;999:::is_imagec^tjjdd}|rd|dSd|dS)ziGenerate a filename for the analysis artifacts based on the file type. Adds a random uuid suffix Nimage_z.pngpdf_z.pdf)uuiduuid4hex)rsuffixs r_generate_filenamer&/sEZ\\ bqb !F%$$$$$ &   rFTpng)filenamefile skip_bboxes skip_dump_od draw_grid draw_captionresizeformatlayout_dumpersanalyzed_image_output_dir_pathr(r)r*r+r,r-r.r/c |st|}|s|rdSt|} | dd|s?t|| } | D]} | | | |sVt |||| |||| }| D]&} t| }||'|dSdS)aSave the analysis artifacts for a given file. Loads some settings from the environment configuration. Args: layout_dumpers: The layout dumpers to save and use for bboxes rendering is_image: Flag for the file type (pdf/image) analyzed_image_output_dir_path: The directory to save the analysis artifacts filename: The filename of the sources analyzed file (pdf/image). Only one of filename or file should be provided. file: The file object for the analyzed file. Only one of filename or file should be provided. draw_grid: Flag for drawing the analysis bboxes on a single image (as grid) draw_caption: Flag for drawing the caption above the analyzed page (for e.g. layout source) resize: Output image resize value. If not provided, the image will not be resized. format: The format for analyzed pages with bboxes drawn on them. Default is 'png'. NTparentsexist_ok)r(save_dir)r(r)rr6r,r-r.r/) r&rmkdirradd_layout_dumperprocessrr add_drawer)rr1r(r)r*r+r,r-r.r/r0 output_pathjson_layout_dumper layout_dumperanalysis_drawerdrawers rsave_analysis_artifiactsr@9s>: 0%h//l566KdT222 %-    , @ @M  0 0 ? ? ? ?""$$$ "( %    , / /M+M::F  & &v . . . .!!!!!""rrenders_output_dir_pathc 2t|j}t|jd }t|dz |z dz } | sdSg} | D] } | st| 5} tj | } dddn #1swxYwY| jdkr#| t| | jdkr#| t| | jdkr#| t| | jd kr#| t|  | r|st|dz |z d z }nt|}|d d t!||||||| }| D]}|||dSdS)aRender the bounding boxes for a given layout dimp file. To be used for analysis after the partition is performed for only dumping the layouts - the bboxes can be rendered later. Expects that the analyzed_image_output_dir_path keeps the structure that was created by the save_analysis_artifacts function. Args: filename: The filename of the sources analyzed file (pdf/image) analyzed_image_output_dir_path: The directory where the analysis artifacts (layout dumps) are saved. It should be the root directory of the structure created by the save_analysis_artifacts function. renders_output_dir_path: Optional directory to save the rendered bboxes - if not provided, it will be saved in the analysis directory. draw_grid: Flag for drawing the analysis bboxes on a single image (as grid) draw_caption: Flag for drawing the caption above the analyzed page (for e.g. layout source) resize: Output image resize value. If not provided, the image will not be resized. format: The format for analyzed pages with bboxes drawn on them. Default is 'png'. pdfanalysisrNfinalrobject_detectionocrpdfminerbboxesTr3)r(r6rr,r-r.r/)rstemr%endswithexistsiterdiris_fileopenjsonloadappendrr r r r7rr:r9)r(r1rAr,r-r.r/ filename_stemranalysis_dumps_dirlayout_drawersanalysis_dump_filenamefrr;r>r?s rrender_bboxes_for_filerXxs8NN'M>>(11%888H +,,z9MIMY  $ $ & &N"4"<"<">"> Q Q%--//   ( ) ) 'Q)A,,K ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ! &' 1 1  ! !"3 "L"L"L M M M ! &*< < <  ! !"5+"N"N"N O O O ! &% / /  ! !/k"J"J"J K K K ! &* 4 4  ! !"6;"O"O"O P P P"& 8344zAMQT\\ K677K$666( %   % / /F  & &v . . . .!!!!!)""s(C  C C )NFTNr')rPr"iorpathlibrtypingrrbsk ;<;H\4J;;;;,#""<"<"<"!<"<"%(<"sm <" 7  <"  <"<"<"<" UO<" <"<"<"<"D.2"F"F"F"$'F"&c]F" F"  F" UO F" F"F"F"F"F"F"r