kh3OdZddlZddlZddlZddlmZmZddlmZddl m Z m Z m Z ddl Z ddlmZddlmZdd lmZeeee j.j1d Zegd ksgd ecxkrgd kr nnedej6ej8dGddZGddeZGddeZ y)zPDF to Docx Converter.N)Pool cpu_count) perf_counter)AnyStrIOUnion)Document)Page)Pages.)r r)r )r rzJ1.19.0 <= PyMuPDF <= 1.23.8, or PyMuPDF>=1.23.16 is required for pdf2docx.z[%(levelname)s] %(message)s)levelformatc feZdZdZ d$dededefdZedZedZ d Z ed Z d%d e d e d e fdZd%d e d e d e fdZdZdZd&dZdZdefdZdefdZdefdZd$de dededefdZd'deeeefd e d e d e fdZd%d e d e d e fdZded e d e fd Zed!Z ed"Z!ed#Z"y)( ConverteraThe ``PDF`` to ``docx`` converter. * Read PDF file with ``PyMuPDF`` to get raw layout data page by page, including text, image, drawing and its properties, e.g. boundary box, font, size, image width, height. * Analyze layout in document level, e.g. page header, footer and margin. * Parse page layout to docx structure, e.g. paragraph and its properties like indentation, spacing, text alignment; table and its properties like border, shading, merging. * Finally, generate docx with ``python-docx``. Npdf_filepasswordstreamc||_t|xsd|_|s |s td|rt j ||_nt j ||_t|_y)zInitialize fitz object with given pdf file path. Args: pdf_file (str): pdf file path. stream (bytes): pdf file in memory. password (str): Password for encrypted pdf. Default to None if not encrypted. z(Either pdf_file or stream must be given.)rN) filename_pdfstrr ValueErrorfitzr _fitz_docr _pages)selfrrrs N/var/www/teggl/fontify/venv/lib/python3.12/site-packages/pdf2docx/converter.py__init__zConverter.__init__&s]%HN+ GH H !]]&9DN"]]84DNg c|jSN)rr!s r"fitz_doczConverter.fitz_docAs #~~-r$c|jSr&)r r's r"pageszConverter.pagesDs  KK'r$c8|jjyr&)rcloser's r"r,zConverter.closeHsT^^))+r$ciddddddddddd d d d d ddddddd dd dddddddd ddddddddd dddddddd!S)"zDefault parsing parameters.debugFocrrignore_page_errorTmulti_processingrmin_section_heightg4@connected_border_toleranceg?max_border_widthg@min_border_clearanceg@float_image_ignorable_gapg@page_margin_factor_toppage_margin_factor_bottomshape_min_dimensionmax_line_spacing_ratiog?line_overlap_thresholdg?line_break_width_ratioline_break_free_space_ratiog?g333333?g?g@g.@)line_separate_thresholdnew_paragraph_free_space_ratiolines_left_aligned_thresholdlines_right_aligned_thresholdlines_center_aligned_thresholdclip_image_res_ratiomin_svg_gap_dxmin_svg_gap_dy min_svg_w min_svg_hextract_stream_tableparse_lattice_tableparse_stream_tabledelete_end_line_hyphenr's r"default_settingszConverter.default_settingsKs  u  q   t   u  q  !t   )s   s   #s   (s   %s   (s   "s   %s   %s  %s!  " *s#  $03/3/2/2/2/2/3/2/2/2/4/3/3/4?  r$startendr*c h|j|||jdi|jdi|S)a2Parse pages in three steps: * open PDF file with ``PyMuPDF`` * analyze whole document, e.g. page section, header/footer and margin * parse specified pages, e.g. paragraph, image and table Args: start (int, optional): First page to process. Defaults to 0, the first page. end (int, optional): Last page to process. Defaults to None, the last page. pages (list, optional): Range of page indexes to parse. Defaults to None. kwargs (dict, optional): Configuration parameters. rL) load_pagesparse_document parse_pages)r!rNrOr*kwargss r"parsezConverter.parsetsDtuc51 ^&$& [#!# #r$c 0tj|jd|jjrU|j st d|jd|jj|j s t dt|j}|jjt|Dcgc]}t|dc}|j||||}|D]}d|j|_|Scc}w)aStep 1 of converting process: open PDF file with ``PyMuPDF``, especially for password encrypted file. Args: start (int, optional): First page to process. Defaults to 0, the first page. end (int, optional): Last page to process. Defaults to None, the last page. pages (list, optional): Range of page indexes to parse. Defaults to None. z[1/4] Opening document...zRequire password for r zIncorrect password.Tid skip_parsingF)logginginfo _color_outputr needs_passrConversionExceptionr authenticatelenr resetranger _page_indexesrY)r!rNrOr*numi page_indexess r"rQzConverter.load_pagess  T''(CDE >> $ $==),A$BSBSATTU*VWW^^00?)*?@@$..! %*MQ4148MN))%eSA  0A*/DKKN ' 0 NsDc tj|jd|jj|j fi||S)zjStep 2 of converting process: analyze whole document, e.g. page section, header/footer and margin.z[2/4] Analyzing document...)rZr[r\r rUr()r!rTs r"rRzConverter.parse_documents>  T''(EFG $--262 r$c tj|jd|jDcgc]}|jr|}}t |}t |dD]?\}}|jdz}tjd||| |jd i|A|Scc}w#t$r=}|ds|drtjd||ntd|d |Yd }~d }~wwxYw) zKStep 3 of converting process: parse pages, e.g. paragraph, image and table.z[3/4] Parsing pages...r rN(%d/%d) Page %dr.r0z,Ignore page %d due to parsing page error: %szError when parsing page : NrL) rZr[r\r rYr` enumeraterXrU Exceptionerrorr^)r!rTpager* num_pagesrepides r"rSzConverter.parse_pagess T''(@AB"&++G$T5F5FGGJ  a0 UGAt''A+C LL*Ay# > U $V$ U H Ug62E+FMM"PRUWXY-0HRPQs.STTZ Us#B'B'B,, C253C--C2c tj|jdtt d|j }|s t d|sh|jrQ|jdtd d}tjj|r!tj|n t dt}t|}t|d D]K\}}|js|j dz}tjd ||| |j#|M|j+|y#t$$r=} |d s|d rtj&d || nt)d|d| Yd} ~ d} ~ wwxYw)zStep 4 of converting process: create docx file with converted pages. Args: filename_or_stream (str, file-like): docx file to write. kwargs (dict, optional): Configuration parameters. z[4/4] Creating pages...c|jSr&) finalized)ros r"z%Converter.make_docx..s r$z)No parsed pages. Please parse page first.rz.pdfz.docxz?Please specify a docx file name or a file-like object to write.r rirjr.r0z+Ignore page %d due to making page error: %szError when make page rkN)rZr[r\listfilterr r^rr`ospathexistsremover rlrurX make_docxrmrnMakedocxExceptionsave) r!filename_or_streamrT parsed_pages docx_filerprerorqrrs r"r}zConverter.make_docxsm  T''(ABCF '  %&QR R!  (,(9(9!S[L(I'J%%P"77>>"45ryyAS7T)*kllJ  %  Q7 PGAt>>8''A+C LL*Ay# > Py) P )* Pg62E+FMM"OQTVWX+.CC51#,NOOY PsD>> F3E??Fctjj|jt |j |j Dcgc]}|j s|j!c}dScc}w)z"Store parsed pages in dict format.)filenamepage_cntr*)ryrzbasenamerr`r rustore)r!ros r"rzConverter.storesV(():):;DKK(26++P$P  Qs A4A4datac R|jsO|jdd}|jjt|Dcgc]}t |dc}|jdgD]2}|jdd}|j|j |4ycc}w) z"Restore pages from parsed results.rdTrWr*rXN)r getrarbr restore)r!rrdreraw_pageidxs r"rzConverter.restores{{((:s+C KK  eCjQtqt<Q R"- /H,,tR(C KK  $ $X . /RsB$rct|dd5}|jtj|j ddddy#1swYyxYw)z*Write parsed pages to specified JSON file.wzutf-8)encoding)indentN)openwritejsondumpsr)r!rfs r" serializezConverter.serializesB (C' 2 8a GGDJJtzz|A6 7 8 8 8s 5A  Act|d5}tj|}ddd|jy#1swYxYw)z+Load parsed pages from specified JSON file.rN)rrloadr)r!rrrs r" deserializezConverter.deserialize s< (C  A99Q(( X01%'KL L & ' /D / / uc VX V ?JDJJuc5 5H 5 ? ? ZQY Z +\^B->?r$c |j}|j||j|||fi|g}|jD]0}|js|j |j di|2|S)aExtract table contents from specified PDF pages. Args: start (int, optional): First page to process. Defaults to 0, the first page. end (int, optional): Last page to process. Defaults to None, the last page. pages (list, optional): Range of page indexes. Defaults to None. kwargs (dict, optional): Configuration parameters. Defaults to None. Returns: list: A list of parsed table content. rL)rMrrUr ruextendextract_tables)r!rNrOr*rTrtablesros r"rzConverter.extract_tablesbsx(( 5#u11KK ND~~v}}-@T-@-@-L8-LM N r$c |drt|dtn t}d}t|Dcgc]&}|||||j|j||d|df(}}t } | j |j|dt|D]P}|d|d} tjj| s+|j| tj| R|j|fi|ycc}w)zParse and create pages based on page indexes with multi-processing. Reference: https://pymupdf.readthedocs.io/en/latest/faq.html#multiprocessing rr*-z.jsonr N)minrrbrrrmap_parse_pages_per_cpuryrzr{rr|r}) r!rrNrOrTcpuprefixrevectorspoolrs r"rz(Converter._convert_with_multi_processing{s8>k7Jc&%y{3PYP[CH:O>?sE3(9(94=="vhas%$8:OOv **GQ7s A 1#U+H77>>(+X   X & IIh    }//Os+C<cr|\}}}}}}}}t||} | j|xst| j}t ||} t| } t | |z } | |z} | t || kz}| dz|zt | |z dz}t ||z| }t ||Dcgc]}| | }}| jD] }d|_ |D]}d| j|_| jdi|jdi|j|| jycc}w)aRender a page range of a document. Args: vector (list): A list containing required parameters. * 0 : segment number for current process * 1 : count of CPUs * 2,3: whole pages range to process * 4 : pdf filename * 5 : password for encrypted pdf * 6 : configuration parameters * 7 : json filename storing parsed results r rTFNrL) rrQr`r(rbintrr*rYrRrSrr,)vectorrrsrr pdf_filenamerrT json_filenamecv all_indexesrpmnseg_sizeseg_fromseg_torerfros r"rzConverter._parse_pages_per_cpusLIOES!Q h |X .   !R[[!Aqk  $   #  Os3q5z>aC9s1S5!},X()405h0GH1 AH HHH6D$d/6 -A',BHHQK $ - #F# [ #! # Y} %  Is1 D4c|r|Dcgc] }t|}}|S|xs|}tt|t|}t||}|Scc}w)zParsing arguments.)rslicerb)rNrOr*pdf_lenxindexesrs r"rczConverter._page_indexess\ ',-!s1v-G-  .Cc%j#c(+AGnQ'G .sAcd|dS)NzzrL)msgs r"r\zConverter._color_outputs%/uG#<((+^  /4 /88 3$3$c$#$X[$6-@U36 ?%;-@3-@Y\-@lp-@`3cd2030c0s08++\  <rs  +$$  S$""((- ./x<8A'i' a bb ,, (* y=y=x  )  + r$