vlj9dZddlZddlZddlZddlZddlmZddlm Z m Z ddl m Z m Z mZddlZddlmZmZmZddlmZmZmZmZddlmZdd lmZdd lmZmZm Z m!Z!dd l"m#Z#m$Z$m%Z%dd l&m'Z'ejPejRe*Z+ejXd Z-de.e/zde.fdZ0d)de de1de.dzddfdZ2 d*de dede3ddfdZ4 d+de dede.dzde3ddf dZ5 d,de de.de de e6de.de3de.dzde.dzddfd Z7e#d!Z8e#d"Z9de.de.de.ddfd#Z: d-de de.de e6de e6de.de3de.dzde.dzde3ddfd$Z;defd%Ze*d(k(re>yy).z#Extract pdf structure in XML formatN)ArgumentParser) ContainerIterable)AnyTextIOcast) PDFDocument PDFNoOutlinesPDFXRefFallback) PDFIOErrorPDFObjectNotFound PDFTypeError PDFValueError)PDFPage) PDFParser) PDFObjRef PDFStreamresolve1 stream_value)LIT PSKeyword PSLiteral)isnumberz&[\000-\037&<>()"\042\047\134\177-\377]sreturnclt|tr t|dn|}tj d|S)Nzlatin-1c>dt|jddS)Nz&#r;)ordgroup)ms J/var/www/html/pp/venv/lib/python3.12/site-packages/../../../bin/dumppdf.pyzescape.."s2c!''!*o%6a!8) isinstancebytesstrESC_PATsub)russ r"escaper+ s+(E2Q B ;;8" ==r$outobjcodecc||jdyt|tr|jdt|d|j D]H\}}|jd|d|jdt |||jdJ|jdyt|t rT|jd t|d|D]}t |||jd !|jd yt|ttfr+|jd t|d t|dyt|tr|dk(r |j|jy|dk(r |j|jy|jdt ||j|jd|dk(r:|j}|jdt|d t|d|jdyt|tr |jd|jdyt|t r |jd|j"dyt|t$r |jd|j"dyt'|r|jd|dyt)|)Nzz zz zz zz z rawbinaryz z textz z z z z zz )writer%dictlenitemsdumpxmllistr'r&r+r get_rawdataget_dataattrsrobjidrnamerrr)r,r-r.kvdatas r"r9r9%sw { *#t LS $/0IIK $DAq IIaS) * IIi CO IIl #  $ )#t LS $/0 A CO IIdO  )#U|$ N3s8*Bvc{m9EF#y! E> IIcoo' ( h  IIclln %  II+ , C # IIn %||~ LT 2fTl^9MN IIk "#y! Icii[-.#y! IchhZz23#y! IchhZz23} HSE+, s r$docshow_fallback_xrefc6|jD]Q}t|tr|s|jdt ||j |jdSt d|jD}|r|sd}tj|yyy)Nz z c3<K|]}t|tywN)r%r ).0xrefs r" zdumptrailers..psK:dO4KszThis PDF does not have an xref. Use --show-fallback-xref if you want to display the content of a fallback xref that contains all objects.) xrefsr%r r5r9 get_trailerallloggerwarning)r,rCrDrIno_xrefsmsgs r" dumptrailersrRfs  *$04F IIm $ C))+ , II( ) * KKKH* $ s +xr$ct}|jd|jD]v}|jD]a}||vr|j | |j |}|.|jd|dt ||||jdcxt||||jdy#t$r}td|Yd}~d}~wwxYw)Nzz .resolve_destsr$e - T 23D), TYY 78$%Cy$ *||~Kr$z rSz/'GoTo'rizG3G! G! GG!GG!!G*Filespec EmbeddedFilec dtdtttfddf fd }t |d5}t |}t || t} jD]p}|jD][} j|} ||vst| ts*| jdtusB|j|||| ]r dddy#1swYyxYw)Nr>r-rctjj|jdxs-t t |jdj }|djdxs|djd}j|j}t|tsd|d}t||jdturtd|dtjj |dd |}tjj|rtd |t!d |tj"tjj%|d t'|d5}|j)|j+dddy#1swYyxYw)NUFFEFz%unable to process PDF: reference for z is not a PDFStreamTypez is not an EmbeddedFile06d-z file exists: z extracting: T)exist_okwb)ospathbasenamertrr&decoderXr>r%rrLITERAL_EMBEDDEDFILEjoinexistsr rYmakedirsdirnamernr5r<) r>r-filenamefilereffileobj error_msgrr,rCrcs r"extract1z!extractembedded..extract1sq77##CGGDM$WT%5N5U5U5WXd)--%;Ts);**W]]+'9-7|CVW  * * ;;v &: :7|D)) ww||J5+Qxj(AB 77>>$ }TH56 6 TH%& BGGOOD)D9 $  * IIg&&( ) * * *s F55F>rer)intr6r'rrnrr rUrKrVrXr%rtLITERAL_FILESPECrW) r^rarcrrwrxextracted_objidsrIr>r-rCs ` @r"extractembeddedrs**$sCx.*T*, eT  )b2&(+5II )D* )jj'!11"3-+;;$((/UC( ) ) ) ) )sAC CC4CC%c t|d5} t| } t| |} |r&|D]!} | j| } t || |#|rnt t j| D]M\}}||vs |r+|jD]} t| } t || |8t ||jO|rt|| |||s|s|s t|| |ddd|dvr|jdyy#1swY xYw)NrerT)r2r3r1)rnrr rXr9rorrpcontentsrr=r\rRr5)r]r^r_r`rarbr.rcrDrwrxrCr>r-ryrzs r"dumppdfrs eT 9b2&(+  1jj's%0 1  )'*>*>s*C D 3 W$#'===C".s"3C#E3e<= tzz2 3  sE+= >7 %7 8'9( %% D&)99sA*C?8A(C??Dcttd}|jdtddd|jddd d tj  |jd d ddd|j }|jddddd|jddtd|jdd}|jdtddd|jddtd |jd!d"td#|jd$d%ddd&|jd'dd()|jd*d+td,d-.|jd/d0}|jd1d2td3d4.|j }|jd5d6ddd7|jd8d9ddd:|jd;dNT) descriptionadd_helpfiles+zOne or more paths to PDF files.)typedefaultnargshelpz --versionz-vversionzpdfminer.six v)rrz--debugz-dF store_truezUse debug logging level.)rrrz --extract-tocz-TzExtract structure of outlinez--extract-embeddedz-EzExtract embedded files)rrParserzUsed during PDF parsing)rz--page-numbersz0A space-seperated list of page numbers to parse.z --pagenosz-pzA comma-separated list of page numbers to parse. Included for legacy applications, use --page-numbers for more idiomatic argument entry.z --objectsz-iz1Comma separated list of object numbers to extractz--allz-az3If the structure of all objects should be extractedz--show-fallback-xrefzAdditionally show the fallback xref. Use this if the PDF has zero or only invalid xref's. This setting is ignored if --extract-toc or --extract-embedded is used.)rrz --passwordz-Pz,The password to use for decrypting PDF file.)rrrOutputzUsed during output generation.z --outfilez-orzJPath to file where output is written. Or "-" (default) to write to stdout.z --raw-streamz-rz%Write stream objects without encodingz--binary-streamz-bz)Write stream objects with binary encodingz --text-streamz-tz"Write stream objects as plain text) r__doc__ add_argumentr'pdfminer __version__add_mutually_exclusive_groupadd_argument_groupr)rxprocedure_parser parse_params output_params codec_parsers r" create_parserrs $ ?F   .    !5!5 67     ' ::<!!  + "!!  % ",,--L  ?     @    B 7  ; --4.M  !==?L  4   8   1  Mr$argvc Xt}|j|}|jr1tjj tj |jr2|jjdDcgc] }t|c}ng}|jr|jDchc]}|dz  }}nK|jr5|jjdDchc]}t|dz }}n t}|j}|jrd}n |jrd}n|j rd}nd}|j"dk(rt$j&nt)|j"d5}|j*D]y} |j,rt/|| ||||j0|d -|j2rt5| ||j2 Rt7|| ||||j0|d|j8 { dddycc}wcc}wcc}w#1swYyxYw) N)args,rfr2r3r4rw)rarbr.rc)rarc)rarbr.rcrD)r parse_argsdebuglogging getLoggersetLevelDEBUGobjectssplitr page_numbersr`rUra raw_stream binary_stream text_streamoutfilesysstdoutrnr extract_tocrrMextract_embeddedrrrD) rrxrxr_r`rar.r]r^s r"mainrs _F   $  'D zz$$W]]3:>,,dll005 6c!f 6BF "&"3"34Q1q544 '+||'9'9#'>?!3q6A:??%}}H !    ||s*T\\30G5ZZ E% HH# &&H9N9N% HH#'+'>'> # +75? sH4 H,H>B H  H)__main__rG)F)NF)rFNN)rFNNF)?rros.pathrrerargparsercollections.abcrrtypingrrrrpdfminer.pdfdocumentr r r pdfminer.pdfexceptionsr r rrpdfminer.pdfpagerpdfminer.pdfparserrpdfminer.pdftypesrrrrpdfminer.psparserrrrpdfminer.utilsr basicConfigr__name__rNcompiler(r'r&r+rrr9boolrRr\rrrrrrrr:rr$r"rs) #/$$LL %(JJ77#   8 $ "**> ?>cEk>c> >>f>S4Z>4>H %   .$    :   >!8 8 8 8s^ 8  8  8 :8d 8 8vz?>*$)3$)#$)3$)4$)X!$      SM s^      : d     Fx~xv8tCy4848v zFr$