B |b7 @sdZddlZddlZddlZddlZddlmZmZm Z m Z m Z m Z m Z mZmZddlmZddlZddlmZmZmZddlmZddlmZddlmZmZdd lmZmZm Z m!Z!dd l"m#Z#m$Z$m%Z%dd l&m'Z'e(e)e*Z+e,d Z-ee.e/fe.d ddZ0d-e e1e e.ddddZ2d.e ee3ddddZ4d/e ee e.e3ddddZ5d0e e.eee6e.e3e e.e e.dd ddZ7e%dZ8e%dZ9e.e.e.dd d!d"Z:d1e e.e e6ee6e.e3e e.e e.e3dd# d$d%Z;ed&d'd(Z()"\042\047\134\177-\377])sreturncCs*t|trt|d}n|}tdd|S)Nzlatin-1cSsdt|dS)Nz&#%d;r)ordgroup)mr W/home/ankuromar296_gmail_com/.local/lib/python3.7/site-packages/../../../bin/dumppdf.pyzescape..) isinstancebytesstrESC_PATsub)rusr r r!escapes  r*)outobjcodecrcCs|dkr|ddSt|tr|dt|x@|D]4\}}|d||dt|||dqz z%s zz zz  zz%srawbinaryz z textz%s z zz%sz%sz%s)writer$dictlenitemsdumpxmllistr&r%r*rZ get_rawdataget_dataattrsrobjidrnamerr TypeError)r+r,r-kvdatar r r!r6!s`                     r6F)r+docshow_fallback_xrefrcCsnx>|jD]4}t|tr|r|dt|||dqWtdd|jD}|rj|sjd}t|dS)Nz z css|]}t|tVqdS)N)r$r).0xrefr r r! jszdumptrailers..zThis PDF does not have an xref. Use --show-fallback-xref if you want to display the content of a fallback xref that contains all objects.) xrefsr$rr2r6Z get_trailerallloggerwarning)r+r@rArCZno_xrefsmsgr r r! dumptrailersbs   rJ)r+r@r-rArc Cst}|dx|jD]}x|D]}||kr4q&||y>||}|dkrTw&|d|t|||d|dWq&tk r}ztd|Wdd}~XYq&Xq&WqWt ||||ddS)Nzz )r-z z not found: %rz) setr2rE get_objidsaddgetobjr6rprintrJ) r+r@r-rAvisitedrCr:r,er r r! dumpallobjsus&    &  rR) outfpfnameobjidspagenospassworddumpallr- extractdirrcst|d}t|} t| |ddttdD} ttdfdd } y } | dx| D]\} }}}}d}|r| |}| |d j }nP|r|}t |t r| d }|rt|d kr| d r| |d }| |d j }t|}| d | ||dk r&| dt||| d|dk r>| d|| dqfW| dWntk rnYnX| |dS)NrbcSsi|]\}}||jqSr )Zpageid)rBpagenopager r r! szdumpoutline..)destrcs`t|ttfrt|}nt|tr8t|j}t|trJ|d}t|tr\| }|S)ND) r$r&r%rZget_destrr;r3rresolve)r`)r@r r! resolve_dests   z!dumpoutline..resolve_destz rSz/'GoTo'raz" zz z%r z z )openrr enumerater create_pagesobjectrZ get_outlinesr2r:r$r3getreprr*formatr6r close)rTrUrVrWrXrYr-rZfpparserZpagesrcZoutlinesleveltitler`aser\actionsubtyperr )r@r! dumpoutlinesH            ruZFilespecZ EmbeddedFile)rUrXrZrc sttttfddfdd }t|d~}t|}t||t}x^jD]T}xN| D]B} |} ||krZt | t rZ| dtkrZ||||| qZWqLWWdQRXdS)N)r:r,rcstj|dp"tt|d}|ddp@|dd}|j}t |t shd|}t ||dt k rt d|tj d||f}tj|rtd|td |tjtj|d d t|d }|||dS) NZUFFZEFz:unable to process PDF: reference for %r is not a PDFStreamTypez>unable to process PDF: reference for %r is not an EmbeddedFilez%.6d-%szfile exists: %rzextracting: %rT)exist_okwb)ospathbasenamerir r%decoderNr:r$rrLITERAL_EMBEDDEDFILEjoinexistsIOErrorrOmakedirsdirnamerer2r8rl)r:r,filenameZfilereffileobj error_msgr{r+)r@rZr r!extract1s(&      z!extractembedded..extract1r[rw)intrr&rrerr rKrErLrNr$r3riLITERAL_FILESPECrM) rUrXrZrrmrnZextracted_objidsrCr:r,r )r@rZr!extractembeddeds       r) rTrUrVrWrXrYr-rZrArc Cst|d} t| } t| |} |rFx$|D]} | | } t|| |dq&W|rxZtt| D]H\}}||krZ|rx2|jD]} t | } t|| |dqvWqZt||j qZW|rt || |||s|s|st || || |dkr|ddS)Nr[)r-)r/r0r.)rerr rNr6rfrrgcontentsrr9rRrJrlr2)rTrUrVrWrXrYr-rZrArmrnr@r:r,r\r]r r r!dumppdfs.        r)rcCsdttdd}|jdtdddd|jddd d tjd |jd d dddd|}|jdddddd|jddtdd|jddd}|jdt dddd|jddtd d|jd!d"td#d|jd$d%ddd&d|jd'dd(d)|jd*d+td,d-d.|jd/d0d}|jd1d2td3d4d.|}|jd5d6ddd7d|jd8d9ddd:d|jd;dNT) descriptionadd_helpfiles+zOne or more paths to PDF files.)typedefaultnargshelpz --versionz-vversionzpdfminer.six v{})rsrz--debugz-dF store_truezUse debug logging level.)rrsrz --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.)rsrz --passwordz-PrSz,The password to use for decrypting PDF file.)rrrOutputzUsed during output generation.z --outfilez-o-zJPath 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&rkpdfminer __version__add_mutually_exclusive_groupadd_argument_groupr)rnZprocedure_parserZ parse_paramsZ output_paramsZ codec_parserr r r! create_parser!s   r)argvrc CsJt}|j|d}|jr(ttj|jdkr:tj }n t |jd}|j rddd|j dD}ng}|j rdd|j D}n$|jrd d|j dD}nt}|j}|jrd }n|jrd }n|jrd }nd}xj|jD]`}|jrt||||||j|dd q|jrt|||jdqt||||||j|d|jd qW|dS)N)argsrwcSsg|] }t|qSr )r)rBxr r r! szmain..,cSsh|] }|dqS)r_r )rBrr r r! szmain..cSsh|]}t|dqS)r_)r)rBrr r r!rsr/r0r1)rXrYr-rZ)rXrZ)rXrYr-rZrA)r parse_argsdebuglogging getLoggersetLevelDEBUGoutfilesysstdoutreobjectssplitZ page_numbersrWrKrXZ raw_streamZ binary_streamZ text_streamrZ extract_tocrurFZextract_embeddedrrrArl) rrnrrTrVrWrXr-rUr r r!mains^     r__main__)N)F)NF)rSFNN)rSFNNF)N)>rros.pathrzrertypingrrrrrrrr r argparser rZpdfminer.pdfdocumentr r rZpdfminer.pdfpagerZpdfminer.pdfparserrZpdfminer.pdftypesrrrrrrZpdfminer.psparserrrrZpdfminer.utilsr basicConfigr__name__rGcompiler'r&r%r*rhr6boolrJrRrrurr~rrrrr r r r!sT,      B40$vA