NgI&#dZddlZddlZddlZddlmZmZmZmZm Z ddl Z ddl m Z ddlmZejdZdede efd Zgd dd d d dddddddd d fdeedede e dedededede eededededede ededed edef"d!Zdejfd"Zd#e eedejfd$Zd'd#e eedefd%Zed&krejedSdS)(zkA command line tool for extracting text and images from PDF and output it to plain text, html, xml or tags.N)Any ContainerIterableListOptional)LAParams)AnyIO))z.htmhtml)z.htmlr )z.xmlxml)z.tagtagxreturnc|dkrdS t|S#t$r(t jd|wxYw)Ndisabledzinvalid float value: {})lowerstripfloat ValueErrorargparseArgumentTypeErrorformat)r s P/var/www/html/ai-engine/env/lib/python3.11/site-packages/../../../bin/pdf2txt.pyfloat_or_disabledrstwwyyJ&&tNQxx NNN()B)I)I!)L)LMMMNs =2A/-textutf-8F?normalfilesoutfilelaparams output_typecodec strip_controlmaxpages page_numberspasswordscalerotation layoutmode output_dirdebugdisable_cachingkwargsc |std|dkr*|dkr$tD]\}}||r|}|dkr tj}tjjd}nt |d}|D]M}t |d5}tjj |fitdddn #1swxYwYN|S)Nz Must provide files to work upon!rrrwbrb) r OUTPUT_TYPESendswithsysstdoutencodingopenpdfminer high_levelextract_text_to_fplocals)r r!r"r#r$r%r&r'r(r)r*r+r,r-r.r/overridealttypeoutfpfnamefps r extract_textrBs7$ =;<<<fC!- & & Hg)) &% #~~z :  *EWd##CC %   C"   22 B B B B B C C C C C C C C C C C C C C C Ls $B<<C C ctjtd}|dtddd|ddd d t j |d d ddd|ddddd|dd}|dtddd|ddtd|ddtd d!"|d#d$td%d&"|d'd(d td)*t}|d+d,}|d-d.ddd/|d0d1|j dd2|d3t|j d4"|d5d6t|jd7"|d8d9t|jd:"|d;dd?t"|jd@"|dAdB|jddC|dDdE}|dFdGtdHdI"|dJdKtdLdM"|dNdOtdPdQ"|dRdSddTU|dVdWdXtdY*|dZd[td\d]"|d^d_ddd`|S)aNT) descriptionadd_helpr +zOne or more paths to PDF files.)typedefaultnargshelpz --versionz-vversionzpdfminer.six v{})actionrKz--debugz-dF store_truezUse debug logging level.)rHrLrJz--disable-cachingz-Cz;If caching or resources, such as fonts, should be disabled.ParserzUsed during PDF parsing)rDz--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.)rGrJz --maxpagesz-mrz%The maximum number of pages to parse.)rGrHrJz --passwordz-Prz,The password to use for decrypting PDF file.z --rotationz-RzIThe number of degrees to rotate the PDF before other types of processing.)rHrGrJzLayout analysiszUsed during layout analysis.z --no-laparamsz-nz0If layout analysis parameters should be ignored.z--detect-verticalz-Vz??     '    J ,,7-L  ?  -   4   ;   ,  I..'E/N  ?   ) K   &=   %<   %?     %,   $"     # I  -->.M /   >   3  $   9     .  -  Margsc t|}|jrd|_n>t |j|j|j|j|j |j |j |_|j rd|j D|_ |j r)d|j dD|_ |jdkr9|jdkr.t"D]&\}}|j|r||_'|S)N)re)rXrYr[rZr\rWr]ch|]}|dz S.0r s r zparse_args..(s#L#L#LaAE#L#L#Lrdc2h|]}t|dz Srh)rVrks rrmzparse_args..+s"#W#W#W1CFFQJ#W#W#Wrd,rr)rc parse_args no_laparamsr"rrXrYr[rZr\rWr]r'pagenossplitr#r!r3r4)re parsed_argsr=r>s rrprps'//,,$,77K  # '$1#/#/#/"-'7!+   M#L#L;3K#L#L#L  X#W#W 8K8Q8QRU8V8V#W#W#W  &(([-@C-G-G!- 2 2 Hg"++H55 2*1 ' rdc~t|}tdit|}|dS)Nrrj)rprBvarsclose)rertr?s rmainrx5s<T""K  - -4 ,, - -E KKMMM 1rd__main__)N)rQrloggingr5typingrrrrrpdfminer.high_levelr9pdfminer.layoutrpdfminer.utilsr basicConfigr3rSrrboolrVrBrPrc Namespacerprx__name__exitrjrdrrs// ;;;;;;;;;;;;;;$$$$$$ V NN%NNNN#'-1 $!$$ C=$ $x $ $  $  $$9S>*$$ $$$ $ $$ !$" #$$$$NQx.QQQQhXd3i(X-?>  xS " c     z CHTTVVrd