o Cji;@sddlZddlZddlZddlZddlZddlmZmZmZm Z ddl Z ddl m Z m Z mZddZddZd2d d Zd d ZddZd3ddZddZddZddZddZddZddZdd Zd!d"Zd#d$Zd%d&Zd'd(Zd)d*Z d+d,Z!d-d.Z"d/d0Z#e$d1kre#dSdS)4N)DictListSetTuple)TEXT_INHIBIT_SPACESTEXT_PRESERVE_LIGATURESTEXT_PRESERVE_WHITESPACEcCsd|ddS)Nz %s K-)center)xr h/var/www/html/Resume-Parser/resume-parser-inhouse/venv/lib/python3.10/site-packages/fitz_new/__main__.pyrcCs|d}|d}|dkr||Sdd}t||}t||} |j|jkr;|j|jkr4dkr;nn|jdksMtd||ft|d}||St|}||jd}}||S)zReturn image for a given XREF.rcSs"|jjdkr|Sttj|}|S)N) colorspacenfitzPixmapcsRGB)pixtpixr r rgetimage s zrecoverpix..getimagez&Warning: unsupported /SMask %i for %i:N) extract_imagerrirectalpharprint set_alphasamples)docitemr srpix1pix2rr r r recoverpixs"   2  r&FTcCst|}|js|durtdd}|js|S|r8||}|s&td|dur6t|dkr3ddnd|Std |j|S) z!Open and authenticate a document.Tz$this command supports PDF files onlyzauthentication unsuccessfulzauthenticated as %sowneruserz'%s' requires a password) ropenis_pdfsysexit needs_pass authenticatername)filenamepasswordshowpdfr!rcr r r open_file=s    r7cCsJtdd|Dd}|D]\}}d|||f}t|qdS)zPrint a Python dictionary.cSsg|]}t|qSr )len).0kr r r Rzprint_dict..rz%s: %sN)maxkeysitemsrjustr)r"lr:vmsgr r r print_dictPs  rDcCstd|||}t|||r=|}z|dd}||}|dr+d}Wnd}Ytd|tdtdd S) zPrint an object given by XREF number. Simulate the PDF source in "pretty" format. For a stream also print its size. z%i 0 objz/Lengthrz0 Runknownzstream ...%s bytes endstreamendobjN)r xref_objectxref_is_streamsplitindexendswith)r!xrefxref_strtempidxsizer r r print_xrefYs      rRpagec Csbt|d}|d|dd}|d}g}t|D]\}}|d}|rIt|} d| kr5|kr?nn|t|n td||fqz|d\} } t| } t| } Wn td||fYd| krr|krnn d| kr~|ksntd||f| | kr|| q| | kr|t t | | d7}q|t t | | dd 7}q|S) aKTransform a page / xref specification into a list of integers. Args ---- rlist: (str) the specification limit: maximum number, i.e. number of pages, number of objects what: a string to be used in error messages Returns ------- A list of integers representing the specification. rN ,zbad %s specification at item %ir z%bad %s range specification at item %ir') strreplacerJ enumerate isdecimalintappendr-r.listrange) rlistlimitwhatrT rlist_arrout_listseqr"rii1i2r r rget_listps6   . ric Cst|j|jd}tj|jd}d}|dkr|d}d}t|d}|j}td|j|j | d|||d|d f|j }|d krU| }td ||d krPd ndf| }|d krctd|t|jr{ttd|}t||t|jrttdt|jt|jrttdt|j| dd}|D] }t||tq|jrttdt|j|j d} | D]} | d}||}td| t||tq|jrttdt|t|dS)NTiKBiMBrz7'%s', pages: %i, objects: %i, %g %s, %s, encryption: %sformat encryptionrz5document contains %i root form fields and is %ssignedznot rVz#document contains %i embedded filesz PDF catalogz PDF metadatazobject informationrM)rbzpage informationzPage %i:z PDF trailer)r7inputr3ospathgetsizeroundmetadatar page_count xref_length is_form_pdf get_sigflags embfile_countcatalogmycenter pdf_catalogrRrDxrefsripages page_xreftrailer pdf_trailerclose) argsr!rQflagmetarr#rMxreflpagelpnor r rr4sz                 r4c Cst|j|jdd}|j}d|}|js0|j|j|j|j |j |j |j |j ||j|j|jd dSt|j|jd}t}|D]}|d}|j|||dq?|j|j|j|j |j |j |j |j ||j|j|jd ||dS)NTr5keepnonezrc4-40zrc4-128zaes-128zaes-256) garbagedeflateprettycleanasciilinearrmowner_pwuser_pw permissionsr from_pageto_page)r7ror3rmrKr~saveoutputrcompressrsanitizerrr)r* permissionrirurr+ insert_pdfr)rr!rmencryptr~outdocrrr r rrsR  rc Cs|j}t}|D]T}|d}t|dkr|dnd}t|d|dd}d|dd}|r@td|dd|jd}nt d|jd}|D]} |j || d| ddqJ| q |j |j d dd | dS) z&Join pages from several PDF documents.rWrNrTrr(rr)rr)rorr+rJr8r7joinrirur_rrrr) rdoc_listr!src_itemsrc_listr3srcr~ page_listrfr r rdoc_join s    rc CsFt|j|jdd}|s|jr|j|jkrtdt|j|j}|j r*t |j nt }t | }|r?||ks>tdn|}|sHtd|t | @}|r[tdt ||D]%}| |}||}|j|||d|d|d d td ||j fq]||jr|j|jkr|j|jd d n||dS)z!Copy embedded files between PDFs.Trcannot save PDF incrementallyz%not all names are contained in sourceznothing to copyz2following names already exist in receiving PDF: %sr2 ufilenamedescr2rrzcopied entry '%s' from '%s'rnrN)r7ror3can_save_incrementallyrr-r.source pwdsourcer1set embfile_namesrX embfile_info embfile_get embfile_addrrrsaveIncr) rr!rnames src_names intersectr"infobuffr r r embedded_copy"sF        rcCst|j|jdd}|s|jr|j|jkrtdz||jWnt y4td|jYnw|jr>|j|jkrC| n|j |jdd| dS)zDelete an embedded file entry.Trrno such embedded file '%s'rrN) r7ror3rrr-r. embfile_delr1 ValueError save_incrrr)rr!r r r embedded_delJs     rcCst|j|jdd}z||j}||j}Wnty(td|jYnw|j r/|j n|d}t |d}| || t d|j|f| dS)z&Retrieve contents of an embedded file.Trrr2wbzsaved entry '%s' as '%s'N)r7ror3rr1rrr-r.rr+writerr)rr!streamdr2rr r r embedded_get]s     rcCs t|j|jdd}|s|jdus|j|jkrtdz||jtd|jWnYt j |j r@t j |j sHtd|j t |j d}|j }|}|js[|}n|j}|j|j||||d|jrr|j|jkrw|n|j|jd d |dS) zInsert a new embedded file.TrNrzentry '%s' already existszno such file '%s'rbrrnr)r7ror3rrr-r.rr1rprqexistsisfiler+readrrrrrrr!rr2rrr r r embedded_addms0     rcCs6t|j|jdd}|s|jdus|j|jkrtdz||jWn td|jY|j durNt j |j rNt j |j rNt |j d}nd}|jrW|j}nd}|jr`|j}n |jrg|j}nd}|jrp|j}nd}|j|j||||d|jdus|j|jkr|n|j|jdd |dS) z0Update contents or metadata of an embedded file.TrNrrrrrnr)r7ror3rrr-r.rr1rqrprrr+rr2rr embfile_updrrrrr r r embedded_updsB       rcCst|j|jdd}|}|jdurE|j|vr td|jn%ttdt|t|dkr0dndftt | |jtdS|sPtd |jdSt|dkr`d |jt|f}nd |j}t|t|D]}|j sxt|qn| |}t | |tqn| dS) zList embedded files.TrNrz!printing 1 of %i embedded file%s:rr#rVz'%s' contains no embedded filesz-'%s' contains the following %i embedded filesz)'%s' contains the following embedded file) r7ror3rr1r-r.rr8rDrdetailr)rr!rrCr1_r r r embedded_lists>      rc Cs\|js |js tdt|j|jdd}|jr!t|j|j d}nt d|j d}|j s4t j t j}n|j }t j |rCt j |sJtd|t}t}|D]}|jr||d}|D]B}|d} | |vr|| || \} } } } | dks~| sq`t j || dd d | d | }t|d }|| |d } q`|jr ||d}|D]Z}|d} | |vr || t||}t|tur|d } |d}t j |d| | f}t|d }|||qt j |d| }|jj dkr|nt!"t!j#|}|$|qqR|jrt%dt&||f|jr(t%dt&||f|d S)z)Extract images and / or fonts from a PDF.z"neither fonts nor images requestedTrrz"output directory %s does not existrzn/arUr .rNextimagez img-%i.%sz img-%i.pngrzsaved %i fonts to '%s'zsaved %i images to '%s')'fontsimagesr-r.r7ror3r~rirur_rrprqabspathcurdirrisdirrget_page_fontsadd extract_fontrrYr+rrget_page_imagesr&typedictrrrrrrrr8)rr!r~out_dir font_xrefs image_xrefsritemlistr"rMfontnamerrbufferoutnameoutfilerimgdatar%r r rextract_objectssr                 rc CsX|rdntdg}|jd|d}|s|s||dS||jddd||dS)N textflagsutf8 surrogatepasserrors)bytesget_textrencode) rStextoutGRIDfontsize noformfeed skip_emptyreoprr r r page_simple's  rc Csz|rdntdg}|jd|d}|gkr|s||dS|jddd|D]} || djd d d q'||dS) NrrblocksrcSs|d|dfS)Nrnrr )br r rr:rz page_blocksort..keyrrrr)rrrsortr) rSrrrrrrrrrr r rpage_blocksort3s  rc!s@|rdntdg}dttdtdtfdd}dttdtfd d } d ttd tjffd d } dtdtfdddd} |jd|dd } | | |\} }}}}| gkr]|s[| |dS| ||}| j dddi}| D]}|\}}}}|||}| |g}| ||||<qnt |}| ||}i}|D]/}||}t|}|dkrd||<qdd|D}| t|}||kr|}|d||<q||d|d|t|d}|d}| d|D].}||kr| d||7}||ks| ||||||} | | d jd!d"d#||}q| |dS)$NrrvaluesvaluereturncSs,t||}|r||dStd||f)zFind the right row coordinate. Args: values: (list) y-coordinates of rows. value: (int) lookup for this value (y-origin of char). Returns: y-ccordinate of appropriate line for value. rzLine for %g not found in %s)bisect bisect_right RuntimeError)rrrfr r rfind_line_indexEs  z$page_layout..find_line_indexrowscSsJt|}||dg}|ddD]}||d|kr"||q|S)Nrrr')r^rr])r rnrowshr r r curate_rowsTs  z page_layout..curate_rowsrrSc st}|jj}|jj}|}|}d}g}|D]} | dD]} | ddkr$q| d\} } } }|dks6| |jjkr7q|| }||krA|}| dD]}|dkrNqE|dD]z}|d\} }} }| | }|d \}}tt|}|||d }||kr|d kr|}|| kr| }|dkr|gkr|d \}}}}||kr|td kr||}n|dkrtd}n |dkrtd}n|}||||f|d <qR|||||fqRqEqq|||||fS)Nrlinesdir)rrbboxspansrQcharsorigincrUr'rfrA) rrectwidthheightr\rsrchrr])rrSr  page_width page_height rowheightleftrightrblocklinex0y0x1y1rspanrrcwidthoxoychold_chold_oxold_oy old_cwidthligr joinligaturer rprocess_blocks]s`           )z#page_layout..process_blocksr/cSst|dkrtdS|dkrtdS|dkrtdS|dkr tdS|d kr(td S|d kr0td S|d kr8tdS|S)zReturn ligature character for a given pair / triple of characters. Args: lig: (str) 2/3 characters, e.g. "ff" Returns: Ligature, e.g. "ff" -> chr(0xFB00) ffrfiifliffirfflrftisti)r)r/r r rr1s z!page_layout..joinligaturecSsd}d}d}d}|tjkrtd||D]`}|\} } } } | |} | | } || kr2| || dkr2q| dkr?|| | dkr?q| }| ||krP|| 7}| }| }q| dkrUqt| |t|}| |krm|dkrm|d|7}|| 7}| }| }q|S)aProduce the text of one output line. Args: left: (float) left most coordinate used on page slot: (float) avg width of one character in any font in use. minslot: (float) min width for the characters in this line. chars: (list[tuple]) characters of this line. Returns: text: (str) text string for this line rVrz%program error: minslot too small = %gg?rUg?r)rEPSILONrr\r8rstrip)rslotminslotlcharsrold_charold_x1r,rcharr(rr'r$deltar r r make_textlines:      z"page_layout..make_textlinerawdictrcSs|dS)Nrr )rr r rrszpage_layout..rr(rcSsg|]}|dqS)rnr )r9rr r rr; r<zpage_layout..rr'g333333? rrr)rrr\rrrPagerXrrrgetr]r^r>r8 statisticsmedianr)!rSrrrrrrrrr r2rCrrr rrrr rrr)yr>r>r<minslotsr:ccountwidths this_slotrowposrr r0r page_layoutAsb 46          $   rPc Cst|j|jdd}t|j|jd}|j}|dkr&tj |j \}}|d}t |d}t t B}|jr6|t N}|jr=|t N}|jrD|tN}tttd}|D]} || d} ||j| ||j|j|j|j|dqL|dS)NFrrz.txtrsimplerlayoutr)r7ror3rir~rurrprqsplitextr1r+rr convert_white noligatures extra_spacesrrrrPmodegridrrrr) rr!rrr2rrrfuncrrSr r rgettext"s<   r[c Cstjdtdd}|jddd}|jdtdd }|jd td d |jd dd|jdddd|jdddd|jdddd|jdtdd |jdtdd |jtd|jdtdd }|jd td d |jdtd d |jd dd|jd!d"d#d$d%|jd&td'd |jd(td)d |jd*t d+t d,d-d.|jd/dd0d1d2|jd3dd0d4d2|jd5dd0d6d2|jd7t d8d9d:|jd;dd0dd2|jdd?d|jt d|jd@tdAdBdC}|jd dDdEdF|jdGdHdIdJ|jt d|jdKtdLd }|jd td d |jdMddNd|jdOddPd|jdGdQd|jd dd|jdtdRd |jt d|jdStdTd }|jd d d|jdUdVd|jdWddXd|jd dd|jtd|jdYtdZd }|jd d d|jd dd|jdGd[d|jdUdHd\dJ|jd]dHd^dJ|jd_d`d|jtd|jdatdbd }|jd d d|jd dd|jdGd[d|jdUdHdcdJ|jtd|jddtdedfdC}|jd d d|jdUdHdgdJ|jd dd|jdGdhd|jd]did|jdjdkd|jdldmd|jd_dnd|jtd|jdotdpd } | jd td d | jdUdHdgdJ| jd dd| jdGdqd| jtd|jdrtdsd } | jd tdtd | jd dud| jdGdvd| jdwdHdxdJ| jdydzd| jdUdDd{dF| jtd|jd|td}d } | jd td~d | jd dd| jdtdddd.| jdtddd| jdddd0d| jdddd0d| jdddd0d| jdddd0d| jdddd0d| jdGdd| jdtddd| jdtddd| jtd|} t| ds|dS| | dS)zDefine command configurations.rzBasic PyMuPDF Functions)prog description Subcommandsz/Enter 'command -h' for subcommand specific help)titlehelpr4zdisplay PDF information)r]roz PDF filename)rr`z -passwordr3)r`z-catalog store_truezshow PDF catalog)actionr`z-trailerzshow PDF trailerz -metadatazshow PDF metadataz-xrefsz&show selected objects, format: 1,5-7,Nz-pagesz'show selected pages, format: 1,5-7,50-N)rZrz.optimize PDF, or create sub-PDF if pages givenrzoutput PDF filenamez -encryptionzencryption methodrr)r`choicesdefaultz-ownerzowner passwordz-userz user passwordz-garbagezgarbage collection levelr)rr`rcrdz -compressFzcompress (deflate) output)rbrdr`z-asciizASCII encode binary dataz-linearzformat for fast web displayz -permissionr'zinteger with permission levels)rrdr`z -sanitizezsanitize / clean contentsz-prettyzprettify PDF structurez/output selected pages pages, format: 1,5-7,50-Nrzjoin PDF documentsz3specify each input as 'filename[,password[,pages]]')r]epilog*zinput filenames)nargsr`z-outputTzoutput filename)requiredr`extractz extract images and fonts to diskz-imageszextract imagesz-fontsz extract fontsz-folder to receive output, defaults to currentz-consider these pages only, format: 1,5-7,50-Nz embed-infozlist embedded filesz-namezif given, report only this onez-detailzdetail informationz embed-addzadd embedded filez-output PDF filename, incremental save if nonezname of new entryz-pathzpath to data for new entryz-desczdescription of new entryz embed-delzdelete embedded filezname of entry to deletez embed-updzupdate embedded filez*except '-name' all parameters are optionalz name of entryz-Output PDF filename, incremental save if nonezpath to new data for entryz -filenameznew filename to store in entryz -ufilenamez&new unicode filename to store in entryz!new description to store in entryz embed-extractzextract embedded file to diskz'output filename, default is stored namez embed-copyz copy embedded files between PDFszPDF to receive embedded fileszpassword of inputz2output PDF, incremental save to 'input' if omittedz-sourcezcopy embedded files from herez -pwdsourcezpassword of 'source' PDFzrestrict copy to these entriesr[z(extract text in various formatting modeszinput document filenamezpassword for input documentz-modez-mode: simple, block sort, or layout (default)rQrSz select pages, format: 1,5-7,50-Nz1-N)rr`rdz -noligaturesz*expand ligature characters (default False))rbr`rdz-convert-whitez6convert whitespace characters to white (default False)z -extra-spacesz%fill gaps with spaces (default False)z -noformfeedz-write linefeeds, no formfeeds (default False)z -skip-emptyz+suppress pages with no text (default False)z3store text in this file (default inputfilename.txt)z-gridz+merge lines if closer than this (default 2)r(z -fontsizez4only include text with a larger fontsize (default 3)rnrZN)argparseArgumentParserr{add_subparsers add_parser add_argumentrX set_defaultsr4r\r_rrrrrrrrrfloatr[ parse_argshasattr print_helprZ) parsersubpsps_showps_cleanps_join ps_extract ps_embed_add ps_embed_del ps_embed_updps_embed_extract ps_embed_copy ps_gettextrr r rmainEs             r__main__)FT)rS)%rkrrpr-rHtypingrrrrr fitz.fitzrrrr{r&r7rDrRrir4rrrrrrrrrrrrPr[r__name__r r r rsF $ /?/(!1$D b#,