U  3g– ã@sddlZddlZddlZddlZddlZddlmZmZmZddl m Z dd„Z dd„Z d5d d „Z d d„Zdd„Zd6dd„Zdd„Zdd„Zdd„Zdd„Zdd„Zdd„Zd d!„Zd"d#„Zd$d%„Zd&d'„Zd(d)„Zd*d+„Zd,d-„Zd.d/„Zd0d1„Zd2d3„Z e!d4kr e ƒdS)7éN)ÚDictÚListÚSeté)ÚpymupdfcCsd| dd¡S)Nz %s éKú-)Úcenter)Úx©r ú4/tmp/pip-unpacked-wheel-8jfb_caj/pymupdf/__main__.pyÚmycentersr cCsÂ|d}|d}|dkr"| |¡Sdd„}t ||¡}t ||¡}|j|jkrt|j|jkrfdkrtnn |jdksœt d||f¡t |¡d}||ƒSt |¡}| |j¡d}}||ƒS)zReturn image for a given XREF.rrcSs"|jjdkr|St tj|¡}|S)Né)Ú colorspaceÚnrÚPixmapÚcsRGB)ÚpixZtpixr r r Úgetimages zrecoverpix..getimagez&Warning: unsupported /SMask %i for %i:N) Z extract_imagerrZirectÚalpharÚmessageZ set_alphaZsamples)ÚdocÚitemr ÚsrZpix1Úpix2rr r r Ú recoverpixs    2   rFTcCs„t |¡}|js"|dkr"t d¡d}|js0|S|rp| |¡}|sLt d¡|dkr€t |dkrhddnd¡nt d |j¡|S) z!Open and authenticate a document.Tz$this command supports PDF files onlyéÿÿÿÿzauthentication unsuccessfulézauthenticated as %sÚownerÚuserz'%s' requires a password) rÚopenZis_pdfÚsysÚexitZ needs_passZ authenticaterÚname)ÚfilenameÚpasswordÚshowÚpdfrÚrcr r r Ú open_file9s    r)cCsLtdd„| ¡Dƒƒd}| ¡D]$\}}d| |¡|f}t |¡q"dS)zPrint a Python dictionary.cSsg|] }t|ƒ‘qSr )Úlen)Ú.0Úkr r r Ú Nszprint_dict..rz%s: %sN)ÚmaxÚkeysÚitemsÚrjustrr)rÚlr,ÚvÚmsgr r r Ú print_dictLsr5cCsœt d|¡| |¡}t |¡| |¡rŽ| ¡}z(| d¡d}||}| d¡rZd}Wntk rtd}YnXt d|¡t d¡t d¡d S) zPrint an object given by XREF number. Simulate the PDF source in "pretty" format. For a stream also print its size. z%i 0 objz/Lengthrz0 RÚunknownzstream ...%s bytesZ endstreamZendobjN)rrZ xref_objectZxref_is_streamÚsplitÚindexÚendswithÚ Exception)rÚxrefZxref_strÚtempÚidxÚsizer r r Ú print_xrefTs      r?Úpagec Cszt|dƒ}| d|¡ dd¡}| d¡}g}t|ƒD]<\}}|d}| ¡r”t|ƒ} d| krl|kr€nn| t|ƒ¡q6t d||f¡q6z"| d¡\} } t| ƒ} t| ƒ} Wn&t k rÜt d||f¡YnXd| krô|krnnd| kr|ks$nt d||f¡| | kr:| | ¡q6| | kr\|t t | | dƒƒ7}q6|t t | | dd ƒƒ7}q6|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. rÚNú Úú,zbad %s specification at item %irz%bad %s range specification at item %ir) ÚstrÚreplacer7Ú enumerateÚ isdecimalÚintÚappendr!r"r:ÚlistÚrange) ZrlistÚlimitÚwhatrAZ rlist_arrZout_listÚseqrrÚiÚi1Úi2r r r Úget_listks6   4   rSc Cst|j|jdƒ}tj |j¡d}d}|dkr:|d}d}t|dƒ}|j}t  d|j|j |  ¡d|||d|d f¡|j }|d kr®|  ¡}t  d ||d kr¤d ndf¡| ¡}|d krÌt  d|¡t  ¡|jrt  tdƒ¡| ¡}t||ƒt  ¡|jr,t  tdƒ¡t|jƒt  ¡|jrtt  tdƒ¡t|j|  ¡dd}|D]}t||ƒt  ¡qZ|jrÚt  tdƒ¡t|j|j dƒ} | D]8} | d}| |¡}t  d| ¡t||ƒt  ¡q |jrt  tdƒ¡t  | ¡¡t  ¡| ¡dS)NTiZKBièZMBrz7'%s', pages: %i, objects: %i, %g %s, %s, encryption: %sÚformatÚ encryptionrz5document contains %i root form fields and is %ssignedéznot rCz#document contains %i embedded filesz PDF catalogz PDF metadatazobject informationr;)rNzpage informationzPage %i:z PDF trailer)r)Úinputr%ÚosÚpathÚgetsizeÚroundÚmetadatarrÚ page_countZ xref_lengthZ is_form_pdfZ get_sigflagsZ embfile_countÚcatalogr Z pdf_catalogr?r5ZxrefsrSÚpagesZ page_xrefÚtrailerZ pdf_trailerÚclose) Úargsrr>ÚflagÚmetarrr;ZxreflÚpagelÚpnor r r r&šsz  ùÿÿ ÿÿ       r&c Csät|j|jdd}|j}d |¡}|js`|j|j|j|j |j |j |j |j ||j|j|jd dSt|j|jdƒ}t ¡}|D]}|d}|j|||dq~|j|j|j|j |j |j |j |j ||j|j|jd | ¡| ¡dS)NT©r'©ZkeepÚnonezrc4-40zrc4-128zaes-128zaes-256) ÚgarbageÚdeflateÚprettyÚcleanÚasciiÚlinearrUZowner_pwZuser_pwÚ permissionsr©Z from_pageZto_page)r)rWr%rUr8r_ÚsaveÚoutputrjÚcompressrlÚsanitizernrorrZ permissionrSr]rr Ú insert_pdfra)rbrrUZencryptr_Zoutdocrfrr r r rmÙsRÿõ õ rmc CsÚ|j}t ¡}|D]¨}| d¡}t|ƒdkr4|dnd}t|d|dd}d |dd…¡}|r€td |dd…¡|jdƒ}nt d|jdƒ}|D]} |j || d| ddq”|  ¡q|j |j d dd |  ¡dS) z&Join pages from several PDF documents.rDrNrTrgrrqr)rjrk)rWrr r7r*r)ÚjoinrSr]rLrvrarrrs) rbZdoc_listrZsrc_itemZsrc_listr%Úsrcr_Z page_listrPr r r Údoc_joins   ryc CsLt|j|jdd}| ¡s6|jr,|j|jkr6t d¡t|j|jƒ}|j rTt |j ƒnt ƒ}t |  ¡ƒ}|r~||ks‚t d¡n|}|st d¡|t |  ¡ƒ@}|r¶t dt |ƒ¡|D]L}|  |¡}| |¡}|j|||d|d|d d t d ||j f¡qº| ¡|jr8|j|jkr8|j|jd d n| ¡| ¡dS)z!Copy embedded files between PDFs.Trgúcannot save PDF incrementallyz%not all names are contained in sourceznothing to copyz2following names already exist in receiving PDF: %sr$Ú ufilenameÚdesc©r$r{r|zcopied entry '%s' from '%s'rV©rjN)r)rWr%Úcan_save_incrementallyrsr!r"ÚsourceZ pwdsourcer#ÚsetÚ embfile_namesrEÚ embfile_infoÚ embfile_getÚ embfile_addrrrarrÚsaveIncr) rbrrxÚnamesZ src_namesZ intersectrÚinfoZbuffr r r Ú embedded_copysFÿ ÿ      ûr‰c CsÔt|j|jdd}| ¡s6|jr,|j|jkr6t d¡ttj j f}tj dkrPt}z|  |j ¡Wn:|k rš}zt d|j ›d|›¡W5d}~XYnX|jr®|j|jkr¸| ¡n|j|jdd | ¡dS) zDelete an embedded file entry.Trgrz©réúno such embedded file ú: Nrr~)r)rWr%rrsr!r"Ú ValueErrorrÚmupdfÚ FzErrorBaseÚmupdf_version_tupleÚ embfile_delr#r†rrra)rbrÚexception_typesÚer r r Ú embedded_delEs$ÿ ÿ   * r•c CsØt|j|jdd}ttjjf}tjdkr,t}z| |j ¡}|  |j ¡}Wn:|k r‚}zt   d|j ›d|›¡W5d}~XYnX|j r|j n|d}t|dƒ}| |¡W5QRXt d |j |f¡| ¡dS) z&Retrieve contents of an embedded file.TrgrŠrŒrNr$Úwbzsaved entry '%s' as '%s')r)rWr%rŽrrrr‘r„r#rƒr!r"rsr Úwriterra)rbrr“ÚstreamÚdr”r$rsr r r Ú embedded_get[s   * ršc Cs*t|j|jdd}| ¡s:|jdks0|j|jkr:t d¡z | |j¡t d|j¡Wnt k rnYnXt j   |j ¡rŒt j   |j ¡sœt d|j ¡t|j dƒ}| ¡}W5QRX|j }|}|jsÒ|}n|j}|j|j||||d|jr|j|jkr| ¡n|j|jd d | ¡dS) zInsert a new embedded file.TrgNrzzentry '%s' already existszno such file '%s'Úrbr}rVr~)r)rWr%rrsr!r"r’r#r:rXrYÚexistsÚisfiler Úreadr|r…r†rrra©rbrÚfr˜r$r{r|r r r Ú embedded_addms@ÿ ÿ  ÿ r¡c CsVt|j|jdd}| ¡s:|jdks0|j|jkr:t d¡z| |j¡Wn$t k rnt d|j¡YnX|j dk r¸t j   |j ¡r¸t j   |j ¡r¸t|j dƒ}| ¡}W5QRXnd}|jrÊ|j}nd}|jrÜ|j}n|jrê|j}nd}|jrü|j}nd}|j|j||||d|jdks0|j|jkr:| ¡n|j|jdd | ¡dS) z0Update contents or metadata of an embedded file.TrgNrzúno such embedded file '%s'r›r}rVr~)r)rWr%rrsr!r"rƒr#r:rYrXrœrr ržr$r{r|Z embfile_updr†rrrarŸr r r Ú embedded_updsTÿ ÿ ÿ þ ýÿ r£cCs0t|j|jdd}| ¡}|jdk r’|j|kr@t d|j¡nRt ¡t dt |ƒt |ƒdkrddndf¡t ¡t |  |j¡ƒt ¡dS|sªt d |j¡dSt |ƒdkrÊd |jt |ƒf}n d |j}t |¡t ¡|D]8}|j st |¡qê|  |¡}t |  |¡ƒt ¡qê|  ¡dS) zList embedded files.TrgNr¢z!printing 1 of %i embedded file%s:rrrCz'%s' contains no embedded filesz-'%s' contains the following %i embedded filesz)'%s' contains the following embedded file)r)rWr%r‚r#r!r"rrr*r5rƒZdetailra)rbrr‡r4r#Ú_r r r Ú embedded_listÁs>  ÿÿ      r¥c Csr|js|jst d¡t|j|jdd}|jrBt|j|j dƒ}nt d|j dƒ}|j sht j  t j¡}n,|j }t j  |¡r†t j  |¡s”t d|¡tƒ}tƒ}|D]‚}|jrP| |d¡}|D]Š}|d} | |krÄ| | ¡| | ¡\} } } } | dksÄ| sqÄt j  ||  dd ¡›d | ›d | ›¡}t|d ƒ}| | ¡W5QRXd } qÄ|jr¤| |d¡}|D]¾}|d} | |krh| | ¡t||ƒ}t|ƒtkrè|d } |d}t j  |d| | f¡}t|d ƒ}| |¡W5QRXn:óz page_blocksort..©Úkeyrr¸r¹rº)r¼r½r—Úsortr¾) r@r¿rÀrÁrÂrÃr·rÄrÆrÇr r r Úpage_blocksort3s  rÍc!s>|rdntdgƒ}ttttdœdd„}tttdœdd„} tttjd œ‡‡fd d „ } ttd œd d„‰dd„} |jd|dd} | | |ƒ\} }}}}| gkr°|s¬|  |¡dS| ||ƒ}| j dd„di}| D]8}|\}}}}|||ƒ}|  |g¡}|  |¡|||<qÒt | ¡ƒ}|  ¡||}i}|D]f}||}t|ƒ}|dkrZd||<q0dd„|Dƒ}|  ¡t |¡}||krˆ|}|d||<q0||d|d|t|ƒd}|d}|  d¡|D]\}||krö|  d¡||7}qÖ| ||||||ƒ} |  | djdd d!¡||}qÒ|  |¡dS)"Nr³r´)ÚvaluesÚvalueÚreturncSs0t ||¡}|r||dStd||fƒ‚dS)zôFind 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 %sN)ÚbisectÚ bisect_rightÚ RuntimeError)rÎrÏrPr r r Úfind_line_indexEs  z$page_layout..find_line_index)ÚrowsrÐcSsJt|ƒ}| ¡|dg}|dd…D]}||d|kr&| |¡q&|S)Nrrr)rKrÌrJ)rÕrÀZnrowsÚhr r r Ú curate_rowsTs  z page_layout..curate_rows)rÆr@c sÈtƒ}|jj}|jj}|}|}d}g}|D]Œ} | dD]|} | ddkrLq8| d\} } } }|dks8| |jjkrrq8|| }||kr†|}| dD]$}|dˆkr¢qŽ|dD]}|d\} }} }| | }|d \}}tt|ƒƒ}| |¡|d }||kr |d kr |}|| kr| }|dkr |gkr |d \}}}}||kr |td ƒkrbˆ||ƒ}n,|dkrvtdƒ}n|dkrŠtdƒ}n|}||||f|d <qª| ||||f¡qªqŽq8q*|||||fS)NrÚlinesÚdir)rrZbboxZspansr>ÚcharsÚoriginÚcrBréûrPéûr2éû) rZrectÚwidthÚheightrIr[r¬ÚchrrJ)rÆr@rÕZ page_widthZ page_heightÚ rowheightÚleftÚrightrÚÚblockÚlineZx0Zy0Úx1Úy1ráÚspanrÜr¤ÚcwidthÚoxÚoyÚchZold_chÚold_oxZold_oyZ old_cwidthÚlig©rÁÚ joinligaturer r Úprocess_blocks]sZ            z#page_layout..process_blocks)rðrÐcSst|dkrtdƒS|dkr tdƒS|dkr0tdƒS|dkr@tdƒS|d krPtd ƒS|d kr`td ƒS|d krptdƒS|S)zËReturn 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) ÚffrÝÚfiiûÚfliûÚffirÞÚfflrßÚftiûÚstiû)râ)rðr r r rò‘s z!page_layout..joinligaturecSsôd}d}d}d}|tjkr&td|ƒ‚|D]À}|\} } } } | |} | | } || krd| || dkrdq*| dkr~|| | dkr~q*| }| ||kr || 7}| }| }q*| dkrªq*t| |ƒt|ƒ}| |krÚ|dkrÚ|d|7}|| 7}| }| }q*| ¡S)a†Produce 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 rCrz%program error: minslot too small = %ggš™™™™™É?rBgš™™™™™é?r)rZEPSILONrÓrIr*Úrstrip)räÚslotZminslotÚlcharsrµZold_charZold_x1rïrÜÚcharrìr¤rërèÚdeltar r r Ú make_textline«s:      z"page_layout..make_textlineZrawdictr¶rÆcSs|dS)Nrr )rÜr r r rÈírÉzpage_layout..rÊrrcSsg|] }|d‘qS)rVr )r+rÜr r r r- szpage_layout..rrg333333ó?Ú r¸r¹rº)r¼rrIrrrZPagerEr½r—rÌÚgetrJrKr/r*Ú statisticsZmedianr¾)!r@r¿rÀrÁrÂrÃr·rÄrÔr×rórrÆrÚrÕrärårãrØrÜr¤ríÚyrýr/rüZminslotsr,ZccountZwidthsZ this_slotZrowposrµr rñr Ú page_layoutAs` 46            $     rc Csèt|j|jdd}t|j|jdƒ}|j}|dkrLtj  |j ¡\}}|d}t |dƒˆ}t j t jB}|jrt|t jN}|jr„|t j N}|jr”|t jN}tttdœ}|D]4} || d} ||j| ||j|j|j|j|dq¤W5QRXdS)NFrgrz.txtr–©ÚsimplerÆÚlayoutr¶)r)rWr%rSr_r]rsrXrYÚsplitextr#r rZTEXT_PRESERVE_LIGATURESZTEXT_PRESERVE_WHITESPACEZ convert_whiteZ noligaturesZ extra_spacesZTEXT_INHIBIT_SPACESrÅrÍrÚmodeZgridrÁrÂrÃ) rbrrersr$r¤r¿r·Úfuncrfr@r r r Úgettext"s:     ý ùr cCst d¡t d¡dS)NzThis is from PyMuPDF message().zThis is from PyMuPDF log().)rrÚlog)rbr r r Ú _internalCs rcCs8tjdtdƒd}|jddd}|jdtdƒd }|jd td d |jd dd|jdddd|jdddd|jdddd|jdtdd |jdtdd |jtd|jdtdƒd }|jd td d |jdtd d |jd dd|jd!d"d#d$d%|jd&td'd |jd(td)d |jd*t d+t d,ƒd-d.|jd/dd0d1d2|jd3dd0d4d2|jd5dd0d6d2|jd7t d8d9d:|jd;dd0dd2|jdd?d|jt d|jd@tdAƒdBdC}|jd dDdEdF|jdGdHdIdJ|jt d|jdKtdLƒd }|jd td d |jdMddNd|jdOddPd|jdGdQd|jd dd|jdtdRd |jt d|jdStdTƒd }|jd d d|jdUdVd|jdWddXd|jd dd|jtd|jdYtdZƒd }|jd d d|jd dd|jdGd[d|jdUdHd\dJ|jd]dHd^dJ|jd_d`d|jtd|jdatdbƒd }|jd d d|jd dd|jdGd[d|jdUdHdcdJ|jtd|jddtdeƒdfdC}|jd d d|jdUdHdgdJ|jd dd|jdGdhd|jd]did|jdjdkd|jdldmd|jd_dnd|jtd|jdotdpƒd } | jd td d | jdUdHdgdJ| jd dd| jdGdqd| jtd|jdrtdsƒd } | jd tdtd | jd dud| jdGdvd| jdwdHdxdJ| jdydzd| jdUdDd{dF| jtd|jd|td}ƒd } | jd td~d | jd dd| jd€tdd‚dƒd.| jdtd„d…d†| jd‡ddˆd0d‰| jdŠdd‹d0d‰| jdŒddd0d‰| jdŽddd0d‰| jddd‘d0d‰| jdGd’d| jd“td”d•d†| jd–td—d˜d†| jtd|jd™tdšƒd } | jtd| ¡} t| d›ƒs*| ¡n |  | ¡dœS)zDefine command configurations.rzBasic PyMuPDF Functions)ÚprogÚ descriptionZ Subcommandsz/Enter 'command -h' for subcommand specific help)ÚtitleÚhelpr&zdisplay PDF information)rrWz PDF filename)r­rz -passwordr%)rz-catalogÚ store_truezshow PDF catalog)Úactionrz-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)r rmz.optimize PDF, or create sub-PDF if pages givenrszoutput PDF filenamez -encryptionzencryption methodrhri)rÚchoicesÚdefaultz-ownerzowner passwordz-userz user passwordz-garbagezgarbage collection levelér)r­rrrz -compressFzcompress (deflate) output)rrrz-asciizASCII encode binary dataz-linearzformat for fast web displayz -permissionrzinteger with permission levels)r­rrz -sanitizezsanitize / clean contentsz-prettyzprettify PDF structurez/output selected pages pages, format: 1,5-7,50-Nrwzjoin PDF documentsz3specify each input as 'filename[,password[,pages]]')rÚepilogÚ*zinput filenames)Únargsrz-outputTzoutput filename)ÚrequiredrÚextractz 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)rrz select pages, format: 1,5-7,50-Nz1-N)r­rrz -noligaturesz*expand ligature characters (default False))rrrz-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)rz -fontsizez4only include text with a larger fontsize (default 3)rVZinternalzinternal testingr N)ÚargparseÚArgumentParserr Úadd_subparsersÚ add_parserÚ add_argumentrEÚ set_defaultsr&rIrLrmryr²r¥r¡r•r£ršr‰Úfloatr rÚ parse_argsÚhasattrÚ print_helpr )ÚparserZsubpsZps_showZps_cleanZps_joinZ ps_extractZ ps_embed_addZ ps_embed_delZ ps_embed_updZps_embed_extractZ ps_embed_copyZ ps_gettextZ ps_internalrbr r r ÚmainGs,þÿÿÿ ÿüûüÿüÿüÿÿ ý ÿÿÿ ÿ ÿÿ ÿÿ ýÿÿ ÿÿ ÿÿÿÿ ÿûüüüüüüþüü ÿ   r(Ú__main__)FT)r@)"rrÑrXr!rÚtypingrrrrCrr rr)r5r?rSr&rmryr‰r•ršr¡r£r¥r²rÅrÍrr rr(Ú__name__r r r r ÚsB $  /?/("2$B b!4