o 0׾gB`0@sdZddlZddlZddlZddlZddlZddlZddlmZmZm Z ddl m Z m Z m Z mZmZmZddlmZddlmZddlmZmZmZddlmZmZmZmZdd lmZm Z m!Z!m"Z"d d gZ#d gZ$d ddddddddddddedddddddfde e%efde&de e%efde&de&de%dede&de%de%de'de'de'de'de d e e%efd!e'd"e e e&fd#e'd$e'd%e&d&e'd'eejf.d(d)Z(d ddddddddddddedddddddfd*e)de&de e%efde&de&de%dede&de%de%de'de'de'de'de e%efd e e%efd!e'd"e e e&fd#e'd$e'd%e&d&e'd'eejf.d+d,Z*d-ede%de&de&de%dede%de%de%de'de'de'd!e'd"e e&e e&e&ffd&e'd'ee%f d.d/Z+dCde%d!e'd'e e%e%ee'ffd0d1Z,ded'e%fd2d3Z-dDd4e%d e%d'e%fd5d6Z. dEd4e%d e%d%e&d'e e&e&ffd7d8Z/       dFde%de%de%d e%d9e'd%e&de&de&d'efd:d;Z0       dFdZ1 dCde%de%d?e%d#e'd@e'd'eejf dAdBZ2dS)Gzq pdf2image is a light wrapper for the poppler-utils tools that can convert your PDFs into Pillow images. N)PopenPIPETimeoutExpired)AnyUnionTupleListDictCallable)PurePath)Image)uuid_generatorcounter_generatorThreadSafeGenerator)parse_buffer_to_pgmparse_buffer_to_ppmparse_buffer_to_jpegparse_buffer_to_png)PDFInfoNotInstalledErrorPDFPageCountErrorPDFSyntaxErrorPDFPopplerTimeoutErrorpngtiffPagesppmFpdf_pathdpi output_folder first_page last_pagefmtjpegopt thread_countuserpwownerpw use_cropboxstrict transparent single_file output_file poppler_path grayscalesize paths_onlyuse_pdftocairotimeouthide_annotationsreturnc/Csf|r|dkrd}t|tr|}t|tr|}t|tr#|}t||| |dd}t||\}}}}|p?|p?| o?|tv}t|rEdnd|d\}}|dkrV|dkrVd }|dkr`|d kr`d }t|tjsyt|t sy| rut |g}d }nt |}|d krd }|d us|d krd }|d us||kr|}||krgSzd }|d ur|rt }d }||d }||kr|}||}|} g}!t|D]}"t|}#||t|dk}$tdt||g|| | |$d |||#|| | | | |||}%|r|rtdtd|g|%}%ntd|g|%}%| |$} |t|dk8}tj}&|d ur'|d|&dd|&d<d }'tdkr|d||dvrM|rM|d t|g| rT|d |durb|tj|||durm|d |g|durx|d |g| r|d | dur |St| t rt | dkr| ddur|dtt | dgn |dtdg| ddur|dtt | dg|S|dtdg|St| t rt | dkr|dtt | dg|St| t st| t r|dtt | g|St d| d)Nz-cropboxz-hide-annotationsz-transp-f-l)pgmr-jpegjpgz-jpegoptz -singlefile-upw-opwz-grayrz -scale-to-xrz -scale-to-yz -scale-tozSize z is not a tuple or an integer)rarIextendrT_parse_jpegoptrWpathjoinrEtuplelenrRfloat ValueError)rwr r!r"r#r$r,r&r'r(r*r+r.r/r3rrrrSsV          rScCsx|}|ddkr|dd}|dvrddtdfS|dkr$ddtdfS|d vr*d S|d kr6|r6d d tdfSd d tdfS) Nr.rrrrFr)tifr)rrNTrr)lowerrrrr)r#r.rrrrHs       rHcCsJg}|D]\}}|durd}|durd}|d||qd|S)NTyFnz{}={},)itemsraformatr)r$partskvrrrrs rcommandcCs.tdkr |d}|durtj||}|S)Nr<z.exe)r[r\rWrr)rr-rrrrVs rVc Cst||dg}tj}|dur|d|dd|d<t||ttd}z |j|d\}}WntyC| |\}}t dwz | dd  d d  d d  d} t | d t | dfWSYdS)Nz-vr9r:r;r=r>r?rARun poppler poppler timeout.rBrC r rrr)r)rVrWrXrYrZrrrbrrcrrdsplitrR) rr-r2r=rzr{r|r}r~versionrrrrJs"   &rJrawdatesc Csztd||g}|dur|d|g|dur|d|g|r&|dg|r1|dt|g|r<|dt|gtj} |durQ|d| d d | d <t|| ttd } z | j |d \} } Wnt yx| | \} }t d wi}| dddD])}|d}|dd|dd}}|d kr|tvrt|n|||<qd|vrt|WStytdtytd| ddw)aFunction wrapping poppler's pdfinfo utility and returns the result as a dictionary. :param pdf_path: Path to the PDF that you want to convert :type pdf_path: str :param userpw: PDF's password, defaults to None :type userpw: str, optional :param ownerpw: PDF's owner password, defaults to None :type ownerpw: str, optional :param poppler_path: Path to look for poppler binaries, defaults to None :type poppler_path: Union[str, PurePath], optional :param rawdates: Return the undecoded data strings, defaults to False :type rawdates: bool, optional :param timeout: Raise PDFPopplerTimeoutError after the given time, defaults to None :type timeout: int, optional :param first_page: First page to process, defaults to None :type first_page: int, optional :param last_page: Last page to process before stopping, defaults to None :type last_page: int, optional :raises PDFPopplerTimeoutError: Raised after the timeout for the image processing is exceeded :raises PDFInfoNotInstalledError: Raised if pdfinfo is not installed :raises PDFPageCountError: Raised if the output could not be parsed :return: Dictionary containing various information on the PDF :rtype: Dict pdfinfoNrrz -rawdatesrrr9r:r;rrArrBrCrrrrz;Unable to get page count. Is poppler installed and in PATH?zUnable to get page count. )rVrrTrWrXrYrZrrrbrrcrrdrrPDFINFO_CONVERT_TO_INTrRstriprOSErrorrr)rr&r'r-rr2r!r"rr=rzoutr|r}r~dfieldsfkeyvaluerrrrG sZ"        rG pdf_bytesc Cst\}} z5t| d} | || Wdn1s wYt| |||||||dWt|t| St|t| w)aFunction wrapping poppler's pdfinfo utility and returns the result as a dictionary. :param pdf_bytes: Bytes of the PDF that you want to convert :type pdf_bytes: bytes :param userpw: PDF's password, defaults to None :type userpw: str, optional :param ownerpw: PDF's owner password, defaults to None :type ownerpw: str, optional :param poppler_path: Path to look for poppler binaries, defaults to None :type poppler_path: Union[str, PurePath], optional :param rawdates: Return the undecoded data strings, defaults to False :type rawdates: bool, optional :param timeout: Raise PDFPopplerTimeoutError after the given time, defaults to None :type timeout: int, optional :param first_page: First page to process, defaults to None :type first_page: int, optional :param last_page: Last page to process before stopping, defaults to None :type last_page: int, optional :return: Dictionary containing various information on the PDF :rtype: Dict rN)r&r'r-rr2r!r") rNrrrrrGrWrr) rr&r'r-rr2r!r"rrrrrrpdfinfo_from_byteshs(      rextrDc Cs|g}tt|D]2}||r;|dd|kr;|r&|tj||q |t tj|||r;|d q |S)Nrr) sortedrWlistdir startswithrrarrr rload)r r,rr0rDrxrrrrres re)F)N)NN)NNNFNNN)3__doc__rWr[rNrKrfr]rrrtypingrrrrr r pathlibr PILr pdf2image.generatorsr rrpdf2image.parsersrrrrpdf2image.exceptionsrrrrrIrrTrRboolrbytesrrSrHrrVrJrGrrerrrrs             s          o      $J     ] 8