G/Ph@\dZddlmZddlZddlZddlmZddlmZddl m Z dZ dZ gd Z d d d d d d dZejejeZeddZeddZGddeZejedfdCdZdDdZe\ZZZZdEdZe\Z Z!dFdGd!Z"dFdHd%Z#dFdId)Z$dJd+Z%d,Z&d-Z'dKd.Z(dLdMd0Z)dNdOd4Z*dNdPd5Z+dQd6Z,dFdRd7Z-dFdRd8Z.d9Z/d d d d d d d:Z0dSdTdAZ1e2dBkr e/dSdS)Uu puremagic is a pure python module that will identify a file based off it's magic numbers. It is designed to be minimalistic and inherently cross platform compatible, with no imports when used as a module. © 2013-2025 Chris Griffith - License: MIT (see LICENSE) Acknowledgements Gary C. Kessler For use of his File Signature Tables, available at: http://www.garykessler.net/library/file_sigs.html ) annotationsN) unhexlify) namedtuple)chainzChris Griffithz1.29) magic_file magic_string magic_stream from_file from_string from_streamext_from_filename PureErrormagic_footer_arraymagic_header_arraymulti_part_dictwhat PureMagicPureMagicWithConfidencebmpjpegrasttiff)dibjfifjpgrstsuntifr byte_matchoffset extension mime_typenamer)r r!r"r#r$ confidenceceZdZdZdS)rz.Do not have that type of file in our databanksN)__name__ __module__ __qualname____doc__N/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/puremagic/main.pyrrFs8888r,rzmagic_data.jsonfilenameos.PathLike | strreturnVtuple[list[PureMagic], list[PureMagic], list[PureMagic], dict[bytes, list[PureMagic]]]ct|d5}tj|}dddn #1swxYwYtd|dDd}td|d Dd }d |d D}i}|d D]4\}}d|D|t |d<5||||fS)zRead the magic fileutf-8)encodingNc34K|]}t|VdSN_create_puremagic.0xs r- z_magic_data..P+DDq'**DDDDDDr,headersc|jSr6r r;s r-z_magic_data..P TUT`r,)keyc34K|]}t|VdSr6r7r9s r-r<z_magic_data..Qr=r,footersc|jSr6r@rAs r-rBz_magic_data..QrCr,c,g|]}t|Sr+r7r9s r- z_magic_data..Rs!GGG1#A&&GGGr,extension_onlyz multi-partc,g|]}t|Sr+r7r9s r-rIz_magic_data..Us#GrGrGrabHYZ[H\H\GrGrGrr,ascii)openjsonloadsorteditemsrencode) r.fdatar>rF extensionsmulti_part_extensions file_match option_lists r- _magic_datarYJsQ h ) ) )Qy||DDDODDDJ`J`aaaGDDDODDDJ`J`aaaGGG5E0FGGGJ#' #5#;#;#=#=ss KGrGrfqGrGrGri (9(9'(B(BCCDD GZ)> >>s 377r;listctt|dd|d|d|d|dS)NrrLr)rrrRrAs r-r8r8YsP QqT[[1122tA$A$ qT    r,tuple[int, int]c tdtD}tdtD}tD]t}|D]o}|jdkr8t|t |jt|jz}Et|t |j|jz}pu||fS)z3The length of the largest magic string + its offsetcFg|]}t|j|jzSr+)lenr r!r9s r-rIz _max_lengths..hs)VVVaS..9VVVr,c`g|]+}t|jt|jz,Sr+)rcr absr!r9s r-rIz _max_lengths..is/[[[1S..QX>[[[r,r) maxrrrvaluesr!rcr re)max_header_lengthmax_footer_lengthoptionsoptions r- _max_lengthsrlfsVVCUVVVWW[[HZ[[[\\"))++cc c cF}q  $'(93v?P;Q;QTWX^XeTfTf;f$g$g!!$'(93v?P;Q;QTZTa;a$b$b!!  c / //r,list[PureMagicWithConfidence]c g}|D]}t|jdkrdn#tdt|j}|dkrr |jkrd}|t d d|i||srfdtD}|stdt|d d S) z:Rough confidence based on string length and file extension g?z0.皙?g?r%cfg|]-}|jk tdddi|.S)r%rpr+)r"r_asdict)r: magic_rowexts r-rIz_confidence..sP   i))) $ J Js Ji6G6G6I6I J J)))r,zCould not identify filec8|jt|jfSr6)r%rcr rAs r-rBz_confidence..s!,AL8I8I)Jr,TrDreverser+) rcr floatr"appendrrrextension_only_arrayrrP)matchesrtresultsmatchcons ` r- _confidencerxs GSS)**a//ccU;WEDT@U@U;W;W5X5X #::#:#"8"8C.QQ#QQQRRRR  s     1    31222 'JJTX Y Y YYr,headerbytesfooterc g}tD]a}|j}|jt|jz}|t|kr9||||jkr||bt D]_}|j}|jt|jz}|dkr |||n ||d}||jkr||`t }|D]4} | jtvr"t| jD]}|j}|jt|jz}|jdkrn|dkr |||n ||d}||jkrH|t| j|jz|j|j |j |j |t|kr||||jkrH|t|| j||j|j |j |j 6| t|t||S)z/Attempt to identify 'data' by its magic numbersrNr)rr!rcr ryrsetraddrr"r#r$extendrZr) rrrtr{rsstartend match_area new_matchesmatcheds r- _identify_allrs| G'&&  Y%9!:!:: V    %)  4 4 4 NN9 % % %'&&  Y%9!:!::*-((VE#I&&uvv - - - NN9 % % %%%K   0 0,W-?@   !(&Y-A)B)BB#a''69Qhhc !2!2F566NJ!Y%999#%+2+= @T+T'0'7*3*=*3*=%.^ S[[(( eCi(I,@@@#%+1'.32F+G'0'7*3*=*3*=%.^  NN4 $$%%% w $ $$r,mimeboolstrc|stdt|||d}|r|jSt|jt s|jn |djS)z=Discover what type of file it is based on the incoming stringInput was emptyr) ValueErrorrr# isinstancer"rZ)rrrrtinfos r-_magicrsc ,*+++  - -a 0D ~!+DND!A!A X4>>tAwGXXr,tuple[bytes, bytes]cPt|d5}|t} |t t jn%#t$r|dYnwxYw|}dddn #1swxYwY||fS)z"Grab the start and end of the filerbrN)rMreadmax_headseekmax_footosSEEK_ENDOSError)r.finheadfoots r- _file_detailsrs h  xx!!  HHhY , , , ,    HHQKKKKK xxzz  :s4B&ABA63B5A66BB Bc@|dt|t dfS)z$Grab the start and end of the stringN)rr)strings r-_string_detailsrs! )8) fhYZZ0 00r,c8|t} |t tjn,#t tf$r|dYnwxYw|}|d||fS)z$Grab the start and end of the streamr)rrrrrrrr)streamrrs r-_stream_detailsrs ;;x D XIr{++++ Z  A ;;==D KKNNN :s&A&A,+A,cN t|dd\}}n#t$rYdSwxYwd|}dt t t D}|dddr|dd|z}||vr|S|S)zScan a filename for its extension. :param filename: string of the filename :return: the extension off the end (empty string if it can't find one) .r\cg|] }|j Sr+)r"r9s r-rIz%ext_from_filename..sSSS SSSr,N)rlowerrsplitrrrr startswith)r.basertall_extslong_exts r-r r s MM''))00a88 cc rr c))CSSU+=?Q%R%RSSSH BCCyC  9s? x  O Js8; A A Fcdt|\}}t|||t|S)aOpens file, attempts to identify content based off magic number and will return the file extension. If mime is True it will return the mime type instead. :param filename: path to file :param mime: Return mime, not extension :return: guessed extension or mime )rrr )r.rrrs r-r r s2x((JD$ $d$5h$?$? @ @@r,r str | bytesos.PathLike | str | Nonect|tr|d}t|\}}|rt |nd}t ||||S)aReads in string, attempts to identify content based off magic number and will return the file extension. If mime is True it will return the mime type instead. If filename is provided it will be used in the computation. :param string: string representation to check :param mime: Return mime, not extension :param filename: original filename :return: guessed extension or mime r3N)rrrRrr r)rrr.rrrts r-r r sd&#(w'' ((JD$)1 ; H % % %tC $dC ( ((r,cpt|\}}|rt|nd}t||||S)aReads in stream, attempts to identify content based off magic number and will return the file extension. If mime is True it will return the mime type instead. If filename is provided it will be used in the computation. :param stream: stream representation to check :param mime: Return mime, not extension :param filename: original filename :return: guessed extension or mime N)rr r)rrr.rrrts r-r r $sA!((JD$)1 ; H % % %tC $dC ( ((r,ct|\}}|std t||t|}n#t$rg}YnwxYw|dd|S)z Returns list of (num_of_matches, array_of_matches) arranged highest confidence match first. :param filename: path to file :return: list of possible matches, highest confidence first rc|jSr6r%rAs r-rBzmagic_file..CALr,Trv)rrrr rsort)r.rrrs r-rr4sx((JD$ ,*+++T4):8)D)DEE II(($I777 KsA AAc|stdt|\}}|rt|nd}t|||}|dd|S)aC Returns tuple of (num_of_matches, array_of_matches) arranged highest confidence match first If filename is provided it will be used in the computation. :param string: string representation to check :param filename: original filename :return: list of possible matches, highest confidence first rNc|jSr6rrAs r-rBzmagic_string..Vrr,Trv)rrr rr)rr.rrrtrs r-rrGst ,*+++ ((JD$)1 ; H % % %tC tS ) )DII(($I777 Kr,ct|\}}|std|rt|nd}t|||}|dd|S)a>Returns tuple of (num_of_matches, array_of_matches) arranged highest confidence match first If filename is provided it will be used in the computation. :param stream: stream representation to check :param filename: original filename :return: list of possible matches, highest confidence first rNc|jSr6rrAs r-rBzmagic_stream..hrr,Trv)rrr rr)rr.rrrtrs r-r r Zst!((JD$ ,*+++)1 ; H % % %tC tS ) )DII(($I777 Kr,c Jddl}ddlm}|d}|ddddd |d d dd d |dd||r|n|jdd}|jD]}tj |std|d6 td|dt||j n##t$rtd|dYwxYw|jrt|}tdt!|t#|D]\}}|dkrtdntd|td|jtdt'|jdzdtd|jtd |jtd!|jtd"|jd#ʐdS)$Nr)ArgumentParserzpuremagic is a pure python file identification module.It looks for matching magic numbers in the file to locate the file type. ) descriptionz-mz--mime store_truerz)Return the mime type instead of file type)actiondesthelpz-vz--vverbosezPrint verbose outputfiles+)nargsr\zFile 'z' does not exist!'z' : z' : could not be IdentifiedzTotal Possible Matches: z Best Matchz Alertnative Match #z Name: z Confidence: d%z Extension: z Mime Type: z Byte Match: z Offset:  )sysargparser add_argument parse_argsargvrrpathexistsprintr rrrrrc enumerater$intr%r"r#r r!)argsrrparserfnr{iresults r-command_line_entryrlsJJJ'''''' ^ XF     8  eLyOefff s+++   T;TTsx| < > B  B __main__)r.r/r0r1)r;rZr0r)r0r`r6)r0rm)rrrrr0rm)rrrrrrr0r)r.r/r0r)r.r/r0r)F)r.r/rrr0r)FN)rrrrr.rr0r)rrr.rr0r)r.r/r0rm)r.rr0rm)NT)rrrrrrr0r)3r* __future__rrNrbinasciir collectionsr itertoolsr __author__ __version____all__rrabspathdirname__file__hererr LookupErrorrjoinrYr8rrrzrrlrrrrrrrrr r r r rrr rrrr'r+r,r-rsM  #""""" """"""      $VF6RXaghh  wrwx0011 J   %*  99999 999 #%',,t5F"G"G ? ? ? ? ?Q\P[P]P]M&(