b) &dZddlZddlZddlZddlmZddlmZddlm Z m Z m Z dZ dZ gdZejejeZed d Zed d ZGd deZejedfde ejefde e ee ee effdZde defdZe\ZZZde e e ffdZ!d*de efdZ"d*de#de#de efdZ$d*de#de#de%defdZ&de ejefde e#e#ffdZ'dZ(dZ)de ejefdefdZ*d+de ejefde%defd!Z+d,d"e ee#fde%de ejefdefd#Z,d,de%de ejefdefd$Z-de ejefde efd%Z.d*de ejefde efd&Z/d*de ejefde efd'Z0d(Z1e2d)kr e1dSdS)-u 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. © 2013-2022 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 N)chain) namedtuple)UnionTupleListzChris Griffithz1.14) magic_file magic_string magic_stream from_file from_string from_streamext_from_filename PureErrormagic_footer_arraymagic_header_array PureMagic byte_matchoffset extension mime_typenamePureMagicWithConfidence)rrrrr confidenceceZdZdZdS)rz.Do not have that type of file in our databanksN)__name__ __module__ __qualname____doc__0/usr/lib/python3/dist-packages/puremagic/main.pyrr>s8888r!rzmagic_data.jsonfilenamereturnc.t|5}tj|}dddn #1swxYwYtd|dDd}td|dDd}d |d D}|||fS) zRead the magic fileNc34K|]}t|VdSN_create_puremagic.0xs r" z_magic_data..H+DDq'**DDDDDDr!headersc|jSr'rr,s r"z_magic_data..H TUT`r!)keyc34K|]}t|VdSr'r(r*s r"r-z_magic_data..Ir.r!footersc|jSr'r1r2s r"r3z_magic_data..Ir4r!c,g|]}t|Sr r(r*s r" z_magic_data..Js!GGG1#A&&GGGr!extension_only)openjsonloadsorted)r#fdatar/r7 extensionss r" _magic_datarCBs h1y||DDDODDDJ`J`aaaGDDDODDDJ`J`aaaGGG5E0FGGGJ GZ ''s 155r,cttj|dd|d|d|d|dS)Nrasciir)rbinascii unhexlifyencoder2s r"r)r)NsX %adkk'&:&:;;AaDTUVWTXdefgdhopqros   r!ctdtD}tdtD}||fS)z3The length of the largest magic string + its offsetcFg|]}t|j|jzSr )lenrrr*s r"r:z _max_lengths..Ys)VVVaS..9VVVr!c`g|]+}t|jt|jz,Sr )rOrabsrr*s r"r:z _max_lengths..Zs/[[[1S..QX>[[[r!)maxrr)max_header_lengthmax_footer_lengths r" _max_lengthsrUWsJVVCUVVVWW[[HZ[[[\\ / //r!c g}|D]}t|jdkrdn3tdt|j}|dkr|r ||jkrd}|t d d|i||sJ|rHtD]@}||jkr3|t d ddi|A|stdt|dd S) z:Rough confidence based on string length and file extension g?z0.{0}g?g?rzCould not identify filec|jSr'rr2s r"r3z_confidence..osr!Tr5reverser ) rOrfloatformatrappendr_asdictextension_only_arrayrr?)matchesextresultsmatchcon magic_rows r" _confidencerg^s7GSS)**a//ccU7>>#eN^J_J_;`;`5a5a #::#:#"8"8C.QQ#QQQRRRR _s_- _ _Ii)))6]]#]IZIZI\I\]]^^^ 31222 '55t D D DDr!headerfooterct}tD]a}|j}|jt|jz}|t|kr9||||jkr||bt D]1}|j}||d|jkr||2t||S)z/Attempt to identify 'data' by its magic numbersN)listrrrOrr^rrg)rhrirbrarfstartends r" _identify_allrnrs ffG'&&  Y%9!:!:: V    %)  4 4 4 NN9 % % %'&&   %&&>Y1 1 1 NN9 % % % w $ $$r!mimec|stdt|||d}|r|jSt|jt s|jn |djS)z=Discover what type of file it is based on the incoming stringInput was emptyr) ValueErrorrnr isinstancerrk)rhrirorbinfos r"_magicrusc ,*+++  - -a 0D ~!+DND!A!A X4>>tAwGXXr!c^t\}}t|d5}||} || tjn%#t $r|dYnwxYw|}dddn #1swxYwY||fS)z"Grab the start and end of the filerbrN)rUr<readseekosSEEK_ENDIOError)r#max_headmax_footfinheadfoots r" _file_detailsrs%Hh h  xx!!  HHhY , , , ,    HHQKKKKK xxzz  :s4B !AB A=:B <A==B  B$'B$cNt\}}|d||| dfS)z$Grab the start and end of the stringN)rU)stringr}r~s r"_string_detailsrs/%Hh )8) fhYZZ0 00r!ct\}}||}|| tj|}||fS)z$Grab the start and end of the stream)rUrxryrzr{)streamr}r~rrs r"_stream_detailsrsP%Hh ;;x D KK 2;''' ;;==D :r!cn t|dd\}}n#t$rYdSwxYwd|}dt t tD}|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) .rFz.{0}cg|] }|j Sr )rr*s r"r:z%ext_from_filename..sSSS SSSr!N) strlowerrsplitrrr]rrr startswith)r#baserball_extslong_exts r"rrs MM''))00a88 cc rr --  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 )rrur)r#rorrs r"r r s2x((JD$ $d$5h$?$? @ @@r!rct|tr|d}t|\}}|rt |nd}t ||||S)aReads in string, attempts to identify content based off magic number and will return the file extension. If mime is True it will return the mime type instead. If filename is provided it will be used in the computation. :param string: string representation to check :param mime: Return mime, not extension :param filename: original filename :return: guessed extension or mime zutf-8N)rsrrLrrru)rror#rrrbs r"r r sd&#(w'' ((JD$)1 ; H % % %tC $dC ( ((r!cpt|\}}|rt|nd}t||||S)aReads in stream, attempts to identify content based off magic number and will return the file extension. If mime is True it will return the mime type instead. If filename is provided it will be used in the computation. :param stream: stream representation to check :param mime: Return mime, not extension :param filename: original filename :return: guessed extension or mime N)rrru)rror#rrrbs r"r r sA!((JD$)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 rqc|jSr'rYr2s r"r3zmagic_file..ALr!TrZ)rrrrnrrsort)r#rrrts r"rrsx((JD$ ,*+++T4):8)D)DEE II(($I777 KsA AAc|stdt|\}}|rt|nd}t|||}|dd|S)aC Returns tuple of (num_of_matches, array_of_matches) arranged highest confidence match first If filename is provided it will be used in the computation. :param string: string representation to check :param filename: original filename :return: list of possible matches, highest confidence first rqNc|jSr'rYr2s r"r3zmagic_string..rr!TrZ)rrrrrnr)rr#rrrbrts r"r r st ,*+++ ((JD$)1 ; H % % %tC tS ) )DII(($I777 Kr!ct|\}}|std|rt|nd}t|||}|dd|S)a>Returns tuple of (num_of_matches, array_of_matches) arranged highest confidence match first If filename is provided it will be used in the computation. :param stream: stream representation to check :param filename: original filename :return: list of possible matches, highest confidence first rqNc|jSr'rYr2s r"r3zmagic_stream..'rr!TrZ)rrrrrnr)rr#rrrbrts r"r r st!((JD$ ,*+++)1 ; H % % %tC tS ) )DII(($I777 Kr!c Jddlm}ddl}|d}|ddddd |d d ||r|n|jdd}|jD]}tj |s#td |D td |t||j |#t$r%td |YwxYwdS)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_trueroz)Return the mime type instead of file type)actiondesthelpfiles+)nargsrFzFile '{0}' does not exist!z '{0}' : {1}z'{0}' : could not be Identified)argparsersys add_argument parse_argsargvrrzpathexistsprintr]r ror)argsrrparserfns r"command_line_entryr+so''''''JJJ ^ XF   h|&?j s+++   T;TTsx| < > ? ? ? ? ? @ @@s:6C11,D D __main__r')F)FN)3rrzr=rJ itertoolsr collectionsrtypingrrr __author__ __version____all__rabspathdirname__file__hererr LookupErrorrjoinPathLikerrCr)rrr`intrUrgbytesrnboolrurrrrr r r rr r rrr r!r"rs   """"""%%%%%%%%%%       wrwx0011 J   %*  99999 999 )+ T;L(M(M ( (BK$% ( 4 ?DOT)_ <= ( ( ( () @K{}}<&(<0eCHo0000EEd+B&CEEEE(%%%%%TBY=Z%%%%,YY5Y%YtY#YYYY E"+s"23 eUl8K    111 bk3&6 7C* A Abk3./ At A A A A A))c5j)))r{\_O_I`)lo))))$ ) )d )eBK