|jn UdZddlmZddlZddlZddlZddlZddlmZddl m Z ddl m Z ddl mZe dZd Zged RZd Zd Zd ZdZdZejddGddZejd dddZgeddejd d!"ed#d$ejd d!"ed%d&ejd d!"ed'd(ejd d!"ed)d*ejd d!"ed+d,ejd d!"ed-d.ejd d!"ed/d0ejd d!"ed1d2ejd d!"ed3d4ejd d!"ed5d6ejdd7"ed8d9ejdd:"ed;dd?ejdd:"ed@dAejdd="edBdCejdd7"edDdEejdd7"edFdGejdd:"edHdIejdd:"edJdKejdd:"edLdMejdd="edNdOejdd:"edPdQejd dR"edSdTejd gedUR"edVdWejd e"edXdYejd e"edZd[ejd d\"ed]d^ejd d_"ed`daejd db"edcddejd e"ededfejd e"edgdhejd di"edjdkejd dl"edmdnejd do"edpdqejd dR"edrdsejd e"edtduejd e"edvdwejd dx"edydzejd e"ed{d|ejd e"ed}d~ejd e"eddejd d\"eddejd db"eddejd d_"eddejd d"eddejd e"eddejd d"eddejd e"eddejd d"eddejdd="eddejd e"eddejd d\"eddejd d"eddejd e"eddejd e"eddejd d_"eddejd e"eddejd d"eddejd do"eddejd d"eddejd d"eddejd d"eddejd e"eddejd d"eddejd d\"eddejd e"eddejd e"eddejd e"eddejd e"eddejd db"eddejd d_"eddejd e"eddejd dǬ"eddejd d"eddejd db"eddejd dά"eddejd dѬ"eddejd dԬ"eddejd e"eddejd d\"eddej d e"eddej d db"eddej d e"eddej d d\"eddej d d_"eddej d d"RZ!e de!DZ"de#d<ejdddZ$ddZ%ddZ&dS)zz!get_candidates..s.DD#cgmD#DDDDDDrNc3.K|]}|jv |VdSrr)rrrs rrz!get_candidates..s0QQc38?P3P3Pc3P3P3P3PQQrc3.K|]}|jv |VdSrr)rrrs rrz!get_candidates..s0UUc38CT3T3Tc3T3T3T3TUUr)REGISTRYvaluestuple)rrr candidatess``` rget_candidatesrswEDDD!2!2DDDJ$QQQQZQQQ $UUUUZUUU   rr)zus-asciiFr)rrrrrr\)r\utf8csutf8zunicode-1-1-utf-8 unicode11utf8 unicode20utf8zx-unicode20utf8r])z UTF-8-SIGz utf-8-bomrU)zUTF-16utf16csutf16rV)z UTF-16-BEzutf-16be csutf16berW)z UTF-16-LEzutf-16le csutf16lerX)zUTF-32utf32csutf32rY)z UTF-32-BEzutf-32be csutf32berZ)z UTF-32-LEzutf-32le csutf32ler[)zUTF-7utf7csutf7r ) z Big5-HKSCS Big5HKSCSbig5zbig5-twcsbig5cp950zcn-big5zx-x-big5 csbig5hkscs)zhr.)CP932ms932mskanjizms-kanji cswindows31jz windows-31j)jar/) CP949ms949uhcz windows-949 csksc56011987z iso-ir-149zks_c_5601-1987zks_c_5601-1989ksc5601ksc_5601)kor0)z EUC-JIS-2004zeuc-jpeucjpujiszu-jisz euc-jisx0213cseucpkdfmtjapanesezx-euc-jpr1)zEUC-KReuckrcseuckrr2)GB18030zgb-18030gb2312gbkcsgb2312gb_2312z gb_2312-80zx-gbkcsiso58gb231280z iso-ir-58 csgb18030csgbkcp936ms936z windows-936r4)z HZ-GB-2312r4r5)z ISO-2022-JP-2z iso-2022-jp csiso2022jpz iso2022-jp-1 csiso2022jp2r6)zISO-2022-JP-2004z iso2022-jp-3r7)zISO-2022-JP-EXTr8)z ISO-2022-KR csiso2022krrS) zShift-JIS-2004Shift_JIS_2004 shift_jissjisshiftjiss_jiszshift-jisx0213zx-sjis csshiftjisms_kanjir,)CP874z windows-874zdos-874)thr)z Windows-1250rzx-cp1250 cswindows1250rsr)z Windows-1251rzx-cp1251 cswindows1251r)z Windows-1252rzx-cp1252 cswindows1252r)z Windows-1253rzx-cp1253 cswindows1253)elr)z Windows-1254rzx-cp1254 cswindows1254)ror)z Windows-1255rzx-cp1255 cswindows1255)her)z Windows-1256rzx-cp1256 cswindows1256r)z Windows-1257rzx-cp1257 cswindows1257r)z Windows-1258rzx-cp1258 cswindows1258)virH)zKOI8-Rkoi8rkoikoi8cskoi8r)rprJ)zKOI8-Ukoi8uzkoi8-rucskoi8u)rrrT)zTIS-620tis620z iso-8859-11z iso8859-11 iso885911cstis620r9)z ISO-8859-1zlatin-1latin1r9iso88591r?)z ISO-8859-2zlatin-2latin2r?iso88592r@)z ISO-8859-3zlatin-3latin3r@iso88593)eomtrorA)z ISO-8859-4zlatin-4latin4rAiso88594rB)z ISO-8859-5rBcyrilliciso88595rC)z ISO-8859-6rCarabiciso88596z iso-8859-6-ez iso-8859-6-i csiso88596e csiso88596irD)z ISO-8859-7rDgreekiso88597 sun_eu_greekrE) z ISO-8859-8rEhebrewiso88598z iso-8859-8-ez iso-8859-8-i csiso88598e csiso88598ivisuallogicalrF)z ISO-8859-9zlatin-5latin5rFiso88599r:)z ISO-8859-10zlatin-6latin6r: iso885910)rhrdr;)z ISO-8859-13zlatin-7latin7r; iso885913 csiso885913r<) z ISO-8859-14zlatin-8latin8r< iso885914 csiso885914z iso-ir-199z iso-celticl8)r_rfr^gdr=)z ISO-8859-15zlatin-9latin9r= iso885915 csisolatin9 csiso885915l9r>)z ISO-8859-16zlatin-10latin10r> iso885916 csiso885916z iso-ir-226l10)rzrvryrxr{r|rG)JohabrL)z Mac-Cyrillic MacCyrillic maccyrilliczx-mac-cyrilliczx-mac-ukrainianrM)z Mac-GreekMacGreekmacgreekrN)z Mac-Iceland MacIceland maciceland)rhrO)z Mac-Latin2 MacLatin2 maclatin2maccentraleuroperP)z Mac-RomanMacRomanmacroman macintosh csmacintoshmacz x-mac-romanrQ)z Mac-Turkish MacTurkish macturkishr)CP720r )CP1006)urr )CP1125rI)zKOI8-T)tgrK)zKZ-1048rKz strk1048-2002rk1048)kkrR)PTCP154pt154cp154r3)z HP-Roman8roman8r8 csHPRoman8r)CP437) rbrerarcrmrirkr`rnrdrfr)CP737r)CP775r)CP850r)CP852r )CP855r!)CP856r")CP857r#)CP858r$)CP860)rmr%)CP861r&)CP862r')CP863)rer()CP864)rr))CP865)r`rlr*)CP866r+)CP869r )CP1140cp037cp01140ibm01140ibm1140 csibm01140r)CP424r)CP500r-)CP875r )CP1026r)CP273)raci|] }|j| Srr)res r rms***1QVQ***rz#MappingProxyType[str, EncodingInfo]rirstrEncodingName | Nonecx|}tD]C}|j|kr |jcS|jD]%}||kr |jccS&D t j|j}n#ttf$rYdSwxYw||krt|SdS)zConvert an encoding name string to the canonical EncodingName. Handles arbitrary casing, aliases, and Python codec names. :param name: Any encoding name string. :returns: The canonical :data:`EncodingName`, or ``None`` if unknown. N) lowerrrrrcodecslookup LookupError ValueErrorlookup_encoding)rloweredentryalias codec_names rrvrvsjjllG"""" : :   ] " "E{{}}''z!!!!!( "]4((-  $ttWz*** 4s3B B"!B" param_namecVt|}|d|d|}t||S)aValidate and normalize a single encoding name. :param name: The encoding name to validate. :param param_name: Parameter name for error messages. :returns: The canonical encoding name. :raises ValueError: If the encoding name is unknown. NzUnknown encoding z in )rvru)rr{ canonicalmsgs r_validate_encodingrs> %%I:$::j::oo r encodingsIterable[str] | Nonecr|dStfd|D}|sd}t||S)a8Normalize an iterable of encoding names to canonical forms. :param encodings: Encoding names to normalize, or ``None``. :param param_name: Parameter name for error messages. :returns: A frozenset of canonical encoding names, or ``None``. :raises ValueError: If any encoding name is unknown. Nc38K|]}t|VdSr)r)rrr{s rrz&normalize_encodings..s.RR)$ ;;RRRRRRrzG must not be empty; omit the argument or pass None to disable filtering) frozensetru)rr{resultr~s ` rnormalize_encodingsrsYt RRRR RRR R RF dddoo Mr)NN)rrrrrrrr)rrnrro)rrnr{rnrrn)rrr{rnrr)'r __future__rrr dataclasses functoolsrcollections.abcrtypesrtypingr chardet.enumsrr_WESTERN _WESTERN_TR _CYRILLIC _CENTRAL_EU_CENTRAL_EU_NO_RO_BALTIC_ARABIC dataclassr lru_cacher MODERN_WEBLEGACY_REGIONAL LEGACY_ISO LEGACY_MACDOS MAINFRAME_REGISTRY_ENTRIESrrrvrrrrrrsrBBB"""""" $$$$$$""""""%%%%%%UW t & $ 0 8 8  d$///%%%%%%%0/%:S!!!04/3"!0c L   " c L    "c 0L *  " 1c >L .  " ?c LL 6  " Mc ZL 6  " [c hL .  " ic vL 6  " wc DL 6  " Ec RL +  ' Sc bL     "cc DL    "   Ec `L     ")ac LL     "Mc lL .  " mc zL "  "9{c vL $  ' wc DL    "   Ec ^L 4  " _c lL $  " mc zL .  ' {c HL     "!Ic nL 3  " oc |L G  "&K&&& }c JL G  " Kc XL G  " Yc fL G  " gc tL G  " uc B L G  " C c P L G  " Q c ^ L G  " _ c l L G  " m c | L =  " } c J L 9  " K c Z L    "   [ c x L L  " y c F L L  " G c T L L  "$ U c b L L  " c c p L C  " q c ~ L     "' c h L    "   i c B L     "+C c p L L  " q c ~ L O  "  c LL    "   Mc hL     "*!ic LL     "Mc lL     "6mc NL   " Oc ^L    "   _c xL 5  " yc FL ;  " Gc TL L  "# Uc bL    "cc @L ;  " Ac PL   ' Qc ^L   ' _c lL   ' mc zL   ' {c HL @  ' Ic VL -  ' Wc dL ;  ' ec tL  OT uc BL  O Cc PL  O Qc ^L  O _c lL  O mc zL  O {c HL  O Ic VL  O Wc dL  O ec rL  O sc @L  O Ac NL  O Oc \L  O ]c jL  O kc xL  O yc FL  O Gc TL  O Uc dL    !   ec @L   ! Ac NL   ! Oc \L   ! ]c jL   ! kc xL   ! yc c J1A0@**)***11T"""#"2    r