o n~bOO@sddlZddlmZmZddlmZmZmZmZzddl m Z Wn e y+e Z Ynwddl mZmZmZmZddlmZmZmZmZddlmZdd lmZmZdd lmZmZmZm Z m!Z!m"Z"e#d Z$e%Z&e&'e(d       d%de)de*de*de+dee dee de,de,defddZ-      d%dede*de*de+dee dee de,de,defddZ.      d%d e de*de*de+dee dee de,de,defd!d"Z/     d&d e de*de*de+dee dee de,defd#d$Z0dS)'N)basenamesplitext)BinaryIOListOptionalSet)PathLike)coherence_ratioencoding_languagesmb_encoding_languagesmerge_coherence_ratios)IANA_SUPPORTEDTOO_BIG_SEQUENCETOO_SMALL_SEQUENCETRACE) mess_ratio) CharsetMatchCharsetMatches)any_specified_encoding iana_nameidentify_sig_or_bom is_cp_similaris_multi_byte_encodingshould_strip_sig_or_bomZcharset_normalizerz)%(asctime)s | %(levelname)s | %(message)s皙?TF sequencessteps chunk_size threshold cp_isolation cp_exclusionpreemptive_behaviourexplainreturnc1 Cs t|ttfstdt||rtj}tt t t t |} | dkrGt d|r;tt t |p9tjtt|dddgdgS|dur]tt d d |d d |D}ng}|durutt d d |dd |D}ng}| ||krtt d||| d}| }|dkr| ||krt| |}t |tk} t |tk} | rtt d| n | rtt d| g} |rt|nd} | dur| | tt d| t}g}g}d}d}d}t}t|\}}|dur| |tt dt ||| dd| vr| d| tD],}|r!||vr!q|r+||vr+q||vr2q||d}||k}|oCt|}|dvrU|sUtt d|qzt|}Wnt t!fyott d|Yqwz9| r|durt"|dur|dtdn |t |td|dnt"|dur|n|t |d|d}Wn+t#t$fy}zt|t$stt d|t"|||WYd}~qd}~wwd}|D] }t%||rd}nq|rtt d||qt&|sdnt || t| |}|o|duot || k} | rtt d|tt |d}!t'|!d }!d}"d}#g}$g}%|D]}&|&|| d!krAq4||&|&|}'|rU|durU||'}'z|'j(||r^d"nd#d$}(Wn"t#y}ztt d%|t"||!}"d}#WYd}~nd}~ww|r|&dkr||&d&krt)|d'})|r|(d|)|vrt&|&|&dd(D])}*||*|&|}'|r|dur||'}'|'j(|d"d$}(|(d|)|vrnq|$|(|%t*|(||%d(|kr|"d7}"|"|!ks|r|durnq4|#s=| r=|s=z|td)dj(|d#d$Wn#t#y<}ztt d*|t"|||WYd}~qd}~ww|%rHt+|%t |%nd}+|+|ksT|"|!kr||tt d+||"t,|+d,d-d.|dd| fvr|#st|||dg|},|| kr|,}n |dkr|,}n|,}qtt d/|t,|+d,d-d.|st-|}-nt.|}-|-rtt d0|t"|-g}.|dkr|$D]}(t/|(d1|-rd2|-nd}/|.|/qt0|.}0|0rtt d3|0||t|||+||0||| ddfvr|+d1krt d4||rtt t |t||gS||kr@t d5||r7tt t |t||gSqt |dkr|sR|sR|rXtt d6|rht d7|j1||n2|rp|dus|r}|r}|j2|j2ks|durt d8||n |rt d9|||rt d:|3j1t |dnt d;|rtt t ||S)]zfrom_bytes..zacp_exclusion is set. use this flag for debugging purpose. limited list of encoding excluded : %s.cSr)r*r+r,r/r/r0r1hr2z^override steps (%i) and chunk_size (%i) as content does not fit (%i byte(s) given) parameters.r z>Trying to detect encoding from a tiny portion of ({}) byte(s).zIUsing lazy str decoding because the payload is quite large, ({}) byte(s).z@Detected declarative mark in sequence. Priority +1 given for %s.zIDetected a SIG or BOM mark on first %i byte(s). Priority +1 given for %s.ascii>utf_16utf_32z[Encoding %s wont be tested as-is because it require a BOM. Will try some sub-encoder LE/BE.z2Encoding %s does not provide an IncrementalDecodergA)encodingz9Code page %s does not fit given bytes sequence at ALL. %sTzW%s is deemed too similar to code page %s and was consider unsuited already. Continuing!zpCode page %s is a multi byte encoding table and it appear that at least one character was encoded using n-bytes.ignorestrict)errorszaLazyStr Loading: After MD chunk decode, code page %s does not fit given bytes sequence at ALL. %sgj@z^LazyStr Loading: After final lookup, code page %s does not fit given bytes sequence at ALL. %szc%s was excluded because of initial chaos probing. Gave up %i time(s). Computed mean chaos is %f %%.d)ndigitsz=%s passed initial chaos probing. Mean measured chaos is %f %%z&{} should target any language(s) of {}g?,z We detected language {} using {}z.Encoding detection: %s is most likely the one.zoEncoding detection: %s is most likely the one as we detected a BOM or SIG within the beginning of the sequence.zONothing got out of the detection process. Using ASCII/UTF-8/Specified fallback.z7Encoding detection: %s will be used as a fallback matchz:Encoding detection: utf_8 will be used as a fallback matchz:Encoding detection: ascii will be used as a fallback matchz]Encoding detection: Found %s as plausible (best-candidate) for content. With %i alternatives.z=Encoding detection: Unable to determine any suitable charset.)4 isinstance bytearraybytes TypeErrorformattypeloggerlevel addHandlerexplain_handlersetLevelrlendebug removeHandlerloggingWARNINGrrlogjoinintrrrappendsetrraddrrModuleNotFoundError ImportErrorstrUnicodeDecodeError LookupErrorrrangemaxdecodeminrsumroundr r r r r6 fingerprintbest)1rrr r!r"r#r$r%Zprevious_logger_levellengthZis_too_small_sequenceZis_too_large_sequenceZprioritized_encodingsZspecified_encodingZtestedZtested_but_hard_failureZtested_but_soft_failureZfallback_asciiZ fallback_u8Zfallback_specifiedresultsZ sig_encodingZ sig_payloadZ encoding_ianaZdecoded_payloadZbom_or_sig_availableZstrip_sig_or_bomZis_multi_byte_decodereZsimilar_soft_failure_testZencoding_soft_failedZr_Zmulti_byte_bonusZmax_chunk_gave_upZearly_stop_countZlazy_str_hard_failureZ md_chunksZ md_ratiosiZ cut_sequencechunkZchunk_partial_size_chkjZmean_mess_ratioZfallback_entryZtarget_languagesZ cd_ratiosZchunk_languagesZcd_ratios_mergedr/r/r0 from_bytes%s                                                       rmfpc Cst||||||||S)z Same thing than the function from_bytes but using a file pointer that is already ready. Will not close the file pointer. )rmread)rnrr r!r"r#r$r%r/r/r0from_fpsrppathc CsDt|d}t||||||||WdS1swYdS)z Same thing than the function from_bytes but with one extra step. Opening and reading given file path in binary mode. Can raise IOError. rbN)openrp) rqrr r!r"r#r$r%rnr/r/r0 from_paths $rtc Cst|||||||}t|}tt|} t|dkr!td||} | dd| j7<t dt | |d | d} | | Wd| S1sTwY| S)zi Take a (text-based) file path and try to create another file next to it, this time using UTF-8. rz;Unable to normalize "{}", no encoding charset seems to fit.-z{}r(wbN)rtrlistrrOIOErrorrHrfr6rsr\replacerUwriteoutput) rqrr r!r"r#r$rhfilenameZtarget_extensionsresultrnr/r/r0 normalize7s8    r~)rrrNNTF)rrrNNT)1rRos.pathrrtypingrrrrosrr[r\cdr r r r Zconstantrrrrmdrmodelsrrutilsrrrrrr getLoggerrJ StreamHandlerrM setFormatter FormatterrFrVfloatboolrmrprtr~r/r/r/r0s     b