o cJ@sddlZddlmZddlmZmZmZmZmZddl m Z m Z m Z m Z ddlmZmZmZmZddlmZddlmZmZdd lmZmZmZmZmZmZmZe d Z!e"Z#e#$e%d       d%de&de'de'de(deee)deee)de*de*de(defddZ+      d%dede'de'de(deee)deee)de*de*de(defdd Z,      d%d!d"de'de'de(deee)deee)de*de*de(defd#d$Z-dS)&N)PathLike)AnyBinaryIOListOptionalSet)coherence_ratioencoding_languagesmb_encoding_languagesmerge_coherence_ratios)IANA_SUPPORTEDTOO_BIG_SEQUENCETOO_SMALL_SEQUENCETRACE) mess_ratio) CharsetMatchCharsetMatches)any_specified_encodingcut_sequence_chunks iana_nameidentify_sig_or_bom is_cp_similaris_multi_byte_encodingshould_strip_sig_or_bomcharset_normalizerz)%(asctime)s | %(levelname)s | %(message)s皙?TF皙? sequencessteps chunk_size threshold cp_isolation cp_exclusionpreemptive_behaviourexplainlanguage_thresholdreturnc . Cs t|ttfstdt||r tj} tt t t t |} | dkrHt d|rt d5||n |rKt d6|||r]t d7|3j1t |dnt d8|rott t | |S)9Nz4Expected object of type bytes or bytearray, got: {0}rzZzfrom_bytes..zacp_exclusion is set. use this flag for debugging purpose. limited list of encoding excluded : %s.cSr,r-r.r/r2r2r3r4er5z^override steps (%i) and chunk_size (%i) as content does not fit (%i byte(s) given) parameters.rz>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_32utf_16z\Encoding %s won't be tested as-is because it require a BOM. Will try some sub-encoder LE/BE.>utf_7zREncoding %s won't be tested as-is because detection is unreliable without BOM/SIG.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.zaLazyStr Loading: After MD chunk decode, code page %s does not fit given bytes sequence at ALL. %sgj@strict)errorsz^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 {},z We detected language {} using {}rz.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 removeHandlerloggingWARNINGrrlogjoinintrrrappendsetrr addrrModuleNotFoundError ImportErrorstrUnicodeDecodeError LookupErrorrrangemaxrrdecodesumroundr r r r r: fingerprintbest).r r!r"r#r$r%r&r'r(Zprevious_logger_levellengthZis_too_small_sequenceZis_too_large_sequenceZprioritized_encodingsspecified_encodingZtestedZtested_but_hard_failureZtested_but_soft_failureZfallback_asciiZ fallback_u8Zfallback_specifiedresultsZ sig_encoding sig_payload encoding_ianadecoded_payloadbom_or_sig_availablestrip_sig_or_bomis_multi_byte_decodereZsimilar_soft_failure_testZencoding_soft_failedZr_Zmulti_byte_bonusZmax_chunk_gave_upZearly_stop_countZlazy_str_hard_failureZ md_chunksZ md_ratioschunkmean_mess_ratioZfallback_entryZtarget_languagesZ cd_ratiosZchunk_languagesZcd_ratios_mergedr2r2r3 from_bytes!s                             &                              rrfpc Cs t||||||||| S)N)rrread) rsr!r"r#r$r%r&r'r(r2r2r3from_fps rupathz PathLike[Any]c CsH t|d} t| |||||||| WdS1swYdS)Nrb)openru) rvr!r"r#r$r%r&r'r(rsr2r2r3 from_paths $ry)rrrNNTFr).rRosrtypingrrrrrcdr r r r constantr rrrmdrmodelsrrutilsrrrrrrr getLoggerrJ StreamHandlerrM setFormatter FormatterrFrVfloatr\boolrrruryr2r2r2r3s  $     W