VY$@s*ddlmZmZmZddlmZmZmZddlm Z m Z m Z ddl m Z mZmZddlmZmZmZGdddeZGdd d eZGd d d eZGd d d eZGdddeZGdddeZGdddeZdS))EUCTW_CHAR_TO_FREQ_ORDEREUCTW_TABLE_SIZE EUCTW_TYPICAL_DISTRIBUTION_RATIO)EUCKR_CHAR_TO_FREQ_ORDEREUCKR_TABLE_SIZE EUCKR_TYPICAL_DISTRIBUTION_RATIO)GB2312_CHAR_TO_FREQ_ORDERGB2312_TABLE_SIZE!GB2312_TYPICAL_DISTRIBUTION_RATIO)BIG5_CHAR_TO_FREQ_ORDERBIG5_TABLE_SIZEBIG5_TYPICAL_DISTRIBUTION_RATIO)JIS_CHAR_TO_FREQ_ORDERJIS_TABLE_SIZEJIS_TYPICAL_DISTRIBUTION_RATIOc@speZdZdZdZdZdZddZddZd d Z d d Z d dZ ddZ dS)CharDistributionAnalysisigGz?g{Gz?cCsDd|_d|_d|_d|_d|_d|_|jdS)N)_char_to_freq_order _table_sizetypical_distribution_ratio_done _total_chars _freq_charsreset)selfrN/tmp/pip-build-04bmskau/requests/requests/packages/chardet/chardistribution.py__init__.s      z!CharDistributionAnalysis.__init__cCsd|_d|_d|_dS)zreset analyser, clear any stateFN)rrr)rrrrr=s  zCharDistributionAnalysis.resetcCst|dkr|j|}nd}|dkrp|jd7_||jkrpd|j|krp|jd7_dS)z"feed a character with known lengthrriN) get_orderrrrr)rcharZchar_lenorderrrrfeedFs  zCharDistributionAnalysis.feedcCsr|jdks!|j|jkr(|jS|j|jkrk|j|j|j|j}||jkrk|S|jS)z(return confidence based on existing datar)rrMINIMUM_DATA_THRESHOLDSURE_NOrSURE_YES)rrrrrget_confidenceTs! z'CharDistributionAnalysis.get_confidencecCs|j|jkS)N)rENOUGH_DATA_THRESHOLD)rrrrgot_enough_datadsz(CharDistributionAnalysis.got_enough_datacCsdS)Nrr r)rbyte_strrrrr!isz"CharDistributionAnalysis.get_orderN) __name__ __module__ __qualname__r*r'r&r%rrr$r)r+r!rrrrr(s     rcs.eZdZfddZddZS)EUCTWDistributionAnalysiscs2tt|jt|_t|_t|_dS)N) superr0rrrrrrr)r) __class__rrrrs  z"EUCTWDistributionAnalysis.__init__cCs6|d}|dkr.d|d|ddSdSdS)Nr^rr r)rr, first_charrrrr!xs  z#EUCTWDistributionAnalysis.get_order)r-r.r/rr!rr)r2rr0qs r0cs.eZdZfddZddZS)EUCKRDistributionAnalysiscs2tt|jt|_t|_t|_dS)N) r1r7rrrrrrr)r)r2rrrs  z"EUCKRDistributionAnalysis.__init__cCs6|d}|dkr.d|d|ddSdSdS)Nrr4rr5r r)rr,r6rrrr!s  z#EUCKRDistributionAnalysis.get_order)r-r.r/rr!rr)r2rr7s r7cs.eZdZfddZddZS)GB2312DistributionAnalysiscs2tt|jt|_t|_t|_dS)N) r1r9rrrr rr r)r)r2rrrs  z#GB2312DistributionAnalysis.__init__cCsI|d|d}}|dkrA|dkrAd|d|dSdSdS)Nrrr8r5r4r r)rr,r6 second_charrrrr!sz$GB2312DistributionAnalysis.get_order)r-r.r/rr!rr)r2rr9s r9cs.eZdZfddZddZS)Big5DistributionAnalysiscs2tt|jt|_t|_t|_dS)N) r1r;rr rr rr r)r)r2rrrs  z!Big5DistributionAnalysis.__init__cCsd|d|d}}|dkr\|dkrEd|d|ddSd|d|dSndSdS) Nrrr5?@r r)rr,r6r:rrrr!s   z"Big5DistributionAnalysis.get_order)r-r.r/rr!rr)r2rr;s r;cs.eZdZfddZddZS)SJISDistributionAnalysiscs2tt|jt|_t|_t|_dS)N) r1r@rrrrrrr)r)r2rrrs  z!SJISDistributionAnalysis.__init__cCs|d|d}}|dkr>|dkr>d|d}n1|dkrk|dkrkd|dd}nd S||d }|d krd }|S) Nrrr?r r r)rr,r6r:r#rrrr!s z"SJISDistributionAnalysis.get_order)r-r.r/rr!rr)r2rr@s r@cs.eZdZfddZddZS)EUCJPDistributionAnalysiscs2tt|jt|_t|_t|_dS)N) r1rHrrrrrrr)r)r2rrrs  z"EUCJPDistributionAnalysis.__init__cCs6|d}|dkr.d|d|ddSdSdS)Nrr4r5rr r)rr,r"rrrr!s  z#EUCJPDistributionAnalysis.get_order)r-r.r/rr!rr)r2rrHs rHN)Z euctwfreqrrrZ euckrfreqrrrZ gb2312freqrr r Zbig5freqr r r Zjisfreqrrrobjectrr0r7r9r;r@rHrrrrsI