U
|
H=®d ã @ s4 d dl mZ d dlmZmZmZ G dd deZdS )é )Ú CharSetProber)ÚCharacterCategoryÚ ProbingStateÚSequenceLikelihoodc sb e Zd ZdZdZdZdZd fdd Z fd d
|
Ze d d Z
|
e d d Z dd Z dd Z ZS )ÚSingleByteCharSetProberé@ i gffffffî?g©?FNc sJ t t| ¡ || _|| _|| _d | _d | _d | _d | _ d | _
|
| ¡ d S )N) Úsuperr Ú__init__Ú_modelÚ _reversedÚ _name_proberÚ _last_orderÚ _seq_countersÚ _total_seqsÚ _total_charÚ
|
_freq_charÚreset)ÚselfÚmodelÚreversedÚ name_prober©Ú __class__© úZD:\z\workplace\VsCode\pyvenv\venv\Lib\site-packages\pip/_vendor/chardet/sbcharsetprober.pyr ' s z SingleByteCharSetProber.__init__c s: t t| ¡ d| _dgt ¡ | _d| _d| _d| _ d S )Néÿ é )
|
r r r r r Úget_num_categoriesr r r r ©r r r r r 5 s z SingleByteCharSetProber.resetc C s | j r| j jS | jd S d S )NÚ charset_name)r r r
|
r r r r r ? s z$SingleByteCharSetProber.charset_namec C s | j r| j jS | j d¡S d S )NÚlanguage)r r r
|
Úgetr r r r r F s z SingleByteCharSetProber.languagec C sj | j d s| |¡}|s | jS | j d }t|D ]¶\}}|| }|tjk rX| jd7 _|| jk rà| jd7 _| j | jk rà| j
|
d7 _
|
| j s°| j | j | }| j d | }n || j | j }| j d | }| j | d7 < || _ q0| j d }| jt jkrd| j
|
| jkrd| ¡ }|| jkr<| j d||¡ t j| _n(|| jk rd| j d||| j¡ t j| _| jS )NÚkeep_english_letterÚchar_to_order_mapr Úprecedence_matrixr z$%s confidence = %s, we have a winnerz9%s confidence = %s, below negative shortcut threshhold %s)r
|
Úfilter_international_wordsÚstateÚ enumerater ÚCONTROLr Ú SAMPLE_SIZEr r r r r r Ú DETECTINGÚSB_ENOUGH_REL_THRESHOLDÚget_confidenceÚPOSITIVE_SHORTCUT_THRESHOLDÚloggerÚdebugÚFOUND_ITÚ_stateÚNEGATIVE_SHORTCUT_THRESHOLDÚNOT_ME) r Úbyte_strr# ÚiÚcÚorderr r Ú
|
confidencer r r ÚfeedM sL
|
|
|
|
|
|
ÿ
|
ýz SingleByteCharSetProber.feedc C sN d}| j dkrJd| jtj | j | jd }|| j | j }|dkrJd}|S )Ng{®Gáz?r g ð?Útypical_positive_ratiog®Gáz®ï?)r r r ÚPOSITIVEr
|
r r )r Úrr r r r, | s
|
ÿÿz&SingleByteCharSetProber.get_confidence)FN)Ú__name__Ú
|
__module__Ú __qualname__r) r+ r- r2 r r Úpropertyr r r9 r, Ú __classcell__r r r r r ! s
|
|
|
/r N)Ú charsetproberr Úenumsr r r r r r r r Ú<module> s
|
