3
Re @ s4 d dl mZ d dlmZmZmZ G dd deZdS ) )
CharSetProber)CharacterCategory ProbingStateSequenceLikelihoodc sb e Zd ZdZdZdZdZd fdd Z fd d
Ze d d Z
e d
d Z dd Z dd Z
ZS )SingleByteCharSetProber@ i gffffff?g?FNc sJ t t| j || _|| _|| _d | _d | _d | _d | _ d | _
| j d S )N) superr __init___model _reversed _name_prober _last_order
_seq_counters _total_seqs _total_char
_freq_charreset)selfmodelreversed name_prober) __class__ /builddir/build/BUILDROOT/alt-python36-pip-20.2.4-5.el9.x86_64/opt/alt/python36/lib/python3.6/site-packages/pip/_vendor/chardet/sbcharsetprober.pyr ' s z SingleByteCharSetProber.__init__c s: t t| j d| _dgtj | _d| _d| _d| _ d S )N )
r r r r
r get_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 )N charset_name)r r r
)r r r r r ? s z$SingleByteCharSetProber.charset_namec C s | j r| j jS | jjdS d S )Nlanguage)r r r
get)r r r r r F s z SingleByteCharSetProber.languagec C sn | j d s| j|}|s | jS | j d }xt|D ]\}}|| }|tjk rZ| jd7 _|| jk r| jd7 _| j | jk r| j
d7 _
| j s| j | j | }| j d | }n || j | j }| j d | }| j | d7 < || _ q2W | j d }| jt
jkrh| j
| jkrh| j }|| jkr@| jjd|| t
j| _n(|| jk rh| jjd||| j t
j| _| jS )Nkeep_english_letterchar_to_order_mapr precedence_matrixr z$%s confidence = %s, we have a winnerz9%s confidence = %s, below negative shortcut threshhold %s)r
filter_international_wordsstate enumerater CONTROLr SAMPLE_SIZEr r
r r r r DETECTINGSB_ENOUGH_REL_THRESHOLDget_confidencePOSITIVE_SHORTCUT_THRESHOLDloggerdebugFOUND_IT_stateNEGATIVE_SHORTCUT_THRESHOLDNOT_ME) r byte_strr! icorderr r
confidencer r r feedM sF
z SingleByteCharSetProber.feedc C sN d}| j dkrJd| jtj | j | jd }|| j | j }|dkrJd}|S )Ng{Gz?r g ?typical_positive_ratiogGz?)r r r POSITIVEr
r r )r rr r r r* | s
z&SingleByteCharSetProber.get_confidence)FN)__name__
__module__ __qualname__r' r) r+ r0 r r propertyr r r7 r*
__classcell__r r )r r r ! s
/r N)
charsetproberr enumsr r r r r r r r s