o
ai @ sN d dl mZ ddlmZ ddlmZmZmZ edg dZG dd deZ d S )
)
namedtuple )
CharSetProber)CharacterCategory ProbingStateSequenceLikelihoodSingleByteCharSetModel) charset_namelanguagechar_to_order_maplanguage_modeltypical_positive_ratiokeep_ascii_lettersalphabetc 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| || _|| _|| _d | _d | _d | _d | _ d | _
| 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-python310-pip-21.3.1-5.el8.x86_64/opt/alt/python310/lib/python3.10/site-packages/pip/_vendor/chardet/sbcharsetprober.pyr 3 s z SingleByteCharSetProber.__init__c s: t t| d| _dgt | _d| _d| _d| _ d S )N r )
r r r r r get_num_categoriesr r r r r r" r$ r% r A s
z SingleByteCharSetProber.resetc C | j r| j jS | jjS r )r r r r( r$ r$ r% r K z$SingleByteCharSetProber.charset_namec C r) r )r r
r r( r$ r$ r% r
R r* z SingleByteCharSetProber.languagec C sF | j js | |}|s| jS | j j}| j j}|D ]L}||tj}|tj k r-| j
d7 _
|| j k ra| j d7 _ | j
| j k ra| jd7 _| jsQ|| j
| }n|| | j
}| j| d7 < || _
q| j j}| jtjkr| j| jkr| }|| jkr| jd|| tj| _| jS || jk r| jd||| j tj | _| jS )Nr z$%s confidence = %s, we have a winnerz9%s confidence = %s, below negative shortcut threshhold %s) r r filter_international_wordsstater r getr UNDEFINEDCONTROLr SAMPLE_SIZEr r r r r r r DETECTINGSB_ENOUGH_REL_THRESHOLDget_confidencePOSITIVE_SHORTCUT_THRESHOLDloggerdebugFOUND_IT_stateNEGATIVE_SHORTCUT_THRESHOLDNOT_ME) r byte_strr r charorderlm_catr
confidencer$ r$ r% feedY sJ
z SingleByteCharSetProber.feedc C sL d}| j dkr$d| jtj | j | jj }|| j | j }|dkr$d}|S )Ng{Gz?r g ?gGz?)r r r POSITIVEr r
r r )r rr$ r$ r% r3 s
z&SingleByteCharSetProber.get_confidence)FN)__name__
__module__ __qualname__r0 r2 r4 r9 r r propertyr r
r@ r3
__classcell__r$ r$ r" r% r - s
0r N)
collectionsr
charsetproberr enumsr r r r r r$ r$ r$ r% s