U
<_9Y � @ s4 d dl mZ d dlmZmZmZ G dd� de�ZdS )� )�
CharSetProber)�CharacterCategory� ProbingState�SequenceLikelihoodc 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_char�reset)�self�model�reversedZ name_prober�� __class__� �H/opt/alt/python38/lib/python3.8/site-packages/chardet/sbcharsetprober.pyr ' s z SingleByteCharSetProber.__init__c s: t t| ��� d| _dgt�� | _d| _d| _d| _ d S )N� � )
r r r r
r Zget_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$SingleByteCharSetProber.charset_namec C s | j r| j jS | j�d�S d S )N�language)r r r
�getr r r r r F s z SingleByteCharSetProber.languagec C sj | j d s| �|�}|s | jS | j d }t|�D ]�\}}|| }|tjk rX| jd7 _|| jk r�| jd7 _| j | jk r�| j
d7 _
| j s�| j | j | }| j d | }n || j | j }| j d | }| j | d7 < || _ q0| j d }| jt
jk�rd| j
| jk�rd| �� }|| jk�r<| j�d||� t
j| _n(|| jk �rd| j�d||| j� t
j| _| jS )NZkeep_english_letter�char_to_order_mapr Zprecedence_matrixr z$%s confidence = %s, we have a winnerz9%s confidence = %s, below negative shortcut threshhold %s)r
Zfilter_international_words�state� enumerater ZCONTROLr � SAMPLE_SIZEr r
r r r r Z DETECTING�SB_ENOUGH_REL_THRESHOLD�get_confidence�POSITIVE_SHORTCUT_THRESHOLDZlogger�debugZFOUND_ITZ_state�NEGATIVE_SHORTCUT_THRESHOLDZNOT_ME) r Zbyte_strr �i�c�orderr r Z
confidencer r r �feedM sL
�
�z SingleByteCharSetProber.feedc C sN d}| j dkrJd| jtj | j | jd }|| j | j }|dkrJd}|S )Ng{�G�z�?r g �?Ztypical_positive_ratiog�G�z��?)r r r ZPOSITIVEr
r r )r �rr r r r% | s
��z&SingleByteCharSetProber.get_confidence)FN)�__name__�
__module__� __qualname__r# r$ r& r( r r �propertyr r r, r% �
__classcell__r r r r r ! s
/r N)Z
charsetproberr Zenumsr r r r r r r r �<module> s