관리-도구
편집 파일: sbcharsetprober.cpython-36.pyc
3 ��Ud� � @ sX d dl mZ ddlmZ ddlmZmZmZ eddddd d d d g�ZG d d� de�Z dS )� )� namedtuple� )� CharSetProber)�CharacterCategory� ProbingState�SequenceLikelihood�SingleByteCharSetModel� charset_name�language�char_to_order_map�language_model�typical_positive_ratio�keep_ascii_lettersZalphabetc 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_char�reset)�self�model�reversed� name_prober)� __class__� �:/tmp/pip-build-nvpq87xi/chardet/chardet/sbcharsetprober.pyr 3 s z SingleByteCharSetProber.__init__c s: t t| �j� d| _dgtj� | _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 s | j r| j jS | jjS d S )N)r r r )r r! r! r" r K s z$SingleByteCharSetProber.charset_namec C s | j r| j jS | jjS d S )N)r r r )r r! r! r" r R s z SingleByteCharSetProber.languagec C sN | j js| j|�}|s | jS | j j}| j j}x�|D ]�}|j|tj�}|tj k r\| j d7 _ || j k r�| j d7 _ | j | j k r�| jd7 _| js�|| j | }n|| | j }| j| d7 < || _ q2W | j j}| jtjk�rH| j| jk�rH| j� }|| jk�r | jjd||� tj| _n(|| jk �rH| jjd||| j� tj | _| jS )Nr z$%s confidence = %s, we have a winnerz9%s confidence = %s, below negative shortcut threshhold %s) r r �filter_international_words�stater r �getr � UNDEFINED�CONTROLr � SAMPLE_SIZEr r r r r r r � DETECTING�SB_ENOUGH_REL_THRESHOLD�get_confidence�POSITIVE_SHORTCUT_THRESHOLD�logger�debug�FOUND_IT�_state�NEGATIVE_SHORTCUT_THRESHOLD�NOT_ME) r �byte_strr r �char�orderZlm_catr � confidencer! r! r" �feedY sD z SingleByteCharSetProber.feedc C sL d}| j dkrHd| jtj | j | jj }|| j | j }|dkrHd}|S )Ng{�G�z�?r g �?g�G�z��?)r r r �POSITIVEr r r r )r �rr! r! r" r- � s z&SingleByteCharSetProber.get_confidence)FN)�__name__� __module__� __qualname__r* r, r. r3 r r �propertyr r r9 r- � __classcell__r! r! )r r" r - s 0r N) � collectionsr � charsetproberr �enumsr r r r r r! r! r! r" �<module> s