| Server IP : 127.0.0.1 / Your IP : 216.73.216.109 Web Server : Apache/2.4.54 (Win64) OpenSSL/1.1.1q PHP/8.1.10 System : Windows NT DESKTOP-E5T4RUN 10.0 build 19045 (Windows 10) AMD64 User : SERVERWEB ( 0) PHP Version : 8.1.10 Disable Function : NONE MySQL : OFF | cURL : ON | WGET : OFF | Perl : OFF | Python : OFF | Sudo : OFF | Pkexec : OFF Directory : C:/cygwin64/lib/python3.9/site-packages/chardet/__pycache__/ |
Upload File : |
a
3<�a� � @ s0 d dl Z d dlZddlmZ G dd� de�ZdS )� N� )� ProbingStatec @ sn e Zd ZdZddd�Zdd� Zedd� �Zd d
� Zed d � �Z d
d� Z
e dd� �Z e dd� �Z
e dd� �ZdS )�
CharSetProbergffffff�?Nc C s d | _ || _t�t�| _d S �N)�_state� lang_filter�loggingZ getLogger�__name__�logger)�selfr � r �9/usr/lib/python3.9/site-packages/chardet/charsetprober.py�__init__' s zCharSetProber.__init__c C s t j| _d S r )r Z DETECTINGr �r r r r
�reset, s zCharSetProber.resetc C s d S r r r r r r
� charset_name/ s zCharSetProber.charset_namec C s d S r r )r �bufr r r
�feed3 s zCharSetProber.feedc C s | j S r )r r r r r
�state6 s zCharSetProber.statec C s dS )Ng r r r r r
�get_confidence: s z CharSetProber.get_confidencec C s t �dd| �} | S )Ns ([ -])+� )�re�sub)r r r r
�filter_high_byte_only= s z#CharSetProber.filter_high_byte_onlyc C s\ t � }t�d| �}|D ]@}|�|d d� � |dd � }|�� sL|dk rLd}|�|� q|S )Ns% [a-zA-Z]*[�-�]+[a-zA-Z]*[^a-zA-Z�-�]?���� �r )� bytearrayr �findall�extend�isalpha)r �filtered�wordsZwordZ last_charr r r
�filter_international_wordsB s � z(CharSetProber.filter_international_wordsc C s� t � }d}d}tt| ��D ]n}| ||d � }|dkr<d}n |dkrHd}|dk r|�� s||kr�|s�|�| ||� � |�d� |d }q|s�|�| |d � � |S ) NFr r � >� <Tr r )r �range�lenr r )r r Zin_tag�prevZcurrZbuf_charr r r
�filter_with_english_lettersg s"
z)CharSetProber.filter_with_english_letters)N)r �
__module__� __qualname__ZSHORTCUT_THRESHOLDr r �propertyr r r r � staticmethodr r" r( r r r r
r # s
$r )r r Zenumsr �objectr r r r r
�<module> s