| 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�0 � @ s| d dl Z d dlZd dlZddlmZ ddlmZmZmZ ddl m
Z
ddlmZ ddl
mZ ddlmZ G d d
� d
e�ZdS )� N� )�CharSetGroupProber)�
InputState�LanguageFilter�ProbingState)�EscCharSetProber)�Latin1Prober)�MBCSGroupProber)�SBCSGroupProberc @ sj e Zd ZdZe�d�Ze�d�Ze�d�Zddddd d
ddd
�Z e
jfdd�Zdd� Z
dd� Zdd� ZdS )�UniversalDetectorg�������?s [�-�]s (|~{)s [�-�]zWindows-1252zWindows-1250zWindows-1251zWindows-1256zWindows-1253zWindows-1255zWindows-1254zWindows-1257)z
iso-8859-1z
iso-8859-2z
iso-8859-5z
iso-8859-6z
iso-8859-7z
iso-8859-8z
iso-8859-9ziso-8859-13c C sN d | _ g | _d | _d | _d | _d | _d | _|| _t� t
�| _d | _| �
� d S )N)�_esc_charset_prober�_charset_probers�result�done� _got_data�_input_state�
_last_char�lang_filter�loggingZ getLogger�__name__�logger�_has_win_bytes�reset)�selfr � r �=/usr/lib/python3.9/site-packages/chardet/universaldetector.py�__init__Q s zUniversalDetector.__init__c C sV d dd d�| _ d| _d| _d| _tj| _d| _| jr>| j� � | j
D ]}|� � qDd S )N� ��encoding�
confidence�languageF� )r r r r r �
PURE_ASCIIr r r r r
)r �proberr r r r ^ s
zUniversalDetector.resetc C s> | j r
d S t|�sd S t|t�s(t|�}| js�|�tj�rJdddd�| _nv|�tj tj
f�rldddd�| _nT|�d�r�dddd�| _n:|�d�r�d ddd�| _n |�tjtjf�r�d
ddd�| _d| _| jd d ur�d| _ d S | j
tjk�r.| j�|��rtj| _
n*| j
tjk�r.| j�| j| ��r.tj| _
|d
d � | _| j
tjk�r�| j�s^t| j�| _| j�|�tjk�r:| jj| j�� | jjd�| _d| _ n�| j
tjk�r:| j�s�t | j�g| _| jt!j"@ �r�| j�#t$� � | j�#t%� � | jD ]:}|�|�tjk�r�|j|�� |jd�| _d| _ �q&�q�| j&�|��r:d| _'d S )Nz UTF-8-SIG� �?� r zUTF-32s �� zX-ISO-10646-UCS-4-3412s ��zX-ISO-10646-UCS-4-2143zUTF-16Tr ���)(r �len�
isinstance� bytearrayr �
startswith�codecs�BOM_UTF8r �BOM_UTF32_LE�BOM_UTF32_BE�BOM_LE�BOM_BEr r r# �HIGH_BYTE_DETECTOR�search� HIGH_BYTE�ESC_DETECTORr Z ESC_ASCIIr r r �feedr ZFOUND_IT�charset_name�get_confidencer! r
r r ZNON_CJK�appendr
r �WIN_BYTE_DETECTORr )r Zbyte_strr$ r r r r6 o s�
�
��
�
�
�
��
�
zUniversalDetector.feedc C st | j r| jS d| _ | js&| j�d� n�| jtjkrBdddd�| _n�| jtjkr�d }d}d }| j D ]"}|sjq`|�
� }||kr`|}|}q`|r�|| jkr�|j}|j�
� }|�
� }|�d�r�| jr�| j�||�}|||jd�| _| j�� tjk�rn| jd d u �rn| j�d
� | j D ]`}|�s�qt|t��rP|jD ] }| j�d|j|j|�
� � �q,n| j�d|j|j|�
� � �q| jS )NTzno data received!�asciir% r&