관리-도구
편집 파일: chardistribution.cpython-38.pyc
U ʗRe�% � @ s� d dl mZmZmZ d dlmZmZmZ d dlm Z m Z m Z d dl m Z mZmZ d dlmZmZmZ d dlmZ G dd� d�ZG d d � d e�ZG d d � d e�ZG d d� de�ZG dd� de�ZG dd� de�ZG dd� de�Z G dd� de�Z dS )� )�BIG5_CHAR_TO_FREQ_ORDER�BIG5_TABLE_SIZE� BIG5_TYPICAL_DISTRIBUTION_RATIO)�EUCKR_CHAR_TO_FREQ_ORDER�EUCKR_TABLE_SIZE� EUCKR_TYPICAL_DISTRIBUTION_RATIO)�EUCTW_CHAR_TO_FREQ_ORDER�EUCTW_TABLE_SIZE� EUCTW_TYPICAL_DISTRIBUTION_RATIO)�GB2312_CHAR_TO_FREQ_ORDER�GB2312_TABLE_SIZE�!GB2312_TYPICAL_DISTRIBUTION_RATIO)�JIS_CHAR_TO_FREQ_ORDER�JIS_TABLE_SIZE� JIS_TYPICAL_DISTRIBUTION_RATIO)�JOHAB_TO_EUCKR_ORDER_TABLEc @ sL e Zd ZdZdZdZdZdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dS )�CharDistributionAnalysisi g�G�z��?g{�G�z�?� c C s2 t � | _d | _d | _d | _d | _d | _| �� d S �N)�tuple�_char_to_freq_order� _table_size�typical_distribution_ratio�_done� _total_chars� _freq_chars�reset��self� r ��/builddir/build/BUILDROOT/alt-python38-pip-22.2.1-2.el8.x86_64/opt/alt/python38/lib/python3.8/site-packages/pip/_vendor/chardet/chardistribution.py�__init__> s z!CharDistributionAnalysis.__init__c C s d| _ d| _d| _dS )z reset analyser, clear any stateF� N)r r r r r r r r M s z CharDistributionAnalysis.resetc C sX |dkr| � |�}nd}|dkrT| jd7 _|| jk rTd| j| krT| jd7 _dS )z"feed a character with known length� ���r" r i N)� get_orderr r r r )r �char�char_len�orderr r r �feedV s z CharDistributionAnalysis.feedc C sT | j dks| j| jkr | jS | j | jkrN| j| j | j | j }|| jk rN|S | jS )z(return confidence based on existing datar" )r r �MINIMUM_DATA_THRESHOLD�SURE_NOr �SURE_YES)r �rr r r �get_confidenced s � z'CharDistributionAnalysis.get_confidencec C s | j | jkS r )r �ENOUGH_DATA_THRESHOLDr r r r �got_enough_datau s z(CharDistributionAnalysis.got_enough_datac C s dS )Nr$ r )r �_r r r r% z s z"CharDistributionAnalysis.get_orderN) �__name__� __module__� __qualname__r/ r, r+ r* r! r r) r. r0 r% r r r r r 8 s r c s$ e Zd Z� fdd�Zdd� Z� ZS )�EUCTWDistributionAnalysisc s t � �� t| _t| _t| _d S r )�superr! r r r r r r r �� __class__r r r! � s z"EUCTWDistributionAnalysis.__init__c C s, |d }|dkr(d|d |d d S dS )Nr" �� �^ r � r$ r �r �byte_str� first_charr r r r% � s z#EUCTWDistributionAnalysis.get_order�r2 r3 r4 r! r% � __classcell__r r r7 r r5 � s r5 c s$ e Zd Z� fdd�Zdd� Z� ZS )�EUCKRDistributionAnalysisc s t � �� t| _t| _t| _d S r �r6 r! r r r r r r r r7 r r r! � s z"EUCKRDistributionAnalysis.__init__c C s, |d }|dkr(d|d |d d S dS )Nr" � r: r r; r$ r r< r r r r% � s z#EUCKRDistributionAnalysis.get_orderr? r r r7 r rA � s rA c s$ e Zd Z� fdd�Zdd� Z� ZS )�JOHABDistributionAnalysisc s t � �� t| _t| _t| _d S r rB r r7 r r r! � s z"JOHABDistributionAnalysis.__init__c C s@ |d }d| kr dk r<n n |d |d }t �|d�S dS )Nr" � �� � r r$ )r �get)r r= r> �coder r r r% � s z#JOHABDistributionAnalysis.get_orderr? r r r7 r rD � s rD c s$ e Zd Z� fdd�Zdd� Z� ZS )�GB2312DistributionAnalysisc s t � �� t| _t| _t| _d S r )r6 r! r r r r r r r r7 r r r! � s z#GB2312DistributionAnalysis.__init__c C s: |d |d }}|dkr6|dkr6d|d | d S dS )Nr" r rC r; r: r$ r �r r= r> � second_charr r r r% � s z$GB2312DistributionAnalysis.get_orderr? r r r7 r rJ � s rJ c s$ e Zd Z� fdd�Zdd� Z� ZS )�Big5DistributionAnalysisc s t � �� t| _t| _t| _d S r )r6 r! r r r r r r r r7 r r r! � s z!Big5DistributionAnalysis.__init__c C sR |d |d }}|dkrN|dkr:d|d | d d S d|d | d S dS ) Nr" r � r; � �? �@ r$ r rK r r r r% � s z"Big5DistributionAnalysis.get_orderr? r r r7 r rM � s rM c s$ e Zd Z� fdd�Zdd� Z� ZS )�SJISDistributionAnalysisc s t � �� t| _t| _t| _d S r �r6 r! r r r r r r r r7 r r r! � s z!SJISDistributionAnalysis.__init__c C s� |d |d }}d| kr&dkr8n nd|d }n.d| krLdkrbn nd|d d }nd S || d }|d kr~d }|S ) Nr" r � � � �� �� � r$ rQ � r )r r= r> rL r( r r r r% � s z"SJISDistributionAnalysis.get_orderr? r r r7 r rR � s rR c s$ e Zd Z� fdd�Zdd� Z� ZS )�EUCJPDistributionAnalysisc s t � �� t| _t| _t| _d S r rS r r7 r r r! � s z"EUCJPDistributionAnalysis.__init__c C s, |d }|dkr(d|d |d d S dS )Nr" � r: r; r r$ r )r r= r& r r r r% � s z#EUCJPDistributionAnalysis.get_orderr? r r r7 r r[ � s r[ N) �big5freqr r r � euckrfreqr r r � euctwfreqr r r � gb2312freqr r r �jisfreqr r r � johabfreqr r r5 rA rD rJ rM rR r[ r r r r �<module> s J