U
|
L ±d0' ã @ s d dl mZmZ ddlmZmZmZ ddlmZm Z m
|
Z
|
ddl m Z m Z mZ ddlmZmZmZ ddlmZmZmZ ddlmZ G d d
|
d
|
ZG d d d eZG d d deZG dd deZ G dd deZ G dd deZ G dd deZ G dd deZ dS )é )ÚTupleÚUnioné )ÚBIG5_CHAR_TO_FREQ_ORDERÚBIG5_TABLE_SIZEÚ BIG5_TYPICAL_DISTRIBUTION_RATIO)ÚEUCKR_CHAR_TO_FREQ_ORDERÚEUCKR_TABLE_SIZEÚ EUCKR_TYPICAL_DISTRIBUTION_RATIO)ÚEUCTW_CHAR_TO_FREQ_ORDERÚEUCTW_TABLE_SIZEÚ EUCTW_TYPICAL_DISTRIBUTION_RATIO)ÚGB2312_CHAR_TO_FREQ_ORDERÚGB2312_TABLE_SIZEÚ!GB2312_TYPICAL_DISTRIBUTION_RATIO)ÚJIS_CHAR_TO_FREQ_ORDERÚJIS_TABLE_SIZEÚ JIS_TYPICAL_DISTRIBUTION_RATIO)ÚJOHAB_TO_EUCKR_ORDER_TABLEc @ s e Zd ZdZdZdZdZddddZddd d
|
Ze e
|
e f e dd d d Z edddZedddZe e
|
e f e dddZdS )ÚCharDistributionAnalysisi g®Gáz®ï?g{®Gáz?é N©Úreturnc C s2 t | _d| _d| _d| _d| _d| _| ¡ d S )Nr g F)ÚtupleÚ_char_to_freq_orderÚ _table_sizeÚtypical_distribution_ratioÚ_doneÚ _total_charsÚ _freq_charsÚreset©Úself© r# úOd:\z\workplace\vscode\pyvenv\venv\Lib\site-packages\chardet/chardistribution.pyÚ__init__@ s z!CharDistributionAnalysis.__init__c C s d| _ d| _d| _dS )z reset analyser, clear any stateFr N)r r r r! r# r# r$ r O s z CharDistributionAnalysis.reset)ÚcharÚchar_lenr c C sX |dkr| |¡}nd}|dkrT| jd7 _|| jk rTd| j| krT| jd7 _dS )z"feed a character with known lengthé éÿÿÿÿr r i N)Ú get_orderr r r r )r" r& r' Úorderr# r# r$ ÚfeedX 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 ÚMINIMUM_DATA_THRESHOLDÚSURE_NOr ÚSURE_YES)r" Úrr# r# r$ Úget_confidencef s ÿ
|
z'CharDistributionAnalysis.get_confidencec C s | j | jkS ©N)r ÚENOUGH_DATA_THRESHOLDr! r# r# r$ Úgot_enough_dataw s z(CharDistributionAnalysis.got_enough_data)Ú_r c C s dS )Nr) r# )r" r5 r# r# r$ r* | s z"CharDistributionAnalysis.get_order)Ú__name__Ú
|
__module__Ú __qualname__r3 r/ r. r- r% r r ÚbytesÚ bytearrayÚintr, Úfloatr1 Úboolr4 r* r# r# r# r$ r : s r c s: e Zd Zdd fdd Zeeef edddZ Z S )ÚEUCTWDistributionAnalysisNr c s t ¡ t| _t| _t| _d S r2 )Úsuperr% r r r r r r r! ©Ú __class__r# r$ r%
s
|
z"EUCTWDistributionAnalysis.__init__©Úbyte_strr c C s, |d }|dkr(d|d |d d S dS )Nr éÄ é^ r é¡ r) r# ©r" rC Ú
|
first_charr# r# r$ r* s z#EUCTWDistributionAnalysis.get_order©
|
r6 r7 r8 r% r r9 r: r; r* Ú __classcell__r# r# r@ r$ r> s r> c s: e Zd Zdd fdd Zeeef edddZ Z S )ÚEUCKRDistributionAnalysisNr c s t ¡ t| _t| _t| _d S r2 ©r? r% r r r r r
|
r r! r@ r# r$ r% s
|
z"EUCKRDistributionAnalysis.__init__rB c C s, |d }|dkr(d|d |d d S dS )Nr é° rE r rF r) r# rG r# r# r$ r* s z#EUCKRDistributionAnalysis.get_orderrI r# r# r@ r$ rK s rK c s: e Zd Zdd fdd Zeeef edddZ Z S )ÚJOHABDistributionAnalysisNr c s t ¡ t| _t| _t| _d S r2 rL r! r@ r# r$ r% © s
|
z"JOHABDistributionAnalysis.__init__rB c C s@ |d }d| kr dk r<n n |d |d }t |d¡S dS )Nr é éÔ é r r) )r Úget)r" rC rH Úcoder# r# r$ r* ¯ s
|
z#JOHABDistributionAnalysis.get_orderrI r# r# r@ r$ rN ¨ s rN c s: e Zd Zdd fdd Zeeef edddZ Z S )ÚGB2312DistributionAnalysisNr c s t ¡ t| _t| _t| _d S r2 )r? r% r r r r r r r! r@ r# r$ r% ¸ s
|
z#GB2312DistributionAnalysis.__init__rB c C s: |d |d }}|dkr6|dkr6d|d | d S dS )Nr r rM rF rE r) r# ©r" rC rH Ú second_charr# r# r$ r* ¾ s z$GB2312DistributionAnalysis.get_orderrI r# r# r@ r$ rT · s rT c s: e Zd Zdd fdd Zeeef edddZ Z S )ÚBig5DistributionAnalysisNr c s t ¡ t| _t| _t| _d S r2 )r? r% r r r r r r r! r@ r# r$ r% Ê s
|
z!Big5DistributionAnalysis.__init__rB c C sR |d |d }}|dkrN|dkr:d|d | d d S d|d | d S dS ) Nr r é¤ rF é é? é@ r) r# rU r# r# r$ r* Ð s z"Big5DistributionAnalysis.get_orderrI r# r# r@ r$ rW É s rW c s: e Zd Zdd fdd Zeeef edddZ Z S )ÚSJISDistributionAnalysisNr c s t ¡ t| _t| _t| _d S r2 ©r? r% r r r r r r r! r@ r# r$ r% Þ s
|
z!SJISDistributionAnalysis.__init__rB 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 é é é¼ éà éï é r) r[ é r# )r" rC rH rV r+ r# r# r$ r* ä s z"SJISDistributionAnalysis.get_orderrI r# r# r@ r$ r\ Ý s r\ c s: e Zd Zdd fdd Zeeef edddZ Z S )ÚEUCJPDistributionAnalysisNr c s t ¡ t| _t| _t| _d S r2 r] r! r@ r# r$ r% ÷ s
|
z"EUCJPDistributionAnalysis.__init__rB c C s, |d }|dkr(d|d |d d S dS )Nr é rE rF r r) r# )r" rC r& r# r# r$ r* ý s z#EUCJPDistributionAnalysis.get_orderrI r# r# r@ r$ re ö s re N)!Útypingr r Úbig5freqr r r Ú euckrfreqr r r
|
Ú euctwfreqr r r Ú
|
gb2312freqr r r Újisfreqr r r Z johabfreqr r r> rK rN rT rW r\ re r# r# r# r$ Ú<module> s J
|
