JFIF $
d d 7
Viewing File: /usr/lib/python3.9/site-packages/chardet/__pycache__/charsetprober.cpython-39.opt-1.pyc
a
��=b� � @ 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 ]@}|�|dd� � |dd� }|�� sL|dk rLd}|�|� q|S )u9
We define three types of bytes:
alphabet: english alphabets [a-zA-Z]
international: international characters [-ÿ]
marker: everything else [^a-zA-Z-ÿ]
The input buffer can be thought to contain a series of words delimited
by markers. This function works to filter all words that contain at
least one international character. All contiguous sequences of markers
are replaced by a single space ascii character.
This filter applies to all scripts which do not use English characters.
s% [a-zA-Z]*[�-�]+[a-zA-Z]*[^a-zA-Z�-�]?N���� �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 )
a�
Returns a copy of ``buf`` that retains only the sequences of English
alphabet and high byte characters that are not between <> characters.
Also retains English alphabet and high byte characters immediately
before occurrences of >.
This filter can be applied to all scripts which contain both English
characters and extended ASCII characters, but is currently only used by
``Latin1Prober``.
Fr r � >� <Tr r N)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 Back to Directory
nL+D550H?Mx ,D"v]qv;6*Zqn)ZP 0 !1 A "#a$2Qr D8 a Ri[f\mIykIw0 cuFcRı?lO7к_f˓[C$殷WF<_W ԣs KcëIzyQy /_LKℂ;C",pFA: /]=H ~,ls/9ć:[=/#f;)x {ٛEQ )~ =𘙲r*2~ a _V=' kumF D}KYYC)({ *g&f`툪ry`=^cJ.I](*`wq 1dđ#̩͑0;H]u搂@:~וKL Nsh}OIR*8:2 !lDJVo( 3=M(zȰ+i*NA r6KnSl )!JJӁ* %݉?|D}d5:eP0R;{$X'xF@.ÊB {,WJuQɲRI;9QE琯62fT.DUJ;*cP A\ILNj!J۱+O\ ͔]ޒ SJȧc%ANolՎprULZԛerE2=XDXgVQeӓk yP7U*omQIs,K`)6\G3t?pgjrmۛجwluGtf h9uyP0D;Uڽ"OXlif$)&|ML0Zrm1[HXPlPR0'G=i2N +0e2]] 9VTPO7h(F*癈 '=QVZDF,d߬~TX
G[`le69CR(!S2!P
<0x < !1AQ "Raq02Br#SCTb
?Ζ"]mH5WR7k.ۛ!}Q~+yԏz|@T20S~Kek *zFf^2X*(@8r?CIuI|֓>^ExLgNUY+{.RѪ
τVYTD I62'8Y2 7'\TP.6d&˦@Vqi|8- OΕ]ʔ U=TL8=;6c| !qfF3aů&~$l}'NWUs$Uk^SV:U# 6w++s&r+nڐ{@29 gL
u"TÙM=6 (^"7r} =6YݾlCuhquympǦ
G jhsǜNlɻ}o7 #S6aw4!OSrD57%|?x>L
|/nD6?/8w#[)L7+6〼T ATg ! %5MmZ/c-{1_Je"|^$'O&ޱմ Trb$ w)R$&
N1EtdU3Uȉ1pM" N*(DN yd96.(jQ)X
5cQɎMyW? Q*!R>6=7)Xj5`J]e 8%t!+'!1Q 5 !1 AQaqё#2"0BRb ? Gt^## .llQT $v,,m 㵜5 ubV =sY+@d{N! dnO< .-B; _wJt6;QJd.Qc%p{ 1,sNDdFHI0ГoXшe黅XۢF:)[FGXƹ/w_cMeD,ʡcc. WDtA$j@:) -# u
c1<@ۗ9F)KJ-hpP]_x[qBlbp ʖw q"LFGd ƶ*s+ډ_Zc"? %t[IP 6J]#=ɺVvvCGsGh1 >)6|ey?Lӣm,4GWUi` ]uJVoV DG< SB 6ϏQ@ TiUly OU0kfV~~}SZ@*WUUi##; s/[ =!7}" WN]'(L! ~y5g9T̅JkbM' +s:S + B)v@Mj
e Cf jE 0Y\QnzG1д~Wo{T 9?`Rmyhsy3!HAD]mc1~2L Su7xT;j$`}4->L#vzŏ ILS ֭ T { rjGKC; bpU= -`BsK.SFw4Mq]ZdHS0)tLg