| Server IP : 172.67.201.108 / Your IP : 216.73.216.55 Web Server : Apache/2.4.68 (Amazon Linux) OpenSSL/3.5.7 System : Linux ip-172-31-69-123.ec2.internal 6.1.177-224.371.amzn2023.x86_64 #1 SMP PREEMPT_DYNAMIC Mon Jul 27 20:28:29 UTC 2026 x86_64 User : ec2-user ( 1000) PHP Version : 8.4.24 Disable Function : NONE MySQL : OFF | cURL : ON | WGET : ON | Perl : ON | Python : OFF | Sudo : ON | Pkexec : ON Directory : /lib/python3.9/site-packages/elementpath/regex/__pycache__/ |
Upload File : |
a
���c"K � a @ s d Z ddlmZ ddlmZmZ ddlmZmZ ddl m
Z
ddl m Z m
Z
mZmZmZ G dd � d e�Zdxeee d �d d
�ZG dd� de�Zdd� e
�� D �Zed�ed�ed�ed�ed�ed�ed�ed�ed�ed�ed �ed �ed �ed �ed �ed!�ed"�ed#�ed$�ed%�ed&�ed'�ed(�ed)�ed*�ed+�ed,�ed-�ed.�ed/�ed0�ed1�ed2�ed3�ed4�ed5�ed6�ed7�ed8�ed9�ed:�ed;�ed<�ed=�ed>�ed?�ed@�edA�edB�edC�edD�edE�edF�edG�edH�edI�edJ�edK�edL�edM�edN�edO�edP�edQ�edR�edS�edT�edU�edV�edW�edX�edY�edZ�ed[�ed\�ed]�ed^�ed_�ed`�eda�edb�edc�edd�ede�edf�edg�edh�edi�edj�edk�edl�edm�edn�edo�edp�edq�dr�`Zeds �dt�f dudv� ZdwS )yz>
This module defines Unicode character categories and blocks.
� )�
maxunicode)�Iterable�
MutableSet)�Iterator�List� )�RAW_UNICODE_CATEGORIES)� CodePoint�code_point_order�code_point_repr�iter_code_points�get_code_point_rangec @ s e Zd ZdZdS )�
RegexErrorz�
Error in a regular expression or in a character class specification.
This exception is derived from `Exception` base class and is raised
only by the regex subpackage.
N)�__name__�
__module__� __qualname__�__doc__� r r �E/usr/lib/python3.9/site-packages/elementpath/regex/unicode_subsets.pyr s r F)�s�returnc
c s d}d}d}t | �}ttt | ���}|D �]�}|dkr�| d }|dkrLd}nL|dv rj|dkrjtd| ��n.|rzt|�V n |d ks�| d d
kr�t|�V q(| | d
k�r�|s�||d kr�| | }t|�V d}�q�|r�| | }t|�V d}n�d}t|�}| | }|dk�rt||d k �rt| |d d v �r>t|�}| | }n6| |d d v �rtd
} t| || |d |d | f ��t|�t|�k�r�d} t| |||d | f ��n>|�r�tt|�d t|�d �E dH nt|�t|�d fV q(| | dv �r|�r�d}d}| | }t|�V q(| | dv �r�|�sF|dk�rFtd| | |f ��d }}| | }||d k�sv| |d d
k�r�t|�V q(| | dk�r�|�r�d }}d}t|�V nd}q(|�r�d}td�V d}| | }||d k�s�| |d d
kr(t|�V q(|�rtd�V dS )a�
Parses a regex character subset, generating a sequence of code points
and code points ranges. An unescaped hyphen (-) that is not at the
start or at the and is interpreted as range specifier.
:param s: a string representing the character subset.
:param expand_ranges: if set to `True` then expands character ranges.
:return: yields integers or couples of integers.
F� r �\Tz[]r z bad character %r at position 0� �-z
-|.^?*+{}()[]Z sSdDiIcCwWpPz/bad character range '%s-\%s' at position %d: %rz.bad character range '%s-%s' at position %d: %rNz
|.^?*+{}()z bad character %r at position %d)�len�iter�ranger �ord�next)
r Z
expand_rangesZescapedZon_range�charZlengthZsubset_index_iterator�kZend_char�msgr r r �iterparse_character_subset s�
"
r# c @ s� e Zd ZU dZdZee ed< d9dd�Ze dd� �Z
d d
� Z d d � Z d
d� Z
dd� Zdd� Zdd� Zdd� Zdd� Zdd� Zdd � Zd d � Zd d � Zd!d"� Zd#d$� Zd%d&� Zd'd(� Zd)d*� Zd+d,� Z d-d.� Z d/d0� Z e Z d1d2� Z d3d4� Z!d5d6� Z"d7d8� Z#dS ):�
UnicodeSubseta�
Represents a subset of Unicode code points, implemented with an ordered list of
integer values and ranges. Codepoints can be added or discarded using sequences
of integer values and ranges or with strings equivalent to regex character set.
:param codepoints: a sequence of integer values and ranges, another UnicodeSubset instance ora a string equivalent of a regex character set.
�� _codepointsr& Nc C sV |st � | _nDt|t �r(t|td�| _n*t|t�r@|j�� | _nt � | _| �|� d S )N)�key) �listr& �
isinstance�sortedr
r$ �
codepoints�copy�update)�selfr+ r r r �__init__� s
zUnicodeSubset.__init__c C s | j S �Nr% �r. r r r r+ � s zUnicodeSubset.codepointsc C s d| j jt| �f S )Nz%s(%r))� __class__r �strr1 r r r �__repr__� s zUnicodeSubset.__repr__c C s d� dd� | jD ��S )Nr c s s | ]}t |�V qd S r0 )r )�.0�cpr r r � <genexpr>� � z(UnicodeSubset.__str__.<locals>.<genexpr>)�joinr& r1 r r r �__str__� s zUnicodeSubset.__str__c C s | � � S r0 )�__copy__r1 r r r r, � s zUnicodeSubset.copyc C s
t | j�S r0 )r$ r& r1 r r r r; � s zUnicodeSubset.__copy__c c sB t | j�D ]2}t|t�r |V q
t t|d |d ��E d H q
d S �Nr r )�reversedr& r) �intr )r. �itemr r r � __reversed__� s
zUnicodeSubset.__reversed__c c s� d}| j D ]�}t|t�r$||d f}|d | }|dkrH||d fV n<|dkrb|V |d V n"|dkrr|V n|r�td�| ���|d }q
|tk r�|td fV n|tkr�tV d S )Nr r r z#unordered code points found in {!r})r& r) r> �
ValueError�formatr )r. Zlast_cpr6 �diffr r r �
complement� s&
zUnicodeSubset.complementc C s t t| �� �S r0 )�map�chr�__iter__r1 r r r �iter_characters� s z UnicodeSubset.iter_charactersc C s� t |t�s,z t|�}W n ty* Y dS 0 | jD ]T}t |t�sj|d |krR dS |d |krbq2q� dS q2||krx dS ||kr2 dS q2dS )NFr r T)r) r> r � TypeErrorr& )r. �valuer6 r r r � __contains__� s"
zUnicodeSubset.__contains__c c s0 | j D ]$}t|t�r |V qt|� E d H qd S r0 )r& r) r> r )r. r6 r r r rG � s
zUnicodeSubset.__iter__c C s d}| D ] }|d7 }q|S r< r )r. r! �_r r r �__len__� s
zUnicodeSubset.__len__c G sV |D ]L}t |t�r4tt|�dd�D ]}| �|� q"qt|dd�D ]}| �|� q@qd S �NT)�reverse)r) r3 r r# �add�r. ZothersrJ r6 r r r r- � s
zUnicodeSubset.updatec
C s� zt |�\}}W n ty0 td�|���Y n0 | j}t|�d }t|�D �]\}}t|t�rl||d f}||d k r�|� |t|t�r�|nt
|�� n�||d kr�qLn�||d k�r8||kr�t |d |�|f||< n`||d }t|t�r�|n|d } || k�rt |d |�|f||< n t |d |�| f||< | }qLn ||d k �rV||d f||< �q~qL| j� t|t
��rxt
|�n|� d S �Nz,{!r} is not a Unicode code point value/ranger r )r
rI rA rB r& r � enumerater) r> �insert�tuple�min�appendr( )
r. rJ Z start_valueZ end_value� code_pointsZ
last_indexr! r6 Znext_cpZ higher_boundr r r rP � s6
zUnicodeSubset.addc G sV |D ]L}t |t�r4tt|�dd�D ]}| �|� q"qt|dd�D ]}| �|� q@qd S rN )r) r3 r r# �discardrQ r r r �difference_update
s
z UnicodeSubset.difference_updatec C s� zt |�\}}W n ty0 td�|���Y n0 | j}ttt|���D �]J}|| }t|t �rl||d f}||d kr� �q�qH||d kr�||d kr�||= n.||d dkr�|d |f||< n |d ||< qH||d krH||d k�r |d | dk�r||d f||< n|d d ||< qH|d | dk�rL|�
|d ||d f� n|�
|d |d d � ||d dk�r�|d |f||< qH|d ||< qHd S rR ) r
rI rA rB r& r= r r r) r> rT )r. rJ Zstart_cpZend_cprX r! r6 r r r rY s8
zUnicodeSubset.discardc C s | j d d �= d S r0 r% r1 r r r �clear= s zUnicodeSubset.clearc C s2 t |t�stS t |t�r$| j|jkS | j|kS d S r0 )r) r �NotImplementedr$ r& )r. �otherr r r �__eq__@ s
zUnicodeSubset.__eq__c C sb t |t�stS t |t�r$t|j�}n&t |t�r>tt|�j�}n t|dd�}|D ]}| �|� qN| S rN ) r) r r\ r$ r= r& r3 r rP �r. r] r6 r r r �__ior__H s
zUnicodeSubset.__ior__c C s | � � }|�|�S r0 )r, r` �r. r] �objr r r �__or__V s zUnicodeSubset.__or__c C sb t |t�stS t |t�r$t|j�}n&t |t�r>tt|�j�}n t|dd�}|D ]}| �|� qN| S rN ) r) r r\ r$ r= r& r3 r rY r_ r r r �__isub__Z s
zUnicodeSubset.__isub__c C s | � � }|�|�S r0 )r, rd ra r r r �__sub__h s zUnicodeSubset.__sub__c C s | | D ]}| � |� q| S r0 )rY �r. r] rJ r r r �__iand__n s zUnicodeSubset.__iand__c C s | � � }|�|�S r0 )r, rg ra r r r �__and__s s zUnicodeSubset.__and__c C s` || u r| � � | S t|t�s"tS t|t�s4t|�}|D ]"}|| v rP| �|� q8| �|� q8| S r0 )r[ r) r r\ r$ rY rP rf r r r �__ixor__w s
zUnicodeSubset.__ixor__c C s | � � }|�|�S r0 )r, ri ra r r r �__xor__� s zUnicodeSubset.__xor__)N)$r r r r � __slots__r r �__annotations__r/ �propertyr+ r4 r: r, r; r@ rD rH rK rG rM r- rP rZ rY r[ r^ r` rc rd re �__rsub__rg rh ri rj r r r r r$ v s>
" 'r$ c C s i | ]\}}|t |��qS r )r$ )r5 r! �vr r r �
<dictcomp>� r8 rp z -u -ÿu Ā-ſu ƀ-ɏu ɐ-ʯu ʰ-˿u ̀-ͯu Ͱ-Ͽu Ѐ-ӿu -֏u -u -ۿu ܀-ݏu ހ-u ऀ-ॿu ঀ-u -u -૿u -u -u ఀ-౿u ಀ-u ഀ-ൿu -u -u -u ༀ-u က-႟u Ⴀ-ჿu ᄀ-ᇿu ሀ-u Ꭰ-u ᐀-ᙿu -u ᚠ-u ក-u ᠀-u Ḁ-ỿu ἀ-u -u ⁰-u ₠-u ⃐-u ℀-⅏u ⅐-u ←-⇿u ∀-⋿u ⌀-⏿u ␀-u ⑀-u ①-⓿u ─-╿u ▀-▟u ■-◿u ☀-⛿u ✀-➿u ⠀-⣿u ⺀-u ⼀-u ⿰-⿿u -〿u -ゟu ゠-ヿu -ㄯu -u ㆐-㆟u ㆠ-ㆿu ㈀-㋿u ㌀-㏿u 㐀-䶵u 一-鿿u ꀀ-u ꒐-u 가-힣u �-�u �-�u �-�u ---u 豈-u ff-ﭏu ﭐ-﷿u ︠-︯u ︰-﹏u ﹐-u ﹰ-u
-�u -u 𐌀-𐌯u 𐌰-u 𐐀-𐑏u 𝀀-u 𝄀-u 𝐀-𝟿u 𠀀-𪛖u 丽-u -)`Z IsBasicLatinzIsLatin-1SupplementzIsLatinExtended-AzIsLatinExtended-BZIsIPAExtensionsZIsSpacingModifierLettersZIsCombiningDiacriticalMarksZIsGreekZ
IsCyrillicZ
IsArmenianZIsHebrewZIsArabicZIsSyriacZIsThaanaZ IsDevanagariZ IsBengaliZ
IsGurmukhiZ
IsGujaratiZIsOriyaZIsTamilZIsTeluguZ IsKannadaZ IsMalayalamZ IsSinhalaZIsThaiZIsLaoZ IsTibetanZ IsMyanmarZ
IsGeorgianZ IsHangulJamoZ
IsEthiopicZ
IsCherokeeZ$IsUnifiedCanadianAboriginalSyllabicsZIsOghamZIsRunicZIsKhmerZ IsMongolianZIsLatinExtendedAdditionalZIsGreekExtendedZIsGeneralPunctuationZIsSuperscriptsandSubscriptsZIsCurrencySymbolsZIsCombiningMarksforSymbolsZIsLetterlikeSymbolsZ
IsNumberFormsZIsArrowsZIsMathematicalOperatorsZIsMiscellaneousTechnicalZIsControlPicturesZ IsOpticalCharacterRecognitionZIsEnclosedAlphanumericsZ IsBoxDrawingZIsBlockElementsZIsGeometricShapesZIsMiscellaneousSymbolsZ
IsDingbatsZIsBraillePatternsZIsCJKRadicalsSupplementZIsKangxiRadicalsZ"IsIdeographicDescriptionCharactersZIsCJKSymbolsandPunctuationZ
IsHiraganaZ
IsKatakanaZ
IsBopomofoZIsHangulCompatibilityJamoZIsKanbunZIsBopomofoExtendedZ IsEnclosedCJKLettersandMonthsZIsCJKCompatibilityZ IsCJKUnifiedIdeographsExtensionAZIsCJKUnifiedIdeographsZ
IsYiSyllablesZ IsYiRadicalsZIsHangulSyllablesZIsHighSurrogatesZIsHighPrivateUseSurrogatesZIsLowSurrogates� IsPrivateUseZ IsCJKCompatibilityIdeographsZ IsAlphabeticPresentationFormszIsArabicPresentationForms-AZIsCombiningHalfMarksZIsCJKCompatibilityFormsZIsSmallFormVariantszIsArabicPresentationForms-BZ
IsSpecialsZ IsHalfwidthandFullwidthFormsZ IsOldItalicZIsGothicZ IsDeseretZIsByzantineMusicalSymbolsZIsMusicalSymbolsZ!IsMathematicalAlphanumericSymbolsZ IsCJKUnifiedIdeographsExtensionBZ&IsCJKCompatibilityIdeographsSupplementZIsTagsrq u -c C sd | � d�r6z
t| W S ty2 td| ��Y q`0 n*z
t| W S ty^ td| ��Y n0 d S )NZIsz&%r doesn't match to any Unicode block.z)%r doesn't match to any Unicode category.)�
startswith�UNICODE_BLOCKS�KeyErrorr �UNICODE_CATEGORIES)�namer r r �unicode_subset� s
rw N)F)r �sysr �collections.abcr r �typingr r Zunicode_categoriesr r+ r r
r r r
� Exceptionr r3 r# r$ �itemsru rs r- rw r r r r �<module>
s� X �c