|
Server : Apache System : Linux iad1-shared-b8-43 6.6.49-grsec-jammy+ #10 SMP Thu Sep 12 23:23:08 UTC 2024 x86_64 User : dh_edsupp ( 6597262) PHP Version : 8.2.26 Disable Function : NONE Directory : /lib/python3/dist-packages/future/backports/__pycache__/ |
Upload File : |
o
,�]W? � @ sX d Z ddlZe�d�jZe�d�jZe�d�Ze�d�Ze�d�Z[G dd � d e �Z
dS )
a Shared support for scanning document type declarations in HTML and XHTML.
Backported for python-future from Python 3.3. Reason: ParserBase is an
old-style class in the Python 2.7 source of markupbase.py, which I suspect
might be the cause of sporadic unit-test failures on travis-ci.org with
test_htmlparser.py. The test failures look like this:
======================================================================
ERROR: test_attr_entity_replacement (future.tests.test_htmlparser.AttributesStrictTestCase)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/travis/build/edschofield/python-future/future/tests/test_htmlparser.py", line 661, in test_attr_entity_replacement
[("starttag", "a", [("b", "&><"'")])])
File "/home/travis/build/edschofield/python-future/future/tests/test_htmlparser.py", line 93, in _run_check
collector = self.get_collector()
File "/home/travis/build/edschofield/python-future/future/tests/test_htmlparser.py", line 617, in get_collector
return EventCollector(strict=True)
File "/home/travis/build/edschofield/python-future/future/tests/test_htmlparser.py", line 27, in __init__
html.parser.HTMLParser.__init__(self, *args, **kw)
File "/home/travis/build/edschofield/python-future/future/backports/html/parser.py", line 135, in __init__
self.reset()
File "/home/travis/build/edschofield/python-future/future/backports/html/parser.py", line 143, in reset
_markupbase.ParserBase.reset(self)
TypeError: unbound method reset() must be called with ParserBase instance as first argument (got EventCollector instance instead)
This module is used as a foundation for the html.parser module. It has no
documented public API and should not be used directly.
� Nz[a-zA-Z][-_.a-zA-Z0-9]*\s*z(\'[^\']*\'|"[^"]*")\s*z--\s*>z ]\s*]\s*>z]\s*>c @ s� e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
d � Zd Z d
d� Z
d#dd�Z d#dd�Z dd� Z
dd� Zdd� Zdd� Zd d � Zd d � Zd d!� Zd"S )$�
ParserBaseziParser base class which provides some common support methods used
by the SGML/HTML and XHTML parsers.c C s | j tu r td��d S )Nz)_markupbase.ParserBase must be subclassed)� __class__r � RuntimeError��self� r �>/usr/lib/python3/dist-packages/future/backports/_markupbase.py�__init__6 s
��zParserBase.__init__c C s t d��)Nz.subclasses of ParserBase must override error())�NotImplementedError)r �messager r r �error; s �zParserBase.errorc C s d| _ d| _d S )N� r ��lineno�offsetr r r r �reset? s
zParserBase.resetc C s | j | jfS )z&Return current line number and offset.r r r r r �getposC s zParserBase.getposc C sd ||kr|S | j }|�d||�}|r(| j| | _|�d||�}||d | _|S | j| | | _|S )N�
r
)�rawdata�countr �rindexr )r �i�jr �nlines�posr r r � updateposK s �zParserBase.updatepos� c
C s� | j }|d }|||� dksJ d��|||d � dkr!|d S |||d � dv r-dS t|�}|||d � dkr@| �|�S || d krK| �|�S | �||�\}}|d
k rY|S |d kr`d | _||k r�|| }|dkr�||d |� }|d kr�| �|� |d S | �|� |d S |d
v r�t||�}|s�dS |� � }nG|dv r�| �||�\} }n:|| jv r�|d }n0|d kr�|d kr�| �
|d |�}n |t g d��v r�| � d| � n| � d� n | � d|| � |d
k r�|S ||k sddS )N� �<!z$unexpected call to parse_declarationr
�>)�-r ���z--�[r �doctyper z"'�4abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ)�attlist�linktype�link�elementz&unsupported '[' char in %s declarationz"unexpected '[' char in declarationz!unexpected %r char in declaration)
r �len�
parse_comment�parse_marked_section�
_scan_name�_decl_otherchars� handle_decl� unknown_decl�_declstringlit_match�end�_parse_doctype_subset�setr )
r r r r �n�decltype�c�data�m�namer r r �parse_declaration[ s`
�
��(z ParserBase.parse_declarationr
c C s� | j }|||d � dksJ d��| �|d |�\}}|dk r!|S |tg d��v r2t�||d �}n |tg d��v rCt�||d �}n
| �d||d |� � |sTdS |rf|�d�}| �||d |� � |� d�S ) N� z<![z)unexpected call to parse_marked_section()r )�temp�cdata�ignore�include�rcdata)�if�else�endifz+unknown status keyword %r in marked sectionr! )
r r, r3 �_markedsectionclose�search�_msmarkedsectioncloser �startr/ r1 )r r �reportr �sectNamer �matchr r r r+ � s
z ParserBase.parse_marked_sectionc C sj | j }|||d � dkr| �d� t�||d �}|s dS |r0|�d�}| �||d |� � |�d�S )N� �<!--z"unexpected call to parse_comment()r! r )r r �
_commentcloserE rG �handle_commentr1 )r r rH r rJ r r r r r* � s
zParserBase.parse_commentc
C s. | j }t|�}|}||k �r|| }|dkr�|||d � }|dkr$dS |dkr7| �||d � | �d| � |d |kr?dS |d |krGdS |||d � dkr_| j|d d
�}|d k r^|S q | �|d |�\}}|dkrodS |tg d ��vr�| �||d � | �d | � t| d
| �} | ||�}|d k r�|S nx|dkr�|d |kr�dS | �|d |�\}}|d k r�|S || dkr�|d }nQ|dkr�|d }||k r�|| �� r�|d }||k r�|| �� s�||k r�|| dkr�|S | �||� | �d� ndS |�� �r|d }n
| �||� | �d| � ||k sdS )N�<r r! r r
z*unexpected char in internal subset (in %r)rK rL r )rH )r% r( �entity�notationz)unknown declaration %r in internal subset�_parse_doctype_�%�;�]r z%unexpected char after internal subsetz%unexpected char %r in internal subset) r r) r r r* r, r3 �getattr�isspace)
r r � declstartposr r4 r r6 �sr9 �methr r r r2 � sx
�
� ��
�<z ParserBase._parse_doctype_subsetc C sF | � ||�\}}|dkrdS | j}d||d � v r!|�d|�d S dS )Nr! r r
)r, r �find)r r rX r9 r r r r r �_parse_doctype_element s z!ParserBase._parse_doctype_elementc C s� | j }| �||�\}}|||d � }|dkrdS |dkr!|d S | �||�\}}|dk r0|S |||d � }|dkr>dS |dkrvd||d � v rS|�d|�d }ndS |||d � �� rm|d }|||d � �� s_||d � sudS n| �||�\}}|||d � }|s�dS |dv r�t||�}|r�|�� }ndS |||d � }|s�dS |d kr�||d � d kr�dS | �|d |�\}}|dk r�|S |||d � }|s�dS |dkr�|d S q")
Nr
r r! r r �(�)�'"�#)r r, r[ rW r0 r1 )r r rX r r9 r r6 r8 r r r �_parse_doctype_attlist s` � �
�z!ParserBase._parse_doctype_attlistc C s� | � ||�\}}|dk r|S | j} |||d � }|s dS |dkr&|d S |dv r8t||�}|s3dS |�� }n| � ||�\}}|dk rF|S q)Nr r
r! r r_ )r, r r0 r1 )r r rX r9 r r r6 r8 r r r �_parse_doctype_notationX s&
�z"ParserBase._parse_doctype_notationc C s� | j }|||d � dkr*|d } |||d � }|s dS |�� r'|d }nnqn|}| �||�\}}|dk r:|S | j ||d � }|sHdS |dv rZt||�}|rX|�� }ndS |dkrb|d S | �||�\}}|dk rp|S q;)Nr
rS r! r r_ r )r rW r, r0 r1 )r r rX r r r6 r9 r8 r r r �_parse_doctype_entityo s>
�
�z ParserBase._parse_doctype_entityc C s� | j }t|�}||kr
dS t||�}|r.|�� }|�� }|t|� |kr&dS |�� |�� fS | �||� | �d|||d � � d S )N)Nr! zexpected name token at %r� ) r r) �_declname_match�group�strip�lowerr1 r r )r r rX r r4 r8 rY r9 r r r r, � s
�zParserBase._scan_namec C s d S )Nr )r r7 r r r r/ � s zParserBase.unknown_declN)r
)�__name__�
__module__� __qualname__�__doc__r r r r r r- r: r+ r* r2 r\ ra rb rc r, r/ r r r r r 2 s$
R
C 9$ r ) rl �re�compilerJ re r0 rM rD rF �objectr r r r r �<module> s "