U
e5d9 � @ sV d Z ddlZe�d�jZe�d�jZe�d�Ze�d�Ze�d�Z[G dd � d �Z dS )
z�Shared support for scanning document type declarations in HTML and XHTML.
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$dd�Z dd� Z
dd� Zdd� Zdd� Zd d � Zd d � Zd d!� Zd"S )%�
ParserBaseziParser base class which provides some common support methods used
by the SGML/HTML and XHTML parsers.c C s | j tkrtd��d S )Nz)_markupbase.ParserBase must be subclassed)� __class__r � RuntimeError��self� r �#/usr/lib64/python3.8/_markupbase.py�__init__ s
�zParserBase.__init__c C s t d��d S )Nz.subclasses of ParserBase must override error())�NotImplementedError)r �messager r r �error s �zParserBase.errorc C s d| _ d| _d S )N� r ��lineno�offsetr r r r �reset$ s zParserBase.resetc C s | j | jfS )z&Return current line number and offset.r r r r r �getpos( s zParserBase.getposc C sb ||kr |S | j }|�d||�}|rN| j| | _|�d||�}||d | _n| j| | | _|S )N�
r
)�rawdata�countr �rindexr )r �i�jr Znlines�posr r r � updatepos0 s zParserBase.updatepos� c
C s� | j }|d }|||� dks&td��|||d � dkrB|d S |||d � dkrZdS t|�}|||d � dkr�| �|�S || d kr�| �|�S | �||�\}}|d
k r�|S |d kr�d | _||k �r�|| }|dk�r||d |� }|d k�r| �|� n
| �|� |d S |d
k�r)�-r �����z--�[r Zdoctyper z"'Z4abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ> �linktype�attlist�element�linkz&unsupported '[' char in %s declarationz"unexpected '[' char in declarationz!unexpected %r char in declaration)
r �AssertionError�len�
parse_comment�parse_marked_section�
_scan_name�_decl_othercharsZ handle_decl� unknown_decl�_declstringlit_match�end�_parse_doctype_subsetr )
r r r r �nZdecltype�c�data�m�namer r r �parse_declaration@ s\
�z ParserBase.parse_declarationr
c C s� | j }|||d � dks"td��| �|d |�\}}|dk rB|S |dkr\t�||d �}n4|dkrvt�||d �}n| �d||d |� � |s�dS |r�|�d�}| �||d |� � |� d�S ) N� z �rcdata�temp�ignore�cdata�include> �endif�else�ifz+unknown status keyword %r in marked sectionr )
r r& r* �_markedsectionclose�search�_msmarkedsectioncloser �startr, r. )r r �reportr ZsectNamer �matchr r r r) � s
z ParserBase.parse_marked_sectionc C sj | j }|||d � dkr$| �d� t�||d �}|s r$ r# �notation�entityz)unknown declaration %r in internal subsetZ_parse_doctype_�%�;�]r z%unexpected char after internal subsetz%unexpected char %r in internal subset)r r' r r r( r* �getattr�isspace)
r r � declstartposr r0 r r1 �sr4 Zmethr r r r/ � sp
�
z ParserBase._parse_doctype_subsetc C sF | � ||�\}}|dkr dS | j}d||d � krB|�d|�d S dS )Nr r r
)r* r �find)r r rP r4 r r r r r �_parse_doctype_element� s z!ParserBase._parse_doctype_elementc C s� | j }| �||�\}}|||d � }|dkr2dS |dkrB|d S | �||�\}}|dk r^|S |||d � }|dkrzdS |dkr�d||d � kr�|�d|�d }ndS |||d � �� r�|d }q�||d � s�dS n| �||�\}}|||d � }|�sdS |dk�rDt||�}|�r&|�� }ndS |||d � }|�sDdS |d k�r�||d � d k�rddS | �|d |�\}}|dk �r�|S |||d � }|�s�dS |dkrB|d S qBd S )
Nr
r r r r �(�)�'"�#)r r* rR rO r- r. )r r rP r r4 r r1 r3 r r r �_parse_doctype_attlist sX
z!ParserBase._parse_doctype_attlistc C s� | � ||�\}}|dk r |S | j}|||d � }|s:dS |dkrJ|d S |dkrnt||�}|sddS |�� }q"| � ||�\}}|dk r"|S q"d S )Nr r
r r rV )r* r r- r. )r r rP r4 r r r1 r3 r r r �_parse_doctype_notation= s"
z"ParserBase._parse_doctype_notationc C s� | j }|||d � dkrR|d }|||d � }|s:dS |�� rP|d }q"qVq"n|}| �||�\}}|dk rr|S | j ||d � }|s�dS |dkr�t||�}|r�|�� }q�dS qr|dkr�|d S | �||�\}}|dk rr|S qrd S )Nr
rK r r rV r )r rO r* r- r. )r r rP r r r1 r4 r3 r r r �_parse_doctype_entityT s4
z ParserBase._parse_doctype_entityc C s� | j }t|�}||krdS t||�}|r\|�� }|�� }|t|� |krLdS |�� |�� fS | �||� | �d|||d � � d S )N)Nr zexpected name token at %r� ) r r' �_declname_match�group�strip�lowerr. r r )r r rP r r0 r3 rQ r4 r r r r* x s
�zParserBase._scan_namec C s d S )Nr )r r2 r r r r, � s zParserBase.unknown_declN)r
)r
)�__name__�
__module__� __qualname__�__doc__r r r r r r+ r5 r) r( r/ rS rX rY rZ r* r, r r r r r s"
R
C 9$r )
rc �re�compilerD r\ r- rG r? rA r r r r r � s