3
\9 � @ sV d Z ddlZejd�jZejd�jZejd�Zejd�Zejd�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$dd�Z dd� Z
dd� Zdd� Zdd� Zd d � Zd d � Zd d!� Zd"S )%�
ParserBaseziParser base class which provides some common support methods used
by the SGML/HTML and XHTML parsers.c C s | j tkrtd��d S )Nz)_markupbase.ParserBase must be subclassed)� __class__r � RuntimeError)�self� r �#/usr/lib64/python3.6/_markupbase.py�__init__ s
zParserBase.__init__c C s t d��d S )Nz.subclasses of ParserBase must override error())�NotImplementedError)r �messager r r �error s zParserBase.errorc C s d| _ d| _d S )N� r )�lineno�offset)r r r r �reset$ s zParserBase.resetc C s | j | jfS )z&Return current line number and offset.)r
r )r r r r �getpos( s zParserBase.getposc C sb ||kr |S | j }|jd||�}|rN| j| | _|jd||�}||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�| j|�S || d kr�| j|�S | j||�\}}|d
k r�|S |d kr�d| _�x ||k �r�|| }|dk�r||d |� }|d k�r| j|� n
| j|� |d S |d k�r@t ||�}|�s6dS |j
� }n�|d
k�r\| j||�\} }nt|| jk�rr|d }n^|d k�r�|d k�r�| j |d |�}n$|dk�r�| j d| � n
| j d� n| j d|| � |d
k r�|S q�W dS )N� z�-r z--�[r Zdoctypez"'Z4abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ�attlist�linktype�link�elementz&unsupported '[' char in %s declarationz"unexpected '[' char in declarationz!unexpected %r char in declaration)r r �����r" > r r r r! r" )
r �AssertionError�len�
parse_comment�parse_marked_section�
_scan_name�_decl_othercharsZ handle_decl� unknown_decl�_declstringlit_match�end�_parse_doctype_subsetr )
r r r r �nZdecltype�c�data�m�namer r r �parse_declaration@ sZ
z ParserBase.parse_declarationr c C s� | j }|||d � dks"td��| j|d |�\}}|dk rB|S |dkr\tj||d �}n4|dkrvtj||d �}n| jd
||d |� � |s�dS |r�|jd�}| j||d |� � |j d�S )N� z r7 r5 r8 r4 r6 > r9 r: r; r" )
r r# r' �_markedsectionclose�search�_msmarkedsectioncloser �startr) r+ )r r �reportr ZsectNamer �matchr r r r& � s
z ParserBase.parse_marked_sectionc C sj | j }|||d � dkr$| jd� tj||d �}|s rE rF r r! r" r" r" )r r$ r r r% r' �getattr�isspace)
r r � declstartposr r- r r. �sr1 �methr r r r, � sn
z ParserBase._parse_doctype_subsetc C sF | j ||�\}}|dkr dS | j}d||d � krB|jd|�d S dS )Nr r r" r" r" )r' r �find)r r rL r1 r r r r r �_parse_doctype_element� s z!ParserBase._parse_doctype_elementc C s� | j }| j||�\}}|||d � }|dkr2d S |dkrB|d S �xv| j||�\}}|dk rb|S |||d � }|dkr~d
S |dkr�d||d � kr�|jd|�d }nd S x |||d � j� r�|d }q�W ||d � s�d S n| j||�\}}|||d � }|�s
d
S |dk�rLt||�}|�r.|j� }ndS |||d � }|�sLdS |dk�r�||d � dk�rldS | j|d |�\}}|dk �r�|S |||d � }|�s�dS |dkrF|d S qFW d S )Nr r r r �(�)z'"�#r" r" r" r" r" r" r" r" r" )r r' rO rK r* r+ )r r rL r r1 r r. r0 r r r �_parse_doctype_attlist sZ
z!ParserBase._parse_doctype_attlistc C s� | j ||�\}}|dk r |S | j}xl|||d � }|s r r r r r � s