3
�L�j�I � @ s� d Z ddlZddlZddlZddlmZ dgZejd�Zejd�Z ejd�Z
ejd�Z ejd �Z ejd
�Z
ejd �Zejd �Zejd
�Zejdej�Zejd
�Zejd�ZG dd� dej�ZdS )z A parser for HTML and XHTML.� N)�unescape�
HTMLParserz[&<]z
&[a-zA-Z#]z%&([a-zA-Z][-.a-zA-Z0-9]*)[^a-zA-Z0-9]z)&#(?:[0-9]+|[xX][0-9a-fA-F]+)[^0-9a-fA-F]z <[a-zA-Z]�>z--\s*>z+([a-zA-Z][^\t\n\r\f />\x00]*)(?:\s|/(?!>))*z]((?<=[\'"\s/])[^\s/>][^\s/=>]*)(\s*=+\s*(\'[^\']*\'|"[^"]*"|(?![\'"])[^>\s]*))?(?:\s|/(?!>))*aF
<[a-zA-Z][^\t\n\r\f />\x00]* # tag name
(?:[\s/]* # optional whitespace before attribute name
(?:(?<=['"\s/])[^\s/>][^\s/=>]* # attribute name
(?:\s*=+\s* # value indicator
(?:'[^']*' # LITA-enclosed value
|"[^"]*" # LIT-enclosed value
|(?!['"])[^>\s]* # bare value
)
(?:\s*,)* # possibly followed by a comma
)?(?:\s|/(?!>))*
)*
)?
\s* # trailing whitespace
z#c @ s� e Zd ZdZd:Zdd�dd�Zdd � Zd
d � Zd d
� ZdZ dd� Z
dd� Z dd� Z dd� Z
dd� Zd;dd�Zd d � Zd d � Zd d!� Zd"d#� Zd$d%� Zd&d'� Zd(d)� Zd*d+� Zd,d-� Zd.d/� Zd0d1� Zd2d3� Zd4d5� Z d6d7� Z d8d9� Z dS )'.)�_HTMLParser__starttag_text)r r
r
r �get_starttag_text� s z HTMLParser.get_starttag_textc C s$ |j � | _tjd| j tj�| _d S )Nz )�lowerr �re�compile�Ir )r �elemr
r
r �set_cdata_mode� s
zHTMLParser.set_cdata_modec C s t | _d | _d S )N)r r r )r r
r
r �clear_cdata_mode� s zHTMLParser.clear_cdata_modec C sZ | j }d}t|�}�x�||k �r�| jr|| j r||jd|�}|dk r�|jdt||d ��}|dkrvtjd�j ||� rvP |}n(| j
j ||�}|r�|j � }n | jr�P |}||k r�| jr�| j r�| j t
|||� �� n| j |||� � | j||�}||kr�P |j}|d|��rLtj||��r&| j|�} n�|d|��r>| j|�} nl|d|��rV| j|�} nT|d|��rn| j|�} n<|d |��r�| j|�} n$|d
|k �r�| j d� |d
} nP | dk �r>|�s�P |jd |d
�} | dk �r�|jd|d
�} | dk �r|d
} n| d
7 } | j�r,| j �r,| j t
||| � �� n| j ||| � � | j|| �}q|d |��r�tj||�}|�r�|j� d
d� }
| j|
� |j� } |d| d
��s�| d
} | j|| �}qn:d||d � k�r�| j |||d
� � | j||d
�}P q|d|��r�tj||�}|�rP|jd
�}
| j |
� |j� } |d| d
��sB| d
} | j|| �}qt j||�}|�r�|�r�|j� ||d � k�r�|j� } | |k�r�|} | j||d
�}P n,|d
|k �r�| j d� | j||d
�}nP qdst d��qW |�rH||k �rH| j �rH| j�r*| j �r*| j t
|||� �� n| j |||� � | j||�}||d � | _ d S )Nr �<�&�" z[\s;]z s
z!HTMLParser.parse_html_declarationr c C s` | j }|||d � dks"td��|jd|d �}|dkr>d S |rX| j||d |� � |d S )
Nr. ��
r0 r0 r0 )r rY )r" �check_for_whole_start_tagr �tagfind_tolerantr9 rF rB r@ r$ r �attrfind_tolerantr r �stripZgetpos�countr r2 r6 �endswith�handle_startendtag�handle_starttag�CDATA_CONTENT_ELEMENTSr) )r rG �endposr �attrsr9 rI �tag�mZattrname�restZ attrvaluerB �lineno�offsetr
r
r r: I sR
(*
zHTMLParser.parse_starttagc C s� | j }tj||�}|r�|j� }|||d � }|dkr>|d S |dkr~|jd|�rZ|d S |jd|�rjd S ||krv|S |d S |dkr�d
S |dkr�d S ||kr�|S |d S td��d S ) Nr r �/z/>r. r
z6abcdefghijklmnopqrstuvwxyz=/ABCDEFGHIJKLMNOPQRSTUVWXYZzwe should not get here!r0 r0 r0 )r �locatestarttagend_tolerantr9 rB r7 rF )r rG r rg rH �nextr
r
r r[ | s. z$HTMLParser.check_for_whole_start_tagc C s2 | j }|||d � dks"td��tj||d �}|s:dS |j� }tj||�}|s�| jd k rr| j|||� � |S t j||d �}|s�|||d � dkr�|d S | j
|�S |j d�j � }|j
d|j� �}| j|� |d S |j d�j � }| jd k �r|| jk�r| j|||� � |S | j|j � � | j� |S ) Nr. zr r0 )r rF � endendtagr4 rB �
endtagfindr9 r r6 r\ rO r@ r$ r1 �
handle_endtagr* )r rG r r9 rP Z namematchZtagnamer( r
r
r r; � s8
zHTMLParser.parse_endtagc C s | j ||� | j|� d S )N)rb rp )r rf re r
r
r ra � s z HTMLParser.handle_startendtagc C s d S )Nr
)r rf re r
r
r rb � s zHTMLParser.handle_starttagc C s d S )Nr
)r rf r
r
r rp � s zHTMLParser.handle_endtagc C s d S )Nr
)r rJ r
r
r rA � s zHTMLParser.handle_charrefc C s d S )Nr
)r rJ r
r
r rD � s zHTMLParser.handle_entityrefc C s d S )Nr
)r r r
r
r r6 � s zHTMLParser.handle_datac C s d S )Nr
)r r r
r
r rS � s zHTMLParser.handle_commentc C s d S )Nr
)r Zdeclr
r
r rN � s zHTMLParser.handle_declc C s d S )Nr
)r r r
r
r rV � s zHTMLParser.handle_pic C s d S )Nr
)r r r
r
r � unknown_decl� s zHTMLParser.unknown_declc C s t jdtdd� t|�S )NzZThe unescape method is deprecated and will be removed in 3.5, use html.unescape() instead.r. )�
stacklevel)�warnings�warn�DeprecationWarningr )r �sr
r
r r � s
zHTMLParser.unescape)r r )r ) �__name__�
__module__� __qualname__�__doc__rc r r r r! r" r# r) r* r r> rO r= r: r[ r; ra rb rp rA rD r6 rS rN rV rq r r
r
r
r r ? s8 z
3"()rz r% rs r Zhtmlr �__all__r& r rE rC r? r8 rU Z commentcloser\ r] �VERBOSErl rn ro r r r
r
r
r � s(