Current File : /opt/alt/python310/lib64/python3.10/html/__pycache__//parser.cpython-310.opt-2.pyc
o
0�i�R � @ s� d dl Z d dlZd dlmZ dgZe �d�Ze �d�Ze �d�Ze �d�Z e �d�Z
e �d �Z e �d
�Z e �d �Z
e �d �Ze �d
�Ze �de j�Ze �de j�Ze �de j�Ze �d
�Ze �d�ZG dd� dej�ZdS )� 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
</[a-zA-Z]�>z--!?>z-?>z0([a-zA-Z][^\t\n\r\f />]*)(?:[\t\n\r\f ]|/(?!>))*a{
(
(?<=['"\t\n\r\f /])[^\t\n\r\f />][^\t\n\r\f /=>]* # attribute name
)
([\t\n\r\f ]*=[\t\n\r\f ]* # value indicator
('[^']*' # LITA-enclosed value
|"[^"]*" # LIT-enclosed value
|(?!['"])[^>\t\n\r\f ]* # bare value
)
)?
(?:[\t\n\r\f ]|/(?!>))* # possibly followed by a space
a
[a-zA-Z][^\t\n\r\f />]* # tag name
[\t\n\r\f /]* # optional whitespace before attribute name
(?:(?<=['"\t\n\r\f /])[^\t\n\r\f />][^\t\n\r\f /=>]* # attribute name
(?:[\t\n\r\f ]*=[\t\n\r\f ]* # value indicator
(?:'[^']*' # LITA-enclosed value
|"[^"]*" # LIT-enclosed value
|(?!['"])[^>\t\n\r\f ]* # bare value
)
)?
[\t\n\r\f /]* # possibly followed by a space
)*
>?
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 space
)?(?:\s|/(?!>))*
)*
)?
\s* # trailing whitespace
z#</\s*([a-zA-Z][-.a-zA-Z0-9:_]*)\s*>c @ s� e Zd Z dZdZddd�dd�Zdd � Zd
d � Zd d
� ZdZ dd� Z
dd�dd�Z dd� Z d=dd�Z
dd� Zdd� Zd=d d �Zd>d d �Zd!d"� Zd#d$� Zd%d&� Zd'd(� Zd)d*� Zd+d,� Zd-d.� Zd/d0� Zd1d2� Zd3d4� Zd5d6� Z d7d8� Z d9d:� Z d;d<� Z dS )?r )ZscriptZstyleZxmpZiframeZnoembedZnoframes)Ztextarea�titleTF)�convert_charrefs� scriptingc C s || _ || _| �� d S �N)r r �reset)�selfr r � r �2/opt/alt/python310/lib64/python3.10/html/parser.py�__init__v s zHTMLParser.__init__c C s6 d| _ d| _t| _d | _d| _d| _tj� | � d S )N� z???T)
�rawdata�lasttag�interesting_normal� interesting�
cdata_elem�_support_cdata�
_escapable� _markupbase�
ParserBaser �r
r r r r � s zHTMLParser.resetc C s | j | | _ | �d� d S )Nr )r �goahead�r
�datar r r �feed� s zHTMLParser.feedc C s | � d� d S )N� )r r r r r �close� s zHTMLParser.closeNc C s | j S r )�_HTMLParser__starttag_textr r r r �get_starttag_text� s z HTMLParser.get_starttag_text�� escapablec C st |� � | _|| _| jdkrt�d�| _d S |r*| js*t�d| j tjtjB �| _d S t�d| j tjtjB �| _d S )N� plaintextz\Zz&|</%s(?=[\t\n\r\f />])z</%s(?=[\t\n\r\f />])) �lowerr r �re�compiler r �
IGNORECASE�ASCII)r
�elemr" r r r �set_cdata_mode� s
�
�zHTMLParser.set_cdata_modec C s t | _d | _d| _d S )NT)r r r r r r r r �clear_cdata_mode� s
zHTMLParser.clear_cdata_modec C s || _ d S r )r )r
�flagr r r �_set_support_cdata� s
z HTMLParser._set_support_cdatac C s | j }d}t|�}||k �rU| jr;| js;|�d|�}|dk r:|�dt||d ��}|dkr8t�d�� ||�s8�n |}n| j
� ||�}|rI|� � }n| jrN�n|}||k ro| jrf| j rf| �
t|||� �� n | �
|||� � | �||�}||kr{�n�|j}|d|��r�t�||�r�| �|�} n@|d|�r�| �|�} n5|d|�r�| �|�} n*|d|�r�| �|�} n |d |�r�| �|�} n|d
|k s�|r�| �
d� |d
} n�n�| dk �r�|sِn|t�||�r�n�|d|��r|d |kr�| �
d� n�t�||�r�n�| �||d d � � n�|d|��r0|}d D ]}
|�|
|d
��r"|t|
�8 } n�q| �||d
|� � nW|d|��rF| j�rF| � ||d d � � nA|||d � � � dk�r_| � ||d d � � n(|d |��rq| �||d d � � n|d|��r�| � ||d d � � nt d��|} | �|| �}n�|d|��r�t!�||�}|�r�|�"� d d� } | �#| � |�$� } |d| d
��s�| d
} | �|| �}q d||d � v �r�| �
|||d � � | �||d �}nu|d|��rQt%�||�}|�r|�"d
�} | �&| � |�$� } |d| d
��s | d
} | �|| �}q t'�||�}|�r;|�r:|�"� ||d � k�r:|�$� } | |k�r2|} | �||d
�}n|d
|k �rP| �
d� | �||d
�}nn||k s|�r�||k �r�| j�rq| j �rq| �
t|||� �� n | �
|||� � | �||�}||d � | _ d S )Nr �<�&�" z [\t\n\r\f ;]z</�<!--z<?z<!r � )z--!z--�-� � <![CDATA[� � � <!doctypezwe should not get here!z&#����;)(r �lenr r �find�rfind�maxr% r& �searchr �startr � handle_datar Z updatepos�
startswith� starttagopen�match�parse_starttag� parse_endtag�
parse_comment�parse_pi�parse_html_declaration�
endtagopen�handle_comment�endswithr � unknown_declr$ � handle_decl� handle_pi�AssertionError�charref�group�handle_charref�end� entityref�handle_entityref�
incomplete) r
rT r �i�n�jZampposrD rB �k�suffix�namer r r r � s�
��
�
�} zHTMLParser.goaheadc C sT | j }|||d � dkr| �|�S |||d � dkr<| jr<|�d|d �}|dk r-dS | �||d |� � |d S |||d � �� d kre|�d
|d �}|dkrVdS | �||d |� � |d S |||d � d
kr�|�d
|d �}|dk r}dS ||d dkr�| �||d |d � � |d S | �||d |� � |d S | �|�S )Nr4 r1 r7 r5 z]]>r r9 r6 r8 r r2 r z<![�]) r rG r r< rM r$ rN rK �parse_bogus_comment)r
rX r rZ Zgtposr r r rI N s2
�
z!HTMLParser.parse_html_declarationc C s\ | j }t�||d �}|st�||d �}|sdS |r*|�� }| �||d |� � |�� S )Nr4 r9 )r � commentcloser? �commentabruptcloserD r@ rK rT )r
rX �reportr rD rZ r r r rG p s zHTMLParser.parse_commentr c C sD | j }|�d|d �}|dkrdS |r | �||d |� � |d S )Nr r2 r9 r )r r< rK )r
rX rb r �posr r r r_ s z HTMLParser.parse_bogus_commentc C sH | j }t�||d �}|sdS |�� }| �||d |� � |�� }|S )Nr2 r9 )r �picloser? r@ rO rT �r
rX r rD rZ r r r rH � s zHTMLParser.parse_pic C s d | _ | �|�}|dk r|S | j}|||� | _ g }t�||d �}|�� }|�d��� | _}||k r�t �||�}|s=nS|�ddd�\} }
} |
sLd } n-| d d� d kr^| dd � kssn | d d� d krq| dd � kryn n| dd� } | rt
| �} |� | �� | f� |�� }||k s4|||� � � } | dvr�| �
� \}
}d | j v r�|
| j �d � }
t| j �| j �d � }n|t| j � }| �|||� � |S | �d
�r�| �||� |S | �||� || jv s�| jr�|d ks�|d kr�| j|d
d� |S || jv �r| j|dd� |S )Nr r r2 r6 �'r9 �")r �/>�
rh Znoscriptr# Fr! T)r �check_for_whole_start_tagr �tagfind_tolerantrD rT rR r$ r �attrfind_tolerantr �append�stripZgetpos�countr; r= rA rL �handle_startendtag�handle_starttag�CDATA_CONTENT_ELEMENTSr r* �RCDATA_CONTENT_ELEMENTS)r
rX �endposr �attrsrD r[ �tag�m�attrname�restZ attrvaluerT �lineno�offsetr r r rE � sd
&( �
�
�
� �zHTMLParser.parse_starttagc C s6 | j }t�||d �}|�� }||d dkrdS |S )Nr r r9 )r � locatetagendrD rT re r r r rj � s z$HTMLParser.check_for_whole_start_tagc C s� | j }|�d|d �dk rdS t�||�s*||d |d � dkr%|d S | �|�S t�||d �}|�� }||d dkr@dS t�||d �}|�d�� � }| �
|� | � � |S )Nr r2 r r9 r6 r ) r r< rJ rD r_ r| rT rk rR r$ �
handle_endtagr+ )r
rX r rD rZ rv r r r rF � s
zHTMLParser.parse_endtagc C s | � ||� | �|� d S r )rq r} �r
rv ru r r r rp � s z HTMLParser.handle_startendtagc C � d S r r r~ r r r rq � � zHTMLParser.handle_starttagc C r r r )r
rv r r r r} r� zHTMLParser.handle_endtagc C r r r �r
r] r r r rS r� zHTMLParser.handle_charrefc C r r r r� r r r rV r� zHTMLParser.handle_entityrefc C r r r r r r r rA
r� zHTMLParser.handle_datac C r r r r r r r rK r� zHTMLParser.handle_commentc C r r r )r
Zdeclr r r rN r� zHTMLParser.handle_declc C r r r r r r r rO r� zHTMLParser.handle_pic C r r r r r r r rM r� zHTMLParser.unknown_decl)T)r ) �__name__�
__module__� __qualname__rr rs r
r r r r r r* r+ r- r rI rG r_ rH rE rj rF rp rq r} rS rV rA rK rN rO rM r r r r r Z s@
"
9
)r% r Zhtmlr �__all__r& r rW rU rQ rC rJ rd r` ra rk �VERBOSErl r| Zlocatestarttagend_tolerantZ endendtagZ
endtagfindr r r r r r �<module> s6
�
��
