SEO

File: //usr/lib64/python3.6/html/__pycache__/parser.cpython-36.opt-2.pyc
3

�_Vj�I�@s�ddlZddlZddlZddlmZdgZejd�Zejd�Zejd�Z	ejd�Z
ejd�Zejd	�Zejd
�Z
ejd�Zejd�Zejd
ej�Zejd	�Zejd�ZGdd�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--\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#</\s*([a-zA-Z][-.a-zA-Z0-9:_]*)\s*>c@s�eZdZd9Zdd�dd�Zdd�Zd	d
�Zdd�Zd
Zdd�Z	dd�Z
dd�Zdd�Zdd�Z
d:dd�Zdd�Zdd�Zdd �Zd!d"�Zd#d$�Zd%d&�Zd'd(�Zd)d*�Zd+d,�Zd-d.�Zd/d0�Zd1d2�Zd3d4�Zd5d6�Zd7d8�Zd
S);r�script�styleT)�convert_charrefscCs||_|j�dS)N)r�reset)�selfr�r
�#/usr/lib64/python3.6/html/parser.py�__init__WszHTMLParser.__init__cCs:d|_d|_t|_d|_g|_d|_d|_tj	j
|�dS)N�z???r�)�rawdata�lasttag�interesting_normal�interesting�
cdata_elem�_pending�_pending_len�_parse_threshold�_markupbase�
ParserBaser)r	r
r
rr`szHTMLParser.resetcCs�|jt|�7_|j|jkr,|jj|�n~|jsB|j|7_n,|jj|�|jdj|j�7_|jj�d|_t|j�}|jd�t|j�|kr�d|_nt|j�|_dS)Nr
rr)	r�lenrr�appendr�join�clear�goahead)r	�data�nr
r
r�feedks



zHTMLParser.feedcCs:|jr,|jdj|j�7_|jj�d|_|jd�dS)Nr
rr)rrrrrr)r	r
r
r�close�s

zHTMLParser.closeNcCs|jS)N)�_HTMLParser__starttag_text)r	r
r
r�get_starttag_text�szHTMLParser.get_starttag_textcCs$|j�|_tjd|jtj�|_dS)Nz</\s*%s\s*>)�lowerr�re�compile�Ir)r	�elemr
r
r�set_cdata_mode�s
zHTMLParser.set_cdata_modecCst|_d|_dS)N)rrr)r	r
r
r�clear_cdata_mode�szHTMLParser.clear_cdata_modecCsL|j}d}t|�}�x�||k�r�|jr||jr||jd|�}|dkr�|jdt||d��}|dkrvtjd�j	||�rvP|}n(|j
j	||�}|r�|j�}n|jr�P|}||kr�|jr�|jr�|jt
|||���n|j|||��|j||�}||kr�P|j}|d|��rLtj||��r&|j|�}	n�|d|��r>|j|�}	nl|d|��rV|j|�}	nT|d|��rn|j|�}	n<|d	|��r�|j|�}	n$|d
|k�r�|jd�|d
}	nP|	dk�r>|�s�P|jd|d
�}	|	dk�r�|jd|d
�}	|	dk�r|d
}	n|	d
7}	|j�r,|j�r,|jt
|||	���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
�}Pq|d|�rtj||�}|�rN|jd
�}
|j|
�|j�}	|d|	d
��s@|	d
}	|j||	�}qtj||�}|�r�|�r�|j�||d�k�r�|j�}	|	|k�r�|}	|j||d
�}Pn,|d
|k�r�|jd�|j||d
�}nPqqW|�r:||k�r:|j�r:|j�r|j�r|jt
|||���n|j|||��|j||�}||d�|_dS)Nr�<�&�"z[\s;]z</z<!--z<?z<!rrz&#��;���)rrrr�find�rfind�maxr%r&�searchr�start�handle_datarZ	updatepos�
startswith�starttagopen�match�parse_starttag�parse_endtag�
parse_comment�parse_pi�parse_html_declaration�charref�group�handle_charref�end�	entityref�handle_entityref�
incomplete)r	rBr�ir�jZampposr9r7�k�namer
r
rr�s�












zHTMLParser.goaheadcCs�|j}|||d�dkr$|j|�S|||d�dkrB|j|�S|||d�j�dkr�|jd|d�}|d
krvdS|j||d	|��|dS|j|�SdS)N�z<!--�z<![�	z	<!doctyperrr.r0r0)rr<Zparse_marked_sectionr$r1�handle_decl�parse_bogus_comment)r	rFr�gtposr
r
rr>s

z!HTMLParser.parse_html_declarationrcCsD|j}|jd|d�}|dkr"dS|r<|j||d|��|dS)Nrr.rr0r0)rr1�handle_comment)r	rFZreportr�posr
r
rrN1szHTMLParser.parse_bogus_commentcCsH|j}tj||d�}|sdS|j�}|j||d|��|j�}|S)Nr.rr0)r�picloser4r5�	handle_pirB)r	rFrr9rGr
r
rr==szHTMLParser.parse_picCs�d|_|j|�}|dkr|S|j}|||�|_g}tj||d�}|j�}|jd�j�|_}x�||k�r$t	j||�}|s~P|jddd�\}	}
}|
s�d}n^|dd�dko�|d
d�kns�|dd�dko�|dd�kn�r�|dd�}|�rt
|�}|j|	j�|f�|j�}qbW|||�j�}|d
k�r�|j
�\}
}d	|jk�r�|
|jjd	�}
t|j�|jjd	�}n|t|j�}|j|||��|S|jd��r�|j||�n"|j||�||jk�r�|j|�|S)Nrrr.rK�'�"r�/>�
r0r0r0)rrV)r"�check_for_whole_start_tagr�tagfind_tolerantr9rBr@r$r�attrfind_tolerantrr�stripZgetpos�countrr2r6�endswith�handle_startendtag�handle_starttag�CDATA_CONTENT_ELEMENTSr))r	rF�endposr�attrsr9rH�tag�mZattrname�restZ	attrvaluerB�lineno�offsetr
r
rr:IsP
(*

zHTMLParser.parse_starttagcCs�|j}tj||�}|r�|j�}|||d�}|dkr>|dS|dkr~|jd|�rZ|dS|jd|�rjd	S||krv|S|dS|dkr�d
S|dkr�dS||kr�|S|dStd��dS)Nrr�/z/>r.r
z6abcdefghijklmnopqrstuvwxyz=/ABCDEFGHIJKLMNOPQRSTUVWXYZzwe should not get here!r0r0r0)r�locatestarttagend_tolerantr9rBr7�AssertionError)r	rFrrdrG�nextr
r
rrX|s.z$HTMLParser.check_for_whole_start_tagcCs|j}tj||d�}|sdS|j�}tj||�}|s�|jdk	rV|j|||��|Stj||d�}|s�|||d�dkr�|dS|j	|�S|j
d�j�}|jd|j��}|j
|�|dS|j
d�j�}|jdk	�r�||jk�r�|j|||��|S|j
|j��|j�|S)Nrr.rKz</>rr0)r�	endendtagr4rB�
endtagfindr9rr6rYrNr@r$r1�
handle_endtagr*)r	rFrr9rOZ	namematchZtagnamer(r
r
rr;�s6


zHTMLParser.parse_endtagcCs|j||�|j|�dS)N)r_rn)r	rcrbr
r
rr^�szHTMLParser.handle_startendtagcCsdS)Nr
)r	rcrbr
r
rr_�szHTMLParser.handle_starttagcCsdS)Nr
)r	rcr
r
rrn�szHTMLParser.handle_endtagcCsdS)Nr
)r	rIr
r
rrA�szHTMLParser.handle_charrefcCsdS)Nr
)r	rIr
r
rrD�szHTMLParser.handle_entityrefcCsdS)Nr
)r	rr
r
rr6�szHTMLParser.handle_datacCsdS)Nr
)r	rr
r
rrP�szHTMLParser.handle_commentcCsdS)Nr
)r	Zdeclr
r
rrM�szHTMLParser.handle_declcCsdS)Nr
)r	rr
r
rrS�szHTMLParser.handle_picCsdS)Nr
)r	rr
r
r�unknown_decl�szHTMLParser.unknown_declcCstjdtdd�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
rr�s
zHTMLParser.unescape)rr)r)�__name__�
__module__�__qualname__r`rrr r!r"r#r)r*rr>rNr=r:rXr;r^r_rnrArDr6rPrMrSrorr
r
r
rr?s6	z
3"()r%rqrZhtmlr�__all__r&rrErCr?r8rRZcommentcloserYrZ�VERBOSErirlrmrrr
r
r
r�<module>s&