File: //usr/local/lib64/python3.6/site-packages/lxml/html/__pycache__/soupparser.cpython-36.pyc
3
��?e�' � @ sT d Z dddgZddlZddlmZmZ y0ddlmZmZm Z m
Z
m Z m Z m
Z
e e
fZW n8 ek
r� ddlmZmZm Z m
Z
m Z m Z e ZY nX dd d�Zdd
d�Zdd d�Zd d
� Zejdej�jZG dd� d�Zdd� Zdd� Zyddlm Z W n" ek
�r ddl m Z Y nX ejd�j Z ye W n e!k
�rF e"Z Y nX dd� Z#dS ) z5External interface to the BeautifulSoup HTML parser.
�
fromstring�parse� convert_tree� N)�etree�html)�
BeautifulSoup�Tag�Comment�ProcessingInstruction�NavigableString� Declaration�Doctype)r r r r
r r c K s t | ||f|�S )a� Parse a string of HTML data into an Element tree using the
BeautifulSoup parser.
Returns the root ``<html>`` Element of the tree.
You can pass a different BeautifulSoup parser through the
`beautifulsoup` keyword, and a diffent Element factory function
through the `makeelement` keyword. By default, the standard
``BeautifulSoup`` class and the default factory of `lxml.html` are
used.
)�_parse)�data�
beautifulsoup� makeelement�bsargs� r �4/tmp/pip-build-_veujq9v/lxml/lxml/html/soupparser.pyr s c K s, t | d�st| �} t| ||f|�}tj|�S )aY Parse a file into an ElemenTree using the BeautifulSoup parser.
You can pass a different BeautifulSoup parser through the
`beautifulsoup` keyword, and a diffent Element factory function
through the `makeelement` keyword. By default, the standard
``BeautifulSoup`` class and the default factory of `lxml.html` are
used.
�read)�hasattr�openr r � ElementTree)�filer r r �rootr r r r $ s
c C s. t | |�}|j� }x|D ]}|j|� qW |S )a Convert a BeautifulSoup tree to a list of Element trees.
Returns a list instead of a single root Element to support
HTML-like soup with more than one root element.
You can pass a different Element factory through the `makeelement`
keyword.
)�
_convert_tree� getchildren�remove)�beautiful_soup_treer r �children�childr r r r 3 s
c K s� |d kr t }t|d�r&d|kr&d|d< t|d�r@d|kr@d|d< || f|�}t||�}t|�dkrx|d jdkrx|d S d|_|S ) NZ
HTML_ENTITIESZconvertEntitiesr ZDEFAULT_BUILDER_FEATURES�featuresz html.parser� r )r r r �len�tag)�sourcer r r �treer r r r r E s
r z`(?:\s|[<!])*DOCTYPE\s*HTML(?:\s+PUBLIC)?(?:\s+(\'[^\']*\'|"[^"]*"))?(?:\s+(\'[^\']*\'|"[^"]*"))?c @ s e Zd Zdd� Zdd� ZdS )�
_PseudoTagc C s d| _ g | _|| _d S )Nr )�name�attrs�contents)�selfr* r r r �__init__` s z_PseudoTag.__init__c C s
| j j� S )N)r* �__iter__)r+ r r r r- e s z_PseudoTag.__iter__N)�__name__�
__module__� __qualname__r, r- r r r r r' ^ s r' c C s |d krt jj}d }}d }}xft| �D ]Z\}}t|t�rn|d krH|}|}|d kr�|jr�|jj� dkr�|}q*|d kr*t|t�r*|}q*W |d kr�g }} | j }
n2| j d |� }| j ||d � }
| j |d d � } |d k �r |
j
|�}|
d |� |j |
|d d � |_ nt |
�}t |�} | |�} | }
x2t
|�D ]&}| |�}|d k �r2|
j|� |}
�q2W | }
x.| D ]&}| |�}|d k �rf|
j|� |}
�qfW |d k �ry |j� }W n tk
�r� |j}Y nX t|�}|�s�n>|j� \}}| j� j}|�o�|dd� |_|�o|dd� |_| S )Nr r" ���r1 )r � html_parserr � enumerate�
isinstancer r( �lower�_DECLARATION_OR_DOCTYPEr* �indexr' �_init_node_converters�reversedZ addpreviousZaddnextZ output_ready�AttributeError�string�_parse_doctype_declaration�groupsZ getroottree�docinfo� public_idZ
system_url)r r Zfirst_element_idxZlast_element_idxZ html_rootZ declaration�i�eZpre_rootZ post_root�roots� convert_nodeZres_root�prevZ convertedZdoctype_string�matchZ external_idZsys_urir> r r r r i sd
&
r c s� i �g ���fdd�}��fdd��d��fdd� �dd� �d d
� � |t t�����fd d ��}|t�d
d� �}|t�dd� �}|t�� fdd��}�S )Nc s ��� fdd�}|S )Nc s$ x �D ]}| � |< �j |� qW | S )N)�append)�handler�t)�
converters�ordered_node_types�typesr r �add� s
z5_init_node_converters.<locals>.converter.<locals>.addr )rK rL )rI rJ )rK r � converter� s z(_init_node_converters.<locals>.converterc s$ x �D ]}t | |�r� | S qW d S )N)r4 )�noderH )rI rJ r r �find_best_converter� s
z2_init_node_converters.<locals>.find_best_converterc sP y� t | � }W n( tk
r8 �| � }� t | �< Y nX |d krFd S || |�S )N)�type�KeyError)�bs_node�parentrG )rI rO r r rC � s z+_init_node_converters.<locals>.convert_nodec S s\ t | t�rFi }xH| j� D ](\}}t |t�r4dj|�}t|�||< qW ntdd� | D ��}|S )N� c s s | ]\}}|t |�fV qd S )N)�unescape)�.0�k�vr r r � <genexpr>� s z;_init_node_converters.<locals>.map_attrs.<locals>.<genexpr>)r4 �dict�items�list�joinrU )Zbs_attrs�attribsrW rX r r r � map_attrs� s
z(_init_node_converters.<locals>.map_attrsc S s: t | �dkr | jpd| | _n| d jp*d| | d _d S )Nr � r" r1 r1 )r# �text�tail)rS ra r r r � append_text� s z*_init_node_converters.<locals>.append_textc s� | j }|d k r2|r�|�nd }tj|| j|d�}n |r>�|�ni }�| j|d�}xP| D ]H}y�t|� }W n tk
r~ Y nX |d k rV|||� qV� ||� qVW |S )N)�attrib)r) r �
SubElementr( rP rQ )rR rS r) r^ �resr rG )rC rI r r_ r r � convert_tag� s
z*_init_node_converters.<locals>.convert_tagc S s t j| �}|d k r |j|� |S )N)r Z HtmlCommentrF )rR rS rf r r r �convert_comment s
z._init_node_converters.<locals>.convert_commentc S s> | j d�r| d d� } tj| jdd�� }|d k r:|j|� |S )N�?r" rT r1 )�endswithr r
�splitrF )rR rS rf r r r �
convert_pi s
z)_init_node_converters.<locals>.convert_pic s |d k r� |t | �� d S )N)rU )rR rS )rc r r � convert_text s z+_init_node_converters.<locals>.convert_text)N)r r' r r
r )r rM rg rh rl rm r )rc rC rI rO r r_ rJ r r8 � s
r8 )�name2codepointz&(\w+);c C s | sdS dd� }t || �S )Nr` c S s2 yt t| jd� �S tk
r, | jd�S X d S )Nr" r )�unichrrn �grouprQ )�mr r r �unescape_entity5 s z!unescape.<locals>.unescape_entity)�handle_entities)r; rr r r r rU 1 s rU )NN)NN)N)$�__doc__�__all__�re�lxmlr r Zbs4r r r r
r r r
r6 � ImportErrorr r r r �compile�
IGNORECASErE r< r' r r8 �
html.entitiesrn �htmlentitydefs�subrs ro � NameError�chrrU r r r r �<module> s:
$
Uc