| Dir : /lib/python3.6/site-packages/bs4/builder/__pycache__/ |
| Server: Linux server1.ngambekcore.com 4.18.0-553.51.1.el8_10.x86_64 #1 SMP Wed Apr 30 04:00:07 EDT 2025 x86_64 IP: 159.198.77.92 |
| Dir : //lib/python3.6/site-packages/bs4/builder/__pycache__/_html5lib.cpython-36.opt-1.pyc |
3
6]0A � @ s d gZ ddlZddlZddlmZmZmZmZ ddlm Z m
Z
ddl Z ddl m
Z
mZ ddlmZmZmZmZ yddlmZ dZW n2 ek
r� Z zdd lmZ d
ZW Y ddZ[X nX G d d � d e�ZG d d
� d
ej�Z G dd� de �Z G dd� dej �Z G dd� de �Z!dS )�HTML5TreeBuilder� N)�
PERMISSIVE�HTML�HTML_5�HTMLTreeBuilder)�NamespacedAttribute�
whitespace_re)�
namespaces�prefixes)�Comment�Doctype�NavigableString�Tag)�_baseF)�baseTc @ sB e Zd ZdZdZeeeegZd dd�Z dd� Z
dd � Z d
d � Z dS )
r z Use html5lib to build a tree.�html5libNc c s& || _ |rtjd� |d d dfV d S )NzjYou provided a value for exclude_encoding, but the html5lib tree builder doesn't support exclude_encoding.F)�user_specified_encoding�warnings�warn)�self�markupr Zdocument_declared_encodingZexclude_encodings� r � /usr/lib/python3.6/_html5lib.py�prepare_markup0 s
z HTML5TreeBuilder.prepare_markupc C s� | j jd k rtjd� tj| jd�}t� }t|t �sNt
rD| j |d< n
| j |d< |j |f|�}t|t �rnd |_
n$|jjjd }t|t �s�|j}||_
d S )Nz�You provided a value for parse_only, but the html5lib tree builder doesn't support parse_only. The entire document will be parsed.)ZtreeZoverride_encoding�encodingr )�soupZ
parse_onlyr r r Z
HTMLParser�create_treebuilder�dict�
isinstance�str� new_html5libr �parse�original_encodingZ tokenizer�streamZ charEncoding�name)r r �parserZ extra_kwargs�docr" r r r �feed= s
zHTML5TreeBuilder.feedc C s t || j�| _| jS )N)�TreeBuilderForHtml5libr Zunderlying_builder)r �namespaceHTMLElementsr r r r X s z#HTML5TreeBuilder.create_treebuilderc C s d| S )zSee `TreeBuilder`.z)<html><head></head><body>%s</body></html>r )r Zfragmentr r r �test_fragment_to_document] s z*HTML5TreeBuilder.test_fragment_to_document)NN)
�__name__�
__module__� __qualname__�__doc__�NAMEr r r Zfeaturesr r' r r* r r r r r ) s
c sf e Zd Zd� fdd� Zdd� Zdd� Zdd � Zd
d � Zd d
� Zdd� Z dd� Z
dd� Z dd� Z � Z
S )r( Nc s8 |r || _ nddlm} |dd�| _ tt| �j|� d S )Nr )�
BeautifulSoup� z html.parser)r �bs4r0 �superr( �__init__)r r) r r0 )� __class__r r r4 d s
z TreeBuilderForHtml5lib.__init__c C s | j j� t| j | j d �S )N)r �reset�Element)r r r r �
documentClassl s
z$TreeBuilderForHtml5lib.documentClassc C s6 |d }|d }|d }t j|||�}| jj|� d S )Nr$ �publicId�systemId)r Zfor_name_and_idsr �object_was_parsed)r �tokenr$ r9 r: Zdoctyper r r �
insertDoctypep s
z$TreeBuilderForHtml5lib.insertDoctypec C s | j j||�}t|| j |�S )N)r �new_tagr7 )r r$ � namespace�tagr r r � elementClassx s z#TreeBuilderForHtml5lib.elementClassc C s t t|�| j�S )N)�TextNoder r )r �datar r r � commentClass| s z#TreeBuilderForHtml5lib.commentClassc C s0 ddl m} |dd�| _d| j_t| j| jd �S )Nr )r0 r1 z html.parserz[document_fragment])r2 r0 r r$ r7 )r r0 r r r �
fragmentClass s z$TreeBuilderForHtml5lib.fragmentClassc C s | j j|j� d S )N)r �append�element)r �noder r r � appendChild� s z"TreeBuilderForHtml5lib.appendChildc C s | j S )N)r )r r r r � getDocument� s z"TreeBuilderForHtml5lib.getDocumentc C s t jj| �jS )N)�treebuilder_base� TreeBuilder� getFragmentrG )r r r r rM � s z"TreeBuilderForHtml5lib.getFragmentc sB ddl m� g �tjd��d� ���fdd� ��|d� dj��S )Nr )r0 z8^(.*?)(?: PUBLIC "(.*?)"(?: "(.*?)")?| SYSTEM "(.*?)")?$c s� t | � �r
t | t�r��j| �}|r�|jd�}|jdkrx|jd�pBd}|jd�pZ|jd�pZd}�jdd| |||f � q��jdd| |f � n�jd d| f � �nHt | t�r̈jd
d| | f � �n$t | t�r�jd d| | f � �n | j�rd t | j | j
f }n| j
}�jd
d| |f � | j �r�g }x`t | j j
� �D ]N\}}t |t��rnd t |j |j
f }t |t ��r�dj|�}|j||f� �qFW x2t|�D ]&\}}�jdd|d ||f � �q�W |d7 }x| jD ]}�||� �q�W d S )N� � r1 � � z|%s<!DOCTYPE %s "%s" "%s">� z|%s<!DOCTYPE %s>z|%s<!DOCTYPE >z|%s<!-- %s -->z|%s"%s"z%s %sz|%s<%s>z
|%s%s="%s")r r �match�group� lastindexrF r r
r? r
r$ �attrs�list�itemsr �join�sortedZchildren) rG �indent�mr$ r9 r: �
attributes�value�child)r0 �
doctype_re�rv�serializeElementr r rb � sD
" z?TreeBuilderForHtml5lib.testSerializer.<locals>.serializeElement�
)r )r2 r0 �re�compilerY )r rG r )r0 r` ra rb r �testSerializer� s
)
z%TreeBuilderForHtml5lib.testSerializer)N)r+ r, r- r4 r8 r= rA rD rE rI rJ rM rf �
__classcell__r r )r5 r r( b s r( c @ sL e Zd Zdd� Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dS )�AttrListc C s || _ t| j j�| _d S )N)rG r rV )r rG r r r r4 � s zAttrList.__init__c C s t | jj� �j� S )N)rW rV rX �__iter__)r r r r ri � s zAttrList.__iter__c C sP t j}||d ks.| jj|krB||| jj krBt|t�sBtj|�}|| j|< d S )N�*)r Zcdata_list_attributesrG r$ r rW r �split)r r$ r^ Z list_attrr r r � __setitem__� s
zAttrList.__setitem__c C s t | jj� �S )N)rW rV rX )r r r r rX � s zAttrList.itemsc C s t | jj� �S )N)rW rV �keys)r r r r rm � s z
AttrList.keysc C s
t | j�S )N)�lenrV )r r r r �__len__� s zAttrList.__len__c C s
| j | S )N)rV )r r$ r r r � __getitem__� s zAttrList.__getitem__c C s |t | jj� �kS )N)rW rV rm )r r$ r r r � __contains__� s zAttrList.__contains__N) r+ r, r- r4 ri rl rX rm ro rp rq r r r r rh � s rh c @ sx e Zd Zdd� Zdd� Zdd� Zdd� Zeee�Zdd
d �Z d d
� Z
dd� Z dd� Z dd� Z
dd� Zdd� Zee�Zd S )r7 c C s& t jj| |j� || _|| _|| _d S )N)rK �Noder4 r$ rG r r? )r rG r r? r r r r4 � s zElement.__init__c C s( d }}t |t�r | }}n:t |t�r,|}n*|jjtkrJ|j }}| |_n |j}| |_t |t� rv|jd k rv|jj� |r�| jjr�| jjd jtkr�| jjd }| j j
|| �}|j |� || j _ n`t |t�r�| j j
|�}| jjr�| jj
d�}n | jjd k �r
| j j
� }n| j}| j j|| j|d� d S )NrN F)�parent�most_recent_element���ru )r r r rG r5 r
rs �extract�contentsr �
new_string� replace_withZ_most_recent_element�_last_descendant� next_elementr; )r rH Z string_childr_ Z old_elementZ new_elementrt r r r rI � s8
zElement.appendChildc C s t | jt�ri S t| j�S )N)r rG r rh )r r r r �
getAttributes s zElement.getAttributesc C s� |d k r�t |�dkr�g }x8t|j� �D ](\}}t|t�r&t|� }||= |||< q&W | jjj| j |� x"t|j� �D ]\}}|| j
|< qrW | jjj | j
� d S )Nr ) rn rW rX r �tupler r ZbuilderZ$_replace_cdata_list_attribute_valuesr$ rG Zset_up_substitutions)r r] Zconverted_attributesr$ r^ �new_namer r r �
setAttributes! s
zElement.setAttributesNc C s4 t | jj|�| j�}|r&| j||� n
| j|� d S )N)rB r rx � insertBeforerI )r rC r� �textr r r �
insertText9 s zElement.insertTextc C s� | j j|j �}|j jtkrf| j jrf| j j|d jtkrf| j j|d }| jj||j �}|j|� n| j j||j � | |_ d S )NrN )
rG �indexr5 r
rw r rx ry �insertrs )r rH ZrefNoder� Zold_nodeZnew_strr r r r� @ s zElement.insertBeforec C s |j j� d S )N)rG rv )r rH r r r � removeChildL s zElement.removeChildc C s� | j }|j }|j}|jdd�}t|j�dkr>|jd }|j}n
d}|j}|j}t|�dkr�|d } |rn|| _n|| _|| _|r�| |_n| |_|r�| |_|d jdd�}
||
_|r�|
|_d|
_x |D ]} || _|jj | � q�W g |_||_dS )z1Move all of this tag's children into another tag.Fr rN NTru ru )
rG Z next_siblingrz rn rw r{ Zprevious_elementZprevious_siblingrs rF ) r Z
new_parentrG Znew_parent_elementZfinal_next_elementZnew_parents_last_descendantZnew_parents_last_childZ(new_parents_last_descendant_next_elementZ to_appendZ first_childZlast_childs_last_descendantr_ r r r �reparentChildrenO s>
zElement.reparentChildrenc C sF | j j| jj| j�}t|| j | j�}x | jD ]\}}||j|< q,W |S )N)r r> rG r$ r? r7 r] )r r@ rH �keyr^ r r r � cloneNode� s
zElement.cloneNodec C s | j jS )N)rG rw )r r r r �
hasContent� s zElement.hasContentc C s( | j d krtd | jfS | j | jfS d S )NZhtml)r? r r$ )r r r r � getNameTuple� s
zElement.getNameTuple)N)r+ r, r- r4 rI r| r �propertyr] r� r� r� r� r� r� r� Z nameTupler r r r r7 � s 6
Br7 c @ s e Zd Zdd� Zdd� ZdS )rB c C s t jj| d � || _|| _d S )N)rK rr r4 rG r )r rG r r r r r4 � s zTextNode.__init__c C s t �d S )N)�NotImplementedError)r r r r r� � s zTextNode.cloneNodeN)r+ r, r- r4 r� r r r r rB � s rB )"�__all__r rd Z bs4.builderr r r r Z bs4.elementr r r Zhtml5lib.constantsr r
r r r
r Zhtml5lib.treebuildersr rK r � ImportError�er r rL r( �objectrh rr r7 rB r r r r �<module> s& 9_ E