U
��.e�� � @ s d dl mZmZmZ d dlmZmZ d dlZd dlm Z ddl
m Z ddl
m Z ddl
m
Z
dd lmZ dd
l
mZ dd lmZmZmZmZmZmZmZmZmZmZm Z m Z m Z m Z m!Z!m"Z" d!dd�Z#d"dd�Z$dd� Z%G dd� de&�Z'ej(dd� �Z)dd� Z*d#d d �Z+G d d � d e,�Z-dS )$� )�absolute_import�division�unicode_literals)�with_metaclass�viewkeysN)� OrderedDict� )� _inputstream)�
_tokenizer)� treebuilders)�Marker)�_utils)�spaceCharacters�asciiUpper2Lower�specialElements�headingElements�
cdataElements�rcdataElements�
tokenTypes�
tagTokenTypes�
namespaces� htmlIntegrationPointElements�"mathmlTextIntegrationPointElements�adjustForeignAttributes�adjustMathMLAttributes�adjustSVGAttributes�E�_ReparseException�etreeTc K s$ t �|�}t||d�}|j| f|�S )a� Parse an HTML document as a string or file-like object into a tree
:arg doc: the document to parse as a string or file-like object
:arg treebuilder: the treebuilder to use when parsing
:arg namespaceHTMLElements: whether or not to namespace HTML elements
:returns: parsed tree
Example:
>>> from html5lib.html5parser import parse
>>> parse('This is a doc
')
��namespaceHTMLElements)r �getTreeBuilder�
HTMLParser�parse)�doc� treebuilderr �kwargs�tb�p� r) �D/usr/lib/python3.8/site-packages/pip/_vendor/html5lib/html5parser.pyr# s
r# �divc K s, t �|�}t||d�}|j| fd|i|��S )a# Parse an HTML fragment as a string or file-like object into a tree
:arg doc: the fragment to parse as a string or file-like object
:arg container: the container context to parse the fragment in
:arg treebuilder: the treebuilder to use when parsing
:arg namespaceHTMLElements: whether or not to namespace HTML elements
:returns: parsed tree
Example:
>>> from html5lib.html5libparser import parseFragment
>>> parseFragment('this is a fragment')
r � container)r r! r" �
parseFragment)r$ r, r% r r& r' r( r) r) r* r- 2 s
r- c s G � fdd�dt �}|S )Nc s e Zd Z� fdd�ZdS )z-method_decorator_metaclass..Decoratedc s> |� � D ]$\}}t|tj�r$� |�}|||< qt�| |||�S �N)�items�
isinstance�types� FunctionType�type�__new__)�metaZ classname�basesZ classDictZ
attributeNameZ attribute��functionr) r* r4 M s
z5method_decorator_metaclass..Decorated.__new__N)�__name__�
__module__� __qualname__r4 r) r7 r) r* � DecoratedL s r< )r3 )r8 r< r) r7 r* �method_decorator_metaclassK s r= c @ s� e Zd ZdZd+dd�Zd,dd �Zd
d � Zed d
� �Zdd� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zd-dd �Zd d � Zd d � Zd!d"� Zd#d$� Zd%d&� Zd'd(� Zd)d*� ZdS ).r" z]HTML parser
Generates a tree structure from a stream of (possibly malformed) HTML.
NFTc sL |� _ |dkrt�d�}||�� _g � _t� fdd�t|��� D ��� _dS )a�
:arg tree: a treebuilder class controlling the type of tree that will be
returned. Built in treebuilders can be accessed through
html5lib.treebuilders.getTreeBuilder(treeType)
:arg strict: raise an exception when a parse error is encountered
:arg namespaceHTMLElements: whether or not to namespace HTML elements
:arg debug: whether or not to enable debug mode which logs things
Example:
>>> from html5lib.html5parser import HTMLParser
>>> parser = HTMLParser() # generates parser with etree builder
>>> parser = HTMLParser('lxml', strict=True) # generates parser with lxml builder which is strict
Nr c s g | ]\}}||� � j �f�qS r) )�tree)�.0�name�cls��selfr) r* �
z s z'HTMLParser.__init__..) �strictr r! r> �errors�dict� getPhasesr/ �phases)rC r> rE r �debugr) rB r* �__init__^ s
�zHTMLParser.__init__r+ c K sh || _ || _|| _tj|fd| i|��| _| �� z | �� W n$ tk
rb | �� | �� Y nX d S )N�parser) �
innerHTMLModer, � scriptingr
Z
HTMLTokenizer� tokenizer�reset�mainLoopr )rC �stream� innerHTMLr, rN r& r) r) r* �_parse} s zHTMLParser._parsec C s� | j �� d| _g | _g | _d| _| jr�| j�� | _ | j t
krL| j j | j _
n0| j tkrd| j j| j _
n| j dkr|| j j| j _
n | jd | _| j�� | �� nd| _ | jd | _d | _d | _d| _d S )NFz no quirks� plaintext�
beforeHtml�initialT)r> rP �
firstStartTagrF �log�
compatModerM r, �lowerrS r rO � rcdataState�stater � rawtextState�plaintextStaterI �phase�insertHtmlElement�resetInsertionModeZ lastPhaseZbeforeRCDataPhase�
framesetOKrB r) r) r* rP � s*
zHTMLParser.resetc C s t | d�sdS | jjjd jS )z�Name of the character encoding that was used to decode the input stream, or
:obj:`None` if that is not determined yet
rO Nr )�hasattrrO rR � charEncodingr@ rB r) r) r* �documentEncoding� s
zHTMLParser.documentEncodingc C sJ |j dkr6|jtd kr6d|jko4|jd �t�dkS |j|j ftkS d S )N�annotation-xml�mathml�encoding)z text/htmlzapplication/xhtml+xml)r@ � namespacer �
attributes� translater r �rC �elementr) r) r* �isHTMLIntegrationPoint� s
�
���z!HTMLParser.isHTMLIntegrationPointc C s |j |jftkS r. )rj r@ r rm r) r) r* � isMathMLTextIntegrationPoint� s z'HTMLParser.isMathMLTextIntegrationPointc C sh t d }t d }t d }t d }t d }t d }t d }| �� D �]�}d } |}
|
d k �r�|
} | jjrp| jjd nd } | r~| jnd } | r�| jnd }
|
d }||kr�| �|
d
|
�d i �� d }
qNt| jj�d k�sd| | jj k�sd| �
| ��r||k�r
|d
t ddg�k�sd|||fk�sd| t d k�rH|
dk�rH||k�rH|d
dk�sd| �
| ��rl||||fk�rl| j}n
| jd }||k�r�|�|
�}
qN||k�r�|�|
�}
qN||k�r�|�|
�}
qN||k�r�|�|
�}
qN||k�r�|�|
�}
qN||krN|�|
�}
qN||kr@| d r@| d s@| �dd
| d
i� q@d}g }|�rd|�| j� | j�� }|�r.| j|k�s.t��q.d S )N�
CharactersZSpaceCharacters�StartTag�EndTag�CommentZDoctype�
ParseError�����r3 �data�datavarsr r@ ZmglyphZ
malignmarkrh rg �svg�inForeignContent� selfClosing�selfClosingAcknowledgedz&non-void-element-with-trailing-solidusT)r �normalizedTokensr> � openElementsrj r@ �
parseError�get�len�defaultNamespacerp � frozensetr ro r` rI �processCharacters�processSpaceCharacters�processStartTag�
processEndTag�processComment�processDoctype�append�
processEOF�AssertionError)rC ZCharactersTokenZSpaceCharactersTokenZ
StartTagTokenZ EndTagTokenZ CommentTokenZ DoctypeTokenZParseErrorToken�tokenZ
prev_token� new_token� currentNodeZcurrentNodeNamespaceZcurrentNodeNamer3 r` Z reprocessrI r) r) r* rQ � s�
����
�
���
�
� �
�
�
zHTMLParser.mainLoopc c s | j D ]}| �|�V qd S r. )rO �normalizeToken�rC r� r) r) r* r} s
zHTMLParser.normalizedTokensc O s | j |ddf|�|� | j�� S )a� Parse a HTML document into a well-formed tree
:arg stream: a file-like object or string containing the HTML to be parsed
The optional encoding parameter must be a string that indicates
the encoding. If specified, that encoding will be used,
regardless of any BOM or later declaration (such as in a meta
element).
:arg scripting: treat noscript elements as if JavaScript was turned on
:returns: parsed tree
Example:
>>> from html5lib.html5parser import HTMLParser
>>> parser = HTMLParser()
>>> parser.parse('This is a doc
')
FN)rT r> Z getDocument�rC rR �argsr& r) r) r* r# s zHTMLParser.parsec O s | j |df|�|� | j�� S )aZ Parse a HTML fragment into a well-formed tree fragment
:arg container: name of the element we're setting the innerHTML
property if set to None, default to 'div'
:arg stream: a file-like object or string containing the HTML to be parsed
The optional encoding parameter must be a string that indicates
the encoding. If specified, that encoding will be used,
regardless of any BOM or later declaration (such as in a meta
element)
:arg scripting: treat noscript elements as if JavaScript was turned on
:returns: parsed tree
Example:
>>> from html5lib.html5libparser import HTMLParser
>>> parser = HTMLParser()
>>> parser.parseFragment('this is a fragment')
T)rT r> Z getFragmentr� r) r) r* r- $ s zHTMLParser.parseFragment�XXX-undefined-errorc C s@ |d kr i }| j �| jj�� ||f� | jr r~ r@ rS r� rj r� rI r` )rC ZlastZnewModes�nodeZnodeNameZ new_phaser) r) r* rb ` sD �
z HTMLParser.resetInsertionModec C sR |dks t �| j�|� |dkr.| jj| j_n | jj| j_| j| _| j d | _d S )N)�RAWTEXT�RCDATAr� �text)
r� r> �
insertElementrO r^ r] r\ r` �
originalPhaserI )rC r� Z contentTyper) r) r* �parseRCDataRawtext� s z HTMLParser.parseRCDataRawtext)NFTF)Fr+ F)r� N)r9 r: r; �__doc__rK rT rP �propertyrf ro rp rQ r} r# r- r r� r r r r� rb r� r) r) r) r* r" W s(
"
C
,r" c s" dd� }dd� }G dd� dt || |���� G dd� d� �}G d d
� d
� �}G � fd d �d � �}G � fd
d�d� �}G � fdd�d� �}G � fdd�d� �}G � fdd�d� �} G � fdd�d� �}
G � fdd�d� �} G � fdd�d� �} G � fdd �d � �}
G � fd d �d � �}G � fd d �d � �}G � fd!d"�d"� �}G � fd#d$�d$� �}G � fd%d&�d&� �}G � fd'd(�d(� �}G � fd)d*�d*� �}G � fd+d,�d,� �}G � fd-d.�d.� �}G � fd/d0�d0� �}G � fd1d2�d2� �}G � fd3d4�d4� �}||||||| |
| | |
||||||||||||d5�S )6Nc s( t dd� t�� D ���� �fdd�}|S )z4Logger that records which phase processes each tokenc s s | ]\}}||fV qd S r. r) )r? �key�valuer) r) r* � � s z)getPhases..log..c s� � j �d�r�t|�dkr�|d }zd�|d i}W n � Y nX |d tkr\|d |d< | jj�| jjjj | jj j
j | j
j � j |f� � | f|�|�S � | f|�|�S d S )NZprocessr r3 r@ ) r9 �
startswithr� r rL rY r� rO r] r` � __class__)rC r� r& r� �info�r8 Z
type_namesr) r* �wrapped� s
�z'getPhases..log..wrapped)rG r r/ )r8 r� r) r� r* rY � s
�zgetPhases..logc S s | r t |�S tS d S r. )r= r3 )Z
use_metaclassZmetaclass_funcr) r) r* � getMetaclass� s z getPhases..getMetaclassc @ sX e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
d � Zd d
� Z dd� Z
dd� Z dd� Z dS )zgetPhases..PhasezNBase class for helper object that implements each phase of processing
c S s || _ || _d S r. )rL r> �rC rL r> r) r) r* rK � s z!getPhases..Phase.__init__c S s t �d S r. )�NotImplementedErrorrB r) r) r* r� � s z#getPhases..Phase.processEOFc S s | j �|| j jd � d S )Nrv �r> �
insertCommentr~ r� r) r) r* r� � s z'getPhases..Phase.processCommentc S s | j �d� d S )Nzunexpected-doctype�rL r r� r) r) r* r� � s z'getPhases..Phase.processDoctypec S s | j �|d � d S �Nrw �r> �
insertTextr� r) r) r* r� � s z*getPhases..Phase.processCharactersc S s | j �|d � d S r� r� r� r) r) r* r� � s z/getPhases..Phase.processSpaceCharactersc S s | j |d |�S �Nr@ )�startTagHandlerr� r) r) r* r� � s z(getPhases..Phase.processStartTagc S sf | j js |d dkr | j �d� |d �� D ],\}}|| jjd jkr,|| jjd j|< q,d| j _d S )Nr@ r� z
non-html-rootrw r F)rL rX r r/ r> r~ rk �rC r� �attrr� r) r) r* � startTagHtml� s z%getPhases..Phase.startTagHtmlc S s | j |d |�S r� )�
endTagHandlerr� r) r) r* r� � s z&getPhases..Phase.processEndTagN)
r9 r: r; r� rK r� r� r� r� r� r� r� r� r) r) r) r* �Phase� s
r� c @ sL e Zd Zdd� Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dS )z getPhases..InitialPhasec S s d S r. r) r� r) r) r* r� � s z6getPhases..InitialPhase.processSpaceCharactersc S s | j �|| j j� d S r. �r> r� Zdocumentr� r) r) r* r� � s z.getPhases..InitialPhase.processCommentc S s� |d }|d }|d }|d }|dks@|d k s@|d k rL|dkrL| j �d� |d krXd}| j�|� |dkrv|�t�}|r�|d dks�|�d �s�|d
ks�|�d �r�|d ks�|r�|�� d kr�d
| j _n$|�d�s�|�d �r�|d k r�d| j _| j j d | j _
d S )Nr@ �publicId�systemId�correctr� zabout:legacy-compatzunknown-doctype� )7z*+//silmaril//dtd html pro v0r11 19970101//z4-//advasoft ltd//dtd html 3.0 aswedit + extensions//z*-//as//dtd html 3.0 aswedit + extensions//z -//ietf//dtd html 2.0 level 1//z -//ietf//dtd html 2.0 level 2//z&-//ietf//dtd html 2.0 strict level 1//z&-//ietf//dtd html 2.0 strict level 2//z -//ietf//dtd html 2.0 strict//z-//ietf//dtd html 2.0//z-//ietf//dtd html 2.1e//z-//ietf//dtd html 3.0//z -//ietf//dtd html 3.2 final//z-//ietf//dtd html 3.2//z-//ietf//dtd html 3//z-//ietf//dtd html level 0//z-//ietf//dtd html level 1//z-//ietf//dtd html level 2//z-//ietf//dtd html level 3//z"-//ietf//dtd html strict level 0//z"-//ietf//dtd html strict level 1//z"-//ietf//dtd html strict level 2//z"-//ietf//dtd html strict level 3//z-//ietf//dtd html strict//z-//ietf//dtd html//z(-//metrius//dtd metrius presentational//z5-//microsoft//dtd internet explorer 2.0 html strict//z.-//microsoft//dtd internet explorer 2.0 html//z0-//microsoft//dtd internet explorer 2.0 tables//z5-//microsoft//dtd internet explorer 3.0 html strict//z.-//microsoft//dtd internet explorer 3.0 html//z0-//microsoft//dtd internet explorer 3.0 tables//z#-//netscape comm. corp.//dtd html//z*-//netscape comm. corp.//dtd strict html//z*-//o'reilly and associates//dtd html 2.0//z3-//o'reilly and associates//dtd html extended 1.0//z;-//o'reilly and associates//dtd html extended relaxed 1.0//zN-//softquad software//dtd hotmetal pro 6.0::19990601::extensions to html 4.0//zE-//softquad//dtd hotmetal pro 4.0::19971010::extensions to html 4.0//z$-//spyglass//dtd html 2.0 extended//z+-//sq//dtd html 2.0 hotmetal + extensions//z--//sun microsystems corp.//dtd hotjava html//z4-//sun microsystems corp.//dtd hotjava strict html//z -//w3c//dtd html 3 1995-03-24//z -//w3c//dtd html 3.2 draft//z -//w3c//dtd html 3.2 final//z-//w3c//dtd html 3.2//z -//w3c//dtd html 3.2s draft//z -//w3c//dtd html 4.0 frameset//z#-//w3c//dtd html 4.0 transitional//z(-//w3c//dtd html experimental 19960712//z&-//w3c//dtd html experimental 970421//z-//w3c//dtd w3 html//z-//w3o//dtd w3 html 3.0//z#-//webtechs//dtd mozilla html 2.0//z -//webtechs//dtd mozilla html//)z$-//w3o//dtd w3 html strict 3.0//en//z"-/w3c/dtd html 4.0 transitional/enr� )z -//w3c//dtd html 4.01 frameset//z$-//w3c//dtd html 4.01 transitional//z:http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd�quirks)z -//w3c//dtd xhtml 1.0 frameset//z$-//w3c//dtd xhtml 1.0 transitional//zlimited quirksrV ) rL r r> Z
insertDoctyperl r r� r[ rZ rI r` )rC r� r@ r� r� r� r) r) r* r� � sX ��
��9�<��?�@�@
�A
����z.getPhases..InitialPhase.processDoctypec S s d| j _| j jd | j _d S )Nr� rV )rL rZ rI r` rB r) r) r* � anythingElseM s z,getPhases..InitialPhase.anythingElsec S s | j �d� | �� |S )Nz expected-doctype-but-got-chars�rL r r� r� r) r) r* r� Q s z1getPhases..InitialPhase.processCharactersc S s" | j �dd|d i� | �� |S )Nz"expected-doctype-but-got-start-tagr@ r� r� r) r) r* r� V s
�z/getPhases..InitialPhase.processStartTagc S s" | j �dd|d i� | �� |S )Nz expected-doctype-but-got-end-tagr@ r� r� r) r) r* r� \ s
�z-getPhases..InitialPhase.processEndTagc S s | j �d� | �� dS )Nz expected-doctype-but-got-eofTr� rB r) r) r* r� b s z*getPhases..InitialPhase.processEOFN) r9 r: r; r� r� r� r� r� r� r� r� r) r) r) r* � InitialPhase� s _r� c @ sD e Zd Zdd� Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dS )z"getPhases..BeforeHtmlPhasec S s& | j �tdd�� | jjd | j_d S )Nr� rr r� )r> Z
insertRoot�impliedTagTokenrL rI r` rB r) r) r* ra i s z4getPhases..BeforeHtmlPhase.insertHtmlElementc S s | � � dS �NT�ra rB r) r) r* r� n s z-getPhases..BeforeHtmlPhase.processEOFc S s | j �|| j j� d S r. r� r� r) r) r* r� r s z1getPhases..BeforeHtmlPhase.processCommentc S s d S r. r) r� r) r) r* r� u s z9getPhases..BeforeHtmlPhase.processSpaceCharactersc S s | � � |S r. r� r� r) r) r* r� x s z4getPhases..BeforeHtmlPhase.processCharactersc S s |d dkrd| j _| �� |S )Nr@ r� T)rL rX ra r� r) r) r* r� | s z2getPhases..BeforeHtmlPhase.processStartTagc S s4 |d dkr$| j �dd|d i� n | �� |S d S )Nr@ �r� r� r� �brz unexpected-end-tag-before-html)rL r ra r� r) r) r* r� � s
�z0getPhases..BeforeHtmlPhase.processEndTagN)
r9 r: r; ra r� r� r� r� r� r� r) r) r) r* �BeforeHtmlPhaseg s r� c sX e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dS )z"getPhases..BeforeHeadPhasec sV � � | ||� t�d| jfd| jfg�| _| j| j_t�d| jfg�| _ | j
| j _d S )Nr� r� r� ) rK r
�MethodDispatcherr� � startTagHeadr� �
startTagOther�default�endTagImplyHeadr� � endTagOtherr� �r� r) r* rK � s �
�z+getPhases..BeforeHeadPhase.__init__c S s | � tdd�� dS )Nr� rr T�r� r� rB r) r) r* r� � s z-getPhases..BeforeHeadPhase.processEOFc S s d S r. r) r� r) r) r* r� � s z9getPhases..BeforeHeadPhase.processSpaceCharactersc S s | � tdd�� |S �Nr� rr r� r� r) r) r* r� � s z4getPhases..BeforeHeadPhase.processCharactersc S s | j jd �|�S �Nr� �rL rI r� r� r) r) r* r� � s z/getPhases..BeforeHeadPhase.startTagHtmlc S s0 | j �|� | j jd | j _| jjd | j_d S )Nrv �inHead)r> r� r~ � headPointerrL rI r` r� r) r) r* r� � s z/getPhases..BeforeHeadPhase.startTagHeadc S s | � tdd�� |S r� r� r� r) r) r* r� � s z0getPhases..BeforeHeadPhase.startTagOtherc S s | � tdd�� |S r� r� r� r) r) r* r� � s z2getPhases..BeforeHeadPhase.endTagImplyHeadc S s | j �dd|d i� d S )Nzend-tag-after-implied-rootr@ r� r� r) r) r* r� � s
�z.getPhases..BeforeHeadPhase.endTagOtherN) r9 r: r; rK r� r� r� r� r� r� r� r� r) r� r) r* �BeforeHeadPhase� s r� c s� e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� Zdd � Zd d � Zd d � Zd!S )"z getPhases..InHeadPhasec s� � � | ||� t�d| jfd| jfd| jfd| jfd| jfd| jfd| j fd| j
fg�| _ | j | j _
t�d| jfd | jfg�| _| j| j_
d S )
Nr� �title)�noframes�style�noscript�script)�base�basefont�bgsound�command�linkr5 r� )r� r� r� )rK r
r� r� �
startTagTitle�startTagNoFramesStyle�startTagNoscript�startTagScript�startTagBaseLinkCommand� startTagMetar� r� r� r� �
endTagHead�endTagHtmlBodyBrr� r� r� r� r) r* rK � s& ��
�z'getPhases..InHeadPhase.__init__c S s | � � dS r� �r� rB r) r) r* r� � s z)getPhases..InHeadPhase.processEOFc S s | � � |S r. r r� r) r) r* r� � s z0getPhases..InHeadPhase.processCharactersc S s | j jd �|�S r� r� r� r) r) r* r� � s z+getPhases..InHeadPhase.startTagHtmlc S s | j �d� d S )Nz!two-heads-are-not-better-than-oner� r� r) r) r* r� � s z+getPhases..InHeadPhase.startTagHeadc S s$ | j �|� | j j�� d|d< d S �NTr| �r> r� r~ �popr� r) r) r* r� � s z6getPhases..InHeadPhase.startTagBaseLinkCommandc S s� | j �|� | j j�� d|d< |d }| jjjjd dkr�d|krZ| jjj�|d � nVd|kr�d|kr�|d � � d kr�t
� |d � d
��}t
�
|�}|�� }| jjj�|� d S ) NTr| rw r Z tentative�charsetZcontentz
http-equivz content-typezutf-8)r> r� r~ r rL rO rR re ZchangeEncodingr[ r Z
EncodingBytes�encodeZContentAttrParserr# )rC r� rk rw rL �codecr) r) r* r� � s ��
z+getPhases..InHeadPhase.startTagMetac S s | j �|d� d S )Nr� �rL r� r� r) r) r* r� � s z,getPhases..InHeadPhase.startTagTitlec S s | j �|d� d S )Nr� r r� r) r) r* r� � s z4getPhases..InHeadPhase.startTagNoFramesStylec S s8 | j jr| j �|d� n | j�|� | j jd | j _d S )Nr� �inHeadNoscript)rL rN r� r> r� rI r` r� r) r) r* r� � s z/getPhases..InHeadPhase.startTagNoscriptc S s< | j �|� | jjj| jj_| jj| j_| jjd | j_d S )Nr� ) r> r� rL rO ZscriptDataStater] r` r� rI r� r) r) r* r� s z-getPhases..InHeadPhase.startTagScriptc S s | � � |S r. r r� r) r) r* r� s z,getPhases..InHeadPhase.startTagOtherc S s: | j jj�� }|jdks&td|j ��| j jd | j _d S )Nr� zExpected head got %s� afterHead�rL r> r~ r r@ r� rI r` �rC r� r� r) r) r* r s z)getPhases..InHeadPhase.endTagHeadc S s | � � |S r. r r� r) r) r* r s z/getPhases..InHeadPhase.endTagHtmlBodyBrc S s | j �dd|d i� d S �N�unexpected-end-tagr@ r� r� r) r) r* r� s z*getPhases..InHeadPhase.endTagOtherc S s | � td�� d S )Nr� )r r� rB r) r) r* r� s z+getPhases..InHeadPhase.anythingElseN)r9 r: r; rK r� r� r� r� r� r� r� r� r� r� r� r r r� r� r) r� r) r* � InHeadPhase� s r c sx e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� ZdS ) z&getPhases..InHeadNoscriptPhasec sf � � | ||� t�d| jfd| jfd| jfg�| _| j| j_t�d| j fd| j
fg�| _ | j | j _d S )Nr� )r� r� r� r5 r� r� )r� r� r� r� )
rK r
r� r� r� �startTagHeadNoscriptr� r� r� �endTagNoscript�endTagBrr� r� r� r� r) r* rK s �
�z/getPhases..InHeadNoscriptPhase.__init__c S s | j �d� | �� dS )Nzeof-in-head-noscriptTr� rB r) r) r* r� 0 s z1getPhases..InHeadNoscriptPhase.processEOFc S s | j jd �|�S �Nr� )rL rI r� r� r) r) r* r� 5 s z5getPhases..InHeadNoscriptPhase.processCommentc S s | j �d� | �� |S )Nzchar-in-head-noscriptr� r� r) r) r* r� 8 s z8getPhases..InHeadNoscriptPhase.processCharactersc S s | j jd �|�S r �rL rI r� r� r) r) r* r� = s z=getPhases..InHeadNoscriptPhase.processSpaceCharactersc S s | j jd �|�S r� r� r� r) r) r* r� @ s z3getPhases..InHeadNoscriptPhase.startTagHtmlc S s | j jd �|�S r r� r� r) r) r* r� C s z>getPhases..InHeadNoscriptPhase.startTagBaseLinkCommandc S s | j �dd|d i� d S �N�unexpected-start-tagr@ r� r� r) r) r* r F s z;getPhases..InHeadNoscriptPhase.startTagHeadNoscriptc S s" | j �dd|d i� | �� |S �Nz unexpected-inhead-noscript-tagr@ r� r� r) r) r* r� I s z4getPhases..InHeadNoscriptPhase.startTagOtherc S s: | j jj�� }|jdks&td|j ��| j jd | j _d S )Nr� zExpected noscript got %sr� r r
r) r) r* r N s z5getPhases..InHeadNoscriptPhase.endTagNoscriptc S s" | j �dd|d i� | �� |S r r� r� r) r) r* r S s z/getPhases..InHeadNoscriptPhase.endTagBrc S s | j �dd|d i� d S r r� r� r) r) r* r� X s z2getPhases..InHeadNoscriptPhase.endTagOtherc S s | � td�� d S )Nr� )r r� rB r) r) r* r� [ s z3getPhases..InHeadNoscriptPhase.anythingElseN)r9 r: r; rK r� r� r� r� r� r� r r� r r r� r� r) r� r) r* �InHeadNoscriptPhase s r c sp e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� ZdS )z!getPhases..AfterHeadPhasec sn � � | ||� t�d| jfd| jfd| jfd| jfd| jfg�| _| j | j_
t�d| j fg�| _ | j
| j _
d S )Nr� r� r� ) r� r� r� r� r5 r� r� r� r� r� )r� r� r� )rK r
r� r� � startTagBody�startTagFrameset�startTagFromHeadr� r� r� r� r r� r� r� r� r) r* rK ` s ��
�
z*getPhases..AfterHeadPhase.__init__c S s | � � dS r� r rB r) r) r* r� q s z,getPhases..AfterHeadPhase.processEOFc S s | � � |S r. r r� r) r) r* r� u s z3getPhases..AfterHeadPhase.processCharactersc S s | j jd �|�S r� r� r� r) r) r* r� y s z.getPhases..AfterHeadPhase.startTagHtmlc S s( d| j _| j�|� | j jd | j _d S )NFr� )rL rc r> r� rI r` r� r) r) r* r | s z.getPhases..AfterHeadPhase.startTagBodyc S s | j �|� | jjd | j_d S )Nr� )r> r� rL rI r` r� r) r) r* r � s z2getPhases..AfterHeadPhase.startTagFramesetc S sr | j �dd|d i� | jj�| jj� | j jd �|� | jjd d d� D ] }|jdkrL| jj� |� qnqLd S )Nz#unexpected-start-tag-out-of-my-headr@ r� rv r� )
rL r r> r~ r� r� rI r� r@ �remover
r) r) r* r � s
�
z2getPhases..AfterHeadPhase.startTagFromHeadc S s | j �dd|d i� d S r r� r� r) r) r* r� � s z.getPhases..AfterHeadPhase.startTagHeadc S s | � � |S r. r r� r) r) r* r� � s z/getPhases..AfterHeadPhase.startTagOtherc S s | � � |S r. r r� r) r) r* r � s z2getPhases..AfterHeadPhase.endTagHtmlBodyBrc S s | j �dd|d i� d S r r� r� r) r) r* r� � s z-getPhases..AfterHeadPhase.endTagOtherc S s. | j �tdd�� | jjd | j_d| j_d S )Nr� rr r� T)r> r� r� rL rI r` rc rB r) r) r* r� � s z.getPhases..AfterHeadPhase.anythingElseN)r9 r: r; rK r� r� r� r r r r� r� r r� r� r) r� r) r* �AfterHeadPhase_ s
r c s� e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� Zdd � Zd d � Zd d � Zd!d"� Zd#d$� Zd%d&� Zd'd(� Zd)d*� Zd+d,� Zd-d.� Zd/d0� Zd1d2� Zd3d4� Z d5d6� Z d7d8� Z d9d:� Z d;d<� Z d=d>� Z!d?d@� Z"dAdB� Z#dCdD� Z$dEdF� Z%dGdH� Z&dIdJ� Z'dKdL� Z(dMdN� Z)dOdP� Z*dQdR� Z+dSdT� Z,dUdV� Z-dWdX� Z.dYdZ� Z/d[d\� Z0d]d^� Z1d_d`� Z2dadb� Z3dcdd� Z4dedf� Z5dgS )hz getPhases..InBodyPhasec $ s� � � | ||� | j| _t�d| jfd| jfd| jfd| jfd| j ft
| j fd| j fd| j
fd| jfd | jfd
| jfd | jfd | jfd
| jfd| jfd| jfd| jfd| jfd| jfd| jfd| jfd| jfd| j fd| j fd| j fd| j fd| j fd| j!fd | j"fd | j#fd | j$fd | j%fd | j&fg!�| _'| j(| j'_)t�d| j*fd| j+fd!| j,fd| j-fd"| j.fd#| j/ft
| j0fd$| j1fd| j2fd%| j3fg
�| _4| j5| j4_)d S )&Nr� ) r� r� r� r� r� r5 r� r� r� r� r� )�address�article�aside�
blockquote�center�details�dirr+ �dl�fieldset�
figcaption�figure�footer�header�hgroup�main�menu�nav�olr( �section�summary�ul)�pre�listing�form)�li�dd�dtrU �a) �b�big�code�em�font�i�s�small�strike�strong�tt�u�nobr�button)ZappletZmarquee�objectZxmpr� )Zarear� �embed�img�keygenZwbr)Zparam�sourceZtrack�input�hr�image�isindex�textareaZiframer� )Znoembedr� r� )�rp�rt)�option�optgroupZmathry ) r� �colr� �framer� r� r� r� r� r� r� )r r r! r" rH r# r$ Zdialogr% r+ r& r' r( r) r* r+ r, r5 r- r. r/ r0 r4 r1 r2 r3 r( )r8 r9 r7 )r: r; r< r= r> r? r@ rG rA rB rC rD rE rF r� )6rK � processSpaceCharactersNonPrer� r
r� r� �startTagProcessInHeadr r �startTagClosePr �startTagHeading�startTagPreListing� startTagForm�startTagListItem�startTagPlaintext� startTagA�startTagFormatting� startTagNobr�startTagButton�startTagAppletMarqueeObject� startTagXmp�
startTagTable�startTagVoidFormatting�startTagParamSource�
startTagInput�
startTagHr�
startTagImage�startTagIsIndex�startTagTextarea�startTagIFramer� �startTagRawtext�startTagSelect� startTagRpRt� startTagOpt� startTagMath� startTagSvg�startTagMisplacedr� r� r� �
endTagBody�
endTagHtml� endTagBlock�
endTagForm�endTagP�endTagListItem�
endTagHeading�endTagFormatting�endTagAppletMarqueeObjectr r� r� r� r� r) r* rK � s� ������-
���z'getPhases..InBodyPhase.__init__c S s$ |j |j ko"|j|jko"|j|jkS r. )r@ rj rk )rC Znode1Znode2r) r) r* �isMatchingFormattingElement� s
�
�z:getPhases..InBodyPhase.isMatchingFormattingElementc S s� | j �|� | j jd }g }| j jd d d� D ](}|tkr@ qXq.| �||�r.|�|� q.t|�dksht�t|�dkr�| j j� |d � | j j�|� d S )Nrv � )
r> r� r~ �activeFormattingElementsr r� r� r� r� r )rC r� rn ZmatchingElementsr� r) r) r* �addFormattingElement� s z3getPhases..InBodyPhase.addFormattingElementc S s> t d�}| jjd d d� D ] }|j|kr| j�d� q:qd S )N) r8 r9 r7 r( r� r� r� r� r� r� r� r� rv z expected-closing-tag-but-got-eof)r� r> r~ r@ rL r )rC Zallowed_elementsr� r) r) r* r� s
z)getPhases..InBodyPhase.processEOFc S sf |d }| j | _|�d�rH| jjd jdkrH| jjd �� sH|dd � }|rb| j�� | j�|� d S )Nrw �
rv )r4 r5 rR r ) rY r� r� r> r~ r@ Z
hasContent�#reconstructActiveFormattingElementsr� )rC r� rw r) r) r* �!processSpaceCharactersDropNewline
s
��
z@getPhases..InBodyPhase.processSpaceCharactersDropNewlinec S sT |d dkrd S | j �� | j �|d � | jjrPtdd� |d D ��rPd| j_d S )Nrw � c S s g | ] }|t k�qS r) �r �r? �charr) r) r* rD " s �zDgetPhases..InBodyPhase.processCharacters..F)r> r� r� rL rc �anyr� r) r) r* r� s
��z0getPhases..InBodyPhase.processCharactersc S s | j �� | j �|d � d S r� )r> r� r� r� r) r) r* rY & s
z;getPhases..InBodyPhase.processSpaceCharactersNonPrec S s | j jd �|�S r r� r� r) r) r* rZ * s z4getPhases..InBodyPhase.startTagProcessInHeadc S s� | j �dddi� t| jj�dks4| jjd jdkrB| j js�t�nBd| j _|d � � D ],\}}|| jjd j
krV|| jjd j
|< qVd S )Nr r@ r� r Frw ) rL r r� r> r~ r@ rS r� rc r/ rk r� r) r) r* r - s �z+getPhases..InBodyPhase.startTagBodyc S s� | j �dddi� t| jj�dks4| jjd jdkrB| j js�t�np| j jsLnf| jjd j rv| jjd j �
| jjd � | jjd jdkr�| jj� � qv| j� |� | j j
d | j _d S ) Nr r@ r� r r� rv r� r� )rL r r� r> r~ r@ rS r� rc �parent� removeChildr r� rI r` r� r) r) r* r 8 s " z/getPhases..InBodyPhase.startTagFramesetc S s. | j jddd�r | �td�� | j �|� d S �Nr( rH ��variant)r> �elementInScoper{ r� r� r� r) r) r* r[ F s z-getPhases..InBodyPhase.startTagClosePc S s> | j jddd�r | �td�� | j �|� d| j_| j| _d S )Nr( rH r� F) r> r� r{ r� r� rL rc r� r� r� r) r) r* r] K s
z1getPhases..InBodyPhase.startTagPreListingc S sZ | j jr | j�dddi� n:| j jddd�r:| �td�� | j �|� | j jd | j _d S )Nr r@ r6 r( rH r� rv ) r> � formPointerrL r r� r{ r� r� r~ r� r) r) r* r^ R s z+getPhases..InBodyPhase.startTagFormc S s� d| j _dgddgddgd�}||d }t| jj�D ]@}|j|kr^| j j�t|jd�� qx|j t
kr6|jdkr6 qxq6| jj d d
d �r�| j j�td d�� | j� |� d S ) NFr7 r9 r8 )r7 r9 r8 r@ rs )r r+ r( r( rH r� )
rL rc �reversedr> r~ r@ r` r� r� � nameTupler r� r� )rC r� Z stopNamesMapZ stopNamesr� r) r) r* r_ [ s* �
�
��z/getPhases..InBodyPhase.startTagListItemc S s> | j jddd�r | �td�� | j �|� | jjj| jj_d S r� ) r> r� r{ r� r� rL rO r_ r] r� r) r) r* r` q s z0getPhases..InBodyPhase.startTagPlaintextc S sb | j jddd�r | �td�� | j jd jtkrR| j�dd|d i� | j j� � | j �
|� d S )Nr( rH r� rv r r@ ) r> r� r{ r� r~ r@ r rL r r r� r� r) r) r* r\ w s z.getPhases..InBodyPhase.startTagHeadingc S s~ | j �d�}|rf| j�dddd�� | �td�� || j jkrL| j j�|� || j jkrf| j j�|� | j � � | �
|� d S )Nr: �$unexpected-start-tag-implies-end-tag�Z startNameZendName) r> �!elementInActiveFormattingElementsrL r r~ r� r~ r r� r� r� )rC r� Z afeAElementr) r) r* ra s �
z(getPhases..InBodyPhase.startTagAc S s | j �� | �|� d S r. )r> r� r� r� r) r) r* rb � s
z1getPhases..InBodyPhase.startTagFormattingc S sP | j �� | j �d�rB| j�dddd�� | �td�� | j �� | �|� d S )NrG r� r� )r> r� r� rL r r� r� r� r� r) r) r* rc � s
�
z+getPhases..InBodyPhase.startTagNobrc S sT | j �d�r2| j�dddd�� | �td�� |S | j �� | j �|� d| j_d S )NrH r� r� F) r> r� rL r r� r� r� r� rc r� r) r) r* rd � s �
z-getPhases..InBodyPhase.startTagButtonc S s0 | j �� | j �|� | j j�t� d| j_d S �NF)r> r� r� r� r� r rL rc r� r) r) r* re � s
z:getPhases..InBodyPhase.startTagAppletMarqueeObjectc S sB | j jddd�r | �td�� | j �� d| j_| j�|d� d S )Nr( rH r� Fr� )r> r� r{ r� r� rL rc r� r� r) r) r* rf � s
z*getPhases..InBodyPhase.startTagXmpc S sR | j jdkr*| jjddd�r*| �td�� | j�|� d| j _| j jd | j _ d S )Nr� r( rH r� Fr� )
rL rZ r> r� r� r� r� rc rI r` r� r) r) r* rg � s z,getPhases..InBodyPhase.startTagTablec S s6 | j �� | j �|� | j j�� d|d< d| j_d S )NTr| F)r> r� r� r~ r rL rc r� r) r) r* rh � s
z5getPhases..InBodyPhase.startTagVoidFormattingc S s@ | j j}| �|� d|d kr<|d d �t�dkr<|| j _d S )Nr3 rw �hidden)rL rc rh rl r )rC r� rc r) r) r* rj � s
�z,getPhases..InBodyPhase.startTagInputc S s$ | j �|� | j j�� d|d< d S r r r� r) r) r* ri � s z2getPhases..InBodyPhase.startTagParamSourcec S sJ | j jddd�r | �td�� | j �|� | j j�� d|d< d| j_d S )Nr( rH r� Tr| F) r> r� r{ r� r� r~ r rL rc r� r) r) r* rk � s z)getPhases..InBodyPhase.startTagHrc S s6 | j �dddd�� | �tdd|d |d d�� d S ) Nz unexpected-start-tag-treated-asrP rK �Z originalNameZnewNamerr rw r{ �rk r{ )rL r r� r� r� r) r) r* rl � s �
�z,getPhases..InBodyPhase.startTagImagec S s | j �dddi� | jjr d S i }d|d kr>|d d |d< | �tdd|d�� | �td d�� | �td
d�� d |d kr�|d d }nd }| �td
|d�� |d �� }d|kr�|d= d |kr�|d = d|d< | �tdd||d d�� | � td
�� | �td d�� | � td�� d S )Nzdeprecated-tagr@ rQ �actionrw r6 rr )rk rO Zlabel�promptz3This is a searchable index. Enter search keywords: rq �r3 rw rN r{ r� )
rL r r> r� r� r� r� r �copyr� )rC r� Z
form_attrsr� rk r) r) r* rm � s<
� �
�z.getPhases..InBodyPhase.startTagIsIndexc S s0 | j �|� | jjj| jj_| j| _d| j_d S r� ) r> r� rL rO r\ r] r� r� rc r� r) r) r* rn � s z/getPhases..InBodyPhase.startTagTextareac S s d| j _| �|� d S r� )rL rc rp r� r) r) r* ro s z-getPhases..InBodyPhase.startTagIFramec S s" | j jr| �|� n
| �|� d S r. )rL rN rp r� r� r) r) r* r� s z/getPhases..InBodyPhase.startTagNoscriptc S s | j �|d� dS )z8iframe, noembed noframes, noscript(if scripting enabled)r� Nr r� r) r) r* rp
s z.getPhases..InBodyPhase.startTagRawtextc S s@ | j jd jdkr$| jj�td�� | j �� | jj �|� d S �Nrv rU ) r> r~ r@ rL r` r� r� r� r� r� r) r) r* rs s
z*getPhases..InBodyPhase.startTagOptc S s� | j �� | j �|� d| j_| jj| jjd | jjd | jjd | jjd | jjd | jjd fkrx| jjd | j_n| jjd | j_d S )
NFr� r� r� r� r� r� �inSelectInTabler� )r> r� r� rL rc r` rI r� r) r) r* rq s
�z-getPhases..InBodyPhase.startTagSelectc S sB | j �d�r2| j �� | j jd jdkr2| j�� | j �|� d S )N�rubyrv )r> r� �generateImpliedEndTagsr~ r@ rL r r� r� r) r) r* rr % s
z+getPhases..InBodyPhase.startTagRpRtc S sZ | j �� | j�|� | j�|� td |d< | j �|� |d rV| j j�� d|d< d S )Nrh rj r{ Tr| ) r> r� rL r r r r� r~ r r� r) r) r* rt , s
z+getPhases..InBodyPhase.startTagMathc S sZ | j �� | j�|� | j�|� td |d< | j �|� |d rV| j j�� d|d< d S )Nry rj r{ Tr| ) r> r� rL r r r r� r~ r r� r) r) r* ru 8 s
z*getPhases..InBodyPhase.startTagSvgc S s | j �dd|d i� dS )a5 Elements that should be children of other elements that have a
different insertion mode; here they are ignored
"caption", "col", "colgroup", "frame", "frameset", "head",
"option", "optgroup", "tbody", "td", "tfoot", "th", "thead",
"tr", "noscript"
z unexpected-start-tag-ignoredr@ Nr� r� r) r) r* rv D s z0getPhases..InBodyPhase.startTagMisplacedc S s | j �� | j �|� d S r. )r> r� r� r� r) r) r* r� M s
z,getPhases..InBodyPhase.startTagOtherc S s� | j jddd�sD| �tdd�� | j�dddi� | �tdd�� nT| j �d� | j jd j dkrt| j�dddi� | j j�
� }|j dkr�| j j�
� }q�d S ) Nr( rH r� rr r r@ rs rv ) r> r� r[ r� rL r r{ r� r~ r@ r r
r) r) r* r{ Q s
z&getPhases..InBodyPhase.endTagPc S s~ | j �d�s| j�� d S | j jd jdkrj| j jdd � D ],}|jtd�kr<| j�dd|jd�� qjq<| jjd | j_d S )Nr� rv � )r8 r9 r7 rV rU r( rS rT r� r� r� r� r� r� r� r� �$expected-one-end-tag-but-got-another�ZgotNameZ expectedName� afterBody) r> r� rL r r~ r@ r� rI r` r
r) r) r* rw ^ s
�z)getPhases..InBodyPhase.endTagBodyc S s" | j �d�r | �td�� |S d S )Nr� )r> r� rw r� r� r) r) r* rx p s z)getPhases..InBodyPhase.endTagHtmlc S s� |d dkr| j | _| j�|d �}|r2| j�� | jjd j|d kr^| j�dd|d i� |r�| jj� � }|j|d kr�| jj� � }qnd S )Nr@ r4 rv �end-tag-too-early)
rY r� r> r� r� r~ r@ rL r r )rC r� ZinScoper� r) r) r* ry v s
z*getPhases..InBodyPhase.endTagBlockc S sv | j j}d | j _|d ks$| j �|�s8| j�dddi� n:| j �� | j jd |krd| j�dddi� | j j�|� d S )Nr r@ r6 rv zend-tag-too-early-ignored)r> r� r� rL r r� r~ r r
r) r) r* rz � s �
�z)getPhases..InBodyPhase.endTagFormc S s� |d dkrd}nd }| j j|d |d�sB| j�dd|d i� nf| j j|d d� | j jd j|d kr�| j�dd|d i� | j j�� }|j|d kr�| j j�� }q�d S ) Nr@ r7 �listr� r ��excluderv r� )r> r� rL r r� r~ r@ r )rC r� r� r� r) r) r* r| � s
� z-getPhases..InBodyPhase.endTagListItemc S s� t D ] }| j�|�r| j�� q$q| jjd j|d krP| j�dd|d i� t D ]8}| j�|�rT| jj�� }|jt kr�| jj�� }qp q�qTd S )Nrv r@ r� ) r r> r� r� r~ r@ rL r r )rC r� �itemr) r) r* r} � s
z,getPhases..InBodyPhase.endTagHeadingc S s d}|dk �r
|d7 }| j �|d �}|rD|| j jkrR| j �|j�sR| �|� dS || j jkr�| j�dd|d i� | j j� |� dS | j �|j�s�| j�dd|d i� dS || j jd kr�| j�d d|d i� | j j�
|�}d}| j j|d� D ]}|j t kr�|} �qq�|dk�rR| j j�
� }||k�r@| j j�
� }�q&| j j� |� dS | j j|d }| j j�
|�}| } }
d} | j j�
|
�} | d
k �rX| d7 } | d8 } | j j| }
|
| j jk�r�| j j� |
� �q�|
|k�rސqX| |k�r�| j j�
|
�d }|
�� }
|
| j j| j j�
|
�< |
| j j| j j�
|
�< |
}
| j�rF| j�| � |
�| � |
} �q�| j�rl| j�| � |jtd �k�r�| j �� \}}|�| |� n
|�| � |�� }
|�|
� |�|
� | j j� |� | j j�||
� | j j� |� | j j�| j j�
|�d |
� qdS ) z)The much-feared adoption agency algorithmr � r r@ Nzadoption-agency-1.2zadoption-agency-4.4rv zadoption-agency-1.3r� �r� r� r� r� r� )r> r� r~ r� r@ r� rL r r� r �indexr� r r Z cloneNoder� r� Z appendChildr� Z getTableMisnestedNodePosition� insertBeforeZreparentChildren�insert)rC r� ZouterLoopCounterZformattingElementZafeIndexZ
furthestBlockrn ZcommonAncestorZbookmarkZlastNoder� ZinnerLoopCounterr� Zcloner� r� r) r) r* r~ � s�
�
� �
� �� �
�z/getPhases..InBodyPhase.endTagFormattingc S s� | j �|d �r| j �� | j jd j|d krF| j�dd|d i� | j �|d �r�| j j�� }|j|d kr~| j j�� }qb| j �� d S )Nr@ rv r� ) r> r� r� r~ r@ rL r r � clearActiveFormattingElements)rC r� rn r) r) r* r T s
z8getPhases..InBodyPhase.endTagAppletMarqueeObjectc S s@ | j �dddd�� | j�� | j�tdd�� | jj�� d S )Nz unexpected-end-tag-treated-asr� z
br elementr� rr )rL r r> r� r� r� r~ r r� r) r) r* r ` s �
z'getPhases..InBodyPhase.endTagBrc S s� | j jd d d� D ]�}|j|d krz| j j|d d� | j jd j|d krb| j�dd|d i� | j j�� |krtqb q�q|jtkr| j�dd|d i� q�qd S )Nrv r@ r� r ) r> r~ r@ r� rL r r r� r r
r) r) r* r� g s
z*getPhases..InBodyPhase.endTagOtherN)6r9 r: r; rK r� r� r� r� r� rY rZ r r r[ r] r^ r_ r` r\ ra rb rc rd re rf rg rh rj ri rk rl rm rn ro r� rp rs rq rr rt ru rv r� r{ rw rx ry rz r| r} r~ r r r� r) r� r) r* � InBodyPhase� sh G
$ r� c s@ e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
S )z getPhases..TextPhasec sF � � | ||� t�g �| _| j| j_t�d| jfg�| _| j| j_d S �Nr� ) rK r
r� r� r� r� � endTagScriptr� r� r� r� r) r* rK v s
�z%getPhases..TextPhase.__init__c S s | j �|d � d S r� r� r� r) r) r* r� ~ s z.getPhases..TextPhase.processCharactersc S s8 | j �dd| jjd ji� | jj�� | j j| j _dS )Nz&expected-named-closing-tag-but-got-eofr@ rv T)rL r r> r~ r@ r r� r` rB r) r) r* r� � s � z'getPhases..TextPhase.processEOFc S s dst d|d ��d S )NFz4Tried to process start tag %s in RCDATA/RAWTEXT moder@ )r� r� r) r) r* r� � s z*getPhases..TextPhase.startTagOtherc S s* | j j�� }|jdkst�| jj| j_d S r� )r> r~ r r@ r� rL r� r` r
r) r) r* r� � s z)getPhases..TextPhase.endTagScriptc S s | j j�� | jj| j_d S r. )r> r~ r rL r� r` r� r) r) r* r� � s z(getPhases..TextPhase.endTagOtherN) r9 r: r; rK r� r� r� r� r� r) r� r) r* � TextPhaseu s r� c s� e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� Zdd � Zd d � Zd d � Zd!d"� Zd#d$� Zd%d&� Zd'S )(z getPhases..InTablePhasec
s� � � | ||� t�d| jfd| jfd| jfd| jfd| jfd| jfd| j fd| j
fd | j fd
| j fg
�| _
| j| j
_t�d| jfd | jfg�| _| j| j_d S ) Nr� r� r� rW �r� r� r� )r� r� r� r� )r� r� rN r6 ) r� r� rW r� r� r� r� r� r� r� r� )rK r
r� r� �startTagCaption�startTagColgroup� startTagCol�startTagRowGroup�startTagImplyTbodyrg �startTagStyleScriptrj r^ r� r� r� � endTagTable� endTagIgnorer� r� r� r� r) r* rK � s* �
��z(getPhases..InTablePhase.__init__c S s$ | j jd jdkr | j j�� q d S )Nrv )r� r� )r> r~ r@ r rB r) r) r* �clearStackToTableContext� s z8getPhases..InTablePhase.clearStackToTableContextc S s0 | j jd jdkr | j�d� n | jjs,t�d S )Nrv r� z eof-in-table�r> r~ r@ rL r rS r� rB r) r) r* r� � s z*getPhases..InTablePhase.processEOFc S s4 | j j}| j jd | j _|| j j_| j j�|� d S �N� inTableText)rL r` rI r� r� �rC r� r� r) r) r* r� � s
z6getPhases..InTablePhase.processSpaceCharactersc S s4 | j j}| j jd | j _|| j j_| j j�|� d S r� )rL r` rI r� r� r� r) r) r* r� � s
z1getPhases..InTablePhase.processCharactersc S s& d| j _| jjd �|� d| j _d S )NTr� F)r> �insertFromTablerL rI r� r� r) r) r* r� � s z*getPhases..InTablePhase.insertTextc S s6 | � � | jj�t� | j�|� | jjd | j_d S )Nr� ) r� r> r� r� r r� rL rI r` r� r) r) r* r� � s z/getPhases..InTablePhase.startTagCaptionc S s( | � � | j�|� | jjd | j_d S )Nr� �r� r> r� rL rI r` r� r) r) r* r� � s z0getPhases..InTablePhase.startTagColgroupc S s | � tdd�� |S )Nr� rr )r� r� r� r) r) r* r� � s z+getPhases..InTablePhase.startTagColc S s( | � � | j�|� | jjd | j_d S �Nr� r� r� r) r) r* r� � s z0getPhases..InTablePhase.startTagRowGroupc S s | � tdd�� |S )Nr� rr )r� r� r� r) r) r* r� � s z2getPhases..InTablePhase.startTagImplyTbodyc S s6 | j �dddd�� | j j�td�� | j js2|S d S )Nr� r� r� )rL r r` r� r� rS r� r) r) r* rg � s �z-getPhases..InTablePhase.startTagTablec S s | j jd �|�S r r� r� r) r) r* r� � s z3getPhases..InTablePhase.startTagStyleScriptc S sV d|d krH|d d � t�dkrH| j�d� | j�|� | jj�� n
| �|� d S )Nr3 rw r� z unexpected-hidden-input-in-table) rl r rL r r> r� r~ r r� r� r) r) r* rj � s � z-getPhases..InTablePhase.startTagInputc S sD | j �d� | jjd kr@| j�|� | jjd | j_| jj�� d S )Nzunexpected-form-in-tablerv )rL r r> r� r� r~ r r� r) r) r* r^ � s
z,getPhases..InTablePhase.startTagFormc S s< | j �dd|d i� d| j_| j jd �|� d| j_d S )Nz)unexpected-start-tag-implies-table-voodoor@ Tr� F)rL r r> r� rI r� r� r) r) r* r� s z-getPhases..InTablePhase.startTagOtherc S s� | j jddd�r�| j �� | j jd jdkrJ| j�dd| j jd jd�� | j jd jdkrj| j j�� qJ| j j�� | j�� n| jj s�t
�| j�� d S )Nr� r� rv zend-tag-too-early-namedr� ) r> r� r� r~ r@ rL r r rb rS r� r� r) r) r* r�
s
�� z+getPhases..InTablePhase.endTagTablec S s | j �dd|d i� d S r r� r� r) r) r* r� s z,getPhases..InTablePhase.endTagIgnorec S s< | j �dd|d i� d| j_| j jd �|� d| j_d S )Nz'unexpected-end-tag-implies-table-voodoor@ Tr� F)rL r r> r� rI r� r� r) r) r* r� s z+getPhases..InTablePhase.endTagOtherN)r9 r: r; rK r� r� r� r� r� r� r� r� r� r� rg r� rj r^ r� r� r� r� r) r� r) r* � InTablePhase� s&
r� c sP e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dS )z#getPhases..InTableTextPhasec s � � | ||� d | _g | _d S r. )rK r� �characterTokensr� r� r) r* rK ( s z,getPhases..InTableTextPhase.__init__c S sd d� dd� | jD ��}tdd� |D ��rJtd |d�}| jjd �|� n|rZ| j�|� g | _d S )Nr� c S s g | ] }|d �qS )rw r) �r? r� r) r) r* rD . s zGgetPhases..InTableTextPhase.flushCharacters..c S s g | ] }|t k�qS r) r� r� r) r) r* rD / s rq r� r� )�joinr� r� r rL rI r� r> )rC rw r� r) r) r* �flushCharacters- s z3getPhases..InTableTextPhase.flushCharactersc S s | � � | j| j_|S r. �r� r� rL r` r� r) r) r* r� 6 s
z2getPhases..InTableTextPhase.processCommentc S s | � � | j| j_dS r� r� rB r) r) r* r� ; s
z.getPhases..InTableTextPhase.processEOFc S s |d dkrd S | j �|� d S �Nrw r� �r� r� r� r) r) r* r� @ s z5getPhases..InTableTextPhase.processCharactersc S s | j �|� d S r. r� r� r) r) r* r� E s z:getPhases..InTableTextPhase.processSpaceCharactersc S s | � � | j| j_|S r. r� r� r) r) r* r� J s
z3getPhases..InTableTextPhase.processStartTagc S s | � � | j| j_|S r. r� r� r) r) r* r� O s
z1getPhases..InTableTextPhase.processEndTagN) r9 r: r; rK r� r� r� r� r� r� r� r) r� r) r* �InTableTextPhase' s r� c s` e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dS )z!getPhases..InCaptionPhasec sf � � | ||� t�d| jfd| jfg�| _| j| j_t�d| jfd| j fd| j
fg�| _ | j | j _d S )Nr� � r� rW r� r� r� r� r� r� r� r� r� )
r� rW r� r� r� r� r� r� r� r� )
rK r
r� r� �startTagTableElementr� r� r� �
endTagCaptionr� r� r� r� r� r� r) r* rK V s ��
��z*getPhases..InCaptionPhase.__init__c S s | j jddd� S )Nr� r� r� �r> r� rB r) r) r* �ignoreEndTagCaptionh s z5getPhases..InCaptionPhase.ignoreEndTagCaptionc S s | j jd �� d S r� �rL rI r� rB r) r) r* r� k s z,getPhases..InCaptionPhase.processEOFc S s | j jd �|�S r� �rL rI r� r� r) r) r* r� n s z3getPhases..InCaptionPhase.processCharactersc S s0 | j �� | �� }| j j�td�� |s,|S d S �Nr� �rL r r� r` r� r� �rC r� � ignoreEndTagr) r) r* r� q s
z6getPhases..InCaptionPhase.startTagTableElementc S s | j jd �|�S r� r� r� r) r) r* r� y s z/getPhases..InCaptionPhase.startTagOtherc S s� | � � s�| j�� | jjd jdkrB| j�dd| jjd jd�� | jjd jdkrb| jj�� qB| jj�� | j�� | jj d | j_
n| jj s�t �| j�� d S )Nrv r� r� r� r� )
r� r> r� r~ r@ rL r r r� rI r` rS r� r� r) r) r* r� | s
��
z/getPhases..InCaptionPhase.endTagCaptionc S s0 | j �� | �� }| j j�td�� |s,|S d S r� r� r� r) r) r* r� � s
z-getPhases..InCaptionPhase.endTagTablec S s | j �dd|d i� d S r r� r� r) r) r* r� � s z.getPhases..InCaptionPhase.endTagIgnorec S s | j jd �|�S r� �rL rI r� r� r) r) r* r� � s z-getPhases..InCaptionPhase.endTagOtherN)
r9 r: r; rK r� r� r� r� r� r� r� r� r� r) r� r) r* �InCaptionPhaseT s r� c sX e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dS )z%getPhases..InColumnGroupPhasec s^ � � | ||� t�d| jfd| jfg�| _| j| j_t�d| jfd| j fg�| _
| j | j
_d S )Nr� rW r� ) rK r
r� r� r� r� r� r� �endTagColgroup� endTagColr� r� r� r� r) r* rK � s �
�z.getPhases..InColumnGroupPhase.__init__c S s | j jd jdkS )Nrv r� )r> r~ r@ rB r) r) r* �ignoreEndTagColgroup� s z:getPhases..InColumnGroupPhase.ignoreEndTagColgroupc S sD | j jd jdkr"| jjs t�d S | �� }| �td�� |s@dS d S )Nrv r� r� T) r> r~ r@ rL rS r� r� r� r� )rC r� r) r) r* r� � s z0getPhases..InColumnGroupPhase.processEOFc S s" | � � }| �td�� |s |S d S �Nr� �r� r� r� r� r) r) r* r� � s z7getPhases..InColumnGroupPhase.processCharactersc S s$ | j �|� | j j�� d|d< d S r r r� r) r) r* r� � s z1getPhases..InColumnGroupPhase.startTagColc S s" | � � }| �td�� |s |S d S r� r� r� r) r) r* r� � s z3getPhases..InColumnGroupPhase.startTagOtherc S s@ | � � r | jjst�| j�� n | jj�� | jjd | j_ d S �Nr� )
r� rL rS r� r r> r~ r rI r` r� r) r) r* r� � s
z4getPhases..InColumnGroupPhase.endTagColgroupc S s | j �dddi� d S )Nz
no-end-tagr@ rW r� r� r) r) r* r� � s z/getPhases..InColumnGroupPhase.endTagColc S s" | � � }| �td�� |s |S d S r� r� r� r) r) r* r� � s z1getPhases..InColumnGroupPhase.endTagOtherN) r9 r: r; rK r� r� r� r� r� r� r� r� r) r� r) r* �InColumnGroupPhase� s
r� c sx e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� ZdS ) z#getPhases..InTableBodyPhasec sv � � | ||� t�d| jfd| jfd| jfd| jfg�| _| j| j_ t�d| j
fd| j fd| j fg�| _
| j| j
_ d S )Nr� r� �r� r� )r� rW r� r� r� r� r� r� )r� r� rW r� r� r� r� r� )rK r
r� r� �
startTagTr�startTagTableCell�startTagTableOtherr� r� r� �endTagTableRowGroupr� r� r� r� r� r� r) r* rK � s$ ��
��z,getPhases..InTableBodyPhase.__init__c S sB | j jd jdkr | j j�� q | j jd jdkr>| jjs>t�d S )Nrv )r� r� r� r� r� )r> r~ r@ r rL rS r� rB r) r) r* � clearStackToTableBodyContext� s z@getPhases..InTableBodyPhase.clearStackToTableBodyContextc S s | j jd �� d S r� r� rB r) r) r* r� � s z.getPhases..InTableBodyPhase.processEOFc S s | j jd �|�S r� r r� r) r) r* r� s z:getPhases..InTableBodyPhase.processSpaceCharactersc S s | j jd �|�S r� r� r� r) r) r* r� s z5getPhases..InTableBodyPhase.processCharactersc S s( | � � | j�|� | jjd | j_d S )Nr� )r� r> r� rL rI r` r� r) r) r* r� s z.getPhases..InTableBodyPhase.startTagTrc S s* | j �dd|d i� | �tdd�� |S )Nz unexpected-cell-in-table-bodyr@ r� rr )rL r r� r� r� r) r) r* r� s
�z5getPhases..InTableBodyPhase.startTagTableCellc S sn | j jddd�s0| j jddd�s0| j jddd�rT| �� | �t| j jd j�� |S | jjs`t �| j�
� d S �Nr� r� r� r� r� rv � r> r� r� r� r� r~ r@ rL rS r� r r� r) r) r* r� s ��� z6getPhases..InTableBodyPhase.startTagTableOtherc S s | j jd �|�S r� r� r� r) r) r* r� s z1getPhases..InTableBodyPhase.startTagOtherc S sT | j j|d dd�r:| �� | j j�� | jjd | j_n| j�dd|d i� d S )Nr@ r� r� r� � unexpected-end-tag-in-table-body) r> r� r� r~ r rL rI r` r r� r) r) r* r� " s
�z7getPhases..InTableBodyPhase.endTagTableRowGroupc S sn | j jddd�s0| j jddd�s0| j jddd�rT| �� | �t| j jd j�� |S | jjs`t �| j�
� d S r� r� r� r) r) r* r� + s ��� z/getPhases..InTableBodyPhase.endTagTablec S s | j �dd|d i� d S )Nr� r@ r� r� r) r) r* r� 8 s
�z0getPhases..InTableBodyPhase.endTagIgnorec S s | j jd �|�S r� r� r� r) r) r* r� < s z/getPhases..InTableBodyPhase.endTagOtherN)r9 r: r; rK r� r� r� r� r� r� r� r� r� r� r� r� r) r� r) r* �InTableBodyPhase� s
r� c s� e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� Zdd � Zd S ) z getPhases..InRowPhasec sv � � | ||� t�d| jfd| jfd| jfg�| _| j| j_t�d| j fd| j
fd| j fd| j fg�| _
| j| j
_d S )Nr� r� )r� rW r� r� r� r� r� r� r� r� )r� r� rW r� r� r� r� )rK r
r� r� r� r� r� r� r� �endTagTrr� r� r� r� r� r� r� r) r* rK A s$ ��
��z&getPhases..InRowPhase.__init__c S s@ | j jd jdkr<| j�dd| j jd ji� | j j�� q d S )Nrv )r� r� z'unexpected-implied-end-tag-in-table-rowr@ )r> r~ r@ rL r r rB r) r) r* �clearStackToTableRowContextU s
�z9getPhases..InRowPhase.clearStackToTableRowContextc S s | j jddd� S )Nr� r� r� r� rB r) r) r* �ignoreEndTagTr[ s z,getPhases..InRowPhase.ignoreEndTagTrc S s | j jd �� d S r� r� rB r) r) r* r� _ s z(getPhases..InRowPhase.processEOFc S s | j jd �|�S r� r r� r) r) r* r� b s z4getPhases..InRowPhase.processSpaceCharactersc S s | j jd �|�S r� r� r� r) r) r* r� e s z/getPhases..InRowPhase.processCharactersc S s6 | � � | j�|� | jjd | j_| jj�t� d S )Nr� ) r� r> r� rL rI r` r� r� r r� r) r) r* r� h s z/getPhases..InRowPhase.startTagTableCellc S s" | � � }| �td�� |s |S d S �Nr� �r� r� r� r� r) r) r* r� n s z0getPhases..InRowPhase.startTagTableOtherc S s | j jd �|�S r� r� r� r) r) r* r� u s z+getPhases..InRowPhase.startTagOtherc S sH | � � s.| �� | jj�� | jjd | j_n| jjs:t �| j�
� d S r� ) r� r� r> r~ r rL rI r` rS r� r r� r) r) r* r� x s z&getPhases..InRowPhase.endTagTrc S s" | � � }| �td�� |s |S d S r� r� r� r) r) r* r� � s z)getPhases..InRowPhase.endTagTablec S s4 | j j|d dd�r&| �td�� |S | j�� d S )Nr@ r� r� r� )r> r� r� r� rL r r� r) r) r* r� � s z1getPhases..InRowPhase.endTagTableRowGroupc S s | j �dd|d i� d S )Nz unexpected-end-tag-in-table-rowr@ r� r� r) r) r* r� � s
�z*getPhases..InRowPhase.endTagIgnorec S s | j jd �|�S r� r� r� r) r) r* r� � s z)getPhases..InRowPhase.endTagOtherN)r9 r: r; rK r� r� r� r� r� r� r� r� r� r� r� r� r� r) r� r) r* �
InRowPhase? s
r� c s` e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dS )z getPhases..InCellPhasec sf � � | ||� t�d| jfd| jfg�| _| j| j_t�d| jfd| j fd| j
fg�| _ | j | j _d S )Nr� r� r� )r� r� rW r� r� r� )
rK r
r� r� r� r� r� r� �endTagTableCellr� � endTagImplyr� r� r� r� r) r* rK � s ��
�z'getPhases..InCellPhase.__init__c S sB | j jddd�r | �td�� n | j jddd�r>| �td�� d S �Nr� r� r� r� )r> r� r� r� rB r) r) r* � closeCell� s z(getPhases..InCellPhase.closeCellc S s | j jd �� d S r� r� rB r) r) r* r� � s z)getPhases..InCellPhase.processEOFc S s | j jd �|�S r� r� r� r) r) r* r� � s z0getPhases..InCellPhase.processCharactersc S sF | j jddd�s | j jddd�r,| �� |S | jjs8t�| j�� d S r� )r> r� r� rL rS r� r r� r) r) r* r� � s � z1getPhases..InCellPhase.startTagTableOtherc S s | j jd �|�S r� r� r� r) r) r* r� � s z,getPhases..InCellPhase.startTagOtherc S s� | j j|d dd�r�| j �|d � | j jd j|d krp| j�dd|d i� | j j�� }|j|d krPq|qPn | j j�� | j �� | jj d | j_
n| j�dd|d i� d S )Nr@ r� r� rv zunexpected-cell-end-tagr� r ) r> r� r� r~ r@ rL r r r� rI r` r
r) r) r* r� � s
�
z.getPhases..InCellPhase.endTagTableCellc S s | j �dd|d i� d S r r� r� r) r) r* r� � s z+getPhases..InCellPhase.endTagIgnorec S s. | j j|d dd�r | �� |S | j�� d S )Nr@ r� r� )r> r� r� rL r r� r) r) r* r� � s z*getPhases..InCellPhase.endTagImplyc S s | j jd �|�S r� r� r� r) r) r* r� � s z*getPhases..InCellPhase.endTagOtherN)
r9 r: r; rK r� r� r� r� r� r� r� r� r� r) r� r) r* � InCellPhase� s
r� c sx e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� ZdS ) z getPhases.