관리-도구
편집 파일: html5parser.cpython-36.pyc
3 �Re�� � @ s d dl mZmZmZ d dlmZmZ d dlZddlm Z ddlm Z ddlm Z ddl m Z dd lmZ dd lmZmZmZmZmZmZmZmZmZmZmZmZ m Z m Z m Z m Z d d d�Z!d!dd�Z"dd� Z#G dd� de$�Z%ej&dd� �Z'dd� Z(d"d d �Z)G d d � d e*�Z+dS )#� )�absolute_import�division�unicode_literals)�with_metaclass�viewkeysN� )� _inputstream)� _tokenizer)� treebuilders)�Marker)�_utils)�spaceCharacters�asciiUpper2Lower�specialElements�headingElements� cdataElements�rcdataElements� tokenTypes� tagTokenTypes� namespaces� htmlIntegrationPointElements�"mathmlTextIntegrationPointElements�adjustForeignAttributes�adjustMathMLAttributes�adjustSVGAttributes�E�_ReparseException�etreeTc K s$ t j|�}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('<html><body><p>This is a doc</p></body></html>') <Element u'{http://www.w3.org/1999/xhtml}html' at 0x7feac4909db0> )�namespaceHTMLElements)r �getTreeBuilder� HTMLParser�parse)�doc� treebuilderr �kwargs�tb�p� r' ��/builddir/build/BUILDROOT/alt-python36-pip-20.2.4-5.el8.x86_64/opt/alt/python36/lib/python3.6/site-packages/pip/_vendor/html5lib/html5parser.pyr! s r! �divc K s, t j|�}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('<b>this is a fragment</b>') <Element u'DOCUMENT_FRAGMENT' at 0x7feac484b090> )r � container)r r r � parseFragment)r" r* r# r r$ r% r& r' r' r( r+ 1 s r+ c s G � fdd�dt �}|S )Nc s e Zd Z� fdd�ZdS )z-method_decorator_metaclass.<locals>.Decoratedc sB x0|j � D ]$\}}t|tj�r&� |�}|||< q W tj| |||�S )N)�items� isinstance�types� FunctionType�type�__new__)�meta� classname�bases� classDict� attributeName� attribute)�functionr' r( r1 L s z5method_decorator_metaclass.<locals>.Decorated.__new__N)�__name__� __module__� __qualname__r1 r' )r8 r' r( � DecoratedK s r< )r0 )r8 r< r' )r8 r( �method_decorator_metaclassJ s r= c @ s� e Zd ZdZd'dd�Zd(dd �Zd d � Zed d � �Zdd� Z dd� Z dd� Z dd� Z dd� Z d)dd�Zdd � Zd d � Zd d � Zd!d"� Zd#d$� Zd%d&� ZdS )*r z]HTML parser Generates a tree structure from a stream of (possibly malformed) HTML. NFTc sH |� _ |dkrtjd�}||�� _g � _� fdd�t|�j� 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 i | ]\}}|� � j �|�qS r' )�tree)�.0�name�cls)�selfr' r( � <dictcomp>y s z'HTMLParser.__init__.<locals>.<dictcomp>)�strictr r r> �errors� getPhasesr, �phases)rB r> rD r �debugr' )rB r( �__init__] s zHTMLParser.__init__r) c K sh || _ || _|| _tj|fd| i|��| _| j� y | j� W n$ tk rb | j� | j� Y nX d S )N�parser) � innerHTMLModer* � scriptingr � HTMLTokenizer� tokenizer�reset�mainLoopr )rB �stream� innerHTMLr* rL r$ r' r' r( �_parse| s zHTMLParser._parsec C s� | j j� d| _g | _g | _d| _| jr�| jj� | _ | j t krL| j j | j _ n0| j tkrd| j j| j _ n| j dkr|| j j| j _ n | jd | _| jj� | j� nd| _ | jd | _d | _d | _d| _d S )NFz no quirks� plaintext� beforeHtml�initialT)r> rO � firstStartTagrE �log� compatModerK r* �lowerrR r rN � rcdataState�stater � rawtextState�plaintextStaterG �phase�insertHtmlElement�resetInsertionMode� lastPhase�beforeRCDataPhase� framesetOK)rB r' r' r( rO � 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 rN Nr )�hasattrrN rQ � charEncodingr@ )rB r' r' r( �documentEncoding� s zHTMLParser.documentEncodingc C sJ |j dkr6|jtd kr6d|jko4|jd jt�dkS |j|j ftkS d S )Nzannotation-xml�mathml�encoding� text/html�application/xhtml+xml)rj rk )r@ � namespacer � attributes� translater r )rB �elementr' r' r( �isHTMLIntegrationPoint� s z!HTMLParser.isHTMLIntegrationPointc C s |j |jftkS )N)rl r@ r )rB ro r' r' r( � isMathMLTextIntegrationPoint� s z'HTMLParser.isMathMLTextIntegrationPointc C sx t d }t d }t d }t d }t d }t d }t d }�x�| jD �]�}d } |} �x�| d k �r�| } | jjrv| jjd nd } | r�| jnd } | r�| jnd } | d }||kr�| j| d | jd i �� d } qTt| jj�d k�sj| | jj k�sj| j | ��r ||k�r|d t ddg�k�sj|||fk�sj| t d k�rN| dk�rN||k�rN|d dk�sj| j | ��rr||||fk�rr| j}n | jd }||k�r�|j| �} qT||k�r�|j| �} qT||k�r�|j| �} qT||k�r�|j| �} qT||k�r�|j| �} qT||krT|j| �} qTW ||krB| d rB| d rB| jdd | d i� qBW d}g }x8|�rr|j| j� | jj� }|�r<| j|k�s<t��q<W d S )N� Characters�SpaceCharacters�StartTag�EndTag�Comment�Doctype� ParseErrorr r0 �data�datavarsr r@ �mglyph� malignmarkrh zannotation-xml�svg�inForeignContent� selfClosing�selfClosingAcknowledgedz&non-void-element-with-trailing-solidusT���)r rN r> � openElementsrl r@ � parseError�get�len�defaultNamespacerq � frozensetr rp r_ rG �processCharacters�processSpaceCharacters�processStartTag� processEndTag�processComment�processDoctype�append� processEOF�AssertionError)rB �CharactersToken�SpaceCharactersToken� StartTagToken� EndTagToken� CommentToken� DoctypeToken�ParseErrorToken�token� prev_token� new_token� currentNode�currentNodeNamespace�currentNodeNamer0 r_ � reprocessrG r' r' r( rP � sp zHTMLParser.mainLoopc O s | j |ddf|�|� | jj� 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('<html><body><p>This is a doc</p></body></html>') <Element u'{http://www.w3.org/1999/xhtml}html' at 0x7feac4909db0> FN)rS r> � getDocument)rB rQ �argsr$ r' r' r( r! s zHTMLParser.parsec O s | j |df|�|� | jj� 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('<b>this is a fragment</b>') <Element u'DOCUMENT_FRAGMENT' at 0x7feac484b090> T)rS r> � getFragment)rB rQ r� r$ r' r' r( r+ s zHTMLParser.parseFragment�XXX-undefined-errorc C s@ |d kr i }| j j| jjj� ||f� | jr<tt| | ��d S )N)rE r� rN rQ �positionrD rx r )rB � errorcoderz r' r' r( r� ; s zHTMLParser.parseErrorc C s t |t� d S )N)�adjust_attributesr )rB r� r' r' r( r C s z!HTMLParser.adjustMathMLAttributesc C s t |t� d S )N)r� r )rB r� r' r' r( r F s z HTMLParser.adjustSVGAttributesc C s t |t� d S )N)r� �adjustForeignAttributesMap)rB r� r' r' r( r I s z"HTMLParser.adjustForeignAttributesc C s | j j� d S )N)rJ r_ )rB r� r' r' r( �reparseTokenNormalL s z HTMLParser.reparseTokenNormalc C s� d}ddddddddddd d d d d �}x�| j jd d d� D ]�}|j}d }|| j jd krl| jsbt�d}| j}|dkr~| js~t�| r�|j| j jkr�q:||kr�| j|| }P q:|r:| jd }P q:W || _d S )NF�inSelect�inCell�inRow� inTableBody� inCaption� inColumnGroup�inTable�inBody� inFrameset� beforeHead)�select�td�th�tr�tbody�thead�tfoot�caption�colgroup�table�head�body�frameset�htmlr r Tr� r� r� r� r� )r� r� r� r� ) r> r� r@ rR r� rl r� rG r_ )rB �last�newModes�node�nodeName� new_phaser' r' r( ra P sB z HTMLParser.resetInsertionModec C sR |dks t �| jj|� |dkr.| jj| j_n | jj| j_| j| _| j d | _d S )N�RAWTEXT�RCDATA�text)r� r� ) r� r> � insertElementrN r] r\ r[ r_ � originalPhaserG )rB r� � contentTyper' r' r( �parseRCDataRawtext| s z HTMLParser.parseRCDataRawtext)NFTF)Fr) F)r� N)r9 r: r; �__doc__rI rS rO �propertyrg rp rq rP r! r+ r� r r r r� ra r� r' r' r' r( r V s" " C ,r c s dd� }dd� }G dd� dt || |����G dd� d��}G d d � d ��}G d d � d ��}G d d� d��}G dd� d��}G dd� d��}G � �fdd�d��� G dd� 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 d'd(� d(��}G �fd)d*�d*��}G d+d,� d,��}G �fd-d.�d.��}G �fd/d0�d0��}G d1d2� d2��}G d3d4� d4��}||||||� | | �| | | ||||||||||d5�S )6Nc s$ dd� t j� D ��� �fdd�}|S )z4Logger that records which phase processes each tokenc S s i | ]\}}||�qS r' r' )r? �key�valuer' r' r( rC � s z*getPhases.<locals>.log.<locals>.<dictcomp>c s� � j jd�r�t|�dkr�|d }d�|d i}|d tkrH|d |d< | jjj| jjjj | jj j j | j j � j |f� � | f|�|�S � | f|�|�S d S )N�processr r0 r@ ) r9 � startswithr� r rJ rX r� rN r\ r_ � __class__)rB r� r$ r� �info)r8 � type_namesr' r( �wrapped� s z'getPhases.<locals>.log.<locals>.wrapped)r r, )r8 r� r' )r8 r� r( rX � s zgetPhases.<locals>.logc S s | r t |�S tS d S )N)r= r0 )� use_metaclass�metaclass_funcr' r' r( � getMetaclass� s z getPhases.<locals>.getMetaclassc @ s\ e Zd ZdZdZdd� Zdd � Zd d � Zd d � Zdd� Z dd� Z dd� Z dd� Z dd� Z dS )zgetPhases.<locals>.PhasezNBase class for helper object that implements each phase of processing rJ r> �__startTagCache� __endTagCachec S s || _ || _i | _i | _d S )N)rJ r> �_Phase__startTagCache�_Phase__endTagCache)rB rJ r> r' r' r( rI � s z!getPhases.<locals>.Phase.__init__c S s t �d S )N)�NotImplementedError)rB r' r' r( r� � s z#getPhases.<locals>.Phase.processEOFc S s | j j|| j jd � d S )Nr r� )r> � insertCommentr� )rB r� r' r' r( r� � s z'getPhases.<locals>.Phase.processCommentc S s | j jd� d S )Nzunexpected-doctype)rJ r� )rB r� r' r' r( r� � s z'getPhases.<locals>.Phase.processDoctypec S s | j j|d � d S )Nry )r> � insertText)rB r� r' r' r( r� � s z*getPhases.<locals>.Phase.processCharactersc S s | j j|d � d S )Nry )r> r� )rB r� r' r' r( r� � s z/getPhases.<locals>.Phase.processSpaceCharactersc S sn |d }|| j kr | j | }nH| j| }| j |< x2t| j �t| j�d krd| j jtt| j ��� q4W ||�S )Nr@ g�������?)r� �startTagHandlerr� �pop�next�iter)rB r� r@ �funcr' r' r( r� � s z(getPhases.<locals>.Phase.processStartTagc S sl | j j r"|d dkr"| j jd� x<|d j� D ],\}}|| jjd jkr0|| jjd j|< q0W d| j _d S )Nr@ r� z non-html-rootry r F)rJ rW r� r, r> r� rm )rB r� �attrr� r' r' r( � startTagHtml� s z%getPhases.<locals>.Phase.startTagHtmlc S sn |d }|| j kr | j | }nH| j| }| j |< x2t| j �t| j�d krd| j jtt| j ��� q4W ||�S )Nr@ g�������?)r� � endTagHandlerr� r� r� r� )rB r� r@ r� r' r' r( r� � s z&getPhases.<locals>.Phase.processEndTagN)rJ r> r� r� )r9 r: r; r� � __slots__rI r� r� r� r� r� r� r� r� r' r' r' r( �Phase� s r� c @ sR e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dS )z getPhases.<locals>.InitialPhasec S s d S )Nr' )rB r� r' r' r( r� � s z6getPhases.<locals>.InitialPhase.processSpaceCharactersc S s | j j|| j j� d S )N)r> r� �document)rB r� r' r' r( r� � s z.getPhases.<locals>.InitialPhase.processCommentc 8 S s |d }|d }|d }|d }|dks@|d k s@|d k rL|dkrL| j jd� |d krXd}| jj|� |dkrv|jt�}| � s�|d dk� s�|jdJ�� s�|dKk� s�|jdL�� r�|d k� s�|� r�|j� dDk� r�dE| j _n*|jdM�� s�|jdN��r |d k �r dH| j _| j j dI | j _ d S )ONr@ �publicId�systemId�correctr� zabout:legacy-compatzunknown-doctype� �*+//silmaril//dtd html pro v0r11 19970101//�4-//advasoft ltd//dtd html 3.0 aswedit + extensions//�*-//as//dtd html 3.0 aswedit + extensions//� -//ietf//dtd html 2.0 level 1//� -//ietf//dtd html 2.0 level 2//�&-//ietf//dtd html 2.0 strict level 1//�&-//ietf//dtd html 2.0 strict level 2//� -//ietf//dtd html 2.0 strict//�-//ietf//dtd html 2.0//�-//ietf//dtd html 2.1e//�-//ietf//dtd html 3.0//� -//ietf//dtd html 3.2 final//�-//ietf//dtd html 3.2//�-//ietf//dtd html 3//�-//ietf//dtd html level 0//�-//ietf//dtd html level 1//�-//ietf//dtd html level 2//�-//ietf//dtd html level 3//�"-//ietf//dtd html strict level 0//�"-//ietf//dtd html strict level 1//�"-//ietf//dtd html strict level 2//�"-//ietf//dtd html strict level 3//�-//ietf//dtd html strict//�-//ietf//dtd html//�(-//metrius//dtd metrius presentational//�5-//microsoft//dtd internet explorer 2.0 html strict//�.-//microsoft//dtd internet explorer 2.0 html//�0-//microsoft//dtd internet explorer 2.0 tables//�5-//microsoft//dtd internet explorer 3.0 html strict//�.-//microsoft//dtd internet explorer 3.0 html//�0-//microsoft//dtd internet explorer 3.0 tables//�#-//netscape comm. corp.//dtd html//�*-//netscape comm. corp.//dtd strict html//�*-//o'reilly and associates//dtd html 2.0//�3-//o'reilly and associates//dtd html extended 1.0//�;-//o'reilly and associates//dtd html extended relaxed 1.0//�N-//softquad software//dtd hotmetal pro 6.0::19990601::extensions to html 4.0//�E-//softquad//dtd hotmetal pro 4.0::19971010::extensions to html 4.0//�$-//spyglass//dtd html 2.0 extended//�+-//sq//dtd html 2.0 hotmetal + extensions//�--//sun microsystems corp.//dtd hotjava html//�4-//sun microsystems corp.//dtd hotjava strict html//� -//w3c//dtd html 3 1995-03-24//� -//w3c//dtd html 3.2 draft//� -//w3c//dtd html 3.2 final//�-//w3c//dtd html 3.2//� -//w3c//dtd html 3.2s draft//� -//w3c//dtd html 4.0 frameset//�#-//w3c//dtd html 4.0 transitional//�(-//w3c//dtd html experimental 19960712//�&-//w3c//dtd html experimental 970421//�-//w3c//dtd w3 html//�-//w3o//dtd w3 html 3.0//�#-//webtechs//dtd mozilla html 2.0//� -//webtechs//dtd mozilla html//�$-//w3o//dtd w3 html strict 3.0//en//�"-/w3c/dtd html 4.0 transitional/en� -//w3c//dtd html 4.01 frameset//�$-//w3c//dtd html 4.01 transitional//z:http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd�quirks� -//w3c//dtd xhtml 1.0 frameset//�$-//w3c//dtd xhtml 1.0 transitional//zlimited quirksrU )7r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r! r" r# r$ r% )r& r' r� )r( r) )r+ r, )r( r) ) rJ r� r> � insertDoctypern r r� rZ rY rG r_ )rB r� r@ r� r� r� r' r' r( r� � s� z.getPhases.<locals>.InitialPhase.processDoctypec S s d| j _| j jd | j _d S )Nr* rU )rJ rY rG r_ )rB r' r' r( � anythingElseZ s z,getPhases.<locals>.InitialPhase.anythingElsec S s | j jd� | j� |S )Nz expected-doctype-but-got-chars)rJ r� r. )rB r� r' r' r( r� ^ s z1getPhases.<locals>.InitialPhase.processCharactersc S s" | j jdd|d i� | j� |S )Nz"expected-doctype-but-got-start-tagr@ )rJ r� r. )rB r� r' r' r( r� c s z/getPhases.<locals>.InitialPhase.processStartTagc S s" | j jdd|d i� | j� |S )Nz expected-doctype-but-got-end-tagr@ )rJ r� r. )rB r� r' r' r( r� i s z-getPhases.<locals>.InitialPhase.processEndTagc S s | j jd� | j� dS )Nz expected-doctype-but-got-eofT)rJ r� r. )rB r' r' r( r� o s z*getPhases.<locals>.InitialPhase.processEOFN) r9 r: r; �tupler� r� r� r� r. r� r� r� r� r' r' r' r( � InitialPhase� s _r0 c @ sJ e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dS )z"getPhases.<locals>.BeforeHtmlPhasec S s& | j jtdd�� | jjd | j_d S )Nr� rt r� )r> � insertRoot�impliedTagTokenrJ rG r_ )rB r' r' r( r` x s z4getPhases.<locals>.BeforeHtmlPhase.insertHtmlElementc S s | j � dS )NT)r` )rB r' r' r( r� } s z-getPhases.<locals>.BeforeHtmlPhase.processEOFc S s | j j|| j j� d S )N)r> r� r� )rB r� r' r' r( r� � s z1getPhases.<locals>.BeforeHtmlPhase.processCommentc S s d S )Nr' )rB r� r' r' r( r� � s z9getPhases.<locals>.BeforeHtmlPhase.processSpaceCharactersc S s | j � |S )N)r` )rB r� r' r' r( r� � s z4getPhases.<locals>.BeforeHtmlPhase.processCharactersc S s |d dkrd| j _| j� |S )Nr@ r� T)rJ rW r` )rB r� r' r' r( r� � s z2getPhases.<locals>.BeforeHtmlPhase.processStartTagc S s4 |d dkr$| j jdd|d i� n | j� |S d S )Nr@ r� r� r� �brz unexpected-end-tag-before-html)r� r� r� r3 )rJ r� r` )rB r� r' r' r( r� � s z0getPhases.<locals>.BeforeHtmlPhase.processEndTagN) r9 r: r; r/ r� r` r� r� r� r� r� r� r' r' r' r( �BeforeHtmlPhaset s r4 c @ s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z e jdefde fg�Ze e_e jde fg�Ze e_dS )z"getPhases.<locals>.BeforeHeadPhasec S s | j tdd�� dS )Nr� rt T)� startTagHeadr2 )rB r' r' r( r� � s z-getPhases.<locals>.BeforeHeadPhase.processEOFc S s d S )Nr' )rB r� r' r' r( r� � s z9getPhases.<locals>.BeforeHeadPhase.processSpaceCharactersc S s | j tdd�� |S )Nr� rt )r5 r2 )rB r� r' r' r( r� � s z4getPhases.<locals>.BeforeHeadPhase.processCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z/getPhases.<locals>.BeforeHeadPhase.startTagHtmlc S s0 | j j|� | j jd | j _| jjd | j_d S )Nr �inHeadr� )r> r� r� � headPointerrJ rG r_ )rB r� r' r' r( r5 � s z/getPhases.<locals>.BeforeHeadPhase.startTagHeadc S s | j tdd�� |S )Nr� rt )r5 r2 )rB r� r' r' r( � startTagOther� s z0getPhases.<locals>.BeforeHeadPhase.startTagOtherc S s | j tdd�� |S )Nr� rt )r5 r2 )rB r� r' r' r( �endTagImplyHead� s z2getPhases.<locals>.BeforeHeadPhase.endTagImplyHeadc S s | j jdd|d i� d S )Nzend-tag-after-implied-rootr@ )rJ r� )rB r� r' r' r( � endTagOther� s z.getPhases.<locals>.BeforeHeadPhase.endTagOtherr� r� r� r3 N)r� r� r� r3 )r9 r: r; r/ r� r� r� r� r� r5 r8 r9 r: r �MethodDispatcherr� �defaultr� r' r' r' r( �BeforeHeadPhase� s r= c @ s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z dd� Zdd� Zdd� Zdd� Zdd � Zd d � Zejd efd e fd/e fd#e fd$efd0e fd*e fd+efg�Zee_ejd+efd1efg�Zee_d.S )2z getPhases.<locals>.InHeadPhasec S s | j � dS )NT)r. )rB r' r' r( r� � s z)getPhases.<locals>.InHeadPhase.processEOFc S s | j � |S )N)r. )rB r� r' r' r( r� � s z0getPhases.<locals>.InHeadPhase.processCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z+getPhases.<locals>.InHeadPhase.startTagHtmlc S s | j jd� d S )Nz!two-heads-are-not-better-than-one)rJ r� )rB r� r' r' r( r5 � s z+getPhases.<locals>.InHeadPhase.startTagHeadc S s$ | j j|� | j jj� d|d< d S )NTr� )r> r� r� r� )rB r� r' r' r( �startTagBaseLinkCommand� s z6getPhases.<locals>.InHeadPhase.startTagBaseLinkCommandc S s� | j j|� | j jj� d|d< |d }| jjjjd dkr�d|krZ| jjjj|d � nVd|kr�d|kr�|d j � d kr�t j |d j d ��}t j |�}|j� }| jjjj|� d S ) NTr� ry r � tentative�charset�contentz http-equivz content-typezutf-8)r> r� r� r� rJ rN rQ rf �changeEncodingrZ r � EncodingBytes�encode�ContentAttrParserr! )rB r� rm ry rJ �codecr' r' r( � startTagMeta� s z+getPhases.<locals>.InHeadPhase.startTagMetac S s | j j|d� d S )Nr� )rJ r� )rB r� r' r' r( � startTagTitle� s z,getPhases.<locals>.InHeadPhase.startTagTitlec S s | j j|d� d S )Nr� )rJ r� )rB r� r' r' r( �startTagNoFramesStyle� s z4getPhases.<locals>.InHeadPhase.startTagNoFramesStylec S s8 | j jr| j j|d� n | jj|� | j jd | j _d S )Nr� �inHeadNoscript)rJ rL r� r> r� rG r_ )rB r� r' r' r( �startTagNoscript� s z/getPhases.<locals>.InHeadPhase.startTagNoscriptc S s< | j j|� | jjj| jj_| jj| j_| jjd | j_d S )Nr� ) r> r� rJ rN �scriptDataStater\ r_ r� rG )rB r� r' r' r( �startTagScript s z-getPhases.<locals>.InHeadPhase.startTagScriptc S s | j � |S )N)r. )rB r� r' r' r( r8 s z,getPhases.<locals>.InHeadPhase.startTagOtherc S s: | j jjj� }|jdks&td|j ��| j jd | j _d S )Nr� zExpected head got %s� afterHead)rJ r> r� r� r@ r� rG r_ )rB r� r� r' r' r( � endTagHead s z)getPhases.<locals>.InHeadPhase.endTagHeadc S s | j � |S )N)r. )rB r� r' r' r( �endTagHtmlBodyBr s z/getPhases.<locals>.InHeadPhase.endTagHtmlBodyBrc S s | j jdd|d i� d S )Nzunexpected-end-tagr@ )rJ r� )rB r� r' r' r( r: s z*getPhases.<locals>.InHeadPhase.endTagOtherc S s | j td�� d S )Nr� )rO r2 )rB r' r' r( r. s z+getPhases.<locals>.InHeadPhase.anythingElser� �title�noframes�style�noscript�script�base�basefont�bgsound�command�linkr2 r� r3 r� N)rR rS )rV rW rX rY rZ )r3 r� r� )r9 r: r; r/ r� r� r� r� r5 r> rG rH rI rK rM r8 rO rP r: r. r r; r� r< r� r' r' r' r( � InHeadPhase� s> r[ c @ s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z dd� Zdd� Zdd� Zejde fd$e fd%e fg�Ze e_ejd!e fd"efg�Zee_d#S )&z&getPhases.<locals>.InHeadNoscriptPhasec S s | j jd� | j� dS )Nzeof-in-head-noscriptT)rJ r� r. )rB r' r' r( r� / s z1getPhases.<locals>.InHeadNoscriptPhase.processEOFc S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( r� 4 s z5getPhases.<locals>.InHeadNoscriptPhase.processCommentc S s | j jd� | j� |S )Nzchar-in-head-noscript)rJ r� r. )rB r� r' r' r( r� 7 s z8getPhases.<locals>.InHeadNoscriptPhase.processCharactersc S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( r� < s z=getPhases.<locals>.InHeadNoscriptPhase.processSpaceCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� ? s z3getPhases.<locals>.InHeadNoscriptPhase.startTagHtmlc S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( r> B s z>getPhases.<locals>.InHeadNoscriptPhase.startTagBaseLinkCommandc S s | j jdd|d i� d S )Nzunexpected-start-tagr@ )rJ r� )rB r� r' r' r( �startTagHeadNoscriptE s z;getPhases.<locals>.InHeadNoscriptPhase.startTagHeadNoscriptc S s" | j jdd|d i� | j� |S )Nz unexpected-inhead-noscript-tagr@ )rJ r� r. )rB r� r' r' r( r8 H s z4getPhases.<locals>.InHeadNoscriptPhase.startTagOtherc S s: | j jjj� }|jdks&td|j ��| j jd | j _d S )NrT zExpected noscript got %sr6 )rJ r> r� r� r@ r� rG r_ )rB r� r� r' r' r( �endTagNoscriptM s z5getPhases.<locals>.InHeadNoscriptPhase.endTagNoscriptc S s" | j jdd|d i� | j� |S )Nz unexpected-inhead-noscript-tagr@ )rJ r� r. )rB r� r' r' r( �endTagBrR s z/getPhases.<locals>.InHeadNoscriptPhase.endTagBrc S s | j jdd|d i� d S )Nzunexpected-end-tagr@ )rJ r� )rB r� r' r' r( r: W s z2getPhases.<locals>.InHeadNoscriptPhase.endTagOtherc S s | j td�� d S )NrT )r] r2 )rB r' r' r( r. Z s z3getPhases.<locals>.InHeadNoscriptPhase.anythingElser� rW rX rZ r2 rR rS r� rT r3 N)rW rX rZ r2 rR rS )r� rT )r9 r: r; r/ r� r� r� r� r� r� r> r\ r8 r] r^ r: r. r r; r� r< r� r' r' r' r( �InHeadNoscriptPhase, s, r_ c @ s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z dd� Zdd� Zejdefdefde fd&e fd#e fg�Ze e_ejd'e fg�Zee_d%S )(z!getPhases.<locals>.AfterHeadPhasec S s | j � dS )NT)r. )rB r' r' r( r� n s z,getPhases.<locals>.AfterHeadPhase.processEOFc S s | j � |S )N)r. )rB r� r' r' r( r� r s z3getPhases.<locals>.AfterHeadPhase.processCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� v s z.getPhases.<locals>.AfterHeadPhase.startTagHtmlc S s( d| j _| jj|� | j jd | j _d S )NFr� )rJ rd r> r� rG r_ )rB r� r' r' r( � startTagBodyy s z.getPhases.<locals>.AfterHeadPhase.startTagBodyc S s | j j|� | jjd | j_d S )Nr� )r> r� rJ rG r_ )rB r� r' r' r( �startTagFrameset~ s z2getPhases.<locals>.AfterHeadPhase.startTagFramesetc S st | j jdd|d i� | jjj| jj� | j jd j|� x4| jjd d d� D ] }|jdkrN| jjj |� P qNW d S )Nz#unexpected-start-tag-out-of-my-headr@ r6 r r� r� ) rJ r� r> r� r� r7 rG r� r@ �remove)rB r� r� r' r' r( �startTagFromHead� s z2getPhases.<locals>.AfterHeadPhase.startTagFromHeadc S s | j jdd|d i� d S )Nzunexpected-start-tagr@ )rJ r� )rB r� r' r' r( r5 � s z.getPhases.<locals>.AfterHeadPhase.startTagHeadc S s | j � |S )N)r. )rB r� r' r' r( r8 � s z/getPhases.<locals>.AfterHeadPhase.startTagOtherc S s | j � |S )N)r. )rB r� r' r' r( rP � s z2getPhases.<locals>.AfterHeadPhase.endTagHtmlBodyBrc S s | j jdd|d i� d S )Nzunexpected-end-tagr@ )rJ r� )rB r� r' r' r( r: � s z-getPhases.<locals>.AfterHeadPhase.endTagOtherc S s. | j jtdd�� | jjd | j_d| j_d S )Nr� rt r� T)r> r� r2 rJ rG r_ rd )rB r' r' r( r. � s z.getPhases.<locals>.AfterHeadPhase.anythingElser� r� r� rV rW rX rZ r2 rR rU rS rQ r� r3 N) rV rW rX rZ r2 rR rU rS rQ )r� r� r3 )r9 r: r; r/ r� r� r� r� r` ra rc r5 r8 rP r: r. r r; r� r< r� r' r' r' r( �AfterHeadPhasek s0 rd c , s� e Zd Zd�Z�� fdd�Zdd� Zdd� Zdd � 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.d/� Zd0d1� Zd2d3� Z d4d5� Z d6d7� Z d8d9� Z d:d;� Z d<d=� Z!d>d?� Z"d@dA� Z#dBdC� Z$dDdE� Z%dFdG� Z&dHdI� Z'dJdK� Z(dLdM� Z)dNdO� Z*dPdQ� Z+dRdS� Z,dTdU� Z-dVdW� Z.dXdY� Z/dZd[� Z0d\d]� Z1d^d_� Z2d`da� Z3dbdc� Z4ddde� Z5dfdg� Z6e7j8dh�j9fd�e fdre fdse fd�efe:efd�efd�efd�efd�efd�efd�efd�efd�efd�efd�efd�efd�efd�e fd�e fd�e fd�e fd�e fd�e!fd�e"fd�e#fd�e$fd�e&fd�e'fd�e%fd�e(fd�e)fd�e*fg!�Z;e+e;_<e7j8dre-fdhe.fd�e/fd�e0fd�e,fd�e1fe:e2fd�e3fd�e4fd�e5fg �Z=e6e=_<� Z>S )�z getPhases.<locals>.InBodyPhaser� c s t � | �j||� | j| _d S )N)�superrI � processSpaceCharactersNonPrer� )rB r� r$ )� InBodyPhaser� r' r( rI � s z'getPhases.<locals>.InBodyPhase.__init__c S s$ |j |j ko"|j|jko"|j|jkS )N)r@ rl rm )rB �node1�node2r' r' r( �isMatchingFormattingElement� s z:getPhases.<locals>.InBodyPhase.isMatchingFormattingElementc S s� | j j|� | j jd }g }x<| j jd d d� D ]&}|tkr@P q0| j||�r0|j|� q0W t|�dksjt�t|�dkr�| j jj |d � | j jj|� d S )Nr � r� r� r� ) r> r� r� �activeFormattingElementsr rj r� r� r� rb )rB r� ro �matchingElementsr� r' r' r( �addFormattingElement� s z3getPhases.<locals>.InBodyPhase.addFormattingElementc S s@ t d�}x2| jjd d d� D ] }|j|kr | jjd� P q W d S )N�dd�dt�lir& r� r� r� r� r� r� r� r� r z expected-closing-tag-but-got-eof) ro rp rq r& r� r� r� r� r� r� r� r� r� )r� r> r� r@ rJ r� )rB �allowed_elementsr� r' r' r( r� � s z)getPhases.<locals>.InBodyPhase.processEOFc S sh |d }| j | _|jd�rJ| jjd jdkrJ| jjd j� rJ|dd � }|rd| jj� | jj|� d S ) Nry � r �pre�listing�textarear� )rt ru rv r� ) rf r� r� r> r� r@ � hasContent�#reconstructActiveFormattingElementsr� )rB r� ry r' r' r( �!processSpaceCharactersDropNewline� s z@getPhases.<locals>.InBodyPhase.processSpaceCharactersDropNewlinec S sT |d dkrd S | j j� | j j|d � | jjrPtdd� |d D ��rPd| j_d S )Nry � c S s g | ] }|t k�qS r' )r )r? �charr' r' r( � <listcomp>� s zDgetPhases.<locals>.InBodyPhase.processCharacters.<locals>.<listcomp>F)r> rx r� rJ rd �any)rB r� r' r' r( r� � s z0getPhases.<locals>.InBodyPhase.processCharactersc S s | j j� | j j|d � d S )Nry )r> rx r� )rB r� r' r' r( rf � s z;getPhases.<locals>.InBodyPhase.processSpaceCharactersNonPrec S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( �startTagProcessInHead� s z4getPhases.<locals>.InBodyPhase.startTagProcessInHeadc S s� | j jdddi� t| jj�dks4| jjd jdkrB| j js�t�nFd| j _x<|d j � D ],\}}|| jjd j krX|| jjd j |< qXW d S )Nzunexpected-start-tagr@ r� r Fry ) rJ r� r� r> r� r@ rR r� rd r, rm )rB r� r� r� r' r' r( r` � s z+getPhases.<locals>.InBodyPhase.startTagBodyc S s� | j jdddi� t| jj�dks4| jjd jdkrB| j js�t�nt| j jsLnj| jjd j rv| jjd j j | jjd � x"| jjd jdkr�| jjj � qxW | jj |� | j j d | j _d S ) Nzunexpected-start-tagr@ r� r r� r� r� r� )rJ r� r� r> r� r@ rR r� rd �parent� removeChildr� r� rG r_ )rB r� r' r' r( ra s " z/getPhases.<locals>.InBodyPhase.startTagFramesetc S s. | j jddd�r | jtd�� | j j|� d S )Nr& �button)�variant)r> �elementInScope�endTagPr2 r� )rB r� r' r' r( �startTagCloseP s z-getPhases.<locals>.InBodyPhase.startTagClosePc S s> | j jddd�r | jtd�� | j j|� d| j_| j| _d S )Nr& r� )r� F) r> r� r� r2 r� rJ rd ry r� )rB r� r' r' r( �startTagPreListing s z1getPhases.<locals>.InBodyPhase.startTagPreListingc S sZ | j jr | jjdddi� n:| j jddd�r:| jtd�� | j j|� | j jd | j _d S ) Nzunexpected-start-tagr@ �formr& r� )r� r r� ) r> � formPointerrJ r� r� r� r2 r� r� )rB r� r' r' r( � startTagForm s z+getPhases.<locals>.InBodyPhase.startTagFormc S s� d| j _dgddgddgd�}||d }xLt| jj�D ]<}|j|kr^| j jjt|jd�� P |j t kr8|jd kr8P q8W | jj d d d �r�| j jjtd d�� | jj |� d S )NFrq rp ro )rq rp ro r@ ru �addressr) r& r� )r� )r� r) r& ) rJ rd �reversedr> r� r@ r_ r� r2 � nameTupler r� r� )rB r� � stopNamesMap� stopNamesr� r' r' r( �startTagListItem& s" z/getPhases.<locals>.InBodyPhase.startTagListItemc S s> | j jddd�r | jtd�� | j j|� | jjj| jj_d S )Nr& r� )r� ) r> r� r� r2 r� rJ rN r^ r\ )rB r� r' r' r( �startTagPlaintext< s z0getPhases.<locals>.InBodyPhase.startTagPlaintextc S sb | j jddd�r | jtd�� | j jd jtkrR| jjdd|d i� | j jj � | j j |� d S )Nr& r� )r� r zunexpected-start-tagr@ r� ) r> r� r� r2 r� r@ r rJ r� r� r� )rB r� r' r' r( �startTagHeadingB s z.getPhases.<locals>.InBodyPhase.startTagHeadingc S s~ | j jd�}|rf| jjdddd�� | jtd�� || j jkrL| j jj|� || j jkrf| j jj|� | j j � | j |� d S )N�az$unexpected-start-tag-implies-end-tag)� startName�endName) r> �!elementInActiveFormattingElementsrJ r� �endTagFormattingr2 r� rb rl rx rn )rB r� � afeAElementr' r' r( � startTagAJ s z(getPhases.<locals>.InBodyPhase.startTagAc S s | j j� | j|� d S )N)r> rx rn )rB r� r' r' r( �startTagFormattingW s z1getPhases.<locals>.InBodyPhase.startTagFormattingc S sP | j j� | j jd�rB| jjdddd�� | jtd�� | j j� | j|� d S )N�nobrz$unexpected-start-tag-implies-end-tag)r� r� )r> rx r� rJ r� r� r2 rn )rB r� r' r' r( � startTagNobr[ s z+getPhases.<locals>.InBodyPhase.startTagNobrc S sT | j jd�r2| jjdddd�� | jtd�� |S | j j� | j j|� d| j_d S )Nr� z$unexpected-start-tag-implies-end-tag)r� r� F) r> r� rJ r� r� r2 rx r� rd )rB r� r' r' r( �startTagButtone s z-getPhases.<locals>.InBodyPhase.startTagButtonc S s0 | j j� | j j|� | j jjt� d| j_d S )NF)r> rx r� rl r� r rJ rd )rB r� r' r' r( �startTagAppletMarqueeObjectp s z:getPhases.<locals>.InBodyPhase.startTagAppletMarqueeObjectc S sB | j jddd�r | jtd�� | j j� d| j_| jj|d� d S )Nr& r� )r� Fr� )r> r� r� r2 rx rJ rd r� )rB r� r' r' r( � startTagXmpv s z*getPhases.<locals>.InBodyPhase.startTagXmpc S sR | j jdkr*| jjddd�r*| jtd�� | jj|� d| j _| j jd | j _ d S )Nr* r& r� )r� Fr� ) rJ rY r> r� r� r2 r� rd rG r_ )rB r� r' r' r( � startTagTable} s z,getPhases.<locals>.InBodyPhase.startTagTablec S s6 | j j� | j j|� | j jj� d|d< d| j_d S )NTr� F)r> rx r� r� r� rJ rd )rB r� r' r' r( �startTagVoidFormatting� s z5getPhases.<locals>.InBodyPhase.startTagVoidFormattingc S s@ | j j}| j|� d|d kr<|d d jt�dkr<|| j _d S )Nr0 ry �hidden)rJ rd r� rn r )rB r� rd r' r' r( � startTagInput� s z,getPhases.<locals>.InBodyPhase.startTagInputc S s$ | j j|� | j jj� d|d< d S )NTr� )r> r� r� r� )rB r� r' r' r( �startTagParamSource� s z2getPhases.<locals>.InBodyPhase.startTagParamSourcec S sJ | j jddd�r | jtd�� | j j|� | j jj� d|d< d| j_d S )Nr& r� )r� Tr� F) r> r� r� r2 r� r� r� rJ rd )rB r� r' r' r( � startTagHr� s z)getPhases.<locals>.InBodyPhase.startTagHrc S s6 | j jdddd�� | jtdd|d |d d�� d S ) Nz unexpected-start-tag-treated-as�image�img)� originalName�newNamert ry r )rm r )rJ r� r� r2 )rB r� r' r' r( � startTagImage� s z,getPhases.<locals>.InBodyPhase.startTagImagec S s | j jdddi� | jjr d S i }d|d kr>|d d |d< | jtdd|d�� | jtd d�� | jtd d�� d |d kr�|d d }nd }| jtd |d�� |d j� }d|kr�|d= d |kr�|d = d|d< | jtdd||d d�� | j td �� | jtd d�� | j td�� d S )Nzdeprecated-tagr@ �isindex�actionry r� rt )rm �hr�label�promptz3This is a searchable index. Enter search keywords: rr )r0 ry �inputr )rm r ) rJ r� r> r� r� r2 r� r �copyr� )rB r� � form_attrsr� rm r' r' r( �startTagIsIndex� s6 z.getPhases.<locals>.InBodyPhase.startTagIsIndexc S s0 | j j|� | jjj| jj_| j| _d| j_d S )NF) r> r� rJ rN r[ r\ ry r� rd )rB r� r' r' r( �startTagTextarea� s z/getPhases.<locals>.InBodyPhase.startTagTextareac S s d| j _| j|� d S )NF)rJ rd �startTagRawtext)rB r� r' r' r( �startTagIFrame� s z-getPhases.<locals>.InBodyPhase.startTagIFramec S s" | j jr| j|� n | j|� d S )N)rJ rL r� r8 )rB r� r' r' r( rK � s z/getPhases.<locals>.InBodyPhase.startTagNoscriptc S s | j j|d� dS )z8iframe, noembed noframes, noscript(if scripting enabled)r� N)rJ r� )rB r� r' r' r( r� � s z.getPhases.<locals>.InBodyPhase.startTagRawtextc S s@ | j jd jdkr$| jjjtd�� | j j� | jj j|� d S )Nr �optionr� ) r> r� r@ rJ r_ r� r2 rx r� )rB r� r' r' r( � startTagOpt� s z*getPhases.<locals>.InBodyPhase.startTagOptc S s� | j j� | j j|� d| j_| jj| jjd | jjd | jjd | jjd | jjd | jjd fkrx| jjd | j_n| jjd | j_d S ) NFr� r� r� r� r� r� �inSelectInTabler� )r> rx r� rJ rd r_ rG )rB r� r' r' r( �startTagSelect� s z-getPhases.<locals>.InBodyPhase.startTagSelectc S sB | j jd�r2| j j� | j jd jdkr2| jj� | j j|� d S )N�rubyr r� )r> r� �generateImpliedEndTagsr� r@ rJ r� r� )rB r� r' r' r( � startTagRpRt� s z+getPhases.<locals>.InBodyPhase.startTagRpRtc S sZ | j j� | jj|� | jj|� td |d< | j j|� |d rV| j jj� d|d< d S )Nrh rl r Tr� ) r> rx rJ r r r r� r� r� )rB r� r' r' r( � startTagMath� s z+getPhases.<locals>.InBodyPhase.startTagMathc S sZ | j j� | jj|� | jj|� td |d< | j j|� |d rV| j jj� d|d< d S )Nr} rl r Tr� ) r> rx rJ r r r r� r� r� )rB r� r' r' r( � startTagSvg s z*getPhases.<locals>.InBodyPhase.startTagSvgc S s | j jdd|d i� dS )a5 Elements that should be children of other elements that have a different insertion mode; here they are ignored "caption", "col", "colgroup", "frame", "frameset", "head", "option", "optgroup", "tbody", "td", "tfoot", "th", "thead", "tr", "noscript" z unexpected-start-tag-ignoredr@ N)rJ r� )rB r� r' r' r( �startTagMisplaced s z0getPhases.<locals>.InBodyPhase.startTagMisplacedc S s | j j� | j j|� d S )N)r> rx r� )rB r� r' r' r( r8 s z,getPhases.<locals>.InBodyPhase.startTagOtherc S s� | j jddd�sD| jtdd�� | jjdddi� | jtdd�� nX| j jd� | j jd j dkrt| jjdddi� | j jj � }x|j dkr�| j jj � }q�W d S ) Nr& r� )r� rt zunexpected-end-tagr@ ru r r� ) r> r� r� r2 rJ r� r� r� r� r@ r� )rB r� r� r' r' r( r� s z&getPhases.<locals>.InBodyPhase.endTagPc S s� | j jd�s| jj� d S | j jd jdkrlx>| j jdd � D ]*}|jtd�kr>| jjdd|jd�� P q>W | jjd | j_d S )Nr� r � ro rp rq �optgroupr� r& �rp�rtr� r� r� r� r� r� r� z$expected-one-end-tag-but-got-another)�gotName� expectedName� afterBodyr� )ro rp rq r� r� r& r� r� r� r� r� r� r� r� r� r� ) r> r� rJ r� r� r@ r� rG r_ )rB r� r� r' r' r( � endTagBody) s z)getPhases.<locals>.InBodyPhase.endTagBodyc S s" | j jd�r | jtd�� |S d S )Nr� )r> r� r� r2 )rB r� r' r' r( � endTagHtml; s z)getPhases.<locals>.InBodyPhase.endTagHtmlc S s� |d dkr| j | _| jj|d �}|r2| jj� | jjd j|d kr^| jjdd|d i� |r�| jjj � }x |j|d kr�| jjj � }qpW d S )Nr@ rt r zend-tag-too-earlyr� ) rf r� r> r� r� r� r@ rJ r� r� )rB r� �inScoper� r' r' r( � endTagBlockA s z*getPhases.<locals>.InBodyPhase.endTagBlockc S sx | j j}d | j _|d ks&| j j|� r:| jjdddi� n:| j j� | j jd |krf| jjdddi� | j jj|� d S )Nzunexpected-end-tagr@ r� r zend-tag-too-early-ignoredr� )r> r� r� rJ r� r� r� rb )rB r� r� r' r' r( � endTagFormO s z)getPhases.<locals>.InBodyPhase.endTagFormc S s� |d dkrd}nd }| j j|d |d�sB| jjdd|d i� nj| j j|d d� | j jd j|d kr�| jjdd|d i� | j jj� }x |j|d kr�| j jj� }q�W d S ) Nr@ rq �list)r� zunexpected-end-tag)�excluder zend-tag-too-earlyr� )r> r� rJ r� r� r� r@ r� )rB r� r� r� r' r' r( �endTagListItem\ s z-getPhases.<locals>.InBodyPhase.endTagListItemc S s� x$t D ] }| jj|�r| jj� P qW | jjd j|d krR| jjdd|d i� xBt D ]:}| jj|�rX| jjj� }x|jt kr�| jjj� }qvW P qXW d S )Nr r@ zend-tag-too-earlyr� ) r r> r� r� r� r@ rJ r� r� )rB r� �itemr' r' r( � endTagHeadingm s z,getPhases.<locals>.InBodyPhase.endTagHeadingc S s" d}�x|dk �r |d7 }| j j|d �}| sL|| j jkrZ| j j|j� rZ| j|� dS || j jkr�| jjdd|d i� | j jj |� dS | j j|j�s�| jjdd|d i� dS || j jd kr�| jjdd|d i� | j jj |�}d}x,| j j|d� D ]}|j t k�r |}P �q W |dk�rb| j jj � }x ||k�rN| j jj � }�q4W | j jj |� dS | j j|d }| j jj |�}| } } d} | j jj | �} x�| d k �rh| d7 } | d8 } | j j| } | | j jk�r�| j jj | � �q�| |k�r�P | |k�r | j jj | �d }| j� } | | j j| j jj | �< | | j j| j jj | �< | } | j�rV| jj| � | j| � | } �q�W | j�r~| jj| � |jtd�k�r�| j j� \}}|j| |� n |j| � |j� } |j| � |j| � | j jj |� | j jj|| � | j jj |� | j jj| j jj |�d | � qW dS )z)The much-feared adoption agency algorithmr � r r@ Nzadoption-agency-1.2zadoption-agency-4.4zadoption-agency-1.3rk r� r� r� r� r� r� )r� r� r� r� r� )r> r� r� r� r@ r: rJ r� rl rb �indexr� r r� � cloneNoder r� � appendChildr� � getTableMisnestedNodePosition� insertBefore�reparentChildren�insert)rB r� �outerLoopCounter�formattingElement�afeIndex� furthestBlockro �commonAncestor�bookmark�lastNoder� �innerLoopCounterr� �cloner r� r' r' r( r� | s� z/getPhases.<locals>.InBodyPhase.endTagFormattingc S s� | j j|d �r| j j� | j jd j|d krF| jjdd|d i� | j j|d �r�| j jj� }x |j|d kr�| j jj� }qdW | j j� d S )Nr@ r zend-tag-too-earlyr� ) r> r� r� r� r@ rJ r� r� � clearActiveFormattingElements)rB r� ro r' r' r( �endTagAppletMarqueeObject s z8getPhases.<locals>.InBodyPhase.endTagAppletMarqueeObjectc S s@ | j jdddd�� | jj� | jjtdd�� | jjj� d S )Nz unexpected-end-tag-treated-asr3 z br element)r� r� rt )rJ r� r> rx r� r2 r� r� )rB r� r' r' r( r^ + s z'getPhases.<locals>.InBodyPhase.endTagBrc S s� x�| j jd d d� D ]�}|j|d kr~| j j|d d� | j jd j|d krd| jjdd|d i� x| j jj� |krxqfW P q|jtkr| jjdd|d i� P qW d S )Nr r@ )r� zunexpected-end-tagr� r� ) r> r� r@ r� rJ r� r� r� r )rB r� r� r' r' r( r: 2 s z*getPhases.<locals>.InBodyPhase.endTagOtherr� rV rW rX rY rZ r2 rU rS rQ r� r� r� �article�aside� blockquote�center�details�dirr) �dl�fieldset� figcaption�figure�footer�header�hgroup�main�menu�nav�olr& �section�summary�ulrt ru r� rq ro rp rT r� �b�big�code�em�font�i�s�small�strike�strong�tt�ur� r� �applet�marquee�object�xmpr� �arear3 �embedr� �keygen�wbr�param�source�trackr� r� r� r� rv �iframerT �noembedrR r� r� r� r� r� �mathr} r� �colr� �framer� r� r� r� r� r� r� �dialog)r� ) rV rW rX rY rZ r2 rU rS rQ )r� r� r� r� r� r� r� r) r� r� r� r� r� r� r� r� r� r� r� r& r� r� r� )rt ru )rq ro rp ) r� r� r� r� r� r� r� r� r r r r )r r r )r r3 r r� r r )r r r )r rR )r� r� )r� r� ) r� r r� r r� r� r� r� r� r� r� )r� r� r� r� r� r� r� r r� r) r� r� r� r� r� r� r� ru r� r� r� r� rt r� r� r� )ro rp rq )r� r� r� r� r� r� r� r� r� r� r r r r )r r r )?r9 r: r; r� rI rj rn r� ry r� rf r~ r` ra r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� rK r� r� r� r� r� r� r� r8 r� r� r� r� r� r� r� r� r� r^ r: r r; r� r r� r< r� � __classcell__r' )rg r� )r� r( rg � s� $ rg c @ s` e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z e j g �Z ee _ e j d efg�Ze e_ d S ) z getPhases.<locals>.TextPhasec S s | j j|d � d S )Nry )r> r� )rB r� r' r' r( r� � s z.getPhases.<locals>.TextPhase.processCharactersc S s8 | j jdd| jjd ji� | jjj� | j j| j _dS )Nz&expected-named-closing-tag-but-got-eofr@ r Tr� )rJ r� r> r� r@ r� r� r_ )rB r' r' r( r� � s z'getPhases.<locals>.TextPhase.processEOFc S s dst d|d ��d S )NFz4Tried to process start tag %s in RCDATA/RAWTEXT moder@ )r� )rB r� r' r' r( r8 � s z*getPhases.<locals>.TextPhase.startTagOtherc S s* | j jj� }|jdkst�| jj| j_d S )NrU )r> r� r� r@ r� rJ r� r_ )rB r� r� r' r' r( � endTagScript� s z)getPhases.<locals>.TextPhase.endTagScriptc S s | j jj� | jj| j_d S )N)r> r� r� rJ r� r_ )rB r� r' r' r( r: � s z(getPhases.<locals>.TextPhase.endTagOtherrU N)r9 r: r; r/ r� r� r� r8 r r: r r; r� r< r� r' r' r' r( � TextPhase� s r c s e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z dd� Zdd� Zdd� Zdd� Zdd � Zd d � Zd d � Zd!d"� Zd#d$� Zejd%� jfd&e fd'e fd(e fd6e fd7efd/efd8efd2efd3efg �Zee_ejd/efd9efg�Z ee _d5S ):z getPhases.<locals>.InTablePhasec S s( x"| j jd jdkr"| j jj� qW d S )Nr r� r� r� )r� r� )r> r� r@ r� )rB r' r' r( �clearStackToTableContext� s z8getPhases.<locals>.InTablePhase.clearStackToTableContextc S s0 | j jd jdkr | jjd� n | jjs,t�d S )Nr r� z eof-in-tabler� )r> r� r@ rJ r� rR r� )rB r' r' r( r� � s z*getPhases.<locals>.InTablePhase.processEOFc S s4 | j j}| j jd | j _|| j j_| j jj|� d S )N� inTableText)rJ r_ rG r� r� )rB r� r� r' r' r( r� � s z6getPhases.<locals>.InTablePhase.processSpaceCharactersc S s4 | j j}| j jd | j _|| j j_| j jj|� d S )Nr )rJ r_ rG r� r� )rB r� r� r' r' r( r� � s z1getPhases.<locals>.InTablePhase.processCharactersc S s& d| j _| jjd j|� d| j _d S )NTr� F)r> �insertFromTablerJ rG r� )rB r� r' r' r( r� � s z*getPhases.<locals>.InTablePhase.insertTextc S s6 | j � | jjjt� | jj|� | jjd | j_d S )Nr� ) r r> rl r� r r� rJ rG r_ )rB r� r' r' r( �startTagCaption� s z/getPhases.<locals>.InTablePhase.startTagCaptionc S s( | j � | jj|� | jjd | j_d S )Nr� )r r> r� rJ rG r_ )rB r� r' r' r( �startTagColgroup� s z0getPhases.<locals>.InTablePhase.startTagColgroupc S s | j tdd�� |S )Nr� rt )r r2 )rB r� r' r' r( � startTagCol� s z+getPhases.<locals>.InTablePhase.startTagColc S s( | j � | jj|� | jjd | j_d S )Nr� )r r> r� rJ rG r_ )rB r� r' r' r( �startTagRowGroup� s z0getPhases.<locals>.InTablePhase.startTagRowGroupc S s | j tdd�� |S )Nr� rt )r r2 )rB r� r' r' r( �startTagImplyTbody� s z2getPhases.<locals>.InTablePhase.startTagImplyTbodyc S s6 | j jdddd�� | j jjtd�� | j js2|S d S )Nz$unexpected-start-tag-implies-end-tagr� )r� r� )rJ r� r_ r� r2 rR )rB r� r' r' r( r� � s z-getPhases.<locals>.InTablePhase.startTagTablec S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( �startTagStyleScript� s z3getPhases.<locals>.InTablePhase.startTagStyleScriptc S sV d|d krH|d d j t�dkrH| jjd� | jj|� | jjj� n | j|� d S )Nr0 ry r� z unexpected-hidden-input-in-table) rn r rJ r� r> r� r� r� r8 )rB r� r' r' r( r� � s z-getPhases.<locals>.InTablePhase.startTagInputc S sD | j jd� | jjd kr@| jj|� | jjd | j_| jjj� d S )Nzunexpected-form-in-tabler r� )rJ r� r> r� r� r� r� )rB r� r' r' r( r� � s z,getPhases.<locals>.InTablePhase.startTagFormc S s< | j jdd|d i� d| j_| j jd j|� d| j_d S )Nz)unexpected-start-tag-implies-table-voodoor@ Tr� F)rJ r� r> r rG r� )rB r� r' r' r( r8 � s z-getPhases.<locals>.InTablePhase.startTagOtherc S s� | j jddd�r�| j j� | j jd jdkrJ| jjdd| j jd jd�� x"| j jd jdkrl| j jj� qLW | j jj� | jj� n| jj s�t �| jj� d S ) Nr� )r� r zend-tag-too-early-named)r� r� r� r� r� ) r> r� r� r� r@ rJ r� r� ra rR r� )rB r� r' r' r( � endTagTable s z+getPhases.<locals>.InTablePhase.endTagTablec S s | j jdd|d i� d S )Nzunexpected-end-tagr@ )rJ r� )rB r� r' r' r( � endTagIgnore s z,getPhases.<locals>.InTablePhase.endTagIgnorec S s< | j jdd|d i� d| j_| j jd j|� d| j_d S )Nz'unexpected-end-tag-implies-table-voodoor@ Tr� F)rJ r� r> r rG r� )rB r� r' r' r( r: s z+getPhases.<locals>.InTablePhase.endTagOtherr� r� r� r r� r� r� r� r� r� r� rS rU r� r� r� N)r� r� r� )r� r� r� )rS rU ) r� r� r r� r� r� r� r� r� r� r� ) r9 r: r; r/ r� r r� r� r� r� r r r r r r� r r� r� r8 r! r" r: r r; r� r� r< r� r' )r� r' r( � InTablePhase� sH r# c sZ e Zd ZdZ�� fdd�Zdd� Zdd� Zd d � Zd d � Zd d� Z dd� Z dd� Z � Z S )z#getPhases.<locals>.InTableTextPhaser� �characterTokensc s" t � | �j||� d | _g | _d S )N)re rI r� r$ )rB r� r$ )�InTableTextPhaser� r' r( rI 6 s z,getPhases.<locals>.InTableTextPhase.__init__c S sd dj dd� | jD ��}tdd� |D ��rJtd |d�}| jjd j|� n|rZ| jj|� g | _d S )Nr� c S s g | ] }|d �qS )ry r' )r? r� r' r' r( r| < s zGgetPhases.<locals>.InTableTextPhase.flushCharacters.<locals>.<listcomp>c S s g | ] }|t k�qS r' )r )r? r� r' r' r( r| = s rr )r0 ry r� )�joinr$ r} r rJ rG r� r> )rB ry r� r' r' r( �flushCharacters; s z3getPhases.<locals>.InTableTextPhase.flushCharactersc S s | j � | j| j_|S )N)r' r� rJ r_ )rB r� r' r' r( r� D s z2getPhases.<locals>.InTableTextPhase.processCommentc S s | j � | j| j_dS )NT)r' r� rJ r_ )rB r' r' r( r� I s z.getPhases.<locals>.InTableTextPhase.processEOFc S s |d dkrd S | j j|� d S )Nry rz )r$ r� )rB r� r' r' r( r� N s z5getPhases.<locals>.InTableTextPhase.processCharactersc S s | j j|� d S )N)r$ r� )rB r� r' r' r( r� S s z:getPhases.<locals>.InTableTextPhase.processSpaceCharactersc S s | j � | j| j_|S )N)r' r� rJ r_ )rB r� r' r' r( r� X s z3getPhases.<locals>.InTableTextPhase.processStartTagc S s | j � | j| j_|S )N)r' r� rJ r_ )rB r� r' r' r( r� ] s z1getPhases.<locals>.InTableTextPhase.processEndTag)r� r$ ) r9 r: r; r� rI r' r� r� r� r� r� r� r r' )r% )r� r( r% 3 s r% c s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z ejd� jfd efg�Ze e_ejde fd e fd!e fg�Ze e_d S )"z!getPhases.<locals>.InCaptionPhasec S s | j jddd� S )Nr� r� )r� )r> r� )rB r' r' r( �ignoreEndTagCaptionf s z5getPhases.<locals>.InCaptionPhase.ignoreEndTagCaptionc S s | j jd j� d S )Nr� )rJ rG r� )rB r' r' r( r� i s z,getPhases.<locals>.InCaptionPhase.processEOFc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� l s z3getPhases.<locals>.InCaptionPhase.processCharactersc S s0 | j j� | j� }| j jjtd�� |s,|S d S )Nr� )rJ r� r( r_ r� r2 )rB r� � ignoreEndTagr' r' r( �startTagTableElemento s z6getPhases.<locals>.InCaptionPhase.startTagTableElementc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r8 w s z/getPhases.<locals>.InCaptionPhase.startTagOtherc S s� | j � s�| jj� | jjd jdkrB| jjdd| jjd jd�� x"| jjd jdkrd| jjj� qDW | jjj� | jj� | jj d | j_ n| jj s�t �| jj� d S ) Nr r� z$expected-one-end-tag-but-got-another)r� r� r� r� r� r� ) r( r> r� r� r@ rJ r� r� r� rG r_ rR r� )rB r� r' r' r( � endTagCaptionz s z/getPhases.<locals>.InCaptionPhase.endTagCaptionc S s0 | j j� | j� }| j jjtd�� |s,|S d S )Nr� )rJ r� r( r_ r� r2 )rB r� r) r' r' r( r! � s z-getPhases.<locals>.InCaptionPhase.endTagTablec S s | j jdd|d i� d S )Nzunexpected-end-tagr@ )rJ r� )rB r� r' r' r( r" � s z.getPhases.<locals>.InCaptionPhase.endTagIgnorec S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r: � s z-getPhases.<locals>.InCaptionPhase.endTagOtherr� r� r r� r� r� r� r� r� r� r� r� N) r� r r� r� r� r� r� r� r� ) r� r r� r� r� r� r� r� r� r� )r9 r: r; r/ r� r( r� r� r* r8 r+ r! r" r: r r; r� r� r< r� r' )r� r' r( �InCaptionPhaseb s* r, c s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z e jd� jfdefg�Ze e_e jde fde fg�Ze e_dS )z%getPhases.<locals>.InColumnGroupPhasec S s | j jd jdkS )Nr r� r� )r> r� r@ )rB r' r' r( �ignoreEndTagColgroup� s z:getPhases.<locals>.InColumnGroupPhase.ignoreEndTagColgroupc S sD | j jd jdkr"| jjs t�d S | j� }| jtd�� |s@dS d S )Nr r� r� Tr� ) r> r� r@ rJ rR r� r- �endTagColgroupr2 )rB r) r' r' r( r� � s z0getPhases.<locals>.InColumnGroupPhase.processEOFc S s" | j � }| jtd�� |s |S d S )Nr� )r- r. r2 )rB r� r) r' r' r( r� � s z7getPhases.<locals>.InColumnGroupPhase.processCharactersc S s$ | j j|� | j jj� d|d< d S )NTr� )r> r� r� r� )rB r� r' r' r( r � s z1getPhases.<locals>.InColumnGroupPhase.startTagColc S s" | j � }| jtd�� |s |S d S )Nr� )r- r. r2 )rB r� r) r' r' r( r8 � s z3getPhases.<locals>.InColumnGroupPhase.startTagOtherc S s@ | j � r | jjst�| jj� n | jjj� | jjd | j_ d S )Nr� ) r- rJ rR r� r� r> r� r� rG r_ )rB r� r' r' r( r. � s z4getPhases.<locals>.InColumnGroupPhase.endTagColgroupc S s | j jdddi� d S )Nz no-end-tagr@ r )rJ r� )rB r� r' r' r( � endTagCol� s z/getPhases.<locals>.InColumnGroupPhase.endTagColc S s" | j � }| jtd�� |s |S d S )Nr� )r- r. r2 )rB r� r) r' r' r( r: � s z1getPhases.<locals>.InColumnGroupPhase.endTagOtherr� r r� N)r9 r: r; r/ r� r- r� r� r r8 r. r/ r: r r; r� r� r< r� r' )r� r' r( �InColumnGroupPhase� s" r0 c s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z dd� Zdd� Zdd� Zejd� jfde fd&e fd'e fg�Ze e_ejd(e fd#efd)efg�Zee_d%S )*z#getPhases.<locals>.InTableBodyPhasec S sF x"| j jd jdkr"| j jj� qW | j jd jdkrB| jjsBt�d S ) Nr r� r� r� r� r� )r� r� r� r� r� )r> r� r@ r� rJ rR r� )rB r' r' r( � clearStackToTableBodyContext� s z@getPhases.<locals>.InTableBodyPhase.clearStackToTableBodyContextc S s | j jd j� d S )Nr� )rJ rG r� )rB r' r' r( r� � s z.getPhases.<locals>.InTableBodyPhase.processEOFc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z:getPhases.<locals>.InTableBodyPhase.processSpaceCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z5getPhases.<locals>.InTableBodyPhase.processCharactersc S s( | j � | jj|� | jjd | j_d S )Nr� )r1 r> r� rJ rG r_ )rB r� r' r' r( � startTagTr s z.getPhases.<locals>.InTableBodyPhase.startTagTrc S s* | j jdd|d i� | jtdd�� |S )Nz unexpected-cell-in-table-bodyr@ r� rt )rJ r� r2 r2 )rB r� r' r' r( �startTagTableCell s z5getPhases.<locals>.InTableBodyPhase.startTagTableCellc S sn | j jddd�s0| j jddd�s0| j jddd�rT| j� | jt| j jd j�� |S | jjs`t �| jj � d S )Nr� r� )r� r� r� r r� ) r> r� r1 �endTagTableRowGroupr2 r� r@ rJ rR r� r� )rB r� r' r' r( �startTagTableOther s z6getPhases.<locals>.InTableBodyPhase.startTagTableOtherc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r8 s z1getPhases.<locals>.InTableBodyPhase.startTagOtherc S sT | j j|d dd�r:| j� | j jj� | jjd | j_n| jjdd|d i� d S )Nr@ r� )r� r� z unexpected-end-tag-in-table-body) r> r� r1 r� r� rJ rG r_ r� )rB r� r' r' r( r4 s z7getPhases.<locals>.InTableBodyPhase.endTagTableRowGroupc S sn | j jddd�s0| j jddd�s0| j jddd�rT| j� | jt| j jd j�� |S | jjs`t �| jj � d S )Nr� r� )r� r� r� r r� ) r> r� r1 r4 r2 r� r@ rJ rR r� r� )rB r� r' r' r( r! % s z/getPhases.<locals>.InTableBodyPhase.endTagTablec S s | j jdd|d i� d S )Nz unexpected-end-tag-in-table-bodyr@ )rJ r� )rB r� r' r' r( r" 2 s z0getPhases.<locals>.InTableBodyPhase.endTagIgnorec S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r: 6 s z/getPhases.<locals>.InTableBodyPhase.endTagOtherr� r� r� r� r� r r� r� r� r� r� r� N)r� r� )r� r r� r� r� r� )r� r� r� )r� r� r r� r� r� r� r� )r9 r: r; r/ r� r1 r� r� r� r2 r3 r5 r8 r4 r! r" r: r r; r� r� r< r� r' )r� r' r( �InTableBodyPhase� s4 r6 c s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z dd� Zdd� Zdd� Zdd� Zejd� jfd(e fd)e fg�Ze e_ejd$e fd%efd*efd+efg�Zee_d'S ),z getPhases.<locals>.InRowPhasec S sD x>| j jd jdkr>| jjdd| j jd ji� | j jj� qW d S ) Nr r� r� z'unexpected-implied-end-tag-in-table-rowr@ r� )r� r� r� )r> r� r@ rJ r� r� )rB r' r' r( �clearStackToTableRowContextO s z9getPhases.<locals>.InRowPhase.clearStackToTableRowContextc S s | j jddd� S )Nr� r� )r� )r> r� )rB r' r' r( �ignoreEndTagTrU s z,getPhases.<locals>.InRowPhase.ignoreEndTagTrc S s | j jd j� d S )Nr� )rJ rG r� )rB r' r' r( r� Y s z(getPhases.<locals>.InRowPhase.processEOFc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� \ s z4getPhases.<locals>.InRowPhase.processSpaceCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� _ s z/getPhases.<locals>.InRowPhase.processCharactersc S s6 | j � | jj|� | jjd | j_| jjjt� d S )Nr� ) r7 r> r� rJ rG r_ rl r� r )rB r� r' r' r( r3 b s z/getPhases.<locals>.InRowPhase.startTagTableCellc S s" | j � }| jtd�� |s |S d S )Nr� )r8 �endTagTrr2 )rB r� r) r' r' r( r5 h s z0getPhases.<locals>.InRowPhase.startTagTableOtherc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r8 o s z+getPhases.<locals>.InRowPhase.startTagOtherc S sH | j � s.| j� | jjj� | jjd | j_n| jjs:t �| jj � d S )Nr� ) r8 r7 r> r� r� rJ rG r_ rR r� r� )rB r� r' r' r( r9 r s z&getPhases.<locals>.InRowPhase.endTagTrc S s" | j � }| jtd�� |s |S d S )Nr� )r8 r9 r2 )rB r� r) r' r' r( r! | s z)getPhases.<locals>.InRowPhase.endTagTablec S s4 | j j|d dd�r&| jtd�� |S | jj� d S )Nr@ r� )r� r� )r> r� r9 r2 rJ r� )rB r� r' r' r( r4 � s z1getPhases.<locals>.InRowPhase.endTagTableRowGroupc S s | j jdd|d i� d S )Nz unexpected-end-tag-in-table-rowr@ )rJ r� )rB r� r' r' r( r" � s z*getPhases.<locals>.InRowPhase.endTagIgnorec S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r: � s z)getPhases.<locals>.InRowPhase.endTagOtherr� r� r� r� r r� r� r� r� r� r� r� N)r� r� )r� r r� r� r� r� r� )r� r� r� )r� r� r r� r� r� r� )r9 r: r; r/ r� r7 r8 r� r� r� r3 r5 r8 r9 r! r4 r" r: r r; r� r� r< r� r' )r� r' r( � InRowPhaseJ s6 r: c s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z ejd� jfd efg�Ze e_ejd!e fd"e fd#e fg�Ze e_d S )$z getPhases.<locals>.InCellPhasec S sB | j jddd�r | jtd�� n | j jddd�r>| jtd�� d S )Nr� r� )r� r� )r> r� �endTagTableCellr2 )rB r' r' r( � closeCell� s z(getPhases.<locals>.InCellPhase.closeCellc S s | j jd j� d S )Nr� )rJ rG r� )rB r' r' r( r� � s z)getPhases.<locals>.InCellPhase.processEOFc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z0getPhases.<locals>.InCellPhase.processCharactersc S sF | j jddd�s | j jddd�r,| j� |S | jjs8t�| jj� d S )Nr� r� )r� r� )r> r� r< rJ rR r� r� )rB r� r' r' r( r5 � s z1getPhases.<locals>.InCellPhase.startTagTableOtherc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r8 � s z,getPhases.<locals>.InCellPhase.startTagOtherc S s� | j j|d dd�r�| j j|d � | j jd j|d krt| jjdd|d i� x.| j jj� }|j|d krRP qRW n | j jj� | j j� | jj d | j_ n| jjdd|d i� d S ) Nr@ r� )r� r zunexpected-cell-end-tagr� zunexpected-end-tagr� ) r> r� r� r� r@ rJ r� r� r� rG r_ )rB r� r� r' r' r( r; � s z.getPhases.<locals>.InCellPhase.endTagTableCellc S s | j jdd|d i� d S )Nzunexpected-end-tagr@ )rJ r� )rB r� r' r' r( r" � s z+getPhases.<locals>.InCellPhase.endTagIgnorec S s. | j j|d dd�r | j� |S | jj� d S )Nr@ r� )r� )r> r� r< rJ r� )rB r� r' r' r( � endTagImply� s z*getPhases.<locals>.InCellPhase.endTagImplyc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r: � s z*getPhases.<locals>.InCellPhase.endTagOtherr� r� r r� r� r� r� r� r� r� r� r� N) r� r r� r� r� r� r� r� r� )r� r� )r� r� r r� r� )r� r� r� r� r� )r9 r: r; r/ r� r< r� r� r5 r8 r; r" r= r: r r; r� r� r< r� r' )r� r' r( � InCellPhase� s( r> c s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z dd� Z dd� Zdd� Zdd� Zejd� jfdefdefd e fd"e fd e fg�Ze e_ejde fdefd efg�Zee_d!S )#z getPhases.<locals>.InSelectPhasec S s0 | j jd jdkr | jjd� n | jjs,t�d S )Nr r� z eof-in-selectr� )r> r� r@ rJ r� rR r� )rB r' r' r( r� � s z+getPhases.<locals>.InSelectPhase.processEOFc S s$ |d dkrd S | j j|d � d S )Nry rz )r> r� )rB r� r' r' r( r� � s z2getPhases.<locals>.InSelectPhase.processCharactersc S s. | j jd jdkr | j jj� | j j|� d S )Nr r� r� )r> r� r@ r� r� )rB r� r' r' r( �startTagOption� s z/getPhases.<locals>.InSelectPhase.startTagOptionc S sL | j jd jdkr | j jj� | j jd jdkr<| j jj� | j j|� d S )Nr r� r� r� r� )r> r� r@ r� r� )rB r� r' r' r( �startTagOptgroup s z1getPhases.<locals>.InSelectPhase.startTagOptgroupc S s | j jd� | jtd�� d S )Nzunexpected-select-in-selectr� )rJ r� � endTagSelectr2 )rB r� r' r' r( r� s z/getPhases.<locals>.InSelectPhase.startTagSelectc S s> | j jd� | jjddd�r.| jtd�� |S | j js:t�d S )Nzunexpected-input-in-selectr� )r� )rJ r� r> r� rA r2 rR r� )rB r� r' r' r( r� s z.getPhases.<locals>.InSelectPhase.startTagInputc S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( rM s z/getPhases.<locals>.InSelectPhase.startTagScriptc S s | j jdd|d i� d S )Nz unexpected-start-tag-in-selectr@ )rJ r� )rB r� r' r' r( r8 s z.getPhases.<locals>.InSelectPhase.startTagOtherc S s6 | j jd jdkr | j jj� n| jjdddi� d S )Nr r� z unexpected-end-tag-in-selectr@ r� )r> r� r@ r� rJ r� )rB r� r' r' r( � endTagOption s z-getPhases.<locals>.InSelectPhase.endTagOptionc S sf | j jd jdkr0| j jd jdkr0| j jj� | j jd jdkrP| j jj� n| jjdddi� d S ) Nr r� r� r� z unexpected-end-tag-in-selectr@ r� ���r� )r> r� r@ r� rJ r� )rB r� r' r' r( �endTagOptgroup% s z/getPhases.<locals>.InSelectPhase.endTagOptgroupc S s^ | j jddd�rD| j jj� }x|jdkr6| j jj� }q W | jj� n| jjsPt�| jj � d S )Nr� )r� ) r> r� r� r� r@ rJ ra rR r� r� )rB r� r� r' r' r( rA 2 s z-getPhases.<locals>.InSelectPhase.endTagSelectc S s | j jdd|d i� d S )Nz unexpected-end-tag-in-selectr@ )rJ r� )rB r� r' r' r( r: = s z,getPhases.<locals>.InSelectPhase.endTagOtherr� r� r� r� r� r rv rU N)r� r rv )r9 r: r; r/ r� r� r� r? r@ r� r� rM r8 rB rD rA r: r r; r� r� r< r� r' )r� r' r( � InSelectPhase� s4 rE c @ sn e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z e j defg�Z ee _e j de fg�Ze e_dS )z'getPhases.<locals>.InSelectInTablePhasec S s | j jd j� d S )Nr� )rJ rG r� )rB r' r' r( r� U s z2getPhases.<locals>.InSelectInTablePhase.processEOFc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� X s z9getPhases.<locals>.InSelectInTablePhase.processCharactersc S s( | j jdd|d i� | jtd�� |S )Nz5unexpected-table-element-start-tag-in-select-in-tabler@ r� )rJ r� r: r2 )rB r� r' r' r( r� [ s z5getPhases.<locals>.InSelectInTablePhase.startTagTablec S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r8 ` s z5getPhases.<locals>.InSelectInTablePhase.startTagOtherc S s@ | j jdd|d i� | jj|d dd�r<| jtd�� |S d S )Nz3unexpected-table-element-end-tag-in-select-in-tabler@ r� )r� r� )rJ r� r> r� r: r2 )rB r� r' r' r( r! c s z3getPhases.<locals>.InSelectInTablePhase.endTagTablec S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r: i s z3getPhases.<locals>.InSelectInTablePhase.endTagOtherr� r� r� r� r� r� r� r� N)r� r� r� r� r� r� r� r� )r� r� r� r� r� r� r� r� )r9 r: r; r/ r� r� r� r� r8 r! r: r r; r� r< r� r' r' r' r( �InSelectInTablePhaseR s rF c - s� e Zd Ze� Zeddddddddd d d d d ddddddddddddddd d d d d d!d"d#d$d%d&d'd(d)d*d+d,g,�Zd-d.� Z� fd/d0�Zd1d2� Z d3d4� Z d5S )6z(getPhases.<locals>.InForeignContentPhaser� r� r� r� r3 r� r� ro r) r� rp r� r �h1�h2�h3�h4�h5�h6r� r� r� r� rq ru r� r2 r� r� r& rt r� r� r� �spanr r �sub�supr� r r r� �varc % S sn ddddddddd d d d d ddddddddddddddd d d d d d!d"d#d$d%�$}|d& |krj||d& |d&< d S )'N�altGlyph� altGlyphDef� altGlyphItem� animateColor� animateMotion�animateTransform�clipPath�feBlend� feColorMatrix�feComponentTransfer� feComposite�feConvolveMatrix�feDiffuseLighting�feDisplacementMap�feDistantLight�feFlood�feFuncA�feFuncB�feFuncG�feFuncR�feGaussianBlur�feImage�feMerge� feMergeNode� feMorphology�feOffset� fePointLight�feSpecularLighting� feSpotLight�feTile� feTurbulence� foreignObject�glyphRef�linearGradient�radialGradient�textPath)$�altglyph� altglyphdef� altglyphitem� animatecolor� animatemotion�animatetransform�clippath�feblend� fecolormatrix�fecomponenttransfer� fecomposite�feconvolvematrix�fediffuselighting�fedisplacementmap�fedistantlight�feflood�fefunca�fefuncb�fefuncg�fefuncr�fegaussianblur�feimage�femerge� femergenode� femorphology�feoffset� fepointlight�fespecularlighting� fespotlight�fetile� feturbulence� foreignobject�glyphref�lineargradient�radialgradient�textpathr@ r' )rB r� � replacementsr' r' r( �adjustSVGTagNames� sL z:getPhases.<locals>.InForeignContentPhase.adjustSVGTagNamesc sL |d dkrd|d< n&| j jr<tdd� |d D ��r<d| j _� j| |� d S )Nry rz u �c s s | ]}|t kV qd S )N)r )r? r{ r' r' r( � <genexpr>� s zMgetPhases.<locals>.InForeignContentPhase.processCharacters.<locals>.<genexpr>F)rJ rd r} r� )rB r� )r� r' r( r� � s z:getPhases.<locals>.InForeignContentPhase.processCharactersc S s2 | j jd }|d | jks@|d dkr�t|d j� �dddh@ r�| jjdd|d i� xR| j jd j| j jkr�| jj | j jd � r�| jj | j jd � r�| j jj � qXW |S |jt d kr�| jj |� n$|jt d kr�| j|� | jj|� | jj|� |j|d < | j j|� |d �r.| j jj � d |d< d S )Nr r@ r� ry �color�face�sizez*unexpected-html-element-in-foreign-contentrh r} rl r Tr� r� r� r� r� )r> r� �breakoutElements�set�keysrJ r� rl r� rp rq r� r r r� r r r� )rB r� r� r' r' r( r� � s. z8getPhases.<locals>.InForeignContentPhase.processStartTagc S s� t | jj�d }| jjd }|jjt�|d krF| jjdd|d i� x�|jjt�|d kr�| jj| jj d kr�| jjj � | jjj | j_x | jjj � |kr�| jjs�t �q�W d }P |d8 }| jj| }|j| jjkr�qHqH| jjj|�}P qHW |S )Nr r@ zunexpected-end-tagr r� )r� r> r� r@ rn r rJ r� r_ rG r' r� r� r� rl r� r� )rB r� � nodeIndexr� r� r' r' r( r� � s( z6getPhases.<locals>.InForeignContentPhase.processEndTagN) r9 r: r; r/ r� r� r� r� r� r� r� r' )r� r' r( �InForeignContentPhasex s ) r� c @ sv e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z e j defg�Ze e_e j de fg�Ze e_dS )z!getPhases.<locals>.AfterBodyPhasec S s d S )Nr' )rB r' r' r( r� � s z,getPhases.<locals>.AfterBodyPhase.processEOFc S s | j j|| j jd � d S )Nr )r> r� r� )rB r� r' r' r( r� � s z0getPhases.<locals>.AfterBodyPhase.processCommentc S s | j jd� | j jd | j _|S )Nzunexpected-char-after-bodyr� )rJ r� rG r_ )rB r� r' r' r( r� � s z3getPhases.<locals>.AfterBodyPhase.processCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z.getPhases.<locals>.AfterBodyPhase.startTagHtmlc S s* | j jdd|d i� | j jd | j _|S )Nz unexpected-start-tag-after-bodyr@ r� )rJ r� rG r_ )rB r� r' r' r( r8 � s z/getPhases.<locals>.AfterBodyPhase.startTagOtherc S s* | j jr| j jd� n| j jd | j _d S )Nz'unexpected-end-tag-after-body-innerhtml�afterAfterBody)rJ rR r� rG r_ )rB r@ r' r' r( r� s z,getPhases.<locals>.AfterBodyPhase.endTagHtmlc S s* | j jdd|d i� | j jd | j _|S )Nz unexpected-end-tag-after-bodyr@ r� )rJ r� rG r_ )rB r� r' r' r( r: s z-getPhases.<locals>.AfterBodyPhase.endTagOtherr� N)r9 r: r; r/ r� r� r� r� r� r8 r� r: r r; r� r< r� r' r' r' r( �AfterBodyPhase� s r� c s� e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z e jd� jfdefdefde fg�Ze e_e jde fg�Ze e_dS )z"getPhases.<locals>.InFramesetPhasec S s0 | j jd jdkr | jjd� n | jjs,t�d S )Nr r� zeof-in-framesetr� )r> r� r@ rJ r� rR r� )rB r' r' r( r� s z-getPhases.<locals>.InFramesetPhase.processEOFc S s | j jd� d S )Nzunexpected-char-in-frameset)rJ r� )rB r� r' r' r( r� " s z4getPhases.<locals>.InFramesetPhase.processCharactersc S s | j j|� d S )N)r> r� )rB r� r' r' r( ra % s z3getPhases.<locals>.InFramesetPhase.startTagFramesetc S s | j j|� | j jj� d S )N)r> r� r� r� )rB r� r' r' r( � startTagFrame( s z0getPhases.<locals>.InFramesetPhase.startTagFramec S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( �startTagNoframes, s z3getPhases.<locals>.InFramesetPhase.startTagNoframesc S s | j jdd|d i� d S )Nz unexpected-start-tag-in-framesetr@ )rJ r� )rB r� r' r' r( r8 / s z0getPhases.<locals>.InFramesetPhase.startTagOtherc S s\ | j jd jdkr | jjd� n | j jj� | jj rX| j jd jdkrX| jjd | j_d S )Nr r� z)unexpected-frameset-in-frameset-innerhtmlr� � afterFramesetr� r� ) r> r� r@ rJ r� r� rR rG r_ )rB r� r' r' r( �endTagFrameset3 s z1getPhases.<locals>.InFramesetPhase.endTagFramesetc S s | j jdd|d i� d S )Nz unexpected-end-tag-in-framesetr@ )rJ r� )rB r� r' r' r( r: ? s z.getPhases.<locals>.InFramesetPhase.endTagOtherr� r� r rR N)r9 r: r; r/ r� r� r� ra r� r� r8 r� r: r r; r� r� r< r� r' )r� r' r( �InFramesetPhase s$ r� c sv e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z e j d � j fdefg�Zee_e j d e fg�Ze e_dS )z%getPhases.<locals>.AfterFramesetPhasec S s d S )Nr' )rB r' r' r( r� T s z0getPhases.<locals>.AfterFramesetPhase.processEOFc S s | j jd� d S )Nz unexpected-char-after-frameset)rJ r� )rB r� r' r' r( r� X s z7getPhases.<locals>.AfterFramesetPhase.processCharactersc S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( r� [ s z6getPhases.<locals>.AfterFramesetPhase.startTagNoframesc S s | j jdd|d i� d S )Nz#unexpected-start-tag-after-framesetr@ )rJ r� )rB r� r' r' r( r8 ^ s z3getPhases.<locals>.AfterFramesetPhase.startTagOtherc S s | j jd | j _d S )N�afterAfterFrameset)rJ rG r_ )rB r� r' r' r( r� b s z0getPhases.<locals>.AfterFramesetPhase.endTagHtmlc S s | j jdd|d i� d S )Nz!unexpected-end-tag-after-framesetr@ )rJ r� )rB r� r' r' r( r: e s z1getPhases.<locals>.AfterFramesetPhase.endTagOtherr� rR N)r9 r: r; r/ r� r� r� r� r8 r� r: r r; r� r� r< r� r' )r� r' r( �AfterFramesetPhaseP s r� c @ s` e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z e j de fg�Ze e_dS )z&getPhases.<locals>.AfterAfterBodyPhasec S s d S )Nr' )rB r' r' r( r� w s z1getPhases.<locals>.AfterAfterBodyPhase.processEOFc S s | j j|| j j� d S )N)r> r� r� )rB r� r' r' r( r� z s z5getPhases.<locals>.AfterAfterBodyPhase.processCommentc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� } s z=getPhases.<locals>.AfterAfterBodyPhase.processSpaceCharactersc S s | j jd� | j jd | j _|S )Nzexpected-eof-but-got-charr� )rJ r� rG r_ )rB r� r' r' r( r� � s z8getPhases.<locals>.AfterAfterBodyPhase.processCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z3getPhases.<locals>.AfterAfterBodyPhase.startTagHtmlc S s* | j jdd|d i� | j jd | j _|S )Nz expected-eof-but-got-start-tagr@ r� )rJ r� rG r_ )rB r� r' r' r( r8 � s z4getPhases.<locals>.AfterAfterBodyPhase.startTagOtherc S s* | j jdd|d i� | j jd | j _|S )Nz expected-eof-but-got-end-tagr@ r� )rJ r� rG r_ )rB r� r' r' r( r� � s z4getPhases.<locals>.AfterAfterBodyPhase.processEndTagr� N)r9 r: r; r/ r� r� r� r� r� r� r8 r� r r; r� r< r' r' r' r( �AfterAfterBodyPhaset s r� c @ sn e Zd Ze� Zdd� Zdd� Zdd� Zdd� Zd d � Z d d � Z d d� Z dd� Z e jde fde fg�Ze e_dS )z*getPhases.<locals>.AfterAfterFramesetPhasec S s d S )Nr' )rB r' r' r( r� � s z5getPhases.<locals>.AfterAfterFramesetPhase.processEOFc S s | j j|| j j� d S )N)r> r� r� )rB r� r' r' r( r� � s z9getPhases.<locals>.AfterAfterFramesetPhase.processCommentc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s zAgetPhases.<locals>.AfterAfterFramesetPhase.processSpaceCharactersc S s | j jd� d S )Nzexpected-eof-but-got-char)rJ r� )rB r� r' r' r( r� � s z<getPhases.<locals>.AfterAfterFramesetPhase.processCharactersc S s | j jd j|�S )Nr� )rJ rG r� )rB r� r' r' r( r� � s z7getPhases.<locals>.AfterAfterFramesetPhase.startTagHtmlc S s | j jd j|�S )Nr6 )rJ rG r� )rB r� r' r' r( �startTagNoFrames� s z;getPhases.<locals>.AfterAfterFramesetPhase.startTagNoFramesc S s | j jdd|d i� d S )Nz expected-eof-but-got-start-tagr@ )rJ r� )rB r� r' r' r( r8 � s z8getPhases.<locals>.AfterAfterFramesetPhase.startTagOtherc S s | j jdd|d i� d S )Nz expected-eof-but-got-end-tagr@ )rJ r� )rB r� r' r' r( r� � s z8getPhases.<locals>.AfterAfterFramesetPhase.processEndTagr� rR N)r9 r: r; r/ r� r� r� r� r� r� r� r8 r� r r; r� r< r' r' r' r( �AfterAfterFramesetPhase� s r� )rV rU r� r6 rJ rN r� r� r� r r� r� r� r� r� r� r� r~ r� r� r� r� r� )r )rH rX r� r0 r4 r= r[ r_ rd r r# r, r0 r6 r: r> rE rF r� r� r� r� r� r� r' )rg r% r� r( rF � sp H %-f?B Y! /F@bYLc&r.8$%&rF c sF t | d �t � �@ }|rBt| d �� fdd�| d j� D ��| d< d S )Nry c 3 s" | ]\}}� j ||�|fV qd S )N)r� )r? �k�v)r� r' r( r� � s z$adjust_attributes.<locals>.<genexpr>)r r0 r, )r� r� �needs_adjustmentr' )r� r( r� � s r� ru Fc C s |d kr i }t | | ||d�S )N)r0 r@ ry r )r )r@ r0 rm r r' r' r( r2 � s r2 c @ s e Zd ZdZdS )rx zError in parsed documentN)r9 r: r; r� r' r' r' r( rx � s rx )r T)r) r T)ru NF),� __future__r r r Zpip._vendor.sixr r r. r� r r r Ztreebuilders.baser r � constantsr r r r r r r r r r r r r� r r r r r! r+ r= r r �memoizerF r� r2 � Exceptionrx r' r' r' r( �<module> sJ H 8 `