관리-도구
편집 파일: html5parser.cpython-36.opt-1.pyc
3 �Pf�� � @ sF d dl mZmZmZ d dlmZmZmZ d dlZyd dl m Z W n e k r` d dl m Z Y nX ddl mZ ddl mZ ddl mZ dd lmZ dd l mZ dd lmZmZmZmZmZmZmZm Z m Z m Z m 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/�Z0dS )$� )�absolute_import�division�unicode_literals)�with_metaclass�viewkeys�PY3N)� OrderedDict� )� _inputstream)� _tokenizer)� treebuilders)�Marker)�_utils)�spaceCharacters�asciiUpper2Lower�specialElements�headingElements� cdataElements�rcdataElements� tokenTypes� tagTokenTypes� namespaces� htmlIntegrationPointElements�"mathmlTextIntegrationPointElements�adjustForeignAttributes�adjustMathMLAttributes�adjustSVGAttributes�E�ReparseException�etreeTc K s$ t j|�}t||d�}|j| f|�S )z.Parse a string or file-like object into a tree)�namespaceHTMLElements)r �getTreeBuilder� HTMLParser�parse)�doc� treebuilderr �kwargs�tb�p� r) �!/usr/lib/python3.6/html5parser.pyr# s r# �divc K s, t j|�}t||d�}|j| fd|i|��S )N)r � container)r r! r" � parseFragment)r$ r, r% r r& r' r( r) r) r* r- & 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 ]$\}}t|tj�r&� |�}|||< q W tj| |||�S )N)�items� isinstance�types� FunctionType�type�__new__)�metaZ classname�basesZ classDictZ attributeNameZ attribute)�functionr) r* r3 . s z5method_decorator_metaclass.<locals>.Decorated.__new__N)�__name__� __module__� __qualname__r3 r) )r6 r) r* � Decorated- s r: )r2 )r6 r: r) )r6 r* �method_decorator_metaclass, s r; c @ s� e Zd ZdZd+dd�Zd,dd �Zd d � Zed d � �Zdd� Z dd� Z dd� Z dd� Z dd� Z dd� Zd-dd �Zd d � Zd d � Zd!d"� Zd#d$� Zd%d&� Zd'd(� Zd)d*� ZdS ).r" zZHTML parser. Generates a tree structure from a stream of (possibly malformed) HTMLNFTc sL |� _ |dkrtjd�}||�� _g � _t� fdd�t|�j� D ��� _dS )a strict - raise an exception when a parse error is encountered tree - a treebuilder class controlling the type of tree that will be returned. Built in treebuilders can be accessed through html5lib.treebuilders.getTreeBuilder(treeType) Nr c s g | ]\}}||� � j �f�qS r) )�tree)�.0�name�cls)�selfr) r* � <listcomp>M s z'HTMLParser.__init__.<locals>.<listcomp>) �strictr r! r< �errors�dict� getPhasesr. �phases)r@ r< rB r �debugr) )r@ 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 Z HTMLTokenizer� tokenizer�reset�mainLoopr )r@ �stream� innerHTMLr, rK r& r) r) r* �_parseP 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< rM � firstStartTagrC �log� compatModerJ r, �lowerrP r rL � rcdataState�stater � rawtextState�plaintextStaterF �phase�insertHtmlElement�resetInsertionModeZ lastPhaseZbeforeRCDataPhase� framesetOK)r@ r) r) r* rM ^ s* zHTMLParser.resetc C s t | d�sdS | jjjd jS )z�The name of the character encoding that was used to decode the input stream, or :obj:`None` if that is not determined yet. rL Nr )�hasattrrL rO � charEncodingr> )r@ 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)rf rg )r> � namespacer � attributes� translater r )r@ �elementr) r) r* �isHTMLIntegrationPoint� s z!HTMLParser.isHTMLIntegrationPointc C s |j |jftkS )N)rh r> r )r@ rk r) r) r* � isMathMLTextIntegrationPoint� s z'HTMLParser.isMathMLTextIntegrationPointc C sj t d }t d }t d }t d }t d }t d }t d }�x�| j� D �]�}d } |} �x�| d k �r | } | jjrx| jjd nd } | r�| jnd } | r�| jnd } | d }||kr�| j| d | jd i �� d } qVt| jj�d k�sl| | jj k�sl| j | ��r ||k�r|d t ddg�k�sl|||fk�sl| t d k�rP| dk�rP||k�rP|d dk�sl| j | ��rt||||fk�rt| j}n | jd }||k�r�|j| �} qV||k�r�|j| �} qV||k�r�|j| �} qV||k�r�|j| �} qV||k�r�|j| �} qV||krV|j| �} qVW ||krD| d rD| d rD| jdd | d i� qDW d}g }x(|�rd|j| j� | jj� }|�r>�q>W d S )N� CharactersZSpaceCharacters�StartTag�EndTag�CommentZDoctype� ParseErrorr r2 �data�datavarsr r> ZmglyphZ malignmarkrd zannotation-xml�svg�inForeignContent� selfClosing�selfClosingAcknowledgedz&non-void-element-with-trailing-solidusT���)r �normalizedTokensr< � openElementsrh r> � parseError�get�len�defaultNamespacerm � frozensetr rl r] rF �processCharacters�processSpaceCharacters�processStartTag� processEndTag�processComment�processDoctype�append� processEOF)r@ ZCharactersTokenZSpaceCharactersTokenZ StartTagTokenZ EndTagTokenZ CommentTokenZ DoctypeTokenZParseErrorToken�tokenZ prev_token� new_token� currentNodeZcurrentNodeNamespaceZcurrentNodeNamer2 r] Z reprocessrF r) r) r* rN � sp zHTMLParser.mainLoopc c s x| j D ]}| j|�V qW d S )N)rL �normalizeToken)r@ r� r) r) r* rz � s zHTMLParser.normalizedTokensc O s | j |ddf|�|� | jj� S )a� Parse a HTML document into a well-formed tree stream - a filelike 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) scripting - treat noscript elements as if javascript was turned on FN)rQ r< Z getDocument)r@ rO �argsr& r) r) r* r# � s zHTMLParser.parsec O s | j |df|�|� | jj� S )a2 Parse a HTML fragment into a well-formed tree fragment container - name of the element we're setting the innerHTML property if set to None, default to 'div' stream - a filelike 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) scripting - treat noscript elements as if javascript was turned on T)rQ r< Z getFragment)r@ rO 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)rC r� rL rO ZpositionrB rr r )r@ � errorcodert r) r) r* r| s zHTMLParser.parseErrorc C sT |d t d krP|d }t|�|d< t|�t|d �krP|d j|ddd� � |S )z3 HTML5 specific normalizations to the token stream r2 ro rs Nr ry )r r r~ �update)r@ r� �rawr) r) r* r� s zHTMLParser.normalizeTokenc C s t |t� d S )N)�adjust_attributesr )r@ r� r) r) r* r s z!HTMLParser.adjustMathMLAttributesc C s t |t� d S )N)r� r )r@ r� r) r) r* r s z HTMLParser.adjustSVGAttributesc C s t |t� d S )N)r� �adjustForeignAttributesMap)r@ r� r) r) r* r s z"HTMLParser.adjustForeignAttributesc C s | j j� d S )N)rI r] )r@ r� r) r) r* �reparseTokenNormal s z HTMLParser.reparseTokenNormalc C s� d}ddddddddddd d d d d �}x�| j jd d d� D ]p}|j}d }|| j jd krbd}| j}|dkrj| 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� ry )r� r� r� r� )r< r{ r> rP rh r rF r] )r@ ZlastZnewModes�nodeZnodeNameZ new_phaser) r) r* r_ ! s> z HTMLParser.resetInsertionModec C sF | j j|� |dkr"| jj| j_n | jj| j_| j| _| jd | _dS )zYGeneric RCDATA/RAWTEXT Parsing algorithm contentType - RCDATA or RAWTEXT �RAWTEXT�textN) r< � insertElementrL r[ rZ rY r] � originalPhaserF )r@ r� Z contentTyper) r) r* �parseRCDataRawtextM s z HTMLParser.parseRCDataRawtext)NFTF)Fr+ F)r� N)r7 r8 r9 �__doc__rH rQ rM �propertyrc rl rm rN rz r# r- r| r� r r r r� r_ r� r) r) r) r* r" 8 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� tj� D ���� �fdd�}|S )z4Logger that records which phase processes each tokenc s s | ]\}}||fV qd S )Nr) )r= �key�valuer) r) r* � <genexpr>c s z)getPhases.<locals>.log.<locals>.<genexpr>c s� � j jd�r�t|�dkr�|d }yd�|d i}W n � Y nX |d tkr\|d |d< | jjj| jjjj | jj j j | j j � j |f� � | f|�|�S � | f|�|�S d S )NZprocessr r2 r> ) r7 � startswithr~ r rI rV r� rL rZ r] � __class__)r@ r� r& r� �info)r6 � type_namesr) r* �wrappedf s z'getPhases.<locals>.log.<locals>.wrapped)rD r r. )r6 r� r) )r6 r� r* rV a s zgetPhases.<locals>.logc S s | r t |�S tS d S )N)r; r2 )Z use_metaclassZmetaclass_funcr) r) r* � getMetaclassz s z getPhases.<locals>.getMetaclassc @ sX e Zd ZdZdd� Zdd� Zdd� Zdd � Zd d � Zd d � Z dd� Z dd� Z dd� Z dS )zgetPhases.<locals>.PhasezNBase class for helper object that implements each phase of processing c S s || _ || _d S )N)rI r< )r@ rI r< r) r) r* rH � s z!getPhases.<locals>.Phase.__init__c S s t �d S )N)�NotImplementedError)r@ r) r) r* r� � s z#getPhases.<locals>.Phase.processEOFc S s | j j|| j jd � d S )Nr ry )r< � insertCommentr{ )r@ r� r) r) r* r� � s z'getPhases.<locals>.Phase.processCommentc S s | j jd� d S )Nzunexpected-doctype)rI r| )r@ r� r) r) r* r� � s z'getPhases.<locals>.Phase.processDoctypec S s | j j|d � d S )Nrs )r< � insertText)r@ r� r) r) r* r� � s z*getPhases.<locals>.Phase.processCharactersc S s | j j|d � d S )Nrs )r< r� )r@ r� r) r) r* r� � s z/getPhases.<locals>.Phase.processSpaceCharactersc S s | j |d |�S )Nr> )�startTagHandler)r@ r� r) 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-rootrs r F)rI rU r| r. r< r{ ri )r@ r� �attrr� r) r) r* � startTagHtml� s z%getPhases.<locals>.Phase.startTagHtmlc S s | j |d |�S )Nr> )� endTagHandler)r@ r� r) r) r* r� � s z&getPhases.<locals>.Phase.processEndTagN) r7 r8 r9 r� rH 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.<locals>.InitialPhasec S s d S )Nr) )r@ r� r) r) r* r� � s z6getPhases.<locals>.InitialPhase.processSpaceCharactersc S s | j j|| j j� d S )N)r< r� �document)r@ 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 quirksrS )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 ) rI r| r< Z insertDoctyperj r r� rX rW rF r] )r@ 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 rS )rI rW rF r] )r@ r) r) r* � anythingElse s z,getPhases.<locals>.InitialPhase.anythingElsec S s | j jd� | j� |S )Nz expected-doctype-but-got-chars)rI r| r )r@ 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> )rI r| r )r@ r� r) r) r* r� s z/getPhases.<locals>.InitialPhase.processStartTagc S s" | j jdd|d i� | j� |S )Nz expected-doctype-but-got-end-tagr> )rI r| r )r@ r� r) r) r* r� s z-getPhases.<locals>.InitialPhase.processEndTagc S s | j jd� | j� dS )Nz expected-doctype-but-got-eofT)rI r| r )r@ r) r) r* r� % s z*getPhases.<locals>.InitialPhase.processEOFN) r7 r8 r9 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.<locals>.BeforeHtmlPhasec S s& | j jtdd�� | jjd | j_d S )Nr� ro r� )r< Z insertRoot�impliedTagTokenrI rF r] )r@ r) r) r* r^ , s z4getPhases.<locals>.BeforeHtmlPhase.insertHtmlElementc S s | j � dS )NT)r^ )r@ r) r) r* r� 1 s z-getPhases.<locals>.BeforeHtmlPhase.processEOFc S s | j j|| j j� d S )N)r< r� r� )r@ r� r) r) r* r� 5 s z1getPhases.<locals>.BeforeHtmlPhase.processCommentc S s d S )Nr) )r@ r� r) r) r* r� 8 s z9getPhases.<locals>.BeforeHtmlPhase.processSpaceCharactersc S s | j � |S )N)r^ )r@ r� r) r) r* r� ; s z4getPhases.<locals>.BeforeHtmlPhase.processCharactersc S s |d dkrd| j _| j� |S )Nr> r� T)rI rU r^ )r@ 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� r )rI r| r^ )r@ r� r) r) r* r� E s z0getPhases.<locals>.BeforeHtmlPhase.processEndTagN) r7 r8 r9 r^ r� r� r� r� r� r� r) r) r) r* �BeforeHtmlPhase* 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.<locals>.BeforeHeadPhasec sV � j | ||� tjd| jfd| jfg�| _| j| j_tjd| jfg�| _ | j | j _d S )Nr� r� r� r )r� r� r� r ) rH r �MethodDispatcherr� � startTagHeadr� � startTagOther�default�endTagImplyHeadr� � endTagOther)r@ rI r< )r� r) r* rH N s z+getPhases.<locals>.BeforeHeadPhase.__init__c S s | j tdd�� dS )Nr� ro T)r r )r@ r) r) r* r� \ s z-getPhases.<locals>.BeforeHeadPhase.processEOFc S s d S )Nr) )r@ r� r) r) r* r� ` s z9getPhases.<locals>.BeforeHeadPhase.processSpaceCharactersc S s | j tdd�� |S )Nr� ro )r r )r@ r� r) r) r* r� c s z4getPhases.<locals>.BeforeHeadPhase.processCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� g s z/getPhases.<locals>.BeforeHeadPhase.startTagHtmlc S s0 | j j|� | j jd | j _| jjd | j_d S )Nr �inHeadry )r< r� r{ � headPointerrI rF r] )r@ r� r) r) r* r j s z/getPhases.<locals>.BeforeHeadPhase.startTagHeadc S s | j tdd�� |S )Nr� ro )r r )r@ r� r) r) r* r o s z0getPhases.<locals>.BeforeHeadPhase.startTagOtherc S s | j tdd�� |S )Nr� ro )r r )r@ r� r) r) r* r s s z2getPhases.<locals>.BeforeHeadPhase.endTagImplyHeadc S s | j jdd|d i� d S )Nzend-tag-after-implied-rootr> )rI r| )r@ r� r) r) r* r w s z.getPhases.<locals>.BeforeHeadPhase.endTagOtherN) r7 r8 r9 rH r� r� r� r� r r r r r) )r� r) r* �BeforeHeadPhaseM 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.<locals>.InHeadPhasec s� � j | ||� tjd| jfd| jfd| jfd| jfd| jfd| jfd | j fd | j fg�| _ | j | j _ tjd | jfd| jfg�| _| j| j_ d S )Nr� �title�noframes�style�noscript�script�base�basefont�bgsound�command�linkr4 r� r r� )r r )r r r r r )r r� r� )rH r r r� � startTagTitle�startTagNoFramesStyle�startTagNoscript�startTagScript�startTagBaseLinkCommand� startTagMetar r� r r � endTagHead�endTagHtmlBodyBrr� r )r@ rI r< )r� r) r* rH | s z'getPhases.<locals>.InHeadPhase.__init__c S s | j � dS )NT)r )r@ r) r) r* r� � s z)getPhases.<locals>.InHeadPhase.processEOFc S s | j � |S )N)r )r@ r� r) r) r* r� � s z0getPhases.<locals>.InHeadPhase.processCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ 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)rI r| )r@ r� r) r) r* r � s z+getPhases.<locals>.InHeadPhase.startTagHeadc S s$ | j j|� | j jj� d|d< d S )NTrx )r< r� r{ �pop)r@ r� r) r) r* r% � 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 ) NTrx rs r Z tentative�charsetZcontentz http-equivz content-typezutf-8)r< r� r{ r) rI rL rO rb ZchangeEncodingrX r Z EncodingBytes�encodeZContentAttrParserr# )r@ r� ri rs rI �codecr) r) r* r& � s z+getPhases.<locals>.InHeadPhase.startTagMetac S s | j j|d� d S )NZRCDATA)rI r� )r@ r� r) r) r* r! � s z,getPhases.<locals>.InHeadPhase.startTagTitlec S s | j j|d� d S )Nr� )rI r� )r@ r� r) r) r* r" � s z4getPhases.<locals>.InHeadPhase.startTagNoFramesStylec S s8 | j jr| j j|d� n | jj|� | j jd | j _d S )Nr� �inHeadNoscript)rI rK r� r< r� rF r] )r@ r� r) r) r* r# � s z/getPhases.<locals>.InHeadPhase.startTagNoscriptc S s< | j j|� | jjj| jj_| jj| j_| jjd | j_d S )Nr� ) r< r� rI rL ZscriptDataStaterZ r] r� rF )r@ r� r) r) r* r$ � s z-getPhases.<locals>.InHeadPhase.startTagScriptc S s | j � |S )N)r )r@ r� r) r) r* r � s z,getPhases.<locals>.InHeadPhase.startTagOtherc S s" | j jjj� }| j jd | j _d S )N� afterHead)rI r< r{ r) rF r] )r@ r� r� r) r) r* r' � s z)getPhases.<locals>.InHeadPhase.endTagHeadc S s | j � |S )N)r )r@ r� r) r) r* r( � s z/getPhases.<locals>.InHeadPhase.endTagHtmlBodyBrc S s | j jdd|d i� d S )Nzunexpected-end-tagr> )rI r| )r@ r� r) r) r* r � s z*getPhases.<locals>.InHeadPhase.endTagOtherc S s | j td�� d S )Nr� )r' r )r@ r) r) r* r � s z+getPhases.<locals>.InHeadPhase.anythingElseN)r7 r8 r9 rH 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.<locals>.InHeadNoscriptPhasec sf � j | ||� tjd| jfd | jfd | jfg�| _| j| j_tjd | j fd | j fg�| _ | j | j _d S ) Nr� r r r r4 r r r� r r )r r r r4 r r )r� r ) rH r r r� r% �startTagHeadNoscriptr� r r �endTagNoscript�endTagBrr� r )r@ rI r< )r� r) r* rH � s z/getPhases.<locals>.InHeadNoscriptPhase.__init__c S s | j jd� | j� dS )Nzeof-in-head-noscriptT)rI r| r )r@ r) r) r* r� � s z1getPhases.<locals>.InHeadNoscriptPhase.processEOFc S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r� � s z5getPhases.<locals>.InHeadNoscriptPhase.processCommentc S s | j jd� | j� |S )Nzchar-in-head-noscript)rI r| r )r@ r� r) r) r* r� � s z8getPhases.<locals>.InHeadNoscriptPhase.processCharactersc S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r� s z=getPhases.<locals>.InHeadNoscriptPhase.processSpaceCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� s z3getPhases.<locals>.InHeadNoscriptPhase.startTagHtmlc S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r% s z>getPhases.<locals>.InHeadNoscriptPhase.startTagBaseLinkCommandc S s | j jdd|d i� d S )Nzunexpected-start-tagr> )rI r| )r@ r� r) r) r* r0 s z;getPhases.<locals>.InHeadNoscriptPhase.startTagHeadNoscriptc S s" | j jdd|d i� | j� |S )Nz unexpected-inhead-noscript-tagr> )rI r| r )r@ r� r) r) r* r s z4getPhases.<locals>.InHeadNoscriptPhase.startTagOtherc S s" | j jjj� }| j jd | j _d S )Nr )rI r< r{ r) rF r] )r@ r� r� r) r) r* r1 s z5getPhases.<locals>.InHeadNoscriptPhase.endTagNoscriptc S s" | j jdd|d i� | j� |S )Nz unexpected-inhead-noscript-tagr> )rI r| r )r@ r� r) r) r* r2 s z/getPhases.<locals>.InHeadNoscriptPhase.endTagBrc S s | j jdd|d i� d S )Nzunexpected-end-tagr> )rI r| )r@ r� r) r) r* r s z2getPhases.<locals>.InHeadNoscriptPhase.endTagOtherc S s | j td�� d S )Nr )r1 r )r@ r) r) r* r s z3getPhases.<locals>.InHeadNoscriptPhase.anythingElseN)r7 r8 r9 rH r� r� r� r� r� r% r0 r r1 r2 r r r) )r� r) r* �InHeadNoscriptPhase� s r3 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.<locals>.AfterHeadPhasec sn � j | ||� tjd| jfd| jfd| jfd| jfd | jfg�| _| j | j_ tjd| j fg�| _ | j | j _ d S )Nr� r� r� r r r r r4 r r r r r� r ) r r r r r4 r r r r )r� r� r )rH r r r� � startTagBody�startTagFrameset�startTagFromHeadr r� r r r( r� r )r@ rI r< )r� r) r* rH # s z*getPhases.<locals>.AfterHeadPhase.__init__c S s | j � dS )NT)r )r@ r) r) r* r� 4 s z,getPhases.<locals>.AfterHeadPhase.processEOFc S s | j � |S )N)r )r@ r� r) r) r* r� 8 s z3getPhases.<locals>.AfterHeadPhase.processCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� < s z.getPhases.<locals>.AfterHeadPhase.startTagHtmlc S s( d| j _| jj|� | j jd | j _d S )NFr� )rI r` r< r� rF r] )r@ r� r) r) r* r4 ? s z.getPhases.<locals>.AfterHeadPhase.startTagBodyc S s | j j|� | jjd | j_d S )Nr� )r< r� rI rF r] )r@ r� r) r) r* r5 D 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> r r r� ry ) rI r| r< r{ r� r rF r� r> �remove)r@ r� r� r) r) r* r6 H s z2getPhases.<locals>.AfterHeadPhase.startTagFromHeadc S s | j jdd|d i� d S )Nzunexpected-start-tagr> )rI r| )r@ r� r) r) r* r R s z.getPhases.<locals>.AfterHeadPhase.startTagHeadc S s | j � |S )N)r )r@ r� r) r) r* r U s z/getPhases.<locals>.AfterHeadPhase.startTagOtherc S s | j � |S )N)r )r@ r� r) r) r* r( Y s z2getPhases.<locals>.AfterHeadPhase.endTagHtmlBodyBrc S s | j jdd|d i� d S )Nzunexpected-end-tagr> )rI r| )r@ r� r) r) r* r ] s z-getPhases.<locals>.AfterHeadPhase.endTagOtherc S s. | j jtdd�� | jjd | j_d| j_d S )Nr� ro r� T)r< r� r rI rF r] r` )r@ r) r) r* r ` s z.getPhases.<locals>.AfterHeadPhase.anythingElseN)r7 r8 r9 rH r� r� r� r4 r5 r6 r r r( r r r) )r� r) r* �AfterHeadPhase" s r8 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.<locals>.InBodyPhasec , s� � j | ||� | j| _tjd| jfdd| jfd | jfd | jfde| j ft | j fdf| j fd&| j fdg| jfd*| jfd+| jfdh| jfd8| jfd9| jfdi| jfd=| jfd>| jfdj| jfdk| jfdH| jfdI| jfdJ| jfdK| j fdL| j fdM| j fdN| j fdl| j fdQ| j!fdm| j"fdn| j#fdV| j$fdW| j%fdo| j&fg!�| _'| j(| j'_)tjd | j*fd| j+fdp| j,fd&| j-fd | j.fdq| j/ft | j0fdr| j1fds| j2fd@| j3fg �| _4| j5| j4_)d S )tNr� r r r r r r4 r r r r� r� �address�article�aside� blockquote�center�details�dirr+ �dl�fieldset� figcaption�figure�footer�header�hgroup�main�menu�nav�olr( �section�summary�ul�pre�listing�form�li�dd�dtrR �a�b�big�code�em�font�i�s�small�strike�strong�tt�u�nobr�button�applet�marquee�objectZxmpr� �arear �embed�img�keygen�wbr�param�source�track�input�hr�image�isindex�textareaZiframer �noembedr r� �rp�rt�option�optgroupZmathru r� �colr� �framer� r� r� r� r� r� r� �dialog) r r r r r r4 r r r )r9 r: r; r< r= r> r? r+ r@ rA rB rC rD rE rF rG rH rI rJ r( rK rL rM )rN rO )rQ rR rS ) rU rV rW rX rY rZ r[ r\ r] r^ r_ r` )rc rd re )rf r rg rh ri rj )rk rl rm )rs r )rt ru )rv rw ) r� rx r� ry r� r� r� r� r� r� r� )r9 r: r; r< rb r= r> rz r? r+ r@ rA rB rC rD rE rF rO rG rH rI rJ rN rK rL rM )rR rS rQ )rT rU rV rW rX rY rZ ra r[ r\ r] r^ r_ r` )rc rd re )6rH � processSpaceCharactersNonPrer� r r r� �startTagProcessInHeadr4 r5 �startTagClosePr �startTagHeading�startTagPreListing� startTagForm�startTagListItem�startTagPlaintext� startTagA�startTagFormatting� startTagNobr�startTagButton�startTagAppletMarqueeObject� startTagXmp� startTagTable�startTagVoidFormatting�startTagParamSource� startTagInput� startTagHr� startTagImage�startTagIsIndex�startTagTextarea�startTagIFramer# �startTagRawtext�startTagSelect� startTagRpRt� startTagOpt� startTagMath� startTagSvg�startTagMisplacedr� r r � endTagBody� endTagHtml� endTagBlock� endTagForm�endTagP�endTagListItem� endTagHeading�endTagFormatting�endTagAppletMarqueeObjectr2 r� r )r@ rI r< )r� r) r* rH h s~ z'getPhases.<locals>.InBodyPhase.__init__c S s$ |j |j ko"|j|jko"|j|jkS )N)r> rh ri )r@ Znode1Znode2r) 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|�dkrx| j jj|d � | j jj|� d S )Nr � ry ry ry ) r< r� r{ �activeFormattingElementsr r� r� r~ r7 )r@ r� rk ZmatchingElementsr� 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 )NrR rS rQ r( r� r� r� r� r� r� r� r� r z expected-closing-tag-but-got-eof) rR rS rQ r( r� r� r� r� r� r� r� r� ry )r� r< r{ r> rI r| )r@ Zallowed_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 ) Nrs � r rN rO rr ry )rN rO rr ry ) r{ r� r� r< r{ r> Z hasContent�#reconstructActiveFormattingElementsr� )r@ r� rs 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 )Nrs � c S s g | ] }|t k�qS r) )r )r= �charr) r) r* rA � s zDgetPhases.<locals>.InBodyPhase.processCharacters.<locals>.<listcomp>F)r< r� r� rI r` �any)r@ r� r) r) r* r� � s z0getPhases.<locals>.InBodyPhase.processCharactersc S s | j j� | j j|d � d S )Nrs )r< r� r� )r@ r� r) r) r* r{ � s z;getPhases.<locals>.InBodyPhase.processSpaceCharactersNonPrec S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r| � s z4getPhases.<locals>.InBodyPhase.startTagProcessInHeadc S s� | j jdddi� t| jj�dks|| jjd jdkr6nFd| j _x<|d j� D ],\}}|| jjd jkrL|| jjd j|< qLW d S )Nzunexpected-start-tagr> r� r Frs ) rI r| r~ r< r{ r> r` r. ri )r@ r� r� r� r) r) r* r4 � s z+getPhases.<locals>.InBodyPhase.startTagBodyc S s� | j jdddi� t| jj�dks�| jjd jdkr6nt| j js@nj| jjd jrj| jjd jj| jjd � x"| jjd jdkr�| jjj � qlW | jj |� | j j d | j _ d S ) Nzunexpected-start-tagr> r� r r� r� r� ry ) rI r| r~ r< r{ r> r` �parent� removeChildr) r� rF r] )r@ r� r) r) r* r5 � s " z/getPhases.<locals>.InBodyPhase.startTagFramesetc S s. | j jddd�r | jtd�� | j j|� d S )Nr( rb )�variant)r< �elementInScoper� r r� )r@ r� r) r) r* r} s z-getPhases.<locals>.InBodyPhase.startTagClosePc S s> | j jddd�r | jtd�� | j j|� d| j_| j| _d S )Nr( rb )r� F) r< r� r� r r� rI r` r� r� )r@ r� r) r) r* r 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> rP r( rb )r� r ry ) r< � formPointerrI r| r� r� r r� r{ )r@ r� r) r) r* r� 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 rS rR )rQ rS rR r> rp r9 r+ r( rb )r� )r9 r+ r( ) rI r` �reversedr< r{ r> r] r� r � nameTupler r� r� )r@ r� Z stopNamesMapZ stopNamesr� r) r) r* r� s" z/getPhases.<locals>.InBodyPhase.startTagListItemc S s> | j jddd�r | jtd�� | j j|� | jjj| jj_d S )Nr( rb )r� ) r< r� r� r r� rI rL r\ rZ )r@ r� r) r) r* r� 4 s z0getPhases.<locals>.InBodyPhase.startTagPlaintextc S sb | j jddd�r | jtd�� | j jd jtkrR| jjdd|d i� | j jj � | j j |� d S )Nr( rb )r� r zunexpected-start-tagr> ry ) r< r� r� r r{ r> r rI r| r) r� )r@ r� r) r) r* r~ : 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 )NrT z$unexpected-start-tag-implies-end-tag)� startName�endName) r< �!elementInActiveFormattingElementsrI r| r� r r{ r7 r� r� r� )r@ r� Z afeAElementr) r) r* r� B s z(getPhases.<locals>.InBodyPhase.startTagAc S s | j j� | j|� d S )N)r< r� r� )r@ r� r) r) r* r� O s z1getPhases.<locals>.InBodyPhase.startTagFormattingc S sP | j j� | j jd�rB| jjdddd�� | jtd�� | j j� | j|� d S )Nra z$unexpected-start-tag-implies-end-tag)r� r� )r< r� r� rI r| r� r r� )r@ r� r) r) r* r� S s z+getPhases.<locals>.InBodyPhase.startTagNobrc S sT | j jd�r2| jjdddd�� | jtd�� |S | j j� | j j|� d| j_d S )Nrb z$unexpected-start-tag-implies-end-tag)r� r� F) r< r� rI r| r� r r� r� r` )r@ r� r) r) r* r� ] s z-getPhases.<locals>.InBodyPhase.startTagButtonc S s0 | j j� | j j|� | j jjt� d| j_d S )NF)r< r� r� r� r� r rI r` )r@ r� r) r) r* r� h s z:getPhases.<locals>.InBodyPhase.startTagAppletMarqueeObjectc S sB | j jddd�r | jtd�� | j j� d| j_| jj|d� d S )Nr( rb )r� Fr� )r< r� r� r r� rI r` r� )r@ r� r) r) r* r� n s z*getPhases.<locals>.InBodyPhase.startTagXmpc S sR | j jdkr*| jjddd�r*| jtd�� | jj|� d| j _| j jd | j _ d S )Nr r( rb )r� Fr� ) rI rW r< r� r� r r� r` rF r] )r@ r� r) r) r* r� u s z,getPhases.<locals>.InBodyPhase.startTagTablec S s6 | j j� | j j|� | j jj� d|d< d| j_d S )NTrx F)r< r� r� r{ r) rI r` )r@ r� r) r) r* r� } s z5getPhases.<locals>.InBodyPhase.startTagVoidFormattingc S s@ | j j}| j|� d|d kr<|d d jt�dkr<|| j _d S )Nr2 rs �hidden)rI r` r� rj r )r@ r� r` r) r) r* r� � s z,getPhases.<locals>.InBodyPhase.startTagInputc S s$ | j j|� | j jj� d|d< d S )NTrx )r< r� r{ r) )r@ r� r) r) r* r� � s z2getPhases.<locals>.InBodyPhase.startTagParamSourcec S sJ | j jddd�r | jtd�� | j j|� | j jj� d|d< d| j_d S )Nr( rb )r� Trx F) r< r� r� r r� r{ r) rI r` )r@ r� r) r) r* r� � s z)getPhases.<locals>.InBodyPhase.startTagHrc S s6 | j jdddd�� | jtdd|d |d d�� d S ) Nz unexpected-start-tag-treated-asrp rh )� originalName�newNamero rs rw )ri rw )rI r| r� r )r@ r� r) r) r* r� � 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> rq �actionrs rP ro )ri ro Zlabel�promptz3This is a searchable index. Enter search keywords: rn )r2 rs rn rw )ri rw ) rI r| r< r� r� r r� r �copyr� )r@ r� Z form_attrsr� ri r) r) r* r� � s6 z.getPhases.<locals>.InBodyPhase.startTagIsIndexc S s0 | j j|� | jjj| jj_| j| _d| j_d S )NF) r< r� rI rL rY rZ r� r� r` )r@ r� r) r) r* r� � s z/getPhases.<locals>.InBodyPhase.startTagTextareac S s d| j _| j|� d S )NF)rI r` r� )r@ r� r) r) r* r� � s z-getPhases.<locals>.InBodyPhase.startTagIFramec S s" | j jr| j|� n | j|� d S )N)rI rK r� r )r@ r� r) r) r* r# � s z/getPhases.<locals>.InBodyPhase.startTagNoscriptc S s | j j|d� dS )z8iframe, noembed noframes, noscript(if scripting enabled)r� N)rI r� )r@ r� r) r) r* r� � s z.getPhases.<locals>.InBodyPhase.startTagRawtextc S s@ | j jd jdkr$| jjjtd�� | j j� | jj j|� d S )Nr rv ry ) r< r{ r> rI r] r� r r� r� )r@ r� r) r) r* r� � s z*getPhases.<locals>.InBodyPhase.startTagOptc S s� | j j� | j j|� d| j_| jj| jjd | jjd | jjd | jjd | jjd | jjd fkrx| jjd | j_n| jjd | j_d S ) NFr� r� r� r� r� r� �inSelectInTabler� )r< r� r� rI r` r] rF )r@ r� r) r) r* r� � s z-getPhases.<locals>.InBodyPhase.startTagSelectc S sB | j jd�r2| j j� | j jd jdkr2| jj� | j j|� d S )N�rubyr ry )r< r� �generateImpliedEndTagsr{ r> rI r| r� )r@ r� r) r) r* r� � s z+getPhases.<locals>.InBodyPhase.startTagRpRtc S sZ | j j� | jj|� | jj|� td |d< | j j|� |d rV| j jj� d|d< d S )Nrd rh rw Trx ) r< r� rI r r r r� r{ r) )r@ r� r) r) r* r� � s z+getPhases.<locals>.InBodyPhase.startTagMathc S sZ | j j� | jj|� | jj|� td |d< | j j|� |d rV| j jj� d|d< d S )Nru rh rw Trx ) r< r� rI r r r r� r{ r) )r@ r� r) r) r* r� � s z*getPhases.<locals>.InBodyPhase.startTagSvgc S s | j jdd|d i� dS )a5 Elements that should be children of other elements that have a different insertion mode; here they are ignored "caption", "col", "colgroup", "frame", "frameset", "head", "option", "optgroup", "tbody", "td", "tfoot", "th", "thead", "tr", "noscript" z unexpected-start-tag-ignoredr> N)rI r| )r@ r� r) r) r* r� s z0getPhases.<locals>.InBodyPhase.startTagMisplacedc S s | j j� | j j|� d S )N)r< r� r� )r@ r� r) r) r* r s z,getPhases.<locals>.InBodyPhase.startTagOtherc S s� | j jddd�sD| jtdd�� | jjdddi� | jtdd�� nX| j jd� | j jd j dkrt| jjdddi� | j jj � }x|j dkr�| j jj � }q�W d S ) Nr( rb )r� ro zunexpected-end-tagr> rp r ry ) r< r� r} r rI r| r� r� r{ r> r) )r@ r� r� r) r) r* r� s z&getPhases.<locals>.InBodyPhase.endTagPc S s� | j jd�s| jj� d S | j jd jdkrlx>| j jdd � D ]*}|jtd�kr>| jjdd|jd�� P q>W | jjd | j_d S )Nr� r � rR rS rQ rw rv r( rt ru r� r� r� r� r� r� r� z$expected-one-end-tag-but-got-another)�gotName� expectedName� afterBodyry )rR rS rQ rw rv r( rt ru r� r� r� r� r� r� r� r� ) r< r� rI r| r{ r> r� rF r] )r@ r� r� r) r) r* r� ! s z)getPhases.<locals>.InBodyPhase.endTagBodyc S s" | j jd�r | jtd�� |S d S )Nr� )r< r� r� r )r@ r� r) r) r* r� 3 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> rN r zend-tag-too-earlyry ) r{ r� r< r� r� r{ r> rI r| r) )r@ r� ZinScoper� r) r) r* r� 9 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> rP r zend-tag-too-early-ignoredry )r< r� r� rI r| r� r{ r7 )r@ r� r� r) r) r* r� G 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-earlyry )r< r� rI r| r� r{ r> r) )r@ r� r� r� r) r) r* r� T 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-earlyry ) r r< r� r� r{ r> rI r| r) )r@ r� �itemr) r) r* r� e 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-1.3r� r� r� r� r� r� ry )r� r� r� r� r� )r< r� r{ r� r> r rI r| r� r7 �indexr� r r) Z cloneNoder� r� Z appendChildr� Z getTableMisnestedNodePosition� insertBeforeZreparentChildren�insert)r@ r� ZouterLoopCounterZformattingElementZafeIndexZ furthestBlockrk ZcommonAncestorZbookmarkZlastNoder� ZinnerLoopCounterr� Zcloner� r� r) r) r* r� t 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-earlyry ) r< r� r� r{ r> rI r| r) � clearActiveFormattingElements)r@ r� rk r) r) r* r� s z8getPhases.<locals>.InBodyPhase.endTagAppletMarqueeObjectc S s@ | j jdddd�� | jj� | jjtdd�� | jjj� d S )Nz unexpected-end-tag-treated-asr z br element)r� r� ro )rI r| r< r� r� r r{ r) )r@ r� r) r) r* r2 # 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-tagry ry ) r< r{ r> r� rI r| r) r� r )r@ r� r� r) r) r* r * s z*getPhases.<locals>.InBodyPhase.endTagOtherN)6r7 r8 r9 rH r� r� r� r� r� r{ r| r4 r5 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� r2 r r) )r� r) r* � InBodyPhasee sh G $ r� c s@ e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d � Zd d � Zd S )z getPhases.<locals>.TextPhasec sF � j | ||� tjg �| _| j| j_tjd| jfg�| _| j| j_d S )Nr ) rH r r r� r r � endTagScriptr� r )r@ rI r< )r� r) r* rH 9 s z%getPhases.<locals>.TextPhase.__init__c S s | j j|d � d S )Nrs )r< r� )r@ r� r) r) r* r� A 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 Try )rI r| r< r{ r> r) r� r] )r@ r) r) r* r� D s z'getPhases.<locals>.TextPhase.processEOFc S s d S )Nr) )r@ r� r) r) r* r K s z*getPhases.<locals>.TextPhase.startTagOtherc S s | j jj� }| jj| j_d S )N)r< r{ r) rI r� r] )r@ r� r� r) r) r* r� N s z)getPhases.<locals>.TextPhase.endTagScriptc S s | j jj� | jj| j_d S )N)r< r{ r) rI r� r] )r@ r� r) r) r* r U s z(getPhases.<locals>.TextPhase.endTagOtherN) r7 r8 r9 rH r� r� r r� r r) )r� r) r* � TextPhase8 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$� Zd%d&� Zd'S )(z getPhases.<locals>.InTablePhasec s� � j | ||� tjd| jfd| jfd| jfd| jfd| jfd| jfd | j fd| j fd| j fd| j fg �| _ | j| j _tjd | jfd| jfg�| _| j| j_d S )Nr� r� r� rx r� r� r� r� r� r� r� r r rn rP r� )r� r� r� )r� r� r� )r r ) r� r� rx r� r� r� r� r� r� r� r� )rH r r r� �startTagCaption�startTagColgroup� startTagCol�startTagRowGroup�startTagImplyTbodyr� �startTagStyleScriptr� r� r� r r � endTagTable� endTagIgnorer� r )r@ rI r< )r� r) r* rH [ s$ z(getPhases.<locals>.InTablePhase.__init__c S s( x"| j jd jdkr"| j jj� qW d S )Nr r� r� ry )r� r� )r< r{ r> r) )r@ r) r) r* �clearStackToTableContexts s z8getPhases.<locals>.InTablePhase.clearStackToTableContextc S s$ | j jd jdkr | jjd� n d S )Nr r� z eof-in-tablery )r< r{ r> rI r| )r@ 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)rI r] rF r� r� )r@ 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� )rI r] rF r� r� )r@ r� r� r) r) r* r� � s z1getPhases.<locals>.InTablePhase.processCharactersc S s& d| j _| jjd j|� d| j _d S )NTr� F)r< �insertFromTablerI rF r� )r@ r� r) r) r* r� � s z*getPhases.<locals>.InTablePhase.insertTextc S s6 | j � | jjjt� | jj|� | jjd | j_d S )Nr� ) r� r< r� r� r r� rI rF r] )r@ r� r) r) r* r� � s z/getPhases.<locals>.InTablePhase.startTagCaptionc S s( | j � | jj|� | jjd | j_d S )Nr� )r� r< r� rI rF r] )r@ r� r) r) r* r� � s z0getPhases.<locals>.InTablePhase.startTagColgroupc S s | j tdd�� |S )Nr� ro )r� r )r@ r� r) r) r* r� � s z+getPhases.<locals>.InTablePhase.startTagColc S s( | j � | jj|� | jjd | j_d S )Nr� )r� r< r� rI rF r] )r@ r� r) r) r* r� � s z0getPhases.<locals>.InTablePhase.startTagRowGroupc S s | j tdd�� |S )Nr� ro )r� r )r@ r� r) r) r* r� � 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� )rI r| r] r� r rP )r@ r� r) r) r* r� � s z-getPhases.<locals>.InTablePhase.startTagTablec S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r� � s z3getPhases.<locals>.InTablePhase.startTagStyleScriptc S sV d|d krH|d d j t�dkrH| jjd� | jj|� | jjj� n | j|� d S )Nr2 rs r� z unexpected-hidden-input-in-table) rj r rI r| r< r� r{ r) r )r@ 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 ry )rI r| r< r� r� r{ r) )r@ 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)rI r| r< r� rF r� )r@ r� r) r) r* r � 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� d S ) Nr� )r� r zend-tag-too-early-named)r� r� ry ry ry ) r< r� r� r{ r> rI r| r) r_ )r@ r� r) r) r* r� � s z+getPhases.<locals>.InTablePhase.endTagTablec S s | j jdd|d i� d S )Nzunexpected-end-tagr> )rI r| )r@ r� r) r) r* r� � 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)rI r| r< r� rF r� )r@ r� r) r) r* r � s z+getPhases.<locals>.InTablePhase.endTagOtherN)r7 r8 r9 rH r� r� r� r� r� r� r� r� r� r� r� r� r� r� r r� r� r r) )r� r) r* � InTablePhaseY s& r� c sP e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d � Zd d � Zd d� Z dd� Z dS )z#getPhases.<locals>.InTableTextPhasec s � j | ||� d | _g | _d S )N)rH r� �characterTokens)r@ rI r< )r� r) r* rH � 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 )rs r) )r= r� r) r) r* rA � s zGgetPhases.<locals>.InTableTextPhase.flushCharacters.<locals>.<listcomp>c S s g | ] }|t k�qS r) )r )r= r� r) r) r* rA � s rn )r2 rs r� )�joinr� r� r rI rF r� r< )r@ rs r� r) r) r* �flushCharacters� s z3getPhases.<locals>.InTableTextPhase.flushCharactersc S s | j � | j| j_|S )N)r� r� rI r] )r@ r� r) r) r* r� � s z2getPhases.<locals>.InTableTextPhase.processCommentc S s | j � | j| j_dS )NT)r� r� rI r] )r@ r) r) r* r� � s z.getPhases.<locals>.InTableTextPhase.processEOFc S s |d dkrd S | j j|� d S )Nrs r� )r� r� )r@ r� r) r) r* r� s z5getPhases.<locals>.InTableTextPhase.processCharactersc S s | j j|� d S )N)r� r� )r@ r� r) r) r* r� s z:getPhases.<locals>.InTableTextPhase.processSpaceCharactersc S s | j � | j| j_|S )N)r� r� rI r] )r@ r� r) r) r* r� s z3getPhases.<locals>.InTableTextPhase.processStartTagc S s | j � | j| j_|S )N)r� r� rI r] )r@ r� r) r) r* r� s z1getPhases.<locals>.InTableTextPhase.processEndTagN) r7 r8 r9 rH 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.<locals>.InCaptionPhasec sf � j | ||� tjd| jfd | jfg�| _| j| j_tjd| jfd | j fd| j fg�| _ | j | j _d S )Nr� r� rx r� r� r� r� r� r� r� r� r� ) r� rx r� r� r� r� r� r� r� ) r� rx r� r� r� r� r� r� r� r� ) rH r r r� �startTagTableElementr� r r � endTagCaptionr� r� r� r )r@ rI r< )r� r) r* rH s z*getPhases.<locals>.InCaptionPhase.__init__c S s | j jddd� S )Nr� r� )r� )r< r� )r@ r) r) r* �ignoreEndTagCaption+ s z5getPhases.<locals>.InCaptionPhase.ignoreEndTagCaptionc S s | j jd j� d S )Nr� )rI rF r� )r@ r) r) r* r� . s z,getPhases.<locals>.InCaptionPhase.processEOFc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� 1 s z3getPhases.<locals>.InCaptionPhase.processCharactersc S s0 | j j� | j� }| j jjtd�� |s,|S d S )Nr� )rI r| r� r] r� r )r@ r� � ignoreEndTagr) r) r* r� 4 s z6getPhases.<locals>.InCaptionPhase.startTagTableElementc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r < 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� d S ) Nr r� z$expected-one-end-tag-but-got-another)r� r� r� ry ry ry ) r� r< r� r{ r> rI r| r) r� rF r] )r@ r� r) r) r* r� ? s z/getPhases.<locals>.InCaptionPhase.endTagCaptionc S s0 | j j� | j� }| j jjtd�� |s,|S d S )Nr� )rI r| r� r] r� r )r@ r� r� r) r) r* r� Q s z-getPhases.<locals>.InCaptionPhase.endTagTablec S s | j jdd|d i� d S )Nzunexpected-end-tagr> )rI r| )r@ r� r) r) r* r� X s z.getPhases.<locals>.InCaptionPhase.endTagIgnorec S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r [ s z-getPhases.<locals>.InCaptionPhase.endTagOtherN) r7 r8 r9 rH r� r� r� r� r r� r� r� r r) )r� r) r* �InCaptionPhase 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.<locals>.InColumnGroupPhasec s^ � j | ||� tjd| jfd| jfg�| _| j| j_tjd| jfd| j fg�| _ | j | j _d S )Nr� rx r� ) rH r r r� r� r� r r �endTagColgroup� endTagColr� r )r@ rI r< )r� r) r* rH a s z.getPhases.<locals>.InColumnGroupPhase.__init__c S s | j jd jdkS )Nr r� ry )r< r{ r> )r@ r) r) r* �ignoreEndTagColgroupp s z:getPhases.<locals>.InColumnGroupPhase.ignoreEndTagColgroupc S s8 | j jd jdkrd S | j� }| jtd�� |s4dS d S )Nr r� r� Try )r< r{ r> r� r� r )r@ r� r) r) r* r� s s z0getPhases.<locals>.InColumnGroupPhase.processEOFc S s" | j � }| jtd�� |s |S d S )Nr� )r� r� r )r@ r� r� r) r) r* r� } s z7getPhases.<locals>.InColumnGroupPhase.processCharactersc S s$ | j j|� | j jj� d|d< d S )NTrx )r< r� r{ r) )r@ r� r) r) r* r� � s z1getPhases.<locals>.InColumnGroupPhase.startTagColc S s" | j � }| jtd�� |s |S d S )Nr� )r� r� r )r@ r� r� r) r) r* r � s z3getPhases.<locals>.InColumnGroupPhase.startTagOtherc S s4 | j � r| jj� n | jjj� | jjd | j_d S )Nr� )r� rI r| r< r{ r) rF r] )r@ r� r) r) r* r� � s z4getPhases.<locals>.InColumnGroupPhase.endTagColgroupc S s | j jdddi� d S )Nz no-end-tagr> rx )rI r| )r@ r� r) r) r* r� � s z/getPhases.<locals>.InColumnGroupPhase.endTagColc S s" | j � }| jtd�� |s |S d S )Nr� )r� r� r )r@ r� r� r) r) r* r � s z1getPhases.<locals>.InColumnGroupPhase.endTagOtherN) r7 r8 r9 rH 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.<locals>.InTableBodyPhasec sv � j | ||� tjd| jfd| jfd | jfd| jfg�| _| j| j_ tjd| j fd | j fd| j fg�| _ | j| j _ d S )Nr� r� r� r� r� rx r� r� r� r� r� r� )r� r� )r� rx r� r� r� r� )r� r� r� )r� r� rx r� r� r� r� r� )rH r r r� � startTagTr�startTagTableCell�startTagTableOtherr� r r �endTagTableRowGroupr� r� r� r )r@ rI r< )r� r) r* rH � s z,getPhases.<locals>.InTableBodyPhase.__init__c S s: x"| j jd jdkr"| j jj� qW | j jd jdkr6d S ) Nr r� r� r� r� ry )r� r� r� r� ry )r< r{ r> r) )r@ r) r) r* � clearStackToTableBodyContext� s z@getPhases.<locals>.InTableBodyPhase.clearStackToTableBodyContextc S s | j jd j� d S )Nr� )rI rF r� )r@ r) r) r* r� � s z.getPhases.<locals>.InTableBodyPhase.processEOFc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� � s z:getPhases.<locals>.InTableBodyPhase.processSpaceCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� � s z5getPhases.<locals>.InTableBodyPhase.processCharactersc S s( | j � | jj|� | jjd | j_d S )Nr� )r� r< r� rI rF r] )r@ r� r) r) r* r� � s z.getPhases.<locals>.InTableBodyPhase.startTagTrc S s* | j jdd|d i� | jtdd�� |S )Nz unexpected-cell-in-table-bodyr> r� ro )rI r| r� r )r@ r� r) r) r* r� � s z5getPhases.<locals>.InTableBodyPhase.startTagTableCellc S sb | j jddd�s0| j jddd�s0| j jddd�rT| j� | jt| j jd j�� |S | jj� d S )Nr� r� )r� r� r� r ry ) r< r� r� r� r r{ r> rI r| )r@ r� r) r) r* r� � s z6getPhases.<locals>.InTableBodyPhase.startTagTableOtherc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r � 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� r� r{ r) rI rF r] r| )r@ r� r) r) r* r� � s z7getPhases.<locals>.InTableBodyPhase.endTagTableRowGroupc S sb | j jddd�s0| j jddd�s0| j jddd�rT| j� | jt| j jd j�� |S | jj� d S )Nr� r� )r� r� r� r ry ) r< r� r� r� r r{ r> rI r| )r@ 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> )rI r| )r@ r� r) r) r* r� � s z0getPhases.<locals>.InTableBodyPhase.endTagIgnorec S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r � s z/getPhases.<locals>.InTableBodyPhase.endTagOtherN)r7 r8 r9 rH 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.<locals>.InRowPhasec sv � j | ||� tjd| jfd | jfd| jfg�| _| j| j_tjd | j fd | j fd| j fd| j fg�| _ | j| j _d S )Nr� r� r� r� rx r� r� r� r� r� r� r� )r� r� )r� rx r� r� r� r� r� )r� r� r� )r� r� rx r� r� r� r� )rH r r r� r� r� r� r r �endTagTrr� r� r� r� r )r@ rI r< )r� r) r* rH s z&getPhases.<locals>.InRowPhase.__init__c 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> ry )r� r� ry )r< r{ r> rI r| r) )r@ r) r) r* �clearStackToTableRowContext s z9getPhases.<locals>.InRowPhase.clearStackToTableRowContextc S s | j jddd� S )Nr� r� )r� )r< r� )r@ r) r) r* �ignoreEndTagTr s z,getPhases.<locals>.InRowPhase.ignoreEndTagTrc S s | j jd j� d S )Nr� )rI rF r� )r@ r) r) r* r� " s z(getPhases.<locals>.InRowPhase.processEOFc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� % s z4getPhases.<locals>.InRowPhase.processSpaceCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� ( s z/getPhases.<locals>.InRowPhase.processCharactersc S s6 | j � | jj|� | jjd | j_| jjjt� d S )Nr� ) r� r< r� rI rF r] r� r� r )r@ r� r) r) r* r� + s z/getPhases.<locals>.InRowPhase.startTagTableCellc S s" | j � }| jtd�� |s |S d S )Nr� )r� r� r )r@ r� r� r) r) r* r� 1 s z0getPhases.<locals>.InRowPhase.startTagTableOtherc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r 8 s z+getPhases.<locals>.InRowPhase.startTagOtherc S s< | j � s.| j� | jjj� | jjd | j_n | jj� d S )Nr� ) r� r� r< r{ r) rI rF r] r| )r@ r� r) r) r* r� ; s z&getPhases.<locals>.InRowPhase.endTagTrc S s" | j � }| jtd�� |s |S d S )Nr� )r� r� r )r@ r� r� r) r) r* r� E s z)getPhases.<locals>.InRowPhase.endTagTablec S s4 | j j|d dd�r&| jtd�� |S | jj� d S )Nr> r� )r� r� )r< r� r� r rI r| )r@ r� r) r) r* r� M s z1getPhases.<locals>.InRowPhase.endTagTableRowGroupc S s | j jdd|d i� d S )Nz unexpected-end-tag-in-table-rowr> )rI r| )r@ r� r) r) r* r� T s z*getPhases.<locals>.InRowPhase.endTagIgnorec S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r X s z)getPhases.<locals>.InRowPhase.endTagOtherN)r7 r8 r9 rH 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.<locals>.InCellPhasec sf � j | ||� tjd| jfd | jfg�| _| j| j_tjd| jfd| j fd| j fg�| _ | j | j _d S )Nr� r� rx r� r� r� r� r� r� r� r� r� ) r� rx r� r� r� r� r� r� r� )r� r� )r� r� rx r� r� )r� r� r� r� r� ) rH r r r� r� r� r r �endTagTableCellr� � endTagImplyr� r )r@ rI r< )r� r) r* rH ] s z'getPhases.<locals>.InCellPhase.__init__c S sB | j jddd�r | jtd�� n | j jddd�r>| jtd�� d S )Nr� r� )r� r� )r< r� r� r )r@ r) r) r* � closeCelln s z(getPhases.<locals>.InCellPhase.closeCellc S s | j jd j� d S )Nr� )rI rF r� )r@ r) r) r* r� u s z)getPhases.<locals>.InCellPhase.processEOFc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� x s z0getPhases.<locals>.InCellPhase.processCharactersc S s: | j jddd�s | j jddd�r,| j� |S | jj� d S )Nr� r� )r� r� )r< r� r� rI r| )r@ r� r) r) r* r� { s z1getPhases.<locals>.InCellPhase.startTagTableOtherc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r � 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-tagry ) r< r� r� r{ r> rI r| r) r� rF r] )r@ r� r� r) r) r* r� � s z.getPhases.<locals>.InCellPhase.endTagTableCellc S s | j jdd|d i� d S )Nzunexpected-end-tagr> )rI r| )r@ 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� rI r| )r@ r� r) r) r* r� � s z*getPhases.<locals>.InCellPhase.endTagImplyc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r � s z*getPhases.<locals>.InCellPhase.endTagOtherN) r7 r8 r9 rH 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.<locals>.InSelectPhasec s� � j | ||� tjd| jfd| jfd| jfd| jfd | jfd| jfg�| _ | j | j _ tjd| j fd| j fd| jfg�| _| j| j_ d S ) Nr� rv rw r� rn ri rr r )rn ri rr )rH r r r� �startTagOption�startTagOptgroupr� r� r$ r� r r � endTagOption�endTagOptgroup� endTagSelectr� r )r@ rI r< )r� r) r* rH � s z)getPhases.<locals>.InSelectPhase.__init__c S s$ | j jd jdkr | jjd� n d S )Nr r� z eof-in-selectry )r< r{ r> rI r| )r@ r) r) r* r� � s z+getPhases.<locals>.InSelectPhase.processEOFc S s$ |d dkrd S | j j|d � d S )Nrs r� )r< r� )r@ r� r) r) r* r� � s z2getPhases.<locals>.InSelectPhase.processCharactersc S s. | j jd jdkr | j jj� | j j|� d S )Nr rv ry )r< r{ r> r) r� )r@ r� r) r) r* r� � s z/getPhases.<locals>.InSelectPhase.startTagOptionc S sL | j jd jdkr | j jj� | j jd jdkr<| j jj� | j j|� d S )Nr rv rw ry ry )r< r{ r> r) r� )r@ r� r) r) r* r� � s z1getPhases.<locals>.InSelectPhase.startTagOptgroupc S s | j jd� | jtd�� d S )Nzunexpected-select-in-selectr� )rI r| r� r )r@ r� r) r) r* r� � s z/getPhases.<locals>.InSelectPhase.startTagSelectc S s2 | j jd� | jjddd�r.| jtd�� |S d S )Nzunexpected-input-in-selectr� )r� )rI r| r< r� r� r )r@ r� r) r) r* r� � s z.getPhases.<locals>.InSelectPhase.startTagInputc S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r$ � s z/getPhases.<locals>.InSelectPhase.startTagScriptc S s | j jdd|d i� d S )Nz unexpected-start-tag-in-selectr> )rI r| )r@ r� r) r) r* r � s z.getPhases.<locals>.InSelectPhase.startTagOtherc S s6 | j jd jdkr | j jj� n| jjdddi� d S )Nr rv z unexpected-end-tag-in-selectr> ry )r< r{ r> r) rI r| )r@ r� r) r) r* r� � 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 rv r� rw z unexpected-end-tag-in-selectr> ry ���ry )r< r{ r> r) rI r| )r@ r� r) r) r* r� � s z/getPhases.<locals>.InSelectPhase.endTagOptgroupc S sR | j jddd�rD| j jj� }x|jdkr6| j jj� }q W | jj� n | jj� d S )Nr� )r� )r< r� r{ r) r> rI r_ r| )r@ r� r� r) r) r* r� � s z-getPhases.<locals>.InSelectPhase.endTagSelectc S s | j jdd|d i� d S )Nz unexpected-end-tag-in-selectr> )rI r| )r@ r� r) r) r* r s z,getPhases.<locals>.InSelectPhase.endTagOtherN)r7 r8 r9 rH r� r� r� r� r� r� r$ r r� r� r� r r) )r� r) r* � InSelectPhase� s r� c sH e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d � Zd d � Zd d� Z dS )z'getPhases.<locals>.InSelectInTablePhasec sN � j | ||� tjd | jfg�| _| j| j_tjd | jfg�| _| j | j_d S ) Nr� r� r� r� r� r� r� r� )r� r� r� r� r� r� r� r� )r� r� r� r� r� r� r� r� ) rH r r r� r� r r r� r� r )r@ rI r< )r� r) r* rH s z0getPhases.<locals>.InSelectInTablePhase.__init__c S s | j jd j� d S )Nr� )rI rF r� )r@ r) r) r* r� s z2getPhases.<locals>.InSelectInTablePhase.processEOFc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� s z9getPhases.<locals>.InSelectInTablePhase.processCharactersc S s( | j jdd|d i� | jtd�� |S )Nz5unexpected-table-element-start-tag-in-select-in-tabler> r� )rI r| r r )r@ r� r) r) r* r� ! s z5getPhases.<locals>.InSelectInTablePhase.startTagTablec S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r & 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� )rI r| r< r� r r )r@ r� r) r) r* r� ) s z3getPhases.<locals>.InSelectInTablePhase.endTagTablec S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r / s z3getPhases.<locals>.InSelectInTablePhase.endTagOtherN) r7 r8 r9 rH r� r� r� r r� r r) )r� r) r* �InSelectInTablePhase s r� c - s� e Zd Zeddddddddd d d d d ddddddddddddddd d d d d d!d"d#d$d%d&d'd(d)d*d+d,g,�Z� fd-d.�Zd/d0� Z� fd1d2�Zd3d4� Zd5d6� Z d7S )8z(getPhases.<locals>.InForeignContentPhaserU rV r< r� r r= rW rR r+ r@ rS rX rg Zh1Zh2Zh3Zh4Zh5Zh6r� ro rZ rh rQ rO rH r4 ra rJ r( rN r� r[ r\ �spanr^ r] �subZsupr� r_ r` rM �varc s � j | ||� d S )N)rH )r@ rI r< )r� r) r* rH < s z1getPhases.<locals>.InForeignContentPhase.__init__c % S sn ddddddddd d d d d ddddddddddddddd d d d d d!d"d#d$d%�$}|d& |krj||d& |d&< d S )'NZaltGlyphZ altGlyphDefZ altGlyphItemZ animateColorZ animateMotionZanimateTransformZclipPathZfeBlendZ feColorMatrixZfeComponentTransferZ feCompositeZfeConvolveMatrixZfeDiffuseLightingZfeDisplacementMapZfeDistantLightZfeFloodZfeFuncAZfeFuncBZfeFuncGZfeFuncRZfeGaussianBlurZfeImageZfeMergeZ feMergeNodeZ feMorphologyZfeOffsetZ fePointLightZfeSpecularLightingZ feSpotLightZfeTileZ feTurbulenceZ foreignObjectZglyphRefZlinearGradientZradialGradientZtextPath)$ZaltglyphZ altglyphdefZ altglyphitemZ animatecolorZ animatemotionZanimatetransformZclippathZfeblendZ fecolormatrixZfecomponenttransferZ fecompositeZfeconvolvematrixZfediffuselightingZfedisplacementmapZfedistantlightZfefloodZfefuncaZfefuncbZfefuncgZfefuncrZfegaussianblurZfeimageZfemergeZ femergenodeZ femorphologyZfeoffsetZ fepointlightZfespecularlightingZ fespotlightZfetileZ feturbulenceZ foreignobjectZglyphrefZlineargradientZradialgradientZtextpathr> r) )r@ 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 )Nrs r� u �c s s | ]}|t kV qd S )N)r )r= r� r) r) r* r� l s zMgetPhases.<locals>.InForeignContentPhase.processCharacters.<locals>.<genexpr>F)rI r` r� r� )r@ r� )r� r) r* r� h s z:getPhases.<locals>.InForeignContentPhase.processCharactersc S s6 | j jd }|d | jksD|d dkr�t|d j� �tdddg�@ r�| jjdd|d i� xR| j jd j| j jkr�| jj | j jd � r�| jj | j jd � r�| j jj � q\W |S |jt d kr�| jj |� n$|jt d kr�| j|� | jj|� | jj|� |j|d < | j j|� |d �r2| j jj � d |d< d S )Nr r> rY rs ZcolorZface�sizez*unexpected-html-element-in-foreign-contentrd ru rh rw Trx ry ry ry ry )r< r{ �breakoutElements�set�keysrI r| rh r rl rm r) r r r r r r� )r@ r� r� r) r) r* r� p 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�q�W d }P |d8 }| jj| }|j | jjkr�qHqH| jjj|�}P qHW |S )Nr r> zunexpected-end-tagr� ry )r~ r< r{ r> rj r rI r| r] rF r� r� r) rh r r� )r@ r� Z nodeIndexr� r� r) r) r* r� � s( z6getPhases.<locals>.InForeignContentPhase.processEndTagN) r7 r8 r9 r� r rH r r� r� r� r) )r� r) r* �InForeignContentPhase2 s ) r c sP e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d � Zd d � Zd d� Z dd� Z dS )z!getPhases.<locals>.AfterBodyPhasec sN � j | ||� tjd| jfg�| _| j| j_tjd| jfg�| _| j | j_d S )Nr� ) rH r r r� r� r r r� r� r )r@ rI r< )r� r) r* rH � s z*getPhases.<locals>.AfterBodyPhase.__init__c S s d S )Nr) )r@ r) r) r* r� � s z,getPhases.<locals>.AfterBodyPhase.processEOFc S s | j j|| j jd � d S )Nr )r< r� r{ )r@ r� r) r) r* r� � s z0getPhases.<locals>.AfterBodyPhase.processCommentc S s | j jd� | j jd | j _|S )Nzunexpected-char-after-bodyr� )rI r| rF r] )r@ r� r) r) r* r� � s z3getPhases.<locals>.AfterBodyPhase.processCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ 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� )rI r| rF r] )r@ r� r) r) r* r � 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)rI rP r| rF r] )r@ 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� )rI r| rF r] )r@ r� r) r) r* r � s z-getPhases.<locals>.AfterBodyPhase.endTagOtherN) r7 r8 r9 rH r� r� r� r� r r� r r) )r� r) r* �AfterBodyPhase� 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.<locals>.InFramesetPhasec sf � j | ||� tjd| jfd| jfd| jfd| jfg�| _| j| j_ tjd| j fg�| _ | j | j _ d S )Nr� r� ry r ) rH r r r� r5 � startTagFrame�startTagNoframesr� r r �endTagFramesetr� r )r@ rI r< )r� r) r* rH � s z+getPhases.<locals>.InFramesetPhase.__init__c S s$ | j jd jdkr | jjd� n d S )Nr r� zeof-in-framesetry )r< r{ r> rI r| )r@ r) r) r* r� � s z-getPhases.<locals>.InFramesetPhase.processEOFc S s | j jd� d S )Nzunexpected-char-in-frameset)rI r| )r@ r� r) r) r* r� � s z4getPhases.<locals>.InFramesetPhase.processCharactersc S s | j j|� d S )N)r< r� )r@ r� r) r) r* r5 � s z3getPhases.<locals>.InFramesetPhase.startTagFramesetc S s | j j|� | j jj� d S )N)r< r� r{ r) )r@ r� r) r) r* r � s z0getPhases.<locals>.InFramesetPhase.startTagFramec S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r � s z3getPhases.<locals>.InFramesetPhase.startTagNoframesc S s | j jdd|d i� d S )Nz unexpected-start-tag-in-framesetr> )rI r| )r@ r� r) r) r* r � 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� � afterFramesetry ry ) r< r{ r> rI r| r) rP rF r] )r@ r� r) r) r* r � s z1getPhases.<locals>.InFramesetPhase.endTagFramesetc S s | j jdd|d i� d S )Nz unexpected-end-tag-in-framesetr> )rI r| )r@ r� r) r) r* r s z.getPhases.<locals>.InFramesetPhase.endTagOtherN) r7 r8 r9 rH r� r� r5 r r r r r r) )r� r) r* �InFramesetPhase� s r c sH e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d � Zd d � Zd d� Z dS )z%getPhases.<locals>.AfterFramesetPhasec sV � j | ||� tjd| jfd| jfg�| _| j| j_tjd| jfg�| _ | j | j _d S )Nr� r ) rH r r r� r r� r r r� r� r )r@ rI r< )r� r) r* rH s z.getPhases.<locals>.AfterFramesetPhase.__init__c S s d S )Nr) )r@ r) r) r* r� s z0getPhases.<locals>.AfterFramesetPhase.processEOFc S s | j jd� d S )Nz unexpected-char-after-frameset)rI r| )r@ r� r) r) r* r� ! s z7getPhases.<locals>.AfterFramesetPhase.processCharactersc S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r $ s z6getPhases.<locals>.AfterFramesetPhase.startTagNoframesc S s | j jdd|d i� d S )Nz#unexpected-start-tag-after-framesetr> )rI r| )r@ r� r) r) r* r ' s z3getPhases.<locals>.AfterFramesetPhase.startTagOtherc S s | j jd | j _d S )N�afterAfterFrameset)rI rF r] )r@ r� r) r) r* r� + s z0getPhases.<locals>.AfterFramesetPhase.endTagHtmlc S s | j jdd|d i� d S )Nz!unexpected-end-tag-after-framesetr> )rI r| )r@ r� r) r) r* r . s z1getPhases.<locals>.AfterFramesetPhase.endTagOtherN) r7 r8 r9 rH r� r� r r r� r r) )r� r) r* �AfterFramesetPhase s r c sP e Zd Z� fdd�Zdd� Zdd� Zdd� Zd d � Zd d � Zd d� Z dd� Z dS )z&getPhases.<locals>.AfterAfterBodyPhasec s0 � j | ||� tjd| jfg�| _| j| j_d S )Nr� )rH r r r� r� r r )r@ rI r< )r� r) r* rH 3 s z/getPhases.<locals>.AfterAfterBodyPhase.__init__c S s d S )Nr) )r@ r) r) r* r� ; s z1getPhases.<locals>.AfterAfterBodyPhase.processEOFc S s | j j|| j j� d S )N)r< r� r� )r@ r� r) r) r* r� > s z5getPhases.<locals>.AfterAfterBodyPhase.processCommentc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� A s z=getPhases.<locals>.AfterAfterBodyPhase.processSpaceCharactersc S s | j jd� | j jd | j _|S )Nzexpected-eof-but-got-charr� )rI r| rF r] )r@ r� r) r) r* r� D s z8getPhases.<locals>.AfterAfterBodyPhase.processCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� I s z3getPhases.<locals>.AfterAfterBodyPhase.startTagHtmlc S s* | j jdd|d i� | j jd | j _|S )Nz expected-eof-but-got-start-tagr> r� )rI r| rF r] )r@ r� r) r) r* r L s z4getPhases.<locals>.AfterAfterBodyPhase.startTagOtherc S s* | j jdd|d i� | j jd | j _|S )Nz expected-eof-but-got-end-tagr> r� )rI r| rF r] )r@ r� r) r) r* r� R s z4getPhases.<locals>.AfterAfterBodyPhase.processEndTagN) r7 r8 r9 rH r� r� r� r� r� r r� r) )r� r) r* �AfterAfterBodyPhase2 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.<locals>.AfterAfterFramesetPhasec s8 � j | ||� tjd| jfd| jfg�| _| j| j_d S )Nr� r )rH r r r� �startTagNoFramesr� r r )r@ rI r< )r� r) r* rH Y s z3getPhases.<locals>.AfterAfterFramesetPhase.__init__c S s d S )Nr) )r@ r) r) r* r� b s z5getPhases.<locals>.AfterAfterFramesetPhase.processEOFc S s | j j|| j j� d S )N)r< r� r� )r@ r� r) r) r* r� e s z9getPhases.<locals>.AfterAfterFramesetPhase.processCommentc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� h s zAgetPhases.<locals>.AfterAfterFramesetPhase.processSpaceCharactersc S s | j jd� d S )Nzexpected-eof-but-got-char)rI r| )r@ r� r) r) r* r� k s z<getPhases.<locals>.AfterAfterFramesetPhase.processCharactersc S s | j jd j|�S )Nr� )rI rF r� )r@ r� r) r) r* r� n s z7getPhases.<locals>.AfterAfterFramesetPhase.startTagHtmlc S s | j jd j|�S )Nr )rI rF r� )r@ r� r) r) r* r q s z;getPhases.<locals>.AfterAfterFramesetPhase.startTagNoFramesc S s | j jdd|d i� d S )Nz expected-eof-but-got-start-tagr> )rI r| )r@ r� r) r) r* r t s z8getPhases.<locals>.AfterAfterFramesetPhase.startTagOtherc S s | j jdd|d i� d S )Nz expected-eof-but-got-end-tagr> )rI r| )r@ r� r) r) r* r� x s z8getPhases.<locals>.AfterAfterFramesetPhase.processEndTagN) r7 r8 r9 rH r� r� r� r� r� r r r� r) )r� r) r* �AfterAfterFramesetPhaseX s r )rT rS r� r r- r. r� r� r� r� r� r� r� r� r� r� r� rv r� r� r r r )r )rG rV r� r r r r/ r3 r8 r� r� r� r� r� r� r� r� r� r� r� r r r r r r r) )r� r* rE _ sp ) #.g@C X! -GBbYLd's/9%&&rE c s^ t s tjr t| d �t� �@ }nt| d �t� �@ }|rZt� fdd�| d j� D ��| d< d S )Nrs c 3 s" | ]\}}� j ||�|fV qd S )N)r} )r= �k�v)r r) r* r� � s z$adjust_attributes.<locals>.<genexpr>)r r ZPY27r r� r r. )r� r Zneeds_adjustmentr) )r r* r� � s r� rp Fc C s |d kr i }t | | ||d�S )N)r2 r> rs rw )r )r> r2 ri rw r) r) r* r � s r c @ s e Zd ZdZdS )rr zError in parsed documentN)r7 r8 r9 r� r) r) r) r* rr � s rr )r T)r+ r T)rp NF)1Z __future__r r r Zpip._vendor.sixr r r r0 � collectionsr � ImportErrorZpip._vendor.ordereddictr� r r r Ztreebuilders.baser r Z constantsr r r r r r r r r r r r r� r r r r r# r- r; re r" ZmemoizerE r� r � Exceptionrr r) r) r) r* �<module> sR H ) L