� �Re%��@s�ddlmZmZmZddlmZddlZddlmZddlm Z ddlm Z ddlmZdd lm Z dd lmZddlmZmZddlmZdd lmZddlmZmZddlmZmZmZddlmZmZddlmZddddd�Zdddddd�Zdd�ZGdd�de�Z dd�Z!ddd d!d"�Z"Gd#d$�d$e#�Z$dS)%�)�absolute_import�division�unicode_literals)�with_metaclassN�)�inputstream)� tokenizer)�treebuilders)�Marker)�utils)� constants)�spaceCharacters�asciiUpper2Lower)�specialElements)�headingElements)� cdataElements�rcdataElements)� tokenTypes�ReparseException� namespaces)�htmlIntegrationPointElements�"mathmlTextIntegrationPointElements)�adjustForeignAttributes�etreeTcCs4tj|�}t|d|�}|j|d|�S)z.Parse a string or file-like object into a tree�namespaceHTMLElements�encoding)r �getTreeBuilder� HTMLParser�parse)�doc�treebuilderrr�tb�p�r#�;/tmp/pip-ztkk0jow-build/pip/_vendor/html5lib/html5parser.pyrsr�divcCs:tj|�}t|d|�}|j|d|d|�S)Nr� containerr)r rr� parseFragment)rr&r rrr!r"r#r#r$r'sr'cs G�fdd�dt�}|S)Ncs"eZdZ�fdd�ZdS)z-method_decorator_metaclass..Decoratedcs^xE|j�D]7\}}t|tj�r:�|�}n|||.Decorated.__new__N)�__name__� __module__�__qualname__r-r#)r4r#r$� Decorated'sr8)r,)r4r8r#)r4r$�method_decorator_metaclass&sr9c@seZdZdZdejddddd�Zddddddd �Zd d�Zdd �Z dd�Z dd�Zdd�Zddddd�Z dddddd�Zdidd�Zdd�Zdd�Zdd �Zd!d"�Zd#d$�Zd%d&�Zd'd(�ZdS))rzZHTML parser. Generates a tree structure from a stream of (possibly malformed) HTMLNFTcsz|�_|dkr'tjd�}n||��_|�_g�_t�fdd�t|�j�D��_ dS)a� strict - raise an exception when a parse error is encountered tree - a treebuilder class controlling the type of tree that will be returned. Built in treebuilders can be accessed through html5lib.treebuilders.getTreeBuilder(treeType) tokenizer - a class that provides a stream of tokens to the treebuilder. This may be replaced for e.g. a sanitizer which converts some tags to text Nrcs.g|]$\}}||��j�f�qSr#)�tree)�.0�name�cls)�selfr#r$� Ms z'HTMLParser.__init__..) �strictr rr:�tokenizer_class�errors�dict� getPhasesr(�phases)r>r:rr@r�debugr#)r>r$�__init__6s zHTMLParser.__init__r%cKs�||_||_|j|d|d|d|d||�|_|j�x2y|j�PWqLtk ry|j�YqLXqLWdS)Nr� parseMeta� useChardet�parser)� innerHTMLModer&rAr�reset�mainLoopr)r>�stream� innerHTMLr&rrHrI�kwargsr#r#r$�_parsePs zHTMLParser._parsecCs|jj�d|_g|_g|_d|_|jr�|jj�|_ |j t krp|jj|j_ nH|j tkr�|jj|j_ n$|j dkr�|jj|j_ n|jd|_|jj�|j�nd|_ |jd|_d|_d|_d|_dS)NFz no quirks� plaintext� beforeHtml�initialT)r:rL� firstStartTagrB�log� compatModerKr&�lowerrOrr�rcdataState�stater�rawtextState�plaintextStaterE�phase�insertHtmlElement�resetInsertionMode� lastPhase�beforeRCDataPhase� framesetOK)r>r#r#r$rLbs* zHTMLParser.resetcCse|jdkrK|jtdkrKd|jkoJ|jdjt�dkS|j|jftkSdS)Nzannotation-xml�mathmlr� text/html�application/xhtml+xml)rdre)r<� namespacer� attributes� translaterr)r>�elementr#r#r$�isHTMLIntegrationPoint�s z!HTMLParser.isHTMLIntegrationPointcCs|j|jftkS)N)rfr<r)r>rir#r#r$�isMathMLTextIntegrationPoint�sz'HTMLParser.isMathMLTextIntegrationPointcCs*td}td}td}td}td}td}td}x�|j�D]y}|} x+| dk r�|jjr�|jjdnd} | r�| jnd}| r�| jnd}| d } | |kr�|j| d | jdi��d} qbt|jj�dks�||jj ks�|j | �rl| |krZ|d tddg�ks�| ||fks�|tdkr�|dkr�|d dks�|j | �r�| |||fkr�|j}n |jd}| |kr�|j| �} qb| |kr|j| �} qb| |kr/|j| �} qb| |krM|j| �} qb| |krk|j| �} qb| |krb|j| �} qbqbW| |krS|drS|drS|jdi|d d 6�qSqSWd}g}xG|r%|j|j�|jj�}|r�|j|ks"t�q�q�WdS)N� Characters�SpaceCharacters�StartTag�EndTag�Comment�Doctype� ParseErrorrr,�data�datavarsrr<�mglyph� malignmarkrczannotation-xml�svg�inForeignContent�selfClosing�selfClosingAcknowledgedz&non-void-element-with-trailing-solidusT��)r�normalizedTokensr:�openElementsrfr<� parseError�get�len�defaultNamespacerk� frozensetrrjr]rE�processCharacters�processSpaceCharacters�processStartTag� processEndTag�processComment�processDoctype�append� processEOF�AssertionError)r>�CharactersToken�SpaceCharactersToken� StartTagToken�EndTagToken�CommentToken�DoctypeToken�ParseErrorToken�token� new_token�currentNode�currentNodeNamespace�currentNodeNamer,r]� reprocessrEr#r#r$rM�sj " zHTMLParser.mainLoopccs&x|jD]}|j|�Vq WdS)N)r�normalizeToken)r>r�r#r#r$r|�szHTMLParser.normalizedTokensc Cs2|j|ddd|d|d|�|jj�S)afParse a HTML document into a well-formed tree stream - a filelike object or string containing the HTML to be parsed The optional encoding parameter must be a string that indicates the encoding. If specified, that encoding will be used, regardless of any BOM or later declaration (such as in a meta element) rOFrrHrI)rQr:�getDocument)r>rNrrHrIr#r#r$r�s zHTMLParser.parsecCs)|j|dd|d|�|jj�S)a�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) Tr&r)rQr:�getFragment)r>rNr&rrHrIr#r#r$r'�szHTMLParser.parseFragmentzXXX-undefined-errorcCs;|jj|jjj�||f�|jr7t�ndS)N)rBr�rrN�positionr@rr)r>� errorcodertr#r#r$r~�s% zHTMLParser.parseErrorcCs<|dtdkr8t|dddd��|dr�r#r#r$r��s$zHTMLParser.normalizeTokencCsbidd6}xN|j�D]@\}}||dkr|d||d|<|d|=qqWdS)N� definitionURL� definitionurlrs)r()r>r��replacements�k�vr#r#r$�adjustMathMLAttributess z!HTMLParser.adjustMathMLAttributescCsi>dd6dd6dd6dd6d d 6dd6d d6dd6dd6dd6dd6dd6dd6dd6dd6dd 6d!d"6d#d$6d%d&6d'd(6d)d*6d+d,6d-d.6d/d06d1d26d3d46d5d66d7d86d9d:6d;d<6d=d>6d?d@6dAdB6dCdD6dEdF6dGdH6dIdJ6dKdL6dMdN6dOdP6dQdR6dSdT6dUdV6dWdX6dYdZ6d[d\6d]d^6d_d`6dadb6dcdd6dedf6dgdh6didj6dkdl6dmdn6dodp6dqdr6dsdt6dudv6dwdx6dydz6d{d|6}xXt|d}j��D]@}||kr�||}|d}||d}|<|d}|=q�q�WdS)~Nr2� attributename� attributeType� attributetype� baseFrequency� basefrequency�baseProfile�baseprofile�calcMode�calcmode� clipPathUnits� clippathunits�contentScriptType�contentscripttype�contentStyleType�contentstyletype�diffuseConstant�diffuseconstant�edgeMode�edgemode�externalResourcesRequired�externalresourcesrequired� filterRes� filterres�filterUnits�filterunits�glyphRef�glyphref�gradientTransform�gradienttransform� gradientUnits� gradientunits�kernelMatrix�kernelmatrix�kernelUnitLength�kernelunitlength� keyPoints� keypoints� keySplines� keysplines�keyTimes�keytimes�lengthAdjust�lengthadjust�limitingConeAngle�limitingconeangle�markerHeight�markerheight�markerUnits�markerunits�markerWidth�markerwidth�maskContentUnits�maskcontentunits� maskUnits� maskunits� numOctaves� numoctaves� pathLength� pathlength�patternContentUnits�patterncontentunits�patternTransform�patterntransform�patternUnits�patternunits� pointsAtX� pointsatx� pointsAtY� pointsaty� pointsAtZ� pointsatz� preserveAlpha� preservealpha�preserveAspectRatio�preserveaspectratio�primitiveUnits�primitiveunits�refX�refx�refY�refy�repeatCount�repeatcount� repeatDur� repeatdur�requiredExtensions�requiredextensions�requiredFeatures�requiredfeatures�specularConstant�specularconstant�specularExponent�specularexponent�spreadMethod�spreadmethod�startOffset�startoffset�stdDeviation�stddeviation�stitchTiles�stitchtiles�surfaceScale�surfacescale�systemLanguage�systemlanguage�tableValues�tablevalues�targetX�targetx�targetY�targety� textLength� textlength�viewBox�viewbox� viewTarget� viewtarget�xChannelSelector�xchannelselector�yChannelSelector�ychannelselector� zoomAndPan� zoomandpanrs)�list�keys)r>r�r��originalName�svgNamer#r#r$�adjustSVGAttributes s� zHTMLParser.adjustSVGAttributescCs_t}xR|dj�D]@}||kr||}|d||d|<|d|=qqWdS)Nrs)�adjustForeignAttributesMapr )r>r�r�r!�foreignNamer#r#r$rPs z"HTMLParser.adjustForeignAttributescCs|jj�dS)N)rJr])r>r�r#r#r$�reparseTokenNormalYszHTMLParser.reparseTokenNormalcCs]d}idd6dd6dd6dd6d d 6d d6d d6d d6dd6dd6dd6dd6dd6dd6}x�|jjddd�D]�}|j}d}||jjdkr�|js�t�d}|j}n|dkr�|js�t�n|r|j|jjkrq�n||kr5|j||}Pq�|r�|jd}Pq�q�W||_dS)NF�inSelect�select�inCell�td�th�inRow�tr�inTableBody�tbody�thead�tfoot� inCaption�caption� inColumnGroup�colgroup�inTable�table�inBody�head�body� inFrameset�frameset� beforeHead�htmlrrTr{)zselectzcolgroupzheadzhtml) r:r}r<rOr�rfr�rEr])r>�last�newModes�node�nodeName� new_phaser#r#r$r_\sD zHTMLParser.resetInsertionModecCsu|dkst�|jj|�|dkrC|jj|j_n|jj|j_|j|_|j d|_dS)zYGeneric RCDATA/RAWTEXT Parsing algorithm contentType - RCDATA or RAWTEXT �RAWTEXT�RCDATA�textN)zRAWTEXTzRCDATA) r�r:� insertElementrr[rZrYr]� originalPhaserE)r>r��contentTyper#r#r$�parseRCDataRawtext�szHTMLParser.parseRCDataRawtext)r5r6r7�__doc__r� HTMLTokenizerrGrQrLrjrkrMr|rr'r~r�r�r#rr&r_rJr#r#r#r$r2s* " @G ,rcs7dd�}dd�}Gdd�dt|||��Gdd�d��}Gd d �d ��}G�fdd�d��}G�fd d�d��}G�fdd�d��}G�fdd�d��}G�fdd�d��} G�fdd�d��} G�fdd�d��}G�fdd�d��}G�fdd�d��} G�fdd�d��}G�fdd �d ��}G�fd!d"�d"��}G�fd#d$�d$��}G�fd%d&�d&��}G�fd'd(�d(��}G�fd)d*�d*��}G�fd+d,�d,��}G�fd-d.�d.��}G�fd/d0�d0��}G�fd1d2�d2��}i|d36|d46|d56|d66|d76|d86| d96| d:6|d;6|d<6| d=6|d>6|d?6|d@6|dA6|dB6|dC6|dD6|dE6|dF6|dG6|dH6S)INcs;tdd�tjj�D��fdd�}|S)z4Logger that records which phase processes each tokencss!|]\}}||fVqdS)Nr#)r;�key�valuer#r#r$� �sz)getPhases..log..cs��jjd�r�t|�dkr�|d}yi�|dd6}Wn�YnX|dtjkry|d|d�argsrPr��info)r4� type_namesr#r$�wrapped�s$ z'getPhases..log..wrapped)rCrrr()r4rWr#)r4rVr$rV�szgetPhases..logcSs|rt|�StSdS)N)r9r,)� use_metaclass�metaclass_funcr#r#r$�getMetaclass�s zgetPhases..getMetaclassc@s�eZdZdZdd�Zdd�Zdd�Zdd �Zd d�Zdd �Z dd�Z dd�Zdd�ZdS)zgetPhases..PhasezNBase class for helper object that implements each phase of processing cSs||_||_dS)N)rJr:)r>rJr:r#r#r$rG�s z!getPhases..Phase.__init__cSs t�dS)N)�NotImplementedError)r>r#r#r$r��sz#getPhases..Phase.processEOFcSs!|jj||jjd�dS)Nrr{)r:� insertCommentr})r>r�r#r#r$r��sz'getPhases..Phase.processCommentcSs|jjd�dS)Nzunexpected-doctype)rJr~)r>r�r#r#r$r��sz'getPhases..Phase.processDoctypecSs|jj|d�dS)Nrs)r:� insertText)r>r�r#r#r$r��sz*getPhases..Phase.processCharacterscSs|jj|d�dS)Nrs)r:r])r>r�r#r#r$r��sz/getPhases..Phase.processSpaceCharacterscSs|j|d|�S)Nr<)�startTagHandler)r>r�r#r#r$r��sz(getPhases..Phase.processStartTagcSs�|jjr0|ddkr0|jjd�nxQ|dj�D]?\}}||jjdjkrA||jjdj|z non-html-rootrsrF)rJrUr~r(r:r}rg)r>r��attrrNr#r#r$�startTagHtml�sz%getPhases..Phase.startTagHtmlcSs|j|d|�S)Nr<)� endTagHandler)r>r�r#r#r$r��sz&getPhases..Phase.processEndTagN) r5r6r7rKrGr�r�r�r�r�r�r`r�r#r#r#r$�Phase�s rbc@speZdZdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dS)zgetPhases..InitialPhasecSsdS)Nr#)r>r�r#r#r$r��sz6getPhases..InitialPhase.processSpaceCharacterscSs|jj||jj�dS)N)r:r\�document)r>r�r#r#r$r��sz.getPhases..InitialPhase.processCommentc8Ssu|d}|d}|d}|d}|dksX|dk sX|dk rk|dkrk|jjd�n|dkr�d}n|jj|�|dkr�|jt�}n|s|ddks|jdJ�s|dKks|jdL�r�|dks|r"|j�dDkr"dE|j_n9|jdM�sL|jdN�r[|dk r[dH|j_n|jj dI|j_ dS)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)7rhrirjrkrlrmrnrorprqrrrsrtrurvrwrxryrzr{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�zhtml)r�r�)r�r�)r�r�)rJr~r:� insertDoctyperhrrQrXrWrEr])r>r�r<rdrerfr#r#r$r��s� z.getPhases..InitialPhase.processDoctypecSs&d|j_|jjd|j_dS)Nr�rS)rJrWrEr])r>r#r#r$�anythingElseJsz,getPhases..InitialPhase.anythingElsecSs|jjd�|j�|S)Nzexpected-doctype-but-got-chars)rJr~r�)r>r�r#r#r$r�Ns z1getPhases..InitialPhase.processCharacterscSs,|jjdi|dd6�|j�|S)Nz"expected-doctype-but-got-start-tagr<)rJr~r�)r>r�r#r#r$r�Ss z/getPhases..InitialPhase.processStartTagcSs,|jjdi|dd6�|j�|S)Nz expected-doctype-but-got-end-tagr<)rJr~r�)r>r�r#r#r$r�Ys z-getPhases..InitialPhase.processEndTagcSs|jjd�|j�dS)Nzexpected-doctype-but-got-eofT)rJr~r�)r>r#r#r$r�_s z*getPhases..InitialPhase.processEOFN)r5r6r7r�r�r�r�r�r�r�r�r#r#r#r$�InitialPhase�s`r�c@sdeZdZdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dS)z"getPhases..BeforeHtmlPhasecSs3|jjtdd��|jjd|j_dS)Nr>rnr=)r:� insertRoot�impliedTagTokenrJrEr])r>r#r#r$r^fsz4getPhases..BeforeHtmlPhase.insertHtmlElementcSs|j�dS)NT)r^)r>r#r#r$r�ks z-getPhases..BeforeHtmlPhase.processEOFcSs|jj||jj�dS)N)r:r\rc)r>r�r#r#r$r�osz1getPhases..BeforeHtmlPhase.processCommentcSsdS)Nr#)r>r�r#r#r$r�rsz9getPhases..BeforeHtmlPhase.processSpaceCharacterscSs|j�|S)N)r^)r>r�r#r#r$r�us z4getPhases..BeforeHtmlPhase.processCharacterscSs-|ddkrd|j_n|j�|S)Nr<r>T)rJrUr^)r>r�r#r#r$r�ys z2getPhases..BeforeHtmlPhase.processStartTagcSsC|ddkr1|jjdi|dd6�n|j�|SdS)Nr<r9r:r>�brzunexpected-end-tag-before-html)zheadzbodyzhtmlzbr)rJr~r^)r>r�r#r#r$r�s z0getPhases..BeforeHtmlPhase.processEndTagN) r5r6r7r^r�r�r�r�r�r�r#r#r#r$�BeforeHtmlPhasedsr�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�ZdS)z"getPhases..BeforeHeadPhasecs}�j|||�tjd|jfd|jfg�|_|j|j_tjd|jfg�|_ |j |j _dS)Nr>r9r:r�)zheadzbodyzhtmlzbr)rGr�MethodDispatcherr`�startTagHeadr^� startTagOther�default�endTagImplyHeadra�endTagOther)r>rJr:)rbr#r$rG�sz+getPhases..BeforeHeadPhase.__init__cSs|jtdd��dS)Nr9rnT)r�r�)r>r#r#r$r��sz-getPhases..BeforeHeadPhase.processEOFcSsdS)Nr#)r>r�r#r#r$r��sz9getPhases..BeforeHeadPhase.processSpaceCharacterscSs|jtdd��|S)Nr9rn)r�r�)r>r�r#r#r$r��sz4getPhases..BeforeHeadPhase.processCharacterscSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r`�sz/getPhases..BeforeHeadPhase.startTagHtmlcSs@|jj|�|jjd|j_|jjd|j_dS)Nr�inHeadr{)r:rGr}�headPointerrJrEr])r>r�r#r#r$r��sz/getPhases..BeforeHeadPhase.startTagHeadcSs|jtdd��|S)Nr9rn)r�r�)r>r�r#r#r$r��sz0getPhases..BeforeHeadPhase.startTagOthercSs|jtdd��|S)Nr9rn)r�r�)r>r�r#r#r$r��sz2getPhases..BeforeHeadPhase.endTagImplyHeadcSs"|jjdi|dd6�dS)Nzend-tag-after-implied-rootr<)rJr~)r>r�r#r#r$r��sz.getPhases..BeforeHeadPhase.endTagOtherN)r5r6r7rGr�r�r�r`r�r�r�r�r#)rbr#r$�BeforeHeadPhase�sr�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�Zdd�Z dd�Zdd�Zdd�Zdd�ZdS) zgetPhases..InHeadPhasec s��j|||�tjd|jfd|jfd|jfd|jfd|jfd|jfd |j fg�|_ |j|j _tjd |j fd|jfg�|_|j|j_dS)Nr>�title�noscript�noframes�style�script�base�basefont�bgsound�command�linkr.r9r�r:)znoscriptznoframeszstyle)zbasezbasefontzbgsoundzcommandzlink)zbrzhtmlzbody)rGrr�r`� startTagTitle�startTagNoScriptNoFramesStyle�startTagScript�startTagBaseLinkCommand�startTagMetar�r^r�r�� endTagHead�endTagHtmlBodyBrrar�)r>rJr:)rbr#r$rG�s z'getPhases..InHeadPhase.__init__cSs|j�dS)NT)r�)r>r#r#r$r��s z)getPhases..InHeadPhase.processEOFcSs|j�|S)N)r�)r>r�r#r#r$r��s z0getPhases..InHeadPhase.processCharacterscSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r`�sz+getPhases..InHeadPhase.startTagHtmlcSs|jjd�dS)Nz!two-heads-are-not-better-than-one)rJr~)r>r�r#r#r$r��sz+getPhases..InHeadPhase.startTagHeadcSs.|jj|�|jjj�d|dr�r#r#r$r��sz6getPhases..InHeadPhase.startTagBaseLinkCommandcSs�|jj|�|jjj�d|d<|d}|jjjjddkr�d|kry|jjjj|d�q�d|kr�d|kr�|dj �d kr�t j|djd ��}t j |�}|j�}|jjjj|�q�ndS)NTrzrsr� tentative�charset�contentz http-equivzcontent-typezutf-8)r:rGr}r�rJrrN�charEncoding�changeEncodingrXr� EncodingBytes�encode�ContentAttrParserr)r>r�rgrsrJ�codecr#r#r$r��s z+getPhases..InHeadPhase.startTagMetacSs|jj|d�dS)NrE)rJrJ)r>r�r#r#r$r��sz,getPhases..InHeadPhase.startTagTitlecSs|jj|d�dS)NrD)rJrJ)r>r�r#r#r$r��sz.InHeadPhase.startTagNoScriptNoFramesStylecSsT|jj|�|jjj|jj_|jj|j_|jjd|j_dS)NrF) r:rGrJr�scriptDataStaterZr]rHrE)r>r�r#r#r$r��sz-getPhases..InHeadPhase.startTagScriptcSs|j�|S)N)r�)r>r�r#r#r$r�s z,getPhases..InHeadPhase.startTagOthercSsQ|jjjj�}|jdks7td|j��|jjd|j_dS)Nr9zExpected head got %s� afterHead)rJr:r}r�r<r�rEr])r>r�rAr#r#r$r�s"z)getPhases..InHeadPhase.endTagHeadcSs|j�|S)N)r�)r>r�r#r#r$r� s z/getPhases..InHeadPhase.endTagHtmlBodyBrcSs"|jjdi|dd6�dS)Nzunexpected-end-tagr<)rJr~)r>r�r#r#r$r�sz*getPhases..InHeadPhase.endTagOthercSs|jtd��dS)Nr9)r�r�)r>r#r#r$r�sz+getPhases..InHeadPhase.anythingElseN)r5r6r7rGr�r�r`r�r�r�r�r�r�r�r�r�r�r�r#)rbr#r$�InHeadPhase�sr�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�Zdd�Z dd�ZdS)z!getPhases..AfterHeadPhasec s��j|||�tjd|jfd|jfd|jfd|jfd |jfg�|_|j |j_ tjd|jfg�|_|j |j_ dS)Nr>r:r<r�r�r�r�r.r�r�r�r�r9r�) zbasezbasefontzbgsoundzlinkzmetaznoframeszscriptzstyleztitle)zbodyzhtmlzbr)rGrr�r`�startTagBody�startTagFrameset�startTagFromHeadr�r^r�r�r�rar�)r>rJr:)rbr#r$rGs z*getPhases..AfterHeadPhase.__init__cSs|j�dS)NT)r�)r>r#r#r$r�*s z,getPhases..AfterHeadPhase.processEOFcSs|j�|S)N)r�)r>r�r#r#r$r�.s z3getPhases..AfterHeadPhase.processCharacterscSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r`2sz.getPhases..AfterHeadPhase.startTagHtmlcSs6d|j_|jj|�|jjd|j_dS)NFr8)rJrbr:rGrEr])r>r�r#r#r$r�5sz.getPhases..AfterHeadPhase.startTagBodycSs*|jj|�|jjd|j_dS)Nr;)r:rGrJrEr])r>r�r#r#r$r�:sz2getPhases..AfterHeadPhase.startTagFramesetcSs�|jjdi|dd6�|jjj|jj�|jjdj|�xG|jjddd�D],}|jdkrh|jjj |�PqhqhWdS)Nz#unexpected-start-tag-out-of-my-headr<r�rr9r{) rJr~r:r}r�r�rEr�r<�remove)r>r�rAr#r#r$r�>s z2getPhases..AfterHeadPhase.startTagFromHeadcSs"|jjdi|dd6�dS)Nzunexpected-start-tagr<)rJr~)r>r�r#r#r$r�Hsz.getPhases..AfterHeadPhase.startTagHeadcSs|j�|S)N)r�)r>r�r#r#r$r�Ks z/getPhases..AfterHeadPhase.startTagOthercSs|j�|S)N)r�)r>r�r#r#r$r�Os z2getPhases..AfterHeadPhase.endTagHtmlBodyBrcSs"|jjdi|dd6�dS)Nzunexpected-end-tagr<)rJr~)r>r�r#r#r$r�Ssz-getPhases..AfterHeadPhase.endTagOthercSs?|jjtdd��|jjd|j_d|j_dS)Nr:rnr8T)r:rGr�rJrEr]rb)r>r#r#r$r�Vsz.getPhases..AfterHeadPhase.anythingElseN)r5r6r7rGr�r�r`r�r�r�r�r�r�r�r�r#)rbr#r$�AfterHeadPhases r�csneZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�Zdd�Z dd�Zdd�Zdd�Zdd�Zdd �Zd!d"�Zd#d$�Zd%d&�Zd'd(�Zd)d*�Zd+d,�Zd-d.�Zd/d0�Zd1d2�Zd3d4�Zd5d6�Zd7d8�Zd9d:�Zd;d<�Z d=d>�Z!d?d@�Z"dAdB�Z#dCdD�Z$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�Z4deS)fzgetPhases..InBodyPhasec+s]�j|||�|j|_tjd|jfdd|jfd|jfd |jfde|j ft |jfdf|jfd'|j fdg|jfd+|jfd,|jfdh|jfd9|jfd:|jfdi|jfd>|jfd?|jfdj|jfdk|jfdI|jfdJ|jfdK|jfdL|jfdM|jfdN|jfdl|jfdQ|j fdm|j!fdn|j"fdV|j#fdW|j$fdo|j%fg �|_&|j'|j&_(tjd|j)fd|j*fdp|j+fd'|j,fd!|j-fdq|j.ft |j/fdr|j0fds|j1fdA|j2fg �|_3|j4|j3_(dS)tNr>r�r�r�r�r�r.r�r�r�r�r:r<�address�article�aside� blockquote�center�details�dirr%�dl�fieldset� figcaption�figure�footer�header�hgroup�main�menu�nav�olr"�section�summary�ul�pre�listing�form�li�dd�dtrR�a�b�big�code�em�font�i�s�small�strike�strong�tt�u�nobr�button�applet�marquee�object�xmpr7�arear��embed�img�keygen�wbr�param�source�track�input�hr�image�isindex�textarea�iframe�noembedr�r(�rp�rt�option�optgroup�mathrwr3�colr5�framer9r/r*r1r+r0r-�dialog) zbasezbasefontzbgsoundzcommandzlinkzmeta�noframeszscriptzstyleztitle)�address�article�aside� blockquote�center�detailsr$�dir�div�dl�fieldset� figcaption�figure�footer�header�hgroup�main�menu�nav�olr"�section�summary�ul)�pre�listing)�li�dd�dt)r��big�code�em�fontr�r��small�strike�strong�ttr)�applet�marquee�object)zareazbrzembedzimgzkeygenzwbr)zparamzsourceztrack)znoembedrznoscript)zrpzrt)zoptionzoptgroup)zcaptionzcolzcolgroupzframezheadztbodyztdztfootzthztheadztr)rr r!r"zbuttonr#r$zdialogr%r&r'r(r)r*r+r,r-r6r.r/r0r1r5r2r3r4)r8r9r7)r�r�r:r;r<r=r�znobrr�r>r?r@rAr)rBrCrD)5rGr��processSpaceCharactersNonPrerr�r`�startTagProcessInHeadr�r��startTagClosePr�startTagHeading�startTagPreListing�startTagForm�startTagListItem�startTagPlaintext� startTagA�startTagFormatting�startTagNobr�startTagButton�startTagAppletMarqueeObject�startTagXmp� startTagTable�startTagVoidFormatting�startTagParamSource� startTagInput� startTagHr� startTagImage�startTagIsIndex�startTagTextarea�startTagIFrame�startTagRawtext�startTagSelect�startTagRpRt�startTagOpt�startTagMath�startTagSvg�startTagMisplacedr^r�r�� endTagBody� endTagHtml�endTagBlock� endTagForm�endTagP�endTagListItem� endTagHeading�endTagFormatting�endTagAppletMarqueeObject�endTagBrrar�)r>rJr:)rbr#r$rG^s| z'getPhases..InBodyPhase.__init__cSs�|j|jks$|j|jkr(dSt|j�t|j�krJdSt|jj��}t|jj��}x-t||�D]\}}||kr�dSq�WdS)NFT)r<rfr�rg�sortedr(�zip)r>�node1�node2�attributes1�attributes2�attr1�attr2r#r#r$�isMatchingFormattingElement�s$z:getPhases..InBodyPhase.isMatchingFormattingElementcSs�|jj|�|jjd}g}xS|jjddd�D]8}|tkrVPq@|j||�r@|j|�q@q@Wt|�dks�t�t|�dkr�|jjj |d�n|jjj|�dS)Nr�r{r{r{) r:rGr}�activeFormattingElementsr rur�r�r�r�)r>r�ri�matchingElementsrAr#r#r$�addFormattingElement�s z3getPhases..InBodyPhase.addFormattingElementc SsWtd�}xD|jjddd�D])}|j|kr&|jjd�Pq&q&WdS)Nr�r�r�r"r/r*r1r+r0r-r:r>rz expected-closing-tag-but-got-eof)zddzdtzlir"ztbodyztdztfootzthztheadztrzbodyzhtmlr{)r�r:r}r<rJr~)r>�allowed_elementsrAr#r#r$r��s z)getPhases..InBodyPhase.processEOFcSs�|d}|j|_|jd�rh|jjdjdkrh|jjd j�rh|dd�}n|r�|jj�|jj|�ndS) Nrs� rr�r�rr{)zprezlistingztextarear{) rEr�rQr:r}r<� hasContent�#reconstructActiveFormattingElementsr])r>r�rsr#r#r$�!processSpaceCharactersDropNewline�s z@getPhases..InBodyPhase.processSpaceCharactersDropNewlinecSsq|ddkrdS|jj�|jj|d�|jjrmtdd�|dD��rmd|j_ndS)Nrs�cSsg|]}|tk�qSr#)r )r;�charr#r#r$r?�s zDgetPhases..InBodyPhase.processCharacters..F)r:r}r]rJrb�any)r>r�r#r#r$r��s z0getPhases..InBodyPhase.processCharacterscSs%|jj�|jj|d�dS)Nrs)r:r}r])r>r�r#r#r$r��s z5getPhases..InBodyPhase.processSpaceCharacterscSs|jjdj|�S)Nr�)rJrEr�)r>r�r#r#r$rF�sz4getPhases..InBodyPhase.startTagProcessInHeadcSs�|jjdidd6�t|jj�dksK|jjdjdkr`|jjs�t�n`d|j_xQ|dj �D]?\}}||jjdj kr}||jjdj |r�r_rNr#r#r$r��sz+getPhases..InBodyPhase.startTagBodycSs|jjdidd6�t|jj�dksK|jjdjdkr`|jjs�t�n�|jjson�|jjdj r�|jjdj j |jjd�nx-|jjdjdkr�|jjj�q�W|jj|�|jj d|j_dS) Nzunexpected-start-tagr<r<rr:r>r;r{)rJr~r�r:r}r<rOr�rb�parent�removeChildr�rGrEr])r>r�r#r#r$r��s1'z/getPhases..InBodyPhase.startTagFramesetcSsB|jjddd�r.|jtd��n|jj|�dS)Nr"�variantr)r:�elementInScopergr�rG)r>r�r#r#r$rGsz-getPhases..InBodyPhase.startTagClosePcSsZ|jjddd�r.|jtd��n|jj|�d|j_|j|_dS)Nr"r�rF) r:r�rgr�rGrJrbr~r�)r>r�r#r#r$rIs z1getPhases..InBodyPhase.startTagPreListingcSs�|jjr)|jjdidd6�nT|jjddd�rW|jtd��n|jj|�|jjd|j_dS) Nzunexpected-start-tagr�r<r"r�rrr{) r:�formPointerrJr~r�rgr�rGr})r>r�r#r#r$rJsz+getPhases..InBodyPhase.startTagFormcSsd|j_idgd6ddgd6ddgd6}||d}xnt|jj�D]Z}|j|kr�|jjjt|jd��Pn|j t krW|jdkrWPqWqWW|jjd d d�r�|jjjtd d��n|jj|�dS) NFr�r�r�r<ror�r%r"r�r)zaddresszdivr") rJrb�reversedr:r}r<r]r�r�� nameTuplerr�rG)r>r��stopNamesMap� stopNamesrAr#r#r$rKs" z/getPhases..InBodyPhase.startTagListItemcSsZ|jjddd�r.|jtd��n|jj|�|jjj|jj_dS)Nr"r�r) r:r�rgr�rGrJrr\rZ)r>r�r#r#r$rL1sz0getPhases..InBodyPhase.startTagPlaintextcSs�|jjddd�r.|jtd��n|jjdjtkrx|jjdi|dd6�|jjj �n|jj |�dS)Nr"r�rrzunexpected-start-tagr<r{)r:r�rgr�r}r<rrJr~r�rG)r>r�r#r#r$rH7sz.getPhases..InBodyPhase.startTagHeadingcSs�|jjd�}|r�|jjdidd6dd6�|jtd��||jjkrt|jjj|�n||jjkr�|jjj|�q�n|jj �|j |�dS)Nr�z$unexpected-start-tag-implies-end-tag� startName�endName)r:�!elementInActiveFormattingElementsrJr~rjr�r}r�rwr}ry)r>r��afeAElementr#r#r$rM?s z(getPhases..InBodyPhase.startTagAcSs|jj�|j|�dS)N)r:r}ry)r>r�r#r#r$rNLs z1getPhases..InBodyPhase.startTagFormattingcSst|jj�|jjd�rc|jjdidd6dd6�|jtd��|jj�n|j|�dS)Nrz$unexpected-start-tag-implies-end-tagr�r�)r:r}r�rJr~r�r�ry)r>r�r#r#r$rOPs z+getPhases..InBodyPhase.startTagNobrcSsw|jjd�rJ|jjdidd6dd6�|jtd��|S|jj�|jj|�d|j_dS)Nrz$unexpected-start-tag-implies-end-tagr�r�F) r:r�rJr~r�r�r}rGrb)r>r�r#r#r$rPZs z-getPhases..InBodyPhase.startTagButtoncSs@|jj�|jj|�|jjjt�d|j_dS)NF)r:r}rGrwr�r rJrb)r>r�r#r#r$rQes z:getPhases..InBodyPhase.startTagAppletMarqueeObjectcSs^|jjddd�r.|jtd��n|jj�d|j_|jj|d�dS)Nr"r�rFrD)r:r�rgr�r}rJrbrJ)r>r�r#r#r$rRks z*getPhases..InBodyPhase.startTagXmpcSsy|jjdkrC|jjddd�rC|jtd��qCn|jj|�d|j_|jjd|j_ dS)Nr�r"r�rFr6) rJrWr:r�r�r�rGrbrEr])r>r�r#r#r$rSrsz,getPhases..InBodyPhase.startTagTablecSsG|jj�|jj|�|jjj�d|dr�r#r#r$rTzs z5getPhases..InBodyPhase.startTagVoidFormattingcSsY|jj}|j|�d|dkrU|ddjt�dkrU||j_ndS)Nr,rs�hidden)rJrbrTrhr)r>r�rbr#r#r$rV�s z,getPhases..InBodyPhase.startTagInputcSs.|jj|�|jjj�d|dr�r#r#r$rU�sz2getPhases..InBodyPhase.startTagParamSourcecSsh|jjddd�r.|jtd��n|jj|�|jjj�d|dr�r#r#r$rW�s z)getPhases..InBodyPhase.startTagHrc SsO|jjdidd6dd6�|jtddd|dd |d ��dS) Nzunexpected-start-tag-treated-asrr!r �newNamernrgrsry)rJr~r�r�)r>r�r#r#r$rX�s z,getPhases..InBodyPhase.startTagImagec Ss�|jjdidd6�|jjr*dSi}d|dkrU|dd|dr�� form_attrsr�rgr#r#r$rY�s6 z.getPhases..InBodyPhase.startTagIsIndexcSsD|jj|�|jjj|jj_|j|_d|j_dS)NF) r:rGrJrrYrZr~r�rb)r>r�r#r#r$rZ�sz/getPhases..InBodyPhase.startTagTextareacSsd|j_|j|�dS)NF)rJrbr\)r>r�r#r#r$r[�sz-getPhases..InBodyPhase.startTagIFramecSs|jj|d�dS)z8iframe, noembed noframes, noscript(if scripting enabled)rDN)rJrJ)r>r�r#r#r$r\�sz.getPhases..InBodyPhase.startTagRawtextcSsY|jjdjdkr5|jjjtd��n|jj�|jjj|�dS)Nrrr{) r:r}r<rJr]r�r�r}rG)r>r�r#r#r$r_�s z*getPhases..InBodyPhase.startTagOptcSs�|jj�|jj|�d|j_|jj|jjd|jjd|jjd|jjd|jjd|jjdfkr�|jjd|j_n|jjd |j_dS) NFr6r2r4r.r,r)�inSelectInTabler')r:r}rGrJrbr]rE)r>r�r#r#r$r]�s z-getPhases..InBodyPhase.startTagSelectcSs_|jjd�rK|jj�|jjdjdkrK|jj�qKn|jj|�dS)N�rubyrr{)r:r��generateImpliedEndTagsr}r<rJr~rG)r>r�r#r#r$r^�s z+getPhases..InBodyPhase.startTagRpRtcSsv|jj�|jj|�|jj|�td|d<|jj|�|drr|jjj�d|dr�r#r#r$r`�s z+getPhases..InBodyPhase.startTagMathcSsv|jj�|jj|�|jj|�td|d<|jj|�|drr|jjj�d|dr�r#r#r$ra�s z*getPhases..InBodyPhase.startTagSvgcSs"|jjdi|dd6�dS)a5 Elements that should be children of other elements that have a different insertion mode; here they are ignored "caption", "col", "colgroup", "frame", "frameset", "head", "option", "optgroup", "tbody", "td", "tfoot", "th", "thead", "tr", "noscript" zunexpected-start-tag-ignoredr<N)rJr~)r>r�r#r#r$rb�sz0getPhases..InBodyPhase.startTagMisplacedcSs!|jj�|jj|�dS)N)r:r}rG)r>r�r#r#r$r�s z,getPhases..InBodyPhase.startTagOthercSs�|jjddd�sa|jtdd��|jjdidd6�|jtdd��n�|jjd�|jjd j dkr�|jjdidd6�n|jjj �}x%|j dkr�|jjj �}q�WdS) Nr"r�rrnzunexpected-end-tagr<rorr{)r:r�rGr�rJr~rgr�r}r<r�)r>r�rAr#r#r$rgsz&getPhases..InBodyPhase.endTagPcSs�|jjd�s#|jj�dS|jjdjdkr�x^|jjdd�D]C}|jtd�krS|jjdidd6|jd6�PqSqSWn|jjd|j_dS)Nr:r�r�r�r�rrr"rrr/r*r1r+r0r-r>z$expected-one-end-tag-but-got-another�expectedName�gotName� afterBodyr{)zddzdtzlizoptgroupzoptionr"zrpzrtztbodyztdztfootzthztheadztrzbodyzhtml) r:r�rJr~r}r<r�rEr])r>r�rAr#r#r$rcs z)getPhases..InBodyPhase.endTagBodycSs-|jjd�r)|jtd��|SdS)Nr:)r:r�rcr�)r>r�r#r#r$rd+sz)getPhases..InBodyPhase.endTagHtmlcSs�|ddkr|j|_n|jj|d�}|rK|jj�n|jjdj|dkr�|jjdi|dd6�n|r�|jjj �}x,|j|dkr�|jjj �}q�WndS)Nr<r�rzend-tag-too-earlyr{) rEr�r:r�r�r}r<rJr~r�)r>r��inScoperAr#r#r$re1s!z*getPhases..InBodyPhase.endTagBlockcSs�|jj}d|j_|dks7|jj|�rT|jjdidd6�nS|jj�|jjd|kr�|jjdidd6�n|jjj|�dS)Nzunexpected-end-tagr�r<rzend-tag-too-early-ignoredr{)r:r�r�rJr~r�r}r�)r>r�rAr#r#r$rf?s z)getPhases..InBodyPhase.endTagFormcSs�|ddkrd}nd}|jj|dd|�s\|jjdi|dd6�n�|jjd|d�|jjd j|dkr�|jjdi|dd6�n|jjj�}x)|j|dkr�|jjj�}q�WdS) Nr<r�rr�zunexpected-end-tag�excluderzend-tag-too-earlyr{)r:r�rJr~r�r}r<r�)r>r�r�rAr#r#r$rhLs ! z-getPhases..InBodyPhase.endTagListItemcSs�x1tD])}|jj|�r|jj�PqqW|jjdj|dkrr|jjdi|dd6�nx^tD]V}|jj|�ry|jjj�}x%|jtkr�|jjj�}q�WPqyqyWdS)Nrr<zend-tag-too-earlyr{) rr:r�r�r}r<rJr~r�)r>r��itemr#r#r$ri]s ! z,getPhases..InBodyPhase.endTagHeadingcSs9d}x,|dkr4|d7}|jj|d�}|sd||jjkru|jj|j�ru|j|�dS||jjkr�|jjdi|dd6�|jjj |�dS|jj|j�s�|jjdi|dd6�dS||jjdkr*|jjdi|dd6�n|jjj |�}d}x7|jj|d�D]}|jtkr\|}Pq\q\W|dkr�|jjj �}x"||kr�|jjj �}q�W|jjj |�dS|jj|d}|jjj |�}|} } d}|jjj | �}x|d kr?|d7}|d8}|jj|} | |jjkr�|jjj | �q*n| |kr�Pn| |kr�|jjj | �d}n| j�} | |jj|jjj | �<| |jj|jjj | �<| } | jr)| jj| �n| j| �| } q*W| jr\| jj| �n|jtd�kr�|jj�\}}|j| |�n |j| �|j�} |j| �|j| �|jjj |�|jjj|| �|jjj |�|jjj|jjj |�d| �q WdS)z)The much-feared adoption agency algorithmr�rr<Nzadoption-agency-1.2zadoption-agency-4.4zadoption-agency-1.3rvr7r/r1r0r-r{)ztableztbodyztfootztheadztr)r:r�r}r�r<r�rJr~rwr��indexr�rr�� cloneNoder�r��appendChildr��getTableMisnestedNodePosition�insertBefore�reparentChildren�insert)r>r��outerLoopCounter�formattingElement�afeIndex� furthestBlockri�commonAncestor�bookmark�lastNoderA�innerLoopCounterr��cloner�r�r#r#r$rjls� ! z/getPhases..InBodyPhase.endTagFormattingcSs�|jj|d�r&|jj�n|jjdj|dkrd|jjdi|dd6�n|jj|d�r�|jjj�}x)|j|dkr�|jjj�}q�W|jj�ndS)Nr<rzend-tag-too-earlyr{) r:r�r�r}r<rJr~r��clearActiveFormattingElements)r>r�rir#r#r$rks!z8getPhases..InBodyPhase.endTagAppletMarqueeObjectcSs[|jjdidd6dd6�|jj�|jjtdd��|jjj�dS)Nzunexpected-end-tag-treated-asr�r!z br elementr�rn)rJr~r:r}rGr�r}r�)r>r�r#r#r$rls z'getPhases..InBodyPhase.endTagBrcSs�x�|jjddd�D]�}|j|dkr�|jjd|d�|jjdj|dkr�|jjdi|dd6�nx|jjj�|kr�q�WPq|jtkr|jjdi|dd6�PqqWdS)Nrr<r�zunexpected-end-tagr{r{) r:r}r<r�rJr~r�r�r)r>r�rAr#r#r$r�"s !z*getPhases..InBodyPhase.endTagOtherN)5r5r6r7rGruryr�r~r�r�rFr�r�rGrIrJrKrLrHrMrNrOrPrQrRrSrTrVrUrWrXrYrZr[r\r_r]r^r`rarbr�rgrcrdrerfrhrirjrkrlr�r#)rbr#r$�InBodyPhase[sdF �r�cs^eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd S)zgetPhases..TextPhasecse�j|||�tjg�|_|j|j_tjd|jfg�|_|j|j_dS)Nr�) rGrr�r^r�r��endTagScriptrar�)r>rJr:)rbr#r$rG1sz%getPhases..TextPhase.__init__cSs|jj|d�dS)Nrs)r:r])r>r�r#r#r$r�9sz.getPhases..TextPhase.processCharacterscSsM|jjdi|jjdjd6�|jjj�|jj|j_dS)Nz&expected-named-closing-tag-but-got-eofrr<Tr{)rJr~r:r}r<r�rHr])r>r#r#r$r�<s z'getPhases..TextPhase.processEOFcSsdstd|d��dS)NFz4Tried to process start tag %s in RCDATA/RAWTEXT moder<)r�)r>r�r#r#r$r�Csz*getPhases..TextPhase.startTagOthercSs=|jjj�}|jdks't�|jj|j_dS)Nr�)r:r}r�r<r�rJrHr])r>r�rAr#r#r$r�Fsz)getPhases..TextPhase.endTagScriptcSs&|jjj�|jj|j_dS)N)r:r}r�rJrHr])r>r�r#r#r$r�Msz(getPhases..TextPhase.endTagOtherN) r5r6r7rGr�r�r�r�r�r#)rbr#r$� TextPhase0sr�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�Zdd�Z dd�Zdd�Zdd�Zdd�Zdd �Zd!d"�Zd#d$�Zd%d&�Zd'S)(zgetPhases..InTablePhasec s��j|||�tjd|jfd|jfd|jfd|jfd|jfd|jfd|j fd|j fd|jfd|jfg �|_ |j|j _tjd|jfd|jfg�|_|j|j_dS)Nr>r3r5rr/r1r0r*r+r-r7r�r�rr�r:)�tbody�tfoot�thead)�td�th�tr)zstylezscript)zbodyzcaptionzcolzcolgroupzhtmlr�r�r�r�r�r�)rGrr�r`�startTagCaption�startTagColgroup�startTagCol�startTagRowGroup�startTagImplyTbodyrS�startTagStyleScriptrVrJr^r�r��endTagTable�endTagIgnorerar�)r>rJr:)rbr#r$rGSs$z(getPhases..InTablePhase.__init__cSs4x-|jjdjdkr/|jjj�qWdS)Nrr7r>r{)ztablezhtml)r:r}r<r�)r>r#r#r$�clearStackToTableContextksz8getPhases..InTablePhase.clearStackToTableContextcSsB|jjdjdkr,|jjd�n|jjs>t�dS)Nrr>zeof-in-tabler{)r:r}r<rJr~rOr�)r>r#r#r$r�tsz*getPhases..InTablePhase.processEOFcSsH|jj}|jjd|j_||jj_|jjj|�dS)N�inTableText)rJr]rErHr�)r>r�rHr#r#r$r�{sz6getPhases..InTablePhase.processSpaceCharacterscSsH|jj}|jjd|j_||jj_|jjj|�dS)Nr�)rJr]rErHr�)r>r�rHr#r#r$r��sz1getPhases..InTablePhase.processCharacterscSs3d|j_|jjdj|�d|j_dS)NTr8F)r:�insertFromTablerJrEr�)r>r�r#r#r$r]�sz*getPhases..InTablePhase.insertTextcSsG|j�|jjjt�|jj|�|jjd|j_dS)Nr2) r�r:rwr�r rGrJrEr])r>r�r#r#r$r��s z/getPhases..InTablePhase.startTagCaptioncSs4|j�|jj|�|jjd|j_dS)Nr4)r�r:rGrJrEr])r>r�r#r#r$r��s z0getPhases..InTablePhase.startTagColgroupcSs|jtdd��|S)Nr5rn)r�r�)r>r�r#r#r$r��sz+getPhases..InTablePhase.startTagColcSs4|j�|jj|�|jjd|j_dS)Nr.)r�r:rGrJrEr])r>r�r#r#r$r��s z0getPhases..InTablePhase.startTagRowGroupcSs|jtdd��|S)Nr/rn)r�r�)r>r�r#r#r$r��sz2getPhases..InTablePhase.startTagImplyTbodycSsN|jjdidd6dd6�|jjjtd��|jjsJ|SdS)Nz$unexpected-start-tag-implies-end-tagr7r�r�)rJr~r]r�r�rO)r>r�r#r#r$rS�s z-getPhases..InTablePhase.startTagTablecSs|jjdj|�S)Nr�)rJrEr�)r>r�r#r#r$r��sz3getPhases..InTablePhase.startTagStyleScriptcSsqd|dkr`|ddjt�dkr`|jjd�|jj|�|jjj�n |j|�dS)Nr,rsr�z unexpected-hidden-input-in-table) rhrrJr~r:rGr}r�r�)r>r�r#r#r$rV�sz-getPhases..InTablePhase.startTagInputcSs_|jjd�|jjdkr[|jj|�|jjd|j_|jjj�ndS)Nzunexpected-form-in-tablerr{)rJr~r:r�rGr}r�)r>r�r#r#r$rJ�s z,getPhases..InTablePhase.startTagFormcSsQ|jjdi|dd6�d|j_|jjdj|�d|j_dS)Nz)unexpected-start-tag-implies-table-voodoor<Tr8F)rJr~r:r�rEr�)r>r�r#r#r$r��sz-getPhases..InTablePhase.startTagOthercSs�|jjddd�r�|jj�|jjdjdkro|jjdidd6|jjdjd6�nx-|jjd jdkr�|jjj�qrW|jjj�|jj�n|jj s�t �|jj�dS) Nr7r�rzend-tag-too-early-namedr�r�r{r{r{)r:r�r�r}r<rJr~r�r_rOr�)r>r�r#r#r$r��s z+getPhases..InTablePhase.endTagTablecSs"|jjdi|dd6�dS)Nzunexpected-end-tagr<)rJr~)r>r�r#r#r$r��sz,getPhases..InTablePhase.endTagIgnorecSsQ|jjdi|dd6�d|j_|jjdj|�d|j_dS)Nz'unexpected-end-tag-implies-table-voodoor<Tr8F)rJr~r:r�rEr�)r>r�r#r#r$r��sz+getPhases..InTablePhase.endTagOtherN)r5r6r7rGr�r�r�r�r]r�r�r�r�r�rSr�rVrJr�r�r�r�r#)rbr#r$�InTablePhaseQs& r�csveZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dS)z#getPhases..InTableTextPhasecs)�j|||�d|_g|_dS)N)rGrH�characterTokens)r>rJr:)rbr#r$rG�s z,getPhases..InTableTextPhase.__init__cSs�djdd�|jD��}tdd�|D��rjitdd6|d6}|jjdj|�n|r�|jj|�ng|_dS) NrgcSsg|]}|d�qS)rsr#)r;r�r#r#r$r?�s zGgetPhases..InTableTextPhase.flushCharacters..cSsg|]}|tk�qSr#)r )r;r�r#r#r$r?�s rlr,rsr6)�joinr�r�rrJrEr]r:)r>rsr�r#r#r$�flushCharacters�sz3getPhases..InTableTextPhase.flushCharacterscSs|j�|j|j_|S)N)r�rHrJr])r>r�r#r#r$r��s z2getPhases..InTableTextPhase.processCommentcSs|j�|j|j_dS)NT)r�rHrJr])r>r#r#r$r��s z.getPhases..InTableTextPhase.processEOFcSs(|ddkrdS|jj|�dS)Nrsr)r�r�)r>r�r#r#r$r��sz5getPhases..InTableTextPhase.processCharacterscSs|jj|�dS)N)r�r�)r>r�r#r#r$r�sz:getPhases..InTableTextPhase.processSpaceCharacterscSs|j�|j|j_|S)N)r�rHrJr])r>r�r#r#r$r�s z3getPhases..InTableTextPhase.processStartTagcSs|j�|j|j_|S)N)r�rHrJr])r>r�r#r#r$r� s z1getPhases..InTableTextPhase.processEndTagN)r5r6r7rGr�r�r�r�r�r�r�r#)rbr#r$�InTableTextPhase�s r�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�ZdS)z!getPhases..InCaptionPhasec s��j|||�tjd|jfd |jfg�|_|j|j_tjd|jfd|j fd|j fg�|_|j|j_dS)Nr>r3rr5r/r*r1r+r0r-r7r:) zcaption�col�colgroup�tbody�td�tfoot�th�thead�tr) zbodyr�r�zhtmlr�r�r�r�r�r�) rGrr�r`�startTagTableElementr^r�r�� endTagCaptionr�r�rar�)r>rJr:)rbr#r$rGsz*getPhases..InCaptionPhase.__init__cSs|jjddd�S)Nr3r�r7)r:r�)r>r#r#r$�ignoreEndTagCaption#sz5getPhases..InCaptionPhase.ignoreEndTagCaptioncSs|jjdj�dS)Nr8)rJrEr�)r>r#r#r$r�&sz,getPhases..InCaptionPhase.processEOFcSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r�)sz3getPhases..InCaptionPhase.processCharacterscSs@|jj�|j�}|jjjtd��|s<|SdS)Nr3)rJr~r�r]r�r�)r>r��ignoreEndTagr#r#r$r�,s z6getPhases..InCaptionPhase.startTagTableElementcSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r�4sz/getPhases..InCaptionPhase.startTagOthercSs�|j�s�|jj�|jjdjdkrc|jjdidd6|jjdjd6�nx-|jjd jdkr�|jjj�qfW|jjj�|jj�|jj d|j_ n|jjs�t�|jj�dS) Nrr3z$expected-one-end-tag-but-got-anotherr�r�r6r{r{r{) r�r:r�r}r<rJr~r�r�rEr]rOr�)r>r�r#r#r$r�7s z/getPhases..InCaptionPhase.endTagCaptioncSs@|jj�|j�}|jjjtd��|s<|SdS)Nr3)rJr~r�r]r�r�)r>r�r�r#r#r$r�Is z-getPhases..InCaptionPhase.endTagTablecSs"|jjdi|dd6�dS)Nzunexpected-end-tagr<)rJr~)r>r�r#r#r$r�Psz.getPhases..InCaptionPhase.endTagIgnorecSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r�Ssz-getPhases..InCaptionPhase.endTagOtherN) r5r6r7rGr�r�r�r�r�r�r�r�r�r#)rbr#r$�InCaptionPhasesr�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�ZdS)z%getPhases..InColumnGroupPhasecs��j|||�tjd|jfd|jfg�|_|j|j_tjd|jfd|j fg�|_ |j|j _dS)Nr>rr5)rGrr�r`r�r^r�r��endTagColgroup� endTagColrar�)r>rJr:)rbr#r$rGYsz.getPhases..InColumnGroupPhase.__init__cSs|jjdjdkS)Nrr>r{)r:r}r<)r>r#r#r$�ignoreEndTagColgrouphsz:getPhases..InColumnGroupPhase.ignoreEndTagColgroupcSs\|jjdjdkr/|jjs+t�dS|j�}|jtd��|sXdSdS)Nrr>r5Tr{) r:r}r<rJrOr�r�r�r�)r>r�r#r#r$r�ksz0getPhases..InColumnGroupPhase.processEOFcSs-|j�}|jtd��|s)|SdS)Nr5)r�r�r�)r>r�r�r#r#r$r�usz7getPhases..InColumnGroupPhase.processCharacterscSs$|jj|�|jjj�dS)N)r:rGr}r�)r>r�r#r#r$r�{sz1getPhases..InColumnGroupPhase.startTagColcSs-|j�}|jtd��|s)|SdS)Nr5)r�r�r�)r>r�r�r#r#r$r�sz3getPhases..InColumnGroupPhase.startTagOthercSsX|j�r.|jjst�|jj�n&|jjj�|jjd|j_ dS)Nr6) r�rJrOr�r~r:r}r�rEr])r>r�r#r#r$r��s z4getPhases..InColumnGroupPhase.endTagColgroupcSs|jjdidd6�dS)Nz no-end-tagrr<)rJr~)r>r�r#r#r$r��sz/getPhases..InColumnGroupPhase.endTagColcSs-|j�}|jtd��|s)|SdS)Nr5)r�r�r�)r>r�r�r#r#r$r��sz1getPhases..InColumnGroupPhase.endTagOtherN)r5r6r7rGr�r�r�r�r�r�r�r�r#)rbr#r$�InColumnGroupPhaseVs r�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�Zdd�Z dd�Zdd�ZdS)z#getPhases..InTableBodyPhasecs��j|||�tjd|jfd|jfd |jfd|jfg�|_|j|j_ tjd|j fd|jfd|jfg�|_ |j|j _ dS)Nr>r-r*r+r3rr5r/r1r0r7r:)�td�th)�caption�col�colgroup�tbody�tfoot�thead)r�r�r�)zbodyr�r�r�zhtmlr�r�ztr)rGrr�r`� startTagTr�startTagTableCell�startTagTableOtherr^r�r��endTagTableRowGroupr�r�rar�)r>rJr:)rbr#r$rG�sz,getPhases..InTableBodyPhase.__init__cSsbx-|jjdjdkr/|jjj�qW|jjdjdkr^|jjs^t�ndS) Nrr/r1r0r>r{)ztbodyztfootztheadzhtmlr{)r:r}r<r�rJrOr�)r>r#r#r$�clearStackToTableBodyContext�s z@getPhases..InTableBodyPhase.clearStackToTableBodyContextcSs|jjdj�dS)Nr6)rJrEr�)r>r#r#r$r��sz.getPhases..InTableBodyPhase.processEOFcSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r��sz:getPhases..InTableBodyPhase.processSpaceCharacterscSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r��sz5getPhases..InTableBodyPhase.processCharacterscSs4|j�|jj|�|jjd|j_dS)Nr,)r�r:rGrJrEr])r>r�r#r#r$r��s z.getPhases..InTableBodyPhase.startTagTrcSs8|jjdi|dd6�|jtdd��|S)Nzunexpected-cell-in-table-bodyr<r-rn)rJr~r�r�)r>r�r#r#r$r��sz5getPhases..InTableBodyPhase.startTagTableCellcSs�|jjddd�sH|jjddd�sH|jjddd�rv|j�|jt|jjdj��|S|jjs�t �|jj �dS)Nr/r�r7r0r1rr{)r:r�r�r�r�r}r<rJrOr�r~)r>r�r#r#r$r��s z6getPhases..InTableBodyPhase.startTagTableOthercSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r��sz1getPhases..InTableBodyPhase.startTagOthercSsq|jj|ddd�rO|j�|jjj�|jjd|j_n|jjdi|dd6�dS)Nr<r�r7r6z unexpected-end-tag-in-table-body) r:r�r�r}r�rJrEr]r~)r>r�r#r#r$r��s z7getPhases..InTableBodyPhase.endTagTableRowGroupcSs�|jjddd�sH|jjddd�sH|jjddd�rv|j�|jt|jjdj��|S|jjs�t �|jj �dS)Nr/r�r7r0r1rr{)r:r�r�r�r�r}r<rJrOr�r~)r>r�r#r#r$r��s z/getPhases..InTableBodyPhase.endTagTablecSs"|jjdi|dd6�dS)Nz unexpected-end-tag-in-table-bodyr<)rJr~)r>r�r#r#r$r��sz0getPhases..InTableBodyPhase.endTagIgnorecSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r��sz/getPhases..InTableBodyPhase.endTagOtherN)r5r6r7rGr�r�r�r�r�r�r�r�r�r�r�r�r#)rbr#r$�InTableBodyPhase�s r�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�Zdd�Z dd�Zdd�Zdd�ZdS)zgetPhases..InRowPhasecs��j|||�tjd|jfd |jfd|jfg�|_|j|j_tjd |j fd|j fd|jfd|jfg�|_ |j|j _dS)Nr>r*r+r3rr5r/r1r0r-r7r:)�td�th)�caption�col�colgroup�tbody�tfoot�theadztr)r�r�r�)zbodyr�r�r�zhtmlr�r�)rGrr�r`r�r�r^r�r��endTagTrr�r�r�rar�)r>rJr:)rbr#r$rG�sz&getPhases..InRowPhase.__init__cSs[xT|jjdjdkrV|jjdi|jjdjd6�|jjj�qWdS) Nrr-r>z'unexpected-implied-end-tag-in-table-rowr<r{)ztrzhtmlr{)r:r}r<rJr~r�)r>r#r#r$�clearStackToTableRowContextsz9getPhases..InRowPhase.clearStackToTableRowContextcSs|jjddd�S)Nr-r�r7)r:r�)r>r#r#r$�ignoreEndTagTrsz,getPhases..InRowPhase.ignoreEndTagTrcSs|jjdj�dS)Nr6)rJrEr�)r>r#r#r$r�sz(getPhases..InRowPhase.processEOFcSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r�sz4getPhases..InRowPhase.processSpaceCharacterscSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r�sz/getPhases..InRowPhase.processCharacterscSsG|j�|jj|�|jjd|j_|jjjt�dS)Nr)) r�r:rGrJrEr]rwr�r )r>r�r#r#r$r�"s z/getPhases..InRowPhase.startTagTableCellcSs-|j�}|jtd��|s)|SdS)Nr-)r�r�r�)r>r�r�r#r#r$r�(sz0getPhases..InRowPhase.startTagTableOthercSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r�/sz+getPhases..InRowPhase.startTagOthercSsb|j�s?|j�|jjj�|jjd|j_n|jjsQt �|jj �dS)Nr.)r�r�r:r}r�rJrEr]rOr�r~)r>r�r#r#r$r�2s z&getPhases..InRowPhase.endTagTrcSs-|j�}|jtd��|s)|SdS)Nr-)r�r�r�)r>r�r�r#r#r$r�<sz)getPhases..InRowPhase.endTagTablecSsD|jj|ddd�r3|jtd��|S|jj�dS)Nr<r�r7r-)r:r�r�r�rJr~)r>r�r#r#r$r�Dsz1getPhases..InRowPhase.endTagTableRowGroupcSs"|jjdi|dd6�dS)Nzunexpected-end-tag-in-table-rowr<)rJr~)r>r�r#r#r$r�Ksz*getPhases..InRowPhase.endTagIgnorecSs|jjdj|�S)Nr6)rJrEr�)r>r�r#r#r$r�Osz)getPhases..InRowPhase.endTagOtherN)r5r6r7rGr�r�r�r�r�r�r�r�r�r�r�r�r�r#)rbr#r$� InRowPhase�s r�cs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�ZdS)zgetPhases..InCellPhasecs��j|||�tjd|jfd |jfg�|_|j|j_tjd|jfd|j fd|j fg�|_|j|j_dS)Nr>r3rr5r/r*r1r+r0r-r:r7) �caption�col�colgroup�tbody�td�tfoot�th�thead�tr)r�r�)zbodyr�r�r�zhtml)ztabler�r�r�r�) rGrr�r`r�r^r�r��endTagTableCellr��endTagImplyrar�)r>rJr:)rbr#r$rGTsz'getPhases..InCellPhase.__init__cSs`|jjddd�r.|jtd��n.|jjddd�r\|jtd��ndS)Nr*r�r7r+)r:r�r�r�)r>r#r#r$� closeCellesz(getPhases..InCellPhase.closeCellcSs|jjdj�dS)Nr8)rJrEr�)r>r#r#r$r�lsz)getPhases..InCellPhase.processEOFcSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r�osz0getPhases..InCellPhase.processCharacterscSsa|jjddd�s0|jjddd�r>|j�|S|jjsPt�|jj�dS)Nr*r�r7r+)r:r�rrJrOr�r~)r>r�r#r#r$r�rs z1getPhases..InCellPhase.startTagTableOthercSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r�|sz,getPhases..InCellPhase.startTagOthercSs�|jj|ddd�r�|jj|d�|jjdj|dkr�|jjdi|dd6�x@|jjj�}|j|dkrnPqnqnWn|jjj�|jj�|jj d|j_ n|jjdi|dd6�dS) Nr<r�r7rzunexpected-cell-end-tagr,zunexpected-end-tagr{)r:r�r�r}r<rJr~r�r�rEr])r>r�rAr#r#r$r�s z.getPhases..InCellPhase.endTagTableCellcSs"|jjdi|dd6�dS)Nzunexpected-end-tagr<)rJr~)r>r�r#r#r$r��sz+getPhases..InCellPhase.endTagIgnorecSs;|jj|ddd�r*|j�|S|jj�dS)Nr<r�r7)r:r�rrJr~)r>r�r#r#r$r�s z*getPhases..InCellPhase.endTagImplycSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r��sz*getPhases..InCellPhase.endTagOtherN) r5r6r7rGrr�r�r�r�r�r�rr�r#)rbr#r$�InCellPhaseRs rcs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�Zdd�Zdd�Z dd�Zdd�ZdS)z getPhases..InSelectPhasecs��j|||�tjd|jfd|jfd|jfd|jfd |jfd|jfg�|_ |j |j _tjd|jfd|j fd|jfg�|_|j|j_dS) Nr>rrr(rr rr�)zinputzkeygenztextarea)rGrr�r`�startTagOption�startTagOptgroupr]rVr�r^r�r��endTagOption�endTagOptgroup�endTagSelectrar�)r>rJr:)rbr#r$rG�sz)getPhases..InSelectPhase.__init__cSsB|jjdjdkr,|jjd�n|jjs>t�dS)Nrr>z eof-in-selectr{)r:r}r<rJr~rOr�)r>r#r#r$r��sz+getPhases..InSelectPhase.processEOFcSs,|ddkrdS|jj|d�dS)Nrsr)r:r])r>r�r#r#r$r��sz2getPhases..InSelectPhase.processCharacterscSs@|jjdjdkr,|jjj�n|jj|�dS)Nrrr{)r:r}r<r�rG)r>r�r#r#r$r�sz/getPhases..InSelectPhase.startTagOptioncSsl|jjdjdkr,|jjj�n|jjdjdkrX|jjj�n|jj|�dS)Nrrrr{r{)r:r}r<r�rG)r>r�r#r#r$r�s z1getPhases..InSelectPhase.startTagOptgroupcSs'|jjd�|jtd��dS)Nzunexpected-select-in-selectr()rJr~rr�)r>r�r#r#r$r]�sz/getPhases..InSelectPhase.startTagSelectcSsU|jjd�|jjddd�r?|jtd��|S|jjsQt�dS)Nzunexpected-input-in-selectr(r�)rJr~r:r�rr�rOr�)r>r�r#r#r$rV�s z.getPhases..InSelectPhase.startTagInputcSs|jjdj|�S)Nr�)rJrEr�)r>r�r#r#r$r��sz/getPhases..InSelectPhase.startTagScriptcSs"|jjdi|dd6�dS)Nzunexpected-start-tag-in-selectr<)rJr~)r>r�r#r#r$r��sz.getPhases..InSelectPhase.startTagOthercSsJ|jjdjdkr,|jjj�n|jjdidd6�dS)Nrrzunexpected-end-tag-in-selectr<r{)r:r}r<r�rJr~)r>r�r#r#r$r�sz-getPhases..InSelectPhase.endTagOptioncSs�|jjdjdkrE|jjdjdkrE|jjj�n|jjd jdkrq|jjj�n|jjdidd6�dS) Nrrr�rzunexpected-end-tag-in-selectr<r{��r{)r:r}r<r�rJr~)r>r�r#r#r$r�sz/getPhases..InSelectPhase.endTagOptgroupcSs�|jjddd�rb|jjj�}x%|jdkrQ|jjj�}q-W|jj�n|jjstt�|jj �dS)Nr(r�) r:r�r}r�r<rJr_rOr�r~)r>r�rAr#r#r$r�sz-getPhases..InSelectPhase.endTagSelectcSs"|jjdi|dd6�dS)Nzunexpected-end-tag-in-selectr<)rJr~)r>r�r#r#r$r��sz,getPhases..InSelectPhase.endTagOtherN)r5r6r7rGr�r�rrr]rVr�r�rrrr�r#)rbr#r$� InSelectPhase�s r csjeZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dS)z'getPhases..InSelectInTablePhasec sq�j|||�tjd |jfg�|_|j|j_tjd |jfg�|_|j |j_dS)Nr3r7r/r1r0r-r*r+)�caption�table�tbody�tfoot�thead�tr�td�th)r rrr rrrr) rGrr�rSr^r�r�r�rar�)r>rJr:)rbr#r$rG sz0getPhases..InSelectInTablePhase.__init__cSs|jjdj�dS)Nr')rJrEr�)r>r#r#r$r� sz2getPhases..InSelectInTablePhase.processEOFcSs|jjdj|�S)Nr')rJrEr�)r>r�r#r#r$r� sz9getPhases..InSelectInTablePhase.processCharacterscSs5|jjdi|dd6�|jtd��|S)Nz5unexpected-table-element-start-tag-in-select-in-tabler<r()rJr~r�r�)r>r�r#r#r$rS sz5getPhases..InSelectInTablePhase.startTagTablecSs|jjdj|�S)Nr')rJrEr�)r>r�r#r#r$r� sz5getPhases..InSelectInTablePhase.startTagOthercSsU|jjdi|dd6�|jj|ddd�rQ|jtd��|SdS)Nz3unexpected-table-element-end-tag-in-select-in-tabler<r�r7r()rJr~r:r�r�r�)r>r�r#r#r$r� sz3getPhases..InSelectInTablePhase.endTagTablecSs|jjdj|�S)Nr')rJrEr�)r>r�r#r#r$r�& sz3getPhases..InSelectInTablePhase.endTagOtherN) r5r6r7rGr�r�rSr�r�r�r#)rbr#r$�InSelectInTablePhase src-s�eZdZeddddddddd d ddd ddddddddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,g,�Z�fd-d.�Zd/d0�Z�fd1d2�Zd3d4�Zd5d6�Z d7S)8z(getPhases..InForeignContentPhaser�r�r�r:r�r�r�r�r%r�r�r�r�h1�h2�h3�h4�h5�h6r9rr�r r�r�r�r.rr�r"r�r�r�r��spanr�r��sub�supr7r�rr��varcs�j|||�dS)N)rG)r>rJr:)rbr#r$rG3 sz1getPhases..InForeignContentPhase.__init__cSs+i$dd6dd6dd6dd6d d 6dd6d d6dd6dd6dd6dd6dd6dd6dd6dd6dd 6d!d"6d#d$6d%d&6d'd(6d)d*6d+d,6d-d.6d/d06d1d26d3d46d5d66d7d86d9d:6d;d<6d=d>6d?d@6dAdB6dCdD6dEdF6dGdH6}|dI|kr'||dI|dIr�r�r#r#r$�adjustSVGTagNames6 sL z:getPhases..InForeignContentPhase.adjustSVGTagNamescsi|ddkrd|d.InForeignContentPhase.processCharacters..F)rJrbr�r�)r>r�)rbr#r$r�_ s z:getPhases..InForeignContentPhase.processCharacterscSs�|jjd}|d|jks\|ddkr�t|dj��tdddg�@r�|jjdi|dd6�xm|jjdj|jjkr�|jj |jjd�r�|jj |jjd�r�|jjj�q}W|S|jtd kr|jj |�n3|jtd krG|j|�|jj|�n|jj|�|j|d<|jj|�|dr�|jjj�d |dr�r�r#r#r$r�g s.) z8getPhases..InForeignContentPhase.processStartTagcSsKt|jj�d}|jjd}|j|dkrZ|jjdi|dd6�nx�|jjt�|dkr�|jj|jj dkr�|jjj �|jjj|j_nx.|jjj�|kr�|jjs�t �q�Wd}Pn|d8}|jj|}|j|jjkr-q]q]|jjj|�}Pq]W|S)Nrr<zunexpected-end-tagr�r{)r�r:r}r<rJr~rhrr]rEr�rHr�r�rfr�r�)r>r�� nodeIndexrAr�r#r#r$r�� s(! z6getPhases..InForeignContentPhase.processEndTagN) r5r6r7r�rgrGrcr�r�r�r#)rbr#r$�InForeignContentPhase) s)rjcsveZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dS)z!getPhases..AfterBodyPhasecsq�j|||�tjd|jfg�|_|j|j_tjd|jfg�|_|j |j_dS)Nr>) rGrr�r`r^r�r�rdrar�)r>rJr:)rbr#r$rG� sz*getPhases..AfterBodyPhase.__init__cSsdS)Nr#)r>r#r#r$r�� sz,getPhases..AfterBodyPhase.processEOFcSs!|jj||jjd�dS)Nr)r:r\r})r>r�r#r#r$r�� sz0getPhases..AfterBodyPhase.processCommentcSs*|jjd�|jjd|j_|S)Nzunexpected-char-after-bodyr8)rJr~rEr])r>r�r#r#r$r�� sz3getPhases..AfterBodyPhase.processCharacterscSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r`� sz.getPhases..AfterBodyPhase.startTagHtmlcSs8|jjdi|dd6�|jjd|j_|S)Nzunexpected-start-tag-after-bodyr<r8)rJr~rEr])r>r�r#r#r$r�� sz/getPhases..AfterBodyPhase.startTagOthercSs9|jjr|jjd�n|jjd|j_dS)Nz'unexpected-end-tag-after-body-innerhtml�afterAfterBody)rJrOr~rEr])r>r<r#r#r$rd� sz,getPhases..AfterBodyPhase.endTagHtmlcSs8|jjdi|dd6�|jjd|j_|S)Nzunexpected-end-tag-after-bodyr<r8)rJr~rEr])r>r�r#r#r$r�� sz-getPhases..AfterBodyPhase.endTagOtherN)r5r6r7rGr�r�r�r`r�rdr�r#)rbr#r$�AfterBodyPhase� srlcs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�ZdS)z"getPhases..InFramesetPhasecs��j|||�tjd|jfd|jfd|jfd|jfg�|_|j|j_ tjd|j fg�|_|j|j_ dS)Nr>r<rr�) rGrr�r`r�� startTagFrame�startTagNoframesr^r�r��endTagFramesetrar�)r>rJr:)rbr#r$rG� sz+getPhases..InFramesetPhase.__init__cSsB|jjdjdkr,|jjd�n|jjs>t�dS)Nrr>zeof-in-framesetr{)r:r}r<rJr~rOr�)r>r#r#r$r�� sz-getPhases..InFramesetPhase.processEOFcSs|jjd�dS)Nzunexpected-char-in-frameset)rJr~)r>r�r#r#r$r�� sz4getPhases..InFramesetPhase.processCharacterscSs|jj|�dS)N)r:rG)r>r�r#r#r$r�� sz3getPhases..InFramesetPhase.startTagFramesetcSs$|jj|�|jjj�dS)N)r:rGr}r�)r>r�r#r#r$rm� sz0getPhases..InFramesetPhase.startTagFramecSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$rn� sz3getPhases..InFramesetPhase.startTagNoframescSs"|jjdi|dd6�dS)Nz unexpected-start-tag-in-framesetr<)rJr~)r>r�r#r#r$r�� sz0getPhases..InFramesetPhase.startTagOthercSs|jjdjdkr,|jjd�n|jjj�|jjr{|jjdjdkr{|jjd|j_ndS)Nrr>z)unexpected-frameset-in-frameset-innerhtmlr<� afterFramesetr{r{) r:r}r<rJr~r�rOrEr])r>r�r#r#r$ro� s z1getPhases..InFramesetPhase.endTagFramesetcSs"|jjdi|dd6�dS)Nzunexpected-end-tag-in-framesetr<)rJr~)r>r�r#r#r$r� sz.getPhases..InFramesetPhase.endTagOtherN)r5r6r7rGr�r�r�rmrnr�ror�r#)rbr#r$�InFramesetPhase� srqcsjeZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dS)z%getPhases..AfterFramesetPhasecs}�j|||�tjd|jfd|jfg�|_|j|j_tjd|jfg�|_ |j |j _dS)Nr>r�)rGrr�r`rnr^r�r�rdrar�)r>rJr:)rbr#r$rG sz.getPhases..AfterFramesetPhase.__init__cSsdS)Nr#)r>r#r#r$r� sz0getPhases..AfterFramesetPhase.processEOFcSs|jjd�dS)Nzunexpected-char-after-frameset)rJr~)r>r�r#r#r$r� sz7getPhases..AfterFramesetPhase.processCharacterscSs|jjdj|�S)Nr�)rJrEr�)r>r�r#r#r$rn sz6getPhases..AfterFramesetPhase.startTagNoframescSs"|jjdi|dd6�dS)Nz#unexpected-start-tag-after-framesetr<)rJr~)r>r�r#r#r$r� sz3getPhases..AfterFramesetPhase.startTagOthercSs|jjd|j_dS)N�afterAfterFrameset)rJrEr])r>r�r#r#r$rd" sz0getPhases..AfterFramesetPhase.endTagHtmlcSs"|jjdi|dd6�dS)Nz!unexpected-end-tag-after-framesetr<)rJr~)r>r�r#r#r$r�% sz1getPhases..AfterFramesetPhase.endTagOtherN) r5r6r7rGr�r�rnr�rdr�r#)rbr#r$�AfterFramesetPhase srscsveZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dS)z&getPhases..AfterAfterBodyPhasecsD�j|||�tjd|jfg�|_|j|j_dS)Nr>)rGrr�r`r^r�r�)r>rJr:)rbr#r$rG* sz/getPhases..AfterAfterBodyPhase.__init__cSsdS)Nr#)r>r#r#r$r�2 sz1getPhases..AfterAfterBodyPhase.processEOFcSs|jj||jj�dS)N)r:r\rc)r>r�r#r#r$r�5 sz5getPhases..AfterAfterBodyPhase.processCommentcSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r�8 sz=getPhases..AfterAfterBodyPhase.processSpaceCharacterscSs*|jjd�|jjd|j_|S)Nzexpected-eof-but-got-charr8)rJr~rEr])r>r�r#r#r$r�; sz8getPhases..AfterAfterBodyPhase.processCharacterscSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r`@ sz3getPhases..AfterAfterBodyPhase.startTagHtmlcSs8|jjdi|dd6�|jjd|j_|S)Nzexpected-eof-but-got-start-tagr<r8)rJr~rEr])r>r�r#r#r$r�C sz4getPhases..AfterAfterBodyPhase.startTagOthercSs8|jjdi|dd6�|jjd|j_|S)Nzexpected-eof-but-got-end-tagr<r8)rJr~rEr])r>r�r#r#r$r�I sz4getPhases..AfterAfterBodyPhase.processEndTagN)r5r6r7rGr�r�r�r�r`r�r�r#)rbr#r$�AfterAfterBodyPhase) srtcs�eZdZ�fdd�Zdd�Zdd�Zdd�Zd d �Zdd�Zd d�Z dd�Z dd�ZdS)z*getPhases..AfterAfterFramesetPhasecsP�j|||�tjd|jfd|jfg�|_|j|j_dS)Nr>r�)rGrr�r`�startTagNoFramesr^r�r�)r>rJr:)rbr#r$rGP s z3getPhases..AfterAfterFramesetPhase.__init__cSsdS)Nr#)r>r#r#r$r�Y sz5getPhases..AfterAfterFramesetPhase.processEOFcSs|jj||jj�dS)N)r:r\rc)r>r�r#r#r$r�\ sz9getPhases..AfterAfterFramesetPhase.processCommentcSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r�_ szAgetPhases..AfterAfterFramesetPhase.processSpaceCharacterscSs|jjd�dS)Nzexpected-eof-but-got-char)rJr~)r>r�r#r#r$r�b sz.AfterAfterFramesetPhase.processCharacterscSs|jjdj|�S)Nr8)rJrEr�)r>r�r#r#r$r`e sz7getPhases..AfterAfterFramesetPhase.startTagHtmlcSs|jjdj|�S)Nr�)rJrEr�)r>r�r#r#r$ruh sz;getPhases..AfterAfterFramesetPhase.startTagNoFramescSs"|jjdi|dd6�dS)Nzexpected-eof-but-got-start-tagr<)rJr~)r>r�r#r#r$r�k sz8getPhases..AfterAfterFramesetPhase.startTagOthercSs"|jjdi|dd6�dS)Nzexpected-eof-but-got-end-tagr<)rJr~)r>r�r#r#r$r�o sz8getPhases..AfterAfterFramesetPhase.processEndTagN)r5r6r7rGr�r�r�r�r`rur�r�r#)rbr#r$�AfterAfterFramesetPhaseO s rvrTrSr=r�r�r8rFr6r�r2r4r.r,r)r'r�rxr�r;rprkrr)r)rFrVrZr�r�r�r�r�r�r�r�r�r�r�r�r�rr rrjrlrqrsrtrvr#)rbr$rD�sd%)�#.cC��!�-GAbYLd's/9%&$rDroFcCs9|dkri}nit|d6|d6|d6|d6S)Nr,r<rsry)r)r<r,rgryr#r#r$r�� s r�c@seZdZdZdS)rrzError in parsed documentN)r5r6r7rKr#r#r#r$rr� srr)%� __future__rrrZpip._vendor.sixrr*rgrrr Ztreebuilders._baser rrr rrrrrrrrrrrr$rr'r9rrrDr�� Exceptionrrr#r#r#r$�sD �i��