ó ;×4\cp@s4dZddlmZddlZddlZddlZddlmZmZddlm Z ddl m Z m Z ddl m ZddlmZdd lmZddlmZdd lmZdd lmZdd lmZdd lmZe jZeeƒZee j de j de j dgƒZ!e j dZ"e j dZ#dddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1d2d3d4d5d6d7d8d9d:d;d<d=d>d?d@dAdBdCdDdEdFdGdHdIdJdKdLdMdNdOdPdQdRdSdTdUdVdWdXdYdZd[d\d]d^d_d`dadbdcdddedfdgdhdidjdkdldmdndodpdqdrdsdtdudvdwdxdydzd{d|d}d~dd€dd‚gpZ$dƒe%fd„„ƒYZ&d…efd†„ƒYZ'd‡efdˆ„ƒYZ(d‰„Z)dŠ„Z*d‹„Z+ej,dŒƒZ-d„Z.dŽefd„ƒYZ/dS(u‡ Shim module between Bleach and html5lib. This makes it easier to upgrade the html5lib library without having to change a lot of code. iÿÿÿÿ(tunicode_literalsN(t HTMLParsert getTreeWalker(t constants(t namespacestprefixes(t_ReparseException(tFilter(tallowed_protocols(tHTMLInputStream(tHTMLSerializer(t HTMLTokenizer(tTrieuStartTaguEndTaguEmptyTagu Charactersu ParseErroruauabbruaddressuareauarticleuasideuaudioububaseubdiubdou blockquoteubodyubrubuttonucanvasucaptionuciteucodeucolucolgroupudataudatalistuddudeludetailsudfnudialogudivudludtuemuembedufieldsetu figcaptionufigureufooteruformuh1uh2uh3uh4uh5uh6uheaduheaderuhgroupuhruhtmluiuiframeuimguinputuinsukbdukeygenulabelulegenduliulinkumapumarkumenuumetaumeterunavunoscriptuobjectuoluoptgroupuoptionuoutputupuparamupictureupreuprogressuqurpurturubyususampuscriptusectionuselectuslotusmallusourceuspanustrongustyleusubusummaryusuputableutbodyutdutemplateutextareautfootuthutheadutimeutitleutrutrackuuuuluvaruvideouwbrtInputStreamWithMemorycBsteZdZd„Zed„ƒZed„ƒZed„ƒZd„Ze d„Z d„Z d„Z d „Z RS( uÄWraps an HTMLInputStream to remember characters since last < This wraps existing HTMLInputStream classes to keep track of the stream since the last < which marked an open tag state. cCs4||_|jj|_|jj|_g|_dS(N(t _inner_streamtresettpositiont_buffer(tselft inner_stream((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyt__init__®s cCs |jjS(N(Rterrors(R((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyR´scCs |jjS(N(Rt charEncoding(R((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyR¸scCs |jjS(N(RtchangeEncoding(R((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyR¼scCs,|jjƒ}|r(|jj|ƒn|S(N(RtcharRtappend(Rtc((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyRÀscCs2|jj|d|ƒ}|jjt|ƒƒ|S(Ntopposite(Rt charsUntilRtextendtlist(Rt charactersRtchars((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyRÇscCs,|jr|jjdƒn|jj|ƒS(Niÿÿÿÿ(RtpopRtunget(RR((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyR"Ìs cCstjdƒj|jƒS(uþReturns the stream history since last '<' Since the buffer starts at the last '<' as as seen by tagOpenState(), we know that everything from that point to when this method is called is the "tag" that is being tokenized. u(tsixt text_typetjoinR(R((s3lib/python2.7/site-packages/bleach/html5lib_shim.pytget_tagÑscCsdg|_dS(u¶Resets stream history to just '<' This gets called by tagOpenState() which marks a '<' that denotes an open tag. Any time we see that, we reset the buffer. u  "    cCsh|jr"tt|ƒj||ƒS|rF|jdddcd7 tBleachHTMLParsercBs)eZdZd„Zeded„ZRS(u$Parser that uses BleachHTMLTokenizercKs`|dk r+g|D]}|jƒ^qnd|_||_||_tt|ƒj|dS(uÐ :arg tags: list of allowed tags--everything else is either stripped or escaped; if None, then this doesn't look at tags at all :arg strip: whether to strip disallowed tags (True) or escape them (False); if tags=None, then this doesn't have any effect :arg consume_entities: whether to consume entities (default behavior) or leave them as is when tokenizing (BleachHTMLTokenizer-added behavior) N(R2R6R9R7R/R.RJR(RR9R7R/R1ttag((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyRls 4  udivcKsƒ||_||_||_td|d|jd|||_|jƒy|jƒWn%tk r~|jƒ|jƒnXdS(NR0R/R8( t innerHTMLModet containert scriptingR-R/t tokenizerRtmainLooptReparseException(RR0t innerHTMLRMRNR1((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyt_parse{s        (R(R)R*RR,RS(((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyRJjs cCsd|ddkrT|dd kr:tjt|ddƒƒStjt|ddƒƒStj|d ƒS( u9Convert an entity (minus the & and ; part) into what it represents This handles numeric, hex, and text entities. :arg value: the string (minus the ``&`` and ``;`` part) to convert :returns: unicode character or None if it's an ambiguous ampersand that doesn't match a character entity iu#iuxuXiii (uxuXN(R#tunichrtinttENTITIESR5R2(tvalue((s3lib/python2.7/site-packages/bleach/html5lib_shim.pytconvert_entitys cCsÕd|kr|Sg}x¯t|ƒD]¡}|s5q#n|jdƒr·t|ƒ}|dk r·t|ƒ}|dk r´|j|ƒ|t|ƒd}|r#|j|ƒq#q#q´q·n|j|ƒq#Wdj|ƒS(u‘Converts all found entities in the text :arg text: the text to convert entities in :returns: unicode text with converted entities u&iuN(tnext_possible_entityt startswitht match_entityR2RXRtlenR%(ttexttnew_texttparttentityt convertedt remainder((s3lib/python2.7/site-packages/bleach/html5lib_shim.pytconvert_entities¢s$       cCs€|ddkrtdƒ‚n|d}t|ƒ}d}dtj}|r|ddkrd}|jdƒ|r£|dd kr£d }||jdƒ7}nd }xC|rî|d|krî|jdƒ}||kráPn||7}q¬W|r|r|dd kr|SdSxF|r[|d|kr[|jdƒ}tj|ƒsNPn||7}qW|r||r||dd kr||SdS(uHReturns first entity in stream or None if no entity exists Note: For Bleach purposes, entities must start with a "&" and end with a ";". This ignoresambiguous character entities that have no ";" at the end. :arg stream: the character stream :returns: ``None`` or the entity string without "&" or ";" iu&uStream should begin with "&"iuu<&=;u#uxuXu0123456789abcdefABCDEFu 0123456789u;(uxuXN(t ValueErrorRtstringt whitespaceR!R2t ENTITIES_TRIEthas_keys_with_prefix(R0tpossible_entitytend_characterstallowedR((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyR[Ås:      u(&)ccsZxSttj|ƒƒD]<\}}|dkr6|Vq|ddkrd|VqqWdS(u·Takes a text and generates a list of possible entities :arg text: the text to look at :returns: generator where each part (except the first) starts with an "&" iiu&N(t enumeratet AMP_SPLIT_REtsplit(R]tiR_((s3lib/python2.7/site-packages/bleach/html5lib_shim.pyRYs " tBleachHTMLSerializercBs#eZdZd„Zdd„ZRS(u3HTMLSerializer that undoes & -> & in attributesccs»|jddƒ}x¢t|ƒD]”}|s1qn|jdƒr¢t|ƒ}|dk r¢t|ƒdk r¢d|dV|t|ƒd}|r|Vqqq¢n|jddƒVqWdS(u,Escapes just bare & in HTML attribute valuesu&u&u;iN(treplaceRYRZR[R2RXR\(RtstokenR_R`((s3lib/python2.7/site-packages/bleach/html5lib_shim.pytescape_base_amps   ccsËt}t}x¸tt|ƒj||ƒD]›}|r¦|dkrIt}nU|r‰|dkržx|j|ƒD] }|VqkWt}q(qžn|dkržt}n|Vq(|jdƒr¾t}n|Vq(WdS(uÜWrap HTMLSerializer.serialize and conver & to & in attribute values Note that this converts & to & in attribute values where the & isn't already part of an unambiguous character entity. u>u"u=us           >…%  # :