MhC\dZddlZddlZddlZejddedddlmZmZddlm Z dd l m Z m Z dd l m Zdd lmZdd lmZmZmZmZmZmZdd lmZdd lmZddlmZmZddlmZm Z ddl!m"Z"e j#Z$e"e$Z%e j&de j&de j&dhZ'e j&dZ(e j&dZ)e j&dZ*e j&dZ+e,dZ-e,dZ.GddZ/Gdde Z0GddeZ1dZ2dZ3d Z4ej5d!Z6d"Z7Gd#d$eZ8dS)%z Shim module between Bleach and html5lib. This makes it easier to upgrade the html5lib library without having to change a lot of code. Nignorez"html5lib's sanitizer is deprecatedzbleach._vendor.html5lib)messagecategorymodule) HTMLParser getTreeWalker) constants) namespacesprefixes)_ReparseException)Filter)allowed_protocolsallowed_css_propertiesallowed_svg_propertiesattr_val_is_urisvg_attr_val_allows_refsvg_allow_local_href)HTMLInputStream)escapeHTMLSerializer) attributeMap HTMLTokenizer)TrieStartTagEndTagEmptyTag Characters ParseError)paabbraddressareaarticleasideaudiobbasebdibdo blockquotebodybrbuttoncanvascaptioncitecodecolcolgroupdatadatalistdddeldetailsdfndialogdivdldtemembedfieldset figcaptionfigurefooterformh1h2h3h4h5h6headheaderhgrouphrhtmliiframeimginputinskbdkeygenlabellegendlilinkmapmarkmenumetameternavnoscriptobjectoloptgroupoptionoutputpparampicturepreprogressqrprtrubyssampscriptsectionselectslotsmallsourcespanstrongstylesubsummarysuptabletbodytdtemplatetextareatfootththeadtimetitletrtrackuulvarvideowbr)!r!r#r$r*r8r:r6r;r<r=r@rArBrCrDrErFrGrHrIrJrLrMrNrYmainr`rcrgrjrsr~rczeZdZdZdZedZedZedZdZ d dZ d Z d Z d Z d S)InputStreamWithMemoryzWraps 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. cf||_|jj|_|jj|_g|_dSN) _inner_streamresetposition_buffer)self inner_streams T/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/bleach/html5lib_shim.py__init__zInputStreamWithMemory.__init__s/)'- *3  c|jjSr)rerrorsrs rrzInputStreamWithMemory.errorss!((rc|jjSr)r charEncodingrs rrz"InputStreamWithMemory.charEncodings!..rc|jjSr)rchangeEncodingrs rrz$InputStreamWithMemory.changeEncoding s!00rcp|j}|r|j||Sr)rcharrappend)rcs rrzInputStreamWithMemory.chars9   # # % %  # L   " " "rFc|j||}|jt ||S)N)opposite)r charsUntilrextendlist)r charactersrcharss rrz InputStreamWithMemory.charsUntils>"--j8-LL DKK((( rcx|jr|jd|j|S)N)rpoprunget)rrs rrzInputStreamWithMemory.ungets9 < ! L  R !''---rc6d|jS)zReturns 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. )joinrrs rget_tagzInputStreamWithMemory.get_tagswwt|$$$rcdg|_dS)zResets stream history to just '<' This gets called by tagOpenState() which marks a '<' that denotes an open tag. Any time we see that, we reset the buffer. z/BleachHTMLTokenizer.__iter__..QsZ111Izy00 #9 4 4 #9 4 4 #J/ !5 4 4 4 11rz!expected-closing-tag-but-got-charzeof-in-tag-namerr4)zduplicate-attributezeof-in-attribute-namez eof-in-attribute-value-no-quoteszexpected-end-of-tag-but-got-eof)r__iter__TAG_TOKEN_TYPESgetritemsparsertagslowerstriprrTAG_TOKEN_TYPE_CHARACTERSTAG_TOKEN_TYPE_PARSEERROR)rlast_error_tokentokenrs rrzBleachHTMLTokenizer.__iter__AsHWW%%''B B E+$V,0UUUf 88 &))9%1115:6]5H5H5J5J111%%E&M(,$KKKK%V,0SSS (4f ++--3355T[=MMM%)K$7$7$9$9E&M$=E&M'+$KKKK6]&???+***',$$+***KKK'+$V} 999#( KKKK  ''+<<< 9$+BUBUBWBWXXXXXXX!&). 9$+BUBUBWBWXXXXXXX'&&&&&7 ' 'rNc|jr"t||S|r#|jdddxxdz cc<dS|jt dddS)Nr4r&r)rr consumeEntity currentToken tokenQueuerr)r allowedChar fromAttributers rrz!BleachHTMLTokenizer.consumeEntitys   E77((mDD D  U  f %b )! , , , 3 , , , , , O " ",Es#S#S T T T T Trcv|jtSr)rrr tagOpenState)rrs rrz BleachHTMLTokenizer.tagOpenStates/ ww##%%%rc8|j}|jj|dtvr|d|jjvr|jjr?|jr5|dtkr$|dtvrd}nd}n|j }t|d}|x|_|_|j ||j|_dS|j|_t!dS)Nrname rr)rrrrrrrTAG_TOKEN_TYPE_STARTHTML_TAGS_BLOCK_LEVELrrrrr dataStatestateremitCurrentToken)rrnew_data new_tokenrs rrz$BleachHTMLTokenizer.emitCurrentTokens! K  (f 00f ##%%T[-=== {  1+ "f )===f ++--1FFF $HH "HH ;..00!:HMMI:C CD  7 O " "9 - - -DJ F"&"3   """""rr)NF) rrrrrrrrr __classcell__rs@rrr3s;; ' ' ' ' ' 'b'b'b'b'b'HUUUUUU$&&&&&*#*#*#*#*#*#*#*#*#rrc,eZdZdZfdZ ddZxZS)BleachHTMLParserz$Parser that uses BleachHTMLTokenizerc |td|Dnd|_||_||_t jdi|dS)a :arg tags: set of allowed tags--everything else is either stripped or escaped; if None, then this doesn't look at tags at all :arg strip: whether to strip disallowed tags (True) or escape them (False); if tags=None, then this doesn't have any effect :arg consume_entities: whether to consume entities (default behavior) or leave them as is when tokenizing (BleachHTMLTokenizer-added behavior) Nc3>K|]}|VdSr)r)rtags rrz,BleachHTMLParser.__init__..s*33ssyy{{333333rr) frozensetrrrrr)rrrrrrs rrzBleachHTMLParser.__init__sd9=8HI33d333 4 4 4d  0""6"""""rFr;Tc (||_||_||_td||j|d||_| |dS#t$r,||YdSwxYw)N)rrrr) innerHTMLMode container scriptingrr tokenizerrmainLoopReparseException)rr innerHTMLrrrs r_parsezBleachHTMLParser._parses'"", D,A$  RX     MMOOOOO    JJLLL MMOOOOOO sA2BB)Fr;T)rrrrrrrrs@rrrsY..#####$CGrrc2|ddkrqt|dkrdS|ddvr |ddd}}n |ddd}}|d krdSt||}d|cxkrd krnnt|SdSt|dS) a9Convert 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 r#NrxX ri)lenintchrENTITIESr)value int_as_stringr' code_points rconvert_entityrs Qx3 u::>>4 8z ! !"')R4MM#()R4M B  4-- z $ $ $ $H $ $ $ $ $z?? "4 <<t $ $$rcd|vr|Sg}t|D]}|s|drit|}|Xt|}|G|||t |dzd}|r||||d|S)zConverts all found entities in the text :arg text: the text to convert entities in :returns: unicode text with converted entities rNrr)next_possible_entity startswith match_entityrrr r)textnew_textpartentity converted remainders rconvert_entitiesr:s $ H$T**   ??3   !$''F!*622 (OOI... $S[[1_%6%6 7I3  222 778  rc|ddkrtd|dd}t|}d}dtjz}|r|ddkrd}|d|r%|dd vrd }||dz }nd }|r5|d|vr+|d}||vrn||z }|r |d|v+|r|r|dd kr|SdS|rL|d|vrB|d}||z }t |sdS|r |d|vB|r|r|dd kr|SdS) avReturns first entity in stream or None if no entity exists Note: For Bleach purposes, entities must start with a "&" and end with a ";". This ignores ambiguous character entities that have no ";" at the end. :arg stream: the character stream :returns: the entity string without "&" or ";" if it's a valid character entity; ``None`` otherwise rrzStream should begin with "&"rNrz<&=;rr0123456789abcdefABCDEF 0123456789;) ValueErrorrstring whitespacer ENTITIES_TRIEhas_keys_with_prefix)rpossible_entityend_charactersallowedrs rrr]sayC7888 ABBZF &\\FOf//N&)s"" 1  #fQi:--.G vzz!}} ,OO"G !.88 1 A q O  !.88  #v #&)s*:*:" "t VAYn44 JJqMM111/BB 4 VAYn446fQi3&6&6 4rz(&)c#Ktt|D] \}}|dkr|V|dzdkrd|zV!dS)zTakes 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 "&" rrrN) enumerate AMP_SPLIT_REsplit)rrPrs rrrsm\//55664 66JJJJ UaZZ*    rc.eZdZdZdZdZdfd ZxZS)BleachHTMLSerializerz[HTMLSerializer that undoes & -> & in attributes and sets escape_rcdata to True Tc#LK|dd}t|D]{}|s|drIt|}|8t |)d|dV|t |dzd}|r|Vc|ddV|dS)z,Escapes just bare & in HTML attribute valuesz&rNr#r)replacerrrrr )rstokenrrs rescape_base_ampz$BleachHTMLSerializer.escape_base_amps--)00 - -D s## %d++%.*@*@*L'f---''' F a 1 12D#" ,,sG,, , , , ,% - -rNc#Kd}d}t||D]\}|r=|dkrd}n/|r%|dkr||Ed{Vd}3n|dkrd}|VA|drd}|V]dS)zWrap 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. F>rN=Tr)r serializer5r)r treewalkerencodingin_tag after_equalsr4rs rr9zBleachHTMLSerializer.serializes gg'' H==  F S=="FF!(}}#'#7#7#?#????????',  % s]]#'L $$S))"!F '  rr)rrrr escape_rcdatar5r9rrs@rr1r1s]M--->rr1)9rrer%warningsfilterwarningsDeprecationWarningbleach._vendor.html5librrr !bleach._vendor.html5lib.constantsr r r r$bleach._vendor.html5lib.filters.baser )bleach._vendor.html5lib.filters.sanitizerrrrrrrSanitizerFilter$bleach._vendor.html5lib._inputstreamr"bleach._vendor.html5lib.serializerrr"bleach._vendor.html5lib._tokenizerrrbleach._vendor.html5lib._trierentitiesrr' tokenTypesrrTAG_TOKEN_TYPE_ENDrrr HTML_TAGSrrrrrrrcompiler.rr1rrrrQs   0  $   X $ " $ !+J7)(3%0>%0> Iqs s  r" "$$N<<<<<<<<~v#v#v#v#v#-v#v#v#r)))))z)))X%%%D   F999xrz%      IIIII>IIIIIr