fZdZddlZddlZddlZejddedddlmZmZddlm Z dd l m Z m Z dd l m Zdd lmZdd lmZmZmZmZmZmZdd lmZdd lmZddlmZmZddlmZm Z ddl!m"Z"e jFZ$e"e$Z%e jLde jLde jLdhZ'e jLdZ(e jLdZ)e jLdZ*e jLdZ+e,dZ-e,dZ.GddZ/Gdde Z0GddeZ1dZ2dZ3d Z4ejjd!Z6d"Z7Gd#d$eZ8y)%z Shim module between Bleach and html5lib. This makes it easier to upgrade the html5lib library without having to change a lot of code. Nignorez"html5lib's sanitizer is deprecatedzbleach._vendor.html5lib)messagecategorymodule) HTMLParser getTreeWalker) constants) namespacesprefixes)_ReparseException)Filter)allowed_protocolsallowed_css_propertiesallowed_svg_propertiesattr_val_is_urisvg_attr_val_allows_refsvg_allow_local_href)HTMLInputStream)escapeHTMLSerializer) attributeMap HTMLTokenizer)TrieStartTagEndTagEmptyTag Characters ParseError)paabbraddressareaarticleasideaudiobbasebdibdo blockquotebodybrbuttoncanvascaptioncitecodecolcolgroupdatadatalistdddeldetailsdfndialogdivdldtemembedfieldset figcaptionfigurefooterformh1h2h3h4h5h6headheaderhgrouphrhtmliiframeimginputinskbdkeygenlabellegendlilinkmapmarkmenumetameternavnoscriptobjectoloptgroupoptionoutputpparampicturepreprogressqrprtrubyssampscriptsectionselectslotsmallsourcespanstrongstylesubsummarysuptabletbodytdtemplatetextareatfootththeadtimetitletrtrackuulvarvideowbr)!r!r#r$r*r8r:r6r;r<r=r@rArBrCrDrErFrGrHrIrJrLrMrNrYmainr`rcrgrjrsr~rcfeZdZdZdZedZedZedZdZ d dZ dZ d Z d Z y ) 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. c||_|jj|_|jj|_g|_yN) _inner_streamresetposition_buffer)self inner_streams O/var/www/cvtools/html/venv/lib/python3.12/site-packages/bleach/html5lib_shim.py__init__zInputStreamWithMemory.__init__s7)''-- **33  c.|jjSr)rerrorsrs rrzInputStreamWithMemory.errorss!!(((rc.|jjSr)r charEncodingrs rrz"InputStreamWithMemory.charEncodings!!...rc.|jjSr)rchangeEncodingrs rrz$InputStreamWithMemory.changeEncoding s!!000rct|jj}|r|jj||Sr)rcharrappend)rcs rrzInputStreamWithMemory.chars0    # # % LL   "rc|jj||}|jjt ||S)N)opposite)r charsUntilrextendlist)r charactersrcharss rrz InputStreamWithMemory.charsUntils8""--j8-L DK( rc|jr|jjd|jj|S)N)rpoprunget)rrs rrzInputStreamWithMemory.ungets2 << LL  R !!''--rc8dj|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|_y)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. |jdr-t d|dj D|d<d}|n|ddk(rz|j jd|djj|j jvr-|jj|d<t|d<d}|n|dtk(r||}n ||d}|dtk(r|}| |r\|ddk(r$t|jjdy|ddvr$t|jjdy|yyw) Nr4z#invalid-character-in-attribute-nametypec3BK|]\}}d|vrd|vr d|vr||fyw)"'rNr).0 attr_name attr_values r z/BleachHTMLTokenizer.__iter__..Qs911Izy0 #9 4 #9 4 #J/1sz!expected-closing-tag-but-got-charzeof-in-tag-namerr4)zeof-in-attribute-namez eof-in-attribute-value-no-quotes)r__iter__TAG_TOKEN_TYPESgetritemsparsertagslowerstriprrTAG_TOKEN_TYPE_CHARACTERSTAG_TOKEN_TYPE_PARSEERROR)rlast_error_tokentokenrs rrzBleachHTMLTokenizer.__iter__AsW%'B E+$V,0UUf 8 &)%115:6]5H5H5J1%E&M(,$K%V,0SS ((4f ++-335T[[=M=MM%)KK$7$7$9E&M$=E&M'+$K6]&??+*',$+*K'+$V} 99#( KEB H '+<< 9$++BUBUBWXX!&). 9$++BUBUBWXX&&% sE>Fc|jrt| ||S|r|jdddxxdz cc<y|jj t ddy)Nr4r&r)rr consumeEntity currentToken tokenQueuerr)r allowedChar fromAttributers rrz!BleachHTMLTokenizer.consumeEntitys_  7(mD D    f %b )! , 3 , OO " ",Es#S TrcT|jjt| Sr)rrr tagOpenState)rrs rrz BleachHTMLTokenizer.tagOpenStates# w#%%rcF|j}|jj|dtvr|dj |jjvr|jj r7|j r(|dtk(r|dj tvrd}nd}n|jj}t|d}|x|_|_|jj||j|_y|j|_t |Ey)Nrname rr)rrrrrrrTAG_TOKEN_TYPE_STARTHTML_TAGS_BLOCK_LEVELrrrrr dataStatestateremitCurrentToken)rrnew_data new_tokenrs rrz$BleachHTMLTokenizer.emitCurrentTokens!! KK   (f 0f ##%T[[-=-== {{  ++f )==f ++-1FF $H "H ;;..0!:HMI:C CD  7 OO " "9 -DJ "&"3"3  "rr)NF) rrrrrrrrr __classcell__rs@rrr3s'; 'Y'vU$&*#*#rrc,eZdZdZfdZ ddZxZS)BleachHTMLParserz$Parser that uses BleachHTMLTokenizerc v|td|Dnd|_||_||_t |di|y)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|]}|jywr)r)rtags rrz,BleachHTMLParser.__init__..s3ssyy{3sr) frozensetrrrrr)rrrrrrs rrzBleachHTMLParser.__init__sB9=8HI3d3 4d  0 "6"rc ||_||_||_td||j|d||_|j  |jy#t$r#|j |jYywxYw)N)rrrr) innerHTMLMode container scriptingrr tokenizerrmainLoopReparseException)rr innerHTMLrrrs r_parsezBleachHTMLParser._parsesz'"", D,A,A$ RX    MMO  JJL MMO sA)BB)Fr;T)rrrrrrrrs@rrrs.#$CGrrc|ddk(rPt|dkry|ddvr|ddd}}n|ddd}}|d k(ryt||}d|cxkrd kr t|Syytj |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:> 8z !"')R4M#()R4M B - z $H $z? " % <<t $$rc>d|vr|Sg}t|D]w}|s|jdrPt|}|Ct|}|6|j ||t |dzd}|r|j |g|j |ydj |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_entitiesr1s $ H$T*  ??3 !$'F!*62 (OOI. $S[1_%6 7I  2%( 778 rc|ddk7r td|dd}t|}d}dtjz}|rz|ddk(rrd}|j d|r|dd vrd }||j dz }nd }|r,|d|vr%|j d}||vrn||z }|r|d|vr%|r |r |dd k(r|Sy|r=|d|vr6|j d}||z }t j |sy|r|d|vr6|r |r |dd k(r|Sy) avReturns first entity in stream or None if no entity exists Note: For Bleach purposes, entities must start with a "&" and end with a ";". This ignores ambiguous character entities that have no ";" at the end. :arg stream: the character stream :returns: the entity string without "&" or ";" if it's a valid character entity; ``None`` otherwise rrzStream should begin with "&"rNrz<&=;rr0123456789abcdefABCDEF 0123456789;) ValueErrorrstring whitespacer ENTITIES_TRIEhas_keys_with_prefix)rpossible_entityend_charactersallowedrs rrrTsQayC788 ABZF &\FOf///N&)s" 1 fQi:-.G vzz!} ,O"G.8 1 A q O .8 v&)s*:" " VAYn4 JJqM111/B VAYn46fQi3&6 rz(&)c#Kttj|D]\}}|dk(r||dzdk(sd|z!yw)zTakes a text and generates a list of possible entities :arg text: the text to look at :returns: generator where each part (except the first) starts with an "&" rrrN) enumerate AMP_SPLIT_REsplit)rrPrs rrrsL\//564 6J UaZ*  s 7A Ac.eZdZdZdZdZdfd ZxZS)BleachHTMLSerializerz[HTMLSerializer that undoes & -> & in attributes and sets escape_rcdata to True Tc#K|jdd}t|D]b}|s|jdr8t|}|+t | d|d|t |dzd}|r|O|jdddyw)z,Escapes just bare & in HTML attribute valuesz&rNr#r)replacerrrrr )rstokenrrs rescape_base_ampz$BleachHTMLSerializer.escape_base_amps-)0 -Ds#%d+%.*@*LfXQ-' F a 12D" ,,sG, ,% -sBBc#Kd}d}t|||D]R}|r7|dk(rd}n*|r!|dk7r#|j|Ed{d}0|dk(rd}|<|jdrd}|Ty7-w)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 g' H= FS="F!}#'#7#7#???',  s]#'L $$S)!F ' @s=A0A..A0r)rrrr escape_rcdatar5r9rrs@rr1r1sM->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%00>%00> qs  r""$N<<~m#-m#`)z)X%D F9xrzz%   I>Ir