o eZ@sdZddlZddlZddlZejddeddddlmZmZddlm 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.GdddZ/Gddde Z0GdddeZ1ddZ2d d!Z3d"d#Z4e5d$Z6d%d&Z7Gd'd(d(eZ8dS))z Shim module between Bleach and html5lib. This makes it easier to upgrade the html5lib library without having to change a lot of code. Nignorez"html5lib's sanitizer is deprecatedzbleach._vendor.html5lib)messagecategorymodule) HTMLParser getTreeWalker) constants) namespacesprefixes)_ReparseException)Filter)allowed_protocolsallowed_css_propertiesallowed_svg_propertiesattr_val_is_urisvg_attr_val_allows_refsvg_allow_local_href)HTMLInputStream)escapeHTMLSerializer) attributeMap HTMLTokenizer)TrieStartTagEndTagEmptyTag Characters ParseError)paabbraddressareaarticleasideaudiobbasebdibdo blockquotebodybrbuttoncanvascaptioncitecodecolcolgroupdatadatalistdddeldetailsdfndialogdivdldtemembedfieldset figcaptionfigurefooterformh1h2h3h4h5h6headheaderhgrouphrhtmliiframeimginputinskbdkeygenlabellegendlilinkmapmarkmenumetameternavnoscriptobjectoloptgroupoptionoutputpparampicturepreprogressqrprtrubyssampscriptsectionselectslotsmallsourcespanstrongstylesubsummarysuptabletbodytdtemplatetextareatfootththeadtimetitletrtrackuulvarvideowbr)!r r"r#r)r7r9r5r:r;r<r?r@rArBrCrDrErFrGrHrIrKrLrMrXmainr_rbrfrirrr}rc@sfeZdZdZddZeddZeddZedd Zd d Z dd dZ ddZ ddZ ddZ dS)InputStreamWithMemoryzWraps an HTMLInputStream to remember characters since last < This wraps existing HTMLInputStream classes to keep track of the stream since the last < which marked an open tag state. cCs$||_|jj|_|jj|_g|_dSN) _inner_streamresetposition_buffer)self inner_streamrG/var/www/html/venv/lib/python3.10/site-packages/bleach/html5lib_shim.py__init__s   zInputStreamWithMemory.__init__cC|jjSr)rerrorsrrrrrzInputStreamWithMemory.errorscCrr)r charEncodingrrrrrrz"InputStreamWithMemory.charEncodingcCrr)rchangeEncodingrrrrr rz$InputStreamWithMemory.changeEncodingcCs|j}|r |j||Sr)rcharrappend)rcrrrrs  zInputStreamWithMemory.charFcCs$|jj||d}|jt||S)N)opposite)r charsUntilrextendlist)r charactersrcharsrrrrsz InputStreamWithMemory.charsUntilcCs|jr |jd|j|S)N)rpoprunget)rrrrrrs  zInputStreamWithMemory.ungetcCs d|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. )joinrrrrrget_tags zInputStreamWithMemory.get_tagcCs dg|_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. Qsz/BleachHTMLTokenizer.__iter__..z!expected-closing-tag-but-got-charzeof-in-tag-namerr3)zeof-in-attribute-namez eof-in-attribute-value-no-quotes)r__iter__TAG_TOKEN_TYPESgetritemsparsertagslowerstriprrTAG_TOKEN_TYPE_CHARACTERSTAG_TOKEN_TYPE_PARSEERROR)rlast_error_tokentokenrrrrAsL             zBleachHTMLTokenizer.__iter__NcsL|jr t||S|r|jdddd7<dS|jtdddS)Nr3r&r)rr consumeEntity currentToken tokenQueuerr)r allowedChar fromAttributerrrrs z!BleachHTMLTokenizer.consumeEntitycs|jtSr)rrr tagOpenStaterrrrrs  z BleachHTMLTokenizer.tagOpenStatecs|j}|jjdurP|dtvrP|d|jjvrP|jjr4|jr1|dtkr1|dtvr1d}nd}n|j }t |d}||_|_|j ||j|_dS|j|_tdS)Nrname rr)rrrrrrrTAG_TOKEN_TYPE_STARTHTML_TAGS_BLOCK_LEVELrrrrr dataStatestateremitCurrentToken)rrnew_data new_tokenrrrrs&       z$BleachHTMLTokenizer.emitCurrentTokenr)NF) rrrrrrrrr __classcell__rrrrr3s [ rcs,eZdZdZfddZ d ddZZS) BleachHTMLParserz$Parser that uses BleachHTMLTokenizerc sB|dur tdd|Dnd|_||_||_tjdi|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) Ncss|]}|VqdSr)r)rtagrrrrsz,BleachHTMLParser.__init__..r) frozensetrrrrr)rrrrrrrrrs  zBleachHTMLParser.__init__Fr:TcKsh||_||_||_td||j|d||_|z|WdSty3||YdSw)N)rrrr) innerHTMLMode container scriptingrr tokenizerrmainLoopReparseException)rr innerHTMLrrrrrr_parses  zBleachHTMLParser._parse)Fr:T)rrrrrrrrrrrrs  rcCs|ddkrDt|dkrdS|ddvr|ddd}}n |ddd}}|d kr-dSt||}d|kr?d krBt|SdSdSt|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_pointrrrconvert_entitys    rcCsd|vr|Sg}t|D]6}|sq |dr=t|}|dur=t|}|dur=|||t|dd}|r<||q ||q 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_entityrrrr)textnew_textpartentity converted remainderrrrconvert_entities1s$      rcCs0|ddkr td|dd}t|}d}dtj}|rj|ddkrjd}|d|r<|dd vr       rz(&)ccsDtt|D]\}}|dkr|Vq|ddkrd|VqdS)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)rrOrrrrrs   rcs.eZdZdZdZddZdfdd ZZS) BleachHTMLSerializerz[HTMLSerializer that undoes & -> & in attributes and sets escape_rcdata to True Tccs|dd}t|D]6}|sq |dr:t|}|dur:t|dur:d|dV|t|dd}|r9|Vq |ddVq dS)z,Escapes just bare & in HTML attribute valuesz&rNr r)replacerrrrr)rstokenrrrrrescape_base_amps   z$BleachHTMLSerializer.escape_base_ampNc#sd}d}t||D]1}|r3|dkrd}n|r)|dkr(||EdHd}q n|dkr/d}|Vq |dr:d}|Vq dS)zWrap HTMLSerializer.serialize and conver & to & in attribute values Note that this converts & to & in attribute values where the & isn't already part of an unambiguous character entity. F>rN=Tr)r serializerr)r treewalkerencodingin_tag after_equalsrrrrrs( zBleachHTMLSerializer.serializer)rrrr escape_rcdatarrrrrrrrs  r)9rrer warningsfilterwarningsDeprecationWarningbleach._vendor.html5librrr!bleach._vendor.html5lib.constantsr r r r$bleach._vendor.html5lib.filters.baser )bleach._vendor.html5lib.filters.sanitizerr rrrrrSanitizerFilter$bleach._vendor.html5lib._inputstreamr"bleach._vendor.html5lib.serializerrr"bleach._vendor.html5lib._tokenizerrrbleach._vendor.html5lib._trierentitiesrr tokenTypesrrTAG_TOKEN_TYPE_ENDrrr HTML_TAGSrrrrrrrcompilerrrrrrrs^           y'?1,"# <