B 1dX@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$Z6d%d&Z7Gd'd(d(eZ8dS))z Shim module between Bleach and html5lib. This makes it easier to upgrade the html5lib library without having to change a lot of code. Nignorez"html5lib's sanitizer is deprecatedzbleach._vendor.html5lib)messagecategorymodule) HTMLParser getTreeWalker) constants) namespacesprefixes)_ReparseException)Filter)allowed_protocolsallowed_css_propertiesallowed_svg_propertiesattr_val_is_urisvg_attr_val_allows_refsvg_allow_local_href)HTMLInputStream)escapeHTMLSerializer) attributeMap HTMLTokenizer)TrieZStartTagZEndTagZEmptyTagZ CharactersZ ParseError)paZabbraddressZareaarticleasideZaudiobbaseZbdiZbdo blockquotebodybrZbuttonZcanvascaptionZcitecodecolZcolgroupdataZdatalistdddeldetailsdfndialogdivdldtZemZembedfieldset figcaptionfigurefooterformh1h2h3h4h5h6headheaderhgrouphrhtmliZiframeZimginputZinsZkbdZkeygenlabelZlegendlilinkmapmarkZmenumetaZmeternavZnoscriptobjectolZoptgroupoptionoutputpparamZpicturepreprogressqrprtrubysZsampscriptsectionselectZslotZsmallsourcespanstrongstylesubsummarysuptableZtbodytdtemplateZtextareaZtfootthZtheadtimetitletrtrackuulvarZvideoZwbr)!rrrrr(r*r&r+r,r-r.r/r0r1r2r3r4r5r6r7r8r:r;r<rAmainrFrHrKrMrUr^rgc@sfeZdZdZddZeddZeddZedd Zd d Z dd dZ ddZ ddZ ddZ dS)InputStreamWithMemoryzWraps an HTMLInputStream to remember characters since last < This wraps existing HTMLInputStream classes to keep track of the stream since the last < which marked an open tag state. cCs$||_|jj|_|jj|_g|_dS)N) _inner_streamresetposition_buffer)selfZ inner_streamrpa/work/yifan.wang/ringdown/master-ringdown-env/lib/python3.7/site-packages/bleach/html5lib_shim.py__init__s  zInputStreamWithMemory.__init__cCs|jjS)N)rkerrors)rorprprqrsszInputStreamWithMemory.errorscCs|jjS)N)rk charEncoding)rorprprqrtsz"InputStreamWithMemory.charEncodingcCs|jjS)N)rkchangeEncoding)rorprprqru sz$InputStreamWithMemory.changeEncodingcCs|j}|r|j||S)N)rkcharrnappend)rocrprprqrvs  zInputStreamWithMemory.charFcCs$|jj||d}|jt||S)N)opposite)rk charsUntilrnextendlist)ro charactersrycharsrprprqrzsz InputStreamWithMemory.charsUntilcCs|jr|jd|j|S)N)rnpoprkunget)rorvrprprqrs 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. )joinrn)rorprprqget_tagszInputStreamWithMemory.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. Rsz/BleachHTMLTokenizer.__iter__..z!expected-closing-tag-but-got-charzeof-in-tag-namername)rr%)r__iter__TAG_TOKEN_TYPESgetritemsparsertagslowerstriprrTAG_TOKEN_TYPE_CHARACTERSTAG_TOKEN_TYPE_PARSEERROR currentToken)roZlast_error_tokentoken)rrprqrAsD         zBleachHTMLTokenizer.__iter__NcsJ|jrt||S|r4|jdddd7<n|jtdddS)Nr%r&)rr%)rr consumeEntityr tokenQueuerwr)roZ allowedCharZ fromAttribute)rrprqrs z!BleachHTMLTokenizer.consumeEntitycs|jtS)N)rrr tagOpenState)ro)rrprqrs z BleachHTMLTokenizer.tagOpenStatecs|j}|jjdk r|dtkr|d|jjkr|jjrh|jrb|dtkrb|dtkrbd}qrd}n |j }t |d}||_|_|j ||j|_dS|j|_tdS)Nrr r)rr%)rrrrrrrTAG_TOKEN_TYPE_STARTHTML_TAGS_BLOCK_LEVELrrrrrwZ dataStatestateremitCurrentToken)rorZnew_dataZ new_token)rrprqrs$       z$BleachHTMLTokenizer.emitCurrentToken)F)NF) rrrrrrrrrr __classcell__rprp)rrqr3s  T rcs*eZdZdZfddZd ddZZS) BleachHTMLParserz$Parser that uses BleachHTMLTokenizerc s>|dk rtdd|Dnd|_||_||_tjf|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|]}|VqdS)N)r)rtagrprprqrsz,BleachHTMLParser.__init__..) frozensetrrrrrr)rorrrr)rrprqrrs zBleachHTMLParser.__init__Fr+TcKsj||_||_||_tf||j|d||_|y |Wn$tk rd||YnXdS)N)rrr) Z innerHTMLMode container scriptingrrZ tokenizerrlZmainLoopReparseException)rorZ innerHTMLrrrrprprq_parses  zBleachHTMLParser._parse)Fr+T)rrrrrrrrrprp)rrqrs rcCs|ddkrt|dkrdS|ddkr<|ddd}}n|ddd}}|d krZdSt||}d|krxd krnnt|SdSt|dS) a9Convert an entity (minus the & and ; part) into what it represents This handles numeric, hex, and text entities. :arg value: the string (minus the ``&`` and ``;`` part) to convert :returns: unicode character or None if it's an ambiguous ampersand that doesn't match a character entity r#Nr)xX ri)lenintchrENTITIESr)valueZ int_as_stringr code_pointrprprqconvert_entitys    rcCsd|kr |Sg}xxt|D]l}|s$q|dr|t|}|dk r|t|}|dk r||||t|dd}|r||q||qWd|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_entityrrwrr)textnew_textpartentityZ converted remainderrprprqconvert_entities*s$   rcCs"|ddkrtd|dd}t|}d}dtj}|r|ddkrd}|d|rx|dd krxd }||d7}nd }x0|r|d|kr|d}||krP||7}q~W|r|r|dd kr|SdSx4|r|d|kr|d}||7}t|sdSqW|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<&=;r)rrZ0123456789abcdefABCDEF 0123456789;) ValueErrorr|string whitespacer ENTITIES_TRIEZhas_keys_with_prefix)rZpossible_entityZend_charactersallowedrxrprprqrMs:        rz(&)ccsFx@tt|D].\}}|dkr(|Vq|ddkrd|VqWdS)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)rr>rrprprqrs  rcs.eZdZdZdZddZdfdd ZZS) BleachHTMLSerializerz[HTMLSerializer that undoes & -> & in attributes and sets escape_rcdata to True Tccs|dd}xxt|D]l}|s q|drtt|}|dk rtt|dk rtd|dV|t|dd}|r|Vq|ddVqWdS)z,Escapes just bare & in HTML attribute valuesz&rNrr)replacerrrrr)rostokenrrrprprqescape_base_amps  z$BleachHTMLSerializer.escape_base_ampNc#sd}d}xtt||D]b}|rf|dkr.d}n0|rR|dkr^||EdHd}qn |dkr^d}|Vq|drtd}|VqWdS)zWrap HTMLSerializer.serialize and conver & to & in attribute values Note that this converts & to & in attribute values where the & isn't already part of an unambiguous character entity. F>rN=Tr)r serializerr)roZ treewalkerencodingin_tagZ after_equalsr)rrprqrs" zBleachHTMLSerializer.serialize)N)rrrrZ escape_rcdatarrrrprp)rrqrsr)9rrerwarningsfilterwarningsDeprecationWarningZbleach._vendor.html5librrrZ!bleach._vendor.html5lib.constantsr r r rZ$bleach._vendor.html5lib.filters.baser Z)bleach._vendor.html5lib.filters.sanitizerr rrrrrZSanitizerFilterZ$bleach._vendor.html5lib._inputstreamrZ"bleach._vendor.html5lib.serializerrrZ"bleach._vendor.html5lib._tokenizerrrZbleach._vendor.html5lib._trierentitiesrrZ tokenTypesrrZTAG_TOKEN_TYPE_ENDrrrZ HTML_TAGSrrjrrrrrcompilerrrrprprprqsT            w%?*,"#<