U mu fZ@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$Z6d%d&Z7Gd'd(d(eZ8dS))z Shim module between Bleach and html5lib. This makes it easier to upgrade the html5lib library without having to change a lot of code. Nignorez"html5lib's sanitizer is deprecatedzbleach._vendor.html5lib)messagecategorymodule) HTMLParser getTreeWalker) constants) namespacesprefixes)_ReparseException)Filter)allowed_protocolsallowed_css_propertiesallowed_svg_propertiesattr_val_is_urisvg_attr_val_allows_refsvg_allow_local_href)HTMLInputStream)escapeHTMLSerializer) attributeMap HTMLTokenizer)TrieStartTagEndTagEmptyTag Characters ParseError)paZabbraddressareaarticleasideaudiobbaseZbdiZbdo blockquotebodybrbuttonZcanvascaptionZcitecodecolcolgroupdataZdatalistdddeldetailsdfndialogdivdldtemembedfieldset figcaptionfigurefooterformh1h2h3h4h5h6headheaderhgrouphrhtmliiframeimginputZinsZkbdkeygenlabelZlegendlilinkmapmarkmenumetaZmeternavnoscriptobjectoloptgroupoptionoutputpparamZpicturepreprogressqrprtrubysZsampscriptsectionselectZslotsmallsourcespanstrongstylesubsummarysuptabletbodytdtemplatetextareatfootththeadtimetitletrtrackuulvarvideowbr)!rr!r"r&r1r3r/r4r5r6r9r:r;r<r=r>r?r@rArBrCrErFrGrOmainrUrXr\r^rfrpr}c@sfeZdZdZddZeddZeddZedd Zd d Z dd dZ ddZ ddZ ddZ dS)InputStreamWithMemoryzWraps an HTMLInputStream to remember characters since last < This wraps existing HTMLInputStream classes to keep track of the stream since the last < which marked an open tag state. cCs$||_|jj|_|jj|_g|_dSN) _inner_streamresetposition_buffer)selfZ inner_streamr8/tmp/pip-unpacked-wheel-0vx3v32x/bleach/html5lib_shim.py__init__s  zInputStreamWithMemory.__init__cCs|jjSr)rerrorsrrrrrszInputStreamWithMemory.errorscCs|jjSr)r charEncodingrrrrrsz"InputStreamWithMemory.charEncodingcCs|jjSr)rchangeEncodingrrrrr sz$InputStreamWithMemory.changeEncodingcCs|j}|r|j||Sr)rcharrappend)rcrrrrs  zInputStreamWithMemory.charFcCs$|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_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. Qs z/BleachHTMLTokenizer.__iter__..z!expected-closing-tag-but-got-charzeof-in-tag-namerr.)zeof-in-attribute-namez eof-in-attribute-value-no-quotes)r__iter__TAG_TOKEN_TYPESgetritemsparsertagslowerstriprrTAG_TOKEN_TYPE_CHARACTERSTAG_TOKEN_TYPE_PARSEERROR)rZlast_error_tokentokenrrrrAsR         zBleachHTMLTokenizer.__iter__NcsJ|jrt||S|r4|jdddd7<n|jtdddS)Nr.r&r)rr consumeEntity currentToken tokenQueuerr)r allowedChar fromAttributerrrrs z!BleachHTMLTokenizer.consumeEntitycs|jtSr)rrr tagOpenStaterrrrrs z BleachHTMLTokenizer.tagOpenStatecs|j}|jjdk r|dtkr|d|jjkr|jjrh|jrb|dtkrb|dtkrbd}qrd}n |j }t |d}||_|_|j ||j|_dS|j|_tdS)Nrname rr)rrrrrrrTAG_TOKEN_TYPE_STARTHTML_TAGS_BLOCK_LEVELrrrrr dataStatestateremitCurrentToken)rrZnew_data new_tokenrrrrs0       z$BleachHTMLTokenizer.emitCurrentToken)F)NF) rrrrrrrrr __classcell__rrrrr3s  [ rcs*eZdZdZfddZd ddZZS) BleachHTMLParserz$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|]}|VqdSr)r)rtagrrrrsz,BleachHTMLParser.__init__..) frozensetrrrrr)rrrrrrrrrs zBleachHTMLParser.__init__Fr4TcKsj||_||_||_tf||j|d||_|z |Wn$tk rd||YnXdS)N)rrr) innerHTMLMode container scriptingrr tokenizerrmainLoopReparseException)rr innerHTMLrrrrrr_parses  zBleachHTMLParser._parse)Fr4T)rrrrrrrrrrrrs  rcCs|ddkrt|dkrdS|ddkr<|ddd}}n|ddd}}|d krZdSt||}d|krxd krnnt|SdSt|dS) a9Convert an entity (minus the & and ; part) into what it represents This handles numeric, hex, and text entities. :arg value: the string (minus the ``&`` and ``;`` part) to convert :returns: unicode character or None if it's an ambiguous ampersand that doesn't match a character entity r#NrxX ri)lenintchrENTITIESr)valueZ int_as_stringr% code_pointrrrconvert_entitys    rcCsd|kr |Sg}t|D]l}|s"q|drzt|}|dk rzt|}|dk rz|||t|dd}|r||q||qd|S)zConverts all found entities in the text :arg text: the text to convert entities in :returns: unicode text with converted entities rNrr)next_possible_entity startswith match_entityrrrr)textZnew_textpartentityZ converted remainderrrrconvert_entities1s$     rcCs|ddkrtd|dd}t|}d}dtj}|r|ddkrd}|d|rx|dd krxd }||d7}nd }|r|d|kr|d}||krq||7}q||r|r|dd kr|SdS|r|d|kr|d}||7}t|sdSq|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<&=;rrZ0123456789abcdefABCDEF 0123456789;) ValueErrorrstring whitespacer ENTITIES_TRIEhas_keys_with_prefix)rZpossible_entityZend_charactersallowedrrrrrTs:        rz(&)ccsBtt|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)rrIrrrrrs  rcs.eZdZdZdZddZdfdd ZZS) BleachHTMLSerializerz[HTMLSerializer that undoes & -> & in attributes and sets escape_rcdata to True Tccs|dd}t|D]l}|sq|drrt|}|dk rrt|dk rrd|dV|t|dd}|r|Vq|ddVqdS)z,Escapes just bare & in HTML attribute valuesz&rNrr)replacerrrrr)rstokenrrrrrescape_base_amps   z$BleachHTMLSerializer.escape_base_ampNc#s~d}d}t||D]b}|rd|dkr,d}n0|rP|dkr\||EdHd}qn |dkr\d}|Vq|drrd}|VqdS)zWrap HTMLSerializer.serialize and conver & to & in attribute values Note that this converts & to & in attribute values where the & isn't already part of an unambiguous character entity. F>rN=Tr)r serializerr)r treewalkerencodingin_tagZ after_equalsrrrrrs" zBleachHTMLSerializer.serialize)N)rrrr escape_rcdatarrrrrrrrs r)9rrerwarningsfilterwarningsDeprecationWarningZbleach._vendor.html5librrrZ!bleach._vendor.html5lib.constantsr r r rZ$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._trierentitiesrr tokenTypesrrZTAG_TOKEN_TYPE_ENDrrrZ HTML_TAGSrrrrrrrcompilerrrrrrrs\           y'?1,"#<