B ;4\ Fp@sHdZddlmZddlZddlZddlZddlmZmZddlm Z ddl m Z m Z ddl m ZddlmZdd lmZddlmZdd lmZdd lmZdd lmZdd lmZe jZeeZee j de j de j dgZ!e j dZ"e j dZ#dddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1d2d3d4d5d6d7d8d9d:d;dd?d@dAdBdCdDdEdFdGdHdIdJdKdLdMdNdOdPdQdRdSdTdUdVdWdXdYdZd[d\d]d^d_d`dadbdcdddedfdgdhdidjdkdldmdndodpdqdrdsdtdudvdwdxdydzd{d|d}d~ddddgpZ$Gddde%Z&GdddeZ'GdddeZ(ddZ)ddZ*ddZ+e,dZ-ddZ.GdddeZ/dS)z Shim module between Bleach and html5lib. This makes it easier to upgrade the html5lib library without having to change a lot of code. )unicode_literalsN) HTMLParser getTreeWalker) constants) namespacesprefixes)_ReparseException)Filter)allowed_protocols)HTMLInputStream)HTMLSerializer) HTMLTokenizer)TrieZStartTagZEndTagZEmptyTagZ CharactersZ ParseErroraZabbrZaddressZareaZarticleZasideZaudiobbaseZbdiZbdoZ blockquoteZbodybrZbuttonZcanvasZcaptionZcitecodecolZcolgroupdataZdatalistZdddelZdetailsZdfnZdialogdivZdlZdtZemZembedZfieldsetZ figcaptionZfigureZfooterZformZh1Zh2Zh3Zh4Zh5Zh6headheaderZhgroupZhrZhtmliZiframeZimginputZinsZkbdZkeygenZlabelZlegendZlilinkmapZmarkZmenumetaZmeterZnavZnoscriptobjectZolZoptgroupZoptionoutputpZparamZpictureZpreZprogressqZrpZrtZrubysZsampZscriptZsectionZselectZslotZsmallsourcespanZstrongZstylesubZsummaryZsuptableZtbodyZtdtemplateZtextareaZtfootZthZtheadZtimetitleZtrZtrackuZulvarZvideoZwbrc@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$||_|jj|_|jj|_g|_dS)N) _inner_streamresetZposition_buffer)selfZ inner_streamr13lib/python3.7/site-packages/bleach/html5lib_shim.py__init__s  zInputStreamWithMemory.__init__cCs|jjS)N)r-errors)r0r1r1r2r4szInputStreamWithMemory.errorscCs|jjS)N)r- charEncoding)r0r1r1r2r5sz"InputStreamWithMemory.charEncodingcCs|jjS)N)r-changeEncoding)r0r1r1r2r6sz$InputStreamWithMemory.changeEncodingcCs|j}|r|j||S)N)r-charr/append)r0cr1r1r2r7s  zInputStreamWithMemory.charFcCs$|jj||d}|jt||S)N)opposite)r- charsUntilr/extendlist)r0Z charactersr:charsr1r1r2r;sz InputStreamWithMemory.charsUntilcCs|jr|jd|j|S)N)r/popr-unget)r0r7r1r1r2rAs zInputStreamWithMemory.ungetcCstd|jS)zReturns the stream history since last '<' Since the buffer starts at the last '<' as as seen by tagOpenState(), we know that everything from that point to when this method is called is the "tag" that is being tokenized. )sixZ text_typejoinr/)r0r1r1r2get_tagszInputStreamWithMemory.get_tagcCs dg|_dS)zResets stream history to just '<' This gets called by tagOpenState() which marks a '<' that denotes an open tag. Any time we see that, we reset the buffer. s  z0BleachHTMLTokenizer.__iter__..z!expected-closing-tag-but-got-char) rNrM__iter__TAG_TOKEN_TYPESgetlowerstripparsertagsrPrECHARACTERS_TYPEPARSEERROR_TYPE)r0Zlast_error_tokentoken)rRr1r2rYs8        zBleachHTMLTokenizer.__iter__NcsN|jrtt|||S|r8|jdddd7<n|jtdddS)Nrr?&)rSr)rOrNrM consumeEntity currentToken tokenQueuer8r`)r0Z allowedCharZ fromAttribute)rRr1r2re-s z!BleachHTMLTokenizer.consumeEntitycs|jtt|S)N)rPrGrNrM tagOpenState)r0)rRr1r2rh?s z BleachHTMLTokenizer.tagOpenStatecs|j}|jjdk rr|dtkrr|d|jjkrr|jjr@d}n |j}t|d}||_|j ||j |_ dSt t|dS)NrSnamerB)rSr)rfr^r_rZr\r]rPrEr`rgr8Z dataStatestaterNrMemitCurrentToken)r0rbZnew_dataZ new_token)rRr1r2rkGs    z$BleachHTMLTokenizer.emitCurrentToken)F)NF) rHrIrJrKr3rYrerhrk __classcell__r1r1)rRr2rMs  > rMcs*eZdZdZfddZdddZZS) BleachHTMLParserz$Parser that uses BleachHTMLTokenizerc s>|dk rdd|Dnd|_||_||_tt|jf|dS)a :arg tags: list 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) NcSsg|] }|qSr1)r\)rVtagr1r1r2rXvsz-BleachHTMLParser.__init__..)r_r]rOrNrmr3)r0r_r]rOrQ)rRr1r2r3ls zBleachHTMLParser.__init__FrcKsj||_||_||_tf||j|d||_|y |Wn$tk rd||YnXdS)N)rPrOr^) Z innerHTMLMode container scriptingrMrOZ tokenizerr.ZmainLoopReparseException)r0rPZ innerHTMLrorprQr1r1r2_parse{s  zBleachHTMLParser._parse)FrF)rHrIrJrKr3rrrlr1r1)rRr2rmjs rmcCsT|ddkrH|ddkr0tt|dddStt|dddSt|dS) a9Convert an entity (minus the & and ; part) into what it represents This handles numeric, hex, and text entities. :arg value: the string (minus the ``&`` and ``;`` part) to convert :returns: unicode character or None if it's an ambiguous ampersand that doesn't match a character entity r#rc)xXN )rCZunichrintENTITIESr[)valuer1r1r2convert_entitys  r|cCsd|kr |Sg}xxt|D]l}|s$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 rdNrvrB)next_possible_entity startswith match_entityr|r8lenrD)textZnew_textpartentityZ convertedZ remainderr1r1r2convert_entitiess$   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|SdSx2|r|d|kr|d}t|sP||7}qW|r|r|dd kr|SdS) aHReturns first entity in stream or None if no entity exists Note: For Bleach purposes, entities must start with a "&" and end with a ";". This ignoresambiguous character entities that have no ";" at the end. :arg stream: the character stream :returns: ``None`` or the entity string without "&" or ";" rrdzStream should begin with "&"rcNrBz<&=;rs)rtruZ0123456789abcdefABCDEF 0123456789;) ValueErrorr=stringZ whitespacer@ ENTITIES_TRIEZhas_keys_with_prefix)rPZpossible_entityZend_charactersZallowedr9r1r1r2rs:         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 "&" rrvrdN) enumerate AMP_SPLIT_REsplit)rrrr1r1r2r}s  r}cs*eZdZdZddZdfdd ZZS)BleachHTMLSerializerz3HTMLSerializer that undoes & -> & in attributesccs|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&rdNrrv)replacer}r~rr|r)r0stokenrrr1r1r2escape_base_amps  z$BleachHTMLSerializer.escape_base_ampNc#sd}d}xtt|||D]l}|rt|dkr2d}n:|r`|dkrlx||D] }|VqJWd}qn |dkrld}|Vq|drd}|VqWdS)zWrap HTMLSerializer.serialize and conver & to & in attribute values Note that this converts & to & in attribute values where the & isn't already part of an unambiguous character entity. F>rT=TrFN)rNr serializerr~)r0Z treewalkerencodingZin_tagZ after_equalsrr)rRr1r2r3s$  zBleachHTMLSerializer.serialize)N)rHrIrJrKrrrlr1r1)rRr2rsr)0rKZ __future__rrerrCZbleach._vendor.html5librrrZ!bleach._vendor.html5lib.constantsrrrrqZ$bleach._vendor.html5lib.filters.baser Z)bleach._vendor.html5lib.filters.sanitizerr ZSanitizerFilterZ$bleach._vendor.html5lib._inputstreamr Z"bleach._vendor.html5lib.serializerr Z"bleach._vendor.html5lib._tokenizerr Zbleach._vendor.html5lib._trierZentitiesrzrsetZ tokenTypesrZr`raZ HTML_TAGSrr,rMrmr|rrcompilerr}rr1r1r1r2s"            >%#: