B u9aºã@s–dZddlZddlZddlZGdd„dejjƒZGdd„deƒZGdd„deƒZ Gd d „d ej ƒZ Gd d „d e ƒZ Gd d„de ƒZ edkr’e ¡dS)zTests for HTMLParser.py.éNc@sleZdZdd„Zdd„Zdd„Zdd„Zd d „Zd d „Zd d„Z dd„Z dd„Z dd„Z dd„Z dd„ZdS)ÚEventCollectorcOs*g|_|jj|_tjjj|f|ž|ŽdS)N)ÚeventsÚappendÚhtmlÚparserÚ HTMLParserÚ__init__)ÚselfÚargsÚkw©r ú%/usr/lib/python3.7/test_htmlparser.pyr s zEventCollector.__init__cCsng}d}xZ|jD]P}|d}||kr0dkrRnnd|dd|df|d<n | |¡|}qW||_|S)NrÚdataéÿÿÿÿé)rr)r ÚLZprevtypeZeventÚtyper r r Ú get_eventss  zEventCollector.get_eventscCs| d||f¡dS)NÚstarttag)r)r ÚtagÚattrsr r r Úhandle_starttag szEventCollector.handle_starttagcCs| d||f¡dS)NÚ startendtag)r)r rrr r r Úhandle_startendtag#sz!EventCollector.handle_startendtagcCs| d|f¡dS)NÚendtag)r)r rr r r Ú handle_endtag&szEventCollector.handle_endtagcCs| d|f¡dS)NÚcomment)r)r rr r r Úhandle_comment+szEventCollector.handle_commentcCs| d|f¡dS)NÚcharref)r)r rr r r Úhandle_charref.szEventCollector.handle_charrefcCs| d|f¡dS)Nr)r)r rr r r Ú handle_data1szEventCollector.handle_datacCs| d|f¡dS)NÚdecl)r)r rr r r Ú handle_decl4szEventCollector.handle_declcCs| d|f¡dS)NÚ entityref)r)r rr r r Úhandle_entityref7szEventCollector.handle_entityrefcCs| d|f¡dS)NÚpi)r)r rr r r Ú handle_pi:szEventCollector.handle_picCs| d|f¡dS)Nz unknown decl)r)r r!r r r Ú unknown_decl=szEventCollector.unknown_declN)Ú__name__Ú __module__Ú __qualname__rrrrrrrr r"r$r&r'r r r r rsrc@seZdZdd„ZdS)ÚEventCollectorExtracCs$t |||¡| d| ¡f¡dS)NÚ starttag_text)rrrZget_starttag_text)r rrr r r rCsz#EventCollectorExtra.handle_starttagN)r(r)r*rr r r r r+Asr+c@seZdZdd„Zdd„ZdS)ÚEventCollectorCharrefscCs| d¡dS)Nz6This should never be called with convert_charrefs=True)Úfail)r rr r r rJsz%EventCollectorCharrefs.handle_charrefcCs| d¡dS)Nz6This should never be called with convert_charrefs=True)r.)r rr r r r$Msz'EventCollectorCharrefs.handle_entityrefN)r(r)r*rr$r r r r r-Hsr-c@s&eZdZdd„Zddd„Zdd„ZdS) Ú TestCaseBasecCs tddS)NF)Úconvert_charrefs)r)r r r r Ú get_collectorSszTestCaseBase.get_collectorNcCsv|dkr| ¡}|}x|D]}| |¡qW| ¡| ¡}||krr| dt|ƒdt |¡dt |¡¡dS)Nz6received events did not match expected events Source: z Expected: z Received: )r1ÚfeedÚcloserr.ÚreprÚpprintZpformat)r ÚsourceZexpected_eventsÚ collectorrÚsrr r r Ú _run_checkVs zTestCaseBase._run_checkcCs| ||tdd¡dS)NF)r0)r9r+)r r6rr r r Ú_run_check_extradszTestCaseBase._run_check_extra)N)r(r)r*r1r9r:r r r r r/Qs r/c@seZdZdd„Zdd„Zdd„Zdd„Zd d „Zd d „Zd d„Z dd„Z dd„Z dd„Z dd„Z dd„Zdd„Zdd„Zdd„Zdd „Zd!d"„Zd#d$„Zd%d&„Zd'd(„Zd)d*„Zd+d,„Zd-d.„Zd/d0„Zd1d2„Zd3d4„Zd5d6„Zd7d8„Zd9d:„Zd;d<„Z d=d>„Z!d?d@„Z"dAdB„Z#dCS)DÚHTMLParserTestCasecCs | ddg¡| ddg¡dS)Nz)r%zprocessing instructionz)r%zprocessing instruction ?)r9)r r r r Ú test_processing_instruction_onlyksz3HTMLParserTestCase.test_processing_instruction_onlycCsB| ddddddgfddddddd d d gfd d dddddg¡dS)Nz¿ &entity; sample text “ )rÚ )r!zDOCTYPE html PUBLIC 'foo'rr)r#Úentity)rZ32)rz4comment1a -><&#bad;

rÚp)rz&#bad;)rrBz
&#bad;
Údiv)rrC)r9)r r r r Útest_malformatted_charref“sz,HTMLParserTestCase.test_malformatted_charrefcCs| dddg¡dS)Nz&entityref foo)r#r#)rz foo)r9)r r r r Útest_unclosed_entityref sz*HTMLParserTestCase.test_unclosed_entityrefcCs$| dddgfddgfddg¡dS)NzrÚaÚb)rrF)rrG)r9)r r r r Útest_bad_nesting¦s z#HTMLParserTestCase.test_bad_nestingcCs| ddg¡dS)Nz#this text & contains & ampersands &)rz#this text & contains & ampersands &)r9)r r r r Útest_bare_ampersands±sz'HTMLParserTestCase.test_bare_ampersandscCs| ddg¡dS)Nz.this < text > contains < bare>pointy< brackets)rz.this < text > contains < bare>pointy< brackets)r9)r r r r Útest_bare_pointy_brackets¶sz,HTMLParserTestCase.test_bare_pointy_bracketscCs0| ddddgfg¡| ddddgfg¡dS)Nz rrF)rGú)rGú>)r9)r r r r Útest_starttag_end_boundary»sz-HTMLParserTestCase.test_starttag_end_boundarycCs’dddgfg}| dg|¡| ddg|¡| ddg|¡| d d g|¡| d d g|¡| d dg|¡dddgfg}| dg|¡| ddg|¡| ddg|¡| d dg|¡| dd g|¡| ddg|¡dg}| ddg|¡| ddg|¡| ddg|¡| ddg|¡| dd g|¡| d!d"g|¡| d#d$g|¡| d%d&g|¡| d'd(g|¡| d)dg|¡| ddg|¡dS)*NrrF)rGrKz zzzzzzb='>'>z='>'>z'>'>z')rÚabcÚz rKz !--abc-->zzzzzzz ¬-an-entity-ref;zz

zfoo = "";zfoo = "";zfoo = < /script> z*z //<\/s'+'cript>'); //]]>z zfoo = "";u ÚscriptÚstyleZSCRIPTZSTYLEZScriptZStylez <{element}>{content})ÚelementÚcontentrrr)ÚlowerÚformatr9)r ÚcontentsÚelementsr[rZÚ element_lowerr8r r r Útest_cdata_content s(  z%HTMLParserTestCase.test_cdata_contentcCsjGdd„dtƒ}d}xPdD]H}| ¡ ¡}dj||d}|j|d|gfd|fd |fg|d d d qWdS) Nc@seZdZdd„ZdS)zBHTMLParserTestCase.test_cdata_with_closing_tags..CollectorcSs|jS)N)r)r r r r r.szMHTMLParserTestCase.test_cdata_with_closing_tags..Collector.get_eventsN)r(r)r*rr r r r Ú Collector-srbz‹ ¬-an-entity-ref;

'')z scriptzscript z script z scriptzscript z script z{1}{1}z &quo &# &#xrzno charrefs here)rzno charrefs here)Z assertTruer0r9r]ÚjoinÚrangeÚlen)r r7ZcharrefsrerÚtextrrlr r r Útest_convert_charrefsXs>       z(HTMLParserTestCase.test_convert_charrefsc Cs2| ddddgfdddddd gfd d d d g ¡dS)NzFte>>xt&a< )rz/img)rzhtml<)r9)r r r r Útest_tolerant_parsings  z(HTMLParserTestCase.test_tolerant_parsingcCs| dg¡| ddg¡| ddg¡| ddg¡| dd d gfg¡| d d g¡| d dg¡| ddg¡| ddg¡| ddg¡| ddg¡| ddg¡| dd dgfg¡| dd dgfg¡| dddgfg¡| dd dgfg¡| d ddgfg¡dS)!Nzz)rú$zrza)rzazza$zza$bzrzz)r9)r r r r Útest_starttag_junk_charsŽs" z+HTMLParserTestCase.test_starttag_junk_charsc CsÊ| ddddgfg¡d}ddddd d d d d dgfg}| ||¡d}dddddgfdddddgfg}| ||¡d}ddgfddgfddgfddgfddgfddgfddgfddgfg}| ||¡dS)NzrrF)rSÚvarzjr?)ÚwidthZ902)ZheightZ250px)r@z,/sites/default/files/images/homepage/foo.jpg)z*whatN)ÚamN)ÚiN)ZdoingN)zhere*Nz9)rSN)ú=N)ÚbarNrzDÚmeta)r9)r rrer r r Útest_slashes_in_starttag¡s$   z+HTMLParserTestCase.test_slashes_in_starttagcCs| ddg¡dS)Nz)r!zDOCTYPE foo $ )r9)r r r r Útest_declaration_junk_charsÀsz.HTMLParserTestCase.test_declaration_junk_charscCs| ddg¡dS)Nz")rzspacer type="block" height="25")r9)r r r r Útest_illegal_declarationsÃsz,HTMLParserTestCase.test_illegal_declarationscCsd}ddgfddddgfdddd d gfdd gfdd d gfdddgfddddgfdddgfdddddddgfdddgfddddg}| ||¡dS)Nzï
- software-and-i- library
rrZbody)ZbgcolorZd0ca90)rtZ181008Útable)Ú cellspacingÚ0)Z cellpaddingrU)rzz100%ÚtrZtd)ZalignÚleftZfont)Úsizez-1)rz- rF)rjz/rabota/Úspan)ÚclassÚen)rz software-and-i)rr‰)rrF)rjz/1/)rz library)rrƒ)r9)r rrer r r Útest_with_unquoted_attributesÇs         z0HTMLParserTestCase.test_with_unquoted_attributescCs| ddddddgfg¡dS)Nz7

rÚform)Úactionz/xxx.php?a=1&b=2&)ú,N)ÚmethodZpost)r9)r r r r Útest_comma_between_attributesßsz0HTMLParserTestCase.test_comma_between_attributescCs| ddddgfg¡dS)Nzrr)rŽzbogus|&#()value)r9)r r r r Ú-test_weird_chars_in_unquoted_attribute_valuesæsz@HTMLParserTestCase.test_weird_chars_in_unquoted_attribute_valuescCsZd}ddgfdddgfdddgfdddgfdddgfdddgfddgfg }| ||¡dS)Nzn






rÚbr)rzlabel<)rrC)rzThis confuses the parserrrG)rZThis)rrG)rz"> confuses the parser)r9)r rrer r r Útest_broken_invalid_end_tags z.HTMLParserTestCase.test_broken_invalid_end_tagcCsd}dddgfddgfddddgfd d d d d gfdddgfddddg}| ||¡d}dddddgfddgfddddgfd d g}| ||¡dS)Nz[
The rain
in Spain
rrC)rYrOrG)rzThe rF)rjZsome_url)rZrain)rrF)rú rr“)rz in r‰)rZSpain)rr‰)rrG)rrCz>
The rain)rN)rSr~)r9)r rrer r r Ú$test_correct_detection_of_start_tagss0    z7HTMLParserTestCase.test_correct_detection_of_start_tagscCsLddgfddgfddddgfd ddgfg}x|D]\}}| ||¡q0WdS) Nza&)rza&za&b)rZabza&b )rrF)r#rG)rr–za&b;)r9)r rrrer r r Útest_EOF_in_charref2s  z&HTMLParserTestCase.test_EOF_in_charrefc CsHddlm}| ¡}| t¡ d}| | |¡||ƒ¡WdQRXdS)Nr)Úunescapez%""""""&#bad;)rr™r1Z assertWarnsÚDeprecationWarningÚ assertEqual)r r™rBr8r r r Útest_unescape_method@s   z'HTMLParserTestCase.test_unescape_methodcCs$d}ddddddg}| ||¡dS)NzŒ)rz not really a comment )rz not a comment either --)rz -- close enough --)rrO)rz<-- this was an empty comment)rz!! another bogus comment !!!)r9)r rrer r r Útest_broken_commentsGsz'HTMLParserTestCase.test_broken_commentscCsXd}dddddddgfdd dd d gfdd d dgfdddddd dgfdg}| ||¡dS)Nzábroken condcomfoo)z unknown declzif !(IE))rzbroken condcom)z unknown declZendif)z unknown declzif ! IErÚlink)rjz favicon.tiff)z unknown declzif !IE 6r?)r@z firefox.png)z unknown declzif !ie 6rrG)rrS)rrG)z unknown declzif (!IE)|(lt IE 9))r@z mammoth.bmp)r9)r rrer r r Útest_broken_condcomsWs&    z'HTMLParserTestCase.test_broken_condcomscCs8tdd}| d¡| | ¡dddgfddd g¡dS) NT)r0zfoo link bar & baz)rzfoo rrF)rrž)rrF)rz bar & baz)rr2r›r)r rr r r Ú"test_convert_charrefs_dropped_textxs    z5HTMLParserTestCase.test_convert_charrefs_dropped_textN)$r(r)r*r<rArDrErHrIrJrMrQrRrTrWrardrfrgrurvrxr€rr‚rŒr‘r’r”r•r—r˜rœrrŸr r r r r r;isB    )  !!r;c@s|eZdZdd„Zdd„Zdd„Zdd„Zd d „Zd d „Zd d„Z dd„Z dd„Z dd„Z dd„Z dd„Zdd„Zdd„ZdS)ÚAttributesTestCasecCsHddddddgfg}| d|¡| d|¡| d |¡| d |¡dS) NrrF)rGÚv)Úcr¢)Údr¢)ÚeNzzzz)r9)r rPr r r Útest_attr_syntax‡s    z#AttributesTestCase.test_attr_syntaxcCsb| ddddddgfg¡| ddddd gfg¡| d dd d gfg¡| d dddgfg¡dS)Nz'rrF)rGzxxx xxx)r£zyyy yyy)r¤z xyz z )rGrO)r£rOzr¥)rFz rgb(1,2,3)z)rjzmailto:xyz@example.com)r9)r r r r Útest_attr_valuess z#AttributesTestCase.test_attr_valuescCsL| dddddgfg¡| ddddd gfg¡| d dddd gfg¡dS) Nu!中文rr?)r@z /foo/bar.png)Zaltu中文u+rF)Útitleu テスト)rjuテスト.htmlu+)r9)r r r r Útest_attr_nonasciiŸs  z%AttributesTestCase.test_attr_nonasciicCs| ddddgfg¡dS)Nz!rrF)rGz&><"')r9)r r r r Útest_attr_entity_replacement®sz/AttributesTestCase.test_attr_entity_replacementcCs| ddddddgfg¡dS)NzrrF)za.br¢)zc:dr¢)ze-fr¢)r9)r r r r Útest_attr_funky_names³sz(AttributesTestCase.test_attr_funky_namescCs| ddddgfg¡dS)Nz0rr)rSu€&aa&unsupported;)r9)r r r r Útest_entityrefs_in_attributes¸sz0AttributesTestCase.test_entityrefs_in_attributescCs.| ddddgfdddgfdddgfg¡dS) NzrrF)rwNrG)rwú%r£)ú\ú/)r9)r r r r Útest_attr_funky_names2¾s   z)AttributesTestCase.test_attr_funky_names2cCs`xZdD]R}| d|dddgfg¡| d|dddgfg¡| d|dddgfg¡qWdS)N)ú&z&z&z&z rrF)rjr±z z )r9)r r>r r r Ú test_entities_in_attribute_valueÅs    z3AttributesTestCase.test_entities_in_attribute_valuec CsPd}dddgfdddddgfddddd gfd dddd gfd dg }| ||¡dS) NzÜtest - bad1test - bad2test - bad3test - bad4rrF)rjztest'style='color:red;bad1')rz test - bad1)rrF)rjztest'+style='color:red;ba2')rz test - bad2)rjutest' style='color:red;bad3')rz test - bad3)rjutest' style='color:red;bad4')rz test - bad4)r9)r rrer r r Útest_malformed_attributesÏs    z,AttributesTestCase.test_malformed_attributescCsH| dddgfddddgfdg¡| d ddgfdddd gfdg¡dS) Nzrrlrrm)rnrO)zo""N)rrlz)z""N)r9)r r r r Ú"test_malformed_adjacent_attributesãs  z5AttributesTestCase.test_malformed_adjacent_attributescCs4| dddddgfg¡| dddddgfg¡dS) NzrrF)rzz100%)r„r…z)ÚidrS)rŠr~)r9)r r r r Útest_adjacent_attributesïs z+AttributesTestCase.test_adjacent_attributescCs| ddddgfg¡dS)NzrrF)r¢rO)r9)r r r r Útest_missing_attribute_valueøsz/AttributesTestCase.test_missing_attribute_valuecCs| ddddgfg¡dS)Nz-rrF)rjz$javascript:popup('/popup/help.html'))r9)r r r r Útest_javascript_attribute_valueüsz2AttributesTestCase.test_javascript_attribute_valuecCs| ddddgfddg¡dS)Nz-spamrrF)rjzhttp://www.example.org/">;)rZspam)rrF)r9)r r r r Útest_end_tag_in_attribute_valuesz2AttributesTestCase.test_end_tag_in_attribute_valueN)r(r)r*r¦r§r©rªr«r¬r°r²r³r´r¶r·r¸r¹r r r r r¡…s    r¡Ú__main__)Ú__doc__Z html.parserrr5Zunittestrrrr+r-ZTestCaser/r;r¡r(Úmainr r r r Ús 9