dafxDdZddlZddlZddlZddlmZGddejjZ Gdde Z Gdd e Z Gd d ejZ Gd d e ZGdde ZGddejZedk(rej$yy)zTests for HTMLParser.py.N)patchcTeZdZdZdZdZdZdZdZdZ dZ d Z d Z d Z d Zy )EventCollectorcg|_|jj|_tjjj |g|i|yN)eventsappendhtmlparser HTMLParser__init__)selfargskws //root/Python-3.12.4/Lib/test/test_htmlparser.pyr zEventCollector.__init__ s< kk((  '':t:r:cg}d}|jD]<}|d}||cxk(rdk(rnnd|dd|dzf|d<n|j||}>||_|S)Nrdata)rr )rLprevtypeeventtypes r get_eventszEventCollector.get_eventsst [[ E8Dx)6)2qE!H!45"H   rc,|jd||fy)Nstarttagr rtagattrss rhandle_starttagzEventCollector.handle_starttag"s Ze,-rc,|jd||fy)N startendtagrrs rhandle_startendtagz!EventCollector.handle_startendtag%s ]C/0rc*|jd|fy)Nendtagr)rr s r handle_endtagzEventCollector.handle_endtag(s XsO$rc*|jd|fy)Ncommentrrrs rhandle_commentzEventCollector.handle_comment- Y%&rc*|jd|fy)Ncharrefrr+s rhandle_charrefzEventCollector.handle_charref0r-rc*|jd|fy)Nrrr+s r handle_datazEventCollector.handle_data3 VTN#rc*|jd|fy)Ndeclrr+s r handle_declzEventCollector.handle_decl6r3rc*|jd|fy)N entityrefrr+s rhandle_entityrefzEventCollector.handle_entityref9s [$'(rc*|jd|fy)Npirr+s r handle_pizEventCollector.handle_pi<s T4L!rc*|jd|fy)N unknown declr)rr5s r unknown_declzEventCollector.unknown_decl?s ^T*+rN)__name__ __module__ __qualname__r rr"r%r(r,r0r2r6r9r<r?rrrr s>; ".1% ''$$)",rrceZdZdZy)EventCollectorExtracttj||||jd|jfy)N starttag_text)rr"r get_starttag_textrs rr"z#EventCollectorExtra.handle_starttagEs.&&tS%8 _d&<&<&>?@rN)r@rArBr"rCrrrErECsArrEceZdZdZdZy)EventCollectorCharrefsc&|jdyNz6This should never be called with convert_charrefs=Truefailr+s rr0z%EventCollectorCharrefs.handle_charrefL JKrc&|jdyrLrMr+s rr9z'EventCollectorCharrefs.handle_entityrefOrOrN)r@rArBr0r9rCrrrJrJJsLLrrJc eZdZdZddZdZy) TestCaseBasectdSNFconvert_charrefs)rrs r get_collectorzTestCaseBase.get_collectorUs u55rNcF||j}|}|D]}|j||j|j}||k7rP|j dt |zdzt j|zdzt j|zyy)Nz6received events did not match expected events Source: z Expected: z Received: )rXfeedcloserrNreprpprintpformat)rsourceexpected_events collectorr srs r _run_checkzTestCaseBase._run_checkXs  **,I A KKN  ""$ _ $ II$&*6l3%&(.(GH&&)/v(>? @ %rc>|j||tdyrT)rcrE)rr_rs r_run_check_extrazTestCaseBase._run_check_extrafs +UC Err)r@rArBrXrcrerCrrrRrRSs6 @ErrRceZdZdZdZdZdZdZdZdZ dZ d Z d Z d Z d Zd ZdZdZdZdZdZdZdZdZdZdZdZdZdZdZdZdZy)HTMLParserTestCasecP|jddg|jddgy)Nz)r;zprocessing instructionz)r;zprocessing instruction ?rcrWs r test_processing_instruction_onlyz3HTMLParserTestCase.test_processing_instruction_onlyms2 3 ,6  5 .8 rcZ|jddddddgfddddddd d d gfd d dddddgy)Nz &entity; sample text “ )r )r5zDOCTYPE html PUBLIC 'foo'rr )r8entity)r/32)r*z4comment1a -><&#bad;

rp)rz&#bad;r'rwz
&#bad;
divr'ryrirWs rtest_malformatted_charrefz,HTMLParserTestCase.test_malformatted_charrefsN  b !  *  ,-  #  0  rc,|jdddgy)Nz&entityref foo)r8r8)rz foorirWs rtest_unclosed_entityrefz*HTMLParserTestCase.test_unclosed_entityrefs ( & + rc<|jdddgfddgfddgy)Nzrabr'rr'rrirWs rtest_bad_nestingz#HTMLParserTestCase.test_bad_nestings2 ( b ! b !   + rc*|jddgy)N#this text & contains & ampersands &)rrrirWs rtest_bare_ampersandsz'HTMLParserTestCase.test_bare_ampersandss = ;@ rc*|jddgy)N.this < text > contains < bare>pointy< brackets)rrrirWs rtest_bare_pointy_bracketsz,HTMLParserTestCase.test_bare_pointy_bracketss H FK rc`|jddddgfg|jddddgfgy)N rrr< r>rirWs rtest_starttag_end_boundaryz-HTMLParserTestCase.test_starttag_end_boundarys6 :sZL*I)JK :sZL*I)JKrcdddgfg}|jdg||jddg||jddg||jd d g||jd d g||jd dg|dddgfg}|jdg||jddg||jddg||jd dg||jdd g||jddg|dg}|jddg||jddg||jddg||jddg||jdd g||jd!d"g||jd#d$g||jd%d&g||jd'd(g||jd)dg||jddg|y)*Nrrrrzzzzzz='>'>z'>'>z')r*abcz rz !--abc-->zzzzzz ¬-an-entity-ref;zz

zfoo = "";zfoo = "";zfoo = < /script> z*z //<\/s'+'cript>'); //]]>z zfoo = "";u )scriptstyleSCRIPTSTYLEScriptStylez <{element}>{content}elementcontentrrr')lowerformatrc)rcontentselementsrr element_lowerrbs rtest_cdata_contentz%HTMLParserTestCase.test_cdata_content s (M @G# @ '  6==gGN>PZ$C%+W$5%-}$=$?@  @ @rcGddt}d}dD]X}|jj}dj||}|j |d|gfd|fd |fg|d  Zy) NceZdZdZy)BHTMLParserTestCase.test_cdata_with_closing_tags..Collectorc|jSr)rrWs rrzMHTMLParserTestCase.test_cdata_with_closing_tags..Collector.get_events0s {{"rN)r@rArBrrCrr Collectorr/s #rrz ¬-an-entity-ref;

'')z scriptzscript z script z scriptzscript z script z{1}{1}z &quo &# &#xrno charrefs here)rr) assertTruerVrcrjoinrangelen)rracharrefsrr/textr rs rtest_convert_charrefsz(HTMLParserTestCase.test_convert_charrefsZs4   445J':&;<#_6 =G OO;BB7K$   = ="&LM!?MC  =G OO))/$   = =  =G88WIaK(D%#Xr2VTN,m#Wb1FD>+] te>>xt&a< )r*z/img)r'zhtml.8// ?1 2rc|jdg|jddg|jddg|jddg|jdd d gfg|jd d g|jd dg|jddg|jddg|jddg|jddg|jddg|jdd dgfg|jdd dgfg|jdddgfg|jdd dgfg|jd ddgfgy)!Nzz)r*$ra)r'rr)rr')rr ?  &;%<= *dB!7 89 :ub"9!:; M5"#=">?  Z$;#<=  mUB%?$@Arc |jddddgfgd}ddgdfg}|j||d}ddgd fd dgd fg}|j||d }d d gfd d gfd d gfd d gfdd gfdd gfdd gfdd gfg}|j||y) Nzr$r)rvarzjro))width902)height250px)rpz,/sites/default/files/images/homepage/foo.jpg)z*whatN)amN)iN)doingN)zhere*Nz9))rN)=N)barNrzDmetarirs rtest_slashes_in_starttagz+HTMLParserTestCase.test_slashes_in_starttags (M3@P+Q*RS. 5 /   h'/C!L M I J  h'3 $z62&>  $z62&> FB '-)D FB '-)D   h'rc*|jddgy)Nz)r5zDOCTYPE foo $ rirWs rtest_declaration_junk_charsz.HTMLParserTestCase.test_declaration_junk_charss +.H-IJrc*|jddgy)Nz")r*zspacer type="block" height="25"rirWs rtest_illegal_declarationsz,HTMLParserTestCase.test_illegal_declarationss <GH Jrcrd}ddgfdddgfdddgfdddgfdddgfdddgfddgfg }|j||y)Nzn






rbr)r'zlabeld}ddgfdddg}|j||y)Nz(This confuses the parserrr)rThisr)rz"> confuses the parserrirs rtest_broken_invalid_end_tagz.HTMLParserTestCase.test_broken_invalid_end_tags3 :b)$#68 h'rcd}dddgfddgfddddgfd d d d d gfdddgfddddg}|j||d}ddgdfddgfddddgfd d g}|j||y)Nz[

The rain
in Spain
rryrrr)rzThe r)rsome_url)rrainrr r$r)rz in span)rSpainr'rrrzz>
The rain)r,N)rrrirs r$test_correct_detection_of_start_tagsz7HTMLParserTestCase.test_correct_detection_of_start_tagss9 0 b !  34 5    D" %   $      h'O  L M b !  34 5     h'rcdddgfddgfdgdfddd gfg}|D]\}}|j||y) Na&)rr za&b)rabza&b )rrr8rrza&b;r r ri)rrr rs rtest_EOF_in_charrefz&HTMLParserTestCase.test_EOF_in_charrefs[ N# $ ^$ % G H m%78 9   # ,ND( OOD( + ,rc4d}gd}|j||y)Nz))r*z not really a comment )r*z not a comment either --)r*z -- close enough --r)r*z<-- this was an empty comment)r*z!! another bogus comment !!!rirs rtest_broken_commentsz'HTMLParserTestCase.test_broken_commentss 2   h'rcpd}dddddddgfdd dd d gfdd d dgfdddddd dgfdg}|j||y)Nzbroken condcomfoo)r>zif !(IE))rzbroken condcom)r>endif)r>zif ! IEr$link)rz favicon.tiff)r>zif !IE 6ro)rpz firefox.png)r>zif !ie 6rr)rrr)r>zif (!IE)|(lt IE 9))rpz mammoth.bmprirs rtest_broken_condcomsz'HTMLParserTestCase.test_broken_condcoms.s~M ) & % ' F%=$> ? % ( E$:#; < % ( b !   % 2 E$:#; < %# & h'rctd}|jd|j|jdddgfddd gy) NTrUzfoo link bar & baz)rzfoo rr)rrr)rz bar & baz)rrZ assertEqualr)rr s r"test_convert_charrefs_dropped_textz5HTMLParserTestCase.test_convert_charrefs_dropped_textOsO 6 34       C46F 4 6 rN) r@rArBrjrur{r}rrrrrrrrrrrrrrrrrrrrrrrrrrCrrrgrgks@     L4<:0 # @>I,("(%/R 2B&(>KJ(4 ((B ,( (B  rrgcreZdZdZdZdZdZdZdZdZ dZ d Z d Z d Z d Zd ZdZdZdZdZy)AttributesTestCasecddgdfg}|jd||jd||jd||jd|y)Nrr))rv)cr)dr)eNzzzzrirs rtest_attr_syntaxz#AttributesTestCase.test_attr_syntax^sR sM N  3V< :FC CVL CVLrc|jdddgdfg|jdddddgfg|jddd d gfg|jd ddd gfgy) Nz'rr))rzxxx xxx)rzyyy yyy)rz xyz z )rr)rrzr)rz rgb(1,2,3)z)rzmailto:xyz@example.comrirWs rtest_attr_valuesz#AttributesTestCase.test_attr_valuesgs K$c,>?@ A +$cIy+ABC E *$c,?+@AB D  -# BC D E Grc|jdddddgfg|jddddd gfg|jd dddd gfgy) Nu!中文rro)rpz /foo/bar.png)altu中文u+r)titleu テスト)ruテスト.htmlu+rirWs rtest_attr_nonasciiz%AttributesTestCase.test_attr_nonasciivs  5%"9"9";< = >  K# ? C EF G H  K# ? C EF G Hrc2|jddddgfgy)Nz!rr)rz&><"'rirWs rtest_attr_entity_replacementz/AttributesTestCase.test_attr_entity_replacements!  /#0 1 2 4rc4|jdddgdfgy)Nzrr))za.br)zc:dr)ze-frrirWs rtest_attr_funky_namesz(AttributesTestCase.test_attr_funky_namess   %#I J K Mrc2|jddddgfgy)Nz0rr )ru€&aa&unsupported;rirWs rtest_entityrefs_in_attributesz0AttributesTestCase.test_entityrefs_in_attributess#  >&#D"E F G IrcF|jddddgfdddgfdddgfgy) Nzrr)rNr)r%r)\/rirWs rtest_attr_funky_names2z)AttributesTestCase.test_attr_funky_names2s:  "# } -# | ,# } - / 0rcdD]P}|jd|zdddgfg|jd|zdddgfg|jd|zdddgfgRy)N)&z&z&z&z rr)rr2z z ri)rrms r test_entities_in_attribute_valuez3AttributesTestCase.test_entities_in_attribute_values|7 BF OOOf4(# ?@ B OOOf4(# ?@ B OOMF2(# ?@ B  Brc hd}dddgfdddddgfddddd gfd dddd gfd dg }|j||y) Nztest - bad1test - bad2test - bad3test - bad4rr)rztest'style='color:red;bad1')rz test - bad1r)rztest'+style='color:red;ba2')rz test - bad2)rutest' style='color:red;bad3')rz test - bad3)rutest' style='color:red;bad4')rz test - bad4rirs rtest_malformed_attributesz,AttributesTestCase.test_malformed_attributessr L FG H #_ FG H #_ JK L #_ JK L #_   h'rcx|jdddgfddddgfdg|jd ddgfdddd gfdgy) Nzrrr$r)rr)zo""N)r'rz)z""NrirWs r"test_malformed_adjacent_attributesz5AttributesTestCase.test_malformed_adjacent_attributessd .$c2.*C)]1KL+- . -$c2.*C)\1JK+- .rcd|jdddddgfg|jdddddgfgy) Nzrrrz100% cellspacing0z)idrclassrrirWs rtest_adjacent_attributesz+AttributesTestCase.test_adjacent_attributessT 7$c,.ABDE F 1$c(/:<= >rc2|jddddgfgy)Nzrr)rrrirWs rtest_missing_attribute_valuez/AttributesTestCase.test_missing_attribute_values $cI;78 :rc2|jddddgfgy)Nz-rr)rz$javascript:popup('/popup/help.html')rirWs rtest_javascript_attribute_valuez2AttributesTestCase.test_javascript_attribute_values) G$cKLNO Prc6|jddddgfddgy)Nz-spamrr)rzhttp://www.example.org/">;)rspamrrirWs rtest_end_tag_in_attribute_valuez2AttributesTestCase.test_end_tag_in_attribute_values, H$cBCE)?< =rcd}ddgfddddgfddgdfdd gfdd d gfdd d gfddddgfdddgfdddddddgfdddgfddddg}|j||y)Nz
- software-and-i- library
rr body)bgcolord0ca90)r181008table)r:) cellpaddingrr9trtd)alignleftfont)sizez-1)rz- r)rz/rabota/r)r?en)rz software-and-irr)rz/1/)rz library)r'rMrirs rtest_with_unquoted_attributesz0AttributesTestCase.test_with_unquoted_attributessM  $ "79K!L M O Q r "  12 3 .!1 2 Z/C.DE /!2 35P  Z.?@ /!2 35I 1D  h'rc d}dddgfddddgfddddgfddgd fdddd gfddd d gfddd dgfddddgfddddgfg }|j||y)Nz
rry)r?z bar,baz=asdr>)z,bazasd)r?zbar,)bazzasd,)r>r)rYrXrr)r?z,bar)rYz,asd)r?z,"bar")rYz,"asd")z,classr)zclass,r)zbaz,rXrirs rtest_comma_between_attributesz0AttributesTestCase.test_comma_between_attributess0!9 ; < !1? C D !2O D E !> ? !1; ? @ !2O D E !46G H I !2O D E !2O D E   h'rc2|jddddgfgy)Nz
rform)actionzbogus|&#()valuerirWs r-test_weird_chars_in_unquoted_attribute_valuesz@AttributesTestCase.test_weird_chars_in_unquoted_attribute_valuess) 7'!> ?A:B CrN)r@rArBrr!r%r'r)r+r0r3r5r7r@rBrDrGrVrZr^rCrrrr\s_M G H4 M I 0B(( .>:P =(0(8Crrc>eZdZededdZy)TestInheritancez_markupbase.ParserBase.__init__z_markupbase.ParserBase.resetctd5}t|j|jdddy#1swYyxYw)Nz_markupbase.ParserBase)rrassert_called_once)rsuper_reset_methodsuper_init_method parser_bases rtest_base_class_methods_calledz.TestInheritance.test_base_class_methods_calledsB+ , 4    0 0 2  1 1 3 4 4 4s +AA N)r@rArBrrfrCrrr`r`s( ,- )*4+.4rr`__main__)__doc__ html.parserr r]unittest unittest.mockrr r rrErJTestCaserRrgrr`r@mainrCrrrns 6,T[[++6,rA.AL^LE8$$E0n n bzCzCz4h''4 zHMMOr