3 d @shdZddlZddlZddlmZddlmZddlmZm Z m Z ej j e ddGd d d e e Z dS) zDTests to ensure that the html5lib tree builder generates good trees.N) BeautifulSoup) SoupStrainer)HTML5LIB_PRESENTHTML5TreeBuilderSmokeTestSoupTestz?html5lib seems not to be present, not testing its tree builder.)reasonc@seZdZdZeddZddZddZdd Zd d Z d d Z ddZ ddZ ddZ ddZddZddZddZddZddZd d!Zd"S)#TestHTML5LibBuilderz"See ``HTML5TreeBuilderSmokeTest``.cCsddlm}|S)Nr)HTML5TreeBuilder)Z bs4.builderr )selfr r A/tmp/pip-build-8z3xcdsh/beautifulsoup4/bs4/tests/test_html5lib.pydefault_builders z#TestHTML5LibBuilder.default_builderc Csrtd}d}tjdd}t|d|d}WdQRX|j|j|ksHt|\}|jtks\tdt |j ksntdS)Nbz

A bold statement.

T)recordhtml5lib)Z parse_onlyz4the html5lib tree builder doesn't support parse_only) rwarningscatch_warningsrdecodeZ document_forAssertionErrorfilename__file__strmessage)r Zstrainermarkupwsoupwarningr r r test_soupstrainersz%TestHTML5LibBuilder.test_soupstrainercCsd}|j|d|jddS)z8html5lib inserts tags where other parsers don't.z[z
Here's another table:
foo
Here's another table:
foo
z{
Foo
Bar
Baz
N)Z assert_soup)r rr r r test_correctly_nested_tables&s z0TestHTML5LibBuilder.test_correctly_nested_tablescCs$d}|j|}d|jjks tdS)Nzy

foo

s

foo

)rpencoder)r rrr r r (test_xml_declaration_followed_by_doctype:s  zfoo

bar

zD

foo

bar

r )rbodyrrlenfind_all)r rrr r r test_reparented_markupHs z*TestHTML5LibBuilder.test_reparented_markupcCs:d}|j|}d|jjks tdt|jdks6tdS)Nz&

foo

bar

zE

foo

bar

r#r )rr$rrr%r&)r rrr r r +test_reparented_markup_ends_with_whitespaceOs z?TestHTML5LibBuilder.test_reparented_markup_ends_with_whitespacecCsLd}|j|}|jdd\}}|jd\}}|j|ks:t|j|ksHtdS)zVerify that we keep the two whitespace nodes in this document distinct when reparenting the adjacent tags. z,
 )stringtbodyN)rr& next_elementr)r rrZspace1Zspace2Ztbody1Ztbody2r r r aftermath

aftermath

target)r*Z aftermathr)rnoscriptr,rfindr&Zprevious_element)r rrr0r.Zfinal_aftermathr r r *test_reparented_markup_containing_children`s  z>TestHTML5LibBuilder.test_reparented_markup_containing_childrencCs$d}|j|}t|jds tdS)z(Processing instructions become comments.szN)rr startswithr)r rrr r r test_processing_instructionps z/TestHTML5LibBuilder.test_processing_instructioncCs8d}|j|}|jd\}}||ks(t||k s4tdS)Ns

a)rr&r)r rrZa1Za2r r r test_cloned_multivalue_nodevs   z/TestHTML5LibBuilder.test_cloned_multivalue_nodecCs$d}|j|}d|jjks tdS)NsAz>A
)rr$rr)r rrr r r test_foster_parenting}s z)TestHTML5LibBuilder.test_foster_parentingcCsLd}|j|}dd|dDdd|dDt|jddksHtd S) z Test that extraction does not destroy the tree. https://bugs.launchpad.net/beautifulsoup/+bug/1782928 zW

hello

cSsg|] }|jqSr )extract).0sr r r sz7TestHTML5LibBuilder.test_extraction..scriptcSsg|] }|jqSr )r8)r9r:r r r r;sstyler rN)rr%r&r)r rrr r r test_extractions  z#TestHTML5LibBuilder.test_extractioncCsJd}|j|}g}x"|jdD]}|j|jdqWt|dksFtdS)z Test that empty comment does not break structure. https://bugs.launchpad.net/beautifulsoup/+bug/1806598 zI
forminputrN)rr&extendr%r)r rrinputsr?r r r test_empty_comments  z&TestHTML5LibBuilder.test_empty_commentcCszd}|j|}d|jjkstd|jjks.td|jjdjksDt|j|dd}d|jjjksdtd|jjjksvtdS)Nz=

text

r# sourcelineF)Zstore_line_numbers sourcepos)rr rErrFr1name)r rrr r r test_tracking_line_numberss z.TestHTML5LibBuilder.test_tracking_line_numberscCsdS)Nr )r r r r test_special_string_containerssz2TestHTML5LibBuilder.test_special_string_containersc Csnxhd>D]`\}}}d*|}|j|j}|j}d+|jd,}||ksFt|jd-d.}d+|}||kstqWdS)?N⇄⇄⇄⊧⊧⊧𝔑𝔑𝔑≧̸≧̸≧̸¬¬¬⫬⫬⫬"""∴∴∴ ∴ ∴fjfjfj⊔⊔⊔⊔︀⊔︀⊔︀'''|||z
%s
s
%s
utf8html) formatterrJrKrLrMrNrOrPrQrRrSrTrUrVrWrXrYrZr[r\r]r^r_r`rarbr`rarcr`rardrerfrgrhrirjrkrlrmrnrorprqrr)rvrwrxryrzr{r|r}r~rrrrrr)rdivr!r) r Z input_elementZoutput_unicodeZoutput_elementrrZwithout_elementexpectZ with_elementr r r test_html5_attributess0    z)TestHTML5LibBuilder.test_html5_attributesN)__name__ __module__ __qualname____doc__propertyrrrr"r'r(r-r2r4r6r7r>rCrHrIrr r r r r s"   r )rZpytestrZbs4rZ bs4.elementrrrrmarkZskipifr r r r r s