B
ª `j³ ã @ sÎ d Z dZddlZddlZddlZddlZddlmZ ddlmZ ddl m
Z
m Z m Z m
Z
mZmZmZmZmZ ddlmZ eZdZG d d
„ d
ejƒZG d d „ d eƒZG d
d„ deƒZG dd„ deƒZdd„ Z dS )zHelper classes for tests.ÚMITé N)ÚTestCase)Ú
BeautifulSoup) ÚCharsetMetaAttributeValueÚCommentÚContentMetaAttributeValueÚDoctypeÚPYTHON_SPECIFIC_ENCODINGSÚ SoupStrainerÚScriptÚ
StylesheetÚTag)ÚHTMLParserTreeBuilderu-
A bare string
A tag
A tag that supposedly has contents.
AT&T
This numeric entity is missing the final semicolon:
tag is never designated as an empty-element tag.
Even if the markup shows it as an empty-element tag, it
shouldn't be presented that way.
z
zN)r Z assertFalserb rO r Ústr)r r r r r Ú!test_p_tag_is_never_empty_element‡ s
z:HTMLTreeBuilderSmokeTest.test_p_tag_is_never_empty_elementc C s( | dd¡ | dd¡ | dd¡ dS )zŠA tag that's not closed by the end of the document should be closed.
This applies to all tags except empty-element tags.
z
z
zzz z N)r' )r r r r Ú test_unclosed_tags_get_closed‘ s z6HTMLTreeBuilderSmokeTest.test_unclosed_tags_get_closedc C s. | d¡}| |jj¡ | t|jƒd¡ dS )zÂA tag is designated as an empty-element tag.
Some parsers treat as one tag, some parsers as
two tags, but it should always be an empty-element tag.
z z N)r Ú
assertTruer? rO r r{ )r r r r r Ú#test_br_is_always_empty_element_tag› s
z)r' )r r r r Ú test_nested_formatting_elements¥ s z8HTMLTreeBuilderSmokeTest.test_nested_formatting_elementsc C s( d}| |¡}| d| d¡d ¡ d S )Nz¯
Ordinary HEAD element test
Hello, world!
ztext/javascriptrS Útype)r r Úfind)r rh r r r r Útest_double_head¨ s
z)HTMLTreeBuilderSmokeTest.test_double_headc C sj d}| |¡ | |¡}|jdd}| |jt¡ |jdd}| ||j¡ |jdd}| ||j¡ d S )Nz
foobaz
Zfoobar)Útextr_ Úbaz)r' r r‚ r rY r r( r) )r r r Úcommentr_ r… r r r Ú test_comment¹ s
z%HTMLTreeBuilderSmokeTest.test_commentc C st d}d}| |¡ | |¡ | |¡}| |j ¡ |¡ | |¡}| |j ¡ |¡ | d¡}| |j ¡ d¡ dS )zWhitespace must be preserved in
and zN)r' r r ÚpreZprettifyÚtextarea)r Z
pre_markupZtextarea_markupr r r r Ú-test_preserved_whitespace_in_pre_and_textareaÈ s
zFHTMLTreeBuilderSmokeTest.test_preserved_whitespace_in_pre_and_textareac C s. d}| |¡ d}| |¡ d}| |¡ dS )z+Inline elements can be nested indefinitely.zInside a B tagz!
N)r' )r Zb_tagZ nested_b_tagZdouble_nested_b_tagr r r Útest_nested_inline_elementsÚ s
z4HTMLTreeBuilderSmokeTest.test_nested_inline_elementsc C s6 | d¡}|j}| |jjjd¡ | |jjd¡ dS )z Block elements can be nested.z*
Foo
ZFooN)r Ú
blockquoter rb ÚbrR )r r rŒ r r r Ú test_nested_block_level_elementså s
z9HTMLTreeBuilderSmokeTest.test_nested_block_level_elementsc C s d}| |d¡ | d¡ dS )z$One table can go inside another one.z[
Here's another table:
foo
zh
Here's another table:
foo
z{
Foo
Bar
Baz
N)r' )r r r r r Ú test_correctly_nested_tablesì s z5HTMLTreeBuilderSmokeTest.test_correctly_nested_tablesc C s@ d}| |¡}| ddg|jd ¡ | |j|jddd¡ d S )Nz
r_ ÚbarÚclassÚdivzfoo bar)Úclass_)r r r’ r‚ )r r r r r r Ú*test_multivalued_attribute_with_whitespace s
zCHTMLTreeBuilderSmokeTest.test_multivalued_attribute_with_whitespacec C s( d}| |¡}| dg|jjd ¡ d S )Nz1
Úcssr‘ )r r r’ )r r r r r r Ú(test_deeply_nested_multivalued_attribute s
zAHTMLTreeBuilderSmokeTest.test_deeply_nested_multivalued_attributec C s( d}| |¡}| ddg|jd ¡ d S )NzÚar r‘ )r r rh )r r r r r r Ú"test_multivalued_attribute_on_html s
z;HTMLTreeBuilderSmokeTest.test_multivalued_attribute_on_htmlc C s | dd¡ d S )Nzz)r' )r r r r Ú3test_angle_brackets_in_attribute_values_are_escaped s zLHTMLTreeBuilderSmokeTest.test_angle_brackets_in_attribute_values_are_escapedc C s | dd¡ d S )Nz$
• AT&T is in the s&p 500
u)
• AT&T is in the s&p 500
)r' )r r r r Ú3test_strings_resembling_character_entity_references s zLHTMLTreeBuilderSmokeTest.test_strings_resembling_character_entity_referencesc C s | dd¡ d S )Nz
Bob's Bar
z
Bob's Bar
)r' )r r r r Útest_apos_entity' s z)HTMLTreeBuilderSmokeTest.test_apos_entityc C s" d}| |¡}| d|jj¡ d S )Nz%
“Hello” -☃
u “Hello†-☃)r r rb rR )r r r r r r Ú*test_entities_in_foreign_document_encoding- s
zCHTMLTreeBuilderSmokeTest.test_entities_in_foreign_document_encodingc C s8 d}| d|¡ | d|¡ | d|¡ | d|¡ d S )Nu zzzz)r' )r Úexpectr r r Ú0test_entities_in_attributes_converted_to_unicode9 s
zIHTMLTreeBuilderSmokeTest.test_entities_in_attributes_converted_to_unicodec C s8 d}| d|¡ | d|¡ | d|¡ | d|¡ d S )Nu
piñata
z
piñata
z
piñata
z
piñata
z
piñata
)r' )r r r r r Ú*test_entities_in_text_converted_to_unicode@ s
zCHTMLTreeBuilderSmokeTest.test_entities_in_text_converted_to_unicodec C s | dd¡ d S )Nz#
I said "good day!"
z
I said "good day!"
)r' )r r r r Ú,test_quot_entity_converted_to_quotation_markG s zEHTMLTreeBuilderSmokeTest.test_quot_entity_converted_to_quotation_markc C s, d}| d|¡ | d|¡ | d|¡ d S )Nu �z�z�z
�)r' )r r r r r Útest_out_of_range_entityK s z1HTMLTreeBuilderSmokeTest.test_out_of_range_entityc C s< | d¡}| d|jjjj¡ | d|jj¡ | |¡ dS )zDMostly to prevent a recurrence of a bug in the html5lib treebuilder.z!
foo
rb N)r r Úh2rR r( r rb r, )r r r r r Útest_multipart_stringsQ s
z/HTMLTreeBuilderSmokeTest.test_multipart_stringsc C s | dd¡ | dd¡ dS )zqVerify consistent handling of empty-element tags,
no matter how they come in through the markup.
z
z
N)r' )r r r r rP X s c C s, d}| |¡}| d|jj¡ | |¡ dS )z8Prevent recurrence of a bug in the html5lib treebuilder.z?
foo
N)r ZassertNotEqualrh Úbodyr, )r Úcontentr r r r Ú#test_head_tag_between_head_and_body_ s
z
)r' )r r„ Úexpectedr r r Ú1test_entities_in_strings_converted_during_parsingÀ s zJHTMLTreeBuilderSmokeTest.test_entities_in_strings_converted_during_parsingc C s" d}| |¡}| |jjd¡ d S )Ns
‘Foo’
u ‘Foo’)r r rb rR )r Úquoter r r r Ú)test_smart_quotes_converted_on_the_way_inÇ s
zBHTMLTreeBuilderSmokeTest.test_smart_quotes_converted_on_the_way_inc C s | d¡}| |jjd¡ d S )Nzu   )r r r— rR )r r r r r Ú0test_non_breaking_spaces_converted_on_the_way_inÐ s
zIHTMLTreeBuilderSmokeTest.test_non_breaking_spaces_converted_on_the_way_inc C s0 d}d d¡}| |¡}| |j d¡|¡ d S )Nz-
zutf-8)ra r r rb )r r„ r´ r r r r Ú&test_entities_converted_on_the_way_outÔ s
z?HTMLTreeBuilderSmokeTest.test_entities_converted_on_the_way_outc C sH d}| d¡}| |¡}| d¡}| dd¡}| d¡}| ||¡ d S )Nu„
z
iso-8859-1zutf-8z ISO-Latin-1)ra r rs r )r Ú unicode_htmlZiso_latin_htmlr Úresultr´ r r r Ú test_real_iso_latin_documentÚ s
z5HTMLTreeBuilderSmokeTest.test_real_iso_latin_documentc C sL d}| d¡}| |¡}| | d¡| d¡¡ | | d¡| d¡¡ d S )Nsk
z shift-jiszutf-8Úeuc_jp)r! r r ra )r Úshift_jis_htmlrº r r r r Ú test_real_shift_jis_documentõ s
z5HTMLTreeBuilderSmokeTest.test_real_shift_jis_documentc C sB d}| j |dd}|jdks t‚| | d¡| d¡ d¡¡ d S )Ns› Hebrew (ISO 8859-8) in Visual Directionality
Hebrew (ISO 8859-8) in Visual Directionality
íåìùz iso8859-8)Z
from_encoding)z iso8859-8z
iso-8859-8zutf-8)r Zoriginal_encodingr% r ra r! )r Zhebrew_documentr r r r Útest_real_hebrew_document s
z2HTMLTreeBuilderSmokeTest.test_real_hebrew_documentc C s` d}d| }| |¡}| dddi¡}|d }| d|¡ | t|tƒ¡ | d| d ¡¡ d S )
NzEzj
%s
Shift-JIS markup goes here.rG z
http-equivz Content-typer¥ ztext/html; charset=x-sjisztext/html; charset=utf8r^ )r r‚ r r~ r2 r ra )r Úmeta_tagr¾ r Ú parsed_metar¥ r r r Ú'test_meta_tag_reflects_current_encoding s
z@HTMLTreeBuilderSmokeTest.test_meta_tag_reflects_current_encodingc C s^ d}d| }| |¡}|jddd}|d }| d|¡ | t|tƒ¡ | d| d¡¡ d S ) Nz'zj
%s
Shift-JIS markup goes here.rG Úencoding)ÚidÚcharsetzx-sjisr^ )r r‚ r r~ r2 r ra )r rÁ r¾ r r rÆ r r r Ú3test_html5_style_meta_tag_reflects_current_encoding/ s
zLHTMLTreeBuilderSmokeTest.test_html5_style_meta_tag_reflects_current_encodingc C s\ xVdD ]N}| |¡}x>tD ]6}|dkr(q| |¡}d|ks>t‚| d¡|kst‚qW qW d S )N)sB )ÚidnaÚmbcsÚoemÚ undefinedÚ
string_escapez
string-escapes meta charset=""Úascii)r r ra r% )r r r rÄ Úencodedr r r Ú2test_python_specific_encodings_not_used_in_charsetG s
zKHTMLTreeBuilderSmokeTest.test_python_specific_encodings_not_used_in_charsetc C s* | d¡}d|jd< | d|j ¡ ¡ d S )Nz textr r_ ztext)r r— r r! )r Údatar r r Ú5test_tag_with_no_attributes_can_have_attributes_added^ s
zNHTMLTreeBuilderSmokeTest.test_tag_with_no_attributes_can_have_attributes_addedc C s | d¡}| d|j ¡ ¡ d S )Nz0