o
ek
@ s d Z ddlmZ ddlZddlZddlZddlZddlmZm Z ddl
m Z m Z ddl
mZmZmZmZmZmZmZ ddlZddlmZmZmZ ddlmZmZm Z ddl Z z dd l
m Z m Z d
Z W n e!yv Z" zd Z W Y dZ"["ndZ"["ww ej#d d koej#d
k Z$G dd deZ%G dd deZ&G dd deZ&G dd deZ'G dd dej(Z)G dd deZ*G dd dej(Z+G dd d eZ,G d d d ej(Z-dS ) z#Tests of Beautiful Soup as a whole. ) set_traceN)
BeautifulSoupBeautifulStoneSoup) TreeBuilderParserRejectedMarkup)CharsetMetaAttributeValueCommentContentMetaAttributeValue SoupStrainerNamespacedAttributeTagNavigableString)EntitySubstitution
UnicodeDammitEncodingDetector)default_builderSoupTestskipIf)LXMLTreeBuilderLXMLTreeBuilderForXMLTF )r c @ sD e Zd Zdd Zdd Zdd Zdd Zd d
Zd d Zd
d Z dS )TestConstructorc C " d}| |}| d|jj d S )Nu
éé
u éésoup assertEqualh1stringselfdatar r" CD:\Projects\ConvertPro\env\Lib\site-packages\bs4/tests/test_soup.pytest_short_unicode_input3
z(TestConstructor.test_short_unicode_inputc C r )Nzfoo bar
zfoo barr r r" r" r# test_embedded_null8 r% z"TestConstructor.test_embedded_nullc C s, d d}| j|dgd}| d|j d S N
Räksmörgåsutf-8)Zexclude_encodingsz windows-1252)encoder r original_encoding)r utf8_datar r" r" r# test_exclude_encodings= s
z&TestConstructor.test_exclude_encodingsc C s G dd dt }tddd}tjdd tdd|i|}W d n1 s(w Y t|j|s5J | tdd |jj | d
|jj |di |}tjdd}td|dd }W d n1 sgw Y t
|d j }| d
szJ | ||j | ||j d S )Nc @ sD e Zd Zdd Zdd Zdd Zdd Zd d
Ze ZZ d d Z
d
S )z7TestConstructor.test_custom_builder_class..Mockc [ s" || _ d| _d| _g | _g | _d S )NTF) called_withZis_xmlZstore_line_numbersZcdata_list_attributesZpreserve_whitespace_tags)r kwargsr" r" r# __init__F s
z@TestConstructor.test_custom_builder_class..Mock.__init__c S d S Nr" r r r" r" r# initialize_soupL zGTestConstructor.test_custom_builder_class..Mock.initialize_soupc S s
|| _ d S r2 )fedr markupr" r" r# feedN
z.Mock.feedc S r1 r2 r" r r" r" r# resetP r5 z=TestConstructor.test_custom_builder_class..Mock.resetc S r1 r2 r" )r ignorer" r" r# r= R r5 z>TestConstructor.test_custom_builder_class..Mock.ignorec s dV d S )N)prepared markupzoriginal encodingzdeclared encodingz contains replacement charactersr" r argsr/ r" r" r# prepare_markupU s
zFTestConstructor.test_custom_builder_class..Mock.prepare_markupN) __name__
__module__ __qualname__r0 r4 r9 r< r= Zset_up_substitutionsZcan_be_empty_elementrA r" r" r" r# MockE s rE valueT)varZconvertEntitiesrecord builder)rG r> )rK Z
ignored_valuer zCKeyword arguments to the BeautifulSoup constructor will be ignored.)rJ r" )
objectdictwarningscatch_warningsr
isinstancerK r r. r6 strmessage
startswith)r rE r/ r rK wmsgr" r" r# test_custom_builder_classB s* z)TestConstructor.test_custom_builder_classc C s8 G dd dt }dd }dd l}| jtdtd|d d S ) Nc @ e Zd Zdd ZdS )z:TestConstructor.test_parser_markup_rejection..Mockc _ s t d)NzNope.)r r? r" r" r# r9 u s z?TestConstructor.test_parser_markup_rejection..Mock.feedN)rB rC rD r9 r" r" r" r# rE t s rE c s" t d d dfV t d d dfV d S )NF)r8 r? r" r" r# rA x s zDTestConstructor.test_parser_markup_rejection..prepare_markupr zoThe markup you provided was rejected by the parser. Trying a different parser or a different encoding may help.rJ )rK )r reassertRaisesRegexr r )r rE rA rX r" r" r# test_parser_markup_rejectionq s
z,TestConstructor.test_parser_markup_rejectionc C s d}| |}|j}| d|d | ddg|d | j |td d}| d|jd ddiddifD ]5}tjd d
}| j |d |d}W d n1 sPw Y |j}| d dg|d | d|d q5d S ) Nz&z an id idaclass)rK Zmulti_valued_attributesz a class *TrH Zan)r r\ r r rN rO )r r8 r r\ Z
switcheroorT r" r" r# test_cdata_list_attributes s
z*TestConstructor.test_cdata_list_attributesc sp G dd dt G dd dtG dd dt | jdt tt id}t fd d
| D s6J d S ) Nc @ e Zd ZdS )z9TestConstructor.test_replacement_classes..TagPlusNrB rC rD r" r" r" r# TagPlus rb c @ r` )z.StringPlusNra r" r" r" r#
StringPlus rc rd c @ r` )z=TestConstructor.test_replacement_classes..CommentPlusNra r" r" r" r# CommentPlus rc re z foobar)Zelement_classesc 3 s | ] }t | fV qd S r2 )rP ).0xre rd rb r" r# s
z;TestConstructor.test_replacement_classes..)r r
r r allZrecursiveChildGeneratorr3 r" rh r# test_replacement_classes s z(TestConstructor.test_replacement_classesN)
rB rC rD r$ r& r- rV rZ r_ rk r" r" r" r# r 1 s / r c @ sF e Zd ZdddZdd Zdd Zdd Zd
d Zd d
Zdd Z dS ) TestWarningsTc C s" | tjd d }| | d S )NP )rS r ZNO_PARSER_SPECIFIED_WARNING
assertTrue)r sZis_therevr" r" r# _no_parser_specified s z!TestWarnings._no_parser_specifiedc C sR t jdd
}| d}W d n1 sw Y t|d j}| | d S )NTrH r rN rO r rQ rR Z_assert_no_parser_specifiedr rT r rU r" r" r# #test_warning_if_no_parser_specified s
z0TestWarnings.test_warning_if_no_parser_specifiedc C sT t jdd}| dd}W d n1 sw Y t|d j}| | d S )NTrH rr htmlr rs rt r" r" r# *test_warning_if_parser_specified_too_vague s
z7TestWarnings.test_warning_if_parser_specified_too_vaguec C sH t jdd}| dd}W d n1 sw Y | g | d S )NTrH rr html.parser)rN rO r r )r rT r r" r" r# ,test_no_warning_if_explicit_parser_specified s z9TestWarnings.test_no_warning_if_explicit_parser_specifiedc C s| t jdd}| jdtdd}W d n1 sw Y t|d j}| d|v | d|v | d | d S )
NTrH rr b)parseOnlyTheser r{
parse_onlys ) rN rO r r
rQ rR rn r r* rt r" r" r# )test_parseOnlyThese_renamed_to_parse_only s z6TestWarnings.test_parseOnlyThese_renamed_to_parse_onlyc C sz t jdd}d}| j|dd}W d n1 sw Y t|d j}| d|v | d|v | d|j d S ) NTrH s éutf8) fromEncodingr r Z
from_encoding)rN rO r rQ rR rn r r+ )r rT r~ r rU r" r" r# *test_fromEncoding_renamed_to_from_encoding s z7TestWarnings.test_fromEncoding_renamed_to_from_encodingc C s | j t| jddd d S )NzT)Zno_such_argument) assertRaises TypeErrorr r; r" r" r# "test_unrecognized_keyword_argument s
z/TestWarnings.test_unrecognized_keyword_argumentN)T)
rB rC rD rq ru rw ry r} r r r" r" r" r# rl s
rl c @ s4 e Zd Zdd Zdd Zdd Zdd Zd d
Zd S ) rl c C s t }|j}z/tjdd
}| |}W d n1 s w Y t|d j}| d|v W | n| w tjdd
}| |}W d n1 sRw Y |
dt | d S )NTrH r zlooks like a filename) tempfileNamedTemporaryFilenamerN rO r rQ rR rn closer len)r Z
filehandlefilenamerT r rU r" r" r# test_disk_file_warning s z#TestWarnings.test_disk_file_warningc C R t jdd
}| d}W d n1 sw Y | tdd |D d S )NTrH s http://www.crummybytes.com/c s | ]
}d t |jv V qdS zlooks like a URLNrQ rR rf rT r" r" r# ri z?TestWarnings.test_url_warning_with_bytes_url..rN rO r rn anyr Z warning_listr r" r" r# test_url_warning_with_bytes_url s z,TestWarnings.test_url_warning_with_bytes_urlc C r )NTrH z http://www.crummyunicode.com/c s r r r r r" r" r# ri r zATestWarnings.test_url_warning_with_unicode_url..r r r" r" r# !test_url_warning_with_unicode_url s z.TestWarnings.test_url_warning_with_unicode_urlc C r )NTrH s$ http://www.crummybytes.com/ is greatc s r r r r r" r" r# ri r zETestWarnings.test_url_warning_with_bytes_and_space..rN rO r Z assertFalser r r" r" r# %test_url_warning_with_bytes_and_space z2TestWarnings.test_url_warning_with_bytes_and_spacec C r )NTrH z%http://www.crummyuncode.com/ is greatc s r r r r r" r" r# ri r zGTestWarnings.test_url_warning_with_unicode_and_space..r r r" r" r# 'test_url_warning_with_unicode_and_space r z4TestWarnings.test_url_warning_with_unicode_and_spaceN)rB rC rD r r r r r r" r" r" r# rl s c @ rW )TestSelectiveParsingc C s. d}t d}| j||d}| | d d S )Nz&NoYesNoYes Yesrz )r| s YesYes Yes)r
r r r* )r r8 Zstrainerr r" r" r# test_parse_with_soupstrainer s z1TestSelectiveParsing.test_parse_with_soupstrainerN)rB rC rD r r" r" r" r# r s r c @ sx e Zd ZdZdd Zdd Zdd Zdd Zd
d Zd d
Z dd Z
dd Z dd Z dd Z
dd Zdd Zdd Zd S ) TestEntitySubstitutionz1Standalone tests of the EntitySubstitution class.c C s
t | _d S r2 )r subr; r" r" r# setUp$ r: z TestEntitySubstitution.setUpc C s d}| | j|d d S )Nu foo∀☃õbaru foo∀☃õbarr r substitute_htmlr ro r" r" r# test_simple_html_substitution' s z4TestEntitySubstitution.test_simple_html_substitutionc C s& d}t |}| | j|jd d S )Ns fooz ‘’foo“”)r r r r r8 )r quotesdammitr" r" r# test_smart_quote_substitution. s
z4TestEntitySubstitution.test_smart_quote_substitutionc C s d}| | j|d| d S )NWelcome to "my bar"Fr r Zsubstitute_xmlr r" r" r# Itest_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_false6 s z`TestEntitySubstitution.test_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falsec C s0 | | jddd | | jddd d S )NZWelcomeTz "Welcome"z Bob's Barz "Bob's Bar"r r; r" r" r# 6test_xml_attribute_quoting_normally_uses_double_quotes: s zMTestEntitySubstitution.test_xml_attribute_quoting_normally_uses_double_quotesc C d}| | j|dd d S )Nr Tz'Welcome to "my bar"'r r r" r" r# Otest_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotes@ s zfTestEntitySubstitution.test_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotesc C r )NWelcome to "Bob's Bar"Tz""Welcome to "Bob's Bar""r r r" r" r# btest_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotesE s
zyTestEntitySubstitution.test_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotesc C s d}| | j|| d S )Nr r )r quotedr" r" r# zfoo<bar>r r; r" r" r# 'test_xml_quoting_handles_angle_bracketsO
z>TestEntitySubstitution.test_xml_quoting_handles_angle_bracketsc C r )NzAT&TzAT&Tr r; r" r" r# #test_xml_quoting_handles_ampersandsT s z:TestEntitySubstitution.test_xml_quoting_handles_ampersandsc C r )N ÁT&TzÁT&Tr r; r" r" r# Etest_xml_quoting_including_ampersands_when_they_are_part_of_an_entityW r z\TestEntitySubstitution.test_xml_quoting_including_ampersands_when_they_are_part_of_an_entityc C r )Nr zÁT&T)r r Z"substitute_xml_containing_entitiesr; r" r" r# Dtest_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entity\ r z[TestEntitySubstitution.test_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityc C s d}| | j|| dS )z:There's no need to do this except inside attribute values.z Bob's "bar"Nr )r textr" r" r# test_quotes_not_html_substituteda s z7TestEntitySubstitution.test_quotes_not_html_substitutedN)rB rC rD __doc__r r r r r r r r r r r r r r" r" r" r# r " s r c sN e Zd Z fddZdd Zdd Zdd Zd d
Zee d d d
Z
Z S )TestEncodingConversionc s4 t t| d| _| jd| _| | jd d S )NuU Sacré bleu!r) sU Sacré bleu!)superr r unicode_datar* r, r r; __class__r" r# r k s z TestEncodingConversion.setUpc C s t jj}ttj zTestEncodingConversion.test_ascii_in_unicode_out..noops ar) )bs4r chardet_dammitloggingdisableWARNINGr decodern rP rQ r Z document_forr+ lowerNOTSET)r chardetr asciiZsoup_from_asciiZunicode_outputr" r" r# test_ascii_in_unicode_outt s
z0TestEncodingConversion.test_ascii_in_unicode_outc C s@ | | j}| | | j | |jjd | |jd d S Nu Sacré bleu!)r r r r foor r+ r Zsoup_from_unicoder" r" r# test_unicode_in_unicode_out s z2TestEncodingConversion.test_unicode_in_unicode_outc C s2 | | j}| | | j | |jjd d S r )r r, r r r r r )r Zsoup_from_utf8r" r" r# test_utf8_in_unicode_out s z/TestEncodingConversion.test_utf8_in_unicode_outc C s$ | | j}| |d| j d S )Nr) )r r r r* r, r r" r" r#
test_utf8_out s z$TestEncodingConversion.test_utf8_outzQBad HTMLParser detected; skipping test of non-ASCII characters in attribute name.c C s( d}| | |jd|d d S )Nu r~ )r r divr* r7 r" r" r# 1test_attribute_name_containing_unicode_characters s $zHTestEncodingConversion.test_attribute_name_containing_unicode_characters) rB rC rD r r r r r r PYTHON_3_PRE_3_2r
__classcell__r" r" r r# r g s r c @ s e Zd ZdZdd Zdd Zdd Zdd Zd
d Zd d
Z dd Z
dd Z dd Z dd Z
dd Zdd Zdd Zd d Zd d Zd d! Zd"d# Zd$d% Zd&S )'TestUnicodeDammitz"Standalone tests of UnicodeDammit.c C s d}t |}| |j| d S )Nu I'm already Unicode! ☃r r unicode_markupr r8 r r" r" r# test_unicode_input s z$TestUnicodeDammit.test_unicode_inputc C s d}t |}| |jd d S )N u ‘’“”r r r" r" r# test_smart_quotes_to_unicode s
z.TestUnicodeDammit.test_smart_quotes_to_unicodec C " d}t |dd}| |jd d S )Nr xmlZsmart_quotes_toz+‘’“”r r r" r" r# !test_smart_quotes_to_xml_entities
z3TestUnicodeDammit.test_smart_quotes_to_xml_entitiesc C r )Nr rv r z'‘’“”r r r" r" r# "test_smart_quotes_to_html_entities r z4TestUnicodeDammit.test_smart_quotes_to_html_entitiesc C r )Nr r r z''""r r r" r" r# test_smart_quotes_to_ascii r z,TestUnicodeDammit.test_smart_quotes_to_asciic C s0 d}t |}| |j d | |jd d S )Ns Sacré bleu! ☃r) u Sacré bleu! ☃r r r+ r r )r r~ r r" r" r# test_detect_utf8 s z"TestUnicodeDammit.test_detect_utf8c C s4 d}t |dg}| |j d | |jd d S )Ns
iso-8859-8u םולשr )r hebrewr r" r" r# test_convert_hebrew s z%TestUnicodeDammit.test_convert_hebrewc C s6 d}t |}| |j d | |jd| d S )Ns ケータイ Watchr) )r r r+ r r r* )r utf_8r r" r" r# /test_dont_see_smart_quotes_where_there_are_none s zATestUnicodeDammit.test_dont_see_smart_quotes_where_there_are_nonec C s, d d}t|dg}| |j d d S )Nr( r) r r* r r r+ r r r, r r" r" r# test_ignore_inappropriate_codecs s
z2TestUnicodeDammit.test_ignore_inappropriate_codecsc C s6 d d}dD ]}t||g}| |j d qd S )Nr( r) )z.utf8z...z
utF---16.!r )r r, Z bad_encodingr r" r" r# test_ignore_invalid_codecs s
z,TestUnicodeDammit.test_ignore_invalid_codecsc C sL d d}t|dgd}| |j d t|ddgd}| |jd d S r' r r r" r" r# r- s
z(TestUnicodeDammit.test_exclude_encodingsc C s" t d}t|j}d|v sJ d S )Ns' u utf-�)r list encodings)r Zdetectedr r" r" r# Ptest_encoding_detector_replaces_junk_in_encoding_name_with_replacement_character s
zbTestUnicodeDammit.test_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterc C s( dD ]}t |dd}| d|j qd S )N)s& s& s$ s# Tis_htmleuc-jp)r r r+ r r! r r" r" r# test_detect_html5_style_meta_tag s z2TestUnicodeDammit.test_detect_html5_style_meta_tagc C s d}t jj}ttj z3dd }|t j_t|}| d|j | d|j
v t |d}| |j W ttj |t j_d S ttj |t j_w )NsT
بتر
ѐc S r1 r2 r" r r" r" r# r r5 zBTestUnicodeDammit.test_last_ditch_entity_replacement..noopTu �rx )
r r r r r r r r Z contains_replacement_charactersrn r r r )r docr r r r r" r" r# "test_last_ditch_entity_replacement s
z4TestUnicodeDammit.test_last_ditch_entity_replacementc C s, d}t |}| d|j | d|j d S )Ns < a > < / a > u áézutf-16le)r r r r+ r r" r" r# test_byte_order_mark_removed# s z.TestUnicodeDammit.test_byte_order_mark_removedc C sP d d}d d}|| | }| t|jd t|}| d|d d S )Nu ☃☃☃r~ u “Hi, I like Windows!” windows_1252u+ ☃☃☃“Hi, I like Windows!”☃☃☃)r* r UnicodeDecodeErrorr r detwingler )r r~ r r fixedr" r" r# test_detwingle* s
z TestUnicodeDammit.test_detwinglec C s> dD ]}| d}| |d t|}| || qd S )N)u œu ₓu ðr~ )r* rn endswithr r r )r Ztricky_unicode_charinputoutputr" r" r# +test_detwingle_ignores_multibyte_charactersB s
z=TestUnicodeDammit.test_detwingle_ignores_multibyte_charactersc C s d}| d}d}| d}tj}| d ||dd | d||dd | d||dd | d|| | d|| d d
}| d ||| | d ||| | d||| ddd | d||dd | d |d
| dd d S )Nz0r z,Fr r) Tz
iso-8859-1 i )r search_entire_document)r a)r* r Zfind_declared_encodingr )r Z html_unicodeZ
html_bytesZ xml_unicodeZ xml_bytesmZspacerr" r" r# test_find_declared_encodingQ s0
z-TestUnicodeDammit.test_find_declared_encodingN)rB rC rD r r r r r r r r r r r r- r r r r r r r r" r" r" r# r s( # r c @ s$ e Zd Zdd Zdd Zdd ZdS )TestNamedspacedAttributec C s. t dd }| |d t d}| |d d S )NZxmlnsr r r r\ r" r" r# test_name_may_be_none_or_missing s
z9TestNamedspacedAttribute.test_name_may_be_none_or_missingc C s t dd}| d| d S )Nr\ rz za:br
r r" r" r# 6test_attribute_is_equivalent_to_colon_separated_string s
zOTestNamedspacedAttribute.test_attribute_is_equivalent_to_colon_separated_stringc C sp t ddd}t ddd}| || t ddd }| || t ddd}| || t ddd}| || d S )Nr\ rz cz)r r ZassertNotEqual)r r\ rz r der" r" r# ;test_attributes_are_equivalent_if_prefix_and_name_identical s zTTestNamedspacedAttribute.test_attributes_are_equivalent_if_prefix_and_name_identicalN)rB rC rD r r
r r" r" r" r# r s r c @ s e Zd Zdd Zdd ZdS ))TestAttributeValueWithCharsetSubstitutionc C s8 t d}| d| | d|j | d|d d S )Nr r~ )r r original_valuer* r rF r" r" r# !test_content_meta_attribute_value zKTestAttributeValueWithCharsetSubstitution.test_content_meta_attribute_valuec C s8 t d}| d| | d|j | d|d d S )Nztext/html; charset=euc-jpztext/html; charset=utf8r~ )r r r r* r r" r" r# r r N)rB rC rD r r" r" r" r# r s r ).r pdbr r Zunittestsysr r r r Z bs4.builderr r Z bs4.elementr r r r
r r r
Z
bs4.dammitr r r Z bs4.testingr r r rN r r Z LXML_PRESENT ImportErrorr version_infor r rl r ZTestCaser r r r r r" r" r" r# s@ $
,/ E< ]