B
j9C\YO ã
@ s~ d Z ddlmZ ddlZddlZddlZddlZddlmZm Z ddl
m Z m Z m
Z
mZ ddlZddlmZmZmZ ddlmZmZ ddlZyddlmZmZ d ZW n& ek
rÊ Z zd
ZW ddZ [ X Y nX ej d d koâej d k Z G d
d„ deƒZ G dd„ deƒZ G dd„ deƒZ G dd„ deƒZ!G dd„ dej"ƒZ#G dd„ deƒZ$G dd„ dej"ƒZ%G dd„ deƒZ&G d d „ d ej"ƒZ'dS ) z#Tests of Beautiful Soup as a whole.é )Ú set_traceN)Ú
BeautifulSoupÚBeautifulStoneSoup)ÚCharsetMetaAttributeValueÚContentMetaAttributeValueÚ SoupStrainerÚNamespacedAttribute)ÚEntitySubstitutionÚ
UnicodeDammitÚEncodingDetector)ÚSoupTestÚskipIf)ÚLXMLTreeBuilderÚLXMLTreeBuilderForXMLTFé )r é c @ s$ e Zd Zdd„ Zdd„ Zdd„ ZdS )ÚTestConstructorc C s" d}| |¡}| d|jj¡ d S )Nu
éé
u éé)ÚsoupÚ assertEqualÚh1Ústring)ÚselfÚdatar © r ú2lib/python3.7/site-packages/bs4/tests/test_soup.pyÚtest_short_unicode_input* s
z(TestConstructor.test_short_unicode_inputc C s" d}| |¡}| d|jj¡ d S )Nzfoo bar
zfoo bar)r r r r )r r r r r r Útest_embedded_null/ s
z"TestConstructor.test_embedded_nullc C s, d d¡}| j|dgd}| d|j¡ d S )Nu
RäksmörgÃ¥szutf-8)Úexclude_encodingsz windows-1252)Úencoder r Úoriginal_encoding)r Ú utf8_datar r r r Útest_exclude_encodings4 s
z&TestConstructor.test_exclude_encodingsN)Ú__name__Ú
__module__Ú __qualname__r r r! r r r r r ( s r c @ sF e Zd Zddd„Zdd„ Zdd„ Zdd „ Zd
d „ Zd d
„ Zdd„ Z dS )Ú TestWarningsTc C s" | tjd d… ¡}| |¡ d S )NéP )Ú
startswithr ZNO_PARSER_SPECIFIED_WARNINGÚ
assertTrue)r ÚsZis_thereÚvr r r Ú_no_parser_specified< s z!TestWarnings._no_parser_specifiedc C s> t jdd}| d¡}W d Q R X t|d jƒ}| |¡ d S )NT)Úrecordzr )ÚwarningsÚcatch_warningsr ÚstrÚmessageÚ_assert_no_parser_specified)r Úwr Úmsgr r r Ú#test_warning_if_no_parser_specified@ s z0TestWarnings.test_warning_if_no_parser_specifiedc C s@ t jdd}| dd¡}W d Q R X t|d jƒ}| |¡ d S )NT)r, zÚhtmlr )r- r. r r/ r0 r1 )r r2 r r3 r r r Ú*test_warning_if_parser_specified_too_vagueF s z7TestWarnings.test_warning_if_parser_specified_too_vaguec C s4 t jdd}| dd¡}W d Q R X | g |¡ d S )NT)r, zz html.parser)r- r. r r )r r2 r r r r Ú,test_no_warning_if_explicit_parser_specifiedL s z9TestWarnings.test_no_warning_if_explicit_parser_specifiedc C sh t jdd}| jdtdƒd}W d Q R X t|d jƒ}| d|k¡ | d|k¡ | d | ¡ ¡ d S )
NT)r, zÚb)ÚparseOnlyTheser r9 Ú
parse_onlys ) r- r. r r r/ r0 r( r r )r r2 r r3 r r r Ú)test_parseOnlyThese_renamed_to_parse_onlyQ s z6TestWarnings.test_parseOnlyThese_renamed_to_parse_onlyc C sf t jdd}d}| j|dd}W d Q R X t|d jƒ}| d|k¡ | d|k¡ | d|j¡ d S ) NT)r, s éÚutf8)Ú fromEncodingr r= Z
from_encoding)r- r. r r/ r0 r( r r )r r2 r< r r3 r r r Ú*test_fromEncoding_renamed_to_from_encodingY s z7TestWarnings.test_fromEncoding_renamed_to_from_encodingc C s | j t| jddd d S )NzT)Zno_such_argument)Ú assertRaisesÚ TypeErrorr )r r r r Ú"test_unrecognized_keyword_argumentb s z/TestWarnings.test_unrecognized_keyword_argumentN)T)
r" r# r$ r+ r4 r6 r7 r; r> rA r r r r r% : s
r% c @ s4 e Zd Zdd„ Zdd„ Zdd„ Zdd„ Zd d
„ Zd S ) r% c C s’ t ¡ }|j}zBtjdd}| |¡}W d Q R X t|d jƒ}| d|k¡ W d | ¡ X tjdd}| |¡}W d Q R X |
dt |ƒ¡ d S )NT)r, r zlooks like a filename) ÚtempfileZNamedTemporaryFileÚnamer- r. r r/ r0 r( Úcloser Úlen)r Z
filehandleÚfilenamer2 r r3 r r r Útest_disk_file_warningh s
z#TestWarnings.test_disk_file_warningc C s> t jdd}| d¡}W d Q R X | tdd„ |D ƒƒ¡ d S )NT)r, s http://www.crummybytes.com/c s s | ]}d t |jƒkV qdS )zlooks like a URLN)r/ r0 )Ú.0r2 r r r ú } s z?TestWarnings.test_url_warning_with_bytes_url..)r- r. r r( Úany)r Ú warning_listr r r r Ú test_url_warning_with_bytes_urlx s z,TestWarnings.test_url_warning_with_bytes_urlc C s> t jdd}| d¡}W d Q R X | tdd„ |D ƒƒ¡ d S )NT)r, z http://www.crummyunicode.com/c s s | ]}d t |jƒkV qdS )zlooks like a URLN)r/ r0 )rH r2 r r r rI … s zATestWarnings.test_url_warning_with_unicode_url..)r- r. r r( rJ )r rK r r r r Ú!test_url_warning_with_unicode_url€ s z.TestWarnings.test_url_warning_with_unicode_urlc C s> t jdd}| d¡}W d Q R X | tdd„ |D ƒƒ¡ d S )NT)r, s$ http://www.crummybytes.com/ is greatc s s | ]}d t |jƒkV qdS )zlooks like a URLN)r/ r0 )rH r2 r r r rI ‹ s zETestWarnings.test_url_warning_with_bytes_and_space..)r- r. r Ú assertFalserJ )r rK r r r r Ú%test_url_warning_with_bytes_and_spaceˆ s z2TestWarnings.test_url_warning_with_bytes_and_spacec C s> t jdd}| d¡}W d Q R X | tdd„ |D ƒƒ¡ d S )NT)r, z%http://www.crummyuncode.com/ is greatc s s | ]}d t |jƒkV qdS )zlooks like a URLN)r/ r0 )rH r2 r r r rI ‘ s zGTestWarnings.test_url_warning_with_unicode_and_space..)r- r. r rN rJ )r rK r r r r Ú'test_url_warning_with_unicode_and_spaceŽ s z4TestWarnings.test_url_warning_with_unicode_and_spaceN)r" r# r$ rG rL rM rO rP r r r r r% f s
c @ s e Zd Zdd„ ZdS )ÚTestSelectiveParsingc C s. d}t dƒ}| j||d}| | ¡ d¡ d S )Nz&NoYesNoYes Yesr8 )r: s YesYes Yes)r r r r )r ÚmarkupZstrainerr r r r Ú test_parse_with_soupstrainer— s z1TestSelectiveParsing.test_parse_with_soupstrainerN)r" r# r$ rS r r r r rQ • s rQ 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 ) ÚTestEntitySubstitutionz1Standalone tests of the EntitySubstitution class.c C s
t | _d S )N)r Úsub)r r r r ÚsetUp s z TestEntitySubstitution.setUpc C s d}| | j |¡d¡ d S )Nu foo∀☃õbaru foo∀☃õbar)r rU Úsubstitute_html)r r) r r r Ú test_simple_html_substitution£ s z4TestEntitySubstitution.test_simple_html_substitutionc C s& d}t |ƒ}| | j |j¡d¡ d S )Ns ‘’foo“”z ‘’foo“”)r
r rU rW rR )r ZquotesÚdammitr r r Ú test_smart_quote_substitutionª s z4TestEntitySubstitution.test_smart_quote_substitutionc C s d}| | j |d¡|¡ d S )NzWelcome to "my bar"F)r rU Úsubstitute_xml)r r) r r r ÚItest_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_false² s z`TestEntitySubstitution.test_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falsec C s0 | | j dd¡d¡ | | j dd¡d¡ d S )NZWelcomeTz "Welcome"z Bob's Barz "Bob's Bar")r rU r[ )r r r r Ú6test_xml_attribute_quoting_normally_uses_double_quotes¶ s zMTestEntitySubstitution.test_xml_attribute_quoting_normally_uses_double_quotesc C s d}| | j |d¡d¡ d S )NzWelcome to "my bar"Tz'Welcome to "my bar"')r rU r[ )r r) r r r ÚOtest_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotes¼ s zfTestEntitySubstitution.test_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotesc C s d}| | j |d¡d¡ d S )NzWelcome to "Bob's Bar"Tz""Welcome to "Bob's Bar"")r rU r[ )r r) r r r Úbtest_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotesÁ s zyTestEntitySubstitution.test_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotesc C s d}| | j |¡|¡ d S )NzWelcome to "Bob's Bar")r rU r[ )r Zquotedr r r Úzfoo<bar>)r rU r[ )r r r r Ú'test_xml_quoting_handles_angle_bracketsË s
z>TestEntitySubstitution.test_xml_quoting_handles_angle_bracketsc C s | | j d¡d¡ d S )NzAT&TzAT&T)r rU r[ )r r r r Ú#test_xml_quoting_handles_ampersandsÐ s z:TestEntitySubstitution.test_xml_quoting_handles_ampersandsc C s | | j d¡d¡ d S )Nz ÁT&TzÁT&T)r rU r[ )r r r r ÚEtest_xml_quoting_including_ampersands_when_they_are_part_of_an_entityÓ s
z\TestEntitySubstitution.test_xml_quoting_including_ampersands_when_they_are_part_of_an_entityc C s | | j d¡d¡ d S )Nz ÁT&TzÁT&T)r rU Z"substitute_xml_containing_entities)r r r r ÚDtest_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityØ s
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"N)r rU rW )r Útextr r r Ú test_quotes_not_html_substitutedÝ s z7TestEntitySubstitution.test_quotes_not_html_substitutedN)r" r# r$ Ú__doc__rV rX rZ r\ r] r^ r_ r` ra rb rc rd rf r r r r rT ž s rT c sN e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Zee d ƒd d
„ ƒZ
‡ Z S )ÚTestEncodingConversionc s4 t t| ƒ ¡ d| _| j d¡| _| | jd¡ d S )NuU Sacré bleu!zutf-8sU Sacré bleu!)Úsuperrh rV Ú unicode_datar r r )r )Ú __class__r r rV ç s z TestEncodingConversion.setUpc C s’ t jj}t tj¡ zbdd„ }|t j_d}| |¡}| ¡ }| t |t
ƒ¡ | || | ¡ ¡¡ | |j
¡ d¡ W d t tj¡ |t j_X d S )Nc S s d S )Nr )r/ r r r Únoopö s z>TestEncodingConversion.test_ascii_in_unicode_out..noops azutf-8)Úbs4rY Úchardet_dammitÚloggingÚdisableÚWARNINGr Údecoder( Ú
isinstancer/ r Z document_forr ÚlowerÚNOTSET)r Úchardetrl ÚasciiZsoup_from_asciiZunicode_outputr r r Útest_ascii_in_unicode_outð s
z0TestEncodingConversion.test_ascii_in_unicode_outc C s@ | | j¡}| | ¡ | j¡ | |jjd¡ | |jd ¡ d S )Nu Sacré bleu!)r rj r rr Úfoor r )r Úsoup_from_unicoder r r Útest_unicode_in_unicode_out s z2TestEncodingConversion.test_unicode_in_unicode_outc C s2 | | j¡}| | ¡ | j¡ | |jjd¡ d S )Nu Sacré bleu!)r r r rr rj ry r )r Zsoup_from_utf8r r r Útest_utf8_in_unicode_out s z/TestEncodingConversion.test_utf8_in_unicode_outc C s$ | | j¡}| | d¡| j¡ d S )Nzutf-8)r rj r r r )r rz r r r Ú
test_utf8_out s z$TestEncodingConversion.test_utf8_outzQBad HTMLParser detected; skipping test of non-ASCII characters in attribute name.c C s( d}| | |¡j d¡| d¡¡ d S )Nu r< )r r Zdivr )r rR r r r Ú1test_attribute_name_containing_unicode_characters s zHTestEncodingConversion.test_attribute_name_containing_unicode_characters) r" r# r$ rV rx r{ r| r} r
ÚPYTHON_3_PRE_3_2r~ Ú
__classcell__r r )rk r rh ã s rh 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$S )%ÚTestUnicodeDammitz"Standalone tests of UnicodeDammit.c C s d}t |ƒ}| |j|¡ d S )Nu I'm already Unicode! ☃)r
r Úunicode_markup)r rR rY r r r Útest_unicode_input" s z$TestUnicodeDammit.test_unicode_inputc C s d}t |ƒ}| |jd¡ d S )Ns ‘’“”u ‘’“â€)r
r r‚ )r rR rY r r r Ú test_smart_quotes_to_unicode' s z.TestUnicodeDammit.test_smart_quotes_to_unicodec C s" d}t |dd}| |jd¡ d S )Ns ‘’“”Zxml)Úsmart_quotes_toz+‘’“”)r
r r‚ )r rR rY r r r Ú!test_smart_quotes_to_xml_entities- s z3TestUnicodeDammit.test_smart_quotes_to_xml_entitiesc C s" d}t |dd}| |jd¡ d S )Ns ‘’“”r5 )r… z'‘’“”)r
r r‚ )r rR rY r r r Ú"test_smart_quotes_to_html_entities3 s z4TestUnicodeDammit.test_smart_quotes_to_html_entitiesc C s" d}t |dd}| |jd¡ d S )Ns ‘’“”rw )r… z''"")r
r r‚ )r rR rY r r r Útest_smart_quotes_to_ascii9 s z,TestUnicodeDammit.test_smart_quotes_to_asciic C s0 d}t |ƒ}| |j ¡ d¡ | |jd¡ d S )Ns Sacré bleu! ☃zutf-8u Sacré bleu! ☃)r
r r rt r‚ )r r< rY r r r Útest_detect_utf8? s z"TestUnicodeDammit.test_detect_utf8c C s4 d}t |dgƒ}| |j ¡ d¡ | |jd¡ d S )Ns íåìùz
iso-8859-8u ×ולש)r
r r rt r‚ )r ÚhebrewrY r r r Útest_convert_hebrewF s z%TestUnicodeDammit.test_convert_hebrewc C s6 d}t |ƒ}| |j ¡ d¡ | |j d¡|¡ d S )Ns ケータイ Watchzutf-8)r
r r rt r‚ r )r Úutf_8rY r r r Ú/test_dont_see_smart_quotes_where_there_are_noneL s zATestUnicodeDammit.test_dont_see_smart_quotes_where_there_are_nonec C s, d d¡}t|dgƒ}| |j ¡ d¡ d S )Nu
Räksmörgåszutf-8z
iso-8859-8)r r
r r rt )r r rY r r r Ú test_ignore_inappropriate_codecsR s
z2TestUnicodeDammit.test_ignore_inappropriate_codecsc C s: d d¡}x*dD ]"}t||gƒ}| |j ¡ d¡ qW d S )Nu
Räksmörgåszutf-8)z.utf8z...z
utF---16.!)r r
r r rt )r r Z bad_encodingrY r r r Útest_ignore_invalid_codecsW s
z,TestUnicodeDammit.test_ignore_invalid_codecsc C sL d d¡}t|dgd}| |j ¡ d¡ t|ddgd}| |jd ¡ d S )Nu
Räksmörgåszutf-8)r z windows-1252)r r
r r rt )r r rY r r r r! ] s
z(TestUnicodeDammit.test_exclude_encodingsc C s" t dƒ}t|jƒ}d|ks t‚d S )Ns' u utf-�)r ÚlistÚ encodingsÚAssertionError)r Zdetectedr‘ r r r ÚPtest_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterk s
zbTestUnicodeDammit.test_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterc C s, x&dD ] }t |dd}| d|j¡ qW d S )N)s& s& s$ s# T)Zis_htmlzeuc-jp)r
r r )r r rY r r r Ú test_detect_html5_style_meta_tagq s
z2TestUnicodeDammit.test_detect_html5_style_meta_tagc C s„ d}t jj}t tj¡ zPdd„ }|t j_t|ƒ}| d|j¡ | d|j
k¡ t |dƒ}| |j¡ W d t tj ¡ |t j_X d S )NsT 
بتر
ÈÒÑÊÑíäc S s d S )Nr )r/ r r r rl ’ s zBTestUnicodeDammit.test_last_ditch_entity_replacement..noopTu �z html.parser)
rm rY rn ro rp rq r
r Z contains_replacement_charactersr( r‚ r ru )r Údocrv rl rY r r r r Ú"test_last_ditch_entity_replacement| s
z4TestUnicodeDammit.test_last_ditch_entity_replacementc C s, d}t |ƒ}| d|j¡ | d|j¡ d S )Ns ÿþ< a > á é < / a > u áézutf-16le)r
r r‚ r )r r rY r r r Ú test_byte_order_mark_removedŸ 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!â€Ú windows_1252u+ ☃☃☃“Hi, I like Windows!â€â˜ƒâ˜ƒâ˜ƒ)r r? ÚUnicodeDecodeErrorrr r
Ú detwingler )r r< r˜ r• Zfixedr r r Útest_detwingle¦ s
z TestUnicodeDammit.test_detwinglec C sB x s6 ,/ E< /