3
*6]dO @ s~ d Z ddlmZ ddlZddlZddlZddlZddlmZm Z ddl
m Z m Z m
Z
mZ ddlZddlmZmZmZ ddlmZmZ ddlZyddlmZmZ d ZW n& ek
r Z z
d
ZW Y ddZ [ X nX ej d d koej d k Z G d
d deZ G dd deZ G dd deZ G dd deZ!G dd dej"Z#G dd deZ$G dd dej"Z%G dd deZ&G d d d ej"Z'dS ) z#Tests of Beautiful Soup as a whole. ) set_traceN)
BeautifulSoupBeautifulStoneSoup)CharsetMetaAttributeValueContentMetaAttributeValue SoupStrainerNamespacedAttribute)EntitySubstitution
UnicodeDammitEncodingDetector)SoupTestskipIf)LXMLTreeBuilderLXMLTreeBuilderForXMLTF c @ s$ e Zd Zdd Zdd Zdd ZdS )TestConstructorc C s" d}| j |}| jd|jj d S )Nu
éé
u éé)soup assertEqualh1string)selfdatar r /usr/lib/python3.6/test_soup.pytest_short_unicode_input* s
z(TestConstructor.test_short_unicode_inputc C s" d}| j |}| jd|jj d S )Nzfoo bar
zfoo bar)r r r r )r r r r r r test_embedded_null/ s
z"TestConstructor.test_embedded_nullc C s, dj d}| j|dgd}| jd|j d S )Nu
Räksmörgåszutf-8)exclude_encodingsz windows-1252)encoder r original_encoding)r utf8_datar r r r test_exclude_encodings4 s
z&TestConstructor.test_exclude_encodingsN)__name__
__module__ __qualname__r r r! r r r r r ( s r c @ sF e Zd ZdddZdd Zdd Zdd Zd
d Zd d
Zdd Z dS ) TestWarningsTc C s" |j tjd d }| j| d S )NP )
startswithr ZNO_PARSER_SPECIFIED_WARNING
assertTrue)r sZis_therevr r r _no_parser_specified< s z!TestWarnings._no_parser_specifiedc
C s> t jdd}| jd}W d Q R X t|d j}| j| d S )NT)recordzr )warningscatch_warningsr strmessage_assert_no_parser_specified)r wr msgr r r #test_warning_if_no_parser_specified@ s z0TestWarnings.test_warning_if_no_parser_specifiedc
C s@ t jdd}| jdd}W d Q R X t|d j}| j| d S )NT)r, zhtmlr )r- r. r r/ r0 r1 )r r2 r r3 r r r *test_warning_if_parser_specified_too_vagueF s z7TestWarnings.test_warning_if_parser_specified_too_vaguec
C s4 t jdd}| jdd}W d Q R X | jg | d S )NT)r, zz html.parser)r- r. r r )r r2 r r r r ,test_no_warning_if_explicit_parser_specifiedL s z9TestWarnings.test_no_warning_if_explicit_parser_specifiedc C sh t jdd}| jdtdd}W d Q R X t|d j}| jd|k | jd|k | jd |j d S )
NT)r, zb)parseOnlyTheser r9
parse_onlys ) r- r. r r r/ r0 r( r r )r r2 r r3 r r r )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 j}| jd|k | jd|k | jd|j d S ) NT)r, s éutf8) fromEncodingr r= Z
from_encoding)r- r. r r/ r0 r( r r )r r2 r< r r3 r r r *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) assertRaises TypeErrorr )r r r r "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 }|j}zBtjdd}| j|}W d Q R X t|d j}| jd|k W d |j X tjdd}| j|}W d Q R X | j
dt | d S )NT)r, r zlooks like a filename) tempfileZNamedTemporaryFilenamer- r. r r/ r0 r( closer len)r Z
filehandlefilenamer2 r r3 r r r test_disk_file_warningh s
z#TestWarnings.test_disk_file_warningc C s> t jdd}| jd}W d Q R X | jtdd |D d S )NT)r, s http://www.crummybytes.com/c s s | ]}d t |jkV qdS )zlooks like a URLN)r/ r0 ).0r2 r r r } s z?TestWarnings.test_url_warning_with_bytes_url..)r- r. r r( any)r warning_listr r r r test_url_warning_with_bytes_urlx s z,TestWarnings.test_url_warning_with_bytes_urlc C s> t jdd}| jd}W d Q R X | jtdd |D d S )NT)r, z http://www.crummyunicode.com/c s s | ]}d t |jkV 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 !test_url_warning_with_unicode_url s z.TestWarnings.test_url_warning_with_unicode_urlc C s> t jdd}| jd}W d Q R X | jtdd |D d S )NT)r, s$ http://www.crummybytes.com/ is greatc s s | ]}d t |jkV 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 assertFalserJ )r rK r r r r %test_url_warning_with_bytes_and_space s z2TestWarnings.test_url_warning_with_bytes_and_spacec C s> t jdd}| jd}W d Q R X | jtdd |D d S )NT)r, z%http://www.crummyuncode.com/ is greatc s s | ]}d t |jkV 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 'test_url_warning_with_unicode_and_space 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 )TestSelectiveParsingc C s. d}t d}| j||d}| j|j d d S )Nz&NoYesNoYes Yesr8 )r: s YesYes Yes)r r r r )r markupZstrainerr r r r 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 ) TestEntitySubstitutionz1Standalone tests of the EntitySubstitution class.c C s
t | _d S )N)r sub)r r r r setUp s z TestEntitySubstitution.setUpc C s d}| j | jj|d d S )Nu foo\u2200☃\u00f5baru foo∀☃õbar)r rU substitute_html)r r) r r r test_simple_html_substitution s z4TestEntitySubstitution.test_simple_html_substitutionc C s& d}t |}| j| jj|jd d S )Ns fooz ‘’foo“”)r
r rU rW rR )r Zquotesdammitr r r test_smart_quote_substitution s z4TestEntitySubstitution.test_smart_quote_substitutionc C s d}| j | jj|d| d S )NzWelcome to "my bar"F)r rU substitute_xml)r r) r r r Itest_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_false s z`TestEntitySubstitution.test_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falsec C s0 | j | jjddd | j | jjddd d S )NZWelcomeTz "Welcome"z Bob's Barz "Bob's Bar")r rU r[ )r r r r 6test_xml_attribute_quoting_normally_uses_double_quotes s zMTestEntitySubstitution.test_xml_attribute_quoting_normally_uses_double_quotesc C s d}| j | jj|dd d S )NzWelcome to "my bar"Tz'Welcome to "my bar"')r rU r[ )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 s d}| j | jj|dd d S )NzWelcome to "Bob's Bar"Tz""Welcome to "Bob's Bar"")r rU r[ )r r) r r r btest_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotes s zyTestEntitySubstitution.test_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotesc C s d}| j | jj|| d S )NzWelcome to "Bob's Bar")r rU r[ )r Zquotedr r r zfoo<bar>)r rU r[ )r r r r 'test_xml_quoting_handles_angle_brackets s
z>TestEntitySubstitution.test_xml_quoting_handles_angle_bracketsc C s | j | jjdd d S )NzAT&TzAT&T)r rU r[ )r r r r #test_xml_quoting_handles_ampersands s z:TestEntitySubstitution.test_xml_quoting_handles_ampersandsc C s | j | jjdd d S )Nz ÁT&TzÁT&T)r rU r[ )r r r r Etest_xml_quoting_including_ampersands_when_they_are_part_of_an_entity s
z\TestEntitySubstitution.test_xml_quoting_including_ampersands_when_they_are_part_of_an_entityc C s | j | jjdd d S )Nz ÁT&TzÁT&T)r rU Z"substitute_xml_containing_entities)r r r r Dtest_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entity s
z[TestEntitySubstitution.test_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityc C s d}| j | jj|| dS )z:There's no need to do this except inside attribute values.z Bob's "bar"N)r rU rW )r textr r r test_quotes_not_html_substituted s z7TestEntitySubstitution.test_quotes_not_html_substitutedN)r" r# r$ __doc__rV rX rZ r\ r] r^ r_ r` ra rb rc rd rf r r r r rT s rT c sN e Zd Z fddZdd Zdd Zdd Zd d
Zee d d d
Z
Z S )TestEncodingConversionc s4 t t| j d| _| jjd| _| j| jd d S )NuU Sacré bleu!zutf-8sU Sacré bleu!)superrh rV unicode_datar r r )r ) __class__r r rV s z TestEncodingConversion.setUpc C s t jj}tjtj zbdd }|t j_d}| j|}|j }| jt |t
| j || j |j | j |j
j d W d tjtj |t j_X d S )Nc S s d S )Nr )r/ r r r noop s z>TestEncodingConversion.test_ascii_in_unicode_out..noops azutf-8)bs4rY chardet_dammitloggingdisableWARNINGr decoder(
isinstancer/ r Z document_forr lowerNOTSET)r chardetrl asciiZsoup_from_asciiZunicode_outputr r r test_ascii_in_unicode_out s
z0TestEncodingConversion.test_ascii_in_unicode_outc C s@ | j | j}| j|j | j | j|jjd | j|jd d S )Nu Sacré bleu!)r rj r rr foor r )r soup_from_unicoder r r test_unicode_in_unicode_out s z2TestEncodingConversion.test_unicode_in_unicode_outc C s2 | j | j}| j|j | j | j|jjd d S )Nu Sacré bleu!)r r r rr rj ry r )r Zsoup_from_utf8r r r test_utf8_in_unicode_out s z/TestEncodingConversion.test_utf8_in_unicode_outc C s$ | j | j}| j|jd| j d S )Nzutf-8)r rj r r r )r rz r r r
test_utf8_out s z$TestEncodingConversion.test_utf8_outzQBad HTMLParser detected; skipping test of non-ASCII characters in attribute name.c C s( d}| j | j|jjd|jd d S )Nu r< )r r Zdivr )r rR r r r 1test_attribute_name_containing_unicode_characters s zHTestEncodingConversion.test_attribute_name_containing_unicode_characters) r" r# r$ rV rx r{ r| r} r
PYTHON_3_PRE_3_2r~
__classcell__r r )rk r rh 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$S )%TestUnicodeDammitz"Standalone tests of UnicodeDammit.c C s d}t |}| j|j| d S )Nu I'm already Unicode! ☃)r
r unicode_markup)r rR rY r r r test_unicode_input" s z$TestUnicodeDammit.test_unicode_inputc C s d}t |}| j|jd d S )Ns z#\u2018\u2019\u201c\u201d)r
r r )r rR rY r r r test_smart_quotes_to_unicode' s z.TestUnicodeDammit.test_smart_quotes_to_unicodec C s" d}t |dd}| j|jd d S )Ns Zxml)smart_quotes_toz+‘’“”)r
r r )r rR rY r r r !test_smart_quotes_to_xml_entities- s z3TestUnicodeDammit.test_smart_quotes_to_xml_entitiesc C s" d}t |dd}| j|jd d S )Ns r5 )r z'‘’“”)r
r r )r rR rY r r r "test_smart_quotes_to_html_entities3 s z4TestUnicodeDammit.test_smart_quotes_to_html_entitiesc C s" d}t |dd}| j|jd d S )Ns rw )r z''"")r
r r )r rR rY r r r test_smart_quotes_to_ascii9 s z,TestUnicodeDammit.test_smart_quotes_to_asciic C s0 d}t |}| j|jj d | j|jd d S )Ns Sacré bleu! ☃zutf-8u Sacré bleu! ☃)r
r r rt r )r r< rY r r r test_detect_utf8? s z"TestUnicodeDammit.test_detect_utf8c C s4 d}t |dg}| j|jj d | j|jd d S )Ns z
iso-8859-8z\u05dd\u05d5\u05dc\u05e9)r
r r rt r )r hebrewrY r r r test_convert_hebrewF s z%TestUnicodeDammit.test_convert_hebrewc C s6 d}t |}| j|jj d | j|jjd| d S )Ns ケータイ Watchzutf-8)r
r r rt r r )r utf_8rY r r r /test_dont_see_smart_quotes_where_there_are_noneL s zATestUnicodeDammit.test_dont_see_smart_quotes_where_there_are_nonec C s, dj d}t|dg}| j|jj d d S )Nu
Räksmörgåszutf-8z
iso-8859-8)r r
r r rt )r r rY r r r test_ignore_inappropriate_codecsR s
z2TestUnicodeDammit.test_ignore_inappropriate_codecsc C s: dj d}x*dD ]"}t||g}| j|jj d qW d S )Nu
Räksmörgåszutf-8.utf8...
utF---16.!)r r r )r r
r r rt )r r Z bad_encodingrY r r r test_ignore_invalid_codecsW s
z,TestUnicodeDammit.test_ignore_invalid_codecsc C sL dj d}t|dgd}| j|jj d t|ddgd}| j|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 d}t|j}d|ks td S )Ns' u utf-�)r list encodingsAssertionError)r Zdetectedr r r r Ptest_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}| jd|j qW d S ) N& & $ # T)Zis_htmlzeuc-jp)r r r r )r
r r )r r rY r r r test_detect_html5_style_meta_tagq s z2TestUnicodeDammit.test_detect_html5_style_meta_tagc C s d}t jj}tjtj zPdd }|t j_t|}| jd|j | j d|j
k t |d}| j |j W d tjtj |t j_X d S )NsT
بتر
ѐc S s d S )Nr )r/ r r r rl s zBTestUnicodeDammit.test_last_ditch_entity_replacement..noopTz\ufffdz html.parser)
rm rY rn ro rp rq r
r Z contains_replacement_charactersr( r r ru )r docrv rl rY r r r r "test_last_ditch_entity_replacement| s
z4TestUnicodeDammit.test_last_ditch_entity_replacementc C s, d}t |}| jd|j | jd|j d S )Ns < a > < / a > u áézutf-16le)r
r r r )r r rY r r r test_byte_order_mark_removed s z.TestUnicodeDammit.test_byte_order_mark_removedc C sP dj d}dj d}|| | }| jt|jd tj|}| jd|jd d S )Nu ☃r r< u “Hi, I like Windows!” windows_1252u+ ☃☃☃“Hi, I like Windows!”☃☃☃u ☃☃☃)r r? UnicodeDecodeErrorrr r
detwingler )r r< r r Zfixedr r r test_detwingle s
z TestUnicodeDammit.test_detwinglec C sB x s6 ,/ E< /