U
lu
f[< @ sl d dl Z d dlZd dlZd dlmZ d dlmZmZmZ G dd deZ G dd deZ
G dd d eZ dS )
N)
BeautifulSoup)EntitySubstitutionEncodingDetector
UnicodeDammitc @ sf e Zd ZdZdd Zejdddddgd d
Zd d Z d
d Z
dd Z dd Z dd Z
dd ZdS )TestUnicodeDammitz"Standalone tests of UnicodeDammit.c C s d}t |}|j|kstd S )Nu I'm already Unicode! ☃)r unicode_markupAssertionError)selfmarkupdammit r 9/tmp/pip-unpacked-wheel-pg4vfqbd/bs4/tests/test_dammit.pytest_unicode_input s z$TestUnicodeDammit.test_unicode_inputz smart_quotes_to,expect_converted)Nu ‘’“”)xmlz ‘’“”)htmlz ‘’“”)asciiz''""c C s, d}t |dg|dj}|d|ks(tdS )zbVerify the functionality of the smart_quotes_to argument
to the UnicodeDammit constructor.s windows-1252)known_definite_encodingssmart_quotes_toz
{}N)r r formatr )r r Zexpect_convertedr
Z convertedr r r
test_smart_quotes_to s z&TestUnicodeDammit.test_smart_quotes_toc C s0 d}t |}|j dks t|jdks,td S )Ns Sacré bleu! ☃utf-8u Sacré bleu! ☃r original_encodinglowerr r )r utf8r r r r
test_detect_utf8& s z"TestUnicodeDammit.test_detect_utf8c C s4 d}t |dg}|j dks"t|jdks0td S )N
iso-8859-8u םולשr r hebrewr r r r
test_convert_hebrew, s z%TestUnicodeDammit.test_convert_hebrewc C s6 d}t |}|j dks t|jd|ks2td S )Ns ケータイ Watchr )r r r r r encode)r utf_8r r r r
/test_dont_see_smart_quotes_where_there_are_none2 s zATestUnicodeDammit.test_dont_see_smart_quotes_where_there_are_nonec C s, d d}t|dg}|j dks(td S )N
Räksmörgåsr r r" r r r r r utf8_datar r r r
test_ignore_inappropriate_codecs8 s
z2TestUnicodeDammit.test_ignore_inappropriate_codecsc C s6 d d}dD ]"}t||g}|j dkstqd S )Nr% r )z.utf8z...z
utF---16.!r& )r r( Z bad_encodingr r r r
test_ignore_invalid_codecs= s
z,TestUnicodeDammit.test_ignore_invalid_codecsc C sL d d}t|dgd}|j dks*tt|ddgd}|jd ksHtd S )Nr% r )Zexclude_encodingsr r& r' r r r
test_exclude_encodingsC s
z(TestUnicodeDammit.test_exclude_encodingsN)__name__
__module__ __qualname____doc__r pytestmark parametrizer r r! r$ r) r* r+ r r r r
r s"
r c @ sT e Zd Zdd Zdd Zdd Zdd Zd d
Zd d Zd
d Z dd Z
dd Z dS )TestEncodingDetectorc C s" t d}t|j}d|ks td S )Ns' u utf-�)r list encodingsr )r Zdetectedr5 r r r
Ptest_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterS s
zeTestEncodingDetector.test_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterc C s( dD ] }t |dd}d|jkstqd S )N)s& s& s$ s# Tis_htmlzeuc-jp)r r r r datar r r r
test_detect_html5_style_meta_tagY s z5TestEncodingDetector.test_detect_html5_style_meta_tagc C s d}t jj}ttj zLdd }|t j_t|}d|jks@t d|j
ksNt t |d}|jsbt W 5 ttj |t j_X d S )NsT
بتر
ѐc S s d S Nr )strr r r
noopy s zETestEncodingDetector.test_last_ditch_entity_replacement..noopTu �z html.parser) bs4r Zchardet_dammitloggingdisableWARNINGNOTSETr Z contains_replacement_charactersr r r )r docchardetr> r Zsoupr r r
"test_last_ditch_entity_replacementc s
z7TestEncodingDetector.test_last_ditch_entity_replacementc C s, d}t |}d|jkstd|jks(td S )N < a > < / a > u áéutf-16le)r r r r r9 r r r
test_byte_order_mark_removed s z1TestEncodingDetector.test_byte_order_mark_removedc C s d}t |}t |dgd}d|jks(tt |dgd}d|jksDtdgdd |jD ks^td }t |dgd
gd }d
|jkstdd
gd d |jD kstd S )
NrG zutf-16)r r )user_encodingsrH c S s g | ] }|d qS r r .0xr r r
s zRTestEncodingDetector.test_known_definite_versus_user_encodings..r r )r rJ c S s g | ] }|d qS rK r rL r r r
rO s r r r tried_encodings)r r: r beforeafterr r r r
)test_known_definite_versus_user_encodings s z>TestEncodingDetector.test_known_definite_versus_user_encodingsc C sJ d}t |dgdgdgd}d|jks(tdddgdd |jD ksFtd S )Nr z shift-jisr r )r Zoverride_encodingsrJ c S s g | ] }|d qS rK r rL r r r
rO s zKTestEncodingDetector.test_deprecated_override_encodings..rP r r r r
"test_deprecated_override_encodings s z7TestEncodingDetector.test_deprecated_override_encodingsc C s` d d}d d}|| | }tt |d W 5 Q R X t|}d|dks\td S )Nu ☃☃☃r u “Hi, I like Windows!” windows_1252u+ ☃☃☃“Hi, I like Windows!”☃☃☃)r" r0 ZraisesUnicodeDecodeErrordecoder detwingler )r r rV rD Zfixedr r r
test_detwingle s
z#TestEncodingDetector.test_detwinglec C s< dD ]2}| d}|ds tt|}||kstqd S )N)u œu ₓu ðr )r" endswithr r rY )r Ztricky_unicode_charinputoutputr r r
+test_detwingle_ignores_multibyte_characters s
z@TestEncodingDetector.test_detwingle_ignores_multibyte_charactersc C s d}| d}d}| d}tj}||ddd ks6td||ddksJtd||ddks^td||ksntd||ks~td d
}||| d kst||| d kst||| ddd dkst||dd dkst|d | dd dkst|d
| dd d kstd S )Nz0r z,Fr7 r Tz
iso-8859-1 i )r8 search_entire_document)ra a)r" r Zfind_declared_encodingr )r Z html_unicodeZ
html_bytesZ xml_unicodeZ xml_bytesmZspacerr r r
test_find_declared_encoding s(
z0TestEncodingDetector.test_find_declared_encodingN) r, r- r. r6 r; rF rI rT rU rZ r_ rd r r r r
r3 Q s
# r3 c @ s e Zd ZdZdd Zejdddgdd Zd d
Z d d Z
d
d Z dd Z dd Z
dd Zdd Zdd Zdd Zdd Zd d Zd d Zd!S )"TestEntitySubstitutionz1Standalone tests of the EntitySubstitution class.c C s
t | _d S r< )r subr r r r
setup_method s z#TestEntitySubstitution.setup_methodzoriginal,substituted)u foo∀☃õbaru foo∀☃õbar)u ‘’foo“”z ‘’foo“”c C s | j ||kstd S r< rf Zsubstitute_htmlr )r originalZ substitutedr r r
test_substitute_html s z+TestEntitySubstitution.test_substitute_htmlc C s: dD ]0\}}d}|| }|| }| j ||kstqd S )N)
)z⊧u ⊧)z𝔑u 𝔑)z≧̸u ≧̸)z¬ ¬)z⫬u ⫬z||)fjrm )z>>)z<<)z&&z3 %s 4ri )r entityutemplaterawZ
with_entitiesr r r
test_html5_entity% s
z(TestEntitySubstitution.test_html5_entityc C s< d}d}| j ||ks td}d}| j ||ks8td S )Nu fjords ⊔ penguinszfjords ⊔ penguinsu fjords ⊔︀ penguinszfjords ⊔︀ penguinsri )r r: r
r r r
)test_html5_entity_with_variation_selectorC s z@TestEntitySubstitution.test_html5_entity_with_variation_selectorc C s d}| j |d|kstd S )NWelcome to "my bar"Frf Zsubstitute_xmlr r sr r r
Itest_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falseO s z`TestEntitySubstitution.test_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falsec C s0 | j dddkst| j dddks,td S )NZWelcomeTz "Welcome"z Bob's Barz "Bob's Bar"rx rg r r r
6test_xml_attribute_quoting_normally_uses_double_quotesS s zMTestEntitySubstitution.test_xml_attribute_quoting_normally_uses_double_quotesc C s d}| j |ddkstd S )Nrw Tz'Welcome to "my bar"'rx ry r r r
Otest_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotesW s zfTestEntitySubstitution.test_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotesc C s d}| j |ddkstd S )NWelcome to "Bob's Bar"Tz""Welcome to "Bob's Bar""rx ry 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 ||kstd S )Nr~ rx )r quotedr r r
zfoo<bar>rx rg r r r
'test_xml_quoting_handles_angle_bracketsc s z>TestEntitySubstitution.test_xml_quoting_handles_angle_bracketsc C s | j ddkstd S )NzAT&TzAT&Trx rg r r r
#test_xml_quoting_handles_ampersandsf s z:TestEntitySubstitution.test_xml_quoting_handles_ampersandsc C s | j ddkstd S )N ÁT&TzÁT&Trx rg r r r
Etest_xml_quoting_including_ampersands_when_they_are_part_of_an_entityi s z\TestEntitySubstitution.test_xml_quoting_including_ampersands_when_they_are_part_of_an_entityc C s | j ddkstd S )Nr zÁT&T)rf Z"substitute_xml_containing_entitiesr rg r r r
Dtest_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityl s z[TestEntitySubstitution.test_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityc C s d}| j ||kstdS )z:There's no need to do this except inside attribute values.z Bob's "bar"Nri )r textr r r
test_quotes_not_html_substitutedo s z7TestEntitySubstitution.test_quotes_not_html_substitutedN)r, r- r. r/ rh r0 r1 r2 rk ru rv r{ r| r} r r r r r r r r r r r
re s*
re ) r0 r@ r? r Z
bs4.dammitr r r objectr r3 re r r r r
s E ?