o
Rh[< @ 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 @ sb e Zd ZdZdd Zejdg ddd Zdd Z d
d Z
d d
Z dd Z dd Z
dd ZdS )TestUnicodeDammitz"Standalone tests of UnicodeDammit.c C s d}t |}|j|ks
J d S )Nu I'm already Unicode! ☃)r unicode_markup)selfmarkupdammit r W/home/air/sanwanet/backup_V2/venv/lib/python3.10/site-packages/bs4/tests/test_dammit.pytest_unicode_input s z$TestUnicodeDammit.test_unicode_inputz smart_quotes_to,expect_converted))Nu ‘’“”)xmlz ‘’“”)htmlz ‘’“”)asciiz''""c C s, d}t |dg|dj}|d|ksJ dS )zbVerify the functionality of the smart_quotes_to argument
to the UnicodeDammit constructor.s windows-1252)known_definite_encodingssmart_quotes_toz
{}N)r r format)r r expect_convertedr convertedr r r test_smart_quotes_to s z&TestUnicodeDammit.test_smart_quotes_toc C s0 d}t |}|j dksJ |jdksJ d S )Ns Sacré bleu! ☃utf-8u Sacré bleu! ☃r original_encodinglowerr )r utf8r
r r r test_detect_utf8& s z"TestUnicodeDammit.test_detect_utf8c C s4 d}t |dg}|j dksJ |jdksJ d S )N
iso-8859-8u םולשr r hebrewr
r r r test_convert_hebrew, s z%TestUnicodeDammit.test_convert_hebrewc C s6 d}t |}|j dksJ |jd|ksJ d S )Ns ケータイ Watchr )r r r r encode)r utf_8r
r r r /test_dont_see_smart_quotes_where_there_are_none2 s zATestUnicodeDammit.test_dont_see_smart_quotes_where_there_are_nonec C s, d d}t|dg}|j dksJ d S )N
Räksmörgåsr r r# r r r r utf8_datar
r r r test_ignore_inappropriate_codecs8 s
z2TestUnicodeDammit.test_ignore_inappropriate_codecsc C s6 d d}dD ]}t||g}|j dksJ qd S )Nr& r )z.utf8z...z
utF---16.!r' )r r) bad_encodingr
r r r test_ignore_invalid_codecs= s
z,TestUnicodeDammit.test_ignore_invalid_codecsc C sL d d}t|dgd}|j dksJ t|ddgd}|jd ks$J d S )Nr& r )exclude_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|v sJ d S )Ns' u utf-�)r list encodings)r detectedr8 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|jksJ qd S )N)s& s& s$ s# Tis_htmlzeuc-jp)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 z1dd }|t j_t|}d|jks J d|jv s'J t |d}|js1J W ttj
|t j_d S ttj
|t j_w )NsT
بتر
ѐc S s d S Nr )strr r r noopy s zETestEncodingDetector.test_last_ditch_entity_replacement..noopTu �z html.parser) bs4r
chardet_dammitloggingdisableWARNINGr contains_replacement_charactersr r NOTSET)r docchardetrB r
soupr r r "test_last_ditch_entity_replacementc s
z7TestEncodingDetector.test_last_ditch_entity_replacementc C s, d}t |}d|jks
J d|jksJ d S )N < a > < / a > u áéutf-16le)r r r r= r r r test_byte_order_mark_removed s z1TestEncodingDetector.test_byte_order_mark_removedc C s d}t |}t |dgd}d|jksJ t |dgd}d|jks"J dgdd |jD ks/J d }t |dgd
gd }d
|jksAJ dd
gd d |jD ksOJ d S )
NrN zutf-16)r r )user_encodingsrO c S g | ]}|d qS r r .0xr r r
zRTestEncodingDetector.test_known_definite_versus_user_encodings..r r )r rQ c S rR rS r rT r r r rW rX 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 sH d}t |dgdgdgd}d|jksJ g ddd |jD ks"J d S ) Nr shift-jisr r )r override_encodingsrQ )r^ r r c S rR rS r rT r r r rW rX zKTestEncodingDetector.test_deprecated_override_encodings..rY r r r r "test_deprecated_override_encodings s z7TestEncodingDetector.test_deprecated_override_encodingsc C st d d}d d}|| | }tt
|d W d n1 s%w Y t|}d|dks8J d S )Nu ☃☃☃r u “Hi, I like Windows!” windows_1252u+ ☃☃☃“Hi, I like Windows!”☃☃☃)r# r3 raisesUnicodeDecodeErrordecoder detwingle)r r ra rJ fixedr r r test_detwingle s
z#TestEncodingDetector.test_detwinglec C s< dD ]}| d}|dsJ t|}||ksJ qd S )N)u œu ₓu ðr )r# endswithr re )r tricky_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 u sJ d||ddks%J d||ddks/J d||ks7J d||ks?J d d
}||| d u sMJ ||| d u sWJ ||| ddd dksdJ ||dd dksnJ |d | dd dkszJ |d
| dd d u sJ d S )Nz0r z,Fr; r Tz
iso-8859-1 i )r< search_entire_document)ro a)r# r find_declared_encoding)r html_unicode
html_bytes xml_unicode xml_bytesmspacerr r r test_find_declared_encoding s&
z0TestEncodingDetector.test_find_declared_encodingN) r/ r0 r1 r: r? rM rP r] r` rg rm rx r r r r r6 Q s
# r6 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 ||ks
J d S r@ rz substitute_html)r original substitutedr r r test_substitute_html s z+TestEntitySubstitution.test_substitute_htmlc C s: dD ]\}}d}|| }|| }| j ||ksJ qd S )N)
)z⊧u ⊧)z𝔑u 𝔑)z≧̸u ≧̸)z¬ ¬)z⫬u ⫬z||)fjr )z>>)z<<)z&&z3 %s 4r} )r entityutemplateraw
with_entitiesr r r test_html5_entity% s z(TestEntitySubstitution.test_html5_entityc C s< d}d}| j ||ksJ d}d}| j ||ks J d S )Nu fjords ⊔ penguinszfjords ⊔ penguinsu fjords ⊔︀ penguinszfjords ⊔︀ penguinsr} )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|ks
J d S )NWelcome to "my bar"Frz substitute_xmlr sr r r Itest_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falseO z`TestEntitySubstitution.test_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falsec C s0 | j dddks J | j dddksJ d S )NWelcomeTz "Welcome"z Bob's Barz "Bob's Bar"r r{ r r r 6test_xml_attribute_quoting_normally_uses_double_quotesS s zMTestEntitySubstitution.test_xml_attribute_quoting_normally_uses_double_quotesc C d}| j |ddks
J d S )Nr Tz'Welcome to "my bar"'r r r r r Otest_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotesW r 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_quotes[ r zyTestEntitySubstitution.test_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotesc C s d}| j ||ks J d S )Nr r )r quotedr r r zfoo<bar>r r{ r r r 'test_xml_quoting_handles_angle_bracketsc z>TestEntitySubstitution.test_xml_quoting_handles_angle_bracketsc C r )NzAT&TzAT&Tr r{ r r r #test_xml_quoting_handles_ampersandsf r 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_entityi r z\TestEntitySubstitution.test_xml_quoting_including_ampersands_when_they_are_part_of_an_entityc C r )Nr zÁT&T)rz "substitute_xml_containing_entitiesr{ r r r Dtest_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityl r z[TestEntitySubstitution.test_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityc C s d}| j ||ks 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_substitutedo s z7TestEntitySubstitution.test_quotes_not_html_substitutedN)r/ r0 r1 r2 r| r3 r4 r5 r r r r r r r r r r r r r r r r r ry s,
ry ) r3 rE rC r
bs4.dammitr r r objectr r6 ry r r r r s E ?