n9d[< d dl Z d dlZd dlZd dlmZ d dlmZmZmZ G d de Z G d de Z
G d d e Z dS )
N)
BeautifulSoup)EntitySubstitutionEncodingDetector
UnicodeDammitc e Zd ZdZd Zej dg d d Zd Z d Z
d Z d Z d
Z
d Zd S )
TestUnicodeDammitz"Standalone tests of UnicodeDammit.c B d}t | }|j |k sJ d S )Nu I'm already Unicode! ☃)r unicode_markup)selfmarkupdammits 5lib/python3.11/site-packages/bs4/tests/test_dammit.pytest_unicode_inputz$TestUnicodeDammit.test_unicode_input s. 3 v&&$...... z smart_quotes_to,expect_converted))Nu ‘’“”)xmlz ‘’“”)htmlz ‘’“”)asciiz''""c p d}t |dg| j }|d | k sJ 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 converteds r test_smart_quotes_toz&TestUnicodeDammit.test_smart_quotes_to sX 0! n-= +
O 2 23C D DDDDDDDr c d}t | }|j dk sJ |j dk sJ d S )Ns Sacré bleu! ☃utf-8u Sacré bleu! ☃r original_encodinglowerr
)r utf8r
s r test_detect_utf8z"TestUnicodeDammit.test_detect_utf8& sO 1 t$$'--//7::::$(DDDDDDDr c d}t |dg }|j dk sJ |j dk sJ d S )N
iso-8859-8u םולשr r hebrewr
s r test_convert_hebrewz%TestUnicodeDammit.test_convert_hebrew, sS $ v ~66'--//<????$(BBBBBBBr c d}t | }|j dk sJ |j d |k sJ d S )Ns ケータイ Watchr )r r r r
encode)r utf_8r
s r /test_dont_see_smart_quotes_where_there_are_nonezATestUnicodeDammit.test_dont_see_smart_quotes_where_there_are_none2 s\ I u%%'--//7::::$++G44======r c d d }t |dg }|j dk sJ d S )N
Räksmörgåsr r% r* r r r r utf8_datar
s r test_ignore_inappropriate_codecsz2TestUnicodeDammit.test_ignore_inappropriate_codecs8 sL #**733 y<.99'--//7::::::r c d d }dD ]2}t ||g }|j dk sJ 3d S )Nr. r )z.utf8z...z
utF---16.!r/ )r r1 bad_encodingr
s r test_ignore_invalid_codecsz,TestUnicodeDammit.test_ignore_invalid_codecs= sb #**733 : ? ?L"9|n==F+1133w>>>>> ? ?r c d d }t |dg }|j dk sJ t |ddg }|j d k sJ d S )Nr. r )exclude_encodingsr r/ r0 s r test_exclude_encodingsz(TestUnicodeDammit.test_exclude_encodingsC s #**733 yWIFFF'--//>AAAA '>)BD D D'4//////r N)__name__
__module__ __qualname____doc__r pytestmark parametrizer r" r( r, r2 r5 r8 r r r r s ,,/ / /
[ *
E E EE E E C C C > > > ; ; ;
? ? ? 0 0 0 0 0r r c D e Zd Zd Zd Zd Zd Zd Zd Zd Z d Z
d Z d
S ) TestEncodingDetectorc X t d }t |j }d|v sJ d S )Ns' u utf-�)r list encodings)r detectedrE s r Ptest_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterzeTestEncodingDetector.test_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterS s; # 9; ;+,, .);;;;;;r c L dD ] }t |d }d|j k sJ !d S )N)s& s& s$ s# Tis_htmlzeuc-jp)r r r datar
s r test_detect_html5_style_meta_tagz5TestEncodingDetector.test_detect_html5_style_meta_tagY sH 4 8 8D
#4666Fv 777777
8 8r c d}t j j }t j t j d }|t j _ t
| }d|j k sJ d|j v sJ t |d }|j sJ t j t j
|t j _ d S # t j t j
|t j _ w xY w)NsT
بتر
ѐc d S Nr@ )strs r noopzETestEncodingDetector.test_last_ditch_entity_replacement..noopy s tr Tu �z html.parser) bs4r
chardet_dammitloggingdisableWARNINGr contains_replacement_charactersr
r NOTSET)r docchardetrR r
soups r "test_last_ditch_entity_replacementz7TestEncodingDetector.test_last_ditch_entity_replacementc s 2 *+((( 0
(,CJ %"3''F6AAAAAv 44444 m44D7 7 7 7 7 OGN + + +(/CJ % % %
OGN + + +(/CJ % / / / /s AB9 91C*c \ d}t | }d|j k sJ d|j k sJ d S )N < a > < / a > u áéutf-16le)r r
r rK s r test_byte_order_mark_removedz1TestEncodingDetector.test_byte_order_mark_removed sB M t$$ 55555V 5555555r c L d}t | }t |dg }d|j k sJ t |dg }d|j k sJ dgd |j D k sJ d}t |dgd g
}d |j k sJ dd gd |j D k sJ d S ) Nr_ zutf-16)r r )user_encodingsr` c g | ]
}|d S r r@ .0xs r
zRTestEncodingDetector.test_known_definite_versus_user_encodings.. s E E E! E E Er r$ r% )r rc c g | ]
}|d S re r@ rf s r ri zRTestEncodingDetector.test_known_definite_versus_user_encodings.. s *P*P*PA1Q4*P*P*Pr r r tried_encodings)r rL r
beforeafterr' s r )test_known_definite_versus_user_encodingsz>TestEncodingDetector.test_known_definite_versus_user_encodings s
N t$$ txjIII633333
dG9===U 44444| E Ef.D E E EEEEE % v /;n> > > v 77777 &*P*P9O*P*P*PPPPPPPr c d}t |dgdgdg }d|j k sJ g dd |j D k sJ d S )Nr$ shift-jisr r% )r override_encodingsrc )rq r r% c g | ]
}|d S re r@ rf s r ri zKTestEncodingDetector.test_deprecated_override_encodings.. s 2 2 2aQqT 2 2 2r rk r& s r "test_deprecated_override_encodingsz7TestEncodingDetector.test_deprecated_override_encodings s % &1] 'y (>
v 77777 433 2 261 2 2 2
r c T d d }d d }||z |z }t j t 5 | d d d d n # 1 swxY w Y t j | }d| d k sJ d S )Nu ☃☃☃r! u “Hi, I like Windows!” windows_1252u+ ☃☃☃“Hi, I like Windows!”☃☃☃)r* r= raisesUnicodeDecodeErrordecoder detwingle)r r! rv rZ fixeds r test_detwinglez#TestEncodingDetector.test_detwingle s !))&11
./5vn/E/E
\!D( ]-
.
. JJv ',,< V@T@TTTTTTTs A..A25A2c dD ]J}| d }| d sJ t j | }||k sJ Kd S )N)u œu ₓu ðr! )r* endswithr rz )r tricky_unicode_charinputoutputs r +test_detwingle_ignores_multibyte_charactersz@TestEncodingDetector.test_detwingle_ignores_multibyte_characters sj
$ # #
(..v66E>>'** * * *",U33FU????? # #r c d}| d }d}| d }t j } ||d J d ||d k sJ d ||d k sJ d || k sJ d || k sJ d d
z } |||z J |||z J |||z dd dk sJ ||d dk sJ |d |z d dk sJ |d
|z d J d S )Nz0r z,FrI r Tz
iso-8859-1 i )rJ search_entire_document)r a)r* r find_declared_encoding)r html_unicode
html_bytes xml_unicode xml_bytesmspacers r test_find_declared_encodingz0TestEncodingDetector.test_find_declared_encoding s J !((11
C &&w// 3q u---555!!L$7777777!!J5555555qq~~----qq||++++ q*$%%---q)#$$,,,
Afz!4 M M M
q4888LHHHHq !$???<OOOOq !$???GGGGGr N) r9 r: r; rG rM r] ra ro rt r| r r r@ r r rB rB Q s < < < 8 8 8!0 !0 !0F6 6 6 Q Q Q>
$U U U.
#
#
# (H (H (H (H (Hr rB c e Zd ZdZd Zej dddg d Zd Z d Z
d Z d
Z d Z
d Zd
Zd Zd Zd Zd Zd ZdS )TestEntitySubstitutionz1Standalone tests of the EntitySubstitution class.c t | _ d S rP )r subr s r setup_methodz#TestEntitySubstitution.setup_method s
%r zoriginal,substituted)u foo∀☃õbaru foo∀☃õbar)u ‘’foo“”z ‘’foo“”c F | j | |k sJ d S rP r substitute_html)r original substituteds r test_substitute_htmlz+TestEntitySubstitution.test_substitute_html s+ x''11[@@@@@@r c n dD ]1\ }}d}||z }||z }| j | |k sJ 2d S )N)
)z⊧u ⊧)z𝔑u 𝔑)z≧̸u ≧̸)z¬ ¬)z⫬u ⫬z||)fjr )z>>)z<<)z&&z3 %s 4r )r entityutemplateraw
with_entitiess r test_html5_entityz(TestEntitySubstitution.test_html5_entity% sc
B BIFA0 HQ,C $v -M8++C00MAAAAA7 B Br c d}d}| j | |k sJ d}d}| j | |k sJ d S )Nu fjords ⊔ penguinszfjords ⊔ penguinsu fjords ⊔︀ penguinszfjords ⊔︀ penguinsr )r rL r s r )test_html5_entity_with_variation_selectorz@TestEntitySubstitution.test_html5_entity_with_variation_selectorC s` (*x''--7777-+x''--777777r c L d}| j |d |k sJ d S )NWelcome to "my bar"Fr substitute_xmlr ss r Itest_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falsez`TestEntitySubstitution.test_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_falseO s0 !x&&q%00A555555r c | j dd dk sJ | j dd dk sJ d S )NWelcomeTz "Welcome"z Bob's Barz "Bob's Bar"r r s r 6test_xml_attribute_quoting_normally_uses_double_quoteszMTestEntitySubstitution.test_xml_attribute_quoting_normally_uses_double_quotesS sN x&&y$77;FFFFx&&{D99^KKKKKKr c L d}| j |d dk sJ d S )Nr Tz'Welcome to "my bar"'r r s r Otest_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quoteszfTestEntitySubstitution.test_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotesW s1 !x&&q$//3LLLLLLLr c L d}| j |d dk sJ d S )NWelcome to "Bob's Bar"Tz""Welcome to "Bob's Bar""r r s r btest_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quoteszyTestEntitySubstitution.test_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotes[ s1 %x&&q$//3XXXXXXXr c J d}| j | |k sJ d S )Nr r )r quoteds r zfoo<bar>r r s r 'test_xml_quoting_handles_angle_bracketsz>TestEntitySubstitution.test_xml_quoting_handles_angle_bracketsc s* x&&z226FFFFFFFr c F | j d dk sJ d S )NzAT&TzAT&Tr r s r #test_xml_quoting_handles_ampersandsz:TestEntitySubstitution.test_xml_quoting_handles_ampersandsf s) x&&v..*<<<<< r? r r r r r r r r r r r r r r@ r r r r s8 ;;& & & [
2
C
A A A B B B<
8
8
86 6 6L L LM M MY Y Y9 9 9G G G= = =O O O_ _ _6 6 6 6 6r r ) r= rU rS r
bs4.dammitr r r objectr rB r r@ r r r s
C0 C0 C0 C0 C0 C0 C0 C0J{H {H {H {H {H6 {H {H {H|c6 c6 c6 c6 c6V c6 c6 c6 c6 c6r