U
}gE @ st d dl Z d dlZd dlZd dlZd dlmZ d dlmZmZmZ G dd de Z
G dd de Z G dd d e Z 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-kgiupv3k/bs4/tests/test_dammit.pytest_unicode_input s z$TestUnicodeDammit.test_unicode_inputz smart_quotes_to,expect_converted)Nu ‘’“”)xmlz ‘’“”)htmlz ‘’“”)asciiz''""c C s, d}t |dg|dj}|d|ks(tdS )zbVerify the functionality of the smart_quotes_to argument
to the UnicodeDammit constructor.s windows-1252)known_definite_encodingssmart_quotes_toz
{}N)r r formatr )r r Zexpect_convertedr
Z convertedr r r
test_smart_quotes_to s z&TestUnicodeDammit.test_smart_quotes_toc C s0 d}t |}|j dks t|jdks,td S )Ns Sacré bleu! ☃utf-8u Sacré bleu! ☃r original_encodinglowerr r )r utf8r r r r
test_detect_utf8* s z"TestUnicodeDammit.test_detect_utf8c C s4 d}t |dg}|j dks"t|jdks0td S )N
iso-8859-8u םולשr )r hebrewr r r r
test_convert_hebrew0 s z%TestUnicodeDammit.test_convert_hebrewc C s6 d}t |}|j dks t|jd|ks2td S )Ns ケータイ Watchr )r r r r r encode)r utf_8r r r r
/test_dont_see_smart_quotes_where_there_are_none6 s zATestUnicodeDammit.test_dont_see_smart_quotes_where_there_are_nonec C s, d d}t|dg}|j dks(td S )N
Räksmörgåsr r r! r r r r r utf8_datar r r r
test_ignore_inappropriate_codecs< s
z2TestUnicodeDammit.test_ignore_inappropriate_codecsc C s6 d d}dD ]"}t||g}|j dkstqd S )Nr$ r )z.utf8z...z
utF---16.!r% )r r' Z bad_encodingr r r r
test_ignore_invalid_codecsA 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_encodingsG 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 Zdetectedr4 r r r
Ptest_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterV 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_tag] 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
noop~ s zETestEncodingDetector.test_last_ditch_entity_replacement..noopTu �z html.parser) bs4r Z_chardet_dammitloggingdisableWARNINGNOTSETr Z contains_replacement_charactersr r r )r docchardetr= r Zsoupr r r
"test_last_ditch_entity_replacementg 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 r8 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 )
NrF zutf-16)r r )user_encodingsrG c S s g | ] }|d qS r r .0xr r r
s zRTestEncodingDetector.test_known_definite_versus_user_encodings..r r )r rI c S s g | ] }|d qS rJ r rK r r r
rN s )r r r tried_encodings)r r9 r beforeafterr r r r
)test_known_definite_versus_user_encodings s z>TestEncodingDetector.test_known_definite_versus_user_encodingsc C s d}t jdd }t|dgdgdgd}W 5 Q R X |\}|j}t|tsLt|jtksZtd|j kshtdddgdd |j
D kstd S )
Nr T)recordz shift-jisr r )r Zoverride_encodingsrI c S s g | ] }|d qS rJ r rK r r r
rN s zKTestEncodingDetector.test_deprecated_override_encodings..) warningscatch_warningsr message
isinstanceDeprecationWarningr filename__file__r rO )r r wr warningrV 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! r/ ZraisesUnicodeDecodeErrordecoder detwingler )r r r^ rC 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 ra )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,Fr6 r Tz
iso-8859-1 i )r7 search_entire_document)ri 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- r5 r: rE rH rR r] rb rg rl r r r r
r2 U s
% r2 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 Zejd!d"d#d$d%d&d'd(d)gd*d+ Zejd,d-d.d/d0gd1d2 Zejd3d4gd5d6 Zd7S )8TestEntitySubstitutionz1Standalone 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; rn substitute_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||)fjrv )z>>)z<zfoo<bar>r ro r r r
'test_xml_quoting_handles_angle_bracketsm s z>TestEntitySubstitution.test_xml_quoting_handles_angle_bracketsc C s | j ddkstd S )NzAT&TzAT&Tr ro r r r
#test_xml_quoting_handles_ampersandsp s z:TestEntitySubstitution.test_xml_quoting_handles_ampersandsc C s | j ddkstd S )N ÁT&TzÁT&Tr ro r r r
Etest_xml_quoting_including_ampersands_when_they_are_part_of_an_entitys s z\TestEntitySubstitution.test_xml_quoting_including_ampersands_when_they_are_part_of_an_entityc C s | j ddkstd S )Nr zÁT&T)rn Z"substitute_xml_containing_entitiesr ro r r r
Dtest_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entityv 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"Nrq )r textr r r
test_quotes_not_html_substituted| s z7TestEntitySubstitution.test_quotes_not_html_substitutedz markup, old)z foo & barz
foo & bar)zfoo&zfoo&)z
foo&&& barzfoo&&& bar)zx=1&y=2z x=1&y=2)z&123z&123)z&abcz&abc)z
foo &0 barzfoo &0 bar)zfoo &lolwat barzfoo &lolwat barc C s, | j ||kst| j ||ks(td S r; rn rr r substitute_html5_raw)r r
oldr r r
'test_unambiguous_ampersands_not_escaped s z>TestEntitySubstitution.test_unambiguous_ampersands_not_escapedzmarkup,html,html5,html5raw)÷ ÷r r )z
&nonesuch;&nonesuch;r r )z÷
÷r r )z¡
¡r r c C s@ | j ||kst| j ||ks(t| j ||ksTestEntitySubstitution.test_when_entity_ampersands_are_escapedz
markup,expect)z&nosuchentity;z&nosuchentity;c C s, | j ||kst| j ||ks(td S r; r )r r
expectr r r
!test_ambiguous_ampersands_escaped s z8TestEntitySubstitution.test_ambiguous_ampersands_escapedN)r+ r, r- r. rp r/ r0 r1 rt r} r~ r r r r r r r r r r r r r r r r r
rm s`
rm )
r/ r? rT r> r Z
bs4.dammitr r r objectr r2 rm r r r r
s G G