n9dMdZddlmZddlZddlZddlZddlZddlZddlZddl m Z m Z m Z m Z mZddlmZmZmZddlmZmZmZmZmZddlmZmZmZddlZGd d eZGd d eZGd deZ GddeZ!GddeZ"GddeZ#GddeZ$GddeZ%dS)z#Tests of Beautiful Soup as a whole.) set_traceN) BeautifulSoupBeautifulStoneSoupGuessedAtParserWarningMarkupResemblesLocatorWarningdammit)builder_registry TreeBuilderParserRejectedMarkup)Comment SoupStrainerPYTHON_SPECIFIC_ENCODINGSTagNavigableString)default_builder LXML_PRESENTSoupTestc>eZdZdZdZdZdZdZdZdZ dZ d S) TestConstructorcXd}||}d|jjksJdS)Nu

éé

uéésouph1stringselfdatars 3lib/python3.11/site-packages/bs4/tests/test_soup.pytest_short_unicode_inputz(TestConstructor.test_short_unicode_input)s1yy''''''cXd}||}d|jjksJdS)Nz

foobar

zfoobarrrs rtest_embedded_nullz"TestConstructor.test_embedded_null.s1"yyTW^++++++r!czdd}||dg}d|jksJdS)Nu Räksmörgåsutf-8)exclude_encodingsz windows-1252)encoderoriginal_encoding)r utf8_datars rtest_exclude_encodingsz&TestConstructor.test_exclude_encodings3sF#**733 yywiy@@!7777777r!czGddt}tdd}tjd5t dd|i|}dddn #1swxYwYt |j|sJtd |jjksJd |jjksJ|di|}tjd5}t d|d }dddn #1swxYwYt|d j }| d sJ||jksJ||jksJdS)Nc:eZdZdZdZdZdZdZexZZ dZ dS)7TestConstructor.test_custom_builder_class..MockcZ||_d|_d|_g|_g|_i|_dS)NTF) called_withis_xmlstore_line_numberscdata_list_attributespreserve_whitespace_tagsstring_containers)rkwargss r__init__z@TestConstructor.test_custom_builder_class..Mock.__init__<s7#) " */'-/*02-)+&&&r!cdSNrrs rinitialize_soupzGTestConstructor.test_custom_builder_class..Mock.initialize_soupCr!c||_dSr8)fed)rmarkups rfeedz.Mock.feedEs !r!cdSr8r9rs rresetz=TestConstructor.test_custom_builder_class..Mock.resetGr<r!cdSr8r9)rignores rrEz>TestConstructor.test_custom_builder_class..Mock.ignoreIr<r!c?KdVdS)N)prepared markupzoriginal encodingzdeclared encodingzcontains replacement charactersr9rargsr5s rprepare_markupzFTestConstructor.test_custom_builder_class..Mock.prepare_markupLsttttttr!N) __name__ __module__ __qualname__r6r;r@rCrEset_up_substitutionscan_be_empty_elementrJr9r!rMockr-;s , , ,    " " "      :@ @ #7 u u u u ur!rPvalueT)varconvertEntitiesrecordbuilder)rRrG)rW ignored_valuerzCKeyword arguments to the BeautifulSoup constructor will be ignored.)rVr9) objectdictwarningscatch_warningsr isinstancerWr/r>strmessage startswith)rrPr5rrWwmsgs rtest_custom_builder_classz)TestConstructor.test_custom_builder_class8s u u u u u6 u u u(!      $D 1 1 1 = = <.Mockc td)NzNope.)r rHs rr@z?TestConstructor.test_parser_markup_rejection..Mock.feedls*7333r!N)rKrLrMr@r9r!rrPrfks# 4 4 4 4 4r!rPc?>KtdddfVtdddfVdS)NF)r?rHs rrJzDTestConstructor.test_parser_markup_rejection..prepare_markupos8$e+ + + +$e+ + + + + +r!rrV)rWzoThe markup you provided was rejected by the parser. Trying a different parser or a different encoding may help.)r repytestraisesr rr^rQ)rrPrJriexc_infos rtest_parser_markup_rejectionz,TestConstructor.test_parser_markup_rejectionhs  4 4 4 4 4; 4 4 4 , , ,  ]/ 0 0 ,H "d + + + + , , , , , , , , , , , , , , ,AEHIQIWEXEXXXXXXXsAAAcd}||}|j}d|dksJddg|dksJ||td}d|jdksJddiddifD]l}tjd 5}||d|}dddn #1swxYwY|j}d dg|dksJd|dksJmdS) Nz&z an id idaclass)rWmulti_valued_attributesz a class *TrTan)rrprr[r\)rr?rrp switcherooras rtest_cdata_list_attributesz*TestConstructor.test_cdata_list_attributes{ss:yy   FAdG####W~7++++ yyRVyWWdfWo---- ;d 4 - -J(555 [yyzyZZ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [A$<1T7****!G*,,,,, - -s B11B5 8B5 c 8GddtGddtGddt|dttti}t fd |DsJdS) NceZdZdS)9TestConstructor.test_replacement_classes..TagPlusNrKrLrMr9r!rTagPlusry Dr!r{ceZdZdS).StringPlusNrzr9r!r StringPlusr~r|r!rceZdZdS)=TestConstructor.test_replacement_classes..CommentPlusNrzr9r!r CommentPlusrr|r!rzfoobar)element_classesc3>K|]}t|fVdSr8)r]).0xrrr{s r z;TestConstructor.test_replacement_classes..sH   q7J < = =      r!)rrr rallrecursiveChildGenerator)rrrrr{s @@@rtest_replacement_classesz(TestConstructor.test_replacement_classess     c                '   yy -W        1133          r!ctGddt}Gddt}|d||d}t|jjdtsJt|jjd|sJ|jjD]}t||sJg|jksJdS) NceZdZdS)ATestConstructor.test_alternate_string_containers..PStringNrzr9r!rPStringrr|r!rceZdZdS)ATestConstructor.test_alternate_string_containers..BStringNrzr9r!rBStringrr|r!rz4
Hello.

Here is some bolded text)bp)r4r) rrr]divcontentsrrstringsstring_container_stack)rrrrss r test_alternate_string_containersz0TestConstructor.test_alternate_string_containerss     o        o   yy B!!  $(+A.@@@@@$&/!,g66666 * *Aa)) ) ) ) )T0000000r!N) rKrLrMr r#r*rcrmrvrrr9r!rrr's((( ,,, 888 .-.-.-`XXX&---:   811111r!rceZdZejdddgdZejddeDdgzdZd Z dS) TestOutputz!eventual_encoding,actual_encoding)r%r%)utf-16rc~|d}d|_d|d||ksJdS)N Tz eventual_encoding)rr0decode)rractual_encodingrs rtest_decode_xml_declarationz&TestOutput.test_decode_xml_declarations`yy'' RRRR;;1B;CCDDDDDDr!rcg|]}|Sr9r9)rrs r zTestOutput.sCCCAaCCCr!Ncltdd}d|_d||ksJdS)Nr html.parserTz! r)rr0r)rrrs rMtest_decode_xml_declaration_with_missing_or_python_internal_eventual_encodingzXTestOutput.test_decode_xml_declaration_with_missing_or_python_internal_eventual_encodingsS]M:: 5;;1B;CCDDDDDDr!c|d}d|dksJd|dksJd|ksJd|ksJdS)Nrs r%)encodingz )rr'encode_contentsdecode_contentsprettifyr:s rtestzTestOutput.testsyy''g!>!>>>>>!5!5w!5!G!GGGGG 4 4 6 66666 DMMOO333333r!) rKrLrMrjmark parametrizerrrrr9r!rrrs [+  .  EE  E [CC)BCCCtfLEEE44444r!rceZdZdZdZdZdZdZdZdZ dZ e j d gd d Ze j d gd d ZdZdZdZdZdS) TestWarningsc|D]-}t|j|r|jtksJ|cS.t |d|)Nz warning not found in )r]r_filename__file__ Exception)rr[clsras r_assert_warningzTestWarnings._assert_warningsa  A!)S)) zX---- XXFGGGr!c||t}t|j}|t jddsJdS)N<)rrr^r_r`rNO_PARSER_SPECIFIED_WARNING)rrawarningr_s r_assert_no_parser_specifiedz(TestWarnings._assert_no_parser_specifiedsS&&q*@AAgo&&!!-"KCRC"PQQQQQQQr!ctjd5}td}dddn #1swxYwY||dSNTrTr[r\rrrrars r#test_warning_if_no_parser_specifiedz0TestWarnings.test_warning_if_no_parser_specifieds  $D 1 1 1 3Q !122D 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 ((+++++s 266ctjd5}tdd}dddn #1swxYwY||dS)NTrTrhtmlrrs r*test_warning_if_parser_specified_too_vaguez7TestWarnings.test_warning_if_parser_specified_too_vagues  $D 1 1 1 ;Q !16::D ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ((+++++s 377ctjd5}|d}dddn #1swxYwYg|ksJdSrr[r\rrs r,test_no_warning_if_explicit_parser_specifiedz9TestWarnings.test_no_warning_if_explicit_parser_specifieds  $D 1 1 1 /Q99-..D / / / / / / / / / / / / / / /Qwwwwww 8<<cHtjd5}tddtd}dddn #1swxYwY||t }t |j}d|vsJd|vsJd |ksJdS) NTrTrrr)parseOnlyTheser parse_onlys) r[r\rr rDeprecationWarningr^r_r')rrarrrbs r)test_parseOnlyThese_renamed_to_parse_onlyz6TestWarnings.test_parseOnlyThese_renamed_to_parse_only"s  $D 1 1 1 Q  -+C00D                &&q*<=='/""3&&&&s""""T[[]]******s AA Actjd5}d}t|dd}dddn #1swxYwY||t}t |j}d|vsJd|vsJd|jksJdS) NTrTsérutf8) fromEncodingr from_encoding)r[r\rrrr^r_r()rrarrrrbs r*test_fromEncoding_renamed_to_from_encodingz7TestWarnings.test_fromEncoding_renamed_to_from_encoding.s  $D 1 1 1 QD m&D                &&q*<=='/""$$$$#%%%%///////s 7;;ctjt5|ddddddS#1swxYwYdS)NzT)no_such_argument)rjrk TypeErrorrrBs r"test_unrecognized_keyword_argumentz/TestWarnings.test_unrecognized_keyword_argument:s ]9 % % 4 4 IIedI 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4s?AA extension)z markup.htmlz markup.htmz markup.HTMLz markup.txtz markup.xhtmlz markup.xmlz/home/user/filez c:\user ilectjd5}td|zd}||t}dt |jvsJ ddddS#1swxYwYdS)NTrTr?rzlooks more like a filenamer[r\rrrr^r_)rrrarrs rtest_resembles_filename_warningz,TestWarnings.test_resembles_filename_warning>s $D 1 1 1 HQ I!5}EED**1.KLLG/3w3G3GGGGGG H H H H H H H H H H H H H H H H H HsAA++A/2A/) markuphtmlz markup.comrVz markup.jsctjd5}|d|z}dddn #1swxYwYg|ksJdS)NTrTr?r)rrrars r"test_resembles_filename_no_warningz/TestWarnings.test_resembles_filename_no_warningKs $D 1 1 1 3Q99X 122D 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3Qwwwwwws ;??c@d}tjd5}t|d}dddn #1swxYwY||t}dt |jvsJ|t |jdvsJdS)Nshttp://www.crummybytes.com/TrTrlooks more like a URLr)r[r\rrrr^r_r'rurl warning_listrrs rtest_url_warning_with_bytes_urlz,TestWarnings.test_url_warning_with_bytes_urlWs,  $D 1 1 1 5\ m44D 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5&& 7  '#go*>*>>>>>#go..55f======== 599cd}tjd5}t|d}dddn #1swxYwY||t}dt |jvsJ|t |jvsJdS)Nzhttp://www.crummyunicode.com/TrTrrrrs r!test_url_warning_with_unicode_urlz.TestWarnings.test_url_warning_with_unicode_urlas-  $D 1 1 1 5\!m44D 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5&& 7  '#go*>*>>>>>#go........rctjd5}|d}dddn #1swxYwYtd|DrJdS)NTrTs$http://www.crummybytes.com/ is greatc3BK|]}dt|jvVdSrNr^r_rras rrzETestWarnings.test_url_warning_with_bytes_and_space..rC../#ai..@......r!r[r\ranyrrrs r%test_url_warning_with_bytes_and_spacez2TestWarnings.test_url_warning_with_bytes_and_spacems $D 1 1 1 F\99DEED F F F F F F F F F F F F F F F.. ,..... . . . . .rctjd5}|d}dddn #1swxYwYtd|DrJdS)NTrTz&http://www.crummyunicode.com/ is greatc3BK|]}dt|jvVdSrrrs rrzGTestWarnings.test_url_warning_with_unicode_and_space..xrr!rrs r'test_url_warning_with_unicode_and_spacez4TestWarnings.test_url_warning_with_unicode_and_spaceus  $D 1 1 1 G\99EFFD G G G G G G G G G G G G G G G.. ,..... . . . . .rN)rKrLrMrrrrrrrrrjrrrrrrrrr9r!rrrsRHHHRRR ,,, ,,,  + + + 0 0 0444 [ K K K HH  H [555 >>> / / /........r!rceZdZdZdS)TestSelectiveParsingcd}td}|||}|dksJdS)Nz&NoYesNoYes Yesr)rsYesYes Yes)r rr')rr?strainerrs rtest_parse_with_soupstrainerz1TestSelectiveParsing.test_parse_with_soupstrainer~sI9$$yyHy55{{}} BBBBBBBr!N)rKrLrMrr9r!rrr|s(CCCCCr!rcjeZdZdZdZeje ddZ dZ dS) TestNewTagz(Test the BeautifulSoup.new_tag() method.c|d}|ddddi}t|tsJd|jksJt dd|jksJd|jksJdS)NrVfoobaznameza name)barattrs)rr)rnew_tagr]rrrZrparent)rrrs r test_new_tagzTestNewTag.test_new_tagsyy}},,u%7I,JJ'3''''' $$$$H--->>>>w~%%%%%%r!z-lxml not installed, cannot parse XML document)reasonctdd}|d}|d}d|ksJd|ksJdS)NrVxmlbrr
s

rrr')rxml_soupxml_brxml_ps r5test_xml_tag_inherits_self_closing_rules_from_builderz@TestNewTag.test_xml_tag_inherits_self_closing_rules_from_builderss !U++!!$''  %%6==??****%,,..((((((r!ctdd}|d}|d}d|ksJd|ksJdS)NrVrr rr s

r )r html_souphtml_brhtml_ps r1test_tag_inherits_self_closing_rules_from_builderz>++++++V]]__,,,,,,r!N) rKrLrM__doc__rrjrskipifrrrr9r!rrrsv22&&& [ >)) )-----r!rceZdZdZdZdZdS) TestNewStringz+Test the BeautifulSoup.new_string() method.c|d}|d}d|ksJt|tsJdSNrVr)r new_stringr]rrrrs r'test_new_string_creates_navigablestringz5TestNewString.test_new_string_creates_navigablestringsKyy}} OOE " "zzzz!_-------r!c|d}|dt}d|ksJt|tsJdSr)rrr r]rs r3test_new_string_can_create_navigablestring_subclasszATestNewString.test_new_string_can_create_navigablestring_subclasssMyy}} OOE7 + +zzzz!W%%%%%%%r!N)rKrLrMrrrr9r!rrrs855... &&&&&r!rceZdZdZdZdS) TestPicklec|d}tj|}tj|}d|jjksJdS)Nzsome markup some markup)rpickledumpsloadsrprrrpickled unpickleds rtest_normal_picklezTestPickle.test_normal_picklesNyy-..,t$$L))  2222222r!c|d}d|_tj|}tj|}d|jksJdS)Nr#)rrWr$r%r&rr's rtest_pickle_with_no_builderz&TestPickle.test_pickle_with_no_buildersTyy'' ,t$$L))  0000000r!N)rKrLrMr*r,r9r!rr!r!s2333 11111r!r!c,eZdZdZdZdZdZdZdS)TestEncodingConversioncld|_|jd|_|jdksJdS)NuUSacré bleu!r%sUSacré bleu!) unicode_datar'r)rBs r setup_methodz#TestEncodingConversion.setup_methods?U*11'::~!r!cJtj}tjtj d}|t_d}||}|}t|tsJ|| |ksJ|j dksJ tjtj |t_dS#tjtj |t_wxYw)NcdSr8r9)r^s rnoopz>TestEncodingConversion.test_ascii_in_unicode_out..noopstr!s ar%) rchardet_dammitloggingdisableWARNINGrrr]r^ document_forr(lowerNOTSET)rchardetr4asciisoup_from_asciiunicode_outputs rtest_ascii_in_unicode_outz0TestEncodingConversion.test_ascii_in_unicode_outs '((( ,   %)F !#E"ii..O,3355Nnc22 2 2 2!T%6%6u||~~%F%FFFFF"4::<<GGGGG OGN + + +$+F ! ! ! OGN + + +$+F ! + + + +s BC66,D"c||j}||jksJ|jjdksJ|jdksJdSNu Sacré bleu!)rr0rrrr(rsoup_from_unicodes rtest_unicode_in_unicode_outz2TestEncodingConversion.test_unicode_in_unicode_outsk!IId&788 ''))T->>>>> $+/????? 2d::::::r!c||j}||jksJ|jjdksJdSrB)rr)rr0rr)rsoup_from_utf8s rtest_utf8_in_unicode_outz/TestEncodingConversion.test_utf8_in_unicode_outsU4>22$$&&$*;;;;;!(,<<<<<< IId&788 ''00DNBBBBBBr!N)rKrLrMr1r@rErHrJr9r!rr.r.sg@@@ ,,,(;;;===CCCCCr!r.)&rpdbrr6osr$rjsystempfilebs4rrrrr bs4.builderr r r bs4.elementr r rrrrVrrrr[rrrrrrr!r.r9r!rrRs))   l1l1l1l1l1hl1l1l1^&4&4&4&4&4&4&4&4Rz.z.z.z.z.8z.z.z.zCCCCC8CCC - - - - - - - -D & & & & &H & & &11111111$0C0C0C0C0CX0C0C0C0C0Cr!