3 dsF@sdZddlmZddlZddlZddlZddlZddlZddlZddl m Z m Z m Z m Z mZddlmZmZmZddlmZmZmZmZddlmZmZmZddlZGd d d eZGd d d eZGd ddeZGdddeZ GdddeZ!GdddeZ"GdddeZ#dS)z#Tests of Beautiful Soup as a whole.) set_traceN) BeautifulSoupBeautifulStoneSoupGuessedAtParserWarningMarkupResemblesLocatorWarningdammit)builder_registry TreeBuilderParserRejectedMarkup)Comment SoupStrainerTagNavigableString)default_builder LXML_PRESENTSoupTestc@sLeZdZddZddZddZddZd d Zd d Zd dZ ddZ dS)TestConstructorcCs"d}|j|}d|jjkstdS)Nu

éé

uéé)souph1stringAssertionError)selfdatarr=/tmp/pip-build-8z3xcdsh/beautifulsoup4/bs4/tests/test_soup.pytest_short_unicode_input(s z(TestConstructor.test_short_unicode_inputcCs"d}|j|}d|jjkstdS)Nz

foobar

zfoobar)rrrr)rrrrrrtest_embedded_null-s z"TestConstructor.test_embedded_nullcCs,djd}|j|dgd}d|jks(tdS)Nu Räksmörgåszutf-8)Zexclude_encodingsz windows-1252)encoderoriginal_encodingr)r utf8_datarrrrtest_exclude_encodings2s z&TestConstructor.test_exclude_encodingscCsGdddt}tddd}tjddtdd|i|}WdQRXt|j|sVttdd |jjksltd |jj ks|t|f|}tjdd}td|dd }WdQRXt |d j }|j d st||jkst||jkstdS)Nc@sDeZdZddZddZddZddZd d ZeZZ d d Z d S)z7TestConstructor.test_custom_builder_class..Mockc[s(||_d|_d|_g|_g|_i|_dS)NTF) called_withZis_xmlZstore_line_numbersZcdata_list_attributesZpreserve_whitespace_tagsstring_containers)rkwargsrrr__init__;s z@TestConstructor.test_custom_builder_class..Mock.__init__cSsdS)Nr)rrrrrinitialize_soupBszGTestConstructor.test_custom_builder_class..Mock.initialize_soupcSs ||_dS)N)fed)rmarkuprrrfeedDsz.Mock.feedcSsdS)Nr)rrrrresetFsz=TestConstructor.test_custom_builder_class..Mock.resetcSsdS)Nr)rignorerrrr+Hsz>TestConstructor.test_custom_builder_class..Mock.ignorecs dVdS)Nprepared markuporiginal encodingdeclared encodingcontains replacement characters)r,r-r.r/r)rargsr$rrrprepare_markupKszFTestConstructor.test_custom_builder_class..Mock.prepare_markupN) __name__ __module__ __qualname__r%r&r)r*r+Zset_up_substitutionsZcan_be_empty_elementr1rrrrMock:sr5valueT)varZconvertEntities)recordbuilder)r7zprepared markup)r:Z ignored_valuerzCKeyword arguments to the BeautifulSoup constructor will be ignored.)r9) objectdictwarningscatch_warningsr isinstancer:rr"r'strmessage startswith)rr5r$rr:wmsgrrrtest_custom_builder_class7s" z)TestConstructor.test_custom_builder_classc CsXGdddt}dd}ddl}tjt}td|dWdQRXdt|jksTtdS) Nc@seZdZddZdS)z:TestConstructor.test_parser_markup_rejection..Mockc_s tddS)NzNope.)r )rr0r$rrrr)ksz?TestConstructor.test_parser_markup_rejection..Mock.feedN)r2r3r4r)rrrrr5jsr5cs tdddfVtdddfVdS)NF)r()rr0r$rrrr1nszDTestConstructor.test_parser_markup_rejection..prepare_markuprr9)r:zoThe markup you provided was rejected by the parser. Trying a different parser or a different encoding may help.) r repytestraisesr rr@r6r)rr5r1rFexc_inforrrtest_parser_markup_rejectiongs  z,TestConstructor.test_parser_markup_rejectionc Csd}|j|}|j}d|dks$tddg|dks8t|j|tdd}d|jdksZtxjddiddifD]V}tjd d }|j|d|d}WdQRX|j}d dg|dkstd|dksltqlWdS) Nz&z an id idaclass)r:Zmulti_valued_attributesz a class *T)r8Zan)rrLrrr=r>)rr(rrLZ switcheroorCrrrtest_cdata_list_attributeszs z*TestConstructor.test_cdata_list_attributescspGdddtGdddtGdddt|jdtttid}tfd d |jDsltdS) Nc@s eZdZdS)z9TestConstructor.test_replacement_classes..TagPlusN)r2r3r4rrrrTagPlussrPc@s eZdZdS)z.StringPlusN)r2r3r4rrrr StringPlussrQc@s eZdZdS)z=TestConstructor.test_replacement_classes..CommentPlusN)r2r3r4rrrr CommentPlussrRzfoobar)Zelement_classesc3s|]}t|fVqdS)N)r?).0x)rRrQrPrr sz;TestConstructor.test_replacement_classes..)r rr rallZrecursiveChildGeneratorr)rrr)rRrQrPrtest_replacement_classess z(TestConstructor.test_replacement_classescCsGdddt}Gdddt}|jd||dd}t|jjdtsJtt|jjd|s`tx|jjD]}t||sjtqjWg|j kstdS) Nc@s eZdZdS)zATestConstructor.test_alternate_string_containers..PStringN)r2r3r4rrrrPStringsrXc@s eZdZdS)zATestConstructor.test_alternate_string_containers..BStringN)r2r3r4rrrrBStringsrYz4
Hello.

Here is some bolded text)bp)r#r) rrr?divcontentsrr[rZstringsZstring_container_stack)rrXrYrsrrr test_alternate_string_containerss z0TestConstructor.test_alternate_string_containersN) r2r3r4rrr!rErJrOrWr`rrrrr&s0rc @seZdZddZddZddZddZd d Zd d Zd dZ ddZ e j j dddddddddgddZe j j dddddgd d!Zd"d#Zd$d%Zd&d'Zd(d)Zd*S)+ TestWarningscCs@x*|D]"}t|j|r|jtks$t|SqWtd||fdS)Nz%s warning not found in %r)r?rAfilename__file__r Exception)rr=clsrCrrr_assert_warnings   zTestWarnings._assert_warningcCs2|j|t}t|j}|jtjdds.tdS)N<)rfrr@rArBrZNO_PARSER_SPECIFIED_WARNINGr)rrCwarningrArrr_assert_no_parser_specifieds  z(TestWarnings._assert_no_parser_specifiedc Cs.tjdd}td}WdQRX|j|dS)NT)r8z)r=r>rri)rrCrrrr#test_warning_if_no_parser_specifiedsz0TestWarnings.test_warning_if_no_parser_specifiedc Cs0tjdd}tdd}WdQRX|j|dS)NT)r8zhtml)r=r>rri)rrCrrrr*test_warning_if_parser_specified_too_vaguesz7TestWarnings.test_warning_if_parser_specified_too_vaguec Cs2tjdd}|jd}WdQRXg|ks.tdS)NT)r8z)r=r>rr)rrCrrrr,test_no_warning_if_explicit_parser_specifiedsz9TestWarnings.test_no_warning_if_explicit_parser_specifiedc Csltjdd}tddtdd}WdQRX|j|t}t|j}d|ksLtd|ksXtd |j kshtdS) NT)r8zz html.parserrZ)parseOnlyThesern parse_onlys) r=r>rr rfDeprecationWarningr@rArr)rrCrrhrDrrr)test_parseOnlyThese_renamed_to_parse_onlys    z6TestWarnings.test_parseOnlyThese_renamed_to_parse_onlyc Csjtjdd}d}t|ddd}WdQRX|j|t}t|j}d|ksLtd|ksXtd|jksftdS) NT)r8séz html.parserutf8) fromEncodingrsZ from_encoding) r=r>rrfrpr@rArr)rrCrrrrhrDrrr*test_fromEncoding_renamed_to_from_encodings    z7TestWarnings.test_fromEncoding_renamed_to_from_encodingc Cs(tjt|jdddWdQRXdS)NzT)Zno_such_argument)rGrH TypeErrorr)rrrr"test_unrecognized_keyword_arguments z/TestWarnings.test_unrecognized_keyword_argument extensionz markup.htmlz markup.htmz markup.HTMLz markup.txtz markup.xhtmlz markup.xmlz/home/user/filez c:\user ilec CsHtjdd2}td|d}|j|t}dt|jks:tWdQRXdS)NT)r8r(z html.parserzlooks more like a filename)r=r>rrfrr@rAr)rrwrCrrhrrrtest_resembles_filename_warnings z,TestWarnings.test_resembles_filename_warningZ markuphtmlz markup.comr9z markup.jsc Cs6tjdd}|jd|}WdQRXg|ks2tdS)NT)r8r()r=r>rr)rrwrCrrrr"test_resembles_filename_no_warning!sz/TestWarnings.test_resembles_filename_no_warningc Cs`d}tjdd}t|d}WdQRX|j|t}dt|jksDt|t|jjdks\tdS)Nshttp://www.crummybytes.com/T)r8z html.parserzlooks more like a URLrr) r=r>rrfrr@rArr)rurl warning_listrrhrrrtest_url_warning_with_bytes_url-sz,TestWarnings.test_url_warning_with_bytes_urlc CsZd}tjdd}t|d}WdQRX|j|t}dt|jksDt|t|jksVtdS)Nzhttp://www.crummyunicode.com/T)r8z html.parserzlooks more like a URL)r=r>rrfrr@rAr)rrzr{rrhrrr!test_url_warning_with_unicode_url7sz.TestWarnings.test_url_warning_with_unicode_urlc Cs>tjdd}|jd}WdQRXtdd|D s:tdS)NT)r8s$http://www.crummybytes.com/ is greatcss|]}dt|jkVqdS)zlooks more like a URLN)r@rA)rSrCrrrrUHszETestWarnings.test_url_warning_with_bytes_and_space..)r=r>ranyr)rr{rrrr%test_url_warning_with_bytes_and_spaceCsz2TestWarnings.test_url_warning_with_bytes_and_spacec Cs>tjdd}|jd}WdQRXtdd|D s:tdS)NT)r8z&http://www.crummyunicode.com/ is greatcss|]}dt|jkVqdS)zlooks more like a URLN)r@rA)rSrCrrrrUNszGTestWarnings.test_url_warning_with_unicode_and_space..)r=r>rr~r)rr{rrrr'test_url_warning_with_unicode_and_spaceKsz4TestWarnings.test_url_warning_with_unicode_and_spaceN)r2r3r4rfrirjrlrmrqrtrvrGmarkZ parametrizerxryr|r}rrrrrrras&      rac@seZdZddZdS)TestSelectiveParsingcCs.d}td}|j||d}|jdks*tdS)Nz&NoYesNoYes YesrZ)rosYesYes Yes)r rrr)rr(Zstrainerrrrrtest_parse_with_soupstrainerTsz1TestSelectiveParsing.test_parse_with_soupstrainerN)r2r3r4rrrrrrRsrc@s:eZdZdZddZejje ddddZ dd Z d S) TestNewTagz(Test the BeautifulSoup.new_tag() method.cCsb|jd}|jddddid}t|ts,td|jks:ttddd|jksPtd|jks^tdS)Nr9fooZbaznameza name)barattrs)rr) rnew_tagr?r rrr<rparent)rrrrrr test_new_tag]s  zTestNewTag.test_new_tagz-lxml not installed, cannot parse XML document)reasoncCsBtdd}|jd}|jd}d|jks.td|jks>tdS)Nr9xmlbrr[s
s

)rrrr)rZxml_soupZxml_brZxml_prrr5test_xml_tag_inherits_self_closing_rules_from_builderes    z@TestNewTag.test_xml_tag_inherits_self_closing_rules_from_buildercCsBtdd}|jd}|jd}d|jks.td|jks>tdS)Nr9z html.parserrr[s
s

)rrrr)rZ html_soupZhtml_brZhtml_prrr1test_tag_inherits_self_closing_rules_from_builderss    zsome markupz some markup)rpickledumpsloadsrLrr)rrpickled unpickledrrrtest_normal_pickles   zTestPickle.test_normal_picklecCs6|jd}d|_tj|}tj|}d|jks2tdS)Nz some markup)rr:rrrrr)rrrrrrrtest_pickle_with_no_builders    z&TestPickle.test_pickle_with_no_builderN)r2r3r4rrrrrrrsrc@s4eZdZddZddZddZddZd d Zd S) TestEncodingConversioncCs&d|_|jjd|_|jdks"tdS)NuUSacré bleu!zutf-8sUSacré bleu!) unicode_datarr r)rrrr setup_methodsz#TestEncodingConversion.setup_methodc Cstj}tjtjz^dd}|t_d}|j|}|j}t|tsFt ||j |jks\t |j j dksnt Wdtjtj |t_XdS)NcSsdS)Nr)r@rrrnoopsz>TestEncodingConversion.test_ascii_in_unicode_out..noops azutf-8)rZchardet_dammitloggingdisableWARNINGrdecoder?r@rZ document_forrlowerNOTSET)rchardetrasciiZsoup_from_asciiZunicode_outputrrrtest_ascii_in_unicode_outs   z0TestEncodingConversion.test_ascii_in_unicode_outcCs@|j|j}|j|jkst|jjdks.t|jdkss( 0} "