File: //lib/python2.7/site-packages/bs4/tests/test_soup.pyo
�
�Wc @ s� d Z d d l m Z d d l Z d d l Z d d l Z d d l Z d d l m Z m Z d d l
m Z m Z m
Z
m Z d d l Z d d l m Z m Z m Z d d l m Z m Z d d l Z y d d l m Z m Z e Z Wn e k
rZ e Z n Xe j d d
k o#e j d k Z d e f d
� � YZ! d e f d � � YZ" d e f d � � YZ" d e f d � � YZ# d e j$ f d � � YZ% d e f d � � YZ& d e j$ f d � � YZ' d e f d � � YZ( d e j$ f d � � YZ) d S( s# Tests of Beautiful Soup as a whole.i����( t set_traceN( t
BeautifulSoupt BeautifulStoneSoup( t CharsetMetaAttributeValuet ContentMetaAttributeValuet SoupStrainert NamespacedAttribute( t EntitySubstitutiont
UnicodeDammitt EncodingDetector( t SoupTestt skipIf( t LXMLTreeBuildert LXMLTreeBuilderForXMLi i i t TestConstructorc B s# e Z d � Z d � Z d � Z RS( c C s/ d } | j | � } | j d | j j � d S( Nu
<h1>éé</h1>u éé( t soupt assertEqualt h1t string( t selft dataR ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_short_unicode_input* s c C s/ d } | j | � } | j d | j j � d S( Nu <h1>foo bar</h1>u foo bar( R R R R ( R R R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_embedded_null/ s c C s> d j d � } | j | d d g �} | j d | j � d S( Nu
Räksmörgåss utf-8t exclude_encodingss windows-1252( t encodeR R t original_encoding( R t utf8_dataR ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_exclude_encodings4 s ( t __name__t
__module__R R R ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR ( s t TestWarningsc B sJ e Z e d � Z d � Z d � Z d � Z d � Z d � Z d � Z RS( c C s' | j t j d � } | j | � d S( NiP ( t
startswithR t NO_PARSER_SPECIFIED_WARNINGt
assertTrue( R t st is_theret v( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt _no_parser_specified< s c C sN t j d t � � } | j d � } Wd QXt | d j � } | j | � d S( Nt records <a><b></b></a>i ( t warningst catch_warningst TrueR t strt messaget _assert_no_parser_specified( R t wR t msg( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt# test_warning_if_no_parser_specified@ s c C sQ t j d t � � } | j d d � } Wd QXt | d j � } | j | � d S( NR& s <a><b></b></a>t htmli ( R' R( R) R R* R+ R, ( R R- R R. ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt* test_warning_if_parser_specified_too_vagueF s c C sA t j d t � � } | j d d � } Wd QX| j g | � d S( NR& s <a><b></b></a>s html.parser( R' R( R) R R ( R R- R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt, test_no_warning_if_explicit_parser_specifiedL s c C s� t j d t � �" } | j d d t d � �} Wd QXt | d j � } | j d | k � | j d | k � | j d | j � � d S( NR& s <a><b></b></a>t parseOnlyTheset bi t
parse_onlys <b></b>(
R' R( R) R R R* R+ R! R R ( R R- R R. ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt) test_parseOnlyThese_renamed_to_parse_onlyQ s !c C s� t j d t � �" } d } | j | d d �} Wd QXt | d j � } | j d | k � | j d | k � | j d | j � d S( NR& s ét fromEncodingt utf8i t
from_encoding( R' R( R) R R* R+ R! R R ( R R- R8 R R. ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt* test_fromEncoding_renamed_to_from_encodingY s c C s | j t | j d d t �d S( Ns <a>t no_such_argument( t assertRaisest TypeErrorR R) ( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt" test_unrecognized_keyword_argumentb s (
R R R) R% R/ R1 R2 R6 R: R> ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR : s c B s5 e Z d � Z d � Z d � Z d � Z d � Z RS( c C s� t j � } | j } zT t j d t � � } | j | � } Wd QXt | d j � } | j d | k � Wd | j
� Xt j d t � � } | j | � } Wd QX| j d t | � � d S( NR& i s looks like a filename(
t tempfilet NamedTemporaryFilet nameR' R( R) R R* R+ R! t closeR t len( R t
filehandlet filenameR- R R. ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_disk_file_warningh s c C sK t j d t � � } | j d � } Wd QX| j t d � | D� � � d S( NR& s http://www.crummybytes.com/c s s$ | ] } d t | j � k Vq d S( s looks like a URLN( R* R+ ( t .0R- ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pys <genexpr>} s ( R' R( R) R R! t any( R t warning_listR ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_url_warning_with_bytes_urlx s c C sK t j d t � � } | j d � } Wd QX| j t d � | D� � � d S( NR& u http://www.crummyunicode.com/c s s$ | ] } d t | j � k Vq d S( s looks like a URLN( R* R+ ( RG R- ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pys <genexpr>� s ( R' R( R) R R! RH ( R RI R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt! test_url_warning_with_unicode_url� s c C sK t j d t � � } | j d � } Wd QX| j t d � | D� � � d S( NR& s$ http://www.crummybytes.com/ is greatc s s$ | ] } d t | j � k Vq d S( s looks like a URLN( R* R+ ( RG R- ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pys <genexpr>� s ( R' R( R) R t assertFalseRH ( R RI R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt% test_url_warning_with_bytes_and_space� s c C sK t j d t � � } | j d � } Wd QX| j t d � | D� � � d S( NR& u% http://www.crummyuncode.com/ is greatc s s$ | ] } d t | j � k Vq d S( s looks like a URLN( R* R+ ( RG R- ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pys <genexpr>� s ( R' R( R) R RL RH ( R RI R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt' test_url_warning_with_unicode_and_space� s ( R R RF RJ RK RM RN ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR f s
t TestSelectiveParsingc B s e Z d � Z RS( c C sA d } t d � } | j | d | �} | j | j � d � d S( Ns& No<b>Yes</b><a>No<b>Yes <c>Yes</c></b>R4 R5 s <b>Yes</b><b>Yes <c>Yes</c></b>( R R R R ( R t markupt strainerR ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_parse_with_soupstrainer� s ( R R RR ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyRO � s t TestEntitySubstitutionc B s� e Z d Z d � Z d � Z d � Z d � Z d � Z d � Z d � Z d � Z
d � Z d
� Z d � Z
d � Z d
� Z RS( s1 Standalone tests of the EntitySubstitution class.c C s
t | _ d S( N( R t sub( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt setUp� s c C s& d } | j | j j | � d � d S( Nu foo∀☃õbaru foo∀☃õbar( R RT t substitute_html( R R" ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_simple_html_substitution� s c C s5 d } t | � } | j | j j | j � d � d S( Ns ��foo��s ‘’foo“”( R R RT RV RP ( R t quotest dammit( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_smart_quote_substitution� s c C s) d } | j | j j | t � | � d S( Ns Welcome to "my bar"( R RT t substitute_xmlt False( R R" ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pytI test_xml_converstion_includes_no_quotes_if_make_quoted_attribute_is_false� s c C sB | j | j j d t � d � | j | j j d t � d � d S( Nt Welcomes "Welcome"s Bob's Bars "Bob's Bar"( R RT R[ R) ( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt6 test_xml_attribute_quoting_normally_uses_double_quotes� s c C s) d } | j | j j | t � d � d S( Ns Welcome to "my bar"s 'Welcome to "my bar"'( R RT R[ R) ( R R" ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pytO test_xml_attribute_quoting_uses_single_quotes_when_value_contains_double_quotes� s c C s) d } | j | j j | t � d � d S( Ns Welcome to "Bob's Bar"s" "Welcome to "Bob's Bar""( R RT R[ R) ( R R" ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pytb test_xml_attribute_quoting_escapes_single_quotes_when_value_contains_both_single_and_double_quotes� s c C s& d } | j | j j | � | � d S( Ns Welcome to "Bob's Bar"( R RT R[ ( R t quoted( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt< test_xml_quotes_arent_escaped_when_value_is_not_being_quoted� s c C s | j | j j d � d � d S( Ns foo<bar>s foo<bar>( R RT R[ ( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt' test_xml_quoting_handles_angle_brackets� s c C s | j | j j d � d � d S( Ns AT&Ts AT&T( R RT R[ ( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt# test_xml_quoting_handles_ampersands� s c C s | j | j j d � d � d S( Ns ÁT&Ts &Aacute;T&T( R RT R[ ( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pytE test_xml_quoting_including_ampersands_when_they_are_part_of_an_entity� s c C s | j | j j d � d � d S( Ns ÁT&Ts ÁT&T( R RT t" substitute_xml_containing_entities( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pytD test_xml_quoting_ignoring_ampersands_when_they_are_part_of_an_entity� s c C s& d } | j | j j | � | � d S( s: There's no need to do this except inside attribute values.s Bob's "bar"N( R RT RV ( R t text( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_quotes_not_html_substituted� s ( R R t __doc__RU RW RZ R] R_ R` Ra Rc Rd Re Rf Rh Rj ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyRS � s t TestEncodingConversionc B sM e Z d � Z d � Z d � Z d � Z d � Z e e d � d � � Z RS( c C sH t t | � j � d | _ | j j d � | _ | j | j d � d S( NuU <html><head><meta charset="utf-8"/></head><body><foo>Sacré bleu!</foo></body></html>s utf-8sU <html><head><meta charset="utf-8"/></head><body><foo>Sacré bleu!</foo></body></html>( t superRl RU t unicode_dataR R R ( R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyRU � s c C s� t j j } t j t j � z� d � } | t j _ d } | j | � } | j � } | j t | t
� � | j | | j | j � � � | j | j
j � d � Wd t j t j � | t j _ Xd S( Nc S s d S( N( t None( R* ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt noop� s s <foo>a</foo>s utf-8( t bs4RY t chardet_dammitt loggingt disablet WARNINGR t decodeR! t
isinstancet unicodeR t document_forR t lowert NOTSET( R t chardetRp t asciit soup_from_asciit unicode_output( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_ascii_in_unicode_out� s c C sX | j | j � } | j | j � | j � | j | j j d � | j | j d � d S( Nu Sacré bleu!( R Rn R Rv t fooR R Ro ( R t soup_from_unicode( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_unicode_in_unicode_out s c C sE | j | j � } | j | j � | j � | j | j j d � d S( Nu Sacré bleu!( R R R Rv Rn R� R ( R t soup_from_utf8( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_utf8_in_unicode_out s c C s2 | j | j � } | j | j d � | j � d S( Ns utf-8( R Rn R R R ( R R� ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt
test_utf8_out s sQ Bad HTMLParser detected; skipping test of non-ASCII characters in attribute name.c C s8 d } | j | j | � j j d � | j d � � d S( Nu <div><a ☃="snowman"></a></div>R8 ( R R t divR ( R RP ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt1 test_attribute_name_containing_unicode_characters s (
R R RU R� R� R� R� R t PYTHON_3_PRE_3_2R� ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyRl � s t TestUnicodeDammitc B s� e Z d Z d � Z d � Z d � Z d � Z d � Z d � Z d � Z d � Z
d � Z d
� Z d � Z
d � Z d
� Z d � Z d � Z d � Z d � Z RS( s" Standalone tests of UnicodeDammit.c C s) d } t | � } | j | j | � d S( Nu I'm already Unicode! ☃( R R t unicode_markup( R RP RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_unicode_input" s c C s) d } t | � } | j | j d � d S( Ns <foo>����</foo>u <foo>‘’“”</foo>( R R R� ( R RP RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_smart_quotes_to_unicode' s c C s/ d } t | d d �} | j | j d � d S( Ns <foo>����</foo>t smart_quotes_tot xmls+ <foo>‘’“”</foo>( R R R� ( R RP RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt! test_smart_quotes_to_xml_entities- s c C s/ d } t | d d �} | j | j d � d S( Ns <foo>����</foo>R� R0 s' <foo>‘’“”</foo>( R R R� ( R RP RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt" test_smart_quotes_to_html_entities3 s c C s/ d } t | d d �} | j | j d � d S( Ns <foo>����</foo>R� R} s <foo>''""</foo>( R R R� ( R RP RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_smart_quotes_to_ascii9 s c C sB d } t | � } | j | j j � d � | j | j d � d S( Ns Sacré bleu! ☃s utf-8u Sacré bleu! ☃( R R R Rz R� ( R R8 RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_detect_utf8? s c C sH d } t | d g � } | j | j j � d � | j | j d � d S( Ns ���s
iso-8859-8u םולש( R R R Rz R� ( R t hebrewRY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_convert_hebrewF s c C sK d } t | � } | j | j j � d � | j | j j d � | � d S( Ns ケータイ Watchs utf-8( R R R Rz R� R ( R t utf_8RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt/ test_dont_see_smart_quotes_where_there_are_noneL s c C s> d j d � } t | d g � } | j | j j � d � d S( Nu
Räksmörgåss utf-8s
iso-8859-8( R R R R Rz ( R R RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_ignore_inappropriate_codecsR s c C sX d j d � } xB d d d g D]1 } t | | g � } | j | j j � d � q Wd S( Nu
Räksmörgåss utf-8s .utf8s ...s
utF---16.!( R R R R Rz ( R R t bad_encodingRY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_ignore_invalid_codecsW s c C sl d j d � } t | d d g �} | j | j j � d � t | d d d g �} | j | j d � d S( Nu
Räksmörgåss utf-8R s windows-1252( R R R R Rz Ro ( R R RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR ] s c C s t d � } t | j � } d S( Ns' <?xml version="1.0" encoding="UTF-�" ?>( R t listt encodings( R t detectedR� ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pytP test_encoding_detector_replaces_junk_in_encoding_name_with_replacement_characterk s c C s: x3 d D]+ } t | d t �} | j d | j � q Wd S( Ns& <html><meta charset="euc-jp" /></html>s& <html><meta charset='euc-jp' /></html>s$ <html><meta charset=euc-jp /></html>s# <html><meta charset=euc-jp/></html>t is_htmls euc-jp( s& <html><meta charset="euc-jp" /></html>s& <html><meta charset='euc-jp' /></html>s$ <html><meta charset=euc-jp /></html>s# <html><meta charset=euc-jp/></html>( R R) R R ( R R RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_detect_html5_style_meta_tagq s
c C s� d } t j j } t j t j � zm d � } | t j _ t | � } | j t | j � | j
d | j k � t | d � } | j
| j � Wd t j t j
� | t j _ Xd S( NsT <?xml version="1.0" encoding="UTF-8"?>
<html><b>بتر</b>
<i>��ѐ����</i></html>c S s d S( N( Ro ( R* ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyRp � s u �s html.parser( Rq RY Rr Rs Rt Ru R R R) t contains_replacement_charactersR! R� R R{ ( R t docR| Rp RY R ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt" test_last_ditch_entity_replacement| s c C s< d } t | � } | j d | j � | j d | j � d S( Ns ��< a > � � < / a > u <a>áé</a>s utf-16le( R R R� R ( R R RY ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_byte_order_mark_removed� s c C sn d j d � } d j d � } | | | } | j t | j d � t j | � } | j d | j d � � d S( Nu ☃i R8 u “Hi, I like Windows!”t windows_1252u+ ☃☃☃“Hi, I like Windows!”☃☃☃u ☃☃☃( R R< t UnicodeDecodeErrorRv R t detwingleR ( R R8 R� R� t fixed( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_detwingle� s c C sY xR d D]J } | j d � } | j | j d � � t j | � } | j | | � q Wd S( Nu œu ₓu ðR8 s �( u œu ₓu ð( R R! t endswithR R� R ( R t tricky_unicode_chart inputt output( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt+ test_detwingle_ignores_multibyte_characters� s
( R R Rk R� R� R� R� R� R� R� R� R� R� R R� R� R� R� R� R� ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR� s$ # t TestNamedspacedAttributec B s# e Z d � Z d � Z d � Z RS( c C s# t d d � } | j | d � d S( Nt xmlns( R Ro R ( R t a( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt test_name_may_be_none� s c C s# t d d � } | j d | � d S( NR� R4 s a:b( R R ( R R� ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt6 test_attribute_is_equivalent_to_colon_separated_string� s c C s� t d d d � } t d d d � } | j | | � t d d d � } | j | | � t d d d � } | j | | � t d d d � } | j | | � d S( NR� R4 t ct z( R R Ro t assertNotEqual( R R� R4 R� t dt e( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt; test_attributes_are_equivalent_if_prefix_and_name_identical� s ( R R R� R� R� ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR� � s t) TestAttributeValueWithCharsetSubstitutionc B s e Z d � Z d � Z RS( c C sL t d � } | j d | � | j d | j � | j d | j d � � d S( Ns euc-jpR8 ( R R t original_valueR ( R t value( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt! test_content_meta_attribute_value� s c C sL t d � } | j d | � | j d | j � | j d | j d � � d S( Ns text/html; charset=euc-jps text/html; charset=utf8R8 ( R R R� R ( R R� ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR� � s ( R R R� ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyR� � s ( i i (* Rk t pdbR Rs t unittestt sysR? Rq R R t bs4.elementR R R R t
bs4.dammitR R R t bs4.testingR
R R' t bs4.builderR R
R) t LXML_PRESENTt ImportErrorR� R\ t version_infoR� R R RO t TestCaseRS Rl R� R� R� ( ( ( s7 /usr/lib/python2.7/site-packages/bs4/tests/test_soup.pyt <module> s4 "
",/ E<�