Sync with latest HTML5lib
This commit is contained in:
Jacques Distler
parent
c2bfdefa57
commit
0ddd422059
|
|
@ -181,7 +181,7 @@ module HTML5lib
|
|||
end
|
||||
|
||||
elsif token[:type] == :EndTag
|
||||
parseError(_('End tag contains unexpected attributes.')) if token[:data]
|
||||
parseError(_('End tag contains unexpected attributes.')) unless token[:data].empty?
|
||||
token[:name] = token[:name].downcase
|
||||
end
|
||||
|
||||
|
|
|
|||
51
vendor/plugins/HTML5lib/testdata/encoding/chardet/test_big5.txt
vendored
Normal file
51
vendor/plugins/HTML5lib/testdata/encoding/chardet/test_big5.txt
vendored
Normal file
|
|
@ -0,0 +1,51 @@
|
|||
老子《道德經》 第一~四十章
|
||||
|
||||
老子道經
|
||||
|
||||
第一章
|
||||
|
||||
道可道，非常道。名可名，非常名。無，名天地之始﹔有，名萬物之母。
|
||||
故常無，欲以觀其妙；常有，欲以觀其徼。此兩者，同出而異名，同謂之
|
||||
玄。玄之又玄，眾妙之門。
|
||||
|
||||
第二章
|
||||
|
||||
天下皆知美之為美，斯惡矣﹔皆知善之為善，斯不善矣。故有無相生，難
|
||||
易相成，長短相形，高下相傾，音聲相和，前後相隨。是以聖人處「無為
|
||||
」之事，行「不言」之教。萬物作焉而不辭，生而不有，為而不恃，功成
|
||||
而弗居。夫唯弗居，是以不去。
|
||||
|
||||
第三章
|
||||
|
||||
不尚賢，使民不爭﹔不貴難得之貨，使民不為盜﹔不見可欲，使民心不亂
|
||||
。是以「聖人」之治，虛其心，實其腹，弱其志，強其骨。常使民無知無
|
||||
欲。使夫智者不敢為也。為「無為」，則無不治。
|
||||
|
||||
第四章
|
||||
|
||||
「道」沖，而用之或不盈。淵兮，似萬物之宗﹔挫其銳，解其紛，和其光
|
||||
，同其塵﹔湛兮似或存。吾不知誰之子？象帝之先。
|
||||
|
||||
第五章
|
||||
|
||||
天地不仁，以萬物為芻狗﹔聖人不仁，以百姓為芻狗。天地之間，其猶橐
|
||||
蘥乎？虛而不屈，動而愈出。多言數窮，不如守中。
|
||||
|
||||
第六章
|
||||
|
||||
谷神不死，是謂玄牝。玄牝之門，是謂天地根。綿綿若存，用之不勤。
|
||||
|
||||
第七章
|
||||
|
||||
天長地久。天地所以能長且久者，以其不自生，故能長久。是以聖人後其
|
||||
身而身先，外其身而身存。非以其無私邪？故能成其私。
|
||||
|
||||
第八章
|
||||
|
||||
上善若水。水善利萬物而不爭。處眾人之所惡，故幾於道。居善地，心善
|
||||
淵，與善仁，言善信，政善治，事善能，動善時。夫唯不爭，故無尤。
|
||||
|
||||
第九章
|
||||
|
||||
持而盈之，不如其已﹔揣而銳之，不可長保。金玉滿堂，莫之能守﹔富貴
|
||||
而驕，自遺其咎。功遂身退，天之道。
|
||||
10
vendor/plugins/HTML5lib/testdata/encoding/test-yahoo-jp.dat
vendored
Normal file
10
vendor/plugins/HTML5lib/testdata/encoding/test-yahoo-jp.dat
vendored
Normal file
|
|
@ -0,0 +1,10 @@
|
|||
#data
|
||||
<html>
|
||||
<head>
|
||||
<meta http-equiv="Content-Type" content="text/html; charset=euc-jp">
|
||||
<!--京-->
|
||||
<title>Yahoo! JAPAN</title>
|
||||
<meta name="description" content="日本最大級のポータルサイト。検索、オークション、ニュース、メール、コミュニティ、ショッピング、など80以上のサービスを展開。あなたの生活をより豊かにする「ライフ・エンジン」を目指していきます。">
|
||||
<style type="text/css" media="all">
|
||||
#encoding
|
||||
euc-jp
|
||||
392
vendor/plugins/HTML5lib/testdata/encoding/tests1.dat
vendored
Normal file
392
vendor/plugins/HTML5lib/testdata/encoding/tests1.dat
vendored
Normal file
File diff suppressed because one or more lines are too long
82
vendor/plugins/HTML5lib/testdata/encoding/tests2.dat
vendored
Normal file
82
vendor/plugins/HTML5lib/testdata/encoding/tests2.dat
vendored
Normal file
|
|
@ -0,0 +1,82 @@
|
|||
#data
|
||||
<meta
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<!
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta charset = "
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta charset=EUC-jp
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta <meta charset='EUC-jp'>
|
||||
#encoding
|
||||
EUC-jp
|
||||
|
||||
#data
|
||||
<meta charset = 'EUC-jp'>
|
||||
#encoding
|
||||
EUC-jp
|
||||
|
||||
|
||||
#data
|
||||
<!-- -->
|
||||
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
|
||||
#encoding
|
||||
utf-8
|
||||
|
||||
#data
|
||||
<!-- -->
|
||||
<meta http-equiv="Content-Type" content="text/html; charset=utf
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta http-equiv="Content-Type<meta charset="utf-8">
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta http-equiv="Content-Type" content="text/html; charset='utf-8'">
|
||||
#encoding
|
||||
utf-8
|
||||
|
||||
#data
|
||||
<meta http-equiv="Content-Type" content="text/html; charset='utf-8">
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta charset =
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta charset= utf-8
|
||||
#encoding
|
||||
windows-1252
|
||||
|
||||
#data
|
||||
<meta content = "text/html;
|
||||
#encoding
|
||||
windows-1252
|
||||
409
vendor/plugins/HTML5lib/testdata/sanitizer/tests1.dat
vendored
Normal file
409
vendor/plugins/HTML5lib/testdata/sanitizer/tests1.dat
vendored
Normal file
|
|
@ -0,0 +1,409 @@
|
|||
[
|
||||
{
|
||||
"name": "IE_Comments",
|
||||
"input": "<!--[if gte IE 4]><script>alert('XSS');</script><![endif]-->",
|
||||
"output": ""
|
||||
},
|
||||
|
||||
{
|
||||
"name": "IE_Comments_2",
|
||||
"input": "<![if !IE 5]><script>alert('XSS');</script><![endif]>",
|
||||
"output": "<script>alert('XSS');</script>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "allow_colons_in_path_component",
|
||||
"input": "<a href=\"./this:that\">foo</a>",
|
||||
"output": "<a href='./this:that'>foo</a>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "background_attribute",
|
||||
"input": "<div background=\"javascript:alert('XSS')\"></div>",
|
||||
"output": "<div/>",
|
||||
"xhtml": "<div></div>",
|
||||
"rexml": "<div></div>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "bgsound",
|
||||
"input": "<bgsound src=\"javascript:alert('XSS');\" />",
|
||||
"output": "<bgsound src=\"javascript:alert('XSS');\"/>",
|
||||
"rexml": "<bgsound src=\"javascript:alert('XSS');\"></bgsound>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "div_background_image_unicode_encoded",
|
||||
"input": "<div style=\"background-image:\a5\a2\006C\0028'\006a\0061\a6\0061\a3\0063\a2\0069\a0\a4\003a\0061\006c\0065\a2\a4\0028.1027\0058.1053\0053\0027\0029'\0029\">foo</div>",
|
||||
"output": "<div style=''>foo</div>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "div_expression",
|
||||
"input": "<div style=\"width: expression(alert('XSS'));\">foo</div>",
|
||||
"output": "<div style=''>foo</div>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "double_open_angle_brackets",
|
||||
"input": "<img src=http://ha.ckers.org/scriptlet.html <",
|
||||
"output": "<img src='http://ha.ckers.org/scriptlet.html'/><",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "double_open_angle_brackets_2",
|
||||
"input": "<script src=http://ha.ckers.org/scriptlet.html <",
|
||||
"output": "<script src=\"http://ha.ckers.org/scriptlet.html\"><",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "grave_accents",
|
||||
"input": "<img src=`javascript:alert('XSS')` />",
|
||||
"output": "<img/>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "img_dynsrc_lowsrc",
|
||||
"input": "<img dynsrc=\"javascript:alert('XSS')\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "img_vbscript",
|
||||
"input": "<img src='vbscript:msgbox(\"XSS\")' />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "input_image",
|
||||
"input": "<input type=\"image\" src=\"javascript:alert('XSS');\" />",
|
||||
"output": "<input type='image'/>",
|
||||
"rexml": "<input type='image' />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "link_stylesheets",
|
||||
"input": "<link rel=\"stylesheet\" href=\"javascript:alert('XSS');\" />",
|
||||
"output": "<link rel=\"stylesheet\" href=\"javascript:alert('XSS');\"/>",
|
||||
"rexml": "<link href=\"javascript:alert('XSS');\" rel=\"stylesheet\"/>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "link_stylesheets_2",
|
||||
"input": "<link rel=\"stylesheet\" href=\"http://ha.ckers.org/xss.css\" />",
|
||||
"output": "<link rel=\"stylesheet\" href=\"http://ha.ckers.org/xss.css\"/>",
|
||||
"rexml": "<link href=\"http://ha.ckers.org/xss.css\" rel=\"stylesheet\"/>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "list_style_image",
|
||||
"input": "<li style=\"list-style-image: url(javascript:alert('XSS'))\">foo</li>",
|
||||
"output": "<li style=''>foo</li>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "no_closing_script_tags",
|
||||
"input": "<script src=http://ha.ckers.org/xss.js?<b>",
|
||||
"output": "<script src=\"http://ha.ckers.org/xss.js?\"><b/>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "non_alpha_non_digit",
|
||||
"input": "<script/XSS src=\"http://ha.ckers.org/xss.js\"></script>",
|
||||
"output": "<script XSS=\"\" src=\"http://ha.ckers.org/xss.js\"></script>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "non_alpha_non_digit_2",
|
||||
"input": "<a onclick!\#$%&()*~+-_.,:;?@[/|\\]^`=alert(\"XSS\")>foo</a>",
|
||||
"output": "<a>foo</a>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "non_alpha_non_digit_3",
|
||||
"input": "<img/src=\"http://ha.ckers.org/xss.js\"/>",
|
||||
"output": "<img src='http://ha.ckers.org/xss.js'/>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "non_alpha_non_digit_II",
|
||||
"input": "<a href!\#$%&()*~+-_.,:;?@[/|]^`=alert('XSS')>foo</a>",
|
||||
"output": "<a>foo</a>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "non_alpha_non_digit_III",
|
||||
"input": "<a/href=\"javascript:alert('XSS');\">foo</a>",
|
||||
"output": "<a>foo</a>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "platypus",
|
||||
"input": "<a href=\"http://www.ragingplatypus.com/\" style=\"display:block; position:absolute; left:0; top:0; width:100%; height:100%; z-index:1; background-color:black; background-image:url(http://www.ragingplatypus.com/i/cam-full.jpg); background-x:center; background-y:center; background-repeat:repeat;\">never trust your upstream platypus</a>",
|
||||
"output": "<a href='http://www.ragingplatypus.com/' style='display: block; width: 100%; height: 100%; background-color: black; background-x: center; background-y: center;'>never trust your upstream platypus</a>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "protocol_resolution_in_script_tag",
|
||||
"input": "<script src=//ha.ckers.org/.j></script>",
|
||||
"output": "<script src=\"//ha.ckers.org/.j\"></script>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_allow_anchors",
|
||||
"input": "<a href='foo' onclick='bar'><script>baz</script></a>",
|
||||
"output": "<a href='foo'><script>baz</script></a>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_allow_image_alt_attribute",
|
||||
"input": "<img alt='foo' onclick='bar' />",
|
||||
"output": "<img alt='foo'/>",
|
||||
"rexml": "<img alt='foo' />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_allow_image_height_attribute",
|
||||
"input": "<img height='foo' onclick='bar' />",
|
||||
"output": "<img height='foo'/>",
|
||||
"rexml": "<img height='foo' />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_allow_image_src_attribute",
|
||||
"input": "<img src='foo' onclick='bar' />",
|
||||
"output": "<img src='foo'/>",
|
||||
"rexml": "<img src='foo' />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_allow_image_width_attribute",
|
||||
"input": "<img width='foo' onclick='bar' />",
|
||||
"output": "<img width='foo'/>",
|
||||
"rexml": "<img width='foo' />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_handle_blank_text",
|
||||
"input": "",
|
||||
"output": ""
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_handle_malformed_image_tags",
|
||||
"input": "<img \"\"\"><script>alert(\"XSS\")</script>\">",
|
||||
"output": "<img/><script>alert(\"XSS\")</script>\">",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_handle_non_html",
|
||||
"input": "abc",
|
||||
"output": "abc"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_ridiculous_hack",
|
||||
"input": "<img\nsrc\n=\n\"\nj\na\nv\na\ns\nc\nr\ni\np\nt\n:\na\nl\ne\nr\nt\n(\n'\nX\nS\nS\n'\n)\n\"\n />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_0",
|
||||
"input": "<img src=\"javascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_1",
|
||||
"input": "<img src=javascript:alert('XSS') />",
|
||||
"output": "<img/>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_10",
|
||||
"input": "<img src=\"jav
ascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_11",
|
||||
"input": "<img src=\"jav
ascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_12",
|
||||
"input": "<img src=\"  javascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_13",
|
||||
"input": "<img src=\" javascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_14",
|
||||
"input": "<img src=\" javascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_2",
|
||||
"input": "<img src=\"JaVaScRiPt:alert('XSS')\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_3",
|
||||
"input": "<img src='javascript:alert("XSS")' />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_4",
|
||||
"input": "<img src='javascript:alert(String.fromCharCode(88,83,83))' />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_5",
|
||||
"input": "<img src='javascript:alert('XSS')' />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_6",
|
||||
"input": "<img src='javascript:alert('XSS')' />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_7",
|
||||
"input": "<img src='javascript:alert('XSS')' />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_8",
|
||||
"input": "<img src=\"jav\tascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_not_fall_for_xss_image_hack_9",
|
||||
"input": "<img src=\"jav	ascript:alert('XSS');\" />",
|
||||
"output": "<img/>",
|
||||
"rexml": "<img />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_sanitize_half_open_scripts",
|
||||
"input": "<img src=\"javascript:alert('XSS')\"",
|
||||
"output": "<img/>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_sanitize_invalid_script_tag",
|
||||
"input": "<script/XSS SRC=\"http://ha.ckers.org/xss.js\"></script>",
|
||||
"output": "<script XSS=\"\" SRC=\"http://ha.ckers.org/xss.js\"></script>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_sanitize_script_tag_with_multiple_open_brackets",
|
||||
"input": "<<script>alert(\"XSS\");//<</script>",
|
||||
"output": "<<script>alert(\"XSS\");//<</script>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_sanitize_script_tag_with_multiple_open_brackets_2",
|
||||
"input": "<iframe src=http://ha.ckers.org/scriptlet.html\n<",
|
||||
"output": "<iframe src=\"http://ha.ckers.org/scriptlet.html\"><",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_sanitize_tag_broken_up_by_null",
|
||||
"input": "<scr\u0000ipt>alert(\"XSS\")</scr\u0000ipt>",
|
||||
"output": "<scr\ufffdipt>alert(\"XSS\")</scr\ufffdipt>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_sanitize_unclosed_script",
|
||||
"input": "<script src=http://ha.ckers.org/xss.js?<b>",
|
||||
"output": "<script src=\"http://ha.ckers.org/xss.js?\"><b/>",
|
||||
"rexml": "Ill-formed XHTML!"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_strip_href_attribute_in_a_with_bad_protocols",
|
||||
"input": "<a href=\"javascript:XSS\" title=\"1\">boo</a>",
|
||||
"output": "<a title='1'>boo</a>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_strip_href_attribute_in_a_with_bad_protocols_and_whitespace",
|
||||
"input": "<a href=\" javascript:XSS\" title=\"1\">boo</a>",
|
||||
"output": "<a title='1'>boo</a>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_strip_src_attribute_in_img_with_bad_protocols",
|
||||
"input": "<img src=\"javascript:XSS\" title=\"1\">boo</img>",
|
||||
"output": "<img title='1'/>boo",
|
||||
"rexml": "<img title='1' />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "should_strip_src_attribute_in_img_with_bad_protocols_and_whitespace",
|
||||
"input": "<img src=\" javascript:XSS\" title=\"1\">boo</img>",
|
||||
"output": "<img title='1'/>boo",
|
||||
"rexml": "<img title='1' />"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "xml_base",
|
||||
"input": "<div xml:base=\"javascript:alert('XSS');//\">foo</div>",
|
||||
"output": "<div>foo</div>"
|
||||
},
|
||||
|
||||
{
|
||||
"name": "xul",
|
||||
"input": "<p style=\"-moz-binding:url('http://ha.ckers.org/xssmoz.xml#xss')\">fubar</p>",
|
||||
"output": "<p style=''>fubar</p>"
|
||||
}
|
||||
]
|
||||
88
vendor/plugins/HTML5lib/testdata/serializer/core.test
vendored
Normal file
88
vendor/plugins/HTML5lib/testdata/serializer/core.test
vendored
Normal file
|
|
@ -0,0 +1,88 @@
|
|||
{"tests": [
|
||||
|
||||
{"description": "proper attribute value escaping",
|
||||
"input": [["StartTag", "span", {"title": "test \"with\" ""}]],
|
||||
"expected": ["<span title='test \"with\" &quot;'>"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value non-quoting",
|
||||
"input": [["StartTag", "span", {"title": "foo"}]],
|
||||
"expected": ["<span title=foo>"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with >)",
|
||||
"input": [["StartTag", "span", {"title": "foo>bar"}]],
|
||||
"expected": ["<span title=\"foo>bar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with <)",
|
||||
"input": [["StartTag", "span", {"title": "foo<bar"}]],
|
||||
"expected": ["<span title=\"foo<bar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with \")",
|
||||
"input": [["StartTag", "span", {"title": "foo\"bar"}]],
|
||||
"expected": ["<span title='foo\"bar'>"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with ')",
|
||||
"input": [["StartTag", "span", {"title": "foo'bar"}]],
|
||||
"expected": ["<span title=\"foo'bar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with both \" and ')",
|
||||
"input": [["StartTag", "span", {"title": "foo'bar\"baz"}]],
|
||||
"expected": ["<span title=\"foo'bar"baz\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with space)",
|
||||
"input": [["StartTag", "span", {"title": "foo bar"}]],
|
||||
"expected": ["<span title=\"foo bar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with tab)",
|
||||
"input": [["StartTag", "span", {"title": "foo\tbar"}]],
|
||||
"expected": ["<span title=\"foo\tbar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with LF)",
|
||||
"input": [["StartTag", "span", {"title": "foo\nbar"}]],
|
||||
"expected": ["<span title=\"foo\nbar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with CR)",
|
||||
"input": [["StartTag", "span", {"title": "foo\rbar"}]],
|
||||
"expected": ["<span title=\"foo\rbar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with linetab)",
|
||||
"input": [["StartTag", "span", {"title": "foo\u000Bbar"}]],
|
||||
"expected": ["<span title=\"foo\u000Bbar\">"]
|
||||
},
|
||||
|
||||
{"description": "proper attribute value quoting (with form feed)",
|
||||
"input": [["StartTag", "span", {"title": "foo\u000Cbar"}]],
|
||||
"expected": ["<span title=\"foo\u000Cbar\">"]
|
||||
},
|
||||
|
||||
{"description": "void element (as EmptyTag token)",
|
||||
"input": [["EmptyTag", "img", {}]],
|
||||
"expected": ["<img>"]
|
||||
},
|
||||
|
||||
{"description": "void element (as StartTag token)",
|
||||
"input": [["StartTag", "img", {}]],
|
||||
"expected": ["<img>"]
|
||||
},
|
||||
|
||||
{"description": "doctype in error",
|
||||
"input": [["Doctype", "foo"]],
|
||||
"expected": ["<!DOCTYPE foo>"]
|
||||
},
|
||||
|
||||
{"description": "doctype",
|
||||
"input": [["Doctype", "HTML"]],
|
||||
"expected": ["<!DOCTYPE HTML>"]
|
||||
}
|
||||
|
||||
]}
|
||||
900
vendor/plugins/HTML5lib/testdata/serializer/optionaltags.test
vendored
Normal file
900
vendor/plugins/HTML5lib/testdata/serializer/optionaltags.test
vendored
Normal file
|
|
@ -0,0 +1,900 @@
|
|||
{"tests": [
|
||||
|
||||
{"description": "html start-tag followed by text, with attributes",
|
||||
"input": [["StartTag", "html", {"lang": "en"}], ["Characters", "foo"]],
|
||||
"expected": ["<html lang=en>foo"]
|
||||
},
|
||||
|
||||
|
||||
|
||||
{"description": "html start-tag followed by comment",
|
||||
"input": [["StartTag", "html", {}], ["Comment", "foo"]],
|
||||
"expected": ["<html><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "html start-tag followed by space character",
|
||||
"input": [["StartTag", "html", {}], ["Characters", " foo"]],
|
||||
"expected": ["<html> foo"]
|
||||
},
|
||||
|
||||
{"description": "html start-tag followed by text",
|
||||
"input": [["StartTag", "html", {}], ["Characters", "foo"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "html start-tag followed by start-tag",
|
||||
"input": [["StartTag", "html", {}], ["StartTag", "foo", {}]],
|
||||
"expected": ["<foo>"]
|
||||
},
|
||||
|
||||
{"description": "html start-tag followed by end-tag",
|
||||
"input": [["StartTag", "html", {}], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "html start-tag at EOF (shouldn't ever happen?!)",
|
||||
"input": [["StartTag", "html", {}]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
{"description": "html end-tag followed by comment",
|
||||
"input": [["EndTag", "html"], ["Comment", "foo"]],
|
||||
"expected": ["</html><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "html end-tag followed by space character",
|
||||
"input": [["EndTag", "html"], ["Characters", " foo"]],
|
||||
"expected": ["</html> foo"]
|
||||
},
|
||||
|
||||
{"description": "html end-tag followed by text",
|
||||
"input": [["EndTag", "html"], ["Characters", "foo"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "html end-tag followed by start-tag",
|
||||
"input": [["EndTag", "html"], ["StartTag", "foo", {}]],
|
||||
"expected": ["<foo>"]
|
||||
},
|
||||
|
||||
{"description": "html end-tag followed by end-tag",
|
||||
"input": [["EndTag", "html"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "html end-tag at EOF",
|
||||
"input": [["EndTag", "html"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "head start-tag followed by comment",
|
||||
"input": [["StartTag", "head", {}], ["Comment", "foo"]],
|
||||
"expected": ["<head><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "head start-tag followed by space character",
|
||||
"input": [["StartTag", "head", {}], ["Characters", " foo"]],
|
||||
"expected": ["<head> foo"]
|
||||
},
|
||||
|
||||
{"description": "head start-tag followed by text",
|
||||
"input": [["StartTag", "head", {}], ["Characters", "foo"]],
|
||||
"expected": ["<head>foo"]
|
||||
},
|
||||
|
||||
{"description": "head start-tag followed by start-tag",
|
||||
"input": [["StartTag", "head", {}], ["StartTag", "foo", {}]],
|
||||
"expected": ["<foo>"]
|
||||
},
|
||||
|
||||
{"description": "head start-tag followed by end-tag",
|
||||
"input": [["StartTag", "head", {}], ["EndTag", "foo", {}]],
|
||||
"expected": ["<head></foo>"]
|
||||
},
|
||||
|
||||
{"description": "head start-tag at EOF (shouldn't ever happen?!)",
|
||||
"input": [["StartTag", "head", {}]],
|
||||
"expected": ["<head>"]
|
||||
},
|
||||
|
||||
|
||||
|
||||
{"description": "head end-tag followed by comment",
|
||||
"input": [["EndTag", "head"], ["Comment", "foo"]],
|
||||
"expected": ["</head><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "head end-tag followed by space character",
|
||||
"input": [["EndTag", "head"], ["Characters", " foo"]],
|
||||
"expected": ["</head> foo"]
|
||||
},
|
||||
|
||||
{"description": "head end-tag followed by text",
|
||||
"input": [["EndTag", "head"], ["Characters", "foo"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "head end-tag followed by start-tag",
|
||||
"input": [["EndTag", "head"], ["StartTag", "foo", {}]],
|
||||
"expected": ["<foo>"]
|
||||
},
|
||||
|
||||
{"description": "head end-tag followed by end-tag",
|
||||
"input": [["EndTag", "head"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "head end-tag at EOF",
|
||||
"input": [["EndTag", "head"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "body start-tag followed by comment",
|
||||
"input": [["StartTag", "body", {}], ["Comment", "foo"]],
|
||||
"expected": ["<body><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "body start-tag followed by space character",
|
||||
"input": [["StartTag", "body", {}], ["Characters", " foo"]],
|
||||
"expected": ["<body> foo"]
|
||||
},
|
||||
|
||||
{"description": "body start-tag followed by text",
|
||||
"input": [["StartTag", "body", {}], ["Characters", "foo"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "body start-tag followed by start-tag",
|
||||
"input": [["StartTag", "body", {}], ["StartTag", "foo", {}]],
|
||||
"expected": ["<foo>"]
|
||||
},
|
||||
|
||||
{"description": "body start-tag followed by end-tag",
|
||||
"input": [["StartTag", "body", {}], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "body start-tag at EOF (shouldn't ever happen?!)",
|
||||
"input": [["StartTag", "body", {}]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
{"description": "body end-tag followed by comment",
|
||||
"input": [["EndTag", "body"], ["Comment", "foo"]],
|
||||
"expected": ["</body><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "body end-tag followed by space character",
|
||||
"input": [["EndTag", "body"], ["Characters", " foo"]],
|
||||
"expected": ["</body> foo"]
|
||||
},
|
||||
|
||||
{"description": "body end-tag followed by text",
|
||||
"input": [["EndTag", "body"], ["Characters", "foo"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "body end-tag followed by start-tag",
|
||||
"input": [["EndTag", "body"], ["StartTag", "foo", {}]],
|
||||
"expected": ["<foo>"]
|
||||
},
|
||||
|
||||
{"description": "body end-tag followed by end-tag",
|
||||
"input": [["EndTag", "body"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "body end-tag at EOF",
|
||||
"input": [["EndTag", "body"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "li end-tag followed by comment",
|
||||
"input": [["EndTag", "li"], ["Comment", "foo"]],
|
||||
"expected": ["</li><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "li end-tag followed by space character",
|
||||
"input": [["EndTag", "li"], ["Characters", " foo"]],
|
||||
"expected": ["</li> foo"]
|
||||
},
|
||||
|
||||
{"description": "li end-tag followed by text",
|
||||
"input": [["EndTag", "li"], ["Characters", "foo"]],
|
||||
"expected": ["</li>foo"]
|
||||
},
|
||||
|
||||
{"description": "li end-tag followed by start-tag",
|
||||
"input": [["EndTag", "li"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</li><foo>"]
|
||||
},
|
||||
|
||||
{"description": "li end-tag followed by li start-tag",
|
||||
"input": [["EndTag", "li"], ["StartTag", "li", {}]],
|
||||
"expected": ["<li>"]
|
||||
},
|
||||
|
||||
{"description": "li end-tag followed by end-tag",
|
||||
"input": [["EndTag", "li"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "li end-tag at EOF",
|
||||
"input": [["EndTag", "li"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "dt end-tag followed by comment",
|
||||
"input": [["EndTag", "dt"], ["Comment", "foo"]],
|
||||
"expected": ["</dt><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "dt end-tag followed by space character",
|
||||
"input": [["EndTag", "dt"], ["Characters", " foo"]],
|
||||
"expected": ["</dt> foo"]
|
||||
},
|
||||
|
||||
{"description": "dt end-tag followed by text",
|
||||
"input": [["EndTag", "dt"], ["Characters", "foo"]],
|
||||
"expected": ["</dt>foo"]
|
||||
},
|
||||
|
||||
{"description": "dt end-tag followed by start-tag",
|
||||
"input": [["EndTag", "dt"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</dt><foo>"]
|
||||
},
|
||||
|
||||
{"description": "dt end-tag followed by dt start-tag",
|
||||
"input": [["EndTag", "dt"], ["StartTag", "dt", {}]],
|
||||
"expected": ["<dt>"]
|
||||
},
|
||||
|
||||
{"description": "dt end-tag followed by dd start-tag",
|
||||
"input": [["EndTag", "dt"], ["StartTag", "dd", {}]],
|
||||
"expected": ["<dd>"]
|
||||
},
|
||||
|
||||
{"description": "dt end-tag followed by end-tag",
|
||||
"input": [["EndTag", "dt"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</dt></foo>"]
|
||||
},
|
||||
|
||||
{"description": "dt end-tag at EOF",
|
||||
"input": [["EndTag", "dt"]],
|
||||
"expected": ["</dt>"]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "dd end-tag followed by comment",
|
||||
"input": [["EndTag", "dd"], ["Comment", "foo"]],
|
||||
"expected": ["</dd><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "dd end-tag followed by space character",
|
||||
"input": [["EndTag", "dd"], ["Characters", " foo"]],
|
||||
"expected": ["</dd> foo"]
|
||||
},
|
||||
|
||||
{"description": "dd end-tag followed by text",
|
||||
"input": [["EndTag", "dd"], ["Characters", "foo"]],
|
||||
"expected": ["</dd>foo"]
|
||||
},
|
||||
|
||||
{"description": "dd end-tag followed by start-tag",
|
||||
"input": [["EndTag", "dd"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</dd><foo>"]
|
||||
},
|
||||
|
||||
{"description": "dd end-tag followed by dd start-tag",
|
||||
"input": [["EndTag", "dd"], ["StartTag", "dd", {}]],
|
||||
"expected": ["<dd>"]
|
||||
},
|
||||
|
||||
{"description": "dd end-tag followed by dt start-tag",
|
||||
"input": [["EndTag", "dd"], ["StartTag", "dt", {}]],
|
||||
"expected": ["<dt>"]
|
||||
},
|
||||
|
||||
{"description": "dd end-tag followed by end-tag",
|
||||
"input": [["EndTag", "dd"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "dd end-tag at EOF",
|
||||
"input": [["EndTag", "dd"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "p end-tag followed by comment",
|
||||
"input": [["EndTag", "p"], ["Comment", "foo"]],
|
||||
"expected": ["</p><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by space character",
|
||||
"input": [["EndTag", "p"], ["Characters", " foo"]],
|
||||
"expected": ["</p> foo"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by text",
|
||||
"input": [["EndTag", "p"], ["Characters", "foo"]],
|
||||
"expected": ["</p>foo"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</p><foo>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by address start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "address", {}]],
|
||||
"expected": ["<address>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by blockquote start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "blockquote", {}]],
|
||||
"expected": ["<blockquote>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by dl start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "dl", {}]],
|
||||
"expected": ["<dl>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by fieldset start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "fieldset", {}]],
|
||||
"expected": ["<fieldset>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by form start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "form", {}]],
|
||||
"expected": ["<form>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by h1 start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "h1", {}]],
|
||||
"expected": ["<h1>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by h2 start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "h2", {}]],
|
||||
"expected": ["<h2>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by h3 start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "h3", {}]],
|
||||
"expected": ["<h3>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by h4 start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "h4", {}]],
|
||||
"expected": ["<h4>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by h5 start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "h5", {}]],
|
||||
"expected": ["<h5>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by h6 start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "h6", {}]],
|
||||
"expected": ["<h6>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by hr start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "hr", {}]],
|
||||
"expected": ["<hr>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by menu start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "menu", {}]],
|
||||
"expected": ["<menu>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by ol start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "ol", {}]],
|
||||
"expected": ["<ol>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by p start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "p", {}]],
|
||||
"expected": ["<p>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by pre start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "pre", {}]],
|
||||
"expected": ["<pre>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by table start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "table", {}]],
|
||||
"expected": ["<table>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by ul start-tag",
|
||||
"input": [["EndTag", "p"], ["StartTag", "ul", {}]],
|
||||
"expected": ["<ul>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag followed by end-tag",
|
||||
"input": [["EndTag", "p"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "p end-tag at EOF",
|
||||
"input": [["EndTag", "p"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "optgroup end-tag followed by comment",
|
||||
"input": [["EndTag", "optgroup"], ["Comment", "foo"]],
|
||||
"expected": ["</optgroup><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "optgroup end-tag followed by space character",
|
||||
"input": [["EndTag", "optgroup"], ["Characters", " foo"]],
|
||||
"expected": ["</optgroup> foo"]
|
||||
},
|
||||
|
||||
{"description": "optgroup end-tag followed by text",
|
||||
"input": [["EndTag", "optgroup"], ["Characters", "foo"]],
|
||||
"expected": ["</optgroup>foo"]
|
||||
},
|
||||
|
||||
{"description": "optgroup end-tag followed by start-tag",
|
||||
"input": [["EndTag", "optgroup"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</optgroup><foo>"]
|
||||
},
|
||||
|
||||
{"description": "optgroup end-tag followed by optgroup start-tag",
|
||||
"input": [["EndTag", "optgroup"], ["StartTag", "optgroup", {}]],
|
||||
"expected": ["<optgroup>"]
|
||||
},
|
||||
|
||||
{"description": "optgroup end-tag followed by end-tag",
|
||||
"input": [["EndTag", "optgroup"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "optgroup end-tag at EOF",
|
||||
"input": [["EndTag", "optgroup"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "option end-tag followed by comment",
|
||||
"input": [["EndTag", "option"], ["Comment", "foo"]],
|
||||
"expected": ["</option><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "option end-tag followed by space character",
|
||||
"input": [["EndTag", "option"], ["Characters", " foo"]],
|
||||
"expected": ["</option> foo"]
|
||||
},
|
||||
|
||||
{"description": "option end-tag followed by text",
|
||||
"input": [["EndTag", "option"], ["Characters", "foo"]],
|
||||
"expected": ["</option>foo"]
|
||||
},
|
||||
|
||||
{"description": "option end-tag followed by start-tag",
|
||||
"input": [["EndTag", "option"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</option><foo>"]
|
||||
},
|
||||
|
||||
{"description": "option end-tag followed by option start-tag",
|
||||
"input": [["EndTag", "option"], ["StartTag", "option", {}]],
|
||||
"expected": ["<option>"]
|
||||
},
|
||||
|
||||
{"description": "option end-tag followed by end-tag",
|
||||
"input": [["EndTag", "option"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "option end-tag at EOF",
|
||||
"input": [["EndTag", "option"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "colgroup start-tag followed by comment",
|
||||
"input": [["StartTag", "colgroup", {}], ["Comment", "foo"]],
|
||||
"expected": ["<colgroup><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "colgroup start-tag followed by space character",
|
||||
"input": [["StartTag", "colgroup", {}], ["Characters", " foo"]],
|
||||
"expected": ["<colgroup> foo"]
|
||||
},
|
||||
|
||||
{"description": "colgroup start-tag followed by text",
|
||||
"input": [["StartTag", "colgroup", {}], ["Characters", "foo"]],
|
||||
"expected": ["<colgroup>foo"]
|
||||
},
|
||||
|
||||
{"description": "colgroup start-tag followed by start-tag",
|
||||
"input": [["StartTag", "colgroup", {}], ["StartTag", "foo", {}]],
|
||||
"expected": ["<colgroup><foo>"]
|
||||
},
|
||||
|
||||
{"description": "first colgroup in a table with a col child",
|
||||
"input": [["StartTag", "table", {}], ["StartTag", "colgroup", {}], ["StartTag", "col", {}]],
|
||||
"expected": ["<table><col>"]
|
||||
},
|
||||
|
||||
{"description": "colgroup with a col child, following another colgroup",
|
||||
"input": [["EndTag", "colgroup", {}], ["StartTag", "colgroup", {}], ["StartTag", "col", {}]],
|
||||
"expected": ["</colgroup><col>", "<colgroup><col>"]
|
||||
},
|
||||
|
||||
{"description": "colgroup start-tag followed by end-tag",
|
||||
"input": [["StartTag", "colgroup", {}], ["EndTag", "foo", {}]],
|
||||
"expected": ["<colgroup></foo>"]
|
||||
},
|
||||
|
||||
{"description": "colgroup start-tag at EOF",
|
||||
"input": [["StartTag", "colgroup", {}]],
|
||||
"expected": ["<colgroup>"]
|
||||
},
|
||||
|
||||
|
||||
|
||||
{"description": "colgroup end-tag followed by comment",
|
||||
"input": [["EndTag", "colgroup"], ["Comment", "foo"]],
|
||||
"expected": ["</colgroup><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "colgroup end-tag followed by space character",
|
||||
"input": [["EndTag", "colgroup"], ["Characters", " foo"]],
|
||||
"expected": ["</colgroup> foo"]
|
||||
},
|
||||
|
||||
{"description": "colgroup end-tag followed by text",
|
||||
"input": [["EndTag", "colgroup"], ["Characters", "foo"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "colgroup end-tag followed by start-tag",
|
||||
"input": [["EndTag", "colgroup"], ["StartTag", "foo", {}]],
|
||||
"expected": ["<foo>"]
|
||||
},
|
||||
|
||||
{"description": "colgroup end-tag followed by end-tag",
|
||||
"input": [["EndTag", "colgroup"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "colgroup end-tag at EOF",
|
||||
"input": [["EndTag", "colgroup"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "thead end-tag followed by comment",
|
||||
"input": [["EndTag", "thead"], ["Comment", "foo"]],
|
||||
"expected": ["</thead><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "thead end-tag followed by space character",
|
||||
"input": [["EndTag", "thead"], ["Characters", " foo"]],
|
||||
"expected": ["</thead> foo"]
|
||||
},
|
||||
|
||||
{"description": "thead end-tag followed by text",
|
||||
"input": [["EndTag", "thead"], ["Characters", "foo"]],
|
||||
"expected": ["</thead>foo"]
|
||||
},
|
||||
|
||||
{"description": "thead end-tag followed by start-tag",
|
||||
"input": [["EndTag", "thead"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</thead><foo>"]
|
||||
},
|
||||
|
||||
{"description": "thead end-tag followed by tbody start-tag",
|
||||
"input": [["EndTag", "thead"], ["StartTag", "tbody", {}]],
|
||||
"expected": ["<tbody>"]
|
||||
},
|
||||
|
||||
{"description": "thead end-tag followed by tfoot start-tag",
|
||||
"input": [["EndTag", "thead"], ["StartTag", "tfoot", {}]],
|
||||
"expected": ["<tfoot>"]
|
||||
},
|
||||
|
||||
{"description": "thead end-tag followed by end-tag",
|
||||
"input": [["EndTag", "thead"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</thead></foo>"]
|
||||
},
|
||||
|
||||
{"description": "thead end-tag at EOF",
|
||||
"input": [["EndTag", "thead"]],
|
||||
"expected": ["</thead>"]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "tbody start-tag followed by comment",
|
||||
"input": [["StartTag", "tbody", {}], ["Comment", "foo"]],
|
||||
"expected": ["<tbody><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "tbody start-tag followed by space character",
|
||||
"input": [["StartTag", "tbody", {}], ["Characters", " foo"]],
|
||||
"expected": ["<tbody> foo"]
|
||||
},
|
||||
|
||||
{"description": "tbody start-tag followed by text",
|
||||
"input": [["StartTag", "tbody", {}], ["Characters", "foo"]],
|
||||
"expected": ["<tbody>foo"]
|
||||
},
|
||||
|
||||
{"description": "tbody start-tag followed by start-tag",
|
||||
"input": [["StartTag", "tbody", {}], ["StartTag", "foo", {}]],
|
||||
"expected": ["<tbody><foo>"]
|
||||
},
|
||||
|
||||
{"description": "first tbody in a table with a tr child",
|
||||
"input": [["StartTag", "table", {}], ["StartTag", "tbody", {}], ["StartTag", "tr", {}]],
|
||||
"expected": ["<table><tr>"]
|
||||
},
|
||||
|
||||
{"description": "tbody with a tr child, following another tbody",
|
||||
"input": [["EndTag", "tbody", {}], ["StartTag", "tbody", {}], ["StartTag", "tr", {}]],
|
||||
"expected": ["<tbody><tr>", "</tbody><tr>"]
|
||||
},
|
||||
|
||||
{"description": "tbody with a tr child, following a thead",
|
||||
"input": [["EndTag", "thead", {}], ["StartTag", "tbody", {}], ["StartTag", "tr", {}]],
|
||||
"expected": ["<tbody><tr>", "</thead><tr>"]
|
||||
},
|
||||
|
||||
{"description": "tbody with a tr child, following a tfoot",
|
||||
"input": [["EndTag", "tfoot", {}], ["StartTag", "tbody", {}], ["StartTag", "tr", {}]],
|
||||
"expected": ["<tbody><tr>", "</tfoot><tr>"]
|
||||
},
|
||||
|
||||
{"description": "tbody start-tag followed by end-tag",
|
||||
"input": [["StartTag", "tbody", {}], ["EndTag", "foo", {}]],
|
||||
"expected": ["<tbody></foo>"]
|
||||
},
|
||||
|
||||
{"description": "tbody start-tag at EOF",
|
||||
"input": [["StartTag", "tbody", {}]],
|
||||
"expected": ["<tbody>"]
|
||||
},
|
||||
|
||||
|
||||
|
||||
{"description": "tbody end-tag followed by comment",
|
||||
"input": [["EndTag", "tbody"], ["Comment", "foo"]],
|
||||
"expected": ["</tbody><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "tbody end-tag followed by space character",
|
||||
"input": [["EndTag", "tbody"], ["Characters", " foo"]],
|
||||
"expected": ["</tbody> foo"]
|
||||
},
|
||||
|
||||
{"description": "tbody end-tag followed by text",
|
||||
"input": [["EndTag", "tbody"], ["Characters", "foo"]],
|
||||
"expected": ["</tbody>foo"]
|
||||
},
|
||||
|
||||
{"description": "tbody end-tag followed by start-tag",
|
||||
"input": [["EndTag", "tbody"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</tbody><foo>"]
|
||||
},
|
||||
|
||||
{"description": "tbody end-tag followed by tbody start-tag",
|
||||
"input": [["EndTag", "tbody"], ["StartTag", "tbody", {}]],
|
||||
"expected": ["<tbody>", "</tbody>"]
|
||||
},
|
||||
|
||||
{"description": "tbody end-tag followed by tfoot start-tag",
|
||||
"input": [["EndTag", "tbody"], ["StartTag", "tfoot", {}]],
|
||||
"expected": ["<tfoot>"]
|
||||
},
|
||||
|
||||
{"description": "tbody end-tag followed by end-tag",
|
||||
"input": [["EndTag", "tbody"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "tbody end-tag at EOF",
|
||||
"input": [["EndTag", "tbody"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "tfoot end-tag followed by comment",
|
||||
"input": [["EndTag", "tfoot"], ["Comment", "foo"]],
|
||||
"expected": ["</tfoot><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "tfoot end-tag followed by space character",
|
||||
"input": [["EndTag", "tfoot"], ["Characters", " foo"]],
|
||||
"expected": ["</tfoot> foo"]
|
||||
},
|
||||
|
||||
{"description": "tfoot end-tag followed by text",
|
||||
"input": [["EndTag", "tfoot"], ["Characters", "foo"]],
|
||||
"expected": ["</tfoot>foo"]
|
||||
},
|
||||
|
||||
{"description": "tfoot end-tag followed by start-tag",
|
||||
"input": [["EndTag", "tfoot"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</tfoot><foo>"]
|
||||
},
|
||||
|
||||
{"description": "tfoot end-tag followed by tbody start-tag",
|
||||
"input": [["EndTag", "tfoot"], ["StartTag", "tbody", {}]],
|
||||
"expected": ["<tbody>", "</tfoot>"]
|
||||
},
|
||||
|
||||
{"description": "tfoot end-tag followed by end-tag",
|
||||
"input": [["EndTag", "tfoot"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "tfoot end-tag at EOF",
|
||||
"input": [["EndTag", "tfoot"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "tr end-tag followed by comment",
|
||||
"input": [["EndTag", "tr"], ["Comment", "foo"]],
|
||||
"expected": ["</tr><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "tr end-tag followed by space character",
|
||||
"input": [["EndTag", "tr"], ["Characters", " foo"]],
|
||||
"expected": ["</tr> foo"]
|
||||
},
|
||||
|
||||
{"description": "tr end-tag followed by text",
|
||||
"input": [["EndTag", "tr"], ["Characters", "foo"]],
|
||||
"expected": ["</tr>foo"]
|
||||
},
|
||||
|
||||
{"description": "tr end-tag followed by start-tag",
|
||||
"input": [["EndTag", "tr"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</tr><foo>"]
|
||||
},
|
||||
|
||||
{"description": "tr end-tag followed by tr start-tag",
|
||||
"input": [["EndTag", "tr"], ["StartTag", "tr", {}]],
|
||||
"expected": ["<tr>", "</tr>"]
|
||||
},
|
||||
|
||||
{"description": "tr end-tag followed by end-tag",
|
||||
"input": [["EndTag", "tr"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "tr end-tag at EOF",
|
||||
"input": [["EndTag", "tr"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "td end-tag followed by comment",
|
||||
"input": [["EndTag", "td"], ["Comment", "foo"]],
|
||||
"expected": ["</td><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "td end-tag followed by space character",
|
||||
"input": [["EndTag", "td"], ["Characters", " foo"]],
|
||||
"expected": ["</td> foo"]
|
||||
},
|
||||
|
||||
{"description": "td end-tag followed by text",
|
||||
"input": [["EndTag", "td"], ["Characters", "foo"]],
|
||||
"expected": ["</td>foo"]
|
||||
},
|
||||
|
||||
{"description": "td end-tag followed by start-tag",
|
||||
"input": [["EndTag", "td"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</td><foo>"]
|
||||
},
|
||||
|
||||
{"description": "td end-tag followed by td start-tag",
|
||||
"input": [["EndTag", "td"], ["StartTag", "td", {}]],
|
||||
"expected": ["<td>", "</td>"]
|
||||
},
|
||||
|
||||
{"description": "td end-tag followed by th start-tag",
|
||||
"input": [["EndTag", "td"], ["StartTag", "th", {}]],
|
||||
"expected": ["<th>", "</td>"]
|
||||
},
|
||||
|
||||
{"description": "td end-tag followed by end-tag",
|
||||
"input": [["EndTag", "td"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "td end-tag at EOF",
|
||||
"input": [["EndTag", "td"]],
|
||||
"expected": [""]
|
||||
},
|
||||
|
||||
|
||||
|
||||
|
||||
{"description": "th end-tag followed by comment",
|
||||
"input": [["EndTag", "th"], ["Comment", "foo"]],
|
||||
"expected": ["</th><!--foo-->"]
|
||||
},
|
||||
|
||||
{"description": "th end-tag followed by space character",
|
||||
"input": [["EndTag", "th"], ["Characters", " foo"]],
|
||||
"expected": ["</th> foo"]
|
||||
},
|
||||
|
||||
{"description": "th end-tag followed by text",
|
||||
"input": [["EndTag", "th"], ["Characters", "foo"]],
|
||||
"expected": ["</th>foo"]
|
||||
},
|
||||
|
||||
{"description": "th end-tag followed by start-tag",
|
||||
"input": [["EndTag", "th"], ["StartTag", "foo", {}]],
|
||||
"expected": ["</th><foo>"]
|
||||
},
|
||||
|
||||
{"description": "th end-tag followed by th start-tag",
|
||||
"input": [["EndTag", "th"], ["StartTag", "th", {}]],
|
||||
"expected": ["<th>", "</th>"]
|
||||
},
|
||||
|
||||
{"description": "th end-tag followed by td start-tag",
|
||||
"input": [["EndTag", "th"], ["StartTag", "td", {}]],
|
||||
"expected": ["<td>", "</th>"]
|
||||
},
|
||||
|
||||
{"description": "th end-tag followed by end-tag",
|
||||
"input": [["EndTag", "th"], ["EndTag", "foo", {}]],
|
||||
"expected": ["</foo>"]
|
||||
},
|
||||
|
||||
{"description": "th end-tag at EOF",
|
||||
"input": [["EndTag", "th"]],
|
||||
"expected": [""]
|
||||
}
|
||||
|
||||
]}
|
||||
45
vendor/plugins/HTML5lib/testdata/serializer/options.test
vendored
Normal file
45
vendor/plugins/HTML5lib/testdata/serializer/options.test
vendored
Normal file
|
|
@ -0,0 +1,45 @@
|
|||
{"tests":[
|
||||
|
||||
{"description": "quote_char=\"'\"",
|
||||
"options": {"quote_char": "'"},
|
||||
"input": [["StartTag", "span", {"title": "test 'with' quote_char"}]],
|
||||
"expected": ["<span title='test 'with' quote_char'>"]
|
||||
},
|
||||
|
||||
{"description": "quote_attr_values=true",
|
||||
"options": {"quote_attr_values": true},
|
||||
"input": [["StartTag", "button", {"disabled": "disabled"}]],
|
||||
"expected": ["<button disabled>"]
|
||||
},
|
||||
|
||||
{"description": "quote_attr_values=true with irrelevant",
|
||||
"options": {"quote_attr_values": true},
|
||||
"input": [["StartTag", "div", {"irrelevant": "irrelevant"}]],
|
||||
"expected": ["<div irrelevant>"]
|
||||
},
|
||||
|
||||
{"description": "use_trailing_solidus=true with void element",
|
||||
"options": {"use_trailing_solidus": true},
|
||||
"input": [["EmptyTag", "img", {}]],
|
||||
"expected": ["<img />"]
|
||||
},
|
||||
|
||||
{"description": "use_trailing_solidus=true with non-void element",
|
||||
"options": {"use_trailing_solidus": true},
|
||||
"input": [["StartTag", "div", {}]],
|
||||
"expected": ["<div>"]
|
||||
},
|
||||
|
||||
{"description": "minimize_boolean_attributes=false",
|
||||
"options": {"minimize_boolean_attributes": false},
|
||||
"input": [["StartTag", "div", {"irrelevant": "irrelevant"}]],
|
||||
"expected": ["<div irrelevant=irrelevant>"]
|
||||
},
|
||||
|
||||
{"description": "minimize_boolean_attributes=false with empty value",
|
||||
"options": {"minimize_boolean_attributes": false},
|
||||
"input": [["StartTag", "div", {"irrelevant": ""}]],
|
||||
"expected": ["<div irrelevant=\"\">"]
|
||||
}
|
||||
|
||||
]}
|
||||
51
vendor/plugins/HTML5lib/testdata/serializer/whitespace.test
vendored
Normal file
51
vendor/plugins/HTML5lib/testdata/serializer/whitespace.test
vendored
Normal file
|
|
@ -0,0 +1,51 @@
|
|||
{"tests": [
|
||||
|
||||
{"description": "bare text with leading spaces",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["Characters", "\t\r\n\u000B\u000C foo"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "bare text with trailing spaces",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["Characters", "foo \t\r\n\u000B\u000C"]],
|
||||
"expected": ["foo"]
|
||||
},
|
||||
|
||||
{"description": "bare text with inner spaces",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["Characters", "foo \t\r\n\u000B\u000C bar"]],
|
||||
"expected": ["foo bar"]
|
||||
},
|
||||
|
||||
{"description": "text within <pre>",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["StartTag", "pre", {}], ["Characters", "\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C"], ["EndTag", "pre"]],
|
||||
"expected": ["<pre>\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C</pre>"]
|
||||
},
|
||||
|
||||
{"description": "text within <pre>, with inner markup",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["StartTag", "pre", {}], ["Characters", "\t\r\n\u000B\u000C fo"], ["StartTag", "span", {}], ["Characters", "o \t\r\n\u000B\u000C b"], ["EndTag", "span"], ["Characters", "ar \t\r\n\u000B\u000C"], ["EndTag", "pre"]],
|
||||
"expected": ["<pre>\t\r\n\u000B\u000C fo<span>o \t\r\n\u000B\u000C b</span>ar \t\r\n\u000B\u000C</pre>"]
|
||||
},
|
||||
|
||||
{"description": "text within <textarea>",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["StartTag", "textarea", {}], ["Characters", "\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C"], ["EndTag", "textarea"]],
|
||||
"expected": ["<textarea>\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C</textarea>"]
|
||||
},
|
||||
|
||||
{"description": "text within <script>",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["StartTag", "script", {}], ["Characters", "\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C"], ["EndTag", "script"]],
|
||||
"expected": ["<script>\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C</script>"]
|
||||
},
|
||||
|
||||
{"description": "text within <style>",
|
||||
"options": {"strip_whitespace": true},
|
||||
"input": [["StartTag", "style", {}], ["Characters", "\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C"], ["EndTag", "style"]],
|
||||
"expected": ["<style>\t\r\n\u000B\u000C foo \t\r\n\u000B\u000C bar \t\r\n\u000B\u000C</style>"]
|
||||
}
|
||||
|
||||
]}
|
||||
1
vendor/plugins/HTML5lib/testdata/sites/google-results.htm
vendored
Executable file
1
vendor/plugins/HTML5lib/testdata/sites/google-results.htm
vendored
Executable file
File diff suppressed because one or more lines are too long
1
vendor/plugins/HTML5lib/testdata/sites/python-ref-import.htm
vendored
Executable file
1
vendor/plugins/HTML5lib/testdata/sites/python-ref-import.htm
vendored
Executable file
File diff suppressed because one or more lines are too long
1
vendor/plugins/HTML5lib/testdata/sites/web-apps-old.htm
vendored
Executable file
1
vendor/plugins/HTML5lib/testdata/sites/web-apps-old.htm
vendored
Executable file
File diff suppressed because one or more lines are too long
34275
vendor/plugins/HTML5lib/testdata/sites/web-apps.htm
vendored
Executable file
34275
vendor/plugins/HTML5lib/testdata/sites/web-apps.htm
vendored
Executable file
File diff suppressed because it is too large
Load diff
36
vendor/plugins/HTML5lib/testdata/tokenizer/contentModelFlags.test
vendored
Normal file
36
vendor/plugins/HTML5lib/testdata/tokenizer/contentModelFlags.test
vendored
Normal file
|
|
@ -0,0 +1,36 @@
|
|||
{"tests": [
|
||||
|
||||
{"description":"PLAINTEXT content model flag",
|
||||
"contentModelFlags":["PLAINTEXT"],
|
||||
"input":"<head>&body;",
|
||||
"output":[["Character", "<head>&body;"]]},
|
||||
|
||||
{"description":"End tag closing RCDATA or CDATA",
|
||||
"contentModelFlags":["RCDATA", "CDATA"],
|
||||
"lastStartTag":"bar",
|
||||
"input":"foo</bar>",
|
||||
"output":[["Character", "foo"], ["EndTag", "bar"]]},
|
||||
|
||||
{"description":"End tag with incorrect name in RCDATA or CDATA",
|
||||
"contentModelFlags":["RCDATA", "CDATA"],
|
||||
"lastStartTag":"baz",
|
||||
"input":"</foo>bar</baz>",
|
||||
"output":["ParseError", ["Character", "</foo>bar"], ["EndTag", "baz"]]},
|
||||
|
||||
{"description":"End tag closing RCDATA or CDATA, switching back to PCDATA",
|
||||
"contentModelFlags":["RCDATA", "CDATA"],
|
||||
"lastStartTag":"bar",
|
||||
"input":"foo</bar></baz>",
|
||||
"output":[["Character", "foo"], ["EndTag", "bar"], ["EndTag", "baz"]]},
|
||||
|
||||
{"description":"CDATA w/ something looking like an entity",
|
||||
"contentModelFlags":["CDATA"],
|
||||
"input":"&foo;",
|
||||
"output":[["Character", "&foo;"]]},
|
||||
|
||||
{"description":"RCDATA w/ an entity",
|
||||
"contentModelFlags":["RCDATA"],
|
||||
"input":"<",
|
||||
"output":[["Character", "<"]]}
|
||||
|
||||
]}
|
||||
136
vendor/plugins/HTML5lib/testdata/tokenizer/test1.test
vendored
Normal file
136
vendor/plugins/HTML5lib/testdata/tokenizer/test1.test
vendored
Normal file
|
|
@ -0,0 +1,136 @@
|
|||
{"tests": [
|
||||
|
||||
{"description":"Correct Doctype lowercase",
|
||||
"input":"<!DOCTYPE html>",
|
||||
"output":[["DOCTYPE", "HTML", false]]},
|
||||
|
||||
{"description":"Correct Doctype uppercase",
|
||||
"input":"<!DOCTYPE HTML>",
|
||||
"output":[["DOCTYPE", "HTML", false]]},
|
||||
|
||||
{"description":"Correct Doctype mixed case",
|
||||
"input":"<!DOCTYPE HtMl>",
|
||||
"output":[["DOCTYPE", "HTML", false]]},
|
||||
|
||||
{"description":"Truncated doctype start",
|
||||
"input":"<!DOC>",
|
||||
"output":["ParseError", ["Comment", "DOC"]]},
|
||||
|
||||
{"description":"Doctype in error",
|
||||
"input":"<!DOCTYPE foo>",
|
||||
"output":[["DOCTYPE", "FOO", true]]},
|
||||
|
||||
{"description":"Single Start Tag",
|
||||
"input":"<h>",
|
||||
"output":[["StartTag", "h", {}]]},
|
||||
|
||||
{"description":"Empty end tag",
|
||||
"input":"</>",
|
||||
"output":["ParseError"]},
|
||||
|
||||
{"description":"Empty start tag",
|
||||
"input":"<>",
|
||||
"output":["ParseError", ["Character", "<>"]]},
|
||||
|
||||
{"description":"Start Tag w/attribute",
|
||||
"input":"<h a='b'>",
|
||||
"output":[["StartTag", "h", {"a":"b"}]]},
|
||||
|
||||
{"description":"Start Tag w/attribute no quotes",
|
||||
"input":"<h a=b>",
|
||||
"output":[["StartTag", "h", {"a":"b"}]]},
|
||||
|
||||
{"description":"Start/End Tag",
|
||||
"input":"<h></h>",
|
||||
"output":[["StartTag", "h", {}], ["EndTag", "h"]]},
|
||||
|
||||
{"description":"Two unclosed start tags",
|
||||
"input":"<p>One<p>Two",
|
||||
"output":[["StartTag", "p", {}], ["Character", "One"], ["StartTag", "p", {}], ["Character", "Two"]]},
|
||||
|
||||
{"description":"End Tag w/attribute",
|
||||
"input":"<h></h a='b'>",
|
||||
"output":[["StartTag", "h", {}], "ParseError", ["EndTag", "h"]]},
|
||||
|
||||
{"description":"Multiple atts",
|
||||
"input":"<h a='b' c='d'>",
|
||||
"output":[["StartTag", "h", {"a":"b", "c":"d"}]]},
|
||||
|
||||
{"description":"Multiple atts no space",
|
||||
"input":"<h a='b'c='d'>",
|
||||
"output":[["StartTag", "h", {"a":"b", "c":"d"}]]},
|
||||
|
||||
{"description":"Repeated attr",
|
||||
"input":"<h a='b' a='d'>",
|
||||
"output":["ParseError", ["StartTag", "h", {"a":"b"}]]},
|
||||
|
||||
{"description":"Simple comment",
|
||||
"input":"<!--comment-->",
|
||||
"output":[["Comment", "comment"]]},
|
||||
|
||||
{"description":"Comment, Central dash no space",
|
||||
"input":"<!----->",
|
||||
"output":["ParseError", ["Comment", "-"]]},
|
||||
|
||||
{"description":"Comment, two central dashes",
|
||||
"input":"<!-- --comment -->",
|
||||
"output":["ParseError", ["Comment", " --comment "]]},
|
||||
|
||||
{"description":"Unfinished comment",
|
||||
"input":"<!--comment",
|
||||
"output":["ParseError", ["Comment", "comment"]]},
|
||||
|
||||
{"description":"Start of a comment",
|
||||
"input":"<!-",
|
||||
"output":["ParseError", ["Comment", "-"]]},
|
||||
|
||||
{"description":"Ampersand only",
|
||||
"input":"&",
|
||||
"output":["ParseError", ["Character", "&"]]},
|
||||
|
||||
{"description":"Unfinished entity",
|
||||
"input":"&f",
|
||||
"output":["ParseError", ["Character", "&"], ["Character", "f"]]},
|
||||
|
||||
{"description":"Ampersand, number sign",
|
||||
"input":"&#",
|
||||
"output":["ParseError", ["Character", "&"], ["Character", "#"]]},
|
||||
|
||||
{"description":"Unfinished numeric entity",
|
||||
"input":"&#x",
|
||||
"output":["ParseError", ["Character", "&#x"]]},
|
||||
|
||||
{"description":"Entity with trailing semicolon (1)",
|
||||
"input":"I'm ¬it",
|
||||
"output":[["Character","I'm ¬it"]]},
|
||||
|
||||
{"description":"Entity with trailing semicolon (2)",
|
||||
"input":"I'm ∉",
|
||||
"output":[["Character","I'm ∉"]]},
|
||||
|
||||
{"description":"Entity without trailing semicolon (1)",
|
||||
"input":"I'm ¬it",
|
||||
"output":[["Character","I'm "], "ParseError", ["Character", "¬"],
|
||||
["Character", "it"]]},
|
||||
|
||||
{"description":"Entity without trailing semicolon (2)",
|
||||
"input":"I'm ¬in",
|
||||
"output":[["Character","I'm "], "ParseError", ["Character", "∉"]]},
|
||||
|
||||
{"description":"Partial entity match at end of file",
|
||||
"input":"I'm &no",
|
||||
"output":[["Character","I'm "], "ParseError", ["Character", "&no"]]},
|
||||
|
||||
{"description":"ASCII decimal entity",
|
||||
"input":"$",
|
||||
"output":[["Character","$"]]},
|
||||
|
||||
{"description":"ASCII hexadecimal entity",
|
||||
"input":"?",
|
||||
"output":[["Character","?"]]},
|
||||
|
||||
{"description":"Hexadecimal entity in attribute",
|
||||
"input":"<h a='?'></h>",
|
||||
"output":[["StartTag", "h", {"a":"?"}], ["EndTag", "h"]]}
|
||||
|
||||
]}
|
||||
105
vendor/plugins/HTML5lib/testdata/tokenizer/test2.test
vendored
Normal file
105
vendor/plugins/HTML5lib/testdata/tokenizer/test2.test
vendored
Normal file
|
|
@ -0,0 +1,105 @@
|
|||
{"tests": [
|
||||
|
||||
{"description":"Doctype without a name",
|
||||
"input":"<!DOCTYPE>",
|
||||
"output":["ParseError", "ParseError", ["DOCTYPE", "", true]]},
|
||||
|
||||
{"description":"Correct doctype without a space before name",
|
||||
"input":"<!DOCTYPEhtml>",
|
||||
"output":["ParseError", ["DOCTYPE", "HTML", false]]},
|
||||
|
||||
{"description":"Incorrect doctype without a space before name",
|
||||
"input":"<!DOCTYPEfoo>",
|
||||
"output":["ParseError", ["DOCTYPE", "FOO", true]]},
|
||||
|
||||
{"description":"Bogus doctype",
|
||||
"input":"<!DOCTYPE html PUBLIC \"-//W3C//DTD HTML Transitional 4.01//EN\">",
|
||||
"output":["ParseError", ["DOCTYPE", "HTML", true]]},
|
||||
|
||||
{"description":"Incomplete doctype",
|
||||
"input":"<!DOCTYPE html ",
|
||||
"output":["ParseError", ["DOCTYPE", "HTML", true]]},
|
||||
|
||||
{"description":"Numeric entity representing the NUL character",
|
||||
"input":"�",
|
||||
"output":[["Character", "\uFFFD"]]},
|
||||
|
||||
{"description":"Hexadecimal entity representing the NUL character",
|
||||
"input":"�",
|
||||
"output":[["Character", "\uFFFD"]]},
|
||||
|
||||
{"description":"Numeric entity representing a codepoint after 1114111 (U+10FFFF)",
|
||||
"input":"�",
|
||||
"output":["ParseError", ["Character", "\uFFFD"]]},
|
||||
|
||||
{"description":"Hexadecimal entity representing a codepoint after 1114111 (U+10FFFF)",
|
||||
"input":"�",
|
||||
"output":["ParseError", ["Character", "\uFFFD"]]},
|
||||
|
||||
{"description":"Numeric entity representing a Windows-1252 'codepoint'",
|
||||
"input":"‰",
|
||||
"output":[["Character", "\u2030"]]},
|
||||
|
||||
{"description":"Hexadecimal entity representing a Windows-1252 'codepoint'",
|
||||
"input":"‰",
|
||||
"output":[["Character", "\u2030"]]},
|
||||
|
||||
{"description":"Hexadecimal entity with mixed uppercase and lowercase",
|
||||
"input":"ꯍ",
|
||||
"output":[["Character", "\uABCD"]]},
|
||||
|
||||
{"description":"Entity without a name",
|
||||
"input":"&;",
|
||||
"output":["ParseError", ["Character", "&;"]]},
|
||||
|
||||
{"description":"Unescaped ampersand in attribute value",
|
||||
"input":"<h a='&'>",
|
||||
"output":["ParseError", ["StartTag", "h", { "a":"&" }]]},
|
||||
|
||||
{"description":"StartTag containing <",
|
||||
"input":"<a<b>",
|
||||
"output":["ParseError", ["StartTag", "a", { }], ["StartTag", "b", { }]]},
|
||||
|
||||
{"description":"Non-void element containing trailing /",
|
||||
"input":"<h/>",
|
||||
"output":["ParseError", ["StartTag", "h", { }]]},
|
||||
|
||||
{"description":"Void element with permitted slash",
|
||||
"input":"<br/>",
|
||||
"output":[["StartTag", "br", { }]]},
|
||||
|
||||
{"description":"StartTag containing /",
|
||||
"input":"<h/a='b'>",
|
||||
"output":["ParseError", ["StartTag", "h", { "a":"b" }]]},
|
||||
|
||||
{"description":"Double-quoted attribute value",
|
||||
"input":"<h a=\"b\">",
|
||||
"output":[["StartTag", "h", { "a":"b" }]]},
|
||||
|
||||
{"description":"Unescaped </",
|
||||
"input":"</",
|
||||
"output":["ParseError", ["Character", "</"]]},
|
||||
|
||||
{"description":"Illegal end tag name",
|
||||
"input":"</1>",
|
||||
"output":["ParseError", ["Comment", "1"]]},
|
||||
|
||||
{"description":"Simili processing instruction",
|
||||
"input":"<?namespace>",
|
||||
"output":["ParseError", ["Comment", "?namespace"]]},
|
||||
|
||||
{"description":"A bogus comment stops at >, even if preceeded by two dashes",
|
||||
"input":"<?foo-->",
|
||||
"output":["ParseError", ["Comment", "?foo--"]]},
|
||||
|
||||
{"description":"Unescaped <",
|
||||
"input":"foo < bar",
|
||||
"output":[["Character", "foo "], "ParseError", ["Character", "< bar"]]},
|
||||
|
||||
{"description":"Null Byte Replacement",
|
||||
"input":"\u0000",
|
||||
"output":[["Character", "\ufffd"]]}
|
||||
|
||||
]}
|
||||
|
||||
|
||||
1929
vendor/plugins/HTML5lib/testdata/tree-construction/tests1.dat
vendored
Normal file
1929
vendor/plugins/HTML5lib/testdata/tree-construction/tests1.dat
vendored
Normal file
File diff suppressed because it is too large
Load diff
782
vendor/plugins/HTML5lib/testdata/tree-construction/tests2.dat
vendored
Executable file
782
vendor/plugins/HTML5lib/testdata/tree-construction/tests2.dat
vendored
Executable file
|
|
@ -0,0 +1,782 @@
|
|||
#data
|
||||
<!doctype html>Test
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "Test"
|
||||
|
||||
#data
|
||||
<textarea>test</div>test
|
||||
#errors
|
||||
10: missing document type declaration.
|
||||
17: unescaped '</' in CDATA or RCDATA block.
|
||||
25: unexpected end of file while parsing CDATA section for element textarea.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <textarea>
|
||||
| "test</div>test"
|
||||
|
||||
#data
|
||||
<table><td>
|
||||
#errors
|
||||
7: missing document type declaration.
|
||||
11: required tr element start tag implied by unexpected td element start tag.
|
||||
12: unexpected end of file implied table element end tag.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <table>
|
||||
| <tbody>
|
||||
| <tr>
|
||||
| <td>
|
||||
|
||||
#data
|
||||
<table><td>test</tbody></table>
|
||||
#errors
|
||||
missing document type declarattion
|
||||
Unexpected and of file
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <table>
|
||||
| <tbody>
|
||||
| <tr>
|
||||
| <td>
|
||||
| "test"
|
||||
|
||||
#data
|
||||
<frame>test
|
||||
#errors
|
||||
missing document type declaration
|
||||
frame element can't occur here
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "test"
|
||||
|
||||
#data
|
||||
<!doctype html><frameset>test
|
||||
#errors
|
||||
frameset can't contain text
|
||||
Unexpected end of file
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <frameset>
|
||||
|
||||
#data
|
||||
<!doctype html><frameset><!doctype html>
|
||||
#errors
|
||||
document type declaration can only occur at the start of a document
|
||||
Expected end tag </frameset>
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <frameset>
|
||||
|
||||
#data
|
||||
<!doctype html><font><p><b>test</font>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <font>
|
||||
| <p>
|
||||
| <font>
|
||||
| <b>
|
||||
| "test"
|
||||
|
||||
#data
|
||||
<!DOCTYPE htmL><dt><div><dd>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <dt>
|
||||
| <div>
|
||||
| <dd>
|
||||
|
||||
#data
|
||||
<script></x
|
||||
#errors
|
||||
no document type
|
||||
</ in script
|
||||
Unexpected end of file. Expected </script> end tag.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <script>
|
||||
| "</x"
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<table><plaintext><td>
|
||||
#errors
|
||||
no document type
|
||||
<plaintext> directly inside table
|
||||
Characters inside table.
|
||||
Unexpected end of file.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <plaintext>
|
||||
| "<td>"
|
||||
| <table>
|
||||
|
||||
#data
|
||||
<plaintext></plaintext>
|
||||
#errors
|
||||
No DOCTYPE seen.
|
||||
Unexpected end of file.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <plaintext>
|
||||
| "</plaintext>"
|
||||
|
||||
#data
|
||||
<!doctype html><table><tr>TEST
|
||||
#errors
|
||||
TEST can't occur in <tr>
|
||||
Unexpected end of file.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "TEST"
|
||||
| <table>
|
||||
| <tbody>
|
||||
| <tr>
|
||||
|
||||
#data
|
||||
<!doctype html><body t1=1><body t2=2><body t3=3 t4=4>
|
||||
#errors
|
||||
Unexpected start tag "body"
|
||||
Unexpected start tag "body"
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| t4="4"
|
||||
| t2="2"
|
||||
| t3="3"
|
||||
| t1="1"
|
||||
|
||||
#data
|
||||
</b test
|
||||
#errors
|
||||
Unexpected EOF in attribute
|
||||
Unexpected attribute in end tag.
|
||||
No doctype.
|
||||
Unexpected end tag.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype HtML></b test<b &=&>X
|
||||
#errors
|
||||
Unexpected < in attribute
|
||||
End tag contains attributes.
|
||||
Unexpected end tag.
|
||||
Named entity didn't end with ;
|
||||
Unexpected EOF. Missing closing tag.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <b>
|
||||
| &="&"
|
||||
| "X"
|
||||
|
||||
#data
|
||||
<!doctypehtml><scrIPt type=text/x-foobar;baz>X</SCRipt
|
||||
#errors
|
||||
No space after literal DOCTYPE.
|
||||
Unexpected EOF in (end) tag name
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <script>
|
||||
| type="text/x-foobar;baz"
|
||||
| "X"
|
||||
| <body>
|
||||
|
||||
#data
|
||||
&
|
||||
#errors
|
||||
No doctype.
|
||||
Unfinished entity.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "&"
|
||||
|
||||
#data
|
||||
&#
|
||||
#errors
|
||||
No doctype.
|
||||
Unfinished numeric entity.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "&#"
|
||||
|
||||
#data
|
||||
&#X
|
||||
#errors
|
||||
No doctype.
|
||||
Unfinished hexadecimal entity.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "&#X"
|
||||
|
||||
#data
|
||||
&#x
|
||||
#errors
|
||||
No doctype.
|
||||
Unfinished hexadecimal entity.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "&#x"
|
||||
|
||||
#data
|
||||
-
|
||||
#errors
|
||||
No doctype.
|
||||
Numeric entity didn't end with ;
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "-"
|
||||
|
||||
#data
|
||||
&x-test
|
||||
#errors
|
||||
No doctype.
|
||||
Unfinished named entity.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "&x-test"
|
||||
|
||||
#data
|
||||
<!doctypehtml><p><li>
|
||||
#errors
|
||||
No space after literal DOCTYPE.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <p>
|
||||
| <li>
|
||||
|
||||
#data
|
||||
<!doctypehtml><p><dt>
|
||||
#errors
|
||||
No space after literal DOCTYPE.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <p>
|
||||
| <dt>
|
||||
|
||||
#data
|
||||
<!doctypehtml><p><dd>
|
||||
#errors
|
||||
No space after literal DOCTYPE.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <p>
|
||||
| <dd>
|
||||
|
||||
#data
|
||||
<!doctypehtml><p><form>
|
||||
#errors
|
||||
No space after literal DOCTYPE.
|
||||
Unexpected EOF.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <p>
|
||||
| <form>
|
||||
|
||||
#data
|
||||
<!doctype html><p><b><i><u></p> <p>X
|
||||
#errors
|
||||
Unexpected end tag </p>.
|
||||
Unexpected end EOF. Missing closing tags.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <p>
|
||||
| <b>
|
||||
| <i>
|
||||
| <u>
|
||||
| " "
|
||||
| <p>
|
||||
| <b>
|
||||
| <i>
|
||||
| <u>
|
||||
| "X"
|
||||
|
||||
#data
|
||||
<!doctype html><p></P>X
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <p>
|
||||
| "X"
|
||||
|
||||
#data
|
||||
&
|
||||
#errors
|
||||
No doctype.
|
||||
No closing ; for the entity.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "&"
|
||||
|
||||
#data
|
||||
&AMp;
|
||||
#errors
|
||||
No doctype.
|
||||
Invalid entity.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "&AMp;"
|
||||
|
||||
#data
|
||||
<!doctype html><html><head></head><body><thisISasillyTESTelementNameToMakeSureCrazyTagNamesArePARSEDcorrectLY>
|
||||
#errors
|
||||
Unexpected end of file.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <thisisasillytestelementnametomakesurecrazytagnamesareparsedcorrectly>
|
||||
|
||||
#data
|
||||
<!doctype html>X</body>X
|
||||
#errors
|
||||
Unexpected non-space characters in the after body phase.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "XX"
|
||||
|
||||
#data
|
||||
<!doctype html><!-- X
|
||||
#errors
|
||||
Unexpected end of file in comment.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <!-- X -->
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html><table><caption>test TEST</caption><td>test
|
||||
#errors
|
||||
Unexpected <td> in table body phase.
|
||||
Unexpected end of file.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <table>
|
||||
| <caption>
|
||||
| "test TEST"
|
||||
| <tbody>
|
||||
| <tr>
|
||||
| <td>
|
||||
| "test"
|
||||
|
||||
#data
|
||||
<!doctype html><select><option><optgroup>
|
||||
#errors
|
||||
Unexpected end of file. Missing closing tags.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <select>
|
||||
| <option>
|
||||
| <optgroup>
|
||||
|
||||
#data
|
||||
<!doctype html><select><optgroup><option></optgroup><option><select><option>
|
||||
#errors
|
||||
Unexpected start tag <select> in <select>.
|
||||
Unexpected start tag <option>.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <select>
|
||||
| <optgroup>
|
||||
| <option>
|
||||
| <option>
|
||||
|
||||
#data
|
||||
<!doctype html><select><optgroup><option><optgroup>
|
||||
#errors
|
||||
Unexpected end of file. Missing closing tags.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <select>
|
||||
| <optgroup>
|
||||
| <option>
|
||||
| <optgroup>
|
||||
|
||||
#data
|
||||
<!doctype html><font><input><input></font>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <font>
|
||||
| <input>
|
||||
| <input>
|
||||
|
||||
#data
|
||||
<!DoctypE html><!-- XXX - XXX -->
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <!-- XXX - XXX -->
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!DoctypE html><!-- XXX - XXX
|
||||
#errors
|
||||
Unexpected EOF in comment.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <!-- XXX - XXX -->
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!DoctypE html><!-- XXX - XXX - XXX -->
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <!-- XXX - XXX - XXX -->
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<isindex test=x name=x>
|
||||
#errors
|
||||
No doctype
|
||||
<isindex> is not ok!
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <form>
|
||||
| <hr>
|
||||
| <p>
|
||||
| <label>
|
||||
| "This is a searchable index. Insert your search keywords here:"
|
||||
| <input>
|
||||
| test="x"
|
||||
| name="isindex"
|
||||
| <hr>
|
||||
|
||||
#data
|
||||
test
|
||||
test
|
||||
#errors
|
||||
No doctype
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "test
|
||||
test"
|
||||
|
||||
#data
|
||||
<p><b><i><u></p>
|
||||
<p>X
|
||||
#errors
|
||||
No doctype
|
||||
Unexpected end tag p.
|
||||
Unexpected EOF.
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <p>
|
||||
| <b>
|
||||
| <i>
|
||||
| <u>
|
||||
| "
|
||||
"
|
||||
| <p>
|
||||
| <b>
|
||||
| <i>
|
||||
| <u>
|
||||
| "X"
|
||||
|
||||
#data
|
||||
<!doctype html><body><title>test</body></title>
|
||||
#errors
|
||||
Unexpected start tag that belongs in the head.
|
||||
Expected closing tag after </.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <title>
|
||||
| "test</body>"
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html><body><title>X</title><meta name=z><link rel=foo><style>
|
||||
x { content:"</style" } </style>
|
||||
#errors
|
||||
Unexpected start tag that belongs in head.
|
||||
Unexpected start tag that belongs in head.
|
||||
Unexpected start tag that belongs in head.
|
||||
Expected closing tag after </.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <title>
|
||||
| "X"
|
||||
| <meta>
|
||||
| name="z"
|
||||
| <link>
|
||||
| rel="foo"
|
||||
| <body>
|
||||
| <style>
|
||||
| "
|
||||
x { content:"</style" } "
|
||||
|
||||
#data
|
||||
<!doctype html><select><optgroup></optgroup></select>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <select>
|
||||
| <optgroup>
|
||||
|
||||
#data
|
||||
|
||||
|
||||
#errors
|
||||
No doctype.
|
||||
#document
|
||||
| "
|
||||
"
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html> <html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| " "
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html><script>
|
||||
</script> <title>x</title> </head>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <script>
|
||||
| "
|
||||
"
|
||||
| " "
|
||||
| <title>
|
||||
| "x"
|
||||
| " "
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html><html><body><html id=x>
|
||||
#errors
|
||||
duplicate html start tag
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| id="x"
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html>X</body><html id="x">
|
||||
#errors
|
||||
Unexpected html start tag in the after body phase.
|
||||
html needs to be the first start tag.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| id="x"
|
||||
| <head>
|
||||
| <body>
|
||||
| "X"
|
||||
|
||||
#data
|
||||
<!doctype html><head><html id=x>
|
||||
#errors
|
||||
html start tag too late
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| id="x"
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html>X</html>X
|
||||
#errors
|
||||
Unexpected non-space characters. Expected end of file.
|
||||
Unexpected non-space characters in after body phase. Expected end of file.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "XX"
|
||||
|
||||
#data
|
||||
<!doctype html>X</html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "X "
|
||||
|
||||
#data
|
||||
<!doctype html>X</html><p>X
|
||||
#errors
|
||||
Unexpected start tag <p> in trailing end phase.
|
||||
Unexpected start tag <p> in after body phase.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "X"
|
||||
| <p>
|
||||
| "X"
|
||||
|
||||
#data
|
||||
<!doctype html>X<p/x/y/z>
|
||||
#errors
|
||||
Solidus (/) incorrectly placed.
|
||||
Solidus (/) incorrectly placed.
|
||||
Solidus (/) incorrectly placed.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| "X"
|
||||
| <p>
|
||||
| y=""
|
||||
| x=""
|
||||
| z=""
|
||||
|
||||
#data
|
||||
<!doctype html><!--x--
|
||||
#errors
|
||||
Unexpected end of file in comment.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <!-- x -->
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!doctype html><table><tr><td></p></table>
|
||||
#errors
|
||||
Unexpected </p> end tag.
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <table>
|
||||
| <tbody>
|
||||
| <tr>
|
||||
| <td>
|
||||
210
vendor/plugins/HTML5lib/testdata/tree-construction/tests3.dat
vendored
Normal file
210
vendor/plugins/HTML5lib/testdata/tree-construction/tests3.dat
vendored
Normal file
|
|
@ -0,0 +1,210 @@
|
|||
#data
|
||||
<head></head><style></style>
|
||||
#errors
|
||||
No DOCTYPE
|
||||
<style> in after-head mode
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <style>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<head></head><script></script>
|
||||
#errors
|
||||
No DOCTYPE
|
||||
<script> in after-head mode
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <script>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<head></head><!-- --><style></style><!-- --><script></script>
|
||||
#errors
|
||||
No DOCTYPE
|
||||
<style> in after-head mode
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <style>
|
||||
| <script>
|
||||
| <!-- -->
|
||||
| <!-- -->
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<head></head><!-- -->x<style></style><!-- --><script></script>
|
||||
#errors
|
||||
No DOCTYPE
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <!-- -->
|
||||
| <body>
|
||||
| "x"
|
||||
| <style>
|
||||
| <!-- -->
|
||||
| <script>
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><html><head></head><body><pre>
|
||||
</pre></body></html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <pre>
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><html><head></head><body><pre>
|
||||
|
||||
foo</pre></body></html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <pre>
|
||||
| "foo"
|
||||
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><html><head></head><body><pre>
|
||||
|
||||
foo
|
||||
</pre></body></html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <pre>
|
||||
| "foo
|
||||
"
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><html><head></head><body><pre>x</pre><span>
|
||||
</span></body></html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <pre>
|
||||
| "x"
|
||||
| <span>
|
||||
| "
|
||||
"
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><html><head></head><body><pre>x
|
||||
y</pre></body></html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <pre>
|
||||
| "x
|
||||
y"
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><html><head></head><body><pre>x<div>
|
||||
y</pre></body></html>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <pre>
|
||||
| "x"
|
||||
| <div>
|
||||
| "
|
||||
| y"
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><HTML><META><HEAD></HEAD></HTML>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <meta>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><HTML><HEAD><head></HEAD></HTML>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<textarea>foo<span>bar</span><i>baz
|
||||
#errors
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <textarea>
|
||||
| "foo<span>bar</span><i>baz"
|
||||
|
||||
#data
|
||||
<title>foo<span>bar</em><i>baz
|
||||
#errors
|
||||
#document
|
||||
| <html>
|
||||
| <head>
|
||||
| <title>
|
||||
| "foo<span>bar</em><i>baz"
|
||||
| <body>
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><textarea>
|
||||
</textarea>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <textarea>
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><textarea>
|
||||
|
||||
foo</textarea>
|
||||
#errors
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <textarea>
|
||||
| "foo"
|
||||
|
||||
#data
|
||||
<!DOCTYPE html><html><head></head><body><ul><li><div><p><li></ul></body></html>
|
||||
#errors
|
||||
Missing end tag (div)
|
||||
#document
|
||||
| <!DOCTYPE HTML>
|
||||
| <html>
|
||||
| <head>
|
||||
| <body>
|
||||
| <ul>
|
||||
| <li>
|
||||
| <div>
|
||||
| <p>
|
||||
| <li>
|
||||
44
vendor/plugins/HTML5lib/testdata/tree-construction/tests4.dat
vendored
Normal file
44
vendor/plugins/HTML5lib/testdata/tree-construction/tests4.dat
vendored
Normal file
|
|
@ -0,0 +1,44 @@
|
|||
#data
|
||||
direct div content
|
||||
#errors
|
||||
#document-fragment div
|
||||
| "direct div content"
|
||||
|
||||
#data
|
||||
direct textarea content
|
||||
#errors
|
||||
#document-fragment textarea
|
||||
| "direct textarea content"
|
||||
|
||||
#data
|
||||
textarea content with <em>pseudo</em> <foo>markup
|
||||
#errors
|
||||
#document-fragment textarea
|
||||
| "textarea content with <em>pseudo</em> <foo>markup"
|
||||
|
||||
#data
|
||||
this is CDATA inside a <style> element
|
||||
#errors
|
||||
#document-fragment style
|
||||
| "this is CDATA inside a <style> element"
|
||||
|
||||
#data
|
||||
</plaintext>
|
||||
#errors
|
||||
#document-fragment plaintext
|
||||
| "</plaintext>"
|
||||
|
||||
#data
|
||||
setting html's innerHTML
|
||||
#errors
|
||||
#document-fragment html
|
||||
| <head>
|
||||
| <body>
|
||||
| "setting html's innerHTML"
|
||||
|
||||
#data
|
||||
<title>setting head's innerHTML</title>
|
||||
#errors
|
||||
#document-fragment head
|
||||
| <title>
|
||||
| "setting head's innerHTML"
|
||||
13
vendor/plugins/HTML5lib/tests/preamble.rb
vendored
13
vendor/plugins/HTML5lib/tests/preamble.rb
vendored
|
|
@ -1,17 +1,24 @@
|
|||
require 'test/unit'
|
||||
|
||||
HTML5LIB_BASE = File.dirname(File.dirname(File.dirname(File.expand_path(__FILE__))))
|
||||
HTML5LIB_BASE = File.dirname(File.dirname(File.dirname(File.expand_path(__FILE__))))
|
||||
|
||||
if File.exists?(File.join(HTML5LIB_BASE, 'testdata'))
|
||||
TESTDATA_DIR = File.join(HTML5LIB_BASE, 'testdata')
|
||||
else
|
||||
TESTDATA_DIR = File.join(File.dirname(File.dirname(File.expand_path(__FILE__))), 'testdata')
|
||||
end
|
||||
|
||||
$:.unshift File.join(File.dirname(File.dirname(__FILE__)),'lib')
|
||||
|
||||
$:.unshift File.dirname(__FILE__)
|
||||
|
||||
def html5lib_test_files(subdirectory)
|
||||
Dir[File.join(HTML5LIB_BASE, 'tests', subdirectory, '*.*')]
|
||||
Dir[File.join(TESTDATA_DIR, subdirectory, '*.*')]
|
||||
end
|
||||
|
||||
begin
|
||||
require 'jsonx'
|
||||
require 'rubygems'
|
||||
require 'json'
|
||||
rescue LoadError
|
||||
class JSON
|
||||
def self.parse json
|
||||
|
|
|
|||
11
vendor/plugins/HTML5lib/tests/test_encoding.rb
vendored
11
vendor/plugins/HTML5lib/tests/test_encoding.rb
vendored
|
|
@ -9,13 +9,12 @@ class Html5EncodingTestCase < Test::Unit::TestCase
|
|||
require 'UniversalDetector'
|
||||
|
||||
def test_chardet
|
||||
File.open(File.join(HTML5LIB_BASE, 'tests', 'encoding', 'chardet', 'test_big5.txt')) do |file|
|
||||
stream = HTML5lib::HTMLInputStream.new(file, :chardet => true)
|
||||
assert_equal 'big5', stream.char_encoding.downcase
|
||||
end
|
||||
file = File.open(File.join(TESTDATA_DIR, 'encoding', 'chardet', 'test_big5.txt'), 'r')
|
||||
stream = HTML5lib::HTMLInputStream.new(file, :chardet => true)
|
||||
assert_equal 'big5', stream.char_encoding.downcase
|
||||
rescue LoadError
|
||||
puts "chardet not found, skipping chardet tests"
|
||||
end
|
||||
rescue LoadError
|
||||
puts "chardet not found, skipping chardet tests"
|
||||
end
|
||||
|
||||
html5lib_test_files('encoding').each do |test_file|
|
||||
|
|
|
|||
2
vendor/plugins/HTML5lib/tests/test_parser.rb
vendored
2
vendor/plugins/HTML5lib/tests/test_parser.rb
vendored
|
|
@ -54,7 +54,7 @@ class Html5ParserTestCase < Test::Unit::TestCase
|
|||
actual_errors = parser.errors.map do |(line, col), message|
|
||||
'Line: %i Col: %i %s' % [line, col, message]
|
||||
end
|
||||
assert_equal parser.errors.length, expected_errors.length, [
|
||||
assert_equal expected_errors.length, parser.errors.length, [
|
||||
'Expected errors:', expected_errors.join("\n"),
|
||||
'Actual errors:', actual_errors.join("\n")
|
||||
].join("\n")
|
||||
|
|
|
|||
307
vendor/plugins/HTML5lib/tests/test_sanitizer.rb
vendored
307
vendor/plugins/HTML5lib/tests/test_sanitizer.rb
vendored
|
|
@ -31,14 +31,14 @@ class SanitizeTest < Test::Unit::TestCase
|
|||
:omit_optional_tags => false,
|
||||
:inject_meta_charset => false,
|
||||
:sanitize => true}).gsub(/^<div xmlns='http:\/\/www.w3.org\/1999\/xhtml'>(.*)<\/div>$/, '\1')
|
||||
rescue
|
||||
return "Ill-formed XHTML!"
|
||||
rescue
|
||||
return "Ill-formed XHTML!"
|
||||
end
|
||||
|
||||
def check_sanitization(input, htmloutput, xhtmloutput, rexmloutput)
|
||||
assert_equal htmloutput, sanitize_html(input)
|
||||
assert_equal xhtmloutput, sanitize_xhtml(input)
|
||||
assert_equal rexmloutput, sanitize_rexml(input)
|
||||
assert_equal htmloutput, sanitize_html(input)
|
||||
assert_equal xhtmloutput, sanitize_xhtml(input)
|
||||
assert_equal rexmloutput, sanitize_rexml(input)
|
||||
end
|
||||
|
||||
HTMLSanitizer::ALLOWED_ELEMENTS.each do |tag_name|
|
||||
|
|
@ -113,191 +113,6 @@ class SanitizeTest < Test::Unit::TestCase
|
|||
end
|
||||
end
|
||||
|
||||
def test_should_allow_anchors
|
||||
input = "<a href='foo' onclick='bar'><script>baz</script></a>"
|
||||
output = "<a href='foo'><script>baz</script></a>"
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
# RFC 3986, sec 4.2
|
||||
def test_allow_colons_in_path_component
|
||||
input = "<a href=\"./this:that\">foo</a>"
|
||||
output = "<a href='./this:that'>foo</a>"
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
%w(src width height alt).each do |img_attr|
|
||||
define_method "test_should_allow_image_#{img_attr}_attribute" do
|
||||
input = "<img #{img_attr}='foo' onclick='bar' />"
|
||||
output = "<img #{img_attr}='foo'/>"
|
||||
rexmloutput = "<img #{img_attr}='foo' />"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
end
|
||||
|
||||
def test_should_handle_non_html
|
||||
input = 'abc'
|
||||
output = 'abc'
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_should_handle_blank_text
|
||||
input = ''
|
||||
output = ''
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
[%w(img src), %w(a href)].each do |(tag, attr)|
|
||||
close = VOID_ELEMENTS.include?(tag) ? "/>boo" : ">boo</#{tag}>"
|
||||
xclose = VOID_ELEMENTS.include?(tag) ? " />" : ">boo</#{tag}>"
|
||||
|
||||
input = %(<#{tag} #{attr}="javascript:XSS" title="1">boo</#{tag}>)
|
||||
output = %(<#{tag} title='1'#{close})
|
||||
rexmloutput = %(<#{tag} title='1'#{xclose})
|
||||
define_method "test_should_strip_#{attr}_attribute_in_#{tag}_with_bad_protocols" do
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
define_method "test_should_strip_#{attr}_attribute_in_#{tag}_with_bad_protocols_and_whitespace" do
|
||||
input = %(<#{tag} #{attr}=" javascript:XSS" title="1">boo</#{tag}>)
|
||||
output = %(<#{tag} title='1'#{close})
|
||||
rexmloutput = %(<#{tag} title='1'#{xclose})
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
end
|
||||
|
||||
[%(<img src="javascript:alert('XSS');" />),
|
||||
%(<img src=javascript:alert('XSS') />),
|
||||
%(<img src="JaVaScRiPt:alert('XSS')" />),
|
||||
%(<img src='javascript:alert("XSS")' />),
|
||||
%(<img src='javascript:alert(String.fromCharCode(88,83,83))' />),
|
||||
%(<img src='javascript:alert('XSS')' />),
|
||||
%(<img src='javascript:alert('XSS')' />),
|
||||
%(<img src='javascript:alert('XSS')' />),
|
||||
%(<img src="jav\tascript:alert('XSS');" />),
|
||||
%(<img src="jav	ascript:alert('XSS');" />),
|
||||
%(<img src="jav
ascript:alert('XSS');" />),
|
||||
%(<img src="jav
ascript:alert('XSS');" />),
|
||||
%(<img src="  javascript:alert('XSS');" />),
|
||||
%(<img src=" javascript:alert('XSS');" />),
|
||||
%(<img src=" javascript:alert('XSS');" />)].each_with_index do |img_hack, i|
|
||||
define_method "test_should_not_fall_for_xss_image_hack_#{i}" do
|
||||
output = "<img/>"
|
||||
rexmloutput = "<img />"
|
||||
rexmloutput = "Ill-formed XHTML!" if i == 1
|
||||
check_sanitization(img_hack, output, output, rexmloutput)
|
||||
end
|
||||
end
|
||||
|
||||
def test_should_sanitize_tag_broken_up_by_null
|
||||
input = %(<scr\0ipt>alert(\"XSS\")</scr\0ipt>)
|
||||
output = "<scr\357\277\275ipt>alert(\"XSS\")</scr\357\277\275ipt>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_should_sanitize_invalid_script_tag
|
||||
input = %(<script/XSS SRC="http://ha.ckers.org/xss.js"></script>)
|
||||
output = "<script XSS=\"\" SRC=\"http://ha.ckers.org/xss.js\"></script>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_should_sanitize_script_tag_with_multiple_open_brackets
|
||||
input = %(<<script>alert("XSS");//<</script>)
|
||||
output = "<<script>alert(\"XSS\");//<</script>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
|
||||
input = %(<iframe src=http://ha.ckers.org/scriptlet.html\n<)
|
||||
output = %(<iframe src=\"http://ha.ckers.org/scriptlet.html\"><)
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_should_sanitize_unclosed_script
|
||||
input = %(<script src=http://ha.ckers.org/xss.js?<b>)
|
||||
output = "<script src=\"http://ha.ckers.org/xss.js?\"><b/>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_should_sanitize_half_open_scripts
|
||||
input = %(<img src="javascript:alert('XSS')")
|
||||
output = "<img/>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_should_not_fall_for_ridiculous_hack
|
||||
img_hack = %(<img\nsrc\n=\n"\nj\na\nv\na\ns\nc\nr\ni\np\nt\n:\na\nl\ne\nr\nt\n(\n'\nX\nS\nS\n'\n)\n"\n />)
|
||||
output = "<img/>"
|
||||
rexmloutput = "<img />"
|
||||
check_sanitization(img_hack, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_platypus
|
||||
input = %(<a href="http://www.ragingplatypus.com/" style="display:block; position:absolute; left:0; top:0; width:100%; height:100%; z-index:1; background-color:black; background-image:url(http://www.ragingplatypus.com/i/cam-full.jpg); background-x:center; background-y:center; background-repeat:repeat;">never trust your upstream platypus</a>)
|
||||
output = %(<a href='http://www.ragingplatypus.com/' style='display: block; width: 100%; height: 100%; background-color: black; background-x: center; background-y: center;'>never trust your upstream platypus</a>)
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_xul
|
||||
input = %(<p style="-moz-binding:url('http://ha.ckers.org/xssmoz.xml#xss')">fubar</p>)
|
||||
output = %(<p style=''>fubar</p>)
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_input_image
|
||||
input = %(<input type="image" src="javascript:alert('XSS');" />)
|
||||
output = %(<input type='image'/>)
|
||||
rexmloutput = %(<input type='image' />)
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_non_alpha_non_digit
|
||||
input = %(<script/XSS src="http://ha.ckers.org/xss.js"></script>)
|
||||
output = "<script XSS=\"\" src=\"http://ha.ckers.org/xss.js\"></script>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
|
||||
input = '<a onclick!#$%&()*~+-_.,:;?@[/|\]^`=alert("XSS")>foo</a>'
|
||||
output = "<a>foo</a>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
|
||||
input = '<img/src="http://ha.ckers.org/xss.js"/>'
|
||||
output = "<img src='http://ha.ckers.org/xss.js'/>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_img_dynsrc_lowsrc
|
||||
input = %(<img dynsrc="javascript:alert('XSS')" />)
|
||||
output = "<img/>"
|
||||
rexmloutput = "<img />"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_div_background_image_unicode_encoded
|
||||
input = %(<div style="background-image:\0075\0072\006C\0028'\006a\0061\0076\0061\0073\0063\0072\0069\0070\0074\003a\0061\006c\0065\0072\0074\0028.1027\0058.1053\0053\0027\0029'\0029">foo</div>)
|
||||
output = "<div style=''>foo</div>"
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_div_expression
|
||||
input = %(<div style="width: expression(alert('XSS'));">foo</div>)
|
||||
output = "<div style=''>foo</div>"
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_img_vbscript
|
||||
input = %(<img src='vbscript:msgbox("XSS")' />)
|
||||
output = '<img/>'
|
||||
rexmloutput = '<img />'
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_should_handle_astral_plane_characters
|
||||
input = "<p>𝒵 𝔸</p>"
|
||||
output = "<p>\360\235\222\265 \360\235\224\270</p>"
|
||||
|
|
@ -308,67 +123,6 @@ class SanitizeTest < Test::Unit::TestCase
|
|||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_should_handle_malformed_image_tags
|
||||
input = %(<img """><script>alert("XSS")</script>">)
|
||||
output = "<img/><script>alert(\"XSS\")</script>\">"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_non_alpha_non_digit_II
|
||||
input = %(<a href!#\$%&()*~+-_.,:;?@[/|\]^`=alert('XSS')>foo</a>)
|
||||
output = "<a>foo</a>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_non_alpha_non_digit_III
|
||||
input = %(<a/href="javascript:alert('XSS');">foo</a>)
|
||||
output = "<a>foo</a>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_no_closing_script_tags
|
||||
input = %(<script src=http://ha.ckers.org/xss.js?<b>)
|
||||
output = "<script src=\"http://ha.ckers.org/xss.js?\"><b/>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_protocol_resolution_in_script_tag
|
||||
input = %(<script src=//ha.ckers.org/.j></script>)
|
||||
output = "<script src=\"//ha.ckers.org/.j\"></script>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_double_open_angle_brackets
|
||||
input = %(<img src=http://ha.ckers.org/scriptlet.html <)
|
||||
output = "<img src='http://ha.ckers.org/scriptlet.html'/><"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
|
||||
input = %(<script src=http://ha.ckers.org/scriptlet.html <)
|
||||
output = "<script src=\"http://ha.ckers.org/scriptlet.html\"><"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_background_attribute
|
||||
input = %(<div background="javascript:alert('XSS')"></div>)
|
||||
output = "<div/>"
|
||||
xhtmloutput = "<div></div>"
|
||||
check_sanitization(input, output, xhtmloutput, xhtmloutput)
|
||||
end
|
||||
|
||||
def test_bgsound
|
||||
input = %(<bgsound src="javascript:alert('XSS');" />)
|
||||
output = "<bgsound src=\"javascript:alert('XSS');\"/>"
|
||||
rexmloutput = "<bgsound src=\"javascript:alert('XSS');\"></bgsound>"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
# This affects only NS4. Is it worth fixing?
|
||||
# def test_javascript_includes
|
||||
# input = %(<div size="&{alert('XSS')}">foo</div>)
|
||||
|
|
@ -376,45 +130,16 @@ class SanitizeTest < Test::Unit::TestCase
|
|||
# check_sanitization(input, output, output, output)
|
||||
# end
|
||||
|
||||
def test_link_stylesheets
|
||||
input =%(<link rel="stylesheet" href="javascript:alert('XSS');" />)
|
||||
output = "<link rel=\"stylesheet\" href=\"javascript:alert('XSS');\"/>"
|
||||
rexmloutput = "<link href=\"javascript:alert('XSS');\" rel=\"stylesheet\"/>"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
|
||||
input =%(<link rel="stylesheet" href="http://ha.ckers.org/xss.css" />)
|
||||
output = "<link rel=\"stylesheet\" href=\"http://ha.ckers.org/xss.css\"/>"
|
||||
rexmloutput = "<link href=\"http://ha.ckers.org/xss.css\" rel=\"stylesheet\"/>"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_list_style_image
|
||||
input = %(<li style="list-style-image: url\(javascript:alert\('XSS'\)\)">foo</li>)
|
||||
output = "<li style=''>foo</li>"
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_IE_Comments
|
||||
input = %(<!--[if gte IE 4]><script>alert\('XSS'\);</script><![endif]-->)
|
||||
output = ""
|
||||
check_sanitization(input, output, output, output)
|
||||
|
||||
input = %(<![if !IE 5]><script>alert\('XSS'\);</script><![endif]>)
|
||||
output = "<script>alert('XSS');</script>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
end
|
||||
|
||||
def test_xml_base
|
||||
input = %(<div xml:base="javascript:alert('XSS');//">foo</div>)
|
||||
output = "<div>foo</div>"
|
||||
check_sanitization(input, output, output, output)
|
||||
end
|
||||
|
||||
def test_grave_accents
|
||||
input =%(<img src=`javascript:alert('XSS')` />)
|
||||
output = "<img/>"
|
||||
rexmloutput = "Ill-formed XHTML!"
|
||||
check_sanitization(input, output, output, rexmloutput)
|
||||
html5lib_test_files('sanitizer').each do |filename|
|
||||
JSON::parse(open(filename).read).each do |test|
|
||||
define_method "test_#{test['name']}" do
|
||||
check_sanitization(
|
||||
test['input'],
|
||||
test['output'],
|
||||
test['xhtml'] || test['output'],
|
||||
test['rexml'] || test['output']
|
||||
)
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
|
|||
Loading…
Reference in a new issue