test/data/tokenizer/test2.test - html - Git at Google

 {"tests": [

 {"description":"DOCTYPE without name",
 "input":"<!DOCTYPE>",
 "output":["ParseError", "ParseError", ["DOCTYPE", "", null, null, false]]},

 {"description":"DOCTYPE without space before name",
 "input":"<!DOCTYPEhtml>",
 "output":["ParseError", ["DOCTYPE", "html", null, null, true]]},

 {"description":"Incorrect DOCTYPE without a space before name",
 "input":"<!DOCTYPEfoo>",
 "output":["ParseError", ["DOCTYPE", "foo", null, null, true]]},

 {"description":"DOCTYPE with publicId",
 "input":"<!DOCTYPE html PUBLIC \"-//W3C//DTD HTML Transitional 4.01//EN\">",
 "output":[["DOCTYPE", "html", "-//W3C//DTD HTML Transitional 4.01//EN", null, true]]},

 {"description":"DOCTYPE with EOF after PUBLIC",
 "input":"<!DOCTYPE html PUBLIC",
 "output":["ParseError", ["DOCTYPE", "html", null, null, false]]},

 {"description":"DOCTYPE with EOF after PUBLIC '",
 "input":"<!DOCTYPE html PUBLIC '",
 "output":["ParseError", ["DOCTYPE", "html", "", null, false]]},

 {"description":"DOCTYPE with EOF after PUBLIC 'x",
 "input":"<!DOCTYPE html PUBLIC 'x",
 "output":["ParseError", ["DOCTYPE", "html", "x", null, false]]},

 {"description":"DOCTYPE with systemId",
 "input":"<!DOCTYPE html SYSTEM \"-//W3C//DTD HTML Transitional 4.01//EN\">",
 "output":[["DOCTYPE", "html", null, "-//W3C//DTD HTML Transitional 4.01//EN", true]]},

 {"description":"DOCTYPE with publicId and systemId",
 "input":"<!DOCTYPE html PUBLIC \"-//W3C//DTD HTML Transitional 4.01//EN\" \"-//W3C//DTD HTML Transitional 4.01//EN\">",
 "output":[["DOCTYPE", "html", "-//W3C//DTD HTML Transitional 4.01//EN", "-//W3C//DTD HTML Transitional 4.01//EN", true]]},

 {"description":"DOCTYPE with > in double-quoted publicId",
 "input":"<!DOCTYPE html PUBLIC \">x",
 "output":["ParseError", ["DOCTYPE", "html", "", null, false], ["Character", "x"]]},

 {"description":"DOCTYPE with > in single-quoted publicId",
 "input":"<!DOCTYPE html PUBLIC '>x",
 "output":["ParseError", ["DOCTYPE", "html", "", null, false], ["Character", "x"]]},

 {"description":"DOCTYPE with > in double-quoted systemId",
 "input":"<!DOCTYPE html PUBLIC \"foo\" \">x",
 "output":["ParseError", ["DOCTYPE", "html", "foo", "", false], ["Character", "x"]]},

 {"description":"DOCTYPE with > in single-quoted systemId",
 "input":"<!DOCTYPE html PUBLIC 'foo' '>x",
 "output":["ParseError", ["DOCTYPE", "html", "foo", "", false], ["Character", "x"]]},

 {"description":"Incomplete doctype",
 "input":"<!DOCTYPE html ",
 "output":["ParseError", ["DOCTYPE", "html", null, null, false]]},

 {"description":"Numeric entity representing the NUL character",
 "input":"&#0000;",
 "output":["ParseError", ["Character", "\uFFFD"]]},

 {"description":"Hexadecimal entity representing the NUL character",
 "input":"&#x0000;",
 "output":["ParseError", ["Character", "\uFFFD"]]},

 {"description":"Numeric entity representing a codepoint after 1114111 (U+10FFFF)",
 "input":"&#2225222;",
 "output":["ParseError", ["Character", "\uFFFD"]]},

 {"description":"Hexadecimal entity representing a codepoint after 1114111 (U+10FFFF)",
 "input":"&#x1010FFFF;",
 "output":["ParseError", ["Character", "\uFFFD"]]},

 {"description":"Hexadecimal entity pair representing a surrogate pair",
 "input":"&#xD869;&#xDED6;",
 "output":["ParseError", ["Character", "\uFFFD"], "ParseError", ["Character", "\uFFFD"]]},

 {"description":"Hexadecimal entity with mixed uppercase and lowercase",
 "input":"&#xaBcD;",
 "output":[["Character", "\uABCD"]]},

 {"description":"Entity without a name",
 "input":"&;",
 "output":["ParseError", ["Character", "&;"]]},

 {"description":"Unescaped ampersand in attribute value",
 "input":"<h a='&'>",
 "output":[["StartTag", "h", { "a":"&" }]]},

 {"description":"StartTag containing <",
 "input":"<a<b>",
 "output":[["StartTag", "a<b", { }]]},

 {"description":"Non-void element containing trailing /",
 "input":"<h/>",
 "output":[["StartTag","h",{},true]]},

 {"description":"Void element with permitted slash",
 "input":"<br/>",
 "output":[["StartTag","br",{},true]]},

 {"description":"Void element with permitted slash (with attribute)",
 "input":"<br foo='bar'/>",
 "output":[["StartTag","br",{"foo":"bar"},true]]},

 {"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":["ParseError", ["Character", "\u0000"]]},

 {"description":"Comment with dash",
 "input":"<!---x",
 "output":["ParseError", ["Comment", "-x"]]},

 {"description":"Entity + newline",
 "input":"\nx\n&gt;\n",
 "output":[["Character","\nx\n>\n"]]},

 {"description":"Start tag with no attributes but space before the greater-than sign",
 "input":"<h >",
 "output":[["StartTag", "h", {}]]},

 {"description":"Empty attribute followed by uppercase attribute",
 "input":"<h a B=''>",
 "output":[["StartTag", "h", {"a":"", "b":""}]]},

 {"description":"Double-quote after attribute name",
 "input":"<h a \">",
 "output":["ParseError", ["StartTag", "h", {"a":"", "\"":""}]]},

 {"description":"Single-quote after attribute name",
 "input":"<h a '>",
 "output":["ParseError", ["StartTag", "h", {"a":"", "'":""}]]},

 {"description":"Empty end tag with following characters",
 "input":"a</>bc",
 "output":[["Character", "a"], "ParseError", ["Character", "bc"]]},

 {"description":"Empty end tag with following tag",
 "input":"a</><b>c",
 "output":[["Character", "a"], "ParseError", ["StartTag", "b", {}], ["Character", "c"]]},

 {"description":"Empty end tag with following comment",
 "input":"a</><!--b-->c",
 "output":[["Character", "a"], "ParseError", ["Comment", "b"], ["Character", "c"]]},

 {"description":"Empty end tag with following end tag",
 "input":"a</></b>c",
 "output":[["Character", "a"], "ParseError", ["EndTag", "b"], ["Character", "c"]]}

 ]}
	{"tests": [

	{"description":"DOCTYPE without name",
	"input":"<!DOCTYPE>",
	"output":["ParseError", "ParseError", ["DOCTYPE", "", null, null, false]]},

	{"description":"DOCTYPE without space before name",
	"input":"<!DOCTYPEhtml>",
	"output":["ParseError", ["DOCTYPE", "html", null, null, true]]},

	{"description":"Incorrect DOCTYPE without a space before name",
	"input":"<!DOCTYPEfoo>",
	"output":["ParseError", ["DOCTYPE", "foo", null, null, true]]},

	{"description":"DOCTYPE with publicId",
	"input":"<!DOCTYPE html PUBLIC \"-//W3C//DTD HTML Transitional 4.01//EN\">",
	"output":[["DOCTYPE", "html", "-//W3C//DTD HTML Transitional 4.01//EN", null, true]]},

	{"description":"DOCTYPE with EOF after PUBLIC",
	"input":"<!DOCTYPE html PUBLIC",
	"output":["ParseError", ["DOCTYPE", "html", null, null, false]]},

	{"description":"DOCTYPE with EOF after PUBLIC '",
	"input":"<!DOCTYPE html PUBLIC '",
	"output":["ParseError", ["DOCTYPE", "html", "", null, false]]},

	{"description":"DOCTYPE with EOF after PUBLIC 'x",
	"input":"<!DOCTYPE html PUBLIC 'x",
	"output":["ParseError", ["DOCTYPE", "html", "x", null, false]]},

	{"description":"DOCTYPE with systemId",
	"input":"<!DOCTYPE html SYSTEM \"-//W3C//DTD HTML Transitional 4.01//EN\">",
	"output":[["DOCTYPE", "html", null, "-//W3C//DTD HTML Transitional 4.01//EN", true]]},

	{"description":"DOCTYPE with publicId and systemId",
	"input":"<!DOCTYPE html PUBLIC \"-//W3C//DTD HTML Transitional 4.01//EN\" \"-//W3C//DTD HTML Transitional 4.01//EN\">",
	"output":[["DOCTYPE", "html", "-//W3C//DTD HTML Transitional 4.01//EN", "-//W3C//DTD HTML Transitional 4.01//EN", true]]},

	{"description":"DOCTYPE with > in double-quoted publicId",
	"input":"<!DOCTYPE html PUBLIC \">x",
	"output":["ParseError", ["DOCTYPE", "html", "", null, false], ["Character", "x"]]},

	{"description":"DOCTYPE with > in single-quoted publicId",
	"input":"<!DOCTYPE html PUBLIC '>x",
	"output":["ParseError", ["DOCTYPE", "html", "", null, false], ["Character", "x"]]},

	{"description":"DOCTYPE with > in double-quoted systemId",
	"input":"<!DOCTYPE html PUBLIC \"foo\" \">x",
	"output":["ParseError", ["DOCTYPE", "html", "foo", "", false], ["Character", "x"]]},

	{"description":"DOCTYPE with > in single-quoted systemId",
	"input":"<!DOCTYPE html PUBLIC 'foo' '>x",
	"output":["ParseError", ["DOCTYPE", "html", "foo", "", false], ["Character", "x"]]},

	{"description":"Incomplete doctype",
	"input":"<!DOCTYPE html ",
	"output":["ParseError", ["DOCTYPE", "html", null, null, false]]},

	{"description":"Numeric entity representing the NUL character",
	"input":"",
	"output":["ParseError", ["Character", "\uFFFD"]]},

	{"description":"Hexadecimal entity representing the NUL character",
	"input":"",
	"output":["ParseError", ["Character", "\uFFFD"]]},

	{"description":"Numeric entity representing a codepoint after 1114111 (U+10FFFF)",
	"input":"&#2225222;",
	"output":["ParseError", ["Character", "\uFFFD"]]},

	{"description":"Hexadecimal entity representing a codepoint after 1114111 (U+10FFFF)",
	"input":"&#x1010FFFF;",
	"output":["ParseError", ["Character", "\uFFFD"]]},

	{"description":"Hexadecimal entity pair representing a surrogate pair",
	"input":"&#xD869;&#xDED6;",
	"output":["ParseError", ["Character", "\uFFFD"], "ParseError", ["Character", "\uFFFD"]]},

	{"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":[["StartTag", "h", { "a":"&" }]]},

	{"description":"StartTag containing <",
	"input":"<a<b>",
	"output":[["StartTag", "a<b", { }]]},

	{"description":"Non-void element containing trailing /",
	"input":"<h/>",
	"output":[["StartTag","h",{},true]]},

	{"description":"Void element with permitted slash",
	"input":"<br/>",
	"output":[["StartTag","br",{},true]]},

	{"description":"Void element with permitted slash (with attribute)",
	"input":"<br foo='bar'/>",
	"output":[["StartTag","br",{"foo":"bar"},true]]},

	{"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":["ParseError", ["Character", "\u0000"]]},

	{"description":"Comment with dash",
	"input":"<!---x",
	"output":["ParseError", ["Comment", "-x"]]},

	{"description":"Entity + newline",
	"input":"\nx\n>\n",
	"output":[["Character","\nx\n>\n"]]},

	{"description":"Start tag with no attributes but space before the greater-than sign",
	"input":"<h >",
	"output":[["StartTag", "h", {}]]},

	{"description":"Empty attribute followed by uppercase attribute",
	"input":"<h a B=''>",
	"output":[["StartTag", "h", {"a":"", "b":""}]]},

	{"description":"Double-quote after attribute name",
	"input":"<h a \">",
	"output":["ParseError", ["StartTag", "h", {"a":"", "\"":""}]]},

	{"description":"Single-quote after attribute name",
	"input":"<h a '>",
	"output":["ParseError", ["StartTag", "h", {"a":"", "'":""}]]},

	{"description":"Empty end tag with following characters",
	"input":"a</>bc",
	"output":[["Character", "a"], "ParseError", ["Character", "bc"]]},

	{"description":"Empty end tag with following tag",
	"input":"a</><b>c",
	"output":[["Character", "a"], "ParseError", ["StartTag", "b", {}], ["Character", "c"]]},

	{"description":"Empty end tag with following comment",
	"input":"a</><!--b-->c",
	"output":[["Character", "a"], "ParseError", ["Comment", "b"], ["Character", "c"]]},

	{"description":"Empty end tag with following end tag",
	"input":"a</></b>c",
	"output":[["Character", "a"], "ParseError", ["EndTag", "b"], ["Character", "c"]]}

	]}