Updating tokeniser tests.

Achal-Aggarwal · Mar 11, 2014 · 544a307 · 544a307
1 parent 2d1fe2e
commit 544a307
Show file tree

Hide file tree

Showing 10 changed files with 2,231 additions and 2,672 deletions.
diff --git a/test/data/tokeniser2/INDEX b/test/data/tokeniser2/INDEX
@@ -10,6 +10,6 @@ contentModelFlags.test	html5lib content model tests
 entities.test		html5lib entity tests
 escapeFlag.test		html5lib escape flag tests
 numericEntities.test	html5lib numeric entities tests
-#unicodeChars.test	html5lib unicode character tests
+unicodeChars.test	html5lib unicode character tests
 cdata.test		CDATA section tests
 regression.test		Regression tests
diff --git a/test/data/tokeniser2/contentModelFlags.test b/test/data/tokeniser2/contentModelFlags.test
@@ -1,73 +1,73 @@
 {"tests": [
 
 {"description":"PLAINTEXT content model flag",
-"contentModelFlags":["PLAINTEXT"],
+"initialStates":["PLAINTEXT state"],
 "lastStartTag":"plaintext",
 "input":"<head>&body;",
 "output":[["Character", "<head>&body;"]]},
 
-{"description":"End tag closing RCDATA or CDATA",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag closing RCDATA or RAWTEXT",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"foo</xmp>",
 "output":[["Character", "foo"], ["EndTag", "xmp"]]},
 
-{"description":"End tag closing RCDATA or CDATA (case-insensitivity)",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag closing RCDATA or RAWTEXT (case-insensitivity)",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"foo</xMp>",
 "output":[["Character", "foo"], ["EndTag", "xmp"]]},
 
-{"description":"End tag closing RCDATA or CDATA (ending with space)",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag closing RCDATA or RAWTEXT (ending with space)",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"foo</xmp ",
-"output":[["Character", "foo"], "ParseError", ["EndTag", "xmp"]]},
+"output":[["Character", "foo"], "ParseError"]},
 
-{"description":"End tag closing RCDATA or CDATA (ending with EOF)",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag closing RCDATA or RAWTEXT (ending with EOF)",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"foo</xmp",
-"output":[["Character", "foo"], "ParseError", ["EndTag", "xmp"]]},
+"output":[["Character", "foo</xmp"]]},
 
-{"description":"End tag closing RCDATA or CDATA (ending with slash)",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag closing RCDATA or RAWTEXT (ending with slash)",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"foo</xmp/",
-"output":[["Character", "foo"], "ParseError", ["EndTag", "xmp"]]},
+"output":[["Character", "foo"], "ParseError"]},
 
-{"description":"End tag not closing RCDATA or CDATA (ending with left-angle-bracket)",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag not closing RCDATA or RAWTEXT (ending with left-angle-bracket)",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"foo</xmp<",
 "output":[["Character", "foo</xmp<"]]},
 
-{"description":"End tag with incorrect name in RCDATA or CDATA",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag with incorrect name in RCDATA or RAWTEXT",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"</foo>bar</xmp>",
 "output":[["Character", "</foo>bar"], ["EndTag", "xmp"]]},
 
-{"description":"End tag with incorrect name in RCDATA or CDATA (starting like correct name)",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag with incorrect name in RCDATA or RAWTEXT (starting like correct name)",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"</foo>bar</xmpaar>",
 "output":[["Character", "</foo>bar</xmpaar>"]]},
 
-{"description":"End tag closing RCDATA or CDATA, switching back to PCDATA",
-"contentModelFlags":["RCDATA", "CDATA"],
+{"description":"End tag closing RCDATA or RAWTEXT, switching back to PCDATA",
+"initialStates":["RCDATA state", "RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"foo</xmp></baz>",
 "output":[["Character", "foo"], ["EndTag", "xmp"], ["EndTag", "baz"]]},
 
-{"description":"CDATA w/ something looking like an entity",
-"contentModelFlags":["CDATA"],
+{"description":"RAWTEXT w/ something looking like an entity",
+"initialStates":["RAWTEXT state"],
 "lastStartTag":"xmp",
 "input":"&foo;",
 "output":[["Character", "&foo;"]]},
 
 {"description":"RCDATA w/ an entity",
-"contentModelFlags":["RCDATA"],
+"initialStates":["RCDATA state"],
 "lastStartTag":"textarea",
 "input":"&lt;",
 "output":[["Character", "<"]]}