Merge pull request #24 from HTMLParseErrorWG/review-fixes2

Character error fixes
inikulin · May 31, 2017 · 8e457b9 · 8e457b9
2 parents 8eb950a + 5d2483f
commit 8e457b9
Show file tree

Hide file tree

Showing 4 changed files with 16 additions and 14 deletions.
diff --git a/lib/common/error_codes.js b/lib/common/error_codes.js
@@ -2,8 +2,8 @@
 
 module.exports = {
     controlCharacterInInputStream: 'control-character-in-input-stream',
-    undefinedCharacterInInputStream: 'undefined-character-in-input-stream',
-    nonUnicodeCharacterInInputStream: 'non-unicode-character-in-input-stream',
+    noncharacterInInputStream: 'noncharacter-in-input-stream',
+    surrogateInInputStream: 'surrogate-in-input-stream',
     nonVoidHtmlElementStartTagWithTrailingSolidus: 'non-void-html-element-start-tag-with-trailing-solidus',
     endTagWithAttributes: 'end-tag-with-attributes',
     endTagWithTrailingSolidus: 'end-tag-with-trailing-solidus',
@@ -42,9 +42,10 @@ module.exports = {
     eofInCdata: 'eof-in-cdata',
     absenceOfDigitsInNumericCharacterReference: 'absence-of-digits-in-numeric-character-reference',
     nullCharacterReference: 'null-character-reference',
-    nonUnicodeCharacterReference: 'non-unicode-character-reference',
+    surrogateCharacterReference: 'surrogate-character-reference',
+    characterReferenceOutsideUnicodeRange: 'character-reference-outside-unicode-range',
     controlCharacterReference: 'control-character-reference',
-    undefinedCharacterReference: 'undefined-character-reference',
+    noncharacterCharacterReference: 'noncharacter-character-reference',
     missingWhitespaceBeforeDoctypeName: 'missing-whitespace-before-doctype-name',
     missingDoctypeName: 'missing-doctype-name',
     invalidCharacterSequenceAfterDoctypeName: 'invalid-character-sequence-after-doctype-name',

diff --git a/lib/common/unicode.js b/lib/common/unicode.js
@@ -55,7 +55,7 @@ exports.CODE_POINT_SEQUENCES = {
 
 
 //Surrogates
-var isSurrogate = exports.isSurrogate = function (cp) {
+exports.isSurrogate = function (cp) {
     return cp >= 0xD800 && cp <= 0xDFFF;
 };
 
@@ -77,7 +77,3 @@ exports.isControlCodePoint = function (cp) {
 exports.isUndefinedCodePoint = function (cp) {
     return cp >= 0xFDD0 && cp <= 0xFDEF || UNDEFINED_CODE_POINTS.indexOf(cp) > -1;
 };
-
-exports.isNonUnicodeCodePoint = function (cp) {
-    return isSurrogate(cp) || cp > 0x10FFFF;
-};
diff --git a/lib/tokenizer/index.js b/lib/tokenizer/index.js
@@ -2515,13 +2515,18 @@ _[NUMERIC_CHARACTER_REFERENCE_END_STATE] = function numericCharacterReferenceEnd
         this.charRefCode = $.REPLACEMENT_CHARACTER;
     }
 
-    else if (unicode.isNonUnicodeCodePoint(this.charRefCode)) {
-        this._err(ERR.nonUnicodeCharacterReference);
+    else if (this.charRefCode > 0x10FFFF) {
+        this._err(ERR.characterReferenceOutsideUnicodeRange);
+        this.charRefCode = $.REPLACEMENT_CHARACTER;
+    }
+
+    else if (unicode.isSurrogate(this.charRefCode)) {
+        this._err(ERR.surrogateCharacterReference);
         this.charRefCode = $.REPLACEMENT_CHARACTER;
     }
 
     else if (unicode.isUndefinedCodePoint(this.charRefCode))
-        this._err(ERR.undefinedCharacterReference);
+        this._err(ERR.noncharacterCharacterReference);
 
     else if (unicode.isControlCodePoint(this.charRefCode) || this.charRefCode === $.CARRIAGE_RETURN) {
         this._err(ERR.controlCharacterReference);

diff --git a/lib/tokenizer/preprocessor.js b/lib/tokenizer/preprocessor.js
@@ -60,7 +60,7 @@ Preprocessor.prototype._processSurrogate = function (cp) {
     }
 
     //NOTE: isolated surrogate
-    this._err(ERR.nonUnicodeCharacterInInputStream);
+    this._err(ERR.surrogateInInputStream);
 
     return cp;
 };
@@ -141,7 +141,7 @@ Preprocessor.prototype._checkForProblematicCharacters = function (cp) {
         this._err(ERR.controlCharacterInInputStream);
 
     else if (unicode.isUndefinedCodePoint(cp))
-        this._err(ERR.undefinedCharacterInInputStream);
+        this._err(ERR.noncharacterInInputStream);
 };
 
 Preprocessor.prototype.retreat = function () {