Merge pull request #28 from HTMLParseErrorWG/null-bogus-comment

Add null character error reporting for bogus doctype and comment per current spec changes
HTMLParseErrorWG · Aug 22, 2017 · 0b8f660 · 0b8f660
2 parents 2d6e16b + b1f9308
commit 0b8f660
Showing 1 changed file with 11 additions and 6 deletions.
diff --git a/lib/tokenizer/index.js b/lib/tokenizer/index.js
@@ -1592,8 +1592,10 @@ _[BOGUS_COMMENT_STATE] = function bogusCommentState(cp) {
         this._emitEOFToken();
     }
 
-    else if (cp === $.NULL)
+    else if (cp === $.NULL) {
+        this._err(ERR.unexpectedNullCharacter);
         this.currentToken.data += unicode.REPLACEMENT_CHARACTER;
+    }
 
     else
         this.currentToken.data += toChar(cp);
@@ -1943,7 +1945,7 @@ _[AFTER_DOCTYPE_NAME_STATE] = function afterDoctypeNameState(cp) {
     else if (!this._ensureHibernation()) {
         this._err(ERR.invalidCharacterSequenceAfterDoctypeName);
         this.currentToken.forceQuirks = true;
-        this.state = BOGUS_DOCTYPE_STATE;
+        this._reconsumeInState(BOGUS_DOCTYPE_STATE);
     }
 };
 
@@ -1983,7 +1985,7 @@ _[AFTER_DOCTYPE_PUBLIC_KEYWORD_STATE] = function afterDoctypePublicKeywordState(
     else {
         this._err(ERR.missingQuoteBeforeDoctypePublicIdentifier);
         this.currentToken.forceQuirks = true;
-        this.state = BOGUS_DOCTYPE_STATE;
+        this._reconsumeInState(BOGUS_DOCTYPE_STATE);
     }
 };
 
@@ -2118,7 +2120,7 @@ _[AFTER_DOCTYPE_PUBLIC_IDENTIFIER_STATE] = function afterDoctypePublicIdentifier
     else {
         this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
         this.currentToken.forceQuirks = true;
-        this.state = BOGUS_DOCTYPE_STATE;
+        this._reconsumeInState(BOGUS_DOCTYPE_STATE);
     }
 };
 
@@ -2194,7 +2196,7 @@ _[AFTER_DOCTYPE_SYSTEM_KEYWORD_STATE] = function afterDoctypeSystemKeywordState(
     else {
         this._err(ERR.missingQuoteBeforeDoctypeSystemIdentifier);
         this.currentToken.forceQuirks = true;
-        this.state = BOGUS_DOCTYPE_STATE;
+        this._reconsumeInState(BOGUS_DOCTYPE_STATE);
     }
 };
 
@@ -2317,7 +2319,7 @@ _[AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE] = function afterDoctypeSystemIdentifier
 
     else {
         this._err(ERR.unexpectedCharacterAfterDoctypeSystemIdentifier);
-        this.state = BOGUS_DOCTYPE_STATE;
+        this._reconsumeInState(BOGUS_DOCTYPE_STATE);
     }
 };
 
@@ -2330,6 +2332,9 @@ _[BOGUS_DOCTYPE_STATE] = function bogusDoctypeState(cp) {
         this.state = DATA_STATE;
     }
 
+    else if (cp === $.NULL)
+        this._err(ERR.unexpectedNullCharacter);
+
     else if (cp === $.EOF) {
         this._emitCurrentToken();
         this._emitEOFToken();