Fix infinite loop in lexer on malformed XML processing instructions /…

… unparsed
scala-ide · Apr 14, 2011 · 4d1a1a8 · 4d1a1a8
1 parent e3a3ba2
commit 4d1a1a8
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 0 deletions.
diff --git a/CHANGELOG b/CHANGELOG
@@ -6,6 +6,7 @@
 * FIX: incorrect indent behaviour for finally block after a catch block
 * Add SpacesWithinPatternBinders preference (issue #15)
 * Add MultilineScaladocCommentsStartOnFirstLine preference (issue #16)
+* FIX: infinite loop in lexer on malformed XML processing instructions / unparsed
 
 0.0.9 (26/February/11)
 

diff --git a/scalariform/src/main/scala/scalariform/lexer/XmlLexer.scala b/scalariform/src/main/scala/scalariform/lexer/XmlLexer.scala
@@ -244,6 +244,8 @@ trait XmlLexer extends Lexer {
       if (lookaheadIs("</xml:unparsed>")) {
         munch("</xml:unparsed>")
         continue = false
+      } else if (ch == SU) {
+        if (forgiveLexerErrors) continue = false else throw new ScalaLexerException("Malformed Unparsed XML")
       } else
         nextChar()
     }
@@ -257,6 +259,8 @@ trait XmlLexer extends Lexer {
       if (lookaheadIs("?>")) {
         munch("?>")
         continue = false
+      } else if (ch == SU) {
+        if (forgiveLexerErrors) continue = false else throw new ScalaLexerException("Malformed XML processing instruction")
       } else
         nextChar()
     }

diff --git a/scalariform/src/test/scala/scalariform/lexer/ScalaLexerTest.scala b/scalariform/src/test/scala/scalariform/lexer/ScalaLexerTest.scala
@@ -193,6 +193,9 @@ println("foo")""" producesTokens (VARID, LPAREN, STRING_LITERAL, RPAREN, WS, VAR
 
   "Lexer" should "throw a lexer exception" in {
     evaluating { ScalaLexer.rawTokenise("\"\"\"") } should produce[ScalaLexerException]
+    evaluating { ScalaLexer.rawTokenise("<?") } should produce[ScalaLexerException]
+    evaluating { ScalaLexer.rawTokenise("<xml:unparsed>") } should produce[ScalaLexerException]
+
   }
 
   {
@@ -203,6 +206,8 @@ println("foo")""" producesTokens (VARID, LPAREN, STRING_LITERAL, RPAREN, WS, VAR
     "\"unclosed" producesTokens (STRING_LITERAL)
     "\\ufoob" producesTokens (WS)
     "`unclosed" producesTokens (VARID)
+    "<?" producesTokens (XML_PROCESSING_INSTRUCTION)
+    "<xml:unparsed>" producesTokens (XML_UNPARSED)
 
   }