Broken HTML generated by FortranXref

Special characters are not always escaped. (Since the Fortran
analyzer is used for .inc files, which is a pretty generic
extension, it is often used for files which are not really Fortran
files. The broken HTML was seen when using FortranXref to analyze
PHP files.)

Also, if the file does not have a final newline, the last <span>
element won't have an end tag. Use yypush()/yypop() to ensure all
<span> elements are closed.

Author: kahatlen

src/org/opensolaris/opengrok/analysis/fortran/FortranXref.lex

@@ -63,7 +63,11 @@ Number = ([0-9]+\.[0-9]+|[0-9][0-9]*|"0x" [0-9a-fA-F]+ )([udl]+)?
 %%
 <YYINITIAL>{
  ^{Label} { out.write("<span class=\"n\">"); out.write(yytext()); out.write("</span>"); }
- ^[^ \t\f\r\n]+ { String commentStr = yytext(); yybegin(LCOMMENT);out.write("<span class=\"c\">"+commentStr);}
+ ^[^ \t\f\r\n]+ {
+    yypush(LCOMMENT, "</span>");
+    out.write("<span class=\"c\">");
+    Util.htmlize(yytext(), out);
+}
 
 {Identifier} {
     String id = yytext();
@@ -85,14 +89,14 @@ Number = ([0-9]+\.[0-9]+|[0-9][0-9]*|"0x" [0-9a-fA-F]+ )([udl]+)?
 */
 {Number}        { out.write("<span class=\"n\">"); out.write(yytext()); out.write("</span>"); }
 
- \"     { yybegin(STRING);out.write("<span class=\"s\">\"");}
- \'     { yybegin(QSTRING);out.write("<span class=\"s\">\'");}
- \!     { yybegin(SCOMMENT);out.write("<span class=\"c\">!");}
+ \"     { yypush(STRING, "</span>"); out.write("<span class=\"s\">\"");}
+ \'     { yypush(QSTRING, "</span>"); out.write("<span class=\"s\">\'");}
+ \!     { yypush(SCOMMENT, "</span>"); out.write("<span class=\"c\">!");}
 }
 
 <STRING> {
  \" {WhiteSpace} \"  { out.write(yytext());}
- \"     { yybegin(YYINITIAL); out.write("\"</span>"); }
+ \"     { out.write('"'); yypop(); }
  \\\\   { out.write("\\\\"); }
  \\\"   { out.write("\\\""); }
 }
@@ -101,23 +105,23 @@ Number = ([0-9]+\.[0-9]+|[0-9][0-9]*|"0x" [0-9a-fA-F]+ )([udl]+)?
  "\\\\" { out.write("\\\\"); }
  "\\'" { out.write("\\\'"); }
  \' {WhiteSpace} \' { out.write(yytext()); }
- \'     { yybegin(YYINITIAL); out.write("'</span>"); }
+ \'     { out.write('\''); yypop(); }
 }
 
 <COMMENT> {
-"*/"    { yybegin(YYINITIAL); out.write("*/</span>"); }
+"*/"    { out.write("*/"); yypop(); }
 }
 
 <SCOMMENT> {
-{WhiteSpace}*{EOL}      { yybegin(YYINITIAL); out.write("</span>");
+{WhiteSpace}*{EOL}      { yypop();
                   startNewLine();}
 }
 
 <LCOMMENT> {
 "&"     {out.write( "&amp;");}
 "<"     {out.write( "&lt;");}
 ">"     {out.write( "&gt;");}
-{WhiteSpace}*{EOL}      { yybegin(YYINITIAL); out.write("</span>");
+{WhiteSpace}*{EOL}      { yypop();
                   startNewLine();}
  {WhiteSpace}   { out.write(yytext()); }
  [!-~]  { out.write(yycharat(0)); }

test/org/opensolaris/opengrok/analysis/JFlexXrefTest.java

@@ -480,4 +480,17 @@ public void testJavaClassAnalyzer() throws Exception {
         DocumentBuilderFactory.newInstance().newDocumentBuilder().parse(
                 new InputSource(new StringReader("<doc>" + out + "</doc>")));
     }
+
+    /**
+     * Test that special characters in Fortran files are escaped.
+     */
+    @Test
+    public void testFortranSpecialCharacters() throws Exception {
+        FortranXref xref = new FortranXref(new StringReader("<?php?>"));
+        StringWriter out = new StringWriter();
+        xref.write(out);
+        // Used to throw SAXParseException.
+        DocumentBuilderFactory.newInstance().newDocumentBuilder().parse(
+            new InputSource(new StringReader("<doc>" + out + "</doc>")));
+    }
 }