michael@0: // Tests illegal UTF-8 sequences
michael@0: 
michael@0: const Cc = Components.Constructor;
michael@0: const Ci = Components.interfaces;
michael@0: 
michael@0: const tests = [
michael@0: { inStrings: ["%80",                 // Illegal or incomplete sequences
michael@0:               "%8f",
michael@0:               "%90",
michael@0:               "%9f",
michael@0:               "%a0",
michael@0:               "%bf",
michael@0:               "%c0",
michael@0:               "%c1",
michael@0:               "%c2",
michael@0:               "%df",
michael@0:               "%e0",
michael@0:               "%e0%a0",
michael@0:               "%e0%bf",
michael@0:               "%ed%80",
michael@0:               "%ed%9f",
michael@0:               "%ef",
michael@0:               "%ef%bf",
michael@0:               "%f0",
michael@0:               "%f0%90",
michael@0:               "%f0%90%80",
michael@0:               "%f0%90%bf",
michael@0:               "%f0%bf",
michael@0:               "%f0%bf%80",
michael@0:               "%f0%bf%bf",
michael@0:               "%f4",
michael@0:               "%f4%80",
michael@0:               "%f4%80%80",
michael@0:               "%f4%80%bf",
michael@0:               "%f4%8f",
michael@0:               "%f4%8f%80",
michael@0:               "%f4%8f%bf",
michael@0:               "%f5",
michael@0:               "%f7",
michael@0:               "%f8",
michael@0:               "%fb",
michael@0:               "%fc",
michael@0:               "%fd"],
michael@0:   expected: "ABC\ufffdXYZ" },
michael@0: 
michael@0: { inStrings: ["%c0%af",              // Illegal bytes in 2-octet
michael@0:               "%c1%af"],             //  sequences
michael@0:   expected: "ABC\ufffd\ufffdXYZ" },
michael@0: 
michael@0: { inStrings: ["%e0%80%80",           // Illegal bytes in 3-octet
michael@0:               "%e0%80%af",           //  sequences
michael@0:               "%e0%9f%bf",
michael@0:                                      // long surrogates
michael@0:               "%ed%a0%80",           // D800
michael@0:               "%ed%ad%bf",           // DB7F
michael@0:               "%ed%ae%80",           // DB80
michael@0:               "%ed%af%bf",           // DBFF
michael@0:               "%ed%b0%80",           // DC00
michael@0:               "%ed%be%80",           // DF80
michael@0:               "%ed%bf%bf"],          // DFFF
michael@0:   expected: "ABC\ufffd\ufffd\ufffdXYZ" },
michael@0: 
michael@0: { inStrings: ["%f0%80%80%80",        // Illegal bytes in 4-octet
michael@0:               "%f0%80%80%af",        //  sequences
michael@0:               "%f0%8f%bf%bf",
michael@0:               "%f4%90%80%80",
michael@0:               "%f4%bf%bf%bf",
michael@0:               "%f5%80%80%80",
michael@0:               "%f7%bf%bf%bf"],
michael@0:   expected: "ABC\ufffd\ufffd\ufffd\ufffdXYZ" },
michael@0: 
michael@0: { inStrings: ["%f8%80%80%80%80",     // Illegal bytes in 5-octet
michael@0:               "%f8%80%80%80%af",     //  sequences
michael@0:               "%fb%bf%bf%bf%bf"],
michael@0:   expected: "ABC\ufffd\ufffd\ufffd\ufffd\ufffdXYZ" },
michael@0: 
michael@0:                                      // Surrogate pairs
michael@0: { inStrings: ["%ed%a0%80%ed%b0%80",  // D800 DC00
michael@0:               "%ed%a0%80%ed%bf%bf",  // D800 DFFF
michael@0:               "%ed%ad%bf%ed%b0%80",  // DB7F DC00
michael@0:               "%ed%ad%bf%ed%bf%bf",  // DB7F DFFF
michael@0:               "%ed%ae%80%ed%b0%80",  // DB80 DC00
michael@0:               "%ed%ae%80%ed%bf%bf",  // DB80 DFFF
michael@0:               "%ed%af%bf%ed%b0%80",  // DBFF DC00
michael@0:               "%ed%ad%bf%ed%bf%bf",  // DBFF DFFF
michael@0:               "%fc%80%80%80%80%80",  // Illegal bytes in 6-octet
michael@0:               "%fc%80%80%80%80%af",  //  sequences
michael@0:               "%fd%bf%bf%bf%bf%bf"],
michael@0:   expected: "ABC\ufffd\ufffd\ufffd\ufffd\ufffd\ufffdXYZ" },
michael@0: ];
michael@0: 
michael@0: 
michael@0: function testCaseInputStream(inStr, expected)
michael@0: {
michael@0:   var dataURI = "data:text/plain; charset=UTF-8,ABC" + inStr + "XYZ"
michael@0:   dump(inStr + "==>");
michael@0: 
michael@0:   var IOService = Cc("@mozilla.org/network/io-service;1",
michael@0: 		     "nsIIOService");
michael@0:   var ConverterInputStream =
michael@0:       Cc("@mozilla.org/intl/converter-input-stream;1",
michael@0: 	 "nsIConverterInputStream",
michael@0: 	 "init");
michael@0: 
michael@0:   var ios = new IOService();
michael@0:   var channel = ios.newChannel(dataURI, "", null);
michael@0:   var testInputStream = channel.open();
michael@0:   var testConverter = new ConverterInputStream(testInputStream,
michael@0: 					       "UTF-8",
michael@0: 					       16,
michael@0: 					       0xFFFD);
michael@0: 
michael@0:   if (!(testConverter instanceof Ci.nsIUnicharLineInputStream))
michael@0:       throw "not line input stream";
michael@0: 
michael@0:   var outStr = "";
michael@0:   var more;
michael@0:   do {
michael@0:       // read the line and check for eof
michael@0:       var line = {};
michael@0:       more = testConverter.readLine(line);
michael@0:       outStr += line.value;
michael@0:   } while (more);
michael@0: 
michael@0:   dump(outStr + "; expected=" + expected + "\n");
michael@0:   do_check_eq(outStr, expected);
michael@0:   do_check_eq(outStr.length, expected.length);
michael@0: }
michael@0: 
michael@0: function run_test() {
michael@0:   for (var t of tests) {
michael@0:     for (var inStr of t.inStrings) {
michael@0:       testCaseInputStream(inStr, t.expected);
michael@0:     }
michael@0:   }
michael@0: }