New, more validating, parser for URI.

tests/corelib/uri_test.dart

Index: tests/corelib/uri_test.dart
diff --git a/tests/corelib/uri_test.dart b/tests/corelib/uri_test.dart
index 19d4019a875629fdd1feaadf7364630b37806561..6ae45ea4ffa6712723a2596a391c9d238d72b997 100644
--- a/tests/corelib/uri_test.dart
+++ b/tests/corelib/uri_test.dart
@@ -124,6 +124,199 @@ void testResolvePath(String expected, String path) {
       Uri.parse("http://localhost").resolveUri(new Uri(path: path)).toString());
 }
 
+const ALPHA = r"abcdefghijklmnopqrstuvwxuzABCDEFGHIJKLMNOPQRSTUVWXUZ";
+const DIGIT = r"0123456789";
+const PERCENT_ENCODED = "%00%ff";
+const SUBDELIM = r"!$&'()*+,;=";
+
+const SCHEMECHAR = "$ALPHA$DIGIT+-.";
+const UNRESERVED = "$ALPHA$DIGIT-._~";
+const REGNAMECHAR = "$UNRESERVED$SUBDELIM$PERCENT_ENCODED";
+const USERINFOCHAR = "$REGNAMECHAR:";
+
+const PCHAR_NC = "$UNRESERVED$SUBDELIM$PERCENT_ENCODED@";
+const PCHAR = "$PCHAR_NC:";
+const QUERYCHAR = "$PCHAR/?";
+
+void testValidCharacters() {
+  // test that all valid characters are accepted.
+
+  for (var scheme in ["", "$SCHEMECHAR$SCHEMECHAR:"]) {
+    for (var userinfo in ["", "@", "$USERINFOCHAR$USERINFOCHAR@",
+                          "$USERINFOCHAR:$DIGIT@"]) {
+      for (var host in ["", "$REGNAMECHAR$REGNAMECHAR",
+                        "255.255.255.256",  // valid reg-name.
+                        "[ffff::ffff:ffff]", "[ffff::255.255.255.255]"]) {
+        for (var port in ["", ":$DIGIT$DIGIT"]) {
+          var auth = "$userinfo$host$port";
+          if (auth.isNotEmpty) auth = "//$auth";
+          var paths = ["", "/", "/$PCHAR", "/$PCHAR/"];  // Absolute or empty.
+          if (auth.isNotEmpty) {
+            // Initial segment may be empty.
+            paths..add("//$PCHAR");
+          } else {
+            // Path may begin with non-slash.
+            if (scheme.isEmpty) {
+              // Initial segment must not contain colon.
+              paths..add(PCHAR_NC)
+                   ..add("$PCHAR_NC/$PCHAR")
+                   ..add("$PCHAR_NC/$PCHAR/");
+            } else {
+              paths..add(PCHAR)
+                   ..add("$PCHAR/$PCHAR")
+                   ..add("$PCHAR/$PCHAR/");
+            }
+          }
+          for (var path in paths) {
+            for (var query in ["", "?", "?$QUERYCHAR"]) {
+              for (var fragment in ["", "#", "#$QUERYCHAR"]) {
+                var uri = "$scheme$auth$path$query$fragment";
+                // Should not throw.
+                var result = Uri.parse(uri);
+              }
+            }
+          }
+        }
+      }
+    }
+  }
+}
+
+void testInvalidUrls() {
+  void checkInvalid(uri) {
+    try {
+      var result = Uri.parse(uri);
+      Expect.fail("Invalid URI `$uri` parsed to $result\n"
+                  "  Scheme:    ${result.scheme}\n"
+                  "  User-info: ${result.userInfo}\n"
+                  "  Host:      ${result.host}\n"
+                  "  Port:      ${result.port}\n"
+                  "  Path:      ${result.path}\n"
+                  "  Query:     ${result.query}\n"
+                  "  Fragment:  ${result.fragment}\n");
+    } on FormatException {
+      // Success.
+    }
+  }
+
+  // Regression test for http://dartbug.com/16081
+  checkInvalid("http://www.example.org/red%09ros\u00E9#red");
+  checkInvalid("http://r\u00E9sum\u00E9.example.org");
+
+  // Invalid characters. The characters must be rejected, even if normalizing
+  // the input would cause them to be valid (normalization happens after
+  // validation).
+  var invalidChars = [
+    "\xe7",      // Arbitrary non-ASCII letter
+    " ",         // Space, not allowed anywhere.
+    '"',         // Quote, not allowed anywhere
+    "\x7f",      // DEL, not allowed anywhere
+    "\xdf",      // German lower-case scharf-S. Becomes ASCII when upper-cased.
+    "\u0130"     // Latin capital dotted I, becomes ASCII lower-case in Turkish.
+    "%\uFB03",   // % + Ligature ffi, becomes ASCII when upper-cased,
+                 // should not be read as "%FFI".
+    "\u212a",    // Kelvin sign. Becomes ASCII when lower-cased.
+    "%1g",       // Invalid escape.
+  ];
+  for (var invalid in invalidChars) {
+    checkInvalid("A${invalid}b:///");
+    checkInvalid("${invalid}b:///");
+    checkInvalid("s://user${invalid}info@x.x/");
+    checkInvalid("s://reg${invalid}name/");
+    checkInvalid("s://regname:12${invalid}45/");
+    checkInvalid("s://regname/p${invalid}ath/");
+    checkInvalid("/p${invalid}ath/");
+    checkInvalid("p${invalid}ath/");
+    checkInvalid("s://regname/path/?x${invalid}x");
+    checkInvalid("s://regname/path/#x${invalid}x");
+    checkInvalid("s://regname/path/?#x${invalid}x");
+  }
+  checkInvalid("s%41://x.x/");      // No escapes in scheme,
+                                    // and no colon before slash in path.
+  checkInvalid("1a://x.x/");        // Scheme must start with letter,
+                                    // and no colon before slash in path.
+  checkInvalid(".a://x.x/");        // Scheme must start with letter,
+                                    // and no colon before slash in path.
+  checkInvalid("_:");               // Character not valid in scheme,
+                                    // and no colon before slash in path.
+  checkInvalid(":");                // Scheme must start with letter,
+                                    // and no colon before slash in path.
+  checkInvalid("s://x@x@x.x/");     // At most one @ in userinfo.
+  checkInvalid("s://x@x:x/");       // No colon in host except before a port.
+  checkInvalid("s://x@x:9:9/");     // At most one port.
+  checkInvalid("s://x/x#foo#bar");  // At most one #.
+  checkInvalid("s://:/");           // Colon in host requires port,
+                                    // and path may not start with //.
+  checkInvalid("s@://x:9/x?x#x");   // @ not allowed in scheme.
+}
+
+void testNormalization() {
+  // The Uri constructor and the Uri.parse function performs RFC-3986
+  // syntax based normalization.
+
+  var uri;
+
+  // Scheme: Only case normalization. Schemes cannot contain escapes.
+  uri = Uri.parse("A:");
+  Expect.equals("a", uri.scheme);
+  uri = Uri.parse("Z:");
+  Expect.equals("z", uri.scheme);
+  uri = Uri.parse("$SCHEMECHAR:");
+  Expect.equals(SCHEMECHAR.toLowerCase(), uri.scheme);
+
+  // Percent escape normalization.
+  // Escapes of unreserved characters are converted to the character,
+  // subject to case normalization in reg-name.
+  for (var i = 0; i < UNRESERVED.length; i++) {
+    var char = UNRESERVED[i];
+    var escape = "%" + char.codeUnitAt(0).toRadixString(16);  // all > 0xf.
+
+    uri = Uri.parse("s://xX${escape}xX@yY${escape}yY/zZ${escape}zZ"
+                    "?vV${escape}vV#wW${escape}wW");
+    Expect.equals("xX${char}xX", uri.userInfo);
+    Expect.equals("yY${char}yY".toLowerCase(), uri.host);
+    Expect.equals("/zZ${char}zZ", uri.path);
+    Expect.equals("vV${char}vV", uri.query);
+    Expect.equals("wW${char}wW", uri.fragment);
+  }
+
+  // Escapes of reserved characters are kept, but upper-cased.
+  for (var escape in ["%00", "%1f", "%7F", "%fF"]) {
+    uri = Uri.parse("s://xX${escape}xX@yY${escape}yY/zZ${escape}zZ"
+                    "?vV${escape}vV#wW${escape}wW");
+    var normalizedEscape = escape.toUpperCase();
+    Expect.equals("xX${normalizedEscape}xX", uri.userInfo);
+    Expect.equals("yy${normalizedEscape}yy", uri.host);
+    Expect.equals("/zZ${normalizedEscape}zZ", uri.path);
+    Expect.equals("vV${normalizedEscape}vV", uri.query);
+    Expect.equals("wW${normalizedEscape}wW", uri.fragment);
+  }
+
+  // Some host normalization edge cases.
+  uri = Uri.parse("x://x%61X%41x%41X%61x/");
+  Expect.equals("xaxaxaxax", uri.host);
+
+  uri = Uri.parse("x://Xxxxxxxx/");
+  Expect.equals("xxxxxxxx", uri.host);
+
+  uri = Uri.parse("x://xxxxxxxX/");
+  Expect.equals("xxxxxxxx", uri.host);
+
+  uri = Uri.parse("x://xxxxxxxx%61/");
+  Expect.equals("xxxxxxxxa", uri.host);
+
+  uri = Uri.parse("x://%61xxxxxxxx/");
+  Expect.equals("axxxxxxxx", uri.host);
+
+  uri = Uri.parse("x://X/");
+  Expect.equals("x", uri.host);
+
+  uri = Uri.parse("x://%61/");
+  Expect.equals("a", uri.host);
+
+  // TODO(lrn): Also do path normalization: /./ -> / and /x/../ -> /
+}
+
 main() {
   testUri("http:", true);
   testUri("file://", true);
@@ -150,7 +343,6 @@ main() {
                           query: null,
                           fragment: null).toString());
   Expect.stringEquals("file://", Uri.parse("file:").toString());
-
   testResolvePath("/a/g", "/a/b/c/./../../g");
   testResolvePath("/a/g", "/a/b/c/./../../g");
   testResolvePath("/mid/6", "mid/content=5/../6");
@@ -261,4 +453,8 @@ main() {
   Expect.throws(
       () => Uri.parse("file://user@password:host/path"),
       (e) => e is FormatException);
+
+  testValidCharacters();
+  testInvalidUrls();
+  testNormalization();
 }