Fix bug in encoding of non-BMP characters of URI host names.

Encoded the incorrect code point if the character is was
on an even page. (Which excludes emojis, so likely nobody noticed.)

CoreLibraryReviewExempt: Localized bugfix.
Change-Id: I4dcb04d0c8ea2cb85af4a43e1fbfa8c27e5b00f9
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/395241
Reviewed-by: Nate Bosch <[email protected]>
Commit-Queue: Lasse Nielsen <[email protected]>

Author: lrhn

sdk/lib/core/uri.dart

@@ -2276,7 +2276,7 @@ final class _Uri implements _PlatformUri {
         if ((char & 0xFC00) == 0xD800 && (index + 1) < end) {
           int tail = host.codeUnitAt(index + 1);
           if ((tail & 0xFC00) == 0xDC00) {
-            char = 0x10000 | ((char & 0x3ff) << 10) | (tail & 0x3ff);
+            char = 0x10000 + ((char & 0x3ff) << 10) + (tail & 0x3ff);
             sourceLength = 2;
           }
         }
@@ -2354,7 +2354,7 @@ final class _Uri implements _PlatformUri {
         if ((char & 0xFC00) == 0xD800 && (index + 1) < end) {
           int tail = host.codeUnitAt(index + 1);
           if ((tail & 0xFC00) == 0xDC00) {
-            char = 0x10000 | ((char & 0x3ff) << 10) | (tail & 0x3ff);
+            char = 0x10000 + ((char & 0x3ff) << 10) + (tail & 0x3ff);
             sourceLength = 2;
           }
         }
@@ -2580,7 +2580,7 @@ final class _Uri implements _PlatformUri {
   static String _escapeChar(int char) {
     assert(char <= 0x10ffff); // It's a valid unicode code point.
     List<int> codeUnits;
-    if (char < 0x80) {
+    if (char <= 0x7f) {
       // ASCII, a single percent encoded sequence.
       codeUnits = Uint8List(3);
       codeUnits[0] = _PERCENT;
@@ -2692,7 +2692,7 @@ final class _Uri implements _PlatformUri {
               if ((tail & 0xFC00) == 0xDC00) {
                 // Tail surrogate.
                 sourceLength = 2;
-                char = 0x10000 | ((char & 0x3ff) << 10) | (tail & 0x3ff);
+                char = 0x10000 + ((char & 0x3ff) << 10) + (tail & 0x3ff);
               }
             }
           }

tests/corelib/uri_test.dart

@@ -879,6 +879,28 @@ main() {
   testReplace();
   testPackageUris();
   testBackslashes();
+
+  testNonBmpEncodingRegression();
+}
+
+void testNonBmpEncodingRegression() {
+  // Regression test for bug in encoding of some non-BMP characters
+  // in host names. The failing character has to be one that doesn't have
+  // the 0x10000 bit set in its code point, and which is not in the BMP.
+  const char = "\u{2003E}"; // CJK Unified Ideo­graph.
+  const echar = "%F0%A0%80%BE"; // UTF-8 encoding of page2Char, %-encoded.
+  var nonBmpUri =
+      Uri.parse("http://$char.example.com/x${char}x?y${char}y#z${char}z");
+  Expect.equals("http://$echar.example.com/x${echar}x?y${echar}y#z${echar}z",
+      nonBmpUri.toString());
+  Expect.equals("$echar.example.com", nonBmpUri.host);
+  Expect.equals("$char.example.com", Uri.decodeComponent(nonBmpUri.host));
+  Expect.equals("/x${echar}x", nonBmpUri.path);
+  Expect.equals("/x${char}x", Uri.decodeComponent(nonBmpUri.path));
+  Expect.equals("y${echar}y", nonBmpUri.query);
+  Expect.equals("y${char}y", Uri.decodeComponent(nonBmpUri.query));
+  Expect.equals("z${echar}z", nonBmpUri.fragment);
+  Expect.equals("z${char}z", Uri.decodeComponent(nonBmpUri.fragment));
 }
 
 String dump(Uri uri) {