Fix handling of UTF8_STRING

Currently we use a default MIME type of US-ASCII when the type string isn't a
MIME type with a charset attribute. This means that unicode characters sent
as type UTF8_STRING get mangled during conversion. Fix this by special casing
UTF8_STRING in mime_utils.cc

R=oshima@chromium.org

Bug: 994470
Change-Id: I292bc42bce528449ebb1b9e8f5b72c9ea0e3968d
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1757261
Reviewed-by: Mitsuru Oshima <oshima@chromium.org>
Commit-Queue: Fergus Dall <sidereal@google.com>
Auto-Submit: Fergus Dall <sidereal@google.com>
Cr-Commit-Position: refs/heads/master@{#687946}

components/exo/data_source.cc

@@ -32,8 +32,6 @@ constexpr char kTextHTML[] = "text/html";
 constexpr char kUtfPrefix[] = "UTF";
 constexpr char kEncoding16[] = "16";
 constexpr char kEncodingASCII[] = "ASCII";
-constexpr char kEncodingUTF8Legacy[] = "UTF8_STRING";
-constexpr char kEncodingUTF8Charset[] = "UTF-8";
 
 constexpr char kUTF16Unspecified[] = "UTF-16";
 constexpr char kUTF16LittleEndian[] = "UTF-16LE";
@@ -250,11 +248,7 @@ void DataSource::GetDataForPreferredMimeTypes(
         continue;
 
       std::string charset;
-      // We special case UTF8_STRING to provide minimal handling of X11 apps.
-      if (mime_type == kEncodingUTF8Legacy)
-        charset = kEncodingUTF8Charset;
-      else
-        charset = GetCharset(mime_type);
+      charset = GetCharset(mime_type);
       int new_rank = GetCharsetRank(charset);
       if (new_rank < text_rank) {
         text_mime = mime_type;

components/exo/mime_utils.cc

@@ -8,12 +8,17 @@ namespace {
 
 constexpr char kCharset[] = ";charset=";
 constexpr char kDefaultCharset[] = "US-ASCII";
+constexpr char kEncodingUTF8Charset[] = "UTF-8";
 
 }  // namespace
 
 namespace exo {
 
 std::string GetCharset(const std::string& mime_type) {
+  // We special case UTF8_STRING to provide minimal handling of X11 apps.
+  if (mime_type == kEncodingUTF8Legacy)
+    return std::string(kEncodingUTF8Charset);
+
   auto pos = mime_type.find(kCharset);
   if (pos == std::string::npos)
     return std::string(kDefaultCharset);

components/exo/mime_utils.h

@@ -9,6 +9,8 @@
 
 namespace exo {
 
+constexpr char kEncodingUTF8Legacy[] = "UTF8_STRING";
+
 // Takes a text/* mime type and returns the name of the character set specified
 // in the type. If no character set is specified, defaults to US-ASCII.
 std::string GetCharset(const std::string& mime_type);

components/exo/mime_utils_unittest.cc

@@ -10,6 +10,12 @@ namespace exo {
 namespace {
 using MimeUtilsTest = testing::Test;
 
+TEST_F(MimeUtilsTest, LegacyString) {
+  std::string mime_type("UTF8_STRING");
+  std::string expected("UTF-8");
+  EXPECT_EQ(GetCharset(mime_type), expected);
+}
+
 TEST_F(MimeUtilsTest, CharsetNotPresent) {
   std::string mime_type("text/plain");
   std::string expected("US-ASCII");

components/exo/seat_unittest.cc

@@ -144,6 +144,32 @@ TEST_F(SeatTest, SetSelectionTextUTF8) {
   EXPECT_EQ(clipboard, converted_data);
 }
 
+TEST_F(SeatTest, SetSelectionTextUTF8Legacy) {
+  Seat seat;
+
+  // UTF8 encoded data
+  const uint8_t data[] = {
+      0xe2, 0x9d, 0x84,       // SNOWFLAKE
+      0xf0, 0x9f, 0x94, 0xa5  // FIRE
+  };
+  base::string16 converted_data;
+  EXPECT_TRUE(base::UTF8ToUTF16(reinterpret_cast<const char*>(data),
+                                sizeof(data), &converted_data));
+
+  TestDataSourceDelegate delegate;
+  DataSource source(&delegate);
+  source.Offer("UTF8_STRING");
+  delegate.SetData(std::vector<uint8_t>(data, data + sizeof(data)));
+  seat.SetSelection(&source);
+
+  RunReadingTask();
+
+  base::string16 clipboard;
+  ui::Clipboard::GetForCurrentThread()->ReadText(ui::ClipboardType::kCopyPaste,
+                                                 &clipboard);
+  EXPECT_EQ(clipboard, converted_data);
+}
+
 TEST_F(SeatTest, SetSelectionTextUTF16LE) {
   Seat seat;