Rewrite invalid surrogate and non-character codepoints

This CL is writing the invalid codepoint by the replacement
characters.

See https://unicode.org/reports/tr44/
"""
  Noncharacter_Code_Point	: Code points permanently reserved for internal use.
"""


Bug: 1011818
Change-Id: Id39b42f52a8c36887fabcd81e794f46ebedec089
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1902772Reviewed-by: Alexei Svitkine <asvitkine@chromium.org>
Reviewed-by: Robert Liao <robliao@chromium.org>
Commit-Queue: Etienne Bergeron <etienneb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#713486}

ui/gfx/render_text.cc

@@ -333,6 +333,13 @@ void ReplaceControlCharactersWithSymbols(bool multiline, base::string16* text) {
     } else if (codepoint == 0x7F) {
       // Replace the 'del' codepoint by its symbol (u2421).
       (*text)[offset] = kSymbolsCodepoint + 0x21;
+    } else if (!U_IS_UNICODE_CHAR(codepoint)) {
+      // Unicode codepoint that can't be assigned a character. This handles:
+      // - single surrogate codepoints,
+      // - last two codepoints on each plane,
+      // - invalid characters (e.g. u+fdd0..u+fdef)
+      // - codepoints above u+10ffff
+      ReplaceCodepointAtIndex(offset, kReplacementCodepoint, text);
     } else if (codepoint > 0x7F) {
       // Private use codepoints are working with a pair of font and codepoint,
       // but they are not used in Chrome.

ui/gfx/render_text_unittest.cc

@@ -863,14 +863,15 @@ TEST_F(RenderTextTest, RevealObscuredText) {
   render_text->SetText(UTF8ToUTF16("new longer"));
   EXPECT_EQ(GetObscuredString(10), render_text->GetDisplayText());
 
-  // Text with invalid surrogates.
+  // Text with invalid surrogates (surrogates low 0xDC00 and high 0xD800).
+  // Invalid surrogates are replaced by replacement character (e.g. 0xFFFD).
   const base::char16 invalid_surrogates[] = {0xDC00, 0xD800, 'h', 'o', 'p', 0};
   render_text->SetText(invalid_surrogates);
   EXPECT_EQ(GetObscuredString(5), render_text->GetDisplayText());
   render_text->RenderText::SetObscuredRevealIndex(0);
-  EXPECT_EQ(GetObscuredString(5, 0, 0xDC00), render_text->GetDisplayText());
+  EXPECT_EQ(GetObscuredString(5, 0, 0xFFFD), render_text->GetDisplayText());
   render_text->RenderText::SetObscuredRevealIndex(1);
-  EXPECT_EQ(GetObscuredString(5, 1, 0xD800), render_text->GetDisplayText());
+  EXPECT_EQ(GetObscuredString(5, 1, 0xFFFD), render_text->GetDisplayText());
   render_text->RenderText::SetObscuredRevealIndex(2);
   EXPECT_EQ(GetObscuredString(5, 2, 'h'), render_text->GetDisplayText());
 
@@ -4785,7 +4786,7 @@ TEST_F(RenderTextTest, ControlCharacterReplacement) {
 
 TEST_F(RenderTextTest, PrivateUseCharacterReplacement) {
   RenderText* render_text = GetRenderText();
-  render_text->SetText(UTF8ToUTF16("xx\ue78d\ue78fa\U00100042z"));
+  render_text->SetText(WideToUTF16(L"xx\ue78d\ue78fa\U00100042z"));
 
   // The private use characters should have been replaced. If the code point is
   // a surrogate pair, it needs to be replaced by two characters.
@@ -4793,6 +4794,13 @@ TEST_F(RenderTextTest, PrivateUseCharacterReplacement) {
             render_text->GetDisplayText());
 }
 
+TEST_F(RenderTextTest, InvalidSurrogateCharacterReplacement) {
+  // Text with invalid surrogates (surrogates low 0xDC00 and high 0xD800).
+  RenderText* render_text = GetRenderText();
+  render_text->SetText(WideToUTF16(L"\xDC00\xD800"));
+  EXPECT_EQ(WideToUTF16(L"\ufffd\ufffd"), render_text->GetDisplayText());
+}
+
 // Make sure the horizontal positions of runs in a line (left-to-right for
 // LTR languages and right-to-left for RTL languages).
 TEST_F(RenderTextTest, HarfBuzz_HorizontalPositions) {