Replace PUA (Private Use Areas) codepoints by an alternative codepoint

The PUA (Private Use Areas) codepoints are not assigned by unicode.
They can be used with an hardcoded font to display custom made glyphs.
This feature is not used in chrome and will lead to fallback fonts.

based on https://chromium-review.googlesource.com/c/chromium/src/+/829378
see http://crbug.com/795225


see: https://en.wikipedia.org/wiki/Private_Use_Areas
Bug: 1018378
Change-Id: I4b218336dba859eda2729bc9668d866e4ec03211
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1887531
Commit-Queue: Etienne Bergeron <etienneb@chromium.org>
Reviewed-by: Alexei Svitkine <asvitkine@chromium.org>
Reviewed-by: Dominik Röttsches <drott@chromium.org>
Cr-Commit-Position: refs/heads/master@{#711257}

ui/gfx/render_text.cc

@@ -22,6 +22,7 @@
 #include "cc/paint/paint_canvas.h"
 #include "cc/paint/paint_shader.h"
 #include "third_party/icu/source/common/unicode/rbbi.h"
+#include "third_party/icu/source/common/unicode/uchar.h"
 #include "third_party/icu/source/common/unicode/utf16.h"
 #include "third_party/skia/include/core/SkDrawLooper.h"
 #include "third_party/skia/include/core/SkFontStyle.h"
@@ -192,25 +193,45 @@ void RestoreBreakList(RenderText* render_text, BreakList<T>* break_list) {
   }
 }
 
-// Replace the unicode control characters ISO 6429 (block C0) by their
-// corresponsing visual symbols. Control chracters can't be displayed but
-// their visual symbols can.
+// Replace the codepoints not handled by RenderText by an other compatible
+// codepoint. Replace the unicode control characters ISO 6429 (block C0) by
+// their corresponding visual symbols. Control characters can't be displayed but
+// their visual symbols can. Replace PUA (Private Use Areas) codepoints with the
+// 'replacement character'.
 void ReplaceControlCharactersWithSymbols(bool multiline, base::string16* text) {
+  // 'REPLACEMENT CHARACTER' used to replace an unknown, unrecognized or
+  // unrepresentable character.
+  constexpr base::char16 kReplacementCodepoint = 0xFFFD;
   // Control Pictures block (see: https://unicode.org/charts/PDF/U2400.pdf).
   constexpr base::char16 kSymbolsCodepoint = 0x2400;
-  for (size_t offset = 0; offset < text->size(); ++offset) {
-    base::char16 control_codepoint = (*text)[offset];
-    if (control_codepoint >= 0 && control_codepoint <= 0x1F) {
+
+  size_t next_offset = 0;
+  while (next_offset < text->length()) {
+    size_t offset = next_offset;
+    UChar32 codepoint;
+    U16_NEXT(text->c_str(), next_offset, text->length(), codepoint);
+
+    if (codepoint >= 0 && codepoint <= 0x1F) {
       // The newline character should be kept as-is when rendertext is
       // multiline.
-      if (control_codepoint == '\n' && multiline)
+      if (codepoint == '\n' && multiline)
         continue;
       // Replace codepoints with their visual symbols, which are at the same
       // offset from kSymbolsCodepoint.
-      (*text)[offset] = kSymbolsCodepoint + control_codepoint;
-    } else if (control_codepoint == 0x7F) {
+      (*text)[offset] = kSymbolsCodepoint + codepoint;
+    } else if (codepoint == 0x7F) {
       // Replace the 'del' codepoint by its symbol (u2421).
       (*text)[offset] = kSymbolsCodepoint + 0x21;
+    } else if (codepoint > 0x7F) {
+      // Private use codepoints are working with a pair of font and codepoint,
+      // but they are not used in Chrome.
+      const int8_t codepoint_category = u_charType(codepoint);
+      if (codepoint_category == U_PRIVATE_USE_CHAR) {
+        (*text)[offset] = kReplacementCodepoint;
+        // We may need to replace the surrogate pair.
+        if (next_offset != offset + 1)
+          (*text)[offset + 1] = kReplacementCodepoint;
+      }
     }
   }
 }

ui/gfx/render_text_unittest.cc

@@ -4530,6 +4530,16 @@ TEST_F(RenderTextTest, ControlCharacterReplacement) {
   EXPECT_EQ(WideToUTF16(L"␈␍␇␉\n␋␌"), render_text->GetDisplayText());
 }
 
+TEST_F(RenderTextTest, PrivateUseCharacterReplacement) {
+  RenderText* render_text = GetRenderText();
+  render_text->SetText(UTF8ToUTF16("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.
+  EXPECT_EQ(WideToUTF16(L"xx\ufffd\ufffda\ufffd\ufffdz"),
+            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) {