From e724c8e6ff1c4ef8d9f4ed6556c0749e326c4a3d Mon Sep 17 00:00:00 2001
From: Jonathan Kew <jkew@mozilla.com>
Date: Mon, 21 Oct 2024 10:32:41 +0000
Subject: [PATCH] Bug 1925854 - Make gfxScriptItemizer support both 16-bit and
 8-bit text buffers. r=gfx-reviewers,lsalzman

This does not change behavior yet, just prepares for the next patch
which will actually use the 8-bit version.

Differential Revision: https://phabricator.services.mozilla.com/D226286
---
 gfx/thebes/gfxScriptItemizer.cpp |  7 +++++--
 gfx/thebes/gfxScriptItemizer.h   | 25 +++++++++++++++++++++----
 gfx/thebes/gfxTextRun.cpp        |  3 ++-
 3 files changed, 28 insertions(+), 7 deletions(-)

diff --git a/gfx/thebes/gfxScriptItemizer.cpp b/gfx/thebes/gfxScriptItemizer.cpp
index ce64692d40c9..3c6abdc6701c 100644
--- a/gfx/thebes/gfxScriptItemizer.cpp
+++ b/gfx/thebes/gfxScriptItemizer.cpp
@@ -122,6 +122,9 @@ static inline bool SameScript(Script runScript, Script currCharScript,
 }
 
 gfxScriptItemizer::Run gfxScriptItemizer::Next() {
+  MOZ_ASSERT(textLength == 0 || (textIs8bit && textPtr._1b) ||
+             (!textIs8bit && textPtr._2b));
+
   /* if we've fallen off the end of the text, we're done */
   if (scriptLimit >= textLength) {
     return Run{};
@@ -136,11 +139,11 @@ gfxScriptItemizer::Run gfxScriptItemizer::Next() {
     Script sc;
     uint32_t startOfChar = scriptLimit;
 
-    ch = textPtr[scriptLimit];
+    ch = textIs8bit ? textPtr._1b[scriptLimit] : textPtr._2b[scriptLimit];
 
     /* decode UTF-16 (may be surrogate pair) */
     if (NS_IS_HIGH_SURROGATE(ch) && scriptLimit < textLength - 1) {
-      uint32_t low = textPtr[scriptLimit + 1];
+      uint32_t low = textPtr._2b[scriptLimit + 1];
       if (NS_IS_LOW_SURROGATE(low)) {
         ch = SURROGATE_TO_UCS4(ch, low);
         scriptLimit += 1;
diff --git a/gfx/thebes/gfxScriptItemizer.h b/gfx/thebes/gfxScriptItemizer.h
index 7417b0bf1981..1518655c0c26 100644
--- a/gfx/thebes/gfxScriptItemizer.h
+++ b/gfx/thebes/gfxScriptItemizer.h
@@ -60,8 +60,21 @@ class gfxScriptItemizer {
  public:
   using Script = mozilla::intl::Script;
 
-  gfxScriptItemizer(const char16_t* aText, uint32_t aLength)
-      : textPtr(aText), textLength(aLength) {}
+  gfxScriptItemizer() = default;
+  gfxScriptItemizer(const gfxScriptItemizer& aOther) = delete;
+  gfxScriptItemizer(gfxScriptItemizer&& aOther) = delete;
+
+  void SetText(const char16_t* aText, uint32_t aLength) {
+    textPtr._2b = aText;
+    textLength = aLength;
+    textIs8bit = false;
+  }
+
+  void SetText(const unsigned char* aText, uint32_t aLength) {
+    textPtr._1b = aText;
+    textLength = aLength;
+    textIs8bit = true;
+  }
 
   struct Run {
     uint32_t mOffset = 0;
@@ -83,8 +96,12 @@ class gfxScriptItemizer {
     Script scriptCode;
   };
 
-  const char16_t* const textPtr;
-  const uint32_t textLength;
+  union {
+    const char16_t* _2b;
+    const unsigned char* _1b;
+  } textPtr;
+  uint32_t textLength;
+  bool textIs8bit;
 
   uint32_t scriptStart = 0;
   uint32_t scriptLimit = 0;
diff --git a/gfx/thebes/gfxTextRun.cpp b/gfx/thebes/gfxTextRun.cpp
index d772145bdccd..b263d3e84418 100644
--- a/gfx/thebes/gfxTextRun.cpp
+++ b/gfx/thebes/gfxTextRun.cpp
@@ -2642,7 +2642,8 @@ void gfxFontGroup::InitTextRun(DrawTarget* aDrawTarget, gfxTextRun* aTextRun,
 
       // split into script runs so that script can potentially influence
       // the font matching process below
-      gfxScriptItemizer scriptRuns(textPtr, aLength);
+      gfxScriptItemizer scriptRuns;
+      scriptRuns.SetText(textPtr, aLength);
 
       while (gfxScriptItemizer::Run run = scriptRuns.Next()) {
         if (MOZ_UNLIKELY(MOZ_LOG_TEST(log, LogLevel::Warning))) {