Encode unicodes in UTF-16BE in ToUnicode map

Bug: pdfium:667
Change-Id: I811571c334ff28162905a65781ca14f03caf2966
Reviewed-on: https://pdfium-review.googlesource.com/4910
Commit-Queue: Nicolás Peña <npm@chromium.org>
Reviewed-by: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>

3 files changed, 81 insertions, 0 deletions

diff --git a/core/fxcrt/fx_extension.cpp b/core/fxcrt/fx_extension.cpp
index 209584b68d..2b290ed79d 100644
--- a/core/fxcrt/fx_extension.cpp
+++ b/core/fxcrt/fx_extension.cpp
@@ -137,6 +137,31 @@ uint32_t FX_HashCode_GetW(const CFX_WideStringC& str, bool bIgnoreCase) {
   return dwHashCode;
 }
 
+void FXSYS_IntToTwoHexChars(uint8_t n, char* buf) {
+  static const char kHex[] = "0123456789ABCDEF";
+  buf[0] = kHex[n / 16];
+  buf[1] = kHex[n % 16];
+}
+
+void FXSYS_IntToFourHexChars(uint16_t n, char* buf) {
+  FXSYS_IntToTwoHexChars(n / 256, buf);
+  FXSYS_IntToTwoHexChars(n % 256, buf + 2);
+}
+
+size_t FXSYS_ToUTF16BE(uint32_t unicode, char* buf) {
+  ASSERT(unicode <= 0xD7FF || (unicode > 0xDFFF && unicode <= 0x10FFFF));
+  if (unicode <= 0xFFFF) {
+    FXSYS_IntToFourHexChars(unicode, buf);
+    return 4;
+  }
+  unicode -= 0x010000;
+  // High ten bits plus 0xD800
+  FXSYS_IntToFourHexChars(0xD800 + unicode / 0x400, buf);
+  // Low ten bits plus 0xDC00
+  FXSYS_IntToFourHexChars(0xDC00 + unicode % 0x400, buf + 4);
+  return 8;
+}
+
 void* FX_Random_MT_Start(uint32_t dwSeed) {
   FX_MTRANDOMCONTEXT* pContext = FX_Alloc(FX_MTRANDOMCONTEXT, 1);
   pContext->mt[0] = dwSeed;
diff --git a/core/fxcrt/fx_extension.h b/core/fxcrt/fx_extension.h
index f55153c0ad..255ee2e3df 100644
--- a/core/fxcrt/fx_extension.h
+++ b/core/fxcrt/fx_extension.h
@@ -76,6 +76,12 @@ inline int FXSYS_DecimalCharToInt(const wchar_t c) {
   return std::iswdigit(c) ? c - L'0' : 0;
 }
 
+void FXSYS_IntToTwoHexChars(uint8_t c, char* buf);
+
+void FXSYS_IntToFourHexChars(uint16_t c, char* buf);
+
+size_t FXSYS_ToUTF16BE(uint32_t unicode, char* buf);
+
 float FXSYS_FractionalScale(size_t scale_factor, int value);
 int FXSYS_FractionalScaleCount();
 
diff --git a/core/fxcrt/fx_extension_unittest.cpp b/core/fxcrt/fx_extension_unittest.cpp
index 1bc3ec6298..38b66ba2d2 100644
--- a/core/fxcrt/fx_extension_unittest.cpp
+++ b/core/fxcrt/fx_extension_unittest.cpp
@@ -39,3 +39,53 @@ TEST(fxcrt, FX_HashCode_Wide) {
   EXPECT_EQ(97u, FX_HashCode_GetW(L"A", true));
   EXPECT_EQ(1313 * 65u + 66u, FX_HashCode_GetW(L"AB", false));
 }
+
+TEST(fxcrt, FXSYS_IntToTwoHexChars) {
+  char buf[3] = {0};
+  FXSYS_IntToTwoHexChars(0x0, buf);
+  EXPECT_STREQ("00", buf);
+  FXSYS_IntToTwoHexChars(0x9, buf);
+  EXPECT_STREQ("09", buf);
+  FXSYS_IntToTwoHexChars(0xA, buf);
+  EXPECT_STREQ("0A", buf);
+  FXSYS_IntToTwoHexChars(0x8C, buf);
+  EXPECT_STREQ("8C", buf);
+  FXSYS_IntToTwoHexChars(0xBE, buf);
+  EXPECT_STREQ("BE", buf);
+  FXSYS_IntToTwoHexChars(0xD0, buf);
+  EXPECT_STREQ("D0", buf);
+  FXSYS_IntToTwoHexChars(0xFF, buf);
+  EXPECT_STREQ("FF", buf);
+}
+
+TEST(fxcrt, FXSYS_IntToFourHexChars) {
+  char buf[5] = {0};
+  FXSYS_IntToFourHexChars(0x0, buf);
+  EXPECT_STREQ("0000", buf);
+  FXSYS_IntToFourHexChars(0xA23, buf);
+  EXPECT_STREQ("0A23", buf);
+  FXSYS_IntToFourHexChars(0xB701, buf);
+  EXPECT_STREQ("B701", buf);
+  FXSYS_IntToFourHexChars(0xFFFF, buf);
+  EXPECT_STREQ("FFFF", buf);
+}
+
+TEST(fxcrt, FXSYS_ToUTF16BE) {
+  char buf[9] = {0};
+  // Test U+0000 to U+D7FF and U+E000 to U+FFFF
+  EXPECT_EQ(4U, FXSYS_ToUTF16BE(0x0, buf));
+  EXPECT_STREQ("0000", buf);
+  EXPECT_EQ(4U, FXSYS_ToUTF16BE(0xD7FF, buf));
+  EXPECT_STREQ("D7FF", buf);
+  EXPECT_EQ(4U, FXSYS_ToUTF16BE(0xE000, buf));
+  EXPECT_STREQ("E000", buf);
+  EXPECT_EQ(4U, FXSYS_ToUTF16BE(0xFFFF, buf));
+  EXPECT_STREQ("FFFF", buf);
+  // Test U+10000 to U+10FFFF
+  EXPECT_EQ(8U, FXSYS_ToUTF16BE(0x10000, buf));
+  EXPECT_STREQ("D800DC00", buf);
+  EXPECT_EQ(8U, FXSYS_ToUTF16BE(0x10FFFF, buf));
+  EXPECT_STREQ("DBFFDFFF", buf);
+  EXPECT_EQ(8U, FXSYS_ToUTF16BE(0x2003E, buf));
+  EXPECT_STREQ("D840DC3E", buf);
+}