Clean up IFX_BidiChar

- Replace IFX_BidiChar with just CFX_BidiChar
- Document implementation
- Change out parameters to pointers
- Remove dead code
- Add an enum for bidi directions
- Move several externs to a header
- Add unit tests

R=tsepez@chromium.org

Review URL: https://codereview.chromium.org/1197643002 .

12 files changed, 345 insertions, 299 deletions

diff --git a/core/include/fxcrt/fx_arb.h b/core/include/fxcrt/fx_arb.h
deleted file mode 100644
index 0f6de64b32..0000000000
--- a/core/include/fxcrt/fx_arb.h
+++ /dev/null
@@ -1,24 +0,0 @@
-// Copyright 2014 PDFium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
-
-#ifndef CORE_INCLUDE_FXCRT_FX_ARB_H_
-#define CORE_INCLUDE_FXCRT_FX_ARB_H_
-
-#include "fx_system.h"
-
-class IFX_BidiChar {
- public:
-  static IFX_BidiChar* Create();
-  virtual ~IFX_BidiChar() {}
-
-  virtual void SetPolicy(FX_BOOL bSeparateNeutral = TRUE) = 0;
-  virtual FX_BOOL AppendChar(FX_WCHAR wch) = 0;
-  virtual FX_BOOL EndChar() = 0;
-  virtual int32_t GetBidiInfo(int32_t& iStart, int32_t& iCount) = 0;
-  virtual void Reset() = 0;
-};
-
-#endif  // CORE_INCLUDE_FXCRT_FX_ARB_H_
diff --git a/core/include/fxcrt/fx_bidi.h b/core/include/fxcrt/fx_bidi.h
new file mode 100644
index 0000000000..a55ce6cfd2
--- /dev/null
+++ b/core/include/fxcrt/fx_bidi.h
@@ -0,0 +1,59 @@
+// Copyright 2014 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
+
+#ifndef CORE_INCLUDE_FXCRT_FX_BIDI_H_
+#define CORE_INCLUDE_FXCRT_FX_BIDI_H_
+
+#include "fx_system.h"
+
+// Processes characters and group them into segments based on text direction.
+class CFX_BidiChar {
+ public:
+  enum Direction { NEUTRAL, LEFT, RIGHT };
+
+  CFX_BidiChar();
+  ~CFX_BidiChar();
+
+  // Append a character and classify it as left, right, or neutral.
+  // Returns true if the character has a different direction than the
+  // existing direction to indicate there is a segment to process.
+  bool AppendChar(FX_WCHAR wch);
+
+  // Call this after the last character has been appended. AppendChar()
+  // must not be called after this.
+  // Returns true if there is still a segment to process.
+  bool EndChar();
+
+  // Get information about the segment to process.
+  // The segment's start position and character count is returned in |iStart|
+  // and |iCount|, respectively. Pass in null pointers if the information is
+  // not needed.
+  // Returns the segment direction.
+  Direction GetBidiInfo(int32_t* iStart, int32_t* iCount) const;
+
+ private:
+  void SaveCurrentStateToLastState();
+
+  // Position of the current segment.
+  int32_t m_iCurStart;
+
+  // Number of characters in the current segment.
+  int32_t m_iCurCount;
+
+  // Direction of the current segment.
+  Direction m_CurBidi;
+
+  // Number of characters in the last segment.
+  int32_t m_iLastStart;
+
+  // Number of characters in the last segment.
+  int32_t m_iLastCount;
+
+  // Direction of the last segment.
+  Direction m_LastBidi;
+};
+
+#endif  // CORE_INCLUDE_FXCRT_FX_BIDI_H_
diff --git a/core/include/fxcrt/fx_ucd.h b/core/include/fxcrt/fx_ucd.h
index 8bc5930dd4..62c2dfdf67 100644
--- a/core/include/fxcrt/fx_ucd.h
+++ b/core/include/fxcrt/fx_ucd.h
@@ -9,97 +9,41 @@
 
 #include "fx_system.h"
 
-enum FX_CHARBREAKPROP {
-  FX_CBP_OP = 0,
-  FX_CBP_CL = 1,
-  FX_CBP_QU = 2,
-  FX_CBP_GL = 3,
-  FX_CBP_NS = 4,
-  FX_CBP_EX = 5,
-  FX_CBP_SY = 6,
-  FX_CBP_IS = 7,
-  FX_CBP_PR = 8,
-  FX_CBP_PO = 9,
-  FX_CBP_NU = 10,
-  FX_CBP_AL = 11,
-  FX_CBP_ID = 12,
-  FX_CBP_IN = 13,
-  FX_CBP_HY = 14,
-  FX_CBP_BA = 15,
-  FX_CBP_BB = 16,
-  FX_CBP_B2 = 17,
-  FX_CBP_ZW = 18,
-  FX_CBP_CM = 19,
-  FX_CBP_WJ = 20,
-  FX_CBP_H2 = 21,
-  FX_CBP_H3 = 22,
-  FX_CBP_JL = 23,
-  FX_CBP_JV = 24,
-  FX_CBP_JT = 25,
-
-  FX_CBP_BK = 26,
-  FX_CBP_CR = 27,
-  FX_CBP_LF = 28,
-  FX_CBP_NL = 29,
-  FX_CBP_SA = 30,
-  FX_CBP_SG = 31,
-  FX_CBP_CB = 32,
-  FX_CBP_XX = 33,
-  FX_CBP_AI = 34,
-  FX_CBP_SP = 35,
-  FX_CBP_TB = 37,
-  FX_CBP_NONE = 36,
-};
 #define FX_BIDICLASSBITS 6
 #define FX_BIDICLASSBITSMASK (31 << FX_BIDICLASSBITS)
 enum FX_BIDICLASS {
-  FX_BIDICLASS_ON = 0,
-  FX_BIDICLASS_L = 1,
-  FX_BIDICLASS_R = 2,
-  FX_BIDICLASS_AN = 3,
-  FX_BIDICLASS_EN = 4,
-  FX_BIDICLASS_AL = 5,
-  FX_BIDICLASS_NSM = 6,
-  FX_BIDICLASS_CS = 7,
-  FX_BIDICLASS_ES = 8,
-  FX_BIDICLASS_ET = 9,
-  FX_BIDICLASS_BN = 10,
-  FX_BIDICLASS_S = 11,
-  FX_BIDICLASS_WS = 12,
-  FX_BIDICLASS_B = 13,
-  FX_BIDICLASS_RLO = 14,
-  FX_BIDICLASS_RLE = 15,
-  FX_BIDICLASS_LRO = 16,
-  FX_BIDICLASS_LRE = 17,
-  FX_BIDICLASS_PDF = 18,
+  FX_BIDICLASS_ON = 0,    // Other Neutral
+  FX_BIDICLASS_L = 1,     // Left Letter
+  FX_BIDICLASS_R = 2,     // Right Letter
+  FX_BIDICLASS_AN = 3,    // Arabic Number
+  FX_BIDICLASS_EN = 4,    // European Number
+  FX_BIDICLASS_AL = 5,    // Arabic Letter
+  FX_BIDICLASS_NSM = 6,   // Non-spacing Mark
+  FX_BIDICLASS_CS = 7,    // Common Number Separator
+  FX_BIDICLASS_ES = 8,    // European Separator
+  FX_BIDICLASS_ET = 9,    // European Number Terminator
+  FX_BIDICLASS_BN = 10,   // Boundary Neutral
+  FX_BIDICLASS_S = 11,    // Segment Separator
+  FX_BIDICLASS_WS = 12,   // Whitespace
+  FX_BIDICLASS_B = 13,    // Paragraph Separator
+  FX_BIDICLASS_RLO = 14,  // Right-to-Left Override
+  FX_BIDICLASS_RLE = 15,  // Right-to-Left Embedding
+  FX_BIDICLASS_LRO = 16,  // Left-to-Right Override
+  FX_BIDICLASS_LRE = 17,  // Left-to-Right Embedding
+  FX_BIDICLASS_PDF = 18,  // Pop Directional Format
   FX_BIDICLASS_N = FX_BIDICLASS_ON,
 };
-#define FX_CHARTYPEBITS 11
-#define FX_CHARTYPEBITSMASK (15 << FX_CHARTYPEBITS)
-enum FX_CHARTYPE {
-  FX_CHARTYPE_Unknown = 0,
-  FX_CHARTYPE_Tab = (1 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_Space = (2 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_Control = (3 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_Combination = (4 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_Numeric = (5 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_Normal = (6 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_ArabicAlef = (7 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_ArabicSpecial = (8 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_ArabicDistortion = (9 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_ArabicNormal = (10 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_ArabicForm = (11 << FX_CHARTYPEBITS),
-  FX_CHARTYPE_Arabic = (12 << FX_CHARTYPEBITS),
-};
+
+extern const FX_DWORD kTextLayoutCodeProperties[];
+extern const size_t kTextLayoutCodePropertiesSize;
+
+extern const FX_WCHAR kFXTextLayoutVerticalMirror[];
+extern const size_t kFXTextLayoutVerticalMirrorSize;
+
+extern const FX_WCHAR kFXTextLayoutBidiMirror[];
+extern const size_t kFXTextLayoutBidiMirrorSize;
+
 FX_DWORD FX_GetUnicodeProperties(FX_WCHAR wch);
-FX_BOOL FX_IsCtrlCode(FX_WCHAR ch);
-FX_BOOL FX_IsRotationCode(FX_WCHAR ch);
-FX_BOOL FX_IsCombinationChar(FX_WCHAR wch);
-FX_BOOL FX_IsBidiChar(FX_WCHAR wch);
 FX_WCHAR FX_GetMirrorChar(FX_WCHAR wch, FX_BOOL bRTL, FX_BOOL bVertical);
-FX_WCHAR FX_GetMirrorChar(FX_WCHAR wch,
-                          FX_DWORD dwProps,
-                          FX_BOOL bRTL,
-                          FX_BOOL bVertical);
 
 #endif  // CORE_INCLUDE_FXCRT_FX_UCD_H_
diff --git a/core/src/fpdftext/fpdf_text.cpp b/core/src/fpdftext/fpdf_text.cpp
index e7ca6c136f..af944d965e 100644
--- a/core/src/fpdftext/fpdf_text.cpp
+++ b/core/src/fpdftext/fpdf_text.cpp
@@ -9,7 +9,7 @@
 #include "../../include/fpdfapi/fpdf_pageobj.h"
 #include "../../include/fpdfapi/fpdf_resource.h"
 #include "../../include/fpdftext/fpdf_text.h"
-#include "../../include/fxcrt/fx_arb.h"
+#include "../../include/fxcrt/fx_bidi.h"
 #include "../../include/fxcrt/fx_ucd.h"
 #include "text_int.h"
 #include "txtproc.h"
@@ -321,35 +321,35 @@ void NormalizeString(CFX_WideString& str) {
     return;
   }
   CFX_WideString sBuffer;
-  nonstd::unique_ptr<IFX_BidiChar> pBidiChar(IFX_BidiChar::Create());
+  nonstd::unique_ptr<CFX_BidiChar> pBidiChar(new CFX_BidiChar);
   CFX_WordArray order;
   FX_BOOL bR2L = FALSE;
   int32_t start = 0, count = 0, i = 0;
   int nR2L = 0, nL2R = 0;
   for (i = 0; i < str.GetLength(); i++) {
     if (pBidiChar->AppendChar(str.GetAt(i))) {
-      int32_t ret = pBidiChar->GetBidiInfo(start, count);
+      CFX_BidiChar::Direction ret = pBidiChar->GetBidiInfo(&start, &count);
       order.Add(start);
       order.Add(count);
       order.Add(ret);
       if (!bR2L) {
-        if (ret == 2) {
+        if (ret == CFX_BidiChar::RIGHT) {
           nR2L++;
-        } else if (ret == 1) {
+        } else if (ret == CFX_BidiChar::LEFT) {
           nL2R++;
         }
       }
     }
   }
   if (pBidiChar->EndChar()) {
-    int32_t ret = pBidiChar->GetBidiInfo(start, count);
+    CFX_BidiChar::Direction ret = pBidiChar->GetBidiInfo(&start, &count);
     order.Add(start);
     order.Add(count);
     order.Add(ret);
     if (!bR2L) {
-      if (ret == 2) {
+      if (ret == CFX_BidiChar::RIGHT) {
         nR2L++;
-      } else if (ret == 1) {
+      } else if (ret == CFX_BidiChar::LEFT) {
         nL2R++;
       }
     }
diff --git a/core/src/fpdftext/fpdf_text_int.cpp b/core/src/fpdftext/fpdf_text_int.cpp
index ce673447a4..462f1369dd 100644
--- a/core/src/fpdftext/fpdf_text_int.cpp
+++ b/core/src/fpdftext/fpdf_text_int.cpp
@@ -13,7 +13,7 @@
 #include "../../include/fpdfapi/fpdf_pageobj.h"
 #include "../../include/fpdfapi/fpdf_resource.h"
 #include "../../include/fpdftext/fpdf_text.h"
-#include "../../include/fxcrt/fx_arb.h"
+#include "../../include/fxcrt/fx_bidi.h"
 #include "../../include/fxcrt/fx_ucd.h"
 #include "text_int.h"
 
@@ -1020,10 +1020,10 @@ int CPDF_TextPage::GetCharWidth(FX_DWORD charCode, CPDF_Font* pFont) const {
   }
   return w;
 }
-void CPDF_TextPage::OnPiece(IFX_BidiChar* pBidi, CFX_WideString& str) {
+void CPDF_TextPage::OnPiece(CFX_BidiChar* pBidi, CFX_WideString& str) {
   int32_t start, count;
-  int32_t ret = pBidi->GetBidiInfo(start, count);
-  if (ret == 2) {
+  CFX_BidiChar::Direction ret = pBidi->GetBidiInfo(&start, &count);
+  if (ret == CFX_BidiChar::RIGHT) {
     for (int i = start + count - 1; i >= start; i--) {
       m_TextBuf.AppendChar(str.GetAt(i));
       m_charList.Add(*(PAGECHAR_INFO*)m_TempCharList.GetAt(i));
@@ -1104,7 +1104,7 @@ void CPDF_TextPage::CloseTempLine() {
   if (count1 <= 0) {
     return;
   }
-  nonstd::unique_ptr<IFX_BidiChar> pBidiChar(IFX_BidiChar::Create());
+  nonstd::unique_ptr<CFX_BidiChar> pBidiChar(new CFX_BidiChar);
   CFX_WideString str = m_TempTextBuf.GetWideString();
   CFX_WordArray order;
   FX_BOOL bR2L = FALSE;
@@ -1126,28 +1126,28 @@ void CPDF_TextPage::CloseTempLine() {
       bPrevSpace = FALSE;
     }
     if (pBidiChar->AppendChar(str.GetAt(i))) {
-      int32_t ret = pBidiChar->GetBidiInfo(start, count);
+      CFX_BidiChar::Direction ret = pBidiChar->GetBidiInfo(&start, &count);
       order.Add(start);
       order.Add(count);
       order.Add(ret);
       if (!bR2L) {
-        if (ret == 2) {
+        if (ret == CFX_BidiChar::RIGHT) {
           nR2L++;
-        } else if (ret == 1) {
+        } else if (ret == CFX_BidiChar::LEFT) {
           nL2R++;
         }
       }
     }
   }
   if (pBidiChar->EndChar()) {
-    int32_t ret = pBidiChar->GetBidiInfo(start, count);
+    CFX_BidiChar::Direction ret = pBidiChar->GetBidiInfo(&start, &count);
     order.Add(start);
     order.Add(count);
     order.Add(ret);
     if (!bR2L) {
-      if (ret == 2) {
+      if (ret == CFX_BidiChar::RIGHT) {
         nR2L++;
-      } else if (ret == 1) {
+      } else if (ret == CFX_BidiChar::LEFT) {
         nL2R++;
       }
     }
@@ -1762,7 +1762,7 @@ void CPDF_TextPage::SwapTempTextBuf(int32_t iCharListStartAppend,
 FX_BOOL CPDF_TextPage::IsRightToLeft(const CPDF_TextObject* pTextObj,
                                      const CPDF_Font* pFont,
                                      int nItems) const {
-  nonstd::unique_ptr<IFX_BidiChar> pBidiChar(IFX_BidiChar::Create());
+  nonstd::unique_ptr<CFX_BidiChar> pBidiChar(new CFX_BidiChar);
   int32_t nR2L = 0;
   int32_t nL2R = 0;
   int32_t start = 0, count = 0;
@@ -1781,19 +1781,19 @@ FX_BOOL CPDF_TextPage::IsRightToLeft(const CPDF_TextObject* pTextObj,
       continue;
     }
     if (pBidiChar->AppendChar(wChar)) {
-      int32_t ret = pBidiChar->GetBidiInfo(start, count);
-      if (ret == 2) {
+      CFX_BidiChar::Direction ret = pBidiChar->GetBidiInfo(&start, &count);
+      if (ret == CFX_BidiChar::RIGHT) {
         nR2L++;
-      } else if (ret == 1) {
+      } else if (ret == CFX_BidiChar::LEFT) {
         nL2R++;
       }
     }
   }
   if (pBidiChar->EndChar()) {
-    int32_t ret = pBidiChar->GetBidiInfo(start, count);
-    if (ret == 2) {
+    CFX_BidiChar::Direction ret = pBidiChar->GetBidiInfo(&start, &count);
+    if (ret == CFX_BidiChar::RIGHT) {
       nR2L++;
-    } else if (ret == 1) {
+    } else if (ret == CFX_BidiChar::LEFT) {
       nL2R++;
     }
   }
diff --git a/core/src/fpdftext/text_int.h b/core/src/fpdftext/text_int.h
index 091ee241d4..3f1023ac8a 100644
--- a/core/src/fpdftext/text_int.h
+++ b/core/src/fpdftext/text_int.h
@@ -111,7 +111,7 @@ class CPDF_TextPage : public IPDF_TextPage {
                            CPDF_TextObject* pTextObj2);
   int GetCharWidth(FX_DWORD charCode, CPDF_Font* pFont) const;
   void CloseTempLine();
-  void OnPiece(IFX_BidiChar* pBidi, CFX_WideString& str);
+  void OnPiece(CFX_BidiChar* pBidi, CFX_WideString& str);
   int32_t PreMarkedContent(PDFTEXT_Obj pObj);
   void ProcessMarkedContent(PDFTEXT_Obj pObj);
   void CheckMarkedContentObject(int32_t& start, int32_t& nCount) const;
diff --git a/core/src/fxcrt/fx_arabic.cpp b/core/src/fxcrt/fx_arabic.cpp
deleted file mode 100644
index 3b7d0c1c71..0000000000
--- a/core/src/fxcrt/fx_arabic.cpp
+++ /dev/null
@@ -1,84 +0,0 @@
-// Copyright 2014 PDFium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
-
-#include "../../include/fxcrt/fx_ucd.h"
-#include "fx_arabic.h"
-
-extern const FX_DWORD gs_FX_TextLayout_CodeProperties[65536];
-IFX_BidiChar* IFX_BidiChar::Create() {
-  return new CFX_BidiChar;
-}
-CFX_BidiChar::CFX_BidiChar()
-    : m_bSeparateNeutral(TRUE),
-      m_iCurStart(0),
-      m_iCurCount(0),
-      m_iCurBidi(0),
-      m_iLastBidi(0),
-      m_iLastStart(0),
-      m_iLastCount(0) {}
-void CFX_BidiChar::SetPolicy(FX_BOOL bSeparateNeutral) {
-  m_bSeparateNeutral = bSeparateNeutral;
-}
-
-FX_BOOL CFX_BidiChar::AppendChar(FX_WCHAR wch) {
-  FX_DWORD dwProps = gs_FX_TextLayout_CodeProperties[(FX_WORD)wch];
-  int32_t iBidiCls = (dwProps & FX_BIDICLASSBITSMASK) >> FX_BIDICLASSBITS;
-  int32_t iContext = 0;
-  switch (iBidiCls) {
-    case FX_BIDICLASS_L:
-    case FX_BIDICLASS_AN:
-    case FX_BIDICLASS_EN:
-      iContext = 1;
-      break;
-    case FX_BIDICLASS_R:
-    case FX_BIDICLASS_AL:
-      iContext = 2;
-      break;
-  }
-  FX_BOOL bRet = FALSE;
-  if (iContext != m_iCurBidi) {
-    if (m_bSeparateNeutral) {
-      bRet = TRUE;
-    } else {
-      if (m_iCurBidi == 0) {
-        bRet = (m_iCurCount > 0);
-      } else {
-        bRet = (iContext != 0);
-      }
-    }
-    if (bRet) {
-      m_iLastBidi = m_iCurBidi;
-      m_iLastStart = m_iCurStart;
-      m_iCurStart = m_iCurCount;
-      m_iLastCount = m_iCurCount - m_iLastStart;
-    }
-    if (m_bSeparateNeutral || iContext != 0) {
-      m_iCurBidi = iContext;
-    }
-  }
-  m_iCurCount++;
-  return bRet;
-}
-FX_BOOL CFX_BidiChar::EndChar() {
-  m_iLastBidi = m_iCurBidi;
-  m_iLastStart = m_iCurStart;
-  m_iCurStart = m_iCurCount;
-  m_iLastCount = m_iCurCount - m_iLastStart;
-  return m_iLastCount > 0;
-}
-int32_t CFX_BidiChar::GetBidiInfo(int32_t& iStart, int32_t& iCount) {
-  iStart = m_iLastStart;
-  iCount = m_iLastCount;
-  return m_iLastBidi;
-}
-void CFX_BidiChar::Reset() {
-  m_iCurStart = 0;
-  m_iCurCount = 0;
-  m_iCurBidi = 0;
-  m_iLastBidi = 0;
-  m_iLastStart = 0;
-  m_iLastCount = 0;
-}
diff --git a/core/src/fxcrt/fx_arabic.h b/core/src/fxcrt/fx_arabic.h
deleted file mode 100644
index 0230d40579..0000000000
--- a/core/src/fxcrt/fx_arabic.h
+++ /dev/null
@@ -1,33 +0,0 @@
-// Copyright 2014 PDFium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
-
-#ifndef CORE_SRC_FXCRT_FX_ARABIC_H_
-#define CORE_SRC_FXCRT_FX_ARABIC_H_
-
-#include "../../include/fxcrt/fx_arb.h"
-
-class CFX_BidiChar final : public IFX_BidiChar {
- public:
-  CFX_BidiChar();
-  ~CFX_BidiChar() override {}
-
-  void SetPolicy(FX_BOOL bSeparateNeutral = TRUE) override;
-  FX_BOOL AppendChar(FX_WCHAR wch) override;
-  FX_BOOL EndChar() override;
-  int32_t GetBidiInfo(int32_t& iStart, int32_t& iCount) override;
-  void Reset() override;
-
- private:
-  FX_BOOL m_bSeparateNeutral;
-  int32_t m_iCurStart;
-  int32_t m_iCurCount;
-  int32_t m_iCurBidi;
-  int32_t m_iLastBidi;
-  int32_t m_iLastStart;
-  int32_t m_iLastCount;
-};
-
-#endif  // CORE_SRC_FXCRT_FX_ARABIC_H_
diff --git a/core/src/fxcrt/fx_bidi.cpp b/core/src/fxcrt/fx_bidi.cpp
new file mode 100644
index 0000000000..0310fa0e94
--- /dev/null
+++ b/core/src/fxcrt/fx_bidi.cpp
@@ -0,0 +1,66 @@
+// Copyright 2014 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
+
+#include "../../include/fxcrt/fx_bidi.h"
+#include "../../include/fxcrt/fx_ucd.h"
+
+CFX_BidiChar::CFX_BidiChar()
+    : m_iCurStart(0),
+      m_iCurCount(0),
+      m_CurBidi(NEUTRAL),
+      m_iLastStart(0),
+      m_iLastCount(0),
+      m_LastBidi(NEUTRAL) {
+}
+
+CFX_BidiChar::~CFX_BidiChar() {
+}
+
+bool CFX_BidiChar::AppendChar(FX_WCHAR wch) {
+  FX_DWORD dwProps = FX_GetUnicodeProperties(wch);
+  int32_t iBidiCls = (dwProps & FX_BIDICLASSBITSMASK) >> FX_BIDICLASSBITS;
+  Direction bidi = NEUTRAL;
+  switch (iBidiCls) {
+    case FX_BIDICLASS_L:
+    case FX_BIDICLASS_AN:
+    case FX_BIDICLASS_EN:
+      bidi = LEFT;
+      break;
+    case FX_BIDICLASS_R:
+    case FX_BIDICLASS_AL:
+      bidi = RIGHT;
+      break;
+  }
+
+  bool bRet = (bidi != m_CurBidi);
+  if (bRet) {
+    SaveCurrentStateToLastState();
+    m_CurBidi = bidi;
+  }
+  m_iCurCount++;
+  return bRet;
+}
+
+bool CFX_BidiChar::EndChar() {
+  SaveCurrentStateToLastState();
+  return m_iLastCount > 0;
+}
+
+CFX_BidiChar::Direction CFX_BidiChar::GetBidiInfo(int32_t* iStart,
+                                                  int32_t* iCount) const {
+  if (iStart)
+    *iStart = m_iLastStart;
+  if (iCount)
+    *iCount = m_iLastCount;
+  return m_LastBidi;
+}
+
+void CFX_BidiChar::SaveCurrentStateToLastState() {
+  m_LastBidi = m_CurBidi;
+  m_iLastStart = m_iCurStart;
+  m_iCurStart = m_iCurCount;
+  m_iLastCount = m_iCurCount - m_iLastStart;
+}
diff --git a/core/src/fxcrt/fx_bidi_unittest.cpp b/core/src/fxcrt/fx_bidi_unittest.cpp
new file mode 100644
index 0000000000..c629cbbdc6
--- /dev/null
+++ b/core/src/fxcrt/fx_bidi_unittest.cpp
@@ -0,0 +1,136 @@
+// Copyright 2015 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "testing/gtest/include/gtest/gtest.h"
+#include "../../include/fxcrt/fx_bidi.h"
+
+namespace {
+
+const FX_WCHAR kNeutralChar = 32;
+const FX_WCHAR kLeftChar = 65;
+const FX_WCHAR kRightChar = 1424;
+
+}  // namespace
+
+TEST(fxcrt, BidiCharEmpty) {
+  int32_t start = -1;
+  int32_t count = -1;
+  CFX_BidiChar bidi;
+  CFX_BidiChar::Direction dir = bidi.GetBidiInfo(nullptr, nullptr);
+  EXPECT_EQ(CFX_BidiChar::NEUTRAL, dir);
+
+  dir = bidi.GetBidiInfo(&start, nullptr);
+  EXPECT_EQ(CFX_BidiChar::NEUTRAL, dir);
+  EXPECT_EQ(0, start);
+
+  dir = bidi.GetBidiInfo(nullptr, &count);
+  EXPECT_EQ(CFX_BidiChar::NEUTRAL, dir);
+  EXPECT_EQ(0, count);
+
+  start = -1;
+  count = -1;
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(CFX_BidiChar::NEUTRAL, dir);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(0, count);
+
+  EXPECT_FALSE(bidi.EndChar());
+}
+
+TEST(fxcrt, BidiCharLeft) {
+  int32_t start = -1;
+  int32_t count = -1;
+  CFX_BidiChar bidi;
+
+  EXPECT_TRUE(bidi.AppendChar(kLeftChar));
+  CFX_BidiChar::Direction dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(0, count);
+
+  EXPECT_FALSE(bidi.AppendChar(kLeftChar));
+  EXPECT_FALSE(bidi.AppendChar(kLeftChar));
+
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(CFX_BidiChar::NEUTRAL, dir);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(0, count);
+
+  EXPECT_TRUE(bidi.EndChar());
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(CFX_BidiChar::LEFT, dir);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(3, count);
+
+  EXPECT_FALSE(bidi.EndChar());
+}
+
+TEST(fxcrt, BidiCharLeftNeutralRight) {
+  int32_t start = -1;
+  int32_t count = -1;
+  CFX_BidiChar bidi;
+
+  EXPECT_TRUE(bidi.AppendChar(kLeftChar));
+  CFX_BidiChar::Direction dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(0, count);
+
+  EXPECT_FALSE(bidi.AppendChar(kLeftChar));
+  EXPECT_FALSE(bidi.AppendChar(kLeftChar));
+  EXPECT_TRUE(bidi.AppendChar(kNeutralChar));
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(3, count);
+
+  EXPECT_FALSE(bidi.AppendChar(kNeutralChar));
+  EXPECT_FALSE(bidi.AppendChar(kNeutralChar));
+  EXPECT_FALSE(bidi.AppendChar(kNeutralChar));
+  EXPECT_TRUE(bidi.AppendChar(kRightChar));
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(CFX_BidiChar::NEUTRAL, dir);
+  EXPECT_EQ(3, start);
+  EXPECT_EQ(4, count);
+
+  EXPECT_TRUE(bidi.EndChar());
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(CFX_BidiChar::RIGHT, dir);
+  EXPECT_EQ(7, start);
+  EXPECT_EQ(1, count);
+
+  EXPECT_FALSE(bidi.EndChar());
+}
+
+TEST(fxcrt, BidiCharLeftRightLeft) {
+  int32_t start = -1;
+  int32_t count = -1;
+  CFX_BidiChar bidi;
+
+  EXPECT_TRUE(bidi.AppendChar(kLeftChar));
+  CFX_BidiChar::Direction dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(0, count);
+
+  EXPECT_FALSE(bidi.AppendChar(kLeftChar));
+  EXPECT_FALSE(bidi.AppendChar(kLeftChar));
+  EXPECT_TRUE(bidi.AppendChar(kRightChar));
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(0, start);
+  EXPECT_EQ(3, count);
+
+  EXPECT_FALSE(bidi.AppendChar(kRightChar));
+  EXPECT_FALSE(bidi.AppendChar(kRightChar));
+  EXPECT_FALSE(bidi.AppendChar(kRightChar));
+  EXPECT_TRUE(bidi.AppendChar(kLeftChar));
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(CFX_BidiChar::RIGHT, dir);
+  EXPECT_EQ(3, start);
+  EXPECT_EQ(4, count);
+
+  EXPECT_TRUE(bidi.EndChar());
+  dir = bidi.GetBidiInfo(&start, &count);
+  EXPECT_EQ(CFX_BidiChar::LEFT, dir);
+  EXPECT_EQ(7, start);
+  EXPECT_EQ(1, count);
+
+  EXPECT_FALSE(bidi.EndChar());
+}
diff --git a/core/src/fxcrt/fx_ucddata.cpp b/core/src/fxcrt/fx_ucddata.cpp
index 8b35f31e2b..13c5da1419 100644
--- a/core/src/fxcrt/fx_ucddata.cpp
+++ b/core/src/fxcrt/fx_ucddata.cpp
@@ -4,9 +4,10 @@
 
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
-#include "../../include/fxcrt/fx_system.h"
+#include "../../include/fxcrt/fx_basic.h"
+#include "../../include/fxcrt/fx_ucd.h"
 
-extern const FX_DWORD gs_FX_TextLayout_CodeProperties[65536] = {
+const FX_DWORD kTextLayoutCodeProperties[] = {
     0xfffe9a93, 0xfffe9a93, 0xfffe9a93, 0xfffe9a93, 0xfffe9a93, 0xfffe9a93,
     0xfffe9a93, 0xfffe9a93, 0xfffe9a93, 0xfffe8ae5, 0xfffe9b5c, 0xfffe9ada,
     0xfffe9b1a, 0xfffe9b5b, 0xfffe9a93, 0xfffe9a93, 0xfffe9a93, 0xfffe9a93,
@@ -10931,7 +10932,11 @@ extern const FX_DWORD gs_FX_TextLayout_CodeProperties[65536] = {
     0xfffe02a4, 0xfffe02a4, 0xfffe02a4, 0xfffe3013, 0xfffe3013, 0xfffe3013,
     0xfffe3011, 0xfffe3022, 0xfffe1aa4, 0xfffe1aa4,
 };
-extern const FX_WCHAR gs_FX_TextLayout_VerticalMirror[64] = {
+
+const size_t kTextLayoutCodePropertiesSize =
+    FX_ArraySize(kTextLayoutCodeProperties);
+
+const FX_WCHAR kFXTextLayoutVerticalMirror[] = {
     0xFE33, 0xFE32, 0xFE31, 0xFE41, 0xFE42, 0xFE43, 0xFE44, 0xFE3F,
     0xFE40, 0xFE3D, 0xFE3E, 0xFE41, 0xFE42, 0xFE43, 0xFE44, 0xFE3B,
     0xFE3C, 0xFE39, 0xFE3A, 0xFE34, 0xFE35, 0xFE36, 0xFE37, 0xFE38,
@@ -10941,7 +10946,10 @@ extern const FX_WCHAR gs_FX_TextLayout_VerticalMirror[64] = {
     0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF,
     0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF,
 };
-extern const FX_WCHAR gs_FX_TextLayout_BidiMirror[512] = {
+const size_t kFXTextLayoutVerticalMirrorSize =
+    FX_ArraySize(kFXTextLayoutVerticalMirror);
+
+const FX_WCHAR kFXTextLayoutBidiMirror[] = {
     0x0029, 0x0028, 0x003E, 0x003C, 0x005D, 0x005B, 0x007D, 0x007B, 0x00BB,
     0x00AB, 0x0F3B, 0x0F3A, 0x0F3D, 0x0F3C, 0x169C, 0x169B, 0x2019, 0x2018,
     0x201D, 0x201C, 0x203A, 0x2039, 0x2046, 0x2045, 0x207E, 0x207D, 0x208E,
@@ -11000,3 +11008,6 @@ extern const FX_WCHAR gs_FX_TextLayout_BidiMirror[512] = {
     0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF,
     0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF, 0xFEFF,
 };
+
+const size_t kFXTextLayoutBidiMirrorSize =
+    FX_ArraySize(kFXTextLayoutBidiMirror);
diff --git a/core/src/fxcrt/fx_unicode.cpp b/core/src/fxcrt/fx_unicode.cpp
index f05aeb5efb..c7ab618200 100644
--- a/core/src/fxcrt/fx_unicode.cpp
+++ b/core/src/fxcrt/fx_unicode.cpp
@@ -6,58 +6,29 @@
 
 #include "../../include/fxcrt/fx_ucd.h"
 
-extern const FX_DWORD gs_FX_TextLayout_CodeProperties[65536];
-extern const FX_WCHAR gs_FX_TextLayout_VerticalMirror[64];
-extern const FX_WCHAR gs_FX_TextLayout_BidiMirror[512];
 FX_DWORD FX_GetUnicodeProperties(FX_WCHAR wch) {
-  return gs_FX_TextLayout_CodeProperties[(FX_WORD)wch];
-}
-FX_BOOL FX_IsCtrlCode(FX_WCHAR ch) {
-  FX_DWORD dwRet =
-      (gs_FX_TextLayout_CodeProperties[(FX_WORD)ch] & FX_CHARTYPEBITSMASK);
-  return dwRet == FX_CHARTYPE_Tab || dwRet == FX_CHARTYPE_Control;
-}
-FX_BOOL FX_IsRotationCode(FX_WCHAR ch) {
-  return (gs_FX_TextLayout_CodeProperties[(FX_WORD)ch] & 0x8000) != 0;
-}
-FX_BOOL FX_IsCombinationChar(FX_WCHAR wch) {
-  FX_DWORD dwProps =
-      (gs_FX_TextLayout_CodeProperties[(FX_WORD)wch] & FX_CHARTYPEBITSMASK);
-  return dwProps == FX_CHARTYPE_Combination;
-}
-FX_BOOL FX_IsBidiChar(FX_WCHAR wch) {
-  FX_DWORD dwProps = gs_FX_TextLayout_CodeProperties[(FX_WORD)wch];
-  int32_t iBidiCls = (dwProps & FX_BIDICLASSBITSMASK) >> FX_BIDICLASSBITS;
-  return (FX_BIDICLASS_R == iBidiCls || FX_BIDICLASS_AL == iBidiCls);
+  size_t idx = static_cast<size_t>(wch);
+  if (idx < kTextLayoutCodePropertiesSize)
+    return kTextLayoutCodeProperties[(FX_WORD)wch];
+  return 0;
 }
+
 FX_WCHAR FX_GetMirrorChar(FX_WCHAR wch, FX_BOOL bRTL, FX_BOOL bVertical) {
-  FX_DWORD dwProps = gs_FX_TextLayout_CodeProperties[(FX_WORD)wch];
+  FX_DWORD dwProps = FX_GetUnicodeProperties(wch);
   FX_DWORD dwTemp = (dwProps & 0xFF800000);
   if (bRTL && dwTemp < 0xFF800000) {
-    wch = gs_FX_TextLayout_BidiMirror[dwTemp >> 23];
-    dwProps = gs_FX_TextLayout_CodeProperties[(FX_WORD)wch];
-  }
-  if (bVertical) {
-    dwTemp = (dwProps & 0x007E0000);
-    if (dwTemp < 0x007E0000) {
-      wch = gs_FX_TextLayout_VerticalMirror[dwTemp >> 17];
+    size_t idx = dwTemp >> 23;
+    if (idx < kFXTextLayoutBidiMirrorSize) {
+      wch = kFXTextLayoutBidiMirror[idx];
+      dwProps = FX_GetUnicodeProperties(wch);
     }
   }
-  return wch;
-}
-FX_WCHAR FX_GetMirrorChar(FX_WCHAR wch,
-                          FX_DWORD dwProps,
-                          FX_BOOL bRTL,
-                          FX_BOOL bVertical) {
-  FX_DWORD dwTemp = (dwProps & 0xFF800000);
-  if (bRTL && dwTemp < 0xFF800000) {
-    wch = gs_FX_TextLayout_BidiMirror[dwTemp >> 23];
-    dwProps = gs_FX_TextLayout_CodeProperties[(FX_WORD)wch];
-  }
   if (bVertical) {
     dwTemp = (dwProps & 0x007E0000);
     if (dwTemp < 0x007E0000) {
-      wch = gs_FX_TextLayout_VerticalMirror[dwTemp >> 17];
+      size_t idx = dwTemp >> 17;
+      if (idx < kFXTextLayoutVerticalMirrorSize)
+        wch = kFXTextLayoutVerticalMirror[idx];
     }
   }
   return wch;