Combine more RTF and TXT Break code

This Cl conbines several more methods from RTFBreak and TXTBreak into
the common Break baseclass.

Change-Id: Ibe2236eb4fd30faab239e7c9c8b578164014eb39
Reviewed-on: https://pdfium-review.googlesource.com/3108
Reviewed-by: Nicolás Peña <npm@chromium.org>
Commit-Queue: dsinclair <dsinclair@chromium.org>

6 files changed, 80 insertions, 140 deletions

diff --git a/xfa/fgas/layout/cfx_break.cpp b/xfa/fgas/layout/cfx_break.cpp
index a8280b1161..52236f3176 100644
--- a/xfa/fgas/layout/cfx_break.cpp
+++ b/xfa/fgas/layout/cfx_break.cpp
@@ -7,6 +7,9 @@
 #include "xfa/fgas/layout/cfx_break.h"
 
 #include <algorithm>
+#include <vector>
+
+#include "third_party/base/stl_util.h"
 
 namespace {
 
@@ -18,6 +21,7 @@ CFX_Break::CFX_Break(uint32_t dwLayoutStyles)
     : m_eCharType(FX_CHARTYPE_Unknown),
       m_bSingleLine(false),
       m_bCombText(false),
+      m_dwIdentity(0),
       m_dwLayoutStyles(dwLayoutStyles),
       m_iLineStart(0),
       m_iLineWidth(2000000),
@@ -88,6 +92,21 @@ void CFX_Break::SetFontSize(float fFontSize) {
   FontChanged();
 }
 
+void CFX_Break::SetBreakStatus() {
+  ++m_dwIdentity;
+  int32_t iCount = m_pCurLine->CountChars();
+  if (iCount < 1)
+    return;
+
+  CFX_Char* tc = m_pCurLine->GetChar(iCount - 1);
+  if (tc->m_dwStatus == CFX_BreakType::None)
+    tc->m_dwStatus = CFX_BreakType::Piece;
+}
+
+FX_CHARTYPE CFX_Break::GetUnifiedCharType(FX_CHARTYPE chartype) const {
+  return chartype >= FX_CHARTYPE_ArabicAlef ? FX_CHARTYPE_Arabic : chartype;
+}
+
 void CFX_Break::FontChanged() {
   m_iDefChar = 0;
   if (!m_pFont || m_wDefChar == 0xFEFF)
@@ -140,3 +159,43 @@ void CFX_Break::SetLineBoundary(float fLineStart, float fLineEnd) {
   m_pCurLine->m_iStart = std::min(m_pCurLine->m_iStart, m_iLineWidth);
   m_pCurLine->m_iStart = std::max(m_pCurLine->m_iStart, m_iLineStart);
 }
+
+CFX_Char* CFX_Break::GetLastChar(int32_t index,
+                                 bool bOmitChar,
+                                 bool bRichText) const {
+  std::vector<CFX_Char>& tca = m_pCurLine->m_LineChars;
+  if (!pdfium::IndexInBounds(tca, index))
+    return nullptr;
+
+  int32_t iStart = pdfium::CollectionSize<int32_t>(tca) - 1;
+  while (iStart > -1) {
+    CFX_Char* pTC = &tca[iStart--];
+    if (((bRichText && pTC->m_iCharWidth < 0) || bOmitChar) &&
+        pTC->GetCharType() == FX_CHARTYPE_Combination) {
+      continue;
+    }
+    if (--index < 0)
+      return pTC;
+  }
+  return nullptr;
+}
+
+int32_t CFX_Break::CountBreakPieces() const {
+  return HasLine() ? pdfium::CollectionSize<int32_t>(
+                         m_Line[m_iReadyLineIndex].m_LinePieces)
+                   : 0;
+}
+
+const CFX_BreakPiece* CFX_Break::GetBreakPieceUnstable(int32_t index) const {
+  if (!HasLine())
+    return nullptr;
+  if (!pdfium::IndexInBounds(m_Line[m_iReadyLineIndex].m_LinePieces, index))
+    return nullptr;
+  return &m_Line[m_iReadyLineIndex].m_LinePieces[index];
+}
+
+void CFX_Break::ClearBreakPieces() {
+  if (HasLine())
+    m_Line[m_iReadyLineIndex].Clear();
+  m_iReadyLineIndex = -1;
+}
diff --git a/xfa/fgas/layout/cfx_break.h b/xfa/fgas/layout/cfx_break.h
index 631cc5aaa1..4a7185bd70 100644
--- a/xfa/fgas/layout/cfx_break.h
+++ b/xfa/fgas/layout/cfx_break.h
@@ -50,14 +50,23 @@ class CFX_Break {
   void SetParagraphBreakChar(wchar_t wch);
   void SetDefaultChar(wchar_t wch);
 
-  virtual void SetBreakStatus() = 0;
+  int32_t CountBreakPieces() const;
+  const CFX_BreakPiece* GetBreakPieceUnstable(int32_t index) const;
+  void ClearBreakPieces();
+
+  CFX_Char* GetLastChar(int32_t index, bool bOmitChar, bool bRichText) const;
 
  protected:
   explicit CFX_Break(uint32_t dwLayoutStyles);
 
+  void SetBreakStatus();
+  bool HasLine() const { return m_iReadyLineIndex >= 0; }
+  FX_CHARTYPE GetUnifiedCharType(FX_CHARTYPE dwType) const;
+
   FX_CHARTYPE m_eCharType;
   bool m_bSingleLine;
   bool m_bCombText;
+  uint32_t m_dwIdentity;
   uint32_t m_dwLayoutStyles;
   int32_t m_iLineStart;
   int32_t m_iLineWidth;
diff --git a/xfa/fgas/layout/fgas_rtfbreak.cpp b/xfa/fgas/layout/fgas_rtfbreak.cpp
index 81fbea0cc6..07a3cefa69 100644
--- a/xfa/fgas/layout/fgas_rtfbreak.cpp
+++ b/xfa/fgas/layout/fgas_rtfbreak.cpp
@@ -17,8 +17,7 @@ CFX_RTFBreak::CFX_RTFBreak(uint32_t dwLayoutStyles)
     : CFX_Break(dwLayoutStyles),
       m_bPagination(false),
       m_iAlignment(CFX_RTFLineAlignment::Left),
-      m_pUserData(nullptr),
-      m_dwIdentity(0) {
+      m_pUserData(nullptr) {
   SetBreakStatus();
   m_bPagination = !!(m_dwLayoutStyles & FX_LAYOUTSTYLE_Pagination);
 }
@@ -50,40 +49,6 @@ void CFX_RTFBreak::SetUserData(const CFX_RetainPtr<CFX_Retainable>& pUserData) {
   m_pUserData = pUserData;
 }
 
-void CFX_RTFBreak::SetBreakStatus() {
-  ++m_dwIdentity;
-  int32_t iCount = m_pCurLine->CountChars();
-  if (iCount < 1)
-    return;
-
-  CFX_Char* tc = m_pCurLine->GetChar(iCount - 1);
-  if (tc->m_dwStatus == CFX_BreakType::None)
-    tc->m_dwStatus = CFX_BreakType::Piece;
-}
-
-CFX_Char* CFX_RTFBreak::GetLastChar(int32_t index) const {
-  std::vector<CFX_Char>& tca = m_pCurLine->m_LineChars;
-  int32_t iCount = pdfium::CollectionSize<int32_t>(tca);
-  if (index < 0 || index >= iCount)
-    return nullptr;
-
-  int32_t iStart = iCount - 1;
-  while (iStart > -1) {
-    CFX_Char* pTC = &tca[iStart--];
-    if (pTC->m_iCharWidth >= 0 ||
-        pTC->GetCharType() != FX_CHARTYPE_Combination) {
-      if (--index < 0)
-        return pTC;
-    }
-  }
-  return nullptr;
-}
-
-inline FX_CHARTYPE CFX_RTFBreak::GetUnifiedCharType(
-    FX_CHARTYPE chartype) const {
-  return chartype >= FX_CHARTYPE_ArabicAlef ? FX_CHARTYPE_Arabic : chartype;
-}
-
 int32_t CFX_RTFBreak::GetLastPositionedTab() const {
   return m_PositionedTabs.empty() ? m_iLineStart : m_PositionedTabs.back();
 }
@@ -167,7 +132,7 @@ void CFX_RTFBreak::AppendChar_Combination(CFX_Char* pCurChar) {
 
   iCharWidth *= m_iFontSize;
   iCharWidth = iCharWidth * m_iHorizontalScale / 100;
-  CFX_Char* pLastChar = GetLastChar(0);
+  CFX_Char* pLastChar = GetLastChar(0, false, true);
   if (pLastChar && pLastChar->GetCharType() > FX_CHARTYPE_Combination)
     iCharWidth = -iCharWidth;
   else
@@ -224,10 +189,10 @@ CFX_BreakType CFX_RTFBreak::AppendChar_Arabic(CFX_Char* pCurChar) {
   bool bAlef = false;
   if (m_eCharType >= FX_CHARTYPE_ArabicAlef &&
       m_eCharType <= FX_CHARTYPE_ArabicDistortion) {
-    pLastChar = GetLastChar(1);
+    pLastChar = GetLastChar(1, false, true);
     if (pLastChar) {
       m_pCurLine->m_iWidth -= pLastChar->m_iCharWidth;
-      CFX_Char* pPrevChar = GetLastChar(2);
+      CFX_Char* pPrevChar = GetLastChar(2, false, true);
       wForm = pdfium::arabic::GetFormChar(pLastChar, pPrevChar, pCurChar);
       bAlef = (wForm == 0xFEFF &&
                pLastChar->GetCharType() == FX_CHARTYPE_ArabicAlef);
@@ -292,7 +257,7 @@ CFX_BreakType CFX_RTFBreak::EndBreak(CFX_BreakType dwStatus) {
     return m_pCurLine->m_LinePieces.back().m_dwStatus;
   }
 
-  if (HasRTFLine()) {
+  if (HasLine()) {
     if (!m_Line[m_iReadyLineIndex].m_LinePieces.empty()) {
       if (dwStatus != CFX_BreakType::Piece)
         m_Line[m_iReadyLineIndex].m_LinePieces.back().m_dwStatus = dwStatus;
@@ -324,7 +289,7 @@ CFX_BreakType CFX_RTFBreak::EndBreak(CFX_BreakType dwStatus) {
   m_pCurLine = pNextLine;
   m_pCurLine->m_iStart = m_iLineStart;
 
-  CFX_Char* pTC = GetLastChar(0);
+  CFX_Char* pTC = GetLastChar(0, false, true);
   m_eCharType = pTC ? pTC->GetCharType() : FX_CHARTYPE_Unknown;
   return dwStatus;
 }
@@ -698,31 +663,6 @@ void CFX_RTFBreak::SplitTextLine(CFX_BreakLine* pCurLine,
   }
 }
 
-int32_t CFX_RTFBreak::CountBreakPieces() const {
-  return HasRTFLine() ? pdfium::CollectionSize<int32_t>(
-                            m_Line[m_iReadyLineIndex].m_LinePieces)
-                      : 0;
-}
-
-const CFX_BreakPiece* CFX_RTFBreak::GetBreakPieceUnstable(int32_t index) const {
-  if (!HasRTFLine())
-    return nullptr;
-
-  const std::vector<CFX_BreakPiece>& pRTFPieces =
-      m_Line[m_iReadyLineIndex].m_LinePieces;
-  if (!pdfium::IndexInBounds(pRTFPieces, index))
-    return nullptr;
-
-  return &pRTFPieces[index];
-}
-
-void CFX_RTFBreak::ClearBreakPieces() {
-  if (HasRTFLine())
-    m_Line[m_iReadyLineIndex].Clear();
-
-  m_iReadyLineIndex = -1;
-}
-
 int32_t CFX_RTFBreak::GetDisplayPos(const FX_RTFTEXTOBJ* pText,
                                     FXTEXT_CHARPOS* pCharPos,
                                     bool bCharCode) const {
diff --git a/xfa/fgas/layout/fgas_rtfbreak.h b/xfa/fgas/layout/fgas_rtfbreak.h
index 865b17ffb3..7cbd250218 100644
--- a/xfa/fgas/layout/fgas_rtfbreak.h
+++ b/xfa/fgas/layout/fgas_rtfbreak.h
@@ -56,9 +56,6 @@ class CFX_RTFBreak : public CFX_Break {
   void AddPositionedTab(float fTabPos);
 
   CFX_BreakType EndBreak(CFX_BreakType dwStatus);
-  int32_t CountBreakPieces() const;
-  const CFX_BreakPiece* GetBreakPieceUnstable(int32_t index) const;
-  void ClearBreakPieces();
 
   int32_t GetDisplayPos(const FX_RTFTEXTOBJ* pText,
                         FXTEXT_CHARPOS* pCharPos,
@@ -74,10 +71,6 @@ class CFX_RTFBreak : public CFX_Break {
   CFX_BreakType AppendChar_Control(CFX_Char* pCurChar);
   CFX_BreakType AppendChar_Arabic(CFX_Char* pCurChar);
   CFX_BreakType AppendChar_Others(CFX_Char* pCurChar);
-  void SetBreakStatus() override;
-  CFX_Char* GetLastChar(int32_t index) const;
-  bool HasRTFLine() const { return m_iReadyLineIndex >= 0; }
-  FX_CHARTYPE GetUnifiedCharType(FX_CHARTYPE chartype) const;
   int32_t GetLastPositionedTab() const;
   bool GetPositionedTab(int32_t* iTabPos) const;
 
@@ -100,7 +93,6 @@ class CFX_RTFBreak : public CFX_Break {
   std::vector<int32_t> m_PositionedTabs;
   CFX_RTFLineAlignment m_iAlignment;
   CFX_RetainPtr<CFX_Retainable> m_pUserData;
-  uint32_t m_dwIdentity;
 };
 
 #endif  // XFA_FGAS_LAYOUT_FGAS_RTFBREAK_H_
diff --git a/xfa/fgas/layout/fgas_textbreak.cpp b/xfa/fgas/layout/fgas_textbreak.cpp
index 41091700fb..c21d6bba3a 100644
--- a/xfa/fgas/layout/fgas_textbreak.cpp
+++ b/xfa/fgas/layout/fgas_textbreak.cpp
@@ -46,37 +46,6 @@ void CFX_TxtBreak::SetCombWidth(float fCombWidth) {
   m_iCombWidth = FXSYS_round(fCombWidth * 20000.0f);
 }
 
-void CFX_TxtBreak::SetBreakStatus() {
-  int32_t iCount = m_pCurLine->CountChars();
-  if (iCount < 1)
-    return;
-
-  CFX_Char* pTC = m_pCurLine->GetChar(iCount - 1);
-  if (pTC->m_dwStatus == CFX_BreakType::None)
-    pTC->m_dwStatus = CFX_BreakType::Piece;
-}
-CFX_Char* CFX_TxtBreak::GetLastChar(int32_t index, bool bOmitChar) const {
-  std::vector<CFX_Char>& ca = m_pCurLine->m_LineChars;
-  int32_t iCount = pdfium::CollectionSize<int32_t>(ca);
-  if (index < 0 || index >= iCount)
-    return nullptr;
-
-  int32_t iStart = iCount - 1;
-  while (iStart > -1) {
-    CFX_Char* pTC = &ca[iStart--];
-    if (bOmitChar && pTC->GetCharType() == FX_CHARTYPE_Combination)
-      continue;
-    if (--index < 0)
-      return pTC;
-  }
-  return nullptr;
-}
-
-inline FX_CHARTYPE CFX_TxtBreak::GetUnifiedCharType(
-    FX_CHARTYPE chartype) const {
-  return chartype >= FX_CHARTYPE_ArabicAlef ? FX_CHARTYPE_Arabic : chartype;
-}
-
 void CFX_TxtBreak::AppendChar_PageLoad(CFX_Char* pCurChar, uint32_t dwProps) {
   pCurChar->m_dwStatus = CFX_BreakType::None;
   pCurChar->m_dwCharStyles = m_iAlignment | (1 << 8);
@@ -91,7 +60,7 @@ void CFX_TxtBreak::AppendChar_Combination(CFX_Char* pCurChar) {
     iCharWidth = m_iCombWidth;
   } else {
     wForm = wch;
-    CFX_Char* pLastChar = GetLastChar(0, false);
+    CFX_Char* pLastChar = GetLastChar(0, false, false);
     if (pLastChar &&
         (pLastChar->m_dwCharStyles & FX_TXTCHARSTYLE_ArabicShadda) == 0) {
       bool bShadda = false;
@@ -162,13 +131,13 @@ CFX_BreakType CFX_TxtBreak::AppendChar_Arabic(CFX_Char* pCurChar) {
   bool bAlef = false;
   if (!m_bCombText && m_eCharType >= FX_CHARTYPE_ArabicAlef &&
       m_eCharType <= FX_CHARTYPE_ArabicDistortion) {
-    pLastChar = GetLastChar(1);
+    pLastChar = GetLastChar(1, true, false);
     if (pLastChar) {
       iCharWidth = pLastChar->m_iCharWidth;
       if (iCharWidth > 0)
         iLineWidth -= iCharWidth;
 
-      CFX_Char* pPrevChar = GetLastChar(2);
+      CFX_Char* pPrevChar = GetLastChar(2, true, false);
       wForm = pdfium::arabic::GetFormChar(pLastChar, pPrevChar, pCurChar);
       bAlef = (wForm == 0xFEFF &&
                pLastChar->GetCharType() == FX_CHARTYPE_ArabicAlef);
@@ -514,7 +483,7 @@ CFX_BreakType CFX_TxtBreak::EndBreak(CFX_BreakType dwStatus) {
     return m_pCurLine->m_LinePieces.back().m_dwStatus;
   }
 
-  if (HasTxtLine()) {
+  if (HasLine()) {
     if (!m_Line[m_iReadyLineIndex].m_LinePieces.empty()) {
       if (dwStatus != CFX_BreakType::Piece)
         m_Line[m_iReadyLineIndex].m_LinePieces.back().m_dwStatus = dwStatus;
@@ -542,7 +511,7 @@ CFX_BreakType CFX_TxtBreak::EndBreak(CFX_BreakType dwStatus) {
   }
 
   m_pCurLine = pNextLine;
-  CFX_Char* pTC = GetLastChar(0, false);
+  CFX_Char* pTC = GetLastChar(0, false, false);
   m_eCharType = pTC ? pTC->GetCharType() : FX_CHARTYPE_Unknown;
 
   return dwStatus;
@@ -682,28 +651,6 @@ void CFX_TxtBreak::SplitTextLine(CFX_BreakLine* pCurLine,
   pNextLine->m_iWidth = iWidth;
 }
 
-int32_t CFX_TxtBreak::CountBreakPieces() const {
-  return HasTxtLine() ? pdfium::CollectionSize<int32_t>(
-                            m_Line[m_iReadyLineIndex].m_LinePieces)
-                      : 0;
-}
-
-const CFX_BreakPiece* CFX_TxtBreak::GetBreakPiece(int32_t index) const {
-  if (!HasTxtLine())
-    return nullptr;
-
-  if (!pdfium::IndexInBounds(m_Line[m_iReadyLineIndex].m_LinePieces, index))
-    return nullptr;
-
-  return &m_Line[m_iReadyLineIndex].m_LinePieces[index];
-}
-
-void CFX_TxtBreak::ClearBreakPieces() {
-  if (HasTxtLine())
-    m_Line[m_iReadyLineIndex].Clear();
-  m_iReadyLineIndex = -1;
-}
-
 struct FX_FORMCHAR {
   uint16_t wch;
   uint16_t wForm;
diff --git a/xfa/fgas/layout/fgas_textbreak.h b/xfa/fgas/layout/fgas_textbreak.h
index 439a7ee15b..81aaf58219 100644
--- a/xfa/fgas/layout/fgas_textbreak.h
+++ b/xfa/fgas/layout/fgas_textbreak.h
@@ -64,9 +64,6 @@ class CFX_TxtBreak : public CFX_Break {
   void SetAlignment(int32_t iAlignment);
   void SetCombWidth(float fCombWidth);
   CFX_BreakType EndBreak(CFX_BreakType dwStatus);
-  int32_t CountBreakPieces() const;
-  const CFX_BreakPiece* GetBreakPiece(int32_t index) const;
-  void ClearBreakPieces();
 
   int32_t GetDisplayPos(const FX_TXTRUN* pTxtRun,
                         FXTEXT_CHARPOS* pCharPos,
@@ -84,10 +81,6 @@ class CFX_TxtBreak : public CFX_Break {
   CFX_BreakType AppendChar_Arabic(CFX_Char* pCurChar);
   CFX_BreakType AppendChar_Others(CFX_Char* pCurChar);
 
-  void SetBreakStatus() override;
-  CFX_Char* GetLastChar(int32_t index, bool bOmitChar = true) const;
-  bool HasTxtLine() const { return m_iReadyLineIndex >= 0; }
-  FX_CHARTYPE GetUnifiedCharType(FX_CHARTYPE dwType) const;
   void ResetContextCharStyles();
   bool EndBreak_SplitLine(CFX_BreakLine* pNextLine, bool bAllChars);
   void EndBreak_BidiLine(std::deque<FX_TPO>* tpos, CFX_BreakType dwStatus);