Add much-needed Find() method for CFX_*StringC

BUG=pdfium:493

Review-Url: https://codereview.chromium.org/1968233002

11 files changed, 103 insertions, 53 deletions

diff --git a/core/fpdfapi/fpdf_page/fpdf_page_parser.cpp b/core/fpdfapi/fpdf_page/fpdf_page_parser.cpp
index a13fb5021d..da1355724f 100644
--- a/core/fpdfapi/fpdf_page/fpdf_page_parser.cpp
+++ b/core/fpdfapi/fpdf_page/fpdf_page_parser.cpp
@@ -271,18 +271,18 @@ int CPDF_StreamContentParser::GetNextParamPos() {
 }
 
 void CPDF_StreamContentParser::AddNameParam(const FX_CHAR* name, int len) {
+  CFX_ByteStringC bsName(name, len);
   int index = GetNextParamPos();
   if (len > 32) {
     m_ParamBuf[index].m_Type = ContentParam::OBJECT;
-    m_ParamBuf[index].m_pObject =
-        new CPDF_Name(PDF_NameDecode(CFX_ByteStringC(name, len)));
+    m_ParamBuf[index].m_pObject = new CPDF_Name(PDF_NameDecode(bsName));
   } else {
     m_ParamBuf[index].m_Type = ContentParam::NAME;
-    if (!FXSYS_memchr(name, '#', len)) {
+    if (bsName.Find('#') == -1) {
       FXSYS_memcpy(m_ParamBuf[index].m_Name.m_Buffer, name, len);
       m_ParamBuf[index].m_Name.m_Len = len;
     } else {
-      CFX_ByteString str = PDF_NameDecode(CFX_ByteStringC(name, len));
+      CFX_ByteString str = PDF_NameDecode(bsName);
       FXSYS_memcpy(m_ParamBuf[index].m_Name.m_Buffer, str.c_str(),
                    str.GetLength());
       m_ParamBuf[index].m_Name.m_Len = str.GetLength();
diff --git a/core/fpdfapi/fpdf_parser/fpdf_parser_utility.cpp b/core/fpdfapi/fpdf_parser/fpdf_parser_utility.cpp
index 466ccaffa5..03975149b0 100644
--- a/core/fpdfapi/fpdf_parser/fpdf_parser_utility.cpp
+++ b/core/fpdfapi/fpdf_parser/fpdf_parser_utility.cpp
@@ -89,9 +89,9 @@ int32_t GetDirectInteger(CPDF_Dictionary* pDict, const CFX_ByteStringC& key) {
 CFX_ByteString PDF_NameDecode(const CFX_ByteStringC& bstr) {
   int size = bstr.GetLength();
   const FX_CHAR* pSrc = bstr.c_str();
-  if (!FXSYS_memchr(pSrc, '#', size)) {
+  if (bstr.Find('#') == -1)
     return bstr;
-  }
+
   CFX_ByteString result;
   FX_CHAR* pDestStart = result.GetBuffer(size);
   FX_CHAR* pDest = pDestStart;
@@ -109,9 +109,8 @@ CFX_ByteString PDF_NameDecode(const CFX_ByteStringC& bstr) {
 }
 
 CFX_ByteString PDF_NameDecode(const CFX_ByteString& orig) {
-  if (!FXSYS_memchr(orig.c_str(), '#', orig.GetLength())) {
+  if (orig.Find('#') == -1)
     return orig;
-  }
   return PDF_NameDecode(orig.AsStringC());
 }
 
diff --git a/core/fpdftext/fpdf_text_int.cpp b/core/fpdftext/fpdf_text_int.cpp
index 8e5e75853e..f843da3061 100644
--- a/core/fpdftext/fpdf_text_int.cpp
+++ b/core/fpdftext/fpdf_text_int.cpp
@@ -2147,7 +2147,7 @@ FX_BOOL CPDF_TextPageFind::ExtractSubString(CFX_WideString& rString,
     return FALSE;
   }
   while (iSubString--) {
-    lpszFullString = FXSYS_wcschr(lpszFullString, chSep);
+    lpszFullString = wcschr(lpszFullString, chSep);
     if (!lpszFullString) {
       rString.clear();
       return FALSE;
@@ -2157,7 +2157,7 @@ FX_BOOL CPDF_TextPageFind::ExtractSubString(CFX_WideString& rString,
       lpszFullString++;
     }
   }
-  const FX_WCHAR* lpchEnd = FXSYS_wcschr(lpszFullString, chSep);
+  const FX_WCHAR* lpchEnd = wcschr(lpszFullString, chSep);
   int nLen = lpchEnd ? (int)(lpchEnd - lpszFullString)
                      : (int)FXSYS_wcslen(lpszFullString);
   ASSERT(nLen >= 0);
diff --git a/core/fxcrt/fx_basic_bstring.cpp b/core/fxcrt/fx_basic_bstring.cpp
index cd7b616fbe..18e240bc84 100644
--- a/core/fxcrt/fx_basic_bstring.cpp
+++ b/core/fxcrt/fx_basic_bstring.cpp
@@ -673,11 +673,12 @@ FX_STRSIZE CFX_ByteString::Find(FX_CHAR ch, FX_STRSIZE nStart) const {
   if (!m_pData)
     return -1;
 
-  if (nStart >= m_pData->m_nDataLength)
+  if (nStart < 0 || nStart >= m_pData->m_nDataLength)
     return -1;
 
-  const FX_CHAR* pStr = FXSYS_strchr(m_pData->m_String + nStart, ch);
-  return pStr ? (int)(pStr - m_pData->m_String) : -1;
+  const FX_CHAR* pStr = static_cast<const FX_CHAR*>(
+      memchr(m_pData->m_String + nStart, ch, m_pData->m_nDataLength - nStart));
+  return pStr ? pStr - m_pData->m_String : -1;
 }
 
 FX_STRSIZE CFX_ByteString::ReverseFind(FX_CHAR ch) const {
diff --git a/core/fxcrt/fx_basic_bstring_unittest.cpp b/core/fxcrt/fx_basic_bstring_unittest.cpp
index 087c264940..fe98b53b45 100644
--- a/core/fxcrt/fx_basic_bstring_unittest.cpp
+++ b/core/fxcrt/fx_basic_bstring_unittest.cpp
@@ -802,6 +802,31 @@ TEST(fxcrt, ByteStringCGetID) {
   EXPECT_EQ(0u, longer_string.GetID(-1000000));
 }
 
+TEST(fxcrt, ByteStringCFind) {
+  CFX_ByteStringC null_string;
+  EXPECT_EQ(-1, null_string.Find('a'));
+  EXPECT_EQ(-1, null_string.Find(0));
+
+  CFX_ByteStringC empty_string("");
+  EXPECT_EQ(-1, empty_string.Find('a'));
+  EXPECT_EQ(-1, empty_string.Find(0));
+
+  CFX_ByteStringC single_string("a");
+  EXPECT_EQ(0, single_string.Find('a'));
+  EXPECT_EQ(-1, single_string.Find('b'));
+  EXPECT_EQ(-1, single_string.Find(0));
+
+  CFX_ByteStringC longer_string("abccc");
+  EXPECT_EQ(0, longer_string.Find('a'));
+  EXPECT_EQ(2, longer_string.Find('c'));
+  EXPECT_EQ(-1, longer_string.Find(0));
+
+  CFX_ByteStringC hibyte_string(
+      "ab\x8c"
+      "def");
+  EXPECT_EQ(2, hibyte_string.Find('\x8c'));
+}
+
 TEST(fxcrt, ByteStringCMid) {
   CFX_ByteStringC null_string;
   EXPECT_EQ(null_string, null_string.Mid(0, 1));
diff --git a/core/fxcrt/fx_basic_util.cpp b/core/fxcrt/fx_basic_util.cpp
index 1cf966589d..f69185653f 100644
--- a/core/fxcrt/fx_basic_util.cpp
+++ b/core/fxcrt/fx_basic_util.cpp
@@ -95,7 +95,7 @@ void CFX_PrivateData::ClearAll() {
   m_DataList.RemoveAll();
 }
 void FX_atonum(const CFX_ByteStringC& strc, FX_BOOL& bInteger, void* pData) {
-  if (!FXSYS_memchr(strc.raw_str(), '.', strc.GetLength())) {
+  if (strc.Find('.') == -1) {
     bInteger = TRUE;
     int cc = 0, integer = 0;
     const FX_CHAR* str = strc.c_str();
diff --git a/core/fxcrt/fx_basic_wstring.cpp b/core/fxcrt/fx_basic_wstring.cpp
index 354bb5014a..03cb0599d5 100644
--- a/core/fxcrt/fx_basic_wstring.cpp
+++ b/core/fxcrt/fx_basic_wstring.cpp
@@ -638,11 +638,12 @@ FX_STRSIZE CFX_WideString::Find(FX_WCHAR ch, FX_STRSIZE nStart) const {
   if (!m_pData)
     return -1;
 
-  if (nStart >= m_pData->m_nDataLength)
+  if (nStart < 0 || nStart >= m_pData->m_nDataLength)
     return -1;
 
-  const FX_WCHAR* pStr = FXSYS_wcschr(m_pData->m_String + nStart, ch);
-  return pStr ? (int)(pStr - m_pData->m_String) : -1;
+  const FX_WCHAR* pStr =
+      wmemchr(m_pData->m_String + nStart, ch, m_pData->m_nDataLength - nStart);
+  return pStr ? pStr - m_pData->m_String : -1;
 }
 
 FX_STRSIZE CFX_WideString::Find(const CFX_WideStringC& pSub,
@@ -858,19 +859,13 @@ FX_STRSIZE CFX_WideString::WStringLength(const unsigned short* str) {
 }
 
 void CFX_WideString::TrimRight(const CFX_WideStringC& pTargets) {
-  if (!m_pData || pTargets.IsEmpty()) {
+  if (IsEmpty() || pTargets.IsEmpty())
     return;
-  }
+
   FX_STRSIZE pos = GetLength();
-  if (pos < 1) {
-    return;
-  }
-  while (pos) {
-    if (!FXSYS_wcschr(pTargets.c_str(), m_pData->m_String[pos - 1])) {
-      break;
-    }
+  while (pos && pTargets.Find(m_pData->m_String[pos - 1]) != -1)
     pos--;
-  }
+
   if (pos < m_pData->m_nDataLength) {
     ReallocBeforeWrite(m_pData->m_nDataLength);
     m_pData->m_String[pos] = 0;
diff --git a/core/fxcrt/fx_basic_wstring_unittest.cpp b/core/fxcrt/fx_basic_wstring_unittest.cpp
index 02281afd36..4b6f4a5800 100644
--- a/core/fxcrt/fx_basic_wstring_unittest.cpp
+++ b/core/fxcrt/fx_basic_wstring_unittest.cpp
@@ -831,6 +831,31 @@ TEST(fxcrt, WideStringCOperatorNE) {
   EXPECT_TRUE(c_string3 != wide_string_c);
 }
 
+TEST(fxcrt, WideStringCFind) {
+  CFX_WideStringC null_string;
+  EXPECT_EQ(-1, null_string.Find(L'a'));
+  EXPECT_EQ(-1, null_string.Find(0));
+
+  CFX_WideStringC empty_string(L"");
+  EXPECT_EQ(-1, empty_string.Find(L'a'));
+  EXPECT_EQ(-1, empty_string.Find(0));
+
+  CFX_WideStringC single_string(L"a");
+  EXPECT_EQ(0, single_string.Find(L'a'));
+  EXPECT_EQ(-1, single_string.Find(L'b'));
+  EXPECT_EQ(-1, single_string.Find(0));
+
+  CFX_WideStringC longer_string(L"abccc");
+  EXPECT_EQ(0, longer_string.Find(L'a'));
+  EXPECT_EQ(2, longer_string.Find(L'c'));
+  EXPECT_EQ(-1, longer_string.Find(0));
+
+  CFX_WideStringC hibyte_string(
+      L"ab\xff08"
+      L"def");
+  EXPECT_EQ(2, hibyte_string.Find(L'\xff08'));
+}
+
 TEST(fxcrt, WideStringFormatWidth) {
   {
     CFX_WideString str;
diff --git a/core/fxcrt/include/fx_string.h b/core/fxcrt/include/fx_string.h
index 66a8039a8d..fbfb91f759 100644
--- a/core/fxcrt/include/fx_string.h
+++ b/core/fxcrt/include/fx_string.h
@@ -100,6 +100,12 @@ class CFX_ByteStringC {
 
   uint8_t GetAt(FX_STRSIZE index) const { return m_Ptr[index]; }
 
+  FX_STRSIZE Find(FX_CHAR ch) const {
+    const uint8_t* found =
+        static_cast<const uint8_t*>(memchr(m_Ptr, ch, m_Length));
+    return found ? found - m_Ptr : -1;
+  }
+
   CFX_ByteStringC Mid(FX_STRSIZE index, FX_STRSIZE count = -1) const {
     if (index < 0) {
       index = 0;
@@ -417,6 +423,12 @@ class CFX_WideStringC {
     return CFX_WideStringC(m_Ptr, count);
   }
 
+  FX_STRSIZE Find(FX_WCHAR ch) const {
+    const FX_WCHAR* found =
+        static_cast<const FX_WCHAR*>(wmemchr(m_Ptr, ch, m_Length));
+    return found ? found - m_Ptr : -1;
+  }
+
   CFX_WideStringC Mid(FX_STRSIZE index, FX_STRSIZE count = -1) const {
     if (index < 0) {
       index = 0;
diff --git a/core/fxcrt/include/fx_system.h b/core/fxcrt/include/fx_system.h
index 795a0c2794..b4659d14e4 100644
--- a/core/fxcrt/include/fx_system.h
+++ b/core/fxcrt/include/fx_system.h
@@ -123,7 +123,6 @@ void FXSYS_vsnprintf(char* str, size_t size, const char* fmt, va_list ap);
 
 #define FXSYS_sprintf DO_NOT_USE_SPRINTF_DIE_DIE_DIE
 #define FXSYS_vsprintf DO_NOT_USE_VSPRINTF_DIE_DIE_DIE
-#define FXSYS_strchr strchr
 #define FXSYS_strncmp strncmp
 #define FXSYS_strcmp strcmp
 #define FXSYS_strcpy strcpy
@@ -166,7 +165,6 @@ extern "C" {
 #endif
 
 #define FXSYS_wcscmp wcscmp
-#define FXSYS_wcschr wcschr
 #define FXSYS_wcsstr wcsstr
 #define FXSYS_wcsncmp wcsncmp
 #define FXSYS_vswprintf vswprintf
@@ -176,7 +174,6 @@ extern "C" {
 #define FXSYS_memcpy memcpy
 #define FXSYS_memmove memmove
 #define FXSYS_memset memset
-#define FXSYS_memchr memchr
 #define FXSYS_qsort qsort
 #define FXSYS_bsearch bsearch
 
diff --git a/xfa/fgas/localization/fgas_locale.cpp b/xfa/fgas/localization/fgas_locale.cpp
index 521b53536e..2336b20ee3 100644
--- a/xfa/fgas/localization/fgas_locale.cpp
+++ b/xfa/fgas/localization/fgas_locale.cpp
@@ -60,21 +60,11 @@ static const FX_WCHAR gs_wsTimeSymbols[] = L"hHkKMSFAzZ";
 static const FX_WCHAR gs_wsDateSymbols[] = L"DJMEeGgYwW";
 static const FX_WCHAR gs_wsConstChars[] = L",-:/. ";
 
-static FX_STRSIZE FX_Local_Find(const CFX_WideStringC& wsSymbols,
-                                FX_WCHAR ch,
-                                FX_STRSIZE nStart = 0) {
-  FX_STRSIZE nLength = wsSymbols.GetLength();
-  if (nLength < 1 || nStart > nLength) {
-    return -1;
-  }
-  const FX_WCHAR* lpsz =
-      (const FX_WCHAR*)FXSYS_wcschr(wsSymbols.c_str() + nStart, ch);
-  return (lpsz == NULL) ? -1 : (FX_STRSIZE)(lpsz - wsSymbols.c_str());
-}
 static const FX_WCHAR* const gs_LocalNumberSymbols[] = {
     L"decimal", L"grouping",       L"percent",      L"minus",
     L"zero",    L"currencySymbol", L"currencyName",
 };
+
 IFX_Locale* IFX_Locale::Create(CXML_Element* pLocaleData) {
   return new CFX_Locale(pLocaleData);
 }
@@ -575,10 +565,11 @@ FX_LOCALECATEGORY CFX_FormatString::GetCategory(
   int32_t iLenf = wsPattern.GetLength();
   const FX_WCHAR* pStr = wsPattern.c_str();
   FX_BOOL bBraceOpen = FALSE;
+  CFX_WideStringC wsConstChars(gs_wsConstChars);
   while (ccf < iLenf) {
     if (pStr[ccf] == '\'') {
       FX_GetLiteralText(pStr, ccf, iLenf);
-    } else if (!bBraceOpen && FX_Local_Find(gs_wsConstChars, pStr[ccf]) < 0) {
+    } else if (!bBraceOpen && wsConstChars.Find(pStr[ccf]) == -1) {
       CFX_WideString wsCategory(pStr[ccf]);
       ccf++;
       while (TRUE) {
@@ -662,12 +653,13 @@ IFX_Locale* CFX_FormatString::GetTextFormat(const CFX_WideString& wsPattern,
   int32_t iLenf = wsPattern.GetLength();
   const FX_WCHAR* pStr = wsPattern.c_str();
   FX_BOOL bBrackOpen = FALSE;
+  CFX_WideStringC wsConstChars(gs_wsConstChars);
   while (ccf < iLenf) {
     if (pStr[ccf] == '\'') {
       int32_t iCurChar = ccf;
       FX_GetLiteralText(pStr, ccf, iLenf);
       wsPurgePattern += CFX_WideStringC(pStr + iCurChar, ccf - iCurChar + 1);
-    } else if (!bBrackOpen && FX_Local_Find(gs_wsConstChars, pStr[ccf]) < 0) {
+    } else if (!bBrackOpen && wsConstChars.Find(pStr[ccf]) == -1) {
       CFX_WideString wsSearchCategory(pStr[ccf]);
       ccf++;
       while (ccf < iLenf && pStr[ccf] != '{' && pStr[ccf] != '.' &&
@@ -719,12 +711,13 @@ IFX_Locale* CFX_FormatString::GetNumericFormat(const CFX_WideString& wsPattern,
   const FX_WCHAR* pStr = wsPattern.c_str();
   FX_BOOL bFindDot = FALSE;
   FX_BOOL bBrackOpen = FALSE;
+  CFX_WideStringC wsConstChars(gs_wsConstChars);
   while (ccf < iLenf) {
     if (pStr[ccf] == '\'') {
       int32_t iCurChar = ccf;
       FX_GetLiteralText(pStr, ccf, iLenf);
       wsPurgePattern += CFX_WideStringC(pStr + iCurChar, ccf - iCurChar + 1);
-    } else if (!bBrackOpen && FX_Local_Find(gs_wsConstChars, pStr[ccf]) < 0) {
+    } else if (!bBrackOpen && wsConstChars.Find(pStr[ccf]) == -1) {
       CFX_WideString wsCategory(pStr[ccf]);
       ccf++;
       while (ccf < iLenf && pStr[ccf] != '{' && pStr[ccf] != '.' &&
@@ -2153,13 +2146,14 @@ FX_DATETIMETYPE CFX_FormatString::GetDateTimeFormat(
   const FX_WCHAR* pStr = wsPattern.c_str();
   int32_t iFindCategory = 0;
   FX_BOOL bBraceOpen = FALSE;
+  CFX_WideStringC wsConstChars(gs_wsConstChars);
   while (ccf < iLenf) {
     if (pStr[ccf] == '\'') {
       int32_t iCurChar = ccf;
       FX_GetLiteralText(pStr, ccf, iLenf);
       wsTempPattern += CFX_WideStringC(pStr + iCurChar, ccf - iCurChar + 1);
     } else if (!bBraceOpen && iFindCategory != 3 &&
-               FX_Local_Find(gs_wsConstChars, pStr[ccf]) < 0) {
+               wsConstChars.Find(pStr[ccf]) == -1) {
       CFX_WideString wsCategory(pStr[ccf]);
       ccf++;
       while (ccf < iLenf && pStr[ccf] != '{' && pStr[ccf] != '.' &&
@@ -2290,6 +2284,7 @@ static FX_BOOL FX_ParseLocaleDate(const CFX_WideString& wsDate,
   int32_t len = wsDate.GetLength();
   const FX_WCHAR* strf = wsDatePattern.c_str();
   int32_t lenf = wsDatePattern.GetLength();
+  CFX_WideStringC wsDateSymbols(gs_wsDateSymbols);
   while (cc < len && ccf < lenf) {
     if (strf[ccf] == '\'') {
       CFX_WideString wsLiteral = FX_GetLiteralText(strf, ccf, lenf);
@@ -2301,10 +2296,9 @@ static FX_BOOL FX_ParseLocaleDate(const CFX_WideString& wsDate,
       cc += iLiteralLen;
       ccf++;
       continue;
-    } else if (FX_Local_Find(gs_wsDateSymbols, strf[ccf]) < 0) {
-      if (strf[ccf] != str[cc]) {
+    } else if (wsDateSymbols.Find(strf[ccf]) == -1) {
+      if (strf[ccf] != str[cc])
         return FALSE;
-      }
       cc++;
       ccf++;
       continue;
@@ -2509,6 +2503,7 @@ static FX_BOOL FX_ParseLocaleTime(const CFX_WideString& wsTime,
   int lenf = wsTimePattern.GetLength();
   FX_BOOL bHasA = FALSE;
   FX_BOOL bPM = FALSE;
+  CFX_WideStringC wsTimeSymbols(gs_wsTimeSymbols);
   while (cc < len && ccf < lenf) {
     if (strf[ccf] == '\'') {
       CFX_WideString wsLiteral = FX_GetLiteralText(strf, ccf, lenf);
@@ -2520,10 +2515,9 @@ static FX_BOOL FX_ParseLocaleTime(const CFX_WideString& wsTime,
       cc += iLiteralLen;
       ccf++;
       continue;
-    } else if (FX_Local_Find(gs_wsTimeSymbols, strf[ccf]) == -1) {
-      if (strf[ccf] != str[cc]) {
+    } else if (wsTimeSymbols.Find(strf[ccf]) == -1) {
+      if (strf[ccf] != str[cc])
         return FALSE;
-      }
       cc++;
       ccf++;
       continue;
@@ -3947,12 +3941,13 @@ static FX_BOOL FX_DateFormat(const CFX_WideString& wsDatePattern,
   int32_t ccf = 0;
   const FX_WCHAR* strf = wsDatePattern.c_str();
   int32_t lenf = wsDatePattern.GetLength();
+  CFX_WideStringC wsDateSymbols(gs_wsDateSymbols);
   while (ccf < lenf) {
     if (strf[ccf] == '\'') {
       wsResult += FX_GetLiteralText(strf, ccf, lenf);
       ccf++;
       continue;
-    } else if (FX_Local_Find(gs_wsDateSymbols, strf[ccf]) < 0) {
+    } else if (wsDateSymbols.Find(strf[ccf]) == -1) {
       wsResult += strf[ccf++];
       continue;
     }
@@ -4073,12 +4068,13 @@ static FX_BOOL FX_TimeFormat(const CFX_WideString& wsTimePattern,
       bPM = TRUE;
     }
   }
+  CFX_WideStringC wsTimeSymbols(gs_wsTimeSymbols);
   while (ccf < lenf) {
     if (strf[ccf] == '\'') {
       wsResult += FX_GetLiteralText(strf, ccf, lenf);
       ccf++;
       continue;
-    } else if (FX_Local_Find(gs_wsTimeSymbols, strf[ccf]) < 0) {
+    } else if (wsTimeSymbols.Find(strf[ccf]) == -1) {
       wsResult += strf[ccf++];
       continue;
     }