DiffTextBuffer: Create Unicode temp files (makes EscapeControlChars() obsolete)

While transcoding UTF-16 input to a relaxed superset of UTF-8 so diffutils can make sense of it, prepare_text_end() (located in diffutils/src/io.c) translates NUL chars into 2 byte sequences so as to not create bogus string terminators. That was also the single remaining purpose which EscapeControlChars() was still serving after EditLib was enabled to show control characters in hexadecimal notation with a rounded rectangle around them. Now that the NUL char handling built into prepare_text_end() applies, EscapeControlChars() is no longer needed.

diff --git a/Src/DiffTextBuffer.cpp b/Src/DiffTextBuffer.cpp
index f90a091..e1f9905 100644 (file)
--- a/Src/DiffTextBuffer.cpp
+++ b/Src/DiffTextBuffer.cpp
@@ -29,7 +29,6 @@ using Poco::Exception;
 #endif
 
 static bool IsTextFileStylePure(const UniMemFile::txtstats & stats);
-static void EscapeControlChars(String &s);
 static CRLFSTYLE GetTextFileStyle(const UniMemFile::txtstats & stats);
 
 /**
@@ -50,54 +49,6 @@ static bool IsTextFileStylePure(const UniMemFile::txtstats & stats)
 }
 
 /**
- * @brief Escape control characters.
- * @param [in,out] s Line of text excluding eol chars.
- *
- * @note Escape sequences follow the pattern
- * (leadin character, high nibble, low nibble, leadout character).
- * The leadin character is '\x0F'. The leadout character is a backslash.
- */
-static void EscapeControlChars(String &s)
-{
-       // Compute buffer length required for escaping
-       size_t n = s.length();
-       LPCTSTR q = s.c_str();
-       size_t i = n;
-       while (i != 0)
-       {
-               TCHAR c = q[--i];
-               // Is it a control character in the range 0..31 except TAB?
-               if (!(c & ~_T('\x1F')) && c != _T('\t'))
-               {
-                       n += 3; // Need 3 extra characters to escape
-               }
-       }
-       // Reallocate accordingly
-       i = s.length();
-       s.reserve(n + 1);
-       s.resize(n + 1);
-       LPTSTR p = &s[0];
-       // Copy/translate characters starting at end of string
-       while (i != 0)
-       {
-               TCHAR c = p[--i];
-               // Is it a control character in the range 0..31 except TAB?
-               if (!(c & ~_T('\x1F')) && c != _T('\t'))
-               {
-                       p[n - 4 + 1] = _T("0123456789ABCDEF")[(c >> 4) & 0xf];
-                       p[n - 4 + 2] = _T("0123456789ABCDEF")[c & 0xf];
-                       // Replace terminating zero with leadout character
-                       p[n - 1] = _T('\\');
-                       // Prepare to replace 1st hex digit with leadin character
-                       c = _T('\x0F');
-                       n -= 3;
-               }
-               p[--n] = c;
-       }
-       s.resize(s.length() - 1);
-}
-
-/**
  * @brief Get file's EOL type.
  * @param [in] stats File's text stats.
  * @return EOL type.
@@ -507,18 +458,20 @@ int CDiffTextBuffer::SaveToFile (const String& pszFileName,
        bool bSaveSuccess = false;
 
        UniStdioFile file;
-       file.SetUnicoding(m_encoding.m_unicoding);
-       file.SetBom(m_encoding.m_bom);
-       file.SetCodepage(m_encoding.m_codepage);
 
        String sIntermediateFilename; // used when !bTempFile
 
        if (bTempFile)
        {
+               file.SetUnicoding(ucr::UCS2LE);
+               file.SetBom(true);
                bOpenSuccess = !!file.OpenCreate(pszFileName);
        }
        else
        {
+               file.SetUnicoding(m_encoding.m_unicoding);
+               file.SetBom(m_encoding.m_bom);
+               file.SetCodepage(m_encoding.m_codepage);
                sIntermediateFilename = env::GetTemporaryFileName(m_strTempPath,
                        _T("MRG_"), nullptr);
                if (sIntermediateFilename.empty())
@@ -564,9 +517,6 @@ int CDiffTextBuffer::SaveToFile (const String& pszFileName,
                else
                        sLine = _T("");
 
-               if (bTempFile)
-                       EscapeControlChars(sLine);
-
                // last real line ?
                if (line == lastRealLine || lastRealLine == -1 )
                {