CPP/Common/UTFConvert.h

// Common/UTFConvert.h

#ifndef ZIP7_INC_COMMON_UTF_CONVERT_H
#define ZIP7_INC_COMMON_UTF_CONVERT_H

#include "MyBuffer.h"
#include "MyString.h"

struct CUtf8Check
{
  // Byte MaxByte;     // in original src stream
  bool NonUtf;
  bool ZeroChar;
  bool SingleSurrogate;
  bool Escape;
  bool Truncated;
  UInt32 MaxHighPoint;  // only for points >= 0x80

  CUtf8Check() { Clear(); }

  void Clear()
  {
    // MaxByte = 0;
    NonUtf = false;
    ZeroChar = false;
    SingleSurrogate = false;
    Escape = false;
    Truncated = false;
    MaxHighPoint = 0;
  }

  void Update(const CUtf8Check &c)
  {
    if (c.NonUtf) NonUtf = true;
    if (c.ZeroChar) ZeroChar = true;
    if (c.SingleSurrogate) SingleSurrogate = true;
    if (c.Escape) Escape = true;
    if (c.Truncated) Truncated = true;
    if (MaxHighPoint < c.MaxHighPoint) MaxHighPoint = c.MaxHighPoint;
  }

  void PrintStatus(AString &s) const
  {
    s.Empty();

    // s.Add_OptSpaced("MaxByte=");
    // s.Add_UInt32(MaxByte);

    if (NonUtf)          s.Add_OptSpaced("non-UTF8");
    if (ZeroChar)        s.Add_OptSpaced("ZeroChar");
    if (SingleSurrogate) s.Add_OptSpaced("SingleSurrogate");
    if (Escape)          s.Add_OptSpaced("Escape");
    if (Truncated)       s.Add_OptSpaced("Truncated");

    if (MaxHighPoint != 0)
    {
      s.Add_OptSpaced("MaxUnicode=");
      s.Add_UInt32(MaxHighPoint);
    }
  }


  bool IsOK(bool allowReduced = false) const
  {
    if (NonUtf || SingleSurrogate || ZeroChar)
      return false;
    if (MaxHighPoint >= 0x110000)
      return false;
    if (Truncated && !allowReduced)
      return false;
    return true;
  }

  // it checks full buffer as specified in (size) and it doesn't stop on zero char
  void Check_Buf(const char *src, size_t size) throw();

  void Check_AString(const AString &s) throw()
  {
    Check_Buf(s.Ptr(), s.Len());
  }
};

/*
if (allowReduced == false) - all UTF-8 character sequences must be finished.
if (allowReduced == true)  - it allows truncated last character-Utf8-sequence
*/

bool Check_UTF8_Buf(const char *src, size_t size, bool allowReduced) throw();
bool CheckUTF8_AString(const AString &s) throw();

#define Z7_UTF_FLAG_FROM_UTF8_SURROGATE_ERROR    (1 << 0)
#define Z7_UTF_FLAG_FROM_UTF8_USE_ESCAPE         (1 << 1)
#define Z7_UTF_FLAG_FROM_UTF8_BMP_ESCAPE_CONVERT (1 << 2)

/*
Z7_UTF_FLAG_FROM_UTF8_SURROGATE_ERROR

   if (flag is NOT set)
   {
     it processes SINGLE-SURROGATE-8 as valid Unicode point.
     it converts  SINGLE-SURROGATE-8 to SINGLE-SURROGATE-16
     Note: some sequencies of two SINGLE-SURROGATE-8 points
           will generate correct SURROGATE-16-PAIR, and
           that SURROGATE-16-PAIR later will be converted to correct
           UTF8-SURROGATE-21 point. So we don't restore original
           STR-8 sequence in that case.
   }
   
   if (flag is set)
   {
     if (Z7_UTF_FLAG_FROM_UTF8_USE_ESCAPE is defined)
        it generates ESCAPE for SINGLE-SURROGATE-8,
     if (Z7_UTF_FLAG_FROM_UTF8_USE_ESCAPE is not defined)
        it generates U+fffd for SINGLE-SURROGATE-8,
   }


Z7_UTF_FLAG_FROM_UTF8_USE_ESCAPE

   if (flag is NOT set)
     it generates (U+fffd) code for non-UTF-8 (invalid) characters

   if (flag is set)
   {
     It generates (ESCAPE) codes for NON-UTF-8 (invalid) characters.
     And later we can restore original UTF-8-RAW characters from (ESCAPE-16-21) codes.
   }

Z7_UTF_FLAG_FROM_UTF8_BMP_ESCAPE_CONVERT

   if (flag is NOT set)
   {
     it process ESCAPE-8 points as another Unicode points.
     In Linux: ESCAPE-16 will mean two different ESCAPE-8 seqences,
       so we need HIGH-ESCAPE-PLANE-21 to restore UTF-8-RAW -> UTF-16 -> UTF-8-RAW
   }

   if (flag is set)
   {
     it generates ESCAPE-16-21 for ESCAPE-8 points
     so we can restore UTF-8-RAW -> UTF-16 -> UTF-8-RAW without HIGH-ESCAPE-PLANE-21.
   }


Main USE CASES with UTF-8 <-> UTF-16 conversions:

 WIN32:   UTF-16-RAW -> UTF-8 (Archive) -> UTF-16-RAW
   {
            set Z7_UTF_FLAG_FROM_UTF8_USE_ESCAPE
     Do NOT set Z7_UTF_FLAG_FROM_UTF8_SURROGATE_ERROR
     Do NOT set Z7_UTF_FLAG_FROM_UTF8_BMP_ESCAPE_CONVERT
     
     So we restore original SINGLE-SURROGATE-16 from single SINGLE-SURROGATE-8.
   }

 Linux:   UTF-8-RAW -> UTF-16 (Intermediate / Archive) -> UTF-8-RAW
   {
     we want restore original UTF-8-RAW sequence later from that ESCAPE-16.
     Set the flags:
       Z7_UTF_FLAG_FROM_UTF8_SURROGATE_ERROR
       Z7_UTF_FLAG_FROM_UTF8_USE_ESCAPE
       Z7_UTF_FLAG_FROM_UTF8_BMP_ESCAPE_CONVERT
   }

 MacOS:   UTF-8-RAW -> UTF-16 (Intermediate / Archive) -> UTF-8-RAW
   {
     we want to restore correct UTF-8 without any BMP processing:
     Set the flags:
       Z7_UTF_FLAG_FROM_UTF8_SURROGATE_ERROR
       Z7_UTF_FLAG_FROM_UTF8_USE_ESCAPE
   }

*/

// zero char is not allowed in (src) buf
bool Convert_UTF8_Buf_To_Unicode(const char *src, size_t srcSize, UString &dest, unsigned flags = 0);

bool ConvertUTF8ToUnicode_Flags(const AString &src, UString &dest, unsigned flags = 0);
bool ConvertUTF8ToUnicode(const AString &src, UString &dest);

#define Z7_UTF_FLAG_TO_UTF8_SURROGATE_ERROR    (1 << 8)
#define Z7_UTF_FLAG_TO_UTF8_EXTRACT_BMP_ESCAPE (1 << 9)
// #define Z7_UTF_FLAG_TO_UTF8_PARSE_HIGH_ESCAPE  (1 << 10)

/*
Z7_UTF_FLAG_TO_UTF8_SURROGATE_ERROR

  if (flag is NOT set)
  {
     we extract SINGLE-SURROGATE as normal UTF-8
     
     In Windows : for UTF-16-RAW <-> UTF-8 (archive) <-> UTF-16-RAW in .
     
     In Linux :
       use-case-1: UTF-8 -> UTF-16 -> UTF-8  doesn't generate UTF-16 SINGLE-SURROGATE,
                   if (Z7_UTF_FLAG_FROM_UTF8_SURROGATE_ERROR) is used.
       use-case 2: UTF-16-7z (with SINGLE-SURROGATE from Windows) -> UTF-8 (Linux)
                   will generate SINGLE-SURROGATE-UTF-8 here.
  }

  if (flag is set)
  {
     we generate UTF_REPLACEMENT_CHAR (0xfffd) for SINGLE_SURROGATE
     it can be used for compatibility mode with WIN32 UTF function
     or if we want UTF-8 stream without any errors
  }


Z7_UTF_FLAG_TO_UTF8_EXTRACT_BMP_ESCAPE
  
  if (flag is NOT set) it doesn't extract  raw 8-bit symbol from Escape-Plane-16
  if (flag is set)     it         extracts raw 8-bit symbol from Escape-Plane-16

  in Linux we need some way to extract NON-UTF8 RAW 8-bits from BMP (UTF-16 7z archive):
  if (we       use High-Escape-Plane), we can transfer BMP escapes to High-Escape-Plane.
  if (we don't use High-Escape-Plane), we must use Z7_UTF_FLAG_TO_UTF8_EXTRACT_BMP_ESCAPE.
    

Z7_UTF_FLAG_TO_UTF8_PARSE_HIGH_ESCAPE
  // that flag affects the code only if (wchar_t is 32-bit)
  // that mode with high-escape can be disabled now in UTFConvert.cpp
  if (flag is NOT set)
     it doesn't extract raw 8-bit symbol from High-Escape-Plane
  if (flag is set)
     it        extracts raw 8-bit symbol from High-Escape-Plane

Main use cases:

WIN32 : UTF-16-RAW -> UTF-8 (archive) -> UTF-16-RAW
   {
     Do NOT set Z7_UTF_FLAG_TO_UTF8_EXTRACT_BMP_ESCAPE.
     Do NOT set Z7_UTF_FLAG_TO_UTF8_SURROGATE_ERROR.
     So we restore original UTF-16-RAW.
   }

Linix : UTF-8 with Escapes -> UTF-16 (7z archive) -> UTF-8 with Escapes
     set Z7_UTF_FLAG_TO_UTF8_EXTRACT_BMP_ESCAPE to extract non-UTF from 7z archive
     set Z7_UTF_FLAG_TO_UTF8_PARSE_HIGH_ESCAPE for intermediate UTF-16.
     Note: high esacape mode can be ignored now in UTFConvert.cpp

macOS:
     the system doesn't support incorrect UTF-8 in file names.
     set Z7_UTF_FLAG_TO_UTF8_SURROGATE_ERROR
*/

extern unsigned g_Unicode_To_UTF8_Flags;

void ConvertUnicodeToUTF8_Flags(const UString &src, AString &dest, unsigned flags = 0);
void ConvertUnicodeToUTF8(const UString &src, AString &dest);

void Convert_Unicode_To_UTF8_Buf(const UString &src, CByteBuffer &dest);

/*
#ifndef _WIN32
void Convert_UTF16_To_UTF32(const UString &src, UString &dest);
void Convert_UTF32_To_UTF16(const UString &src, UString &dest);
bool UTF32_IsThere_BigPoint(const UString &src);
bool Unicode_IsThere_BmpEscape(const UString &src);
#endif

bool Unicode_IsThere_Utf16SurrogateError(const UString &src);
*/

#ifdef Z7_WCHART_IS_16BIT
#define Convert_UnicodeEsc16_To_UnicodeEscHigh(s)
#else
void Convert_UnicodeEsc16_To_UnicodeEscHigh(UString &s);
#endif

/*
// #include "../../C/CpuArch.h"

// ---------- Utf16 Little endian functions ----------

// We store 16-bit surrogates even in 32-bit WCHARs in Linux.
// So now we don't use the following code:

#if WCHAR_MAX > 0xffff

// void *p     : pointer to src bytes stream
// size_t len  : num Utf16 characters : it can include or not include NULL character

inline size_t Utf16LE__Get_Num_WCHARs(const void *p, size_t len)
{
  #if WCHAR_MAX > 0xffff
  size_t num_wchars = 0;
  for (size_t i = 0; i < len; i++)
  {
    wchar_t c = GetUi16(p);
    p = (const void *)((const Byte *)p + 2);
    if (c >= 0xd800 && c < 0xdc00 && i + 1 != len)
    {
      wchar_t c2 = GetUi16(p);
      if (c2 >= 0xdc00 && c2 < 0xe000)
      {
        c = 0x10000 + ((c & 0x3ff) << 10) + (c2 & 0x3ff);
        p = (const void *)((const Byte *)p + 2);
        i++;
      }
    }
    num_wchars++;
  }
  return num_wchars;
  #else
  UNUSED_VAR(p)
  return len;
  #endif
}

// #include <stdio.h>

inline wchar_t *Utf16LE__To_WCHARs_Sep(const void *p, size_t len, wchar_t *dest)
{
  for (size_t i = 0; i < len; i++)
  {
    wchar_t c = GetUi16(p);
    p = (const void *)((const Byte *)p + 2);
    
    #if WCHAR_PATH_SEPARATOR != L'/'
    if (c == L'/')
      c = WCHAR_PATH_SEPARATOR;
    #endif
    
    #if WCHAR_MAX > 0xffff
    
    if (c >= 0xd800 && c < 0xdc00 && i + 1 != len)
    {
      wchar_t c2 = GetUi16(p);
      if (c2 >= 0xdc00 && c2 < 0xe000)
      {
        // printf("\nSurragate : %4x %4x -> ", (int)c, (int)c2);
        c = 0x10000 + ((c & 0x3ff) << 10) + (c2 & 0x3ff);
        p = (const void *)((const Byte *)p + 2);
        i++;
        // printf("%4x\n", (int)c);
      }
    }
    
    #endif
    
    *dest++ = c;
  }
  return dest;
}


inline size_t Get_Num_Utf16_chars_from_wchar_string(const wchar_t *p)
{
  size_t num = 0;
  for (;;)
  {
    wchar_t c = *p++;
    if (c == 0)
      return num;
    num += ((c >= 0x10000 && c < 0x110000) ? 2 : 1);
  }
  return num;
}

inline Byte *wchars_to_Utf16LE(const wchar_t *p, Byte *dest)
{
  for (;;)
  {
    wchar_t c = *p++;
    if (c == 0)
      return dest;
    if (c >= 0x10000 && c < 0x110000)
    {
      SetUi16(dest    , (UInt16)(0xd800 + ((c >> 10) & 0x3FF)));
      SetUi16(dest + 2, (UInt16)(0xdc00 + ( c        & 0x3FF)));
      dest += 4;
    }
    else
    {
      SetUi16(dest, c);
      dest += 2;
    }
  }
}

#endif
*/

#endif