CryptoByLeo/Src/RSPParser/RSP.mqh

//+------------------------------------------------------------------+
//|                                                          RSP.mqh |
//|                                   Copyright 2026, Niquel Mendoza |
//|                                             https://www.mql5.com |
//+------------------------------------------------------------------+
#property copyright "Copyright 2026, Niquel Mendoza"
#property link      "https://www.mql5.com"
#property strict

#ifndef CRYPTOBYLEO_RSPPARSER_RSP_MQH
#define CRYPTOBYLEO_RSPPARSER_RSP_MQH

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
#include "Def.mqh"

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
namespace TSN
{
//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
class CRSPParser : public CBaseLanFormat
 {
public:
  int                m_root_values;

private:
  //--- last err
  ENUM_RSPPARSER_LAST_ERR m_last_err;

  //--- parseo
  int                m_prev_start_v;

  // el ultimo valor esta entre [] osea meta
  bool               m_last_meta;

  // mascara de fin
#ifndef RSPDATA_DESACTIVE_SWAR
  ulong              m_last_val_mask;
#else
  uchar              m_last_val_char;
#endif // RSPDATA_DESACTIVE_SWAR

public:
                     CRSPParser(void);
                    ~CRSPParser(void) {}

  //---
  __forceinline void CorrectPadding() { CorrectPadingInternal('\n', 8);  }

  //---
  bool               AutoParseByFileExt(const int file_ext_type);
  int                AssingFile(const string& file_name, const bool comon_flag);

  //--
  // verifica si una clave esta limpia
  // En caso no lo este ejeucta la limpiza ahi mismo
  void               VerifyCleanKey(const int cur, const int start, int& len);

  //---
  bool               KeysEqual(const string &str, const int curr);
  bool               StrEquals(const string &str, const int curr) const;

  //---
  __forceinline ENUM_RSPPARSER_LAST_ERR LastErr() const { return m_last_err; }

  //---
  bool               Parse();

  //---
  void               ErrorInfo(bool with_avanze = false) const;

  //---
  bool               SaveCompiled(const string & file_name_out, const bool out_common_flag) const;
  bool               SaveCompiled(const string & file_name_out, const bool out_common_flag, const string & file_name_data, const bool data_common_flag) const;


  //---
  void               PrintCintaTypes(const int start, const int num) const;

  //---
  CRSPNode           GetRoot();
 };

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
CRSPParser::CRSPParser()
  :
#ifndef RSPDATA_DESACTIVE_SWAR
  m_last_val_mask(TSNTABLES_SWAR_MASK_NEW_LINE)
#else //  RSPDATA_DESACTIVE_SWAR
  m_last_val_char('\n')
#endif // RSPDATA_DESACTIVE_SWAR
 {
 }


//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
bool CRSPParser::AutoParseByFileExt(const int file_ext_type)
 {
  switch(file_ext_type)
   {
    case TSN_SLF_RAW_FILE:
     {
      CorrectPadding(); // Auto
      return Parse();
     }
    case TSN_SLF_RAW_EMBEBED:
     {
      return m_cinta_pos > 0 && m_len > 0;
     }
    default:
     {
      m_last_err = RSPPARSER_LAST_ERR_INVALID_EXT_FILE;
      m_pos = 0;
      return false;
     }
   }
 }

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
bool CRSPParser::StrEquals(const string &str, const int curr) const
 {
//---
// m_cinta[m_cinta_pos++] = long(m_pos - 1) | long(start) << TSN_JSON_BIT_STR_START_END;
  const int start = int(m_cinta[curr] >> TSN_SBL_BIT_STR_START);
  const int len = int((m_cinta[curr] >> TSN_SBL_BIT_STR_LEN) & TSN_SBL_BIT_STR_MASK_LEN);
//PrintFormat("Exact = '%s', other = '%s'", CharArrayToString(m_raw, start, len), str);

  const int l = StringLen(str);

//---
  if(l != len)
    return false;
  for(int i = 0; i < l; i++)
   {
    if(str[i] != m_raw[start + i])
      return false;
   }
  return true;
 }

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
void CRSPParser::VerifyCleanKey(const int cur, const int start, int& len)
 {
  const int k = cur + 1; // kpos+1
  if(((m_cinta[k] >> RSPDATA_BITSTART_ISKEY_CLEAN) & RSPDATA_MASK_META_IS) == 0)
   {
    int j = (start + len) - 1;
    while(j >= start && m_raw[j] < 33)
     {
      j--;
     }
    len = (j - start) + 1;
    m_cinta[cur] =  long(RSPDATA_TYPE_PLAIN_KEY)
                    | long(len) << TSN_SBL_BIT_STR_LEN
                    | long(start) << TSN_SBL_BIT_STR_START;

    m_cinta[k] |= RSPDATA_BITSTART_ISKEY_CLEAN_F;
   }
 }

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
bool CRSPParser::KeysEqual(const string &str, const int curr)
 {
//---
  const int start = int(m_cinta[curr] >> TSN_SBL_BIT_STR_START);
  int len = int((m_cinta[curr] >> TSN_SBL_BIT_STR_LEN) & TSN_SBL_BIT_STR_MASK_LEN);
//---
  VerifyCleanKey(curr, start, len);
//---
  const int l = StringLen(str);
//---
  if(l != len)
    return false;
//---
  for(int i = 0; i < l; i++)
    if(str[i] != m_raw[start + i])
      return false;
//---
  return true;
 }


//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
// luego de un valor buscar key
// el |1 lo hacemos para acitvar el bit 1 cosa uqe aumenta en 1 el valor base del token
// En el enum estan definidos de tal forma qeu si le aumentmos 1 para value=key (ahora es key)
// y los demas tambien
// coment aunenta en uno ahora es comentp
// lo que tambien aumenta es cor que pasa a ser corp
// e invalid pasa a (fuera) de rango (defualt en NEXTT)
// maxima velocidad sin branches
#define TSN_RSPPARSER_NEXTV \
        while(true) \
         { \
          if(m_pos >= m_len) \
           { \
            return true; \
           } \
          const uchar cfasd = m_raw[m_pos]; \
          if(cfasd < 33) \
           { \
            m_pos++; \
            continue; \
           } \
           m_next = g_tsn_rsp_t_to_tokesn[cfasd]|1; \
           break;\
         }

//---
#define RSPDATA_PARSE_FIND_CHAR_UNROLLED_8(CH) \
        if(m_raw[m_pos] == CH)\
          break;\
        m_pos++;\
        if(m_raw[m_pos] == CH)\
          break;\
        m_pos++;\
        if(m_raw[m_pos] == CH)\
          break;\
        m_pos++;\
        if(m_raw[m_pos] == CH)\
          break;\
        m_pos++;\
        if(m_raw[m_pos] == CH)\
          break;\
        m_pos++;\
        if(m_raw[m_pos] == CH)\
          break;\
        m_pos++;\
        if(m_raw[m_pos] == CH)\
          break;\
        m_pos++;\
        if(m_raw[m_pos] == CH)\
          break; \
        m_pos++;


//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
bool CRSPParser::Parse(void)
 {
//---
  m_pos = m_offset;
  m_cinta_pos = 0;
  m_last_meta = false;
  m_last_err = RSPPARSER_LAST_ERR_NOT;
  m_root_values = 0;

//--- Initial
  while(true)
   {
    const uchar _c = m_raw[m_pos];
    if(_c < 33)
     {
      m_pos++;
      continue;
     }
    m_next = g_tsn_rsp_t_to_tokesn[_c] | 1;
    break;
   }

//--- Iter
  while(true)
   {
    switch(m_next)
     {
      //---
      case RSPDATA_TOK_TYPE_VALUE:
       {
        // m_prev_start_v = start del key qeu este ultimo nos definio para este valor
#ifndef RSPDATA_DESACTIVE_SWAR
        while(true)
         {
          const ulong chunk = (ulong)m_raw[m_pos]
                              | (ulong)m_raw[m_pos + 1] << 8
                              | (ulong)m_raw[m_pos + 2] << 16
                              | (ulong)m_raw[m_pos + 3] << 24
                              | (ulong)m_raw[m_pos + 4] << 32
                              | (ulong)m_raw[m_pos + 5] << 40
                              | (ulong)m_raw[m_pos + 6] << 48
                              | (ulong)m_raw[m_pos + 7] << 56;

          //---
          const ulong x = chunk ^ m_last_val_mask;
          const ulong has_x = TSNTABLES_SWAR_HAS(x);
          if(has_x == 0)
           {
            m_pos += 8;

            //---
            if((m_pos + 7) >= m_len)
             {
              m_last_err = RSPPARSER_LAST_ERR_SE_BUSCABA_FIN_V;
              return false;
             }

            //---
            continue;
           }

          //---
          // Si hay \n
          const ulong flag = TSNTABLES_SWAR_64_DEJAR_SOLO_MINUS_SIG_BYTE(has_x);
          m_pos += TSNTABLES_SWAR_64_GET_BYTE(flag); // Ahora mismo en \n
          break;
         }

        //---
        m_last_val_mask = TSNTABLES_SWAR_MASK_NEW_LINE;
        
        
#else  // RSPDATA_DESACTIVE_SWAR
        while(true)
         {
          RSPDATA_PARSE_FIND_CHAR_UNROLLED_8(m_last_val_char)
          //---
          if((m_pos + 7) >= m_len)
           {
            m_last_err = RSPPARSER_LAST_ERR_SE_BUSCABA_FIN_V;
            return false;
           }
         }

        //---
        m_last_val_char = '\n';
#endif // RSPDATA_DESACTIVE_SWAR
        // Ahora mismo en \n
        
        
        //---
        m_cinta[m_cinta_pos++] = long(RSPDATA_TYPE_PLAIN_VALUE) | long(m_pos - m_prev_start_v) << TSN_SBL_BIT_STR_LEN
                                 | long(m_prev_start_v) << TSN_SBL_BIT_STR_START;


        //---
        m_pos++; // Luiego del \n
        m_root_values++;

        //---
        TSN_RSPPARSER_NEXTV
        break;
       }



      //---
      case RSPDATA_TOK_TYPE_KEY:
       {
        // ahora mismo justo en kstart
        const int start = m_pos++;

        //---
        // Por deqfecto desactivo en priomedio als key no superan los 8 chars ahi
        // asi que no hay mucho beneficio
#ifdef  RSPDATA_PARSER_KEY_ACTIVE_SWAR
        while(true)
         {
          //Print(CharArrayToString(m_raw, m_pos, 7));
          const ulong chunk = (ulong)m_raw[m_pos]
                              | (ulong)m_raw[m_pos + 1] << 8
                              | (ulong)m_raw[m_pos + 2] << 16
                              | (ulong)m_raw[m_pos + 3] << 24
                              | (ulong)m_raw[m_pos + 4] << 32
                              | (ulong)m_raw[m_pos + 5] << 40
                              | (ulong)m_raw[m_pos + 6] << 48
                              | (ulong)m_raw[m_pos + 7] << 56;

          //---
          const ulong x = chunk ^ TSNTABLES_SWAR_MASK_IGUAL;
          const ulong has_x = TSNTABLES_SWAR_HAS(x);
          if(has_x == 0)
           {
            m_pos += 8;

            //---
            if((m_pos + 7) >= m_len)
             {
              m_last_err = RSPPARSER_LAST_ERR_SE_BUSCABA_EQ;
              return false;
             }
            continue;
           }

          //---
          // Si hay =
          const ulong flag = TSNTABLES_SWAR_64_DEJAR_SOLO_MINUS_SIG_BYTE(has_x);
          m_pos += TSNTABLES_SWAR_64_GET_BYTE(flag);
          break;
         }
#else // RSPDATA_PARSER_KEY_ACTIVE_SWAR
        while(true)
         {
          RSPDATA_PARSE_FIND_CHAR_UNROLLED_8('=')
          //---
          if((m_pos + 7) >= m_len)
           {
            m_last_err = RSPPARSER_LAST_ERR_SE_BUSCABA_EQ;
            return false;
           }
         }
#endif // RSPDATA_PARSER_KEY_ACTIVE_SWAR
        // Ahora mismo justo en =
        m_cinta[m_cinta_pos++] = long(RSPDATA_TYPE_PLAIN_KEY)
                                 | long(m_pos - start) << TSN_SBL_BIT_STR_LEN
                                 | long(start) << TSN_SBL_BIT_STR_START;

        // ahora toca meta
        m_cinta[m_cinta_pos++] = long(RSPDATA_TYPE_META) | long(m_last_meta) << TSN_SBL_BIT_START_ENDTYPE;

        //--- avanzamos
        m_pos++; // Ahora toca value
        m_prev_start_v = m_pos;
        m_last_meta = false;

        //--- Ahora buscamos el valor
        while(true)
         {
          const uchar f = m_raw[m_pos];
          if(f < 33)
           {
            if(f == '\n')
             {
              // No habai nada estamos en\n entonces 0
              m_cinta[m_cinta_pos++] = long(RSPDATA_TYPE_PLAIN_VALUE) | long(0) << TSN_SBL_BIT_STR_LEN
                                       | long(start) << TSN_SBL_BIT_STR_START;

              //---
              TSN_RSPPARSER_NEXTV
              break;
             }

            //---
            m_pos++;
            if(m_pos >= m_len) // o en todo caso fin
             {
              m_cinta[m_cinta_pos++] = long(RSPDATA_TYPE_PLAIN_VALUE) | long(0) << TSN_SBL_BIT_STR_LEN
                                       | long(start) << TSN_SBL_BIT_STR_START;
              return true;
             }

            //---
            m_prev_start_v++; // nada avanzamos..
            continue;
           }

          //--- Justo aqui nos detenemos. Vemos que sigue comentario o valor
          // Valor raw tal cual (sin 1 asi que el value es vlaue ya no se pasa a key)
          m_next = g_tsn_rsp_t_to_tokesn[f];
          break;
         }

        //---
        break;
       }


      //--- [ ]
      case RSPDATA_TOK_TYPE_COR:
      case RSPDATA_TOK_TYPE_COR_P:
       {
        m_pos++; // luego del cor

        //
        m_last_meta = true;

        // saltar ws
        while(true)
         {
          if(m_raw[m_pos] < 33)
            continue;
          break;
         }

        // camibamos las mcaras para que en vez de \n sea ] fin de valor
#ifndef RSPDATA_DESACTIVE_SWAR
        m_last_val_mask = TSNTABLES_SWAR_MASK_COR_END;
#else
        m_last_val_char = ']';
#endif

        // ahora mismo ene l primer char de key
        m_next = RSPDATA_TOK_TYPE_KEY;
        break;
       }


      //---
      case RSPDATA_TOK_TYPE_COMMENT:
      case RSPDATA_TOK_TYPE_COMMENT_P:
       {
        m_pos++; // Siguiente..

        //---
#ifndef RSPDATA_DESACTIVE_SWAR
        while(true)
         {
          //Print(CharArrayToString(m_raw, m_pos, 7));
          const ulong chunk = (ulong)m_raw[m_pos]
                              | (ulong)m_raw[m_pos + 1] << 8
                              | (ulong)m_raw[m_pos + 2] << 16
                              | (ulong)m_raw[m_pos + 3] << 24
                              | (ulong)m_raw[m_pos + 4] << 32
                              | (ulong)m_raw[m_pos + 5] << 40
                              | (ulong)m_raw[m_pos + 6] << 48
                              | (ulong)m_raw[m_pos + 7] << 56;

          //---
          const ulong x = chunk ^ TSNTABLES_SWAR_MASK_NEW_LINE;
          const ulong has_x = TSNTABLES_SWAR_HAS(x);
          if(has_x == 0)
           {
            m_pos += 8;

            //--- Check
            if((m_pos + 7) >= m_len)
             {
              m_last_err = RSPPARSER_LAST_ERR_SE_BUSCABA_FIN_DE_LINEA;
              return false;
             }
            continue;
           }

          //---
          // Si hay \n
          const ulong flag = TSNTABLES_SWAR_64_DEJAR_SOLO_MINUS_SIG_BYTE(has_x);
          m_pos += TSNTABLES_SWAR_64_GET_BYTE(flag); // Ahora mismo en \n+1 \n[aqui]
          break;
         }
#else // RSPDATA_DESACTIVE_SWAR
        while(true)
         {
          RSPDATA_PARSE_FIND_CHAR_UNROLLED_8('\n')
          //---
          if((m_pos + 7) >= m_len)
           {
            m_last_err = RSPPARSER_LAST_ERR_SE_BUSCABA_FIN_DE_LINEA;
            return false;
           }
         }
#endif // RSPDATA_DESACTIVE_SWAR
        //---
        m_pos++;
        // Si o si sigue key
        TSN_RSPPARSER_NEXTV
        break;
       }

      //---
      case RSPDATA_TOK_INVALID:
       {
        m_last_err = RSPPARSER_LAST_ERR_INVALID_TOKEN;
        return false;
       }

      //---
      default:
        m_last_err = RSPPARSER_LAST_ERR_INVALID_NEXTT;
        return false;
     }
   }
 }

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
int CRSPParser::AssingFile(const string &file_name, const bool comon_flag)
 {
//---
  if(FileGetExtension(file_name) == RSPPARSER_TAPECACHE_EXT) // jsontp tape cache
   {
    ::ResetLastError();
    const int fh = FileOpen(file_name, FILE_BIN | FILE_ANSI | (comon_flag ?  FILE_COMMON : 0) | FILE_READ);
    if(fh == INVALID_HANDLE)
     {
      return -1;
     }
    // Metadata (cinta\yaml)
    m_cinta_pos = FileReadInteger(fh); // Nota que cuadno se guarda se hace con el tamaño exacto
    m_cinta_reserve = m_cinta_pos; // Mismo tamaño
    const int type = FileReadInteger(fh); // Tipo ()

    //---
    if(type == TSN_SLF_RAW_EMBEBED)
     {
      m_len = FileReadInteger(fh);
      m_offset = FileReadInteger(fh);
      FileReadArray(fh, m_cinta, 0, m_cinta_pos); // Total
      // Yaml embebido (aqui no hace falta hacer nada ya todo esta hecho.. osea lo de encoding etc..)
      FileReadArray(fh, m_raw, 0, m_len);
     }
    else // Dir
     {
      const bool common = bool(FileReadInteger(fh, CHAR_VALUE));
      const int file_name_len = FileReadInteger(fh);
      if(AssingFile(FileReadString(fh, file_name_len), common) == TSN_SLF_EXT_TYPE_CACHE_FILE)
       {
        FileClose(fh);
        return -1; // fallo
       }
      FileReadArray(fh, m_cinta, 0, m_cinta_pos);
     }

    //---
    FileClose(fh);
    return TSN_SLF_EXT_TYPE_CACHE_FILE;
   }
  else // Ninguno de los dos... json normal O jsonasm
   {
    const int l = (int)FileLoad(file_name, m_raw, (comon_flag ? FILE_COMMON : 0));
    if(l == -1)
      return -1;

    //---
    m_len = l;

    //---
    NormalizeUft();

    //---
    return TSN_SLF_EXT_TYPE_FILE;
   }
 }

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
//--- Ambos
// 4 bytes: m_cinta_pos
// 4 bytes: type


//---
bool CRSPParser::SaveCompiled(const string & file_name_out, const bool out_common_flag) const
 {
  ::ResetLastError();
  const int fh = FileOpen(file_name_out, FILE_BIN | (out_common_flag ?  FILE_COMMON : 0) | FILE_WRITE | FILE_ANSI);
  if(fh == INVALID_HANDLE)
   {
    return false;
   }

//--- Comun
  FileWriteInteger(fh, m_cinta_pos);
  FileWriteInteger(fh, TSN_SLF_RAW_EMBEBED);

//---
// 4 bytes: len
// 4 bytes: offset
// array: cinta
// array yaml

//---
  FileWriteInteger(fh, m_len);
  FileWriteInteger(fh, m_offset);
  FileWriteArray(fh, m_cinta, 0, m_cinta_pos);
  FileWriteArray(fh, m_raw, 0, m_len);


//---
  FileClose(fh);

//---
  return true;
 }

//+------------------------------------------------------------------+
bool CRSPParser::SaveCompiled(const string & file_name_out, const bool out_common_flag,
                              const string & file_name_data, const bool data_common_flag) const
 {
//---
  ::ResetLastError();
  const int fh = FileOpen(file_name_out, FILE_BIN | (out_common_flag ?  FILE_COMMON : 0) | FILE_WRITE | FILE_ANSI);
  if(fh == INVALID_HANDLE)
   {
    return false;
   }

//--- Comun
  FileWriteInteger(fh, m_cinta_pos);
  FileWriteInteger(fh, TSN_SLF_RAW_FILE);

//---
// 1 bytes: comon
// 4 bytes: len
// string: name_file
// array: cinta


//---
  FileWriteInteger(fh, data_common_flag, CHAR_VALUE);
  const int len = StringLen(file_name_data);
  FileWriteInteger(fh, len);
  FileWriteString(fh, file_name_data, len);
  FileWriteArray(fh, m_cinta, 0, m_cinta_pos);

//---
  FileClose(fh);

//---
  return true;
 }

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
void CRSPParser::ErrorInfo(bool with_avanze = false) const
 {
 }

//+------------------------------------------------------------------+
//|                                                                  |
//+------------------------------------------------------------------+
void CRSPParser::PrintCintaTypes(const int start, const int num) const
 {
//---
  int p = fmax(start, 0);
  const int end = (num == WHOLE_ARRAY || (p + num) > m_cinta_pos) ? m_cinta_pos : (p + num);

//---
  while(p < end)
   {
    const ENUM_RSPDATA_TYPE tipo = ENUM_RSPDATA_TYPE(m_cinta[p] & TSN_SBL_BIT_MASK_TYPE);
    string text = "[" + (string)p + "] " + EnumToString(tipo) + " | ";

    //---
    switch(tipo)
     {
      case RSPDATA_TYPE_PLAIN_KEY:
       {
        const int s = int(m_cinta[p] >> TSN_SBL_BIT_STR_START);
        const int slen = int((m_cinta[p] >> TSN_SBL_BIT_STR_LEN) & TSN_SBL_BIT_STR_MASK_LEN);

        //---
        text += "start=" + (string)s + " len=" + (string)slen + " key=\"" + CharArrayToString(m_raw, s, slen) + "\"";
        break;
       }
      case RSPDATA_TYPE_PLAIN_VALUE:
       {
        const int s = int(m_cinta[p] >> TSN_SBL_BIT_STR_START);
        const int slen = int((m_cinta[p] >> TSN_SBL_BIT_STR_LEN) & TSN_SBL_BIT_STR_MASK_LEN);

        //---
        text += "start=" + (string)s + " len=" + (string)slen + " val=\"" + CharArrayToString(m_raw, s, slen) + "\"";
        break;
       }
      case RSPDATA_TYPE_META:
       {
        static BitInterpreter un;
        un.long_value = m_cinta[p + 1];
        text += StringFormat("is_cor=%s is_val_cleaned=%s is_key_cleaned=%s",
                             (((m_cinta[p] >> TSN_SBL_BIT_START_ENDTYPE) & RSPDATA_MASK_META_IS) != 0 ? "true" : "false"),
                             (((m_cinta[p] >> RSPDATA_BITSTART_ISVAL_CLEAN) & RSPDATA_MASK_META_IS) != 0 ? "true" : "false"),
                             (((m_cinta[p] >> RSPDATA_BITSTART_ISKEY_CLEAN) & RSPDATA_MASK_META_IS) != 0 ? "true" : "false")
                            );
        break;
       }
      default:
        break;
     }

    Print(text);
    p++;
   }

 }

//---
}

//+------------------------------------------------------------------+
#endif // CRYPTOBYLEO_RSPPARSER_RSP_MQH