---

stringutil.h

Go to the documentation of this file.

// dataio 0.5 (beta) - Copyright (C) 2001, by Sadao Massago              //
// file: stringutil.h (miscelaneous string related functions) header     //
// ----------------------------------------------------------------------//
// The dataio library and related files is licenced under the term of    //
// GNU Lesser General Public License version 2.1 or latter               //
// (see lesser.txt for detail).                                          //
// For information over GNU or GNU compatible license, visit the site    //
// http://www.gnu.org.                                                   //

// javadoc style documentation in way to generate class documentation   //
// using GNU GPL licensed doxygen utility. For doxygen, see the         //
// http://www.doxygen.org                                               //

// currently, getline and putline is not used by dataio 
// classes
#ifndef  _STRINGUTIL_H_

// #ifdef __plusplus
#define _STRINGUTIL_H_

#include <string>
#include <iostream>
// #include <strstream>
#include <vector>


#ifndef DOXYGEN_SHOULD_SKIP_THIS // doxygen skip this...

#ifdef _DEBUG_ // _DEBUG_ activate debugging mode on stringutil
   #define _STRINGUTILDEBUG_
#endif

// #define _STRINGUTILNOTHROW_ // if error occur, use error message 
// #define _STRINGUTILNOMESSAGE_ // if error occur, use error message 

// #ifdef _STRINGUTILDEBUG_ // activate message in debug mode
//  #undef _STRINGUTILNOMESSAGE_ // if error occur, use error message 
// #endif

#endif // #ifndef DOXYGEN_SHOULD_SKIP_THIS // doxygen skip this...

// defining group for datautil functions (for doxygen)
// it's need to list function (that is not class member)
// define datautil module group
// starting group (for doxygen)
// #include <errno.h> // for C routine error handle

// #define MAXUINT ((unsigned) ~0)

// #define _DEBUG_
// #define _COMMAFLOAT_ // input float that use comma as dot

// getline works same as this function!!
/* string &readlinestring(fstream &f)
// use istream getline(istream &f, string &item, char delimiter='\n');
{   
  #define BUFFLEN 10
  char buff[BUFFLEN+1];    
  static string line;
  line.erase();
  if(unnamed.size() != 0) // process unnamed field 
  while(!f.eof())
  {
    f.get(buff, BUFFLEN);
    line += string(buff);
    if(!f.eof() && (f.get() != '\n'))
      f.unget(); // line reading not finished
    else return line;
  }      
  return line;
} // readlinestring
*/

// status acess
#define _STRINGUTILDEFAULTPRINTERROR true
#define _STRINGUTILDEFAULTTHROWEXCEPTION false

#define _STRINGUTILSTRDELIMITER '\"'

#define _STRINGUTILSCAPECHAR '\\'

#define _STRINGUTILCOMMENTCHAR '#'

#define  _STRINGUTILCOLUMNSEPARATOR '\t'

#define _STRINGUTILCNTRLMARK '^'

#define _STRINGUTILHEXAMARK 'X'

#define _STRINGUTILCHARTABLE { \
                  /* {'\"', '\"'}, */ /* double quote */ \
                  /* {'\'', '\''}, */ /* quote */ \
                  /* {'\\', '\\'}, */ /* backslash */ \
                  {'a', '\a'},  /* Alarm beep  */ \
                  {'b', '\b'},  /* Back space  */ \
                  {'e', '\e'},  /* Escape character  */  \
                  {'f', '\f'},  /* Form Feed (next page)  */  \
                  {'n', '\n'},  /* new line (next line) */ \
                  {'r', '\r'},  /* Carriage Return (start of current line)  */  \
                  {'t', '\t'},  /* Tabulation  */ \
                  {'v', '\v'},  /* Vertical tabulation  */ \
                  {'\0', '\0'}, /* end of table */ \
                  }

bool stringutilprinterror();
bool stringutilprinterror(bool status);

bool stringutilthrowexception();
bool stringutilthrowexception(bool status);

// white char detections

//  The latin and other usuall extended character,
//  if larger than 160?
//
inline bool isextended(char c)
{
  return ((unsigned int)c) > 160;
} // isextended

// obs.: isspace('\0') == false
// isspace (POSIX) are: space,  form-feed ('\f'), newline  ('\n'),  
// carriage  return  ('\r'), horizontal tab ('\t'), and vertical tab ('\v').
// isgraph() is printable except space 
// (space is space, but printable)
// isprint() is all printable character
inline bool iswhite(char c)
{
  return !isgraph(c) && !isextended(c);
} // iswhite

inline char tocntrl(char c)
{
 //  if(isalpha(c) || (('['<=toupper(c)) && (toupper(c) <='_')) ||  )
 if(iscntrl(toupper(c)-'A'+1))
    return toupper(c)-'A'+1;
 else
    return c;
}

// miscelaneous functions

   // find position of s2 on s1, starting of pos
   unsigned stringcasefind(string const &s1, string const &s2, unsigned pos =0);

   // find last position of s2 on s1, starting of pos
   unsigned stringcasefind_last_of(string const &s1, string const &s2, unsigned pos =0);


   // destruct clearing and parsing
   string &parsestring(string &s, char strdelimiter=_STRINGUTILSTRDELIMITER, char scapechar=_STRINGUTILSCAPECHAR);

   vector <string> &parsestring(vector <string> &strlist, char strdelimiter=_STRINGUTILSTRDELIMITER,  char scapechar=_STRINGUTILSCAPECHAR);

   string &reverseparsestring(string &s, char strdelimiter=_STRINGUTILSTRDELIMITER,  char scapechar=_STRINGUTILSCAPECHAR);

   vector <string> &reverseparsestring(vector <string> &strlist, char strdelimiter=_STRINGUTILSTRDELIMITER,  char scapechar=_STRINGUTILSCAPECHAR);

   unsigned findstringdelimiterclose(string const &s, unsigned pos=0, char strdelimiter=_STRINGUTILSTRDELIMITER,  char scapechar=_STRINGUTILSCAPECHAR);

   bool stringdelimiterbalanced(string const &s, char strdelimiter=_STRINGUTILSTRDELIMITER, char scapechar=_STRINGUTILSCAPECHAR, unsigned pos1=0, unsigned pos2=(~0U));
   bool needdelimiter(string const &s, char strdelimiter=_STRINGUTILSTRDELIMITER,  char scapechar=_STRINGUTILSCAPECHAR);

   istream &gettextline(istream &f, string &line, char comment=_STRINGUTILCOMMENTCHAR,
                  char strdelimiter=_STRINGUTILSTRDELIMITER,
                  char scapechar=_STRINGUTILSCAPECHAR, char newlinedelimiter='\0');

   //  \sa enhacedgetline
   // istream &getline(istream &f, vector <string> &dataline,
   // char separator =  _STRINGUTILCOLUMNSEPARATOR, char newline='\0');

   // enhaced version of getline()
   // \sa getline
   istream &getline(istream &f, vector <string> &dataline, char separator =  _STRINGUTILCOLUMNSEPARATOR,
      char comment=_STRINGUTILCOMMENTCHAR, char strdelimiter=_STRINGUTILSTRDELIMITER,
      char scapechar=_STRINGUTILSCAPECHAR, char newline='\0');


   // \sa enhacedgetlines
   // istream &getlines(istream &f, vector <vector<string> > &datalist,
   //         char separator =  _STRINGUTILCOLUMNSEPARATOR, char comment='\0', char newline='\0');

   // enhaced version of getlines()
   // \sa getlines
   istream &getlines(istream &f, vector <vector <string> > &datalist, 
                  char separator =  _STRINGUTILCOLUMNSEPARATOR,
                  char comment=_STRINGUTILCOMMENTCHAR,
                  char strdelimiter=_STRINGUTILSTRDELIMITER,
                  char scapechar=_STRINGUTILSCAPECHAR, char newline='\0');



   // vectorized putline
   ostream &putline(ostream &f, vector <string> &dataline,
   char separator =  _STRINGUTILCOLUMNSEPARATOR, char newline='\0');

   ostream &putlines(ostream &f, vector <vector<string> > &datalist,
                  char separator =  _STRINGUTILCOLUMNSEPARATOR, char newline='\0');

   // vector <vector ,string> > transposition
   // extern  vector <vector <string> > &transpose(vector <vector <string> >&data);
   vector <vector <string> > &transpose(vector <vector <string> >&data);


// endding group (for doxygen)
#endif // #define _STRINGUTIL_H_

---