// microsoft excel spreadsheets exported as .csv
// have some unique quallities:
// - separator is , or ; It seems that this is dependent of you
//    are living in the US or somewhere else
// - text, even with embedded space is not quoted, except when
//    the text contains a separator
//
// This program tries to convert an excel generated .csv into an
// orderly .csv:
// - text is quoted when space or separator or ' is in the text
// - separator is always ,
// - we assume, that a quoted field does not need further handling,
//   it is copied asis
//
// This program reads from stdin, writes to stdout
// One optional parameter: the separator that is used, default ;
//
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
char         sepin,sepout;
int          maxlen;
char*        field;
int          lf;
int          quoted;

void outputfield()
{
   int i, needquote;
   char c;
   needquote = 0;
   // determine if extra quotation is necessary
   if (! quoted)
   {
      for (i=0; i<lf; i++)
      {
	 switch (field[i])
	 {
	    case '"':
	    case ' ':
	    case '\'':
	    case '\t':
	       needquote = 1;
	       break;
	    default:
	       if (field[i] == sepout)
		  needquote = 1;
	       break;
	 }
	 if(needquote)
	    break;
      }
   }

   if (needquote)
      putchar('"');
   for (i=0; i<lf; i++)
   {
      c = field[i];
      putchar(c);
      if(needquote)
	 if (c == '"')
	    putchar(c);
   }
   if (needquote)
      putchar('"');
}

void addtofield(int c)
{
   if (lf >= maxlen-1)
   {
      maxlen = 1.5*maxlen + 1;
      field  = realloc(field,maxlen);
   }
   field[lf++] = c;
}

void usage()
{
   printf("Converts Excel-cvs to cvs by changig delimiter and adding quotes.\n");
   printf("Usage:\n");
   printf("mstocsv [sc] [cs] [ss] [cc]\n");
   printf("mstocsv reads from STDIN and writes to STDOUT\n");
   printf("  sc: input delimiter ';', output delimiter ','  (default)\n");
   printf("  cs: input delimiter ',', output delimiter ';'\n");
   printf("  ss: input delimiter ';', output delimiter ';'\n");
   printf("  cc: input delimiter ',', output delimiter ','\n");
   printf("\n");
   printf("Examples:\n");
   printf("   mstocvs < in.cvs > out.cvs\n");
   printf("   mstocvs ss < in.cvs > out.cvs   # delimiter remains ';', but quotation is done\n");
   printf("   mstocvs sc < in.cvs | my_cvs_processing_program\n");
}
int main(int argc, char*argv[])
{
   sepin  = ';';
   sepout = ',';
   if (argc > 2) 
   {
      usage();
      return 1;
   }
   if (argc == 2 && strlen(argv[1]) == 2 )
   {
      switch (argv[1][0])
      {
	 case 'c':
	    sepin = ',';
	    break;
	 case 's':
	    sepin = ';';
	    break;
	 default:
	    usage();
	    return 1;
      }
      switch (argv[1][1])
      {
	 case 'c':
	    sepout = ',';
	    break;
	 case 's':
	    sepout = ';';
	    break;
	 default:
	    usage();
	    return 1;
      }
   }
   else
   {
      if (argc != 1)
      {
	 usage();
	 return 1;
      }
   }



   maxlen = 20;
   field  = malloc(maxlen);

   int state;
   int c,cc;

   enum {
      START,
      INFIELD,
      INQFIELD,
      LFFOUND,
      CRFOUND,
      EOFFOUND,
   };

   state = START;

   while(1)
   {
      switch (state)
      {
	 case START: 
	    lf = 0;
	    quoted = 0;
	    c = getchar();
	    switch (c)
	    {
	       case EOF:
		  state = EOFFOUND;
		  break;
	       case '"':
		  state = INQFIELD;
		  quoted = 1;
		  addtofield(c);
		  break;
	       case '\n':
		  state = LFFOUND;
		  break;
	       case '\r':
		  state = CRFOUND;
		  break;
	       default:
		  if (c == sepin)
		  {
		     putchar(sepout);
		     break;
		  }
		  addtofield(c);
		  state       = INFIELD;
		  break;
	    }
	    break;
	 case INFIELD:
	    c = getchar();
	    switch(c)
	    {
	       case EOF:
		  state = EOFFOUND;
		  break;
	       case '\n':
		  state = LFFOUND;
		  break;
	       case '\r':
		  state = CRFOUND;
		  break;
	       default:
		  if (c == sepin)
		  {
		     outputfield();
		     putchar(sepout);
		     state = START;
		     break;
		  }
		  addtofield(c);
		  break;
	    }
	    break;
	 case INQFIELD:
	    c = getchar();
	    switch(c)
	    {
	       case EOF:
		  state = EOFFOUND;
		  break;
	       case '\n':
		  state = LFFOUND;
		  break;
	       case '\r':
		  state = CRFOUND;
		  break;
	       case '"':
		  addtofield('"');
		  cc = getchar();
		  if (cc == '"')
		     addtofield('"');
		  else
		  {
		     if (cc != EOF)
			ungetc(cc,stdin);
		     state = INFIELD;
		  }
		  break;
	       default:
		  addtofield(c);
		  break;
	    }
	    break;
	 case LFFOUND:
	    outputfield();
	    putchar('\n');
	    state = START;
	    break;
	 case CRFOUND:
	    c = getchar();  // assuming it is  '\n'
	    outputfield();
	    putchar('\r');
	    putchar(c);
	    state = START;
	    break;
	 case EOFFOUND:
	    outputfield();
	    free(field);
	    return 0;
      }
   }
}