This file has the code of useful routines for read or write csv formatted lines. More...

#include "bufrdeco.h"

Include dependency graph for bufrdeco_csv.c:

Macros
#define	CSV_WAIT_ITEM 1

#define	CSV_FINISHED_ITEM 2

#define	CSV_IS_ITEM 4

#define	CSV_IS_SEPARATOR 8

#define	CSV_IS_CITED 16

#define	CSV_WAIT_SEPARATOR 32

#define	CSV_IS_DOUBLE_QUOTED 64

Functions
char *	csv_quoted_string (char out, char in)
	Transform a string to a quoted string to be inserted in a csv file. More...

int	parse_csv_line2 (int nt, char tk[], char *lin)
	Parse a csv line. More...

int	parse_csv_line (int nt, char tk[], char *lin)
	Parse a csv line. More...

Variables
const char	CSV_SPACE [] = " \t\r"

const char	CSV_SEPARATOR = ','

const char	CSV_CITE = '\"'

const char	CSV_FINAL = '\n'

Detailed Description

This file has the code of useful routines for read or write csv formatted lines.

Definition in file bufrdeco_csv.c.

Macro Definition Documentation

◆ CSV_FINISHED_ITEM

#define CSV_FINISHED_ITEM 2

Definition at line 27 of file bufrdeco_csv.c.

◆ CSV_IS_CITED

#define CSV_IS_CITED 16

Definition at line 30 of file bufrdeco_csv.c.

◆ CSV_IS_DOUBLE_QUOTED

#define CSV_IS_DOUBLE_QUOTED 64

Definition at line 32 of file bufrdeco_csv.c.

◆ CSV_IS_ITEM

#define CSV_IS_ITEM 4

Definition at line 28 of file bufrdeco_csv.c.

◆ CSV_IS_SEPARATOR

#define CSV_IS_SEPARATOR 8

Definition at line 29 of file bufrdeco_csv.c.

◆ CSV_WAIT_ITEM

#define CSV_WAIT_ITEM 1

Definition at line 26 of file bufrdeco_csv.c.

◆ CSV_WAIT_SEPARATOR

#define CSV_WAIT_SEPARATOR 32

Definition at line 31 of file bufrdeco_csv.c.

Function Documentation

◆ csv_quoted_string()

char * csv_quoted_string	(	char *	out,
		char *	in
	)

Transform a string to a quoted string to be inserted in a csv file.

Parameters

out	resulting string
in	input string

Returns: If problem returns NULL, otherwise out

Definition at line 48 of file bufrdeco_csv.c.

{
  size_t i = 0, j = 0;
 
  if ( in == NULL || out == NULL)
      return NULL;
 
  if ( in[0] == 0 )
    return in;
 
  out[j++] = '\"';
  while ( in[i] && j < ( CSV_MAXL - 1 )  && i < CSV_MAXL )
    {
      if ( in[i] == '\"' )
        {
          out[j++] = '\"';
          out[j++] = '\"';
          i++;
        }
      else
        out[j++] = in[i++];
    }
  out[j++] = '\"';
  out[j] = 0; // end of string
  return out;
}

References CSV_MAXL.

Referenced by main().

Here is the caller graph for this function:

◆ parse_csv_line()

int parse_csv_line	(	int *	nt,
		char *	tk[],
		char *	lin
	)

Parse a csv line.

Parameters

nt	pointer to a integer. On success is the number of items found
tk	array of pointers. Every pointer is a item on success
lin	input line which is modified in this routine to be splitted into items

Returns: On success return 0, otherwise -1

This is an optimization of routine parse_csv_line2(). Here we suppose: -All no void items are closed between "" and separed by comas ',' -No " in items. -No blank spaces at the end nor begin of items

NOTE that input line is modified

Definition at line 258 of file bufrdeco_csv.c.

{
  char *cq[CSV_MAXL / 4], *cc[CSV_MAXL / 4], *c0;
  buf_t nc, nq, i, j;
 
  //bufrdeco_assert (lin != NULL && tk != NULL && nt != NULL);
  
  *nt = 0;
 
  // clean final new line)
  if ((c0 = strchr(lin, CSV_FINAL)) != NULL)
    *c0 = '\0';
 
  c0 = lin;
  nq = 0;
  while (c0 != NULL && *c0  && nq < (CSV_MAXL / 4))
  {
      if ((c0 = strchr(c0, CSV_CITE)) != NULL)
     {
       cq[nq++] = c0++;
     }
  }
 
  if (nq % 2)
    return -1; // Problem, number of quotes are non paired
 
  c0 = lin;  
  nc = 0;
  while (c0 != NULL && *c0 && nc < (CSV_MAXL / 4))
  {
     if ((c0 = strchr(c0, CSV_SEPARATOR)) != NULL)
     {
        j = 0;
        if (nq)
        {
          for (; j < nq ; j+=2)
            if ((cq[j] < c0) && (c0 < cq[j + 1]))
            {
              j = nq;// is a comma between two cites. Ignored
              c0++;
            }
        }
        if (j == nq)
        {
           // is a comma between two items.
           cc[nc++] = c0++;
        }
     }
  }
 
  // Now go to tokens, in this pass, still CSV_CITE are in items
  c0 = lin;
  if (nc == 0)
    tk[(*nt)++] = c0;
  else
  {
    for (i = 0; i < nc; i++)
    {
      if (cc[i] == c0)
      {
        tk[(*nt)++] = c0;
        *(c0) = '\0';
        c0++;
      }
      else
      {
        tk[(*nt)++] = c0;
        *(cc[i]) = '\0';
        c0 = cc[i] + 1;
      }
    }
  } 
  tk[(*nt)++] = c0;
  
  // fix the tokens supresing the first CSV_CITE if any
  for (i = 0; i < (buf_t)(*nt); i++)
  {
    if( *(tk[i]) == CSV_CITE)
      (tk[i])++;
  }
  
  // And then supress all CSV_CITE (also in the end of every item)
  for (i = 0; i < nq ; i++)
    *(cq[i]) = '\0';
  
  return 0;
}

References CSV_CITE, CSV_FINAL, CSV_MAXL, and CSV_SEPARATOR.

Referenced by bufr_read_tableB(), bufr_read_tableC(), bufr_read_tableD(), and main().

Here is the caller graph for this function:

◆ parse_csv_line2()

int parse_csv_line2	(	int *	nt,
		char *	tk[],
		char *	lin
	)

Parse a csv line.

Parameters

nt	pointer to a integer. On success is the number of items found
tk	array of pointers. Every pointer is a item on success
lin	input line which is modified in this routine to be splitted into items

Returns: On success return 0, otherwise -1

NOTE that input line is modified

Definition at line 86 of file bufrdeco_csv.c.

{
  size_t i, j, k = 0, l, latest_char = 0;
  int flag;
  char c, caux[CSV_MAXL];
 
  bufrdeco_assert (lin != NULL && tk != NULL && nt != NULL);
  
  l = strlen ( lin );
  if ( l >= CSV_MAXL || l == 0 )
    return -1;
 
  // copy original string. In this case the copy is safe
  strcpy ( caux, lin );
 
  flag = CSV_WAIT_ITEM;
 
  for ( i = 0, j = 0, k = 0; i < l; i++ )
    {
      c = caux[i]; // original char
 
      if ( ( flag & CSV_IS_CITED ) == 0 )
        {
          if ( flag & CSV_WAIT_ITEM )
            {
              if ( strchr ( CSV_SPACE, c ) != NULL )
                {
                  lin[j++] = c; // copy to target
                  continue;
                }
              else if ( c == CSV_CITE )
                {
                  lin[j++] = '\0'; // copy to target
                  flag |= CSV_IS_CITED;
                  continue;
                }
              else if ( c == CSV_SEPARATOR )
                {
                  /* item vacio */
                  tk[k++] = lin + j;
                  lin[j++] = '\0';
                  flag = CSV_WAIT_ITEM;
                  continue;
                }
              else if ( c == CSV_FINAL )
                {
                  if ( flag & CSV_FINISHED_ITEM )
                    {
                      *nt = k;
                      return 0;
                    }
                  /* al menos un token, vacio */
                  tk[k++] = lin + j;
                  lin[j] = '\0';
                  *nt = k;
                  return 0;
                }
              else
                {
                  latest_char = j;
                  tk[k] = lin + j;
                  lin[j++] = c;
                  flag = CSV_IS_ITEM;
                  continue;
                }
            }
          else if ( flag & CSV_IS_ITEM )
            {
              if ( strchr ( CSV_SPACE, c ) != NULL )
                {
                  lin[j++] = c;
                  continue;
                }
              else if ( c == CSV_SEPARATOR )
                {
                  lin[latest_char + 1] = '\0';
                  j = latest_char + 2;
                  k++;
                  flag = CSV_WAIT_ITEM | CSV_FINISHED_ITEM;
                  continue;
                }
              else if ( c == CSV_FINAL )
                {
                  lin[latest_char + 1] = '\0';
                  k++;
                  *nt = k;
                  return 0;
                }
              else
                {
                  latest_char = j++;
                }
            }
          else if ( flag & CSV_WAIT_SEPARATOR )
            {
              if ( c == CSV_SEPARATOR )
                {
                  lin[j++] = '\0';
                  k++;
                  flag = CSV_WAIT_ITEM;
                }
              else if ( c == CSV_FINAL )
                {
                  lin[j++] = '\0';
                  *nt = k + 1;
                  return 0;
                }
              continue;
            }
        }
      else /* CITED */
        {
          if ( flag & CSV_WAIT_ITEM )
            {
              tk[k] = lin + j;
              flag &= ( ~CSV_WAIT_ITEM ); // Clean wait item bit
            }
          if ( c == CSV_CITE )
            {
              if ( caux[i + 1] == CSV_CITE )
                {
                  if ( flag & CSV_IS_DOUBLE_QUOTED )
                    {
                      flag &= ( ~CSV_IS_DOUBLE_QUOTED ) ; // Clean double quoted bit
                      lin[j++] = '"';
                      i++;
                    }
                  else
                    {
                      flag |= CSV_IS_DOUBLE_QUOTED;
                      lin[j++] = '"';
                      i++;
                    }
                  continue;
                }
              else
                {
                  lin[j++] = '\0';
                  flag = CSV_WAIT_SEPARATOR;
                }
            }
          else if ( c == CSV_FINAL )
            {
              lin[j++] = '\0';
              *nt = k + 1;
              return 0;
            }
          else
            {
              lin[j++] = c;
            }
          continue;
        }
    }
  return -1;
}

References bufrdeco_assert, CSV_CITE, CSV_FINAL, CSV_FINISHED_ITEM, CSV_IS_CITED, CSV_IS_DOUBLE_QUOTED, CSV_IS_ITEM, CSV_MAXL, CSV_SEPARATOR, CSV_SPACE, CSV_WAIT_ITEM, and CSV_WAIT_SEPARATOR.

Variable Documentation

◆ CSV_CITE

const char CSV_CITE = '\"'

Definition at line 36 of file bufrdeco_csv.c.

Referenced by parse_csv_line(), and parse_csv_line2().

◆ CSV_FINAL

const char CSV_FINAL = '\n'

Definition at line 37 of file bufrdeco_csv.c.

Referenced by parse_csv_line(), and parse_csv_line2().

◆ CSV_SEPARATOR

const char CSV_SEPARATOR = ','

Definition at line 35 of file bufrdeco_csv.c.

Referenced by parse_csv_line(), and parse_csv_line2().

◆ CSV_SPACE

const char CSV_SPACE[] = " \t\r"

Definition at line 34 of file bufrdeco_csv.c.

Referenced by parse_csv_line2().

Macros

Functions

Variables

Detailed Description

Macro Definition Documentation

◆ CSV_FINISHED_ITEM

◆ CSV_IS_CITED

◆ CSV_IS_DOUBLE_QUOTED

◆ CSV_IS_ITEM

◆ CSV_IS_SEPARATOR

◆ CSV_WAIT_ITEM

◆ CSV_WAIT_SEPARATOR

Function Documentation

◆ csv_quoted_string()

◆ parse_csv_line()

◆ parse_csv_line2()

Variable Documentation

◆ CSV_CITE

◆ CSV_FINAL

◆ CSV_SEPARATOR

◆ CSV_SPACE