tpccsv.h File Reference

Header file for library libtpccsv. More...

#include "tpcclibConfig.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <math.h>
#include "tpcextensions.h"

Go to the source code of this file.

Data Structures
struct	CSV_item
struct	CSV

Functions
void	csvInit (CSV *csv)
void	csvFree (CSV *csv)
int	csvAllocate (CSV *csv, int nr)
int	csvDuplicate (CSV *csv1, CSV *csv2)
	Make a duplicate of CSV structure.
int	csvPutString (CSV *csv, const char *s, int newline)
int	csvPutDouble (CSV *csv, double v, int newline, int tointl)
int	csvPutInt (CSV *csv, int v, int newline)
int	csvRowLength (CSV *csv, int row)
int	csvSetDimensions (CSV *csv)
int	csvIsRegular (CSV *csv)
int	csvTrimRight (CSV *csv)
char *	csvCell (CSV *csv, int row, int col)
int	csvCellReplace (CSV *csv, int row, int col, const char *s)
int	csvRemoveItem (CSV *csv, int i)
int	csvRemoveEmptyRows (CSV *csv)
int	csvRemoveComments (CSV *csv)
int	csvReorg (CSV *d, TPCSTATUS *status)
int	csvTranspose (CSV *d)
int	csvList (CSV *csv, FILE *fp)
int	csvWrite (CSV *csv, int regular, FILE *fp, TPCSTATUS *status)
int	csvRead (CSV *csv, FILE *fp, TPCSTATUS *status)
int	csvPutLine (CSV *csv, const char *line, TPCSTATUS *status)
int	csvPutLineWithSpaces (CSV *csv, const char *line, TPCSTATUS *status)
int	csvCleanSpaces (CSV *csv)
int	csvFindField (CSV *csv, const char *s, int start_index)
int	csvSearchField (CSV *csv, const char *s, int start_index)

Detailed Description

Header file for library libtpccsv.

CSV and TSV data processing and file i/o

Author

Vesa Oikonen

Copyright

(c) Turku PET Centre

Definition in file tpccsv.h.

Function Documentation

csvAllocate()

int csvAllocate	(	CSV *	csv,
		int	nr )

Allocate memory for list of field items for CSV data.

Any previous content is preserved.

Precondition

Before first use initialize the CSV struct with csvInit().

Returns

enum tpcerror (TPCERROR_OK when successful).

Author

Vesa Oikonen

See also

csvInit, csvFree, csvDuplicate, csvSetDimensions, csvList

Parameters

csv	Pointer to CSV.
nr	Nr of field items to add to the list.

Definition at line 58 of file csv.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  if(nr<1) return TPCERROR_OK;
 
  /* If not allocated previously, then just allocate and thats it */
  if(csv->_item_nr==0) {
    csv->c=(CSV_item*)calloc(nr, sizeof(CSV_item));
    if(csv->c==NULL) return TPCERROR_OUT_OF_MEMORY;
    csv->_item_nr=nr;
    return TPCERROR_OK;
  }
 
  /* How many unused places there are already? */
  int unused=csv->_item_nr-csv->nr;
  /* Check if enough memory is already allocated */
  if(unused>=nr) return TPCERROR_OK;
 
  /* Ok, we have to reallocate more memory for the list */
  nr+=csv->nr;
  CSV_item *new_list;
  new_list=(CSV_item*)realloc(csv->c, nr*sizeof(CSV_item));
  if(new_list==NULL) return TPCERROR_OUT_OF_MEMORY;
  else csv->c=new_list;
  /* Initiate the list */
  for(int i=csv->_item_nr; i<nr; i++) {
    csv->c[i].row=0; csv->c[i].col=0; csv->c[i].content=NULL;
  }
  csv->_item_nr=nr;
  return TPCERROR_OK;
}

Referenced by csvDuplicate(), csvPutString(), parWriteCSV(), and tacWriteCSV().

csvCell()

char * csvCell	(	CSV *	csv,
		int	row,
		int	col )

Get the CVS field contents in specified row and column.

Returns

Returns pointer to the content string, or NULL if not found.

Author

Vesa Oikonen

See also

csvRead, csvSearchField, csvFindField, csvCellReplace, csvSetDimensions

Parameters

csv	Pointer to CSV.
row	CSV row index.
col	CSV column index.

Definition at line 358 of file csv.c.

  {
  if(csv==NULL) return((char*)NULL);
  for(int i=0; i<csv->nr; i++) 
    if(csv->c[i].row==row && csv->c[i].col==col)
      return(csv->c[i].content);
  return((char*)NULL); 
}

Referenced by csvDuplicate(), csvRemoveEmptyRows(), csvWrite(), parReadFIT(), parReadRES(), tacRead(), tacReadCarimasTxt(), tacReadInveonCSV(), tacReadMat(), tacReadQView(), tacReadSIF(), and tacReadSimple().

csvCellReplace()

int csvCellReplace	(	CSV *	csv,
		int	row,
		int	col,
		const char *	s )

Replace the value of CVS field.

Returns

tpcerror (TPCERROR_OK when successful).

Author

Vesa Oikonen

See also

csvRead, csvSearchField, csvCell, csvRemoveItem, csvTranspose

Parameters

csv	Pointer to CSV.
row	CSV row index.
col	CSV column index.
s	Field string to add; can be empty ("" or NULL).

Definition at line 380 of file csv.c.

  {
  if(csv==NULL) return(TPCERROR_FAIL);
  int ci=-1;
  for(int i=0; i<csv->nr; i++) if(csv->c[i].row==row && csv->c[i].col==col) {ci=i; break;}
  if(ci<0) return(TPCERROR_MISSING_DATA);
  /* Delete the previous value */
  free(csv->c[ci].content);
  /* Save the new item */
  if(s==NULL || *s=='\0') csv->c[ci].content=strdup(""); 
  else csv->c[ci].content=strdup(s);
  if(csv->c[ci].content==NULL) return(TPCERROR_OUT_OF_MEMORY);
  return(TPCERROR_OK); 
}

csvCleanSpaces()

int csvCleanSpaces

(

CSV *

csv

)

Removes any initial and trailing space characters from CSV. Space characters in the middle of the string are not removed.

Returns

tpcerror (TPCERROR_OK when successful).

Author

Vesa Oikonen

See also

csvRemoveEmptyLines, csvRead, strCleanSpaces

Parameters

csv	Pointer to CSV, the content of which is to be cleaned.

Definition at line 422 of file csvio.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  for(int i=0; i<csv->nr; i++) strCleanSpaces(csv->c[i].content);
  return(TPCERROR_OK);
}

Referenced by parReadCSV().

csvDuplicate()

int csvDuplicate	(	CSV *	csv1,
		CSV *	csv2 )

Make a duplicate of CSV structure.

Previous contents are copied. In the duplicate the space is allocated for row_nr times col_nr cells.

See also

csvInit, csvAllocate, csvFree

Returns

Returns TPCERROR status (0 when successful).

Parameters

csv1	Pointer to the source CSV.
csv2	Pointer to the target CSV; must be initiated; any old contents are deleted.

Definition at line 102 of file csv.c.

  {
  if(csv1==NULL || csv2==NULL) return TPCERROR_FAIL;
  if(csv1->nr<1 || csv1->_item_nr<csv1->nr || csv1->row_nr<1 || csv1->col_nr<1) return TPCERROR_NO_DATA;
 
  /* Empty the duplicate */
  csvFree(csv2);
 
  /* Allocate memory for csv2 */
  int ret=csvAllocate(csv2, csv1->row_nr*csv1->col_nr);
  if(ret!=TPCERROR_OK) return(ret);
  csv2->col_nr=csv1->col_nr;
  csv2->row_nr=csv1->row_nr;
  csv2->nr=csv2->col_nr*csv2->row_nr;
 
  /* Copy the contents */
  csv2->separator=csv1->separator;
  int ni=0, fulln=0;
  for(int ri=0; ri<csv2->row_nr; ri++)
    for(int ci=0; ci<csv2->col_nr; ci++) {
      csv2->c[ni].row=ri;
      csv2->c[ni].col=ci;
      char *cp=csvCell(csv1, ri, ci);
      if(cp==NULL) csv2->c[ni].content=NULL; else {csv2->c[ni].content=strdup(cp); fulln++;}
      ni++;
    }
  if(fulln<1) return TPCERROR_NO_DATA;
  return TPCERROR_OK;
}

csvFindField()

int csvFindField	(	CSV *	csv,
		const char *	s,
		int	start_index )

Find the CSV field with specified content. Search is case-insensitive, but otherwise match must be exact.

See also

csvSearchField, csvCell, csvRead, csvRemoveItem

Returns

field index, or <0 if not found.

Author

Vesa Oikonen

Parameters

csv	Pointer to CSV.
s	Field value to be searched for.
start_index	CSV item index [0..nr-1] from which search is started.

Definition at line 27 of file csvfind.c.

  {
  if(csv==NULL || s==NULL) return -2;
  if(start_index<0 || start_index>=csv->nr) return -2;
  for(int i=start_index; i<csv->nr; i++) if(csv->c[i].content!=NULL)
    if(strcasecmp(csv->c[i].content, s)==0) return i;
  return -1;
}

Referenced by parReadRES(), and tacRead4DM().

csvFree()

void csvFree

(

CSV *

csv

)

Free memory allocated for CSV data. All contents are destroyed.

Precondition

Before first use initialize the CSV struct with csvInit().

Author

Vesa Oikonen

See also

csvInit, csvAllocate, csvWrite

Parameters

csv	Pointer to CSV.

Definition at line 38 of file csv.c.

  {
  if(csv==NULL) return;
  for(int i=0; i<csv->_item_nr; i++) free(csv->c[i].content);
  free(csv->c);
  csvInit(csv);
}

Referenced by csvDuplicate(), parRead(), parWriteCSV(), tacRead(), and tacWriteCSV().

csvInit()

void csvInit

(

CSV *

csv

)

Initiate the CSV struct before any use.

Author

Vesa Oikonen

See also

csvFree, csvAllocate, csvRead

Parameters

csv	Pointer to CSV.

Definition at line 22 of file csv.c.

  {
  if(csv==NULL) return;
  csv->c=NULL; csv->_item_nr=csv->nr=csv->row_nr=csv->col_nr=0;
  csv->separator='\t'; //(char)0;
}

Referenced by csvFree(), parRead(), parWriteCSV(), tacFormatDetermine(), tacRead(), and tacWriteCSV().

csvIsRegular()

int csvIsRegular

(

CSV *

csv

)

Check whether CSV is regular, that is, each row contain the same number of columns.

Author

Vesa Oikonen

See also

csvTranspose, csvTrimRight, csvSetDimensions, csvRowLength, csvRead, csvDuplicate

Returns

Returns 1 if CSV is regular, 0 if not.

Parameters

csv	Pointer to CSV

Definition at line 292 of file csv.c.

  {
  if(csv==NULL) return(0);
  if(csv->nr<2) return(1);
  int i, r, n=0, m=0;
  i=0; r=csv->c[i].row; m++;
  for(i=1; i<csv->nr; i++) {
    if(r==csv->c[i].row) {
      m++; 
      continue;}
    if(n>0 && m!=n) return(0);
    r=csv->c[i].row; n=m; m=1;
  }
  if(n>0 && m!=n) return(0);
  return(1);
}

Referenced by tacRead(), tacReadMat(), and tacReadSimple().

csvList()

int csvList	(	CSV *	csv,
		FILE *	fp )

Write CSV data as a tab separated list into file opened for writing. List contains the cell rows, columns, and values.

Data is not sorted, and cell contents are written as they are, that is, no conversions for decimal separator is done here.

Returns

enum tpcerror (TPCERROR_OK when successful).

Author

Vesa Oikonen

See also

csvWrite, csvSetDimensions, csvTrimRight

Parameters

csv	Pointer to CSV structure, contents of which are to be written.
fp	Output file pointer; usually stdout.

Definition at line 27 of file csvio.c.

  {
  if(fp==NULL) return TPCERROR_CANNOT_WRITE;
  if(csv==NULL || csv->nr<1) return TPCERROR_NO_DATA;
 
  for(int i=0; i<csv->nr; i++)
    if(fprintf(fp, "%d\t%d\t%s\n", 1+csv->c[i].row, 1+csv->c[i].col, csv->c[i].content)<5)
      return TPCERROR_CANNOT_WRITE;
 
  return(TPCERROR_OK);
}

Referenced by tacReadSIF().

csvPutDouble()

int csvPutDouble	(	CSV *	csv,
		double	v,
		int	newline,
		int	tointl )

Add specified double value as string into CSV.

Returns

tpcerror (TPCERROR_OK when successful).

Precondition

Before first use initialize the CSV struct with csvInit().

Author

Vesa Oikonen

See also

csvInit, csvPutString, csvPutInt, csvPutLine

Parameters

csv	Pointer to initiated CSV; previous contents are not changed.
v	Double value to add; NaN is added as an empty field.
newline	New line (1) or same line (0).
tointl	Convert (1) or do not convert (0) commas to semicolons and dots to commas.

Definition at line 193 of file csv.c.

  {
  int ret;
  if(isnan(v)) {
    ret=csvPutString(csv, "", newline);
  } else {
    char s[128];
    sprintf(s, "%g", v); if(tointl) strReplaceChar(s, '.', ',');
    ret=csvPutString(csv, s, newline);
  }
  return ret;
}

Referenced by parWriteCSV(), and tacWriteCSV().

csvPutInt()

int csvPutInt	(	CSV *	csv,
		int	v,
		int	newline )

Add specified integer value as string into CSV.

Returns

tpcerror (TPCERROR_OK when successful).

Precondition

Before first use initialize the CSV struct with csvInit().

Author

Vesa Oikonen

See also

csvInit, csvPutDouble, csvPutString, csvPutLine

Parameters

csv	Pointer to initiated CSV; previous contents are not changed
v	Integer value to add; use csvPutString() to add an empty field
newline	New line (1) or same line (0)

Definition at line 222 of file csv.c.

  {
  int ret;
  char s[128];
  sprintf(s, "%d", v);
  ret=csvPutString(csv, s, newline);
  return ret;
}

Referenced by parWriteCSV().

csvPutLine()

int csvPutLine	(	CSV *	csv,
		const char *	line,
		TPCSTATUS *	status )

Process a given text line (string) to add a new row of fields to CSV, using as field delimiter the character specified in CSV structure.

Returns

tpcerror (TPCERROR_OK when successful).

Precondition

Before first use initialize the CSV structure with csvInit().

Postcondition

Remember to free the memory in CSV after last use with csvFree().

Author

Vesa Oikonen

See also

csvInit, csvFree, csvWrite, csvPutString, csvPutInt, csvPutLineWithSpaces

Parameters

csv	Pointer to initiated CSV; previous contents are not changed.
line	Pointer to the CSV file line to be processed.
status	Pointer to status data; enter NULL if not needed.

Definition at line 251 of file csvio.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  if(line==NULL || strlen(line)<1) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_NO_DATA);
    return TPCERROR_NO_DATA;
  }
  int verbose=0; if(status!=NULL) verbose=status->verbose;
  if(verbose>10) printf("%s():\n", __func__);
  if(verbose>12) printf("'%s'\n", line);
 
  //size_t len=strlen(line);
  char delimiter=csv->separator;
 
  /* Space is not supported here */
  if(delimiter==' ') {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_INVALID_SEPARATOR);
    return TPCERROR_INVALID_SEPARATOR;
  }
 
  int ret, i, last_was_delim=0;
  int field_nr=0;
  char *cptr=(char*)line, *lptr, *s;
  /* If the first character is delimiter, we have had an empty field */
  if(*cptr==delimiter) {
    if(verbose>20) printf("first char is delimiter.\n");
    ret=csvPutString(csv, "", !field_nr); if(ret!=TPCERROR_OK) {
      statusSet(status, __func__, __FILE__, __LINE__, ret);
      return ret;
    }
    last_was_delim=1; cptr++; field_nr++;
  }
  /* Read all fields */
  int single_quotation=0;
  int double_quotation=0;
  lptr=cptr;
  while(*cptr && *lptr) {
    if(verbose>20) printf("cptr='%s'\n", cptr);
    /* Read next field */
    lptr=cptr; i=0;
    while(*lptr) {
      // jump over quoted sequences
      if(*lptr=='\'') {
        if(single_quotation==0 && strchr(lptr+1, '\'')!=NULL) single_quotation=1;
        else single_quotation=0;
        lptr++; i++; continue;
      }
      if(*lptr=='\"') {
        if(double_quotation==0 && strchr(lptr+1, '\"')!=NULL) double_quotation=1;
        else double_quotation=0;
        lptr++; i++; continue;
      }
      if(single_quotation==1 || double_quotation==1) {lptr++; i++; continue;}
      // if this character is the delimiter, then stop
      if(*lptr==delimiter) break;
      // otherwise continue search
      lptr++; i++;
    }
    s=strndup(cptr, i);
    if(verbose>20) printf("  s='%s'\n", s);
    ret=csvPutString(csv, s, !field_nr); if(ret!=TPCERROR_OK) {
      statusSet(status, __func__, __FILE__, __LINE__, ret);
      free(s); return ret;
    }
    free(s); field_nr++;
    if(*lptr==delimiter) {
      last_was_delim=1; cptr+=(i+1);
    } else {last_was_delim=0; cptr+=(i+1);}
  }
  if(verbose>20) printf("line finished.\n");
  /* If the last character is delimiter, we have an empty field in the end */
  if(last_was_delim) {
    if(verbose>20) printf("last char is delimiter.\n");
    ret=csvPutString(csv, "", !field_nr); if(ret!=TPCERROR_OK) {
      statusSet(status, __func__, __FILE__, __LINE__, ret);
      return ret;
    }
    field_nr++;
  }
  if(verbose>20) printf("ending %s()\n", __func__);
  statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_OK);
  return(TPCERROR_OK);
}

Referenced by csvRead().

csvPutLineWithSpaces()

int csvPutLineWithSpaces	(	CSV *	csv,
		const char *	line,
		TPCSTATUS *	status )

Process a given text line (string) to add a new row of fields to CSV, using spaces as field delimiters, independent on what is told in CSV struct.

Returns

tpcerror (TPCERROR_OK when successful).

Precondition

Before first use initialize the CSV struct with csvInit().

Postcondition

Remember to free the memory in CSV after last use with csvFree().

Author

Vesa Oikonen

See also

csvPutLine, csvRemoveEmptyLines, csvRead

Parameters

csv	Pointer to initiated CSV; previous contents are not changed.
line	Pointer to the CSV file line to be processed.
status	Pointer to status data; enter NULL if not needed.

Definition at line 351 of file csvio.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  if(line==NULL || strlen(line)<1) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_NO_DATA);
    return TPCERROR_NO_DATA;
  }
  int verbose=0; if(status!=NULL) verbose=status->verbose;
  if(verbose>10) printf("%s():\n", __func__);
  if(verbose>12) printf("'%s'\n", line);
 
  char *cptr=(char*)line;
  char *lptr=cptr, *s;
  int single_quotation=0;
  int double_quotation=0;
  int ret;
  size_t j;
  int field_nr=0;
  while(*cptr && *lptr) {
    if(verbose>20) printf("cptr='%s'\n", cptr);
    // Pass the spaces
    j=strspn(cptr, " \t\n\r"); cptr+=j; if(!cptr) break;
    // Find the end of token
    lptr=cptr; j=0;
    while(*lptr) {
      // jump over quoted sequences */
      if(*lptr=='\'') {
        if(single_quotation==0 && strchr(lptr+1, '\'')!=NULL) single_quotation=1;
        else single_quotation=0;
        lptr++; j++; continue;
      }
      if(*lptr=='\"') {
        if(double_quotation==0 && strchr(lptr+1, '\"')!=NULL) double_quotation=1;
        else double_quotation=0;
        lptr++; j++; continue;
      }
      if(single_quotation==1 || double_quotation==1) {lptr++; j++; continue;}
      // if this character is the delimiter, then stop
      if(*lptr==' ') break;
      // otherwise continue search
      lptr++; j++;
    }
    if(j==0) break;
    s=strndup(cptr, j);
    if(verbose>20) printf("  s='%s'\n", s);
    ret=csvPutString(csv, s, !field_nr); if(ret!=TPCERROR_OK) {
      statusSet(status, __func__, __FILE__, __LINE__, ret);
      free(s); return ret;
    }
    free(s); cptr+=j; field_nr++;
    if(verbose>20) printf("  csv.nr=%d\n", csv->nr);
  }
 
  statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_OK);
  return(TPCERROR_OK);
}

Referenced by csvRead().

csvPutString()

int csvPutString	(	CSV *	csv,
		const char *	s,
		int	newline )

Add specified string to CSV.

Returns

tpcerror (TPCERROR_OK when successful).

Precondition

Before first use initialize the CSV struct with csvInit().

Author

Vesa Oikonen

See also

csvInit, csvPutDouble, csvPutInt, csvPutLine

Parameters

csv	Pointer to initiated CSV; previous contents are not changed.
s	Field string to add; can be empty ("" or NULL).
newline	New line (1) or same line (0).

Definition at line 144 of file csv.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  /* Allocate memory, if needed, and if needed, then a bigger chunk */
  int avail=csv->_item_nr-csv->nr;
  if(avail<1) {
    int ret=csvAllocate(csv, 10);
    if(ret!=TPCERROR_OK) return ret;
  }
  /* Write the item */
  if(s==NULL || *s=='\0') csv->c[csv->nr].content=strdup(""); 
  else csv->c[csv->nr].content=strdup(s);
  if(csv->c[csv->nr].content==NULL) return TPCERROR_OUT_OF_MEMORY;
  /* Set the row and column number */
  int r, c;
  if(csv->nr==0) {r=c=0;}
  else {
    if(newline!=0) {
      c=0;
      r=csv->c[csv->nr-1].row+1;
    } else {
      c=csv->c[csv->nr-1].col+1;
      r=csv->c[csv->nr-1].row;
    }
  }
  csv->c[csv->nr].col=c; csv->c[csv->nr].row=r;
  /* Update row_nr and max col nr, if necessary */
  c++; r++;
  if(c>csv->col_nr) csv->col_nr=c;
  if(r>csv->row_nr) csv->row_nr=r;
 
  csv->nr++;
  return TPCERROR_OK;
}

Referenced by csvPutDouble(), csvPutInt(), csvPutLine(), csvPutLineWithSpaces(), parWriteCSV(), tacReadSIF(), and tacWriteCSV().

csvRead()

int csvRead	(	CSV *	csv,
		FILE *	fp,
		TPCSTATUS *	status )

Read CSV file contents into CSV structure, allocating memory as needed.

Lines consisting only of space characters, including tabs, are not read. Partial support for spaces as delimiters.

Returns

enum tpcerror (TPCERROR_OK when successful).

Precondition

Before first use initialize the CSV structure with csvInit().

Postcondition

After last use free memory in the CSV structure with csvFree().

Bug

File is assumed to be relatively well-formatted. Specifically, both tabs and spaces must not be used as field delimiters inside one file.

Author

Vesa Oikonen

See also

csvInit, csvFree, csvWrite, csvSearchField, csvCell, csvCellReplace, csvIsRegular

Parameters

csv	Pointer to CSV to read into; any previous contents of CSV are preserved.
fp	Input file pointer.
status	Pointer to status data; enter NULL if not needed.

Definition at line 124 of file csvio.c.

  {
  statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_FAIL);
  if(csv==NULL || fp==NULL) return TPCERROR_FAIL;
  int verbose=0; if(status!=NULL) verbose=status->verbose;
  if(verbose>10) printf("%s()\n", __func__);
 
  /* Get the size of the ASCII part of the file */
  size_t fsize=asciiFileSize(fp, NULL);
  if(verbose>11) printf("  ASCII size := %d\n", (int)fsize);
  /* If ASCII part is too small, then lets consider that an error */
  if(fsize<1) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_NO_DATA);
    return TPCERROR_NO_DATA;
  }
  /* If ASCII part is too large, then lets consider that an error */
  if(fsize>50000000) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_TOO_BIG);
    return TPCERROR_TOO_BIG;
  }
  /* Read that to a string */
  rewind(fp);
  char *data;
  data=asciiFileRead(fp, NULL, fsize+1); rewind(fp);
  if(data==NULL) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_NO_DATA);
    return TPCERROR_NO_DATA;
  }
  if(verbose>20) printf("  ASCII file read\n");
  
  /* Read one line at a time from the string and determine the field and decimal separators */
  int i=0, j;
  int tab_nr=0, sem_nr=0, com_nr=0, dot_nr=0, spa_nr=0;
  char *cptr, *line, *lptr;
  cptr=data;
  while((line=strTokenDup(cptr, "\n\r", &j))!=NULL) {
    if(verbose>80) printf("line='%s'\n", line);
    /* If line starts with '#' then jump over it */
    if(line[0]=='#') {free(line); cptr+=j; continue;}
    /* If line contains only space characters then jump over it */
    if(strIsSpaceOnly(line)) {free(line); cptr+=j; continue;}
    /* Compute the nr of dots, commas etc outside quotes */
    lptr=line; while((lptr=strstrNoQuotation(lptr, "\t"))!=NULL) {tab_nr++;lptr++;}
    lptr=line; while((lptr=strstrNoQuotation(lptr, ";"))!=NULL) {sem_nr++; lptr++;}
    lptr=line; while((lptr=strstrNoQuotation(lptr, ","))!=NULL) {com_nr++; lptr++;}
    lptr=line; while((lptr=strstrNoQuotation(lptr, "."))!=NULL) {dot_nr++; lptr++;}
    lptr=line; while((lptr=strstrNoQuotation(lptr, " "))!=NULL) {spa_nr++; lptr++;}
    free(line); cptr+=j; i++;
  }
  if(verbose>10) {
    printf("dataline_nr := %d\n", i);
    printf("semicolon_nr := %d\n", sem_nr);
    printf("tabulator_nr := %d\n", tab_nr);
    printf("dot_nr := %d\n", dot_nr);
    printf("comma_nr := %d\n", com_nr);
    printf("space_nr := %d\n", spa_nr);
  }
  if(sem_nr==0 && tab_nr==0 && dot_nr==0 && com_nr==0 && spa_nr==0) {
    csv->separator='\t'; // the default
  } else if(sem_nr>0) {
    // If at least one semi-colon, then assume that it is the field separator
    csv->separator=';';
  } else if(tab_nr>0) {
    // If at least one tab, then assume that it is the field separator
    csv->separator='\t';
  } else if(spa_nr==0) {
    // If no spaces, then comma must be the field separator
    csv->separator=',';
  } else {
    // Spaces exist, so is space or comma the field separator ?
    if(com_nr==0) {
      // No commas, thus space is probably field separator
      csv->separator=' ';
    } else if(dot_nr>0) {
      // Dots and commas exist, probably decimal point, and comma as field separator
      csv->separator=',';
    } else {
      // No dots, but commas and spaces; lets assume that the more frequent one is the field separator
      if(com_nr>spa_nr) csv->separator=','; else csv->separator=' ';
    }
  }
  if(verbose>10) {
    if(csv->separator=='\t') printf("field_separator := tab\n");
    else if(csv->separator==' ') printf("field_separator := space\n");
    else printf("field_separator := %c\n", csv->separator);
  }
 
  /* Copy field values into CSV */
  cptr=data; i=0; int ret;
  while((line=strTokenDup(cptr, "\n\r", &j))!=NULL) {
    /* If line starts with '#' then jump over it */
    if(line[0]=='#') {free(line); cptr+=j; continue;}
    /* If line contains only space characters then jump over it */
    if(strIsSpaceOnly(line)) {free(line); cptr+=j; continue;}
    /* Write contents into CSV as a new data row */
    if(csv->separator!=' ') ret=csvPutLine(csv, line, status);
    else ret=csvPutLineWithSpaces(csv, line, status);
    if(verbose>1 && ret!=0) fprintf(stderr, "Warning: cannot read line %d: '%s'.\n", i, line);
    /* Prepare for the next line */
    free(line); cptr+=j; i++;
  }
  if(i==0) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_NO_DATA);
    return TPCERROR_NO_DATA;
  }
 
  free(data);
  statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_OK);
  return(TPCERROR_OK);
}

Referenced by parRead(), tacFormatDetermine(), and tacRead().

csvRemoveComments()

int csvRemoveComments

(

CSV *

csv

)

Remove those cells from CSV structure whose contents start with '#'.

Returns

enum tpcerror (TPCERROR_OK when successful).

See also

csvIsRegular, csvRemoveEmptyRows

Parameters

csv	Pointer to CSV.

Definition at line 470 of file csv.c.

  {
  if(csv==NULL) return(TPCERROR_FAIL);
  if(csv->nr<1) return(TPCERROR_NO_DATA);
 
  int i=csv->nr-1;
  while(i>=0) {
    if(csv->c[i].content!=NULL && csv->c[i].content[0]=='#') csvRemoveItem(csv, i);
    i--;
  }
  return(TPCERROR_OK);
//  return(csvRemoveEmptyRows(csv));
}

csvRemoveEmptyRows()

int csvRemoveEmptyRows

(

CSV *

csv

)

Remove empty rows from CSV. Rows that contain only empty cells or cells that start with '#' are considered empty.

Returns

enum tpcerror (TPCERROR_OK when successful).

See also

csvRemoveComments, csvIsRegular, csvDuplicate, csvCellReplace

Parameters

csv	Pointer to CSV.

Definition at line 434 of file csv.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  if(csv->nr<1 || csv->row_nr<1 || csv->col_nr<1) return TPCERROR_NO_DATA;
 
  /* Check the rows */
  int ri=csv->row_nr-1;
  while(ri>=0) {
    int n=0;
    for(int ci=0; ci<csv->col_nr; ci++) {
      char *cp=csvCell(csv, ri, ci); if(cp==NULL) continue;
      if(*cp=='\0' || *cp=='#') continue;
      n++;
    }
    if(n==0) { // this line is empty
      int i=csv->nr-1;
      while(i>=0) {
        if(csv->c[i].row==ri) csvRemoveItem(csv, i);
        i--;
      }
      for(i=0; i<csv->nr; i++) if(csv->c[i].row>ri) csv->c[i].row--;
      csv->row_nr--;
    }
    ri--;
  }
  return(TPCERROR_OK);
}

csvRemoveItem()

int csvRemoveItem	(	CSV *	csv,
		int	i )

Remove item from CSV data structure.

Returns

enum tpcerror (TPCERROR_OK when successful).

See also

csvCellReplace, csvFree, csvDuplicate, csvRemoveEmptyRows

Parameters

csv	Pointer to CSV.
i	Index of item to delete.

Definition at line 409 of file csv.c.

  {
  if(csv==NULL || i<0) return TPCERROR_FAIL;
  if(csv->nr<1 || i>=csv->nr) return TPCERROR_NO_DATA;
  free(csv->c[i].content); csv->c[i].content=NULL;
  for(int j=i+1; j<csv->nr; j++) {
    csv->c[j-1].row=csv->c[j].row;
    csv->c[j-1].col=csv->c[j].col;
    csv->c[j-1].content=csv->c[j].content;
  }
  csv->nr--; csv->c[csv->nr].content=NULL;
  return(TPCERROR_OK); 
}

Referenced by csvRemoveComments(), csvRemoveEmptyRows(), and csvTrimRight().

csvReorg()

int csvReorg	(	CSV *	d,
		TPCSTATUS *	status )

Sort CSV data array by increasing row and column numbers.

Returns

enum tpcerror (TPCERROR_OK when successful).

See also

csvTranspose, csvList, csvAllocate, csvWrite

Parameters

d	Pointer to CSV structure.
status	Pointer to status data; enter NULL if not needed.

Definition at line 503 of file csv.c.

  {
  int verbose=0; if(status!=NULL) verbose=status->verbose;
  if(verbose>0) printf("%s()\n", __func__);
  /* Check that required data exists */
  if(d==NULL || d->nr<1) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_NO_DATA);
    return TPCERROR_NO_DATA;
  }
  statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_OK);
  if(d->nr<2) return TPCERROR_OK;
  qsort(d->c, d->nr, sizeof(CSV_item), csvReorgQSort);
  return TPCERROR_OK;
}

Referenced by csvTranspose(), and tacReadSIF().

csvRowLength()

int csvRowLength	(	CSV *	csv,
		int	row )

Count the nr of columns on specified CSV row.

Author

Vesa Oikonen

See also

csvCell, csvRead, csvIsRegular, csvSetDimensions

Returns

The number of columns.

Parameters

csv	Pointer to CSV
row	CSV row index

Definition at line 244 of file csv.c.

  {
  if(csv==NULL) return(0);
//  int i, n=0;
//  for(i=0; i<csv->nr; i++) if(csv->c[i].row==row) n++;
//  return(n); 
  int n=-1;
  for(int i=0; i<csv->nr; i++) if(csv->c[i].row==row && csv->c[i].col>n) n=csv->c[i].col;
  return(1+n); 
}

Referenced by csvWrite(), parReadFIT(), parReadRES(), tacReadDFT(), tacReadSIF(), and tacReadSimple().

csvSearchField()

int csvSearchField	(	CSV *	csv,
		const char *	s,
		int	start_index )

Find the CSV field which contains the given search string. Search is case-insensitive.

See also

csvFindField, csvCell, csvRead, csvRemoveItem

Returns

field index, or <0 if not found.

Author

Vesa Oikonen

Parameters

csv	Pointer to CSV.
s	String to be searched for inside field values.
start_index	CSV field index [0..nr-1] from which search is started.

Definition at line 50 of file csvfind.c.

  {
  if(csv==NULL || s==NULL) return -2;
  if(start_index<0 || start_index>=csv->nr) return -2;
  for(int i=start_index; i<csv->nr; i++) {
    if(csv->c[i].content==NULL || csv->c[i].content[0]=='\0') {
      if(*s=='\0') return i; else continue;
    }
    if(*s=='\0') continue;
    if(strcasestr(csv->c[i].content, s)!=NULL) return i;
  }
  return -1;
}

Referenced by parReadFIT(), and tacRead4DM().

csvSetDimensions()

int csvSetDimensions

(

CSV *

csv

)

Set the number of rows and columns in CSV table, based on the row and column stored with each cell. CSV may not be regular, thus maximum row and column numbers are used.

Author

Vesa Oikonen

See also

csvIsRegular, csvRowLength, csvRead, csvRemoveComments

Returns

enum tpcerror (TPCERROR_OK when successful).

Parameters

csv	Pointer to CSV structure.

Definition at line 269 of file csv.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  if(csv->nr<1) {csv->col_nr=0; csv->row_nr=0; return(TPCERROR_OK);}
 
  int maxri=0, maxci=0;
  for(int i=0; i<csv->nr; i++) {
    if(csv->c[i].row>maxri) maxri=csv->c[i].row;
    if(csv->c[i].col>maxci) maxci=csv->c[i].col;
  }
  csv->col_nr=1+maxci; csv->row_nr=1+maxri;
  return(TPCERROR_OK);
}

Referenced by csvTranspose().

csvTranspose()

int csvTranspose

(

CSV *

csv

)

Transpose data in CSV structure.

To work properly, CSV should not contain comment cells/lines.

Returns

enum tpcerror (TPCERROR_OK when successful).

See also

csvIsRegular, csvDuplicate, csvCellReplace, csvReorg, csvRmComments

Parameters

csv	Pointer to CSV.

Definition at line 530 of file csv.c.

  {
  if(csv==NULL) return(TPCERROR_FAIL);
  if(csv->nr<1) return(TPCERROR_NO_DATA);
 
  /* Get dimensions */
  {
    int ret=csvSetDimensions(csv);
    if(ret!=TPCERROR_OK) return(ret);
  }
  if(csv->row_nr<1 || csv->col_nr<1) return(TPCERROR_NO_DATA);
 
  /* If only one row and column, then nothing to do */
  if(csv->row_nr==1 && csv->col_nr==1) return(TPCERROR_OK);
 
  /* Switch column and row numbers */
  for(int i=0; i<csv->nr; i++) {
    int cr=csv->c[i].col;
    csv->c[i].col=csv->c[i].row;
    csv->c[i].row=cr;
  }
  {
    int cn=csv->col_nr;
    csv->col_nr=csv->row_nr;
    csv->row_nr=cn;
  }
 
  /* Sort CSV data by rows and columns */
  {
    int ret=csvReorg(csv, NULL);
    if(ret!=TPCERROR_OK) return(ret);
  }
 
  return(TPCERROR_OK);
}

csvTrimRight()

int csvTrimRight

(

CSV *

csv

)

Remove empty cells from the right side of CVS table until a non-empty column is reached.

Resulting CSV may still not be regular.

Author

Vesa Oikonen

See also

csvIsRegular, csvRowLength, csvRead, csvTranspose

Returns

enum tpcerror (TPCERROR_OK when successful).

Parameters

csv	Pointer to CSV

Definition at line 321 of file csv.c.

  {
  if(csv==NULL) return TPCERROR_FAIL;
  if(csv->nr<1 || csv->col_nr<2 || csv->row_nr<2) return TPCERROR_NO_DATA;
 
  int ci=csv->col_nr-1;
  while(ci>0) {
    /* Does this column contain any data? */
    int n=0;
    for(int i=0; i<csv->nr; i++) if(csv->c[i].col==ci) {
      if(csv->c[i].content!=NULL && csv->c[i].content[0]!='\0' && csv->c[i].content[0]!='#') n++;
    }
    if(n>0) break; // yes
    /* No, so let's delete the last column */
    int i=csv->nr-1;
    while(i>=0) {
      if(csv->c[i].col==ci) {
        if(csvRemoveItem(csv, i)!=TPCERROR_OK) return TPCERROR_FAIL;
      }
      i--;
    }
    csv->col_nr--;
    /* Try the previous column */
    ci--;
  }
  return(TPCERROR_OK);
}

Referenced by tacReadSimple().

csvWrite()

int csvWrite	(	CSV *	csv,
		int	regular,
		FILE *	fp,
		TPCSTATUS *	status )

Write CSV data into file opened for writing, using the column separator specified inside CSV structure.

Field contents are written as they are, that is, no conversions for decimal separator is done here.

Returns

enum tpcerror (TPCERROR_OK when successful).

Author

Vesa Oikonen

See also

csvRead, csvList, csvSetDimensions, csvTrimRight

Parameters

csv	Pointer to CSV structure, contents of which are to be written.
regular	Forced regularization (1), or not (0); if regularized, then row_nr rows are written, each with col_nr columns; otherwise, empty rows are not written, and missing cells are not written to the end of rows.
fp	Output file pointer.
status	Pointer to status data; enter NULL if not needed.

Definition at line 52 of file csvio.c.

  {
  int verbose=0; if(status!=NULL) verbose=status->verbose;
  if(fp==NULL) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_CANNOT_WRITE);
    return TPCERROR_CANNOT_WRITE;
  }
  if(csv==NULL || csv->nr<1) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_NO_DATA);
    return TPCERROR_NO_DATA;
  }
  if(csv->separator!=',' && csv->separator!=';' && csv->separator!='\t' && csv->separator!=' ') {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_INVALID_SEPARATOR);
    return TPCERROR_INVALID_SEPARATOR;
  }
  if(verbose>10) {
    printf("%s():\n", __func__);
    printf("regular := %d\n", regular);
    printf("csv_nr := %d\n", csv->nr);
    printf("csv_row_nr := %d\n", csv->row_nr);
    printf("csv_col_nr := %d\n", csv->col_nr);
    if(csv->separator=='\t') printf("csv_separator := tab\n");
    else if(csv->separator==' ') printf("csv_separator := space\n");
    else printf("csv_separator := '%c'\n", csv->separator);
  }
 
  /* Write in file */
  int wn=0;
  for(int ri=0; ri<csv->row_nr; ri++) {
    int n=csv->col_nr;
    if(regular==0) {n=csvRowLength(csv, ri); if(n==0) continue;}
    for(int ci=0; ci<n; ci++) {
      if(ci>0) wn+=fprintf(fp, "%c", csv->separator);
      char *cptr=csvCell(csv, ri, ci);
      if(cptr!=NULL) wn+=fprintf(fp, "%s", cptr);
    }
    wn+=fprintf(fp, "\n");
  }
  if(wn<1) {
    statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_CANNOT_WRITE);
    return TPCERROR_CANNOT_WRITE;
  }
 
  statusSet(status, __func__, __FILE__, __LINE__, TPCERROR_OK);
  return(TPCERROR_OK);
}

Referenced by parRead(), parReadFIT(), parReadRES(), parWriteCSV(), tacRead(), tacReadSimple(), and tacWriteCSV().