binning/binning.c

/******************************************************************/
/*                                                                */
/* Binning Routine fuer Histogramme                               */
/*                                                                */
/* little program to produce histograms from space separated      */
/* input data files. Output is a space separated histogram.       */
/*                                                                */
/* example:                                                       */
/* binning                                                        */
/*                                                                */
/* compile: gcc binning.c -o binning                              */
/*                                                                */
/*                                                                */
/* (c) Markus Hoffmann 1995        Letzte Bearbeitung: 30.03.1996 */
/* (c) Markus Hoffmann 1995-2010   Letzte Bearbeitung: 30.03.2010 */
/*                                                                */
/******************************************************************/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <ctype.h>
#ifdef WINDOWS
#define EX_OK 0
#define EX_USAGE 0
#define EX_NOINPUT 0
#define EX_CANTCREAT 0
#else
#include <sysexits.h>
#endif
#include <math.h>

#define VERSION "1.03"
#define VERSION_DATE "Mi 12. Mai 10:18:39 CEST 2010"

#define FALSE    0
#define TRUE     (!FALSE)
#define PI       3.141592654
#define E        2.718281828
#define round(a) ((int)(a+0.5))
#define min(a,b) ((a<b)?a:b)
#define max(a,b) ((a>b)?a:b)
#define sgn(x)   ((x<0)?-1:1)
#define rad(x)   (PI*x/180)
#define deg(x)   (180*x/PI)

char *progname="binning";
int verbose=1;
char *ifilename="input.dat";
char *ofilename="output.dat";
int *b;
int n=200;
int col=1;
int doauto=FALSE;
double anfang=-.6;
double ende=.6;

int wort_sep (char *,char,int,char *, char *);
void xtrim(char *,int , char *);

char *lineinput(FILE *n, char *line) {   /* liest eine ganze Zeile aus einem ASCII-File ein */
  char c; int i=0;
 
  while((c=fgetc(n))!=EOF) {
    if(c=='\n') {
      line[i]='\0';
      return line;
    } else line[i++]=c;
  }
  line[i]='\0';
  return line;
}

void usage(){
  printf("Usage: binning [options] filename\n\n");
  printf("--input filename   --- Einlesefile  [%s]\n",ifilename);
  printf("-o, --output file  --- Outputfile   [%s]\n",ofilename);
  printf("--anfang x         --- Startbereich [%g]\n",anfang);
  printf("--ende  x          --- Endbereich   [%g]\n",ende); 
  printf("--auto             --- Auto-Bereich [%s]\n",(doauto?"on":"off"));
  printf("-n      x          --- Anzahl Bins  [%d]\n",n);
  printf("-c, --col   x      --- Spalte       [%d]\n",col);
  puts(  "-h --help          --- Show usage\n"
         "-v                 --- be more verbose\n"
         "-q                 --- be more quiet");
}
void intro(){
  puts("*********************************************************\n"
       "*             BINNING               V." VERSION "              *\n"
       "*		       (c) by Markus Hoffmann 1997-2010 *\n"
       "* version date: " VERSION_DATE "		*\n"
       "*********************************************************\n"); 
}


int main(int argc, char *argv[]) {
  FILE *dptr,*dptr2;
     
  double a,mean=0;
  int anzdata=0,over=0,under=0;
  double dmin=1e300,dmax=-1e300;
  char zz[2560],w1[2560];
  int i,j=0,k;
  int DoExit=0;
  int doversion=0;

  progname=argv[0];

  if(argc>1) {
    int count;
    for(count=1;count<argc;count++) {  /* Kommandozeile bearbeiten   */  
      if(strcmp(argv[count],"-h")==0 || strcmp(argv[count],"--help")==0) {
        usage();
        DoExit=1;
      } 
      else if(strcmp(argv[count],"-v")==0)      verbose++;
      else if(strcmp(argv[count],"-q")==0)      verbose--;
      else if(strcmp(argv[count],"--input")==0) ifilename=argv[++count]; 
      else if(strcmp(argv[count],"-o")==0 || 
              strcmp(argv[count],"--output")==0) ofilename=argv[++count];
      else if(strcmp(argv[count],"-n")==0)       n=atoi(argv[++count]); 
      else if(strcmp(argv[count],"--anfang")==0) anfang=atof(argv[++count]); 
      else if(strcmp(argv[count],"--ende")==0)   ende=atof(argv[++count]);    
      else if(strcmp(argv[count],"--auto")==0)   doauto=TRUE; 
      else if(strcmp(argv[count],"--version")==0)   doversion=TRUE; 
      else if(strcmp(argv[count],"-c")==0 || 
              strcmp(argv[count],"--col")==0)    col=atoi(argv[++count]);
      else if(*(argv[count])=='-')
        printf("%s: Unknown commandline option <%s> ignored.\n",progname,argv[count]);
      else 
        ifilename=argv[count];
    }
  } else {
    intro();
    usage();
    return(EX_USAGE);
  } /* zu weinig Parameter */
  if(doversion) intro();
  if(DoExit) return(EX_OK);


  b=(int *)malloc(n*sizeof(int));
 
  /* ANzal der Daten zaehlen  */
  dptr2=fopen(ifilename,"r");
  if(!dptr2) {
    printf("%s: ERROR, %s could not be read.\n",progname,ifilename);
    return(EX_NOINPUT);
  }
  if(verbose) printf("PASS 1 %s [",ifilename);
  while(feof(dptr2)==0) {
    lineinput(dptr2,zz); 
    xtrim(zz,0,zz);
    if(strlen(zz) && *zz!='%' && *zz!='#') {
      for(i=0;i<col;i++) {
        wort_sep(zz,' ',0,w1,zz);
      }
      a=atof(w1);
      dmin=min(dmin,a);
      dmax=max(dmax,a);
      mean+=a;
      anzdata++;
    } 
  }
  if(anzdata) mean/=anzdata;
  if(verbose) printf("] %d min=%g  max=%g mean=%g\n",anzdata,dmin,dmax,mean);

 
  if(doauto==TRUE) {
    anfang=dmin;
    ende=dmax;
  }

  rewind(dptr2); /* prepare for PASS 2 */

  if(verbose) printf("PASS 2 %s [",ifilename);
  while(feof(dptr2)==0) {
    lineinput(dptr2,zz); 
    xtrim(zz,0,zz);
    if(strlen(zz) && zz[0]!='%' && *zz!='#') {
      for(i=0;i<col;i++) {
        wort_sep(zz,' ',0,w1,zz);
      }
      a=atof(w1);
      i=(int)(n*(a-anfang)/(ende-anfang));
      if(i<0) under++;
      else if (i>=n) over++;
      else b[(int)(n*(a-anfang)/(ende-anfang))]++;
    } 
  }
  fclose(dptr2);
  if(verbose) printf("] %d %d\n",under,over);
 
 
  dptr=fopen(ofilename,"w");j=0;
  if(!dptr) {
    printf("%s: ERROR, %s could not be created.\n",progname,ofilename);
    return(EX_CANTCREAT);
  }
  if(verbose) printf("> %s[",ofilename);
  
  fprintf(dptr,"# binning V." VERSION " (c) Markus Hoffmann 1997-2010\n");
  fprintf(dptr,"# FILE=<%s>  COL=%d\n",ifilename,col);
  fprintf(dptr,"# [%g:%g] %d\n",anfang,ende,n);
  fprintf(dptr,"# ANZDATA=%d\n",anzdata);
  fprintf(dptr,"# OVER=   %d\n",over);
  fprintf(dptr,"# UNDER=  %d\n",under);
  
  for(k=0;k<n;k++) {
    fprintf(dptr,"%g %d \n",(double)k/n*(ende-anfang)+anfang,b[k]);
    printf("."); j++;
  }
  fclose(dptr);
  if(verbose) printf("] %d \n",j);
  free(b);
  return(EX_OK);
}
int wort_sep (char *t,char c,int klamb ,char *w1, char *w2)    {
  int f=0, klam=0;

  /* klamb=0 : keine Klammern werden beruecksichtigt
     klamb=1 : normale Klammern werden beruecksichtigt
     klamb=2 : eckige Klammern werden beruecksichtigt
     klamb=4 : geschweifte Klammern werden beruecksichtigt
  */

  if(!(*t)) return(*w1=*w2=0);

  while(*t && (*t!=c || f || klam>0)) {
    if(*t=='"') f=!f;
    else if(!f && (((klamb&1) && *t=='(') || ((klamb&2) && *t=='[') || ((klamb&4) && *t=='{'))) klam++;
    else if(!f && (((klamb&1) && *t==')') || ((klamb&2) && *t==']') || ((klamb&4) && *t=='}'))) klam--;
    *w1++=*t++;
  }
  if(!(*t)) {
    *w2=*w1=0;
    return(1);
  } else {
    *w1=0;
    strcpy(w2,++t);
    return(2);
  }
}
/*  (c) Markus Hoffmann
  ' Teile in Anf<EFBFBD>hrungszeichen werden nicht ver<EFBFBD>ndert
  ' Ersetzt Tabulator durch Leerzeichen
  ' mehrfache Leerzeichen zwischen W<EFBFBD>rtern werden entfernt
  ' wenn fl&=1 ,  gib Gro<EFBFBD>buchstaben zur<EFBFBD>ck
 */
 
void xtrim(char *t,int f, char *w ) {
  register int a=0,u=0,b=0;
  while(*t) {
    while(*t && (!isspace(*t) || a)) {
      if(*t=='"') a=!a;
      u=1; if(b==1) {*w++=' '; b=0;}
      if(f && !a) *w++=toupper(*t++); else *w++=*t++;
    }
    if(u && !b) b=1;
    if(*t) t++;
  } *w=0;
}