da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#pragma prototyped

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinstatic const char usage[] =

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "from one or more files, contatenating them on standard output.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "list of positive numbers and ranges. Ranges can be of three "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "forms. The first is two positive integers separated by a hyphen "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "(\alow\a\b-\b\ahigh\a), which represents all fields from \alow\a to "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "\ahigh\a. The second is a positive number preceded by a hyphen "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "(\b-\b\ahigh\a), which represents all fields from field \b1\b to "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "\ahigh\a. The last is a positive number followed by a hyphen "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "(\alow\a\b-\b), which represents all fields from \alow\a to the "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "last field, inclusive. Elements in the \alist\a can be repeated, "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "can overlap, and can appear in any order. The order of the "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "output is that of the input.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "cuts from standard input. The start of the file is defined "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "as the current offset.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "to \adelim\a. The default is the \btab\b character.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "character specified with the \b-d\b optiion.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "records of length \areclen\a when used with the \b-b\b or \b-c\b "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "option.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "when used with the \b-f\b option. By default, lines with no "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "delimiters will be passsed in untouched.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "the \b-f\b option is set to \aldelim\a. The default is the "

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "\bnewline\b character.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "with the \b-b\b or \b-c\b option.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "[+0?All files processed successfully.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "[+>0?One or more files failed to open or could not be read.]"

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#include <cmd.h>

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#include <ctype.h>

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chintypedef struct Last_s

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int seqno;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int seq;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int wdelim;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int ldelim;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin} Last_t;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chintypedef struct Cut_s

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int cflag;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int sflag;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int nlflag;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int wdelim;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int ldelim;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int seqno;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int reclen;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin signed char space[UCHAR_MAX];

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin Last_t last;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int list[2]; /* NOTE: must be last member */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin} Cut_t;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define HUGE (1<<14)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define BLOCK 8*1024

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define C_BYTES 1

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define C_CHARS 2

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define C_FIELDS 4

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define C_SUPRESS 8

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define C_NOCHOP 16

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define C_NONEWLINE 32

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinstatic int mycomp(register const void *a,register const void *b)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(*((int*)a) - *((int*)b));

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin}

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinstatic Cut_t *cutinit(int mode,char *str,int wdelim,int ldelim,size_t reclen)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int *lp, c, n=0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int range = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char *cp = str;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin Cut_t *cuthdr;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(cuthdr = (Cut_t*)stakalloc(sizeof(Cut_t)+strlen(cp)*sizeof(int))))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_exit(1), "out of space");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin memset(cuthdr->space, 0, sizeof(cuthdr->space));

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->last.seqno = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->last.seq = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->last.wdelim = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->last.ldelim = '\n';

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->cflag = ((mode&C_CHARS)!=0 && mbwide());

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->sflag = ((mode&C_SUPRESS)!=0);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->nlflag = ((mode&C_NONEWLINE)!=0);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->wdelim = wdelim;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->ldelim = ldelim;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->reclen = reclen;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->seqno = ++cuthdr->last.seqno;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin lp = cuthdr->list;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(1) switch(c= *cp++)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case ' ':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case '\t':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(*cp==' ' || *cp=='\t')

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cp++;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 0:

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case ',':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(range)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin --range;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((n = (n==0?HUGE:n-range)) < 0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_exit(1),"invalid range for c/f option");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *lp++ = range;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *lp++ = n;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *lp++ = --n;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *lp++ = 1;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(c==0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int *dp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *lp = HUGE;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = 1 + (lp-cuthdr->list)/2;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin qsort(lp=cuthdr->list,n,2*sizeof(*lp),mycomp);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* eliminate overlapping regions */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for(n=0,range= -2,dp=lp; *lp!=HUGE; lp+=2)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(lp[0] <= range)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(lp[1]==HUGE)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin dp[-1] = HUGE;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((c = lp[0]+lp[1]-range)>0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin range += c;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin dp[-1] += c;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin range = *dp++ = lp[0];

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(lp[1]==HUGE)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *dp++ = HUGE;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin range += (*dp++ = lp[1]);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *dp = HUGE;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin lp = cuthdr->list;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* convert ranges into gaps */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for(n=0; *lp!=HUGE; lp+=2)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin c = *lp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *lp -= n;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = c+lp[1];

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(cuthdr);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = range = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case '-':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(range)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_exit(1),"bad list for c/f option");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin range = n?n:1;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin default:

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!isdigit(c))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_exit(1),"bad list for c/f option");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = 10*n + (c-'0');

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* NOTREACHED */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin}

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinstatic int advance(const char *str, register int n, register int inlen)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int size, len=inlen;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register const char *cp=str;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(len>0 && n-->0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin size = mblen(cp, len);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(size<0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin size = 1;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cp += size;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin len -= size;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(n>0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(inlen+1);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(cp-str);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin}

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinstatic int cutcols(Cut_t *cuthdr,Sfio_t *fdin,Sfio_t *fdout)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int c, ncol=0,len;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register const int *lp = cuthdr->list;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char *inp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int skip; /* non-zero for don't copy */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(1)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(len = cuthdr->reclen)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin inp = sfreserve(fdin, len, -1);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin inp = sfgetr(fdin, '\n', 0);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!inp && !(inp = sfgetr(fdin, 0, SF_LASTR)))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin len = sfvalue(fdin);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((ncol = skip = *(lp = cuthdr->list)) == 0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin ncol = *++lp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(1)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((c=(cuthdr->cflag?advance(inp,ncol,len):ncol)) > len)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin c = len;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if(c==len && !skip)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin ncol++;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin ncol -= c;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!skip && sfwrite(fdout,(char*)inp,c)<0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(-1);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin inp += c;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(ncol)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin len -= c;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin ncol = *++lp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin skip = !skip;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!cuthdr->nlflag && (skip || cuthdr->reclen))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfputc(fdout,cuthdr->ldelim);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(c);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin}

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin#define endline(c) (((signed char)-1)<0?(c)<0:(c)==((char)-1))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinstatic int cutfields(Cut_t *cuthdr,Sfio_t *fdin,Sfio_t *fdout)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register unsigned char *cp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int c, nfields;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register const int *lp = cuthdr->list;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register unsigned char *copy;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int nodelim, empty, inword=0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register unsigned char *endbuff;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin unsigned char *inbuff, *first;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int lastchar;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin Sfio_t *fdtmp = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin long offset = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(cuthdr->seqno != cuthdr->last.seq)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->space[cuthdr->last.ldelim] = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->space[cuthdr->last.wdelim] = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->space[cuthdr->last.wdelim=cuthdr->wdelim] = 1;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->space[cuthdr->last.ldelim=cuthdr->ldelim] = -1;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr->last.seq = cuthdr->seqno;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* process each buffer */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while ((inbuff = (unsigned char*)sfreserve(fdin, SF_UNBOUND, 0)) && (c = sfvalue(fdin)) > 0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cp = inbuff;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin endbuff = cp + --c;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((lastchar = cp[c]) != cuthdr->ldelim)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *endbuff = cuthdr->ldelim;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* process each line in the buffer */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(cp <= endbuff)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin first = cp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!inword)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin nodelim = empty = 1;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin copy = cp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(nfields = *(lp = cuthdr->list))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin copy = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin nfields = *++lp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if(copy)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin copy = cp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin inword = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(!inword)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* skip over non-delimiter characters */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(!(c=cuthdr->space[*cp++]));

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* check for end-of-line */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(endline(c))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(cp<=endbuff)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((c=cuthdr->space[lastchar]),endline(c))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* restore cuthdr->last. character */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(lastchar != cuthdr->ldelim)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *endbuff = lastchar;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin inword++;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!c)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin nodelim = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(--nfields >0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin continue;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin nfields = *++lp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(copy)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin empty = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((c=(cp-1)-copy)>0 && sfwrite(fdout,(char*)copy,c)< 0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin goto failed;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin copy = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* set to delimiter unless the first field */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin copy = cp -!empty;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!inword)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!copy)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(nodelim)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!cuthdr->sflag)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(offset)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfseek(fdtmp,(Sfoff_t)0,SEEK_SET);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfmove(fdtmp,fdout,offset,-1);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin copy = first;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfputc(fdout,'\n');

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(offset)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfseek(fdtmp,offset=0,SEEK_SET);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(copy && (c=cp-copy)>0 && (!nodelim || !cuthdr->sflag) && sfwrite(fdout,(char*)copy,c)< 0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin goto failed;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* see whether to save in tmp file */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(nodelim && inword && !cuthdr->sflag && (c=cp-first)>0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* copy line to tmpfile in case no fields */

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!fdtmp)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin fdtmp = sftmp(BLOCK);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfwrite(fdtmp,(char*)first,c);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin offset +=c;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinfailed:

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(fdtmp)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfclose(fdtmp);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(0);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin}

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinb_cut(int argc,char *argv[], void* context)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin{

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char *cp = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register Sfio_t *fp;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int n;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin Cut_t *cuthdr;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int mode = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int wdelim = '\t';

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int ldelim = '\n';

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin size_t reclen = 0;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cmdinit(argc, argv, context, ERROR_CATALOG, 0);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (n = optget(argv, usage)) switch (n)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'b':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'c':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(mode&C_FIELDS)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(2, "f option already specified");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cp = opt_info.arg;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(n=='b')

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin mode |= C_BYTES;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin mode |= C_CHARS;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'D':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin ldelim = *(unsigned char*)opt_info.arg;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'd':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin wdelim = *(unsigned char*)opt_info.arg;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'f':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(mode&(C_CHARS|C_BYTES))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(2, "c option already specified");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cp = opt_info.arg;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin mode |= C_FIELDS;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'n':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin mode |= C_NOCHOP;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'N':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin mode |= C_NONEWLINE;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'R':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 'r':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(opt_info.num>0)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin reclen = opt_info.num;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case 's':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin mode |= C_SUPRESS;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case ':':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(2, "%s", opt_info.arg);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case '?':

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_usage(2), "%s", opt_info.arg);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin break;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin argv += opt_info.index;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (error_info.errors)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_usage(2), "%s",optusage(NiL));

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!cp)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(2, "b, c or f option must be specified");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_usage(2), "%s", optusage(NiL));

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!*cp)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(3, "non-empty b, c or f option must be specified");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if((mode & (C_FIELDS|C_SUPRESS)) == C_SUPRESS)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(3, "s option requires f option");

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cuthdr = cutinit(mode,cp,wdelim,ldelim,reclen);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(cp = *argv)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin argv++;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin do

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(!cp || streq(cp,"-"))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin fp = sfstdin;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if(!(fp = sfopen(NiL,cp,"r")))

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin {

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin error(ERROR_system(0),"%s: cannot open",cp);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin continue;

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(mode&C_FIELDS)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cutfields(cuthdr,fp,sfstdout);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin cutcols(cuthdr,fp,sfstdout);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if(fp!=sfstdin)

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfclose(fp);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin }

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while(cp= *argv++);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return(error_info.errors?1:0);

da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin}