static void
cutfields(Cut_t* cut, Sfio_t* fdin, Sfio_t* fdout)
{
register unsigned char *sp = cut->space;
register unsigned char *cp;
register unsigned char *wp;
register int c, nfields;
register const int *lp = cut->list;
register unsigned char *copy;
register int nodelim, empty, inword=0;
register unsigned char *ep;
unsigned char *bp, *first;
int lastchar;
wchar_t w;
Sfio_t *fdtmp = 0;
long offset = 0;
unsigned char mb[8];
/* process each buffer */
while ((bp = (unsigned char*)sfreserve(fdin, SF_UNBOUND, -1)) && (c = sfvalue(fdin)) > 0)
{
cp = bp;
ep = cp + --c;
if((lastchar = cp[c]) != cut->eob)
*ep = cut->eob;
/* process each line in the buffer */
while (cp <= ep)
{
first = cp;
if (!inword)
{
nodelim = empty = 1;
copy = cp;
if (nfields = *(lp = cut->list))
copy = 0;
else
nfields = *++lp;
}
else if (copy)
copy = cp;
inword = 0;
do
{
/* skip over non-delimiter characters */
if (cut->mb)
for (;;)
{
switch (c = sp[*(unsigned char*)cp++])
{
case 0:
continue;
case SP_WIDE:
wp = --cp;
while ((c = mb2wc(w, cp, ep - cp)) <= 0)
{
/* mb char possibly spanning buffer boundary -- fun stuff */
if ((ep - cp) < mbmax())
{
int i;
int j;
int k;
if (lastchar != cut->eob)
{
*ep = lastchar;
if ((c = mb2wc(w, cp, ep - cp)) > 0)
break;
}
if (copy)
{
empty = 0;
if ((c = cp - copy) > 0 && sfwrite(fdout, (char*)copy, c) < 0)
goto failed;
}
for (i = 0; i <= (ep - cp); i++)
mb[i] = cp[i];
if (!(bp = (unsigned char*)sfreserve(fdin, SF_UNBOUND, -1)) || (c = sfvalue(fdin)) <= 0)
goto failed;
cp = bp;
ep = cp + --c;
if ((lastchar = cp[c]) != cut->eob)
*ep = cut->eob;
j = i;
k = 0;
while (j < mbmax())
mb[j++] = cp[k++];
if ((c = mb2wc(w, (char*)mb, j)) <= 0)
{
c = i;
w = 0;
}
first = bp = cp += c - i;
if (copy)
{
copy = bp;
if (w == cut->ldelim.chr)
lastchar = cut->ldelim.chr;
else if (w != cut->wdelim.chr)
{
empty = 0;
if (sfwrite(fdout, (char*)mb, c) < 0)
goto failed;
}
}
c = 0;
}
else
{
w = *cp;
c = 1;
}
break;
}
cp += c;
c = w;
if (c == cut->wdelim.chr)
{
c = SP_WORD;
break;
}
if (c == cut->ldelim.chr)
{
c = SP_LINE;
break;
}
continue;
default:
wp = cp - 1;
break;
}
break;
}
else
{
while (!(c = sp[*cp++]));
wp = cp - 1;
}
/* check for end-of-line */
if (c == SP_LINE)
{
if (cp <= ep)
break;
if (lastchar == cut->ldelim.chr)
break;
/* restore cut->last character */
if (lastchar != cut->eob)
*ep = lastchar;
inword++;
if (!sp[lastchar])
break;
}
nodelim = 0;
if (--nfields > 0)
continue;
nfields = *++lp;
if (copy)
{
empty = 0;
if ((c = wp - copy) > 0 && sfwrite(fdout, (char*)copy, c) < 0)
goto failed;
copy = 0;
}
else
/* set to delimiter unless the first field */
copy = empty ? cp : wp;
} while (!inword);
if (!inword)
{
if (!copy)
{
if (nodelim)
{
if (!cut->sflag)
{
if (offset)
{
sfseek(fdtmp,(Sfoff_t)0,SEEK_SET);
sfmove(fdtmp,fdout,offset,-1);
}
copy = first;
}
}
else
sfputc(fdout,'\n');
}
if (offset)
sfseek(fdtmp,offset=0,SEEK_SET);
}
if (copy && (c=cp-copy)>0 && (!nodelim || !cut->sflag) && sfwrite(fdout,(char*)copy,c)< 0)
goto failed;
}
/* see whether to save in tmp file */
if(inword && nodelim && !cut->sflag && (c=cp-first)>0)
{
/* copy line to tmpfile in case no fields */
if(!fdtmp)
fdtmp = sftmp(BLOCK);
sfwrite(fdtmp,(char*)first,c);
offset +=c;
}
}
failed:
if(fdtmp)
sfclose(fdtmp);
}
/*
* put <string> of length <nbyte> onto lookahead stack
* if <type> is non-zero, the negation of the character is put
* onto the stack so that it can be checked for KEYTRAP
* putstack() returns 1 except when in the middle of a multi-byte char
*/
static int putstack(Edit_t *ep,char string[], register int nbyte, int type)
{
register int c;
#if SHOPT_MULTIBYTE
char *endp, *p=string;
int size, offset = ep->e_lookahead + nbyte;
*(endp = &p[nbyte]) = 0;
endp = &p[nbyte];
do
{
c = (int)((*p) & STRIP);
if(c< 0x80 && c!='<')
{
if (type)
c = -c;
# ifndef CBREAK
if(c == '\0')
{
/*** user break key ***/
ep->e_lookahead = 0;
# if KSHELL
sh_fault(SIGINT);
siglongjmp(ep->e_env, UINTR);
# endif /* KSHELL */
}
# endif /* CBREAK */
}
else
{
again:
if((c=mbchar(p)) >=0)
{
p--; /* incremented below */
if(type)
c = -c;
}
#ifdef EILSEQ
else if(errno == EILSEQ)
errno = 0;
#endif
else if((endp-p) < mbmax())
{
if ((c=ed_read(ep,ep->e_fd,endp, 1,0)) == 1)
{
*++endp = 0;
goto again;
}
return(c);
}
else
{
ed_ringbell();
c = -(int)((*p) & STRIP);
offset += mbmax()-1;
}
}
ep->e_lbuf[--offset] = c;
p++;
}
while (p < endp);
/* shift lookahead buffer if necessary */
if(offset -= ep->e_lookahead)
{
for(size=offset;size < nbyte;size++)
ep->e_lbuf[ep->e_lookahead+size-offset] = ep->e_lbuf[ep->e_lookahead+size];
}
ep->e_lookahead += nbyte-offset;
#else
while (nbyte > 0)
{
c = string[--nbyte] & STRIP;
ep->e_lbuf[ep->e_lookahead++] = (type?-c:c);
# ifndef CBREAK
if( c == '\0' )
{
/*** user break key ***/
ep->e_lookahead = 0;
# if KSHELL
sh_fault(SIGINT);
siglongjmp(ep->e_env, UINTR);
# endif /* KSHELL */
}
# endif /* CBREAK */
}
#endif /* SHOPT_MULTIBYTE */
return(1);
}
