static Join_t*
init(void)
{
register Join_t* jp;
register int i;
setlocale(LC_ALL, "");
if (jp = newof(0, Join_t, 1, 0))
{
if (jp->mb = mbwide())
for (i = 0x80; i <= 0xff; i++)
jp->state[i] = S_WIDE;
jp->state[' '] = jp->state['\t'] = S_SPACE;
jp->state['\n'] = S_NL;
jp->delim = -1;
jp->nullfield = 0;
if (!(jp->file[0].fields = newof(0, Field_t, NFIELD + 1, 0)) ||
!(jp->file[1].fields = newof(0, Field_t, NFIELD + 1, 0)))
{
done(jp);
return 0;
}
jp->file[0].maxfields = NFIELD;
jp->file[1].maxfields = NFIELD;
jp->outmode = C_COMMON;
}
return jp;
}
void sh_trim(register char *sp)
/*@
assume sp!=NULL;
promise strlen(in sp) <= in strlen(sp);
@*/
{
register char *dp;
register int c;
if(sp)
{
dp = sp;
while(c= *sp)
{
#if SHOPT_MULTIBYTE
int len;
if(mbwide() && (len=mbsize(sp))>1)
{
memmove(dp, sp, len);
dp += len;
sp += len;
continue;
}
#endif /* SHOPT_MULTIBYTE */
sp++;
if(c == '\\')
c = *sp++;
if(c)
*dp++ = c;
}
*dp = 0;
}
}
static void
print(register Sfio_t* sp, register char* name, char* delim)
{
if (mbwide())
sfputr(sp, name, -1);
else
{
#if CC_NATIVE != CC_ASCII
register int c;
register unsigned char* n2a;
register unsigned char* a2n;
register int aa;
register int as;
n2a = ccmap(CC_NATIVE, CC_ASCII);
a2n = ccmap(CC_ASCII, CC_NATIVE);
aa = n2a['A'];
as = n2a[' '];
while (c = *name++)
{
c = n2a[c];
if (c & 0200)
{
c &= 0177;
sfputc(sp, '?');
}
if (c < as)
{
c += aa - 1;
sfputc(sp, '^');
}
c = a2n[c];
sfputc(sp, c);
}
#else
register int c;
while (c = *name++)
{
if (c & 0200)
{
c &= 0177;
sfputc(sp, '?');
}
if (c < ' ')
{
c += 'A' - 1;
sfputc(sp, '^');
}
sfputc(sp, c);
}
#endif
}
if (delim)
sfputr(sp, delim, -1);
}
Wc_t* wc_init(int mode)
{
register int n;
register int w;
Wc_t* wp;
if (!(wp = (Wc_t*)stakalloc(sizeof(Wc_t))))
return 0;
if (!mbwide())
wp->mb = 0;
#if _hdr_wchar && _hdr_wctype && _lib_iswctype
else if (!(mode & WC_NOUTF8) && (lcinfo(LC_CTYPE)->lc->flags & LC_utf8))
wp->mb = 1;
#endif
else
wp->mb = -1;
w = mode & WC_WORDS;
for (n = (1<<CHAR_BIT); --n >= 0;)
wp->type[n] = (w && isspace(n)) ? WC_SP : 0;
wp->type['\n'] = WC_SP|WC_NL;
if ((mode & (WC_MBYTE|WC_WORDS)) && wp->mb > 0)
{
for (n = 0; n < 64; n++)
{
wp->type[0x80+n] |= WC_MB;
if (n<32)
wp->type[0xc0+n] |= WC_MB+1;
else if (n<48)
wp->type[0xc0+n] |= WC_MB+2;
else if (n<56)
wp->type[0xc0+n] |= WC_MB+3;
else if (n<60)
wp->type[0xc0+n] |= WC_MB+4;
else if (n<62)
wp->type[0xc0+n] |= WC_MB+5;
}
wp->type[0xc0] = WC_MB|WC_ERR;
wp->type[0xc1] = WC_MB|WC_ERR;
wp->type[0xfe] = WC_MB|WC_ERR;
wp->type[0xff] = WC_MB|WC_ERR;
}
wp->mode = mode;
return wp;
}
int
b_cut(int argc, char** argv, void* context)
{
register char* cp = 0;
register Sfio_t* fp;
char* s;
int n;
Cut_t* cut;
int mode = 0;
Delim_t wdelim;
Delim_t ldelim;
size_t reclen = 0;
cmdinit(argc, argv, context, ERROR_CATALOG, 0);
wdelim.chr = '\t';
ldelim.chr = '\n';
wdelim.len = ldelim.len = 1;
for (;;)
{
switch (n = optget(argv, usage))
{
case 0:
break;
case 'b':
case 'c':
if(mode&C_FIELDS)
{
error(2, "f option already specified");
continue;
}
cp = opt_info.arg;
if(n=='b')
mode |= C_BYTES;
else
mode |= C_CHARS;
continue;
case 'D':
ldelim.str = opt_info.arg;
if (mbwide())
{
s = opt_info.arg;
ldelim.chr = mbchar(s);
if ((n = s - opt_info.arg) > 1)
{
ldelim.len = n;
continue;
}
}
ldelim.chr = *(unsigned char*)opt_info.arg;
ldelim.len = 1;
continue;
case 'd':
wdelim.str = opt_info.arg;
if (mbwide())
{
s = opt_info.arg;
wdelim.chr = mbchar(s);
if ((n = s - opt_info.arg) > 1)
{
wdelim.len = n;
continue;
}
}
wdelim.chr = *(unsigned char*)opt_info.arg;
wdelim.len = 1;
continue;
case 'f':
if(mode&(C_CHARS|C_BYTES))
{
error(2, "c option already specified");
continue;
}
cp = opt_info.arg;
mode |= C_FIELDS;
continue;
case 'n':
mode |= C_NOSPLIT;
continue;
case 'N':
mode |= C_NONEWLINE;
continue;
case 'R':
case 'r':
if(opt_info.num>0)
reclen = opt_info.num;
continue;
case 's':
mode |= C_SUPRESS;
continue;
case ':':
error(2, "%s", opt_info.arg);
break;
case '?':
error(ERROR_usage(2), "%s", opt_info.arg);
break;
}
break;
}
argv += opt_info.index;
if (error_info.errors)
error(ERROR_usage(2), "%s",optusage(NiL));
if(!cp)
{
error(2, "b, c or f option must be specified");
error(ERROR_usage(2), "%s", optusage(NiL));
}
if(!*cp)
error(3, "non-empty b, c or f option must be specified");
if((mode & (C_FIELDS|C_SUPRESS)) == C_SUPRESS)
error(3, "s option requires f option");
cut = cutinit(mode, cp, &wdelim, &ldelim, reclen);
if(cp = *argv)
argv++;
do
{
if(!cp || streq(cp,"-"))
fp = sfstdin;
else if(!(fp = sfopen(NiL,cp,"r")))
{
error(ERROR_system(0),"%s: cannot open",cp);
continue;
}
if(mode&C_FIELDS)
cutfields(cut,fp,sfstdout);
else
cutcols(cut,fp,sfstdout);
if(fp!=sfstdin)
sfclose(fp);
} while(cp = *argv++);
if (sfsync(sfstdout))
error(ERROR_system(0), "write error");
return error_info.errors != 0;
}
static Cut_t*
cutinit(int mode, char* str, Delim_t* wdelim, Delim_t* ldelim, size_t reclen)
{
register int* lp;
register int c;
register int n = 0;
register int range = 0;
register char* cp = str;
Cut_t* cut;
if (!(cut = (Cut_t*)stakalloc(sizeof(Cut_t) + strlen(cp) * sizeof(int))))
error(ERROR_exit(1), "out of space");
if (cut->mb = mbwide())
{
memset(cut->space, 0, sizeof(cut->space) / 2);
memset(cut->space + sizeof(cut->space) / 2, SP_WIDE, sizeof(cut->space) / 2);
}
else
memset(cut->space, 0, sizeof(cut->space));
cut->wdelim = *wdelim;
if (wdelim->len == 1)
cut->space[wdelim->chr] = SP_WORD;
cut->ldelim = *ldelim;
cut->eob = (ldelim->len == 1) ? ldelim->chr : 0;
cut->space[cut->eob] = SP_LINE;
cut->cflag = (mode&C_CHARS) && cut->mb;
cut->nosplit = (mode&(C_BYTES|C_NOSPLIT)) == (C_BYTES|C_NOSPLIT) && cut->mb;
cut->sflag = (mode&C_SUPRESS) != 0;
cut->nlflag = (mode&C_NONEWLINE) != 0;
cut->reclen = reclen;
lp = cut->list;
for (;;)
switch(c = *cp++)
{
case ' ':
case '\t':
while(*cp==' ' || *cp=='\t')
cp++;
/*FALLTHROUGH*/
case 0:
case ',':
if(range)
{
--range;
if((n = (n ? (n-range) : (HUGE-1))) < 0)
error(ERROR_exit(1),"invalid range for c/f option");
*lp++ = range;
*lp++ = n;
}
else
{
*lp++ = --n;
*lp++ = 1;
}
if(c==0)
{
register int *dp;
*lp = HUGE;
n = 1 + (lp-cut->list)/2;
qsort(lp=cut->list,n,2*sizeof(*lp),mycomp);
/* eliminate overlapping regions */
for(n=0,range= -2,dp=lp; *lp!=HUGE; lp+=2)
{
if(lp[0] <= range)
{
if(lp[1]==HUGE)
{
dp[-1] = HUGE;
break;
}
if((c = lp[0]+lp[1]-range)>0)
{
range += c;
dp[-1] += c;
}
}
else
{
range = *dp++ = lp[0];
if(lp[1]==HUGE)
{
*dp++ = HUGE;
break;
}
range += (*dp++ = lp[1]);
}
}
*dp = HUGE;
lp = cut->list;
/* convert ranges into gaps */
for(n=0; *lp!=HUGE; lp+=2)
{
c = *lp;
*lp -= n;
n = c+lp[1];
}
return cut;
}
n = range = 0;
break;
case '-':
if(range)
error(ERROR_exit(1),"bad list for c/f option");
range = n?n:1;
n = 0;
break;
default:
if(!isdigit(c))
error(ERROR_exit(1),"bad list for c/f option");
n = 10*n + (c-'0');
break;
}
/* NOTREACHED */
}
int
b_paste(int argc, char** argv, Shbltin_t* context)
{
register int n, sflag=0;
register Sfio_t *fp, **streams;
register char *cp, *delim;
char *ep;
Delim_t *mp;
int dlen, dsiz;
char defdelim[2];
cmdinit(argc, argv, context, ERROR_CATALOG, 0);
delim = 0;
for (;;)
{
switch (optget(argv, usage))
{
case 'd':
delim = opt_info.arg;
continue;
case 's':
sflag++;
continue;
case ':':
error(2, "%s", opt_info.arg);
break;
case '?':
error(ERROR_usage(2), "%s", opt_info.arg);
break;
}
break;
}
argv += opt_info.index;
if(error_info.errors)
error(ERROR_usage(2),"%s", optusage(NiL));
if(!delim || !*delim)
{
delim = defdelim;
delim[0] = '\t';
delim[1] = 0;
}
if (!(delim = strdup(delim)))
error(ERROR_system(1), "out of space");
dlen = dsiz = stresc(delim);
mp = 0;
if (mbwide())
{
cp = delim;
ep = delim + dlen;
dlen = 0;
while (cp < ep)
{
mbchar(cp);
dlen++;
}
if(dlen < dsiz)
{
if (!(mp = newof(0, Delim_t, dlen, 0)))
{
free(delim);
error(ERROR_system(1), "out of space");
}
cp = delim;
dlen = 0;
while (cp < ep)
{
mp[dlen].chr = cp;
mbchar(cp);
mp[dlen].len = cp - mp[dlen].chr;
dlen++;
}
}
}
if(cp = *argv)
{
n = argc - opt_info.index;
argv++;
}
else
n = 1;
if(!sflag)
{
if (!(streams = (Sfio_t**)stakalloc(n*sizeof(Sfio_t*))))
error(ERROR_exit(1), "out of space");
n = 0;
}
do
{
if(!cp || streq(cp,"-"))
fp = sfstdin;
else if(!(fp = sfopen(NiL,cp,"r")))
error(ERROR_system(0),"%s: cannot open",cp);
if(fp && sflag)
{
if(spaste(fp,sfstdout,delim,dsiz,dlen,mp) < 0)
error(ERROR_system(0),"write failed");
if(fp!=sfstdin)
sfclose(fp);
}
else if(!sflag)
streams[n++] = fp;
} while(cp= *argv++);
if(!sflag)
{
if(error_info.errors==0 && paste(n,streams,sfstdout,delim,dsiz,dlen,mp) < 0)
error(ERROR_system(0),"write failed");
while(--n>=0)
if((fp=streams[n]) && fp!=sfstdin)
sfclose(fp);
}
if (mp)
free(mp);
free(delim);
return(error_info.errors);
}
int chrexp(const char *s, char **p, int *m, int flags) {
const char *t;
int c;
const char *e;
const char *b;
char *r;
int n;
int x;
wchar_t d;
Mbstate_t q;
bool u;
bool w;
u = w = 0;
mbinit(&q);
for (;;) {
b = s;
c = mbchar(&d, (char **)&s, MB_LEN_MAX, &q);
switch (c) {
case 0:
s = b;
break;
case '\\':
b = s;
switch (c = *s++) {
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c -= '0';
t = s + 2;
while (s < t) {
switch (*s) {
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
c = (c << 3) + *s++ - '0';
break;
default:
t = s;
break;
}
}
break;
case 'a':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c = CC_bel;
break;
case 'b':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c = '\b';
break;
case 'c': /*DEPRECATED*/
case 'C':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c = *s;
if (c) {
s++;
if (c == '\\') {
c = chrexp(s - 1, &r, 0, flags);
s = (const char *)r;
}
if (islower(c)) c = toupper(c);
c ^= 0x40;
}
break;
case 'e': /*DEPRECATED*/
case 'E':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c = CC_esc;
break;
case 'f':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c = '\f';
break;
case 'M':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
if (*s == '-') {
s++;
c = CC_esc;
}
break;
case 'n':
if (flags & FMT_EXP_NONL) continue;
if (!(flags & FMT_EXP_LINE)) goto noexpand;
c = '\n';
break;
case 'r':
if (flags & FMT_EXP_NOCR) continue;
if (!(flags & FMT_EXP_LINE)) goto noexpand;
c = '\r';
break;
case 't':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c = '\t';
break;
case 'v':
if (!(flags & FMT_EXP_CHAR)) goto noexpand;
c = CC_vt;
break;
case 'u':
u = 1;
// FALLTHRU
case 'w':
t = s + 4;
goto wex;
case 'U':
u = 1;
// FALLTHRU
case 'W':
t = s + 8;
wex:
if (!(flags & FMT_EXP_WIDE)) goto noexpand;
w = 1;
goto hex;
case 'x':
t = s + 2;
hex:
e = s;
n = 0;
c = 0;
x = 0;
while (!e || !t || s < t) {
switch (*s) {
case 'a':
case 'b':
case 'c':
case 'd':
case 'e':
case 'f':
c = (c << 4) + *s++ - 'a' + 10;
n++;
continue;
case 'A':
case 'B':
case 'C':
case 'D':
case 'E':
case 'F':
c = (c << 4) + *s++ - 'A' + 10;
n++;
continue;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
c = (c << 4) + *s++ - '0';
n++;
continue;
case '{':
case '[':
if (s != e) break;
e = 0;
s++;
if (w && (*s == 'U' || *s == 'W') && *(s + 1) == '+') s += 2;
continue;
case '-':
if (e) break;
if (*(s + 1) != '}' && *(s + 1) != ']') {
if (!*(s + 1) || (*(s + 2) != '}' && *(s + 2) != ']')) {
break;
}
x = *(unsigned char *)(s + 1);
s += 2;
} else {
x = -1;
s++;
}
/*FALLTHROUGH*/
case '}':
case ']':
if (!e) e = ++s;
break;
default:
break;
}
break;
}
if (e) {
if (n < 8 || (n == 8 && c >= 0)) {
if (!w) {
if (n > 2) {
if (!(flags & FMT_EXP_WIDE)) goto noexpand;
w = 1;
} else if (!(flags & FMT_EXP_CHAR)) {
goto noexpand;
} else {
break;
}
}
if (!mbwide()) w = 0;
if (c <= 0x7f) break;
if (u) {
uint32_t i = c;
wchar_t o;
if (!utf32invalid(i) && utf32stowcs(&o, &i, 1) > 0) {
c = o;
break;
}
} else if (w || c <= ast.byte_max) {
break;
}
}
if (x) {
c = x;
w = 0;
break;
}
}
/*FALLTHROUGH*/
case 0:
goto noexpand;
}
break;
default:
if ((s - b) > 1) w = 1;
break;
noexpand:
s = b;
w = 0;
c = '\\';
break;
}
break;
}
if (m) *m = w;
if (p) *p = (char *)s;
return c;
}
