/*XXX
- if no magic,
if dest given, copy to dst
return ?
- if magic && (no globbing || syntax error)
debunk to dst
return ?
- return ?
*/
int
has_globbing(const char *xp, const char *xpe)
{
const unsigned char *p = (const unsigned char *) xp;
const unsigned char *pe = (const unsigned char *) xpe;
int c;
int nest = 0, bnest = 0;
int saw_glob = 0;
int in_bracket = 0; /* inside [...] */
for (; p < pe; p++) {
if (!ISMAGIC(*p))
continue;
if ((c = *++p) == '*' || c == '?')
saw_glob = 1;
else if (c == '[') {
if (!in_bracket) {
saw_glob = 1;
in_bracket = 1;
if (ISMAGIC(p[1]) && p[2] == NOT)
p += 2;
if (ISMAGIC(p[1]) && p[2] == ']')
p += 2;
}
/* XXX Do we need to check ranges here? POSIX Q */
} else if (c == ']') {
if (in_bracket) {
if (bnest) /* [a*(b]) */
return 0;
in_bracket = 0;
}
} else if ((c & 0x80) && strchr("*+?@! ", c & 0x7f)) {
saw_glob = 1;
if (in_bracket)
bnest++;
else
nest++;
} else if (c == '|') {
if (in_bracket && !bnest) /* *(a[foo|bar]) */
return 0;
} else if (c == /*(*/ ')') {
if (in_bracket) {
if (!bnest--) /* *(a[b)c] */
return 0;
} else if (nest)
nest--;
}
/* else must be a MAGIC-MAGIC, or MAGIC-!, MAGIC--, MAGIC-]
MAGIC-{, MAGIC-,, MAGIC-} */
}
return saw_glob && !in_bracket && !nest;
}
static const unsigned char *
cclass(const unsigned char *p, int sub)
{
int c, d, rv, not, found = 0;
const unsigned char *orig_p = p;
if ((not = (ISMAGIC(*p) && *++p == NOT)))
p++;
do {
/* check for POSIX character class (e.g. [[:alpha:]]) */
if ((p[0] == MAGIC && p[1] == '[' && p[2] == ':') ||
(p[0] == '[' && p[1] == ':')) {
do {
const char *pp = p + (*p == MAGIC) + 2;
rv = posix_cclass(pp, sub, &p);
switch (rv) {
case 1:
found = 1;
break;
case -2:
return NULL;
}
} while (rv != -1 && p[0] == MAGIC && p[1] == '[' && p[2] == ':');
if (p[0] == MAGIC && p[1] == ']')
break;
}
c = *p++;
if (ISMAGIC(c)) {
c = *p++;
if ((c & 0x80) && !ISMAGIC(c)) {
c &= 0x7f;/* extended pattern matching: *+?@! */
/* XXX the ( char isn't handled as part of [] */
if (c == ' ') /* simile for @: plain (..) */
c = '(' /*)*/;
}
}
if (c == '\0')
/* No closing ] - act as if the opening [ was quoted */
return sub == '[' ? orig_p : NULL;
if (ISMAGIC(p[0]) && p[1] == '-' &&
(!ISMAGIC(p[2]) || p[3] != ']')) {
p += 2; /* MAGIC- */
d = *p++;
if (ISMAGIC(d)) {
d = *p++;
if ((d & 0x80) && !ISMAGIC(d))
d &= 0x7f;
}
/* POSIX says this is an invalid expression */
if (c > d)
return NULL;
} else
d = c;
if (c == sub || (c <= sub && sub <= d))
found = 1;
} while (!(ISMAGIC(p[0]) && p[1] == ']'));
return (found != not) ? p+2 : NULL;
}
static const unsigned char *
gmatch_cclass(const unsigned char *p, unsigned char sub)
{
unsigned char c, d;
bool notp, found = false;
const unsigned char *orig_p = p;
if ((notp = tobool(ISMAGIC(*p) && *++p == '!')))
p++;
do {
c = *p++;
if (ISMAGIC(c)) {
c = *p++;
if ((c & 0x80) && !ISMAGIC(c)) {
/* extended pattern matching: *+?@! */
c &= 0x7F;
/* XXX the ( char isn't handled as part of [] */
if (c == ' ')
/* simile for @: plain (..) */
c = '(' /*)*/;
}
}
if (c == '\0')
/* No closing ] - act as if the opening [ was quoted */
return (sub == '[' ? orig_p : NULL);
if (ISMAGIC(p[0]) && p[1] == '-' &&
(!ISMAGIC(p[2]) || p[3] != ']')) {
/* MAGIC- */
p += 2;
d = *p++;
if (ISMAGIC(d)) {
d = *p++;
if ((d & 0x80) && !ISMAGIC(d))
d &= 0x7f;
}
/* POSIX says this is an invalid expression */
if (c > d)
return (NULL);
} else
d = c;
if (c == sub || (c <= sub && sub <= d))
found = true;
} while (!(ISMAGIC(p[0]) && p[1] == ']'));
return ((found != notp) ? p+2 : NULL);
}
/* Look for next ) or | (if match_sep) in *(foo|bar) pattern */
const unsigned char *
pat_scan(const unsigned char *p, const unsigned char *pe, int match_sep)
{
int nest = 0;
for (; p < pe; p++) {
if (!ISMAGIC(*p))
continue;
if ((*++p == /*(*/ ')' && nest-- == 0) ||
(*p == '|' && match_sep && nest == 0))
return ++p;
if ((*p & 0x80) && strchr("*+?@! ", *p & 0x7f))
nest++;
}
return (const unsigned char *) 0;
}
static const unsigned char *
cclass(const unsigned char *p, int sub)
{
int c, d, not, found = 0;
const unsigned char *orig_p = p;
if ((not = (ISMAGIC(*p) && *++p == NOT)))
p++;
do {
c = *p++;
if (ISMAGIC(c)) {
c = *p++;
if ((c & 0x80) && !ISMAGIC(c)) {
c &= 0x7f;/* extended pattern matching: *+?@! */
/* XXX the ( char isn't handled as part of [] */
if (c == ' ') /* simile for @: plain (..) */
c = '(' /*)*/;
}
}
if (c == '\0')
/* No closing ] - act as if the opening [ was quoted */
return sub == '[' ? orig_p : NULL;
if (ISMAGIC(p[0]) && p[1] == '-' &&
(!ISMAGIC(p[2]) || p[3] != ']'))
{
p += 2; /* MAGIC- */
d = *p++;
if (ISMAGIC(d)) {
d = *p++;
if ((d & 0x80) && !ISMAGIC(d))
d &= 0x7f;
}
/* POSIX says this is an invalid expression */
if (c > d)
return NULL;
} else
d = c;
if (c == sub || (c <= sub && sub <= d))
found = 1;
} while (!(ISMAGIC(p[0]) && p[1] == ']'));
return (found != not) ? p+2 : NULL;
}
/* variant of fputs for ptreef and wdstrip */
static const char *
wdvarput(struct shf *shf, const char *wp, int quotelevel, int opmode)
{
int c;
const char *cs;
/*-
* problems:
* `...` -> $(...)
* 'foo' -> "foo"
* x${foo:-"hi"} -> x${foo:-hi} unless WDS_TPUTS
* x${foo:-'hi'} -> x${foo:-hi} unless WDS_KEEPQ
* could change encoding to:
* OQUOTE ["'] ... CQUOTE ["']
* COMSUB [(`] ...\0 (handle $ ` \ and maybe " in `...` case)
*/
while (/* CONSTCOND */ 1)
switch (*wp++) {
case EOS:
return (--wp);
case ADELIM:
case CHAR:
c = *wp++;
if ((opmode & WDS_MAGIC) &&
(ISMAGIC(c) || c == '[' || c == '!' ||
c == '-' || c == ']' || c == '*' || c == '?'))
shf_putc(MAGIC, shf);
shf_putc(c, shf);
break;
case QCHAR: {
bool doq;
c = *wp++;
doq = (c == '"' || c == '`' || c == '$' || c == '\\');
if (opmode & WDS_TPUTS) {
if (quotelevel == 0)
doq = true;
} else {
if (!(opmode & WDS_KEEPQ))
doq = false;
}
if (doq)
shf_putc('\\', shf);
shf_putc(c, shf);
break;
}
case COMSUB:
shf_puts("$(", shf);
cs = ")";
pSUB:
while ((c = *wp++) != 0)
shf_putc(c, shf);
shf_puts(cs, shf);
break;
case FUNSUB:
c = ' ';
if (0)
/* FALLTHROUGH */
case VALSUB:
c = '|';
shf_putc('$', shf);
shf_putc('{', shf);
shf_putc(c, shf);
cs = ";}";
goto pSUB;
case EXPRSUB:
shf_puts("$((", shf);
cs = "))";
goto pSUB;
case OQUOTE:
if (opmode & WDS_TPUTS) {
quotelevel++;
shf_putc('"', shf);
}
break;
case CQUOTE:
if (opmode & WDS_TPUTS) {
if (quotelevel)
quotelevel--;
shf_putc('"', shf);
}
break;
case OSUBST:
shf_putc('$', shf);
if (*wp++ == '{')
shf_putc('{', shf);
while ((c = *wp++) != 0)
shf_putc(c, shf);
wp = wdvarput(shf, wp, 0, opmode);
break;
case CSUBST:
if (*wp++ == '}')
shf_putc('}', shf);
return (wp);
case OPAT:
if (opmode & WDS_MAGIC) {
shf_putc(MAGIC, shf);
shf_putchar(*wp++ | 0x80, shf);
} else {
shf_putchar(*wp++, shf);
shf_putc('(', shf);
}
break;
case SPAT:
c = '|';
if (0)
case CPAT:
c = /*(*/ ')';
if (opmode & WDS_MAGIC)
shf_putc(MAGIC, shf);
shf_putc(c, shf);
break;
}
}
/* Function must return either 0 or 1 (assumed by code for 0x80|'!') */
static int
do_gmatch(const unsigned char *s, const unsigned char *se,
const unsigned char *p, const unsigned char *pe)
{
int sc, pc;
const unsigned char *prest, *psub, *pnext;
const unsigned char *srest;
if (s == NULL || p == NULL)
return 0;
while (p < pe) {
pc = *p++;
sc = s < se ? *s : '\0';
s++;
if (!ISMAGIC(pc)) {
if (sc != pc)
return 0;
continue;
}
switch (*p++) {
case '[':
if (sc == 0 || (p = cclass(p, sc)) == NULL)
return 0;
break;
case '?':
if (sc == 0)
return 0;
break;
case '*':
if (p == pe)
return 1;
s--;
do {
if (do_gmatch(s, se, p, pe))
return 1;
} while (s++ < se);
return 0;
/*
* [*+?@!](pattern|pattern|..)
*
* Not ifdef'd KSH as this is needed for ${..%..}, etc.
*/
case 0x80|'+': /* matches one or more times */
case 0x80|'*': /* matches zero or more times */
if (!(prest = pat_scan(p, pe, 0)))
return 0;
s--;
/* take care of zero matches */
if (p[-1] == (0x80 | '*') &&
do_gmatch(s, se, prest, pe))
return 1;
for (psub = p; ; psub = pnext) {
pnext = pat_scan(psub, pe, 1);
for (srest = s; srest <= se; srest++) {
if (do_gmatch(s, srest, psub, pnext - 2) &&
(do_gmatch(srest, se, prest, pe) ||
(s != srest && do_gmatch(srest,
se, p - 2, pe))))
return 1;
}
if (pnext == prest)
break;
}
return 0;
case 0x80|'?': /* matches zero or once */
case 0x80|'@': /* matches one of the patterns */
case 0x80|' ': /* simile for @ */
if (!(prest = pat_scan(p, pe, 0)))
return 0;
s--;
/* Take care of zero matches */
if (p[-1] == (0x80 | '?') &&
do_gmatch(s, se, prest, pe))
return 1;
for (psub = p; ; psub = pnext) {
pnext = pat_scan(psub, pe, 1);
srest = prest == pe ? se : s;
for (; srest <= se; srest++) {
if (do_gmatch(s, srest, psub, pnext - 2) &&
do_gmatch(srest, se, prest, pe))
return 1;
}
if (pnext == prest)
break;
}
return 0;
case 0x80|'!': /* matches none of the patterns */
if (!(prest = pat_scan(p, pe, 0)))
return 0;
s--;
for (srest = s; srest <= se; srest++) {
int matched = 0;
for (psub = p; ; psub = pnext) {
pnext = pat_scan(psub, pe, 1);
if (do_gmatch(s, srest, psub,
pnext - 2))
{
matched = 1;
break;
}
if (pnext == prest)
break;
}
if (!matched &&
do_gmatch(srest, se, prest, pe))
return 1;
}
return 0;
default:
if (sc != p[-1])
return 0;
break;
}
}
return s == se;
}
/**
* pattern simplifications:
* - @(x) -> x (not @(x|y) though)
* - ** -> *
*/
static void *
simplify_gmatch_pattern(const unsigned char *sp)
{
uint8_t c;
unsigned char *cp, *dp;
const unsigned char *ps, *se;
cp = alloc(strlen((const void *)sp) + 1, ATEMP);
goto simplify_gmatch_pat1a;
/* foo@(b@(a)r)b@(a|a)z -> foobarb@(a|a)z */
simplify_gmatch_pat1:
sp = cp;
simplify_gmatch_pat1a:
dp = cp;
se = sp + strlen((const void *)sp);
while ((c = *sp++)) {
if (!ISMAGIC(c)) {
*dp++ = c;
continue;
}
switch ((c = *sp++)) {
case 0x80|'@':
/* simile for @ */
case 0x80|' ':
/* check whether it has only one clause */
ps = pat_scan(sp, se, true);
if (!ps || ps[-1] != /*(*/ ')')
/* nope */
break;
/* copy inner clause until matching close */
ps -= 2;
while ((const unsigned char *)sp < ps)
*dp++ = *sp++;
/* skip MAGIC and closing parenthesis */
sp += 2;
/* copy the rest of the pattern */
memmove(dp, sp, strlen((const void *)sp) + 1);
/* redo from start */
goto simplify_gmatch_pat1;
}
*dp++ = MAGIC;
*dp++ = c;
}
*dp = '\0';
/* collapse adjacent asterisk wildcards */
sp = dp = cp;
while ((c = *sp++)) {
if (!ISMAGIC(c)) {
*dp++ = c;
continue;
}
switch ((c = *sp++)) {
case '*':
while (ISMAGIC(sp[0]) && sp[1] == c)
sp += 2;
break;
}
*dp++ = MAGIC;
*dp++ = c;
}
*dp = '\0';
/* return the result, allocated from ATEMP */
return (cp);
}
static void
glob_path(int flags, const char *pat, XPtrV *wp, const char *path)
{
const char *sp, *p;
char *xp;
int staterr;
int pathlen;
int patlen;
int oldsize, newsize, i, j;
char **words;
XString xs;
patlen = strlen(pat) + 1;
sp = path;
Xinit(xs, xp, patlen + 128, ATEMP);
while (sp) {
xp = Xstring(xs, xp);
if (!(p = strchr(sp, ':')))
p = sp + strlen(sp);
pathlen = p - sp;
if (pathlen) {
/* Copy sp into xp, stuffing any MAGIC characters
* on the way
*/
const char *s = sp;
XcheckN(xs, xp, pathlen * 2);
while (s < p) {
if (ISMAGIC(*s))
*xp++ = MAGIC;
*xp++ = *s++;
}
*xp++ = '/';
pathlen++;
}
sp = p;
XcheckN(xs, xp, patlen);
memcpy(xp, pat, patlen);
oldsize = XPsize(*wp);
glob_str(Xstring(xs, xp), wp, 1); /* mark dirs */
newsize = XPsize(*wp);
/* Check that each match is executable... */
words = (char **) XPptrv(*wp);
for (i = j = oldsize; i < newsize; i++) {
staterr = 0;
if ((search_access(words[i], X_OK, &staterr) >= 0) ||
(staterr == EISDIR)) {
words[j] = words[i];
if (!(flags & XCF_FULLPATH))
memmove(words[j], words[j] + pathlen,
strlen(words[j] + pathlen) + 1);
j++;
} else
afree(words[i], ATEMP);
}
wp->cur = (void **) &words[j];
if (!*sp++)
break;
}
Xfree(xs, xp);
}
