static void push_onecapture (MatchState *ms, int i, const char *s,
const char *e) {
if (i >= ms->level) {
if (i == 0) { /* ms->level == 0, too */
buffer_addstring(ms->buff, s, e - s); /* add whole match */
} else
gsub_error("invalid capture index");
}
else {
ptrdiff_t l = ms->capture[i].len;
if (l == CAP_UNFINISHED) gsub_error("unfinished capture");
if (l == CAP_POSITION) {
// lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
assert(0);
} else
buffer_addstring(ms->buff, ms->capture[i].init, l);
}
}
static const char *start_capture (MatchState *ms, const char *s,
const char *p, int what) {
const char *res;
int level = ms->level;
if (level >= LUA_MAXCAPTURES) gsub_error("too many captures");
ms->capture[level].init = s;
ms->capture[level].len = what;
ms->level = level+1;
if ((res=match(ms, s, p)) == NULL) /* match failed? */
ms->level--; /* undo capture */
return res;
}
static const char *classend (MatchState *ms, const char *p) {
switch (*p++) {
case L_ESC: {
if (*p == '\0')
gsub_error("malformed pattern (ends with " LUA_QL("%%") ")");
return p+1;
}
case '[': {
if (*p == '^') p++;
do { /* look for a `]' */
if (*p == '\0')
gsub_error("malformed pattern (missing " LUA_QL("]") ")");
if (*(p++) == L_ESC && *p != '\0')
p++; /* skip escapes (e.g. `%]') */
} while (*p != ']');
return p+1;
}
default: {
return p;
}
}
}
static const char *matchbalance (MatchState *ms, const char *s,
const char *p) {
if (*p == 0 || *(p+1) == 0)
gsub_error("unbalanced pattern");
if (*s != *p) return NULL;
else {
int b = *p;
int e = *(p+1);
int cont = 1;
while (++s < ms->src_end) {
if (*s == e) {
if (--cont == 0) return s+1;
}
else if (*s == b) cont++;
}
}
return NULL; /* string ends out of balance */
}
static int capture_to_close (MatchState *ms) {
int level = ms->level;
for (level--; level>=0; level--)
if (ms->capture[level].len == CAP_UNFINISHED) return level;
return gsub_error("invalid pattern capture");
}
static int check_capture (MatchState *ms, int l) {
l -= '1';
if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
return gsub_error("invalid capture index");
return l;
}
static const char *match (MatchState *ms, const char *s, const char *p) {
init: /* using goto's to optimize tail recursion */
switch (*p) {
case '(': { /* start capture */
if (*(p+1) == ')') /* position capture? */
return start_capture(ms, s, p+2, CAP_POSITION);
else
return start_capture(ms, s, p+1, CAP_UNFINISHED);
}
case ')': { /* end capture */
return end_capture(ms, s, p+1);
}
case L_ESC: {
switch (*(p+1)) {
case 'b': { /* balanced string? */
s = matchbalance(ms, s, p+2);
if (s == NULL) return NULL;
p+=4; goto init; /* else return match(ms, s, p+4); */
}
case 'f': { /* frontier? */
const char *ep; char previous;
p += 2;
if (*p != '[')
gsub_error("missing " LUA_QL("[") " after "
LUA_QL("%%f") " in pattern");
ep = classend(ms, p); /* points to what is next */
previous = (s == ms->src_init) ? '\0' : *(s-1);
if (matchbracketclass(uchar(previous), p, ep-1) ||
!matchbracketclass(uchar(*s), p, ep-1)) return NULL;
p=ep; goto init; /* else return match(ms, s, ep); */
}
default: {
if (isdigit(uchar(*(p+1)))) { /* capture results (%0-%9)? */
s = match_capture(ms, s, uchar(*(p+1)));
if (s == NULL) return NULL;
p+=2; goto init; /* else return match(ms, s, p+2) */
}
goto dflt; /* case default */
}
}
}
case '\0': { /* end of pattern */
return s; /* match succeeded */
}
case '$': {
if (*(p+1) == '\0') /* is the `$' the last char in pattern? */
return (s == ms->src_end) ? s : NULL; /* check end of string */
else goto dflt;
}
default: dflt: { /* it is a pattern item */
const char *ep = classend(ms, p); /* points to what is next */
int m = s<ms->src_end && singlematch(uchar(*s), p, ep);
switch (*ep) {
case '?': { /* optional */
const char *res;
if (m && ((res=match(ms, s+1, ep+1)) != NULL))
return res;
p=ep+1; goto init; /* else return match(ms, s, ep+1); */
}
case '*': { /* 0 or more repetitions */
return max_expand(ms, s, p, ep);
}
case '+': { /* 1 or more repetitions */
return (m ? max_expand(ms, s+1, p, ep) : NULL);
}
case '-': { /* 0 or more repetitions (minimum) */
return min_expand(ms, s, p, ep);
}
default: {
if (!m) return NULL;
s++; p=ep; goto init; /* else return match(ms, s+1, ep); */
}
}
}
}
}