static int read_pattern (lua_State *L, FILE *f, const char *p) {
int inskip = 0; /* {skip} level */
int c = NEED_OTHER;
luaL_Buffer b;
luaL_buffinit(L, &b);
while (*p != '\0') {
switch (*p) {
case '{':
inskip++;
p++;
continue;
case '}':
if (!inskip) lua_error(L, "unbalanced braces in read pattern");
inskip--;
p++;
continue;
default: {
const char *ep = luaI_classend(L, p); /* get what is next */
int m; /* match result */
if (c == NEED_OTHER) c = getc(f);
m = (c==EOF) ? 0 : luaI_singlematch(c, p, ep);
if (m) {
if (!inskip) luaL_putchar(&b, c);
c = NEED_OTHER;
}
switch (*ep) {
case '+': /* repetition (1 or more) */
if (!m) goto break_while; /* pattern fails? */
/* else go through */
case '*': /* repetition (0 or more) */
while (m) { /* reads the same item until it fails */
c = getc(f);
m = (c==EOF) ? 0 : luaI_singlematch(c, p, ep);
if (m && !inskip) luaL_putchar(&b, c);
}
/* go through to continue reading the pattern */
case '?': /* optional */
p = ep+1; /* continues reading the pattern */
continue;
default:
if (!m) goto break_while; /* pattern fails? */
p = ep; /* else continues reading the pattern */
}
}
}
} break_while:
if (c != NEED_OTHER) ungetc(c, f);
luaL_pushresult(&b); /* close buffer */
return (*p == '\0');
}
static const char *matchitem (const char *s, const char *p, struct Capture *cap, const char **ep)
{
if (*p == ESC) {
p++;
if (isdigit((byte)*p)) { /* capture */
int32 l = check_cap(*p, cap);
int32 len = cap->capture[l].len;
*ep = p+1;
if (cap->src_end-s >= len && memcmp(cap->capture[l].init, s, len) == 0)
return s+len;
else return NULL;
}
else if (*p == 'b') { /* balanced string */
p++;
if (*p == 0 || *(p+1) == 0)
lua_error("unbalanced pattern");
*ep = p+2;
return matchbalance(s, *p, *(p+1), cap);
}
else p--; /* and go through */
}
/* "luaI_singlematch" sets "ep" (so must be called even when *s == 0) */
return (luaI_singlematch((byte)*s, p, ep) && s<cap->src_end) ?
s+1 : NULL;
}
static void io_read() {
int32 arg = FIRSTARG;
LuaFile *f = (LuaFile *)getfileparam(FINPUT, &arg);
char *buff;
const char *p = luaL_opt_string(arg, "[^\n]*{\n}");
int inskip = 0; // to control {skips}
int c = NEED_OTHER;
luaL_resetbuffer();
while (*p) {
if (*p == '{') {
inskip++;
p++;
} else if (*p == '}') {
if (inskip == 0)
lua_error("unbalanced braces in read pattern");
inskip--;
p++;
} else {
const char *ep; // get what is next
int m; // match result
if (c == NEED_OTHER) {
char z;
if (f->read(&z, 1) != 1)
c = EOF;
else
c = z;
}
m = luaI_singlematch((c == EOF) ? 0 : (char)c, p, &ep);
if (m) {
if (inskip == 0)
luaL_addchar(c);
c = NEED_OTHER;
}
switch (*ep) {
case '*': // repetition
if (!m)
p = ep + 1; // else stay in (repeat) the same item
break;
case '?': // optional
p = ep + 1; // continues reading the pattern
break;
default:
if (m)
p = ep; // continues reading the pattern
else
goto break_while; // pattern fails
}
}
}
break_while:
if (c >= 0) // not EOF nor NEED_OTHER?
f->seek(-1, SEEK_CUR);
luaL_addchar(0);
buff = luaL_buffer();
if (*buff != 0 || *p == 0) // read something or did not fail?
lua_pushstring(buff);
}
static const char *match (lua_State *L, const char *s, const char *p,
struct Capture *cap) {
init: /* using goto's to optimize tail recursion */
switch (*p) {
case '(': /* start capture */
return start_capture(L, s, p, cap);
case ')': /* end capture */
return end_capture(L, s, p, cap);
case ESC: /* may be %[0-9] or %b */
if (isdigit((unsigned char)(*(p+1)))) { /* capture? */
s = match_capture(L, s, *(p+1), cap);
if (s == NULL) return NULL;
p+=2;
goto init; /* else return match(L, s, p+2, cap) */
}
else if (*(p+1) == 'b') { /* balanced string? */
s = matchbalance(L, s, p+2, cap);
if (s == NULL) return NULL;
p+=4;
goto init; /* else return match(L, s, p+4, cap); */
}
else 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 == cap->src_end) ? s : NULL; /* check end of string */
else goto dflt;
default:
dflt: { /* it is a pattern item */
const char *ep = luaI_classend(L, p); /* points to what is next */
int m = s<cap->src_end && luaI_singlematch((unsigned char)*s, p, ep);
switch (*ep) {
case '?': { /* optional */
const char *res;
if (m && ((res=match(L, s+1, ep+1, cap)) != NULL))
return res;
p=ep+1;
goto init; /* else return match(L, s, ep+1, cap); */
}
case '*': /* 0 or more repetitions */
return max_expand(L, s, p, ep, cap);
case '+': /* 1 or more repetitions */
return (m ? max_expand(L, s+1, p, ep, cap) : NULL);
case '-': /* 0 or more repetitions (minimum) */
return min_expand(L, s, p, ep, cap);
default:
if (!m) return NULL;
s++;
p=ep;
goto init; /* else return match(L, s+1, ep, cap); */
}
}
}
}
static const lua_WChar *min_expand (MatchState *ms, const lua_WChar *s,
const lua_WChar *p, const lua_WChar *ep) {
for (;;) {
const lua_WChar *res = match(ms, s, ep+1);
if (res != NULL)
return res;
else if (s<ms->src_end && luaI_singlematch(*s, p, ep))
s++; /* try with one more repetition */
else return NULL;
}
}
static const char *min_expand (lua_State *L, const char *s, const char *p,
const char *ep, struct Capture *cap) {
for (;;) {
const char *res = match(L, s, ep+1, cap);
if (res != NULL)
return res;
else if (s<cap->src_end && luaI_singlematch((unsigned char)*s, p, ep))
s++; /* try with one more repetition */
else return NULL;
}
}
static const lua_WChar *max_expand (MatchState *ms, const lua_WChar *s,
const lua_WChar *p, const lua_WChar *ep) {
sint32 i = 0; /* counts maximum expand for item */
while ((s+i)<ms->src_end && luaI_singlematch(*(s+i), p, ep))
i++;
/* keeps trying to match with the maximum repetitions */
while (i>=0) {
const lua_WChar *res = match(ms, (s+i), ep+1);
if (res) return res;
i--; /* else didn't match; reduce 1 repetition to try again */
}
return NULL;
}
static const char *max_expand (lua_State *L, const char *s, const char *p,
const char *ep, struct Capture *cap) {
long i = 0; /* counts maximum expand for item */
while ((s+i)<cap->src_end && luaI_singlematch((unsigned char)*(s+i), p, ep))
i++;
/* keeps trying to match with the maximum repetitions */
while (i>=0) {
const char *res = match(L, (s+i), ep+1, cap);
if (res) return res;
i--; /* else didn't match; reduce 1 repetition to try again */
}
return NULL;
}
static const lua_WChar *match (MatchState *ms, const lua_WChar *s, const lua_WChar *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 ESC: /* may be %[0-9] or %b */
if (iswdigit(*(p+1))) { /* capture? */
s = match_capture(ms, s, *(p+1));
if (s == NULL) return NULL;
p+=2; goto init; /* else return match(ms, s, p+2) */
}
else if (*(p+1) == '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); */
}
else 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 lua_WChar *ep = luaI_classend(ms, p); /* points to what is next */
int m = s<ms->src_end && luaI_singlematch(*s, p, ep);
switch (*ep) {
case '?': { /* optional */
const lua_WChar *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); */
}
}
}
}
static void io_read (void)
{
int arg = FIRSTARG;
FILE *f = getfileparam(FINPUT, &arg);
char *buff;
char *p = luaL_opt_string(arg, "[^\n]*{\n}");
int inskip = 0; /* to control {skips} */
int c = NEED_OTHER;
luaL_resetbuffer();
while (*p) {
if (*p == '{') {
inskip++;
p++;
}
else if (*p == '}') {
if (inskip == 0)
lua_error("unbalanced braces in read pattern");
inskip--;
p++;
}
else {
char *ep; /* get what is next */
int m; /* match result */
if (c == NEED_OTHER) c = getc(f);
m = luaI_singlematch((c == EOF) ? 0 : (char)c, p, &ep);
if (m) {
if (inskip == 0) luaL_addchar(c);
c = NEED_OTHER;
}
switch (*ep) {
case '*': /* repetition */
if (!m) p = ep+1; /* else stay in (repeat) the same item */
break;
case '?': /* optional */
p = ep+1; /* continues reading the pattern */
break;
default:
if (m) p = ep; /* continues reading the pattern */
else
goto break_while; /* pattern fails */
}
}
} break_while:
if (c >= 0) /* not EOF nor NEED_OTHER? */
ungetc(c, f);
luaL_addchar(0);
buff = luaL_buffer();
if (*buff != 0 || *p == 0) /* read something or did not fail? */
lua_pushstring(buff);
}
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 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 != '[')
luaL_error(ms->L, "missing `[' after `%%f' in pattern");
ep = luaI_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, *(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 = luaI_classend(ms, p); /* points to what is next */
int m = s<ms->src_end && luaI_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); */
}
}
}
}
}
