rlm@1: /*
rlm@1: ** $Id: lstrlib.c,v 1.132.1.4 2008/07/11 17:27:21 roberto Exp $
rlm@1: ** Standard library for string operations and pattern-matching
rlm@1: ** See Copyright Notice in lua.h
rlm@1: */
rlm@1: 
rlm@1: 
rlm@1: #include <ctype.h>
rlm@1: #include <stddef.h>
rlm@1: #include <stdio.h>
rlm@1: #include <stdlib.h>
rlm@1: #include <string.h>
rlm@1: 
rlm@1: #define lstrlib_c
rlm@1: #define LUA_LIB
rlm@1: 
rlm@1: #include "lua.h"
rlm@1: 
rlm@1: #include "lauxlib.h"
rlm@1: #include "lualib.h"
rlm@1: 
rlm@1: 
rlm@1: /* macro to `unsign' a character */
rlm@1: #define uchar(c)        ((unsigned char)(c))
rlm@1: 
rlm@1: 
rlm@1: 
rlm@1: static int str_len (lua_State *L) {
rlm@1:   size_t l;
rlm@1:   luaL_checklstring(L, 1, &l);
rlm@1:   lua_pushinteger(L, l);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static ptrdiff_t posrelat (ptrdiff_t pos, size_t len) {
rlm@1:   /* relative string position: negative means back from end */
rlm@1:   if (pos < 0) pos += (ptrdiff_t)len + 1;
rlm@1:   return (pos >= 0) ? pos : 0;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_sub (lua_State *L) {
rlm@1:   size_t l;
rlm@1:   const char *s = luaL_checklstring(L, 1, &l);
rlm@1:   ptrdiff_t start = posrelat(luaL_checkinteger(L, 2), l);
rlm@1:   ptrdiff_t end = posrelat(luaL_optinteger(L, 3, -1), l);
rlm@1:   if (start < 1) start = 1;
rlm@1:   if (end > (ptrdiff_t)l) end = (ptrdiff_t)l;
rlm@1:   if (start <= end)
rlm@1:     lua_pushlstring(L, s+start-1, end-start+1);
rlm@1:   else lua_pushliteral(L, "");
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_reverse (lua_State *L) {
rlm@1:   size_t l;
rlm@1:   luaL_Buffer b;
rlm@1:   const char *s = luaL_checklstring(L, 1, &l);
rlm@1:   luaL_buffinit(L, &b);
rlm@1:   while (l--) luaL_addchar(&b, s[l]);
rlm@1:   luaL_pushresult(&b);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_lower (lua_State *L) {
rlm@1:   size_t l;
rlm@1:   size_t i;
rlm@1:   luaL_Buffer b;
rlm@1:   const char *s = luaL_checklstring(L, 1, &l);
rlm@1:   luaL_buffinit(L, &b);
rlm@1:   for (i=0; i<l; i++)
rlm@1:     luaL_addchar(&b, tolower(uchar(s[i])));
rlm@1:   luaL_pushresult(&b);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_upper (lua_State *L) {
rlm@1:   size_t l;
rlm@1:   size_t i;
rlm@1:   luaL_Buffer b;
rlm@1:   const char *s = luaL_checklstring(L, 1, &l);
rlm@1:   luaL_buffinit(L, &b);
rlm@1:   for (i=0; i<l; i++)
rlm@1:     luaL_addchar(&b, toupper(uchar(s[i])));
rlm@1:   luaL_pushresult(&b);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: static int str_rep (lua_State *L) {
rlm@1:   size_t l;
rlm@1:   luaL_Buffer b;
rlm@1:   const char *s = luaL_checklstring(L, 1, &l);
rlm@1:   int n = luaL_checkint(L, 2);
rlm@1:   luaL_buffinit(L, &b);
rlm@1:   while (n-- > 0)
rlm@1:     luaL_addlstring(&b, s, l);
rlm@1:   luaL_pushresult(&b);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_byte (lua_State *L) {
rlm@1:   size_t l;
rlm@1:   const char *s = luaL_checklstring(L, 1, &l);
rlm@1:   ptrdiff_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
rlm@1:   ptrdiff_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
rlm@1:   int n, i;
rlm@1:   if (posi <= 0) posi = 1;
rlm@1:   if ((size_t)pose > l) pose = l;
rlm@1:   if (posi > pose) return 0;  /* empty interval; return no values */
rlm@1:   n = (int)(pose -  posi + 1);
rlm@1:   if (posi + n <= pose)  /* overflow? */
rlm@1:     luaL_error(L, "string slice too long");
rlm@1:   luaL_checkstack(L, n, "string slice too long");
rlm@1:   for (i=0; i<n; i++)
rlm@1:     lua_pushinteger(L, uchar(s[posi+i-1]));
rlm@1:   return n;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_char (lua_State *L) {
rlm@1:   int n = lua_gettop(L);  /* number of arguments */
rlm@1:   int i;
rlm@1:   luaL_Buffer b;
rlm@1:   luaL_buffinit(L, &b);
rlm@1:   for (i=1; i<=n; i++) {
rlm@1:     int c = luaL_checkint(L, i);
rlm@1:     luaL_argcheck(L, uchar(c) == c, i, "invalid value");
rlm@1:     luaL_addchar(&b, uchar(c));
rlm@1:   }
rlm@1:   luaL_pushresult(&b);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int writer (lua_State *L, const void* b, size_t size, void* B) {
rlm@1:   (void)L;
rlm@1:   luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);
rlm@1:   return 0;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_dump (lua_State *L) {
rlm@1:   luaL_Buffer b;
rlm@1:   luaL_checktype(L, 1, LUA_TFUNCTION);
rlm@1:   lua_settop(L, 1);
rlm@1:   luaL_buffinit(L,&b);
rlm@1:   if (lua_dump(L, writer, &b) != 0)
rlm@1:     luaL_error(L, "unable to dump given function");
rlm@1:   luaL_pushresult(&b);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: 
rlm@1: /*
rlm@1: ** {======================================================
rlm@1: ** PATTERN MATCHING
rlm@1: ** =======================================================
rlm@1: */
rlm@1: 
rlm@1: 
rlm@1: #define CAP_UNFINISHED	(-1)
rlm@1: #define CAP_POSITION	(-2)
rlm@1: 
rlm@1: typedef struct MatchState {
rlm@1:   const char *src_init;  /* init of source string */
rlm@1:   const char *src_end;  /* end (`\0') of source string */
rlm@1:   lua_State *L;
rlm@1:   int level;  /* total number of captures (finished or unfinished) */
rlm@1:   struct {
rlm@1:     const char *init;
rlm@1:     ptrdiff_t len;
rlm@1:   } capture[LUA_MAXCAPTURES];
rlm@1: } MatchState;
rlm@1: 
rlm@1: 
rlm@1: #define L_ESC		'%'
rlm@1: #define SPECIALS	"^$*+?.([%-"
rlm@1: 
rlm@1: 
rlm@1: static int check_capture (MatchState *ms, int l) {
rlm@1:   l -= '1';
rlm@1:   if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
rlm@1:     return luaL_error(ms->L, "invalid capture index");
rlm@1:   return l;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int capture_to_close (MatchState *ms) {
rlm@1:   int level = ms->level;
rlm@1:   for (level--; level>=0; level--)
rlm@1:     if (ms->capture[level].len == CAP_UNFINISHED) return level;
rlm@1:   return luaL_error(ms->L, "invalid pattern capture");
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *classend (MatchState *ms, const char *p) {
rlm@1:   switch (*p++) {
rlm@1:     case L_ESC: {
rlm@1:       if (*p == '\0')
rlm@1:         luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
rlm@1:       return p+1;
rlm@1:     }
rlm@1:     case '[': {
rlm@1:       if (*p == '^') p++;
rlm@1:       do {  /* look for a `]' */
rlm@1:         if (*p == '\0')
rlm@1:           luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
rlm@1:         if (*(p++) == L_ESC && *p != '\0')
rlm@1:           p++;  /* skip escapes (e.g. `%]') */
rlm@1:       } while (*p != ']');
rlm@1:       return p+1;
rlm@1:     }
rlm@1:     default: {
rlm@1:       return p;
rlm@1:     }
rlm@1:   }
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int match_class (int c, int cl) {
rlm@1:   int res;
rlm@1:   switch (tolower(cl)) {
rlm@1:     case 'a' : res = isalpha(c); break;
rlm@1:     case 'c' : res = iscntrl(c); break;
rlm@1:     case 'd' : res = isdigit(c); break;
rlm@1:     case 'l' : res = islower(c); break;
rlm@1:     case 'p' : res = ispunct(c); break;
rlm@1:     case 's' : res = isspace(c); break;
rlm@1:     case 'u' : res = isupper(c); break;
rlm@1:     case 'w' : res = isalnum(c); break;
rlm@1:     case 'x' : res = isxdigit(c); break;
rlm@1:     case 'z' : res = (c == 0); break;
rlm@1:     default: return (cl == c);
rlm@1:   }
rlm@1:   return (islower(cl) ? res : !res);
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int matchbracketclass (int c, const char *p, const char *ec) {
rlm@1:   int sig = 1;
rlm@1:   if (*(p+1) == '^') {
rlm@1:     sig = 0;
rlm@1:     p++;  /* skip the `^' */
rlm@1:   }
rlm@1:   while (++p < ec) {
rlm@1:     if (*p == L_ESC) {
rlm@1:       p++;
rlm@1:       if (match_class(c, uchar(*p)))
rlm@1:         return sig;
rlm@1:     }
rlm@1:     else if ((*(p+1) == '-') && (p+2 < ec)) {
rlm@1:       p+=2;
rlm@1:       if (uchar(*(p-2)) <= c && c <= uchar(*p))
rlm@1:         return sig;
rlm@1:     }
rlm@1:     else if (uchar(*p) == c) return sig;
rlm@1:   }
rlm@1:   return !sig;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int singlematch (int c, const char *p, const char *ep) {
rlm@1:   switch (*p) {
rlm@1:     case '.': return 1;  /* matches any char */
rlm@1:     case L_ESC: return match_class(c, uchar(*(p+1)));
rlm@1:     case '[': return matchbracketclass(c, p, ep-1);
rlm@1:     default:  return (uchar(*p) == c);
rlm@1:   }
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *match (MatchState *ms, const char *s, const char *p);
rlm@1: 
rlm@1: 
rlm@1: static const char *matchbalance (MatchState *ms, const char *s,
rlm@1:                                    const char *p) {
rlm@1:   if (*p == 0 || *(p+1) == 0)
rlm@1:     luaL_error(ms->L, "unbalanced pattern");
rlm@1:   if (*s != *p) return NULL;
rlm@1:   else {
rlm@1:     int b = *p;
rlm@1:     int e = *(p+1);
rlm@1:     int cont = 1;
rlm@1:     while (++s < ms->src_end) {
rlm@1:       if (*s == e) {
rlm@1:         if (--cont == 0) return s+1;
rlm@1:       }
rlm@1:       else if (*s == b) cont++;
rlm@1:     }
rlm@1:   }
rlm@1:   return NULL;  /* string ends out of balance */
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *max_expand (MatchState *ms, const char *s,
rlm@1:                                  const char *p, const char *ep) {
rlm@1:   ptrdiff_t i = 0;  /* counts maximum expand for item */
rlm@1:   while ((s+i)<ms->src_end && singlematch(uchar(*(s+i)), p, ep))
rlm@1:     i++;
rlm@1:   /* keeps trying to match with the maximum repetitions */
rlm@1:   while (i>=0) {
rlm@1:     const char *res = match(ms, (s+i), ep+1);
rlm@1:     if (res) return res;
rlm@1:     i--;  /* else didn't match; reduce 1 repetition to try again */
rlm@1:   }
rlm@1:   return NULL;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *min_expand (MatchState *ms, const char *s,
rlm@1:                                  const char *p, const char *ep) {
rlm@1:   for (;;) {
rlm@1:     const char *res = match(ms, s, ep+1);
rlm@1:     if (res != NULL)
rlm@1:       return res;
rlm@1:     else if (s<ms->src_end && singlematch(uchar(*s), p, ep))
rlm@1:       s++;  /* try with one more repetition */
rlm@1:     else return NULL;
rlm@1:   }
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *start_capture (MatchState *ms, const char *s,
rlm@1:                                     const char *p, int what) {
rlm@1:   const char *res;
rlm@1:   int level = ms->level;
rlm@1:   if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
rlm@1:   ms->capture[level].init = s;
rlm@1:   ms->capture[level].len = what;
rlm@1:   ms->level = level+1;
rlm@1:   if ((res=match(ms, s, p)) == NULL)  /* match failed? */
rlm@1:     ms->level--;  /* undo capture */
rlm@1:   return res;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *end_capture (MatchState *ms, const char *s,
rlm@1:                                   const char *p) {
rlm@1:   int l = capture_to_close(ms);
rlm@1:   const char *res;
rlm@1:   ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
rlm@1:   if ((res = match(ms, s, p)) == NULL)  /* match failed? */
rlm@1:     ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
rlm@1:   return res;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *match_capture (MatchState *ms, const char *s, int l) {
rlm@1:   size_t len;
rlm@1:   l = check_capture(ms, l);
rlm@1:   len = ms->capture[l].len;
rlm@1:   if ((size_t)(ms->src_end-s) >= len &&
rlm@1:       memcmp(ms->capture[l].init, s, len) == 0)
rlm@1:     return s+len;
rlm@1:   else return NULL;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const char *match (MatchState *ms, const char *s, const char *p) {
rlm@1:   init: /* using goto's to optimize tail recursion */
rlm@1:   switch (*p) {
rlm@1:     case '(': {  /* start capture */
rlm@1:       if (*(p+1) == ')')  /* position capture? */
rlm@1:         return start_capture(ms, s, p+2, CAP_POSITION);
rlm@1:       else
rlm@1:         return start_capture(ms, s, p+1, CAP_UNFINISHED);
rlm@1:     }
rlm@1:     case ')': {  /* end capture */
rlm@1:       return end_capture(ms, s, p+1);
rlm@1:     }
rlm@1:     case L_ESC: {
rlm@1:       switch (*(p+1)) {
rlm@1:         case 'b': {  /* balanced string? */
rlm@1:           s = matchbalance(ms, s, p+2);
rlm@1:           if (s == NULL) return NULL;
rlm@1:           p+=4; goto init;  /* else return match(ms, s, p+4); */
rlm@1:         }
rlm@1:         case 'f': {  /* frontier? */
rlm@1:           const char *ep; char previous;
rlm@1:           p += 2;
rlm@1:           if (*p != '[')
rlm@1:             luaL_error(ms->L, "missing " LUA_QL("[") " after "
rlm@1:                                LUA_QL("%%f") " in pattern");
rlm@1:           ep = classend(ms, p);  /* points to what is next */
rlm@1:           previous = (s == ms->src_init) ? '\0' : *(s-1);
rlm@1:           if (matchbracketclass(uchar(previous), p, ep-1) ||
rlm@1:              !matchbracketclass(uchar(*s), p, ep-1)) return NULL;
rlm@1:           p=ep; goto init;  /* else return match(ms, s, ep); */
rlm@1:         }
rlm@1:         default: {
rlm@1:           if (isdigit(uchar(*(p+1)))) {  /* capture results (%0-%9)? */
rlm@1:             s = match_capture(ms, s, uchar(*(p+1)));
rlm@1:             if (s == NULL) return NULL;
rlm@1:             p+=2; goto init;  /* else return match(ms, s, p+2) */
rlm@1:           }
rlm@1:           goto dflt;  /* case default */
rlm@1:         }
rlm@1:       }
rlm@1:     }
rlm@1:     case '\0': {  /* end of pattern */
rlm@1:       return s;  /* match succeeded */
rlm@1:     }
rlm@1:     case '$': {
rlm@1:       if (*(p+1) == '\0')  /* is the `$' the last char in pattern? */
rlm@1:         return (s == ms->src_end) ? s : NULL;  /* check end of string */
rlm@1:       else goto dflt;
rlm@1:     }
rlm@1:     default: dflt: {  /* it is a pattern item */
rlm@1:       const char *ep = classend(ms, p);  /* points to what is next */
rlm@1:       int m = s<ms->src_end && singlematch(uchar(*s), p, ep);
rlm@1:       switch (*ep) {
rlm@1:         case '?': {  /* optional */
rlm@1:           const char *res;
rlm@1:           if (m && ((res=match(ms, s+1, ep+1)) != NULL))
rlm@1:             return res;
rlm@1:           p=ep+1; goto init;  /* else return match(ms, s, ep+1); */
rlm@1:         }
rlm@1:         case '*': {  /* 0 or more repetitions */
rlm@1:           return max_expand(ms, s, p, ep);
rlm@1:         }
rlm@1:         case '+': {  /* 1 or more repetitions */
rlm@1:           return (m ? max_expand(ms, s+1, p, ep) : NULL);
rlm@1:         }
rlm@1:         case '-': {  /* 0 or more repetitions (minimum) */
rlm@1:           return min_expand(ms, s, p, ep);
rlm@1:         }
rlm@1:         default: {
rlm@1:           if (!m) return NULL;
rlm@1:           s++; p=ep; goto init;  /* else return match(ms, s+1, ep); */
rlm@1:         }
rlm@1:       }
rlm@1:     }
rlm@1:   }
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: 
rlm@1: static const char *lmemfind (const char *s1, size_t l1,
rlm@1:                                const char *s2, size_t l2) {
rlm@1:   if (l2 == 0) return s1;  /* empty strings are everywhere */
rlm@1:   else if (l2 > l1) return NULL;  /* avoids a negative `l1' */
rlm@1:   else {
rlm@1:     const char *init;  /* to search for a `*s2' inside `s1' */
rlm@1:     l2--;  /* 1st char will be checked by `memchr' */
rlm@1:     l1 = l1-l2;  /* `s2' cannot be found after that */
rlm@1:     while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
rlm@1:       init++;   /* 1st char is already checked */
rlm@1:       if (memcmp(init, s2+1, l2) == 0)
rlm@1:         return init-1;
rlm@1:       else {  /* correct `l1' and `s1' to try again */
rlm@1:         l1 -= init-s1;
rlm@1:         s1 = init;
rlm@1:       }
rlm@1:     }
rlm@1:     return NULL;  /* not found */
rlm@1:   }
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static void push_onecapture (MatchState *ms, int i, const char *s,
rlm@1:                                                     const char *e) {
rlm@1:   if (i >= ms->level) {
rlm@1:     if (i == 0)  /* ms->level == 0, too */
rlm@1:       lua_pushlstring(ms->L, s, e - s);  /* add whole match */
rlm@1:     else
rlm@1:       luaL_error(ms->L, "invalid capture index");
rlm@1:   }
rlm@1:   else {
rlm@1:     ptrdiff_t l = ms->capture[i].len;
rlm@1:     if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
rlm@1:     if (l == CAP_POSITION)
rlm@1:       lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
rlm@1:     else
rlm@1:       lua_pushlstring(ms->L, ms->capture[i].init, l);
rlm@1:   }
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int push_captures (MatchState *ms, const char *s, const char *e) {
rlm@1:   int i;
rlm@1:   int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
rlm@1:   luaL_checkstack(ms->L, nlevels, "too many captures");
rlm@1:   for (i = 0; i < nlevels; i++)
rlm@1:     push_onecapture(ms, i, s, e);
rlm@1:   return nlevels;  /* number of strings pushed */
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_find_aux (lua_State *L, int find) {
rlm@1:   size_t l1, l2;
rlm@1:   const char *s = luaL_checklstring(L, 1, &l1);
rlm@1:   const char *p = luaL_checklstring(L, 2, &l2);
rlm@1:   ptrdiff_t init = posrelat(luaL_optinteger(L, 3, 1), l1) - 1;
rlm@1:   if (init < 0) init = 0;
rlm@1:   else if ((size_t)(init) > l1) init = (ptrdiff_t)l1;
rlm@1:   if (find && (lua_toboolean(L, 4) ||  /* explicit request? */
rlm@1:       strpbrk(p, SPECIALS) == NULL)) {  /* or no special characters? */
rlm@1:     /* do a plain search */
rlm@1:     const char *s2 = lmemfind(s+init, l1-init, p, l2);
rlm@1:     if (s2) {
rlm@1:       lua_pushinteger(L, s2-s+1);
rlm@1:       lua_pushinteger(L, s2-s+l2);
rlm@1:       return 2;
rlm@1:     }
rlm@1:   }
rlm@1:   else {
rlm@1:     MatchState ms;
rlm@1:     int anchor = (*p == '^') ? (p++, 1) : 0;
rlm@1:     const char *s1=s+init;
rlm@1:     ms.L = L;
rlm@1:     ms.src_init = s;
rlm@1:     ms.src_end = s+l1;
rlm@1:     do {
rlm@1:       const char *res;
rlm@1:       ms.level = 0;
rlm@1:       if ((res=match(&ms, s1, p)) != NULL) {
rlm@1:         if (find) {
rlm@1:           lua_pushinteger(L, s1-s+1);  /* start */
rlm@1:           lua_pushinteger(L, res-s);   /* end */
rlm@1:           return push_captures(&ms, NULL, 0) + 2;
rlm@1:         }
rlm@1:         else
rlm@1:           return push_captures(&ms, s1, res);
rlm@1:       }
rlm@1:     } while (s1++ < ms.src_end && !anchor);
rlm@1:   }
rlm@1:   lua_pushnil(L);  /* not found */
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_find (lua_State *L) {
rlm@1:   return str_find_aux(L, 1);
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_match (lua_State *L) {
rlm@1:   return str_find_aux(L, 0);
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int gmatch_aux (lua_State *L) {
rlm@1:   MatchState ms;
rlm@1:   size_t ls;
rlm@1:   const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
rlm@1:   const char *p = lua_tostring(L, lua_upvalueindex(2));
rlm@1:   const char *src;
rlm@1:   ms.L = L;
rlm@1:   ms.src_init = s;
rlm@1:   ms.src_end = s+ls;
rlm@1:   for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
rlm@1:        src <= ms.src_end;
rlm@1:        src++) {
rlm@1:     const char *e;
rlm@1:     ms.level = 0;
rlm@1:     if ((e = match(&ms, src, p)) != NULL) {
rlm@1:       lua_Integer newstart = e-s;
rlm@1:       if (e == src) newstart++;  /* empty match? go at least one position */
rlm@1:       lua_pushinteger(L, newstart);
rlm@1:       lua_replace(L, lua_upvalueindex(3));
rlm@1:       return push_captures(&ms, src, e);
rlm@1:     }
rlm@1:   }
rlm@1:   return 0;  /* not found */
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int gmatch (lua_State *L) {
rlm@1:   luaL_checkstring(L, 1);
rlm@1:   luaL_checkstring(L, 2);
rlm@1:   lua_settop(L, 2);
rlm@1:   lua_pushinteger(L, 0);
rlm@1:   lua_pushcclosure(L, gmatch_aux, 3);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int gfind_nodef (lua_State *L) {
rlm@1:   return luaL_error(L, LUA_QL("string.gfind") " was renamed to "
rlm@1:                        LUA_QL("string.gmatch"));
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
rlm@1:                                                    const char *e) {
rlm@1:   size_t l, i;
rlm@1:   const char *news = lua_tolstring(ms->L, 3, &l);
rlm@1:   for (i = 0; i < l; i++) {
rlm@1:     if (news[i] != L_ESC)
rlm@1:       luaL_addchar(b, news[i]);
rlm@1:     else {
rlm@1:       i++;  /* skip ESC */
rlm@1:       if (!isdigit(uchar(news[i])))
rlm@1:         luaL_addchar(b, news[i]);
rlm@1:       else if (news[i] == '0')
rlm@1:           luaL_addlstring(b, s, e - s);
rlm@1:       else {
rlm@1:         push_onecapture(ms, news[i] - '1', s, e);
rlm@1:         luaL_addvalue(b);  /* add capture to accumulated result */
rlm@1:       }
rlm@1:     }
rlm@1:   }
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
rlm@1:                                                        const char *e) {
rlm@1:   lua_State *L = ms->L;
rlm@1:   switch (lua_type(L, 3)) {
rlm@1:     case LUA_TNUMBER:
rlm@1:     case LUA_TSTRING: {
rlm@1:       add_s(ms, b, s, e);
rlm@1:       return;
rlm@1:     }
rlm@1:     case LUA_TFUNCTION: {
rlm@1:       int n;
rlm@1:       lua_pushvalue(L, 3);
rlm@1:       n = push_captures(ms, s, e);
rlm@1:       lua_call(L, n, 1);
rlm@1:       break;
rlm@1:     }
rlm@1:     case LUA_TTABLE: {
rlm@1:       push_onecapture(ms, 0, s, e);
rlm@1:       lua_gettable(L, 3);
rlm@1:       break;
rlm@1:     }
rlm@1:   }
rlm@1:   if (!lua_toboolean(L, -1)) {  /* nil or false? */
rlm@1:     lua_pop(L, 1);
rlm@1:     lua_pushlstring(L, s, e - s);  /* keep original text */
rlm@1:   }
rlm@1:   else if (!lua_isstring(L, -1))
rlm@1:     luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1)); 
rlm@1:   luaL_addvalue(b);  /* add result to accumulator */
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_gsub (lua_State *L) {
rlm@1:   size_t srcl;
rlm@1:   const char *src = luaL_checklstring(L, 1, &srcl);
rlm@1:   const char *p = luaL_checkstring(L, 2);
rlm@1:   int  tr = lua_type(L, 3);
rlm@1:   int max_s = luaL_optint(L, 4, srcl+1);
rlm@1:   int anchor = (*p == '^') ? (p++, 1) : 0;
rlm@1:   int n = 0;
rlm@1:   MatchState ms;
rlm@1:   luaL_Buffer b;
rlm@1:   luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
rlm@1:                    tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
rlm@1:                       "string/function/table expected");
rlm@1:   luaL_buffinit(L, &b);
rlm@1:   ms.L = L;
rlm@1:   ms.src_init = src;
rlm@1:   ms.src_end = src+srcl;
rlm@1:   while (n < max_s) {
rlm@1:     const char *e;
rlm@1:     ms.level = 0;
rlm@1:     e = match(&ms, src, p);
rlm@1:     if (e) {
rlm@1:       n++;
rlm@1:       add_value(&ms, &b, src, e);
rlm@1:     }
rlm@1:     if (e && e>src) /* non empty match? */
rlm@1:       src = e;  /* skip it */
rlm@1:     else if (src < ms.src_end)
rlm@1:       luaL_addchar(&b, *src++);
rlm@1:     else break;
rlm@1:     if (anchor) break;
rlm@1:   }
rlm@1:   luaL_addlstring(&b, src, ms.src_end-src);
rlm@1:   luaL_pushresult(&b);
rlm@1:   lua_pushinteger(L, n);  /* number of substitutions */
rlm@1:   return 2;
rlm@1: }
rlm@1: 
rlm@1: /* }====================================================== */
rlm@1: 
rlm@1: 
rlm@1: /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
rlm@1: #define MAX_ITEM	512
rlm@1: /* valid flags in a format specification */
rlm@1: #define FLAGS	"-+ #0"
rlm@1: /*
rlm@1: ** maximum size of each format specification (such as '%-099.99d')
rlm@1: ** (+10 accounts for %99.99x plus margin of error)
rlm@1: */
rlm@1: #define MAX_FORMAT	(sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
rlm@1: 
rlm@1: 
rlm@1: static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
rlm@1:   size_t l;
rlm@1:   const char *s = luaL_checklstring(L, arg, &l);
rlm@1:   luaL_addchar(b, '"');
rlm@1:   while (l--) {
rlm@1:     switch (*s) {
rlm@1:       case '"': case '\\': case '\n': {
rlm@1:         luaL_addchar(b, '\\');
rlm@1:         luaL_addchar(b, *s);
rlm@1:         break;
rlm@1:       }
rlm@1:       case '\r': {
rlm@1:         luaL_addlstring(b, "\\r", 2);
rlm@1:         break;
rlm@1:       }
rlm@1:       case '\0': {
rlm@1:         luaL_addlstring(b, "\\000", 4);
rlm@1:         break;
rlm@1:       }
rlm@1:       default: {
rlm@1:         luaL_addchar(b, *s);
rlm@1:         break;
rlm@1:       }
rlm@1:     }
rlm@1:     s++;
rlm@1:   }
rlm@1:   luaL_addchar(b, '"');
rlm@1: }
rlm@1: 
rlm@1: static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
rlm@1:   const char *p = strfrmt;
rlm@1:   while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++;  /* skip flags */
rlm@1:   if ((size_t)(p - strfrmt) >= sizeof(FLAGS))
rlm@1:     luaL_error(L, "invalid format (repeated flags)");
rlm@1:   if (isdigit(uchar(*p))) p++;  /* skip width */
rlm@1:   if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
rlm@1:   if (*p == '.') {
rlm@1:     p++;
rlm@1:     if (isdigit(uchar(*p))) p++;  /* skip precision */
rlm@1:     if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
rlm@1:   }
rlm@1:   if (isdigit(uchar(*p)))
rlm@1:     luaL_error(L, "invalid format (width or precision too long)");
rlm@1:   *(form++) = '%';
rlm@1:   strncpy(form, strfrmt, p - strfrmt + 1);
rlm@1:   form += p - strfrmt + 1;
rlm@1:   *form = '\0';
rlm@1:   return p;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static void addintlen (char *form) {
rlm@1:   size_t l = strlen(form);
rlm@1:   char spec = form[l - 1];
rlm@1:   strcpy(form + l - 1, LUA_INTFRMLEN);
rlm@1:   form[l + sizeof(LUA_INTFRMLEN) - 2] = spec;
rlm@1:   form[l + sizeof(LUA_INTFRMLEN) - 1] = '\0';
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static int str_format (lua_State *L) {
rlm@1:   int arg = 1;
rlm@1:   size_t sfl;
rlm@1:   const char *strfrmt = luaL_checklstring(L, arg, &sfl);
rlm@1:   const char *strfrmt_end = strfrmt+sfl;
rlm@1:   luaL_Buffer b;
rlm@1:   luaL_buffinit(L, &b);
rlm@1:   while (strfrmt < strfrmt_end) {
rlm@1:     if (*strfrmt != L_ESC)
rlm@1:       luaL_addchar(&b, *strfrmt++);
rlm@1:     else if (*++strfrmt == L_ESC)
rlm@1:       luaL_addchar(&b, *strfrmt++);  /* %% */
rlm@1:     else { /* format item */
rlm@1:       char form[MAX_FORMAT];  /* to store the format (`%...') */
rlm@1:       char buff[MAX_ITEM];  /* to store the formatted item */
rlm@1:       arg++;
rlm@1:       strfrmt = scanformat(L, strfrmt, form);
rlm@1:       switch (*strfrmt++) {
rlm@1:         case 'c': {
rlm@1:           sprintf(buff, form, (int)luaL_checknumber(L, arg));
rlm@1:           break;
rlm@1:         }
rlm@1:         case 'd':  case 'i': {
rlm@1:           addintlen(form);
rlm@1:           sprintf(buff, form, (LUA_INTFRM_T)luaL_checknumber(L, arg));
rlm@1:           break;
rlm@1:         }
rlm@1:         case 'o':  case 'u':  case 'x':  case 'X': {
rlm@1:           addintlen(form);
rlm@1:           sprintf(buff, form, (unsigned LUA_INTFRM_T)luaL_checknumber(L, arg));
rlm@1:           break;
rlm@1:         }
rlm@1:         case 'e':  case 'E': case 'f':
rlm@1:         case 'g': case 'G': {
rlm@1:           sprintf(buff, form, (double)luaL_checknumber(L, arg));
rlm@1:           break;
rlm@1:         }
rlm@1:         case 'q': {
rlm@1:           addquoted(L, &b, arg);
rlm@1:           continue;  /* skip the 'addsize' at the end */
rlm@1:         }
rlm@1:         case 's': {
rlm@1:           size_t l;
rlm@1:           const char *s = luaL_checklstring(L, arg, &l);
rlm@1:           if (!strchr(form, '.') && l >= 100) {
rlm@1:             /* no precision and string is too long to be formatted;
rlm@1:                keep original string */
rlm@1:             lua_pushvalue(L, arg);
rlm@1:             luaL_addvalue(&b);
rlm@1:             continue;  /* skip the `addsize' at the end */
rlm@1:           }
rlm@1:           else {
rlm@1:             sprintf(buff, form, s);
rlm@1:             break;
rlm@1:           }
rlm@1:         }
rlm@1:         default: {  /* also treat cases `pnLlh' */
rlm@1:           return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
rlm@1:                                LUA_QL("format"), *(strfrmt - 1));
rlm@1:         }
rlm@1:       }
rlm@1:       luaL_addlstring(&b, buff, strlen(buff));
rlm@1:     }
rlm@1:   }
rlm@1:   luaL_pushresult(&b);
rlm@1:   return 1;
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: static const luaL_Reg strlib[] = {
rlm@1:   {"byte", str_byte},
rlm@1:   {"char", str_char},
rlm@1:   {"dump", str_dump},
rlm@1:   {"find", str_find},
rlm@1:   {"format", str_format},
rlm@1:   {"gfind", gfind_nodef},
rlm@1:   {"gmatch", gmatch},
rlm@1:   {"gsub", str_gsub},
rlm@1:   {"len", str_len},
rlm@1:   {"lower", str_lower},
rlm@1:   {"match", str_match},
rlm@1:   {"rep", str_rep},
rlm@1:   {"reverse", str_reverse},
rlm@1:   {"sub", str_sub},
rlm@1:   {"upper", str_upper},
rlm@1:   {NULL, NULL}
rlm@1: };
rlm@1: 
rlm@1: 
rlm@1: static void createmetatable (lua_State *L) {
rlm@1:   lua_createtable(L, 0, 1);  /* create metatable for strings */
rlm@1:   lua_pushliteral(L, "");  /* dummy string */
rlm@1:   lua_pushvalue(L, -2);
rlm@1:   lua_setmetatable(L, -2);  /* set string metatable */
rlm@1:   lua_pop(L, 1);  /* pop dummy string */
rlm@1:   lua_pushvalue(L, -2);  /* string library... */
rlm@1:   lua_setfield(L, -2, "__index");  /* ...is the __index metamethod */
rlm@1:   lua_pop(L, 1);  /* pop metatable */
rlm@1: }
rlm@1: 
rlm@1: 
rlm@1: /*
rlm@1: ** Open string library
rlm@1: */
rlm@1: LUALIB_API int luaopen_string (lua_State *L) {
rlm@1:   luaL_register(L, LUA_STRLIBNAME, strlib);
rlm@1: #if defined(LUA_COMPAT_GFIND)
rlm@1:   lua_getfield(L, -1, "gmatch");
rlm@1:   lua_setfield(L, -2, "gfind");
rlm@1: #endif
rlm@1:   createmetatable(L);
rlm@1:   return 1;
rlm@1: }
rlm@1: