18e3e3a7aSWarner Losh /*
20495ed39SKyle Evans ** $Id: lstrlib.c $
38e3e3a7aSWarner Losh ** Standard library for string operations and pattern-matching
48e3e3a7aSWarner Losh ** See Copyright Notice in lua.h
58e3e3a7aSWarner Losh */
68e3e3a7aSWarner Losh
78e3e3a7aSWarner Losh #define lstrlib_c
88e3e3a7aSWarner Losh #define LUA_LIB
98e3e3a7aSWarner Losh
108e3e3a7aSWarner Losh #include "lprefix.h"
118e3e3a7aSWarner Losh
128e3e3a7aSWarner Losh
138e3e3a7aSWarner Losh #include <ctype.h>
148e3e3a7aSWarner Losh #include <float.h>
158e3e3a7aSWarner Losh #include <limits.h>
168e3e3a7aSWarner Losh #include <locale.h>
170495ed39SKyle Evans #include <math.h>
188e3e3a7aSWarner Losh #include <stddef.h>
198e3e3a7aSWarner Losh #include <stdio.h>
208e3e3a7aSWarner Losh #include <stdlib.h>
218e3e3a7aSWarner Losh #include <string.h>
228e3e3a7aSWarner Losh
238e3e3a7aSWarner Losh #include "lua.h"
248e3e3a7aSWarner Losh
258e3e3a7aSWarner Losh #include "lauxlib.h"
268e3e3a7aSWarner Losh #include "lualib.h"
278e3e3a7aSWarner Losh
288e3e3a7aSWarner Losh
298e3e3a7aSWarner Losh /*
308e3e3a7aSWarner Losh ** maximum number of captures that a pattern can do during
318e3e3a7aSWarner Losh ** pattern-matching. This limit is arbitrary, but must fit in
328e3e3a7aSWarner Losh ** an unsigned char.
338e3e3a7aSWarner Losh */
348e3e3a7aSWarner Losh #if !defined(LUA_MAXCAPTURES)
358e3e3a7aSWarner Losh #define LUA_MAXCAPTURES 32
368e3e3a7aSWarner Losh #endif
378e3e3a7aSWarner Losh
388e3e3a7aSWarner Losh
398e3e3a7aSWarner Losh /* macro to 'unsign' a character */
408e3e3a7aSWarner Losh #define uchar(c) ((unsigned char)(c))
418e3e3a7aSWarner Losh
428e3e3a7aSWarner Losh
438e3e3a7aSWarner Losh /*
448e3e3a7aSWarner Losh ** Some sizes are better limited to fit in 'int', but must also fit in
458e3e3a7aSWarner Losh ** 'size_t'. (We assume that 'lua_Integer' cannot be smaller than 'int'.)
468e3e3a7aSWarner Losh */
478e3e3a7aSWarner Losh #define MAX_SIZET ((size_t)(~(size_t)0))
488e3e3a7aSWarner Losh
498e3e3a7aSWarner Losh #define MAXSIZE \
508e3e3a7aSWarner Losh (sizeof(size_t) < sizeof(int) ? MAX_SIZET : (size_t)(INT_MAX))
518e3e3a7aSWarner Losh
528e3e3a7aSWarner Losh
538e3e3a7aSWarner Losh
548e3e3a7aSWarner Losh
str_len(lua_State * L)558e3e3a7aSWarner Losh static int str_len (lua_State *L) {
568e3e3a7aSWarner Losh size_t l;
578e3e3a7aSWarner Losh luaL_checklstring(L, 1, &l);
588e3e3a7aSWarner Losh lua_pushinteger(L, (lua_Integer)l);
598e3e3a7aSWarner Losh return 1;
608e3e3a7aSWarner Losh }
618e3e3a7aSWarner Losh
628e3e3a7aSWarner Losh
630495ed39SKyle Evans /*
640495ed39SKyle Evans ** translate a relative initial string position
650495ed39SKyle Evans ** (negative means back from end): clip result to [1, inf).
660495ed39SKyle Evans ** The length of any string in Lua must fit in a lua_Integer,
670495ed39SKyle Evans ** so there are no overflows in the casts.
680495ed39SKyle Evans ** The inverted comparison avoids a possible overflow
690495ed39SKyle Evans ** computing '-pos'.
700495ed39SKyle Evans */
posrelatI(lua_Integer pos,size_t len)710495ed39SKyle Evans static size_t posrelatI (lua_Integer pos, size_t len) {
720495ed39SKyle Evans if (pos > 0)
730495ed39SKyle Evans return (size_t)pos;
740495ed39SKyle Evans else if (pos == 0)
750495ed39SKyle Evans return 1;
760495ed39SKyle Evans else if (pos < -(lua_Integer)len) /* inverted comparison */
770495ed39SKyle Evans return 1; /* clip to 1 */
780495ed39SKyle Evans else return len + (size_t)pos + 1;
790495ed39SKyle Evans }
800495ed39SKyle Evans
810495ed39SKyle Evans
820495ed39SKyle Evans /*
830495ed39SKyle Evans ** Gets an optional ending string position from argument 'arg',
840495ed39SKyle Evans ** with default value 'def'.
850495ed39SKyle Evans ** Negative means back from end: clip result to [0, len]
860495ed39SKyle Evans */
getendpos(lua_State * L,int arg,lua_Integer def,size_t len)870495ed39SKyle Evans static size_t getendpos (lua_State *L, int arg, lua_Integer def,
880495ed39SKyle Evans size_t len) {
890495ed39SKyle Evans lua_Integer pos = luaL_optinteger(L, arg, def);
900495ed39SKyle Evans if (pos > (lua_Integer)len)
910495ed39SKyle Evans return len;
920495ed39SKyle Evans else if (pos >= 0)
930495ed39SKyle Evans return (size_t)pos;
940495ed39SKyle Evans else if (pos < -(lua_Integer)len)
950495ed39SKyle Evans return 0;
960495ed39SKyle Evans else return len + (size_t)pos + 1;
978e3e3a7aSWarner Losh }
988e3e3a7aSWarner Losh
998e3e3a7aSWarner Losh
str_sub(lua_State * L)1008e3e3a7aSWarner Losh static int str_sub (lua_State *L) {
1018e3e3a7aSWarner Losh size_t l;
1028e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &l);
1030495ed39SKyle Evans size_t start = posrelatI(luaL_checkinteger(L, 2), l);
1040495ed39SKyle Evans size_t end = getendpos(L, 3, -1, l);
1058e3e3a7aSWarner Losh if (start <= end)
1060495ed39SKyle Evans lua_pushlstring(L, s + start - 1, (end - start) + 1);
1078e3e3a7aSWarner Losh else lua_pushliteral(L, "");
1088e3e3a7aSWarner Losh return 1;
1098e3e3a7aSWarner Losh }
1108e3e3a7aSWarner Losh
1118e3e3a7aSWarner Losh
str_reverse(lua_State * L)1128e3e3a7aSWarner Losh static int str_reverse (lua_State *L) {
1138e3e3a7aSWarner Losh size_t l, i;
1148e3e3a7aSWarner Losh luaL_Buffer b;
1158e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &l);
1168e3e3a7aSWarner Losh char *p = luaL_buffinitsize(L, &b, l);
1178e3e3a7aSWarner Losh for (i = 0; i < l; i++)
1188e3e3a7aSWarner Losh p[i] = s[l - i - 1];
1198e3e3a7aSWarner Losh luaL_pushresultsize(&b, l);
1208e3e3a7aSWarner Losh return 1;
1218e3e3a7aSWarner Losh }
1228e3e3a7aSWarner Losh
1238e3e3a7aSWarner Losh
str_lower(lua_State * L)1248e3e3a7aSWarner Losh static int str_lower (lua_State *L) {
1258e3e3a7aSWarner Losh size_t l;
1268e3e3a7aSWarner Losh size_t i;
1278e3e3a7aSWarner Losh luaL_Buffer b;
1288e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &l);
1298e3e3a7aSWarner Losh char *p = luaL_buffinitsize(L, &b, l);
1308e3e3a7aSWarner Losh for (i=0; i<l; i++)
1318e3e3a7aSWarner Losh p[i] = tolower(uchar(s[i]));
1328e3e3a7aSWarner Losh luaL_pushresultsize(&b, l);
1338e3e3a7aSWarner Losh return 1;
1348e3e3a7aSWarner Losh }
1358e3e3a7aSWarner Losh
1368e3e3a7aSWarner Losh
str_upper(lua_State * L)1378e3e3a7aSWarner Losh static int str_upper (lua_State *L) {
1388e3e3a7aSWarner Losh size_t l;
1398e3e3a7aSWarner Losh size_t i;
1408e3e3a7aSWarner Losh luaL_Buffer b;
1418e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &l);
1428e3e3a7aSWarner Losh char *p = luaL_buffinitsize(L, &b, l);
1438e3e3a7aSWarner Losh for (i=0; i<l; i++)
1448e3e3a7aSWarner Losh p[i] = toupper(uchar(s[i]));
1458e3e3a7aSWarner Losh luaL_pushresultsize(&b, l);
1468e3e3a7aSWarner Losh return 1;
1478e3e3a7aSWarner Losh }
1488e3e3a7aSWarner Losh
1498e3e3a7aSWarner Losh
str_rep(lua_State * L)1508e3e3a7aSWarner Losh static int str_rep (lua_State *L) {
1518e3e3a7aSWarner Losh size_t l, lsep;
1528e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &l);
1538e3e3a7aSWarner Losh lua_Integer n = luaL_checkinteger(L, 2);
1548e3e3a7aSWarner Losh const char *sep = luaL_optlstring(L, 3, "", &lsep);
1558c784bb8SWarner Losh if (n <= 0)
1568c784bb8SWarner Losh lua_pushliteral(L, "");
1578c784bb8SWarner Losh else if (l_unlikely(l + lsep < l || l + lsep > MAXSIZE / n))
1588e3e3a7aSWarner Losh return luaL_error(L, "resulting string too large");
1598e3e3a7aSWarner Losh else {
1608e3e3a7aSWarner Losh size_t totallen = (size_t)n * l + (size_t)(n - 1) * lsep;
1618e3e3a7aSWarner Losh luaL_Buffer b;
1628e3e3a7aSWarner Losh char *p = luaL_buffinitsize(L, &b, totallen);
1638e3e3a7aSWarner Losh while (n-- > 1) { /* first n-1 copies (followed by separator) */
1648e3e3a7aSWarner Losh memcpy(p, s, l * sizeof(char)); p += l;
1658e3e3a7aSWarner Losh if (lsep > 0) { /* empty 'memcpy' is not that cheap */
1668e3e3a7aSWarner Losh memcpy(p, sep, lsep * sizeof(char));
1678e3e3a7aSWarner Losh p += lsep;
1688e3e3a7aSWarner Losh }
1698e3e3a7aSWarner Losh }
1708e3e3a7aSWarner Losh memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */
1718e3e3a7aSWarner Losh luaL_pushresultsize(&b, totallen);
1728e3e3a7aSWarner Losh }
1738e3e3a7aSWarner Losh return 1;
1748e3e3a7aSWarner Losh }
1758e3e3a7aSWarner Losh
1768e3e3a7aSWarner Losh
str_byte(lua_State * L)1778e3e3a7aSWarner Losh static int str_byte (lua_State *L) {
1788e3e3a7aSWarner Losh size_t l;
1798e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &l);
1800495ed39SKyle Evans lua_Integer pi = luaL_optinteger(L, 2, 1);
1810495ed39SKyle Evans size_t posi = posrelatI(pi, l);
1820495ed39SKyle Evans size_t pose = getendpos(L, 3, pi, l);
1838e3e3a7aSWarner Losh int n, i;
1848e3e3a7aSWarner Losh if (posi > pose) return 0; /* empty interval; return no values */
1858c784bb8SWarner Losh if (l_unlikely(pose - posi >= (size_t)INT_MAX)) /* arithmetic overflow? */
1868e3e3a7aSWarner Losh return luaL_error(L, "string slice too long");
1878e3e3a7aSWarner Losh n = (int)(pose - posi) + 1;
1888e3e3a7aSWarner Losh luaL_checkstack(L, n, "string slice too long");
1898e3e3a7aSWarner Losh for (i=0; i<n; i++)
1908e3e3a7aSWarner Losh lua_pushinteger(L, uchar(s[posi+i-1]));
1918e3e3a7aSWarner Losh return n;
1928e3e3a7aSWarner Losh }
1938e3e3a7aSWarner Losh
1948e3e3a7aSWarner Losh
str_char(lua_State * L)1958e3e3a7aSWarner Losh static int str_char (lua_State *L) {
1968e3e3a7aSWarner Losh int n = lua_gettop(L); /* number of arguments */
1978e3e3a7aSWarner Losh int i;
1988e3e3a7aSWarner Losh luaL_Buffer b;
1998e3e3a7aSWarner Losh char *p = luaL_buffinitsize(L, &b, n);
2008e3e3a7aSWarner Losh for (i=1; i<=n; i++) {
2010495ed39SKyle Evans lua_Unsigned c = (lua_Unsigned)luaL_checkinteger(L, i);
2020495ed39SKyle Evans luaL_argcheck(L, c <= (lua_Unsigned)UCHAR_MAX, i, "value out of range");
2038e3e3a7aSWarner Losh p[i - 1] = uchar(c);
2048e3e3a7aSWarner Losh }
2058e3e3a7aSWarner Losh luaL_pushresultsize(&b, n);
2068e3e3a7aSWarner Losh return 1;
2078e3e3a7aSWarner Losh }
2088e3e3a7aSWarner Losh
2098e3e3a7aSWarner Losh
2100495ed39SKyle Evans /*
2110495ed39SKyle Evans ** Buffer to store the result of 'string.dump'. It must be initialized
2120495ed39SKyle Evans ** after the call to 'lua_dump', to ensure that the function is on the
2130495ed39SKyle Evans ** top of the stack when 'lua_dump' is called. ('luaL_buffinit' might
2140495ed39SKyle Evans ** push stuff.)
2150495ed39SKyle Evans */
2160495ed39SKyle Evans struct str_Writer {
2170495ed39SKyle Evans int init; /* true iff buffer has been initialized */
2180495ed39SKyle Evans luaL_Buffer B;
2190495ed39SKyle Evans };
2200495ed39SKyle Evans
2210495ed39SKyle Evans
writer(lua_State * L,const void * b,size_t size,void * ud)2220495ed39SKyle Evans static int writer (lua_State *L, const void *b, size_t size, void *ud) {
2230495ed39SKyle Evans struct str_Writer *state = (struct str_Writer *)ud;
2240495ed39SKyle Evans if (!state->init) {
2250495ed39SKyle Evans state->init = 1;
2260495ed39SKyle Evans luaL_buffinit(L, &state->B);
2270495ed39SKyle Evans }
2280495ed39SKyle Evans luaL_addlstring(&state->B, (const char *)b, size);
2298e3e3a7aSWarner Losh return 0;
2308e3e3a7aSWarner Losh }
2318e3e3a7aSWarner Losh
2328e3e3a7aSWarner Losh
str_dump(lua_State * L)2338e3e3a7aSWarner Losh static int str_dump (lua_State *L) {
2340495ed39SKyle Evans struct str_Writer state;
2358e3e3a7aSWarner Losh int strip = lua_toboolean(L, 2);
2368e3e3a7aSWarner Losh luaL_checktype(L, 1, LUA_TFUNCTION);
2370495ed39SKyle Evans lua_settop(L, 1); /* ensure function is on the top of the stack */
2380495ed39SKyle Evans state.init = 0;
2398c784bb8SWarner Losh if (l_unlikely(lua_dump(L, writer, &state, strip) != 0))
2408e3e3a7aSWarner Losh return luaL_error(L, "unable to dump given function");
2410495ed39SKyle Evans luaL_pushresult(&state.B);
2428e3e3a7aSWarner Losh return 1;
2438e3e3a7aSWarner Losh }
2448e3e3a7aSWarner Losh
2458e3e3a7aSWarner Losh
2468e3e3a7aSWarner Losh
2478e3e3a7aSWarner Losh /*
2488e3e3a7aSWarner Losh ** {======================================================
2490495ed39SKyle Evans ** METAMETHODS
2500495ed39SKyle Evans ** =======================================================
2510495ed39SKyle Evans */
2520495ed39SKyle Evans
2530495ed39SKyle Evans #if defined(LUA_NOCVTS2N) /* { */
2540495ed39SKyle Evans
2550495ed39SKyle Evans /* no coercion from strings to numbers */
2560495ed39SKyle Evans
2570495ed39SKyle Evans static const luaL_Reg stringmetamethods[] = {
2580495ed39SKyle Evans {"__index", NULL}, /* placeholder */
2590495ed39SKyle Evans {NULL, NULL}
2600495ed39SKyle Evans };
2610495ed39SKyle Evans
2620495ed39SKyle Evans #else /* }{ */
2630495ed39SKyle Evans
tonum(lua_State * L,int arg)2640495ed39SKyle Evans static int tonum (lua_State *L, int arg) {
2650495ed39SKyle Evans if (lua_type(L, arg) == LUA_TNUMBER) { /* already a number? */
2660495ed39SKyle Evans lua_pushvalue(L, arg);
2670495ed39SKyle Evans return 1;
2680495ed39SKyle Evans }
2690495ed39SKyle Evans else { /* check whether it is a numerical string */
2700495ed39SKyle Evans size_t len;
2710495ed39SKyle Evans const char *s = lua_tolstring(L, arg, &len);
2720495ed39SKyle Evans return (s != NULL && lua_stringtonumber(L, s) == len + 1);
2730495ed39SKyle Evans }
2740495ed39SKyle Evans }
2750495ed39SKyle Evans
2760495ed39SKyle Evans
trymt(lua_State * L,const char * mtname)2770495ed39SKyle Evans static void trymt (lua_State *L, const char *mtname) {
2780495ed39SKyle Evans lua_settop(L, 2); /* back to the original arguments */
2798c784bb8SWarner Losh if (l_unlikely(lua_type(L, 2) == LUA_TSTRING ||
2808c784bb8SWarner Losh !luaL_getmetafield(L, 2, mtname)))
2810495ed39SKyle Evans luaL_error(L, "attempt to %s a '%s' with a '%s'", mtname + 2,
2820495ed39SKyle Evans luaL_typename(L, -2), luaL_typename(L, -1));
2830495ed39SKyle Evans lua_insert(L, -3); /* put metamethod before arguments */
2840495ed39SKyle Evans lua_call(L, 2, 1); /* call metamethod */
2850495ed39SKyle Evans }
2860495ed39SKyle Evans
2870495ed39SKyle Evans
arith(lua_State * L,int op,const char * mtname)2880495ed39SKyle Evans static int arith (lua_State *L, int op, const char *mtname) {
2890495ed39SKyle Evans if (tonum(L, 1) && tonum(L, 2))
2900495ed39SKyle Evans lua_arith(L, op); /* result will be on the top */
2910495ed39SKyle Evans else
2920495ed39SKyle Evans trymt(L, mtname);
2930495ed39SKyle Evans return 1;
2940495ed39SKyle Evans }
2950495ed39SKyle Evans
2960495ed39SKyle Evans
arith_add(lua_State * L)2970495ed39SKyle Evans static int arith_add (lua_State *L) {
2980495ed39SKyle Evans return arith(L, LUA_OPADD, "__add");
2990495ed39SKyle Evans }
3000495ed39SKyle Evans
arith_sub(lua_State * L)3010495ed39SKyle Evans static int arith_sub (lua_State *L) {
3020495ed39SKyle Evans return arith(L, LUA_OPSUB, "__sub");
3030495ed39SKyle Evans }
3040495ed39SKyle Evans
arith_mul(lua_State * L)3050495ed39SKyle Evans static int arith_mul (lua_State *L) {
3060495ed39SKyle Evans return arith(L, LUA_OPMUL, "__mul");
3070495ed39SKyle Evans }
3080495ed39SKyle Evans
arith_mod(lua_State * L)3090495ed39SKyle Evans static int arith_mod (lua_State *L) {
3100495ed39SKyle Evans return arith(L, LUA_OPMOD, "__mod");
3110495ed39SKyle Evans }
3120495ed39SKyle Evans
arith_pow(lua_State * L)3130495ed39SKyle Evans static int arith_pow (lua_State *L) {
3140495ed39SKyle Evans return arith(L, LUA_OPPOW, "__pow");
3150495ed39SKyle Evans }
3160495ed39SKyle Evans
arith_div(lua_State * L)3170495ed39SKyle Evans static int arith_div (lua_State *L) {
3180495ed39SKyle Evans return arith(L, LUA_OPDIV, "__div");
3190495ed39SKyle Evans }
3200495ed39SKyle Evans
arith_idiv(lua_State * L)3210495ed39SKyle Evans static int arith_idiv (lua_State *L) {
3220495ed39SKyle Evans return arith(L, LUA_OPIDIV, "__idiv");
3230495ed39SKyle Evans }
3240495ed39SKyle Evans
arith_unm(lua_State * L)3250495ed39SKyle Evans static int arith_unm (lua_State *L) {
3260495ed39SKyle Evans return arith(L, LUA_OPUNM, "__unm");
3270495ed39SKyle Evans }
3280495ed39SKyle Evans
3290495ed39SKyle Evans
3300495ed39SKyle Evans static const luaL_Reg stringmetamethods[] = {
3310495ed39SKyle Evans {"__add", arith_add},
3320495ed39SKyle Evans {"__sub", arith_sub},
3330495ed39SKyle Evans {"__mul", arith_mul},
3340495ed39SKyle Evans {"__mod", arith_mod},
3350495ed39SKyle Evans {"__pow", arith_pow},
3360495ed39SKyle Evans {"__div", arith_div},
3370495ed39SKyle Evans {"__idiv", arith_idiv},
3380495ed39SKyle Evans {"__unm", arith_unm},
3390495ed39SKyle Evans {"__index", NULL}, /* placeholder */
3400495ed39SKyle Evans {NULL, NULL}
3410495ed39SKyle Evans };
3420495ed39SKyle Evans
3430495ed39SKyle Evans #endif /* } */
3440495ed39SKyle Evans
3450495ed39SKyle Evans /* }====================================================== */
3460495ed39SKyle Evans
3470495ed39SKyle Evans /*
3480495ed39SKyle Evans ** {======================================================
3498e3e3a7aSWarner Losh ** PATTERN MATCHING
3508e3e3a7aSWarner Losh ** =======================================================
3518e3e3a7aSWarner Losh */
3528e3e3a7aSWarner Losh
3538e3e3a7aSWarner Losh
3548e3e3a7aSWarner Losh #define CAP_UNFINISHED (-1)
3558e3e3a7aSWarner Losh #define CAP_POSITION (-2)
3568e3e3a7aSWarner Losh
3578e3e3a7aSWarner Losh
3588e3e3a7aSWarner Losh typedef struct MatchState {
3598e3e3a7aSWarner Losh const char *src_init; /* init of source string */
3608e3e3a7aSWarner Losh const char *src_end; /* end ('\0') of source string */
3618e3e3a7aSWarner Losh const char *p_end; /* end ('\0') of pattern */
3628e3e3a7aSWarner Losh lua_State *L;
3638e3e3a7aSWarner Losh int matchdepth; /* control for recursive depth (to avoid C stack overflow) */
3648e3e3a7aSWarner Losh unsigned char level; /* total number of captures (finished or unfinished) */
3658e3e3a7aSWarner Losh struct {
3668e3e3a7aSWarner Losh const char *init;
3678e3e3a7aSWarner Losh ptrdiff_t len;
3688e3e3a7aSWarner Losh } capture[LUA_MAXCAPTURES];
3698e3e3a7aSWarner Losh } MatchState;
3708e3e3a7aSWarner Losh
3718e3e3a7aSWarner Losh
3728e3e3a7aSWarner Losh /* recursive function */
3738e3e3a7aSWarner Losh static const char *match (MatchState *ms, const char *s, const char *p);
3748e3e3a7aSWarner Losh
3758e3e3a7aSWarner Losh
3768e3e3a7aSWarner Losh /* maximum recursion depth for 'match' */
3778e3e3a7aSWarner Losh #if !defined(MAXCCALLS)
3788e3e3a7aSWarner Losh #define MAXCCALLS 200
3798e3e3a7aSWarner Losh #endif
3808e3e3a7aSWarner Losh
3818e3e3a7aSWarner Losh
3828e3e3a7aSWarner Losh #define L_ESC '%'
3838e3e3a7aSWarner Losh #define SPECIALS "^$*+?.([%-"
3848e3e3a7aSWarner Losh
3858e3e3a7aSWarner Losh
check_capture(MatchState * ms,int l)3868e3e3a7aSWarner Losh static int check_capture (MatchState *ms, int l) {
3878e3e3a7aSWarner Losh l -= '1';
3888c784bb8SWarner Losh if (l_unlikely(l < 0 || l >= ms->level ||
3898c784bb8SWarner Losh ms->capture[l].len == CAP_UNFINISHED))
3908e3e3a7aSWarner Losh return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
3918e3e3a7aSWarner Losh return l;
3928e3e3a7aSWarner Losh }
3938e3e3a7aSWarner Losh
3948e3e3a7aSWarner Losh
capture_to_close(MatchState * ms)3958e3e3a7aSWarner Losh static int capture_to_close (MatchState *ms) {
3968e3e3a7aSWarner Losh int level = ms->level;
3978e3e3a7aSWarner Losh for (level--; level>=0; level--)
3988e3e3a7aSWarner Losh if (ms->capture[level].len == CAP_UNFINISHED) return level;
3998e3e3a7aSWarner Losh return luaL_error(ms->L, "invalid pattern capture");
4008e3e3a7aSWarner Losh }
4018e3e3a7aSWarner Losh
4028e3e3a7aSWarner Losh
classend(MatchState * ms,const char * p)4038e3e3a7aSWarner Losh static const char *classend (MatchState *ms, const char *p) {
4048e3e3a7aSWarner Losh switch (*p++) {
4058e3e3a7aSWarner Losh case L_ESC: {
4068c784bb8SWarner Losh if (l_unlikely(p == ms->p_end))
4078e3e3a7aSWarner Losh luaL_error(ms->L, "malformed pattern (ends with '%%')");
4088e3e3a7aSWarner Losh return p+1;
4098e3e3a7aSWarner Losh }
4108e3e3a7aSWarner Losh case '[': {
4118e3e3a7aSWarner Losh if (*p == '^') p++;
4128e3e3a7aSWarner Losh do { /* look for a ']' */
4138c784bb8SWarner Losh if (l_unlikely(p == ms->p_end))
4148e3e3a7aSWarner Losh luaL_error(ms->L, "malformed pattern (missing ']')");
4158e3e3a7aSWarner Losh if (*(p++) == L_ESC && p < ms->p_end)
4168e3e3a7aSWarner Losh p++; /* skip escapes (e.g. '%]') */
4178e3e3a7aSWarner Losh } while (*p != ']');
4188e3e3a7aSWarner Losh return p+1;
4198e3e3a7aSWarner Losh }
4208e3e3a7aSWarner Losh default: {
4218e3e3a7aSWarner Losh return p;
4228e3e3a7aSWarner Losh }
4238e3e3a7aSWarner Losh }
4248e3e3a7aSWarner Losh }
4258e3e3a7aSWarner Losh
4268e3e3a7aSWarner Losh
match_class(int c,int cl)4278e3e3a7aSWarner Losh static int match_class (int c, int cl) {
4288e3e3a7aSWarner Losh int res;
4298e3e3a7aSWarner Losh switch (tolower(cl)) {
4308e3e3a7aSWarner Losh case 'a' : res = isalpha(c); break;
4318e3e3a7aSWarner Losh case 'c' : res = iscntrl(c); break;
4328e3e3a7aSWarner Losh case 'd' : res = isdigit(c); break;
4338e3e3a7aSWarner Losh case 'g' : res = isgraph(c); break;
4348e3e3a7aSWarner Losh case 'l' : res = islower(c); break;
4358e3e3a7aSWarner Losh case 'p' : res = ispunct(c); break;
4368e3e3a7aSWarner Losh case 's' : res = isspace(c); break;
4378e3e3a7aSWarner Losh case 'u' : res = isupper(c); break;
4388e3e3a7aSWarner Losh case 'w' : res = isalnum(c); break;
4398e3e3a7aSWarner Losh case 'x' : res = isxdigit(c); break;
4408e3e3a7aSWarner Losh case 'z' : res = (c == 0); break; /* deprecated option */
4418e3e3a7aSWarner Losh default: return (cl == c);
4428e3e3a7aSWarner Losh }
4438e3e3a7aSWarner Losh return (islower(cl) ? res : !res);
4448e3e3a7aSWarner Losh }
4458e3e3a7aSWarner Losh
4468e3e3a7aSWarner Losh
matchbracketclass(int c,const char * p,const char * ec)4478e3e3a7aSWarner Losh static int matchbracketclass (int c, const char *p, const char *ec) {
4488e3e3a7aSWarner Losh int sig = 1;
4498e3e3a7aSWarner Losh if (*(p+1) == '^') {
4508e3e3a7aSWarner Losh sig = 0;
4518e3e3a7aSWarner Losh p++; /* skip the '^' */
4528e3e3a7aSWarner Losh }
4538e3e3a7aSWarner Losh while (++p < ec) {
4548e3e3a7aSWarner Losh if (*p == L_ESC) {
4558e3e3a7aSWarner Losh p++;
4568e3e3a7aSWarner Losh if (match_class(c, uchar(*p)))
4578e3e3a7aSWarner Losh return sig;
4588e3e3a7aSWarner Losh }
4598e3e3a7aSWarner Losh else if ((*(p+1) == '-') && (p+2 < ec)) {
4608e3e3a7aSWarner Losh p+=2;
4618e3e3a7aSWarner Losh if (uchar(*(p-2)) <= c && c <= uchar(*p))
4628e3e3a7aSWarner Losh return sig;
4638e3e3a7aSWarner Losh }
4648e3e3a7aSWarner Losh else if (uchar(*p) == c) return sig;
4658e3e3a7aSWarner Losh }
4668e3e3a7aSWarner Losh return !sig;
4678e3e3a7aSWarner Losh }
4688e3e3a7aSWarner Losh
4698e3e3a7aSWarner Losh
singlematch(MatchState * ms,const char * s,const char * p,const char * ep)4708e3e3a7aSWarner Losh static int singlematch (MatchState *ms, const char *s, const char *p,
4718e3e3a7aSWarner Losh const char *ep) {
4728e3e3a7aSWarner Losh if (s >= ms->src_end)
4738e3e3a7aSWarner Losh return 0;
4748e3e3a7aSWarner Losh else {
4758e3e3a7aSWarner Losh int c = uchar(*s);
4768e3e3a7aSWarner Losh switch (*p) {
4778e3e3a7aSWarner Losh case '.': return 1; /* matches any char */
4788e3e3a7aSWarner Losh case L_ESC: return match_class(c, uchar(*(p+1)));
4798e3e3a7aSWarner Losh case '[': return matchbracketclass(c, p, ep-1);
4808e3e3a7aSWarner Losh default: return (uchar(*p) == c);
4818e3e3a7aSWarner Losh }
4828e3e3a7aSWarner Losh }
4838e3e3a7aSWarner Losh }
4848e3e3a7aSWarner Losh
4858e3e3a7aSWarner Losh
matchbalance(MatchState * ms,const char * s,const char * p)4868e3e3a7aSWarner Losh static const char *matchbalance (MatchState *ms, const char *s,
4878e3e3a7aSWarner Losh const char *p) {
4888c784bb8SWarner Losh if (l_unlikely(p >= ms->p_end - 1))
4898e3e3a7aSWarner Losh luaL_error(ms->L, "malformed pattern (missing arguments to '%%b')");
4908e3e3a7aSWarner Losh if (*s != *p) return NULL;
4918e3e3a7aSWarner Losh else {
4928e3e3a7aSWarner Losh int b = *p;
4938e3e3a7aSWarner Losh int e = *(p+1);
4948e3e3a7aSWarner Losh int cont = 1;
4958e3e3a7aSWarner Losh while (++s < ms->src_end) {
4968e3e3a7aSWarner Losh if (*s == e) {
4978e3e3a7aSWarner Losh if (--cont == 0) return s+1;
4988e3e3a7aSWarner Losh }
4998e3e3a7aSWarner Losh else if (*s == b) cont++;
5008e3e3a7aSWarner Losh }
5018e3e3a7aSWarner Losh }
5028e3e3a7aSWarner Losh return NULL; /* string ends out of balance */
5038e3e3a7aSWarner Losh }
5048e3e3a7aSWarner Losh
5058e3e3a7aSWarner Losh
max_expand(MatchState * ms,const char * s,const char * p,const char * ep)5068e3e3a7aSWarner Losh static const char *max_expand (MatchState *ms, const char *s,
5078e3e3a7aSWarner Losh const char *p, const char *ep) {
5088e3e3a7aSWarner Losh ptrdiff_t i = 0; /* counts maximum expand for item */
5098e3e3a7aSWarner Losh while (singlematch(ms, s + i, p, ep))
5108e3e3a7aSWarner Losh i++;
5118e3e3a7aSWarner Losh /* keeps trying to match with the maximum repetitions */
5128e3e3a7aSWarner Losh while (i>=0) {
5138e3e3a7aSWarner Losh const char *res = match(ms, (s+i), ep+1);
5148e3e3a7aSWarner Losh if (res) return res;
5158e3e3a7aSWarner Losh i--; /* else didn't match; reduce 1 repetition to try again */
5168e3e3a7aSWarner Losh }
5178e3e3a7aSWarner Losh return NULL;
5188e3e3a7aSWarner Losh }
5198e3e3a7aSWarner Losh
5208e3e3a7aSWarner Losh
min_expand(MatchState * ms,const char * s,const char * p,const char * ep)5218e3e3a7aSWarner Losh static const char *min_expand (MatchState *ms, const char *s,
5228e3e3a7aSWarner Losh const char *p, const char *ep) {
5238e3e3a7aSWarner Losh for (;;) {
5248e3e3a7aSWarner Losh const char *res = match(ms, s, ep+1);
5258e3e3a7aSWarner Losh if (res != NULL)
5268e3e3a7aSWarner Losh return res;
5278e3e3a7aSWarner Losh else if (singlematch(ms, s, p, ep))
5288e3e3a7aSWarner Losh s++; /* try with one more repetition */
5298e3e3a7aSWarner Losh else return NULL;
5308e3e3a7aSWarner Losh }
5318e3e3a7aSWarner Losh }
5328e3e3a7aSWarner Losh
5338e3e3a7aSWarner Losh
start_capture(MatchState * ms,const char * s,const char * p,int what)5348e3e3a7aSWarner Losh static const char *start_capture (MatchState *ms, const char *s,
5358e3e3a7aSWarner Losh const char *p, int what) {
5368e3e3a7aSWarner Losh const char *res;
5378e3e3a7aSWarner Losh int level = ms->level;
5388e3e3a7aSWarner Losh if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
5398e3e3a7aSWarner Losh ms->capture[level].init = s;
5408e3e3a7aSWarner Losh ms->capture[level].len = what;
5418e3e3a7aSWarner Losh ms->level = level+1;
5428e3e3a7aSWarner Losh if ((res=match(ms, s, p)) == NULL) /* match failed? */
5438e3e3a7aSWarner Losh ms->level--; /* undo capture */
5448e3e3a7aSWarner Losh return res;
5458e3e3a7aSWarner Losh }
5468e3e3a7aSWarner Losh
5478e3e3a7aSWarner Losh
end_capture(MatchState * ms,const char * s,const char * p)5488e3e3a7aSWarner Losh static const char *end_capture (MatchState *ms, const char *s,
5498e3e3a7aSWarner Losh const char *p) {
5508e3e3a7aSWarner Losh int l = capture_to_close(ms);
5518e3e3a7aSWarner Losh const char *res;
5528e3e3a7aSWarner Losh ms->capture[l].len = s - ms->capture[l].init; /* close capture */
5538e3e3a7aSWarner Losh if ((res = match(ms, s, p)) == NULL) /* match failed? */
5548e3e3a7aSWarner Losh ms->capture[l].len = CAP_UNFINISHED; /* undo capture */
5558e3e3a7aSWarner Losh return res;
5568e3e3a7aSWarner Losh }
5578e3e3a7aSWarner Losh
5588e3e3a7aSWarner Losh
match_capture(MatchState * ms,const char * s,int l)5598e3e3a7aSWarner Losh static const char *match_capture (MatchState *ms, const char *s, int l) {
5608e3e3a7aSWarner Losh size_t len;
5618e3e3a7aSWarner Losh l = check_capture(ms, l);
5628e3e3a7aSWarner Losh len = ms->capture[l].len;
5638e3e3a7aSWarner Losh if ((size_t)(ms->src_end-s) >= len &&
5648e3e3a7aSWarner Losh memcmp(ms->capture[l].init, s, len) == 0)
5658e3e3a7aSWarner Losh return s+len;
5668e3e3a7aSWarner Losh else return NULL;
5678e3e3a7aSWarner Losh }
5688e3e3a7aSWarner Losh
5698e3e3a7aSWarner Losh
match(MatchState * ms,const char * s,const char * p)5708e3e3a7aSWarner Losh static const char *match (MatchState *ms, const char *s, const char *p) {
5718c784bb8SWarner Losh if (l_unlikely(ms->matchdepth-- == 0))
5728e3e3a7aSWarner Losh luaL_error(ms->L, "pattern too complex");
573*a9490b81SWarner Losh init: /* using goto to optimize tail recursion */
5748e3e3a7aSWarner Losh if (p != ms->p_end) { /* end of pattern? */
5758e3e3a7aSWarner Losh switch (*p) {
5768e3e3a7aSWarner Losh case '(': { /* start capture */
5778e3e3a7aSWarner Losh if (*(p + 1) == ')') /* position capture? */
5788e3e3a7aSWarner Losh s = start_capture(ms, s, p + 2, CAP_POSITION);
5798e3e3a7aSWarner Losh else
5808e3e3a7aSWarner Losh s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
5818e3e3a7aSWarner Losh break;
5828e3e3a7aSWarner Losh }
5838e3e3a7aSWarner Losh case ')': { /* end capture */
5848e3e3a7aSWarner Losh s = end_capture(ms, s, p + 1);
5858e3e3a7aSWarner Losh break;
5868e3e3a7aSWarner Losh }
5878e3e3a7aSWarner Losh case '$': {
5888e3e3a7aSWarner Losh if ((p + 1) != ms->p_end) /* is the '$' the last char in pattern? */
5898e3e3a7aSWarner Losh goto dflt; /* no; go to default */
5908e3e3a7aSWarner Losh s = (s == ms->src_end) ? s : NULL; /* check end of string */
5918e3e3a7aSWarner Losh break;
5928e3e3a7aSWarner Losh }
5938e3e3a7aSWarner Losh case L_ESC: { /* escaped sequences not in the format class[*+?-]? */
5948e3e3a7aSWarner Losh switch (*(p + 1)) {
5958e3e3a7aSWarner Losh case 'b': { /* balanced string? */
5968e3e3a7aSWarner Losh s = matchbalance(ms, s, p + 2);
5978e3e3a7aSWarner Losh if (s != NULL) {
5988e3e3a7aSWarner Losh p += 4; goto init; /* return match(ms, s, p + 4); */
5998e3e3a7aSWarner Losh } /* else fail (s == NULL) */
6008e3e3a7aSWarner Losh break;
6018e3e3a7aSWarner Losh }
6028e3e3a7aSWarner Losh case 'f': { /* frontier? */
6038e3e3a7aSWarner Losh const char *ep; char previous;
6048e3e3a7aSWarner Losh p += 2;
6058c784bb8SWarner Losh if (l_unlikely(*p != '['))
6068e3e3a7aSWarner Losh luaL_error(ms->L, "missing '[' after '%%f' in pattern");
6078e3e3a7aSWarner Losh ep = classend(ms, p); /* points to what is next */
6088e3e3a7aSWarner Losh previous = (s == ms->src_init) ? '\0' : *(s - 1);
6098e3e3a7aSWarner Losh if (!matchbracketclass(uchar(previous), p, ep - 1) &&
6108e3e3a7aSWarner Losh matchbracketclass(uchar(*s), p, ep - 1)) {
6118e3e3a7aSWarner Losh p = ep; goto init; /* return match(ms, s, ep); */
6128e3e3a7aSWarner Losh }
6138e3e3a7aSWarner Losh s = NULL; /* match failed */
6148e3e3a7aSWarner Losh break;
6158e3e3a7aSWarner Losh }
6168e3e3a7aSWarner Losh case '0': case '1': case '2': case '3':
6178e3e3a7aSWarner Losh case '4': case '5': case '6': case '7':
6188e3e3a7aSWarner Losh case '8': case '9': { /* capture results (%0-%9)? */
6198e3e3a7aSWarner Losh s = match_capture(ms, s, uchar(*(p + 1)));
6208e3e3a7aSWarner Losh if (s != NULL) {
6218e3e3a7aSWarner Losh p += 2; goto init; /* return match(ms, s, p + 2) */
6228e3e3a7aSWarner Losh }
6238e3e3a7aSWarner Losh break;
6248e3e3a7aSWarner Losh }
6258e3e3a7aSWarner Losh default: goto dflt;
6268e3e3a7aSWarner Losh }
6278e3e3a7aSWarner Losh break;
6288e3e3a7aSWarner Losh }
6298e3e3a7aSWarner Losh default: dflt: { /* pattern class plus optional suffix */
6308e3e3a7aSWarner Losh const char *ep = classend(ms, p); /* points to optional suffix */
6318e3e3a7aSWarner Losh /* does not match at least once? */
6328e3e3a7aSWarner Losh if (!singlematch(ms, s, p, ep)) {
6338e3e3a7aSWarner Losh if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */
6348e3e3a7aSWarner Losh p = ep + 1; goto init; /* return match(ms, s, ep + 1); */
6358e3e3a7aSWarner Losh }
6368e3e3a7aSWarner Losh else /* '+' or no suffix */
6378e3e3a7aSWarner Losh s = NULL; /* fail */
6388e3e3a7aSWarner Losh }
6398e3e3a7aSWarner Losh else { /* matched once */
6408e3e3a7aSWarner Losh switch (*ep) { /* handle optional suffix */
6418e3e3a7aSWarner Losh case '?': { /* optional */
6428e3e3a7aSWarner Losh const char *res;
6438e3e3a7aSWarner Losh if ((res = match(ms, s + 1, ep + 1)) != NULL)
6448e3e3a7aSWarner Losh s = res;
6458e3e3a7aSWarner Losh else {
6468e3e3a7aSWarner Losh p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */
6478e3e3a7aSWarner Losh }
6488e3e3a7aSWarner Losh break;
6498e3e3a7aSWarner Losh }
6508e3e3a7aSWarner Losh case '+': /* 1 or more repetitions */
6518e3e3a7aSWarner Losh s++; /* 1 match already done */
6528e3e3a7aSWarner Losh /* FALLTHROUGH */
6538e3e3a7aSWarner Losh case '*': /* 0 or more repetitions */
6548e3e3a7aSWarner Losh s = max_expand(ms, s, p, ep);
6558e3e3a7aSWarner Losh break;
6568e3e3a7aSWarner Losh case '-': /* 0 or more repetitions (minimum) */
6578e3e3a7aSWarner Losh s = min_expand(ms, s, p, ep);
6588e3e3a7aSWarner Losh break;
6598e3e3a7aSWarner Losh default: /* no suffix */
6608e3e3a7aSWarner Losh s++; p = ep; goto init; /* return match(ms, s + 1, ep); */
6618e3e3a7aSWarner Losh }
6628e3e3a7aSWarner Losh }
6638e3e3a7aSWarner Losh break;
6648e3e3a7aSWarner Losh }
6658e3e3a7aSWarner Losh }
6668e3e3a7aSWarner Losh }
6678e3e3a7aSWarner Losh ms->matchdepth++;
6688e3e3a7aSWarner Losh return s;
6698e3e3a7aSWarner Losh }
6708e3e3a7aSWarner Losh
6718e3e3a7aSWarner Losh
6728e3e3a7aSWarner Losh
lmemfind(const char * s1,size_t l1,const char * s2,size_t l2)6738e3e3a7aSWarner Losh static const char *lmemfind (const char *s1, size_t l1,
6748e3e3a7aSWarner Losh const char *s2, size_t l2) {
6758e3e3a7aSWarner Losh if (l2 == 0) return s1; /* empty strings are everywhere */
6768e3e3a7aSWarner Losh else if (l2 > l1) return NULL; /* avoids a negative 'l1' */
6778e3e3a7aSWarner Losh else {
6788e3e3a7aSWarner Losh const char *init; /* to search for a '*s2' inside 's1' */
6798e3e3a7aSWarner Losh l2--; /* 1st char will be checked by 'memchr' */
6808e3e3a7aSWarner Losh l1 = l1-l2; /* 's2' cannot be found after that */
6818e3e3a7aSWarner Losh while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
6828e3e3a7aSWarner Losh init++; /* 1st char is already checked */
6838e3e3a7aSWarner Losh if (memcmp(init, s2+1, l2) == 0)
6848e3e3a7aSWarner Losh return init-1;
6858e3e3a7aSWarner Losh else { /* correct 'l1' and 's1' to try again */
6868e3e3a7aSWarner Losh l1 -= init-s1;
6878e3e3a7aSWarner Losh s1 = init;
6888e3e3a7aSWarner Losh }
6898e3e3a7aSWarner Losh }
6908e3e3a7aSWarner Losh return NULL; /* not found */
6918e3e3a7aSWarner Losh }
6928e3e3a7aSWarner Losh }
6938e3e3a7aSWarner Losh
6948e3e3a7aSWarner Losh
6950495ed39SKyle Evans /*
6960495ed39SKyle Evans ** get information about the i-th capture. If there are no captures
6970495ed39SKyle Evans ** and 'i==0', return information about the whole match, which
6980495ed39SKyle Evans ** is the range 's'..'e'. If the capture is a string, return
6990495ed39SKyle Evans ** its length and put its address in '*cap'. If it is an integer
7000495ed39SKyle Evans ** (a position), push it on the stack and return CAP_POSITION.
7010495ed39SKyle Evans */
get_onecapture(MatchState * ms,int i,const char * s,const char * e,const char ** cap)7020495ed39SKyle Evans static size_t get_onecapture (MatchState *ms, int i, const char *s,
7030495ed39SKyle Evans const char *e, const char **cap) {
7048e3e3a7aSWarner Losh if (i >= ms->level) {
7058c784bb8SWarner Losh if (l_unlikely(i != 0))
7068e3e3a7aSWarner Losh luaL_error(ms->L, "invalid capture index %%%d", i + 1);
7070495ed39SKyle Evans *cap = s;
7080495ed39SKyle Evans return e - s;
7098e3e3a7aSWarner Losh }
7108e3e3a7aSWarner Losh else {
7110495ed39SKyle Evans ptrdiff_t capl = ms->capture[i].len;
7120495ed39SKyle Evans *cap = ms->capture[i].init;
7138c784bb8SWarner Losh if (l_unlikely(capl == CAP_UNFINISHED))
7140495ed39SKyle Evans luaL_error(ms->L, "unfinished capture");
7150495ed39SKyle Evans else if (capl == CAP_POSITION)
7168e3e3a7aSWarner Losh lua_pushinteger(ms->L, (ms->capture[i].init - ms->src_init) + 1);
7170495ed39SKyle Evans return capl;
7188e3e3a7aSWarner Losh }
7198e3e3a7aSWarner Losh }
7208e3e3a7aSWarner Losh
7218e3e3a7aSWarner Losh
7220495ed39SKyle Evans /*
7230495ed39SKyle Evans ** Push the i-th capture on the stack.
7240495ed39SKyle Evans */
push_onecapture(MatchState * ms,int i,const char * s,const char * e)7250495ed39SKyle Evans static void push_onecapture (MatchState *ms, int i, const char *s,
7260495ed39SKyle Evans const char *e) {
7270495ed39SKyle Evans const char *cap;
7280495ed39SKyle Evans ptrdiff_t l = get_onecapture(ms, i, s, e, &cap);
7290495ed39SKyle Evans if (l != CAP_POSITION)
7300495ed39SKyle Evans lua_pushlstring(ms->L, cap, l);
7310495ed39SKyle Evans /* else position was already pushed */
7320495ed39SKyle Evans }
7330495ed39SKyle Evans
7340495ed39SKyle Evans
push_captures(MatchState * ms,const char * s,const char * e)7358e3e3a7aSWarner Losh static int push_captures (MatchState *ms, const char *s, const char *e) {
7368e3e3a7aSWarner Losh int i;
7378e3e3a7aSWarner Losh int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
7388e3e3a7aSWarner Losh luaL_checkstack(ms->L, nlevels, "too many captures");
7398e3e3a7aSWarner Losh for (i = 0; i < nlevels; i++)
7408e3e3a7aSWarner Losh push_onecapture(ms, i, s, e);
7418e3e3a7aSWarner Losh return nlevels; /* number of strings pushed */
7428e3e3a7aSWarner Losh }
7438e3e3a7aSWarner Losh
7448e3e3a7aSWarner Losh
7458e3e3a7aSWarner Losh /* check whether pattern has no special characters */
nospecials(const char * p,size_t l)7468e3e3a7aSWarner Losh static int nospecials (const char *p, size_t l) {
7478e3e3a7aSWarner Losh size_t upto = 0;
7488e3e3a7aSWarner Losh do {
7498e3e3a7aSWarner Losh if (strpbrk(p + upto, SPECIALS))
7508e3e3a7aSWarner Losh return 0; /* pattern has a special character */
7518e3e3a7aSWarner Losh upto += strlen(p + upto) + 1; /* may have more after \0 */
7528e3e3a7aSWarner Losh } while (upto <= l);
7538e3e3a7aSWarner Losh return 1; /* no special chars found */
7548e3e3a7aSWarner Losh }
7558e3e3a7aSWarner Losh
7568e3e3a7aSWarner Losh
prepstate(MatchState * ms,lua_State * L,const char * s,size_t ls,const char * p,size_t lp)7578e3e3a7aSWarner Losh static void prepstate (MatchState *ms, lua_State *L,
7588e3e3a7aSWarner Losh const char *s, size_t ls, const char *p, size_t lp) {
7598e3e3a7aSWarner Losh ms->L = L;
7608e3e3a7aSWarner Losh ms->matchdepth = MAXCCALLS;
7618e3e3a7aSWarner Losh ms->src_init = s;
7628e3e3a7aSWarner Losh ms->src_end = s + ls;
7638e3e3a7aSWarner Losh ms->p_end = p + lp;
7648e3e3a7aSWarner Losh }
7658e3e3a7aSWarner Losh
7668e3e3a7aSWarner Losh
reprepstate(MatchState * ms)7678e3e3a7aSWarner Losh static void reprepstate (MatchState *ms) {
7688e3e3a7aSWarner Losh ms->level = 0;
7698e3e3a7aSWarner Losh lua_assert(ms->matchdepth == MAXCCALLS);
7708e3e3a7aSWarner Losh }
7718e3e3a7aSWarner Losh
7728e3e3a7aSWarner Losh
str_find_aux(lua_State * L,int find)7738e3e3a7aSWarner Losh static int str_find_aux (lua_State *L, int find) {
7748e3e3a7aSWarner Losh size_t ls, lp;
7758e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &ls);
7768e3e3a7aSWarner Losh const char *p = luaL_checklstring(L, 2, &lp);
7770495ed39SKyle Evans size_t init = posrelatI(luaL_optinteger(L, 3, 1), ls) - 1;
7780495ed39SKyle Evans if (init > ls) { /* start after string's end? */
7790495ed39SKyle Evans luaL_pushfail(L); /* cannot find anything */
7808e3e3a7aSWarner Losh return 1;
7818e3e3a7aSWarner Losh }
7828e3e3a7aSWarner Losh /* explicit request or no special characters? */
7838e3e3a7aSWarner Losh if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
7848e3e3a7aSWarner Losh /* do a plain search */
7850495ed39SKyle Evans const char *s2 = lmemfind(s + init, ls - init, p, lp);
7868e3e3a7aSWarner Losh if (s2) {
7878e3e3a7aSWarner Losh lua_pushinteger(L, (s2 - s) + 1);
7888e3e3a7aSWarner Losh lua_pushinteger(L, (s2 - s) + lp);
7898e3e3a7aSWarner Losh return 2;
7908e3e3a7aSWarner Losh }
7918e3e3a7aSWarner Losh }
7928e3e3a7aSWarner Losh else {
7938e3e3a7aSWarner Losh MatchState ms;
7940495ed39SKyle Evans const char *s1 = s + init;
7958e3e3a7aSWarner Losh int anchor = (*p == '^');
7968e3e3a7aSWarner Losh if (anchor) {
7978e3e3a7aSWarner Losh p++; lp--; /* skip anchor character */
7988e3e3a7aSWarner Losh }
7998e3e3a7aSWarner Losh prepstate(&ms, L, s, ls, p, lp);
8008e3e3a7aSWarner Losh do {
8018e3e3a7aSWarner Losh const char *res;
8028e3e3a7aSWarner Losh reprepstate(&ms);
8038e3e3a7aSWarner Losh if ((res=match(&ms, s1, p)) != NULL) {
8048e3e3a7aSWarner Losh if (find) {
8058e3e3a7aSWarner Losh lua_pushinteger(L, (s1 - s) + 1); /* start */
8068e3e3a7aSWarner Losh lua_pushinteger(L, res - s); /* end */
8078e3e3a7aSWarner Losh return push_captures(&ms, NULL, 0) + 2;
8088e3e3a7aSWarner Losh }
8098e3e3a7aSWarner Losh else
8108e3e3a7aSWarner Losh return push_captures(&ms, s1, res);
8118e3e3a7aSWarner Losh }
8128e3e3a7aSWarner Losh } while (s1++ < ms.src_end && !anchor);
8138e3e3a7aSWarner Losh }
8140495ed39SKyle Evans luaL_pushfail(L); /* not found */
8158e3e3a7aSWarner Losh return 1;
8168e3e3a7aSWarner Losh }
8178e3e3a7aSWarner Losh
8188e3e3a7aSWarner Losh
str_find(lua_State * L)8198e3e3a7aSWarner Losh static int str_find (lua_State *L) {
8208e3e3a7aSWarner Losh return str_find_aux(L, 1);
8218e3e3a7aSWarner Losh }
8228e3e3a7aSWarner Losh
8238e3e3a7aSWarner Losh
str_match(lua_State * L)8248e3e3a7aSWarner Losh static int str_match (lua_State *L) {
8258e3e3a7aSWarner Losh return str_find_aux(L, 0);
8268e3e3a7aSWarner Losh }
8278e3e3a7aSWarner Losh
8288e3e3a7aSWarner Losh
8298e3e3a7aSWarner Losh /* state for 'gmatch' */
8308e3e3a7aSWarner Losh typedef struct GMatchState {
8318e3e3a7aSWarner Losh const char *src; /* current position */
8328e3e3a7aSWarner Losh const char *p; /* pattern */
8338e3e3a7aSWarner Losh const char *lastmatch; /* end of last match */
8348e3e3a7aSWarner Losh MatchState ms; /* match state */
8358e3e3a7aSWarner Losh } GMatchState;
8368e3e3a7aSWarner Losh
8378e3e3a7aSWarner Losh
gmatch_aux(lua_State * L)8388e3e3a7aSWarner Losh static int gmatch_aux (lua_State *L) {
8398e3e3a7aSWarner Losh GMatchState *gm = (GMatchState *)lua_touserdata(L, lua_upvalueindex(3));
8408e3e3a7aSWarner Losh const char *src;
8418e3e3a7aSWarner Losh gm->ms.L = L;
8428e3e3a7aSWarner Losh for (src = gm->src; src <= gm->ms.src_end; src++) {
8438e3e3a7aSWarner Losh const char *e;
8448e3e3a7aSWarner Losh reprepstate(&gm->ms);
8458e3e3a7aSWarner Losh if ((e = match(&gm->ms, src, gm->p)) != NULL && e != gm->lastmatch) {
8468e3e3a7aSWarner Losh gm->src = gm->lastmatch = e;
8478e3e3a7aSWarner Losh return push_captures(&gm->ms, src, e);
8488e3e3a7aSWarner Losh }
8498e3e3a7aSWarner Losh }
8508e3e3a7aSWarner Losh return 0; /* not found */
8518e3e3a7aSWarner Losh }
8528e3e3a7aSWarner Losh
8538e3e3a7aSWarner Losh
gmatch(lua_State * L)8548e3e3a7aSWarner Losh static int gmatch (lua_State *L) {
8558e3e3a7aSWarner Losh size_t ls, lp;
8568e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, 1, &ls);
8578e3e3a7aSWarner Losh const char *p = luaL_checklstring(L, 2, &lp);
8580495ed39SKyle Evans size_t init = posrelatI(luaL_optinteger(L, 3, 1), ls) - 1;
8598e3e3a7aSWarner Losh GMatchState *gm;
8600495ed39SKyle Evans lua_settop(L, 2); /* keep strings on closure to avoid being collected */
8610495ed39SKyle Evans gm = (GMatchState *)lua_newuserdatauv(L, sizeof(GMatchState), 0);
8620495ed39SKyle Evans if (init > ls) /* start after string's end? */
8630495ed39SKyle Evans init = ls + 1; /* avoid overflows in 's + init' */
8648e3e3a7aSWarner Losh prepstate(&gm->ms, L, s, ls, p, lp);
8650495ed39SKyle Evans gm->src = s + init; gm->p = p; gm->lastmatch = NULL;
8668e3e3a7aSWarner Losh lua_pushcclosure(L, gmatch_aux, 3);
8678e3e3a7aSWarner Losh return 1;
8688e3e3a7aSWarner Losh }
8698e3e3a7aSWarner Losh
8708e3e3a7aSWarner Losh
add_s(MatchState * ms,luaL_Buffer * b,const char * s,const char * e)8718e3e3a7aSWarner Losh static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
8728e3e3a7aSWarner Losh const char *e) {
8730495ed39SKyle Evans size_t l;
8748e3e3a7aSWarner Losh lua_State *L = ms->L;
8758e3e3a7aSWarner Losh const char *news = lua_tolstring(L, 3, &l);
8760495ed39SKyle Evans const char *p;
8770495ed39SKyle Evans while ((p = (char *)memchr(news, L_ESC, l)) != NULL) {
8780495ed39SKyle Evans luaL_addlstring(b, news, p - news);
8790495ed39SKyle Evans p++; /* skip ESC */
8800495ed39SKyle Evans if (*p == L_ESC) /* '%%' */
8810495ed39SKyle Evans luaL_addchar(b, *p);
8820495ed39SKyle Evans else if (*p == '0') /* '%0' */
8838e3e3a7aSWarner Losh luaL_addlstring(b, s, e - s);
8840495ed39SKyle Evans else if (isdigit(uchar(*p))) { /* '%n' */
8850495ed39SKyle Evans const char *cap;
8860495ed39SKyle Evans ptrdiff_t resl = get_onecapture(ms, *p - '1', s, e, &cap);
8870495ed39SKyle Evans if (resl == CAP_POSITION)
8880495ed39SKyle Evans luaL_addvalue(b); /* add position to accumulated result */
8890495ed39SKyle Evans else
8900495ed39SKyle Evans luaL_addlstring(b, cap, resl);
8918e3e3a7aSWarner Losh }
8920495ed39SKyle Evans else
8930495ed39SKyle Evans luaL_error(L, "invalid use of '%c' in replacement string", L_ESC);
8940495ed39SKyle Evans l -= p + 1 - news;
8950495ed39SKyle Evans news = p + 1;
8968e3e3a7aSWarner Losh }
8970495ed39SKyle Evans luaL_addlstring(b, news, l);
8988e3e3a7aSWarner Losh }
8998e3e3a7aSWarner Losh
9008e3e3a7aSWarner Losh
9010495ed39SKyle Evans /*
9020495ed39SKyle Evans ** Add the replacement value to the string buffer 'b'.
9030495ed39SKyle Evans ** Return true if the original string was changed. (Function calls and
9040495ed39SKyle Evans ** table indexing resulting in nil or false do not change the subject.)
9050495ed39SKyle Evans */
add_value(MatchState * ms,luaL_Buffer * b,const char * s,const char * e,int tr)9060495ed39SKyle Evans static int add_value (MatchState *ms, luaL_Buffer *b, const char *s,
9078e3e3a7aSWarner Losh const char *e, int tr) {
9088e3e3a7aSWarner Losh lua_State *L = ms->L;
9098e3e3a7aSWarner Losh switch (tr) {
9100495ed39SKyle Evans case LUA_TFUNCTION: { /* call the function */
9118e3e3a7aSWarner Losh int n;
9120495ed39SKyle Evans lua_pushvalue(L, 3); /* push the function */
9130495ed39SKyle Evans n = push_captures(ms, s, e); /* all captures as arguments */
9140495ed39SKyle Evans lua_call(L, n, 1); /* call it */
9158e3e3a7aSWarner Losh break;
9168e3e3a7aSWarner Losh }
9170495ed39SKyle Evans case LUA_TTABLE: { /* index the table */
9180495ed39SKyle Evans push_onecapture(ms, 0, s, e); /* first capture is the index */
9198e3e3a7aSWarner Losh lua_gettable(L, 3);
9208e3e3a7aSWarner Losh break;
9218e3e3a7aSWarner Losh }
9228e3e3a7aSWarner Losh default: { /* LUA_TNUMBER or LUA_TSTRING */
9230495ed39SKyle Evans add_s(ms, b, s, e); /* add value to the buffer */
9240495ed39SKyle Evans return 1; /* something changed */
9258e3e3a7aSWarner Losh }
9268e3e3a7aSWarner Losh }
9278e3e3a7aSWarner Losh if (!lua_toboolean(L, -1)) { /* nil or false? */
9280495ed39SKyle Evans lua_pop(L, 1); /* remove value */
9290495ed39SKyle Evans luaL_addlstring(b, s, e - s); /* keep original text */
9300495ed39SKyle Evans return 0; /* no changes */
9318e3e3a7aSWarner Losh }
9328c784bb8SWarner Losh else if (l_unlikely(!lua_isstring(L, -1)))
9330495ed39SKyle Evans return luaL_error(L, "invalid replacement value (a %s)",
9340495ed39SKyle Evans luaL_typename(L, -1));
9350495ed39SKyle Evans else {
9368e3e3a7aSWarner Losh luaL_addvalue(b); /* add result to accumulator */
9370495ed39SKyle Evans return 1; /* something changed */
9380495ed39SKyle Evans }
9398e3e3a7aSWarner Losh }
9408e3e3a7aSWarner Losh
9418e3e3a7aSWarner Losh
str_gsub(lua_State * L)9428e3e3a7aSWarner Losh static int str_gsub (lua_State *L) {
9438e3e3a7aSWarner Losh size_t srcl, lp;
9448e3e3a7aSWarner Losh const char *src = luaL_checklstring(L, 1, &srcl); /* subject */
9458e3e3a7aSWarner Losh const char *p = luaL_checklstring(L, 2, &lp); /* pattern */
9468e3e3a7aSWarner Losh const char *lastmatch = NULL; /* end of last match */
9478e3e3a7aSWarner Losh int tr = lua_type(L, 3); /* replacement type */
9488e3e3a7aSWarner Losh lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1); /* max replacements */
9498e3e3a7aSWarner Losh int anchor = (*p == '^');
9508e3e3a7aSWarner Losh lua_Integer n = 0; /* replacement count */
9510495ed39SKyle Evans int changed = 0; /* change flag */
9528e3e3a7aSWarner Losh MatchState ms;
9538e3e3a7aSWarner Losh luaL_Buffer b;
9540495ed39SKyle Evans luaL_argexpected(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
9558e3e3a7aSWarner Losh tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
9560495ed39SKyle Evans "string/function/table");
9578e3e3a7aSWarner Losh luaL_buffinit(L, &b);
9588e3e3a7aSWarner Losh if (anchor) {
9598e3e3a7aSWarner Losh p++; lp--; /* skip anchor character */
9608e3e3a7aSWarner Losh }
9618e3e3a7aSWarner Losh prepstate(&ms, L, src, srcl, p, lp);
9628e3e3a7aSWarner Losh while (n < max_s) {
9638e3e3a7aSWarner Losh const char *e;
9648e3e3a7aSWarner Losh reprepstate(&ms); /* (re)prepare state for new match */
9658e3e3a7aSWarner Losh if ((e = match(&ms, src, p)) != NULL && e != lastmatch) { /* match? */
9668e3e3a7aSWarner Losh n++;
9670495ed39SKyle Evans changed = add_value(&ms, &b, src, e, tr) | changed;
9688e3e3a7aSWarner Losh src = lastmatch = e;
9698e3e3a7aSWarner Losh }
9708e3e3a7aSWarner Losh else if (src < ms.src_end) /* otherwise, skip one character */
9718e3e3a7aSWarner Losh luaL_addchar(&b, *src++);
9728e3e3a7aSWarner Losh else break; /* end of subject */
9738e3e3a7aSWarner Losh if (anchor) break;
9748e3e3a7aSWarner Losh }
9750495ed39SKyle Evans if (!changed) /* no changes? */
9760495ed39SKyle Evans lua_pushvalue(L, 1); /* return original string */
9770495ed39SKyle Evans else { /* something changed */
9788e3e3a7aSWarner Losh luaL_addlstring(&b, src, ms.src_end-src);
9790495ed39SKyle Evans luaL_pushresult(&b); /* create and return new string */
9800495ed39SKyle Evans }
9818e3e3a7aSWarner Losh lua_pushinteger(L, n); /* number of substitutions */
9828e3e3a7aSWarner Losh return 2;
9838e3e3a7aSWarner Losh }
9848e3e3a7aSWarner Losh
9858e3e3a7aSWarner Losh /* }====================================================== */
9868e3e3a7aSWarner Losh
9878e3e3a7aSWarner Losh
9888e3e3a7aSWarner Losh
9898e3e3a7aSWarner Losh /*
9908e3e3a7aSWarner Losh ** {======================================================
9918e3e3a7aSWarner Losh ** STRING FORMAT
9928e3e3a7aSWarner Losh ** =======================================================
9938e3e3a7aSWarner Losh */
9948e3e3a7aSWarner Losh
9958e3e3a7aSWarner Losh #if !defined(lua_number2strx) /* { */
9968e3e3a7aSWarner Losh
9978e3e3a7aSWarner Losh /*
9988e3e3a7aSWarner Losh ** Hexadecimal floating-point formatter
9998e3e3a7aSWarner Losh */
10008e3e3a7aSWarner Losh
10018e3e3a7aSWarner Losh #define SIZELENMOD (sizeof(LUA_NUMBER_FRMLEN)/sizeof(char))
10028e3e3a7aSWarner Losh
10038e3e3a7aSWarner Losh
10048e3e3a7aSWarner Losh /*
10058e3e3a7aSWarner Losh ** Number of bits that goes into the first digit. It can be any value
10068e3e3a7aSWarner Losh ** between 1 and 4; the following definition tries to align the number
10078e3e3a7aSWarner Losh ** to nibble boundaries by making what is left after that first digit a
10088e3e3a7aSWarner Losh ** multiple of 4.
10098e3e3a7aSWarner Losh */
10100495ed39SKyle Evans #define L_NBFD ((l_floatatt(MANT_DIG) - 1)%4 + 1)
10118e3e3a7aSWarner Losh
10128e3e3a7aSWarner Losh
10138e3e3a7aSWarner Losh /*
10148e3e3a7aSWarner Losh ** Add integer part of 'x' to buffer and return new 'x'
10158e3e3a7aSWarner Losh */
adddigit(char * buff,int n,lua_Number x)10168e3e3a7aSWarner Losh static lua_Number adddigit (char *buff, int n, lua_Number x) {
10178e3e3a7aSWarner Losh lua_Number dd = l_mathop(floor)(x); /* get integer part from 'x' */
10188e3e3a7aSWarner Losh int d = (int)dd;
10198e3e3a7aSWarner Losh buff[n] = (d < 10 ? d + '0' : d - 10 + 'a'); /* add to buffer */
10208e3e3a7aSWarner Losh return x - dd; /* return what is left */
10218e3e3a7aSWarner Losh }
10228e3e3a7aSWarner Losh
10238e3e3a7aSWarner Losh
num2straux(char * buff,int sz,lua_Number x)10248e3e3a7aSWarner Losh static int num2straux (char *buff, int sz, lua_Number x) {
10258e3e3a7aSWarner Losh /* if 'inf' or 'NaN', format it like '%g' */
10268e3e3a7aSWarner Losh if (x != x || x == (lua_Number)HUGE_VAL || x == -(lua_Number)HUGE_VAL)
10278e3e3a7aSWarner Losh return l_sprintf(buff, sz, LUA_NUMBER_FMT, (LUAI_UACNUMBER)x);
10288e3e3a7aSWarner Losh else if (x == 0) { /* can be -0... */
10298e3e3a7aSWarner Losh /* create "0" or "-0" followed by exponent */
10308e3e3a7aSWarner Losh return l_sprintf(buff, sz, LUA_NUMBER_FMT "x0p+0", (LUAI_UACNUMBER)x);
10318e3e3a7aSWarner Losh }
10328e3e3a7aSWarner Losh else {
10338e3e3a7aSWarner Losh int e;
10348e3e3a7aSWarner Losh lua_Number m = l_mathop(frexp)(x, &e); /* 'x' fraction and exponent */
10358e3e3a7aSWarner Losh int n = 0; /* character count */
10368e3e3a7aSWarner Losh if (m < 0) { /* is number negative? */
10370495ed39SKyle Evans buff[n++] = '-'; /* add sign */
10388e3e3a7aSWarner Losh m = -m; /* make it positive */
10398e3e3a7aSWarner Losh }
10408e3e3a7aSWarner Losh buff[n++] = '0'; buff[n++] = 'x'; /* add "0x" */
10418e3e3a7aSWarner Losh m = adddigit(buff, n++, m * (1 << L_NBFD)); /* add first digit */
10428e3e3a7aSWarner Losh e -= L_NBFD; /* this digit goes before the radix point */
10438e3e3a7aSWarner Losh if (m > 0) { /* more digits? */
10448e3e3a7aSWarner Losh buff[n++] = lua_getlocaledecpoint(); /* add radix point */
10458e3e3a7aSWarner Losh do { /* add as many digits as needed */
10468e3e3a7aSWarner Losh m = adddigit(buff, n++, m * 16);
10478e3e3a7aSWarner Losh } while (m > 0);
10488e3e3a7aSWarner Losh }
10498e3e3a7aSWarner Losh n += l_sprintf(buff + n, sz - n, "p%+d", e); /* add exponent */
10508e3e3a7aSWarner Losh lua_assert(n < sz);
10518e3e3a7aSWarner Losh return n;
10528e3e3a7aSWarner Losh }
10538e3e3a7aSWarner Losh }
10548e3e3a7aSWarner Losh
10558e3e3a7aSWarner Losh
lua_number2strx(lua_State * L,char * buff,int sz,const char * fmt,lua_Number x)10568e3e3a7aSWarner Losh static int lua_number2strx (lua_State *L, char *buff, int sz,
10578e3e3a7aSWarner Losh const char *fmt, lua_Number x) {
10588e3e3a7aSWarner Losh int n = num2straux(buff, sz, x);
10598e3e3a7aSWarner Losh if (fmt[SIZELENMOD] == 'A') {
10608e3e3a7aSWarner Losh int i;
10618e3e3a7aSWarner Losh for (i = 0; i < n; i++)
10628e3e3a7aSWarner Losh buff[i] = toupper(uchar(buff[i]));
10638e3e3a7aSWarner Losh }
10648c784bb8SWarner Losh else if (l_unlikely(fmt[SIZELENMOD] != 'a'))
1065e112e9d2SKyle Evans return luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented");
10668e3e3a7aSWarner Losh return n;
10678e3e3a7aSWarner Losh }
10688e3e3a7aSWarner Losh
10698e3e3a7aSWarner Losh #endif /* } */
10708e3e3a7aSWarner Losh
10718e3e3a7aSWarner Losh
10728e3e3a7aSWarner Losh /*
10730495ed39SKyle Evans ** Maximum size for items formatted with '%f'. This size is produced
10748e3e3a7aSWarner Losh ** by format('%.99f', -maxfloat), and is equal to 99 + 3 ('-', '.',
10758e3e3a7aSWarner Losh ** and '\0') + number of decimal digits to represent maxfloat (which
10760495ed39SKyle Evans ** is maximum exponent + 1). (99+3+1, adding some extra, 110)
10778e3e3a7aSWarner Losh */
10780495ed39SKyle Evans #define MAX_ITEMF (110 + l_floatatt(MAX_10_EXP))
10790495ed39SKyle Evans
10800495ed39SKyle Evans
10810495ed39SKyle Evans /*
10820495ed39SKyle Evans ** All formats except '%f' do not need that large limit. The other
10830495ed39SKyle Evans ** float formats use exponents, so that they fit in the 99 limit for
10840495ed39SKyle Evans ** significant digits; 's' for large strings and 'q' add items directly
10850495ed39SKyle Evans ** to the buffer; all integer formats also fit in the 99 limit. The
10860495ed39SKyle Evans ** worst case are floats: they may need 99 significant digits, plus
10870495ed39SKyle Evans ** '0x', '-', '.', 'e+XXXX', and '\0'. Adding some extra, 120.
10880495ed39SKyle Evans */
10890495ed39SKyle Evans #define MAX_ITEM 120
10908e3e3a7aSWarner Losh
10918e3e3a7aSWarner Losh
10928e3e3a7aSWarner Losh /* valid flags in a format specification */
10938c784bb8SWarner Losh #if !defined(L_FMTFLAGSF)
10948c784bb8SWarner Losh
10958c784bb8SWarner Losh /* valid flags for a, A, e, E, f, F, g, and G conversions */
10968c784bb8SWarner Losh #define L_FMTFLAGSF "-+#0 "
10978c784bb8SWarner Losh
10988c784bb8SWarner Losh /* valid flags for o, x, and X conversions */
10998c784bb8SWarner Losh #define L_FMTFLAGSX "-#0"
11008c784bb8SWarner Losh
11018c784bb8SWarner Losh /* valid flags for d and i conversions */
11028c784bb8SWarner Losh #define L_FMTFLAGSI "-+0 "
11038c784bb8SWarner Losh
11048c784bb8SWarner Losh /* valid flags for u conversions */
11058c784bb8SWarner Losh #define L_FMTFLAGSU "-0"
11068c784bb8SWarner Losh
11078c784bb8SWarner Losh /* valid flags for c, p, and s conversions */
11088c784bb8SWarner Losh #define L_FMTFLAGSC "-"
11098c784bb8SWarner Losh
11100495ed39SKyle Evans #endif
11110495ed39SKyle Evans
11128e3e3a7aSWarner Losh
11138e3e3a7aSWarner Losh /*
11148c784bb8SWarner Losh ** Maximum size of each format specification (such as "%-099.99d"):
11158c784bb8SWarner Losh ** Initial '%', flags (up to 5), width (2), period, precision (2),
11168c784bb8SWarner Losh ** length modifier (8), conversion specifier, and final '\0', plus some
11178c784bb8SWarner Losh ** extra.
11188e3e3a7aSWarner Losh */
11198e3e3a7aSWarner Losh #define MAX_FORMAT 32
11208e3e3a7aSWarner Losh
11218e3e3a7aSWarner Losh
addquoted(luaL_Buffer * b,const char * s,size_t len)11228e3e3a7aSWarner Losh static void addquoted (luaL_Buffer *b, const char *s, size_t len) {
11238e3e3a7aSWarner Losh luaL_addchar(b, '"');
11248e3e3a7aSWarner Losh while (len--) {
11258e3e3a7aSWarner Losh if (*s == '"' || *s == '\\' || *s == '\n') {
11268e3e3a7aSWarner Losh luaL_addchar(b, '\\');
11278e3e3a7aSWarner Losh luaL_addchar(b, *s);
11288e3e3a7aSWarner Losh }
11298e3e3a7aSWarner Losh else if (iscntrl(uchar(*s))) {
11308e3e3a7aSWarner Losh char buff[10];
11318e3e3a7aSWarner Losh if (!isdigit(uchar(*(s+1))))
11328e3e3a7aSWarner Losh l_sprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s));
11338e3e3a7aSWarner Losh else
11348e3e3a7aSWarner Losh l_sprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s));
11358e3e3a7aSWarner Losh luaL_addstring(b, buff);
11368e3e3a7aSWarner Losh }
11378e3e3a7aSWarner Losh else
11388e3e3a7aSWarner Losh luaL_addchar(b, *s);
11398e3e3a7aSWarner Losh s++;
11408e3e3a7aSWarner Losh }
11418e3e3a7aSWarner Losh luaL_addchar(b, '"');
11428e3e3a7aSWarner Losh }
11438e3e3a7aSWarner Losh
11448e3e3a7aSWarner Losh
11458c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
11468e3e3a7aSWarner Losh /*
11470495ed39SKyle Evans ** Serialize a floating-point number in such a way that it can be
11480495ed39SKyle Evans ** scanned back by Lua. Use hexadecimal format for "common" numbers
11490495ed39SKyle Evans ** (to preserve precision); inf, -inf, and NaN are handled separately.
11500495ed39SKyle Evans ** (NaN cannot be expressed as a numeral, so we write '(0/0)' for it.)
11518e3e3a7aSWarner Losh */
quotefloat(lua_State * L,char * buff,lua_Number n)11520495ed39SKyle Evans static int quotefloat (lua_State *L, char *buff, lua_Number n) {
11530495ed39SKyle Evans const char *s; /* for the fixed representations */
11540495ed39SKyle Evans if (n == (lua_Number)HUGE_VAL) /* inf? */
11550495ed39SKyle Evans s = "1e9999";
11560495ed39SKyle Evans else if (n == -(lua_Number)HUGE_VAL) /* -inf? */
11570495ed39SKyle Evans s = "-1e9999";
11580495ed39SKyle Evans else if (n != n) /* NaN? */
11590495ed39SKyle Evans s = "(0/0)";
11600495ed39SKyle Evans else { /* format number as hexadecimal */
11610495ed39SKyle Evans int nb = lua_number2strx(L, buff, MAX_ITEM,
11620495ed39SKyle Evans "%" LUA_NUMBER_FRMLEN "a", n);
11630495ed39SKyle Evans /* ensures that 'buff' string uses a dot as the radix character */
11648e3e3a7aSWarner Losh if (memchr(buff, '.', nb) == NULL) { /* no dot? */
11658e3e3a7aSWarner Losh char point = lua_getlocaledecpoint(); /* try locale point */
11668e3e3a7aSWarner Losh char *ppoint = (char *)memchr(buff, point, nb);
11678e3e3a7aSWarner Losh if (ppoint) *ppoint = '.'; /* change it to a dot */
11688e3e3a7aSWarner Losh }
11690495ed39SKyle Evans return nb;
11700495ed39SKyle Evans }
11710495ed39SKyle Evans /* for the fixed representations */
11720495ed39SKyle Evans return l_sprintf(buff, MAX_ITEM, "%s", s);
11738e3e3a7aSWarner Losh }
117407d397d7SWarner Losh #endif
11758e3e3a7aSWarner Losh
11768c784bb8SWarner Losh
addliteral(lua_State * L,luaL_Buffer * b,int arg)11778e3e3a7aSWarner Losh static void addliteral (lua_State *L, luaL_Buffer *b, int arg) {
11788e3e3a7aSWarner Losh switch (lua_type(L, arg)) {
11798e3e3a7aSWarner Losh case LUA_TSTRING: {
11808e3e3a7aSWarner Losh size_t len;
11818e3e3a7aSWarner Losh const char *s = lua_tolstring(L, arg, &len);
11828e3e3a7aSWarner Losh addquoted(b, s, len);
11838e3e3a7aSWarner Losh break;
11848e3e3a7aSWarner Losh }
11858e3e3a7aSWarner Losh case LUA_TNUMBER: {
11868e3e3a7aSWarner Losh char *buff = luaL_prepbuffsize(b, MAX_ITEM);
11878e3e3a7aSWarner Losh int nb;
11888c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
11890495ed39SKyle Evans if (!lua_isinteger(L, arg)) /* float? */
11900495ed39SKyle Evans nb = quotefloat(L, buff, lua_tonumber(L, arg));
11918e3e3a7aSWarner Losh else { /* integers */
1192919cf86cSWarner Losh #else
1193919cf86cSWarner Losh {
1194919cf86cSWarner Losh #endif
11958e3e3a7aSWarner Losh lua_Integer n = lua_tointeger(L, arg);
11968e3e3a7aSWarner Losh const char *format = (n == LUA_MININTEGER) /* corner case? */
11970495ed39SKyle Evans ? "0x%" LUA_INTEGER_FRMLEN "x" /* use hex */
11988e3e3a7aSWarner Losh : LUA_INTEGER_FMT; /* else use default format */
11998e3e3a7aSWarner Losh nb = l_sprintf(buff, MAX_ITEM, format, (LUAI_UACINT)n);
12008e3e3a7aSWarner Losh }
12018e3e3a7aSWarner Losh luaL_addsize(b, nb);
12028e3e3a7aSWarner Losh break;
12038e3e3a7aSWarner Losh }
12048e3e3a7aSWarner Losh case LUA_TNIL: case LUA_TBOOLEAN: {
12058e3e3a7aSWarner Losh luaL_tolstring(L, arg, NULL);
12068e3e3a7aSWarner Losh luaL_addvalue(b);
12078e3e3a7aSWarner Losh break;
12088e3e3a7aSWarner Losh }
12098e3e3a7aSWarner Losh default: {
12108e3e3a7aSWarner Losh luaL_argerror(L, arg, "value has no literal form");
12118e3e3a7aSWarner Losh }
12128e3e3a7aSWarner Losh }
12138e3e3a7aSWarner Losh }
12148e3e3a7aSWarner Losh
12158e3e3a7aSWarner Losh
12168c784bb8SWarner Losh static const char *get2digits (const char *s) {
12178c784bb8SWarner Losh if (isdigit(uchar(*s))) {
12188c784bb8SWarner Losh s++;
12198c784bb8SWarner Losh if (isdigit(uchar(*s))) s++; /* (2 digits at most) */
12208e3e3a7aSWarner Losh }
12218c784bb8SWarner Losh return s;
12228c784bb8SWarner Losh }
12238c784bb8SWarner Losh
12248c784bb8SWarner Losh
12258c784bb8SWarner Losh /*
12268c784bb8SWarner Losh ** Check whether a conversion specification is valid. When called,
12278c784bb8SWarner Losh ** first character in 'form' must be '%' and last character must
12288c784bb8SWarner Losh ** be a valid conversion specifier. 'flags' are the accepted flags;
12298c784bb8SWarner Losh ** 'precision' signals whether to accept a precision.
12308c784bb8SWarner Losh */
12318c784bb8SWarner Losh static void checkformat (lua_State *L, const char *form, const char *flags,
12328c784bb8SWarner Losh int precision) {
12338c784bb8SWarner Losh const char *spec = form + 1; /* skip '%' */
12348c784bb8SWarner Losh spec += strspn(spec, flags); /* skip flags */
12358c784bb8SWarner Losh if (*spec != '0') { /* a width cannot start with '0' */
12368c784bb8SWarner Losh spec = get2digits(spec); /* skip width */
12378c784bb8SWarner Losh if (*spec == '.' && precision) {
12388c784bb8SWarner Losh spec++;
12398c784bb8SWarner Losh spec = get2digits(spec); /* skip precision */
12408c784bb8SWarner Losh }
12418c784bb8SWarner Losh }
12428c784bb8SWarner Losh if (!isalpha(uchar(*spec))) /* did not go to the end? */
12438c784bb8SWarner Losh luaL_error(L, "invalid conversion specification: '%s'", form);
12448c784bb8SWarner Losh }
12458c784bb8SWarner Losh
12468c784bb8SWarner Losh
12478c784bb8SWarner Losh /*
12488c784bb8SWarner Losh ** Get a conversion specification and copy it to 'form'.
12498c784bb8SWarner Losh ** Return the address of its last character.
12508c784bb8SWarner Losh */
12518c784bb8SWarner Losh static const char *getformat (lua_State *L, const char *strfrmt,
12528c784bb8SWarner Losh char *form) {
12538c784bb8SWarner Losh /* spans flags, width, and precision ('0' is included as a flag) */
12548c784bb8SWarner Losh size_t len = strspn(strfrmt, L_FMTFLAGSF "123456789.");
12558c784bb8SWarner Losh len++; /* adds following character (should be the specifier) */
12568c784bb8SWarner Losh /* still needs space for '%', '\0', plus a length modifier */
12578c784bb8SWarner Losh if (len >= MAX_FORMAT - 10)
12588c784bb8SWarner Losh luaL_error(L, "invalid format (too long)");
12598e3e3a7aSWarner Losh *(form++) = '%';
12608c784bb8SWarner Losh memcpy(form, strfrmt, len * sizeof(char));
12618c784bb8SWarner Losh *(form + len) = '\0';
12628c784bb8SWarner Losh return strfrmt + len - 1;
12638e3e3a7aSWarner Losh }
12648e3e3a7aSWarner Losh
12658e3e3a7aSWarner Losh
12668e3e3a7aSWarner Losh /*
12678e3e3a7aSWarner Losh ** add length modifier into formats
12688e3e3a7aSWarner Losh */
12698e3e3a7aSWarner Losh static void addlenmod (char *form, const char *lenmod) {
12708e3e3a7aSWarner Losh size_t l = strlen(form);
12718e3e3a7aSWarner Losh size_t lm = strlen(lenmod);
12728e3e3a7aSWarner Losh char spec = form[l - 1];
12738e3e3a7aSWarner Losh strcpy(form + l - 1, lenmod);
12748e3e3a7aSWarner Losh form[l + lm - 1] = spec;
12758e3e3a7aSWarner Losh form[l + lm] = '\0';
12768e3e3a7aSWarner Losh }
12778e3e3a7aSWarner Losh
12788e3e3a7aSWarner Losh
12798e3e3a7aSWarner Losh static int str_format (lua_State *L) {
12808e3e3a7aSWarner Losh int top = lua_gettop(L);
12818e3e3a7aSWarner Losh int arg = 1;
12828e3e3a7aSWarner Losh size_t sfl;
12838e3e3a7aSWarner Losh const char *strfrmt = luaL_checklstring(L, arg, &sfl);
12848e3e3a7aSWarner Losh const char *strfrmt_end = strfrmt+sfl;
12858c784bb8SWarner Losh const char *flags;
12868e3e3a7aSWarner Losh luaL_Buffer b;
12878e3e3a7aSWarner Losh luaL_buffinit(L, &b);
12888e3e3a7aSWarner Losh while (strfrmt < strfrmt_end) {
12898e3e3a7aSWarner Losh if (*strfrmt != L_ESC)
12908e3e3a7aSWarner Losh luaL_addchar(&b, *strfrmt++);
12918e3e3a7aSWarner Losh else if (*++strfrmt == L_ESC)
12928e3e3a7aSWarner Losh luaL_addchar(&b, *strfrmt++); /* %% */
12938e3e3a7aSWarner Losh else { /* format item */
12948e3e3a7aSWarner Losh char form[MAX_FORMAT]; /* to store the format ('%...') */
12958c784bb8SWarner Losh int maxitem = MAX_ITEM; /* maximum length for the result */
12968c784bb8SWarner Losh char *buff = luaL_prepbuffsize(&b, maxitem); /* to put result */
12978c784bb8SWarner Losh int nb = 0; /* number of bytes in result */
12988e3e3a7aSWarner Losh if (++arg > top)
12990495ed39SKyle Evans return luaL_argerror(L, arg, "no value");
13008c784bb8SWarner Losh strfrmt = getformat(L, strfrmt, form);
13018e3e3a7aSWarner Losh switch (*strfrmt++) {
13028e3e3a7aSWarner Losh case 'c': {
13038c784bb8SWarner Losh checkformat(L, form, L_FMTFLAGSC, 0);
13040495ed39SKyle Evans nb = l_sprintf(buff, maxitem, form, (int)luaL_checkinteger(L, arg));
13058e3e3a7aSWarner Losh break;
13068e3e3a7aSWarner Losh }
13078e3e3a7aSWarner Losh case 'd': case 'i':
13088c784bb8SWarner Losh flags = L_FMTFLAGSI;
13098c784bb8SWarner Losh goto intcase;
13108c784bb8SWarner Losh case 'u':
13118c784bb8SWarner Losh flags = L_FMTFLAGSU;
13128c784bb8SWarner Losh goto intcase;
13138c784bb8SWarner Losh case 'o': case 'x': case 'X':
13148c784bb8SWarner Losh flags = L_FMTFLAGSX;
13158c784bb8SWarner Losh intcase: {
13168e3e3a7aSWarner Losh lua_Integer n = luaL_checkinteger(L, arg);
13178c784bb8SWarner Losh checkformat(L, form, flags, 1);
13188e3e3a7aSWarner Losh addlenmod(form, LUA_INTEGER_FRMLEN);
13190495ed39SKyle Evans nb = l_sprintf(buff, maxitem, form, (LUAI_UACINT)n);
13208e3e3a7aSWarner Losh break;
13218e3e3a7aSWarner Losh }
13228e3e3a7aSWarner Losh case 'a': case 'A':
13238c784bb8SWarner Losh checkformat(L, form, L_FMTFLAGSF, 1);
13248e3e3a7aSWarner Losh addlenmod(form, LUA_NUMBER_FRMLEN);
13250495ed39SKyle Evans nb = lua_number2strx(L, buff, maxitem, form,
13268e3e3a7aSWarner Losh luaL_checknumber(L, arg));
13278e3e3a7aSWarner Losh break;
13288c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
13290495ed39SKyle Evans case 'f':
13300495ed39SKyle Evans maxitem = MAX_ITEMF; /* extra space for '%f' */
13310495ed39SKyle Evans buff = luaL_prepbuffsize(&b, maxitem);
13320495ed39SKyle Evans /* FALLTHROUGH */
13330495ed39SKyle Evans #endif
13340495ed39SKyle Evans case 'e': case 'E': case 'g': case 'G': {
13358e3e3a7aSWarner Losh lua_Number n = luaL_checknumber(L, arg);
13368c784bb8SWarner Losh checkformat(L, form, L_FMTFLAGSF, 1);
13378e3e3a7aSWarner Losh addlenmod(form, LUA_NUMBER_FRMLEN);
13380495ed39SKyle Evans nb = l_sprintf(buff, maxitem, form, (LUAI_UACNUMBER)n);
13390495ed39SKyle Evans break;
13400495ed39SKyle Evans }
13410495ed39SKyle Evans case 'p': {
13420495ed39SKyle Evans const void *p = lua_topointer(L, arg);
13438c784bb8SWarner Losh checkformat(L, form, L_FMTFLAGSC, 0);
13440495ed39SKyle Evans if (p == NULL) { /* avoid calling 'printf' with argument NULL */
13450495ed39SKyle Evans p = "(null)"; /* result */
13460495ed39SKyle Evans form[strlen(form) - 1] = 's'; /* format it as a string */
13470495ed39SKyle Evans }
13480495ed39SKyle Evans nb = l_sprintf(buff, maxitem, form, p);
13498e3e3a7aSWarner Losh break;
13508e3e3a7aSWarner Losh }
13518e3e3a7aSWarner Losh case 'q': {
13520495ed39SKyle Evans if (form[2] != '\0') /* modifiers? */
13530495ed39SKyle Evans return luaL_error(L, "specifier '%%q' cannot have modifiers");
13548e3e3a7aSWarner Losh addliteral(L, &b, arg);
13558e3e3a7aSWarner Losh break;
13568e3e3a7aSWarner Losh }
13578e3e3a7aSWarner Losh case 's': {
13588e3e3a7aSWarner Losh size_t l;
13598e3e3a7aSWarner Losh const char *s = luaL_tolstring(L, arg, &l);
13608e3e3a7aSWarner Losh if (form[2] == '\0') /* no modifiers? */
13618e3e3a7aSWarner Losh luaL_addvalue(&b); /* keep entire string */
13628e3e3a7aSWarner Losh else {
13638e3e3a7aSWarner Losh luaL_argcheck(L, l == strlen(s), arg, "string contains zeros");
13648c784bb8SWarner Losh checkformat(L, form, L_FMTFLAGSC, 1);
13658c784bb8SWarner Losh if (strchr(form, '.') == NULL && l >= 100) {
13668e3e3a7aSWarner Losh /* no precision and string is too long to be formatted */
13678e3e3a7aSWarner Losh luaL_addvalue(&b); /* keep entire string */
13688e3e3a7aSWarner Losh }
13698e3e3a7aSWarner Losh else { /* format the string into 'buff' */
13700495ed39SKyle Evans nb = l_sprintf(buff, maxitem, form, s);
13718e3e3a7aSWarner Losh lua_pop(L, 1); /* remove result from 'luaL_tolstring' */
13728e3e3a7aSWarner Losh }
13738e3e3a7aSWarner Losh }
13748e3e3a7aSWarner Losh break;
13758e3e3a7aSWarner Losh }
13768e3e3a7aSWarner Losh default: { /* also treat cases 'pnLlh' */
13770495ed39SKyle Evans return luaL_error(L, "invalid conversion '%s' to 'format'", form);
13788e3e3a7aSWarner Losh }
13798e3e3a7aSWarner Losh }
13800495ed39SKyle Evans lua_assert(nb < maxitem);
13818e3e3a7aSWarner Losh luaL_addsize(&b, nb);
13828e3e3a7aSWarner Losh }
13838e3e3a7aSWarner Losh }
13848e3e3a7aSWarner Losh luaL_pushresult(&b);
13858e3e3a7aSWarner Losh return 1;
13868e3e3a7aSWarner Losh }
13878e3e3a7aSWarner Losh
13888e3e3a7aSWarner Losh /* }====================================================== */
13898e3e3a7aSWarner Losh
13908e3e3a7aSWarner Losh
13918e3e3a7aSWarner Losh /*
13928e3e3a7aSWarner Losh ** {======================================================
13938e3e3a7aSWarner Losh ** PACK/UNPACK
13948e3e3a7aSWarner Losh ** =======================================================
13958e3e3a7aSWarner Losh */
13968e3e3a7aSWarner Losh
13978e3e3a7aSWarner Losh
13988e3e3a7aSWarner Losh /* value used for padding */
13998e3e3a7aSWarner Losh #if !defined(LUAL_PACKPADBYTE)
14008e3e3a7aSWarner Losh #define LUAL_PACKPADBYTE 0x00
14018e3e3a7aSWarner Losh #endif
14028e3e3a7aSWarner Losh
14038e3e3a7aSWarner Losh /* maximum size for the binary representation of an integer */
14048e3e3a7aSWarner Losh #define MAXINTSIZE 16
14058e3e3a7aSWarner Losh
14068e3e3a7aSWarner Losh /* number of bits in a character */
14078e3e3a7aSWarner Losh #define NB CHAR_BIT
14088e3e3a7aSWarner Losh
14098e3e3a7aSWarner Losh /* mask for one character (NB 1's) */
14108e3e3a7aSWarner Losh #define MC ((1 << NB) - 1)
14118e3e3a7aSWarner Losh
14128e3e3a7aSWarner Losh /* size of a lua_Integer */
14138e3e3a7aSWarner Losh #define SZINT ((int)sizeof(lua_Integer))
14148e3e3a7aSWarner Losh
14158e3e3a7aSWarner Losh
14168e3e3a7aSWarner Losh /* dummy union to get native endianness */
14178e3e3a7aSWarner Losh static const union {
14188e3e3a7aSWarner Losh int dummy;
14198e3e3a7aSWarner Losh char little; /* true iff machine is little endian */
14208e3e3a7aSWarner Losh } nativeendian = {1};
14218e3e3a7aSWarner Losh
14228e3e3a7aSWarner Losh
14238e3e3a7aSWarner Losh /*
14248e3e3a7aSWarner Losh ** information to pack/unpack stuff
14258e3e3a7aSWarner Losh */
14268e3e3a7aSWarner Losh typedef struct Header {
14278e3e3a7aSWarner Losh lua_State *L;
14288e3e3a7aSWarner Losh int islittle;
14298e3e3a7aSWarner Losh int maxalign;
14308e3e3a7aSWarner Losh } Header;
14318e3e3a7aSWarner Losh
14328e3e3a7aSWarner Losh
14338e3e3a7aSWarner Losh /*
14348e3e3a7aSWarner Losh ** options for pack/unpack
14358e3e3a7aSWarner Losh */
14368e3e3a7aSWarner Losh typedef enum KOption {
14378e3e3a7aSWarner Losh Kint, /* signed integers */
14388e3e3a7aSWarner Losh Kuint, /* unsigned integers */
14398c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
14408c784bb8SWarner Losh Kfloat, /* single-precision floating-point numbers */
14418c784bb8SWarner Losh Kdouble, /* double-precision floating-point numbers */
14428c784bb8SWarner Losh #endif
14438c784bb8SWarner Losh Knumber, /* Lua "native" floating-point numbers */
14448e3e3a7aSWarner Losh Kchar, /* fixed-length strings */
14458e3e3a7aSWarner Losh Kstring, /* strings with prefixed length */
14468e3e3a7aSWarner Losh Kzstr, /* zero-terminated strings */
14478e3e3a7aSWarner Losh Kpadding, /* padding */
14488e3e3a7aSWarner Losh Kpaddalign, /* padding for alignment */
14498e3e3a7aSWarner Losh Knop /* no-op (configuration or spaces) */
14508e3e3a7aSWarner Losh } KOption;
14518e3e3a7aSWarner Losh
14528e3e3a7aSWarner Losh
14538e3e3a7aSWarner Losh /*
14548e3e3a7aSWarner Losh ** Read an integer numeral from string 'fmt' or return 'df' if
14558e3e3a7aSWarner Losh ** there is no numeral
14568e3e3a7aSWarner Losh */
14578e3e3a7aSWarner Losh static int digit (int c) { return '0' <= c && c <= '9'; }
14588e3e3a7aSWarner Losh
14598e3e3a7aSWarner Losh static int getnum (const char **fmt, int df) {
14608e3e3a7aSWarner Losh if (!digit(**fmt)) /* no number? */
14618e3e3a7aSWarner Losh return df; /* return default value */
14628e3e3a7aSWarner Losh else {
14638e3e3a7aSWarner Losh int a = 0;
14648e3e3a7aSWarner Losh do {
14658e3e3a7aSWarner Losh a = a*10 + (*((*fmt)++) - '0');
14668e3e3a7aSWarner Losh } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10);
14678e3e3a7aSWarner Losh return a;
14688e3e3a7aSWarner Losh }
14698e3e3a7aSWarner Losh }
14708e3e3a7aSWarner Losh
14718e3e3a7aSWarner Losh
14728e3e3a7aSWarner Losh /*
14738e3e3a7aSWarner Losh ** Read an integer numeral and raises an error if it is larger
14748e3e3a7aSWarner Losh ** than the maximum size for integers.
14758e3e3a7aSWarner Losh */
14768e3e3a7aSWarner Losh static int getnumlimit (Header *h, const char **fmt, int df) {
14778e3e3a7aSWarner Losh int sz = getnum(fmt, df);
14788c784bb8SWarner Losh if (l_unlikely(sz > MAXINTSIZE || sz <= 0))
1479e112e9d2SKyle Evans return luaL_error(h->L, "integral size (%d) out of limits [1,%d]",
14808e3e3a7aSWarner Losh sz, MAXINTSIZE);
14818e3e3a7aSWarner Losh return sz;
14828e3e3a7aSWarner Losh }
14838e3e3a7aSWarner Losh
14848e3e3a7aSWarner Losh
14858e3e3a7aSWarner Losh /*
14868e3e3a7aSWarner Losh ** Initialize Header
14878e3e3a7aSWarner Losh */
14888e3e3a7aSWarner Losh static void initheader (lua_State *L, Header *h) {
14898e3e3a7aSWarner Losh h->L = L;
14908e3e3a7aSWarner Losh h->islittle = nativeendian.little;
14918e3e3a7aSWarner Losh h->maxalign = 1;
14928e3e3a7aSWarner Losh }
14938e3e3a7aSWarner Losh
14948e3e3a7aSWarner Losh
14958e3e3a7aSWarner Losh /*
14968e3e3a7aSWarner Losh ** Read and classify next option. 'size' is filled with option's size.
14978e3e3a7aSWarner Losh */
14988e3e3a7aSWarner Losh static KOption getoption (Header *h, const char **fmt, int *size) {
14998c784bb8SWarner Losh /* dummy structure to get native alignment requirements */
15008c784bb8SWarner Losh struct cD { char c; union { LUAI_MAXALIGN; } u; };
15018e3e3a7aSWarner Losh int opt = *((*fmt)++);
15028e3e3a7aSWarner Losh *size = 0; /* default */
15038e3e3a7aSWarner Losh switch (opt) {
15048e3e3a7aSWarner Losh case 'b': *size = sizeof(char); return Kint;
15058e3e3a7aSWarner Losh case 'B': *size = sizeof(char); return Kuint;
15068e3e3a7aSWarner Losh case 'h': *size = sizeof(short); return Kint;
15078e3e3a7aSWarner Losh case 'H': *size = sizeof(short); return Kuint;
15088e3e3a7aSWarner Losh case 'l': *size = sizeof(long); return Kint;
15098e3e3a7aSWarner Losh case 'L': *size = sizeof(long); return Kuint;
15108e3e3a7aSWarner Losh case 'j': *size = sizeof(lua_Integer); return Kint;
15118e3e3a7aSWarner Losh case 'J': *size = sizeof(lua_Integer); return Kuint;
15128e3e3a7aSWarner Losh case 'T': *size = sizeof(size_t); return Kuint;
15138c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
15148e3e3a7aSWarner Losh case 'f': *size = sizeof(float); return Kfloat;
15158c784bb8SWarner Losh case 'd': *size = sizeof(double); return Kdouble;
151607c17b2bSWarner Losh #endif
15178c784bb8SWarner Losh case 'n': *size = sizeof(lua_Number); return Knumber;
15188e3e3a7aSWarner Losh case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint;
15198e3e3a7aSWarner Losh case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint;
15208e3e3a7aSWarner Losh case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring;
15218e3e3a7aSWarner Losh case 'c':
15228e3e3a7aSWarner Losh *size = getnum(fmt, -1);
15238c784bb8SWarner Losh if (l_unlikely(*size == -1))
15248e3e3a7aSWarner Losh luaL_error(h->L, "missing size for format option 'c'");
15258e3e3a7aSWarner Losh return Kchar;
15268e3e3a7aSWarner Losh case 'z': return Kzstr;
15278e3e3a7aSWarner Losh case 'x': *size = 1; return Kpadding;
15288e3e3a7aSWarner Losh case 'X': return Kpaddalign;
15298e3e3a7aSWarner Losh case ' ': break;
15308e3e3a7aSWarner Losh case '<': h->islittle = 1; break;
15318e3e3a7aSWarner Losh case '>': h->islittle = 0; break;
15328e3e3a7aSWarner Losh case '=': h->islittle = nativeendian.little; break;
15338c784bb8SWarner Losh case '!': {
15348c784bb8SWarner Losh const int maxalign = offsetof(struct cD, u);
15358c784bb8SWarner Losh h->maxalign = getnumlimit(h, fmt, maxalign);
15368c784bb8SWarner Losh break;
15378c784bb8SWarner Losh }
15388e3e3a7aSWarner Losh default: luaL_error(h->L, "invalid format option '%c'", opt);
15398e3e3a7aSWarner Losh }
15408e3e3a7aSWarner Losh return Knop;
15418e3e3a7aSWarner Losh }
15428e3e3a7aSWarner Losh
15438e3e3a7aSWarner Losh
15448e3e3a7aSWarner Losh /*
15458e3e3a7aSWarner Losh ** Read, classify, and fill other details about the next option.
15468e3e3a7aSWarner Losh ** 'psize' is filled with option's size, 'notoalign' with its
15478e3e3a7aSWarner Losh ** alignment requirements.
15488e3e3a7aSWarner Losh ** Local variable 'size' gets the size to be aligned. (Kpadal option
15498e3e3a7aSWarner Losh ** always gets its full alignment, other options are limited by
15508e3e3a7aSWarner Losh ** the maximum alignment ('maxalign'). Kchar option needs no alignment
15518e3e3a7aSWarner Losh ** despite its size.
15528e3e3a7aSWarner Losh */
15538e3e3a7aSWarner Losh static KOption getdetails (Header *h, size_t totalsize,
15548e3e3a7aSWarner Losh const char **fmt, int *psize, int *ntoalign) {
15558e3e3a7aSWarner Losh KOption opt = getoption(h, fmt, psize);
15568e3e3a7aSWarner Losh int align = *psize; /* usually, alignment follows size */
15578e3e3a7aSWarner Losh if (opt == Kpaddalign) { /* 'X' gets alignment from following option */
15588e3e3a7aSWarner Losh if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0)
15598e3e3a7aSWarner Losh luaL_argerror(h->L, 1, "invalid next option for option 'X'");
15608e3e3a7aSWarner Losh }
15618e3e3a7aSWarner Losh if (align <= 1 || opt == Kchar) /* need no alignment? */
15628e3e3a7aSWarner Losh *ntoalign = 0;
15638e3e3a7aSWarner Losh else {
15648e3e3a7aSWarner Losh if (align > h->maxalign) /* enforce maximum alignment */
15658e3e3a7aSWarner Losh align = h->maxalign;
15668c784bb8SWarner Losh if (l_unlikely((align & (align - 1)) != 0)) /* not a power of 2? */
15678e3e3a7aSWarner Losh luaL_argerror(h->L, 1, "format asks for alignment not power of 2");
15688e3e3a7aSWarner Losh *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1);
15698e3e3a7aSWarner Losh }
15708e3e3a7aSWarner Losh return opt;
15718e3e3a7aSWarner Losh }
15728e3e3a7aSWarner Losh
15738e3e3a7aSWarner Losh
15748e3e3a7aSWarner Losh /*
15758e3e3a7aSWarner Losh ** Pack integer 'n' with 'size' bytes and 'islittle' endianness.
15768e3e3a7aSWarner Losh ** The final 'if' handles the case when 'size' is larger than
15778e3e3a7aSWarner Losh ** the size of a Lua integer, correcting the extra sign-extension
15788e3e3a7aSWarner Losh ** bytes if necessary (by default they would be zeros).
15798e3e3a7aSWarner Losh */
15808e3e3a7aSWarner Losh static void packint (luaL_Buffer *b, lua_Unsigned n,
15818e3e3a7aSWarner Losh int islittle, int size, int neg) {
15828e3e3a7aSWarner Losh char *buff = luaL_prepbuffsize(b, size);
15838e3e3a7aSWarner Losh int i;
15848e3e3a7aSWarner Losh buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */
15858e3e3a7aSWarner Losh for (i = 1; i < size; i++) {
15868e3e3a7aSWarner Losh n >>= NB;
15878e3e3a7aSWarner Losh buff[islittle ? i : size - 1 - i] = (char)(n & MC);
15888e3e3a7aSWarner Losh }
15898e3e3a7aSWarner Losh if (neg && size > SZINT) { /* negative number need sign extension? */
15908e3e3a7aSWarner Losh for (i = SZINT; i < size; i++) /* correct extra bytes */
15918e3e3a7aSWarner Losh buff[islittle ? i : size - 1 - i] = (char)MC;
15928e3e3a7aSWarner Losh }
15938e3e3a7aSWarner Losh luaL_addsize(b, size); /* add result to buffer */
15948e3e3a7aSWarner Losh }
15958e3e3a7aSWarner Losh
15968e3e3a7aSWarner Losh
15978e3e3a7aSWarner Losh /*
15988e3e3a7aSWarner Losh ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if
15998e3e3a7aSWarner Losh ** given 'islittle' is different from native endianness.
16008e3e3a7aSWarner Losh */
16010495ed39SKyle Evans static void copywithendian (char *dest, const char *src,
16028e3e3a7aSWarner Losh int size, int islittle) {
16030495ed39SKyle Evans if (islittle == nativeendian.little)
16040495ed39SKyle Evans memcpy(dest, src, size);
16058e3e3a7aSWarner Losh else {
16068e3e3a7aSWarner Losh dest += size - 1;
16078e3e3a7aSWarner Losh while (size-- != 0)
16088e3e3a7aSWarner Losh *(dest--) = *(src++);
16098e3e3a7aSWarner Losh }
16108e3e3a7aSWarner Losh }
16118e3e3a7aSWarner Losh
16128e3e3a7aSWarner Losh
16138e3e3a7aSWarner Losh static int str_pack (lua_State *L) {
16148e3e3a7aSWarner Losh luaL_Buffer b;
16158e3e3a7aSWarner Losh Header h;
16168e3e3a7aSWarner Losh const char *fmt = luaL_checkstring(L, 1); /* format string */
16178e3e3a7aSWarner Losh int arg = 1; /* current argument to pack */
16188e3e3a7aSWarner Losh size_t totalsize = 0; /* accumulate total size of result */
16198e3e3a7aSWarner Losh initheader(L, &h);
16208e3e3a7aSWarner Losh lua_pushnil(L); /* mark to separate arguments from string buffer */
16218e3e3a7aSWarner Losh luaL_buffinit(L, &b);
16228e3e3a7aSWarner Losh while (*fmt != '\0') {
16238e3e3a7aSWarner Losh int size, ntoalign;
16248e3e3a7aSWarner Losh KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
16258e3e3a7aSWarner Losh totalsize += ntoalign + size;
16268e3e3a7aSWarner Losh while (ntoalign-- > 0)
16278e3e3a7aSWarner Losh luaL_addchar(&b, LUAL_PACKPADBYTE); /* fill alignment */
16288e3e3a7aSWarner Losh arg++;
16298e3e3a7aSWarner Losh switch (opt) {
16308e3e3a7aSWarner Losh case Kint: { /* signed integers */
16318e3e3a7aSWarner Losh lua_Integer n = luaL_checkinteger(L, arg);
16328e3e3a7aSWarner Losh if (size < SZINT) { /* need overflow check? */
16338e3e3a7aSWarner Losh lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1);
16348e3e3a7aSWarner Losh luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow");
16358e3e3a7aSWarner Losh }
16368e3e3a7aSWarner Losh packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0));
16378e3e3a7aSWarner Losh break;
16388e3e3a7aSWarner Losh }
16398e3e3a7aSWarner Losh case Kuint: { /* unsigned integers */
16408e3e3a7aSWarner Losh lua_Integer n = luaL_checkinteger(L, arg);
16418e3e3a7aSWarner Losh if (size < SZINT) /* need overflow check? */
16428e3e3a7aSWarner Losh luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)),
16438e3e3a7aSWarner Losh arg, "unsigned overflow");
16448e3e3a7aSWarner Losh packint(&b, (lua_Unsigned)n, h.islittle, size, 0);
16458e3e3a7aSWarner Losh break;
16468e3e3a7aSWarner Losh }
16478c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
16488c784bb8SWarner Losh case Kfloat: { /* C float */
16498c784bb8SWarner Losh float f = (float)luaL_checknumber(L, arg); /* get argument */
16508c784bb8SWarner Losh char *buff = luaL_prepbuffsize(&b, sizeof(f));
16518c784bb8SWarner Losh /* move 'f' to final result, correcting endianness if needed */
16528c784bb8SWarner Losh copywithendian(buff, (char *)&f, sizeof(f), h.islittle);
16538e3e3a7aSWarner Losh luaL_addsize(&b, size);
16548e3e3a7aSWarner Losh break;
16558e3e3a7aSWarner Losh }
16568c784bb8SWarner Losh #endif
16578c784bb8SWarner Losh case Knumber: { /* Lua float */
16588c784bb8SWarner Losh lua_Number f = luaL_checknumber(L, arg); /* get argument */
16598c784bb8SWarner Losh char *buff = luaL_prepbuffsize(&b, sizeof(f));
16608c784bb8SWarner Losh /* move 'f' to final result, correcting endianness if needed */
16618c784bb8SWarner Losh copywithendian(buff, (char *)&f, sizeof(f), h.islittle);
16628c784bb8SWarner Losh luaL_addsize(&b, size);
16638c784bb8SWarner Losh break;
16648c784bb8SWarner Losh }
16658c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
16668c784bb8SWarner Losh case Kdouble: { /* C double */
16678c784bb8SWarner Losh double f = (double)luaL_checknumber(L, arg); /* get argument */
16688c784bb8SWarner Losh char *buff = luaL_prepbuffsize(&b, sizeof(f));
16698c784bb8SWarner Losh /* move 'f' to final result, correcting endianness if needed */
16708c784bb8SWarner Losh copywithendian(buff, (char *)&f, sizeof(f), h.islittle);
16718c784bb8SWarner Losh luaL_addsize(&b, size);
16728c784bb8SWarner Losh break;
16738c784bb8SWarner Losh }
16748c784bb8SWarner Losh #endif
16758e3e3a7aSWarner Losh case Kchar: { /* fixed-size string */
16768e3e3a7aSWarner Losh size_t len;
16778e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, arg, &len);
16788e3e3a7aSWarner Losh luaL_argcheck(L, len <= (size_t)size, arg,
16798e3e3a7aSWarner Losh "string longer than given size");
16808e3e3a7aSWarner Losh luaL_addlstring(&b, s, len); /* add string */
16818e3e3a7aSWarner Losh while (len++ < (size_t)size) /* pad extra space */
16828e3e3a7aSWarner Losh luaL_addchar(&b, LUAL_PACKPADBYTE);
16838e3e3a7aSWarner Losh break;
16848e3e3a7aSWarner Losh }
16858e3e3a7aSWarner Losh case Kstring: { /* strings with length count */
16868e3e3a7aSWarner Losh size_t len;
16878e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, arg, &len);
16888e3e3a7aSWarner Losh luaL_argcheck(L, size >= (int)sizeof(size_t) ||
16898e3e3a7aSWarner Losh len < ((size_t)1 << (size * NB)),
16908e3e3a7aSWarner Losh arg, "string length does not fit in given size");
16918e3e3a7aSWarner Losh packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */
16928e3e3a7aSWarner Losh luaL_addlstring(&b, s, len);
16938e3e3a7aSWarner Losh totalsize += len;
16948e3e3a7aSWarner Losh break;
16958e3e3a7aSWarner Losh }
16968e3e3a7aSWarner Losh case Kzstr: { /* zero-terminated string */
16978e3e3a7aSWarner Losh size_t len;
16988e3e3a7aSWarner Losh const char *s = luaL_checklstring(L, arg, &len);
16998e3e3a7aSWarner Losh luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros");
17008e3e3a7aSWarner Losh luaL_addlstring(&b, s, len);
17018e3e3a7aSWarner Losh luaL_addchar(&b, '\0'); /* add zero at the end */
17028e3e3a7aSWarner Losh totalsize += len + 1;
17038e3e3a7aSWarner Losh break;
17048e3e3a7aSWarner Losh }
17058e3e3a7aSWarner Losh case Kpadding: luaL_addchar(&b, LUAL_PACKPADBYTE); /* FALLTHROUGH */
17068e3e3a7aSWarner Losh case Kpaddalign: case Knop:
17078e3e3a7aSWarner Losh arg--; /* undo increment */
17088e3e3a7aSWarner Losh break;
17098e3e3a7aSWarner Losh }
17108e3e3a7aSWarner Losh }
17118e3e3a7aSWarner Losh luaL_pushresult(&b);
17128e3e3a7aSWarner Losh return 1;
17138e3e3a7aSWarner Losh }
17148e3e3a7aSWarner Losh
17158e3e3a7aSWarner Losh
17168e3e3a7aSWarner Losh static int str_packsize (lua_State *L) {
17178e3e3a7aSWarner Losh Header h;
17188e3e3a7aSWarner Losh const char *fmt = luaL_checkstring(L, 1); /* format string */
17198e3e3a7aSWarner Losh size_t totalsize = 0; /* accumulate total size of result */
17208e3e3a7aSWarner Losh initheader(L, &h);
17218e3e3a7aSWarner Losh while (*fmt != '\0') {
17228e3e3a7aSWarner Losh int size, ntoalign;
17238e3e3a7aSWarner Losh KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
17240495ed39SKyle Evans luaL_argcheck(L, opt != Kstring && opt != Kzstr, 1,
17250495ed39SKyle Evans "variable-length format");
17268e3e3a7aSWarner Losh size += ntoalign; /* total space used by option */
17278e3e3a7aSWarner Losh luaL_argcheck(L, totalsize <= MAXSIZE - size, 1,
17288e3e3a7aSWarner Losh "format result too large");
17298e3e3a7aSWarner Losh totalsize += size;
17308e3e3a7aSWarner Losh }
17318e3e3a7aSWarner Losh lua_pushinteger(L, (lua_Integer)totalsize);
17328e3e3a7aSWarner Losh return 1;
17338e3e3a7aSWarner Losh }
17348e3e3a7aSWarner Losh
17358e3e3a7aSWarner Losh
17368e3e3a7aSWarner Losh /*
17378e3e3a7aSWarner Losh ** Unpack an integer with 'size' bytes and 'islittle' endianness.
17388e3e3a7aSWarner Losh ** If size is smaller than the size of a Lua integer and integer
17398e3e3a7aSWarner Losh ** is signed, must do sign extension (propagating the sign to the
17408e3e3a7aSWarner Losh ** higher bits); if size is larger than the size of a Lua integer,
17418e3e3a7aSWarner Losh ** it must check the unread bytes to see whether they do not cause an
17428e3e3a7aSWarner Losh ** overflow.
17438e3e3a7aSWarner Losh */
17448e3e3a7aSWarner Losh static lua_Integer unpackint (lua_State *L, const char *str,
17458e3e3a7aSWarner Losh int islittle, int size, int issigned) {
17468e3e3a7aSWarner Losh lua_Unsigned res = 0;
17478e3e3a7aSWarner Losh int i;
17488e3e3a7aSWarner Losh int limit = (size <= SZINT) ? size : SZINT;
17498e3e3a7aSWarner Losh for (i = limit - 1; i >= 0; i--) {
17508e3e3a7aSWarner Losh res <<= NB;
17518e3e3a7aSWarner Losh res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i];
17528e3e3a7aSWarner Losh }
17538e3e3a7aSWarner Losh if (size < SZINT) { /* real size smaller than lua_Integer? */
17548e3e3a7aSWarner Losh if (issigned) { /* needs sign extension? */
17558e3e3a7aSWarner Losh lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1);
17568e3e3a7aSWarner Losh res = ((res ^ mask) - mask); /* do sign extension */
17578e3e3a7aSWarner Losh }
17588e3e3a7aSWarner Losh }
17598e3e3a7aSWarner Losh else if (size > SZINT) { /* must check unread bytes */
17608e3e3a7aSWarner Losh int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC;
17618e3e3a7aSWarner Losh for (i = limit; i < size; i++) {
17628c784bb8SWarner Losh if (l_unlikely((unsigned char)str[islittle ? i : size - 1 - i] != mask))
17638e3e3a7aSWarner Losh luaL_error(L, "%d-byte integer does not fit into Lua Integer", size);
17648e3e3a7aSWarner Losh }
17658e3e3a7aSWarner Losh }
17668e3e3a7aSWarner Losh return (lua_Integer)res;
17678e3e3a7aSWarner Losh }
17688e3e3a7aSWarner Losh
17698e3e3a7aSWarner Losh
17708e3e3a7aSWarner Losh static int str_unpack (lua_State *L) {
17718e3e3a7aSWarner Losh Header h;
17728e3e3a7aSWarner Losh const char *fmt = luaL_checkstring(L, 1);
17738e3e3a7aSWarner Losh size_t ld;
17748e3e3a7aSWarner Losh const char *data = luaL_checklstring(L, 2, &ld);
17750495ed39SKyle Evans size_t pos = posrelatI(luaL_optinteger(L, 3, 1), ld) - 1;
17768e3e3a7aSWarner Losh int n = 0; /* number of results */
17778e3e3a7aSWarner Losh luaL_argcheck(L, pos <= ld, 3, "initial position out of string");
17788e3e3a7aSWarner Losh initheader(L, &h);
17798e3e3a7aSWarner Losh while (*fmt != '\0') {
17808e3e3a7aSWarner Losh int size, ntoalign;
17818e3e3a7aSWarner Losh KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign);
17820495ed39SKyle Evans luaL_argcheck(L, (size_t)ntoalign + size <= ld - pos, 2,
17830495ed39SKyle Evans "data string too short");
17848e3e3a7aSWarner Losh pos += ntoalign; /* skip alignment */
17858e3e3a7aSWarner Losh /* stack space for item + next position */
17868e3e3a7aSWarner Losh luaL_checkstack(L, 2, "too many results");
17878e3e3a7aSWarner Losh n++;
17888e3e3a7aSWarner Losh switch (opt) {
17898e3e3a7aSWarner Losh case Kint:
17908e3e3a7aSWarner Losh case Kuint: {
17918e3e3a7aSWarner Losh lua_Integer res = unpackint(L, data + pos, h.islittle, size,
17928e3e3a7aSWarner Losh (opt == Kint));
17938e3e3a7aSWarner Losh lua_pushinteger(L, res);
17948e3e3a7aSWarner Losh break;
17958e3e3a7aSWarner Losh }
17968c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
17978e3e3a7aSWarner Losh case Kfloat: {
17988c784bb8SWarner Losh float f;
17998c784bb8SWarner Losh copywithendian((char *)&f, data + pos, sizeof(f), h.islittle);
18008c784bb8SWarner Losh lua_pushnumber(L, (lua_Number)f);
18018e3e3a7aSWarner Losh break;
18028e3e3a7aSWarner Losh }
18038c784bb8SWarner Losh #endif
18048c784bb8SWarner Losh case Knumber: {
18058c784bb8SWarner Losh lua_Number f;
18068c784bb8SWarner Losh copywithendian((char *)&f, data + pos, sizeof(f), h.islittle);
18078c784bb8SWarner Losh lua_pushnumber(L, f);
18088c784bb8SWarner Losh break;
18098c784bb8SWarner Losh }
18108c784bb8SWarner Losh #ifndef LUA_AVOID_FLOAT
18118c784bb8SWarner Losh case Kdouble: {
18128c784bb8SWarner Losh double f;
18138c784bb8SWarner Losh copywithendian((char *)&f, data + pos, sizeof(f), h.islittle);
18148c784bb8SWarner Losh lua_pushnumber(L, (lua_Number)f);
18158c784bb8SWarner Losh break;
18168c784bb8SWarner Losh }
18178c784bb8SWarner Losh #endif
18188e3e3a7aSWarner Losh case Kchar: {
18198e3e3a7aSWarner Losh lua_pushlstring(L, data + pos, size);
18208e3e3a7aSWarner Losh break;
18218e3e3a7aSWarner Losh }
18228e3e3a7aSWarner Losh case Kstring: {
18238e3e3a7aSWarner Losh size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0);
18240495ed39SKyle Evans luaL_argcheck(L, len <= ld - pos - size, 2, "data string too short");
18258e3e3a7aSWarner Losh lua_pushlstring(L, data + pos + size, len);
18268e3e3a7aSWarner Losh pos += len; /* skip string */
18278e3e3a7aSWarner Losh break;
18288e3e3a7aSWarner Losh }
18298e3e3a7aSWarner Losh case Kzstr: {
18300495ed39SKyle Evans size_t len = strlen(data + pos);
18310495ed39SKyle Evans luaL_argcheck(L, pos + len < ld, 2,
18320495ed39SKyle Evans "unfinished string for format 'z'");
18338e3e3a7aSWarner Losh lua_pushlstring(L, data + pos, len);
18348e3e3a7aSWarner Losh pos += len + 1; /* skip string plus final '\0' */
18358e3e3a7aSWarner Losh break;
18368e3e3a7aSWarner Losh }
18378e3e3a7aSWarner Losh case Kpaddalign: case Kpadding: case Knop:
18388e3e3a7aSWarner Losh n--; /* undo increment */
18398e3e3a7aSWarner Losh break;
18408e3e3a7aSWarner Losh }
18418e3e3a7aSWarner Losh pos += size;
18428e3e3a7aSWarner Losh }
18438e3e3a7aSWarner Losh lua_pushinteger(L, pos + 1); /* next position */
18448e3e3a7aSWarner Losh return n + 1;
18458e3e3a7aSWarner Losh }
18468e3e3a7aSWarner Losh
18478e3e3a7aSWarner Losh /* }====================================================== */
18488e3e3a7aSWarner Losh
18498e3e3a7aSWarner Losh
18508e3e3a7aSWarner Losh static const luaL_Reg strlib[] = {
18518e3e3a7aSWarner Losh {"byte", str_byte},
18528e3e3a7aSWarner Losh {"char", str_char},
18538e3e3a7aSWarner Losh {"dump", str_dump},
18548e3e3a7aSWarner Losh {"find", str_find},
18558e3e3a7aSWarner Losh {"format", str_format},
18568e3e3a7aSWarner Losh {"gmatch", gmatch},
18578e3e3a7aSWarner Losh {"gsub", str_gsub},
18588e3e3a7aSWarner Losh {"len", str_len},
18598e3e3a7aSWarner Losh {"lower", str_lower},
18608e3e3a7aSWarner Losh {"match", str_match},
18618e3e3a7aSWarner Losh {"rep", str_rep},
18628e3e3a7aSWarner Losh {"reverse", str_reverse},
18638e3e3a7aSWarner Losh {"sub", str_sub},
18648e3e3a7aSWarner Losh {"upper", str_upper},
18658e3e3a7aSWarner Losh {"pack", str_pack},
18668e3e3a7aSWarner Losh {"packsize", str_packsize},
18678e3e3a7aSWarner Losh {"unpack", str_unpack},
18688e3e3a7aSWarner Losh {NULL, NULL}
18698e3e3a7aSWarner Losh };
18708e3e3a7aSWarner Losh
18718e3e3a7aSWarner Losh
18728e3e3a7aSWarner Losh static void createmetatable (lua_State *L) {
18730495ed39SKyle Evans /* table to be metatable for strings */
18740495ed39SKyle Evans luaL_newlibtable(L, stringmetamethods);
18750495ed39SKyle Evans luaL_setfuncs(L, stringmetamethods, 0);
18768e3e3a7aSWarner Losh lua_pushliteral(L, ""); /* dummy string */
18778e3e3a7aSWarner Losh lua_pushvalue(L, -2); /* copy table */
18788e3e3a7aSWarner Losh lua_setmetatable(L, -2); /* set table as metatable for strings */
18798e3e3a7aSWarner Losh lua_pop(L, 1); /* pop dummy string */
18808e3e3a7aSWarner Losh lua_pushvalue(L, -2); /* get string library */
18818e3e3a7aSWarner Losh lua_setfield(L, -2, "__index"); /* metatable.__index = string */
18828e3e3a7aSWarner Losh lua_pop(L, 1); /* pop metatable */
18838e3e3a7aSWarner Losh }
18848e3e3a7aSWarner Losh
18858e3e3a7aSWarner Losh
18868e3e3a7aSWarner Losh /*
18878e3e3a7aSWarner Losh ** Open string library
18888e3e3a7aSWarner Losh */
18898e3e3a7aSWarner Losh LUAMOD_API int luaopen_string (lua_State *L) {
18908e3e3a7aSWarner Losh luaL_newlib(L, strlib);
18918e3e3a7aSWarner Losh createmetatable(L);
18928e3e3a7aSWarner Losh return 1;
18938e3e3a7aSWarner Losh }
18948e3e3a7aSWarner Losh
1895