18e3e3a7aSWarner Losh /*
2*e112e9d2SKyle Evans ** $Id: lstring.c,v 2.56.1.1 2017/04/19 17:20:42 roberto Exp $
38e3e3a7aSWarner Losh ** String table (keeps all strings handled by Lua)
48e3e3a7aSWarner Losh ** See Copyright Notice in lua.h
58e3e3a7aSWarner Losh */
68e3e3a7aSWarner Losh
78e3e3a7aSWarner Losh #define lstring_c
88e3e3a7aSWarner Losh #define LUA_CORE
98e3e3a7aSWarner Losh
108e3e3a7aSWarner Losh #include "lprefix.h"
118e3e3a7aSWarner Losh
128e3e3a7aSWarner Losh
138e3e3a7aSWarner Losh #include <string.h>
148e3e3a7aSWarner Losh
158e3e3a7aSWarner Losh #include "lua.h"
168e3e3a7aSWarner Losh
178e3e3a7aSWarner Losh #include "ldebug.h"
188e3e3a7aSWarner Losh #include "ldo.h"
198e3e3a7aSWarner Losh #include "lmem.h"
208e3e3a7aSWarner Losh #include "lobject.h"
218e3e3a7aSWarner Losh #include "lstate.h"
228e3e3a7aSWarner Losh #include "lstring.h"
238e3e3a7aSWarner Losh
248e3e3a7aSWarner Losh
258e3e3a7aSWarner Losh #define MEMERRMSG "not enough memory"
268e3e3a7aSWarner Losh
278e3e3a7aSWarner Losh
288e3e3a7aSWarner Losh /*
298e3e3a7aSWarner Losh ** Lua will use at most ~(2^LUAI_HASHLIMIT) bytes from a string to
308e3e3a7aSWarner Losh ** compute its hash
318e3e3a7aSWarner Losh */
328e3e3a7aSWarner Losh #if !defined(LUAI_HASHLIMIT)
338e3e3a7aSWarner Losh #define LUAI_HASHLIMIT 5
348e3e3a7aSWarner Losh #endif
358e3e3a7aSWarner Losh
368e3e3a7aSWarner Losh
378e3e3a7aSWarner Losh /*
388e3e3a7aSWarner Losh ** equality for long strings
398e3e3a7aSWarner Losh */
luaS_eqlngstr(TString * a,TString * b)408e3e3a7aSWarner Losh int luaS_eqlngstr (TString *a, TString *b) {
418e3e3a7aSWarner Losh size_t len = a->u.lnglen;
428e3e3a7aSWarner Losh lua_assert(a->tt == LUA_TLNGSTR && b->tt == LUA_TLNGSTR);
438e3e3a7aSWarner Losh return (a == b) || /* same instance or... */
448e3e3a7aSWarner Losh ((len == b->u.lnglen) && /* equal length and ... */
458e3e3a7aSWarner Losh (memcmp(getstr(a), getstr(b), len) == 0)); /* equal contents */
468e3e3a7aSWarner Losh }
478e3e3a7aSWarner Losh
488e3e3a7aSWarner Losh
luaS_hash(const char * str,size_t l,unsigned int seed)498e3e3a7aSWarner Losh unsigned int luaS_hash (const char *str, size_t l, unsigned int seed) {
508e3e3a7aSWarner Losh unsigned int h = seed ^ cast(unsigned int, l);
518e3e3a7aSWarner Losh size_t step = (l >> LUAI_HASHLIMIT) + 1;
528e3e3a7aSWarner Losh for (; l >= step; l -= step)
538e3e3a7aSWarner Losh h ^= ((h<<5) + (h>>2) + cast_byte(str[l - 1]));
548e3e3a7aSWarner Losh return h;
558e3e3a7aSWarner Losh }
568e3e3a7aSWarner Losh
578e3e3a7aSWarner Losh
luaS_hashlongstr(TString * ts)588e3e3a7aSWarner Losh unsigned int luaS_hashlongstr (TString *ts) {
598e3e3a7aSWarner Losh lua_assert(ts->tt == LUA_TLNGSTR);
608e3e3a7aSWarner Losh if (ts->extra == 0) { /* no hash? */
618e3e3a7aSWarner Losh ts->hash = luaS_hash(getstr(ts), ts->u.lnglen, ts->hash);
628e3e3a7aSWarner Losh ts->extra = 1; /* now it has its hash */
638e3e3a7aSWarner Losh }
648e3e3a7aSWarner Losh return ts->hash;
658e3e3a7aSWarner Losh }
668e3e3a7aSWarner Losh
678e3e3a7aSWarner Losh
688e3e3a7aSWarner Losh /*
698e3e3a7aSWarner Losh ** resizes the string table
708e3e3a7aSWarner Losh */
luaS_resize(lua_State * L,int newsize)718e3e3a7aSWarner Losh void luaS_resize (lua_State *L, int newsize) {
728e3e3a7aSWarner Losh int i;
738e3e3a7aSWarner Losh stringtable *tb = &G(L)->strt;
748e3e3a7aSWarner Losh if (newsize > tb->size) { /* grow table if needed */
758e3e3a7aSWarner Losh luaM_reallocvector(L, tb->hash, tb->size, newsize, TString *);
768e3e3a7aSWarner Losh for (i = tb->size; i < newsize; i++)
778e3e3a7aSWarner Losh tb->hash[i] = NULL;
788e3e3a7aSWarner Losh }
798e3e3a7aSWarner Losh for (i = 0; i < tb->size; i++) { /* rehash */
808e3e3a7aSWarner Losh TString *p = tb->hash[i];
818e3e3a7aSWarner Losh tb->hash[i] = NULL;
828e3e3a7aSWarner Losh while (p) { /* for each node in the list */
838e3e3a7aSWarner Losh TString *hnext = p->u.hnext; /* save next */
848e3e3a7aSWarner Losh unsigned int h = lmod(p->hash, newsize); /* new position */
858e3e3a7aSWarner Losh p->u.hnext = tb->hash[h]; /* chain it */
868e3e3a7aSWarner Losh tb->hash[h] = p;
878e3e3a7aSWarner Losh p = hnext;
888e3e3a7aSWarner Losh }
898e3e3a7aSWarner Losh }
908e3e3a7aSWarner Losh if (newsize < tb->size) { /* shrink table if needed */
918e3e3a7aSWarner Losh /* vanishing slice should be empty */
928e3e3a7aSWarner Losh lua_assert(tb->hash[newsize] == NULL && tb->hash[tb->size - 1] == NULL);
938e3e3a7aSWarner Losh luaM_reallocvector(L, tb->hash, tb->size, newsize, TString *);
948e3e3a7aSWarner Losh }
958e3e3a7aSWarner Losh tb->size = newsize;
968e3e3a7aSWarner Losh }
978e3e3a7aSWarner Losh
988e3e3a7aSWarner Losh
998e3e3a7aSWarner Losh /*
1008e3e3a7aSWarner Losh ** Clear API string cache. (Entries cannot be empty, so fill them with
1018e3e3a7aSWarner Losh ** a non-collectable string.)
1028e3e3a7aSWarner Losh */
luaS_clearcache(global_State * g)1038e3e3a7aSWarner Losh void luaS_clearcache (global_State *g) {
1048e3e3a7aSWarner Losh int i, j;
1058e3e3a7aSWarner Losh for (i = 0; i < STRCACHE_N; i++)
1068e3e3a7aSWarner Losh for (j = 0; j < STRCACHE_M; j++) {
1078e3e3a7aSWarner Losh if (iswhite(g->strcache[i][j])) /* will entry be collected? */
1088e3e3a7aSWarner Losh g->strcache[i][j] = g->memerrmsg; /* replace it with something fixed */
1098e3e3a7aSWarner Losh }
1108e3e3a7aSWarner Losh }
1118e3e3a7aSWarner Losh
1128e3e3a7aSWarner Losh
1138e3e3a7aSWarner Losh /*
1148e3e3a7aSWarner Losh ** Initialize the string table and the string cache
1158e3e3a7aSWarner Losh */
luaS_init(lua_State * L)1168e3e3a7aSWarner Losh void luaS_init (lua_State *L) {
1178e3e3a7aSWarner Losh global_State *g = G(L);
1188e3e3a7aSWarner Losh int i, j;
1198e3e3a7aSWarner Losh luaS_resize(L, MINSTRTABSIZE); /* initial size of string table */
1208e3e3a7aSWarner Losh /* pre-create memory-error message */
1218e3e3a7aSWarner Losh g->memerrmsg = luaS_newliteral(L, MEMERRMSG);
1228e3e3a7aSWarner Losh luaC_fix(L, obj2gco(g->memerrmsg)); /* it should never be collected */
1238e3e3a7aSWarner Losh for (i = 0; i < STRCACHE_N; i++) /* fill cache with valid strings */
1248e3e3a7aSWarner Losh for (j = 0; j < STRCACHE_M; j++)
1258e3e3a7aSWarner Losh g->strcache[i][j] = g->memerrmsg;
1268e3e3a7aSWarner Losh }
1278e3e3a7aSWarner Losh
1288e3e3a7aSWarner Losh
1298e3e3a7aSWarner Losh
1308e3e3a7aSWarner Losh /*
1318e3e3a7aSWarner Losh ** creates a new string object
1328e3e3a7aSWarner Losh */
createstrobj(lua_State * L,size_t l,int tag,unsigned int h)1338e3e3a7aSWarner Losh static TString *createstrobj (lua_State *L, size_t l, int tag, unsigned int h) {
1348e3e3a7aSWarner Losh TString *ts;
1358e3e3a7aSWarner Losh GCObject *o;
1368e3e3a7aSWarner Losh size_t totalsize; /* total size of TString object */
1378e3e3a7aSWarner Losh totalsize = sizelstring(l);
1388e3e3a7aSWarner Losh o = luaC_newobj(L, tag, totalsize);
1398e3e3a7aSWarner Losh ts = gco2ts(o);
1408e3e3a7aSWarner Losh ts->hash = h;
1418e3e3a7aSWarner Losh ts->extra = 0;
1428e3e3a7aSWarner Losh getstr(ts)[l] = '\0'; /* ending 0 */
1438e3e3a7aSWarner Losh return ts;
1448e3e3a7aSWarner Losh }
1458e3e3a7aSWarner Losh
1468e3e3a7aSWarner Losh
luaS_createlngstrobj(lua_State * L,size_t l)1478e3e3a7aSWarner Losh TString *luaS_createlngstrobj (lua_State *L, size_t l) {
1488e3e3a7aSWarner Losh TString *ts = createstrobj(L, l, LUA_TLNGSTR, G(L)->seed);
1498e3e3a7aSWarner Losh ts->u.lnglen = l;
1508e3e3a7aSWarner Losh return ts;
1518e3e3a7aSWarner Losh }
1528e3e3a7aSWarner Losh
1538e3e3a7aSWarner Losh
luaS_remove(lua_State * L,TString * ts)1548e3e3a7aSWarner Losh void luaS_remove (lua_State *L, TString *ts) {
1558e3e3a7aSWarner Losh stringtable *tb = &G(L)->strt;
1568e3e3a7aSWarner Losh TString **p = &tb->hash[lmod(ts->hash, tb->size)];
1578e3e3a7aSWarner Losh while (*p != ts) /* find previous element */
1588e3e3a7aSWarner Losh p = &(*p)->u.hnext;
1598e3e3a7aSWarner Losh *p = (*p)->u.hnext; /* remove element from its list */
1608e3e3a7aSWarner Losh tb->nuse--;
1618e3e3a7aSWarner Losh }
1628e3e3a7aSWarner Losh
1638e3e3a7aSWarner Losh
1648e3e3a7aSWarner Losh /*
1658e3e3a7aSWarner Losh ** checks whether short string exists and reuses it or creates a new one
1668e3e3a7aSWarner Losh */
internshrstr(lua_State * L,const char * str,size_t l)1678e3e3a7aSWarner Losh static TString *internshrstr (lua_State *L, const char *str, size_t l) {
1688e3e3a7aSWarner Losh TString *ts;
1698e3e3a7aSWarner Losh global_State *g = G(L);
1708e3e3a7aSWarner Losh unsigned int h = luaS_hash(str, l, g->seed);
1718e3e3a7aSWarner Losh TString **list = &g->strt.hash[lmod(h, g->strt.size)];
1728e3e3a7aSWarner Losh lua_assert(str != NULL); /* otherwise 'memcmp'/'memcpy' are undefined */
1738e3e3a7aSWarner Losh for (ts = *list; ts != NULL; ts = ts->u.hnext) {
1748e3e3a7aSWarner Losh if (l == ts->shrlen &&
1758e3e3a7aSWarner Losh (memcmp(str, getstr(ts), l * sizeof(char)) == 0)) {
1768e3e3a7aSWarner Losh /* found! */
1778e3e3a7aSWarner Losh if (isdead(g, ts)) /* dead (but not collected yet)? */
1788e3e3a7aSWarner Losh changewhite(ts); /* resurrect it */
1798e3e3a7aSWarner Losh return ts;
1808e3e3a7aSWarner Losh }
1818e3e3a7aSWarner Losh }
1828e3e3a7aSWarner Losh if (g->strt.nuse >= g->strt.size && g->strt.size <= MAX_INT/2) {
1838e3e3a7aSWarner Losh luaS_resize(L, g->strt.size * 2);
1848e3e3a7aSWarner Losh list = &g->strt.hash[lmod(h, g->strt.size)]; /* recompute with new size */
1858e3e3a7aSWarner Losh }
1868e3e3a7aSWarner Losh ts = createstrobj(L, l, LUA_TSHRSTR, h);
1878e3e3a7aSWarner Losh memcpy(getstr(ts), str, l * sizeof(char));
1888e3e3a7aSWarner Losh ts->shrlen = cast_byte(l);
1898e3e3a7aSWarner Losh ts->u.hnext = *list;
1908e3e3a7aSWarner Losh *list = ts;
1918e3e3a7aSWarner Losh g->strt.nuse++;
1928e3e3a7aSWarner Losh return ts;
1938e3e3a7aSWarner Losh }
1948e3e3a7aSWarner Losh
1958e3e3a7aSWarner Losh
1968e3e3a7aSWarner Losh /*
1978e3e3a7aSWarner Losh ** new string (with explicit length)
1988e3e3a7aSWarner Losh */
luaS_newlstr(lua_State * L,const char * str,size_t l)1998e3e3a7aSWarner Losh TString *luaS_newlstr (lua_State *L, const char *str, size_t l) {
2008e3e3a7aSWarner Losh if (l <= LUAI_MAXSHORTLEN) /* short string? */
2018e3e3a7aSWarner Losh return internshrstr(L, str, l);
2028e3e3a7aSWarner Losh else {
2038e3e3a7aSWarner Losh TString *ts;
2048e3e3a7aSWarner Losh if (l >= (MAX_SIZE - sizeof(TString))/sizeof(char))
2058e3e3a7aSWarner Losh luaM_toobig(L);
2068e3e3a7aSWarner Losh ts = luaS_createlngstrobj(L, l);
2078e3e3a7aSWarner Losh memcpy(getstr(ts), str, l * sizeof(char));
2088e3e3a7aSWarner Losh return ts;
2098e3e3a7aSWarner Losh }
2108e3e3a7aSWarner Losh }
2118e3e3a7aSWarner Losh
2128e3e3a7aSWarner Losh
2138e3e3a7aSWarner Losh /*
2148e3e3a7aSWarner Losh ** Create or reuse a zero-terminated string, first checking in the
2158e3e3a7aSWarner Losh ** cache (using the string address as a key). The cache can contain
2168e3e3a7aSWarner Losh ** only zero-terminated strings, so it is safe to use 'strcmp' to
2178e3e3a7aSWarner Losh ** check hits.
2188e3e3a7aSWarner Losh */
luaS_new(lua_State * L,const char * str)2198e3e3a7aSWarner Losh TString *luaS_new (lua_State *L, const char *str) {
2208e3e3a7aSWarner Losh unsigned int i = point2uint(str) % STRCACHE_N; /* hash */
2218e3e3a7aSWarner Losh int j;
2228e3e3a7aSWarner Losh TString **p = G(L)->strcache[i];
2238e3e3a7aSWarner Losh for (j = 0; j < STRCACHE_M; j++) {
2248e3e3a7aSWarner Losh if (strcmp(str, getstr(p[j])) == 0) /* hit? */
2258e3e3a7aSWarner Losh return p[j]; /* that is it */
2268e3e3a7aSWarner Losh }
2278e3e3a7aSWarner Losh /* normal route */
2288e3e3a7aSWarner Losh for (j = STRCACHE_M - 1; j > 0; j--)
2298e3e3a7aSWarner Losh p[j] = p[j - 1]; /* move out last element */
2308e3e3a7aSWarner Losh /* new element is first in the list */
2318e3e3a7aSWarner Losh p[0] = luaS_newlstr(L, str, strlen(str));
2328e3e3a7aSWarner Losh return p[0];
2338e3e3a7aSWarner Losh }
2348e3e3a7aSWarner Losh
2358e3e3a7aSWarner Losh
luaS_newudata(lua_State * L,size_t s)2368e3e3a7aSWarner Losh Udata *luaS_newudata (lua_State *L, size_t s) {
2378e3e3a7aSWarner Losh Udata *u;
2388e3e3a7aSWarner Losh GCObject *o;
2398e3e3a7aSWarner Losh if (s > MAX_SIZE - sizeof(Udata))
2408e3e3a7aSWarner Losh luaM_toobig(L);
2418e3e3a7aSWarner Losh o = luaC_newobj(L, LUA_TUSERDATA, sizeludata(s));
2428e3e3a7aSWarner Losh u = gco2u(o);
2438e3e3a7aSWarner Losh u->len = s;
2448e3e3a7aSWarner Losh u->metatable = NULL;
2458e3e3a7aSWarner Losh setuservalue(L, u, luaO_nilobject);
2468e3e3a7aSWarner Losh return u;
2478e3e3a7aSWarner Losh }
2488e3e3a7aSWarner Losh
249