18e3e3a7aSWarner Losh /*
2*0495ed39SKyle Evans ** $Id: lstring.c $
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 /*
26*0495ed39SKyle Evans ** Maximum size for string table.
278e3e3a7aSWarner Losh */
28*0495ed39SKyle Evans #define MAXSTRTB cast_int(luaM_limitN(MAX_INT, TString*))
298e3e3a7aSWarner Losh
308e3e3a7aSWarner Losh
318e3e3a7aSWarner Losh /*
328e3e3a7aSWarner Losh ** equality for long strings
338e3e3a7aSWarner Losh */
luaS_eqlngstr(TString * a,TString * b)348e3e3a7aSWarner Losh int luaS_eqlngstr (TString *a, TString *b) {
358e3e3a7aSWarner Losh size_t len = a->u.lnglen;
36*0495ed39SKyle Evans lua_assert(a->tt == LUA_VLNGSTR && b->tt == LUA_VLNGSTR);
378e3e3a7aSWarner Losh return (a == b) || /* same instance or... */
388e3e3a7aSWarner Losh ((len == b->u.lnglen) && /* equal length and ... */
398e3e3a7aSWarner Losh (memcmp(getstr(a), getstr(b), len) == 0)); /* equal contents */
408e3e3a7aSWarner Losh }
418e3e3a7aSWarner Losh
428e3e3a7aSWarner Losh
luaS_hash(const char * str,size_t l,unsigned int seed)438e3e3a7aSWarner Losh unsigned int luaS_hash (const char *str, size_t l, unsigned int seed) {
44*0495ed39SKyle Evans unsigned int h = seed ^ cast_uint(l);
45*0495ed39SKyle Evans for (; l > 0; l--)
468e3e3a7aSWarner Losh h ^= ((h<<5) + (h>>2) + cast_byte(str[l - 1]));
478e3e3a7aSWarner Losh return h;
488e3e3a7aSWarner Losh }
498e3e3a7aSWarner Losh
508e3e3a7aSWarner Losh
luaS_hashlongstr(TString * ts)518e3e3a7aSWarner Losh unsigned int luaS_hashlongstr (TString *ts) {
52*0495ed39SKyle Evans lua_assert(ts->tt == LUA_VLNGSTR);
538e3e3a7aSWarner Losh if (ts->extra == 0) { /* no hash? */
54*0495ed39SKyle Evans size_t len = ts->u.lnglen;
55*0495ed39SKyle Evans ts->hash = luaS_hash(getstr(ts), len, ts->hash);
568e3e3a7aSWarner Losh ts->extra = 1; /* now it has its hash */
578e3e3a7aSWarner Losh }
588e3e3a7aSWarner Losh return ts->hash;
598e3e3a7aSWarner Losh }
608e3e3a7aSWarner Losh
618e3e3a7aSWarner Losh
tablerehash(TString ** vect,int osize,int nsize)62*0495ed39SKyle Evans static void tablerehash (TString **vect, int osize, int nsize) {
638e3e3a7aSWarner Losh int i;
64*0495ed39SKyle Evans for (i = osize; i < nsize; i++) /* clear new elements */
65*0495ed39SKyle Evans vect[i] = NULL;
66*0495ed39SKyle Evans for (i = 0; i < osize; i++) { /* rehash old part of the array */
67*0495ed39SKyle Evans TString *p = vect[i];
68*0495ed39SKyle Evans vect[i] = NULL;
69*0495ed39SKyle Evans while (p) { /* for each string in the list */
708e3e3a7aSWarner Losh TString *hnext = p->u.hnext; /* save next */
71*0495ed39SKyle Evans unsigned int h = lmod(p->hash, nsize); /* new position */
72*0495ed39SKyle Evans p->u.hnext = vect[h]; /* chain it into array */
73*0495ed39SKyle Evans vect[h] = p;
748e3e3a7aSWarner Losh p = hnext;
758e3e3a7aSWarner Losh }
768e3e3a7aSWarner Losh }
778e3e3a7aSWarner Losh }
78*0495ed39SKyle Evans
79*0495ed39SKyle Evans
80*0495ed39SKyle Evans /*
81*0495ed39SKyle Evans ** Resize the string table. If allocation fails, keep the current size.
82*0495ed39SKyle Evans ** (This can degrade performance, but any non-zero size should work
83*0495ed39SKyle Evans ** correctly.)
84*0495ed39SKyle Evans */
luaS_resize(lua_State * L,int nsize)85*0495ed39SKyle Evans void luaS_resize (lua_State *L, int nsize) {
86*0495ed39SKyle Evans stringtable *tb = &G(L)->strt;
87*0495ed39SKyle Evans int osize = tb->size;
88*0495ed39SKyle Evans TString **newvect;
89*0495ed39SKyle Evans if (nsize < osize) /* shrinking table? */
90*0495ed39SKyle Evans tablerehash(tb->hash, osize, nsize); /* depopulate shrinking part */
91*0495ed39SKyle Evans newvect = luaM_reallocvector(L, tb->hash, osize, nsize, TString*);
92*0495ed39SKyle Evans if (unlikely(newvect == NULL)) { /* reallocation failed? */
93*0495ed39SKyle Evans if (nsize < osize) /* was it shrinking table? */
94*0495ed39SKyle Evans tablerehash(tb->hash, nsize, osize); /* restore to original size */
95*0495ed39SKyle Evans /* leave table as it was */
96*0495ed39SKyle Evans }
97*0495ed39SKyle Evans else { /* allocation succeeded */
98*0495ed39SKyle Evans tb->hash = newvect;
99*0495ed39SKyle Evans tb->size = nsize;
100*0495ed39SKyle Evans if (nsize > osize)
101*0495ed39SKyle Evans tablerehash(newvect, osize, nsize); /* rehash for new size */
102*0495ed39SKyle Evans }
1038e3e3a7aSWarner Losh }
1048e3e3a7aSWarner Losh
1058e3e3a7aSWarner Losh
1068e3e3a7aSWarner Losh /*
1078e3e3a7aSWarner Losh ** Clear API string cache. (Entries cannot be empty, so fill them with
1088e3e3a7aSWarner Losh ** a non-collectable string.)
1098e3e3a7aSWarner Losh */
luaS_clearcache(global_State * g)1108e3e3a7aSWarner Losh void luaS_clearcache (global_State *g) {
1118e3e3a7aSWarner Losh int i, j;
1128e3e3a7aSWarner Losh for (i = 0; i < STRCACHE_N; i++)
1138e3e3a7aSWarner Losh for (j = 0; j < STRCACHE_M; j++) {
1148e3e3a7aSWarner Losh if (iswhite(g->strcache[i][j])) /* will entry be collected? */
1158e3e3a7aSWarner Losh g->strcache[i][j] = g->memerrmsg; /* replace it with something fixed */
1168e3e3a7aSWarner Losh }
1178e3e3a7aSWarner Losh }
1188e3e3a7aSWarner Losh
1198e3e3a7aSWarner Losh
1208e3e3a7aSWarner Losh /*
1218e3e3a7aSWarner Losh ** Initialize the string table and the string cache
1228e3e3a7aSWarner Losh */
luaS_init(lua_State * L)1238e3e3a7aSWarner Losh void luaS_init (lua_State *L) {
1248e3e3a7aSWarner Losh global_State *g = G(L);
1258e3e3a7aSWarner Losh int i, j;
126*0495ed39SKyle Evans stringtable *tb = &G(L)->strt;
127*0495ed39SKyle Evans tb->hash = luaM_newvector(L, MINSTRTABSIZE, TString*);
128*0495ed39SKyle Evans tablerehash(tb->hash, 0, MINSTRTABSIZE); /* clear array */
129*0495ed39SKyle Evans tb->size = MINSTRTABSIZE;
1308e3e3a7aSWarner Losh /* pre-create memory-error message */
1318e3e3a7aSWarner Losh g->memerrmsg = luaS_newliteral(L, MEMERRMSG);
1328e3e3a7aSWarner Losh luaC_fix(L, obj2gco(g->memerrmsg)); /* it should never be collected */
1338e3e3a7aSWarner Losh for (i = 0; i < STRCACHE_N; i++) /* fill cache with valid strings */
1348e3e3a7aSWarner Losh for (j = 0; j < STRCACHE_M; j++)
1358e3e3a7aSWarner Losh g->strcache[i][j] = g->memerrmsg;
1368e3e3a7aSWarner Losh }
1378e3e3a7aSWarner Losh
1388e3e3a7aSWarner Losh
1398e3e3a7aSWarner Losh
1408e3e3a7aSWarner Losh /*
1418e3e3a7aSWarner Losh ** creates a new string object
1428e3e3a7aSWarner Losh */
createstrobj(lua_State * L,size_t l,int tag,unsigned int h)1438e3e3a7aSWarner Losh static TString *createstrobj (lua_State *L, size_t l, int tag, unsigned int h) {
1448e3e3a7aSWarner Losh TString *ts;
1458e3e3a7aSWarner Losh GCObject *o;
1468e3e3a7aSWarner Losh size_t totalsize; /* total size of TString object */
1478e3e3a7aSWarner Losh totalsize = sizelstring(l);
1488e3e3a7aSWarner Losh o = luaC_newobj(L, tag, totalsize);
1498e3e3a7aSWarner Losh ts = gco2ts(o);
1508e3e3a7aSWarner Losh ts->hash = h;
1518e3e3a7aSWarner Losh ts->extra = 0;
1528e3e3a7aSWarner Losh getstr(ts)[l] = '\0'; /* ending 0 */
1538e3e3a7aSWarner Losh return ts;
1548e3e3a7aSWarner Losh }
1558e3e3a7aSWarner Losh
1568e3e3a7aSWarner Losh
luaS_createlngstrobj(lua_State * L,size_t l)1578e3e3a7aSWarner Losh TString *luaS_createlngstrobj (lua_State *L, size_t l) {
158*0495ed39SKyle Evans TString *ts = createstrobj(L, l, LUA_VLNGSTR, G(L)->seed);
1598e3e3a7aSWarner Losh ts->u.lnglen = l;
1608e3e3a7aSWarner Losh return ts;
1618e3e3a7aSWarner Losh }
1628e3e3a7aSWarner Losh
1638e3e3a7aSWarner Losh
luaS_remove(lua_State * L,TString * ts)1648e3e3a7aSWarner Losh void luaS_remove (lua_State *L, TString *ts) {
1658e3e3a7aSWarner Losh stringtable *tb = &G(L)->strt;
1668e3e3a7aSWarner Losh TString **p = &tb->hash[lmod(ts->hash, tb->size)];
1678e3e3a7aSWarner Losh while (*p != ts) /* find previous element */
1688e3e3a7aSWarner Losh p = &(*p)->u.hnext;
1698e3e3a7aSWarner Losh *p = (*p)->u.hnext; /* remove element from its list */
1708e3e3a7aSWarner Losh tb->nuse--;
1718e3e3a7aSWarner Losh }
1728e3e3a7aSWarner Losh
1738e3e3a7aSWarner Losh
growstrtab(lua_State * L,stringtable * tb)174*0495ed39SKyle Evans static void growstrtab (lua_State *L, stringtable *tb) {
175*0495ed39SKyle Evans if (unlikely(tb->nuse == MAX_INT)) { /* too many strings? */
176*0495ed39SKyle Evans luaC_fullgc(L, 1); /* try to free some... */
177*0495ed39SKyle Evans if (tb->nuse == MAX_INT) /* still too many? */
178*0495ed39SKyle Evans luaM_error(L); /* cannot even create a message... */
179*0495ed39SKyle Evans }
180*0495ed39SKyle Evans if (tb->size <= MAXSTRTB / 2) /* can grow string table? */
181*0495ed39SKyle Evans luaS_resize(L, tb->size * 2);
182*0495ed39SKyle Evans }
183*0495ed39SKyle Evans
184*0495ed39SKyle Evans
1858e3e3a7aSWarner Losh /*
186*0495ed39SKyle Evans ** Checks whether short string exists and reuses it or creates a new one.
1878e3e3a7aSWarner Losh */
internshrstr(lua_State * L,const char * str,size_t l)1888e3e3a7aSWarner Losh static TString *internshrstr (lua_State *L, const char *str, size_t l) {
1898e3e3a7aSWarner Losh TString *ts;
1908e3e3a7aSWarner Losh global_State *g = G(L);
191*0495ed39SKyle Evans stringtable *tb = &g->strt;
1928e3e3a7aSWarner Losh unsigned int h = luaS_hash(str, l, g->seed);
193*0495ed39SKyle Evans TString **list = &tb->hash[lmod(h, tb->size)];
1948e3e3a7aSWarner Losh lua_assert(str != NULL); /* otherwise 'memcmp'/'memcpy' are undefined */
1958e3e3a7aSWarner Losh for (ts = *list; ts != NULL; ts = ts->u.hnext) {
196*0495ed39SKyle Evans if (l == ts->shrlen && (memcmp(str, getstr(ts), l * sizeof(char)) == 0)) {
1978e3e3a7aSWarner Losh /* found! */
1988e3e3a7aSWarner Losh if (isdead(g, ts)) /* dead (but not collected yet)? */
1998e3e3a7aSWarner Losh changewhite(ts); /* resurrect it */
2008e3e3a7aSWarner Losh return ts;
2018e3e3a7aSWarner Losh }
2028e3e3a7aSWarner Losh }
203*0495ed39SKyle Evans /* else must create a new string */
204*0495ed39SKyle Evans if (tb->nuse >= tb->size) { /* need to grow string table? */
205*0495ed39SKyle Evans growstrtab(L, tb);
206*0495ed39SKyle Evans list = &tb->hash[lmod(h, tb->size)]; /* rehash with new size */
2078e3e3a7aSWarner Losh }
208*0495ed39SKyle Evans ts = createstrobj(L, l, LUA_VSHRSTR, h);
2098e3e3a7aSWarner Losh memcpy(getstr(ts), str, l * sizeof(char));
2108e3e3a7aSWarner Losh ts->shrlen = cast_byte(l);
2118e3e3a7aSWarner Losh ts->u.hnext = *list;
2128e3e3a7aSWarner Losh *list = ts;
213*0495ed39SKyle Evans tb->nuse++;
2148e3e3a7aSWarner Losh return ts;
2158e3e3a7aSWarner Losh }
2168e3e3a7aSWarner Losh
2178e3e3a7aSWarner Losh
2188e3e3a7aSWarner Losh /*
2198e3e3a7aSWarner Losh ** new string (with explicit length)
2208e3e3a7aSWarner Losh */
luaS_newlstr(lua_State * L,const char * str,size_t l)2218e3e3a7aSWarner Losh TString *luaS_newlstr (lua_State *L, const char *str, size_t l) {
2228e3e3a7aSWarner Losh if (l <= LUAI_MAXSHORTLEN) /* short string? */
2238e3e3a7aSWarner Losh return internshrstr(L, str, l);
2248e3e3a7aSWarner Losh else {
2258e3e3a7aSWarner Losh TString *ts;
226*0495ed39SKyle Evans if (unlikely(l >= (MAX_SIZE - sizeof(TString))/sizeof(char)))
2278e3e3a7aSWarner Losh luaM_toobig(L);
2288e3e3a7aSWarner Losh ts = luaS_createlngstrobj(L, l);
2298e3e3a7aSWarner Losh memcpy(getstr(ts), str, l * sizeof(char));
2308e3e3a7aSWarner Losh return ts;
2318e3e3a7aSWarner Losh }
2328e3e3a7aSWarner Losh }
2338e3e3a7aSWarner Losh
2348e3e3a7aSWarner Losh
2358e3e3a7aSWarner Losh /*
2368e3e3a7aSWarner Losh ** Create or reuse a zero-terminated string, first checking in the
2378e3e3a7aSWarner Losh ** cache (using the string address as a key). The cache can contain
2388e3e3a7aSWarner Losh ** only zero-terminated strings, so it is safe to use 'strcmp' to
2398e3e3a7aSWarner Losh ** check hits.
2408e3e3a7aSWarner Losh */
luaS_new(lua_State * L,const char * str)2418e3e3a7aSWarner Losh TString *luaS_new (lua_State *L, const char *str) {
2428e3e3a7aSWarner Losh unsigned int i = point2uint(str) % STRCACHE_N; /* hash */
2438e3e3a7aSWarner Losh int j;
2448e3e3a7aSWarner Losh TString **p = G(L)->strcache[i];
2458e3e3a7aSWarner Losh for (j = 0; j < STRCACHE_M; j++) {
2468e3e3a7aSWarner Losh if (strcmp(str, getstr(p[j])) == 0) /* hit? */
2478e3e3a7aSWarner Losh return p[j]; /* that is it */
2488e3e3a7aSWarner Losh }
2498e3e3a7aSWarner Losh /* normal route */
2508e3e3a7aSWarner Losh for (j = STRCACHE_M - 1; j > 0; j--)
2518e3e3a7aSWarner Losh p[j] = p[j - 1]; /* move out last element */
2528e3e3a7aSWarner Losh /* new element is first in the list */
2538e3e3a7aSWarner Losh p[0] = luaS_newlstr(L, str, strlen(str));
2548e3e3a7aSWarner Losh return p[0];
2558e3e3a7aSWarner Losh }
2568e3e3a7aSWarner Losh
2578e3e3a7aSWarner Losh
luaS_newudata(lua_State * L,size_t s,int nuvalue)258*0495ed39SKyle Evans Udata *luaS_newudata (lua_State *L, size_t s, int nuvalue) {
2598e3e3a7aSWarner Losh Udata *u;
260*0495ed39SKyle Evans int i;
2618e3e3a7aSWarner Losh GCObject *o;
262*0495ed39SKyle Evans if (unlikely(s > MAX_SIZE - udatamemoffset(nuvalue)))
2638e3e3a7aSWarner Losh luaM_toobig(L);
264*0495ed39SKyle Evans o = luaC_newobj(L, LUA_VUSERDATA, sizeudata(nuvalue, s));
2658e3e3a7aSWarner Losh u = gco2u(o);
2668e3e3a7aSWarner Losh u->len = s;
267*0495ed39SKyle Evans u->nuvalue = nuvalue;
2688e3e3a7aSWarner Losh u->metatable = NULL;
269*0495ed39SKyle Evans for (i = 0; i < nuvalue; i++)
270*0495ed39SKyle Evans setnilvalue(&u->uv[i].uv);
2718e3e3a7aSWarner Losh return u;
2728e3e3a7aSWarner Losh }
2738e3e3a7aSWarner Losh
274