source/libs/luajit/LuaJIT-2.0.4/src/lj_str.c

   1 /*
   2 ** String handling.
   3 ** Copyright (C) 2005-2015 Mike Pall. See Copyright Notice in luajit.h
   4 **
   5 ** Portions taken verbatim or adapted from the Lua interpreter.
   6 ** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
   7 */
   8
   9 #include <stdio.h>
  10
  11 #define lj_str_c
  12 #define LUA_CORE
  13
  14 #include "lj_obj.h"
  15 #include "lj_gc.h"
  16 #include "lj_err.h"
  17 #include "lj_str.h"
  18 #include "lj_state.h"
  19 #include "lj_char.h"
  20
  21 /* -- String interning ---------------------------------------------------- */
  22
  23 /* Ordered compare of strings. Assumes string data is 4-byte aligned. */
  24 int32_t LJ_FASTCALL lj_str_cmp(GCstr *a, GCstr *b)
  25 {
  26   MSize i, n = a->len > b->len ? b->len : a->len;
  27   for (i = 0; i < n; i += 4) {
  28     /* Note: innocuous access up to end of string + 3. */
  29     uint32_t va = *(const uint32_t *)(strdata(a)+i);
  30     uint32_t vb = *(const uint32_t *)(strdata(b)+i);
  31     if (va != vb) {
  32 #if LJ_LE
  33       va = lj_bswap(va); vb = lj_bswap(vb);
  34 #endif
  35       i -= n;
  36       if ((int32_t)i >= -3) {
  37         va >>= 32+(i<<3); vb >>= 32+(i<<3);
  38         if (va == vb) break;
  39       }
  40       return va < vb ? -1 : 1;
  41     }
  42   }
  43   return (int32_t)(a->len - b->len);
  44 }
  45
  46 /* Fast string data comparison. Caveat: unaligned access to 1st string! */
  47 static LJ_AINLINE int str_fastcmp(const char *a, const char *b, MSize len)
  48 {
  49   MSize i = 0;
  50   lua_assert(len > 0);
  51   lua_assert((((uintptr_t)a+len-1) & (LJ_PAGESIZE-1)) <= LJ_PAGESIZE-4);
  52   do {  /* Note: innocuous access up to end of string + 3. */
  53     uint32_t v = lj_getu32(a+i) ^ *(const uint32_t *)(b+i);
  54     if (v) {
  55       i -= len;
  56 #if LJ_LE
  57       return (int32_t)i >= -3 ? (v << (32+(i<<3))) : 1;
  58 #else
  59       return (int32_t)i >= -3 ? (v >> (32+(i<<3))) : 1;
  60 #endif
  61     }
  62     i += 4;
  63   } while (i < len);
  64   return 0;
  65 }
  66
  67 /* Resize the string hash table (grow and shrink). */
  68 void lj_str_resize(lua_State *L, MSize newmask)
  69 {
  70   global_State *g = G(L);
  71   GCRef *newhash;
  72   MSize i;
  73   if (g->gc.state == GCSsweepstring || newmask >= LJ_MAX_STRTAB-1)
  74     return;  /* No resizing during GC traversal or if already too big. */
  75   newhash = lj_mem_newvec(L, newmask+1, GCRef);
  76   memset(newhash, 0, (newmask+1)*sizeof(GCRef));
  77   for (i = g->strmask; i != ~(MSize)0; i--) {  /* Rehash old table. */
  78     GCobj *p = gcref(g->strhash[i]);
  79     while (p) {  /* Follow each hash chain and reinsert all strings. */
  80       MSize h = gco2str(p)->hash & newmask;
  81       GCobj *next = gcnext(p);
  82       /* NOBARRIER: The string table is a GC root. */
  83       setgcrefr(p->gch.nextgc, newhash[h]);
  84       setgcref(newhash[h], p);
  85       p = next;
  86     }
  87   }
  88   lj_mem_freevec(g, g->strhash, g->strmask+1, GCRef);
  89   g->strmask = newmask;
  90   g->strhash = newhash;
  91 }
  92
  93 /*
  94 ** Lua will use at most ~(2^LUAI_HASHLIMIT) bytes from a string to
  95 ** compute its hash
  96 */
  97 #if !defined(LUAI_HASHLIMIT)
  98 #define LUAI_HASHLIMIT          5
  99 #endif
 100
 101 #define cast(t, exp)    ((t)(exp))
 102 int luajittex_choose_hash_function = 0 ;
 103 /* Intern a string and return string object. */
 104 GCstr *lj_str_new(lua_State *L, const char *str, size_t lenx)
 105 {
 106   global_State *g;
 107   GCstr *s;
 108   GCobj *o;
 109   MSize len = (MSize)lenx;
 110   MSize a, b, h = len;
 111   size_t step ;
 112   size_t l1 ;
 113   if (lenx >= LJ_MAX_STR)
 114     lj_err_msg(L, LJ_ERR_STROV);
 115   g = G(L);
 116
 117   if (len==0)
 118     return &g->strempty;
 119   if (luajittex_choose_hash_function==0) {
 120     /* Lua 5.1.5 hash function */
 121     /* for 5.2 max methods we also need to patch the vm eq */
 122     step = (len>>LUAI_HASHLIMIT)+1;  /* if string is too long, don't hash all its chars */
 123     for (l1=len; l1>=step; l1-=step)  /* compute hash */
 124       h = h ^ ((h<<5)+(h>>2)+cast(unsigned char, str[l1-1]));
 125    } else {
 126   /* LuaJIT 2.0.2 hash function */
 127   /* Compute string hash. Constants taken from lookup3 hash by Bob Jenkins. */
 128     if (len >= 4) {  /* Caveat: unaligned access! */
 129       a = lj_getu32(str);
 130       h ^= lj_getu32(str+len-4);
 131       b = lj_getu32(str+(len>>1)-2);
 132       h ^= b; h -= lj_rol(b, 14);
 133       b += lj_getu32(str+(len>>2)-1);
 134     } else if (len > 0) {
 135       a = *(const uint8_t *)str;
 136       h ^= *(const uint8_t *)(str+len-1);
 137       b = *(const uint8_t *)(str+(len>>1));
 138       h ^= b; h -= lj_rol(b, 14);
 139     } else {
 140        /* Already done, kept for reference */
 141        return &g->strempty;
 142     }
 143     a ^= h; a -= lj_rol(h, 11);
 144     b ^= a; b -= lj_rol(a, 25);
 145     h ^= b; h -= lj_rol(b, 16);
 146   }
 147
 148
 149   /* Check if the string has already been interned. */
 150   o = gcref(g->strhash[h & g->strmask]);
 151   if (LJ_LIKELY((((uintptr_t)str+len-1) & (LJ_PAGESIZE-1)) <= LJ_PAGESIZE-4)) {
 152     while (o != NULL) {
 153       GCstr *sx = gco2str(o);
 154       if (sx->len == len && str_fastcmp(str, strdata(sx), len) == 0) {
 155         /* Resurrect if dead. Can only happen with fixstring() (keywords). */
 156         if (isdead(g, o)) flipwhite(o);
 157         return sx;  /* Return existing string. */
 158       }
 159       o = gcnext(o);
 160     }
 161   } else {  /* Slow path: end of string is too close to a page boundary. */
 162     while (o != NULL) {
 163       GCstr *sx = gco2str(o);
 164       if (sx->len == len && memcmp(str, strdata(sx), len) == 0) {
 165         /* Resurrect if dead. Can only happen with fixstring() (keywords). */
 166         if (isdead(g, o)) flipwhite(o);
 167         return sx;  /* Return existing string. */
 168       }
 169       o = gcnext(o);
 170     }
 171   }
 172   /* Nope, create a new string. */
 173   s = lj_mem_newt(L, sizeof(GCstr)+len+1, GCstr);
 174   newwhite(g, s);
 175   s->gct = ~LJ_TSTR;
 176   s->len = len;
 177   s->hash = h;
 178   s->reserved = 0;
 179   memcpy(strdatawr(s), str, len);
 180   strdatawr(s)[len] = '\0';  /* Zero-terminate string. */
 181   /* Add it to string hash table. */
 182   h &= g->strmask;
 183   s->nextgc = g->strhash[h];
 184   /* NOBARRIER: The string table is a GC root. */
 185   setgcref(g->strhash[h], obj2gco(s));
 186   if (g->strnum++ > g->strmask)  /* Allow a 100% load factor. */
 187     lj_str_resize(L, (g->strmask<<1)+1);  /* Grow string table. */
 188   return s;  /* Return newly interned string. */
 189 }
 190
 191 void LJ_FASTCALL lj_str_free(global_State *g, GCstr *s)
 192 {
 193   g->strnum--;
 194   lj_mem_free(g, s, sizestring(s));
 195 }
 196
 197 /* -- Type conversions ---------------------------------------------------- */
 198
 199 /* Print number to buffer. Canonicalizes non-finite values. */
 200 size_t LJ_FASTCALL lj_str_bufnum(char *s, cTValue *o)
 201 {
 202   if (LJ_LIKELY((o->u32.hi << 1) < 0xffe00000)) {  /* Finite? */
 203     lua_Number n = o->n;
 204 #if __BIONIC__
 205     if (tvismzero(o)) { s[0] = '-'; s[1] = '0'; return 2; }
 206 #endif
 207     return (size_t)lua_number2str(s, n);
 208   } else if (((o->u32.hi & 0x000fffff) | o->u32.lo) != 0) {
 209     s[0] = 'n'; s[1] = 'a'; s[2] = 'n'; return 3;
 210   } else if ((o->u32.hi & 0x80000000) == 0) {
 211     s[0] = 'i'; s[1] = 'n'; s[2] = 'f'; return 3;
 212   } else {
 213     s[0] = '-'; s[1] = 'i'; s[2] = 'n'; s[3] = 'f'; return 4;
 214   }
 215 }
 216
 217 /* Print integer to buffer. Returns pointer to start. */
 218 char * LJ_FASTCALL lj_str_bufint(char *p, int32_t k)
 219 {
 220   uint32_t u = (uint32_t)(k < 0 ? -k : k);
 221   p += 1+10;
 222   do { *--p = (char)('0' + u % 10); } while (u /= 10);
 223   if (k < 0) *--p = '-';
 224   return p;
 225 }
 226
 227 /* Convert number to string. */
 228 GCstr * LJ_FASTCALL lj_str_fromnum(lua_State *L, const lua_Number *np)
 229 {
 230   char buf[LJ_STR_NUMBUF];
 231   size_t len = lj_str_bufnum(buf, (TValue *)np);
 232   return lj_str_new(L, buf, len);
 233 }
 234
 235 /* Convert integer to string. */
 236 GCstr * LJ_FASTCALL lj_str_fromint(lua_State *L, int32_t k)
 237 {
 238   char s[1+10];
 239   char *p = lj_str_bufint(s, k);
 240   return lj_str_new(L, p, (size_t)(s+sizeof(s)-p));
 241 }
 242
 243 GCstr * LJ_FASTCALL lj_str_fromnumber(lua_State *L, cTValue *o)
 244 {
 245   return tvisint(o) ? lj_str_fromint(L, intV(o)) : lj_str_fromnum(L, &o->n);
 246 }
 247
 248 /* -- String formatting --------------------------------------------------- */
 249
 250 static void addstr(lua_State *L, SBuf *sb, const char *str, MSize len)
 251 {
 252   char *p;
 253   MSize i;
 254   if (sb->n + len > sb->sz) {
 255     MSize sz = sb->sz * 2;
 256     while (sb->n + len > sz) sz = sz * 2;
 257     lj_str_resizebuf(L, sb, sz);
 258   }
 259   p = sb->buf + sb->n;
 260   sb->n += len;
 261   for (i = 0; i < len; i++) p[i] = str[i];
 262 }
 263
 264 static void addchar(lua_State *L, SBuf *sb, int c)
 265 {
 266   if (sb->n + 1 > sb->sz) {
 267     MSize sz = sb->sz * 2;
 268     lj_str_resizebuf(L, sb, sz);
 269   }
 270   sb->buf[sb->n++] = (char)c;
 271 }
 272
 273 /* Push formatted message as a string object to Lua stack. va_list variant. */
 274 const char *lj_str_pushvf(lua_State *L, const char *fmt, va_list argp)
 275 {
 276   SBuf *sb = &G(L)->tmpbuf;
 277   lj_str_needbuf(L, sb, (MSize)strlen(fmt));
 278   lj_str_resetbuf(sb);
 279   for (;;) {
 280     const char *e = strchr(fmt, '%');
 281     if (e == NULL) break;
 282     addstr(L, sb, fmt, (MSize)(e-fmt));
 283     /* This function only handles %s, %c, %d, %f and %p formats. */
 284     switch (e[1]) {
 285     case 's': {
 286       const char *s = va_arg(argp, char *);
 287       if (s == NULL) s = "(null)";
 288       addstr(L, sb, s, (MSize)strlen(s));
 289       break;
 290       }
 291     case 'c':
 292       addchar(L, sb, va_arg(argp, int));
 293       break;
 294     case 'd': {
 295       char buf[LJ_STR_INTBUF];
 296       char *p = lj_str_bufint(buf, va_arg(argp, int32_t));
 297       addstr(L, sb, p, (MSize)(buf+LJ_STR_INTBUF-p));
 298       break;
 299       }
 300     case 'f': {
 301       char buf[LJ_STR_NUMBUF];
 302       TValue tv;
 303       MSize len;
 304       tv.n = (lua_Number)(va_arg(argp, LUAI_UACNUMBER));
 305       len = (MSize)lj_str_bufnum(buf, &tv);
 306       addstr(L, sb, buf, len);
 307       break;
 308       }
 309     case 'p': {
 310 #define FMTP_CHARS      (2*sizeof(ptrdiff_t))
 311       char buf[2+FMTP_CHARS];
 312       ptrdiff_t p = (ptrdiff_t)(va_arg(argp, void *));
 313       ptrdiff_t i, lasti = 2+FMTP_CHARS;
 314       if (p == 0) {
 315         addstr(L, sb, "NULL", 4);
 316         break;
 317       }
 318 #if LJ_64
 319       /* Shorten output for 64 bit pointers. */
 320       lasti = 2+2*4+((p >> 32) ? 2+2*(lj_fls((uint32_t)(p >> 32))>>3) : 0);
 321 #endif
 322       buf[0] = '0';
 323       buf[1] = 'x';
 324       for (i = lasti-1; i >= 2; i--, p >>= 4)
 325         buf[i] = "0123456789abcdef"[(p & 15)];
 326       addstr(L, sb, buf, (MSize)lasti);
 327       break;
 328       }
 329     case '%':
 330       addchar(L, sb, '%');
 331       break;
 332     default:
 333       addchar(L, sb, '%');
 334       addchar(L, sb, e[1]);
 335       break;
 336     }
 337     fmt = e+2;
 338   }
 339   addstr(L, sb, fmt, (MSize)strlen(fmt));
 340   setstrV(L, L->top, lj_str_new(L, sb->buf, sb->n));
 341   incr_top(L);
 342   return strVdata(L->top - 1);
 343 }
 344
 345 /* Push formatted message as a string object to Lua stack. Vararg variant. */
 346 const char *lj_str_pushf(lua_State *L, const char *fmt, ...)
 347 {
 348   const char *msg;
 349   va_list argp;
 350   va_start(argp, fmt);
 351   msg = lj_str_pushvf(L, fmt, argp);
 352   va_end(argp);
 353   return msg;
 354 }
 355
 356 /* -- Buffer handling ----------------------------------------------------- */
 357
 358 char *lj_str_needbuf(lua_State *L, SBuf *sb, MSize sz)
 359 {
 360   if (sz > sb->sz) {
 361     if (sz < LJ_MIN_SBUF) sz = LJ_MIN_SBUF;
 362     lj_str_resizebuf(L, sb, sz);
 363   }
 364   return sb->buf;
 365 }
 366