src/lib/lstrlib.c

   1 /*
   2 ** $Id: lstrlib.c,v 1.32 1999/06/17 17:04:03 roberto Exp $
   3 ** Standard library for strings and pattern-matching
   4 ** See Copyright Notice in lua.h
   5 */
   6
   7
   8 #include <ctype.h>
   9 #include <stdio.h>
  10 #include <stdlib.h>
  11 #include <string.h>
  12
  13 #include "lauxlib.h"
  14 #include "lua.h"
  15 #include "lualib.h"
  16
  17
  18
  19 static void addnchar (char *s, int n)
  20 {
  21   char *b = luaL_openspace(n);
  22   memcpy(b, s, n);
  23   luaL_addsize(n);
  24 }
  25
  26
  27 static void str_len (void)
  28 {
  29   long l;
  30   luaL_check_lstr(1, &l);
  31   lua_pushnumber(l);
  32 }
  33
  34
  35 static void closeandpush (void) {
  36   lua_pushlstring(luaL_buffer(), luaL_getsize());
  37 }
  38
  39
  40 static long posrelat (long pos, long len) {
  41   /* relative string position: negative means back from end */
  42   return (pos>=0) ? pos : len+pos+1;
  43 }
  44
  45
  46 static void str_sub (void) {
  47   long l;
  48   char *s = luaL_check_lstr(1, &l);
  49   long start = posrelat(luaL_check_long(2), l);
  50   long end = posrelat(luaL_opt_long(3, -1), l);
  51   if (start < 1) start = 1;
  52   if (end > l) end = l;
  53   if (start <= end)
  54     lua_pushlstring(s+start-1, end-start+1);
  55   else lua_pushstring("");
  56 }
  57
  58
  59 static void str_lower (void) {
  60   long l;
  61   int i;
  62   char *s = luaL_check_lstr(1, &l);
  63   luaL_resetbuffer();
  64   for (i=0; i<l; i++)
  65     luaL_addchar(tolower((unsigned char)(s[i])));
  66   closeandpush();
  67 }
  68
  69
  70 static void str_upper (void) {
  71   long l;
  72   int i;
  73   char *s = luaL_check_lstr(1, &l);
  74   luaL_resetbuffer();
  75   for (i=0; i<l; i++)
  76     luaL_addchar(toupper((unsigned char)(s[i])));
  77   closeandpush();
  78 }
  79
  80 static void str_rep (void)
  81 {
  82   long l;
  83   char *s = luaL_check_lstr(1, &l);
  84   int n = luaL_check_int(2);
  85   luaL_resetbuffer();
  86   while (n-- > 0)
  87     addnchar(s, l);
  88   closeandpush();
  89 }
  90
  91
  92 static void str_byte (void) {
  93   long l;
  94   char *s = luaL_check_lstr(1, &l);
  95   long pos = posrelat(luaL_opt_long(2, 1), l);
  96   luaL_arg_check(0<pos && pos<=l, 2,  "out of range");
  97   lua_pushnumber((unsigned char)s[pos-1]);
  98 }
  99
 100
 101 static void str_char (void) {
 102   int i = 0;
 103   luaL_resetbuffer();
 104   while (lua_getparam(++i) != LUA_NOOBJECT) {
 105     double c = luaL_check_number(i);
 106     luaL_arg_check((unsigned char)c == c, i, "invalid value");
 107     luaL_addchar((unsigned char)c);
 108   }
 109   closeandpush();
 110 }
 111
 112
 113
 114 /*
 115 ** {======================================================
 116 ** PATTERN MATCHING
 117 ** =======================================================
 118 */
 119
 120 #ifndef MAX_CAPT
 121 #define MAX_CAPT 32  /* arbitrary limit */
 122 #endif
 123
 124
 125 struct Capture {
 126   char *src_end;  /* end ('\0') of source string */
 127   int level;  /* total number of captures (finished or unfinished) */
 128   struct {
 129     char *init;
 130     int len;  /* -1 signals unfinished capture */
 131   } capture[MAX_CAPT];
 132 };
 133
 134
 135 #define ESC     '%'
 136 #define SPECIALS  "^$*+?.([%-"
 137
 138
 139 static void push_captures (struct Capture *cap) {
 140   int i;
 141   for (i=0; i<cap->level; i++) {
 142     int l = cap->capture[i].len;
 143     if (l == -1) lua_error("unfinished capture");
 144     lua_pushlstring(cap->capture[i].init, l);
 145   }
 146 }
 147
 148
 149 static int check_cap (int l, struct Capture *cap) {
 150   l -= '1';
 151   if (!(0 <= l && l < cap->level && cap->capture[l].len != -1))
 152     lua_error("invalid capture index");
 153   return l;
 154 }
 155
 156
 157 static int capture_to_close (struct Capture *cap) {
 158   int level = cap->level;
 159   for (level--; level>=0; level--)
 160     if (cap->capture[level].len == -1) return level;
 161   lua_error("invalid pattern capture");
 162   return 0;  /* to avoid warnings */
 163 }
 164
 165
 166 char *luaI_classend (char *p) {
 167   switch (*p++) {
 168     case ESC:
 169       if (*p == '\0')
 170         luaL_verror("incorrect pattern (ends with `%c')", ESC);
 171       return p+1;
 172     case '[':
 173       if (*p == '^') p++;
 174       if (*p == ']') p++;
 175       p = strchr(p, ']');
 176       if (!p) lua_error("incorrect pattern (missing `]')");
 177       return p+1;
 178     default:
 179       return p;
 180   }
 181 }
 182
 183
 184 static int matchclass (int c, int cl) {
 185   int res;
 186   switch (tolower(cl)) {
 187     case 'a' : res = isalpha(c); break;
 188     case 'c' : res = iscntrl(c); break;
 189     case 'd' : res = isdigit(c); break;
 190     case 'l' : res = islower(c); break;
 191     case 'p' : res = ispunct(c); break;
 192     case 's' : res = isspace(c); break;
 193     case 'u' : res = isupper(c); break;
 194     case 'w' : res = isalnum(c); break;
 195     case 'x' : res = isxdigit(c); break;
 196     case 'z' : res = (c == '\0'); break;
 197     default: return (cl == c);
 198   }
 199   return (islower(cl) ? res : !res);
 200 }
 201
 202
 203
 204 static int matchbracketclass (int c, char *p, char *end) {
 205   int sig = 1;
 206   if (*(p+1) == '^') {
 207     sig = 0;
 208     p++;  /* skip the '^' */
 209   }
 210   while (++p < end) {
 211     if (*p == ESC) {
 212       p++;
 213       if ((p < end) && matchclass(c, (unsigned char)*p))
 214         return sig;
 215     }
 216     else if ((*(p+1) == '-') && (p+2 < end)) {
 217       p+=2;
 218       if ((int)(unsigned char)*(p-2) <= c && c <= (int)(unsigned char)*p)
 219         return sig;
 220     }
 221     else if ((unsigned char)*p == c) return sig;
 222   }
 223   return !sig;
 224 }
 225
 226
 227
 228 int luaI_singlematch (int c, char *p, char *ep) {
 229   switch (*p) {
 230     case '.':  /* matches any char */
 231       return 1;
 232     case ESC:
 233       return matchclass(c, (unsigned char)*(p+1));
 234     case '[':
 235       return matchbracketclass(c, p, ep-1);
 236     default:
 237       return ((unsigned char)*p == c);
 238   }
 239 }
 240
 241
 242 static char *match (char *s, char *p, struct Capture *cap);
 243
 244
 245 static char *matchbalance (char *s, char *p, struct Capture *cap) {
 246   if (*p == 0 || *(p+1) == 0)
 247     lua_error("unbalanced pattern");
 248   if (*s != *p) return NULL;
 249   else {
 250     int b = *p;
 251     int e = *(p+1);
 252     int cont = 1;
 253     while (++s < cap->src_end) {
 254       if (*s == e) {
 255         if (--cont == 0) return s+1;
 256       }
 257       else if (*s == b) cont++;
 258     }
 259   }
 260   return NULL;  /* string ends out of balance */
 261 }
 262
 263
 264 static char *max_expand (char *s, char *p, char *ep, struct Capture *cap) {
 265   int i = 0;  /* counts maximum expand for item */
 266   while ((s+i)<cap->src_end && luaI_singlematch((unsigned char)*(s+i), p, ep))
 267     i++;
 268   /* keeps trying to match mith the maximum repetitions */
 269   while (i>=0) {
 270     char *res = match((s+i), ep+1, cap);
 271     if (res) return res;
 272     i--;  /* else didn't match; reduce 1 repetition to try again */
 273   }
 274   return NULL;
 275 }
 276
 277
 278 static char *min_expand (char *s, char *p, char *ep, struct Capture *cap) {
 279   for (;;) {
 280     char *res = match(s, ep+1, cap);
 281     if (res != NULL)
 282       return res;
 283     else if (s<cap->src_end && luaI_singlematch((unsigned char)*s, p, ep))
 284       s++;  /* try with one more repetition */
 285     else return NULL;
 286   }
 287 }
 288
 289
 290 static char *start_capt (char *s, char *p, struct Capture *cap) {
 291   char *res;
 292   int level = cap->level;
 293   if (level >= MAX_CAPT) lua_error("too many captures");
 294   cap->capture[level].init = s;
 295   cap->capture[level].len = -1;
 296   cap->level = level+1;
 297   if ((res=match(s, p+1, cap)) == NULL)  /* match failed? */
 298     cap->level--;  /* undo capture */
 299   return res;
 300 }
 301
 302
 303 static char *end_capt (char *s, char *p, struct Capture *cap) {
 304   int l = capture_to_close(cap);
 305   char *res;
 306   cap->capture[l].len = s - cap->capture[l].init;  /* close capture */
 307   if ((res = match(s, p+1, cap)) == NULL)  /* match failed? */
 308     cap->capture[l].len = -1;  /* undo capture */
 309   return res;
 310 }
 311
 312
 313 static char *match_capture (char *s, int level, struct Capture *cap) {
 314   int l = check_cap(level, cap);
 315   int len = cap->capture[l].len;
 316   if (cap->src_end-s >= len &&
 317       memcmp(cap->capture[l].init, s, len) == 0)
 318     return s+len;
 319   else return NULL;
 320 }
 321
 322
 323 static char *match (char *s, char *p, struct Capture *cap) {
 324   init: /* using goto's to optimize tail recursion */
 325   switch (*p) {
 326     case '(':  /* start capture */
 327       return start_capt(s, p, cap);
 328     case ')':  /* end capture */
 329       return end_capt(s, p, cap);
 330     case ESC:  /* may be %[0-9] or %b */
 331       if (isdigit((unsigned char)(*(p+1)))) {  /* capture? */
 332         s = match_capture(s, *(p+1), cap);
 333         if (s == NULL) return NULL;
 334         p+=2; goto init;  /* else return match(p+2, s, cap) */
 335       }
 336       else if (*(p+1) == 'b') {  /* balanced string? */
 337         s = matchbalance(s, p+2, cap);
 338         if (s == NULL) return NULL;
 339         p+=4; goto init;  /* else return match(p+4, s, cap); */
 340       }
 341       else goto dflt;  /* case default */
 342     case '\0':  /* end of pattern */
 343       return s;  /* match succeeded */
 344     case '$':
 345       if (*(p+1) == '\0')  /* is the '$' the last char in pattern? */
 346         return (s == cap->src_end) ? s : NULL;  /* check end of string */
 347       else goto dflt;
 348     default: dflt: {  /* it is a pattern item */
 349       char *ep = luaI_classend(p);  /* points to what is next */
 350       int m = s<cap->src_end && luaI_singlematch((unsigned char)*s, p, ep);
 351       switch (*ep) {
 352         case '?': {  /* optional */
 353           char *res;
 354           if (m && ((res=match(s+1, ep+1, cap)) != NULL))
 355             return res;
 356           p=ep+1; goto init;  /* else return match(s, ep+1, cap); */
 357         }
 358         case '*':  /* 0 or more repetitions */
 359           return max_expand(s, p, ep, cap);
 360         case '+':  /* 1 or more repetitions */
 361           return (m ? max_expand(s+1, p, ep, cap) : NULL);
 362         case '-':  /* 0 or more repetitions (minimum) */
 363           return min_expand(s, p, ep, cap);
 364         default:
 365           if (!m) return NULL;
 366           s++; p=ep; goto init;  /* else return match(s+1, ep, cap); */
 367       }
 368     }
 369   }
 370 }
 371
 372
 373 static void str_find (void) {
 374   long l;
 375   char *s = luaL_check_lstr(1, &l);
 376   char *p = luaL_check_string(2);
 377   long init = posrelat(luaL_opt_long(3, 1), l) - 1;
 378   struct Capture cap;
 379   luaL_arg_check(0 <= init && init <= l, 3, "out of range");
 380   if (lua_getparam(4) != LUA_NOOBJECT ||
 381       strpbrk(p, SPECIALS) == NULL) {  /* no special characters? */
 382     char *s2 = strstr(s+init, p);
 383     if (s2) {
 384       lua_pushnumber(s2-s+1);
 385       lua_pushnumber(s2-s+strlen(p));
 386       return;
 387     }
 388   }
 389   else {
 390     int anchor = (*p == '^') ? (p++, 1) : 0;
 391     char *s1=s+init;
 392     cap.src_end = s+l;
 393     do {
 394       char *res;
 395       cap.level = 0;
 396       if ((res=match(s1, p, &cap)) != NULL) {
 397         lua_pushnumber(s1-s+1);  /* start */
 398         lua_pushnumber(res-s);   /* end */
 399         push_captures(&cap);
 400         return;
 401       }
 402     } while (s1++<cap.src_end && !anchor);
 403   }
 404   lua_pushnil();  /* if arrives here, it didn't find */
 405 }
 406
 407
 408 static void add_s (lua_Object newp, struct Capture *cap) {
 409   if (lua_isstring(newp)) {
 410     char *news = lua_getstring(newp);
 411     int l = lua_strlen(newp);
 412     int i;
 413     for (i=0; i<l; i++) {
 414       if (news[i] != ESC)
 415         luaL_addchar(news[i]);
 416       else {
 417         i++;  /* skip ESC */
 418         if (!isdigit((unsigned char)news[i]))
 419           luaL_addchar(news[i]);
 420         else {
 421           int level = check_cap(news[i], cap);
 422           addnchar(cap->capture[level].init, cap->capture[level].len);
 423         }
 424       }
 425     }
 426   }
 427   else {  /* is a function */
 428     lua_Object res;
 429     int status;
 430     int oldbuff;
 431     lua_beginblock();
 432     push_captures(cap);
 433     /* function may use buffer, so save it and create a new one */
 434     oldbuff = luaL_newbuffer(0);
 435     status = lua_callfunction(newp);
 436     /* restore old buffer */
 437     luaL_oldbuffer(oldbuff);
 438     if (status != 0) {
 439       lua_endblock();
 440       lua_error(NULL);
 441     }
 442     res = lua_getresult(1);
 443     if (lua_isstring(res))
 444       addnchar(lua_getstring(res), lua_strlen(res));
 445     lua_endblock();
 446   }
 447 }
 448
 449
 450 static void str_gsub (void) {
 451   long srcl;
 452   char *src = luaL_check_lstr(1, &srcl);
 453   char *p = luaL_check_string(2);
 454   lua_Object newp = lua_getparam(3);
 455   int max_s = luaL_opt_int(4, srcl+1);
 456   int anchor = (*p == '^') ? (p++, 1) : 0;
 457   int n = 0;
 458   struct Capture cap;
 459   luaL_arg_check(lua_isstring(newp) || lua_isfunction(newp), 3,
 460                  "string or function expected");
 461   luaL_resetbuffer();
 462   cap.src_end = src+srcl;
 463   while (n < max_s) {
 464     char *e;
 465     cap.level = 0;
 466     e = match(src, p, &cap);
 467     if (e) {
 468       n++;
 469       add_s(newp, &cap);
 470     }
 471     if (e && e>src) /* non empty match? */
 472       src = e;  /* skip it */
 473     else if (src < cap.src_end)
 474       luaL_addchar(*src++);
 475     else break;
 476     if (anchor) break;
 477   }
 478   addnchar(src, cap.src_end-src);
 479   closeandpush();
 480   lua_pushnumber(n);  /* number of substitutions */
 481 }
 482
 483 /* }====================================================== */
 484
 485
 486 static void luaI_addquoted (int arg) {
 487   long l;
 488   char *s = luaL_check_lstr(arg, &l);
 489   luaL_addchar('"');
 490   while (l--) {
 491     switch (*s) {
 492       case '"':  case '\\':  case '\n':
 493         luaL_addchar('\\');
 494         luaL_addchar(*s);
 495         break;
 496       case '\0': addnchar("\\000", 4); break;
 497       default: luaL_addchar(*s);
 498     }
 499     s++;
 500   }
 501   luaL_addchar('"');
 502 }
 503
 504 /* maximum size of each format specification (such as '%-099.99d') */
 505 #define MAX_FORMAT 20  /* arbitrary limit */
 506
 507 static void str_format (void) {
 508   int arg = 1;
 509   char *strfrmt = luaL_check_string(arg);
 510   luaL_resetbuffer();
 511   while (*strfrmt) {
 512     if (*strfrmt != '%')
 513       luaL_addchar(*strfrmt++);
 514     else if (*++strfrmt == '%')
 515       luaL_addchar(*strfrmt++);  /* %% */
 516     else { /* format item */
 517       struct Capture cap;
 518       char form[MAX_FORMAT];  /* to store the format ('%...') */
 519       char *buff;  /* to store the formatted item */
 520       char *initf = strfrmt;
 521       form[0] = '%';
 522       if (isdigit((unsigned char)*initf) && *(initf+1) == '$') {
 523         arg = *initf - '0';
 524         initf += 2;  /* skip the 'n$' */
 525       }
 526       arg++;
 527       cap.src_end = strfrmt+strlen(strfrmt)+1;
 528       cap.level = 0;
 529       strfrmt = match(initf, "[-+ #0]*(%d*)%.?(%d*)", &cap);
 530       if (cap.capture[0].len > 2 || cap.capture[1].len > 2 ||  /* < 100? */
 531           strfrmt-initf > MAX_FORMAT-2)
 532         lua_error("invalid format (width or precision too long)");
 533       strncpy(form+1, initf, strfrmt-initf+1); /* +1 to include conversion */
 534       form[strfrmt-initf+2] = 0;
 535       buff = luaL_openspace(512);  /* 512 > size of format('%99.99f', -1e308) */
 536       switch (*strfrmt++) {
 537         case 'c':  case 'd':  case 'i':
 538           sprintf(buff, form, luaL_check_int(arg));
 539           break;
 540         case 'o':  case 'u':  case 'x':  case 'X':
 541           sprintf(buff, form, (unsigned int)luaL_check_number(arg));
 542           break;
 543         case 'e':  case 'E': case 'f': case 'g': case 'G':
 544           sprintf(buff, form, luaL_check_number(arg));
 545           break;
 546         case 'q':
 547           luaI_addquoted(arg);
 548           continue;  /* skip the "addsize" at the end */
 549         case 's': {
 550           long l;
 551           char *s = luaL_check_lstr(arg, &l);
 552           if (cap.capture[1].len == 0 && l >= 100) {
 553             /* no precision and string is too big to be formatted;
 554                keep original string */
 555             addnchar(s, l);
 556             continue;  /* skip the "addsize" at the end */
 557           }
 558           else {
 559             sprintf(buff, form, s);
 560             break;
 561           }
 562         }
 563         default:  /* also treat cases 'pnLlh' */
 564           lua_error("invalid option in `format'");
 565       }
 566       luaL_addsize(strlen(buff));
 567     }
 568   }
 569   closeandpush();  /* push the result */
 570 }
 571
 572
 573 static struct luaL_reg strlib[] = {
 574 {"strlen", str_len},
 575 {"strsub", str_sub},
 576 {"strlower", str_lower},
 577 {"strupper", str_upper},
 578 {"strchar", str_char},
 579 {"strrep", str_rep},
 580 {"ascii", str_byte},  /* for compatibility with 3.0 and earlier */
 581 {"strbyte", str_byte},
 582 {"format", str_format},
 583 {"strfind", str_find},
 584 {"gsub", str_gsub}
 585 };
 586
 587
 588 /*
 589 ** Open string library
 590 */
 591 void strlib_open (void)
 592 {
 593   luaL_openlib(strlib, (sizeof(strlib)/sizeof(strlib[0])));
 594 }