util-src/encodings.c

   1 /* Prosody IM
   2 -- Copyright (C) 2008-2010 Matthew Wild
   3 -- Copyright (C) 2008-2010 Waqas Hussain
   4 -- Copyright (C) 1994-2015 Lua.org, PUC-Rio.
   5 --
   6 -- This project is MIT/X11 licensed. Please see the
   7 -- COPYING file in the source package for more information.
   8 --
   9 */
  10
  11 /*
  12 * encodings.c
  13 * Lua library for base64, stringprep and idna encodings
  14 */
  15
  16 /* Newer MSVC compilers deprecate strcpy as unsafe, but we use it in a safe way */
  17 #define _CRT_SECURE_NO_DEPRECATE
  18
  19 #include <string.h>
  20 #include <stdlib.h>
  21 #include "lua.h"
  22 #include "lauxlib.h"
  23
  24 #if (LUA_VERSION_NUM == 501)
  25 #define luaL_setfuncs(L, R, N) luaL_register(L, NULL, R)
  26 #endif
  27
  28 /***************** BASE64 *****************/
  29
  30 static const char code[] =
  31     "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
  32
  33 static void base64_encode(luaL_Buffer *b, unsigned int c1, unsigned int c2, unsigned int c3, int n) {
  34         unsigned long tuple = c3 + 256UL * (c2 + 256UL * c1);
  35         int i;
  36         char s[4];
  37
  38         for(i = 0; i < 4; i++) {
  39                 s[3 - i] = code[tuple % 64];
  40                 tuple /= 64;
  41         }
  42
  43         for(i = n + 1; i < 4; i++) {
  44                 s[i] = '=';
  45         }
  46
  47         luaL_addlstring(b, s, 4);
  48 }
  49
  50 static int Lbase64_encode(lua_State *L) {       /** encode(s) */
  51         size_t l;
  52         const unsigned char *s = (const unsigned char *)luaL_checklstring(L, 1, &l);
  53         luaL_Buffer b;
  54         int n;
  55         luaL_buffinit(L, &b);
  56
  57         for(n = l / 3; n--; s += 3) {
  58                 base64_encode(&b, s[0], s[1], s[2], 3);
  59         }
  60
  61         switch(l % 3) {
  62                 case 1:
  63                         base64_encode(&b, s[0], 0, 0, 1);
  64                         break;
  65
  66                 case 2:
  67                         base64_encode(&b, s[0], s[1], 0, 2);
  68                         break;
  69         }
  70
  71         luaL_pushresult(&b);
  72         return 1;
  73 }
  74
  75 static void base64_decode(luaL_Buffer *b, int c1, int c2, int c3, int c4, int n) {
  76         unsigned long tuple = c4 + 64L * (c3 + 64L * (c2 + 64L * c1));
  77         char s[3];
  78
  79         switch(--n) {
  80                 case 3:
  81                         s[2] = (char) tuple;
  82                         /* Falls through. */
  83
  84                 case 2:
  85                         s[1] = (char)(tuple >> 8);
  86                         /* Falls through. */
  87
  88                 case 1:
  89                         s[0] = (char)(tuple >> 16);
  90         }
  91
  92         luaL_addlstring(b, s, n);
  93 }
  94
  95 static int Lbase64_decode(lua_State *L) {       /** decode(s) */
  96         size_t l;
  97         const char *s = luaL_checklstring(L, 1, &l);
  98         luaL_Buffer b;
  99         int n = 0;
 100         char t[4];
 101         luaL_buffinit(L, &b);
 102
 103         for(;;) {
 104                 int c = *s++;
 105
 106                 switch(c) {
 107                                 const char *p;
 108
 109                         default:
 110                                 p = strchr(code, c);
 111
 112                                 if(p == NULL) {
 113                                         return 0;
 114                                 }
 115
 116                                 t[n++] = (char)(p - code);
 117
 118                                 if(n == 4) {
 119                                         base64_decode(&b, t[0], t[1], t[2], t[3], 4);
 120                                         n = 0;
 121                                 }
 122
 123                                 break;
 124
 125                         case '=':
 126
 127                                 switch(n) {
 128                                         case 1:
 129                                                 base64_decode(&b, t[0], 0, 0, 0, 1);
 130                                                 break;
 131
 132                                         case 2:
 133                                                 base64_decode(&b, t[0], t[1], 0, 0, 2);
 134                                                 break;
 135
 136                                         case 3:
 137                                                 base64_decode(&b, t[0], t[1], t[2], 0, 3);
 138                                                 break;
 139                                 }
 140
 141                                 n = 0;
 142                                 break;
 143
 144                         case 0:
 145                                 luaL_pushresult(&b);
 146                                 return 1;
 147
 148                         case '\n':
 149                         case '\r':
 150                         case '\t':
 151                         case ' ':
 152                         case '\f':
 153                         case '\b':
 154                                 break;
 155                 }
 156         }
 157 }
 158
 159 static const luaL_Reg Reg_base64[] = {
 160         { "encode",     Lbase64_encode  },
 161         { "decode",     Lbase64_decode  },
 162         { NULL,         NULL    }
 163 };
 164
 165 /******************* UTF-8 ********************/
 166
 167 /*
 168  * Adapted from Lua 5.3
 169  * Needed because libidn does not validate that input is valid UTF-8
 170  */
 171
 172 #define MAXUNICODE      0x10FFFF
 173
 174 /*
 175  * Decode one UTF-8 sequence, returning NULL if byte sequence is invalid.
 176  */
 177 static const char *utf8_decode(const char *o, int *val) {
 178         static const unsigned int limits[] = {0xFF, 0x7F, 0x7FF, 0xFFFF};
 179         const unsigned char *s = (const unsigned char *)o;
 180         unsigned int c = s[0];
 181         unsigned int res = 0;  /* final result */
 182
 183         if(c < 0x80) { /* ascii? */
 184                 res = c;
 185         } else {
 186                 int count = 0;  /* to count number of continuation bytes */
 187
 188                 while(c & 0x40) {   /* still have continuation bytes? */
 189                         int cc = s[++count];  /* read next byte */
 190
 191                         if((cc & 0xC0) != 0x80) { /* not a continuation byte? */
 192                                 return NULL;    /* invalid byte sequence */
 193                         }
 194
 195                         res = (res << 6) | (cc & 0x3F);  /* add lower 6 bits from cont. byte */
 196                         c <<= 1;  /* to test next bit */
 197                 }
 198
 199                 res |= ((c & 0x7F) << (count * 5));  /* add first byte */
 200
 201                 if(count > 3 || res > MAXUNICODE || res <= limits[count] || (0xd800 <= res && res <= 0xdfff)) {
 202                         return NULL;    /* invalid byte sequence */
 203                 }
 204
 205                 s += count;  /* skip continuation bytes read */
 206         }
 207
 208         if(val) {
 209                 *val = res;
 210         }
 211
 212         return (const char *)s + 1;  /* +1 to include first byte */
 213 }
 214
 215 /*
 216  * Check that a string is valid UTF-8
 217  * Returns NULL if not
 218  */
 219 const char *check_utf8(lua_State *L, int idx, size_t *l) {
 220         size_t pos, len;
 221         const char *s = luaL_checklstring(L, idx, &len);
 222         pos = 0;
 223
 224         while(pos <= len) {
 225                 const char *s1 = utf8_decode(s + pos, NULL);
 226
 227                 if(s1 == NULL) {   /* conversion error? */
 228                         return NULL;
 229                 }
 230
 231                 pos = s1 - s;
 232         }
 233
 234         if(l != NULL) {
 235                 *l = len;
 236         }
 237
 238         return s;
 239 }
 240
 241 static int Lutf8_valid(lua_State *L) {
 242         lua_pushboolean(L, check_utf8(L, 1, NULL) != NULL);
 243         return 1;
 244 }
 245
 246 static int Lutf8_length(lua_State *L) {
 247         size_t len;
 248
 249         if(!check_utf8(L, 1, &len)) {
 250                 lua_pushnil(L);
 251                 lua_pushliteral(L, "invalid utf8");
 252                 return 2;
 253         }
 254
 255         lua_pushinteger(L, len);
 256         return 1;
 257 }
 258
 259 static const luaL_Reg Reg_utf8[] = {
 260         { "valid",      Lutf8_valid     },
 261         { "length",     Lutf8_length    },
 262         { NULL,         NULL    }
 263 };
 264
 265 /***************** STRINGPREP *****************/
 266 #ifdef USE_STRINGPREP_ICU
 267
 268 #include <unicode/usprep.h>
 269 #include <unicode/ustring.h>
 270 #include <unicode/utrace.h>
 271 #include <unicode/uspoof.h>
 272
 273 static int icu_stringprep_prep(lua_State *L, const UStringPrepProfile *profile) {
 274         size_t input_len;
 275         int32_t unprepped_len, prepped_len, output_len;
 276         const char *input;
 277         char output[1024];
 278
 279         UChar unprepped[1024]; /* Temporary unicode buffer (1024 characters) */
 280         UChar prepped[1024];
 281
 282         UErrorCode err = U_ZERO_ERROR;
 283
 284         if(!lua_isstring(L, 1)) {
 285                 lua_pushnil(L);
 286                 return 1;
 287         }
 288
 289         input = lua_tolstring(L, 1, &input_len);
 290
 291         if(input_len >= 1024) {
 292                 lua_pushnil(L);
 293                 return 1;
 294         }
 295
 296         u_strFromUTF8(unprepped, 1024, &unprepped_len, input, input_len, &err);
 297
 298         if(U_FAILURE(err)) {
 299                 lua_pushnil(L);
 300                 return 1;
 301         }
 302
 303         prepped_len = usprep_prepare(profile, unprepped, unprepped_len, prepped, 1024, USPREP_ALLOW_UNASSIGNED, NULL, &err);
 304
 305         if(U_FAILURE(err)) {
 306                 lua_pushnil(L);
 307                 return 1;
 308         } else {
 309                 u_strToUTF8(output, 1024, &output_len, prepped, prepped_len, &err);
 310
 311                 if(U_SUCCESS(err) && output_len < 1024) {
 312                         lua_pushlstring(L, output, output_len);
 313                 } else {
 314                         lua_pushnil(L);
 315                 }
 316
 317                 return 1;
 318         }
 319 }
 320
 321 UStringPrepProfile *icu_nameprep;
 322 UStringPrepProfile *icu_nodeprep;
 323 UStringPrepProfile *icu_resourceprep;
 324 UStringPrepProfile *icu_saslprep;
 325 USpoofChecker *icu_spoofcheck;
 326
 327 #if (U_ICU_VERSION_MAJOR_NUM < 58)
 328 /* COMPAT */
 329 #define USPOOF_CONFUSABLE (USPOOF_SINGLE_SCRIPT_CONFUSABLE | USPOOF_MIXED_SCRIPT_CONFUSABLE | USPOOF_WHOLE_SCRIPT_CONFUSABLE)
 330 #endif
 331
 332 /* initialize global ICU stringprep profiles */
 333 void init_icu(void) {
 334         UErrorCode err = U_ZERO_ERROR;
 335         utrace_setLevel(UTRACE_VERBOSE);
 336         icu_nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, &err);
 337         icu_nodeprep = usprep_openByType(USPREP_RFC3920_NODEPREP, &err);
 338         icu_resourceprep = usprep_openByType(USPREP_RFC3920_RESOURCEPREP, &err);
 339         icu_saslprep = usprep_openByType(USPREP_RFC4013_SASLPREP, &err);
 340         icu_spoofcheck = uspoof_open(&err);
 341         uspoof_setChecks(icu_spoofcheck, USPOOF_CONFUSABLE, &err);
 342
 343         if(U_FAILURE(err)) {
 344                 fprintf(stderr, "[c] util.encodings: error: %s\n", u_errorName((UErrorCode)err));
 345         }
 346 }
 347
 348 #define MAKE_PREP_FUNC(myFunc, prep) \
 349 static int myFunc(lua_State *L) { return icu_stringprep_prep(L, prep); }
 350
 351 MAKE_PREP_FUNC(Lstringprep_nameprep, icu_nameprep)              /** stringprep.nameprep(s) */
 352 MAKE_PREP_FUNC(Lstringprep_nodeprep, icu_nodeprep)              /** stringprep.nodeprep(s) */
 353 MAKE_PREP_FUNC(Lstringprep_resourceprep, icu_resourceprep)              /** stringprep.resourceprep(s) */
 354 MAKE_PREP_FUNC(Lstringprep_saslprep, icu_saslprep)              /** stringprep.saslprep(s) */
 355
 356 static const luaL_Reg Reg_stringprep[] = {
 357         { "nameprep",   Lstringprep_nameprep    },
 358         { "nodeprep",   Lstringprep_nodeprep    },
 359         { "resourceprep",       Lstringprep_resourceprep        },
 360         { "saslprep",   Lstringprep_saslprep    },
 361         { NULL,         NULL    }
 362 };
 363 #else /* USE_STRINGPREP_ICU */
 364
 365 /****************** libidn ********************/
 366
 367 #include <stringprep.h>
 368
 369 static int stringprep_prep(lua_State *L, const Stringprep_profile *profile) {
 370         size_t len;
 371         const char *s;
 372         char string[1024];
 373         int ret;
 374
 375         if(!lua_isstring(L, 1)) {
 376                 lua_pushnil(L);
 377                 return 1;
 378         }
 379
 380         s = check_utf8(L, 1, &len);
 381
 382         if(s == NULL || len >= 1024 || len != strlen(s)) {
 383                 lua_pushnil(L);
 384                 return 1; /* TODO return error message */
 385         }
 386
 387         strcpy(string, s);
 388         ret = stringprep(string, 1024, (Stringprep_profile_flags)0, profile);
 389
 390         if(ret == STRINGPREP_OK) {
 391                 lua_pushstring(L, string);
 392                 return 1;
 393         } else {
 394                 lua_pushnil(L);
 395                 return 1; /* TODO return error message */
 396         }
 397 }
 398
 399 #define MAKE_PREP_FUNC(myFunc, prep) \
 400 static int myFunc(lua_State *L) { return stringprep_prep(L, prep); }
 401
 402 MAKE_PREP_FUNC(Lstringprep_nameprep, stringprep_nameprep)               /** stringprep.nameprep(s) */
 403 MAKE_PREP_FUNC(Lstringprep_nodeprep, stringprep_xmpp_nodeprep)          /** stringprep.nodeprep(s) */
 404 MAKE_PREP_FUNC(Lstringprep_resourceprep, stringprep_xmpp_resourceprep)          /** stringprep.resourceprep(s) */
 405 MAKE_PREP_FUNC(Lstringprep_saslprep, stringprep_saslprep)               /** stringprep.saslprep(s) */
 406
 407 static const luaL_Reg Reg_stringprep[] = {
 408         { "nameprep",   Lstringprep_nameprep    },
 409         { "nodeprep",   Lstringprep_nodeprep    },
 410         { "resourceprep",       Lstringprep_resourceprep        },
 411         { "saslprep",   Lstringprep_saslprep    },
 412         { NULL,         NULL    }
 413 };
 414 #endif
 415
 416 /***************** IDNA *****************/
 417 #ifdef USE_STRINGPREP_ICU
 418 #include <unicode/ustdio.h>
 419 #include <unicode/uidna.h>
 420 /* IDNA2003 or IDNA2008 ? ? ? */
 421 static int Lidna_to_ascii(lua_State *L) {       /** idna.to_ascii(s) */
 422         size_t len;
 423         int32_t ulen, dest_len, output_len;
 424         const char *s = luaL_checklstring(L, 1, &len);
 425         UChar ustr[1024];
 426         UErrorCode err = U_ZERO_ERROR;
 427         UChar dest[1024];
 428         char output[1024];
 429
 430         u_strFromUTF8(ustr, 1024, &ulen, s, len, &err);
 431
 432         if(U_FAILURE(err)) {
 433                 lua_pushnil(L);
 434                 return 1;
 435         }
 436
 437         dest_len = uidna_IDNToASCII(ustr, ulen, dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err);
 438
 439         if(U_FAILURE(err)) {
 440                 lua_pushnil(L);
 441                 return 1;
 442         } else {
 443                 u_strToUTF8(output, 1024, &output_len, dest, dest_len, &err);
 444
 445                 if(U_SUCCESS(err) && output_len < 1024) {
 446                         lua_pushlstring(L, output, output_len);
 447                 } else {
 448                         lua_pushnil(L);
 449                 }
 450
 451                 return 1;
 452         }
 453 }
 454
 455 static int Lidna_to_unicode(lua_State *L) {     /** idna.to_unicode(s) */
 456         size_t len;
 457         int32_t ulen, dest_len, output_len;
 458         const char *s = luaL_checklstring(L, 1, &len);
 459         UChar ustr[1024];
 460         UErrorCode err = U_ZERO_ERROR;
 461         UChar dest[1024];
 462         char output[1024];
 463
 464         u_strFromUTF8(ustr, 1024, &ulen, s, len, &err);
 465
 466         if(U_FAILURE(err)) {
 467                 lua_pushnil(L);
 468                 return 1;
 469         }
 470
 471         dest_len = uidna_IDNToUnicode(ustr, ulen, dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err);
 472
 473         if(U_FAILURE(err)) {
 474                 lua_pushnil(L);
 475                 return 1;
 476         } else {
 477                 u_strToUTF8(output, 1024, &output_len, dest, dest_len, &err);
 478
 479                 if(U_SUCCESS(err) && output_len < 1024) {
 480                         lua_pushlstring(L, output, output_len);
 481                 } else {
 482                         lua_pushnil(L);
 483                 }
 484
 485                 return 1;
 486         }
 487 }
 488
 489 static int Lskeleton(lua_State *L) {
 490         size_t len;
 491         int32_t ulen, dest_len, output_len;
 492         const char *s = luaL_checklstring(L, 1, &len);
 493         UErrorCode err = U_ZERO_ERROR;
 494         UChar ustr[1024];
 495         UChar dest[1024];
 496         char output[1024];
 497
 498         u_strFromUTF8(ustr, 1024, &ulen, s, len, &err);
 499
 500         if(U_FAILURE(err)) {
 501                 lua_pushnil(L);
 502                 return 1;
 503         }
 504
 505         dest_len = uspoof_getSkeleton(icu_spoofcheck, 0, ustr, ulen, dest, 1024, &err);
 506
 507         if(U_FAILURE(err)) {
 508                 lua_pushnil(L);
 509                 return 1;
 510         }
 511
 512         u_strToUTF8(output, 1024, &output_len, dest, dest_len, &err);
 513
 514         if(U_SUCCESS(err)) {
 515                 lua_pushlstring(L, output, output_len);
 516                 return 1;
 517         }
 518
 519         lua_pushnil(L);
 520         return 1;
 521 }
 522
 523 #else /* USE_STRINGPREP_ICU */
 524 /****************** libidn ********************/
 525
 526 #include <idna.h>
 527 #include <idn-free.h>
 528
 529 static int Lidna_to_ascii(lua_State *L) {       /** idna.to_ascii(s) */
 530         size_t len;
 531         const char *s = check_utf8(L, 1, &len);
 532         char *output = NULL;
 533         int ret;
 534
 535         if(s == NULL || len != strlen(s)) {
 536                 lua_pushnil(L);
 537                 return 1; /* TODO return error message */
 538         }
 539
 540         ret = idna_to_ascii_8z(s, &output, IDNA_USE_STD3_ASCII_RULES);
 541
 542         if(ret == IDNA_SUCCESS) {
 543                 lua_pushstring(L, output);
 544                 idn_free(output);
 545                 return 1;
 546         } else {
 547                 lua_pushnil(L);
 548                 idn_free(output);
 549                 return 1; /* TODO return error message */
 550         }
 551 }
 552
 553 static int Lidna_to_unicode(lua_State *L) {     /** idna.to_unicode(s) */
 554         size_t len;
 555         const char *s = luaL_checklstring(L, 1, &len);
 556         char *output = NULL;
 557         int ret = idna_to_unicode_8z8z(s, &output, 0);
 558
 559         if(ret == IDNA_SUCCESS) {
 560                 lua_pushstring(L, output);
 561                 idn_free(output);
 562                 return 1;
 563         } else {
 564                 lua_pushnil(L);
 565                 idn_free(output);
 566                 return 1; /* TODO return error message */
 567         }
 568 }
 569 #endif
 570
 571 static const luaL_Reg Reg_idna[] = {
 572         { "to_ascii",   Lidna_to_ascii  },
 573         { "to_unicode", Lidna_to_unicode        },
 574         { NULL,         NULL    }
 575 };
 576
 577 /***************** end *****************/
 578
 579 LUALIB_API int luaopen_util_encodings(lua_State *L) {
 580 #if (LUA_VERSION_NUM > 501)
 581         luaL_checkversion(L);
 582 #endif
 583 #ifdef USE_STRINGPREP_ICU
 584         init_icu();
 585 #endif
 586         lua_newtable(L);
 587
 588         lua_newtable(L);
 589         luaL_setfuncs(L, Reg_base64, 0);
 590         lua_setfield(L, -2, "base64");
 591
 592         lua_newtable(L);
 593         luaL_setfuncs(L, Reg_stringprep, 0);
 594         lua_setfield(L, -2, "stringprep");
 595
 596         lua_newtable(L);
 597         luaL_setfuncs(L, Reg_idna, 0);
 598         lua_setfield(L, -2, "idna");
 599
 600         lua_newtable(L);
 601         luaL_setfuncs(L, Reg_utf8, 0);
 602         lua_setfield(L, -2, "utf8");
 603
 604 #ifdef USE_STRINGPREP_ICU
 605         lua_newtable(L);
 606         lua_pushcfunction(L, Lskeleton);
 607         lua_setfield(L, -2, "skeleton");
 608         lua_setfield(L, -2, "confusable");
 609 #endif
 610
 611         lua_pushliteral(L, "-3.14");
 612         lua_setfield(L, -2, "version");
 613         return 1;
 614 }