Software / code / prosody
Comparison
util-src/encodings.cpp @ 3882:4f88ea5f05ea
util.encodings: Support for ICU for IDNA operations.
| author | Tobias Markmann <tm@ayena.de> |
|---|---|
| date | Tue, 14 Dec 2010 15:03:37 +0100 |
| parent | 2923:util-src/encodings.c@b7049746bd29 |
| child | 3764:323169f229fa |
comparison
equal
deleted
inserted
replaced
| 3881:5f3e9d7064a4 | 3882:4f88ea5f05ea |
|---|---|
| 1 /* Prosody IM | |
| 2 -- Copyright (C) 2008-2010 Matthew Wild | |
| 3 -- Copyright (C) 2008-2010 Waqas Hussain | |
| 4 -- | |
| 5 -- This project is MIT/X11 licensed. Please see the | |
| 6 -- COPYING file in the source package for more information. | |
| 7 -- | |
| 8 */ | |
| 9 | |
| 10 /* | |
| 11 * encodings.c | |
| 12 * Lua library for base64, stringprep and idna encodings | |
| 13 */ | |
| 14 | |
| 15 // Newer MSVC compilers deprecate strcpy as unsafe, but we use it in a safe way | |
| 16 #define _CRT_SECURE_NO_DEPRECATE | |
| 17 | |
| 18 extern "C" { | |
| 19 #include <string.h> | |
| 20 #include <stdlib.h> | |
| 21 #include "lua.h" | |
| 22 #include "lauxlib.h" | |
| 23 } | |
| 24 /***************** BASE64 *****************/ | |
| 25 | |
| 26 #define LUALIB_API extern "C" | |
| 27 | |
| 28 static const char code[]= | |
| 29 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; | |
| 30 | |
| 31 static void base64_encode(luaL_Buffer *b, unsigned int c1, unsigned int c2, unsigned int c3, int n) | |
| 32 { | |
| 33 unsigned long tuple=c3+256UL*(c2+256UL*c1); | |
| 34 int i; | |
| 35 char s[4]; | |
| 36 for (i=0; i<4; i++) { | |
| 37 s[3-i] = code[tuple % 64]; | |
| 38 tuple /= 64; | |
| 39 } | |
| 40 for (i=n+1; i<4; i++) s[i]='='; | |
| 41 luaL_addlstring(b,s,4); | |
| 42 } | |
| 43 | |
| 44 static int Lbase64_encode(lua_State *L) /** encode(s) */ | |
| 45 { | |
| 46 size_t l; | |
| 47 const unsigned char *s=(const unsigned char*)luaL_checklstring(L,1,&l); | |
| 48 luaL_Buffer b; | |
| 49 int n; | |
| 50 luaL_buffinit(L,&b); | |
| 51 for (n=l/3; n--; s+=3) base64_encode(&b,s[0],s[1],s[2],3); | |
| 52 switch (l%3) | |
| 53 { | |
| 54 case 1: base64_encode(&b,s[0],0,0,1); break; | |
| 55 case 2: base64_encode(&b,s[0],s[1],0,2); break; | |
| 56 } | |
| 57 luaL_pushresult(&b); | |
| 58 return 1; | |
| 59 } | |
| 60 | |
| 61 static void base64_decode(luaL_Buffer *b, int c1, int c2, int c3, int c4, int n) | |
| 62 { | |
| 63 unsigned long tuple=c4+64L*(c3+64L*(c2+64L*c1)); | |
| 64 char s[3]; | |
| 65 switch (--n) | |
| 66 { | |
| 67 case 3: s[2]=(char) tuple; | |
| 68 case 2: s[1]=(char) (tuple >> 8); | |
| 69 case 1: s[0]=(char) (tuple >> 16); | |
| 70 } | |
| 71 luaL_addlstring(b,s,n); | |
| 72 } | |
| 73 | |
| 74 static int Lbase64_decode(lua_State *L) /** decode(s) */ | |
| 75 { | |
| 76 size_t l; | |
| 77 const char *s=luaL_checklstring(L,1,&l); | |
| 78 luaL_Buffer b; | |
| 79 int n=0; | |
| 80 char t[4]; | |
| 81 luaL_buffinit(L,&b); | |
| 82 for (;;) | |
| 83 { | |
| 84 int c=*s++; | |
| 85 switch (c) | |
| 86 { | |
| 87 const char *p; | |
| 88 default: | |
| 89 p=strchr(code,c); if (p==NULL) return 0; | |
| 90 t[n++]= (char) (p-code); | |
| 91 if (n==4) | |
| 92 { | |
| 93 base64_decode(&b,t[0],t[1],t[2],t[3],4); | |
| 94 n=0; | |
| 95 } | |
| 96 break; | |
| 97 case '=': | |
| 98 switch (n) | |
| 99 { | |
| 100 case 1: base64_decode(&b,t[0],0,0,0,1); break; | |
| 101 case 2: base64_decode(&b,t[0],t[1],0,0,2); break; | |
| 102 case 3: base64_decode(&b,t[0],t[1],t[2],0,3); break; | |
| 103 } | |
| 104 n=0; | |
| 105 break; | |
| 106 case 0: | |
| 107 luaL_pushresult(&b); | |
| 108 return 1; | |
| 109 case '\n': case '\r': case '\t': case ' ': case '\f': case '\b': | |
| 110 break; | |
| 111 } | |
| 112 } | |
| 113 } | |
| 114 | |
| 115 static const luaL_Reg Reg_base64[] = | |
| 116 { | |
| 117 { "encode", Lbase64_encode }, | |
| 118 { "decode", Lbase64_decode }, | |
| 119 { NULL, NULL } | |
| 120 }; | |
| 121 | |
| 122 /***************** STRINGPREP *****************/ | |
| 123 #ifndef USE_STRINGPREP_ICU | |
| 124 /****************** libidn ********************/ | |
| 125 #include <stringprep.h> | |
| 126 | |
| 127 static int stringprep_prep(lua_State *L, const Stringprep_profile *profile) | |
| 128 { | |
| 129 size_t len; | |
| 130 const char *s; | |
| 131 char string[1024]; | |
| 132 int ret; | |
| 133 if(!lua_isstring(L, 1)) { | |
| 134 lua_pushnil(L); | |
| 135 return 1; | |
| 136 } | |
| 137 s = lua_tolstring(L, 1, &len); | |
| 138 if (len >= 1024) { | |
| 139 lua_pushnil(L); | |
| 140 return 1; // TODO return error message | |
| 141 } | |
| 142 strcpy(string, s); | |
| 143 ret = stringprep(string, 1024, 0, profile); | |
| 144 if (ret == STRINGPREP_OK) { | |
| 145 lua_pushstring(L, string); | |
| 146 return 1; | |
| 147 } else { | |
| 148 lua_pushnil(L); | |
| 149 return 1; // TODO return error message | |
| 150 } | |
| 151 } | |
| 152 | |
| 153 #define MAKE_PREP_FUNC(myFunc, prep) \ | |
| 154 static int myFunc(lua_State *L) { return stringprep_prep(L, prep); } | |
| 155 | |
| 156 MAKE_PREP_FUNC(Lstringprep_nameprep, stringprep_nameprep) /** stringprep.nameprep(s) */ | |
| 157 MAKE_PREP_FUNC(Lstringprep_nodeprep, stringprep_xmpp_nodeprep) /** stringprep.nodeprep(s) */ | |
| 158 MAKE_PREP_FUNC(Lstringprep_resourceprep, stringprep_xmpp_resourceprep) /** stringprep.resourceprep(s) */ | |
| 159 MAKE_PREP_FUNC(Lstringprep_saslprep, stringprep_saslprep) /** stringprep.saslprep(s) */ | |
| 160 | |
| 161 static const luaL_Reg Reg_stringprep[] = | |
| 162 { | |
| 163 { "nameprep", Lstringprep_nameprep }, | |
| 164 { "nodeprep", Lstringprep_nodeprep }, | |
| 165 { "resourceprep", Lstringprep_resourceprep }, | |
| 166 { "saslprep", Lstringprep_saslprep }, | |
| 167 { NULL, NULL } | |
| 168 }; | |
| 169 | |
| 170 #else | |
| 171 #include <unicode/usprep.h> | |
| 172 //#include <unicode/stringpiece.h> | |
| 173 #include <unicode/unistr.h> | |
| 174 #include <unicode/utrace.h> | |
| 175 | |
| 176 static int icu_stringprep_prep(lua_State *L, const UStringPrepProfile *profile) | |
| 177 { | |
| 178 size_t len; | |
| 179 const char *s; | |
| 180 UnicodeString ustr; | |
| 181 UErrorCode err = U_ZERO_ERROR; | |
| 182 UChar dest[1024]; | |
| 183 char output[1024]; | |
| 184 if(!lua_isstring(L, 1)) { | |
| 185 lua_pushnil(L); | |
| 186 return 1; | |
| 187 } | |
| 188 s = lua_tolstring(L, 1, &len); | |
| 189 if (len >= 1024) { | |
| 190 lua_pushnil(L); | |
| 191 return 1; // TODO return error message | |
| 192 } | |
| 193 ustr = UnicodeString::fromUTF8(s); | |
| 194 len = usprep_prepare(profile, ustr.getBuffer(), ustr.length(), dest, 1024, 0, NULL, &err); | |
| 195 if (U_FAILURE(err)) { | |
| 196 lua_pushnil(L); | |
| 197 return 1; | |
| 198 } else { | |
| 199 CheckedArrayByteSink output_sink(output, 1024); | |
| 200 UnicodeString dest_str(TRUE, dest, len); | |
| 201 dest_str.toUTF8(output_sink); | |
| 202 lua_pushstring(L, output); | |
| 203 return 1; | |
| 204 } | |
| 205 } | |
| 206 | |
| 207 UStringPrepProfile *icu_nameprep; | |
| 208 UStringPrepProfile *icu_nodeprep; | |
| 209 UStringPrepProfile *icu_resourceprep; | |
| 210 UStringPrepProfile *icu_saslprep; | |
| 211 | |
| 212 /* initialize global ICU stringprep profiles */ | |
| 213 void init_icu() | |
| 214 { | |
| 215 UErrorCode err = U_ZERO_ERROR; | |
| 216 utrace_setLevel(UTRACE_VERBOSE); | |
| 217 icu_nameprep = usprep_openByType(USPREP_RFC3491_NAMEPREP, &err); | |
| 218 icu_nodeprep = usprep_openByType(USPREP_RFC3920_NODEPREP, &err); | |
| 219 icu_resourceprep = usprep_openByType(USPREP_RFC3920_RESOURCEPREP, &err); | |
| 220 icu_saslprep = usprep_openByType(USPREP_RFC4013_SASLPREP, &err); | |
| 221 if (U_FAILURE(err)) fprintf(stderr, "[c] util.encodings: error: %s\n", u_errorName((UErrorCode)err)); | |
| 222 } | |
| 223 | |
| 224 #define MAKE_PREP_FUNC(myFunc, prep) \ | |
| 225 static int myFunc(lua_State *L) { return icu_stringprep_prep(L, prep); } | |
| 226 | |
| 227 MAKE_PREP_FUNC(Lstringprep_nameprep, icu_nameprep) /** stringprep.nameprep(s) */ | |
| 228 MAKE_PREP_FUNC(Lstringprep_nodeprep, icu_nodeprep) /** stringprep.nodeprep(s) */ | |
| 229 MAKE_PREP_FUNC(Lstringprep_resourceprep, icu_resourceprep) /** stringprep.resourceprep(s) */ | |
| 230 MAKE_PREP_FUNC(Lstringprep_saslprep, icu_saslprep) /** stringprep.saslprep(s) */ | |
| 231 | |
| 232 static const luaL_Reg Reg_stringprep[] = | |
| 233 { | |
| 234 { "nameprep", Lstringprep_nameprep }, | |
| 235 { "nodeprep", Lstringprep_nodeprep }, | |
| 236 { "resourceprep", Lstringprep_resourceprep }, | |
| 237 { "saslprep", Lstringprep_saslprep }, | |
| 238 { NULL, NULL } | |
| 239 }; | |
| 240 #endif | |
| 241 | |
| 242 /***************** IDNA *****************/ | |
| 243 #ifndef USE_STRINGPREP_ICU | |
| 244 /****************** libidn ********************/ | |
| 245 #include <idna.h> | |
| 246 #include <idn-free.h> | |
| 247 static int Lidna_to_ascii(lua_State *L) /** idna.to_ascii(s) */ | |
| 248 { | |
| 249 size_t len; | |
| 250 const char *s = luaL_checklstring(L, 1, &len); | |
| 251 char* output = NULL; | |
| 252 int ret = idna_to_ascii_8z(s, &output, IDNA_USE_STD3_ASCII_RULES); | |
| 253 if (ret == IDNA_SUCCESS) { | |
| 254 lua_pushstring(L, output); | |
| 255 idn_free(output); | |
| 256 return 1; | |
| 257 } else { | |
| 258 lua_pushnil(L); | |
| 259 idn_free(output); | |
| 260 return 1; // TODO return error message | |
| 261 } | |
| 262 } | |
| 263 | |
| 264 static int Lidna_to_unicode(lua_State *L) /** idna.to_unicode(s) */ | |
| 265 { | |
| 266 size_t len; | |
| 267 const char *s = luaL_checklstring(L, 1, &len); | |
| 268 char* output = NULL; | |
| 269 int ret = idna_to_unicode_8z8z(s, &output, 0); | |
| 270 if (ret == IDNA_SUCCESS) { | |
| 271 lua_pushstring(L, output); | |
| 272 idn_free(output); | |
| 273 return 1; | |
| 274 } else { | |
| 275 lua_pushnil(L); | |
| 276 idn_free(output); | |
| 277 return 1; // TODO return error message | |
| 278 } | |
| 279 } | |
| 280 #else | |
| 281 #include <unicode/ustdio.h> | |
| 282 #include <unicode/uidna.h> | |
| 283 /* IDNA2003 or IDNA2008 ? ? ? */ | |
| 284 static int Lidna_to_ascii(lua_State *L) /** idna.to_ascii(s) */ | |
| 285 { | |
| 286 size_t len; | |
| 287 int32_t out_len; | |
| 288 const char *s = luaL_checklstring(L, 1, &len); | |
| 289 UnicodeString ustr; | |
| 290 UErrorCode err = U_ZERO_ERROR; | |
| 291 UChar dest[1024]; | |
| 292 char output[1024]; | |
| 293 | |
| 294 ustr = UnicodeString::fromUTF8(s); | |
| 295 out_len = uidna_IDNToASCII(ustr.getBuffer(), ustr.length(), dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err); | |
| 296 if (U_FAILURE(err)) { | |
| 297 lua_pushnil(L); | |
| 298 return 1; | |
| 299 } else { | |
| 300 CheckedArrayByteSink output_sink(output, 1024); | |
| 301 UnicodeString dest_str(TRUE, dest, out_len); | |
| 302 dest_str.toUTF8(output_sink); | |
| 303 lua_pushstring(L, output); | |
| 304 return 1; | |
| 305 } | |
| 306 } | |
| 307 | |
| 308 static int Lidna_to_unicode(lua_State *L) /** idna.to_unicode(s) */ | |
| 309 { | |
| 310 size_t len; | |
| 311 int32_t out_len; | |
| 312 const char *s = luaL_checklstring(L, 1, &len); | |
| 313 UnicodeString ustr; | |
| 314 UErrorCode err = U_ZERO_ERROR; | |
| 315 UChar dest[1024]; | |
| 316 char output[1024]; | |
| 317 | |
| 318 ustr = UnicodeString::fromUTF8(s); | |
| 319 out_len = uidna_IDNToUnicode(ustr.getBuffer(), ustr.length(), dest, 1024, UIDNA_USE_STD3_RULES, NULL, &err); | |
| 320 if (U_FAILURE(err)) { | |
| 321 lua_pushnil(L); | |
| 322 return 1; | |
| 323 } else { | |
| 324 CheckedArrayByteSink output_sink(output, 1024); | |
| 325 UnicodeString dest_str(TRUE, dest, out_len); | |
| 326 dest_str.toUTF8(output_sink); | |
| 327 lua_pushstring(L, output); | |
| 328 return 1; | |
| 329 } | |
| 330 } | |
| 331 #endif | |
| 332 | |
| 333 static const luaL_Reg Reg_idna[] = | |
| 334 { | |
| 335 { "to_ascii", Lidna_to_ascii }, | |
| 336 { "to_unicode", Lidna_to_unicode }, | |
| 337 { NULL, NULL } | |
| 338 }; | |
| 339 | |
| 340 /***************** end *****************/ | |
| 341 | |
| 342 static const luaL_Reg Reg[] = | |
| 343 { | |
| 344 { NULL, NULL } | |
| 345 }; | |
| 346 | |
| 347 LUALIB_API int luaopen_util_encodings(lua_State *L) | |
| 348 { | |
| 349 #ifdef USE_STRINGPREP_ICU | |
| 350 init_icu(); | |
| 351 #endif | |
| 352 luaL_register(L, "encodings", Reg); | |
| 353 | |
| 354 lua_pushliteral(L, "base64"); | |
| 355 lua_newtable(L); | |
| 356 luaL_register(L, NULL, Reg_base64); | |
| 357 lua_settable(L,-3); | |
| 358 | |
| 359 lua_pushliteral(L, "stringprep"); | |
| 360 lua_newtable(L); | |
| 361 luaL_register(L, NULL, Reg_stringprep); | |
| 362 lua_settable(L,-3); | |
| 363 | |
| 364 lua_pushliteral(L, "idna"); | |
| 365 lua_newtable(L); | |
| 366 luaL_register(L, NULL, Reg_idna); | |
| 367 lua_settable(L,-3); | |
| 368 | |
| 369 lua_pushliteral(L, "version"); /** version */ | |
| 370 lua_pushliteral(L, "-3.14"); | |
| 371 lua_settable(L,-3); | |
| 372 return 1; | |
| 373 } |