src/util/serialize.cpp

   1 /*
   2 Minetest
   3 Copyright (C) 2010-2013 celeron55, Perttu Ahola <celeron55@gmail.com>
   4
   5 This program is free software; you can redistribute it and/or modify
   6 it under the terms of the GNU Lesser General Public License as published by
   7 the Free Software Foundation; either version 2.1 of the License, or
   8 (at your option) any later version.
   9
  10 This program is distributed in the hope that it will be useful,
  11 but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 GNU Lesser General Public License for more details.
  14
  15 You should have received a copy of the GNU Lesser General Public License along
  16 with this program; if not, write to the Free Software Foundation, Inc.,
  17 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  18 */
  19
  20 #include "serialize.h"
  21 #include "pointer.h"
  22 #include "porting.h"
  23 #include "util/string.h"
  24 #include "../exceptions.h"
  25 #include "../irrlichttypes.h"
  26
  27 #include <sstream>
  28 #include <iomanip>
  29 #include <vector>
  30
  31 SerializationError eof_ser_err("Attempted read past end of data");
  32
  33 ////
  34 //// BufReader
  35 ////
  36
  37 bool BufReader::getStringNoEx(std::string *val)
  38 {
  39         u16 num_chars;
  40         if (!getU16NoEx(&num_chars))
  41                 return false;
  42
  43         if (pos + num_chars > size) {
  44                 pos -= sizeof(num_chars);
  45                 return false;
  46         }
  47
  48         val->assign((const char *)data + pos, num_chars);
  49         pos += num_chars;
  50
  51         return true;
  52 }
  53
  54 bool BufReader::getWideStringNoEx(std::wstring *val)
  55 {
  56         u16 num_chars;
  57         if (!getU16NoEx(&num_chars))
  58                 return false;
  59
  60         if (pos + num_chars * 2 > size) {
  61                 pos -= sizeof(num_chars);
  62                 return false;
  63         }
  64
  65         for (size_t i = 0; i != num_chars; i++) {
  66                 val->push_back(readU16(data + pos));
  67                 pos += 2;
  68         }
  69
  70         return true;
  71 }
  72
  73 bool BufReader::getLongStringNoEx(std::string *val)
  74 {
  75         u32 num_chars;
  76         if (!getU32NoEx(&num_chars))
  77                 return false;
  78
  79         if (pos + num_chars > size) {
  80                 pos -= sizeof(num_chars);
  81                 return false;
  82         }
  83
  84         val->assign((const char *)data + pos, num_chars);
  85         pos += num_chars;
  86
  87         return true;
  88 }
  89
  90 bool BufReader::getRawDataNoEx(void *val, size_t len)
  91 {
  92         if (pos + len > size)
  93                 return false;
  94
  95         memcpy(val, data + pos, len);
  96         pos += len;
  97
  98         return true;
  99 }
 100
 101
 102 ////
 103 //// String
 104 ////
 105
 106 std::string serializeString(const std::string &plain)
 107 {
 108         std::string s;
 109         char buf[2];
 110
 111         if (plain.size() > STRING_MAX_LEN)
 112                 throw SerializationError("String too long for serializeString");
 113
 114         writeU16((u8 *)&buf[0], plain.size());
 115         s.append(buf, 2);
 116
 117         s.append(plain);
 118         return s;
 119 }
 120
 121 std::string deSerializeString(std::istream &is)
 122 {
 123         std::string s;
 124         char buf[2];
 125
 126         is.read(buf, 2);
 127         if (is.gcount() != 2)
 128                 throw SerializationError("deSerializeString: size not read");
 129
 130         u16 s_size = readU16((u8 *)buf);
 131         if (s_size == 0)
 132                 return s;
 133
 134         Buffer<char> buf2(s_size);
 135         is.read(&buf2[0], s_size);
 136         if (is.gcount() != s_size)
 137                 throw SerializationError("deSerializeString: couldn't read all chars");
 138
 139         s.reserve(s_size);
 140         s.append(&buf2[0], s_size);
 141         return s;
 142 }
 143
 144 ////
 145 //// Wide String
 146 ////
 147
 148 std::string serializeWideString(const std::wstring &plain)
 149 {
 150         std::string s;
 151         char buf[2];
 152
 153         if (plain.size() > WIDE_STRING_MAX_LEN)
 154                 throw SerializationError("String too long for serializeWideString");
 155
 156         writeU16((u8 *)buf, plain.size());
 157         s.append(buf, 2);
 158
 159         for (u32 i = 0; i < plain.size(); i++) {
 160                 writeU16((u8 *)buf, plain[i]);
 161                 s.append(buf, 2);
 162         }
 163         return s;
 164 }
 165
 166 std::wstring deSerializeWideString(std::istream &is)
 167 {
 168         std::wstring s;
 169         char buf[2];
 170
 171         is.read(buf, 2);
 172         if (is.gcount() != 2)
 173                 throw SerializationError("deSerializeWideString: size not read");
 174
 175         u16 s_size = readU16((u8 *)buf);
 176         if (s_size == 0)
 177                 return s;
 178
 179         s.reserve(s_size);
 180         for (u32 i = 0; i < s_size; i++) {
 181                 is.read(&buf[0], 2);
 182                 if (is.gcount() != 2) {
 183                         throw SerializationError(
 184                                 "deSerializeWideString: couldn't read all chars");
 185                 }
 186
 187                 wchar_t c16 = readU16((u8 *)buf);
 188                 s.append(&c16, 1);
 189         }
 190         return s;
 191 }
 192
 193 ////
 194 //// Long String
 195 ////
 196
 197 std::string serializeLongString(const std::string &plain)
 198 {
 199         char buf[4];
 200
 201         if (plain.size() > LONG_STRING_MAX_LEN)
 202                 throw SerializationError("String too long for serializeLongString");
 203
 204         writeU32((u8*)&buf[0], plain.size());
 205         std::string s;
 206         s.append(buf, 4);
 207         s.append(plain);
 208         return s;
 209 }
 210
 211 std::string deSerializeLongString(std::istream &is)
 212 {
 213         std::string s;
 214         char buf[4];
 215
 216         is.read(buf, 4);
 217         if (is.gcount() != 4)
 218                 throw SerializationError("deSerializeLongString: size not read");
 219
 220         u32 s_size = readU32((u8 *)buf);
 221         if (s_size == 0)
 222                 return s;
 223
 224         // We don't really want a remote attacker to force us to allocate 4GB...
 225         if (s_size > LONG_STRING_MAX_LEN) {
 226                 throw SerializationError("deSerializeLongString: "
 227                         "string too long: " + itos(s_size) + " bytes");
 228         }
 229
 230         Buffer<char> buf2(s_size);
 231         is.read(&buf2[0], s_size);
 232         if ((u32)is.gcount() != s_size)
 233                 throw SerializationError("deSerializeLongString: couldn't read all chars");
 234
 235         s.reserve(s_size);
 236         s.append(&buf2[0], s_size);
 237         return s;
 238 }
 239
 240 ////
 241 //// JSON
 242 ////
 243
 244 std::string serializeJsonString(const std::string &plain)
 245 {
 246         std::ostringstream os(std::ios::binary);
 247         os << "\"";
 248
 249         for (size_t i = 0; i < plain.size(); i++) {
 250                 char c = plain[i];
 251                 switch (c) {
 252                         case '"':
 253                                 os << "\\\"";
 254                                 break;
 255                         case '\\':
 256                                 os << "\\\\";
 257                                 break;
 258                         case '/':
 259                                 os << "\\/";
 260                                 break;
 261                         case '\b':
 262                                 os << "\\b";
 263                                 break;
 264                         case '\f':
 265                                 os << "\\f";
 266                                 break;
 267                         case '\n':
 268                                 os << "\\n";
 269                                 break;
 270                         case '\r':
 271                                 os << "\\r";
 272                                 break;
 273                         case '\t':
 274                                 os << "\\t";
 275                                 break;
 276                         default: {
 277                                 if (c >= 32 && c <= 126) {
 278                                         os << c;
 279                                 } else {
 280                                         u32 cnum = (u8)c;
 281                                         os << "\\u" << std::hex << std::setw(4)
 282                                                 << std::setfill('0') << cnum;
 283                                 }
 284                                 break;
 285                         }
 286                 }
 287         }
 288
 289         os << "\"";
 290         return os.str();
 291 }
 292
 293 std::string deSerializeJsonString(std::istream &is)
 294 {
 295         std::ostringstream os(std::ios::binary);
 296         char c, c2;
 297
 298         // Parse initial doublequote
 299         is >> c;
 300         if (c != '"')
 301                 throw SerializationError("JSON string must start with doublequote");
 302
 303         // Parse characters
 304         for (;;) {
 305                 c = is.get();
 306                 if (is.eof())
 307                         throw SerializationError("JSON string ended prematurely");
 308
 309                 if (c == '"') {
 310                         return os.str();
 311                 } else if (c == '\\') {
 312                         c2 = is.get();
 313                         if (is.eof())
 314                                 throw SerializationError("JSON string ended prematurely");
 315                         switch (c2) {
 316                                 case 'b':
 317                                         os << '\b';
 318                                         break;
 319                                 case 'f':
 320                                         os << '\f';
 321                                         break;
 322                                 case 'n':
 323                                         os << '\n';
 324                                         break;
 325                                 case 'r':
 326                                         os << '\r';
 327                                         break;
 328                                 case 't':
 329                                         os << '\t';
 330                                         break;
 331                                 case 'u': {
 332                                         int hexnumber;
 333                                         char hexdigits[4 + 1];
 334
 335                                         is.read(hexdigits, 4);
 336                                         if (is.eof())
 337                                                 throw SerializationError("JSON string ended prematurely");
 338                                         hexdigits[4] = 0;
 339
 340                                         std::istringstream tmp_is(hexdigits, std::ios::binary);
 341                                         tmp_is >> std::hex >> hexnumber;
 342                                         os << (char)hexnumber;
 343                                         break;
 344                                 }
 345                                 default:
 346                                         os << c2;
 347                                         break;
 348                         }
 349                 } else {
 350                         os << c;
 351                 }
 352         }
 353
 354         return os.str();
 355 }
 356
 357 ////
 358 //// String/Struct conversions
 359 ////
 360
 361 bool deSerializeStringToStruct(std::string valstr,
 362         std::string format, void *out, size_t olen)
 363 {
 364         size_t len = olen;
 365         std::vector<std::string *> strs_alloced;
 366         std::string *str;
 367         char *f, *snext;
 368         size_t pos;
 369
 370         char *s = &valstr[0];
 371         char *buf = new char[len];
 372         char *bufpos = buf;
 373
 374         char *fmtpos, *fmt = &format[0];
 375         while ((f = strtok_r(fmt, ",", &fmtpos)) && s) {
 376                 fmt = NULL;
 377
 378                 bool is_unsigned = false;
 379                 int width = 0;
 380                 char valtype = *f;
 381
 382                 width = (int)strtol(f + 1, &f, 10);
 383                 if (width && valtype == 's')
 384                         valtype = 'i';
 385
 386                 switch (valtype) {
 387                         case 'u':
 388                                 is_unsigned = true;
 389                                 /* FALLTHROUGH */
 390                         case 'i':
 391                                 if (width == 16) {
 392                                         bufpos += PADDING(bufpos, u16);
 393                                         if ((bufpos - buf) + sizeof(u16) <= len) {
 394                                                 if (is_unsigned)
 395                                                         *(u16 *)bufpos = (u16)strtoul(s, &s, 10);
 396                                                 else
 397                                                         *(s16 *)bufpos = (s16)strtol(s, &s, 10);
 398                                         }
 399                                         bufpos += sizeof(u16);
 400                                 } else if (width == 32) {
 401                                         bufpos += PADDING(bufpos, u32);
 402                                         if ((bufpos - buf) + sizeof(u32) <= len) {
 403                                                 if (is_unsigned)
 404                                                         *(u32 *)bufpos = (u32)strtoul(s, &s, 10);
 405                                                 else
 406                                                         *(s32 *)bufpos = (s32)strtol(s, &s, 10);
 407                                         }
 408                                         bufpos += sizeof(u32);
 409                                 } else if (width == 64) {
 410                                         bufpos += PADDING(bufpos, u64);
 411                                         if ((bufpos - buf) + sizeof(u64) <= len) {
 412                                                 if (is_unsigned)
 413                                                         *(u64 *)bufpos = (u64)strtoull(s, &s, 10);
 414                                                 else
 415                                                         *(s64 *)bufpos = (s64)strtoll(s, &s, 10);
 416                                         }
 417                                         bufpos += sizeof(u64);
 418                                 }
 419                                 s = strchr(s, ',');
 420                                 break;
 421                         case 'b':
 422                                 snext = strchr(s, ',');
 423                                 if (snext)
 424                                         *snext++ = 0;
 425
 426                                 bufpos += PADDING(bufpos, bool);
 427                                 if ((bufpos - buf) + sizeof(bool) <= len)
 428                                         *(bool *)bufpos = is_yes(std::string(s));
 429                                 bufpos += sizeof(bool);
 430
 431                                 s = snext;
 432                                 break;
 433                         case 'f':
 434                                 bufpos += PADDING(bufpos, float);
 435                                 if ((bufpos - buf) + sizeof(float) <= len)
 436                                         *(float *)bufpos = strtof(s, &s);
 437                                 bufpos += sizeof(float);
 438
 439                                 s = strchr(s, ',');
 440                                 break;
 441                         case 's':
 442                                 while (*s == ' ' || *s == '\t')
 443                                         s++;
 444                                 if (*s++ != '"') //error, expected string
 445                                         goto fail;
 446                                 snext = s;
 447
 448                                 while (snext[0] && !(snext[-1] != '\\' && snext[0] == '"'))
 449                                         snext++;
 450                                 *snext++ = 0;
 451
 452                                 bufpos += PADDING(bufpos, std::string *);
 453
 454                                 str = new std::string(s);
 455                                 pos = 0;
 456                                 while ((pos = str->find("\\\"", pos)) != std::string::npos)
 457                                         str->erase(pos, 1);
 458
 459                                 if ((bufpos - buf) + sizeof(std::string *) <= len)
 460                                         *(std::string **)bufpos = str;
 461                                 bufpos += sizeof(std::string *);
 462                                 strs_alloced.push_back(str);
 463
 464                                 s = *snext ? snext + 1 : NULL;
 465                                 break;
 466                         case 'v':
 467                                 while (*s == ' ' || *s == '\t')
 468                                         s++;
 469                                 if (*s++ != '(') //error, expected vector
 470                                         goto fail;
 471
 472                                 if (width == 2) {
 473                                         bufpos += PADDING(bufpos, v2f);
 474
 475                                         if ((bufpos - buf) + sizeof(v2f) <= len) {
 476                                         v2f *v = (v2f *)bufpos;
 477                                                 v->X = strtof(s, &s);
 478                                                 s++;
 479                                                 v->Y = strtof(s, &s);
 480                                         }
 481
 482                                         bufpos += sizeof(v2f);
 483                                 } else if (width == 3) {
 484                                         bufpos += PADDING(bufpos, v3f);
 485                                         if ((bufpos - buf) + sizeof(v3f) <= len) {
 486                                                 v3f *v = (v3f *)bufpos;
 487                                                 v->X = strtof(s, &s);
 488                                                 s++;
 489                                                 v->Y = strtof(s, &s);
 490                                                 s++;
 491                                                 v->Z = strtof(s, &s);
 492                                         }
 493
 494                                         bufpos += sizeof(v3f);
 495                                 }
 496                                 s = strchr(s, ',');
 497                                 break;
 498                         default: //error, invalid format specifier
 499                                 goto fail;
 500                 }
 501
 502                 if (s && *s == ',')
 503                         s++;
 504
 505                 if ((size_t)(bufpos - buf) > len) //error, buffer too small
 506                         goto fail;
 507         }
 508
 509         if (f && *f) { //error, mismatched number of fields and values
 510 fail:
 511                 for (size_t i = 0; i != strs_alloced.size(); i++)
 512                         delete strs_alloced[i];
 513                 delete[] buf;
 514                 return false;
 515         }
 516
 517         memcpy(out, buf, olen);
 518         delete[] buf;
 519         return true;
 520 }
 521
 522 // Casts *buf to a signed or unsigned fixed-width integer of 'w' width
 523 #define SIGN_CAST(w, buf) (is_unsigned ? *((u##w *) buf) : *((s##w *) buf))
 524
 525 bool serializeStructToString(std::string *out,
 526         std::string format, void *value)
 527 {
 528         std::ostringstream os;
 529         std::string str;
 530         char *f;
 531         size_t strpos;
 532
 533         char *bufpos = (char *) value;
 534         char *fmtpos, *fmt = &format[0];
 535         while ((f = strtok_r(fmt, ",", &fmtpos))) {
 536                 fmt = NULL;
 537                 bool is_unsigned = false;
 538                 int width = 0;
 539                 char valtype = *f;
 540
 541                 width = (int)strtol(f + 1, &f, 10);
 542                 if (width && valtype == 's')
 543                         valtype = 'i';
 544
 545                 switch (valtype) {
 546                         case 'u':
 547                                 is_unsigned = true;
 548                                 /* FALLTHROUGH */
 549                         case 'i':
 550                                 if (width == 16) {
 551                                         bufpos += PADDING(bufpos, u16);
 552                                         os << SIGN_CAST(16, bufpos);
 553                                         bufpos += sizeof(u16);
 554                                 } else if (width == 32) {
 555                                         bufpos += PADDING(bufpos, u32);
 556                                         os << SIGN_CAST(32, bufpos);
 557                                         bufpos += sizeof(u32);
 558                                 } else if (width == 64) {
 559                                         bufpos += PADDING(bufpos, u64);
 560                                         os << SIGN_CAST(64, bufpos);
 561                                         bufpos += sizeof(u64);
 562                                 }
 563                                 break;
 564                         case 'b':
 565                                 bufpos += PADDING(bufpos, bool);
 566                                 os << std::boolalpha << *((bool *) bufpos);
 567                                 bufpos += sizeof(bool);
 568                                 break;
 569                         case 'f':
 570                                 bufpos += PADDING(bufpos, float);
 571                                 os << *((float *) bufpos);
 572                                 bufpos += sizeof(float);
 573                                 break;
 574                         case 's':
 575                                 bufpos += PADDING(bufpos, std::string *);
 576                                 str = **((std::string **) bufpos);
 577
 578                                 strpos = 0;
 579                                 while ((strpos = str.find('"', strpos)) != std::string::npos) {
 580                                         str.insert(strpos, 1, '\\');
 581                                         strpos += 2;
 582                                 }
 583
 584                                 os << str;
 585                                 bufpos += sizeof(std::string *);
 586                                 break;
 587                         case 'v':
 588                                 if (width == 2) {
 589                                         bufpos += PADDING(bufpos, v2f);
 590                                         v2f *v = (v2f *) bufpos;
 591                                         os << '(' << v->X << ", " << v->Y << ')';
 592                                         bufpos += sizeof(v2f);
 593                                 } else {
 594                                         bufpos += PADDING(bufpos, v3f);
 595                                         v3f *v = (v3f *) bufpos;
 596                                         os << '(' << v->X << ", " << v->Y << ", " << v->Z << ')';
 597                                         bufpos += sizeof(v3f);
 598                                 }
 599                                 break;
 600                         default:
 601                                 return false;
 602                 }
 603                 os << ", ";
 604         }
 605         *out = os.str();
 606
 607         // Trim off the trailing comma and space
 608         if (out->size() >= 2)
 609                 out->resize(out->size() - 2);
 610
 611         return true;
 612 }
 613
 614 #undef SIGN_CAST
 615
 616 ////
 617 //// Other
 618 ////
 619
 620 std::string serializeHexString(const std::string &data, bool insert_spaces)
 621 {
 622         std::string result;
 623         result.reserve(data.size() * (2 + insert_spaces));
 624
 625         static const char hex_chars[] = "0123456789abcdef";
 626
 627         const size_t len = data.size();
 628         for (size_t i = 0; i != len; i++) {
 629                 u8 byte = data[i];
 630                 result.push_back(hex_chars[(byte >> 4) & 0x0F]);
 631                 result.push_back(hex_chars[(byte >> 0) & 0x0F]);
 632                 if (insert_spaces && i != len - 1)
 633                         result.push_back(' ');
 634         }
 635
 636         return result;
 637 }