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 ////
  32 //// BufReader
  33 ////
  34
  35 bool BufReader::getStringNoEx(std::string *val)
  36 {
  37         u16 num_chars;
  38         if (!getU16NoEx(&num_chars))
  39                 return false;
  40
  41         if (pos + num_chars > size) {
  42                 pos -= sizeof(num_chars);
  43                 return false;
  44         }
  45
  46         val->assign((const char *)data + pos, num_chars);
  47         pos += num_chars;
  48
  49         return true;
  50 }
  51
  52 bool BufReader::getWideStringNoEx(std::wstring *val)
  53 {
  54         u16 num_chars;
  55         if (!getU16NoEx(&num_chars))
  56                 return false;
  57
  58         if (pos + num_chars * 2 > size) {
  59                 pos -= sizeof(num_chars);
  60                 return false;
  61         }
  62
  63         for (size_t i = 0; i != num_chars; i++) {
  64                 val->push_back(readU16(data + pos));
  65                 pos += 2;
  66         }
  67
  68         return true;
  69 }
  70
  71 bool BufReader::getLongStringNoEx(std::string *val)
  72 {
  73         u32 num_chars;
  74         if (!getU32NoEx(&num_chars))
  75                 return false;
  76
  77         if (pos + num_chars > size) {
  78                 pos -= sizeof(num_chars);
  79                 return false;
  80         }
  81
  82         val->assign((const char *)data + pos, num_chars);
  83         pos += num_chars;
  84
  85         return true;
  86 }
  87
  88 bool BufReader::getRawDataNoEx(void *val, size_t len)
  89 {
  90         if (pos + len > size)
  91                 return false;
  92
  93         memcpy(val, data + pos, len);
  94         pos += len;
  95
  96         return true;
  97 }
  98
  99
 100 ////
 101 //// String
 102 ////
 103
 104 std::string serializeString(const std::string &plain)
 105 {
 106         std::string s;
 107         char buf[2];
 108
 109         if (plain.size() > STRING_MAX_LEN)
 110                 throw SerializationError("String too long for serializeString");
 111
 112         writeU16((u8 *)&buf[0], plain.size());
 113         s.append(buf, 2);
 114
 115         s.append(plain);
 116         return s;
 117 }
 118
 119 std::string deSerializeString(std::istream &is)
 120 {
 121         std::string s;
 122         char buf[2];
 123
 124         is.read(buf, 2);
 125         if (is.gcount() != 2)
 126                 throw SerializationError("deSerializeString: size not read");
 127
 128         u16 s_size = readU16((u8 *)buf);
 129         if (s_size == 0)
 130                 return s;
 131
 132         Buffer<char> buf2(s_size);
 133         is.read(&buf2[0], s_size);
 134         if (is.gcount() != s_size)
 135                 throw SerializationError("deSerializeString: couldn't read all chars");
 136
 137         s.reserve(s_size);
 138         s.append(&buf2[0], s_size);
 139         return s;
 140 }
 141
 142 ////
 143 //// Wide String
 144 ////
 145
 146 std::string serializeWideString(const std::wstring &plain)
 147 {
 148         std::string s;
 149         char buf[2];
 150
 151         if (plain.size() > WIDE_STRING_MAX_LEN)
 152                 throw SerializationError("String too long for serializeWideString");
 153
 154         writeU16((u8 *)buf, plain.size());
 155         s.append(buf, 2);
 156
 157         for (wchar_t i : plain) {
 158                 writeU16((u8 *)buf, i);
 159                 s.append(buf, 2);
 160         }
 161         return s;
 162 }
 163
 164 std::wstring deSerializeWideString(std::istream &is)
 165 {
 166         std::wstring s;
 167         char buf[2];
 168
 169         is.read(buf, 2);
 170         if (is.gcount() != 2)
 171                 throw SerializationError("deSerializeWideString: size not read");
 172
 173         u16 s_size = readU16((u8 *)buf);
 174         if (s_size == 0)
 175                 return s;
 176
 177         s.reserve(s_size);
 178         for (u32 i = 0; i < s_size; i++) {
 179                 is.read(&buf[0], 2);
 180                 if (is.gcount() != 2) {
 181                         throw SerializationError(
 182                                 "deSerializeWideString: couldn't read all chars");
 183                 }
 184
 185                 wchar_t c16 = readU16((u8 *)buf);
 186                 s.append(&c16, 1);
 187         }
 188         return s;
 189 }
 190
 191 ////
 192 //// Long String
 193 ////
 194
 195 std::string serializeLongString(const std::string &plain)
 196 {
 197         char buf[4];
 198
 199         if (plain.size() > LONG_STRING_MAX_LEN)
 200                 throw SerializationError("String too long for serializeLongString");
 201
 202         writeU32((u8*)&buf[0], plain.size());
 203         std::string s;
 204         s.append(buf, 4);
 205         s.append(plain);
 206         return s;
 207 }
 208
 209 std::string deSerializeLongString(std::istream &is)
 210 {
 211         std::string s;
 212         char buf[4];
 213
 214         is.read(buf, 4);
 215         if (is.gcount() != 4)
 216                 throw SerializationError("deSerializeLongString: size not read");
 217
 218         u32 s_size = readU32((u8 *)buf);
 219         if (s_size == 0)
 220                 return s;
 221
 222         // We don't really want a remote attacker to force us to allocate 4GB...
 223         if (s_size > LONG_STRING_MAX_LEN) {
 224                 throw SerializationError("deSerializeLongString: "
 225                         "string too long: " + itos(s_size) + " bytes");
 226         }
 227
 228         Buffer<char> buf2(s_size);
 229         is.read(&buf2[0], s_size);
 230         if ((u32)is.gcount() != s_size)
 231                 throw SerializationError("deSerializeLongString: couldn't read all chars");
 232
 233         s.reserve(s_size);
 234         s.append(&buf2[0], s_size);
 235         return s;
 236 }
 237
 238 ////
 239 //// JSON
 240 ////
 241
 242 std::string serializeJsonString(const std::string &plain)
 243 {
 244         std::ostringstream os(std::ios::binary);
 245         os << "\"";
 246
 247         for (char c : plain) {
 248                 switch (c) {
 249                         case '"':
 250                                 os << "\\\"";
 251                                 break;
 252                         case '\\':
 253                                 os << "\\\\";
 254                                 break;
 255                         case '/':
 256                                 os << "\\/";
 257                                 break;
 258                         case '\b':
 259                                 os << "\\b";
 260                                 break;
 261                         case '\f':
 262                                 os << "\\f";
 263                                 break;
 264                         case '\n':
 265                                 os << "\\n";
 266                                 break;
 267                         case '\r':
 268                                 os << "\\r";
 269                                 break;
 270                         case '\t':
 271                                 os << "\\t";
 272                                 break;
 273                         default: {
 274                                 if (c >= 32 && c <= 126) {
 275                                         os << c;
 276                                 } else {
 277                                         u32 cnum = (u8)c;
 278                                         os << "\\u" << std::hex << std::setw(4)
 279                                                 << std::setfill('0') << cnum;
 280                                 }
 281                                 break;
 282                         }
 283                 }
 284         }
 285
 286         os << "\"";
 287         return os.str();
 288 }
 289
 290 std::string deSerializeJsonString(std::istream &is)
 291 {
 292         std::ostringstream os(std::ios::binary);
 293         char c, c2;
 294
 295         // Parse initial doublequote
 296         is >> c;
 297         if (c != '"')
 298                 throw SerializationError("JSON string must start with doublequote");
 299
 300         // Parse characters
 301         for (;;) {
 302                 c = is.get();
 303                 if (is.eof())
 304                         throw SerializationError("JSON string ended prematurely");
 305
 306                 if (c == '"') {
 307                         return os.str();
 308                 }
 309
 310                 if (c == '\\') {
 311                         c2 = is.get();
 312                         if (is.eof())
 313                                 throw SerializationError("JSON string ended prematurely");
 314                         switch (c2) {
 315                                 case 'b':
 316                                         os << '\b';
 317                                         break;
 318                                 case 'f':
 319                                         os << '\f';
 320                                         break;
 321                                 case 'n':
 322                                         os << '\n';
 323                                         break;
 324                                 case 'r':
 325                                         os << '\r';
 326                                         break;
 327                                 case 't':
 328                                         os << '\t';
 329                                         break;
 330                                 case 'u': {
 331                                         int hexnumber;
 332                                         char hexdigits[4 + 1];
 333
 334                                         is.read(hexdigits, 4);
 335                                         if (is.eof())
 336                                                 throw SerializationError("JSON string ended prematurely");
 337                                         hexdigits[4] = 0;
 338
 339                                         std::istringstream tmp_is(hexdigits, std::ios::binary);
 340                                         tmp_is >> std::hex >> hexnumber;
 341                                         os << (char)hexnumber;
 342                                         break;
 343                                 }
 344                                 default:
 345                                         os << c2;
 346                                         break;
 347                         }
 348                 } else {
 349                         os << c;
 350                 }
 351         }
 352
 353         return os.str();
 354 }
 355
 356 std::string serializeJsonStringIfNeeded(const std::string &s)
 357 {
 358         for (size_t i = 0; i < s.size(); ++i) {
 359                 if (s[i] <= 0x1f || s[i] >= 0x7f || s[i] == ' ' || s[i] == '\"')
 360                         return serializeJsonString(s);
 361         }
 362         return s;
 363 }
 364
 365 std::string deSerializeJsonStringIfNeeded(std::istream &is)
 366 {
 367         std::ostringstream tmp_os;
 368         bool expect_initial_quote = true;
 369         bool is_json = false;
 370         bool was_backslash = false;
 371         for (;;) {
 372                 char c = is.get();
 373                 if (is.eof())
 374                         break;
 375
 376                 if (expect_initial_quote && c == '"') {
 377                         tmp_os << c;
 378                         is_json = true;
 379                 } else if(is_json) {
 380                         tmp_os << c;
 381                         if (was_backslash)
 382                                 was_backslash = false;
 383                         else if (c == '\\')
 384                                 was_backslash = true;
 385                         else if (c == '"')
 386                                 break; // Found end of string
 387                 } else {
 388                         if (c == ' ') {
 389                                 // Found end of word
 390                                 is.unget();
 391                                 break;
 392                         }
 393
 394                         tmp_os << c;
 395                 }
 396                 expect_initial_quote = false;
 397         }
 398         if (is_json) {
 399                 std::istringstream tmp_is(tmp_os.str(), std::ios::binary);
 400                 return deSerializeJsonString(tmp_is);
 401         }
 402
 403         return tmp_os.str();
 404 }
 405
 406 ////
 407 //// String/Struct conversions
 408 ////
 409
 410 bool deSerializeStringToStruct(std::string valstr,
 411         std::string format, void *out, size_t olen)
 412 {
 413         size_t len = olen;
 414         std::vector<std::string *> strs_alloced;
 415         std::string *str;
 416         char *f, *snext;
 417         size_t pos;
 418
 419         char *s = &valstr[0];
 420         char *buf = new char[len];
 421         char *bufpos = buf;
 422
 423         char *fmtpos, *fmt = &format[0];
 424         while ((f = strtok_r(fmt, ",", &fmtpos)) && s) {
 425                 fmt = nullptr;
 426
 427                 bool is_unsigned = false;
 428                 int width = 0;
 429                 char valtype = *f;
 430
 431                 width = (int)strtol(f + 1, &f, 10);
 432                 if (width && valtype == 's')
 433                         valtype = 'i';
 434
 435                 switch (valtype) {
 436                         case 'u':
 437                                 is_unsigned = true;
 438                                 /* FALLTHROUGH */
 439                         case 'i':
 440                                 if (width == 16) {
 441                                         bufpos += PADDING(bufpos, u16);
 442                                         if ((bufpos - buf) + sizeof(u16) <= len) {
 443                                                 if (is_unsigned)
 444                                                         *(u16 *)bufpos = (u16)strtoul(s, &s, 10);
 445                                                 else
 446                                                         *(s16 *)bufpos = (s16)strtol(s, &s, 10);
 447                                         }
 448                                         bufpos += sizeof(u16);
 449                                 } else if (width == 32) {
 450                                         bufpos += PADDING(bufpos, u32);
 451                                         if ((bufpos - buf) + sizeof(u32) <= len) {
 452                                                 if (is_unsigned)
 453                                                         *(u32 *)bufpos = (u32)strtoul(s, &s, 10);
 454                                                 else
 455                                                         *(s32 *)bufpos = (s32)strtol(s, &s, 10);
 456                                         }
 457                                         bufpos += sizeof(u32);
 458                                 } else if (width == 64) {
 459                                         bufpos += PADDING(bufpos, u64);
 460                                         if ((bufpos - buf) + sizeof(u64) <= len) {
 461                                                 if (is_unsigned)
 462                                                         *(u64 *)bufpos = (u64)strtoull(s, &s, 10);
 463                                                 else
 464                                                         *(s64 *)bufpos = (s64)strtoll(s, &s, 10);
 465                                         }
 466                                         bufpos += sizeof(u64);
 467                                 }
 468                                 s = strchr(s, ',');
 469                                 break;
 470                         case 'b':
 471                                 snext = strchr(s, ',');
 472                                 if (snext)
 473                                         *snext++ = 0;
 474
 475                                 bufpos += PADDING(bufpos, bool);
 476                                 if ((bufpos - buf) + sizeof(bool) <= len)
 477                                         *(bool *)bufpos = is_yes(std::string(s));
 478                                 bufpos += sizeof(bool);
 479
 480                                 s = snext;
 481                                 break;
 482                         case 'f':
 483                                 bufpos += PADDING(bufpos, float);
 484                                 if ((bufpos - buf) + sizeof(float) <= len)
 485                                         *(float *)bufpos = strtof(s, &s);
 486                                 bufpos += sizeof(float);
 487
 488                                 s = strchr(s, ',');
 489                                 break;
 490                         case 's':
 491                                 while (*s == ' ' || *s == '\t')
 492                                         s++;
 493                                 if (*s++ != '"') //error, expected string
 494                                         goto fail;
 495                                 snext = s;
 496
 497                                 while (snext[0] && !(snext[-1] != '\\' && snext[0] == '"'))
 498                                         snext++;
 499                                 *snext++ = 0;
 500
 501                                 bufpos += PADDING(bufpos, std::string *);
 502
 503                                 str = new std::string(s);
 504                                 pos = 0;
 505                                 while ((pos = str->find("\\\"", pos)) != std::string::npos)
 506                                         str->erase(pos, 1);
 507
 508                                 if ((bufpos - buf) + sizeof(std::string *) <= len)
 509                                         *(std::string **)bufpos = str;
 510                                 bufpos += sizeof(std::string *);
 511                                 strs_alloced.push_back(str);
 512
 513                                 s = *snext ? snext + 1 : nullptr;
 514                                 break;
 515                         case 'v':
 516                                 while (*s == ' ' || *s == '\t')
 517                                         s++;
 518                                 if (*s++ != '(') //error, expected vector
 519                                         goto fail;
 520
 521                                 if (width == 2) {
 522                                         bufpos += PADDING(bufpos, v2f);
 523
 524                                         if ((bufpos - buf) + sizeof(v2f) <= len) {
 525                                         v2f *v = (v2f *)bufpos;
 526                                                 v->X = strtof(s, &s);
 527                                                 s++;
 528                                                 v->Y = strtof(s, &s);
 529                                         }
 530
 531                                         bufpos += sizeof(v2f);
 532                                 } else if (width == 3) {
 533                                         bufpos += PADDING(bufpos, v3f);
 534                                         if ((bufpos - buf) + sizeof(v3f) <= len) {
 535                                                 v3f *v = (v3f *)bufpos;
 536                                                 v->X = strtof(s, &s);
 537                                                 s++;
 538                                                 v->Y = strtof(s, &s);
 539                                                 s++;
 540                                                 v->Z = strtof(s, &s);
 541                                         }
 542
 543                                         bufpos += sizeof(v3f);
 544                                 }
 545                                 s = strchr(s, ',');
 546                                 break;
 547                         default: //error, invalid format specifier
 548                                 goto fail;
 549                 }
 550
 551                 if (s && *s == ',')
 552                         s++;
 553
 554                 if ((size_t)(bufpos - buf) > len) //error, buffer too small
 555                         goto fail;
 556         }
 557
 558         if (f && *f) { //error, mismatched number of fields and values
 559 fail:
 560                 for (size_t i = 0; i != strs_alloced.size(); i++)
 561                         delete strs_alloced[i];
 562                 delete[] buf;
 563                 return false;
 564         }
 565
 566         memcpy(out, buf, olen);
 567         delete[] buf;
 568         return true;
 569 }
 570
 571 // Casts *buf to a signed or unsigned fixed-width integer of 'w' width
 572 #define SIGN_CAST(w, buf) (is_unsigned ? *((u##w *) buf) : *((s##w *) buf))
 573
 574 bool serializeStructToString(std::string *out,
 575         std::string format, void *value)
 576 {
 577         std::ostringstream os;
 578         std::string str;
 579         char *f;
 580         size_t strpos;
 581
 582         char *bufpos = (char *) value;
 583         char *fmtpos, *fmt = &format[0];
 584         while ((f = strtok_r(fmt, ",", &fmtpos))) {
 585                 fmt = nullptr;
 586                 bool is_unsigned = false;
 587                 int width = 0;
 588                 char valtype = *f;
 589
 590                 width = (int)strtol(f + 1, &f, 10);
 591                 if (width && valtype == 's')
 592                         valtype = 'i';
 593
 594                 switch (valtype) {
 595                         case 'u':
 596                                 is_unsigned = true;
 597                                 /* FALLTHROUGH */
 598                         case 'i':
 599                                 if (width == 16) {
 600                                         bufpos += PADDING(bufpos, u16);
 601                                         os << SIGN_CAST(16, bufpos);
 602                                         bufpos += sizeof(u16);
 603                                 } else if (width == 32) {
 604                                         bufpos += PADDING(bufpos, u32);
 605                                         os << SIGN_CAST(32, bufpos);
 606                                         bufpos += sizeof(u32);
 607                                 } else if (width == 64) {
 608                                         bufpos += PADDING(bufpos, u64);
 609                                         os << SIGN_CAST(64, bufpos);
 610                                         bufpos += sizeof(u64);
 611                                 }
 612                                 break;
 613                         case 'b':
 614                                 bufpos += PADDING(bufpos, bool);
 615                                 os << std::boolalpha << *((bool *) bufpos);
 616                                 bufpos += sizeof(bool);
 617                                 break;
 618                         case 'f':
 619                                 bufpos += PADDING(bufpos, float);
 620                                 os << *((float *) bufpos);
 621                                 bufpos += sizeof(float);
 622                                 break;
 623                         case 's':
 624                                 bufpos += PADDING(bufpos, std::string *);
 625                                 str = **((std::string **) bufpos);
 626
 627                                 strpos = 0;
 628                                 while ((strpos = str.find('"', strpos)) != std::string::npos) {
 629                                         str.insert(strpos, 1, '\\');
 630                                         strpos += 2;
 631                                 }
 632
 633                                 os << str;
 634                                 bufpos += sizeof(std::string *);
 635                                 break;
 636                         case 'v':
 637                                 if (width == 2) {
 638                                         bufpos += PADDING(bufpos, v2f);
 639                                         v2f *v = (v2f *) bufpos;
 640                                         os << '(' << v->X << ", " << v->Y << ')';
 641                                         bufpos += sizeof(v2f);
 642                                 } else {
 643                                         bufpos += PADDING(bufpos, v3f);
 644                                         v3f *v = (v3f *) bufpos;
 645                                         os << '(' << v->X << ", " << v->Y << ", " << v->Z << ')';
 646                                         bufpos += sizeof(v3f);
 647                                 }
 648                                 break;
 649                         default:
 650                                 return false;
 651                 }
 652                 os << ", ";
 653         }
 654         *out = os.str();
 655
 656         // Trim off the trailing comma and space
 657         if (out->size() >= 2)
 658                 out->resize(out->size() - 2);
 659
 660         return true;
 661 }
 662
 663 #undef SIGN_CAST
 664
 665 ////
 666 //// Other
 667 ////
 668
 669 std::string serializeHexString(const std::string &data, bool insert_spaces)
 670 {
 671         std::string result;
 672         result.reserve(data.size() * (2 + insert_spaces));
 673
 674         static const char hex_chars[] = "0123456789abcdef";
 675
 676         const size_t len = data.size();
 677         for (size_t i = 0; i != len; i++) {
 678                 u8 byte = data[i];
 679                 result.push_back(hex_chars[(byte >> 4) & 0x0F]);
 680                 result.push_back(hex_chars[(byte >> 0) & 0x0F]);
 681                 if (insert_spaces && i != len - 1)
 682                         result.push_back(' ');
 683         }
 684
 685         return result;
 686 }