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 FloatType g_serialize_f32_type = FLOATTYPE_UNKNOWN;
  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         s.reserve(2 + plain.size());
 114
 115         writeU16((u8 *)&buf[0], plain.size());
 116         s.append(buf, 2);
 117
 118         s.append(plain);
 119         return s;
 120 }
 121
 122 std::string deSerializeString(std::istream &is)
 123 {
 124         std::string s;
 125         char buf[2];
 126
 127         is.read(buf, 2);
 128         if (is.gcount() != 2)
 129                 throw SerializationError("deSerializeString: size not read");
 130
 131         u16 s_size = readU16((u8 *)buf);
 132         if (s_size == 0)
 133                 return s;
 134
 135         s.resize(s_size);
 136         is.read(&s[0], s_size);
 137         if (is.gcount() != s_size)
 138                 throw SerializationError("deSerializeString: couldn't read all chars");
 139
 140         return s;
 141 }
 142
 143 ////
 144 //// Wide String
 145 ////
 146
 147 std::string serializeWideString(const std::wstring &plain)
 148 {
 149         std::string s;
 150         char buf[2];
 151
 152         if (plain.size() > WIDE_STRING_MAX_LEN)
 153                 throw SerializationError("String too long for serializeWideString");
 154         s.reserve(2 + 2 * plain.size());
 155
 156         writeU16((u8 *)buf, plain.size());
 157         s.append(buf, 2);
 158
 159         for (wchar_t i : plain) {
 160                 writeU16((u8 *)buf, 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         std::string s;
 200         char buf[4];
 201
 202         if (plain.size() > LONG_STRING_MAX_LEN)
 203                 throw SerializationError("String too long for serializeLongString");
 204         s.reserve(4 + plain.size());
 205
 206         writeU32((u8*)&buf[0], plain.size());
 207         s.append(buf, 4);
 208         s.append(plain);
 209         return s;
 210 }
 211
 212 std::string deSerializeLongString(std::istream &is)
 213 {
 214         std::string s;
 215         char buf[4];
 216
 217         is.read(buf, 4);
 218         if (is.gcount() != 4)
 219                 throw SerializationError("deSerializeLongString: size not read");
 220
 221         u32 s_size = readU32((u8 *)buf);
 222         if (s_size == 0)
 223                 return s;
 224
 225         // We don't really want a remote attacker to force us to allocate 4GB...
 226         if (s_size > LONG_STRING_MAX_LEN) {
 227                 throw SerializationError("deSerializeLongString: "
 228                         "string too long: " + itos(s_size) + " bytes");
 229         }
 230
 231         s.resize(s_size);
 232         is.read(&s[0], s_size);
 233         if ((u32)is.gcount() != s_size)
 234                 throw SerializationError("deSerializeLongString: couldn't read all chars");
 235
 236         return s;
 237 }
 238
 239 ////
 240 //// JSON
 241 ////
 242
 243 std::string serializeJsonString(const std::string &plain)
 244 {
 245         std::ostringstream os(std::ios::binary);
 246         os << "\"";
 247
 248         for (char c : plain) {
 249                 switch (c) {
 250                         case '"':
 251                                 os << "\\\"";
 252                                 break;
 253                         case '\\':
 254                                 os << "\\\\";
 255                                 break;
 256                         case '/':
 257                                 os << "\\/";
 258                                 break;
 259                         case '\b':
 260                                 os << "\\b";
 261                                 break;
 262                         case '\f':
 263                                 os << "\\f";
 264                                 break;
 265                         case '\n':
 266                                 os << "\\n";
 267                                 break;
 268                         case '\r':
 269                                 os << "\\r";
 270                                 break;
 271                         case '\t':
 272                                 os << "\\t";
 273                                 break;
 274                         default: {
 275                                 if (c >= 32 && c <= 126) {
 276                                         os << c;
 277                                 } else {
 278                                         u32 cnum = (u8)c;
 279                                         os << "\\u" << std::hex << std::setw(4)
 280                                                 << std::setfill('0') << cnum;
 281                                 }
 282                                 break;
 283                         }
 284                 }
 285         }
 286
 287         os << "\"";
 288         return os.str();
 289 }
 290
 291 std::string deSerializeJsonString(std::istream &is)
 292 {
 293         std::ostringstream os(std::ios::binary);
 294         char c, c2;
 295
 296         // Parse initial doublequote
 297         is >> c;
 298         if (c != '"')
 299                 throw SerializationError("JSON string must start with doublequote");
 300
 301         // Parse characters
 302         for (;;) {
 303                 c = is.get();
 304                 if (is.eof())
 305                         throw SerializationError("JSON string ended prematurely");
 306
 307                 if (c == '"') {
 308                         return os.str();
 309                 }
 310
 311                 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 std::string serializeJsonStringIfNeeded(const std::string &s)
 358 {
 359         for (size_t i = 0; i < s.size(); ++i) {
 360                 if (s[i] <= 0x1f || s[i] >= 0x7f || s[i] == ' ' || s[i] == '\"')
 361                         return serializeJsonString(s);
 362         }
 363         return s;
 364 }
 365
 366 std::string deSerializeJsonStringIfNeeded(std::istream &is)
 367 {
 368         std::ostringstream tmp_os;
 369         bool expect_initial_quote = true;
 370         bool is_json = false;
 371         bool was_backslash = false;
 372         for (;;) {
 373                 char c = is.get();
 374                 if (is.eof())
 375                         break;
 376
 377                 if (expect_initial_quote && c == '"') {
 378                         tmp_os << c;
 379                         is_json = true;
 380                 } else if(is_json) {
 381                         tmp_os << c;
 382                         if (was_backslash)
 383                                 was_backslash = false;
 384                         else if (c == '\\')
 385                                 was_backslash = true;
 386                         else if (c == '"')
 387                                 break; // Found end of string
 388                 } else {
 389                         if (c == ' ') {
 390                                 // Found end of word
 391                                 is.unget();
 392                                 break;
 393                         }
 394
 395                         tmp_os << c;
 396                 }
 397                 expect_initial_quote = false;
 398         }
 399         if (is_json) {
 400                 std::istringstream tmp_is(tmp_os.str(), std::ios::binary);
 401                 return deSerializeJsonString(tmp_is);
 402         }
 403
 404         return tmp_os.str();
 405 }
 406
 407 ////
 408 //// String/Struct conversions
 409 ////
 410
 411 bool deSerializeStringToStruct(std::string valstr,
 412         std::string format, void *out, size_t olen)
 413 {
 414         size_t len = olen;
 415         std::vector<std::string *> strs_alloced;
 416         std::string *str;
 417         char *f, *snext;
 418         size_t pos;
 419
 420         char *s = &valstr[0];
 421         char *buf = new char[len];
 422         char *bufpos = buf;
 423
 424         char *fmtpos, *fmt = &format[0];
 425         while ((f = strtok_r(fmt, ",", &fmtpos)) && s) {
 426                 fmt = nullptr;
 427
 428                 bool is_unsigned = false;
 429                 int width = 0;
 430                 char valtype = *f;
 431
 432                 width = (int)strtol(f + 1, &f, 10);
 433                 if (width && valtype == 's')
 434                         valtype = 'i';
 435
 436                 switch (valtype) {
 437                         case 'u':
 438                                 is_unsigned = true;
 439                                 /* FALLTHROUGH */
 440                         case 'i':
 441                                 if (width == 16) {
 442                                         bufpos += PADDING(bufpos, u16);
 443                                         if ((bufpos - buf) + sizeof(u16) <= len) {
 444                                                 if (is_unsigned)
 445                                                         *(u16 *)bufpos = (u16)strtoul(s, &s, 10);
 446                                                 else
 447                                                         *(s16 *)bufpos = (s16)strtol(s, &s, 10);
 448                                         }
 449                                         bufpos += sizeof(u16);
 450                                 } else if (width == 32) {
 451                                         bufpos += PADDING(bufpos, u32);
 452                                         if ((bufpos - buf) + sizeof(u32) <= len) {
 453                                                 if (is_unsigned)
 454                                                         *(u32 *)bufpos = (u32)strtoul(s, &s, 10);
 455                                                 else
 456                                                         *(s32 *)bufpos = (s32)strtol(s, &s, 10);
 457                                         }
 458                                         bufpos += sizeof(u32);
 459                                 } else if (width == 64) {
 460                                         bufpos += PADDING(bufpos, u64);
 461                                         if ((bufpos - buf) + sizeof(u64) <= len) {
 462                                                 if (is_unsigned)
 463                                                         *(u64 *)bufpos = (u64)strtoull(s, &s, 10);
 464                                                 else
 465                                                         *(s64 *)bufpos = (s64)strtoll(s, &s, 10);
 466                                         }
 467                                         bufpos += sizeof(u64);
 468                                 }
 469                                 s = strchr(s, ',');
 470                                 break;
 471                         case 'b':
 472                                 snext = strchr(s, ',');
 473                                 if (snext)
 474                                         *snext++ = 0;
 475
 476                                 bufpos += PADDING(bufpos, bool);
 477                                 if ((bufpos - buf) + sizeof(bool) <= len)
 478                                         *(bool *)bufpos = is_yes(std::string(s));
 479                                 bufpos += sizeof(bool);
 480
 481                                 s = snext;
 482                                 break;
 483                         case 'f':
 484                                 bufpos += PADDING(bufpos, float);
 485                                 if ((bufpos - buf) + sizeof(float) <= len)
 486                                         *(float *)bufpos = strtof(s, &s);
 487                                 bufpos += sizeof(float);
 488
 489                                 s = strchr(s, ',');
 490                                 break;
 491                         case 's':
 492                                 while (*s == ' ' || *s == '\t')
 493                                         s++;
 494                                 if (*s++ != '"') //error, expected string
 495                                         goto fail;
 496                                 snext = s;
 497
 498                                 while (snext[0] && !(snext[-1] != '\\' && snext[0] == '"'))
 499                                         snext++;
 500                                 *snext++ = 0;
 501
 502                                 bufpos += PADDING(bufpos, std::string *);
 503
 504                                 str = new std::string(s);
 505                                 pos = 0;
 506                                 while ((pos = str->find("\\\"", pos)) != std::string::npos)
 507                                         str->erase(pos, 1);
 508
 509                                 if ((bufpos - buf) + sizeof(std::string *) <= len)
 510                                         *(std::string **)bufpos = str;
 511                                 bufpos += sizeof(std::string *);
 512                                 strs_alloced.push_back(str);
 513
 514                                 s = *snext ? snext + 1 : nullptr;
 515                                 break;
 516                         case 'v':
 517                                 while (*s == ' ' || *s == '\t')
 518                                         s++;
 519                                 if (*s++ != '(') //error, expected vector
 520                                         goto fail;
 521
 522                                 if (width == 2) {
 523                                         bufpos += PADDING(bufpos, v2f);
 524
 525                                         if ((bufpos - buf) + sizeof(v2f) <= len) {
 526                                         v2f *v = (v2f *)bufpos;
 527                                                 v->X = strtof(s, &s);
 528                                                 s++;
 529                                                 v->Y = strtof(s, &s);
 530                                         }
 531
 532                                         bufpos += sizeof(v2f);
 533                                 } else if (width == 3) {
 534                                         bufpos += PADDING(bufpos, v3f);
 535                                         if ((bufpos - buf) + sizeof(v3f) <= len) {
 536                                                 v3f *v = (v3f *)bufpos;
 537                                                 v->X = strtof(s, &s);
 538                                                 s++;
 539                                                 v->Y = strtof(s, &s);
 540                                                 s++;
 541                                                 v->Z = strtof(s, &s);
 542                                         }
 543
 544                                         bufpos += sizeof(v3f);
 545                                 }
 546                                 s = strchr(s, ',');
 547                                 break;
 548                         default: //error, invalid format specifier
 549                                 goto fail;
 550                 }
 551
 552                 if (s && *s == ',')
 553                         s++;
 554
 555                 if ((size_t)(bufpos - buf) > len) //error, buffer too small
 556                         goto fail;
 557         }
 558
 559         if (f && *f) { //error, mismatched number of fields and values
 560 fail:
 561                 for (size_t i = 0; i != strs_alloced.size(); i++)
 562                         delete strs_alloced[i];
 563                 delete[] buf;
 564                 return false;
 565         }
 566
 567         memcpy(out, buf, olen);
 568         delete[] buf;
 569         return true;
 570 }
 571
 572 // Casts *buf to a signed or unsigned fixed-width integer of 'w' width
 573 #define SIGN_CAST(w, buf) (is_unsigned ? *((u##w *) buf) : *((s##w *) buf))
 574
 575 bool serializeStructToString(std::string *out,
 576         std::string format, void *value)
 577 {
 578         std::ostringstream os;
 579         std::string str;
 580         char *f;
 581         size_t strpos;
 582
 583         char *bufpos = (char *) value;
 584         char *fmtpos, *fmt = &format[0];
 585         while ((f = strtok_r(fmt, ",", &fmtpos))) {
 586                 fmt = nullptr;
 587                 bool is_unsigned = false;
 588                 int width = 0;
 589                 char valtype = *f;
 590
 591                 width = (int)strtol(f + 1, &f, 10);
 592                 if (width && valtype == 's')
 593                         valtype = 'i';
 594
 595                 switch (valtype) {
 596                         case 'u':
 597                                 is_unsigned = true;
 598                                 /* FALLTHROUGH */
 599                         case 'i':
 600                                 if (width == 16) {
 601                                         bufpos += PADDING(bufpos, u16);
 602                                         os << SIGN_CAST(16, bufpos);
 603                                         bufpos += sizeof(u16);
 604                                 } else if (width == 32) {
 605                                         bufpos += PADDING(bufpos, u32);
 606                                         os << SIGN_CAST(32, bufpos);
 607                                         bufpos += sizeof(u32);
 608                                 } else if (width == 64) {
 609                                         bufpos += PADDING(bufpos, u64);
 610                                         os << SIGN_CAST(64, bufpos);
 611                                         bufpos += sizeof(u64);
 612                                 }
 613                                 break;
 614                         case 'b':
 615                                 bufpos += PADDING(bufpos, bool);
 616                                 os << std::boolalpha << *((bool *) bufpos);
 617                                 bufpos += sizeof(bool);
 618                                 break;
 619                         case 'f':
 620                                 bufpos += PADDING(bufpos, float);
 621                                 os << *((float *) bufpos);
 622                                 bufpos += sizeof(float);
 623                                 break;
 624                         case 's':
 625                                 bufpos += PADDING(bufpos, std::string *);
 626                                 str = **((std::string **) bufpos);
 627
 628                                 strpos = 0;
 629                                 while ((strpos = str.find('"', strpos)) != std::string::npos) {
 630                                         str.insert(strpos, 1, '\\');
 631                                         strpos += 2;
 632                                 }
 633
 634                                 os << str;
 635                                 bufpos += sizeof(std::string *);
 636                                 break;
 637                         case 'v':
 638                                 if (width == 2) {
 639                                         bufpos += PADDING(bufpos, v2f);
 640                                         v2f *v = (v2f *) bufpos;
 641                                         os << '(' << v->X << ", " << v->Y << ')';
 642                                         bufpos += sizeof(v2f);
 643                                 } else {
 644                                         bufpos += PADDING(bufpos, v3f);
 645                                         v3f *v = (v3f *) bufpos;
 646                                         os << '(' << v->X << ", " << v->Y << ", " << v->Z << ')';
 647                                         bufpos += sizeof(v3f);
 648                                 }
 649                                 break;
 650                         default:
 651                                 return false;
 652                 }
 653                 os << ", ";
 654         }
 655         *out = os.str();
 656
 657         // Trim off the trailing comma and space
 658         if (out->size() >= 2)
 659                 out->resize(out->size() - 2);
 660
 661         return true;
 662 }
 663
 664 #undef SIGN_CAST
 665
 666 ////
 667 //// Other
 668 ////
 669
 670 std::string serializeHexString(const std::string &data, bool insert_spaces)
 671 {
 672         std::string result;
 673         result.reserve(data.size() * (2 + insert_spaces));
 674
 675         static const char hex_chars[] = "0123456789abcdef";
 676
 677         const size_t len = data.size();
 678         for (size_t i = 0; i != len; i++) {
 679                 u8 byte = data[i];
 680                 result.push_back(hex_chars[(byte >> 4) & 0x0F]);
 681                 result.push_back(hex_chars[(byte >> 0) & 0x0F]);
 682                 if (insert_spaces && i != len - 1)
 683                         result.push_back(' ');
 684         }
 685
 686         return result;
 687 }