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 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                         } else {
 394                                 tmp_os << c;
 395                         }
 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         } else
 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 = NULL;
 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 : NULL;
 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 = NULL;
 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 }