src/SafeFile.cpp

   1 //
   2 // This file is part of the aMule Project.
   3 //
   4 // Copyright (c) 2003-2008 aMule Team ( admin@amule.org / http://www.amule.org )
   5 // Copyright (c) 2002-2008 Merkur ( devs@emule-project.net / http://www.emule-project.net )
   6 //
   7 // Any parts of this program derived from the xMule, lMule or eMule project,
   8 // or contributed by third-party developers are copyrighted by their
   9 // respective authors.
  10 //
  11 // This program is free software; you can redistribute it and/or modify
  12 // it under the terms of the GNU General Public License as published by
  13 // the Free Software Foundation; either version 2 of the License, or
  14 // (at your option) any later version.
  15 //
  16 // This program is distributed in the hope that it will be useful,
  17 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 // GNU General Public License for more details.
  20 //
  21 // You should have received a copy of the GNU General Public License
  22 // along with this program; if not, write to the Free Software
  23 // Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301, USA
  24 //
  25
  26 #include "SafeFile.h"                           // Interface declarations.
  27 #include "MD4Hash.h"                            // Needed for CMD4Hash
  28 #include "kademlia/utils/UInt128.h"     // Needed for CUInt128
  29 #include "ScopedPtr.h"                          // Needed for CScopedPtr and CScopedArray
  30
  31 #if defined(__SUNPRO_CC)
  32 #define __FUNCTION__ __FILE__+__LINE__
  33 #endif
  34
  35 #define CHECK_BOM(size, x) ((size >= 3)  && (x[0] == (char)0xEF) && (x[1] == (char)0xBB) && (x[2] == (char)0xBF))
  36
  37 const char BOMHeader[3] = { '\xEF', '\xBB', '\xBF'};
  38
  39
  40 CSafeIOException::CSafeIOException(const wxString& type, const wxString& desc)
  41         : CMuleException(wxT("SafeIO::") + type, desc) {}
  42
  43
  44 CEOFException::CEOFException(const wxString& desc)
  45         : CSafeIOException(wxT("EOF"), desc) {}
  46
  47
  48 CIOFailureException::CIOFailureException(const wxString& desc)
  49         : CSafeIOException(wxT("IOFailure"), desc) {}
  50
  51 CIOFailureException::CIOFailureException(const wxString& type, const wxString& desc)
  52         : CSafeIOException(wxT("IOFailure::") + type, desc) {}
  53
  54
  55 ///////////////////////////////////////////////////////////////////////////////
  56 // CFileDataIO
  57
  58
  59 CFileDataIO::~CFileDataIO()
  60 {
  61 }
  62
  63
  64 bool CFileDataIO::Eof() const
  65 {
  66         return GetPosition() >= GetLength();
  67 }
  68
  69
  70 void CFileDataIO::Read(void *buffer, size_t count) const
  71 {
  72         MULE_VALIDATE_PARAMS(buffer, wxT("Attempting to write to NULL buffer."));
  73
  74         // Check that we read everything we wanted.
  75         if (doRead(buffer, count) == (signed)count) {
  76                 return;
  77         }
  78
  79         // To reduce potential system calls, we only do EOF checks when reads fail.
  80         if (Eof()) {
  81                 throw CEOFException(wxT("Attempt to read past end of file."));
  82         } else {
  83                 throw CIOFailureException(wxT("Read error, failed to read from file."));
  84         }
  85 }
  86
  87
  88 void CFileDataIO::Write(const void* buffer, size_t count)
  89 {
  90         MULE_VALIDATE_PARAMS(buffer, wxT("Attempting to read from NULL buffer."));
  91
  92         if (doWrite(buffer, count) != (signed)count) {
  93                 throw CIOFailureException(wxT("Write error, failed to write to file."));
  94         }
  95 }
  96
  97
  98 uint64 CFileDataIO::Seek(sint64 offset, wxSeekMode from) const
  99 {
 100         sint64 newpos = 0;
 101         switch (from) {
 102                 case wxFromStart:
 103                         newpos = offset;
 104                         break;
 105
 106                 case wxFromCurrent:
 107                         newpos = GetPosition() + offset;
 108                         break;
 109
 110                 case wxFromEnd:
 111                         newpos = GetLength() + offset;
 112                         break;
 113
 114                 default:
 115                         MULE_VALIDATE_PARAMS(false, wxT("Invalid seek-mode specified."));
 116         }
 117
 118         MULE_VALIDATE_PARAMS(newpos >= 0, wxT("Position after seeking would be less than zero!"));
 119
 120         sint64 result = doSeek(newpos);
 121         MULE_VALIDATE_STATE(result >= 0, wxT("Seeking resulted in invalid offset."));
 122         MULE_VALIDATE_STATE(result == newpos, wxT("Target position and actual position disagree."));
 123
 124         return result;
 125 }
 126
 127
 128 uint8 CFileDataIO::ReadUInt8() const
 129 {
 130         uint8 value = 0;
 131         Read(&value, sizeof(uint8));
 132
 133         return value;
 134 }
 135
 136
 137 uint16 CFileDataIO::ReadUInt16() const
 138 {
 139         uint16 value = 0;
 140         Read(&value, sizeof(uint16));
 141
 142         return ENDIAN_SWAP_16(value);
 143 }
 144
 145
 146 uint32 CFileDataIO::ReadUInt32() const
 147 {
 148         uint32 value = 0;
 149         Read(&value, sizeof(uint32));
 150
 151         return ENDIAN_SWAP_32(value);
 152 }
 153
 154
 155 uint64 CFileDataIO::ReadUInt64() const
 156 {
 157         uint64 value = 0;
 158         Read(&value, sizeof(uint64));
 159
 160         return ENDIAN_SWAP_64(value);
 161 }
 162
 163
 164 // UInt128 values are stored a little weird way...
 165 // Four little-endian 32-bit numbers, stored in
 166 // big-endian order
 167 CUInt128 CFileDataIO::ReadUInt128() const
 168 {
 169         CUInt128 value;
 170         for (int i = 0; i < 4; i++) {
 171                 // Four 32bits chunks
 172                 value.Set32BitChunk(i, ReadUInt32());
 173         }
 174
 175         return value;
 176 }
 177
 178
 179 CMD4Hash CFileDataIO::ReadHash() const
 180 {
 181         CMD4Hash value;
 182         Read(value.GetHash(), MD4HASH_LENGTH);
 183
 184         return value;
 185 }
 186
 187
 188 float CFileDataIO::ReadFloat() const
 189 {
 190         float retVal;
 191         Read(&retVal, sizeof(float));
 192         return retVal;
 193 }
 194
 195
 196 unsigned char* CFileDataIO::ReadBsob(uint8* puSize) const
 197 {
 198         MULE_VALIDATE_PARAMS(puSize, wxT("NULL pointer argument in ReadBsob"));
 199
 200         *puSize = ReadUInt8();
 201
 202         CScopedArray<unsigned char> bsob(new unsigned char[*puSize]);
 203         Read(bsob.get(), *puSize);
 204
 205         return bsob.release();
 206 }
 207
 208
 209 wxString CFileDataIO::ReadString(bool bOptUTF8, uint8 SizeLen, bool SafeRead) const
 210 {
 211         uint32 readLen;
 212         switch (SizeLen) {
 213                 case sizeof(uint16):    readLen = ReadUInt16(); break;
 214                 case sizeof(uint32):    readLen = ReadUInt32(); break;
 215
 216                 default:
 217                         MULE_VALIDATE_PARAMS(false, wxT("Invalid SizeLen value in ReadString"));
 218         }
 219
 220         if (SafeRead) {
 221                 readLen = std::min<uint64>(readLen, GetLength() - GetPosition());
 222         }
 223
 224         return ReadOnlyString(bOptUTF8, readLen);
 225 }
 226
 227
 228 wxString CFileDataIO::ReadOnlyString(bool bOptUTF8, uint16 raw_len) const
 229 {
 230         // We only need to set the the NULL terminator, since we know that
 231         // reads will either succeed or throw an exception, in which case
 232         // we wont be returning anything
 233         std::vector<char> val_array(raw_len + 1);
 234         val_array[raw_len] = 0;
 235
 236         char* val = &(val_array[0]);
 237
 238         Read(val, raw_len);
 239         wxString str;
 240
 241         if (CHECK_BOM(raw_len, val)) {
 242                 // This is a UTF8 string with a BOM header, skip header.
 243                 str = UTF82unicode(val + 3);
 244         } else if (bOptUTF8) {
 245                 str = UTF82unicode(val);
 246                 if (str.IsEmpty()) {
 247                         // Fallback to Latin-1
 248                         str = wxString(val, wxConvISO8859_1, raw_len);
 249                 }
 250         } else {
 251                 // Raw strings are written as Latin-1 (see CFileDataIO::WriteStringCore)
 252                 str = wxString(val, wxConvISO8859_1, raw_len);
 253         }
 254
 255         return str;
 256 }
 257
 258
 259 void CFileDataIO::WriteUInt8(uint8 value)
 260 {
 261         Write(&value, sizeof(uint8));
 262 }
 263
 264
 265 void CFileDataIO::WriteUInt16(uint16 value)
 266 {
 267         ENDIAN_SWAP_I_16(value);
 268
 269         Write(&value, sizeof(uint16));
 270 }
 271
 272
 273 void CFileDataIO::WriteUInt32(uint32 value)
 274 {
 275         ENDIAN_SWAP_I_32(value);
 276
 277         Write(&value, sizeof(uint32));
 278 }
 279
 280
 281 void CFileDataIO::WriteUInt64(uint64 value)
 282 {
 283         ENDIAN_SWAP_I_64(value);
 284
 285         Write(&value, sizeof(uint64));
 286 }
 287
 288
 289 // UInt128 values are stored a little weird way...
 290 // Four little-endian 32-bit numbers, stored in
 291 // big-endian order
 292 void CFileDataIO::WriteUInt128(const Kademlia::CUInt128& value)
 293 {
 294         for (int i = 0; i < 4; i++) {
 295                 // Four 32bits chunks
 296                 WriteUInt32(value.Get32BitChunk(i));
 297         }
 298 }
 299
 300
 301 void CFileDataIO::WriteHash(const CMD4Hash& value)
 302 {
 303         Write(value.GetHash(), MD4HASH_LENGTH);
 304 }
 305
 306
 307 void CFileDataIO::WriteFloat(float value)
 308 {
 309         Write(&value, sizeof(float));
 310 }
 311
 312
 313 void CFileDataIO::WriteBsob(const unsigned char* value, uint8 size)
 314 {
 315         WriteUInt8(size);
 316         Write(value, size);
 317 }
 318
 319
 320 void CFileDataIO::WriteString(const wxString& str, EUtf8Str eEncode, uint8 SizeLen)
 321 {
 322         switch (eEncode) {
 323                 case utf8strRaw:
 324                 case utf8strOptBOM: {
 325                         Unicode2CharBuf s(unicode2UTF8(str));
 326                         if (s) {
 327                                 WriteStringCore(s, eEncode, SizeLen);
 328                                 break;
 329                         }
 330                 }
 331                 default: {
 332                         // Non UTF-8 strings are saved as Latin-1
 333                         wxCharBuffer s1 = wxConvISO8859_1.cWC2MB(str);
 334                         WriteStringCore(s1, utf8strNone, SizeLen);
 335                 }
 336         }
 337 }
 338
 339
 340 void CFileDataIO::WriteStringCore(const char *s, EUtf8Str eEncode, uint8 SizeLen)
 341 {
 342         uint32 sLength = s ? strlen(s) : 0;
 343         uint32 real_length = 0;
 344         if (eEncode == utf8strOptBOM) {
 345                 real_length = sLength + 3; // For BOM header.
 346         } else {
 347                 real_length = sLength;
 348         }
 349
 350         switch (SizeLen) {
 351                 case 0:
 352                         // Don't write size.
 353                         break;
 354
 355                 case sizeof(uint16):
 356                         // We must not allow too long strings to be written,
 357                         // as this would allow for a buggy clients to "poison"
 358                         // us, by sending ISO8859-1 strings that expand to a
 359                         // greater than 16b length when converted as UTF-8.
 360                         if (real_length > 0xFFFF) {
 361                                 wxFAIL_MSG(wxT("String is too long to be saved"));
 362
 363                                 real_length = std::min<uint32>(real_length, 0xFFFF);
 364                                 if (eEncode == utf8strOptBOM) {
 365                                         sLength = real_length - 3;
 366                                 } else {
 367                                         sLength = real_length;
 368                                 }
 369                         }
 370
 371                         WriteUInt16(real_length);
 372                         break;
 373
 374                 case sizeof(uint32):
 375                         WriteUInt32(real_length);
 376                         break;
 377
 378                 default:
 379                         MULE_VALIDATE_PARAMS(false, wxT("Invalid length for string-length field."));
 380         }
 381
 382         // The BOM header must be written even if the string is empty.
 383         if (eEncode == utf8strOptBOM) {
 384                 Write(BOMHeader, 3);
 385         }
 386
 387         // Only attempt to write non-NULL strings.
 388         if (sLength) {
 389                 // No NULL terminator is written since we explicitly specify the length
 390                 Write(s, sLength);
 391         }
 392 }
 393
 394
 395 CTag *CFileDataIO::ReadTag(bool bOptACP) const
 396 {
 397         CTag *retVal = NULL;
 398         wxString name;
 399         byte type = 0;
 400         try {
 401                 type = ReadUInt8();
 402                 name = ReadString(false);
 403
 404                 switch (type)
 405                 {
 406                         // NOTE: This tag data type is accepted and stored only to give us the possibility to upgrade
 407                         // the net in some months.
 408                         //
 409                         // And still.. it doesnt't work this way without breaking backward compatibility. To properly
 410                         // do this without messing up the network the following would have to be done:
 411                         //       -      those tag types have to be ignored by any client, otherwise those tags would also be sent (and
 412                         //              that's really the problem)
 413                         //
 414                         //       -      ignoring means, each client has to read and right throw away those tags, so those tags get
 415                         //              get never stored in any tag list which might be sent by that client to some other client.
 416                         //
 417                         //       -      all calling functions have to be changed to deal with the 'nr. of tags' attribute (which was
 418                         //              already parsed) correctly.. just ignoring those tags here is not enough, any taglists have to
 419                         //              be built with the knowledge that the 'nr. of tags' attribute may get decreased during the tag
 420                         //              reading..
 421                         //
 422                         // If those new tags would just be stored and sent to remote clients, any malicious or just bugged
 423                         // client could let send a lot of nodes "corrupted" packets...
 424                         //
 425                         case TAGTYPE_HASH16:
 426                         {
 427                                 retVal = new CTagHash(name, ReadHash());
 428                                 break;
 429                         }
 430
 431                         case TAGTYPE_STRING:
 432                                 retVal = new CTagString(name, ReadString(bOptACP));
 433                                 break;
 434
 435                         case TAGTYPE_UINT64:
 436                                 retVal = new CTagInt64(name, ReadUInt64());
 437                                 break;
 438
 439                         case TAGTYPE_UINT32:
 440                                 retVal = new CTagInt32(name, ReadUInt32());
 441                                 break;
 442
 443                         case TAGTYPE_UINT16:
 444                                 retVal = new CTagInt16(name, ReadUInt16());
 445                                 break;
 446
 447                         case TAGTYPE_UINT8:
 448                                 retVal = new CTagInt8(name, ReadUInt8());
 449                                 break;
 450
 451                         case TAGTYPE_FLOAT32:
 452                                 retVal = new CTagFloat(name, ReadFloat());
 453                                 break;
 454
 455                         // NOTE: This tag data type is accepted and stored only to give us the possibility to upgrade
 456                         // the net in some months.
 457                         //
 458                         // And still.. it doesnt't work this way without breaking backward compatibility
 459                         case TAGTYPE_BSOB:
 460                         {
 461                                 uint8 size = 0;
 462                                 CScopedArray<unsigned char> value(ReadBsob(&size));
 463
 464                                 retVal = new CTagBsob(name, value.get(), size);
 465                                 break;
 466                         }
 467
 468                         default:
 469                                 throw wxString(wxT("Invalid Kad tag type on packet"));
 470                 }
 471         } catch (...) {
 472                 printf("Invalid Kad tag; type=0x%02x name=%s\n",
 473                         type, (const char *)unicode2char(name));
 474                 delete retVal;
 475                 throw;
 476         }
 477
 478         return retVal;
 479 }
 480
 481
 482 void CFileDataIO::ReadTagPtrList(TagPtrList* taglist, bool bOptACP) const
 483 {
 484         MULE_VALIDATE_PARAMS(taglist, wxT("NULL pointer argument in ReadTagPtrList"));
 485
 486         uint32 count = ReadUInt8();
 487         for (uint32 i = 0; i < count; i++)
 488         {
 489                 CTag* tag = ReadTag(bOptACP);
 490                 taglist->push_back(tag);
 491         }
 492 }
 493
 494
 495 void CFileDataIO::WriteTag(const CTag& tag)
 496 {
 497         try
 498         {
 499                 WriteUInt8(tag.GetType());
 500
 501                 if (!tag.GetName().IsEmpty()) {
 502                         WriteString(tag.GetName(),utf8strNone);
 503                 } else {
 504                         WriteUInt16(1);
 505                         WriteUInt8(tag.GetNameID());
 506                 }
 507
 508                 switch (tag.GetType())
 509                 {
 510                         case TAGTYPE_HASH16:
 511                                 // Do NOT use this to transfer any tags for at least half a year!!
 512                                 WriteHash(CMD4Hash(tag.GetHash()));
 513                                 break;
 514                         case TAGTYPE_STRING:
 515                                 WriteString(tag.GetStr(), utf8strRaw); // Always UTF8
 516                                 break;
 517                         case TAGTYPE_UINT64:
 518                                 WriteUInt64(tag.GetInt());
 519                                 break;
 520                         case TAGTYPE_UINT32:
 521                                 WriteUInt32(tag.GetInt());
 522                                 break;
 523                         case TAGTYPE_FLOAT32:
 524                                 WriteFloat(tag.GetFloat());
 525                                 break;
 526                         case TAGTYPE_BSOB:
 527                                 WriteBsob(tag.GetBsob(), tag.GetBsobSize());
 528                                 break;
 529                         case TAGTYPE_UINT16:
 530                                 WriteUInt16(tag.GetInt());
 531                                 break;
 532                         case TAGTYPE_UINT8:
 533                                 WriteUInt8(tag.GetInt());
 534                                 break;
 535                         case TAGTYPE_BLOB:
 536                                 // NOTE: This will break backward compatibility with met files for eMule versions prior to 0.44a
 537                                 // and any aMule prior to SVN 26/02/2005
 538                                 WriteUInt32(tag.GetBlobSize());
 539                                 Write(tag.GetBlob(), tag.GetBlobSize());
 540                                 break;
 541                         default:
 542                                 //TODO: Support more tag types
 543                                 // With the if above, this should NEVER happen.
 544                                 printf("%s; Unknown tag: type=0x%02X\n", __FUNCTION__, tag.GetType());
 545                                 wxASSERT(0);
 546                                 break;
 547                 }
 548         } catch (...) {
 549                 //AddDebugLogLine(false, wxT("Exception in CDataIO:WriteTag"));
 550                 printf("Exception in CDataIO:WriteTag");
 551                 throw;
 552         }
 553 }
 554
 555
 556 void CFileDataIO::WriteTagPtrList(const TagPtrList& tagList)
 557 {
 558         uint32 count = tagList.size();
 559         wxASSERT( count <= 0xFF );
 560
 561         WriteUInt8(count);
 562         TagPtrList::const_iterator it;
 563         for (it = tagList.begin(); it != tagList.end(); it++) {
 564                 WriteTag(**it);
 565         }
 566 }
 567
 568 uint64 CFileDataIO::GetIntTagValue() const {
 569
 570         uint8 type = ReadUInt8();
 571
 572         ReadString(false);
 573
 574         switch (type) {
 575
 576                 case TAGTYPE_UINT64:
 577                         return ReadUInt64();
 578                         break;
 579
 580                 case TAGTYPE_UINT32:
 581                         return ReadUInt32();
 582                         break;
 583
 584                 case TAGTYPE_UINT16:
 585                         return ReadUInt16();
 586                         break;
 587
 588                 case TAGTYPE_UINT8:
 589                         return ReadUInt8();
 590                         break;
 591
 592                 default:
 593                         throw wxString(wxT("Wrong tag type reading int tag"));
 594         }
 595 }
 596 // File_checked_for_headers