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