diff options
Diffstat (limited to 'CPP/7zip/Archive/ExtHandler.cpp')
| -rw-r--r-- | CPP/7zip/Archive/ExtHandler.cpp | 2610 |
1 files changed, 2610 insertions, 0 deletions
diff --git a/CPP/7zip/Archive/ExtHandler.cpp b/CPP/7zip/Archive/ExtHandler.cpp new file mode 100644 index 00000000..6d8101f5 --- /dev/null +++ b/CPP/7zip/Archive/ExtHandler.cpp @@ -0,0 +1,2610 @@ +// ExtHandler.cpp + +#include "StdAfx.h" + +// #define SHOW_DEBUG_INFO + +#ifdef SHOW_DEBUG_INFO +#include <stdio.h> +#define PRF(x) x +#else +#define PRF(x) +#endif + +#include "../../../C/Alloc.h" +#include "../../../C/CpuArch.h" + +#include "../../Common/ComTry.h" +#include "../../Common/IntToString.h" +#include "../../Common/MyLinux.h" +#include "../../Common/StringConvert.h" +#include "../../Common/UTFConvert.h" + +#include "../../Windows/PropVariantUtils.h" +#include "../../Windows/TimeUtils.h" + +#include "../Common/LimitedStreams.h" +#include "../Common/ProgressUtils.h" +#include "../Common/RegisterArc.h" +#include "../Common/StreamObjects.h" +#include "../Common/StreamUtils.h" + +#include "../Compress/CopyCoder.h" + +using namespace NWindows; + +namespace NArchive { +namespace NExt { + +#define Get16(p) GetUi16(p) +#define Get32(p) GetUi32(p) +#define Get64(p) GetUi64(p) + +#define LE_16(offs, dest) dest = Get16(p + (offs)); +#define LE_32(offs, dest) dest = Get32(p + (offs)); +#define LE_64(offs, dest) dest = Get64(p + (offs)); + +#define HI_16(offs, dest) dest |= (((UInt32)Get16(p + (offs))) << 16); +#define HI_32(offs, dest) dest |= (((UInt64)Get32(p + (offs))) << 32); + +/* +static UInt32 g_Crc32CTable[256]; + +static struct CInitCrc32C +{ + CInitCrc32C() + { + for (unsigned i = 0; i < 256; i++) + { + UInt32 r = i; + unsigned j; + for (j = 0; j < 8; j++) + r = (r >> 1) ^ (0x82F63B78 & ~((r & 1) - 1)); + g_Crc32CTable[i] = r; + } + } +} g_InitCrc32C; + +#define CRC32C_INIT_VAL 0xFFFFFFFF +#define CRC32C_GET_DIGEST(crc) ((crc) ^ CRC_INIT_VAL) +#define CRC32C_UPDATE_BYTE(crc, b) (g_Crc32CTable[((crc) ^ (b)) & 0xFF] ^ ((crc) >> 8)) + +static UInt32 Crc32C_Update(UInt32 crc, Byte const *data, size_t size) +{ + for (size_t i = 0; i < size; i++) + crc = CRC32C_UPDATE_BYTE(crc, data[i]); + return crc; +} + +static UInt32 Crc32C_Calc(Byte const *data, size_t size) +{ + return Crc32C_Update(CRC32C_INIT_VAL, data, size); +} +*/ + + +// CRC-16-ANSI. The poly is 0x8005 (x^16 + x^15 + x^2 + 1) +static UInt16 g_Crc16Table[256]; + +static struct CInitCrc16 +{ + CInitCrc16() + { + for (unsigned i = 0; i < 256; i++) + { + UInt32 r = i; + unsigned j; + for (j = 0; j < 8; j++) + r = (r >> 1) ^ (0xA001 & ~((r & 1) - 1)); + g_Crc16Table[i] = (UInt16)r; + } + } +} g_InitCrc16; + +#define CRC16_UPDATE_BYTE(crc, b) (g_Crc16Table[((crc) ^ (b)) & 0xFF] ^ ((crc) >> 8)) +#define CRC16_INIT_VAL 0xFFFF + +static UInt32 Crc16Update(UInt32 crc, Byte const *data, size_t size) +{ + for (size_t i = 0; i < size; i++) + crc = CRC16_UPDATE_BYTE(crc, data[i]); + return crc; +} + +static UInt32 Crc16Calc(Byte const *data, size_t size) +{ + return Crc16Update(CRC16_INIT_VAL, data, size); +} + + +#define EXT4_GOOD_OLD_INODE_SIZE 128 + +// inodes numbers + +// #define k_INODE_BAD 1 // Bad blocks +#define k_INODE_ROOT 2 // Root dir +// #define k_INODE_USR_QUOTA 3 // User quota +// #define k_INODE_GRP_QUOTA 4 // Group quota +// #define k_INODE_BOOT_LOADER 5 // Boot loader +// #define k_INODE_UNDEL_DIR 6 // Undelete dir +#define k_INODE_RESIZE 7 // Reserved group descriptors +// #define k_INODE_JOURNAL 8 // Journal + +// First non-reserved inode for old ext4 filesystems +#define k_INODE_GOOD_OLD_FIRST 11 + +static const char * const k_SysInode_Names[] = +{ + "0" + , "Bad" + , "Root" + , "UserQuota" + , "GroupQuota" + , "BootLoader" + , "Undelete" + , "Resize" + , "Journal" + , "Exclude" + , "Replica" +}; + +static const char * const kHostOS[] = +{ + "Linux" + , "Hurd" + , "Masix" + , "FreeBSD" + , "Lites" +}; + +static const CUInt32PCharPair g_FeatureCompat_Flags[] = +{ + { 0, "DIR_PREALLOC" }, + { 1, "IMAGIC_INODES" }, + { 2, "HAS_JOURNAL" }, + { 3, "EXT_ATTR" }, + { 4, "RESIZE_INODE" }, + { 5, "DIR_INDEX" }, + { 6, "LAZY_BG" }, // not in Linux + // { 7, "EXCLUDE_INODE" }, // not used + // { 8, "EXCLUDE_BITMAP" }, // not in kernel + { 9, "SPARSE_SUPER2" } +}; + + +#define EXT4_FEATURE_INCOMPAT_FILETYPE (1 << 1) +#define EXT4_FEATURE_INCOMPAT_64BIT (1 << 7) + +static const CUInt32PCharPair g_FeatureIncompat_Flags[] = +{ + { 0, "COMPRESSION" }, + { 1, "FILETYPE" }, + { 2, "RECOVER" }, /* Needs recovery */ + { 3, "JOURNAL_DEV" }, /* Journal device */ + { 4, "META_BG" }, + + { 6, "EXTENTS" }, /* extents support */ + { 7, "64BIT" }, + { 8, "MMP" }, + { 9, "FLEX_BG" }, + { 10, "EA_INODE" }, /* EA in inode */ + + { 12, "DIRDATA" }, /* data in dirent */ + { 13, "BG_USE_META_CSUM" }, /* use crc32c for bg */ + { 14, "LARGEDIR" }, /* >2GB or 3-lvl htree */ + { 15, "INLINE_DATA" }, /* data in inode */ + { 16, "ENCRYPT" } +}; + + +static const UInt32 RO_COMPAT_GDT_CSUM = 1 << 4; +static const UInt32 RO_COMPAT_METADATA_CSUM = 1 << 10; + +static const CUInt32PCharPair g_FeatureRoCompat_Flags[] = +{ + { 0, "SPARSE_SUPER" }, + { 1, "LARGE_FILE" }, + { 2, "BTREE_DIR" }, + { 3, "HUGE_FILE" }, + { 4, "GDT_CSUM" }, + { 5, "DIR_NLINK" }, + { 6, "EXTRA_ISIZE" }, + { 7, "HAS_SNAPSHOT" }, + { 8, "QUOTA" }, + { 9, "BIGALLOC" }, + { 10, "METADATA_CSUM" }, + { 11, "REPLICA" }, + { 12, "READONLY" } +}; + + + +static const UInt32 k_NodeFlags_HUGE = (UInt32)1 << 18; +static const UInt32 k_NodeFlags_EXTENTS = (UInt32)1 << 19; + + +static const CUInt32PCharPair g_NodeFlags[] = +{ + { 0, "SECRM" }, + { 1, "UNRM" }, + { 2, "COMPR" }, + { 3, "SYNC" }, + { 4, "IMMUTABLE" }, + { 5, "APPEND" }, + { 6, "NODUMP" }, + { 7, "NOATIME" }, + { 8, "DIRTY" }, + { 9, "COMPRBLK" }, + { 10, "NOCOMPR" }, + { 11, "ENCRYPT" }, + { 12, "INDEX" }, + { 13, "IMAGIC" }, + { 14, "JOURNAL_DATA" }, + { 15, "NOTAIL" }, + { 16, "DIRSYNC" }, + { 17, "TOPDIR" }, + { 18, "HUGE_FILE" }, + { 19, "EXTENTS" }, + + { 21, "EA_INODE" }, + { 22, "EOFBLOCKS" }, + + { 28, "INLINE_DATA" } +}; + + +static inline char GetHex(unsigned t) { return (char)(((t < 10) ? ('0' + t) : ('A' + (t - 10)))); } + +static inline void PrintHex(unsigned v, char *s) +{ + s[0] = GetHex((v >> 4) & 0xF); + s[1] = GetHex(v & 0xF); +} + + +enum +{ + k_Type_UNKNOWN, + k_Type_REG_FILE, + k_Type_DIR, + k_Type_CHRDEV, + k_Type_BLKDEV, + k_Type_FIFO, + k_Type_SOCK, + k_Type_SYMLINK +}; + +static const UInt16 k_TypeToMode[] = +{ + 0, + MY_LIN_S_IFREG, + MY_LIN_S_IFDIR, + MY_LIN_S_IFCHR, + MY_LIN_S_IFBLK, + MY_LIN_S_IFIFO, + MY_LIN_S_IFSOCK, + MY_LIN_S_IFLNK +}; + + +#define EXT4_GOOD_OLD_REV 0 // old (original) format +// #define EXT4_DYNAMIC_REV 1 // V2 format with dynamic inode sizes + +struct CHeader +{ + unsigned BlockBits; + unsigned ClusterBits; + + UInt32 NumInodes; + UInt64 NumBlocks; + // UInt64 NumBlocksSuper; + UInt64 NumFreeBlocks; + UInt32 NumFreeInodes; + UInt32 FirstDataBlock; + + UInt32 BlocksPerGroup; + UInt32 ClustersPerGroup; + UInt32 InodesPerGroup; + + UInt32 MountTime; + UInt32 WriteTime; + + // UInt16 NumMounts; + // UInt16 NumMountsMax; + + // UInt16 State; + // UInt16 Errors; + // UInt16 MinorRevLevel; + + UInt32 LastCheckTime; + // UInt32 CheckInterval; + UInt32 CreatorOs; + UInt32 RevLevel; + + // UInt16 DefResUid; + // UInt16 DefResGid; + + UInt32 FirstInode; + UInt16 InodeSize; + UInt16 BlockGroupNr; + UInt32 FeatureCompat; + UInt32 FeatureIncompat; + UInt32 FeatureRoCompat; + Byte Uuid[16]; + char VolName[16]; + char LastMount[64]; + // UInt32 BitmapAlgo; + + UInt32 JournalInode; + UInt16 GdSize; // = 64 if 64-bit(); + UInt32 CTime; + UInt16 MinExtraISize; + // UInt16 WantExtraISize; + // UInt32 Flags; + // Byte LogGroupsPerFlex; + // Byte ChecksumType; + + UInt64 WrittenKB; + + bool IsOldRev() const { return RevLevel == EXT4_GOOD_OLD_REV; } + + UInt64 GetNumGroups() const { return (NumBlocks + BlocksPerGroup - 1) / BlocksPerGroup; } + + bool IsThereFileType() const { return (FeatureIncompat & EXT4_FEATURE_INCOMPAT_FILETYPE) != 0; } + bool Is64Bit() const { return (FeatureIncompat & EXT4_FEATURE_INCOMPAT_64BIT) != 0; } + bool UseGdtChecksum() const { return (FeatureRoCompat & RO_COMPAT_GDT_CSUM) != 0; } + bool UseMetadataChecksum() const { return (FeatureRoCompat & RO_COMPAT_METADATA_CSUM) != 0; } + + bool Parse(const Byte *p); +}; + + +static int inline GetLog(UInt32 num) +{ + for (unsigned i = 0; i < 32; i++) + if (((UInt32)1 << i) == num) + return i; + return -1; +} + +bool CHeader::Parse(const Byte *p) +{ + if (GetUi16(p + 0x38) != 0xEF53) + return false; + + LE_32 (0x18, BlockBits); + LE_32 (0x1C, ClusterBits); + + if (ClusterBits != 0 && BlockBits != ClusterBits) + return false; + + if (BlockBits > 16 - 10) return false; + BlockBits += 10; + if (ClusterBits > 16) return false; + + LE_32 (0x00, NumInodes); + LE_32 (0x04, NumBlocks); + // LE_32 (0x08, NumBlocksSuper); + LE_32 (0x0C, NumFreeBlocks); + LE_32 (0x10, NumFreeInodes); + LE_32 (0x14, FirstDataBlock); + + if (FirstDataBlock != 0) + return false; + + LE_32 (0x20, BlocksPerGroup); + LE_32 (0x24, ClustersPerGroup); + + if (BlocksPerGroup != ClustersPerGroup) + return false; + if (BlocksPerGroup != ((UInt32)1 << (BlockBits + 3))) + return false; + + LE_32 (0x28, InodesPerGroup); + + LE_32 (0x2C, MountTime); + LE_32 (0x30, WriteTime); + + // LE_16 (0x34, NumMounts); + // LE_16 (0x36, NumMountsMax); + + // LE_16 (0x3A, State); + // LE_16 (0x3C, Errors); + // LE_16 (0x3E, MinorRevLevel); + + LE_32 (0x40, LastCheckTime); + // LE_32 (0x44, CheckInterval); + LE_32 (0x48, CreatorOs); + LE_32 (0x4C, RevLevel); + + // LE_16 (0x50, DefResUid); + // LE_16 (0x52, DefResGid); + + FirstInode = k_INODE_GOOD_OLD_FIRST; + InodeSize = EXT4_GOOD_OLD_INODE_SIZE; + + if (!IsOldRev()) + { + LE_32 (0x54, FirstInode); + LE_16 (0x58, InodeSize); + if (FirstInode < k_INODE_GOOD_OLD_FIRST) + return false; + if (InodeSize > (UInt32)1 << BlockBits) + return false; + if (GetLog(InodeSize) < 0) + return false; + } + + LE_16 (0x5A, BlockGroupNr); + LE_32 (0x5C, FeatureCompat); + LE_32 (0x60, FeatureIncompat); + LE_32 (0x64, FeatureRoCompat); + + memcpy(Uuid, p + 0x68, sizeof(Uuid)); + memcpy(VolName, p + 0x78, sizeof(VolName)); + memcpy(LastMount, p + 0x88, sizeof(LastMount)); + + // LE_32 (0xC8, BitmapAlgo); + + LE_32 (0xE0, JournalInode); + + LE_16 (0xFE, GdSize); + + LE_32 (0x108, CTime); + + if (Is64Bit()) + { + HI_32(150, NumBlocks); + // HI_32(154, NumBlocksSuper); + HI_32(0x158, NumFreeBlocks); + } + + if (NumBlocks >= (UInt64)1 << (63 - BlockBits)) + return false; + + + LE_16(0x15C, MinExtraISize); + // LE_16(0x15E, WantExtraISize); + // LE_32(0x160, Flags); + // LE_16(0x164, RaidStride); + // LE_16(0x166, MmpInterval); + // LE_64(0x168, MmpBlock); + + // LogGroupsPerFlex = p[0x174]; + // ChecksumType = p[0x175]; + + LE_64 (0x178, WrittenKB); + + // LE_32(0x194, ErrorCount); + // LE_32(0x198, ErrorTime); + // LE_32(0x19C, ErrorINode); + // LE_32(0x1A0, ErrorBlock); + + if (NumBlocks < 1) + return false; + if (NumFreeBlocks > NumBlocks) + return false; + + return true; +} + + +struct CGroupDescriptor +{ + UInt64 BlockBitmap; + UInt64 InodeBitmap; + UInt64 InodeTable; + UInt32 NumFreeBlocks; + UInt32 NumFreeInodes; + UInt32 DirCount; + + UInt16 Flags; + + UInt64 ExcludeBitmap; + UInt32 BlockBitmap_Checksum; + UInt32 InodeBitmap_Checksum; + UInt32 UnusedCount; + UInt16 Checksum; + + void Parse(const Byte *p, unsigned size); +}; + +void CGroupDescriptor::Parse(const Byte *p, unsigned size) +{ + LE_32 (0x00, BlockBitmap); + LE_32 (0x04, InodeBitmap); + LE_32 (0x08, InodeTable); + LE_16 (0x0C, NumFreeBlocks); + LE_16 (0x0E, NumFreeInodes); + LE_16 (0x10, DirCount); + LE_16 (0x12, Flags); + LE_32 (0x14, ExcludeBitmap); + LE_16 (0x18, BlockBitmap_Checksum); + LE_16 (0x1A, InodeBitmap_Checksum); + LE_16 (0x1C, UnusedCount); + LE_16 (0x1E, Checksum); + + if (size >= 64) + { + p += 0x20; + HI_32 (0x00, BlockBitmap); + HI_32 (0x04, InodeBitmap); + HI_32 (0x08, InodeTable); + HI_16 (0x0C, NumFreeBlocks); + HI_16 (0x0E, NumFreeInodes); + HI_16 (0x10, DirCount); + HI_16 (0x12, UnusedCount); // instead of Flags + HI_32 (0x14, ExcludeBitmap); + HI_16 (0x18, BlockBitmap_Checksum); + HI_16 (0x1A, InodeBitmap_Checksum); + // HI_16 (0x1C, Reserved); + } +} + + +static const unsigned kNodeBlockFieldSize = 60; + +struct CExtentTreeHeader +{ + UInt16 NumEntries; + UInt16 MaxEntries; + UInt16 Depth; + // UInt32 Generation; + + bool Parse(const Byte *p) + { + LE_16 (0x02, NumEntries); + LE_16 (0x04, MaxEntries); + LE_16 (0x06, Depth); + // LE_32 (0x08, Generation); + return Get16(p) == 0xF30A; // magic + } +}; + +struct CExtentIndexNode +{ + UInt32 VirtBlock; + UInt64 PhyLeaf; + + void Parse(const Byte *p) + { + LE_32 (0x00, VirtBlock); + LE_32 (0x04, PhyLeaf); + PhyLeaf |= (((UInt64)Get16(p + 8)) << 32); + // unused 16-bit field (at offset 0x0A) can be not zero in some cases. Why? + } +}; + +struct CExtent +{ + UInt32 VirtBlock; + UInt16 Len; + bool IsInited; + UInt64 PhyStart; + + UInt32 GetVirtEnd() const { return VirtBlock + Len; } + bool IsLenOK() const { return VirtBlock + Len >= VirtBlock; } + + void Parse(const Byte *p) + { + LE_32 (0x00, VirtBlock); + LE_16 (0x04, Len); + IsInited = true; + if (Len > (UInt32)0x8000) + { + IsInited = false; + Len -= (UInt32)0x8000; + } + LE_32 (0x08, PhyStart); + UInt16 hi; + LE_16 (0x06, hi); + PhyStart |= ((UInt64)hi << 32); + } +}; + + + +struct CExtTime +{ + UInt32 Val; + UInt32 Extra; +}; + +struct CNode +{ + Int32 ParentNode; // in _nodes[], -1 if not dir + int ItemIndex; // in _items[] + int SymLinkIndex; // in _symLinks[] + int DirIndex; // in _dirs[] + + UInt16 Mode; + UInt16 Uid; + UInt16 Gid; + // UInt16 Checksum; + bool IsEmpty; + + UInt64 FileSize; + CExtTime MTime; + CExtTime ATime; + CExtTime CTime; + // CExtTime InodeChangeTime; + // CExtTime DTime; + + UInt64 NumBlocks; + UInt32 NumLinks; + UInt32 Flags; + + UInt32 NumLinksCalced; + + Byte Block[kNodeBlockFieldSize]; + + CNode(): + ParentNode(-1), + ItemIndex(-1), + SymLinkIndex(-1), + DirIndex(0), + NumLinksCalced(0) + {} + + bool IsFlags_HUGE() const { return (Flags & k_NodeFlags_HUGE) != 0; } + bool IsFlags_EXTENTS() const { return (Flags & k_NodeFlags_EXTENTS) != 0; } + + bool IsDir() const { return MY_LIN_S_ISDIR(Mode); } + bool IsRegular() const { return MY_LIN_S_ISREG(Mode); } + bool IsLink() const { return MY_LIN_S_ISLNK(Mode); } + + bool Parse(const Byte *p, const CHeader &_h); +}; + + +bool CNode::Parse(const Byte *p, const CHeader &_h) +{ + MTime.Extra = 0; + ATime.Extra = 0; + CTime.Extra = 0; + // InodeChangeTime.Extra = 0; + // DTime.Extra = 0; + + LE_16 (0x00, Mode); + LE_16 (0x02, Uid); + LE_32 (0x04, FileSize); + LE_32 (0x08, ATime.Val); + // LE_32 (0x0C, InodeChangeTime.Val); + LE_32 (0x10, MTime.Val); + // LE_32 (0x14, DTime.Val); + LE_16 (0x18, Gid); + LE_16 (0x1A, NumLinks); + + LE_32 (0x1C, NumBlocks); + LE_32 (0x20, Flags); + // LE_32 (0x24, Union osd1); + + memcpy(Block, p + 0x28, kNodeBlockFieldSize); + + // LE_32 (0x64, Generation); // File version (for NFS) + // LE_32 (0x68, ACL); + + { + UInt32 highSize; + LE_32 (0x6C, highSize); // In ext2/3 this field was named i_dir_acl + + if (IsRegular()) // do we need that check ? + FileSize |= ((UInt64)highSize << 32); + } + + // LE_32 (0x70, fragmentAddress); + + // osd2 + { + // Linux; + UInt32 numBlocksHigh; + LE_16 (0x74, numBlocksHigh); + NumBlocks |= (UInt64)numBlocksHigh << 32; + HI_16 (0x74 + 4, Uid); + HI_16 (0x74 + 6, Gid); + /* + UInt32 checksum; + LE_16 (0x74 + 8, checksum); + checksum = checksum; + */ + } + + // 0x74: Byte Union osd2[12]; + + if (_h.InodeSize > 128) + { + UInt16 extra_isize; + LE_16 (0x80, extra_isize); + if (128 + extra_isize > _h.InodeSize) + return false; + if (extra_isize >= 0x1C) + { + // UInt16 checksumUpper; + // LE_16 (0x82, checksumUpper); + // LE_32 (0x84, InodeChangeTime.Extra); + LE_32 (0x88, MTime.Extra); + LE_32 (0x8C, ATime.Extra); + LE_32 (0x90, CTime.Val); + LE_32 (0x94, CTime.Extra); + // LE_32 (0x98, VersionHi); + } + } + + PRF(printf("size = %5d", (unsigned)FileSize)); + + return true; +} + + +struct CItem +{ + unsigned Node; // in _nodes[] + int ParentNode; // in _nodes[] + int SymLinkItemIndex; // in _items[], if the Node contains SymLink to existing dir + Byte Type; + + AString Name; + + CItem(): + Node(0), + ParentNode(-1), + SymLinkItemIndex(-1), + Type(k_Type_UNKNOWN) + {} + + void Clear() + { + Node = 0; + ParentNode = -1; + SymLinkItemIndex = -1; + Type = k_Type_UNKNOWN; + Name.Empty(); + } + + bool IsDir() const { return Type == k_Type_DIR; } + // bool IsNotDir() const { return Type != k_Type_DIR && Type != k_Type_UNKNOWN; } + +}; + + + +static const unsigned kNumTreeLevelsMax = 6; // must be >= 3 + + +class CHandler: + public IInArchive, + public IArchiveGetRawProps, + public IInArchiveGetStream, + public CMyUnknownImp +{ + CObjectVector<CItem> _items; + CRecordVector<CNode> _nodes; + CObjectVector<CUIntVector> _dirs; // each CUIntVector contains indexes in _items[] only for dir items; + AStringVector _symLinks; + AStringVector _auxItems; + int _auxSysIndex; + int _auxUnknownIndex; + + CMyComPtr<IInStream> _stream; + UInt64 _phySize; + bool _isArc; + bool _headersError; + bool _linksError; + + bool _isUTF; + + CHeader _h; + + IArchiveOpenCallback *_openCallback; + UInt64 _totalRead; + UInt64 _totalReadPrev; + + CByteBuffer _tempBufs[kNumTreeLevelsMax]; + + + HRESULT CheckProgress2() + { + const UInt64 numFiles = _items.Size(); + return _openCallback->SetCompleted(&numFiles, &_totalRead); + } + + HRESULT CheckProgress() + { + HRESULT res = S_OK; + if (_openCallback) + { + if (_totalRead - _totalReadPrev >= ((UInt32)1 << 20)) + { + _totalReadPrev = _totalRead; + res = CheckProgress2(); + } + } + return res; + } + + + const int GetParentAux(const CItem &item) const + { + if (item.Node < _h.FirstInode && _auxSysIndex >= 0) + return _auxSysIndex; + return _auxUnknownIndex; + } + + HRESULT SeekAndRead(IInStream *inStream, UInt64 block, Byte *data, size_t size); + HRESULT ParseDir(const Byte *data, size_t size, unsigned nodeIndex); + int FindTargetItem_for_SymLink(unsigned dirNode, const AString &path) const; + + HRESULT FillFileBlocks2(UInt32 block, unsigned level, unsigned numBlocks, CRecordVector<UInt32> &blocks); + HRESULT FillFileBlocks(const Byte *p, unsigned numBlocks, CRecordVector<UInt32> &blocks); + HRESULT FillExtents(const Byte *p, size_t size, CRecordVector<CExtent> &extents, int parentDepth); + + HRESULT GetStream_Node(UInt32 nodeIndex, ISequentialInStream **stream); + HRESULT ExtractNode(unsigned nodeIndex, CByteBuffer &data); + + void ClearRefs(); + HRESULT Open2(IInStream *inStream); + + void GetPath(unsigned index, AString &s) const; + bool GetPackSize(unsigned index, UInt64 &res) const; + +public: + CHandler() {} + ~CHandler() {} + + MY_UNKNOWN_IMP3(IInArchive, IArchiveGetRawProps, IInArchiveGetStream) + + INTERFACE_IInArchive(;) + INTERFACE_IArchiveGetRawProps(;) + STDMETHOD(GetStream)(UInt32 index, ISequentialInStream **stream); +}; + + + +HRESULT CHandler::ParseDir(const Byte *p, size_t size, unsigned nodeIndex) +{ + bool isThereSelfLink = false; + + PRF(printf("\n\n========= node = %5d size = %5d", nodeIndex, size)); + + CNode &nodeDir = _nodes[nodeIndex]; + nodeDir.DirIndex = _dirs.Size(); + CUIntVector &dir = _dirs.AddNew(); + int parentNode = -1; + + CItem item; + + for (;;) + { + if (size == 0) + break; + if (size < 8) + return S_FALSE; + UInt32 iNode; + LE_32 (0x00, iNode); + unsigned recLen; + LE_16 (0x04, recLen); + unsigned nameLen = p[6]; + Byte type = p[7]; + + if (recLen > size) + return S_FALSE; + if (nameLen + 8 > recLen) + return S_FALSE; + + if (iNode >= _nodes.Size()) + return S_FALSE; + + item.Clear(); + + if (_h.IsThereFileType()) + item.Type = type; + else if (type != 0) + return S_FALSE; + + item.ParentNode = nodeIndex; + item.Node = iNode; + item.Name.SetFrom_CalcLen((const char *)(p + 8), nameLen); + + p += recLen; + size -= recLen; + + if (item.Name.Len() != nameLen) + return S_FALSE; + + if (_isUTF) + _isUTF = CheckUTF8(item.Name); + + if (iNode == 0) + { + /* + ext3 deleted?? + if (item.Name.Len() != 0) + return S_FALSE; + */ + + PRF(printf("\n EMPTY %6d %d %s", recLen, type, (const char *)item.Name)); + if (type == 0xDE) + { + // checksum + } + continue; + } + + CNode &node = _nodes[iNode]; + if (node.IsEmpty) + return S_FALSE; + + if (_h.IsThereFileType() && type != 0) + { + if (type >= ARRAY_SIZE(k_TypeToMode)) + return S_FALSE; + if (k_TypeToMode[type] != (node.Mode & MY_LIN_S_IFMT)) + return S_FALSE; + } + + node.NumLinksCalced++; + + PRF(printf("\n%s %6d %s", item.IsDir() ? "DIR " : " ", item.Node, (const char *)item.Name)); + + if (item.Name[0] == '.') + { + if (item.Name[1] == 0) + { + if (isThereSelfLink) + return S_FALSE; + isThereSelfLink = true; + if (nodeIndex != nodeIndex) + return S_FALSE; + continue; + } + + if (item.Name[1] == '.' && item.Name[2] == 0) + { + if (parentNode >= 0) + return S_FALSE; + if (!node.IsDir()) + return S_FALSE; + if (iNode == nodeIndex && iNode != k_INODE_ROOT) + return S_FALSE; + + parentNode = iNode; + + if (nodeDir.ParentNode < 0) + nodeDir.ParentNode = iNode; + else if ((unsigned)nodeDir.ParentNode != iNode) + return S_FALSE; + + continue; + } + } + + if (iNode == nodeIndex) + return S_FALSE; + + if (parentNode < 0) + return S_FALSE; + + if (node.IsDir()) + { + if (node.ParentNode < 0) + node.ParentNode = nodeIndex; + else if ((unsigned)node.ParentNode != nodeIndex) + return S_FALSE; + const unsigned itemIndex = _items.Size(); + dir.Add(itemIndex); + node.ItemIndex = itemIndex; + } + + _items.Add(item); + } + + if (parentNode < 0 || !isThereSelfLink) + return S_FALSE; + + return S_OK; +} + + +int CHandler::FindTargetItem_for_SymLink(unsigned nodeIndex, const AString &path) const +{ + unsigned pos = 0; + + if (path.IsEmpty()) + return -1; + + if (path[0] == '/') + { + nodeIndex = k_INODE_ROOT; + if (nodeIndex >= _nodes.Size()) + return -1; + pos = 1; + } + + AString s; + + while (pos != path.Len()) + { + const CNode &node = _nodes[nodeIndex]; + int slash = path.Find('/', pos); + + if (slash < 0) + { + s = path.Ptr(pos); + pos = path.Len(); + } + else + { + s.SetFrom(path.Ptr(pos), slash - pos); + pos = slash + 1; + } + + if (s[0] == '.') + { + if (s[1] == 0) + continue; + else if (s[1] == '.' && s[2] == 0) + { + if (node.ParentNode < 0) + return -1; + if (nodeIndex == k_INODE_ROOT) + return -1; + nodeIndex = node.ParentNode; + continue; + } + } + + if (node.DirIndex < 0) + return -1; + + const CUIntVector &dir = _dirs[node.DirIndex]; + + for (unsigned i = 0;; i++) + { + if (i >= dir.Size()) + return -1; + const CItem &item = _items[dir[i]]; + if (item.Name == s) + { + nodeIndex = item.Node; + break; + } + } + } + + return _nodes[nodeIndex].ItemIndex; +} + + +HRESULT CHandler::SeekAndRead(IInStream *inStream, UInt64 block, Byte *data, size_t size) +{ + if (block == 0 || block >= _h.NumBlocks) + return S_FALSE; + if (((size + (1 << _h.BlockBits) + 1) >> _h.BlockBits) > _h.NumBlocks - block) + return S_FALSE; + RINOK(inStream->Seek((UInt64)block << _h.BlockBits, STREAM_SEEK_SET, NULL)); + _totalRead += size; + return ReadStream_FALSE(inStream, data, size); +} + + +static const unsigned kHeaderSize = 2 * 1024; +static const unsigned kHeaderDataOffset = 1024; + +HRESULT CHandler::Open2(IInStream *inStream) +{ + { + Byte buf[kHeaderSize]; + RINOK(ReadStream_FALSE(inStream, buf, kHeaderSize)); + if (!_h.Parse(buf + kHeaderDataOffset)) + return S_FALSE; + if (_h.BlockGroupNr != 0) + return S_FALSE; // it's just copy of super block + } + + { + // ---------- Read groups and nodes ---------- + + unsigned numGroups; + { + UInt64 numGroups64 = _h.GetNumGroups(); + if (numGroups64 > (UInt32)1 << 31) + return S_FALSE; + numGroups = (unsigned)numGroups64; + } + + unsigned gdBits = 5; + if (_h.Is64Bit()) + { + if (_h.GdSize != 64) + return S_FALSE; + gdBits = 6; + } + + _isArc = true; + _phySize = _h.NumBlocks << _h.BlockBits; + + if (_openCallback) + { + RINOK(_openCallback->SetTotal(NULL, &_phySize)); + } + + CRecordVector<CGroupDescriptor> groups; + + { + // ---------- Read groups ---------- + + CByteBuffer gdBuf; + size_t gdBufSize = (size_t)numGroups << gdBits; + if ((gdBufSize >> gdBits) != numGroups) + return S_FALSE; + gdBuf.Alloc(gdBufSize); + RINOK(SeekAndRead(inStream, (_h.BlockBits <= 10 ? 2 : 1), gdBuf, gdBufSize)); + + for (unsigned i = 0; i < numGroups; i++) + { + CGroupDescriptor gd; + + const Byte *p = gdBuf + ((size_t)i << gdBits); + unsigned gd_Size = (unsigned)1 << gdBits; + gd.Parse(p, gd_Size); + + if (_h.UseMetadataChecksum()) + { + // use CRC32c + } + else if (_h.UseGdtChecksum()) + { + UInt32 crc = Crc16Calc(_h.Uuid, sizeof(_h.Uuid)); + Byte i_le[4]; + SetUi32(i_le, i); + crc = Crc16Update(crc, i_le, 4); + crc = Crc16Update(crc, p, 32 - 2); + if (gd_Size != 32) + crc = Crc16Update(crc, p + 32, gd_Size - 32); + if (crc != gd.Checksum) + return S_FALSE; + } + + groups.Add(gd); + } + } + + { + // ---------- Read nodes ---------- + + if (_h.NumInodes < _h.NumFreeInodes) + return S_FALSE; + + UInt32 numReserveInodes = _h.NumInodes - _h.NumFreeInodes + 1; + // numReserveInodes = _h.NumInodes + 1; + if (numReserveInodes != 0) + _nodes.Reserve(numReserveInodes); + + UInt32 numNodes = _h.InodesPerGroup; + if (numNodes > _h.NumInodes) + numNodes = _h.NumInodes; + size_t nodesDataSize = numNodes * _h.InodeSize; + if (nodesDataSize / _h.InodeSize != numNodes) + return S_FALSE; + + CByteBuffer nodesData; + nodesData.Alloc(nodesDataSize); + + CByteBuffer nodesMap; + const size_t blockSize = (size_t)1 << _h.BlockBits; + nodesMap.Alloc(blockSize); + + unsigned globalNodeIndex = 0; + + FOR_VECTOR (gi, groups) + { + if (globalNodeIndex >= _h.NumInodes) + break; + + const CGroupDescriptor &gd = groups[gi]; + + PRF(printf("\n\ng%6d block = %6x\n", gi, gd.InodeTable)); + + RINOK(SeekAndRead(inStream, gd.InodeBitmap, nodesMap, blockSize)); + RINOK(SeekAndRead(inStream, gd.InodeTable, nodesData, nodesDataSize)); + + for (size_t n = 0; n < numNodes && globalNodeIndex < _h.NumInodes; n++, globalNodeIndex++) + { + if ((nodesMap[n >> 3] & ((unsigned)1 << (n & 7))) == 0) + continue; + + const Byte *p = nodesData + (size_t)n * _h.InodeSize; + unsigned j = 0; + for (j = 0; j < _h.InodeSize; j++) + if (p[j] != 0) + break; + + if (j == _h.InodeSize) + { + if (_nodes.Size() >= _h.FirstInode) + { + // return S_FALSE; + } + continue; + } + + CNode node; + node.IsEmpty = false; + + PRF(printf("\nnode = %5d ", (unsigned)n)); + + if (!node.Parse(p, _h)) + return S_FALSE; + + // PRF(printf("\n %6d", n)); + /* + SetUi32(p + 0x7C, 0) + SetUi32(p + 0x82, 0) + + UInt32 crc = Crc32C_Calc(_h.Uuid, sizeof(_h.Uuid)); + Byte i_le[4]; + SetUi32(i_le, n); + crc = Crc32C_Update(crc, i_le, 4); + crc = Crc32C_Update(crc, p, _h.InodeSize); + if (crc != node.Checksum) return S_FALSE; + */ + + while (_nodes.Size() < globalNodeIndex + 1) + { + CNode node2; + node2.IsEmpty = true; + _nodes.Add(node2); + } + + _nodes.Add(node); + } + } + + if (_nodes.Size() <= k_INODE_ROOT) + return S_FALSE; + } + } + + _stream = inStream; // we need stream for dir nodes + + { + // ---------- Read Dirs ---------- + + CByteBuffer dataBuf; + + FOR_VECTOR (i, _nodes) + { + { + const CNode &node = _nodes[i]; + if (node.IsEmpty || !node.IsDir()) + continue; + } + RINOK(ExtractNode(i, dataBuf)); + if (dataBuf.Size() == 0) + { + // _headersError = true; + return S_FALSE; + } + else + { + RINOK(ParseDir(dataBuf, dataBuf.Size(), i)); + } + RINOK(CheckProgress()); + } + + if (_nodes[k_INODE_ROOT].ParentNode != k_INODE_ROOT) + return S_FALSE; + } + + { + // ---------- Check NumLinks and unreferenced dir nodes ---------- + + FOR_VECTOR (i, _nodes) + { + const CNode &node = _nodes[i]; + if (node.IsEmpty) + continue; + + if (node.NumLinks != node.NumLinksCalced) + { + if (node.NumLinks != 1 || node.NumLinksCalced != 0 + // ) && i >= _h.FirstInode + ) + { + _linksError = true; + // return S_FALSE; + } + } + + if (!node.IsDir()) + continue; + + if (node.ParentNode < 0) + { + if (i >= _h.FirstInode) + return S_FALSE; + continue; + } + } + } + + { + // ---------- Check that there is no loops in parents list ---------- + + unsigned numNodes = _nodes.Size(); + CIntArr UsedByNode(numNodes); + { + for (unsigned i = 0; i < numNodes; i++) + UsedByNode[i] = -1; + } + + FOR_VECTOR (i, _nodes) + { + { + CNode &node = _nodes[i]; + if (node.IsEmpty + || node.ParentNode < 0 // not dir + || i == k_INODE_ROOT) + continue; + } + + unsigned c = i; + + for (;;) + { + CNode &node = _nodes[c]; + if (node.IsEmpty) + return S_FALSE; + + if (UsedByNode[c] != -1) + { + if ((unsigned)UsedByNode[c] == i) + return S_FALSE; + break; + } + + UsedByNode[c] = i; + if (node.ParentNode < 0 || node.ParentNode == k_INODE_ROOT) + break; + if ((unsigned)node.ParentNode == i) + return S_FALSE; + c = node.ParentNode; + } + } + } + + { + // ---------- Fill SymLinks data ---------- + + AString s; + CByteBuffer data; + + unsigned i; + for (i = 0; i < _nodes.Size(); i++) + { + CNode &node = _nodes[i]; + if (node.IsEmpty || !node.IsLink()) + continue; + if (node.FileSize > ((UInt32)1 << 14)) + continue; + if (ExtractNode(i, data) == S_OK && data.Size() != 0) + { + s.SetFrom_CalcLen((const char *)(const Byte *)data, (unsigned)data.Size()); + if (s.Len() == data.Size()) + node.SymLinkIndex = _symLinks.Add(s); + RINOK(CheckProgress()); + } + } + + unsigned prev = 0; + unsigned complex = 0; + + for (i = 0; i < _items.Size(); i++) + { + CItem &item = _items[i]; + int sym = _nodes[item.Node].SymLinkIndex; + if (sym >= 0 && item.ParentNode >= 0) + { + item.SymLinkItemIndex = FindTargetItem_for_SymLink(item.ParentNode, _symLinks[sym]); + if (_openCallback) + { + complex++; + if (complex - prev >= (1 << 10)) + { + RINOK(CheckProgress2()); + prev = complex; + } + } + } + } + } + + { + // ---------- Add items and aux folders for unreferenced files ---------- + + bool useSys = false; + bool useUnknown = false; + + for (unsigned i = 0; i < _nodes.Size(); i++) + { + const CNode &node = _nodes[i]; + if (node.IsEmpty) + continue; + + if (node.NumLinksCalced == 0 /* || i > 100 && i < 150 */) // for debug + { + CItem item; + item.Node = i; + + // we don't know how to work with k_INODE_RESIZE node (strange FileSize and Block values). + // so we ignore it; + + if (i == k_INODE_RESIZE) + continue; + + if (node.FileSize == 0) + continue; + + if (i < _h.FirstInode) + { + if (item.Node < ARRAY_SIZE(k_SysInode_Names)) + item.Name = k_SysInode_Names[item.Node]; + useSys = true; + } + else + useUnknown = true; + + if (item.Name.IsEmpty()) + { + char temp[16]; + ConvertUInt32ToString(item.Node, temp); + item.Name = temp; + } + + _items.Add(item); + } + } + + if (useSys) + _auxSysIndex = _auxItems.Add("[SYS]"); + if (useUnknown) + _auxUnknownIndex = _auxItems.Add("[UNKNOWN]"); + } + + return S_OK; +} + + +STDMETHODIMP CHandler::Open(IInStream *stream, const UInt64 *, IArchiveOpenCallback *callback) +{ + COM_TRY_BEGIN + { + Close(); + HRESULT res; + try + { + _openCallback = callback; + res = Open2(stream); + } + catch(...) + { + ClearRefs(); + throw; + } + + if (res != S_OK) + { + ClearRefs(); + return res; + } + _stream = stream; + } + return S_OK; + COM_TRY_END +} + + +void CHandler::ClearRefs() +{ + _stream.Release(); + _items.Clear(); + _nodes.Clear(); + _auxItems.Clear(); + _symLinks.Clear(); + _dirs.Clear(); + _auxSysIndex = -1; + _auxUnknownIndex = -1; +} + + +STDMETHODIMP CHandler::Close() +{ + _totalRead = 0; + _totalReadPrev = 0; + _phySize = 0; + _isArc = false; + _headersError = false; + _linksError = false; + _isUTF = true; + + ClearRefs(); + return S_OK; +} + + +void CHandler::GetPath(unsigned index, AString &s) const +{ + s.Empty(); + + if (index >= _items.Size()) + { + s = _auxItems[index - _items.Size()]; + return; + } + + for (;;) + { + const CItem &item = _items[index]; + if (!s.IsEmpty()) + s.InsertAtFront(CHAR_PATH_SEPARATOR); + s.Insert(0, item.Name); + + if (item.ParentNode == k_INODE_ROOT) + return; + + if (item.ParentNode < 0) + { + int aux = GetParentAux(item); + if (aux < 0) + break; + s.InsertAtFront(CHAR_PATH_SEPARATOR); + s.Insert(0, _auxItems[aux]); + return; + } + + const CNode &node = _nodes[item.ParentNode]; + if (node.ItemIndex < 0) + return; + index = node.ItemIndex; + + if (s.Len() > ((UInt32)1 << 16)) + { + s.Insert(0, "[LONG]" STRING_PATH_SEPARATOR); + return; + } + } +} + + +bool CHandler::GetPackSize(unsigned index, UInt64 &totalPack) const +{ + if (index >= _items.Size()) + { + totalPack = 0; + return false; + } + + const CItem &item = _items[index]; + const CNode &node = _nodes[item.Node]; + + // if (!node.IsFlags_EXTENTS()) + { + totalPack = (UInt64)node.NumBlocks << (node.IsFlags_HUGE() ? _h.BlockBits : 9); + return true; + } + + /* + CExtentTreeHeader eth; + if (!eth.Parse(node.Block)) + return false; + if (eth.NumEntries > 3) + return false; + if (!eth.Depth == 0) + return false; + + UInt64 numBlocks = 0; + { + for (unsigned i = 0; i < eth.NumEntries; i++) + { + CExtent e; + e.Parse(node.Block + 12 + i * 12); + // const CExtent &e = node.leafs[i]; + if (e.IsInited) + numBlocks += e.Len; + } + } + + totalPack = numBlocks << _h.BlockBits; + return true; + */ +} + + +STDMETHODIMP CHandler::GetNumberOfItems(UInt32 *numItems) +{ + *numItems = _items.Size() + _auxItems.Size(); + return S_OK; +} + +enum +{ + kpidMountTime = kpidUserDefined, + kpidLastCheckTime, + kpidRevision, + kpidINodeSize, + kpidLastMount, + kpidFeatureIncompat, + kpidFeatureRoCompat, + kpidWrittenKB + + // kpidGroupSize, + + // kpidChangeTime = kpidUserDefined + 256, + // kpidDTime +}; + +static const UInt32 kProps[] = +{ + kpidPath, + kpidIsDir, + kpidSize, + kpidPackSize, + kpidPosixAttrib, + kpidMTime, + kpidCTime, + kpidATime, + // kpidChangeTime, + // kpidDTime, + kpidINode, + kpidLinks, + kpidSymLink, + kpidCharacts, + kpidUser, + kpidGroup +}; + + +static const CStatProp kArcProps[] = +{ + { NULL, kpidHeadersSize, VT_BSTR }, + // { NULL, kpidFileSystem, VT_BSTR }, + // kpidMethod, + { NULL, kpidClusterSize, VT_UI4 }, + // { "Group Size", kpidGroupSize, VT_UI8 }, + { NULL, kpidFreeSpace, VT_UI8 }, + + { NULL, kpidMTime, VT_FILETIME }, + { NULL, kpidCTime, VT_FILETIME }, + { "Mount Time", kpidMountTime, VT_FILETIME }, + { "Last Check Time", kpidLastCheckTime, VT_FILETIME }, + + { NULL, kpidHostOS, VT_BSTR}, + { "Revision", kpidRevision, VT_UI4}, + { "inode Size", kpidINodeSize, VT_UI4}, + { NULL, kpidCodePage, VT_BSTR}, + { NULL, kpidVolumeName, VT_BSTR}, + { "Last Mounted", kpidLastMount, VT_BSTR}, + { NULL, kpidId, VT_BSTR}, + { NULL, kpidCharacts, VT_BSTR }, + { "Incompatible Features", kpidFeatureIncompat, VT_BSTR }, + { "Readonly-compatible Features", kpidFeatureRoCompat, VT_BSTR }, + { "Written KiB", kpidWrittenKB, VT_UI8 } +}; + +IMP_IInArchive_Props +IMP_IInArchive_ArcProps_WITH_NAME + +static void StringToProp(bool isUTF, const char *s, unsigned size, NCOM::CPropVariant &prop) +{ + UString u; + AString a; + a.SetFrom_CalcLen(s, size); + if (!isUTF || !ConvertUTF8ToUnicode(a, u)) + MultiByteToUnicodeString2(u, a); + prop = u; +} + +static void UnixTimeToProp(UInt32 val, NCOM::CPropVariant &prop) +{ + if (val != 0) + { + FILETIME ft; + NTime::UnixTimeToFileTime(val, ft); + prop = ft; + } +} + +STDMETHODIMP CHandler::GetArchiveProperty(PROPID propID, PROPVARIANT *value) +{ + COM_TRY_BEGIN + + NCOM::CPropVariant prop; + + switch (propID) + { + /* + case kpidFileSystem: + { + AString res = "Ext4"; + prop = res; + break; + } + */ + + case kpidIsTree: prop = true; break; + case kpidIsAux: prop = true; break; + case kpidINode: prop = true; break; + + case kpidClusterSize: prop = (UInt32)1 << _h.BlockBits; break; + // case kpidGroupSize: prop = (UInt64)_h.BlocksPerGroup << _h.BlockBits; break; + + case kpidFreeSpace: prop = (UInt64)_h.NumFreeBlocks << _h.BlockBits; break; + + case kpidCTime: UnixTimeToProp(_h.CTime, prop); break; + case kpidMTime: UnixTimeToProp(_h.WriteTime, prop); break; + case kpidMountTime: UnixTimeToProp(_h.MountTime, prop); break; + case kpidLastCheckTime: UnixTimeToProp(_h.LastCheckTime, prop); break; + + case kpidHostOS: + { + char temp[16]; + const char *s = NULL; + if (_h.CreatorOs < ARRAY_SIZE(kHostOS)) + s = kHostOS[_h.CreatorOs]; + else + { + ConvertUInt32ToString(_h.CreatorOs, temp); + s = temp; + } + prop = s; + break; + } + + case kpidRevision: prop = _h.RevLevel; break; + + case kpidINodeSize: prop = _h.InodeSize; break; + + case kpidId: + { + char s[16 * 2 + 2]; + for (unsigned i = 0; i < 16; i++) + PrintHex(_h.Uuid[i], s + i * 2); + s[16 * 2] = 0; + prop = s; + break; + } + + case kpidCodePage: if (_isUTF) prop = "UTF-8"; break; + case kpidVolumeName: StringToProp(_isUTF, _h.VolName, sizeof(_h.VolName), prop); break; + case kpidLastMount: StringToProp(_isUTF, _h.LastMount, sizeof(_h.LastMount), prop); break; + + case kpidCharacts: FLAGS_TO_PROP(g_FeatureCompat_Flags, _h.FeatureCompat, prop); break; + case kpidFeatureIncompat: FLAGS_TO_PROP(g_FeatureIncompat_Flags, _h.FeatureIncompat, prop); break; + case kpidFeatureRoCompat: FLAGS_TO_PROP(g_FeatureRoCompat_Flags, _h.FeatureRoCompat, prop); break; + case kpidWrittenKB: prop = _h.WrittenKB; break; + + case kpidPhySize: prop = _phySize; break; + + case kpidErrorFlags: + { + UInt32 v = 0; + if (!_isArc) v |= kpv_ErrorFlags_IsNotArc;; + if (_linksError) v |= kpv_ErrorFlags_HeadersError; + if (_headersError) v |= kpv_ErrorFlags_HeadersError; + if (!_stream && v == 0 && _isArc) + v = kpv_ErrorFlags_HeadersError; + if (v != 0) + prop = v; + break; + } + } + + prop.Detach(value); + return S_OK; + + COM_TRY_END +} + + +/* +static const Byte kRawProps[] = +{ + // kpidSha1, +}; +*/ + +STDMETHODIMP CHandler::GetNumRawProps(UInt32 *numProps) +{ + // *numProps = ARRAY_SIZE(kRawProps); + *numProps = 0; + return S_OK; +} + +STDMETHODIMP CHandler::GetRawPropInfo(UInt32 /* index */, BSTR *name, PROPID *propID) +{ + // *propID = kRawProps[index]; + *propID = 0; + *name = 0; + return S_OK; +} + +STDMETHODIMP CHandler::GetParent(UInt32 index, UInt32 *parent, UInt32 *parentType) +{ + *parentType = NParentType::kDir; + *parent = (UInt32)(Int32)-1; + + if (index >= _items.Size()) + return S_OK; + + const CItem &item = _items[index]; + + if (item.ParentNode < 0) + { + int aux = GetParentAux(item); + if (aux >= 0) + *parent = _items.Size() + aux; + } + else + { + int itemIndex = _nodes[item.ParentNode].ItemIndex; + if (itemIndex >= 0) + *parent = itemIndex; + } + + return S_OK; +} + + +STDMETHODIMP CHandler::GetRawProp(UInt32 index, PROPID propID, const void **data, UInt32 *dataSize, UInt32 *propType) +{ + *data = NULL; + *dataSize = 0; + *propType = 0; + + if (propID == kpidName && _isUTF) + { + if (index < _items.Size()) + { + const AString &s = _items[index].Name; + if (!s.IsEmpty()) + { + *data = (void *)(const char *)s; + *dataSize = (UInt32)s.Len() + 1; + *propType = NPropDataType::kUtf8z; + } + return S_OK; + } + else + { + const AString &s = _auxItems[index - _items.Size()]; + { + *data = (void *)(const char *)s; + *dataSize = (UInt32)s.Len() + 1; + *propType = NPropDataType::kUtf8z; + } + return S_OK; + } + } + + return S_OK; +} + + +static void ExtTimeToProp(const CExtTime &t, NCOM::CPropVariant &prop) +{ + /* + UInt32 nano = 0; + if (t.Extra != 0) + { + UInt32 mask = t.Extra & 3; + if (mask != 0) + return; + nano = t.Extra >> 2; + } + UInt64 v; + if (t.Val == 0 && nano == 0) + return; + if (!NTime::UnixTime_to_FileTime64(t.Val, v)) + return; + if (nano != 0) + v += nano / 100; + + FILETIME ft; + ft.dwLowDateTime = (DWORD)v; + ft.dwHighDateTime = (DWORD)(v >> 32); + prop = ft; + */ + if (t.Val == 0) + return; + if (t.Extra != 0) + { + UInt32 mask = t.Extra & 3; + if (mask != 0) + return; + } + FILETIME ft; + if (NTime::UnixTime64ToFileTime(t.Val, ft)) + prop = ft; +} + + +STDMETHODIMP CHandler::GetProperty(UInt32 index, PROPID propID, PROPVARIANT *value) +{ + COM_TRY_BEGIN + NCOM::CPropVariant prop; + + if (index >= _items.Size()) + { + switch (propID) + { + case kpidPath: + case kpidName: + { + AString s = _auxItems[index - _items.Size()]; + prop = s; + break; + } + case kpidIsDir: prop = true; break; + case kpidIsAux: prop = true; break; + } + } + else + { + + const CItem &item = _items[index]; + const CNode &node = _nodes[item.Node]; + bool isDir = node.IsDir(); + + switch (propID) + { + case kpidPath: + { + UString u; + { + AString s; + GetPath(index, s); + if (!_isUTF || !ConvertUTF8ToUnicode(s, u)) + MultiByteToUnicodeString2(u, s); + } + prop = u; + break; + } + + case kpidName: + { + { + UString u; + { + if (!_isUTF || !ConvertUTF8ToUnicode(item.Name, u)) + MultiByteToUnicodeString2(u, item.Name); + } + prop = u; + } + break; + } + + case kpidIsDir: + { + bool isDir2 = isDir; + if (item.SymLinkItemIndex >= 0) + isDir2 = _nodes[_items[item.SymLinkItemIndex].Node].IsDir(); + prop = isDir2; + break; + } + + case kpidSize: if (!isDir) prop = node.FileSize; break; + + case kpidPackSize: + if (!isDir) + { + UInt64 size; + if (GetPackSize(index, size)) + prop = size; + } + break; + + case kpidPosixAttrib: + { + /* + if (node.Type != 0 && node.Type < ARRAY_SIZE(k_TypeToMode)) + prop = (UInt32)(node.Mode & 0xFFF) | k_TypeToMode[node.Type]; + */ + prop = (UInt32)(node.Mode); + break; + } + + case kpidMTime: ExtTimeToProp(node.MTime, prop); break; + case kpidCTime: ExtTimeToProp(node.CTime, prop); break; + case kpidATime: ExtTimeToProp(node.ATime, prop); break; + // case kpidDTime: ExtTimeToProp(node.DTime, prop); break; + // case kpidChangeTime: ExtTimeToProp(node.InodeChangeTime, prop); break; + + case kpidUser: prop = (UInt32)node.Uid; break; + case kpidGroup: prop = (UInt32)node.Gid; break; + case kpidLinks: prop = node.NumLinks; break; + case kpidINode: prop = (UInt32)item.Node; break; + case kpidStreamId: if (!isDir) prop = (UInt32)item.Node; break; + case kpidCharacts: FLAGS_TO_PROP(g_NodeFlags, (UInt32)node.Flags, prop); break; + + case kpidSymLink: + { + if (node.SymLinkIndex >= 0) + { + UString u; + { + const AString &s = _symLinks[node.SymLinkIndex]; + if (!_isUTF || !ConvertUTF8ToUnicode(s, u)) + MultiByteToUnicodeString2(u, s); + } + prop = u; + } + break; + } + } + + } + + prop.Detach(value); + return S_OK; + COM_TRY_END +} + + +class CExtInStream: + public IInStream, + public CMyUnknownImp +{ + UInt64 _virtPos; + UInt64 _phyPos; +public: + UInt64 _size; + unsigned BlockBits; + CMyComPtr<IInStream> Stream; + CRecordVector<CExtent> Extents; + + CExtInStream() {} + + HRESULT StartSeek() + { + _virtPos = 0; + _phyPos = 0; + return Stream->Seek(_phyPos, STREAM_SEEK_SET, NULL); + } + + MY_UNKNOWN_IMP2(ISequentialInStream, IInStream) + STDMETHOD(Read)(void *data, UInt32 size, UInt32 *processedSize); + STDMETHOD(Seek)(Int64 offset, UInt32 seekOrigin, UInt64 *newPosition); +}; + +STDMETHODIMP CExtInStream::Read(void *data, UInt32 size, UInt32 *processedSize) +{ + if (processedSize) + *processedSize = 0; + if (_virtPos >= _size) + return S_OK; + UInt64 rem = _size - _virtPos; + if (size > rem) + size = (UInt32)rem; + if (size == 0) + return S_OK; + + UInt32 blockIndex = (UInt32)(_virtPos >> BlockBits); + + unsigned left = 0, right = Extents.Size(); + for (;;) + { + unsigned mid = (left + right) / 2; + if (mid == left) + break; + if (blockIndex < Extents[mid].VirtBlock) + right = mid; + else + left = mid; + } + + { + const CExtent &extent = Extents[left]; + if (blockIndex < extent.VirtBlock) + return E_FAIL; + UInt32 bo = blockIndex - extent.VirtBlock; + if (bo >= extent.Len) + return E_FAIL; + + UInt32 offset = ((UInt32)_virtPos & (((UInt32)1 << BlockBits) - 1)); + UInt32 remBlocks = extent.Len - bo; + UInt64 remBytes = ((UInt64)remBlocks << BlockBits); + remBytes -= offset; + + if (size > remBytes) + size = (UInt32)remBytes; + + if (!extent.IsInited) + { + memset(data, 0, size); + _virtPos += size; + if (processedSize) + *processedSize = size; + return S_OK; + } + + UInt64 phyBlock = extent.PhyStart + bo; + UInt64 phy = (phyBlock << BlockBits) + offset; + + if (phy != _phyPos) + { + RINOK(Stream->Seek(phy, STREAM_SEEK_SET, NULL)); + _phyPos = phy; + } + + UInt32 realProcessSize = 0; + + HRESULT res = Stream->Read(data, size, &realProcessSize); + + _phyPos += realProcessSize; + _virtPos += realProcessSize; + if (processedSize) + *processedSize = realProcessSize; + return res; + } +} + + +STDMETHODIMP CExtInStream::Seek(Int64 offset, UInt32 seekOrigin, UInt64 *newPosition) +{ + switch (seekOrigin) + { + case STREAM_SEEK_SET: break; + case STREAM_SEEK_CUR: offset += _virtPos; break; + case STREAM_SEEK_END: offset += _size; break; + default: return STG_E_INVALIDFUNCTION; + } + if (offset < 0) + return HRESULT_WIN32_ERROR_NEGATIVE_SEEK; + _virtPos = offset; + if (newPosition) + *newPosition = offset; + return S_OK; +} + + + +HRESULT CHandler::FillFileBlocks2(UInt32 block, unsigned level, unsigned numBlocks, CRecordVector<UInt32> &blocks) +{ + const size_t blockSize = (size_t)1 << _h.BlockBits; + CByteBuffer &tempBuf = _tempBufs[level]; + tempBuf.Alloc(blockSize); + + RINOK(SeekAndRead(_stream, block, tempBuf, blockSize)); + + const Byte *p = tempBuf; + size_t num = (size_t)1 << (_h.BlockBits - 2); + + for (size_t i = 0; i < num; i++) + { + if (blocks.Size() == numBlocks) + break; + UInt32 val = GetUi32(p + 4 * i); + if (val == 0 || val >= _h.NumBlocks) + return S_FALSE; + + if (level != 0) + { + RINOK(FillFileBlocks2(val, level - 1, numBlocks, blocks)); + continue; + } + + PRF(printf("\n i = %3d, start = %5d ", (unsigned)val)); + + blocks.Add(val); + } + + return S_OK; +} + + +static const unsigned kNumDirectNodeBlocks = 12; + +HRESULT CHandler::FillFileBlocks(const Byte *p, unsigned numBlocks, CRecordVector<UInt32> &blocks) +{ + blocks.ClearAndReserve(numBlocks); + + unsigned i; + + for (i = 0; i < kNumDirectNodeBlocks; i++) + { + if (i == numBlocks) + return S_OK; + UInt32 val = GetUi32(p + 4 * i); + if (val == 0 || val >= _h.NumBlocks) + return S_FALSE; + blocks.Add(val); + } + + for (i = 0; i < 3; i++) + { + if (blocks.Size() == numBlocks) + break; + UInt32 val = GetUi32(p + 4 * (kNumDirectNodeBlocks + i)); + if (val == 0 || val >= _h.NumBlocks) + return S_FALSE; + RINOK(FillFileBlocks2(val, i, numBlocks, blocks)); + } + + return S_OK; +} + + +static void AddSkipExtents(CRecordVector<CExtent> &extents, UInt32 virtBlock, UInt32 numBlocks) +{ + while (numBlocks != 0) + { + UInt32 len = numBlocks; + const UInt32 kLenMax = (UInt32)1 << 15; + if (len > kLenMax) + len = kLenMax; + CExtent e; + e.VirtBlock = virtBlock; + e.Len = (UInt16)len; + e.IsInited = false; + e.PhyStart = 0; + extents.Add(e); + virtBlock += len; + numBlocks -= len; + } +} + +static bool UpdateExtents(CRecordVector<CExtent> &extents, UInt32 block) +{ + if (extents.IsEmpty()) + { + if (block == 0) + return true; + AddSkipExtents(extents, 0, block); + return true; + } + + const CExtent &prev = extents.Back(); + if (block < prev.VirtBlock) + return false; + UInt32 prevEnd = prev.GetVirtEnd(); + if (block == prevEnd) + return true; + AddSkipExtents(extents, prevEnd, block - prevEnd); + return true; +} + + +HRESULT CHandler::FillExtents(const Byte *p, size_t size, CRecordVector<CExtent> &extents, int parentDepth) +{ + CExtentTreeHeader eth; + if (!eth.Parse(p)) + return S_FALSE; + + if (parentDepth >= 0 && eth.Depth != parentDepth - 1) // (eth.Depth >= parentDepth) + return S_FALSE; + + if (12 + 12 * (size_t)eth.NumEntries > size) + return S_FALSE; + + if (eth.Depth >= kNumTreeLevelsMax) + return S_FALSE; + + if (eth.Depth == 0) + { + for (unsigned i = 0; i < eth.NumEntries; i++) + { + CExtent e; + e.Parse(p + 12 + i * 12); + if (e.PhyStart == 0 + || e.PhyStart > _h.NumBlocks + || e.PhyStart + e.Len > _h.NumBlocks + || !e.IsLenOK()) + return S_FALSE; + if (!UpdateExtents(extents, e.VirtBlock)) + return S_FALSE; + extents.Add(e); + } + + return S_OK; + } + + const size_t blockSize = (size_t)1 << _h.BlockBits; + CByteBuffer &tempBuf = _tempBufs[eth.Depth]; + tempBuf.Alloc(blockSize); + + for (unsigned i = 0; i < eth.NumEntries; i++) + { + CExtentIndexNode e; + e.Parse(p + 12 + i * 12); + + if (e.PhyLeaf == 0 || e.PhyLeaf >= _h.NumBlocks) + return S_FALSE; + + if (!UpdateExtents(extents, e.VirtBlock)) + return S_FALSE; + + RINOK(SeekAndRead(_stream, e.PhyLeaf, tempBuf, blockSize)); + RINOK(FillExtents(tempBuf, blockSize, extents, eth.Depth)); + } + + return S_OK; +} + + +HRESULT CHandler::GetStream_Node(UInt32 nodeIndex, ISequentialInStream **stream) +{ + COM_TRY_BEGIN + + *stream = NULL; + + const CNode &node = _nodes[nodeIndex]; + + if (!node.IsFlags_EXTENTS()) + { + // maybe sparse file can have NumBlocks == 0 ? + if (node.NumBlocks == 0 && node.FileSize < kNodeBlockFieldSize) + { + Create_BufInStream_WithNewBuffer(node.Block, (size_t)node.FileSize, stream); + return S_OK; + } + } + + if (node.FileSize >= ((UInt64)1 << 63)) + return S_FALSE; + + CMyComPtr<IInStream> streamTemp; + + UInt64 numBlocks64 = (node.FileSize + (UInt64)(((UInt32)1 << _h.BlockBits) - 1)) >> _h.BlockBits; + + if (node.IsFlags_EXTENTS()) + { + if ((UInt32)numBlocks64 != numBlocks64) + return S_FALSE; + + CExtInStream *streamSpec = new CExtInStream; + streamTemp = streamSpec; + + streamSpec->BlockBits = _h.BlockBits; + streamSpec->_size = node.FileSize; + streamSpec->Stream = _stream; + + RINOK(FillExtents(node.Block, kNodeBlockFieldSize, streamSpec->Extents, -1)); + + UInt32 end = 0; + if (!streamSpec->Extents.IsEmpty()) + end = streamSpec->Extents.Back().GetVirtEnd(); + if (end < numBlocks64) + { + AddSkipExtents(streamSpec->Extents, end, (UInt32)(numBlocks64 - end)); + // return S_FALSE; + } + + RINOK(streamSpec->StartSeek()); + } + else + { + { + UInt64 numBlocks2 = numBlocks64; + + if (numBlocks64 > kNumDirectNodeBlocks) + { + UInt64 rem = numBlocks64 - kNumDirectNodeBlocks; + const unsigned refBits = (_h.BlockBits - 2); + const size_t numRefsInBlocks = (size_t)1 << refBits; + numBlocks2++; + if (rem > numRefsInBlocks) + { + numBlocks2++; + const UInt64 numL2 = (rem - 1) >> refBits; + numBlocks2 += numL2; + if (numL2 > numRefsInBlocks) + { + numBlocks2++; + numBlocks2 += (numL2 - 1) >> refBits; + } + } + } + + const unsigned specBits = (node.IsFlags_HUGE() ? 0 : _h.BlockBits - 9); + const UInt32 specMask = ((UInt32)1 << specBits) - 1;; + if ((node.NumBlocks & specMask) != 0) + return S_FALSE; + const UInt64 numBlocks64_from_header = node.NumBlocks >> specBits; + if (numBlocks64_from_header < numBlocks2) + { + // why (numBlocks64_from_header > numBlocks2) in some cases? + // return S_FALSE; + } + } + + unsigned numBlocks = (unsigned)numBlocks64; + if (numBlocks != numBlocks64) + return S_FALSE; + + CClusterInStream *streamSpec = new CClusterInStream; + streamTemp = streamSpec; + + streamSpec->BlockSizeLog = _h.BlockBits; + streamSpec->StartOffset = 0; + streamSpec->Size = node.FileSize; + streamSpec->Stream = _stream; + + RINOK(FillFileBlocks(node.Block, numBlocks, streamSpec->Vector)); + streamSpec->InitAndSeek(); + } + + *stream = streamTemp.Detach(); + + return S_OK; + + COM_TRY_END +} + + +HRESULT CHandler::ExtractNode(unsigned nodeIndex, CByteBuffer &data) +{ + data.Free(); + const CNode &node = _nodes[nodeIndex]; + size_t size = (size_t)node.FileSize; + if (size != node.FileSize) + return S_FALSE; + CMyComPtr<ISequentialInStream> inSeqStream; + RINOK(GetStream_Node(nodeIndex, &inSeqStream)); + if (!inSeqStream) + return S_FALSE; + data.Alloc(size); + _totalRead += size; + return ReadStream_FALSE(inSeqStream, data, size); +} + + +STDMETHODIMP CHandler::Extract(const UInt32 *indices, UInt32 numItems, + Int32 testMode, IArchiveExtractCallback *extractCallback) +{ + COM_TRY_BEGIN + bool allFilesMode = (numItems == (UInt32)(Int32)-1); + if (allFilesMode) + numItems = _items.Size() + _auxItems.Size(); + if (numItems == 0) + return S_OK; + + UInt64 totalSize = 0; + UInt32 i; + + for (i = 0; i < numItems; i++) + { + UInt32 index = allFilesMode ? i : indices[i]; + if (index >= _items.Size()) + continue; + const CItem &item = _items[index]; + const CNode &node = _nodes[item.Node]; + if (!node.IsDir()) + totalSize += node.FileSize; + } + + extractCallback->SetTotal(totalSize); + + UInt64 totalPackSize; + totalSize = totalPackSize = 0; + + NCompress::CCopyCoder *copyCoderSpec = new NCompress::CCopyCoder(); + CMyComPtr<ICompressCoder> copyCoder = copyCoderSpec; + + CLocalProgress *lps = new CLocalProgress; + CMyComPtr<ICompressProgressInfo> progress = lps; + lps->Init(extractCallback, false); + + for (i = 0;; i++) + { + lps->InSize = totalPackSize; + lps->OutSize = totalSize; + RINOK(lps->SetCur()); + + if (i == numItems) + break; + + CMyComPtr<ISequentialOutStream> outStream; + Int32 askMode = testMode ? + NExtract::NAskMode::kTest : + NExtract::NAskMode::kExtract; + + UInt32 index = allFilesMode ? i : indices[i]; + + RINOK(extractCallback->GetStream(index, &outStream, askMode)); + + if (index >= _items.Size()) + { + RINOK(extractCallback->PrepareOperation(askMode)); + RINOK(extractCallback->SetOperationResult(NExtract::NOperationResult::kOK)); + continue; + } + + const CItem &item = _items[index]; + const CNode &node = _nodes[item.Node]; + + if (node.IsDir()) + { + RINOK(extractCallback->PrepareOperation(askMode)); + RINOK(extractCallback->SetOperationResult(NExtract::NOperationResult::kOK)); + continue; + } + + UInt64 unpackSize = node.FileSize; + totalSize += unpackSize; + UInt64 packSize; + if (GetPackSize(index, packSize)) + totalPackSize += packSize; + + if (!testMode && !outStream) + continue; + RINOK(extractCallback->PrepareOperation(askMode)); + + int res = NExtract::NOperationResult::kDataError; + { + CMyComPtr<ISequentialInStream> inSeqStream; + HRESULT hres = GetStream(index, &inSeqStream); + if (hres == S_FALSE || !inSeqStream) + { + if (hres == E_OUTOFMEMORY) + return hres; + res = NExtract::NOperationResult::kUnsupportedMethod; + } + else + { + RINOK(hres); + { + HRESULT hres = copyCoder->Code(inSeqStream, outStream, NULL, NULL, progress); + if (hres == S_OK) + { + if (copyCoderSpec->TotalSize == unpackSize) + res = NExtract::NOperationResult::kOK; + } + else if (hres == E_NOTIMPL) + { + res = NExtract::NOperationResult::kUnsupportedMethod; + } + else if (hres != S_FALSE) + { + RINOK(hres); + } + } + } + } + RINOK(extractCallback->SetOperationResult(res)); + } + + return S_OK; + COM_TRY_END +} + + +STDMETHODIMP CHandler::GetStream(UInt32 index, ISequentialInStream **stream) +{ + *stream = NULL; + if (index >= _items.Size()) + return S_FALSE; + return GetStream_Node(_items[index].Node, stream); +} + + +API_FUNC_static_IsArc IsArc_Ext(const Byte *p, size_t size) +{ + if (size < kHeaderSize) + return k_IsArc_Res_NEED_MORE; + CHeader h; + if (!h.Parse(p + kHeaderDataOffset)) + return k_IsArc_Res_NO; + return k_IsArc_Res_YES; +} +} + +static const Byte k_Signature[] = { 0x53, 0xEF }; + +REGISTER_ARC_I( + "Ext", "ext ext3 ext4", 0, 0xC7, + k_Signature, + 0x438, + 0, + IsArc_Ext) + +}} |