PE文件格式分析心得

PE文件格式最近好像炒得沸沸扬扬，由于我正在做一个这样的程序，索性将自己的心得写出来与大家同享。
  PE文件头分两大部分：
1：DOS ‘MZ’ HEADER
2：IMAGE_NT_HEADERS
  其中IMAGE_NT_HEADERS中包含
PE signature
IMAGE_FILE_HEADER
IMAGE_OPTIONAL_HEADER（其中包含Data Direcotry）
  文件头后紧跟着为
Section Table (array of IMAGE_SECTION_HEADERs)
  在DELPHI的windows.pad中已经有定义的有：
TImageDosHeader;
TImageNtHeaders;
TImageSectionHeader; { size of TIm..der is $28 }
  定义变量后按住Ctrl可以察看具体的项目，这里我就不多说了，这方面的东西也很多。
  而其他的如TImageResourceDirectory等，在DELPHI中却没有定义，察看其他资料，我在这里给出他们的结构和简单说明：
  以下是我写的PEDump.exe的类型说明：

type
  PIMAGE_RESOURCE_DIRECTORY = ^TImageResourceDirectory;
  _IMAGE_RESOURCE_DIRECTORY = packed record
    Characteristics:DWORD;
    TimeDateStamp:DWORD;
    MajorVersion:WORD;
    MinorVersion:WORD;
    NumberOfNamedEntries:WORD;
    NumberOfIdEntries:WORD;
  end;
  TImageResourceDirectory = _IMAGE_RESOURCE_DIRECTORY;
  { 资源目录的格式说明 }

  PIMAGE_RESOURCE_DIRECTORY_ENTRY = ^TImageResourceDirectoryEntry;
  _IMAGE_RESOURCE_DIRECTORY_ENTRY = packed record
    Name:DWORD;         { NameOffset:31,NameIsString:1 }
//    Id:WORD;
    OffsetToData:DWORD; { OffsetToDirectory:31,DataIsDirectory:1 }
  end;
  TImageResourceDirectoryEntry = _IMAGE_RESOURCE_DIRECTORY_ENTRY;
  { 资源目录进入点的格式说明 }

  PIMAGE_RESOURCE_DIRECTORY_STRING = ^TImageResourceDirectoryString;
  _IMAGE_RESOURCE_DIRECTORY_STRING = packed record
    Length:WORD;
    NameString:CHAR;
  end;
  TImageResourceDirectoryString = _IMAGE_RESOURCE_DIRECTORY_STRING;
  { 资源目录名的格式说明 }

  PIMAGE_RESOURCE_DIR_STRING_U = ^TImageResourceDirStringU;
  _IMAGE_RESOURCE_DIR_STRING_U = packed record
    Length:WORD;
    NameString:WCHAR;
  end;
  TImageResourceDirStringU = _IMAGE_RESOURCE_DIR_STRING_U;
  { unicode形式的资源目录名的格式说明 }

  PIMAGE_RESOURCE_DATA_ENTRY = ^TImageResourceDataEntry;
  _IMAGE_RESOURCE_DATA_ENTRY = packed record
    OffsetToData:DWORD;
    Size:DWORD;
    CodePage:DWORD;
    Reserved:DWORD;
  end;
  TImageResourceDataEntry = _IMAGE_RESOURCE_DATA_ENTRY;
  { 资源目录数据进入点的格式说明 }

const
  IMAGE_RESOURCE_NAME_IS_STRING = $80000000;
  { 检测TImageResourceDirectoryEntry.Name的最高为是否设立，
    是则说明剩下的31位指向IMAGE_RESOURCE_DIR_STRING_U的偏移，
    否则说明剩下的31位为一个整数ID。 }
  IMAGE_RESOURCE_DATA_IS_DIRECTORY = $80000000;
  { 检测TImageResourceDirectoryEntry.OffsetToData的最高为是否设立，
    是则说明剩下的31位指向另一个IMAGE_RESOURCE_DIRECTORY的偏移，
    否则说明剩下的31位指向IMAGE_RESOURCE_DATA_ENTRY的偏移。 }

  { 以下是文件属性具体值常量说明 }
  { File Characteristics }
  IMAGE_FILE_RELOCS_STRIPPED           = $0001; // Relocation info stripped from file.
  IMAGE_FILE_EXECUTABLE_IMAGE          = $0002; // File is executable.
  IMAGE_FILE_LINE_NUMS_STRIPPED        = $0004; // Line nunbers stripped from file.
  IMAGE_FILE_LOCAL_SYMS_STRIPPED       = $0008; // Local symbols stripped from file.
  IMAGE_FILE_AGGRESIVE_WS_TRIM         = $0010; // Agressively trim working set
  IMAGE_FILE_LARGE_ADDRESS_AWARE       = $0020; // App can handle >2gb addresses
  IMAGE_FILE_BYTES_REVERSED_LO         = $0080; // Bytes of machine word are reversed.
  IMAGE_FILE_32BIT_MACHINE             = $0100; // 32 bit word machine.
  IMAGE_FILE_DEBUG_STRIPPED            = $0200;  
  // Debugging info stripped from file in .DBG file
  IMAGE_FILE_REMOVABLE_RUN_FROM_SWAP   = $0400;  
  // If Image is on removable media, copy and run from the swap file.
  IMAGE_FILE_NET_RUN_FROM_SWAP         = $0800;  
  // If Image is on Net, copy and run from the swap file.
  IMAGE_FILE_SYSTEM                    = $1000; // System File.
  IMAGE_FILE_DLL                       = $2000; // File is a DLL.
  IMAGE_FILE_UP_SYSTEM_ONLY            = $4000; // File should only be run on a UP machine
  IMAGE_FILE_BYTES_REVERSED_HI         = $8000; // Bytes of machine word are reversed.

  { 以下是文件头机器属性值的具体说明 }
  { Machine }
  IMAGE_FILE_MACHINE_UNKNOWN           = $0;
  IMAGE_FILE_MACHINE_I386              = $014c; // Intel 386.
  IMAGE_FILE_MACHINE_R3000             = $0162; // MIPS little-endian, $160 big-endian
  IMAGE_FILE_MACHINE_R4000             = $0166; // MIPS little-endian
  IMAGE_FILE_MACHINE_R10000            = $0168; // MIPS little-endian
  IMAGE_FILE_MACHINE_WCEMIPSV2         = $0169; // MIPS little-endian WCE v2
  IMAGE_FILE_MACHINE_ALPHA             = $0184; // Alpha_AXP
  IMAGE_FILE_MACHINE_SH3               = $01a2; // SH3 little-endian
  IMAGE_FILE_MACHINE_SH3E              = $01a4; // SH3E little-endian
  IMAGE_FILE_MACHINE_SH4               = $01a6; // SH4 little-endian
  IMAGE_FILE_MACHINE_SH5               = $01a8; // SH5
  IMAGE_FILE_MACHINE_ARM               = $01c0; // ARM Little-Endian
  IMAGE_FILE_MACHINE_THUMB             = $01c2;
  IMAGE_FILE_MACHINE_ARM33             = $01d3;
  IMAGE_FILE_MACHINE_POWERPC           = $01F0; // IBM PowerPC Little-Endian
  IMAGE_FILE_MACHINE_IA64              = $0200; // Intel 64
  IMAGE_FILE_MACHINE_MIPS16            = $0266; // MIPS
  IMAGE_FILE_MACHINE_ALPHA64           = $0284; // ALPHA64
  IMAGE_FILE_MACHINE_MIPSFPU           = $0366; // MIPS
  IMAGE_FILE_MACHINE_MIPSFPU16         = $0466; // MIPS
//  IMAGE_FILE_MACHINE_AXP64             IMAGE_FILE_MACHINE_ALPHA64
  IMAGE_FILE_MACHINE_AMD64             = $0500; // AMD K8
  IMAGE_FILE_MACHINE_TRICORE           = $0520; // Infineon
  IMAGE_FILE_MACHINE_CEF               = $0CEF;

  { 以下是SECTION的属性值具体说明 }
  { Section characteristics }
//  IMAGE_SCN_TYPE_REG                   = $00000000; // Reserved.
//  IMAGE_SCN_TYPE_DSECT                 = $00000001; // Reserved.
//  IMAGE_SCN_TYPE_NOLOAD                = $00000002; // Reserved.
//  IMAGE_SCN_TYPE_GROUP                 = $00000004; // Reserved.
  IMAGE_SCN_TYPE_NO_PAD                = $00000008; // Reserved.
//  IMAGE_SCN_TYPE_COPY                  = $00000010; // Reserved.

  IMAGE_SCN_CNT_CODE                   = $00000020; // Section contains code.
  IMAGE_SCN_CNT_INITIALIZED_DATA       = $00000040; // Section contains initialized data.
  IMAGE_SCN_CNT_UNINITIALIZED_DATA     = $00000080; // Section contains uninitialized data.

  IMAGE_SCN_LNK_OTHER                  = $00000100; // Reserved.
  IMAGE_SCN_LNK_INFO                   = $00000200;  
  // Section contains comments or some other type of information.
//  IMAGE_SCN_TYPE_OVER                  = $00000400; // Reserved.
  IMAGE_SCN_LNK_REMOVE                 = $00000800;  
  // Section contents will not become part of image.
  IMAGE_SCN_LNK_COMDAT                 = $00001000; // Section contents comdat.
//                                       = $00002000; // Reserved.
//  IMAGE_SCN_MEM_PROTECTED - Obsolete   = $00004000;
  IMAGE_SCN_NO_DEFER_SPEC_EXC          = $00004000;  
  // Reset speculative exceptions handling bits in the TLB entries for this section.
  IMAGE_SCN_GPREL                      = $00008000;  
  // Section content can be accessed relative to GP
  IMAGE_SCN_MEM_FARDATA                = $00008000;
//  IMAGE_SCN_MEM_SYSHEAP  - Obsolete    = $00010000;
  IMAGE_SCN_MEM_PURGEABLE              = $00020000;
  IMAGE_SCN_MEM_16BIT                  = $00020000;
  IMAGE_SCN_MEM_LOCKED                 = $00040000;
  IMAGE_SCN_MEM_PRELOAD                = $00080000;

  IMAGE_SCN_ALIGN_1BYTES               = $00100000; //
  IMAGE_SCN_ALIGN_2BYTES               = $00200000; //
  IMAGE_SCN_ALIGN_4BYTES               = $00300000; //
  IMAGE_SCN_ALIGN_8BYTES               = $00400000; //
  IMAGE_SCN_ALIGN_16BYTES              = $00500000;  
  // Default alignment if no others are specified.
  IMAGE_SCN_ALIGN_32BYTES              = $00600000; //
  IMAGE_SCN_ALIGN_64BYTES              = $00700000; //
  IMAGE_SCN_ALIGN_128BYTES             = $00800000; //
  IMAGE_SCN_ALIGN_256BYTES             = $00900000; //
  IMAGE_SCN_ALIGN_512BYTES             = $00A00000; //
  IMAGE_SCN_ALIGN_1024BYTES            = $00B00000; //
  IMAGE_SCN_ALIGN_2048BYTES            = $00C00000; //
  IMAGE_SCN_ALIGN_4096BYTES            = $00D00000; //
  IMAGE_SCN_ALIGN_8192BYTES            = $00E00000; //
// Unused                                = $00F00000;
  IMAGE_SCN_ALIGN_MASK                 = $00F00000;

  IMAGE_SCN_LNK_NRELOC_OVFL            = $01000000; // Section contains extended relocations.
  IMAGE_SCN_MEM_DISCARDABLE            = $02000000; // Section can be discarded.
  IMAGE_SCN_MEM_NOT_CACHED             = $04000000; // Section is not cachable.
  IMAGE_SCN_MEM_NOT_PAGED              = $08000000; // Section is not pageable.
  IMAGE_SCN_MEM_SHARED                 = $10000000; // Section is shareable.
  IMAGE_SCN_MEM_EXECUTE                = $20000000; // Section is executable.
  IMAGE_SCN_MEM_READ                   = $40000000; // Section is readable.
  IMAGE_SCN_MEM_WRITE                  = $80000000; // Section is writeable.

  我写了检测是否包含此属性的函数
  function BeTrue(fg:Cardinal,Value):Boolean;
  begin
    Result:=fg and not Value=0;
  end;
  如果fg的属性值在Value中，则为True,否则为False;
  例如  BeTrue(IMAGE_FILE_RELOCS_STRIPPED,PENTHead.FileHeader.Characteristics);

  至于资源目录的读取，至少需要两重循环来定位，具体实现就要靠你的算法功力了：）

  好了，差不多了，再具体一点我也说不上了，我做了一个PEDump的程序，在我的主页DELPHI盒子
==================== http://www.delphibox.com =====================
可以下载，程序中有更详细的说明，相信能帮助想了解这方面的东西的朋友。