test/Scripts/macho-dump

   1 #!/usr/bin/env python
   2
   3 import struct
   4 import sys
   5 import StringIO
   6
   7 class Reader:
   8    def __init__(self, path):
   9       if path == '-':
  10          # Snarf all the data so we can seek.
  11          self.file = StringIO.StringIO(sys.stdin.read())
  12       else:
  13          self.file = open(path,'rb')
  14       self.isLSB = None
  15
  16       self.string_table = None
  17
  18    def setLSB(self, isLSB):
  19       self.isLSB = bool(isLSB)
  20
  21    def tell(self):
  22       return self.file.tell()
  23
  24    def seek(self, pos):
  25       self.file.seek(pos)
  26
  27    def read(self, N):
  28       data = self.file.read(N)
  29       if len(data) != N:
  30          raise ValueError,"Out of data!"
  31       return data
  32
  33    def read8(self):
  34       return ord(self.read(1))
  35
  36    def read16(self):
  37       return struct.unpack('><'[self.isLSB] + 'H', self.read(2))[0]
  38
  39    def read32(self):
  40       # Force to 32-bit, if possible; otherwise these might be long ints on a
  41       # big-endian platform. FIXME: Why???
  42       Value = struct.unpack('><'[self.isLSB] + 'I', self.read(4))[0]
  43       return int(Value)
  44
  45    def registerStringTable(self, strings):
  46       if self.string_table is not None:
  47          raise ValueError,"%s: warning: multiple string tables" % sys.argv[0]
  48
  49       self.string_table = strings
  50
  51    def getString(self, index):
  52       if self.string_table is None:
  53          raise ValueError,"%s: warning: no string table registered" % sys.argv[0]
  54
  55       end = self.string_table.index('\x00', index)
  56       return self.string_table[index:end]
  57
  58 def dumpmacho(path, opts):
  59    f = Reader(path)
  60
  61    magic = f.read(4)
  62    if magic == '\xFE\xED\xFA\xCE':
  63       f.setLSB(False)
  64    elif magic == '\xCE\xFA\xED\xFE':
  65       f.setLSB(True)
  66    else:
  67       raise ValueError,"Not a Mach-O object file: %r (bad magic)" % path
  68
  69    print "('cputype', %r)" % f.read32()
  70    print "('cpusubtype', %r)" % f.read32()
  71    filetype = f.read32()
  72    print "('filetype', %r)" % filetype
  73
  74    numLoadCommands = f.read32()
  75    print "('num_load_commands', %r)" % filetype
  76
  77    loadCommandsSize = f.read32()
  78    print "('load_commands_size', %r)" % loadCommandsSize
  79
  80    print "('flag', %r)" % f.read32()
  81
  82    start = f.tell()
  83
  84    print "('load_commands', ["
  85    for i in range(numLoadCommands):
  86       dumpLoadCommand(f, i, opts)
  87    print "])"
  88
  89    if f.tell() - start != loadCommandsSize:
  90       raise ValueError,"%s: warning: invalid load commands size: %r" % (sys.argv[0], loadCommandsSize)
  91
  92 def dumpLoadCommand(f, i, opts):
  93    start = f.tell()
  94
  95    print "  # Load Command %r" % i
  96    cmd = f.read32()
  97    print " (('command', %r)" % cmd
  98    cmdSize = f.read32()
  99    print "  ('size', %r)" % cmdSize
 100
 101    if cmd == 1:
 102       dumpSegmentLoadCommand32(f, opts)
 103    elif cmd == 2:
 104       dumpSymtabCommand(f, opts)
 105    elif cmd == 11:
 106       dumpDysymtabCommand(f, opts)
 107    else:
 108       print >>sys.stderr,"%s: warning: unknown load command: %r" % (sys.argv[0], cmd)
 109       f.read(cmdSize - 8)
 110    print " ),"
 111
 112    if f.tell() - start != cmdSize:
 113       raise ValueError,"%s: warning: invalid load command size: %r" % (sys.argv[0], cmdSize)
 114
 115 def dumpSegmentLoadCommand32(f, opts):
 116    print "  ('segment_name', %r)" % f.read(16)
 117    print "  ('vm_addr', %r)" % f.read32()
 118    print "  ('vm_size', %r)" % f.read32()
 119    print "  ('file_offset', %r)" % f.read32()
 120    print "  ('file_size', %r)" % f.read32()
 121    print "  ('maxprot', %r)" % f.read32()
 122    print "  ('initprot', %r)" % f.read32()
 123    numSections = f.read32()
 124    print "  ('num_sections', %r)" % numSections
 125    print "  ('flags', %r)" % f.read32()
 126
 127    print "  ('sections', ["
 128    for i in range(numSections):
 129       dumpSection32(f, i, opts)
 130    print "  ])"
 131
 132 def dumpSymtabCommand(f, opts):
 133    symoff = f.read32()
 134    print "  ('symoff', %r)" % symoff
 135    nsyms = f.read32()
 136    print "  ('nsyms', %r)" % nsyms
 137    stroff = f.read32()
 138    print "  ('stroff', %r)" % stroff
 139    strsize = f.read32()
 140    print "  ('strsize', %r)" % strsize
 141
 142    prev_pos = f.tell()
 143
 144    f.seek(stroff)
 145    string_data = f.read(strsize)
 146    print "  ('_string_data', %r)" % string_data
 147
 148    f.registerStringTable(string_data)
 149
 150    f.seek(symoff)
 151    print "  ('_symbols', ["
 152    for i in range(nsyms):
 153       dumpNlist32(f, i, opts)
 154    print "  ])"
 155
 156    f.seek(prev_pos)
 157
 158 def dumpNlist32(f, i, opts):
 159    print "    # Symbol %r" % i
 160    n_strx = f.read32()
 161    print "   (('n_strx', %r)" % n_strx
 162    n_type = f.read8()
 163    print "    ('n_type', %#x)" % n_type
 164    n_sect = f.read8()
 165    print "    ('n_sect', %r)" % n_sect
 166    n_desc = f.read16()
 167    print "    ('n_desc', %r)" % n_desc
 168    n_value = f.read32()
 169    print "    ('n_value', %r)" % n_value
 170    print "    ('_string', %r)" % f.getString(n_strx)
 171    print "   ),"
 172
 173 def dumpDysymtabCommand(f, opts):
 174    print "  ('ilocalsym', %r)" % f.read32()
 175    print "  ('nlocalsym', %r)" % f.read32()
 176    print "  ('iextdefsym', %r)" % f.read32()
 177    print "  ('nextdefsym', %r)" % f.read32()
 178    print "  ('iundefsym', %r)" % f.read32()
 179    print "  ('nundefsym', %r)" % f.read32()
 180    print "  ('tocoff', %r)" % f.read32()
 181    print "  ('ntoc', %r)" % f.read32()
 182    print "  ('modtaboff', %r)" % f.read32()
 183    print "  ('nmodtab', %r)" % f.read32()
 184    print "  ('extrefsymoff', %r)" % f.read32()
 185    print "  ('nextrefsyms', %r)" % f.read32()
 186    indirectsymoff = f.read32()
 187    print "  ('indirectsymoff', %r)" % indirectsymoff
 188    nindirectsyms = f.read32()
 189    print "  ('nindirectsyms', %r)" % nindirectsyms
 190    print "  ('extreloff', %r)" % f.read32()
 191    print "  ('nextrel', %r)" % f.read32()
 192    print "  ('locreloff', %r)" % f.read32()
 193    print "  ('nlocrel', %r)" % f.read32()
 194
 195    prev_pos = f.tell()
 196
 197    f.seek(indirectsymoff)
 198    print "  ('_indirect_symbols', ["
 199    for i in range(nindirectsyms):
 200       print "    # Indirect Symbol %r" % i
 201       print "    (('symbol_index', %#x),)," % f.read32()
 202    print "  ])"
 203
 204    f.seek(prev_pos)
 205
 206 def dumpSection32(f, i, opts):
 207    print "    # Section %r" % i
 208    print "   (('section_name', %r)" % f.read(16)
 209    print "    ('segment_name', %r)" % f.read(16)
 210    print "    ('address', %r)" % f.read32()
 211    size = f.read32()
 212    print "    ('size', %r)" % size
 213    offset = f.read32()
 214    print "    ('offset', %r)" % offset
 215    print "    ('alignment', %r)" % f.read32()
 216    reloc_offset = f.read32()
 217    print "    ('reloc_offset', %r)" % reloc_offset
 218    num_reloc = f.read32()
 219    print "    ('num_reloc', %r)" % num_reloc
 220    print "    ('flags', %#x)" % f.read32()
 221    print "    ('reserved1', %r)" % f.read32()
 222    print "    ('reserved2', %r)" % f.read32()
 223    print "   ),"
 224
 225    prev_pos = f.tell()
 226
 227    f.seek(reloc_offset)
 228    print "  ('_relocations', ["
 229    for i in range(num_reloc):
 230       print "    # Relocation %r" % i
 231       print "    (('word-0', %#x)," % f.read32()
 232       print "     ('word-1', %#x))," % f.read32()
 233    print "  ])"
 234
 235    if opts.dumpSectionData:
 236       f.seek(offset)
 237       print "  ('_section_data', %r)" % f.read(size)
 238
 239    f.seek(prev_pos)
 240
 241 def main():
 242     from optparse import OptionParser, OptionGroup
 243     parser = OptionParser("usage: %prog [options] {files}")
 244     parser.add_option("", "--dump-section-data", dest="dumpSectionData",
 245                       help="Dump the contents of sections",
 246                       action="store_true", default=False)
 247     (opts, args) = parser.parse_args()
 248
 249     if not args:
 250        args.append('-')
 251
 252     for arg in args:
 253        dumpmacho(arg, opts)
 254
 255 if __name__ == '__main__':
 256    main()