libjava/contrib/classfile.py

   1 ## Copyright (C) 2004, 2005 Free Software Foundation
   2 ## Written by Gary Benson <gbenson@redhat.com>
   3 ##
   4 ## This program is free software; you can redistribute it and/or modify
   5 ## it under the terms of the GNU General Public License as published by
   6 ## the Free Software Foundation; either version 2 of the License, or
   7 ## (at your option) any later version.
   8 ##
   9 ## This program is distributed in the hope that it will be useful,
  10 ## but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 ## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 ## GNU General Public License for more details.
  13
  14 """Read Java(TM) class files."""
  15
  16 import cStringIO as StringIO
  17 import struct
  18
  19 class Class:
  20     def __init__(self, arg):
  21         if hasattr(arg, "read"):
  22             self.fp = arg
  23         elif type(arg) == type(""):
  24             if arg.startswith("\xca\xfe\xba\xbe"):
  25                 self.fp = StringIO.StringIO(arg)
  26             else:
  27                 self.fp = open(arg, "r")
  28         else:
  29             raise TypeError, type(arg)
  30
  31         magic = self._read_int()
  32         assert magic == 0xcafebabeL
  33         minor, major = self._read(">HH")
  34         self.version = (major, minor)
  35
  36         self.pool_integrity_checks = None
  37         try:
  38             assert False
  39         except AssertionError:
  40             self.pool_integrity_checks = []
  41
  42         self._read_constants_pool()
  43
  44         self.access_flags = self._read_short()
  45         self.name = self._read_reference_Class()
  46         self.super = self._read_reference_Class()
  47
  48         self.interfaces = self._read_interfaces()
  49         self.fields = self._read_fieldsormethods()
  50         self.methods = self._read_fieldsormethods()
  51         self.attributes = self._read_attributes()
  52
  53         if self.pool_integrity_checks is not None:
  54             for index, tag in self.pool_integrity_checks:
  55                 assert self.constants[index][0] == tag
  56
  57         del self.fp, self.pool_integrity_checks
  58
  59     def __repr__(self):
  60         result = []
  61         attrs = [attr for attr in dir(self)
  62                  if not attr.startswith("_") and attr != "Member"]
  63         attrs.sort()
  64         for attr in attrs:
  65             result.append("%-13s %s" % (
  66                 attr + ":", attr == "constants" and
  67                 "<ELIDED>" or repr(getattr(self, attr))))
  68         return "\n".join(result)
  69
  70     def _read_constants_pool(self):
  71         self.constants = {}
  72         skip = False
  73         for i in xrange(1, self._read_short()):
  74             if skip:
  75                 skip = False
  76                 continue
  77             tag = {
  78                 1: "Utf8", 3: "Integer", 4: "Float", 5: "Long",
  79                 6: "Double", 7: "Class", 8: "String", 9: "Fieldref",
  80                 10: "Methodref", 11: "InterfaceMethodref",
  81                 12: "NameAndType"}[self._read_byte()]
  82             skip = tag in ("Long", "Double") # crack crack crack!
  83             self.constants[i] = (tag, getattr(self, "_read_constant_" + tag)())
  84
  85     def _read_interfaces(self):
  86         result = []
  87         for i in xrange(self._read_short()):
  88             result.append(self._read_reference_Class())
  89         return result
  90
  91     def _read_fieldsormethods(self):
  92         result = []
  93         for i in xrange(self._read_short()):
  94             result.append(self.Member(self))
  95         return result
  96
  97     class Member:
  98         def __init__(self, source):
  99             self.access_flags = source._read_short()
 100             self.name = source._read_reference_Utf8()
 101             self.descriptor = source._read_reference_Utf8()
 102             self.attributes = source._read_attributes()
 103
 104         def __repr__(self):
 105             result = []
 106             attrs = [attr for attr in dir(self) if not attr.startswith("_")]
 107             attrs.sort()
 108             for attr in attrs:
 109                 value = getattr(self, attr)
 110                 if attr == "attributes" and value.has_key("Code"):
 111                     value = value.copy()
 112                     value.update({"Code": "<ELIDED>"})
 113                 result.append("%-13s %s" % (
 114                     attr + ":", repr(value).replace(
 115                         "'Code': '<ELIDED>'", "'Code': <ELIDED>")))
 116             return ("\n%s" % (15 * " ")).join(result)
 117
 118     def _read_attributes(self):
 119         result = {}
 120         for i in xrange(self._read_short()):
 121             name = self._read_reference_Utf8()
 122             data = self.fp.read(self._read_int())
 123             assert not result.has_key(name)
 124             result[name] = data
 125         return result
 126
 127     # Constants pool reference reader convenience functions
 128
 129     def _read_reference_Utf8(self):
 130         return self._read_references("Utf8")[0]
 131
 132     def _read_reference_Class(self):
 133         return self._read_references("Class")[0]
 134
 135     def _read_reference_Class_NameAndType(self):
 136         return self._read_references("Class", "NameAndType")
 137
 138     def _read_references(self, *args):
 139         result = []
 140         for arg in args:
 141             index = self._read_short()
 142             if self.pool_integrity_checks is not None:
 143                 self.pool_integrity_checks.append((index, arg))
 144             result.append(index)
 145         return result
 146
 147     # Constants pool constant reader functions
 148
 149     def _read_constant_Utf8(self):
 150         constant = self.fp.read(self._read_short())
 151         try:
 152             constant = constant.decode("utf-8")
 153         except UnicodeError:
 154             constant = _bork_utf8_decode(constant)
 155         try:
 156             constant = constant.encode("us-ascii")
 157         except UnicodeError:
 158             pass
 159         return constant
 160
 161     def _read_constant_Integer(self):
 162         return self._read_int()
 163
 164     def _read_constant_Float(self):
 165         return self._read(">f")[0]
 166
 167     def _read_constant_Long(self):
 168         return self._read(">q")[0]
 169
 170     def _read_constant_Double(self):
 171         return self._read(">d")[0]
 172
 173     _read_constant_Class = _read_reference_Utf8
 174     _read_constant_String = _read_reference_Utf8
 175     _read_constant_Fieldref = _read_reference_Class_NameAndType
 176     _read_constant_Methodref = _read_reference_Class_NameAndType
 177     _read_constant_InterfaceMethodref = _read_reference_Class_NameAndType
 178
 179     def _read_constant_NameAndType(self):
 180         return self._read_reference_Utf8(), self._read_reference_Utf8()
 181
 182     # Generic reader functions
 183
 184     def _read_int(self):
 185         # XXX how else to read 32 bits on a 64-bit box?
 186         h, l = map(long, self._read(">HH"))
 187         return (h << 16) + l
 188
 189     def _read_short(self):
 190         return self._read(">H")[0]
 191
 192     def _read_byte(self):
 193         return self._read("B")[0]
 194
 195     def _read(self, fmt):
 196         return struct.unpack(fmt, self.fp.read(struct.calcsize(fmt)))
 197
 198 def _bork_utf8_decode(data):
 199     # more crack!
 200     bytes, unicode = map(ord, data), ""
 201     while bytes:
 202         b1 = bytes.pop(0)
 203         if b1 & 0x80:
 204             assert b1 & 0x40
 205             b2 = bytes.pop(0)
 206             assert b2 & 0xC0 == 0x80
 207             if b1 & 0x20:
 208                 assert not b1 & 0x10
 209                 b3 = bytes.pop(0)
 210                 assert b3 & 0xC0 == 0x80
 211                 unicode += unichr(
 212                     ((b1 & 0x0f) << 12) + ((b2 & 0x3f) << 6) + (b3 & 0x3f))
 213             else:
 214                 unicode += unichr(((b1 & 0x1f) << 6) + (b2 & 0x3f))
 215         else:
 216             unicode += unichr(b1)
 217     return unicode
 218
 219 if __name__ == "__main__":
 220     print Class("/usr/share/katana/build/ListDependentClasses.class")
 221