Guido van Rossum | 54f22ed | 2000-02-04 15:10:34 +0000 | [diff] [blame] | 1 | """Create portable serialized representations of Python objects. |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 2 | |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 3 | See module cPickle for a (much) faster implementation. |
| 4 | See module copy_reg for a mechanism for registering custom picklers. |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 5 | |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 6 | Classes: |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 7 | |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 8 | Pickler |
| 9 | Unpickler |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 10 | |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 11 | Functions: |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 12 | |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 13 | dump(object, file) |
| 14 | dumps(object) -> string |
| 15 | load(file) -> object |
| 16 | loads(string) -> object |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 17 | |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 18 | Misc variables: |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 19 | |
Fred Drake | fe82acc | 1998-02-13 03:24:48 +0000 | [diff] [blame] | 20 | __version__ |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 21 | format_version |
| 22 | compatible_formats |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 23 | |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 24 | """ |
| 25 | |
Guido van Rossum | 743d17e | 1998-09-15 20:25:57 +0000 | [diff] [blame] | 26 | __version__ = "$Revision$" # Code version |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 27 | |
| 28 | from types import * |
Guido van Rossum | 4fb5b28 | 1997-09-12 20:07:24 +0000 | [diff] [blame] | 29 | from copy_reg import dispatch_table, safe_constructors |
Guido van Rossum | d370379 | 1998-10-22 20:15:36 +0000 | [diff] [blame] | 30 | import marshal |
| 31 | import sys |
| 32 | import struct |
Skip Montanaro | 23bafc6 | 2001-02-18 03:10:09 +0000 | [diff] [blame] | 33 | import re |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 34 | |
Skip Montanaro | 352674d | 2001-02-07 23:14:30 +0000 | [diff] [blame] | 35 | __all__ = ["PickleError", "PicklingError", "UnpicklingError", "Pickler", |
| 36 | "Unpickler", "dump", "dumps", "load", "loads"] |
| 37 | |
Guido van Rossum | d370379 | 1998-10-22 20:15:36 +0000 | [diff] [blame] | 38 | format_version = "1.3" # File format version we write |
| 39 | compatible_formats = ["1.0", "1.1", "1.2"] # Old format versions we can read |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 40 | |
| 41 | mdumps = marshal.dumps |
| 42 | mloads = marshal.loads |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 43 | |
Guido van Rossum | 26e89d0 | 2000-06-29 16:15:52 +0000 | [diff] [blame] | 44 | class PickleError(Exception): pass |
| 45 | class PicklingError(PickleError): pass |
| 46 | class UnpicklingError(PickleError): pass |
Guido van Rossum | 7849da8 | 1995-03-09 14:08:35 +0000 | [diff] [blame] | 47 | |
Guido van Rossum | ff87174 | 2000-12-13 18:11:56 +0000 | [diff] [blame] | 48 | class _Stop(Exception): |
| 49 | def __init__(self, value): |
| 50 | self.value = value |
| 51 | |
Jeremy Hylton | 2b9d029 | 1998-05-27 22:38:22 +0000 | [diff] [blame] | 52 | try: |
| 53 | from org.python.core import PyStringMap |
| 54 | except ImportError: |
| 55 | PyStringMap = None |
| 56 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 57 | MARK = '(' |
| 58 | STOP = '.' |
| 59 | POP = '0' |
| 60 | POP_MARK = '1' |
| 61 | DUP = '2' |
| 62 | FLOAT = 'F' |
| 63 | INT = 'I' |
| 64 | BININT = 'J' |
| 65 | BININT1 = 'K' |
| 66 | LONG = 'L' |
| 67 | BININT2 = 'M' |
| 68 | NONE = 'N' |
| 69 | PERSID = 'P' |
| 70 | BINPERSID = 'Q' |
| 71 | REDUCE = 'R' |
| 72 | STRING = 'S' |
| 73 | BINSTRING = 'T' |
| 74 | SHORT_BINSTRING = 'U' |
Guido van Rossum | b5f2f1b | 2000-03-10 23:20:09 +0000 | [diff] [blame] | 75 | UNICODE = 'V' |
| 76 | BINUNICODE = 'X' |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 77 | APPEND = 'a' |
| 78 | BUILD = 'b' |
| 79 | GLOBAL = 'c' |
| 80 | DICT = 'd' |
| 81 | EMPTY_DICT = '}' |
| 82 | APPENDS = 'e' |
| 83 | GET = 'g' |
| 84 | BINGET = 'h' |
| 85 | INST = 'i' |
| 86 | LONG_BINGET = 'j' |
| 87 | LIST = 'l' |
| 88 | EMPTY_LIST = ']' |
| 89 | OBJ = 'o' |
| 90 | PUT = 'p' |
| 91 | BINPUT = 'q' |
| 92 | LONG_BINPUT = 'r' |
| 93 | SETITEM = 's' |
| 94 | TUPLE = 't' |
| 95 | EMPTY_TUPLE = ')' |
| 96 | SETITEMS = 'u' |
Guido van Rossum | d370379 | 1998-10-22 20:15:36 +0000 | [diff] [blame] | 97 | BINFLOAT = 'G' |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 98 | |
Skip Montanaro | 23bafc6 | 2001-02-18 03:10:09 +0000 | [diff] [blame] | 99 | __all__.extend([x for x in dir() if re.match("[A-Z][A-Z0-9_]+$",x)]) |
| 100 | |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 101 | class Pickler: |
| 102 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 103 | def __init__(self, file, bin = 0): |
| 104 | self.write = file.write |
| 105 | self.memo = {} |
| 106 | self.bin = bin |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 107 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 108 | def dump(self, object): |
| 109 | self.save(object) |
| 110 | self.write(STOP) |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 111 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 112 | def put(self, i): |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 113 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 114 | s = mdumps(i)[1:] |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 115 | if i < 256: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 116 | return BINPUT + s[0] |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 117 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 118 | return LONG_BINPUT + s |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 119 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 120 | return PUT + `i` + '\n' |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 121 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 122 | def get(self, i): |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 123 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 124 | s = mdumps(i)[1:] |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 125 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 126 | if i < 256: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 127 | return BINGET + s[0] |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 128 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 129 | return LONG_BINGET + s |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 130 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 131 | return GET + `i` + '\n' |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 132 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 133 | def save(self, object, pers_save = 0): |
| 134 | memo = self.memo |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 135 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 136 | if not pers_save: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 137 | pid = self.persistent_id(object) |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 138 | if pid is not None: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 139 | self.save_pers(pid) |
| 140 | return |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 141 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 142 | d = id(object) |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 143 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 144 | t = type(object) |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 145 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 146 | if (t is TupleType) and (len(object) == 0): |
| 147 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 148 | self.save_empty_tuple(object) |
| 149 | else: |
| 150 | self.save_tuple(object) |
| 151 | return |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 152 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 153 | if memo.has_key(d): |
| 154 | self.write(self.get(memo[d][0])) |
| 155 | return |
| 156 | |
| 157 | try: |
| 158 | f = self.dispatch[t] |
| 159 | except KeyError: |
| 160 | pid = self.inst_persistent_id(object) |
| 161 | if pid is not None: |
| 162 | self.save_pers(pid) |
| 163 | return |
| 164 | |
| 165 | try: |
| 166 | reduce = dispatch_table[t] |
| 167 | except KeyError: |
| 168 | try: |
| 169 | reduce = object.__reduce__ |
| 170 | except AttributeError: |
| 171 | raise PicklingError, \ |
Guido van Rossum | 08a92cb | 1999-10-10 21:14:25 +0000 | [diff] [blame] | 172 | "can't pickle %s object: %s" % (`t.__name__`, |
| 173 | `object`) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 174 | else: |
| 175 | tup = reduce() |
| 176 | else: |
| 177 | tup = reduce(object) |
| 178 | |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 179 | if type(tup) is StringType: |
| 180 | self.save_global(object, tup) |
| 181 | return |
Guido van Rossum | d1f4984 | 1997-12-10 23:40:18 +0000 | [diff] [blame] | 182 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 183 | if type(tup) is not TupleType: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 184 | raise PicklingError, "Value returned by %s must be a " \ |
| 185 | "tuple" % reduce |
| 186 | |
| 187 | l = len(tup) |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 188 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 189 | if (l != 2) and (l != 3): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 190 | raise PicklingError, "tuple returned by %s must contain " \ |
| 191 | "only two or three elements" % reduce |
| 192 | |
| 193 | callable = tup[0] |
| 194 | arg_tup = tup[1] |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 195 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 196 | if l > 2: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 197 | state = tup[2] |
| 198 | else: |
| 199 | state = None |
| 200 | |
Guido van Rossum | d1f4984 | 1997-12-10 23:40:18 +0000 | [diff] [blame] | 201 | if type(arg_tup) is not TupleType and arg_tup is not None: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 202 | raise PicklingError, "Second element of tuple returned " \ |
| 203 | "by %s must be a tuple" % reduce |
| 204 | |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 205 | self.save_reduce(callable, arg_tup, state) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 206 | memo_len = len(memo) |
| 207 | self.write(self.put(memo_len)) |
| 208 | memo[d] = (memo_len, object) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 209 | return |
| 210 | |
| 211 | f(self, object) |
| 212 | |
| 213 | def persistent_id(self, object): |
| 214 | return None |
| 215 | |
| 216 | def inst_persistent_id(self, object): |
| 217 | return None |
| 218 | |
| 219 | def save_pers(self, pid): |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 220 | if not self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 221 | self.write(PERSID + str(pid) + '\n') |
| 222 | else: |
| 223 | self.save(pid, 1) |
| 224 | self.write(BINPERSID) |
| 225 | |
| 226 | def save_reduce(self, callable, arg_tup, state = None): |
| 227 | write = self.write |
| 228 | save = self.save |
| 229 | |
| 230 | save(callable) |
| 231 | save(arg_tup) |
| 232 | write(REDUCE) |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 233 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 234 | if state is not None: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 235 | save(state) |
| 236 | write(BUILD) |
| 237 | |
| 238 | dispatch = {} |
| 239 | |
| 240 | def save_none(self, object): |
| 241 | self.write(NONE) |
| 242 | dispatch[NoneType] = save_none |
| 243 | |
| 244 | def save_int(self, object): |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 245 | if self.bin: |
Tim Peters | 4471400 | 2001-04-10 05:02:52 +0000 | [diff] [blame] | 246 | # If the int is small enough to fit in a signed 4-byte 2's-comp |
| 247 | # format, we can store it more efficiently than the general |
| 248 | # case. |
| 249 | high_bits = object >> 31 # note that Python shift sign-extends |
| 250 | if high_bits == 0 or high_bits == -1: |
| 251 | # All high bits are copies of bit 2**31, so the value |
| 252 | # fits in a 4-byte signed int. |
| 253 | i = mdumps(object)[1:] |
| 254 | assert len(i) == 4 |
| 255 | if i[-2:] == '\000\000': # fits in 2-byte unsigned int |
| 256 | if i[-3] == '\000': # fits in 1-byte unsigned int |
| 257 | self.write(BININT1 + i[0]) |
| 258 | else: |
| 259 | self.write(BININT2 + i[:2]) |
| 260 | else: |
| 261 | self.write(BININT + i) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 262 | return |
Tim Peters | 4471400 | 2001-04-10 05:02:52 +0000 | [diff] [blame] | 263 | # Text pickle, or int too big to fit in signed 4-byte format. |
| 264 | self.write(INT + `object` + '\n') |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 265 | dispatch[IntType] = save_int |
| 266 | |
| 267 | def save_long(self, object): |
| 268 | self.write(LONG + `object` + '\n') |
| 269 | dispatch[LongType] = save_long |
| 270 | |
Guido van Rossum | d370379 | 1998-10-22 20:15:36 +0000 | [diff] [blame] | 271 | def save_float(self, object, pack=struct.pack): |
| 272 | if self.bin: |
| 273 | self.write(BINFLOAT + pack('>d', object)) |
| 274 | else: |
| 275 | self.write(FLOAT + `object` + '\n') |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 276 | dispatch[FloatType] = save_float |
| 277 | |
| 278 | def save_string(self, object): |
| 279 | d = id(object) |
| 280 | memo = self.memo |
| 281 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 282 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 283 | l = len(object) |
| 284 | s = mdumps(l)[1:] |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 285 | if l < 256: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 286 | self.write(SHORT_BINSTRING + s[0] + object) |
| 287 | else: |
| 288 | self.write(BINSTRING + s + object) |
| 289 | else: |
| 290 | self.write(STRING + `object` + '\n') |
| 291 | |
| 292 | memo_len = len(memo) |
| 293 | self.write(self.put(memo_len)) |
| 294 | memo[d] = (memo_len, object) |
| 295 | dispatch[StringType] = save_string |
| 296 | |
Guido van Rossum | b5f2f1b | 2000-03-10 23:20:09 +0000 | [diff] [blame] | 297 | def save_unicode(self, object): |
| 298 | d = id(object) |
| 299 | memo = self.memo |
| 300 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 301 | if self.bin: |
Guido van Rossum | b5f2f1b | 2000-03-10 23:20:09 +0000 | [diff] [blame] | 302 | encoding = object.encode('utf-8') |
| 303 | l = len(encoding) |
| 304 | s = mdumps(l)[1:] |
| 305 | self.write(BINUNICODE + s + encoding) |
| 306 | else: |
Guido van Rossum | a31b4ef | 2000-12-19 01:29:00 +0000 | [diff] [blame] | 307 | object = object.replace(u"\\", u"\\u005c") |
| 308 | object = object.replace(u"\n", u"\\u000a") |
Guido van Rossum | b5f2f1b | 2000-03-10 23:20:09 +0000 | [diff] [blame] | 309 | self.write(UNICODE + object.encode('raw-unicode-escape') + '\n') |
| 310 | |
| 311 | memo_len = len(memo) |
| 312 | self.write(self.put(memo_len)) |
| 313 | memo[d] = (memo_len, object) |
| 314 | dispatch[UnicodeType] = save_unicode |
| 315 | |
Guido van Rossum | 31584cb | 2001-01-22 14:53:29 +0000 | [diff] [blame] | 316 | if StringType == UnicodeType: |
| 317 | # This is true for Jython |
| 318 | def save_string(self, object): |
| 319 | d = id(object) |
| 320 | memo = self.memo |
| 321 | unicode = object.isunicode() |
| 322 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 323 | if self.bin: |
Guido van Rossum | 31584cb | 2001-01-22 14:53:29 +0000 | [diff] [blame] | 324 | if unicode: |
| 325 | object = object.encode("utf-8") |
| 326 | l = len(object) |
| 327 | s = mdumps(l)[1:] |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 328 | if l < 256 and not unicode: |
Guido van Rossum | 31584cb | 2001-01-22 14:53:29 +0000 | [diff] [blame] | 329 | self.write(SHORT_BINSTRING + s[0] + object) |
| 330 | else: |
| 331 | if unicode: |
| 332 | self.write(BINUNICODE + s + object) |
| 333 | else: |
| 334 | self.write(BINSTRING + s + object) |
| 335 | else: |
Tim Peters | 658cba6 | 2001-02-09 20:06:00 +0000 | [diff] [blame] | 336 | if unicode: |
Guido van Rossum | 31584cb | 2001-01-22 14:53:29 +0000 | [diff] [blame] | 337 | object = object.replace(u"\\", u"\\u005c") |
| 338 | object = object.replace(u"\n", u"\\u000a") |
| 339 | object = object.encode('raw-unicode-escape') |
| 340 | self.write(UNICODE + object + '\n') |
| 341 | else: |
| 342 | self.write(STRING + `object` + '\n') |
| 343 | |
| 344 | memo_len = len(memo) |
| 345 | self.write(self.put(memo_len)) |
| 346 | memo[d] = (memo_len, object) |
| 347 | dispatch[StringType] = save_string |
Tim Peters | 658cba6 | 2001-02-09 20:06:00 +0000 | [diff] [blame] | 348 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 349 | def save_tuple(self, object): |
| 350 | |
| 351 | write = self.write |
| 352 | save = self.save |
| 353 | memo = self.memo |
| 354 | |
| 355 | d = id(object) |
| 356 | |
| 357 | write(MARK) |
| 358 | |
| 359 | for element in object: |
| 360 | save(element) |
| 361 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 362 | if len(object) and memo.has_key(d): |
| 363 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 364 | write(POP_MARK + self.get(memo[d][0])) |
| 365 | return |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 366 | |
Guido van Rossum | 599174f | 1998-03-31 16:30:28 +0000 | [diff] [blame] | 367 | write(POP * (len(object) + 1) + self.get(memo[d][0])) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 368 | return |
| 369 | |
| 370 | memo_len = len(memo) |
| 371 | self.write(TUPLE + self.put(memo_len)) |
| 372 | memo[d] = (memo_len, object) |
| 373 | dispatch[TupleType] = save_tuple |
| 374 | |
| 375 | def save_empty_tuple(self, object): |
| 376 | self.write(EMPTY_TUPLE) |
| 377 | |
| 378 | def save_list(self, object): |
| 379 | d = id(object) |
| 380 | |
| 381 | write = self.write |
| 382 | save = self.save |
| 383 | memo = self.memo |
| 384 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 385 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 386 | write(EMPTY_LIST) |
| 387 | else: |
| 388 | write(MARK + LIST) |
| 389 | |
| 390 | memo_len = len(memo) |
| 391 | write(self.put(memo_len)) |
| 392 | memo[d] = (memo_len, object) |
| 393 | |
| 394 | using_appends = (self.bin and (len(object) > 1)) |
| 395 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 396 | if using_appends: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 397 | write(MARK) |
| 398 | |
| 399 | for element in object: |
| 400 | save(element) |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 401 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 402 | if not using_appends: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 403 | write(APPEND) |
| 404 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 405 | if using_appends: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 406 | write(APPENDS) |
| 407 | dispatch[ListType] = save_list |
| 408 | |
| 409 | def save_dict(self, object): |
| 410 | d = id(object) |
| 411 | |
| 412 | write = self.write |
| 413 | save = self.save |
| 414 | memo = self.memo |
| 415 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 416 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 417 | write(EMPTY_DICT) |
| 418 | else: |
| 419 | write(MARK + DICT) |
| 420 | |
| 421 | memo_len = len(memo) |
| 422 | self.write(self.put(memo_len)) |
| 423 | memo[d] = (memo_len, object) |
| 424 | |
| 425 | using_setitems = (self.bin and (len(object) > 1)) |
| 426 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 427 | if using_setitems: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 428 | write(MARK) |
| 429 | |
| 430 | items = object.items() |
| 431 | for key, value in items: |
| 432 | save(key) |
| 433 | save(value) |
| 434 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 435 | if not using_setitems: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 436 | write(SETITEM) |
| 437 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 438 | if using_setitems: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 439 | write(SETITEMS) |
| 440 | |
| 441 | dispatch[DictionaryType] = save_dict |
Jeremy Hylton | 2b9d029 | 1998-05-27 22:38:22 +0000 | [diff] [blame] | 442 | if not PyStringMap is None: |
| 443 | dispatch[PyStringMap] = save_dict |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 444 | |
| 445 | def save_inst(self, object): |
| 446 | d = id(object) |
| 447 | cls = object.__class__ |
| 448 | |
| 449 | memo = self.memo |
| 450 | write = self.write |
| 451 | save = self.save |
| 452 | |
| 453 | if hasattr(object, '__getinitargs__'): |
| 454 | args = object.__getinitargs__() |
| 455 | len(args) # XXX Assert it's a sequence |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 456 | _keep_alive(args, memo) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 457 | else: |
| 458 | args = () |
| 459 | |
| 460 | write(MARK) |
| 461 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 462 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 463 | save(cls) |
| 464 | |
| 465 | for arg in args: |
| 466 | save(arg) |
| 467 | |
| 468 | memo_len = len(memo) |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 469 | if self.bin: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 470 | write(OBJ + self.put(memo_len)) |
| 471 | else: |
Guido van Rossum | 4fb5b28 | 1997-09-12 20:07:24 +0000 | [diff] [blame] | 472 | write(INST + cls.__module__ + '\n' + cls.__name__ + '\n' + |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 473 | self.put(memo_len)) |
| 474 | |
| 475 | memo[d] = (memo_len, object) |
| 476 | |
| 477 | try: |
| 478 | getstate = object.__getstate__ |
| 479 | except AttributeError: |
| 480 | stuff = object.__dict__ |
| 481 | else: |
| 482 | stuff = getstate() |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 483 | _keep_alive(stuff, memo) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 484 | save(stuff) |
| 485 | write(BUILD) |
| 486 | dispatch[InstanceType] = save_inst |
| 487 | |
| 488 | def save_global(self, object, name = None): |
| 489 | write = self.write |
| 490 | memo = self.memo |
| 491 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 492 | if name is None: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 493 | name = object.__name__ |
| 494 | |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 495 | try: |
| 496 | module = object.__module__ |
| 497 | except AttributeError: |
| 498 | module = whichmodule(object, name) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 499 | |
Guido van Rossum | b0a98e9 | 2001-08-17 18:49:52 +0000 | [diff] [blame^] | 500 | try: |
| 501 | __import__(module) |
| 502 | mod = sys.modules[module] |
| 503 | klass = getattr(mod, name) |
| 504 | except (ImportError, KeyError, AttributeError): |
| 505 | raise PicklingError( |
| 506 | "Can't pickle %r: it's not found as %s.%s" % |
| 507 | (object, module, name)) |
| 508 | else: |
| 509 | if klass is not object: |
| 510 | raise PicklingError( |
| 511 | "Can't pickle %r: it's not the same object as %s.%s" % |
| 512 | (object, module, name)) |
| 513 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 514 | memo_len = len(memo) |
| 515 | write(GLOBAL + module + '\n' + name + '\n' + |
| 516 | self.put(memo_len)) |
| 517 | memo[id(object)] = (memo_len, object) |
| 518 | dispatch[ClassType] = save_global |
| 519 | dispatch[FunctionType] = save_global |
| 520 | dispatch[BuiltinFunctionType] = save_global |
Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 521 | dispatch[TypeType] = save_global |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 522 | |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 523 | |
Guido van Rossum | 5ed5c4c | 1997-09-03 00:23:54 +0000 | [diff] [blame] | 524 | def _keep_alive(x, memo): |
| 525 | """Keeps a reference to the object x in the memo. |
| 526 | |
| 527 | Because we remember objects by their id, we have |
| 528 | to assure that possibly temporary objects are kept |
| 529 | alive by referencing them. |
| 530 | We store a reference at the id of the memo, which should |
| 531 | normally not be used unless someone tries to deepcopy |
| 532 | the memo itself... |
| 533 | """ |
| 534 | try: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 535 | memo[id(memo)].append(x) |
Guido van Rossum | 5ed5c4c | 1997-09-03 00:23:54 +0000 | [diff] [blame] | 536 | except KeyError: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 537 | # aha, this is the first one :-) |
| 538 | memo[id(memo)]=[x] |
Guido van Rossum | 5ed5c4c | 1997-09-03 00:23:54 +0000 | [diff] [blame] | 539 | |
| 540 | |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 541 | classmap = {} |
| 542 | |
Guido van Rossum | 4fb5b28 | 1997-09-12 20:07:24 +0000 | [diff] [blame] | 543 | # This is no longer used to find classes, but still for functions |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 544 | def whichmodule(cls, clsname): |
| 545 | """Figure out the module in which a class occurs. |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 546 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 547 | Search sys.modules for the module. |
| 548 | Cache in classmap. |
| 549 | Return a module name. |
| 550 | If the class cannot be found, return __main__. |
| 551 | """ |
| 552 | if classmap.has_key(cls): |
| 553 | return classmap[cls] |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 554 | |
| 555 | for name, module in sys.modules.items(): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 556 | if name != '__main__' and \ |
| 557 | hasattr(module, clsname) and \ |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 558 | getattr(module, clsname) is cls: |
| 559 | break |
| 560 | else: |
| 561 | name = '__main__' |
| 562 | classmap[cls] = name |
| 563 | return name |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 564 | |
| 565 | |
| 566 | class Unpickler: |
| 567 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 568 | def __init__(self, file): |
| 569 | self.readline = file.readline |
| 570 | self.read = file.read |
| 571 | self.memo = {} |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 572 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 573 | def load(self): |
| 574 | self.mark = ['spam'] # Any new unique object |
| 575 | self.stack = [] |
| 576 | self.append = self.stack.append |
| 577 | read = self.read |
| 578 | dispatch = self.dispatch |
| 579 | try: |
| 580 | while 1: |
| 581 | key = read(1) |
| 582 | dispatch[key](self) |
Guido van Rossum | ff87174 | 2000-12-13 18:11:56 +0000 | [diff] [blame] | 583 | except _Stop, stopinst: |
| 584 | return stopinst.value |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 585 | |
| 586 | def marker(self): |
| 587 | stack = self.stack |
| 588 | mark = self.mark |
| 589 | k = len(stack)-1 |
| 590 | while stack[k] is not mark: k = k-1 |
| 591 | return k |
| 592 | |
| 593 | dispatch = {} |
| 594 | |
| 595 | def load_eof(self): |
| 596 | raise EOFError |
| 597 | dispatch[''] = load_eof |
| 598 | |
| 599 | def load_persid(self): |
| 600 | pid = self.readline()[:-1] |
| 601 | self.append(self.persistent_load(pid)) |
| 602 | dispatch[PERSID] = load_persid |
| 603 | |
| 604 | def load_binpersid(self): |
| 605 | stack = self.stack |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 606 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 607 | pid = stack[-1] |
| 608 | del stack[-1] |
| 609 | |
| 610 | self.append(self.persistent_load(pid)) |
| 611 | dispatch[BINPERSID] = load_binpersid |
| 612 | |
| 613 | def load_none(self): |
| 614 | self.append(None) |
| 615 | dispatch[NONE] = load_none |
| 616 | |
| 617 | def load_int(self): |
Guido van Rossum | ff87174 | 2000-12-13 18:11:56 +0000 | [diff] [blame] | 618 | self.append(int(self.readline()[:-1])) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 619 | dispatch[INT] = load_int |
| 620 | |
| 621 | def load_binint(self): |
| 622 | self.append(mloads('i' + self.read(4))) |
| 623 | dispatch[BININT] = load_binint |
| 624 | |
| 625 | def load_binint1(self): |
| 626 | self.append(mloads('i' + self.read(1) + '\000\000\000')) |
| 627 | dispatch[BININT1] = load_binint1 |
| 628 | |
| 629 | def load_binint2(self): |
| 630 | self.append(mloads('i' + self.read(2) + '\000\000')) |
| 631 | dispatch[BININT2] = load_binint2 |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 632 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 633 | def load_long(self): |
Guido van Rossum | ff87174 | 2000-12-13 18:11:56 +0000 | [diff] [blame] | 634 | self.append(long(self.readline()[:-1], 0)) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 635 | dispatch[LONG] = load_long |
| 636 | |
| 637 | def load_float(self): |
Guido van Rossum | ff87174 | 2000-12-13 18:11:56 +0000 | [diff] [blame] | 638 | self.append(float(self.readline()[:-1])) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 639 | dispatch[FLOAT] = load_float |
| 640 | |
Guido van Rossum | d370379 | 1998-10-22 20:15:36 +0000 | [diff] [blame] | 641 | def load_binfloat(self, unpack=struct.unpack): |
| 642 | self.append(unpack('>d', self.read(8))[0]) |
| 643 | dispatch[BINFLOAT] = load_binfloat |
| 644 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 645 | def load_string(self): |
Jeremy Hylton | be467e5 | 2000-09-15 15:14:51 +0000 | [diff] [blame] | 646 | rep = self.readline()[:-1] |
| 647 | if not self._is_string_secure(rep): |
| 648 | raise ValueError, "insecure string pickle" |
| 649 | self.append(eval(rep, |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 650 | {'__builtins__': {}})) # Let's be careful |
| 651 | dispatch[STRING] = load_string |
| 652 | |
Jeremy Hylton | be467e5 | 2000-09-15 15:14:51 +0000 | [diff] [blame] | 653 | def _is_string_secure(self, s): |
| 654 | """Return true if s contains a string that is safe to eval |
| 655 | |
| 656 | The definition of secure string is based on the implementation |
| 657 | in cPickle. s is secure as long as it only contains a quoted |
| 658 | string and optional trailing whitespace. |
| 659 | """ |
| 660 | q = s[0] |
| 661 | if q not in ("'", '"'): |
| 662 | return 0 |
| 663 | # find the closing quote |
| 664 | offset = 1 |
| 665 | i = None |
| 666 | while 1: |
| 667 | try: |
| 668 | i = s.index(q, offset) |
| 669 | except ValueError: |
| 670 | # if there is an error the first time, there is no |
| 671 | # close quote |
| 672 | if offset == 1: |
| 673 | return 0 |
| 674 | if s[i-1] != '\\': |
| 675 | break |
| 676 | # check to see if this one is escaped |
| 677 | nslash = 0 |
| 678 | j = i - 1 |
| 679 | while j >= offset and s[j] == '\\': |
| 680 | j = j - 1 |
| 681 | nslash = nslash + 1 |
| 682 | if nslash % 2 == 0: |
| 683 | break |
| 684 | offset = i + 1 |
| 685 | for c in s[i+1:]: |
| 686 | if ord(c) > 32: |
| 687 | return 0 |
| 688 | return 1 |
| 689 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 690 | def load_binstring(self): |
| 691 | len = mloads('i' + self.read(4)) |
| 692 | self.append(self.read(len)) |
| 693 | dispatch[BINSTRING] = load_binstring |
| 694 | |
Guido van Rossum | b5f2f1b | 2000-03-10 23:20:09 +0000 | [diff] [blame] | 695 | def load_unicode(self): |
| 696 | self.append(unicode(self.readline()[:-1],'raw-unicode-escape')) |
| 697 | dispatch[UNICODE] = load_unicode |
| 698 | |
| 699 | def load_binunicode(self): |
| 700 | len = mloads('i' + self.read(4)) |
| 701 | self.append(unicode(self.read(len),'utf-8')) |
| 702 | dispatch[BINUNICODE] = load_binunicode |
| 703 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 704 | def load_short_binstring(self): |
| 705 | len = mloads('i' + self.read(1) + '\000\000\000') |
| 706 | self.append(self.read(len)) |
| 707 | dispatch[SHORT_BINSTRING] = load_short_binstring |
| 708 | |
| 709 | def load_tuple(self): |
| 710 | k = self.marker() |
| 711 | self.stack[k:] = [tuple(self.stack[k+1:])] |
| 712 | dispatch[TUPLE] = load_tuple |
| 713 | |
| 714 | def load_empty_tuple(self): |
| 715 | self.stack.append(()) |
| 716 | dispatch[EMPTY_TUPLE] = load_empty_tuple |
| 717 | |
| 718 | def load_empty_list(self): |
| 719 | self.stack.append([]) |
| 720 | dispatch[EMPTY_LIST] = load_empty_list |
| 721 | |
| 722 | def load_empty_dictionary(self): |
| 723 | self.stack.append({}) |
| 724 | dispatch[EMPTY_DICT] = load_empty_dictionary |
| 725 | |
| 726 | def load_list(self): |
| 727 | k = self.marker() |
| 728 | self.stack[k:] = [self.stack[k+1:]] |
| 729 | dispatch[LIST] = load_list |
| 730 | |
| 731 | def load_dict(self): |
| 732 | k = self.marker() |
| 733 | d = {} |
| 734 | items = self.stack[k+1:] |
| 735 | for i in range(0, len(items), 2): |
| 736 | key = items[i] |
| 737 | value = items[i+1] |
| 738 | d[key] = value |
| 739 | self.stack[k:] = [d] |
| 740 | dispatch[DICT] = load_dict |
| 741 | |
| 742 | def load_inst(self): |
| 743 | k = self.marker() |
| 744 | args = tuple(self.stack[k+1:]) |
| 745 | del self.stack[k:] |
| 746 | module = self.readline()[:-1] |
| 747 | name = self.readline()[:-1] |
| 748 | klass = self.find_class(module, name) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 749 | instantiated = 0 |
| 750 | if (not args and type(klass) is ClassType and |
| 751 | not hasattr(klass, "__getinitargs__")): |
| 752 | try: |
| 753 | value = _EmptyClass() |
| 754 | value.__class__ = klass |
Guido van Rossum | b19e2a3 | 1998-04-13 18:08:45 +0000 | [diff] [blame] | 755 | instantiated = 1 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 756 | except RuntimeError: |
| 757 | # In restricted execution, assignment to inst.__class__ is |
| 758 | # prohibited |
| 759 | pass |
| 760 | if not instantiated: |
Guido van Rossum | 743d17e | 1998-09-15 20:25:57 +0000 | [diff] [blame] | 761 | try: |
| 762 | value = apply(klass, args) |
| 763 | except TypeError, err: |
| 764 | raise TypeError, "in constructor for %s: %s" % ( |
| 765 | klass.__name__, str(err)), sys.exc_info()[2] |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 766 | self.append(value) |
| 767 | dispatch[INST] = load_inst |
| 768 | |
| 769 | def load_obj(self): |
| 770 | stack = self.stack |
| 771 | k = self.marker() |
| 772 | klass = stack[k + 1] |
| 773 | del stack[k + 1] |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 774 | args = tuple(stack[k + 1:]) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 775 | del stack[k:] |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 776 | instantiated = 0 |
| 777 | if (not args and type(klass) is ClassType and |
| 778 | not hasattr(klass, "__getinitargs__")): |
| 779 | try: |
| 780 | value = _EmptyClass() |
| 781 | value.__class__ = klass |
| 782 | instantiated = 1 |
| 783 | except RuntimeError: |
| 784 | # In restricted execution, assignment to inst.__class__ is |
| 785 | # prohibited |
| 786 | pass |
| 787 | if not instantiated: |
| 788 | value = apply(klass, args) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 789 | self.append(value) |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 790 | dispatch[OBJ] = load_obj |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 791 | |
| 792 | def load_global(self): |
| 793 | module = self.readline()[:-1] |
| 794 | name = self.readline()[:-1] |
| 795 | klass = self.find_class(module, name) |
| 796 | self.append(klass) |
| 797 | dispatch[GLOBAL] = load_global |
| 798 | |
| 799 | def find_class(self, module, name): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 800 | try: |
Guido van Rossum | 397032a | 1999-03-25 21:58:59 +0000 | [diff] [blame] | 801 | __import__(module) |
| 802 | mod = sys.modules[module] |
| 803 | klass = getattr(mod, name) |
| 804 | except (ImportError, KeyError, AttributeError): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 805 | raise SystemError, \ |
| 806 | "Failed to import class %s from module %s" % \ |
| 807 | (name, module) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 808 | return klass |
| 809 | |
| 810 | def load_reduce(self): |
| 811 | stack = self.stack |
| 812 | |
| 813 | callable = stack[-2] |
| 814 | arg_tup = stack[-1] |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 815 | del stack[-2:] |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 816 | |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 817 | if type(callable) is not ClassType: |
| 818 | if not safe_constructors.has_key(callable): |
| 819 | try: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 820 | safe = callable.__safe_for_unpickling__ |
| 821 | except AttributeError: |
| 822 | safe = None |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 823 | |
Tim Peters | c32d824 | 2001-04-10 02:48:53 +0000 | [diff] [blame] | 824 | if not safe: |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 825 | raise UnpicklingError, "%s is not safe for " \ |
| 826 | "unpickling" % callable |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 827 | |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 828 | if arg_tup is None: |
| 829 | value = callable.__basicnew__() |
| 830 | else: |
| 831 | value = apply(callable, arg_tup) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 832 | self.append(value) |
| 833 | dispatch[REDUCE] = load_reduce |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 834 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 835 | def load_pop(self): |
| 836 | del self.stack[-1] |
| 837 | dispatch[POP] = load_pop |
Guido van Rossum | 7b5430f | 1995-03-04 22:25:21 +0000 | [diff] [blame] | 838 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 839 | def load_pop_mark(self): |
| 840 | k = self.marker() |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 841 | del self.stack[k:] |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 842 | dispatch[POP_MARK] = load_pop_mark |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 843 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 844 | def load_dup(self): |
Guido van Rossum | b1062fc | 1998-03-31 17:00:46 +0000 | [diff] [blame] | 845 | self.append(self.stack[-1]) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 846 | dispatch[DUP] = load_dup |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 847 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 848 | def load_get(self): |
| 849 | self.append(self.memo[self.readline()[:-1]]) |
| 850 | dispatch[GET] = load_get |
Guido van Rossum | 7853647 | 1996-04-12 13:36:27 +0000 | [diff] [blame] | 851 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 852 | def load_binget(self): |
| 853 | i = mloads('i' + self.read(1) + '\000\000\000') |
| 854 | self.append(self.memo[`i`]) |
| 855 | dispatch[BINGET] = load_binget |
Guido van Rossum | 7853647 | 1996-04-12 13:36:27 +0000 | [diff] [blame] | 856 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 857 | def load_long_binget(self): |
| 858 | i = mloads('i' + self.read(4)) |
| 859 | self.append(self.memo[`i`]) |
| 860 | dispatch[LONG_BINGET] = load_long_binget |
Guido van Rossum | 7853647 | 1996-04-12 13:36:27 +0000 | [diff] [blame] | 861 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 862 | def load_put(self): |
| 863 | self.memo[self.readline()[:-1]] = self.stack[-1] |
| 864 | dispatch[PUT] = load_put |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 865 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 866 | def load_binput(self): |
| 867 | i = mloads('i' + self.read(1) + '\000\000\000') |
| 868 | self.memo[`i`] = self.stack[-1] |
| 869 | dispatch[BINPUT] = load_binput |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 870 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 871 | def load_long_binput(self): |
| 872 | i = mloads('i' + self.read(4)) |
| 873 | self.memo[`i`] = self.stack[-1] |
| 874 | dispatch[LONG_BINPUT] = load_long_binput |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 875 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 876 | def load_append(self): |
| 877 | stack = self.stack |
| 878 | value = stack[-1] |
| 879 | del stack[-1] |
| 880 | list = stack[-1] |
| 881 | list.append(value) |
| 882 | dispatch[APPEND] = load_append |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 883 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 884 | def load_appends(self): |
| 885 | stack = self.stack |
| 886 | mark = self.marker() |
| 887 | list = stack[mark - 1] |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 888 | for i in range(mark + 1, len(stack)): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 889 | list.append(stack[i]) |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 890 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 891 | del stack[mark:] |
| 892 | dispatch[APPENDS] = load_appends |
Tim Peters | 2344fae | 2001-01-15 00:50:52 +0000 | [diff] [blame] | 893 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 894 | def load_setitem(self): |
| 895 | stack = self.stack |
| 896 | value = stack[-1] |
| 897 | key = stack[-2] |
| 898 | del stack[-2:] |
| 899 | dict = stack[-1] |
| 900 | dict[key] = value |
| 901 | dispatch[SETITEM] = load_setitem |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 902 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 903 | def load_setitems(self): |
| 904 | stack = self.stack |
| 905 | mark = self.marker() |
| 906 | dict = stack[mark - 1] |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 907 | for i in range(mark + 1, len(stack), 2): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 908 | dict[stack[i]] = stack[i + 1] |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 909 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 910 | del stack[mark:] |
| 911 | dispatch[SETITEMS] = load_setitems |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 912 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 913 | def load_build(self): |
| 914 | stack = self.stack |
| 915 | value = stack[-1] |
| 916 | del stack[-1] |
| 917 | inst = stack[-1] |
| 918 | try: |
| 919 | setstate = inst.__setstate__ |
| 920 | except AttributeError: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 921 | try: |
| 922 | inst.__dict__.update(value) |
| 923 | except RuntimeError: |
| 924 | # XXX In restricted execution, the instance's __dict__ is not |
| 925 | # accessible. Use the old way of unpickling the instance |
| 926 | # variables. This is a semantic different when unpickling in |
| 927 | # restricted vs. unrestricted modes. |
| 928 | for k, v in value.items(): |
| 929 | setattr(inst, k, v) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 930 | else: |
| 931 | setstate(value) |
| 932 | dispatch[BUILD] = load_build |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 933 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 934 | def load_mark(self): |
| 935 | self.append(self.mark) |
| 936 | dispatch[MARK] = load_mark |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 937 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 938 | def load_stop(self): |
| 939 | value = self.stack[-1] |
| 940 | del self.stack[-1] |
Guido van Rossum | ff87174 | 2000-12-13 18:11:56 +0000 | [diff] [blame] | 941 | raise _Stop(value) |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 942 | dispatch[STOP] = load_stop |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 943 | |
Guido van Rossum | e467be6 | 1997-12-05 19:42:42 +0000 | [diff] [blame] | 944 | # Helper class for load_inst/load_obj |
| 945 | |
| 946 | class _EmptyClass: |
| 947 | pass |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 948 | |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 949 | # Shorthands |
| 950 | |
Guido van Rossum | c7c5e69 | 1996-07-22 22:26:07 +0000 | [diff] [blame] | 951 | from StringIO import StringIO |
| 952 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 953 | def dump(object, file, bin = 0): |
| 954 | Pickler(file, bin).dump(object) |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 955 | |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 956 | def dumps(object, bin = 0): |
| 957 | file = StringIO() |
| 958 | Pickler(file, bin).dump(object) |
| 959 | return file.getvalue() |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 960 | |
| 961 | def load(file): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 962 | return Unpickler(file).load() |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 963 | |
| 964 | def loads(str): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 965 | file = StringIO(str) |
| 966 | return Unpickler(file).load() |
Guido van Rossum | 0c891ce | 1995-03-14 15:09:05 +0000 | [diff] [blame] | 967 | |
| 968 | |
| 969 | # The rest is used for testing only |
| 970 | |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 971 | class C: |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 972 | def __cmp__(self, other): |
| 973 | return cmp(self.__dict__, other.__dict__) |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 974 | |
| 975 | def test(): |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 976 | fn = 'out' |
| 977 | c = C() |
| 978 | c.foo = 1 |
| 979 | c.bar = 2 |
| 980 | x = [0, 1, 2, 3] |
| 981 | y = ('abc', 'abc', c, c) |
| 982 | x.append(y) |
| 983 | x.append(y) |
| 984 | x.append(5) |
| 985 | f = open(fn, 'w') |
| 986 | F = Pickler(f) |
| 987 | F.dump(x) |
| 988 | f.close() |
| 989 | f = open(fn, 'r') |
| 990 | U = Unpickler(f) |
| 991 | x2 = U.load() |
| 992 | print x |
| 993 | print x2 |
| 994 | print x == x2 |
| 995 | print map(id, x) |
| 996 | print map(id, x2) |
| 997 | print F.memo |
| 998 | print U.memo |
Guido van Rossum | a48061a | 1995-01-10 00:31:14 +0000 | [diff] [blame] | 999 | |
| 1000 | if __name__ == '__main__': |
Guido van Rossum | b72cf2d | 1997-04-09 17:32:51 +0000 | [diff] [blame] | 1001 | test() |