1 #!/usr/bin/env python 2 3 """ 4 Java bytecode conversion. Specification found at the following URL: 5 http://java.sun.com/docs/books/vmspec/2nd-edition/html/Instructions2.doc.html 6 7 NOTE: Synchronized constructs are not actually supported. 8 """ 9 10 import classfile 11 from dis import cmp_op, opname # for access to Python bytecode values and operators 12 try: 13 from dis import opmap 14 except ImportError: 15 opmap = {} 16 for i in range(0, len(opname)): 17 opmap[opname[i]] = i 18 from UserDict import UserDict 19 import new 20 21 # Bytecode production classes. 22 23 class BytecodeWriter: 24 25 "A Python bytecode writer." 26 27 def __init__(self): 28 29 "Initialise the writer." 30 31 # A stack of loop start instructions corresponding to loop blocks. 32 self.loops = [] 33 34 # A stack of loop block or exception block start positions. 35 self.blocks = [] 36 37 # A stack of exception block handler pointers. 38 self.exception_handlers = [] 39 40 # A list of exception offset details. 41 self.exception_offsets = [] 42 43 # A dictionary mapping labels to jump instructions referencing such labels. 44 self.jumps = {} 45 46 # The output values, including "lazy" subvalues which will need evaluating. 47 self.output = [] 48 49 # The current Python bytecode instruction position. 50 self.position = 0 51 52 # Stack depth estimation. 53 self.stack_depth = 0 54 self.max_stack_depth = 0 55 56 # Local variable estimation. 57 self.max_locals = 0 58 59 # Mapping from values to indexes. 60 self.constants = {} 61 62 # Mapping from names to indexes. 63 # NOTE: This may be acquired from elsewhere. 64 #self.globals = {} 65 66 # Mapping from names to indexes. 67 self.names = {} 68 69 # A list of constants used as exception handler return addresses. 70 self.constants_for_exceptions = [] 71 72 # A list of external names. 73 self.external_names = [] 74 75 def get_bytecodes(self): 76 77 "Return the list of bytecodes written to the writer." 78 79 output = [] 80 for element in self.output: 81 if isinstance(element, LazySubValue): 82 value = element.value 83 else: 84 value = element 85 output.append(value) 86 return output 87 88 def get_output(self): 89 90 "Return the output of the writer as a string." 91 92 output = [] 93 for value in self.get_bytecodes(): 94 # NOTE: ValueError gets raised for bad values here. 95 output.append(chr(value)) 96 return "".join(output) 97 98 def get_constants(self): 99 100 """ 101 Return a list of constants with ordering significant to the code 102 employing them. 103 """ 104 105 l = self._get_list(self._invert(self.constants)) 106 result = [] 107 for i in l: 108 if isinstance(i, LazyValue): 109 result.append(i.get_value()) 110 else: 111 result.append(i) 112 return result 113 114 #def get_globals(self): 115 # return self._get_list(self._invert(self.globals)) 116 117 def get_names(self): 118 119 """ 120 Return a list of names with ordering significant to the code employing 121 them. 122 """ 123 124 return self._get_list(self._invert(self.names)) 125 126 def _invert(self, d): 127 128 """ 129 Return a new dictionary whose key-to-value mapping is in the inverse of 130 that found in 'd'. 131 """ 132 133 inverted = {} 134 for k, v in d.items(): 135 inverted[v] = k 136 return inverted 137 138 def _get_list(self, d): 139 140 """ 141 Traverse the dictionary 'd' returning a list whose values appear at the 142 position denoted by each value's key in 'd'. 143 """ 144 145 l = [] 146 for i in range(0, len(d.keys())): 147 l.append(d[i]) 148 return l 149 150 # Administrative methods. 151 152 def update_stack_depth(self, change): 153 154 """ 155 Given the stated 'change' in stack depth, update the maximum stack depth 156 where appropriate. 157 """ 158 159 self.stack_depth += change 160 if self.stack_depth > self.max_stack_depth: 161 self.max_stack_depth = self.stack_depth 162 163 def update_locals(self, index): 164 165 """ 166 Given the stated 'index' of a local variable, update the maximum local 167 variable index where appropriate. 168 """ 169 170 if index > self.max_locals: 171 self.max_locals = index 172 173 # Special methods. 174 175 def _write_value(self, value): 176 177 """ 178 Write the given 'value' at the current output position. 179 """ 180 181 if isinstance(value, LazyValue): 182 # NOTE: Assume a 16-bit value. 183 self.output.append(value.values[0]) 184 self.output.append(value.values[1]) 185 self.position += 2 186 elif value <= 0xffff: 187 self.output.append(value & 0xff) 188 self.output.append((value & 0xff00) >> 8) 189 self.position += 2 190 else: 191 # NOTE: EXTENDED_ARG not yet supported. 192 raise ValueError, value 193 194 def _rewrite_value(self, position, value): 195 196 """ 197 At the given output 'position', rewrite the given 'value'. 198 """ 199 200 # NOTE: Assume a 16-bit value. 201 if value <= 0xffff: 202 self.output[position] = (value & 0xff) 203 self.output[position + 1] = ((value & 0xff00) >> 8) 204 else: 205 # NOTE: EXTENDED_ARG not yet supported. 206 raise ValueError, value 207 208 # Higher level methods. 209 210 def use_external_name(self, name): 211 # NOTE: Remove array and object indicators. 212 self.external_names.append(name) 213 214 def setup_loop(self): 215 self.loops.append(self.position) 216 self.output.append(opmap["SETUP_LOOP"]) 217 self.position += 1 218 self._write_value(0) # To be filled in later 219 220 def end_loop(self): 221 current_loop_start = self.loops.pop() 222 current_loop_real_start = self.blocks.pop() 223 #print "<", self.blocks, current_loop_real_start 224 # Fix the iterator delta. 225 # NOTE: Using 3 as the assumed length of the FOR_ITER instruction. 226 self.jump_absolute(current_loop_real_start) 227 self._rewrite_value(current_loop_real_start + 1, self.position - current_loop_real_start - 3) 228 self.pop_block() 229 # Fix the loop delta. 230 # NOTE: Using 3 as the assumed length of the SETUP_LOOP instruction. 231 self._rewrite_value(current_loop_start + 1, self.position - current_loop_start - 3) 232 233 def jump_to_label(self, status, name): 234 # Record the instruction using the jump. 235 jump_instruction = self.position 236 if status is None: 237 self.jump_forward() 238 elif status: 239 self.jump_if_true() 240 else: 241 self.jump_if_false() 242 # Record the following instruction, too. 243 if not self.jumps.has_key(name): 244 self.jumps[name] = [] 245 self.jumps[name].append((jump_instruction, self.position)) 246 247 def start_label(self, name): 248 # Fill in all jump instructions. 249 for jump_instruction, following_instruction in self.jumps[name]: 250 self._rewrite_value(jump_instruction + 1, self.position - following_instruction) 251 del self.jumps[name] 252 253 def load_const_ret(self, value): 254 self.constants_for_exceptions.append(value) 255 self.load_const(value) 256 257 def ret(self, index): 258 self.load_fast(index) 259 260 # Previously, the constant stored on the stack by jsr/jsr_w was stored 261 # in a local variable. In the JVM, extracting the value from the local 262 # variable and jumping can be done at runtime. In the Python VM, any 263 # jump target must be known in advance and written into the bytecode. 264 265 for constant in self.constants_for_exceptions: 266 self.dup_top() # Stack: actual-address, actual-address 267 self.load_const(constant) # Stack: actual-address, actual-address, suggested-address 268 self.compare_op("==") # Stack: actual-address, result 269 self.jump_to_label(0, "const") 270 self.pop_top() # Stack: actual-address 271 self.pop_top() # Stack: 272 self.jump_absolute(constant) 273 self.start_label("const") 274 self.pop_top() # Stack: actual-address 275 276 # NOTE: If we get here, something is really wrong. 277 278 self.pop_top() # Stack: 279 280 def setup_except(self, target): 281 self.blocks.append(self.position) 282 self.exception_handlers.append(target) 283 #print "-", self.position, target 284 self.output.append(opmap["SETUP_EXCEPT"]) 285 self.position += 1 286 self._write_value(0) # To be filled in later 287 288 def setup_finally(self, target): 289 self.blocks.append(self.position) 290 self.exception_handlers.append(target) 291 #print "-", self.position, target 292 self.output.append(opmap["SETUP_FINALLY"]) 293 self.position += 1 294 self._write_value(0) # To be filled in later 295 296 def end_exception(self): 297 current_exception_start = self.blocks.pop() 298 # Convert the "lazy" absolute value. 299 current_exception_target = self.exception_handlers.pop() 300 # NOTE: Using 3 as the assumed length of the SETUP_* instruction. 301 self.exception_offsets.append((current_exception_start + 1, current_exception_target, current_exception_start)) 302 303 def end_exceptions(self): 304 for position, exception_target, exception_start in self.exception_offsets: 305 #print "*", exception_start, exception_target.get_value() 306 self._rewrite_value(position, exception_target.get_value() - exception_start - 3) 307 308 def start_handler(self, exc_name, class_file): 309 310 # Where handlers are begun, produce bytecode to test the type of 311 # the exception. 312 # NOTE: Since RAISE_VARARGS and END_FINALLY are not really documented, 313 # NOTE: we store the top of the stack and use it later to trigger the 314 # NOTE: magic processes when re-raising. 315 self.use_external_name(str(exc_name)) 316 317 self.rot_two() # Stack: raised-exception, exception 318 self.dup_top() # Stack: raised-exception, exception, exception 319 # Handled exceptions are wrapped before being thrown. 320 self.load_global("Exception") # Stack: raised-exception, exception, exception, Exception 321 self.compare_op("exception match") # Stack: raised-exception, exception, result 322 self.jump_to_label(0, "next") 323 self.pop_top() # Stack: raised-exception, exception 324 self.dup_top() # Stack: raised-exception, exception, exception 325 self.load_attr("args") # Stack: raised-exception, exception, args 326 self.load_const(0) # Stack: raised-exception, exception, args, 0 327 self.binary_subscr() # Stack: raised-exception, exception, exception-object 328 load_class_name(class_file, str(exc_name), self) 329 # Stack: raised-exception, exception, exception-object, handled-exception 330 self.load_global("isinstance") # Stack: raised-exception, exception, exception-object, handled-exception, isinstance 331 self.rot_three() # Stack: raised-exception, exception, isinstance, exception-object, handled-exception 332 self.call_function(2) # Stack: raised-exception, exception, result 333 self.jump_to_label(1, "handler") 334 self.start_label("next") 335 self.pop_top() # Stack: raised-exception, exception 336 self.rot_two() # Stack: exception, raised-exception 337 self.end_finally() 338 self.start_label("handler") 339 self.pop_top() # Stack: raised-exception, exception 340 341 # Complicated methods. 342 343 def load_const(self, value): 344 self.output.append(opmap["LOAD_CONST"]) 345 if not self.constants.has_key(value): 346 self.constants[value] = len(self.constants.keys()) 347 self.position += 1 348 self._write_value(self.constants[value]) 349 self.update_stack_depth(1) 350 351 def load_global(self, name): 352 self.output.append(opmap["LOAD_GLOBAL"]) 353 if not self.names.has_key(name): 354 self.names[name] = len(self.names.keys()) 355 self.position += 1 356 self._write_value(self.names[name]) 357 self.update_stack_depth(1) 358 359 def load_attr(self, name): 360 self.output.append(opmap["LOAD_ATTR"]) 361 if not self.names.has_key(name): 362 self.names[name] = len(self.names.keys()) 363 self.position += 1 364 self._write_value(self.names[name]) 365 366 def load_name(self, name): 367 self.output.append(opmap["LOAD_NAME"]) 368 if not self.names.has_key(name): 369 self.names[name] = len(self.names.keys()) 370 self.position += 1 371 self._write_value(self.names[name]) 372 self.update_stack_depth(1) 373 374 def load_fast(self, index): 375 self.output.append(opmap["LOAD_FAST"]) 376 self.position += 1 377 self._write_value(index) 378 self.update_stack_depth(1) 379 self.update_locals(index) 380 381 def store_attr(self, name): 382 self.output.append(opmap["STORE_ATTR"]) 383 if not self.names.has_key(name): 384 self.names[name] = len(self.names.keys()) 385 self.position += 1 386 self._write_value(self.names[name]) 387 self.update_stack_depth(-1) 388 389 def store_fast(self, index): 390 self.output.append(opmap["STORE_FAST"]) 391 self.position += 1 392 self._write_value(index) 393 self.update_stack_depth(-1) 394 self.update_locals(index) 395 396 def for_iter(self): 397 self.blocks.append(self.position) 398 #print ">", self.blocks 399 self.output.append(opmap["FOR_ITER"]) 400 self.position += 1 401 self._write_value(0) # To be filled in later 402 self.update_stack_depth(1) 403 404 def break_loop(self): 405 self.output.append(opmap["BREAK_LOOP"]) 406 self.position += 1 407 self.jump_absolute(self.blocks[-1]) 408 409 # Normal bytecode generators. 410 411 def get_iter(self): 412 self.output.append(opmap["GET_ITER"]) 413 self.position += 1 414 415 def jump_if_false(self, offset=0): 416 self.output.append(opmap["JUMP_IF_FALSE"]) 417 self.position += 1 418 self._write_value(offset) # May be filled in later 419 420 def jump_if_true(self, offset=0): 421 self.output.append(opmap["JUMP_IF_TRUE"]) 422 self.position += 1 423 self._write_value(offset) # May be filled in later 424 425 def jump_forward(self, offset=0): 426 self.output.append(opmap["JUMP_FORWARD"]) 427 self.position += 1 428 self._write_value(offset) # May be filled in later 429 430 def jump_absolute(self, address=0): 431 self.output.append(opmap["JUMP_ABSOLUTE"]) 432 self.position += 1 433 self._write_value(address) # May be filled in later 434 435 def build_tuple(self, count): 436 self.output.append(opmap["BUILD_TUPLE"]) 437 self.position += 1 438 self._write_value(count) 439 self.update_stack_depth(-(count - 1)) 440 441 def build_list(self, count): 442 self.output.append(opmap["BUILD_LIST"]) 443 self.position += 1 444 self._write_value(count) 445 self.update_stack_depth(-(count - 1)) 446 447 def pop_top(self): 448 self.output.append(opmap["POP_TOP"]) 449 self.position += 1 450 self.update_stack_depth(-1) 451 452 def dup_top(self): 453 self.output.append(opmap["DUP_TOP"]) 454 self.position += 1 455 self.update_stack_depth(1) 456 457 def dup_topx(self, count): 458 self.output.append(opmap["DUP_TOPX"]) 459 self.position += 1 460 self._write_value(count) 461 self.update_stack_depth(count) 462 463 def rot_two(self): 464 self.output.append(opmap["ROT_TWO"]) 465 self.position += 1 466 467 def rot_three(self): 468 self.output.append(opmap["ROT_THREE"]) 469 self.position += 1 470 471 def rot_four(self): 472 self.output.append(opmap["ROT_FOUR"]) 473 self.position += 1 474 475 def call_function(self, count): 476 self.output.append(opmap["CALL_FUNCTION"]) 477 self.position += 1 478 self._write_value(count) 479 self.update_stack_depth(-count) 480 481 def call_function_var(self, count): 482 self.output.append(opmap["CALL_FUNCTION_VAR"]) 483 self.position += 1 484 self._write_value(count) 485 self.update_stack_depth(-count-1) 486 487 def binary_subscr(self): 488 self.output.append(opmap["BINARY_SUBSCR"]) 489 self.position += 1 490 self.update_stack_depth(-1) 491 492 def binary_add(self): 493 self.output.append(opmap["BINARY_ADD"]) 494 self.position += 1 495 self.update_stack_depth(-1) 496 497 def binary_divide(self): 498 self.output.append(opmap["BINARY_DIVIDE"]) 499 self.position += 1 500 self.update_stack_depth(-1) 501 502 def binary_multiply(self): 503 self.output.append(opmap["BINARY_MULTIPLY"]) 504 self.position += 1 505 self.update_stack_depth(-1) 506 507 def binary_modulo(self): 508 self.output.append(opmap["BINARY_MODULO"]) 509 self.position += 1 510 self.update_stack_depth(-1) 511 512 def binary_subtract(self): 513 self.output.append(opmap["BINARY_SUBTRACT"]) 514 self.position += 1 515 self.update_stack_depth(-1) 516 517 def binary_and(self): 518 self.output.append(opmap["BINARY_AND"]) 519 self.position += 1 520 self.update_stack_depth(-1) 521 522 def binary_or(self): 523 self.output.append(opmap["BINARY_XOR"]) 524 self.position += 1 525 self.update_stack_depth(-1) 526 527 def binary_lshift(self): 528 self.output.append(opmap["BINARY_LSHIFT"]) 529 self.position += 1 530 self.update_stack_depth(-1) 531 532 def binary_rshift(self): 533 self.output.append(opmap["BINARY_RSHIFT"]) 534 self.position += 1 535 self.update_stack_depth(-1) 536 537 def binary_xor(self): 538 self.output.append(opmap["BINARY_XOR"]) 539 self.position += 1 540 self.update_stack_depth(-1) 541 542 def store_subscr(self): 543 self.output.append(opmap["STORE_SUBSCR"]) 544 self.position += 1 545 self.update_stack_depth(-3) 546 547 def unary_negative(self): 548 self.output.append(opmap["UNARY_NEGATIVE"]) 549 self.position += 1 550 551 def slice_0(self): 552 self.output.append(opmap["SLICE+0"]) 553 self.position += 1 554 555 def slice_1(self): 556 self.output.append(opmap["SLICE+1"]) 557 self.position += 1 558 559 def compare_op(self, op): 560 self.output.append(opmap["COMPARE_OP"]) 561 self.position += 1 562 self._write_value(list(cmp_op).index(op)) 563 self.update_stack_depth(-1) 564 565 def return_value(self): 566 self.output.append(opmap["RETURN_VALUE"]) 567 self.position += 1 568 self.update_stack_depth(-1) 569 570 def raise_varargs(self, count): 571 self.output.append(opmap["RAISE_VARARGS"]) 572 self.position += 1 573 self._write_value(count) 574 575 def pop_block(self): 576 self.output.append(opmap["POP_BLOCK"]) 577 self.position += 1 578 579 def end_finally(self): 580 self.output.append(opmap["END_FINALLY"]) 581 self.position += 1 582 583 def unpack_sequence(self, count): 584 self.output.append(opmap["UNPACK_SEQUENCE"]) 585 self.position += 1 586 self._write_value(count) 587 588 # Debugging. 589 590 def print_item(self): 591 self.output.append(opmap["PRINT_ITEM"]) 592 self.position += 1 593 594 # Utility classes and functions. 595 596 class LazyDict(UserDict): 597 def __getitem__(self, key): 598 if not self.data.has_key(key): 599 # NOTE: Assume 16-bit value. 600 self.data[key] = LazyValue(2) 601 return self.data[key] 602 def __setitem__(self, key, value): 603 if self.data.has_key(key): 604 existing_value = self.data[key] 605 if isinstance(existing_value, LazyValue): 606 existing_value.set_value(value) 607 return 608 self.data[key] = value 609 610 class LazyValue: 611 def __init__(self, nvalues): 612 self.values = [] 613 for i in range(0, nvalues): 614 self.values.append(LazySubValue()) 615 def set_value(self, value): 616 # NOTE: Assume at least 16-bit value. No "filling" performed. 617 if value <= 0xffff: 618 self.values[0].set_value(value & 0xff) 619 self.values[1].set_value((value & 0xff00) >> 8) 620 else: 621 # NOTE: EXTENDED_ARG not yet supported. 622 raise ValueError, value 623 def get_value(self): 624 value = 0 625 values = self.values[:] 626 for i in range(0, len(values)): 627 value = (value << 8) + values.pop().value 628 return value 629 630 class LazySubValue: 631 def __init__(self): 632 self.value = 0 633 def set_value(self, value): 634 self.value = value 635 636 def signed(value, limit): 637 638 """ 639 Return the signed integer from the unsigned 'value', where 'limit' (a value 640 one greater than the highest possible positive integer) is used to determine 641 whether a negative or positive result is produced. 642 """ 643 644 d, r = divmod(value, limit) 645 if d == 1: 646 mask = limit * 2 - 1 647 return -1 - (value ^ mask) 648 else: 649 return value 650 651 def signed1(value): 652 return signed(value, 0x80) 653 654 def signed2(value): 655 return signed(value, 0x8000) 656 657 def signed4(value): 658 return signed(value, 0x80000000) 659 660 def load_class_name(class_file, full_class_name, program): 661 this_class_name = str(class_file.this_class.get_python_name()) 662 this_class_parts = this_class_name.split(".") 663 class_parts = full_class_name.split(".") 664 665 # Only use the full path if different from this class's path. 666 667 if class_parts[:-1] != this_class_parts[:-1]: 668 program.use_external_name(full_class_name) 669 program.load_global(class_parts[0]) 670 for class_part in class_parts[1:]: 671 program.load_attr(class_part) # Stack: classref 672 else: 673 program.load_global(class_parts[-1]) 674 675 atypes_to_default_values = { 676 4 : 0, # bool (NOTE: Should be False.) 677 5 : u"", # char 678 6 : 0.0, # float 679 7 : 0.0, # double 680 8 : 0, # byte 681 9 : 0, # short 682 10: 0, # int 683 11: 0 # long 684 } 685 686 def get_default_for_atype(atype): 687 global atypes_to_default_values 688 return atypes_to_default_values.get(atype) 689 690 # Bytecode conversion. 691 692 class BytecodeReader: 693 694 "A generic Java bytecode reader." 695 696 def __init__(self, class_file): 697 698 """ 699 Initialise the reader with a 'class_file' containing essential 700 information for any bytecode inspection activity. 701 """ 702 703 self.class_file = class_file 704 self.position_mapping = LazyDict() 705 706 def process(self, method, program): 707 708 """ 709 Process the given 'method' (obtained from the class file), using the 710 given 'program' to write translated Python bytecode instructions. 711 """ 712 713 self.java_position = 0 714 self.in_finally = 0 715 self.method = method 716 717 # NOTE: Potentially unreliable way of getting necessary information. 718 719 code, exception_table = None, None 720 for attribute in method.attributes: 721 if isinstance(attribute, classfile.CodeAttributeInfo): 722 code, exception_table = attribute.code, attribute.exception_table 723 break 724 725 # Where no code was found, write a very simple placeholder routine. 726 # This is useful for interfaces and abstract classes. 727 # NOTE: Assess the correctness of doing this. An exception should really 728 # NOTE: be raised instead. 729 730 if code is None: 731 program.load_const(None) 732 program.return_value() 733 return 734 735 # Produce a structure which permits fast access to exception details. 736 737 exception_block_start = {} 738 exception_block_end = {} 739 exception_block_handler = {} 740 reversed_exception_table = exception_table[:] 741 reversed_exception_table.reverse() 742 743 # Later entries have wider coverage than earlier entries. 744 745 for exception in reversed_exception_table: 746 747 # NOTE: Strange case with javac from JDK 1.4 but not JDK 1.3: 748 # NOTE: start_pc == handler_pc 749 # Merge all finally handlers with the same handler location. 750 751 if exception.catch_type == 0 and exception_block_handler.get(exception.handler_pc, []) != []: 752 753 # Make a new definition. 754 755 new_exception = classfile.ExceptionInfo() 756 new_exception.catch_type = exception.catch_type 757 new_exception.handler_pc = exception.handler_pc 758 new_exception.end_pc = exception.end_pc 759 new_exception.start_pc = exception.start_pc 760 761 # Find the previous exception handler definition. 762 763 for previous_exception in exception_block_handler[exception.handler_pc][:]: 764 if previous_exception.catch_type == 0: 765 new_exception.end_pc = max(new_exception.end_pc, previous_exception.end_pc) 766 new_exception.start_pc = min(new_exception.start_pc, previous_exception.start_pc) 767 768 # Remove this exception from the lists. 769 770 exception_block_handler[previous_exception.handler_pc].remove(previous_exception) 771 exception_block_start[previous_exception.start_pc].remove(previous_exception) 772 exception_block_end[previous_exception.end_pc].remove(previous_exception) 773 break 774 775 # Use the new definition instead. 776 777 exception = new_exception 778 779 # Index start positions. 780 781 if not exception_block_start.has_key(exception.start_pc): 782 exception_block_start[exception.start_pc] = [] 783 exception_block_start[exception.start_pc].append(exception) 784 785 # Index end positions. 786 787 if not exception_block_end.has_key(exception.end_pc): 788 exception_block_end[exception.end_pc] = [] 789 exception_block_end[exception.end_pc].append(exception) 790 791 # Index handler positions. 792 793 if not exception_block_handler.has_key(exception.handler_pc): 794 exception_block_handler[exception.handler_pc] = [] 795 exception_block_handler[exception.handler_pc].append(exception) 796 797 # Process each instruction in the code. 798 799 while self.java_position < len(code): 800 self.position_mapping[self.java_position] = program.position 801 802 # Insert exception handling constructs. 803 804 block_starts = exception_block_start.get(self.java_position, []) 805 for exception in block_starts: 806 807 # Note that the absolute position is used. 808 809 if exception.catch_type == 0: 810 program.setup_finally(self.position_mapping[exception.handler_pc]) 811 else: 812 program.setup_except(self.position_mapping[exception.handler_pc]) 813 814 if block_starts: 815 self.in_finally = 0 816 817 # Insert exception handler details. 818 # NOTE: Ensure that pop_block is reachable by possibly inserting it at the start of finally handlers. 819 # NOTE: Insert a check for the correct exception at the start of each handler. 820 821 for exception in exception_block_handler.get(self.java_position, []): 822 program.end_exception() 823 if exception.catch_type == 0: 824 self.in_finally = 1 825 else: 826 program.start_handler(self.class_file.constants[exception.catch_type - 1].get_python_name(), self.class_file) 827 828 # Process the bytecode at the current position. 829 830 bytecode = ord(code[self.java_position]) 831 mnemonic, number_of_arguments = self.java_bytecodes[bytecode] 832 number_of_arguments = self.process_bytecode(mnemonic, number_of_arguments, code, program) 833 next_java_position = self.java_position + 1 + number_of_arguments 834 835 # Insert exception block end details. 836 837 for exception in exception_block_end.get(next_java_position, []): 838 839 # NOTE: Insert jump beyond handlers. 840 # NOTE: program.jump_forward/absolute(...) 841 # NOTE: Insert end finally at end of handlers as well as where "ret" occurs. 842 843 if exception.catch_type != 0: 844 program.pop_block() 845 846 # Only advance the JVM position after sneaking in extra Python 847 # instructions. 848 849 self.java_position = next_java_position 850 851 # Tidy up exceptions. 852 853 program.end_exceptions() 854 855 def process_bytecode(self, mnemonic, number_of_arguments, code, program): 856 857 """ 858 Process a bytecode instruction with the given 'mnemonic' and 859 'number_of_arguments'. The 'code' parameter contains the full method 860 code so that argument data can be inspected. The 'program' parameter is 861 used to produce a Python translation of the instruction. 862 """ 863 864 if number_of_arguments is not None: 865 arguments = [] 866 for j in range(0, number_of_arguments): 867 arguments.append(ord(code[self.java_position + 1 + j])) 868 869 # Call the handler. 870 871 getattr(self, mnemonic)(arguments, program) 872 return number_of_arguments 873 else: 874 # Call the handler. 875 876 return getattr(self, mnemonic)(code[self.java_position+1:], program) 877 878 java_bytecodes = { 879 # code : (mnemonic, number of following bytes, change in stack) 880 0 : ("nop", 0), 881 1 : ("aconst_null", 0), 882 2 : ("iconst_m1", 0), 883 3 : ("iconst_0", 0), 884 4 : ("iconst_1", 0), 885 5 : ("iconst_2", 0), 886 6 : ("iconst_3", 0), 887 7 : ("iconst_4", 0), 888 8 : ("iconst_5", 0), 889 9 : ("lconst_0", 0), 890 10 : ("lconst_1", 0), 891 11 : ("fconst_0", 0), 892 12 : ("fconst_1", 0), 893 13 : ("fconst_2", 0), 894 14 : ("dconst_0", 0), 895 15 : ("dconst_1", 0), 896 16 : ("bipush", 1), 897 17 : ("sipush", 2), 898 18 : ("ldc", 1), 899 19 : ("ldc_w", 2), 900 20 : ("ldc2_w", 2), 901 21 : ("iload", 1), 902 22 : ("lload", 1), 903 23 : ("fload", 1), 904 24 : ("dload", 1), 905 25 : ("aload", 1), 906 26 : ("iload_0", 0), 907 27 : ("iload_1", 0), 908 28 : ("iload_2", 0), 909 29 : ("iload_3", 0), 910 30 : ("lload_0", 0), 911 31 : ("lload_1", 0), 912 32 : ("lload_2", 0), 913 33 : ("lload_3", 0), 914 34 : ("fload_0", 0), 915 35 : ("fload_1", 0), 916 36 : ("fload_2", 0), 917 37 : ("fload_3", 0), 918 38 : ("dload_0", 0), 919 39 : ("dload_1", 0), 920 40 : ("dload_2", 0), 921 41 : ("dload_3", 0), 922 42 : ("aload_0", 0), 923 43 : ("aload_1", 0), 924 44 : ("aload_2", 0), 925 45 : ("aload_3", 0), 926 46 : ("iaload", 0), 927 47 : ("laload", 0), 928 48 : ("faload", 0), 929 49 : ("daload", 0), 930 50 : ("aaload", 0), 931 51 : ("baload", 0), 932 52 : ("caload", 0), 933 53 : ("saload", 0), 934 54 : ("istore", 1), 935 55 : ("lstore", 1), 936 56 : ("fstore", 1), 937 57 : ("dstore", 1), 938 58 : ("astore", 1), 939 59 : ("istore_0", 0), 940 60 : ("istore_1", 0), 941 61 : ("istore_2", 0), 942 62 : ("istore_3", 0), 943 63 : ("lstore_0", 0), 944 64 : ("lstore_1", 0), 945 65 : ("lstore_2", 0), 946 66 : ("lstore_3", 0), 947 67 : ("fstore_0", 0), 948 68 : ("fstore_1", 0), 949 69 : ("fstore_2", 0), 950 70 : ("fstore_3", 0), 951 71 : ("dstore_0", 0), 952 72 : ("dstore_1", 0), 953 73 : ("dstore_2", 0), 954 74 : ("dstore_3", 0), 955 75 : ("astore_0", 0), 956 76 : ("astore_1", 0), 957 77 : ("astore_2", 0), 958 78 : ("astore_3", 0), 959 79 : ("iastore", 0), 960 80 : ("lastore", 0), 961 81 : ("fastore", 0), 962 82 : ("dastore", 0), 963 83 : ("aastore", 0), 964 84 : ("bastore", 0), 965 85 : ("castore", 0), 966 86 : ("sastore", 0), 967 87 : ("pop", 0), 968 88 : ("pop2", 0), 969 89 : ("dup", 0), 970 90 : ("dup_x1", 0), 971 91 : ("dup_x2", 0), 972 92 : ("dup2", 0), 973 93 : ("dup2_x1", 0), 974 94 : ("dup2_x2", 0), 975 95 : ("swap", 0), 976 96 : ("iadd", 0), 977 97 : ("ladd", 0), 978 98 : ("fadd", 0), 979 99 : ("dadd", 0), 980 100 : ("isub", 0), 981 101 : ("lsub", 0), 982 102 : ("fsub", 0), 983 103 : ("dsub", 0), 984 104 : ("imul", 0), 985 105 : ("lmul", 0), 986 106 : ("fmul", 0), 987 107 : ("dmul", 0), 988 108 : ("idiv", 0), 989 109 : ("ldiv", 0), 990 110 : ("fdiv", 0), 991 111 : ("ddiv", 0), 992 112 : ("irem", 0), 993 113 : ("lrem", 0), 994 114 : ("frem", 0), 995 115 : ("drem", 0), 996 116 : ("ineg", 0), 997 117 : ("lneg", 0), 998 118 : ("fneg", 0), 999 119 : ("dneg", 0), 1000 120 : ("ishl", 0), 1001 121 : ("lshl", 0), 1002 122 : ("ishr", 0), 1003 123 : ("lshr", 0), 1004 124 : ("iushr", 0), 1005 125 : ("lushr", 0), 1006 126 : ("iand", 0), 1007 127 : ("land", 0), 1008 128 : ("ior", 0), 1009 129 : ("lor", 0), 1010 130 : ("ixor", 0), 1011 131 : ("lxor", 0), 1012 132 : ("iinc", 2), 1013 133 : ("i2l", 0), 1014 134 : ("i2f", 0), 1015 135 : ("i2d", 0), 1016 136 : ("l2i", 0), 1017 137 : ("l2f", 0), 1018 138 : ("l2d", 0), 1019 139 : ("f2i", 0), 1020 140 : ("f2l", 0), 1021 141 : ("f2d", 0), 1022 142 : ("d2i", 0), 1023 143 : ("d2l", 0), 1024 144 : ("d2f", 0), 1025 145 : ("i2b", 0), 1026 146 : ("i2c", 0), 1027 147 : ("i2s", 0), 1028 148 : ("lcmp", 0), 1029 149 : ("fcmpl", 0), 1030 150 : ("fcmpg", 0), 1031 151 : ("dcmpl", 0), 1032 152 : ("dcmpg", 0), 1033 153 : ("ifeq", 2), 1034 154 : ("ifne", 2), 1035 155 : ("iflt", 2), 1036 156 : ("ifge", 2), 1037 157 : ("ifgt", 2), 1038 158 : ("ifle", 2), 1039 159 : ("if_icmpeq", 2), 1040 160 : ("if_icmpne", 2), 1041 161 : ("if_icmplt", 2), 1042 162 : ("if_icmpge", 2), 1043 163 : ("if_icmpgt", 2), 1044 164 : ("if_icmple", 2), 1045 165 : ("if_acmpeq", 2), 1046 166 : ("if_acmpne", 2), 1047 167 : ("goto", 2), 1048 168 : ("jsr", 2), 1049 169 : ("ret", 1), 1050 170 : ("tableswitch", None), # variable number of arguments 1051 171 : ("lookupswitch", None), # variable number of arguments 1052 172 : ("ireturn", 0), 1053 173 : ("lreturn", 0), 1054 174 : ("freturn", 0), 1055 175 : ("dreturn", 0), 1056 176 : ("areturn", 0), 1057 177 : ("return_", 0), 1058 178 : ("getstatic", 2), 1059 179 : ("putstatic", 2), 1060 180 : ("getfield", 2), 1061 181 : ("putfield", 2), 1062 182 : ("invokevirtual", 2), 1063 183 : ("invokespecial", 2), 1064 184 : ("invokestatic", 2), 1065 185 : ("invokeinterface", 4), 1066 187 : ("new", 2), 1067 188 : ("newarray", 1), 1068 189 : ("anewarray", 2), 1069 190 : ("arraylength", 0), 1070 191 : ("athrow", 0), 1071 192 : ("checkcast", 2), 1072 193 : ("instanceof", 2), 1073 194 : ("monitorenter", 0), 1074 195 : ("monitorexit", 0), 1075 196 : ("wide", None), # 3 or 5 arguments, stack changes according to modified element 1076 197 : ("multianewarray", 3), 1077 198 : ("ifnull", 2), 1078 199 : ("ifnonnull", 2), 1079 200 : ("goto_w", 4), 1080 201 : ("jsr_w", 4), 1081 } 1082 1083 class BytecodeDisassembler(BytecodeReader): 1084 1085 "A Java bytecode disassembler." 1086 1087 bytecode_methods = [spec[0] for spec in BytecodeReader.java_bytecodes.values()] 1088 1089 def __getattr__(self, name): 1090 if name in self.bytecode_methods: 1091 print "%5s %s" % (self.java_position, name), 1092 return self.generic 1093 else: 1094 raise AttributeError, name 1095 1096 def generic(self, arguments, program): 1097 print arguments 1098 1099 def lookupswitch(self, code, program): 1100 print "%5s lookupswitch" % (self.java_position,), 1101 d, r = divmod(self.java_position + 1, 4) 1102 to_boundary = (4 - r) % 4 1103 code = code[to_boundary:] 1104 default = classfile.u4(code[0:4]) 1105 npairs = classfile.u4(code[4:8]) 1106 print default, npairs 1107 return to_boundary + 8 + npairs * 8 1108 1109 def tableswitch(self, code, program): 1110 print "%5s tableswitch" % (self.java_position,), 1111 d, r = divmod(self.java_position + 1, 4) 1112 to_boundary = (4 - r) % 4 1113 code = code[to_boundary:] 1114 default = classfile.u4(code[0:4]) 1115 low = classfile.u4(code[4:8]) 1116 high = classfile.u4(code[8:12]) 1117 print default, low, high 1118 return to_boundary + 12 + (high - low + 1) * 4 1119 1120 class BytecodeDisassemblerProgram: 1121 position = 0 1122 def setup_except(self, target): 1123 print "(setup_except %s)" % target 1124 def setup_finally(self, target): 1125 print "(setup_finally %s)" % target 1126 def end_exception(self): 1127 print "(end_exception)" 1128 def start_handler(self, exc_name, class_file): 1129 print "(start_handler %s)" % exc_name 1130 def pop_block(self): 1131 print "(pop_block)" 1132 def load_const(self, const): 1133 print "(load_const %s)" % const 1134 def return_value(self): 1135 print "(return_value)" 1136 1137 class BytecodeTranslator(BytecodeReader): 1138 1139 "A Java bytecode translator which uses a Python bytecode writer." 1140 1141 def aaload(self, arguments, program): 1142 # NOTE: No type checking performed. 1143 program.binary_subscr() 1144 1145 def aastore(self, arguments, program): 1146 # NOTE: No type checking performed. 1147 # Stack: arrayref, index, value 1148 program.rot_three() # Stack: value, arrayref, index 1149 program.store_subscr() 1150 1151 def aconst_null(self, arguments, program): 1152 program.load_const(None) 1153 1154 def aload(self, arguments, program): 1155 program.load_fast(arguments[0]) 1156 1157 def aload_0(self, arguments, program): 1158 program.load_fast(0) 1159 1160 def aload_1(self, arguments, program): 1161 program.load_fast(1) 1162 1163 def aload_2(self, arguments, program): 1164 program.load_fast(2) 1165 1166 def aload_3(self, arguments, program): 1167 program.load_fast(3) 1168 1169 def anewarray(self, arguments, program): 1170 # NOTE: Does not raise NegativeArraySizeException. 1171 # NOTE: Not using the index to type the list/array. 1172 index = (arguments[0] << 8) + arguments[1] 1173 type_name = self.class_file.constants[index - 1].get_python_name() 1174 default_value = classfile.get_default_for_type(type_name) 1175 self._newarray(program, type_name) 1176 1177 def _newarray(self, program, default_value): 1178 program.build_list(0) # Stack: count, list 1179 program.rot_two() # Stack: list, count 1180 program.setup_loop() 1181 program.load_global("range") 1182 program.load_const(0) # Stack: list, count, range, 0 1183 program.rot_three() # Stack: list, 0, count, range 1184 program.rot_three() # Stack: list, range, 0, count 1185 program.call_function(2) # Stack: list, range_list 1186 program.get_iter() # Stack: list, iter 1187 program.for_iter() # Stack: list, iter, value 1188 program.pop_top() # Stack: list, iter 1189 program.rot_two() # Stack: iter, list 1190 program.dup_top() # Stack: iter, list, list 1191 program.load_attr("append") # Stack: iter, list, append 1192 program.load_const(default_value) # Stack: iter, list, append, default 1193 program.call_function(1) # Stack: iter, list, default 1194 program.pop_top() # Stack: iter, list 1195 program.rot_two() # Stack: list, iter 1196 program.end_loop() # Back to for_iter above 1197 1198 def areturn(self, arguments, program): 1199 program.return_value() 1200 1201 def arraylength(self, arguments, program): 1202 program.load_global("len") # Stack: arrayref, len 1203 program.rot_two() # Stack: len, arrayref 1204 program.call_function(1) 1205 1206 def astore(self, arguments, program): 1207 program.store_fast(arguments[0]) 1208 1209 def astore_0(self, arguments, program): 1210 program.store_fast(0) 1211 1212 def astore_1(self, arguments, program): 1213 program.store_fast(1) 1214 1215 def astore_2(self, arguments, program): 1216 program.store_fast(2) 1217 1218 def astore_3(self, arguments, program): 1219 program.store_fast(3) 1220 1221 def athrow(self, arguments, program): 1222 # NOTE: NullPointerException not raised where null/None is found on the stack. 1223 # If this instruction appears in a finally handler, use end_finally instead. 1224 if self.in_finally: 1225 program.end_finally() 1226 else: 1227 # Wrap the exception in a Python exception. 1228 program.load_global("Exception") # Stack: objectref, Exception 1229 program.rot_two() # Stack: Exception, objectref 1230 program.call_function(1) # Stack: exception 1231 program.raise_varargs(1) 1232 # NOTE: This seems to put another object on the stack. 1233 1234 baload = aaload 1235 bastore = aastore 1236 1237 def bipush(self, arguments, program): 1238 program.load_const(signed1(arguments[0])) 1239 1240 caload = aaload 1241 castore = aastore 1242 1243 def checkcast(self, arguments, program): 1244 index = (arguments[0] << 8) + arguments[1] 1245 target_name = self.class_file.constants[index - 1].get_python_name() 1246 program.use_external_name(target_name) 1247 program.dup_top() # Stack: objectref, objectref 1248 program.load_const(None) # Stack: objectref, objectref, None 1249 program.compare_op("is") # Stack: objectref, result 1250 program.jump_to_label(1, "next") 1251 program.pop_top() # Stack: objectref 1252 program.dup_top() # Stack: objectref, objectref 1253 program.load_global("isinstance") # Stack: objectref, objectref, isinstance 1254 program.rot_two() # Stack: objectref, isinstance, objectref 1255 load_class_name(self.class_file, target_name, program) 1256 program.call_function(2) # Stack: objectref, result 1257 program.jump_to_label(1, "next") 1258 program.pop_top() # Stack: objectref 1259 program.pop_top() # Stack: 1260 program.use_external_name("java.lang.ClassCastException") 1261 load_class_name(self.class_file, "java.lang.ClassCastException", program) 1262 program.call_function(0) # Stack: exception 1263 # Wrap the exception in a Python exception. 1264 program.load_global("Exception") # Stack: exception, Exception 1265 program.rot_two() # Stack: Exception, exception 1266 program.call_function(1) # Stack: exception 1267 program.raise_varargs(1) 1268 # NOTE: This seems to put another object on the stack. 1269 program.start_label("next") 1270 program.pop_top() # Stack: objectref 1271 1272 def d2f(self, arguments, program): 1273 pass 1274 1275 def d2i(self, arguments, program): 1276 program.load_global("int") # Stack: value, int 1277 program.rot_two() # Stack: int, value 1278 program.call_function(1) # Stack: result 1279 1280 d2l = d2i # Preserving Java semantics 1281 1282 def dadd(self, arguments, program): 1283 # NOTE: No type checking performed. 1284 program.binary_add() 1285 1286 daload = aaload 1287 dastore = aastore 1288 1289 def dcmpg(self, arguments, program): 1290 # NOTE: No type checking performed. 1291 program.compare_op(">") 1292 1293 def dcmpl(self, arguments, program): 1294 # NOTE: No type checking performed. 1295 program.compare_op("<") 1296 1297 def dconst_0(self, arguments, program): 1298 program.load_const(0.0) 1299 1300 def dconst_1(self, arguments, program): 1301 program.load_const(1.0) 1302 1303 def ddiv(self, arguments, program): 1304 # NOTE: No type checking performed. 1305 program.binary_divide() 1306 1307 dload = aload 1308 dload_0 = aload_0 1309 dload_1 = aload_1 1310 dload_2 = aload_2 1311 dload_3 = aload_3 1312 1313 def dmul(self, arguments, program): 1314 # NOTE: No type checking performed. 1315 program.binary_multiply() 1316 1317 def dneg(self, arguments, program): 1318 # NOTE: No type checking performed. 1319 program.unary_negative() 1320 1321 def drem(self, arguments, program): 1322 # NOTE: No type checking performed. 1323 program.binary_modulo() 1324 1325 dreturn = areturn 1326 dstore = astore 1327 dstore_0 = astore_0 1328 dstore_1 = astore_1 1329 dstore_2 = astore_2 1330 dstore_3 = astore_3 1331 1332 def dsub(self, arguments, program): 1333 # NOTE: No type checking performed. 1334 program.binary_subtract() 1335 1336 def dup(self, arguments, program): 1337 program.dup_top() 1338 1339 def dup_x1(self, arguments, program): 1340 # Ignoring computational type categories. 1341 program.dup_top() 1342 program.rot_three() 1343 1344 def dup_x2(self, arguments, program): 1345 # Ignoring computational type categories. 1346 program.dup_top() 1347 program.rot_four() 1348 1349 dup2 = dup # Ignoring computational type categories 1350 dup2_x1 = dup_x1 # Ignoring computational type categories 1351 dup2_x2 = dup_x2 # Ignoring computational type categories 1352 1353 def f2d(self, arguments, program): 1354 pass # Preserving Java semantics 1355 1356 def f2i(self, arguments, program): 1357 program.load_global("int") # Stack: value, int 1358 program.rot_two() # Stack: int, value 1359 program.call_function(1) # Stack: result 1360 1361 f2l = f2i # Preserving Java semantics 1362 fadd = dadd 1363 faload = daload 1364 fastore = dastore 1365 fcmpg = dcmpg 1366 fcmpl = dcmpl 1367 fconst_0 = dconst_0 1368 fconst_1 = dconst_1 1369 1370 def fconst_2(self, arguments, program): 1371 program.load_const(2.0) 1372 1373 fdiv = ddiv 1374 fload = dload 1375 fload_0 = dload_0 1376 fload_1 = dload_1 1377 fload_2 = dload_2 1378 fload_3 = dload_3 1379 fmul = dmul 1380 fneg = dneg 1381 frem = drem 1382 freturn = dreturn 1383 fstore = dstore 1384 fstore_0 = dstore_0 1385 fstore_1 = dstore_1 1386 fstore_2 = dstore_2 1387 fstore_3 = dstore_3 1388 fsub = dsub 1389 1390 def getfield(self, arguments, program): 1391 index = (arguments[0] << 8) + arguments[1] 1392 target_name = self.class_file.constants[index - 1].get_python_name() 1393 # NOTE: Using the string version of the name which may contain incompatible characters. 1394 program.load_attr(str(target_name)) 1395 1396 def getstatic(self, arguments, program): 1397 index = (arguments[0] << 8) + arguments[1] 1398 target = self.class_file.constants[index - 1] 1399 target_name = target.get_python_name() 1400 1401 # Get the class name instead of the fully qualified name. 1402 1403 full_class_name = target.get_class().get_python_name() 1404 program.use_external_name(full_class_name) 1405 load_class_name(self.class_file, full_class_name, program) 1406 # NOTE: Using the string version of the name which may contain incompatible characters. 1407 program.load_attr(str(target_name)) 1408 1409 def goto(self, arguments, program): 1410 offset = signed2((arguments[0] << 8) + arguments[1]) 1411 java_absolute = self.java_position + offset 1412 program.jump_absolute(self.position_mapping[java_absolute]) 1413 1414 def goto_w(self, arguments, program): 1415 offset = signed4((arguments[0] << 24) + (arguments[1] << 16) + (arguments[2] << 8) + arguments[3]) 1416 java_absolute = self.java_position + offset 1417 program.jump_absolute(self.position_mapping[java_absolute]) 1418 1419 def i2b(self, arguments, program): 1420 pass 1421 1422 def i2c(self, arguments, program): 1423 pass 1424 1425 def i2d(self, arguments, program): 1426 program.load_global("float") # Stack: value, float 1427 program.rot_two() # Stack: float, value 1428 program.call_function(1) # Stack: result 1429 1430 i2f = i2d # Not distinguishing between float and double 1431 1432 def i2l(self, arguments, program): 1433 pass # Preserving Java semantics 1434 1435 def i2s(self, arguments, program): 1436 pass # Not distinguishing between int and short 1437 1438 iadd = fadd 1439 iaload = faload 1440 1441 def iand(self, arguments, program): 1442 # NOTE: No type checking performed. 1443 program.binary_and() 1444 1445 iastore = fastore 1446 1447 def iconst_m1(self, arguments, program): 1448 program.load_const(-1) 1449 1450 def iconst_0(self, arguments, program): 1451 program.load_const(0) 1452 1453 def iconst_1(self, arguments, program): 1454 program.load_const(1) 1455 1456 def iconst_2(self, arguments, program): 1457 program.load_const(2) 1458 1459 def iconst_3(self, arguments, program): 1460 program.load_const(3) 1461 1462 def iconst_4(self, arguments, program): 1463 program.load_const(4) 1464 1465 def iconst_5(self, arguments, program): 1466 program.load_const(5) 1467 1468 idiv = fdiv 1469 1470 def _if_xcmpx(self, arguments, program, op): 1471 offset = signed2((arguments[0] << 8) + arguments[1]) 1472 java_absolute = self.java_position + offset 1473 program.compare_op(op) 1474 program.jump_to_label(0, "next") # skip if false 1475 program.pop_top() 1476 program.jump_absolute(self.position_mapping[java_absolute]) 1477 program.start_label("next") 1478 program.pop_top() 1479 1480 def if_acmpeq(self, arguments, program): 1481 # NOTE: No type checking performed. 1482 self._if_xcmpx(arguments, program, "is") 1483 1484 def if_acmpne(self, arguments, program): 1485 # NOTE: No type checking performed. 1486 self._if_xcmpx(arguments, program, "is not") 1487 1488 def if_icmpeq(self, arguments, program): 1489 # NOTE: No type checking performed. 1490 self._if_xcmpx(arguments, program, "==") 1491 1492 def if_icmpne(self, arguments, program): 1493 # NOTE: No type checking performed. 1494 self._if_xcmpx(arguments, program, "!=") 1495 1496 def if_icmplt(self, arguments, program): 1497 # NOTE: No type checking performed. 1498 self._if_xcmpx(arguments, program, "<") 1499 1500 def if_icmpge(self, arguments, program): 1501 # NOTE: No type checking performed. 1502 self._if_xcmpx(arguments, program, ">=") 1503 1504 def if_icmpgt(self, arguments, program): 1505 # NOTE: No type checking performed. 1506 self._if_xcmpx(arguments, program, ">") 1507 1508 def if_icmple(self, arguments, program): 1509 # NOTE: No type checking performed. 1510 self._if_xcmpx(arguments, program, "<=") 1511 1512 def ifeq(self, arguments, program): 1513 # NOTE: No type checking performed. 1514 program.load_const(0) 1515 self._if_xcmpx(arguments, program, "==") 1516 1517 def ifne(self, arguments, program): 1518 # NOTE: No type checking performed. 1519 program.load_const(0) 1520 self._if_xcmpx(arguments, program, "!=") 1521 1522 def iflt(self, arguments, program): 1523 # NOTE: No type checking performed. 1524 program.load_const(0) 1525 self._if_xcmpx(arguments, program, "<") 1526 1527 def ifge(self, arguments, program): 1528 # NOTE: No type checking performed. 1529 program.load_const(0) 1530 self._if_xcmpx(arguments, program, ">=") 1531 1532 def ifgt(self, arguments, program): 1533 # NOTE: No type checking performed. 1534 program.load_const(0) 1535 self._if_xcmpx(arguments, program, ">") 1536 1537 def ifle(self, arguments, program): 1538 # NOTE: No type checking performed. 1539 program.load_const(0) 1540 self._if_xcmpx(arguments, program, "<=") 1541 1542 def ifnonnull(self, arguments, program): 1543 # NOTE: No type checking performed. 1544 program.load_const(None) 1545 self._if_xcmpx(arguments, program, "is not") 1546 1547 def ifnull(self, arguments, program): 1548 # NOTE: No type checking performed. 1549 program.load_const(None) 1550 self._if_xcmpx(arguments, program, "is") 1551 1552 def iinc(self, arguments, program): 1553 # NOTE: No type checking performed. 1554 program.load_fast(arguments[0]) 1555 program.load_const(arguments[1]) 1556 program.binary_add() 1557 program.store_fast(arguments[0]) 1558 1559 iload = fload 1560 iload_0 = fload_0 1561 iload_1 = fload_1 1562 iload_2 = fload_2 1563 iload_3 = fload_3 1564 imul = fmul 1565 ineg = fneg 1566 1567 def instanceof(self, arguments, program): 1568 index = (arguments[0] << 8) + arguments[1] 1569 target_name = self.class_file.constants[index - 1].get_python_name() 1570 program.use_external_name(target_name) 1571 program.load_global("isinstance") # Stack: objectref, isinstance 1572 program.rot_two() # Stack: isinstance, objectref 1573 load_class_name(self.class_file, target_name, program) 1574 program.call_function(2) # Stack: result 1575 1576 def _invoke(self, target_name, program): 1577 # NOTE: Using the string version of the name which may contain incompatible characters. 1578 program.load_attr(str(target_name)) # Stack: tuple, method 1579 program.rot_two() # Stack: method, tuple 1580 program.call_function_var(0) # Stack: result 1581 1582 def invokeinterface(self, arguments, program): 1583 # NOTE: This implementation does not perform the necessary checks for 1584 # NOTE: signature-based polymorphism. 1585 # NOTE: Java rules not specifically obeyed. 1586 index = (arguments[0] << 8) + arguments[1] 1587 # NOTE: "count" == nargs + 1, apparently. 1588 count = arguments[2] - 1 1589 target_name = self.class_file.constants[index - 1].get_python_name() 1590 # Stack: objectref, arg1, arg2, ... 1591 program.build_tuple(count) # Stack: objectref, tuple 1592 program.rot_two() # Stack: tuple, objectref 1593 # NOTE: The interface information is not used to discover the correct 1594 # NOTE: method. 1595 self._invoke(target_name, program) 1596 1597 def invokespecial(self, arguments, program): 1598 # NOTE: This implementation does not perform the necessary checks for 1599 # NOTE: signature-based polymorphism. 1600 # NOTE: Java rules not specifically obeyed. 1601 index = (arguments[0] << 8) + arguments[1] 1602 target = self.class_file.constants[index - 1] 1603 original_name = target.get_name() 1604 target_name = target.get_python_name() 1605 1606 # Get the number of parameters from the descriptor. 1607 1608 count = len(target.get_descriptor()[0]) 1609 1610 # First, we build a tuple of the reference and arguments. 1611 1612 program.build_tuple(count + 1) # Stack: tuple 1613 1614 # Get the class name instead of the fully qualified name. 1615 # NOTE: Not bothering with Object initialisation. 1616 1617 full_class_name = target.get_class().get_python_name() 1618 if full_class_name not in ("java.lang.Object", "java.lang.Exception"): 1619 program.use_external_name(full_class_name) 1620 load_class_name(self.class_file, full_class_name, program) 1621 self._invoke(target_name, program) 1622 1623 # Remove Python None return value. 1624 1625 if str(original_name) == "<init>": 1626 program.pop_top() 1627 1628 def invokestatic(self, arguments, program): 1629 # NOTE: This implementation does not perform the necessary checks for 1630 # NOTE: signature-based polymorphism. 1631 # NOTE: Java rules not specifically obeyed. 1632 index = (arguments[0] << 8) + arguments[1] 1633 target = self.class_file.constants[index - 1] 1634 target_name = target.get_python_name() 1635 1636 # Get the number of parameters from the descriptor. 1637 1638 count = len(target.get_descriptor()[0]) 1639 1640 # Stack: arg1, arg2, ... 1641 1642 program.build_tuple(count) # Stack: tuple 1643 1644 # Use the class to provide access to static methods. 1645 # Get the class name instead of the fully qualified name. 1646 1647 full_class_name = target.get_class().get_python_name() 1648 if full_class_name not in ("java.lang.Object", "java.lang.Exception"): 1649 program.use_external_name(full_class_name) 1650 load_class_name(self.class_file, full_class_name, program) 1651 self._invoke(target_name, program) 1652 1653 def invokevirtual (self, arguments, program): 1654 # NOTE: This implementation does not perform the necessary checks for 1655 # NOTE: signature-based polymorphism. 1656 # NOTE: Java rules not specifically obeyed. 1657 index = (arguments[0] << 8) + arguments[1] 1658 target = self.class_file.constants[index - 1] 1659 target_name = target.get_python_name() 1660 # Get the number of parameters from the descriptor. 1661 count = len(target.get_descriptor()[0]) 1662 # Stack: objectref, arg1, arg2, ... 1663 program.build_tuple(count) # Stack: objectref, tuple 1664 program.rot_two() # Stack: tuple, objectref 1665 self._invoke(target_name, program) 1666 1667 def ior(self, arguments, program): 1668 # NOTE: No type checking performed. 1669 program.binary_or() 1670 1671 irem = frem 1672 ireturn = freturn 1673 1674 def ishl(self, arguments, program): 1675 # NOTE: No type checking performed. 1676 # NOTE: Not verified. 1677 program.binary_lshift() 1678 1679 def ishr(self, arguments, program): 1680 # NOTE: No type checking performed. 1681 # NOTE: Not verified. 1682 program.binary_rshift() 1683 1684 istore = fstore 1685 istore_0 = fstore_0 1686 istore_1 = fstore_1 1687 istore_2 = fstore_2 1688 istore_3 = fstore_3 1689 isub = fsub 1690 iushr = ishr # Ignoring distinctions between arithmetic and logical shifts 1691 1692 def ixor(self, arguments, program): 1693 # NOTE: No type checking performed. 1694 program.binary_xor() 1695 1696 def jsr(self, arguments, program): 1697 offset = signed2((arguments[0] << 8) + arguments[1]) 1698 java_absolute = self.java_position + offset 1699 # Store the address of the next instruction. 1700 program.load_const_ret(self.position_mapping[self.java_position + 3]) 1701 program.jump_absolute(self.position_mapping[java_absolute]) 1702 1703 def jsr_w(self, arguments, program): 1704 offset = signed4((arguments[0] << 24) + (arguments[1] << 16) + (arguments[2] << 8) + arguments[3]) 1705 java_absolute = self.java_position + offset 1706 # Store the address of the next instruction. 1707 program.load_const_ret(self.position_mapping[self.java_position + 5]) 1708 program.jump_absolute(self.position_mapping[java_absolute]) 1709 1710 l2d = i2d 1711 l2f = i2f 1712 1713 def l2i(self, arguments, program): 1714 pass # Preserving Java semantics 1715 1716 ladd = iadd 1717 laload = iaload 1718 land = iand 1719 lastore = iastore 1720 1721 def lcmp(self, arguments, program): 1722 # NOTE: No type checking performed. 1723 program.dup_topx(2) # Stack: value1, value2, value1, value2 1724 program.compare_op(">") # Stack: value1, value2, result 1725 program.jump_to_label(0, "equals") 1726 # True - produce result and branch. 1727 program.pop_top() # Stack: value1, value2 1728 program.pop_top() # Stack: value1 1729 program.pop_top() # Stack: 1730 program.load_const(1) # Stack: 1 1731 program.jump_to_label(None, "next") 1732 # False - test equality. 1733 program.start_label("equals") 1734 program.pop_top() # Stack: value1, value2 1735 program.dup_topx(2) # Stack: value1, value2, value1, value2 1736 program.compare_op("==") # Stack: value1, value2, result 1737 program.jump_to_label(0, "less") 1738 # True - produce result and branch. 1739 program.pop_top() # Stack: value1, value2 1740 program.pop_top() # Stack: value1 1741 program.pop_top() # Stack: 1742 program.load_const(0) # Stack: 0 1743 program.jump_to_label(None, "next") 1744 # False - produce result. 1745 program.start_label("less") 1746 program.pop_top() # Stack: value1, value2 1747 program.pop_top() # Stack: value1 1748 program.pop_top() # Stack: 1749 program.load_const(-1) # Stack: -1 1750 program.start_label("next") 1751 1752 lconst_0 = iconst_0 1753 lconst_1 = iconst_1 1754 1755 def ldc(self, arguments, program): 1756 const = self.class_file.constants[arguments[0] - 1] 1757 if isinstance(const, classfile.StringInfo): 1758 program.use_external_name("java.lang.String") 1759 program.load_global("java") 1760 program.load_attr("lang") 1761 program.load_attr("String") 1762 program.load_const(const.get_value()) 1763 program.call_function(1) 1764 else: 1765 program.load_const(const.get_value()) 1766 1767 def ldc_w(self, arguments, program): 1768 const = self.class_file.constants[(arguments[0] << 8) + arguments[1] - 1] 1769 if isinstance(const, classfile.StringInfo): 1770 program.use_external_name("java.lang.String") 1771 program.load_global("java") 1772 program.load_attr("lang") 1773 program.load_attr("String") 1774 program.load_const(const.get_value()) 1775 program.call_function(1) 1776 else: 1777 program.load_const(const.get_value()) 1778 1779 ldc2_w = ldc_w 1780 ldiv = idiv 1781 lload = iload 1782 lload_0 = iload_0 1783 lload_1 = iload_1 1784 lload_2 = iload_2 1785 lload_3 = iload_3 1786 lmul = imul 1787 lneg = ineg 1788 1789 def lookupswitch(self, code, program): 1790 1791 # Find the offset to the next 4 byte boundary in the code. 1792 1793 d, r = divmod(self.java_position + 1, 4) 1794 to_boundary = (4 - r) % 4 1795 1796 # Get the pertinent arguments. 1797 1798 code = code[to_boundary:] 1799 default = classfile.u4(code[0:4]) 1800 npairs = classfile.u4(code[4:8]) 1801 1802 # Process the pairs. 1803 # NOTE: This is not the most optimal implementation. 1804 1805 pair_index = 8 1806 for pair in range(0, npairs): 1807 match = classfile.u4(code[pair_index:pair_index+4]) 1808 offset = classfile.s4(code[pair_index+4:pair_index+8]) 1809 # Calculate the branch target. 1810 java_absolute = self.java_position + offset 1811 # Generate branching code. 1812 program.dup_top() # Stack: key, key 1813 program.load_const(match) # Stack: key, key, match 1814 program.compare_op("==") # Stack: key, result 1815 program.jump_to_label(0, "end") 1816 program.pop_top() # Stack: key 1817 program.pop_top() # Stack: 1818 program.jump_absolute(self.position_mapping[java_absolute]) 1819 # Generate the label for the end of the branching code. 1820 program.start_label("end") 1821 program.pop_top() # Stack: key 1822 # Update the index. 1823 pair_index += 4 1824 1825 # Generate the default. 1826 1827 java_absolute = self.java_position + default 1828 program.jump_absolute(self.position_mapping[java_absolute]) 1829 return pair_index + to_boundary 1830 1831 lor = ior 1832 lrem = irem 1833 lreturn = ireturn 1834 lshl = ishl 1835 lshr = ishr 1836 lstore = istore 1837 lstore_0 = istore_0 1838 lstore_1 = istore_1 1839 lstore_2 = istore_2 1840 lstore_3 = istore_3 1841 lsub = isub 1842 lushr = iushr 1843 lxor = ixor 1844 1845 def monitorenter(self, arguments, program): 1846 # NOTE: To be implemented. 1847 pass 1848 1849 def monitorexit(self, arguments, program): 1850 # NOTE: To be implemented. 1851 pass 1852 1853 def multianewarray(self, arguments, program): 1854 index = (arguments[0] << 8) + arguments[1] 1855 dimensions = arguments[2] 1856 # Stack: count1, ..., countN-1, countN 1857 type_name = self.class_file.constants[index - 1].get_python_name() 1858 default_value = classfile.get_default_for_type(type_name) 1859 self._newarray(program, default_value) # Stack: count1, ..., countN-1, list 1860 for dimension in range(1, dimensions): 1861 program.rot_two() # Stack: count1, ..., list, countN-1 1862 program.build_list(0) # Stack: count1, ..., list, countN-1, new-list 1863 program.rot_three() # Stack: count1, ..., new-list, list, countN-1 1864 program.setup_loop() 1865 program.load_const(0) # Stack: count1, ..., new-list, list, countN-1, 0 1866 program.rot_two() # Stack: count1, ..., new-list, list, 0, countN-1 1867 program.load_global("range") # Stack: count1, ..., new-list, list, 0, countN-1, range 1868 program.rot_three() # Stack: count1, ..., new-list, list, range, 0, countN-1 1869 program.call_function(2) # Stack: count1, ..., new-list, list, range-list 1870 program.get_iter() # Stack: count1, ..., new-list, list, iter 1871 program.for_iter() # Stack: count1, ..., new-list, list, iter, value 1872 program.pop_top() # Stack: count1, ..., new-list, list, iter 1873 program.rot_three() # Stack: count1, ..., iter, new-list, list 1874 program.slice_0() # Stack: count1, ..., iter, new-list, list[:] 1875 program.dup_top() # Stack: count1, ..., iter, new-list, list[:], list[:] 1876 program.rot_three() # Stack: count1, ..., iter, list[:], new-list, list[:] 1877 program.rot_two() # Stack: count1, ..., iter, list[:], list[:], new-list 1878 program.dup_top() # Stack: count1, ..., iter, list[:], list[:], new-list, new-list 1879 program.load_attr("append") # Stack: count1, ..., iter, list[:], list[:], new-list, append 1880 program.rot_three() # Stack: count1, ..., iter, list[:], append, list[:], new-list 1881 program.rot_three() # Stack: count1, ..., iter, list[:], new-list, append, list[:] 1882 program.call_function(1) # Stack: count1, ..., iter, list[:], new-list, None 1883 program.pop_top() # Stack: count1, ..., iter, list[:], new-list 1884 program.rot_two() # Stack: count1, ..., iter, new-list, list[:] 1885 program.rot_three() # Stack: count1, ..., list[:], iter, new-list 1886 program.rot_three() # Stack: count1, ..., new-list, list[:], iter 1887 program.end_loop() # Stack: count1, ..., new-list, list[:], iter 1888 program.pop_top() # Stack: count1, ..., new-list 1889 1890 def new(self, arguments, program): 1891 # This operation is considered to be the same as the calling of the 1892 # initialisation method of the given class with no arguments. 1893 1894 index = (arguments[0] << 8) + arguments[1] 1895 target_name = self.class_file.constants[index - 1].get_python_name() 1896 program.use_external_name(target_name) 1897 1898 # NOTE: Using the string version of the name which may contain incompatible characters. 1899 program.load_global("object") 1900 program.load_attr("__new__") 1901 load_class_name(self.class_file, target_name, program) 1902 program.call_function(1) 1903 1904 def newarray(self, arguments, program): 1905 # NOTE: Does not raise NegativeArraySizeException. 1906 # NOTE: Not completely using the arguments to type the list/array. 1907 atype = arguments[0] 1908 default_value = get_default_for_atype(atype) 1909 self._newarray(program, default_value) 1910 1911 def nop(self, arguments, program): 1912 pass 1913 1914 def pop(self, arguments, program): 1915 program.pop_top() 1916 1917 pop2 = pop # ignoring Java stack value distinctions 1918 1919 def putfield(self, arguments, program): 1920 index = (arguments[0] << 8) + arguments[1] 1921 target_name = self.class_file.constants[index - 1].get_python_name() 1922 program.rot_two() 1923 # NOTE: Using the string version of the name which may contain incompatible characters. 1924 program.store_attr(str(target_name)) 1925 1926 def putstatic(self, arguments, program): 1927 index = (arguments[0] << 8) + arguments[1] 1928 target = self.class_file.constants[index - 1] 1929 target_name = target.get_python_name() 1930 1931 # Get the class name instead of the fully qualified name. 1932 1933 full_class_name = target.get_class().get_python_name() 1934 program.use_external_name(full_class_name) 1935 load_class_name(self.class_file, full_class_name, program) 1936 # NOTE: Using the string version of the name which may contain incompatible characters. 1937 program.store_attr(str(target_name)) 1938 1939 def ret(self, arguments, program): 1940 program.ret(arguments[0]) 1941 # Indicate that the finally handler is probably over. 1942 # NOTE: This is seemingly not guaranteed. 1943 self.in_finally = 0 1944 1945 def return_(self, arguments, program): 1946 program.load_const(None) 1947 program.return_value() 1948 1949 saload = laload 1950 sastore = lastore 1951 1952 def sipush(self, arguments, program): 1953 program.load_const(signed2((arguments[0] << 8) + arguments[1])) 1954 1955 def swap(self, arguments, program): 1956 program.rot_two() 1957 1958 def tableswitch(self, code, program): 1959 1960 # Find the offset to the next 4 byte boundary in the code. 1961 1962 d, r = divmod(self.java_position + 1, 4) 1963 to_boundary = (4 - r) % 4 1964 1965 # Get the pertinent arguments. 1966 1967 code = code[to_boundary:] 1968 default = classfile.u4(code[0:4]) 1969 low = classfile.u4(code[4:8]) 1970 high = classfile.u4(code[8:12]) 1971 1972 # Process the jump entries. 1973 # NOTE: This is not the most optimal implementation. 1974 1975 jump_index = 12 1976 for jump in range(low, high + 1): 1977 offset = classfile.s4(code[jump_index:jump_index + 4]) 1978 1979 # Calculate the branch target. 1980 1981 java_absolute = self.java_position + offset 1982 1983 # Generate branching code. 1984 1985 program.dup_top() # Stack: key, key 1986 program.load_const(jump) # Stack: key, key, jump 1987 program.compare_op("==") # Stack: key, result 1988 program.jump_to_label(0, "end") 1989 program.pop_top() # Stack: key 1990 program.pop_top() # Stack: 1991 program.jump_absolute(self.position_mapping[java_absolute]) 1992 1993 # Generate the label for the end of the branching code. 1994 1995 program.start_label("end") 1996 program.pop_top() # Stack: key 1997 1998 # Update the index. 1999 2000 jump_index += 4 2001 2002 # Generate the default. 2003 2004 java_absolute = self.java_position + default 2005 program.jump_absolute(self.position_mapping[java_absolute]) 2006 return jump_index + to_boundary 2007 2008 def wide(self, code, program): 2009 # NOTE: To be implemented. 2010 return number_of_arguments 2011 2012 def disassemble(class_file, method): 2013 disassembler = BytecodeDisassembler(class_file) 2014 disassembler.process(method, BytecodeDisassemblerProgram()) 2015 return disassembler 2016 2017 class ClassTranslator: 2018 2019 """ 2020 A class which provides a wrapper around a class file and the means to 2021 translate the represented class into a Python class. 2022 """ 2023 2024 def __init__(self, class_file): 2025 2026 "Initialise the object with the given 'class_file'." 2027 2028 self.class_file = class_file 2029 self.filename = "" 2030 2031 for attribute in self.class_file.attributes: 2032 if isinstance(attribute, classfile.SourceFileAttributeInfo): 2033 self.filename = str(attribute.get_name()) 2034 2035 def translate_method(self, method): 2036 2037 "Translate the given 'method' - an object obtained from the class file." 2038 2039 translator = BytecodeTranslator(self.class_file) 2040 writer = BytecodeWriter() 2041 translator.process(method, writer) 2042 return translator, writer 2043 2044 def make_method(self, real_method_name, methods, global_names, namespace): 2045 2046 """ 2047 Make a dispatcher method with the given 'real_method_name', providing 2048 dispatch to the supplied type-sensitive 'methods', accessing the given 2049 'global_names' where necessary, and storing the new method in the 2050 'namespace' provided. 2051 """ 2052 2053 if real_method_name == "<init>": 2054 method_name = "__init__" 2055 else: 2056 method_name = real_method_name 2057 2058 # Where only one method exists, just make an alias. 2059 2060 if len(methods) == 1: 2061 method, fn = methods[0] 2062 namespace[method_name] = fn 2063 return 2064 2065 # Write a simple bytecode dispatching mechanism. 2066 2067 program = BytecodeWriter() 2068 2069 # Remember whether any of the methods are static. 2070 # NOTE: This should be an all or nothing situation. 2071 2072 method_is_static = 0 2073 2074 # NOTE: The code below should use dictionary-based dispatch for better performance. 2075 2076 for method, fn in methods: 2077 method_is_static = real_method_name != "<init>" and method_is_static or \ 2078 classfile.has_flags(method.access_flags, [classfile.STATIC]) 2079 2080 if method_is_static: 2081 program.load_fast(0) # Stack: arguments 2082 else: 2083 program.load_fast(1) # Stack: arguments 2084 2085 program.setup_loop() 2086 program.load_const(1) # Stack: arguments, 1 2087 2088 if method_is_static: 2089 program.store_fast(1) # Stack: arguments (found = 1) 2090 else: 2091 program.store_fast(2) # Stack: arguments (found = 1) 2092 2093 # Emit a list of parameter types. 2094 2095 descriptor_types = method.get_descriptor()[0] 2096 for descriptor_type in descriptor_types: 2097 base_type, object_type, array_type = descriptor_type 2098 python_type = classfile.descriptor_base_type_mapping[base_type] 2099 if python_type == "instance": 2100 # NOTE: This will need extending. 2101 python_type = object_type 2102 program.load_global(python_type) # Stack: arguments, type, ... 2103 program.build_list(len(descriptor_types)) 2104 # Stack: arguments, types 2105 # Make a map of arguments and types. 2106 program.load_const(None) # Stack: arguments, types, None 2107 program.rot_three() # Stack: None, arguments, types 2108 program.build_tuple(3) # Stack: tuple 2109 program.load_global("map") # Stack: tuple, map 2110 program.rot_two() # Stack: map, tuple 2111 program.call_function_var(0) # Stack: list (mapping arguments to types) 2112 # Loop over each pair. 2113 program.get_iter() # Stack: iter 2114 program.for_iter() # Stack: iter, (argument, type) 2115 program.unpack_sequence(2) # Stack: iter, type, argument 2116 program.dup_top() # Stack: iter, type, argument, argument 2117 program.load_const(None) # Stack: iter, type, argument, argument, None 2118 program.compare_op("is") # Stack: iter, type, argument, result 2119 # Missing argument? 2120 program.jump_to_label(0, "present") 2121 program.pop_top() # Stack: iter, type, argument 2122 program.pop_top() # Stack: iter, type 2123 program.pop_top() # Stack: iter 2124 program.load_const(0) # Stack: iter, 0 2125 2126 if method_is_static: 2127 program.store_fast(1) # Stack: iter (found = 0) 2128 else: 2129 program.store_fast(2) # Stack: iter (found = 0) 2130 2131 program.break_loop() 2132 # Argument was present. 2133 program.start_label("present") 2134 program.pop_top() # Stack: iter, type, argument 2135 program.rot_two() # Stack: iter, argument, type 2136 program.dup_top() # Stack: iter, argument, type, type 2137 program.load_const(None) # Stack: iter, argument, type, type, None 2138 program.compare_op("is") # Stack: iter, argument, type, result 2139 # Missing parameter type? 2140 program.jump_to_label(0, "present") 2141 program.pop_top() # Stack: iter, argument, type 2142 program.pop_top() # Stack: iter, argument 2143 program.pop_top() # Stack: iter 2144 program.load_const(0) # Stack: iter, 0 2145 2146 if method_is_static: 2147 program.store_fast(1) # Stack: iter (found = 0) 2148 else: 2149 program.store_fast(2) # Stack: iter (found = 0) 2150 2151 program.break_loop() 2152 # Parameter was present. 2153 program.start_label("present") 2154 program.pop_top() # Stack: iter, argument, type 2155 program.build_tuple(2) # Stack: iter, (argument, type) 2156 program.load_global("isinstance") # Stack: iter, (argument, type), isinstance 2157 program.rot_two() # Stack: iter, isinstance, (argument, type) 2158 program.call_function_var(0) # Stack: iter, result 2159 program.jump_to_label(1, "match") 2160 program.pop_top() # Stack: iter 2161 program.load_const(0) # Stack: iter, 0 2162 2163 if method_is_static: 2164 program.store_fast(1) # Stack: iter (found = 0) 2165 else: 2166 program.store_fast(2) # Stack: iter (found = 0) 2167 2168 program.break_loop() 2169 # Argument type and parameter type matched. 2170 program.start_label("match") 2171 program.pop_top() # Stack: iter 2172 program.end_loop() # Stack: 2173 # If all the parameters matched, call the method. 2174 2175 if method_is_static: 2176 program.load_fast(1) # Stack: match 2177 else: 2178 program.load_fast(2) # Stack: match 2179 2180 program.jump_to_label(0, "failed") 2181 # All the parameters matched. 2182 program.pop_top() # Stack: 2183 2184 if method_is_static: 2185 program.load_fast(0) # Stack: arguments 2186 program.load_global(str(self.class_file.this_class.get_python_name())) 2187 # Stack: arguments, class 2188 else: 2189 program.load_fast(1) # Stack: arguments 2190 program.load_fast(0) # Stack: arguments, self 2191 2192 program.load_attr(str(method.get_python_name())) 2193 # Stack: arguments, method 2194 program.rot_two() # Stack: method, arguments 2195 program.call_function_var(0) # Stack: result 2196 program.return_value() 2197 # Try the next method if arguments or parameters were missing or incorrect. 2198 program.start_label("failed") 2199 program.pop_top() # Stack: 2200 2201 # Raise an exception if nothing matched. 2202 # NOTE: Improve this. 2203 2204 program.load_const("No matching method") 2205 program.raise_varargs(1) 2206 program.load_const(None) 2207 program.return_value() 2208 2209 # Add the code as a method in the namespace. 2210 # NOTE: One actual parameter, flags as 71 apparently means that a list 2211 # NOTE: parameter is used in a method. 2212 2213 if method_is_static: 2214 nargs = 0 2215 else: 2216 nargs = 1 2217 nlocals = program.max_locals + 1 2218 2219 code = new.code(nargs, nlocals, program.max_stack_depth, 71, program.get_output(), 2220 tuple(program.get_constants()), tuple(program.get_names()), tuple(self.make_varnames(nlocals, method_is_static)), 2221 self.filename, method_name, 0, "") 2222 fn = new.function(code, global_names) 2223 2224 if method_is_static: 2225 fn = staticmethod(fn) 2226 2227 namespace[method_name] = fn 2228 2229 def process(self, global_names): 2230 2231 """ 2232 Process the class, storing it in the 'global_names' dictionary provided. 2233 Return a tuple containing the class and a list of external names 2234 referenced by the class's methods. 2235 """ 2236 2237 namespace = {} 2238 2239 # Make the fields. 2240 2241 for field in self.class_file.fields: 2242 if classfile.has_flags(field.access_flags, [classfile.STATIC]): 2243 field_name = str(field.get_python_name()) 2244 namespace[field_name] = None 2245 2246 # Make the methods. 2247 2248 real_methods = {} 2249 external_names = [] 2250 2251 for method in self.class_file.methods: 2252 real_method_name = str(method.get_name()) 2253 method_name = str(method.get_python_name()) 2254 2255 translator, writer = self.translate_method(method) 2256 2257 # Add external names to the master list. 2258 2259 for external_name in writer.external_names: 2260 if external_name not in external_names: 2261 external_names.append(external_name) 2262 2263 # Fix up special class initialisation methods and static methods. 2264 2265 method_is_static = real_method_name != "<init>" and classfile.has_flags(method.access_flags, [classfile.STATIC]) 2266 if method_is_static: 2267 nargs = len(method.get_descriptor()[0]) 2268 else: 2269 nargs = len(method.get_descriptor()[0]) + 1 2270 nlocals = writer.max_locals + 1 2271 flags = 67 2272 2273 # NOTE: Add line number table later. 2274 2275 code = new.code(nargs, nlocals, writer.max_stack_depth, flags, writer.get_output(), 2276 tuple(writer.get_constants()), tuple(writer.get_names()), 2277 tuple(self.make_varnames(nlocals, method_is_static)), self.filename, method_name, 0, "") 2278 2279 # NOTE: May need more globals. 2280 2281 fn = new.function(code, global_names) 2282 2283 # Fix up special class initialisation methods and static methods. 2284 2285 if method_is_static: 2286 fn = staticmethod(fn) 2287 2288 # Remember the real method name and the corresponding methods produced. 2289 2290 if not real_methods.has_key(real_method_name): 2291 real_methods[real_method_name] = [] 2292 real_methods[real_method_name].append((method, fn)) 2293 2294 # Add the method to the class's namespace. 2295 2296 namespace[method_name] = fn 2297 2298 # Define superclasses. 2299 2300 bases = self.get_base_classes(global_names) 2301 2302 # Define method dispatchers. 2303 2304 for real_method_name, methods in real_methods.items(): 2305 if real_method_name != "<clinit>": 2306 self.make_method(real_method_name, methods, global_names, namespace) 2307 2308 # Use only the last part of the fully qualified name. 2309 2310 full_class_name = str(self.class_file.this_class.get_python_name()) 2311 class_name = full_class_name.split(".")[-1] 2312 cls = new.classobj(class_name, bases, namespace) 2313 global_names[cls.__name__] = cls 2314 2315 return cls, external_names 2316 2317 def get_base_classes(self, global_names): 2318 2319 """ 2320 Identify the superclass, then either load it from the given 2321 'global_names' if available, or import the class from its parent module. 2322 Return a tuple containing all base classes (typically a single element 2323 tuple). 2324 """ 2325 2326 original_name = str(self.class_file.super_class.get_name()) 2327 full_this_class_name = str(self.class_file.this_class.get_python_name()) 2328 this_class_name_parts = full_this_class_name.split(".") 2329 this_class_module_name = ".".join(this_class_name_parts[:-1]) 2330 full_super_class_name = str(self.class_file.super_class.get_python_name()) 2331 super_class_name_parts = full_super_class_name.split(".") 2332 super_class_name = super_class_name_parts[-1] 2333 super_class_module_name = ".".join(super_class_name_parts[:-1]) 2334 if super_class_module_name == "": 2335 obj = global_names[super_class_name] 2336 elif super_class_module_name == this_class_module_name: 2337 obj = global_names[super_class_name] 2338 else: 2339 #print "Importing", super_class_module_name, super_class_name 2340 obj = __import__(super_class_module_name, global_names, {}, []) 2341 for super_class_name_part in super_class_name_parts[1:] or [super_class_name]: 2342 #print "*", obj, super_class_name_part 2343 obj = getattr(obj, super_class_name_part) 2344 return (obj,) 2345 2346 def make_varnames(self, nlocals, method_is_static=0): 2347 2348 """ 2349 A utility method which invents variable names for the given number - 2350 'nlocals' - of local variables in a method. Returns a list of such 2351 variable names. 2352 2353 If the optional 'method_is_static' is set to true, do not use "self" as 2354 the first argument name. 2355 """ 2356 2357 if method_is_static: 2358 l = ["cls"] 2359 else: 2360 l = ["self"] 2361 for i in range(1, nlocals): 2362 l.append("_l%s" % i) 2363 return l[:nlocals] 2364 2365 # Test functions, useful for tracing generated bytecode operations. 2366 2367 def _map(*args): 2368 print args 2369 return apply(__builtins__.map, args) 2370 2371 def _isinstance(*args): 2372 print args 2373 return apply(__builtins__.isinstance, args) 2374 2375 if __name__ == "__main__": 2376 import sys 2377 import dis 2378 global_names = globals() 2379 #global_names["isinstance"] = _isinstance 2380 #global_names["map"] = _map 2381 for filename in sys.argv[1:]: 2382 f = open(filename, "rb") 2383 c = classfile.ClassFile(f.read()) 2384 translator = ClassTranslator(c) 2385 cls, external_names = translator.process(global_names) 2386 2387 # vim: tabstop=4 expandtab shiftwidth=4