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 749 750 if exception.start_pc == exception.handler_pc: 751 continue 752 753 # Index start positions. 754 755 if not exception_block_start.has_key(exception.start_pc): 756 exception_block_start[exception.start_pc] = [] 757 exception_block_start[exception.start_pc].append(exception) 758 759 # Index end positions. 760 761 if not exception_block_end.has_key(exception.end_pc): 762 exception_block_end[exception.end_pc] = [] 763 exception_block_end[exception.end_pc].append(exception) 764 765 # Index handler positions. 766 767 if not exception_block_handler.has_key(exception.handler_pc): 768 exception_block_handler[exception.handler_pc] = [] 769 exception_block_handler[exception.handler_pc].append(exception) 770 771 # Process each instruction in the code. 772 773 while self.java_position < len(code): 774 self.position_mapping[self.java_position] = program.position 775 776 # Insert exception handling constructs. 777 778 block_starts = exception_block_start.get(self.java_position, []) 779 for exception in block_starts: 780 781 # Note that the absolute position is used. 782 783 if exception.catch_type == 0: 784 program.setup_finally(self.position_mapping[exception.handler_pc]) 785 else: 786 program.setup_except(self.position_mapping[exception.handler_pc]) 787 788 if block_starts: 789 self.in_finally = 0 790 791 # Insert exception handler details. 792 # NOTE: Ensure that pop_block is reachable by possibly inserting it at the start of finally handlers. 793 # NOTE: Insert a check for the correct exception at the start of each handler. 794 795 for exception in exception_block_handler.get(self.java_position, []): 796 program.end_exception() 797 if exception.catch_type == 0: 798 self.in_finally = 1 799 else: 800 program.start_handler(self.class_file.constants[exception.catch_type - 1].get_python_name(), self.class_file) 801 802 # Process the bytecode at the current position. 803 804 bytecode = ord(code[self.java_position]) 805 mnemonic, number_of_arguments = self.java_bytecodes[bytecode] 806 number_of_arguments = self.process_bytecode(mnemonic, number_of_arguments, code, program) 807 next_java_position = self.java_position + 1 + number_of_arguments 808 809 # Insert exception block end details. 810 811 for exception in exception_block_end.get(next_java_position, []): 812 813 # NOTE: Insert jump beyond handlers. 814 # NOTE: program.jump_forward/absolute(...) 815 # NOTE: Insert end finally at end of handlers as well as where "ret" occurs. 816 817 if exception.catch_type != 0: 818 program.pop_block() 819 820 # Only advance the JVM position after sneaking in extra Python 821 # instructions. 822 823 self.java_position = next_java_position 824 825 # Tidy up exceptions. 826 827 program.end_exceptions() 828 829 def process_bytecode(self, mnemonic, number_of_arguments, code, program): 830 831 """ 832 Process a bytecode instruction with the given 'mnemonic' and 833 'number_of_arguments'. The 'code' parameter contains the full method 834 code so that argument data can be inspected. The 'program' parameter is 835 used to produce a Python translation of the instruction. 836 """ 837 838 if number_of_arguments is not None: 839 arguments = [] 840 for j in range(0, number_of_arguments): 841 arguments.append(ord(code[self.java_position + 1 + j])) 842 843 # Call the handler. 844 845 getattr(self, mnemonic)(arguments, program) 846 return number_of_arguments 847 else: 848 # Call the handler. 849 850 return getattr(self, mnemonic)(code[self.java_position+1:], program) 851 852 java_bytecodes = { 853 # code : (mnemonic, number of following bytes, change in stack) 854 0 : ("nop", 0), 855 1 : ("aconst_null", 0), 856 2 : ("iconst_m1", 0), 857 3 : ("iconst_0", 0), 858 4 : ("iconst_1", 0), 859 5 : ("iconst_2", 0), 860 6 : ("iconst_3", 0), 861 7 : ("iconst_4", 0), 862 8 : ("iconst_5", 0), 863 9 : ("lconst_0", 0), 864 10 : ("lconst_1", 0), 865 11 : ("fconst_0", 0), 866 12 : ("fconst_1", 0), 867 13 : ("fconst_2", 0), 868 14 : ("dconst_0", 0), 869 15 : ("dconst_1", 0), 870 16 : ("bipush", 1), 871 17 : ("sipush", 2), 872 18 : ("ldc", 1), 873 19 : ("ldc_w", 2), 874 20 : ("ldc2_w", 2), 875 21 : ("iload", 1), 876 22 : ("lload", 1), 877 23 : ("fload", 1), 878 24 : ("dload", 1), 879 25 : ("aload", 1), 880 26 : ("iload_0", 0), 881 27 : ("iload_1", 0), 882 28 : ("iload_2", 0), 883 29 : ("iload_3", 0), 884 30 : ("lload_0", 0), 885 31 : ("lload_1", 0), 886 32 : ("lload_2", 0), 887 33 : ("lload_3", 0), 888 34 : ("fload_0", 0), 889 35 : ("fload_1", 0), 890 36 : ("fload_2", 0), 891 37 : ("fload_3", 0), 892 38 : ("dload_0", 0), 893 39 : ("dload_1", 0), 894 40 : ("dload_2", 0), 895 41 : ("dload_3", 0), 896 42 : ("aload_0", 0), 897 43 : ("aload_1", 0), 898 44 : ("aload_2", 0), 899 45 : ("aload_3", 0), 900 46 : ("iaload", 0), 901 47 : ("laload", 0), 902 48 : ("faload", 0), 903 49 : ("daload", 0), 904 50 : ("aaload", 0), 905 51 : ("baload", 0), 906 52 : ("caload", 0), 907 53 : ("saload", 0), 908 54 : ("istore", 1), 909 55 : ("lstore", 1), 910 56 : ("fstore", 1), 911 57 : ("dstore", 1), 912 58 : ("astore", 1), 913 59 : ("istore_0", 0), 914 60 : ("istore_1", 0), 915 61 : ("istore_2", 0), 916 62 : ("istore_3", 0), 917 63 : ("lstore_0", 0), 918 64 : ("lstore_1", 0), 919 65 : ("lstore_2", 0), 920 66 : ("lstore_3", 0), 921 67 : ("fstore_0", 0), 922 68 : ("fstore_1", 0), 923 69 : ("fstore_2", 0), 924 70 : ("fstore_3", 0), 925 71 : ("dstore_0", 0), 926 72 : ("dstore_1", 0), 927 73 : ("dstore_2", 0), 928 74 : ("dstore_3", 0), 929 75 : ("astore_0", 0), 930 76 : ("astore_1", 0), 931 77 : ("astore_2", 0), 932 78 : ("astore_3", 0), 933 79 : ("iastore", 0), 934 80 : ("lastore", 0), 935 81 : ("fastore", 0), 936 82 : ("dastore", 0), 937 83 : ("aastore", 0), 938 84 : ("bastore", 0), 939 85 : ("castore", 0), 940 86 : ("sastore", 0), 941 87 : ("pop", 0), 942 88 : ("pop2", 0), 943 89 : ("dup", 0), 944 90 : ("dup_x1", 0), 945 91 : ("dup_x2", 0), 946 92 : ("dup2", 0), 947 93 : ("dup2_x1", 0), 948 94 : ("dup2_x2", 0), 949 95 : ("swap", 0), 950 96 : ("iadd", 0), 951 97 : ("ladd", 0), 952 98 : ("fadd", 0), 953 99 : ("dadd", 0), 954 100 : ("isub", 0), 955 101 : ("lsub", 0), 956 102 : ("fsub", 0), 957 103 : ("dsub", 0), 958 104 : ("imul", 0), 959 105 : ("lmul", 0), 960 106 : ("fmul", 0), 961 107 : ("dmul", 0), 962 108 : ("idiv", 0), 963 109 : ("ldiv", 0), 964 110 : ("fdiv", 0), 965 111 : ("ddiv", 0), 966 112 : ("irem", 0), 967 113 : ("lrem", 0), 968 114 : ("frem", 0), 969 115 : ("drem", 0), 970 116 : ("ineg", 0), 971 117 : ("lneg", 0), 972 118 : ("fneg", 0), 973 119 : ("dneg", 0), 974 120 : ("ishl", 0), 975 121 : ("lshl", 0), 976 122 : ("ishr", 0), 977 123 : ("lshr", 0), 978 124 : ("iushr", 0), 979 125 : ("lushr", 0), 980 126 : ("iand", 0), 981 127 : ("land", 0), 982 128 : ("ior", 0), 983 129 : ("lor", 0), 984 130 : ("ixor", 0), 985 131 : ("lxor", 0), 986 132 : ("iinc", 2), 987 133 : ("i2l", 0), 988 134 : ("i2f", 0), 989 135 : ("i2d", 0), 990 136 : ("l2i", 0), 991 137 : ("l2f", 0), 992 138 : ("l2d", 0), 993 139 : ("f2i", 0), 994 140 : ("f2l", 0), 995 141 : ("f2d", 0), 996 142 : ("d2i", 0), 997 143 : ("d2l", 0), 998 144 : ("d2f", 0), 999 145 : ("i2b", 0), 1000 146 : ("i2c", 0), 1001 147 : ("i2s", 0), 1002 148 : ("lcmp", 0), 1003 149 : ("fcmpl", 0), 1004 150 : ("fcmpg", 0), 1005 151 : ("dcmpl", 0), 1006 152 : ("dcmpg", 0), 1007 153 : ("ifeq", 2), 1008 154 : ("ifne", 2), 1009 155 : ("iflt", 2), 1010 156 : ("ifge", 2), 1011 157 : ("ifgt", 2), 1012 158 : ("ifle", 2), 1013 159 : ("if_icmpeq", 2), 1014 160 : ("if_icmpne", 2), 1015 161 : ("if_icmplt", 2), 1016 162 : ("if_icmpge", 2), 1017 163 : ("if_icmpgt", 2), 1018 164 : ("if_icmple", 2), 1019 165 : ("if_acmpeq", 2), 1020 166 : ("if_acmpne", 2), 1021 167 : ("goto", 2), 1022 168 : ("jsr", 2), 1023 169 : ("ret", 1), 1024 170 : ("tableswitch", None), # variable number of arguments 1025 171 : ("lookupswitch", None), # variable number of arguments 1026 172 : ("ireturn", 0), 1027 173 : ("lreturn", 0), 1028 174 : ("freturn", 0), 1029 175 : ("dreturn", 0), 1030 176 : ("areturn", 0), 1031 177 : ("return_", 0), 1032 178 : ("getstatic", 2), 1033 179 : ("putstatic", 2), 1034 180 : ("getfield", 2), 1035 181 : ("putfield", 2), 1036 182 : ("invokevirtual", 2), 1037 183 : ("invokespecial", 2), 1038 184 : ("invokestatic", 2), 1039 185 : ("invokeinterface", 4), 1040 187 : ("new", 2), 1041 188 : ("newarray", 1), 1042 189 : ("anewarray", 2), 1043 190 : ("arraylength", 0), 1044 191 : ("athrow", 0), 1045 192 : ("checkcast", 2), 1046 193 : ("instanceof", 2), 1047 194 : ("monitorenter", 0), 1048 195 : ("monitorexit", 0), 1049 196 : ("wide", None), # 3 or 5 arguments, stack changes according to modified element 1050 197 : ("multianewarray", 3), 1051 198 : ("ifnull", 2), 1052 199 : ("ifnonnull", 2), 1053 200 : ("goto_w", 4), 1054 201 : ("jsr_w", 4), 1055 } 1056 1057 class BytecodeDisassembler(BytecodeReader): 1058 1059 "A Java bytecode disassembler." 1060 1061 bytecode_methods = [spec[0] for spec in BytecodeReader.java_bytecodes.values()] 1062 1063 def __getattr__(self, name): 1064 if name in self.bytecode_methods: 1065 print "%5s %s" % (self.java_position, name), 1066 return self.generic 1067 else: 1068 raise AttributeError, name 1069 1070 def generic(self, arguments, program): 1071 print arguments 1072 1073 def lookupswitch(self, code, program): 1074 print "%5s lookupswitch" % (self.java_position,), 1075 d, r = divmod(self.java_position + 1, 4) 1076 to_boundary = (4 - r) % 4 1077 code = code[to_boundary:] 1078 default = classfile.u4(code[0:4]) 1079 npairs = classfile.u4(code[4:8]) 1080 print default, npairs 1081 return to_boundary + 8 + npairs * 8 1082 1083 def tableswitch(self, code, program): 1084 print "%5s tableswitch" % (self.java_position,), 1085 d, r = divmod(self.java_position + 1, 4) 1086 to_boundary = (4 - r) % 4 1087 code = code[to_boundary:] 1088 default = classfile.u4(code[0:4]) 1089 low = classfile.u4(code[4:8]) 1090 high = classfile.u4(code[8:12]) 1091 print default, low, high 1092 return to_boundary + 12 + (high - low + 1) * 4 1093 1094 class BytecodeDisassemblerProgram: 1095 position = 0 1096 def setup_except(self, target): 1097 print "(setup_except %s)" % target 1098 def setup_finally(self, target): 1099 print "(setup_finally %s)" % target 1100 def end_exception(self): 1101 print "(end_exception)" 1102 def start_handler(self, exc_name, class_file): 1103 print "(start_handler %s)" % exc_name 1104 def pop_block(self): 1105 print "(pop_block)" 1106 def load_const(self, const): 1107 print "(load_const %s)" % const 1108 def return_value(self): 1109 print "(return_value)" 1110 1111 class BytecodeTranslator(BytecodeReader): 1112 1113 "A Java bytecode translator which uses a Python bytecode writer." 1114 1115 def aaload(self, arguments, program): 1116 # NOTE: No type checking performed. 1117 program.binary_subscr() 1118 1119 def aastore(self, arguments, program): 1120 # NOTE: No type checking performed. 1121 # Stack: arrayref, index, value 1122 program.rot_three() # Stack: value, arrayref, index 1123 program.store_subscr() 1124 1125 def aconst_null(self, arguments, program): 1126 program.load_const(None) 1127 1128 def aload(self, arguments, program): 1129 program.load_fast(arguments[0]) 1130 1131 def aload_0(self, arguments, program): 1132 program.load_fast(0) 1133 1134 def aload_1(self, arguments, program): 1135 program.load_fast(1) 1136 1137 def aload_2(self, arguments, program): 1138 program.load_fast(2) 1139 1140 def aload_3(self, arguments, program): 1141 program.load_fast(3) 1142 1143 def anewarray(self, arguments, program): 1144 # NOTE: Does not raise NegativeArraySizeException. 1145 # NOTE: Not using the index to type the list/array. 1146 index = (arguments[0] << 8) + arguments[1] 1147 type_name = self.class_file.constants[index - 1].get_python_name() 1148 default_value = classfile.get_default_for_type(type_name) 1149 self._newarray(program, type_name) 1150 1151 def _newarray(self, program, default_value): 1152 program.build_list(0) # Stack: count, list 1153 program.rot_two() # Stack: list, count 1154 program.setup_loop() 1155 program.load_global("range") 1156 program.load_const(0) # Stack: list, count, range, 0 1157 program.rot_three() # Stack: list, 0, count, range 1158 program.rot_three() # Stack: list, range, 0, count 1159 program.call_function(2) # Stack: list, range_list 1160 program.get_iter() # Stack: list, iter 1161 program.for_iter() # Stack: list, iter, value 1162 program.pop_top() # Stack: list, iter 1163 program.rot_two() # Stack: iter, list 1164 program.dup_top() # Stack: iter, list, list 1165 program.load_attr("append") # Stack: iter, list, append 1166 program.load_const(default_value) # Stack: iter, list, append, default 1167 program.call_function(1) # Stack: iter, list, default 1168 program.pop_top() # Stack: iter, list 1169 program.rot_two() # Stack: list, iter 1170 program.end_loop() # Back to for_iter above 1171 1172 def areturn(self, arguments, program): 1173 program.return_value() 1174 1175 def arraylength(self, arguments, program): 1176 program.load_global("len") # Stack: arrayref, len 1177 program.rot_two() # Stack: len, arrayref 1178 program.call_function(1) 1179 1180 def astore(self, arguments, program): 1181 program.store_fast(arguments[0]) 1182 1183 def astore_0(self, arguments, program): 1184 program.store_fast(0) 1185 1186 def astore_1(self, arguments, program): 1187 program.store_fast(1) 1188 1189 def astore_2(self, arguments, program): 1190 program.store_fast(2) 1191 1192 def astore_3(self, arguments, program): 1193 program.store_fast(3) 1194 1195 def athrow(self, arguments, program): 1196 # NOTE: NullPointerException not raised where null/None is found on the stack. 1197 # If this instruction appears in a finally handler, use end_finally instead. 1198 if self.in_finally: 1199 program.end_finally() 1200 else: 1201 # Wrap the exception in a Python exception. 1202 program.load_global("Exception") # Stack: objectref, Exception 1203 program.rot_two() # Stack: Exception, objectref 1204 program.call_function(1) # Stack: exception 1205 program.raise_varargs(1) 1206 # NOTE: This seems to put another object on the stack. 1207 1208 baload = aaload 1209 bastore = aastore 1210 1211 def bipush(self, arguments, program): 1212 program.load_const(signed1(arguments[0])) 1213 1214 caload = aaload 1215 castore = aastore 1216 1217 def checkcast(self, arguments, program): 1218 index = (arguments[0] << 8) + arguments[1] 1219 target_name = self.class_file.constants[index - 1].get_python_name() 1220 program.use_external_name(target_name) 1221 program.dup_top() # Stack: objectref, objectref 1222 program.load_const(None) # Stack: objectref, objectref, None 1223 program.compare_op("is") # Stack: objectref, result 1224 program.jump_to_label(1, "next") 1225 program.pop_top() # Stack: objectref 1226 program.dup_top() # Stack: objectref, objectref 1227 program.load_global("isinstance") # Stack: objectref, objectref, isinstance 1228 program.rot_two() # Stack: objectref, isinstance, objectref 1229 load_class_name(self.class_file, target_name, program) 1230 program.call_function(2) # Stack: objectref, result 1231 program.jump_to_label(1, "next") 1232 program.pop_top() # Stack: objectref 1233 program.pop_top() # Stack: 1234 program.use_external_name("java.lang.ClassCastException") 1235 load_class_name(self.class_file, "java.lang.ClassCastException", program) 1236 program.call_function(0) # Stack: exception 1237 # Wrap the exception in a Python exception. 1238 program.load_global("Exception") # Stack: exception, Exception 1239 program.rot_two() # Stack: Exception, exception 1240 program.call_function(1) # Stack: exception 1241 program.raise_varargs(1) 1242 # NOTE: This seems to put another object on the stack. 1243 program.start_label("next") 1244 program.pop_top() # Stack: objectref 1245 1246 def d2f(self, arguments, program): 1247 pass 1248 1249 def d2i(self, arguments, program): 1250 program.load_global("int") # Stack: value, int 1251 program.rot_two() # Stack: int, value 1252 program.call_function(1) # Stack: result 1253 1254 d2l = d2i # Preserving Java semantics 1255 1256 def dadd(self, arguments, program): 1257 # NOTE: No type checking performed. 1258 program.binary_add() 1259 1260 daload = aaload 1261 dastore = aastore 1262 1263 def dcmpg(self, arguments, program): 1264 # NOTE: No type checking performed. 1265 program.compare_op(">") 1266 1267 def dcmpl(self, arguments, program): 1268 # NOTE: No type checking performed. 1269 program.compare_op("<") 1270 1271 def dconst_0(self, arguments, program): 1272 program.load_const(0.0) 1273 1274 def dconst_1(self, arguments, program): 1275 program.load_const(1.0) 1276 1277 def ddiv(self, arguments, program): 1278 # NOTE: No type checking performed. 1279 program.binary_divide() 1280 1281 dload = aload 1282 dload_0 = aload_0 1283 dload_1 = aload_1 1284 dload_2 = aload_2 1285 dload_3 = aload_3 1286 1287 def dmul(self, arguments, program): 1288 # NOTE: No type checking performed. 1289 program.binary_multiply() 1290 1291 def dneg(self, arguments, program): 1292 # NOTE: No type checking performed. 1293 program.unary_negative() 1294 1295 def drem(self, arguments, program): 1296 # NOTE: No type checking performed. 1297 program.binary_modulo() 1298 1299 dreturn = areturn 1300 dstore = astore 1301 dstore_0 = astore_0 1302 dstore_1 = astore_1 1303 dstore_2 = astore_2 1304 dstore_3 = astore_3 1305 1306 def dsub(self, arguments, program): 1307 # NOTE: No type checking performed. 1308 program.binary_subtract() 1309 1310 def dup(self, arguments, program): 1311 program.dup_top() 1312 1313 def dup_x1(self, arguments, program): 1314 # Ignoring computational type categories. 1315 program.dup_top() 1316 program.rot_three() 1317 1318 def dup_x2(self, arguments, program): 1319 # Ignoring computational type categories. 1320 program.dup_top() 1321 program.rot_four() 1322 1323 dup2 = dup # Ignoring computational type categories 1324 dup2_x1 = dup_x1 # Ignoring computational type categories 1325 dup2_x2 = dup_x2 # Ignoring computational type categories 1326 1327 def f2d(self, arguments, program): 1328 pass # Preserving Java semantics 1329 1330 def f2i(self, arguments, program): 1331 program.load_global("int") # Stack: value, int 1332 program.rot_two() # Stack: int, value 1333 program.call_function(1) # Stack: result 1334 1335 f2l = f2i # Preserving Java semantics 1336 fadd = dadd 1337 faload = daload 1338 fastore = dastore 1339 fcmpg = dcmpg 1340 fcmpl = dcmpl 1341 fconst_0 = dconst_0 1342 fconst_1 = dconst_1 1343 1344 def fconst_2(self, arguments, program): 1345 program.load_const(2.0) 1346 1347 fdiv = ddiv 1348 fload = dload 1349 fload_0 = dload_0 1350 fload_1 = dload_1 1351 fload_2 = dload_2 1352 fload_3 = dload_3 1353 fmul = dmul 1354 fneg = dneg 1355 frem = drem 1356 freturn = dreturn 1357 fstore = dstore 1358 fstore_0 = dstore_0 1359 fstore_1 = dstore_1 1360 fstore_2 = dstore_2 1361 fstore_3 = dstore_3 1362 fsub = dsub 1363 1364 def getfield(self, arguments, program): 1365 index = (arguments[0] << 8) + arguments[1] 1366 target_name = self.class_file.constants[index - 1].get_python_name() 1367 # NOTE: Using the string version of the name which may contain incompatible characters. 1368 program.load_attr(str(target_name)) 1369 1370 def getstatic(self, arguments, program): 1371 index = (arguments[0] << 8) + arguments[1] 1372 target = self.class_file.constants[index - 1] 1373 target_name = target.get_python_name() 1374 1375 # Get the class name instead of the fully qualified name. 1376 1377 full_class_name = target.get_class().get_python_name() 1378 program.use_external_name(full_class_name) 1379 load_class_name(self.class_file, full_class_name, program) 1380 # NOTE: Using the string version of the name which may contain incompatible characters. 1381 program.load_attr(str(target_name)) 1382 1383 def goto(self, arguments, program): 1384 offset = signed2((arguments[0] << 8) + arguments[1]) 1385 java_absolute = self.java_position + offset 1386 program.jump_absolute(self.position_mapping[java_absolute]) 1387 1388 def goto_w(self, arguments, program): 1389 offset = signed4((arguments[0] << 24) + (arguments[1] << 16) + (arguments[2] << 8) + arguments[3]) 1390 java_absolute = self.java_position + offset 1391 program.jump_absolute(self.position_mapping[java_absolute]) 1392 1393 def i2b(self, arguments, program): 1394 pass 1395 1396 def i2c(self, arguments, program): 1397 pass 1398 1399 def i2d(self, arguments, program): 1400 program.load_global("float") # Stack: value, float 1401 program.rot_two() # Stack: float, value 1402 program.call_function(1) # Stack: result 1403 1404 i2f = i2d # Not distinguishing between float and double 1405 1406 def i2l(self, arguments, program): 1407 pass # Preserving Java semantics 1408 1409 def i2s(self, arguments, program): 1410 pass # Not distinguishing between int and short 1411 1412 iadd = fadd 1413 iaload = faload 1414 1415 def iand(self, arguments, program): 1416 # NOTE: No type checking performed. 1417 program.binary_and() 1418 1419 iastore = fastore 1420 1421 def iconst_m1(self, arguments, program): 1422 program.load_const(-1) 1423 1424 def iconst_0(self, arguments, program): 1425 program.load_const(0) 1426 1427 def iconst_1(self, arguments, program): 1428 program.load_const(1) 1429 1430 def iconst_2(self, arguments, program): 1431 program.load_const(2) 1432 1433 def iconst_3(self, arguments, program): 1434 program.load_const(3) 1435 1436 def iconst_4(self, arguments, program): 1437 program.load_const(4) 1438 1439 def iconst_5(self, arguments, program): 1440 program.load_const(5) 1441 1442 idiv = fdiv 1443 1444 def _if_xcmpx(self, arguments, program, op): 1445 offset = signed2((arguments[0] << 8) + arguments[1]) 1446 java_absolute = self.java_position + offset 1447 program.compare_op(op) 1448 program.jump_to_label(0, "next") # skip if false 1449 program.pop_top() 1450 program.jump_absolute(self.position_mapping[java_absolute]) 1451 program.start_label("next") 1452 program.pop_top() 1453 1454 def if_acmpeq(self, arguments, program): 1455 # NOTE: No type checking performed. 1456 self._if_xcmpx(arguments, program, "is") 1457 1458 def if_acmpne(self, arguments, program): 1459 # NOTE: No type checking performed. 1460 self._if_xcmpx(arguments, program, "is not") 1461 1462 def if_icmpeq(self, arguments, program): 1463 # NOTE: No type checking performed. 1464 self._if_xcmpx(arguments, program, "==") 1465 1466 def if_icmpne(self, arguments, program): 1467 # NOTE: No type checking performed. 1468 self._if_xcmpx(arguments, program, "!=") 1469 1470 def if_icmplt(self, arguments, program): 1471 # NOTE: No type checking performed. 1472 self._if_xcmpx(arguments, program, "<") 1473 1474 def if_icmpge(self, arguments, program): 1475 # NOTE: No type checking performed. 1476 self._if_xcmpx(arguments, program, ">=") 1477 1478 def if_icmpgt(self, arguments, program): 1479 # NOTE: No type checking performed. 1480 self._if_xcmpx(arguments, program, ">") 1481 1482 def if_icmple(self, arguments, program): 1483 # NOTE: No type checking performed. 1484 self._if_xcmpx(arguments, program, "<=") 1485 1486 def ifeq(self, arguments, program): 1487 # NOTE: No type checking performed. 1488 program.load_const(0) 1489 self._if_xcmpx(arguments, program, "==") 1490 1491 def ifne(self, arguments, program): 1492 # NOTE: No type checking performed. 1493 program.load_const(0) 1494 self._if_xcmpx(arguments, program, "!=") 1495 1496 def iflt(self, arguments, program): 1497 # NOTE: No type checking performed. 1498 program.load_const(0) 1499 self._if_xcmpx(arguments, program, "<") 1500 1501 def ifge(self, arguments, program): 1502 # NOTE: No type checking performed. 1503 program.load_const(0) 1504 self._if_xcmpx(arguments, program, ">=") 1505 1506 def ifgt(self, arguments, program): 1507 # NOTE: No type checking performed. 1508 program.load_const(0) 1509 self._if_xcmpx(arguments, program, ">") 1510 1511 def ifle(self, arguments, program): 1512 # NOTE: No type checking performed. 1513 program.load_const(0) 1514 self._if_xcmpx(arguments, program, "<=") 1515 1516 def ifnonnull(self, arguments, program): 1517 # NOTE: No type checking performed. 1518 program.load_const(None) 1519 self._if_xcmpx(arguments, program, "is not") 1520 1521 def ifnull(self, arguments, program): 1522 # NOTE: No type checking performed. 1523 program.load_const(None) 1524 self._if_xcmpx(arguments, program, "is") 1525 1526 def iinc(self, arguments, program): 1527 # NOTE: No type checking performed. 1528 program.load_fast(arguments[0]) 1529 program.load_const(arguments[1]) 1530 program.binary_add() 1531 program.store_fast(arguments[0]) 1532 1533 iload = fload 1534 iload_0 = fload_0 1535 iload_1 = fload_1 1536 iload_2 = fload_2 1537 iload_3 = fload_3 1538 imul = fmul 1539 ineg = fneg 1540 1541 def instanceof(self, arguments, program): 1542 index = (arguments[0] << 8) + arguments[1] 1543 target_name = self.class_file.constants[index - 1].get_python_name() 1544 program.use_external_name(target_name) 1545 program.load_global("isinstance") # Stack: objectref, isinstance 1546 program.rot_two() # Stack: isinstance, objectref 1547 load_class_name(self.class_file, target_name, program) 1548 program.call_function(2) # Stack: result 1549 1550 def _invoke(self, target_name, program): 1551 # NOTE: Using the string version of the name which may contain incompatible characters. 1552 program.load_attr(str(target_name)) # Stack: tuple, method 1553 program.rot_two() # Stack: method, tuple 1554 program.call_function_var(0) # Stack: result 1555 1556 def invokeinterface(self, arguments, program): 1557 # NOTE: This implementation does not perform the necessary checks for 1558 # NOTE: signature-based polymorphism. 1559 # NOTE: Java rules not specifically obeyed. 1560 index = (arguments[0] << 8) + arguments[1] 1561 # NOTE: "count" == nargs + 1, apparently. 1562 count = arguments[2] - 1 1563 target_name = self.class_file.constants[index - 1].get_python_name() 1564 # Stack: objectref, arg1, arg2, ... 1565 program.build_tuple(count) # Stack: objectref, tuple 1566 program.rot_two() # Stack: tuple, objectref 1567 # NOTE: The interface information is not used to discover the correct 1568 # NOTE: method. 1569 self._invoke(target_name, program) 1570 1571 def invokespecial(self, arguments, program): 1572 # NOTE: This implementation does not perform the necessary checks for 1573 # NOTE: signature-based polymorphism. 1574 # NOTE: Java rules not specifically obeyed. 1575 index = (arguments[0] << 8) + arguments[1] 1576 target = self.class_file.constants[index - 1] 1577 original_name = target.get_name() 1578 target_name = target.get_python_name() 1579 1580 # Get the number of parameters from the descriptor. 1581 1582 count = len(target.get_descriptor()[0]) 1583 1584 # First, we build a tuple of the reference and arguments. 1585 1586 program.build_tuple(count + 1) # Stack: tuple 1587 1588 # Get the class name instead of the fully qualified name. 1589 # NOTE: Not bothering with Object initialisation. 1590 1591 full_class_name = target.get_class().get_python_name() 1592 if full_class_name not in ("java.lang.Object", "java.lang.Exception"): 1593 program.use_external_name(full_class_name) 1594 load_class_name(self.class_file, full_class_name, program) 1595 self._invoke(target_name, program) 1596 1597 # Remove Python None return value. 1598 1599 if str(original_name) == "<init>": 1600 program.pop_top() 1601 1602 def invokestatic(self, arguments, program): 1603 # NOTE: This implementation does not perform the necessary checks for 1604 # NOTE: signature-based polymorphism. 1605 # NOTE: Java rules not specifically obeyed. 1606 index = (arguments[0] << 8) + arguments[1] 1607 target = self.class_file.constants[index - 1] 1608 target_name = target.get_python_name() 1609 1610 # Get the number of parameters from the descriptor. 1611 1612 count = len(target.get_descriptor()[0]) 1613 1614 # Stack: arg1, arg2, ... 1615 1616 program.build_tuple(count) # Stack: tuple 1617 1618 # Use the class to provide access to static methods. 1619 # Get the class name instead of the fully qualified name. 1620 1621 full_class_name = target.get_class().get_python_name() 1622 if full_class_name not in ("java.lang.Object", "java.lang.Exception"): 1623 program.use_external_name(full_class_name) 1624 load_class_name(self.class_file, full_class_name, program) 1625 self._invoke(target_name, program) 1626 1627 def invokevirtual (self, arguments, program): 1628 # NOTE: This implementation does not perform the necessary checks for 1629 # NOTE: signature-based polymorphism. 1630 # NOTE: Java rules not specifically obeyed. 1631 index = (arguments[0] << 8) + arguments[1] 1632 target = self.class_file.constants[index - 1] 1633 target_name = target.get_python_name() 1634 # Get the number of parameters from the descriptor. 1635 count = len(target.get_descriptor()[0]) 1636 # Stack: objectref, arg1, arg2, ... 1637 program.build_tuple(count) # Stack: objectref, tuple 1638 program.rot_two() # Stack: tuple, objectref 1639 self._invoke(target_name, program) 1640 1641 def ior(self, arguments, program): 1642 # NOTE: No type checking performed. 1643 program.binary_or() 1644 1645 irem = frem 1646 ireturn = freturn 1647 1648 def ishl(self, arguments, program): 1649 # NOTE: No type checking performed. 1650 # NOTE: Not verified. 1651 program.binary_lshift() 1652 1653 def ishr(self, arguments, program): 1654 # NOTE: No type checking performed. 1655 # NOTE: Not verified. 1656 program.binary_rshift() 1657 1658 istore = fstore 1659 istore_0 = fstore_0 1660 istore_1 = fstore_1 1661 istore_2 = fstore_2 1662 istore_3 = fstore_3 1663 isub = fsub 1664 iushr = ishr # Ignoring distinctions between arithmetic and logical shifts 1665 1666 def ixor(self, arguments, program): 1667 # NOTE: No type checking performed. 1668 program.binary_xor() 1669 1670 def jsr(self, arguments, program): 1671 offset = signed2((arguments[0] << 8) + arguments[1]) 1672 java_absolute = self.java_position + offset 1673 # Store the address of the next instruction. 1674 program.load_const_ret(self.position_mapping[self.java_position + 3]) 1675 program.jump_absolute(self.position_mapping[java_absolute]) 1676 1677 def jsr_w(self, arguments, program): 1678 offset = signed4((arguments[0] << 24) + (arguments[1] << 16) + (arguments[2] << 8) + arguments[3]) 1679 java_absolute = self.java_position + offset 1680 # Store the address of the next instruction. 1681 program.load_const_ret(self.position_mapping[self.java_position + 5]) 1682 program.jump_absolute(self.position_mapping[java_absolute]) 1683 1684 l2d = i2d 1685 l2f = i2f 1686 1687 def l2i(self, arguments, program): 1688 pass # Preserving Java semantics 1689 1690 ladd = iadd 1691 laload = iaload 1692 land = iand 1693 lastore = iastore 1694 1695 def lcmp(self, arguments, program): 1696 # NOTE: No type checking performed. 1697 program.dup_topx(2) # Stack: value1, value2, value1, value2 1698 program.compare_op(">") # Stack: value1, value2, result 1699 program.jump_to_label(0, "equals") 1700 # True - produce result and branch. 1701 program.pop_top() # Stack: value1, value2 1702 program.pop_top() # Stack: value1 1703 program.pop_top() # Stack: 1704 program.load_const(1) # Stack: 1 1705 program.jump_to_label(None, "next") 1706 # False - test equality. 1707 program.start_label("equals") 1708 program.pop_top() # Stack: value1, value2 1709 program.dup_topx(2) # Stack: value1, value2, value1, value2 1710 program.compare_op("==") # Stack: value1, value2, result 1711 program.jump_to_label(0, "less") 1712 # True - produce result and branch. 1713 program.pop_top() # Stack: value1, value2 1714 program.pop_top() # Stack: value1 1715 program.pop_top() # Stack: 1716 program.load_const(0) # Stack: 0 1717 program.jump_to_label(None, "next") 1718 # False - produce result. 1719 program.start_label("less") 1720 program.pop_top() # Stack: value1, value2 1721 program.pop_top() # Stack: value1 1722 program.pop_top() # Stack: 1723 program.load_const(-1) # Stack: -1 1724 program.start_label("next") 1725 1726 lconst_0 = iconst_0 1727 lconst_1 = iconst_1 1728 1729 def ldc(self, arguments, program): 1730 const = self.class_file.constants[arguments[0] - 1] 1731 if isinstance(const, classfile.StringInfo): 1732 program.use_external_name("java.lang.String") 1733 program.load_global("java") 1734 program.load_attr("lang") 1735 program.load_attr("String") 1736 program.load_const(const.get_value()) 1737 program.call_function(1) 1738 else: 1739 program.load_const(const.get_value()) 1740 1741 def ldc_w(self, arguments, program): 1742 const = self.class_file.constants[(arguments[0] << 8) + arguments[1] - 1] 1743 if isinstance(const, classfile.StringInfo): 1744 program.use_external_name("java.lang.String") 1745 program.load_global("java") 1746 program.load_attr("lang") 1747 program.load_attr("String") 1748 program.load_const(const.get_value()) 1749 program.call_function(1) 1750 else: 1751 program.load_const(const.get_value()) 1752 1753 ldc2_w = ldc_w 1754 ldiv = idiv 1755 lload = iload 1756 lload_0 = iload_0 1757 lload_1 = iload_1 1758 lload_2 = iload_2 1759 lload_3 = iload_3 1760 lmul = imul 1761 lneg = ineg 1762 1763 def lookupswitch(self, code, program): 1764 1765 # Find the offset to the next 4 byte boundary in the code. 1766 1767 d, r = divmod(self.java_position + 1, 4) 1768 to_boundary = (4 - r) % 4 1769 1770 # Get the pertinent arguments. 1771 1772 code = code[to_boundary:] 1773 default = classfile.u4(code[0:4]) 1774 npairs = classfile.u4(code[4:8]) 1775 1776 # Process the pairs. 1777 # NOTE: This is not the most optimal implementation. 1778 1779 pair_index = 8 1780 for pair in range(0, npairs): 1781 match = classfile.u4(code[pair_index:pair_index+4]) 1782 offset = classfile.s4(code[pair_index+4:pair_index+8]) 1783 # Calculate the branch target. 1784 java_absolute = self.java_position + offset 1785 # Generate branching code. 1786 program.dup_top() # Stack: key, key 1787 program.load_const(match) # Stack: key, key, match 1788 program.compare_op("==") # Stack: key, result 1789 program.jump_to_label(0, "end") 1790 program.pop_top() # Stack: key 1791 program.pop_top() # Stack: 1792 program.jump_absolute(self.position_mapping[java_absolute]) 1793 # Generate the label for the end of the branching code. 1794 program.start_label("end") 1795 program.pop_top() # Stack: key 1796 # Update the index. 1797 pair_index += 4 1798 1799 # Generate the default. 1800 1801 java_absolute = self.java_position + default 1802 program.jump_absolute(self.position_mapping[java_absolute]) 1803 return pair_index + to_boundary 1804 1805 lor = ior 1806 lrem = irem 1807 lreturn = ireturn 1808 lshl = ishl 1809 lshr = ishr 1810 lstore = istore 1811 lstore_0 = istore_0 1812 lstore_1 = istore_1 1813 lstore_2 = istore_2 1814 lstore_3 = istore_3 1815 lsub = isub 1816 lushr = iushr 1817 lxor = ixor 1818 1819 def monitorenter(self, arguments, program): 1820 # NOTE: To be implemented. 1821 pass 1822 1823 def monitorexit(self, arguments, program): 1824 # NOTE: To be implemented. 1825 pass 1826 1827 def multianewarray(self, arguments, program): 1828 index = (arguments[0] << 8) + arguments[1] 1829 dimensions = arguments[2] 1830 # Stack: count1, ..., countN-1, countN 1831 type_name = self.class_file.constants[index - 1].get_python_name() 1832 default_value = classfile.get_default_for_type(type_name) 1833 self._newarray(program, default_value) # Stack: count1, ..., countN-1, list 1834 for dimension in range(1, dimensions): 1835 program.rot_two() # Stack: count1, ..., list, countN-1 1836 program.build_list(0) # Stack: count1, ..., list, countN-1, new-list 1837 program.rot_three() # Stack: count1, ..., new-list, list, countN-1 1838 program.setup_loop() 1839 program.load_const(0) # Stack: count1, ..., new-list, list, countN-1, 0 1840 program.rot_two() # Stack: count1, ..., new-list, list, 0, countN-1 1841 program.load_global("range") # Stack: count1, ..., new-list, list, 0, countN-1, range 1842 program.rot_three() # Stack: count1, ..., new-list, list, range, 0, countN-1 1843 program.call_function(2) # Stack: count1, ..., new-list, list, range-list 1844 program.get_iter() # Stack: count1, ..., new-list, list, iter 1845 program.for_iter() # Stack: count1, ..., new-list, list, iter, value 1846 program.pop_top() # Stack: count1, ..., new-list, list, iter 1847 program.rot_three() # Stack: count1, ..., iter, new-list, list 1848 program.slice_0() # Stack: count1, ..., iter, new-list, list[:] 1849 program.dup_top() # Stack: count1, ..., iter, new-list, list[:], list[:] 1850 program.rot_three() # Stack: count1, ..., iter, list[:], new-list, list[:] 1851 program.rot_two() # Stack: count1, ..., iter, list[:], list[:], new-list 1852 program.dup_top() # Stack: count1, ..., iter, list[:], list[:], new-list, new-list 1853 program.load_attr("append") # Stack: count1, ..., iter, list[:], list[:], new-list, append 1854 program.rot_three() # Stack: count1, ..., iter, list[:], append, list[:], new-list 1855 program.rot_three() # Stack: count1, ..., iter, list[:], new-list, append, list[:] 1856 program.call_function(1) # Stack: count1, ..., iter, list[:], new-list, None 1857 program.pop_top() # Stack: count1, ..., iter, list[:], new-list 1858 program.rot_two() # Stack: count1, ..., iter, new-list, list[:] 1859 program.rot_three() # Stack: count1, ..., list[:], iter, new-list 1860 program.rot_three() # Stack: count1, ..., new-list, list[:], iter 1861 program.end_loop() # Stack: count1, ..., new-list, list[:], iter 1862 program.pop_top() # Stack: count1, ..., new-list 1863 1864 def new(self, arguments, program): 1865 # This operation is considered to be the same as the calling of the 1866 # initialisation method of the given class with no arguments. 1867 1868 index = (arguments[0] << 8) + arguments[1] 1869 target_name = self.class_file.constants[index - 1].get_python_name() 1870 program.use_external_name(target_name) 1871 1872 # NOTE: Using the string version of the name which may contain incompatible characters. 1873 program.load_global("object") 1874 program.load_attr("__new__") 1875 load_class_name(self.class_file, target_name, program) 1876 program.call_function(1) 1877 1878 def newarray(self, arguments, program): 1879 # NOTE: Does not raise NegativeArraySizeException. 1880 # NOTE: Not completely using the arguments to type the list/array. 1881 atype = arguments[0] 1882 default_value = get_default_for_atype(atype) 1883 self._newarray(program, default_value) 1884 1885 def nop(self, arguments, program): 1886 pass 1887 1888 def pop(self, arguments, program): 1889 program.pop_top() 1890 1891 pop2 = pop # ignoring Java stack value distinctions 1892 1893 def putfield(self, arguments, program): 1894 index = (arguments[0] << 8) + arguments[1] 1895 target_name = self.class_file.constants[index - 1].get_python_name() 1896 program.rot_two() 1897 # NOTE: Using the string version of the name which may contain incompatible characters. 1898 program.store_attr(str(target_name)) 1899 1900 def putstatic(self, arguments, program): 1901 index = (arguments[0] << 8) + arguments[1] 1902 target = self.class_file.constants[index - 1] 1903 target_name = target.get_python_name() 1904 1905 # Get the class name instead of the fully qualified name. 1906 1907 full_class_name = target.get_class().get_python_name() 1908 program.use_external_name(full_class_name) 1909 load_class_name(self.class_file, full_class_name, program) 1910 # NOTE: Using the string version of the name which may contain incompatible characters. 1911 program.store_attr(str(target_name)) 1912 1913 def ret(self, arguments, program): 1914 program.ret(arguments[0]) 1915 # Indicate that the finally handler is probably over. 1916 # NOTE: This is seemingly not guaranteed. 1917 self.in_finally = 0 1918 1919 def return_(self, arguments, program): 1920 program.load_const(None) 1921 program.return_value() 1922 1923 saload = laload 1924 sastore = lastore 1925 1926 def sipush(self, arguments, program): 1927 program.load_const(signed2((arguments[0] << 8) + arguments[1])) 1928 1929 def swap(self, arguments, program): 1930 program.rot_two() 1931 1932 def tableswitch(self, code, program): 1933 1934 # Find the offset to the next 4 byte boundary in the code. 1935 1936 d, r = divmod(self.java_position + 1, 4) 1937 to_boundary = (4 - r) % 4 1938 1939 # Get the pertinent arguments. 1940 1941 code = code[to_boundary:] 1942 default = classfile.u4(code[0:4]) 1943 low = classfile.u4(code[4:8]) 1944 high = classfile.u4(code[8:12]) 1945 1946 # Process the jump entries. 1947 # NOTE: This is not the most optimal implementation. 1948 1949 jump_index = 12 1950 for jump in range(low, high + 1): 1951 offset = classfile.s4(code[jump_index:jump_index + 4]) 1952 1953 # Calculate the branch target. 1954 1955 java_absolute = self.java_position + offset 1956 1957 # Generate branching code. 1958 1959 program.dup_top() # Stack: key, key 1960 program.load_const(jump) # Stack: key, key, jump 1961 program.compare_op("==") # Stack: key, result 1962 program.jump_to_label(0, "end") 1963 program.pop_top() # Stack: key 1964 program.pop_top() # Stack: 1965 program.jump_absolute(self.position_mapping[java_absolute]) 1966 1967 # Generate the label for the end of the branching code. 1968 1969 program.start_label("end") 1970 program.pop_top() # Stack: key 1971 1972 # Update the index. 1973 1974 jump_index += 4 1975 1976 # Generate the default. 1977 1978 java_absolute = self.java_position + default 1979 program.jump_absolute(self.position_mapping[java_absolute]) 1980 return jump_index + to_boundary 1981 1982 def wide(self, code, program): 1983 # NOTE: To be implemented. 1984 return number_of_arguments 1985 1986 def disassemble(class_file, method): 1987 disassembler = BytecodeDisassembler(class_file) 1988 disassembler.process(method, BytecodeDisassemblerProgram()) 1989 return disassembler 1990 1991 class ClassTranslator: 1992 1993 """ 1994 A class which provides a wrapper around a class file and the means to 1995 translate the represented class into a Python class. 1996 """ 1997 1998 def __init__(self, class_file): 1999 2000 "Initialise the object with the given 'class_file'." 2001 2002 self.class_file = class_file 2003 self.filename = "" 2004 2005 for attribute in self.class_file.attributes: 2006 if isinstance(attribute, classfile.SourceFileAttributeInfo): 2007 self.filename = str(attribute.get_name()) 2008 2009 def translate_method(self, method): 2010 2011 "Translate the given 'method' - an object obtained from the class file." 2012 2013 translator = BytecodeTranslator(self.class_file) 2014 writer = BytecodeWriter() 2015 translator.process(method, writer) 2016 return translator, writer 2017 2018 def make_method(self, real_method_name, methods, global_names, namespace): 2019 2020 """ 2021 Make a dispatcher method with the given 'real_method_name', providing 2022 dispatch to the supplied type-sensitive 'methods', accessing the given 2023 'global_names' where necessary, and storing the new method in the 2024 'namespace' provided. 2025 """ 2026 2027 if real_method_name == "<init>": 2028 method_name = "__init__" 2029 else: 2030 method_name = real_method_name 2031 2032 # Where only one method exists, just make an alias. 2033 2034 if len(methods) == 1: 2035 method, fn = methods[0] 2036 namespace[method_name] = fn 2037 return 2038 2039 # Write a simple bytecode dispatching mechanism. 2040 2041 program = BytecodeWriter() 2042 2043 # Remember whether any of the methods are static. 2044 # NOTE: This should be an all or nothing situation. 2045 2046 method_is_static = 0 2047 2048 # NOTE: The code below should use dictionary-based dispatch for better performance. 2049 2050 for method, fn in methods: 2051 method_is_static = real_method_name != "<init>" and method_is_static or \ 2052 classfile.has_flags(method.access_flags, [classfile.STATIC]) 2053 2054 if method_is_static: 2055 program.load_fast(0) # Stack: arguments 2056 else: 2057 program.load_fast(1) # Stack: arguments 2058 2059 program.setup_loop() 2060 program.load_const(1) # Stack: arguments, 1 2061 2062 if method_is_static: 2063 program.store_fast(1) # Stack: arguments (found = 1) 2064 else: 2065 program.store_fast(2) # Stack: arguments (found = 1) 2066 2067 # Emit a list of parameter types. 2068 2069 descriptor_types = method.get_descriptor()[0] 2070 for descriptor_type in descriptor_types: 2071 base_type, object_type, array_type = descriptor_type 2072 python_type = classfile.descriptor_base_type_mapping[base_type] 2073 if python_type == "instance": 2074 # NOTE: This will need extending. 2075 python_type = object_type 2076 program.load_global(python_type) # Stack: arguments, type, ... 2077 program.build_list(len(descriptor_types)) 2078 # Stack: arguments, types 2079 # Make a map of arguments and types. 2080 program.load_const(None) # Stack: arguments, types, None 2081 program.rot_three() # Stack: None, arguments, types 2082 program.build_tuple(3) # Stack: tuple 2083 program.load_global("map") # Stack: tuple, map 2084 program.rot_two() # Stack: map, tuple 2085 program.call_function_var(0) # Stack: list (mapping arguments to types) 2086 # Loop over each pair. 2087 program.get_iter() # Stack: iter 2088 program.for_iter() # Stack: iter, (argument, type) 2089 program.unpack_sequence(2) # Stack: iter, type, argument 2090 program.dup_top() # Stack: iter, type, argument, argument 2091 program.load_const(None) # Stack: iter, type, argument, argument, None 2092 program.compare_op("is") # Stack: iter, type, argument, result 2093 # Missing argument? 2094 program.jump_to_label(0, "present") 2095 program.pop_top() # Stack: iter, type, argument 2096 program.pop_top() # Stack: iter, type 2097 program.pop_top() # Stack: iter 2098 program.load_const(0) # Stack: iter, 0 2099 2100 if method_is_static: 2101 program.store_fast(1) # Stack: iter (found = 0) 2102 else: 2103 program.store_fast(2) # Stack: iter (found = 0) 2104 2105 program.break_loop() 2106 # Argument was present. 2107 program.start_label("present") 2108 program.pop_top() # Stack: iter, type, argument 2109 program.rot_two() # Stack: iter, argument, type 2110 program.dup_top() # Stack: iter, argument, type, type 2111 program.load_const(None) # Stack: iter, argument, type, type, None 2112 program.compare_op("is") # Stack: iter, argument, type, result 2113 # Missing parameter type? 2114 program.jump_to_label(0, "present") 2115 program.pop_top() # Stack: iter, argument, type 2116 program.pop_top() # Stack: iter, argument 2117 program.pop_top() # Stack: iter 2118 program.load_const(0) # Stack: iter, 0 2119 2120 if method_is_static: 2121 program.store_fast(1) # Stack: iter (found = 0) 2122 else: 2123 program.store_fast(2) # Stack: iter (found = 0) 2124 2125 program.break_loop() 2126 # Parameter was present. 2127 program.start_label("present") 2128 program.pop_top() # Stack: iter, argument, type 2129 program.build_tuple(2) # Stack: iter, (argument, type) 2130 program.load_global("isinstance") # Stack: iter, (argument, type), isinstance 2131 program.rot_two() # Stack: iter, isinstance, (argument, type) 2132 program.call_function_var(0) # Stack: iter, result 2133 program.jump_to_label(1, "match") 2134 program.pop_top() # Stack: iter 2135 program.load_const(0) # Stack: iter, 0 2136 2137 if method_is_static: 2138 program.store_fast(1) # Stack: iter (found = 0) 2139 else: 2140 program.store_fast(2) # Stack: iter (found = 0) 2141 2142 program.break_loop() 2143 # Argument type and parameter type matched. 2144 program.start_label("match") 2145 program.pop_top() # Stack: iter 2146 program.end_loop() # Stack: 2147 # If all the parameters matched, call the method. 2148 2149 if method_is_static: 2150 program.load_fast(1) # Stack: match 2151 else: 2152 program.load_fast(2) # Stack: match 2153 2154 program.jump_to_label(0, "failed") 2155 # All the parameters matched. 2156 program.pop_top() # Stack: 2157 2158 if method_is_static: 2159 program.load_fast(0) # Stack: arguments 2160 program.load_global(str(self.class_file.this_class.get_python_name())) 2161 # Stack: arguments, class 2162 else: 2163 program.load_fast(1) # Stack: arguments 2164 program.load_fast(0) # Stack: arguments, self 2165 2166 program.load_attr(str(method.get_python_name())) 2167 # Stack: arguments, method 2168 program.rot_two() # Stack: method, arguments 2169 program.call_function_var(0) # Stack: result 2170 program.return_value() 2171 # Try the next method if arguments or parameters were missing or incorrect. 2172 program.start_label("failed") 2173 program.pop_top() # Stack: 2174 2175 # Raise an exception if nothing matched. 2176 # NOTE: Improve this. 2177 2178 program.load_const("No matching method") 2179 program.raise_varargs(1) 2180 program.load_const(None) 2181 program.return_value() 2182 2183 # Add the code as a method in the namespace. 2184 # NOTE: One actual parameter, flags as 71 apparently means that a list 2185 # NOTE: parameter is used in a method. 2186 2187 if method_is_static: 2188 nargs = 0 2189 else: 2190 nargs = 1 2191 nlocals = program.max_locals + 1 2192 2193 code = new.code(nargs, nlocals, program.max_stack_depth, 71, program.get_output(), 2194 tuple(program.get_constants()), tuple(program.get_names()), tuple(self.make_varnames(nlocals, method_is_static)), 2195 self.filename, method_name, 0, "") 2196 fn = new.function(code, global_names) 2197 2198 if method_is_static: 2199 fn = staticmethod(fn) 2200 2201 namespace[method_name] = fn 2202 2203 def process(self, global_names): 2204 2205 """ 2206 Process the class, storing it in the 'global_names' dictionary provided. 2207 Return a tuple containing the class and a list of external names 2208 referenced by the class's methods. 2209 """ 2210 2211 namespace = {} 2212 2213 # Make the fields. 2214 2215 for field in self.class_file.fields: 2216 if classfile.has_flags(field.access_flags, [classfile.STATIC]): 2217 field_name = str(field.get_python_name()) 2218 namespace[field_name] = None 2219 2220 # Make the methods. 2221 2222 real_methods = {} 2223 external_names = [] 2224 2225 for method in self.class_file.methods: 2226 real_method_name = str(method.get_name()) 2227 method_name = str(method.get_python_name()) 2228 2229 translator, writer = self.translate_method(method) 2230 2231 # Add external names to the master list. 2232 2233 for external_name in writer.external_names: 2234 if external_name not in external_names: 2235 external_names.append(external_name) 2236 2237 # Fix up special class initialisation methods and static methods. 2238 2239 method_is_static = real_method_name != "<init>" and classfile.has_flags(method.access_flags, [classfile.STATIC]) 2240 if method_is_static: 2241 nargs = len(method.get_descriptor()[0]) 2242 else: 2243 nargs = len(method.get_descriptor()[0]) + 1 2244 nlocals = writer.max_locals + 1 2245 flags = 67 2246 2247 # NOTE: Add line number table later. 2248 2249 code = new.code(nargs, nlocals, writer.max_stack_depth, flags, writer.get_output(), 2250 tuple(writer.get_constants()), tuple(writer.get_names()), 2251 tuple(self.make_varnames(nlocals, method_is_static)), self.filename, method_name, 0, "") 2252 2253 # NOTE: May need more globals. 2254 2255 fn = new.function(code, global_names) 2256 2257 # Fix up special class initialisation methods and static methods. 2258 2259 if method_is_static: 2260 fn = staticmethod(fn) 2261 2262 # Remember the real method name and the corresponding methods produced. 2263 2264 if not real_methods.has_key(real_method_name): 2265 real_methods[real_method_name] = [] 2266 real_methods[real_method_name].append((method, fn)) 2267 2268 # Add the method to the class's namespace. 2269 2270 namespace[method_name] = fn 2271 2272 # Define superclasses. 2273 2274 bases = self.get_base_classes(global_names) 2275 2276 # Define method dispatchers. 2277 2278 for real_method_name, methods in real_methods.items(): 2279 if real_method_name != "<clinit>": 2280 self.make_method(real_method_name, methods, global_names, namespace) 2281 2282 # Use only the last part of the fully qualified name. 2283 2284 full_class_name = str(self.class_file.this_class.get_python_name()) 2285 class_name = full_class_name.split(".")[-1] 2286 cls = new.classobj(class_name, bases, namespace) 2287 global_names[cls.__name__] = cls 2288 2289 return cls, external_names 2290 2291 def get_base_classes(self, global_names): 2292 2293 """ 2294 Identify the superclass, then either load it from the given 2295 'global_names' if available, or import the class from its parent module. 2296 Return a tuple containing all base classes (typically a single element 2297 tuple). 2298 """ 2299 2300 original_name = str(self.class_file.super_class.get_name()) 2301 full_this_class_name = str(self.class_file.this_class.get_python_name()) 2302 this_class_name_parts = full_this_class_name.split(".") 2303 this_class_module_name = ".".join(this_class_name_parts[:-1]) 2304 full_super_class_name = str(self.class_file.super_class.get_python_name()) 2305 super_class_name_parts = full_super_class_name.split(".") 2306 super_class_name = super_class_name_parts[-1] 2307 super_class_module_name = ".".join(super_class_name_parts[:-1]) 2308 if super_class_module_name == "": 2309 obj = global_names[super_class_name] 2310 elif super_class_module_name == this_class_module_name: 2311 obj = global_names[super_class_name] 2312 else: 2313 #print "Importing", super_class_module_name, super_class_name 2314 obj = __import__(super_class_module_name, global_names, {}, []) 2315 for super_class_name_part in super_class_name_parts[1:] or [super_class_name]: 2316 #print "*", obj, super_class_name_part 2317 obj = getattr(obj, super_class_name_part) 2318 return (obj,) 2319 2320 def make_varnames(self, nlocals, method_is_static=0): 2321 2322 """ 2323 A utility method which invents variable names for the given number - 2324 'nlocals' - of local variables in a method. Returns a list of such 2325 variable names. 2326 2327 If the optional 'method_is_static' is set to true, do not use "self" as 2328 the first argument name. 2329 """ 2330 2331 if method_is_static: 2332 l = ["cls"] 2333 else: 2334 l = ["self"] 2335 for i in range(1, nlocals): 2336 l.append("_l%s" % i) 2337 return l[:nlocals] 2338 2339 # Test functions, useful for tracing generated bytecode operations. 2340 2341 def _map(*args): 2342 print args 2343 return apply(__builtins__.map, args) 2344 2345 def _isinstance(*args): 2346 print args 2347 return apply(__builtins__.isinstance, args) 2348 2349 if __name__ == "__main__": 2350 import sys 2351 import dis 2352 global_names = globals() 2353 #global_names["isinstance"] = _isinstance 2354 #global_names["map"] = _map 2355 for filename in sys.argv[1:]: 2356 f = open(filename, "rb") 2357 c = classfile.ClassFile(f.read()) 2358 translator = ClassTranslator(c) 2359 cls, external_names = translator.process(global_names) 2360 2361 # vim: tabstop=4 expandtab shiftwidth=4