Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 1 | # -*- coding: utf-8 -*- |
| 2 | """ |
| 3 | codegen |
| 4 | ~~~~~~~ |
| 5 | |
| 6 | Extension to ast that allow ast -> python code generation. |
| 7 | |
| 8 | :copyright: Copyright 2008 by Armin Ronacher. |
| 9 | :license: BSD. |
| 10 | """ |
| 11 | from ast import * |
| 12 | |
| 13 | BOOLOP_SYMBOLS = { |
| 14 | And: 'and', |
| 15 | Or: 'or' |
| 16 | } |
| 17 | |
| 18 | BINOP_SYMBOLS = { |
| 19 | Add: '+', |
| 20 | Sub: '-', |
| 21 | Mult: '*', |
| 22 | Div: '/', |
| 23 | FloorDiv: '//', |
| 24 | Mod: '%', |
| 25 | LShift: '<<', |
| 26 | RShift: '>>', |
| 27 | BitOr: '|', |
| 28 | BitAnd: '&', |
| 29 | BitXor: '^' |
| 30 | } |
| 31 | |
| 32 | CMPOP_SYMBOLS = { |
| 33 | Eq: '==', |
| 34 | Gt: '>', |
| 35 | GtE: '>=', |
| 36 | In: 'in', |
| 37 | Is: 'is', |
| 38 | IsNot: 'is not', |
| 39 | Lt: '<', |
| 40 | LtE: '<=', |
| 41 | NotEq: '!=', |
| 42 | NotIn: 'not in' |
| 43 | } |
| 44 | |
| 45 | UNARYOP_SYMBOLS = { |
| 46 | Invert: '~', |
| 47 | Not: 'not', |
| 48 | UAdd: '+', |
| 49 | USub: '-' |
| 50 | } |
| 51 | |
| 52 | ALL_SYMBOLS = {} |
| 53 | ALL_SYMBOLS.update(BOOLOP_SYMBOLS) |
| 54 | ALL_SYMBOLS.update(BINOP_SYMBOLS) |
| 55 | ALL_SYMBOLS.update(CMPOP_SYMBOLS) |
| 56 | ALL_SYMBOLS.update(UNARYOP_SYMBOLS) |
| 57 | |
| 58 | def to_source(node, indent_with=' ' * 4, add_line_information=False): |
| 59 | """This function can convert a node tree back into python sourcecode. |
| 60 | This is useful for debugging purposes, especially if you're dealing with |
| 61 | custom asts not generated by python itself. |
| 62 | |
| 63 | It could be that the sourcecode is evaluable when the AST itself is not |
| 64 | compilable / evaluable. The reason for this is that the AST contains some |
| 65 | more data than regular sourcecode does, which is dropped during |
| 66 | conversion. |
| 67 | |
| 68 | Each level of indentation is replaced with `indent_with`. Per default this |
| 69 | parameter is equal to four spaces as suggested by PEP 8, but it might be |
| 70 | adjusted to match the application's styleguide. |
| 71 | |
| 72 | If `add_line_information` is set to `True` comments for the line numbers |
| 73 | of the nodes are added to the output. This can be used to spot wrong line |
| 74 | number information of statement nodes. |
| 75 | """ |
| 76 | generator = SourceGenerator(indent_with, add_line_information) |
| 77 | generator.visit(node) |
| 78 | return ''.join(generator.result) |
| 79 | |
| 80 | |
| 81 | class SourceGenerator(NodeVisitor): |
| 82 | """This visitor is able to transform a well formed syntax tree into python |
| 83 | sourcecode. For more details have a look at the docstring of the |
| 84 | `node_to_source` function. |
| 85 | """ |
| 86 | |
| 87 | def __init__(self, indent_with, add_line_information=False): |
| 88 | self.result = [] |
| 89 | self.indent_with = indent_with |
| 90 | self.add_line_information = add_line_information |
| 91 | self.indentation = 0 |
| 92 | self.new_lines = 0 |
| 93 | |
| 94 | def write(self, x): |
| 95 | if self.new_lines: |
| 96 | if self.result: |
| 97 | self.result.append('\n' * self.new_lines) |
| 98 | self.result.append(self.indent_with * self.indentation) |
| 99 | self.new_lines = 0 |
| 100 | self.result.append(x) |
| 101 | |
| 102 | def newline(self, node=None, extra=0): |
| 103 | self.new_lines = max(self.new_lines, 1 + extra) |
| 104 | if node is not None and self.add_line_information: |
| 105 | self.write('# line: %s' % node.lineno) |
| 106 | self.new_lines = 1 |
| 107 | |
| 108 | def body(self, statements): |
| 109 | self.new_line = True |
| 110 | self.indentation += 1 |
| 111 | for stmt in statements: |
| 112 | self.visit(stmt) |
| 113 | self.indentation -= 1 |
| 114 | |
| 115 | def body_or_else(self, node): |
| 116 | self.body(node.body) |
| 117 | if node.orelse: |
| 118 | self.newline() |
| 119 | self.write('else:') |
| 120 | self.body(node.orelse) |
| 121 | |
| 122 | def signature(self, node): |
| 123 | want_comma = [] |
| 124 | def write_comma(): |
| 125 | if want_comma: |
| 126 | self.write(', ') |
| 127 | else: |
| 128 | want_comma.append(True) |
| 129 | |
| 130 | padding = [None] * (len(node.args) - len(node.defaults)) |
| 131 | for arg, default in zip(node.args, padding + node.defaults): |
| 132 | write_comma() |
| 133 | self.visit(arg) |
| 134 | if default is not None: |
| 135 | self.write('=') |
| 136 | self.visit(default) |
| 137 | if node.vararg is not None: |
| 138 | write_comma() |
| 139 | self.write('*' + node.vararg) |
| 140 | if node.kwarg is not None: |
| 141 | write_comma() |
| 142 | self.write('**' + node.kwarg) |
| 143 | |
| 144 | def decorators(self, node): |
| 145 | for decorator in node.decorator_list: |
| 146 | self.newline(decorator) |
| 147 | self.write('@') |
| 148 | self.visit(decorator) |
| 149 | |
| 150 | # Statements |
| 151 | |
| 152 | def visit_Assign(self, node): |
| 153 | self.newline(node) |
| 154 | for idx, target in enumerate(node.targets): |
| 155 | if idx: |
| 156 | self.write(', ') |
| 157 | self.visit(target) |
| 158 | self.write(' = ') |
| 159 | self.visit(node.value) |
| 160 | |
| 161 | def visit_AugAssign(self, node): |
| 162 | self.newline(node) |
| 163 | self.visit(node.target) |
| 164 | self.write(BINOP_SYMBOLS[type(node.op)] + '=') |
| 165 | self.visit(node.value) |
| 166 | |
| 167 | def visit_ImportFrom(self, node): |
| 168 | self.newline(node) |
| 169 | self.write('from %s%s import ' % ('.' * node.level, node.module)) |
| 170 | for idx, item in enumerate(node.names): |
| 171 | if idx: |
| 172 | self.write(', ') |
| 173 | self.write(item) |
| 174 | |
| 175 | def visit_Import(self, node): |
| 176 | self.newline(node) |
| 177 | for item in node.names: |
| 178 | self.write('import ') |
| 179 | self.visit(item) |
| 180 | |
| 181 | def visit_Expr(self, node): |
| 182 | self.newline(node) |
| 183 | self.generic_visit(node) |
| 184 | |
| 185 | def visit_FunctionDef(self, node): |
| 186 | self.newline(extra=1) |
| 187 | self.decorators(node) |
| 188 | self.newline(node) |
| 189 | self.write('def %s(' % node.name) |
| 190 | self.signature(node.args) |
| 191 | self.write('):') |
| 192 | self.body(node.body) |
| 193 | |
| 194 | def visit_ClassDef(self, node): |
| 195 | have_args = [] |
| 196 | def paren_or_comma(): |
| 197 | if have_args: |
| 198 | self.write(', ') |
| 199 | else: |
| 200 | have_args.append(True) |
| 201 | self.write('(') |
| 202 | |
| 203 | self.newline(extra=2) |
| 204 | self.decorators(node) |
| 205 | self.newline(node) |
| 206 | self.write('class %s' % node.name) |
| 207 | for base in node.bases: |
| 208 | paren_or_comma() |
| 209 | self.visit(base) |
| 210 | # XXX: the if here is used to keep this module compatible |
| 211 | # with python 2.6. |
| 212 | if hasattr(node, 'keywords'): |
| 213 | for keyword in node.keywords: |
| 214 | paren_or_comma() |
| 215 | self.write(keyword.arg + '=') |
| 216 | self.visit(keyword.value) |
Patrick Williams | c0f7c04 | 2017-02-23 20:41:17 -0600 | [diff] [blame] | 217 | if hasattr(node, 'starargs') and node.starargs is not None: |
Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 218 | paren_or_comma() |
| 219 | self.write('*') |
| 220 | self.visit(node.starargs) |
Patrick Williams | c0f7c04 | 2017-02-23 20:41:17 -0600 | [diff] [blame] | 221 | if hasattr(node, 'kwargs') and node.kwargs is not None: |
Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 222 | paren_or_comma() |
| 223 | self.write('**') |
| 224 | self.visit(node.kwargs) |
| 225 | self.write(have_args and '):' or ':') |
| 226 | self.body(node.body) |
| 227 | |
| 228 | def visit_If(self, node): |
| 229 | self.newline(node) |
| 230 | self.write('if ') |
| 231 | self.visit(node.test) |
| 232 | self.write(':') |
| 233 | self.body(node.body) |
| 234 | while True: |
| 235 | else_ = node.orelse |
| 236 | if len(else_) == 1 and isinstance(else_[0], If): |
| 237 | node = else_[0] |
| 238 | self.newline() |
| 239 | self.write('elif ') |
| 240 | self.visit(node.test) |
| 241 | self.write(':') |
| 242 | self.body(node.body) |
| 243 | else: |
| 244 | self.newline() |
| 245 | self.write('else:') |
| 246 | self.body(else_) |
| 247 | break |
| 248 | |
| 249 | def visit_For(self, node): |
| 250 | self.newline(node) |
| 251 | self.write('for ') |
| 252 | self.visit(node.target) |
| 253 | self.write(' in ') |
| 254 | self.visit(node.iter) |
| 255 | self.write(':') |
| 256 | self.body_or_else(node) |
| 257 | |
| 258 | def visit_While(self, node): |
| 259 | self.newline(node) |
| 260 | self.write('while ') |
| 261 | self.visit(node.test) |
| 262 | self.write(':') |
| 263 | self.body_or_else(node) |
| 264 | |
| 265 | def visit_With(self, node): |
| 266 | self.newline(node) |
| 267 | self.write('with ') |
| 268 | self.visit(node.context_expr) |
| 269 | if node.optional_vars is not None: |
| 270 | self.write(' as ') |
| 271 | self.visit(node.optional_vars) |
| 272 | self.write(':') |
| 273 | self.body(node.body) |
| 274 | |
| 275 | def visit_Pass(self, node): |
| 276 | self.newline(node) |
| 277 | self.write('pass') |
| 278 | |
| 279 | def visit_Print(self, node): |
| 280 | # XXX: python 2.6 only |
| 281 | self.newline(node) |
| 282 | self.write('print ') |
| 283 | want_comma = False |
| 284 | if node.dest is not None: |
| 285 | self.write(' >> ') |
| 286 | self.visit(node.dest) |
| 287 | want_comma = True |
| 288 | for value in node.values: |
| 289 | if want_comma: |
| 290 | self.write(', ') |
| 291 | self.visit(value) |
| 292 | want_comma = True |
| 293 | if not node.nl: |
| 294 | self.write(',') |
| 295 | |
| 296 | def visit_Delete(self, node): |
| 297 | self.newline(node) |
| 298 | self.write('del ') |
| 299 | for idx, target in enumerate(node): |
| 300 | if idx: |
| 301 | self.write(', ') |
| 302 | self.visit(target) |
| 303 | |
| 304 | def visit_TryExcept(self, node): |
| 305 | self.newline(node) |
| 306 | self.write('try:') |
| 307 | self.body(node.body) |
| 308 | for handler in node.handlers: |
| 309 | self.visit(handler) |
| 310 | |
| 311 | def visit_TryFinally(self, node): |
| 312 | self.newline(node) |
| 313 | self.write('try:') |
| 314 | self.body(node.body) |
| 315 | self.newline(node) |
| 316 | self.write('finally:') |
| 317 | self.body(node.finalbody) |
| 318 | |
| 319 | def visit_Global(self, node): |
| 320 | self.newline(node) |
| 321 | self.write('global ' + ', '.join(node.names)) |
| 322 | |
| 323 | def visit_Nonlocal(self, node): |
| 324 | self.newline(node) |
| 325 | self.write('nonlocal ' + ', '.join(node.names)) |
| 326 | |
| 327 | def visit_Return(self, node): |
| 328 | self.newline(node) |
| 329 | self.write('return ') |
| 330 | self.visit(node.value) |
| 331 | |
| 332 | def visit_Break(self, node): |
| 333 | self.newline(node) |
| 334 | self.write('break') |
| 335 | |
| 336 | def visit_Continue(self, node): |
| 337 | self.newline(node) |
| 338 | self.write('continue') |
| 339 | |
| 340 | def visit_Raise(self, node): |
| 341 | # XXX: Python 2.6 / 3.0 compatibility |
| 342 | self.newline(node) |
| 343 | self.write('raise') |
| 344 | if hasattr(node, 'exc') and node.exc is not None: |
| 345 | self.write(' ') |
| 346 | self.visit(node.exc) |
| 347 | if node.cause is not None: |
| 348 | self.write(' from ') |
| 349 | self.visit(node.cause) |
| 350 | elif hasattr(node, 'type') and node.type is not None: |
| 351 | self.visit(node.type) |
| 352 | if node.inst is not None: |
| 353 | self.write(', ') |
| 354 | self.visit(node.inst) |
| 355 | if node.tback is not None: |
| 356 | self.write(', ') |
| 357 | self.visit(node.tback) |
| 358 | |
| 359 | # Expressions |
| 360 | |
| 361 | def visit_Attribute(self, node): |
| 362 | self.visit(node.value) |
| 363 | self.write('.' + node.attr) |
| 364 | |
| 365 | def visit_Call(self, node): |
| 366 | want_comma = [] |
| 367 | def write_comma(): |
| 368 | if want_comma: |
| 369 | self.write(', ') |
| 370 | else: |
| 371 | want_comma.append(True) |
| 372 | |
| 373 | self.visit(node.func) |
| 374 | self.write('(') |
| 375 | for arg in node.args: |
| 376 | write_comma() |
| 377 | self.visit(arg) |
| 378 | for keyword in node.keywords: |
| 379 | write_comma() |
| 380 | self.write(keyword.arg + '=') |
| 381 | self.visit(keyword.value) |
Patrick Williams | c0f7c04 | 2017-02-23 20:41:17 -0600 | [diff] [blame] | 382 | if hasattr(node, 'starargs') and node.starargs is not None: |
Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 383 | write_comma() |
| 384 | self.write('*') |
| 385 | self.visit(node.starargs) |
Patrick Williams | c0f7c04 | 2017-02-23 20:41:17 -0600 | [diff] [blame] | 386 | if hasattr(node, 'kwargs') and node.kwargs is not None: |
Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 387 | write_comma() |
| 388 | self.write('**') |
| 389 | self.visit(node.kwargs) |
| 390 | self.write(')') |
| 391 | |
| 392 | def visit_Name(self, node): |
| 393 | self.write(node.id) |
| 394 | |
| 395 | def visit_Str(self, node): |
| 396 | self.write(repr(node.s)) |
| 397 | |
| 398 | def visit_Bytes(self, node): |
| 399 | self.write(repr(node.s)) |
| 400 | |
| 401 | def visit_Num(self, node): |
| 402 | self.write(repr(node.n)) |
| 403 | |
| 404 | def visit_Tuple(self, node): |
| 405 | self.write('(') |
| 406 | idx = -1 |
| 407 | for idx, item in enumerate(node.elts): |
| 408 | if idx: |
| 409 | self.write(', ') |
| 410 | self.visit(item) |
| 411 | self.write(idx and ')' or ',)') |
| 412 | |
| 413 | def sequence_visit(left, right): |
| 414 | def visit(self, node): |
| 415 | self.write(left) |
| 416 | for idx, item in enumerate(node.elts): |
| 417 | if idx: |
| 418 | self.write(', ') |
| 419 | self.visit(item) |
| 420 | self.write(right) |
| 421 | return visit |
| 422 | |
| 423 | visit_List = sequence_visit('[', ']') |
| 424 | visit_Set = sequence_visit('{', '}') |
| 425 | del sequence_visit |
| 426 | |
| 427 | def visit_Dict(self, node): |
| 428 | self.write('{') |
| 429 | for idx, (key, value) in enumerate(zip(node.keys, node.values)): |
| 430 | if idx: |
| 431 | self.write(', ') |
| 432 | self.visit(key) |
| 433 | self.write(': ') |
| 434 | self.visit(value) |
| 435 | self.write('}') |
| 436 | |
| 437 | def visit_BinOp(self, node): |
| 438 | self.visit(node.left) |
| 439 | self.write(' %s ' % BINOP_SYMBOLS[type(node.op)]) |
| 440 | self.visit(node.right) |
| 441 | |
| 442 | def visit_BoolOp(self, node): |
| 443 | self.write('(') |
| 444 | for idx, value in enumerate(node.values): |
| 445 | if idx: |
| 446 | self.write(' %s ' % BOOLOP_SYMBOLS[type(node.op)]) |
| 447 | self.visit(value) |
| 448 | self.write(')') |
| 449 | |
| 450 | def visit_Compare(self, node): |
| 451 | self.write('(') |
| 452 | self.write(node.left) |
| 453 | for op, right in zip(node.ops, node.comparators): |
| 454 | self.write(' %s %%' % CMPOP_SYMBOLS[type(op)]) |
| 455 | self.visit(right) |
| 456 | self.write(')') |
| 457 | |
| 458 | def visit_UnaryOp(self, node): |
| 459 | self.write('(') |
| 460 | op = UNARYOP_SYMBOLS[type(node.op)] |
| 461 | self.write(op) |
| 462 | if op == 'not': |
| 463 | self.write(' ') |
| 464 | self.visit(node.operand) |
| 465 | self.write(')') |
| 466 | |
| 467 | def visit_Subscript(self, node): |
| 468 | self.visit(node.value) |
| 469 | self.write('[') |
| 470 | self.visit(node.slice) |
| 471 | self.write(']') |
| 472 | |
| 473 | def visit_Slice(self, node): |
| 474 | if node.lower is not None: |
| 475 | self.visit(node.lower) |
| 476 | self.write(':') |
| 477 | if node.upper is not None: |
| 478 | self.visit(node.upper) |
| 479 | if node.step is not None: |
| 480 | self.write(':') |
| 481 | if not (isinstance(node.step, Name) and node.step.id == 'None'): |
| 482 | self.visit(node.step) |
| 483 | |
| 484 | def visit_ExtSlice(self, node): |
| 485 | for idx, item in node.dims: |
| 486 | if idx: |
| 487 | self.write(', ') |
| 488 | self.visit(item) |
| 489 | |
| 490 | def visit_Yield(self, node): |
| 491 | self.write('yield ') |
| 492 | self.visit(node.value) |
| 493 | |
| 494 | def visit_Lambda(self, node): |
| 495 | self.write('lambda ') |
| 496 | self.signature(node.args) |
| 497 | self.write(': ') |
| 498 | self.visit(node.body) |
| 499 | |
| 500 | def visit_Ellipsis(self, node): |
| 501 | self.write('Ellipsis') |
| 502 | |
| 503 | def generator_visit(left, right): |
| 504 | def visit(self, node): |
| 505 | self.write(left) |
| 506 | self.visit(node.elt) |
| 507 | for comprehension in node.generators: |
| 508 | self.visit(comprehension) |
| 509 | self.write(right) |
| 510 | return visit |
| 511 | |
| 512 | visit_ListComp = generator_visit('[', ']') |
| 513 | visit_GeneratorExp = generator_visit('(', ')') |
| 514 | visit_SetComp = generator_visit('{', '}') |
| 515 | del generator_visit |
| 516 | |
| 517 | def visit_DictComp(self, node): |
| 518 | self.write('{') |
| 519 | self.visit(node.key) |
| 520 | self.write(': ') |
| 521 | self.visit(node.value) |
| 522 | for comprehension in node.generators: |
| 523 | self.visit(comprehension) |
| 524 | self.write('}') |
| 525 | |
| 526 | def visit_IfExp(self, node): |
| 527 | self.visit(node.body) |
| 528 | self.write(' if ') |
| 529 | self.visit(node.test) |
| 530 | self.write(' else ') |
| 531 | self.visit(node.orelse) |
| 532 | |
| 533 | def visit_Starred(self, node): |
| 534 | self.write('*') |
| 535 | self.visit(node.value) |
| 536 | |
| 537 | def visit_Repr(self, node): |
| 538 | # XXX: python 2.6 only |
| 539 | self.write('`') |
| 540 | self.visit(node.value) |
| 541 | self.write('`') |
| 542 | |
| 543 | # Helper Nodes |
| 544 | |
| 545 | def visit_alias(self, node): |
| 546 | self.write(node.name) |
| 547 | if node.asname is not None: |
| 548 | self.write(' as ' + node.asname) |
| 549 | |
| 550 | def visit_comprehension(self, node): |
| 551 | self.write(' for ') |
| 552 | self.visit(node.target) |
| 553 | self.write(' in ') |
| 554 | self.visit(node.iter) |
| 555 | if node.ifs: |
| 556 | for if_ in node.ifs: |
| 557 | self.write(' if ') |
| 558 | self.visit(if_) |
| 559 | |
| 560 | def visit_excepthandler(self, node): |
| 561 | self.newline(node) |
| 562 | self.write('except') |
| 563 | if node.type is not None: |
| 564 | self.write(' ') |
| 565 | self.visit(node.type) |
| 566 | if node.name is not None: |
| 567 | self.write(' as ') |
| 568 | self.visit(node.name) |
| 569 | self.write(':') |
| 570 | self.body(node.body) |