blob: 9d4605a821bdceedae6300b9b36e104b6cece669 [file] [log] [blame]
Jeremy Hylton8b966dc2001-04-09 04:35:35 +00001"""Module symbol-table generator"""
2
3from compiler import ast
Jeremy Hylton364f9b92001-04-12 06:40:42 +00004from compiler.consts import SC_LOCAL, SC_GLOBAL, SC_FREE, SC_CELL, SC_UNKNOWN
Jeremy Hyltonc59e2202001-08-27 22:56:16 +00005from compiler.misc import mangle
Jeremy Hyltonf870c952001-04-09 13:57:32 +00006import types
Jeremy Hylton8b966dc2001-04-09 04:35:35 +00007
Jeremy Hyltonc59e2202001-08-27 22:56:16 +00008
Jeremy Hylton364f9b92001-04-12 06:40:42 +00009import sys
10
Jeremy Hyltonf870c952001-04-09 13:57:32 +000011MANGLE_LEN = 256
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000012
13class Scope:
14 # XXX how much information do I need about each name?
Jeremy Hyltonf870c952001-04-09 13:57:32 +000015 def __init__(self, name, module, klass=None):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000016 self.name = name
Jeremy Hyltonf870c952001-04-09 13:57:32 +000017 self.module = module
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000018 self.defs = {}
19 self.uses = {}
20 self.globals = {}
21 self.params = {}
Jeremy Hylton364f9b92001-04-12 06:40:42 +000022 self.frees = {}
23 self.cells = {}
Jeremy Hyltonf870c952001-04-09 13:57:32 +000024 self.children = []
Jeremy Hylton364f9b92001-04-12 06:40:42 +000025 # nested is true if the class could contain free variables,
26 # i.e. if it is nested within another function.
27 self.nested = None
Jeremy Hyltond4be10d2001-08-29 18:10:51 +000028 self.generator = None
Jeremy Hyltonf870c952001-04-09 13:57:32 +000029 self.klass = None
30 if klass is not None:
31 for i in range(len(klass)):
32 if klass[i] != '_':
33 self.klass = klass[i:]
34 break
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000035
36 def __repr__(self):
37 return "<%s: %s>" % (self.__class__.__name__, self.name)
38
Jeremy Hyltonf870c952001-04-09 13:57:32 +000039 def mangle(self, name):
40 if self.klass is None:
41 return name
Jeremy Hyltonc59e2202001-08-27 22:56:16 +000042 return mangle(name, self.klass)
Jeremy Hyltonf870c952001-04-09 13:57:32 +000043
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000044 def add_def(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000045 self.defs[self.mangle(name)] = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000046
47 def add_use(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000048 self.uses[self.mangle(name)] = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000049
50 def add_global(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000051 name = self.mangle(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000052 if self.uses.has_key(name) or self.defs.has_key(name):
53 pass # XXX warn about global following def/use
54 if self.params.has_key(name):
55 raise SyntaxError, "%s in %s is global and parameter" % \
56 (name, self.name)
57 self.globals[name] = 1
Jeremy Hyltonf870c952001-04-09 13:57:32 +000058 self.module.add_def(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000059
60 def add_param(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000061 name = self.mangle(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000062 self.defs[name] = 1
63 self.params[name] = 1
64
65 def get_names(self):
66 d = {}
67 d.update(self.defs)
68 d.update(self.uses)
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +000069 d.update(self.globals)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000070 return d.keys()
71
Jeremy Hyltonf870c952001-04-09 13:57:32 +000072 def add_child(self, child):
73 self.children.append(child)
74
75 def get_children(self):
76 return self.children
77
Jeremy Hylton364f9b92001-04-12 06:40:42 +000078 def DEBUG(self):
Jeremy Hylton364f9b92001-04-12 06:40:42 +000079 print >> sys.stderr, self.name, self.nested and "nested" or ""
80 print >> sys.stderr, "\tglobals: ", self.globals
81 print >> sys.stderr, "\tcells: ", self.cells
82 print >> sys.stderr, "\tdefs: ", self.defs
83 print >> sys.stderr, "\tuses: ", self.uses
84 print >> sys.stderr, "\tfrees:", self.frees
85
86 def check_name(self, name):
87 """Return scope of name.
88
89 The scope of a name could be LOCAL, GLOBAL, FREE, or CELL.
90 """
91 if self.globals.has_key(name):
92 return SC_GLOBAL
93 if self.cells.has_key(name):
94 return SC_CELL
95 if self.defs.has_key(name):
96 return SC_LOCAL
97 if self.nested and (self.frees.has_key(name) or
98 self.uses.has_key(name)):
99 return SC_FREE
100 if self.nested:
101 return SC_UNKNOWN
102 else:
103 return SC_GLOBAL
104
105 def get_free_vars(self):
106 if not self.nested:
107 return ()
108 free = {}
109 free.update(self.frees)
110 for name in self.uses.keys():
111 if not (self.defs.has_key(name) or
112 self.globals.has_key(name)):
113 free[name] = 1
114 return free.keys()
115
116 def handle_children(self):
117 for child in self.children:
118 frees = child.get_free_vars()
119 globals = self.add_frees(frees)
120 for name in globals:
121 child.force_global(name)
122
123 def force_global(self, name):
124 """Force name to be global in scope.
125
126 Some child of the current node had a free reference to name.
127 When the child was processed, it was labelled a free
128 variable. Now that all its enclosing scope have been
129 processed, the name is known to be a global or builtin. So
130 walk back down the child chain and set the name to be global
131 rather than free.
132
133 Be careful to stop if a child does not think the name is
Tim Peterse0c446b2001-10-18 21:57:37 +0000134 free.
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000135 """
136 self.globals[name] = 1
137 if self.frees.has_key(name):
138 del self.frees[name]
139 for child in self.children:
140 if child.check_name(name) == SC_FREE:
141 child.force_global(name)
142
143 def add_frees(self, names):
144 """Process list of free vars from nested scope.
145
146 Returns a list of names that are either 1) declared global in the
147 parent or 2) undefined in a top-level parent. In either case,
148 the nested scope should treat them as globals.
149 """
150 child_globals = []
151 for name in names:
152 sc = self.check_name(name)
153 if self.nested:
154 if sc == SC_UNKNOWN or sc == SC_FREE \
155 or isinstance(self, ClassScope):
156 self.frees[name] = 1
157 elif sc == SC_GLOBAL:
158 child_globals.append(name)
159 elif isinstance(self, FunctionScope) and sc == SC_LOCAL:
160 self.cells[name] = 1
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000161 elif sc != SC_CELL:
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000162 child_globals.append(name)
163 else:
164 if sc == SC_LOCAL:
165 self.cells[name] = 1
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000166 elif sc != SC_CELL:
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000167 child_globals.append(name)
168 return child_globals
169
170 def get_cell_vars(self):
171 return self.cells.keys()
172
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000173class ModuleScope(Scope):
174 __super_init = Scope.__init__
Tim Peterse0c446b2001-10-18 21:57:37 +0000175
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000176 def __init__(self):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000177 self.__super_init("global", self)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000178
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000179class FunctionScope(Scope):
180 pass
181
Raymond Hettinger354433a2004-05-19 08:20:33 +0000182class GenExprScope(Scope):
183 __super_init = Scope.__init__
184
185 __counter = 1
186
187 def __init__(self, module, klass=None):
188 i = self.__counter
189 self.__counter += 1
190 self.__super_init("generator expression<%d>"%i, module, klass)
191 self.add_param('[outmost-iterable]')
192
193 def get_names(self):
194 keys = Scope.get_names()
195 return keys
196
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000197class LambdaScope(FunctionScope):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000198 __super_init = Scope.__init__
199
200 __counter = 1
Tim Peterse0c446b2001-10-18 21:57:37 +0000201
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000202 def __init__(self, module, klass=None):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000203 i = self.__counter
204 self.__counter += 1
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000205 self.__super_init("lambda.%d" % i, module, klass)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000206
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000207class ClassScope(Scope):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000208 __super_init = Scope.__init__
209
210 def __init__(self, name, module):
211 self.__super_init(name, module, name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000212
213class SymbolVisitor:
214 def __init__(self):
215 self.scopes = {}
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000216 self.klass = None
Tim Peterse0c446b2001-10-18 21:57:37 +0000217
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000218 # node that define new scopes
219
220 def visitModule(self, node):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000221 scope = self.module = self.scopes[node] = ModuleScope()
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000222 self.visit(node.node, scope)
223
Barry Warsaw52acb492001-12-21 20:04:22 +0000224 visitExpression = visitModule
225
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000226 def visitFunction(self, node, parent):
Anthony Baxterc2a5a632004-08-02 06:10:11 +0000227 if node.decorators:
228 self.visit(node.decorators, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000229 parent.add_def(node.name)
230 for n in node.defaults:
231 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000232 scope = FunctionScope(node.name, self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000233 if parent.nested or isinstance(parent, FunctionScope):
234 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000235 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000236 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000237 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000238 self.handle_free_vars(scope, parent)
Tim Peterse0c446b2001-10-18 21:57:37 +0000239
Raymond Hettinger354433a2004-05-19 08:20:33 +0000240 def visitGenExpr(self, node, parent):
241 scope = GenExprScope(self.module, self.klass);
242 if parent.nested or isinstance(parent, FunctionScope) \
243 or isinstance(parent, GenExprScope):
244 scope.nested = 1
245
246 self.scopes[node] = scope
247 self.visit(node.code, scope)
248
249 self.handle_free_vars(scope, parent)
250
251 def visitGenExprInner(self, node, scope):
252 for genfor in node.quals:
253 self.visit(genfor, scope)
254
255 self.visit(node.expr, scope)
256
257 def visitGenExprFor(self, node, scope):
258 self.visit(node.assign, scope, 1)
259 self.visit(node.iter, scope)
260 for if_ in node.ifs:
261 self.visit(if_, scope)
262
263 def visitGenExprIf(self, node, scope):
264 self.visit(node.test, scope)
Tim Peters4e0e1b62004-07-07 20:54:48 +0000265
Jeremy Hyltonead21f52003-08-28 02:09:26 +0000266 def visitLambda(self, node, parent, assign=0):
267 # Lambda is an expression, so it could appear in an expression
268 # context where assign is passed. The transformer should catch
269 # any code that has a lambda on the left-hand side.
Tim Peters4e0e1b62004-07-07 20:54:48 +0000270 assert not assign
271
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000272 for n in node.defaults:
273 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000274 scope = LambdaScope(self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000275 if parent.nested or isinstance(parent, FunctionScope):
276 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000277 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000278 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000279 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000280 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000281
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000282 def _do_args(self, scope, args):
283 for name in args:
284 if type(name) == types.TupleType:
285 self._do_args(scope, name)
286 else:
287 scope.add_param(name)
288
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000289 def handle_free_vars(self, scope, parent):
290 parent.add_child(scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000291 scope.handle_children()
292
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000293 def visitClass(self, node, parent):
294 parent.add_def(node.name)
295 for n in node.bases:
296 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000297 scope = ClassScope(node.name, self.module)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000298 if parent.nested or isinstance(parent, FunctionScope):
299 scope.nested = 1
Jeremy Hyltonaccb62b2002-12-31 18:17:44 +0000300 if node.doc is not None:
301 scope.add_def('__doc__')
302 scope.add_def('__module__')
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000303 self.scopes[node] = scope
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000304 prev = self.klass
305 self.klass = node.name
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000306 self.visit(node.code, scope)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000307 self.klass = prev
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000308 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000309
310 # name can be a def or a use
311
312 # XXX a few calls and nodes expect a third "assign" arg that is
313 # true if the name is being used as an assignment. only
314 # expressions contained within statements may have the assign arg.
315
316 def visitName(self, node, scope, assign=0):
317 if assign:
318 scope.add_def(node.name)
319 else:
320 scope.add_use(node.name)
321
322 # operations that bind new names
323
324 def visitFor(self, node, scope):
325 self.visit(node.assign, scope, 1)
326 self.visit(node.list, scope)
327 self.visit(node.body, scope)
328 if node.else_:
329 self.visit(node.else_, scope)
330
331 def visitFrom(self, node, scope):
332 for name, asname in node.names:
333 if name == "*":
334 continue
335 scope.add_def(asname or name)
336
337 def visitImport(self, node, scope):
338 for name, asname in node.names:
339 i = name.find(".")
340 if i > -1:
341 name = name[:i]
342 scope.add_def(asname or name)
343
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000344 def visitGlobal(self, node, scope):
345 for name in node.names:
346 scope.add_global(name)
347
348 def visitAssign(self, node, scope):
349 """Propagate assignment flag down to child nodes.
350
351 The Assign node doesn't itself contains the variables being
352 assigned to. Instead, the children in node.nodes are visited
353 with the assign flag set to true. When the names occur in
354 those nodes, they are marked as defs.
355
356 Some names that occur in an assignment target are not bound by
357 the assignment, e.g. a name occurring inside a slice. The
358 visitor handles these nodes specially; they do not propagate
359 the assign flag to their children.
360 """
361 for n in node.nodes:
362 self.visit(n, scope, 1)
363 self.visit(node.expr, scope)
364
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000365 def visitAssName(self, node, scope, assign=1):
366 scope.add_def(node.name)
367
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000368 def visitAssAttr(self, node, scope, assign=0):
369 self.visit(node.expr, scope, 0)
370
371 def visitSubscript(self, node, scope, assign=0):
372 self.visit(node.expr, scope, 0)
373 for n in node.subs:
374 self.visit(n, scope, 0)
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000375
376 def visitSlice(self, node, scope, assign=0):
Jeremy Hylton652a2242001-09-14 22:45:57 +0000377 self.visit(node.expr, scope, 0)
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000378 if node.lower:
379 self.visit(node.lower, scope, 0)
380 if node.upper:
381 self.visit(node.upper, scope, 0)
Tim Peterse0c446b2001-10-18 21:57:37 +0000382
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000383 def visitAugAssign(self, node, scope):
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000384 # If the LHS is a name, then this counts as assignment.
385 # Otherwise, it's just use.
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000386 self.visit(node.node, scope)
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000387 if isinstance(node.node, ast.Name):
388 self.visit(node.node, scope, 1) # XXX worry about this
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000389 self.visit(node.expr, scope)
390
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000391 # prune if statements if tests are false
392
393 _const_types = types.StringType, types.IntType, types.FloatType
394
395 def visitIf(self, node, scope):
396 for test, body in node.tests:
397 if isinstance(test, ast.Const):
398 if type(test.value) in self._const_types:
399 if not test.value:
400 continue
401 self.visit(test, scope)
402 self.visit(body, scope)
403 if node.else_:
404 self.visit(node.else_, scope)
405
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000406 # a yield statement signals a generator
407
408 def visitYield(self, node, scope):
Jeremy Hylton652a2242001-09-14 22:45:57 +0000409 scope.generator = 1
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000410 self.visit(node.value, scope)
411
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000412def sort(l):
413 l = l[:]
414 l.sort()
415 return l
416
417def list_eq(l1, l2):
418 return sort(l1) == sort(l2)
419
420if __name__ == "__main__":
421 import sys
422 from compiler import parseFile, walk
423 import symtable
424
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000425 def get_names(syms):
426 return [s for s in [s.get_name() for s in syms.get_symbols()]
Tim Peterse0c446b2001-10-18 21:57:37 +0000427 if not (s.startswith('_[') or s.startswith('.'))]
428
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000429 for file in sys.argv[1:]:
430 print file
431 f = open(file)
432 buf = f.read()
433 f.close()
434 syms = symtable.symtable(buf, file, "exec")
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000435 mod_names = get_names(syms)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000436 tree = parseFile(file)
437 s = SymbolVisitor()
438 walk(tree, s)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000439
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000440 # compare module-level symbols
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000441 names2 = s.scopes[tree].get_names()
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000442
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000443 if not list_eq(mod_names, names2):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000444 print
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000445 print "oops", file
446 print sort(mod_names)
447 print sort(names2)
448 sys.exit(-1)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000449
450 d = {}
451 d.update(s.scopes)
452 del d[tree]
453 scopes = d.values()
454 del d
455
456 for s in syms.get_symbols():
457 if s.is_namespace():
458 l = [sc for sc in scopes
459 if sc.name == s.get_name()]
460 if len(l) > 1:
461 print "skipping", s.get_name()
462 else:
463 if not list_eq(get_names(s.get_namespace()),
464 l[0].get_names()):
465 print s.get_name()
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000466 print sort(get_names(s.get_namespace()))
467 print sort(l[0].get_names())
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000468 sys.exit(-1)