blob: cd7bcebe0f83cbaf84ff906701fdcbdf7e0e01f4 [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
182class LambdaScope(FunctionScope):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000183 __super_init = Scope.__init__
184
185 __counter = 1
Tim Peterse0c446b2001-10-18 21:57:37 +0000186
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000187 def __init__(self, module, klass=None):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000188 i = self.__counter
189 self.__counter += 1
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000190 self.__super_init("lambda.%d" % i, module, klass)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000191
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000192class ClassScope(Scope):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000193 __super_init = Scope.__init__
194
195 def __init__(self, name, module):
196 self.__super_init(name, module, name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000197
198class SymbolVisitor:
199 def __init__(self):
200 self.scopes = {}
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000201 self.klass = None
Tim Peterse0c446b2001-10-18 21:57:37 +0000202
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000203 # node that define new scopes
204
205 def visitModule(self, node):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000206 scope = self.module = self.scopes[node] = ModuleScope()
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000207 self.visit(node.node, scope)
208
Barry Warsaw52acb492001-12-21 20:04:22 +0000209 visitExpression = visitModule
210
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000211 def visitFunction(self, node, parent):
212 parent.add_def(node.name)
213 for n in node.defaults:
214 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000215 scope = FunctionScope(node.name, self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000216 if parent.nested or isinstance(parent, FunctionScope):
217 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000218 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000219 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000220 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000221 self.handle_free_vars(scope, parent)
Tim Peterse0c446b2001-10-18 21:57:37 +0000222
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000223 def visitLambda(self, node, parent):
224 for n in node.defaults:
225 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000226 scope = LambdaScope(self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000227 if parent.nested or isinstance(parent, FunctionScope):
228 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000229 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000230 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000231 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000232 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000233
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000234 def _do_args(self, scope, args):
235 for name in args:
236 if type(name) == types.TupleType:
237 self._do_args(scope, name)
238 else:
239 scope.add_param(name)
240
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000241 def handle_free_vars(self, scope, parent):
242 parent.add_child(scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000243 scope.handle_children()
244
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000245 def visitClass(self, node, parent):
246 parent.add_def(node.name)
247 for n in node.bases:
248 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000249 scope = ClassScope(node.name, self.module)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000250 if parent.nested or isinstance(parent, FunctionScope):
251 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000252 self.scopes[node] = scope
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000253 prev = self.klass
254 self.klass = node.name
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000255 self.visit(node.code, scope)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000256 self.klass = prev
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000257 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000258
259 # name can be a def or a use
260
261 # XXX a few calls and nodes expect a third "assign" arg that is
262 # true if the name is being used as an assignment. only
263 # expressions contained within statements may have the assign arg.
264
265 def visitName(self, node, scope, assign=0):
266 if assign:
267 scope.add_def(node.name)
268 else:
269 scope.add_use(node.name)
270
271 # operations that bind new names
272
273 def visitFor(self, node, scope):
274 self.visit(node.assign, scope, 1)
275 self.visit(node.list, scope)
276 self.visit(node.body, scope)
277 if node.else_:
278 self.visit(node.else_, scope)
279
280 def visitFrom(self, node, scope):
281 for name, asname in node.names:
282 if name == "*":
283 continue
284 scope.add_def(asname or name)
285
286 def visitImport(self, node, scope):
287 for name, asname in node.names:
288 i = name.find(".")
289 if i > -1:
290 name = name[:i]
291 scope.add_def(asname or name)
292
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000293 def visitGlobal(self, node, scope):
294 for name in node.names:
295 scope.add_global(name)
296
297 def visitAssign(self, node, scope):
298 """Propagate assignment flag down to child nodes.
299
300 The Assign node doesn't itself contains the variables being
301 assigned to. Instead, the children in node.nodes are visited
302 with the assign flag set to true. When the names occur in
303 those nodes, they are marked as defs.
304
305 Some names that occur in an assignment target are not bound by
306 the assignment, e.g. a name occurring inside a slice. The
307 visitor handles these nodes specially; they do not propagate
308 the assign flag to their children.
309 """
310 for n in node.nodes:
311 self.visit(n, scope, 1)
312 self.visit(node.expr, scope)
313
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000314 def visitAssName(self, node, scope, assign=1):
315 scope.add_def(node.name)
316
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000317 def visitAssAttr(self, node, scope, assign=0):
318 self.visit(node.expr, scope, 0)
319
320 def visitSubscript(self, node, scope, assign=0):
321 self.visit(node.expr, scope, 0)
322 for n in node.subs:
323 self.visit(n, scope, 0)
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000324
325 def visitSlice(self, node, scope, assign=0):
Jeremy Hylton652a2242001-09-14 22:45:57 +0000326 self.visit(node.expr, scope, 0)
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000327 if node.lower:
328 self.visit(node.lower, scope, 0)
329 if node.upper:
330 self.visit(node.upper, scope, 0)
Tim Peterse0c446b2001-10-18 21:57:37 +0000331
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000332 def visitAugAssign(self, node, scope):
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000333 # If the LHS is a name, then this counts as assignment.
334 # Otherwise, it's just use.
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000335 self.visit(node.node, scope)
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000336 if isinstance(node.node, ast.Name):
337 self.visit(node.node, scope, 1) # XXX worry about this
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000338 self.visit(node.expr, scope)
339
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000340 # prune if statements if tests are false
341
342 _const_types = types.StringType, types.IntType, types.FloatType
343
344 def visitIf(self, node, scope):
345 for test, body in node.tests:
346 if isinstance(test, ast.Const):
347 if type(test.value) in self._const_types:
348 if not test.value:
349 continue
350 self.visit(test, scope)
351 self.visit(body, scope)
352 if node.else_:
353 self.visit(node.else_, scope)
354
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000355 # a yield statement signals a generator
356
357 def visitYield(self, node, scope):
Jeremy Hylton652a2242001-09-14 22:45:57 +0000358 scope.generator = 1
Jeremy Hyltond4be10d2001-08-29 18:10:51 +0000359 self.visit(node.value, scope)
360
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000361def sort(l):
362 l = l[:]
363 l.sort()
364 return l
365
366def list_eq(l1, l2):
367 return sort(l1) == sort(l2)
368
369if __name__ == "__main__":
370 import sys
371 from compiler import parseFile, walk
372 import symtable
373
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000374 def get_names(syms):
375 return [s for s in [s.get_name() for s in syms.get_symbols()]
Tim Peterse0c446b2001-10-18 21:57:37 +0000376 if not (s.startswith('_[') or s.startswith('.'))]
377
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000378 for file in sys.argv[1:]:
379 print file
380 f = open(file)
381 buf = f.read()
382 f.close()
383 syms = symtable.symtable(buf, file, "exec")
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000384 mod_names = get_names(syms)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000385 tree = parseFile(file)
386 s = SymbolVisitor()
387 walk(tree, s)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000388
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000389 # compare module-level symbols
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000390 names2 = s.scopes[tree].get_names()
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000391
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000392 if not list_eq(mod_names, names2):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000393 print
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000394 print "oops", file
395 print sort(mod_names)
396 print sort(names2)
397 sys.exit(-1)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000398
399 d = {}
400 d.update(s.scopes)
401 del d[tree]
402 scopes = d.values()
403 del d
404
405 for s in syms.get_symbols():
406 if s.is_namespace():
407 l = [sc for sc in scopes
408 if sc.name == s.get_name()]
409 if len(l) > 1:
410 print "skipping", s.get_name()
411 else:
412 if not list_eq(get_names(s.get_namespace()),
413 l[0].get_names()):
414 print s.get_name()
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000415 print sort(get_names(s.get_namespace()))
416 print sort(l[0].get_names())
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000417 sys.exit(-1)