blob: 0ef0d12f03bc635005970a619cc58ae178e432e5 [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 Hyltonf870c952001-04-09 13:57:32 +00005import types
Jeremy Hylton8b966dc2001-04-09 04:35:35 +00006
Jeremy Hylton364f9b92001-04-12 06:40:42 +00007import sys
8
Jeremy Hyltonf870c952001-04-09 13:57:32 +00009MANGLE_LEN = 256
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000010
11class Scope:
12 # XXX how much information do I need about each name?
Jeremy Hyltonf870c952001-04-09 13:57:32 +000013 def __init__(self, name, module, klass=None):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000014 self.name = name
Jeremy Hyltonf870c952001-04-09 13:57:32 +000015 self.module = module
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000016 self.defs = {}
17 self.uses = {}
18 self.globals = {}
19 self.params = {}
Jeremy Hylton364f9b92001-04-12 06:40:42 +000020 self.frees = {}
21 self.cells = {}
Jeremy Hyltonf870c952001-04-09 13:57:32 +000022 self.children = []
Jeremy Hylton364f9b92001-04-12 06:40:42 +000023 # nested is true if the class could contain free variables,
24 # i.e. if it is nested within another function.
25 self.nested = None
Jeremy Hyltonf870c952001-04-09 13:57:32 +000026 self.klass = None
27 if klass is not None:
28 for i in range(len(klass)):
29 if klass[i] != '_':
30 self.klass = klass[i:]
31 break
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000032
33 def __repr__(self):
34 return "<%s: %s>" % (self.__class__.__name__, self.name)
35
Jeremy Hyltonf870c952001-04-09 13:57:32 +000036 def mangle(self, name):
37 if self.klass is None:
38 return name
39 if not name.startswith('__'):
40 return name
41 if len(name) + 2 >= MANGLE_LEN:
42 return name
43 if name.endswith('__'):
44 return name
45 return "_%s%s" % (self.klass, name)
46
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000047 def add_def(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000048 self.defs[self.mangle(name)] = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000049
50 def add_use(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000051 self.uses[self.mangle(name)] = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000052
53 def add_global(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000054 name = self.mangle(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000055 if self.uses.has_key(name) or self.defs.has_key(name):
56 pass # XXX warn about global following def/use
57 if self.params.has_key(name):
58 raise SyntaxError, "%s in %s is global and parameter" % \
59 (name, self.name)
60 self.globals[name] = 1
Jeremy Hyltonf870c952001-04-09 13:57:32 +000061 self.module.add_def(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000062
63 def add_param(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000064 name = self.mangle(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000065 self.defs[name] = 1
66 self.params[name] = 1
67
68 def get_names(self):
69 d = {}
70 d.update(self.defs)
71 d.update(self.uses)
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +000072 d.update(self.globals)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000073 return d.keys()
74
Jeremy Hyltonf870c952001-04-09 13:57:32 +000075 def add_child(self, child):
76 self.children.append(child)
77
78 def get_children(self):
79 return self.children
80
Jeremy Hylton364f9b92001-04-12 06:40:42 +000081 def DEBUG(self):
Jeremy Hylton364f9b92001-04-12 06:40:42 +000082 print >> sys.stderr, self.name, self.nested and "nested" or ""
83 print >> sys.stderr, "\tglobals: ", self.globals
84 print >> sys.stderr, "\tcells: ", self.cells
85 print >> sys.stderr, "\tdefs: ", self.defs
86 print >> sys.stderr, "\tuses: ", self.uses
87 print >> sys.stderr, "\tfrees:", self.frees
88
89 def check_name(self, name):
90 """Return scope of name.
91
92 The scope of a name could be LOCAL, GLOBAL, FREE, or CELL.
93 """
94 if self.globals.has_key(name):
95 return SC_GLOBAL
96 if self.cells.has_key(name):
97 return SC_CELL
98 if self.defs.has_key(name):
99 return SC_LOCAL
100 if self.nested and (self.frees.has_key(name) or
101 self.uses.has_key(name)):
102 return SC_FREE
103 if self.nested:
104 return SC_UNKNOWN
105 else:
106 return SC_GLOBAL
107
108 def get_free_vars(self):
109 if not self.nested:
110 return ()
111 free = {}
112 free.update(self.frees)
113 for name in self.uses.keys():
114 if not (self.defs.has_key(name) or
115 self.globals.has_key(name)):
116 free[name] = 1
117 return free.keys()
118
119 def handle_children(self):
120 for child in self.children:
121 frees = child.get_free_vars()
122 globals = self.add_frees(frees)
123 for name in globals:
124 child.force_global(name)
125
126 def force_global(self, name):
127 """Force name to be global in scope.
128
129 Some child of the current node had a free reference to name.
130 When the child was processed, it was labelled a free
131 variable. Now that all its enclosing scope have been
132 processed, the name is known to be a global or builtin. So
133 walk back down the child chain and set the name to be global
134 rather than free.
135
136 Be careful to stop if a child does not think the name is
137 free.
138 """
139 self.globals[name] = 1
140 if self.frees.has_key(name):
141 del self.frees[name]
142 for child in self.children:
143 if child.check_name(name) == SC_FREE:
144 child.force_global(name)
145
146 def add_frees(self, names):
147 """Process list of free vars from nested scope.
148
149 Returns a list of names that are either 1) declared global in the
150 parent or 2) undefined in a top-level parent. In either case,
151 the nested scope should treat them as globals.
152 """
153 child_globals = []
154 for name in names:
155 sc = self.check_name(name)
156 if self.nested:
157 if sc == SC_UNKNOWN or sc == SC_FREE \
158 or isinstance(self, ClassScope):
159 self.frees[name] = 1
160 elif sc == SC_GLOBAL:
161 child_globals.append(name)
162 elif isinstance(self, FunctionScope) and sc == SC_LOCAL:
163 self.cells[name] = 1
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000164 elif sc != SC_CELL:
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000165 child_globals.append(name)
166 else:
167 if sc == SC_LOCAL:
168 self.cells[name] = 1
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000169 elif sc != SC_CELL:
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000170 child_globals.append(name)
171 return child_globals
172
173 def get_cell_vars(self):
174 return self.cells.keys()
175
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000176class ModuleScope(Scope):
177 __super_init = Scope.__init__
178
179 def __init__(self):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000180 self.__super_init("global", self)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000181
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000182class FunctionScope(Scope):
183 pass
184
185class LambdaScope(FunctionScope):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000186 __super_init = Scope.__init__
187
188 __counter = 1
189
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000190 def __init__(self, module, klass=None):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000191 i = self.__counter
192 self.__counter += 1
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000193 self.__super_init("lambda.%d" % i, module, klass)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000194
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000195class ClassScope(Scope):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000196 __super_init = Scope.__init__
197
198 def __init__(self, name, module):
199 self.__super_init(name, module, name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000200
201class SymbolVisitor:
202 def __init__(self):
203 self.scopes = {}
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000204 self.klass = None
205
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000206 # node that define new scopes
207
208 def visitModule(self, node):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000209 scope = self.module = self.scopes[node] = ModuleScope()
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000210 self.visit(node.node, scope)
211
212 def visitFunction(self, node, parent):
213 parent.add_def(node.name)
214 for n in node.defaults:
215 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000216 scope = FunctionScope(node.name, self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000217 if parent.nested or isinstance(parent, FunctionScope):
218 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000219 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000220 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000221 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000222 self.handle_free_vars(scope, parent)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000223
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000224 def visitLambda(self, node, parent):
225 for n in node.defaults:
226 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000227 scope = LambdaScope(self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000228 if parent.nested or isinstance(parent, FunctionScope):
229 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000230 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000231 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000232 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000233 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000234
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000235 def _do_args(self, scope, args):
236 for name in args:
237 if type(name) == types.TupleType:
238 self._do_args(scope, name)
239 else:
240 scope.add_param(name)
241
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000242 def handle_free_vars(self, scope, parent):
243 parent.add_child(scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000244 scope.handle_children()
245
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000246 def visitClass(self, node, parent):
247 parent.add_def(node.name)
248 for n in node.bases:
249 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000250 scope = ClassScope(node.name, self.module)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000251 if parent.nested or isinstance(parent, FunctionScope):
252 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000253 self.scopes[node] = scope
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000254 prev = self.klass
255 self.klass = node.name
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000256 self.visit(node.code, scope)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000257 self.klass = prev
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000258 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000259
260 # name can be a def or a use
261
262 # XXX a few calls and nodes expect a third "assign" arg that is
263 # true if the name is being used as an assignment. only
264 # expressions contained within statements may have the assign arg.
265
266 def visitName(self, node, scope, assign=0):
267 if assign:
268 scope.add_def(node.name)
269 else:
270 scope.add_use(node.name)
271
272 # operations that bind new names
273
274 def visitFor(self, node, scope):
275 self.visit(node.assign, scope, 1)
276 self.visit(node.list, scope)
277 self.visit(node.body, scope)
278 if node.else_:
279 self.visit(node.else_, scope)
280
281 def visitFrom(self, node, scope):
282 for name, asname in node.names:
283 if name == "*":
284 continue
285 scope.add_def(asname or name)
286
287 def visitImport(self, node, scope):
288 for name, asname in node.names:
289 i = name.find(".")
290 if i > -1:
291 name = name[:i]
292 scope.add_def(asname or name)
293
294 def visitAssName(self, node, scope, assign=1):
295 scope.add_def(node.name)
296
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000297 def visitAssAttr(self, node, scope, assign=0):
298 self.visit(node.expr, scope, 0)
299
300 def visitSubscript(self, node, scope, assign=0):
301 self.visit(node.expr, scope, 0)
302 for n in node.subs:
303 self.visit(n, scope, 0)
304
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000305 def visitAugAssign(self, node, scope):
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000306 # If the LHS is a name, then this counts as assignment.
307 # Otherwise, it's just use.
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000308 self.visit(node.node, scope)
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000309 if isinstance(node.node, ast.Name):
310 self.visit(node.node, scope, 1) # XXX worry about this
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000311 self.visit(node.expr, scope)
312
313 def visitAssign(self, node, scope):
314 for n in node.nodes:
315 self.visit(n, scope, 1)
316 self.visit(node.expr, scope)
317
318 def visitGlobal(self, node, scope):
319 for name in node.names:
320 scope.add_global(name)
321
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000322 # prune if statements if tests are false
323
324 _const_types = types.StringType, types.IntType, types.FloatType
325
326 def visitIf(self, node, scope):
327 for test, body in node.tests:
328 if isinstance(test, ast.Const):
329 if type(test.value) in self._const_types:
330 if not test.value:
331 continue
332 self.visit(test, scope)
333 self.visit(body, scope)
334 if node.else_:
335 self.visit(node.else_, scope)
336
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000337def sort(l):
338 l = l[:]
339 l.sort()
340 return l
341
342def list_eq(l1, l2):
343 return sort(l1) == sort(l2)
344
345if __name__ == "__main__":
346 import sys
347 from compiler import parseFile, walk
348 import symtable
349
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000350 def get_names(syms):
351 return [s for s in [s.get_name() for s in syms.get_symbols()]
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000352 if not (s.startswith('_[') or s.startswith('.'))]
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000353
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000354 for file in sys.argv[1:]:
355 print file
356 f = open(file)
357 buf = f.read()
358 f.close()
359 syms = symtable.symtable(buf, file, "exec")
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000360 mod_names = get_names(syms)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000361 tree = parseFile(file)
362 s = SymbolVisitor()
363 walk(tree, s)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000364
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000365 # compare module-level symbols
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000366 names2 = s.scopes[tree].get_names()
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000367
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000368 if not list_eq(mod_names, names2):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000369 print
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000370 print "oops", file
371 print sort(mod_names)
372 print sort(names2)
373 sys.exit(-1)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000374
375 d = {}
376 d.update(s.scopes)
377 del d[tree]
378 scopes = d.values()
379 del d
380
381 for s in syms.get_symbols():
382 if s.is_namespace():
383 l = [sc for sc in scopes
384 if sc.name == s.get_name()]
385 if len(l) > 1:
386 print "skipping", s.get_name()
387 else:
388 if not list_eq(get_names(s.get_namespace()),
389 l[0].get_names()):
390 print s.get_name()
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000391 print sort(get_names(s.get_namespace()))
392 print sort(l[0].get_names())
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000393 sys.exit(-1)