blob: 40fd127bfd19705b765d1e26268454323a20204b [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 Hyltonf870c952001-04-09 13:57:32 +000028 self.klass = None
29 if klass is not None:
30 for i in range(len(klass)):
31 if klass[i] != '_':
32 self.klass = klass[i:]
33 break
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000034
35 def __repr__(self):
36 return "<%s: %s>" % (self.__class__.__name__, self.name)
37
Jeremy Hyltonf870c952001-04-09 13:57:32 +000038 def mangle(self, name):
39 if self.klass is None:
40 return name
Jeremy Hyltonc59e2202001-08-27 22:56:16 +000041 return mangle(name, self.klass)
Jeremy Hyltonf870c952001-04-09 13:57:32 +000042
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000043 def add_def(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000044 self.defs[self.mangle(name)] = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000045
46 def add_use(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000047 self.uses[self.mangle(name)] = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000048
49 def add_global(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000050 name = self.mangle(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000051 if self.uses.has_key(name) or self.defs.has_key(name):
52 pass # XXX warn about global following def/use
53 if self.params.has_key(name):
54 raise SyntaxError, "%s in %s is global and parameter" % \
55 (name, self.name)
56 self.globals[name] = 1
Jeremy Hyltonf870c952001-04-09 13:57:32 +000057 self.module.add_def(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000058
59 def add_param(self, name):
Jeremy Hyltonf870c952001-04-09 13:57:32 +000060 name = self.mangle(name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000061 self.defs[name] = 1
62 self.params[name] = 1
63
64 def get_names(self):
65 d = {}
66 d.update(self.defs)
67 d.update(self.uses)
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +000068 d.update(self.globals)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +000069 return d.keys()
70
Jeremy Hyltonf870c952001-04-09 13:57:32 +000071 def add_child(self, child):
72 self.children.append(child)
73
74 def get_children(self):
75 return self.children
76
Jeremy Hylton364f9b92001-04-12 06:40:42 +000077 def DEBUG(self):
Jeremy Hylton364f9b92001-04-12 06:40:42 +000078 print >> sys.stderr, self.name, self.nested and "nested" or ""
79 print >> sys.stderr, "\tglobals: ", self.globals
80 print >> sys.stderr, "\tcells: ", self.cells
81 print >> sys.stderr, "\tdefs: ", self.defs
82 print >> sys.stderr, "\tuses: ", self.uses
83 print >> sys.stderr, "\tfrees:", self.frees
84
85 def check_name(self, name):
86 """Return scope of name.
87
88 The scope of a name could be LOCAL, GLOBAL, FREE, or CELL.
89 """
90 if self.globals.has_key(name):
91 return SC_GLOBAL
92 if self.cells.has_key(name):
93 return SC_CELL
94 if self.defs.has_key(name):
95 return SC_LOCAL
96 if self.nested and (self.frees.has_key(name) or
97 self.uses.has_key(name)):
98 return SC_FREE
99 if self.nested:
100 return SC_UNKNOWN
101 else:
102 return SC_GLOBAL
103
104 def get_free_vars(self):
105 if not self.nested:
106 return ()
107 free = {}
108 free.update(self.frees)
109 for name in self.uses.keys():
110 if not (self.defs.has_key(name) or
111 self.globals.has_key(name)):
112 free[name] = 1
113 return free.keys()
114
115 def handle_children(self):
116 for child in self.children:
117 frees = child.get_free_vars()
118 globals = self.add_frees(frees)
119 for name in globals:
120 child.force_global(name)
121
122 def force_global(self, name):
123 """Force name to be global in scope.
124
125 Some child of the current node had a free reference to name.
126 When the child was processed, it was labelled a free
127 variable. Now that all its enclosing scope have been
128 processed, the name is known to be a global or builtin. So
129 walk back down the child chain and set the name to be global
130 rather than free.
131
132 Be careful to stop if a child does not think the name is
133 free.
134 """
135 self.globals[name] = 1
136 if self.frees.has_key(name):
137 del self.frees[name]
138 for child in self.children:
139 if child.check_name(name) == SC_FREE:
140 child.force_global(name)
141
142 def add_frees(self, names):
143 """Process list of free vars from nested scope.
144
145 Returns a list of names that are either 1) declared global in the
146 parent or 2) undefined in a top-level parent. In either case,
147 the nested scope should treat them as globals.
148 """
149 child_globals = []
150 for name in names:
151 sc = self.check_name(name)
152 if self.nested:
153 if sc == SC_UNKNOWN or sc == SC_FREE \
154 or isinstance(self, ClassScope):
155 self.frees[name] = 1
156 elif sc == SC_GLOBAL:
157 child_globals.append(name)
158 elif isinstance(self, FunctionScope) and sc == SC_LOCAL:
159 self.cells[name] = 1
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000160 elif sc != SC_CELL:
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000161 child_globals.append(name)
162 else:
163 if sc == SC_LOCAL:
164 self.cells[name] = 1
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000165 elif sc != SC_CELL:
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000166 child_globals.append(name)
167 return child_globals
168
169 def get_cell_vars(self):
170 return self.cells.keys()
171
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000172class ModuleScope(Scope):
173 __super_init = Scope.__init__
174
175 def __init__(self):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000176 self.__super_init("global", self)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000177
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000178class FunctionScope(Scope):
179 pass
180
181class LambdaScope(FunctionScope):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000182 __super_init = Scope.__init__
183
184 __counter = 1
185
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000186 def __init__(self, module, klass=None):
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000187 i = self.__counter
188 self.__counter += 1
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000189 self.__super_init("lambda.%d" % i, module, klass)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000190
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000191class ClassScope(Scope):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000192 __super_init = Scope.__init__
193
194 def __init__(self, name, module):
195 self.__super_init(name, module, name)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000196
197class SymbolVisitor:
198 def __init__(self):
199 self.scopes = {}
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000200 self.klass = None
201
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000202 # node that define new scopes
203
204 def visitModule(self, node):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000205 scope = self.module = self.scopes[node] = ModuleScope()
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000206 self.visit(node.node, scope)
207
208 def visitFunction(self, node, parent):
209 parent.add_def(node.name)
210 for n in node.defaults:
211 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000212 scope = FunctionScope(node.name, self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000213 if parent.nested or isinstance(parent, FunctionScope):
214 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000215 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000216 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000217 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000218 self.handle_free_vars(scope, parent)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000219
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000220 def visitLambda(self, node, parent):
221 for n in node.defaults:
222 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000223 scope = LambdaScope(self.module, self.klass)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000224 if parent.nested or isinstance(parent, FunctionScope):
225 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000226 self.scopes[node] = scope
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000227 self._do_args(scope, node.argnames)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000228 self.visit(node.code, scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000229 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000230
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000231 def _do_args(self, scope, args):
232 for name in args:
233 if type(name) == types.TupleType:
234 self._do_args(scope, name)
235 else:
236 scope.add_param(name)
237
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000238 def handle_free_vars(self, scope, parent):
239 parent.add_child(scope)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000240 scope.handle_children()
241
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000242 def visitClass(self, node, parent):
243 parent.add_def(node.name)
244 for n in node.bases:
245 self.visit(n, parent)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000246 scope = ClassScope(node.name, self.module)
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000247 if parent.nested or isinstance(parent, FunctionScope):
248 scope.nested = 1
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000249 self.scopes[node] = scope
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000250 prev = self.klass
251 self.klass = node.name
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000252 self.visit(node.code, scope)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000253 self.klass = prev
Jeremy Hylton364f9b92001-04-12 06:40:42 +0000254 self.handle_free_vars(scope, parent)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000255
256 # name can be a def or a use
257
258 # XXX a few calls and nodes expect a third "assign" arg that is
259 # true if the name is being used as an assignment. only
260 # expressions contained within statements may have the assign arg.
261
262 def visitName(self, node, scope, assign=0):
263 if assign:
264 scope.add_def(node.name)
265 else:
266 scope.add_use(node.name)
267
268 # operations that bind new names
269
270 def visitFor(self, node, scope):
271 self.visit(node.assign, scope, 1)
272 self.visit(node.list, scope)
273 self.visit(node.body, scope)
274 if node.else_:
275 self.visit(node.else_, scope)
276
277 def visitFrom(self, node, scope):
278 for name, asname in node.names:
279 if name == "*":
280 continue
281 scope.add_def(asname or name)
282
283 def visitImport(self, node, scope):
284 for name, asname in node.names:
285 i = name.find(".")
286 if i > -1:
287 name = name[:i]
288 scope.add_def(asname or name)
289
290 def visitAssName(self, node, scope, assign=1):
291 scope.add_def(node.name)
292
Jeremy Hyltoncd8a1272001-08-27 21:06:35 +0000293 def visitAssAttr(self, node, scope, assign=0):
294 self.visit(node.expr, scope, 0)
295
296 def visitSubscript(self, node, scope, assign=0):
297 self.visit(node.expr, scope, 0)
298 for n in node.subs:
299 self.visit(n, scope, 0)
300
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000301 def visitAugAssign(self, node, scope):
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000302 # If the LHS is a name, then this counts as assignment.
303 # Otherwise, it's just use.
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000304 self.visit(node.node, scope)
Jeremy Hylton5c9aad62001-04-12 07:06:25 +0000305 if isinstance(node.node, ast.Name):
306 self.visit(node.node, scope, 1) # XXX worry about this
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000307 self.visit(node.expr, scope)
308
309 def visitAssign(self, node, scope):
310 for n in node.nodes:
311 self.visit(n, scope, 1)
312 self.visit(node.expr, scope)
313
314 def visitGlobal(self, node, scope):
315 for name in node.names:
316 scope.add_global(name)
317
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000318 # prune if statements if tests are false
319
320 _const_types = types.StringType, types.IntType, types.FloatType
321
322 def visitIf(self, node, scope):
323 for test, body in node.tests:
324 if isinstance(test, ast.Const):
325 if type(test.value) in self._const_types:
326 if not test.value:
327 continue
328 self.visit(test, scope)
329 self.visit(body, scope)
330 if node.else_:
331 self.visit(node.else_, scope)
332
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000333def sort(l):
334 l = l[:]
335 l.sort()
336 return l
337
338def list_eq(l1, l2):
339 return sort(l1) == sort(l2)
340
341if __name__ == "__main__":
342 import sys
343 from compiler import parseFile, walk
344 import symtable
345
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000346 def get_names(syms):
347 return [s for s in [s.get_name() for s in syms.get_symbols()]
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000348 if not (s.startswith('_[') or s.startswith('.'))]
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000349
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000350 for file in sys.argv[1:]:
351 print file
352 f = open(file)
353 buf = f.read()
354 f.close()
355 syms = symtable.symtable(buf, file, "exec")
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000356 mod_names = get_names(syms)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000357 tree = parseFile(file)
358 s = SymbolVisitor()
359 walk(tree, s)
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000360
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000361 # compare module-level symbols
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000362 names2 = s.scopes[tree].get_names()
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000363
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000364 if not list_eq(mod_names, names2):
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000365 print
Jeremy Hylton8b966dc2001-04-09 04:35:35 +0000366 print "oops", file
367 print sort(mod_names)
368 print sort(names2)
369 sys.exit(-1)
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000370
371 d = {}
372 d.update(s.scopes)
373 del d[tree]
374 scopes = d.values()
375 del d
376
377 for s in syms.get_symbols():
378 if s.is_namespace():
379 l = [sc for sc in scopes
380 if sc.name == s.get_name()]
381 if len(l) > 1:
382 print "skipping", s.get_name()
383 else:
384 if not list_eq(get_names(s.get_namespace()),
385 l[0].get_names()):
386 print s.get_name()
Jeremy Hyltondbdb28e2001-04-09 20:11:59 +0000387 print sort(get_names(s.get_namespace()))
388 print sort(l[0].get_names())
Jeremy Hyltonf870c952001-04-09 13:57:32 +0000389 sys.exit(-1)