Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 1 | """Module symbol-table generator""" |
| 2 | |
| 3 | from compiler import ast |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 4 | import types |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 5 | |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 6 | MANGLE_LEN = 256 |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 7 | |
| 8 | class Scope: |
| 9 | # XXX how much information do I need about each name? |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 10 | def __init__(self, name, module, klass=None): |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 11 | self.name = name |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 12 | self.module = module |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 13 | self.defs = {} |
| 14 | self.uses = {} |
| 15 | self.globals = {} |
| 16 | self.params = {} |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 17 | self.children = [] |
| 18 | self.klass = None |
| 19 | if klass is not None: |
| 20 | for i in range(len(klass)): |
| 21 | if klass[i] != '_': |
| 22 | self.klass = klass[i:] |
| 23 | break |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 24 | |
| 25 | def __repr__(self): |
| 26 | return "<%s: %s>" % (self.__class__.__name__, self.name) |
| 27 | |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 28 | def mangle(self, name): |
| 29 | if self.klass is None: |
| 30 | return name |
| 31 | if not name.startswith('__'): |
| 32 | return name |
| 33 | if len(name) + 2 >= MANGLE_LEN: |
| 34 | return name |
| 35 | if name.endswith('__'): |
| 36 | return name |
| 37 | return "_%s%s" % (self.klass, name) |
| 38 | |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 39 | def add_def(self, name): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 40 | self.defs[self.mangle(name)] = 1 |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 41 | |
| 42 | def add_use(self, name): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 43 | self.uses[self.mangle(name)] = 1 |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 44 | |
| 45 | def add_global(self, name): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 46 | name = self.mangle(name) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 47 | if self.uses.has_key(name) or self.defs.has_key(name): |
| 48 | pass # XXX warn about global following def/use |
| 49 | if self.params.has_key(name): |
| 50 | raise SyntaxError, "%s in %s is global and parameter" % \ |
| 51 | (name, self.name) |
| 52 | self.globals[name] = 1 |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 53 | self.module.add_def(name) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 54 | |
| 55 | def add_param(self, name): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 56 | name = self.mangle(name) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 57 | self.defs[name] = 1 |
| 58 | self.params[name] = 1 |
| 59 | |
| 60 | def get_names(self): |
| 61 | d = {} |
| 62 | d.update(self.defs) |
| 63 | d.update(self.uses) |
| 64 | return d.keys() |
| 65 | |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 66 | def add_child(self, child): |
| 67 | self.children.append(child) |
| 68 | |
| 69 | def get_children(self): |
| 70 | return self.children |
| 71 | |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 72 | class ModuleScope(Scope): |
| 73 | __super_init = Scope.__init__ |
| 74 | |
| 75 | def __init__(self): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 76 | self.__super_init("global", self) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 77 | |
| 78 | class LambdaScope(Scope): |
| 79 | __super_init = Scope.__init__ |
| 80 | |
| 81 | __counter = 1 |
| 82 | |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 83 | def __init__(self, module, klass=None): |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 84 | i = self.__counter |
| 85 | self.__counter += 1 |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 86 | self.__super_init("lambda.%d" % i, module, klass) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 87 | |
| 88 | class FunctionScope(Scope): |
| 89 | pass |
| 90 | |
| 91 | class ClassScope(Scope): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 92 | __super_init = Scope.__init__ |
| 93 | |
| 94 | def __init__(self, name, module): |
| 95 | self.__super_init(name, module, name) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 96 | |
| 97 | class SymbolVisitor: |
| 98 | def __init__(self): |
| 99 | self.scopes = {} |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 100 | self.klass = None |
| 101 | |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 102 | # node that define new scopes |
| 103 | |
| 104 | def visitModule(self, node): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 105 | scope = self.module = self.scopes[node] = ModuleScope() |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 106 | self.visit(node.node, scope) |
| 107 | |
| 108 | def visitFunction(self, node, parent): |
| 109 | parent.add_def(node.name) |
| 110 | for n in node.defaults: |
| 111 | self.visit(n, parent) |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 112 | scope = FunctionScope(node.name, self.module, self.klass) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 113 | self.scopes[node] = scope |
| 114 | for name in node.argnames: |
| 115 | scope.add_param(name) |
| 116 | self.visit(node.code, scope) |
| 117 | |
| 118 | def visitLambda(self, node, parent): |
| 119 | for n in node.defaults: |
| 120 | self.visit(n, parent) |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 121 | scope = LambdaScope(self.module, self.klass) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 122 | self.scopes[node] = scope |
| 123 | for name in node.argnames: |
| 124 | scope.add_param(name) |
| 125 | self.visit(node.code, scope) |
| 126 | |
| 127 | def visitClass(self, node, parent): |
| 128 | parent.add_def(node.name) |
| 129 | for n in node.bases: |
| 130 | self.visit(n, parent) |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 131 | scope = ClassScope(node.name, self.module) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 132 | self.scopes[node] = scope |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 133 | prev = self.klass |
| 134 | self.klass = node.name |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 135 | self.visit(node.code, scope) |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 136 | self.klass = prev |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 137 | |
| 138 | # name can be a def or a use |
| 139 | |
| 140 | # XXX a few calls and nodes expect a third "assign" arg that is |
| 141 | # true if the name is being used as an assignment. only |
| 142 | # expressions contained within statements may have the assign arg. |
| 143 | |
| 144 | def visitName(self, node, scope, assign=0): |
| 145 | if assign: |
| 146 | scope.add_def(node.name) |
| 147 | else: |
| 148 | scope.add_use(node.name) |
| 149 | |
| 150 | # operations that bind new names |
| 151 | |
| 152 | def visitFor(self, node, scope): |
| 153 | self.visit(node.assign, scope, 1) |
| 154 | self.visit(node.list, scope) |
| 155 | self.visit(node.body, scope) |
| 156 | if node.else_: |
| 157 | self.visit(node.else_, scope) |
| 158 | |
| 159 | def visitFrom(self, node, scope): |
| 160 | for name, asname in node.names: |
| 161 | if name == "*": |
| 162 | continue |
| 163 | scope.add_def(asname or name) |
| 164 | |
| 165 | def visitImport(self, node, scope): |
| 166 | for name, asname in node.names: |
| 167 | i = name.find(".") |
| 168 | if i > -1: |
| 169 | name = name[:i] |
| 170 | scope.add_def(asname or name) |
| 171 | |
| 172 | def visitAssName(self, node, scope, assign=1): |
| 173 | scope.add_def(node.name) |
| 174 | |
| 175 | def visitAugAssign(self, node, scope): |
| 176 | # basically, the node is referenced and defined by the same expr |
| 177 | self.visit(node.node, scope) |
| 178 | self.visit(node.node, scope, 1) |
| 179 | self.visit(node.expr, scope) |
| 180 | |
| 181 | def visitAssign(self, node, scope): |
| 182 | for n in node.nodes: |
| 183 | self.visit(n, scope, 1) |
| 184 | self.visit(node.expr, scope) |
| 185 | |
| 186 | def visitGlobal(self, node, scope): |
| 187 | for name in node.names: |
| 188 | scope.add_global(name) |
| 189 | |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 190 | # prune if statements if tests are false |
| 191 | |
| 192 | _const_types = types.StringType, types.IntType, types.FloatType |
| 193 | |
| 194 | def visitIf(self, node, scope): |
| 195 | for test, body in node.tests: |
| 196 | if isinstance(test, ast.Const): |
| 197 | if type(test.value) in self._const_types: |
| 198 | if not test.value: |
| 199 | continue |
| 200 | self.visit(test, scope) |
| 201 | self.visit(body, scope) |
| 202 | if node.else_: |
| 203 | self.visit(node.else_, scope) |
| 204 | |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 205 | def sort(l): |
| 206 | l = l[:] |
| 207 | l.sort() |
| 208 | return l |
| 209 | |
| 210 | def list_eq(l1, l2): |
| 211 | return sort(l1) == sort(l2) |
| 212 | |
| 213 | if __name__ == "__main__": |
| 214 | import sys |
| 215 | from compiler import parseFile, walk |
| 216 | import symtable |
| 217 | |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 218 | def get_names(syms): |
| 219 | return [s for s in [s.get_name() for s in syms.get_symbols()] |
| 220 | if not s.startswith('_[')] |
| 221 | |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 222 | for file in sys.argv[1:]: |
| 223 | print file |
| 224 | f = open(file) |
| 225 | buf = f.read() |
| 226 | f.close() |
| 227 | syms = symtable.symtable(buf, file, "exec") |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 228 | mod_names = get_names(syms) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 229 | tree = parseFile(file) |
| 230 | s = SymbolVisitor() |
| 231 | walk(tree, s) |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 232 | |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 233 | # compare module-level symbols |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 234 | names2 = s.scopes[tree].get_names() |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 235 | |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 236 | if not list_eq(mod_names, names2): |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 237 | print |
Jeremy Hylton | 8b966dc | 2001-04-09 04:35:35 +0000 | [diff] [blame] | 238 | print "oops", file |
| 239 | print sort(mod_names) |
| 240 | print sort(names2) |
| 241 | sys.exit(-1) |
Jeremy Hylton | f870c95 | 2001-04-09 13:57:32 +0000 | [diff] [blame^] | 242 | |
| 243 | d = {} |
| 244 | d.update(s.scopes) |
| 245 | del d[tree] |
| 246 | scopes = d.values() |
| 247 | del d |
| 248 | |
| 249 | for s in syms.get_symbols(): |
| 250 | if s.is_namespace(): |
| 251 | l = [sc for sc in scopes |
| 252 | if sc.name == s.get_name()] |
| 253 | if len(l) > 1: |
| 254 | print "skipping", s.get_name() |
| 255 | else: |
| 256 | if not list_eq(get_names(s.get_namespace()), |
| 257 | l[0].get_names()): |
| 258 | print s.get_name() |
| 259 | print get_names(s.get_namespace()) |
| 260 | print l[0].get_names() |
| 261 | sys.exit(-1) |