blob: 5b22ddf8b686e6e579c0e2fb6e72acabf16d3ef5 [file] [log] [blame]
eli.benderskyd5ba3452011-02-18 21:32:47 +02001#-----------------------------------------------------------------
2# pycparser: c-to-c.py
3#
eli.benderskyafcfaac2011-02-25 16:46:01 +02004# Example of a C code generator from pycparser AST nodes, serving
5# as a simplistic translator from C to AST and back to C.
eli.benderskyb36f5bb2011-03-06 07:22:33 +02006# Note: at this stage, the example is "alpha release" and considered
7# experimental. Please file any bugs you find in the Issues page on pycparser's
8# website.
eli.benderskyd5ba3452011-02-18 21:32:47 +02009#
10# Copyright (C) 2008-2011, Eli Bendersky
eli.bendersky84a6a632011-04-29 09:00:43 +030011# License: BSD
eli.benderskyd5ba3452011-02-18 21:32:47 +020012#-----------------------------------------------------------------
13from __future__ import print_function
14import sys
15
16# This is not required if you've installed pycparser into
17# your site-packages/ with setup.py
18#
19sys.path.insert(0, '..')
20
21from pycparser import c_parser, c_ast, parse_file
22
23
eli.benderskyafcfaac2011-02-25 16:46:01 +020024class CGenerator(object):
25 """ Uses the same visitor pattern as c_ast.NodeVisitor, but modified to
26 return a value from each visit method, using string accumulation in
27 generic_visit.
28 """
eli.benderskyd5ba3452011-02-18 21:32:47 +020029 def __init__(self):
30 self.output = ''
31
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020032 # Statements start with indentation of self.indent_level spaces, using
33 # the _make_indent method
eli.benderskyafcfaac2011-02-25 16:46:01 +020034 #
35 self.indent_level = 0
36
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020037 def _make_indent(self):
38 return ' ' * self.indent_level
39
eli.benderskyafcfaac2011-02-25 16:46:01 +020040 def visit(self, node):
41 method = 'visit_' + node.__class__.__name__
42 return getattr(self, method, self.generic_visit)(node)
43
44 def generic_visit(self, node):
eli.benderskydcb859c2011-03-05 17:02:43 +020045 #~ print('generic:', type(node))
eli.benderskyafcfaac2011-02-25 16:46:01 +020046 if node is None:
47 return ''
48 else:
49 return ''.join(self.visit(c) for c in node.children())
eli.benderskyd5ba3452011-02-18 21:32:47 +020050
51 def visit_Constant(self, n):
eli.benderskyafcfaac2011-02-25 16:46:01 +020052 return n.value
53
eli.benderskyd5ba3452011-02-18 21:32:47 +020054 def visit_ID(self, n):
eli.benderskyafcfaac2011-02-25 16:46:01 +020055 return n.name
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020056
57 def visit_ArrayRef(self, n):
eli.benderskyb36f5bb2011-03-06 07:22:33 +020058 arrref = self._parenthesize_unless_simple(n.name)
59 return arrref + '[' + self.visit(n.subscript) + ']'
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020060
eli.bendersky724b1cc2011-03-05 10:45:08 +020061 def visit_StructRef(self, n):
eli.benderskyb36f5bb2011-03-06 07:22:33 +020062 sref = self._parenthesize_unless_simple(n.name)
63 return sref + n.type + self.visit(n.field)
eli.bendersky724b1cc2011-03-05 10:45:08 +020064
65 def visit_FuncCall(self, n):
eli.benderskyb36f5bb2011-03-06 07:22:33 +020066 fref = self._parenthesize_unless_simple(n.name)
67 return fref + '(' + self.visit(n.args) + ')'
eli.bendersky724b1cc2011-03-05 10:45:08 +020068
eli.benderskyae36e962011-02-27 08:25:05 +020069 def visit_UnaryOp(self, n):
eli.benderskyb36f5bb2011-03-06 07:22:33 +020070 operand = self._parenthesize_unless_simple(n.expr)
eli.benderskyae36e962011-02-27 08:25:05 +020071 if n.op == 'p++':
eli.benderskyb36f5bb2011-03-06 07:22:33 +020072 return '%s++' % operand
eli.benderskyd4a99752011-05-26 07:04:19 +030073 elif n.op == 'p--':
74 return '%s--' % operand
eli.bendersky8348a9d2011-05-26 07:01:43 +030075 elif n.op == 'sizeof':
76 # Always parenthesize the argument of sizeof since it can be
77 # a name.
78 return 'sizeof(%s)' % self.visit(n.expr)
eli.benderskyae36e962011-02-27 08:25:05 +020079 else:
eli.benderskyb36f5bb2011-03-06 07:22:33 +020080 return '%s%s' % (n.op, operand)
eli.benderskyae36e962011-02-27 08:25:05 +020081
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020082 def visit_BinaryOp(self, n):
eli.benderskyb36f5bb2011-03-06 07:22:33 +020083 lval_str = self._parenthesize_if(n.left,
84 lambda d: not self._is_simple_node(d))
85 rval_str = self._parenthesize_if(n.right,
86 lambda d: not self._is_simple_node(d))
eli.bendersky724b1cc2011-03-05 10:45:08 +020087 return '%s %s %s' % (lval_str, n.op, rval_str)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020088
89 def visit_Assignment(self, n):
eli.bendersky724b1cc2011-03-05 10:45:08 +020090 rval_str = self._parenthesize_if(
91 n.rvalue,
92 lambda n: isinstance(n, c_ast.Assignment))
93 return '%s %s %s' % (self.visit(n.lvalue), n.op, rval_str)
eli.benderskyd5ba3452011-02-18 21:32:47 +020094
95 def visit_IdentifierType(self, n):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020096 return ' '.join(n.names)
eli.benderskyd5ba3452011-02-18 21:32:47 +020097
eli.benderskyabb96e62011-03-05 16:30:05 +020098 def visit_Decl(self, n, no_type=False):
99 # no_type is used when a Decl is part of a DeclList, where the type is
100 # explicitly only for the first delaration in a list.
101 #
102 s = n.name if no_type else self._generate_decl(n)
eli.benderskyafcfaac2011-02-25 16:46:01 +0200103 if n.bitsize: s += ' : ' + self.visit(n.bitsize)
eli.benderskydcb859c2011-03-05 17:02:43 +0200104 if n.init:
105 if isinstance(n.init, c_ast.ExprList):
106 s += ' = {' + self.visit(n.init) + '}'
107 else:
108 s += ' = ' + self.visit(n.init)
eli.benderskyafcfaac2011-02-25 16:46:01 +0200109 return s
eli.benderskyd5ba3452011-02-18 21:32:47 +0200110
eli.benderskyabb96e62011-03-05 16:30:05 +0200111 def visit_DeclList(self, n):
112 s = self.visit(n.decls[0])
113 if len(n.decls) > 1:
114 s += ', ' + ', '.join(self.visit_Decl(decl, no_type=True)
115 for decl in n.decls[1:])
116 return s
117
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200118 def visit_Typedef(self, n):
119 s = ''
120 if n.storage: s += ' '.join(n.storage) + ' '
121 s += self._generate_type(n.type)
122 return s
123
124 def visit_Cast(self, n):
eli.bendersky6ed80a62011-05-09 18:22:33 +0300125 s = '(' + self._generate_type(n.to_type) + ')'
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200126 return s + ' ' + self.visit(n.expr)
127
eli.bendersky724b1cc2011-03-05 10:45:08 +0200128 def visit_ExprList(self, n):
eli.bendersky8e6c5862011-05-20 12:35:08 +0300129 visited_subexprs = []
130 for expr in n.exprs:
131 if isinstance(expr, c_ast.ExprList):
132 visited_subexprs.append('{' + self.visit(expr) + '}')
133 else:
134 visited_subexprs.append(self.visit(expr))
135 return ', '.join(visited_subexprs)
eli.bendersky724b1cc2011-03-05 10:45:08 +0200136
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200137 def visit_Enum(self, n):
138 s = 'enum'
139 if n.name: s += ' ' + n.name
140 if n.values:
141 s += ' {'
142 for i, enumerator in enumerate(n.values.enumerators):
143 s += enumerator.name
144 if enumerator.value:
145 s += ' = ' + self.visit(enumerator.value)
146 if i != len(n.values.enumerators) - 1:
147 s += ', '
148 s += '}'
149 return s
150
151 def visit_Struct(self, n):
152 s = 'struct'
153 if n.name: s += ' ' + n.name
154 if n.decls:
155 s += ' { \n'
156 for decl in n.decls:
157 s += ' ' + self.visit(decl) + ';\n'
158 s += '}'
159 return s
160
eli.benderskyafcfaac2011-02-25 16:46:01 +0200161 def visit_FuncDef(self, n):
162 decl = self.visit(n.decl)
163 self.indent_level = 0
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200164 # The body is a Compound node
eli.benderskyafcfaac2011-02-25 16:46:01 +0200165 body = self.visit(n.body)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200166 return decl + '\n' + body + '\n'
167
168 def visit_FileAST(self, n):
169 s = ''
170 for ext in n.ext:
171 if isinstance(ext, c_ast.FuncDef):
172 s += self.visit(ext)
173 else:
174 s += self.visit(ext) + ';\n'
175 return s
176
177 def visit_Compound(self, n):
178 s = self._make_indent() + '{\n'
179 self.indent_level += 2
eli.benderskycad1cfd2011-05-26 06:56:27 +0300180 if n.block_items:
181 s += ''.join(self._generate_stmt(stmt) for stmt in n.block_items)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200182 self.indent_level -= 2
183 s += self._make_indent() + '}\n'
184 return s
185
186 def visit_ParamList(self, n):
187 return ', '.join(self.visit(param) for param in n.params)
188
189 def visit_Return(self, n):
190 s = 'return'
191 if n.expr: s += ' ' + self.visit(n.expr)
192 return s + ';'
193
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200194 def visit_Break(self, n):
195 return 'break;'
196
197 def visit_Continue(self, n):
198 return 'continue;'
eli.bendersky724b1cc2011-03-05 10:45:08 +0200199
200 def visit_TernaryOp(self, n):
201 s = self.visit(n.cond) + ' ? '
202 s += self.visit(n.iftrue) + ' : '
203 s += self.visit(n.iffalse)
204 return s
205
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200206 def visit_If(self, n):
207 s = 'if ('
208 if n.cond: s += self.visit(n.cond)
209 s += ')\n'
210 s += self._generate_stmt(n.iftrue, add_indent=True)
211 if n.iffalse:
212 s += self._make_indent() + 'else\n'
213 s += self._generate_stmt(n.iffalse, add_indent=True)
214 return s
215
eli.benderskyae36e962011-02-27 08:25:05 +0200216 def visit_For(self, n):
217 s = 'for ('
218 if n.init: s += self.visit(n.init)
219 s += ';'
220 if n.cond: s += ' ' + self.visit(n.cond)
221 s += ';'
222 if n.next: s += ' ' + self.visit(n.next)
223 s += ')\n'
224 s += self._generate_stmt(n.stmt, add_indent=True)
225 return s
226
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200227 def visit_While(self, n):
228 s = 'while ('
229 if n.cond: s += self.visit(n.cond)
230 s += ')\n'
231 s += self._generate_stmt(n.stmt, add_indent=True)
232 return s
233
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200234 def visit_DoWhile(self, n):
235 s = 'do\n'
236 s += self._generate_stmt(n.stmt, add_indent=True)
237 s += self._make_indent() + 'while ('
238 if n.cond: s += self.visit(n.cond)
239 s += ');'
240 return s
241
eli.benderskyabb96e62011-03-05 16:30:05 +0200242 def visit_Switch(self, n):
243 s = 'switch (' + self.visit(n.cond) + ')\n'
244 s += self._generate_stmt(n.stmt, add_indent=True)
245 return s
246
247 def visit_Case(self, n):
248 s = 'case ' + self.visit(n.expr) + ':\n'
249 s += self._generate_stmt(n.stmt, add_indent=True)
250 return s
251
252 def visit_Default(self, n):
253 return 'default:\n' + self._generate_stmt(n.stmt, add_indent=True)
254
255 def visit_Label(self, n):
256 return n.name + ':\n' + self._generate_stmt(n.stmt)
257
258 def visit_Goto(self, n):
259 return 'goto ' + n.name + ';'
260
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200261 def visit_EllipsisParam(self, n):
262 return '...'
263
eli.benderskyabb96e62011-03-05 16:30:05 +0200264 def visit_Struct(self, n):
265 return self._generate_struct_union(n, 'struct')
266
eli.bendersky8e6c5862011-05-20 12:35:08 +0300267 def visit_Typename(self, n):
268 return self._generate_type(n.type)
269
eli.benderskyabb96e62011-03-05 16:30:05 +0200270 def visit_Union(self, n):
271 return self._generate_struct_union(n, 'union')
272
eli.benderskydcb859c2011-03-05 17:02:43 +0200273 def visit_NamedInitializer(self, n):
274 s = ''
275 for name in n.name:
276 if isinstance(name, c_ast.ID):
277 s += '.' + name.name
278 elif isinstance(name, c_ast.Constant):
279 s += '[' + name.value + ']'
280 s += ' = ' + self.visit(n.expr)
281 return s
282
eli.benderskyabb96e62011-03-05 16:30:05 +0200283 def _generate_struct_union(self, n, name):
284 """ Generates code for structs and unions. name should be either
285 'struct' or union.
286 """
eli.benderskydcb859c2011-03-05 17:02:43 +0200287 s = name + ' ' + (n.name or '')
288 if n.decls:
289 s += '\n'
290 s += self._make_indent()
291 self.indent_level += 2
292 s += '{\n'
293 for decl in n.decls:
294 s += self._generate_stmt(decl)
295 self.indent_level -= 2
296 s += self._make_indent() + '}'
eli.benderskyabb96e62011-03-05 16:30:05 +0200297 return s
298
eli.benderskyae36e962011-02-27 08:25:05 +0200299 def _generate_stmt(self, n, add_indent=False):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200300 """ Generation from a statement node. This method exists as a wrapper
301 for individual visit_* methods to handle different treatment of
302 some statements in this context.
303 """
304 typ = type(n)
eli.benderskyae36e962011-02-27 08:25:05 +0200305 if add_indent: self.indent_level += 2
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200306 indent = self._make_indent()
eli.benderskyae36e962011-02-27 08:25:05 +0200307 if add_indent: self.indent_level -= 2
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200308
eli.bendersky724b1cc2011-03-05 10:45:08 +0200309 if typ in (
310 c_ast.Decl, c_ast.Assignment, c_ast.Cast, c_ast.UnaryOp,
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200311 c_ast.BinaryOp, c_ast.TernaryOp, c_ast.FuncCall, c_ast.ArrayRef,
312 c_ast.StructRef):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200313 # These can also appear in an expression context so no semicolon
314 # is added to them automatically
315 #
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200316 return indent + self.visit(n) + ';\n'
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200317 elif typ in (c_ast.Compound,):
318 # No extra indentation required before the opening brace of a
319 # compound - because it consists of multiple lines it has to
320 # compute its own indentation.
321 #
eli.benderskyae36e962011-02-27 08:25:05 +0200322 return self.visit(n)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200323 else:
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200324 return indent + self.visit(n) + '\n'
eli.benderskyafcfaac2011-02-25 16:46:01 +0200325
eli.benderskyd5ba3452011-02-18 21:32:47 +0200326 def _generate_decl(self, n):
327 """ Generation from a Decl node.
328 """
eli.benderskyafcfaac2011-02-25 16:46:01 +0200329 s = ''
330 if n.funcspec: s = ' '.join(n.funcspec) + ' '
331 if n.storage: s += ' '.join(n.storage) + ' '
332 s += self._generate_type(n.type)
333 return s
eli.benderskyd5ba3452011-02-18 21:32:47 +0200334
335 def _generate_type(self, n, modifiers=[]):
336 """ Recursive generation from a type node. n is the type node.
eli.benderskyafcfaac2011-02-25 16:46:01 +0200337 modifiers collects the PtrDecl, ArrayDecl and FuncDecl modifiers
338 encountered on the way down to a TypeDecl, to allow proper
339 generation from it.
eli.benderskyd5ba3452011-02-18 21:32:47 +0200340 """
341 typ = type(n)
342 #~ print(n, modifiers)
343
344 if typ == c_ast.TypeDecl:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200345 s = ''
346 if n.quals: s += ' '.join(n.quals) + ' '
347 s += self.visit(n.type)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200348
349 nstr = n.declname if n.declname else ''
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200350 # Resolve modifiers.
eli.benderskyafcfaac2011-02-25 16:46:01 +0200351 # Wrap in parens to distinguish pointer to array and pointer to
352 # function syntax.
eli.benderskyd5ba3452011-02-18 21:32:47 +0200353 #
354 for i, modifier in enumerate(modifiers):
355 if isinstance(modifier, c_ast.ArrayDecl):
356 if (i != 0 and isinstance(modifiers[i - 1], c_ast.PtrDecl)):
357 nstr = '(' + nstr + ')'
eli.benderskyafcfaac2011-02-25 16:46:01 +0200358 nstr += '[' + self.visit(modifier.dim) + ']'
359 elif isinstance(modifier, c_ast.FuncDecl):
360 if (i != 0 and isinstance(modifiers[i - 1], c_ast.PtrDecl)):
361 nstr = '(' + nstr + ')'
362 nstr += '(' + self.visit(modifier.args) + ')'
eli.benderskyd5ba3452011-02-18 21:32:47 +0200363 elif isinstance(modifier, c_ast.PtrDecl):
364 nstr = '*' + nstr
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200365 s += ' ' + nstr
eli.benderskyafcfaac2011-02-25 16:46:01 +0200366 return s
eli.bendersky6ed80a62011-05-09 18:22:33 +0300367 elif typ == c_ast.Decl:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200368 return self._generate_decl(n.type)
eli.bendersky6ed80a62011-05-09 18:22:33 +0300369 elif typ == c_ast.Typename:
370 return self._generate_type(n.type)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200371 elif typ == c_ast.IdentifierType:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200372 return ' '.join(n.names) + ' '
373 elif typ in (c_ast.ArrayDecl, c_ast.PtrDecl, c_ast.FuncDecl):
374 return self._generate_type(n.type, modifiers + [n])
eli.benderskyabb96e62011-03-05 16:30:05 +0200375 else:
376 return self.visit(n)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200377
eli.bendersky724b1cc2011-03-05 10:45:08 +0200378 def _parenthesize_if(self, n, condition):
379 """ Visits 'n' and returns its string representation, parenthesized
380 if the condition function applied to the node returns True.
381 """
382 s = self.visit(n)
383 if condition(n):
384 return '(' + s + ')'
385 else:
386 return s
387
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200388 def _parenthesize_unless_simple(self, n):
389 """ Common use case for _parenthesize_if
390 """
391 return self._parenthesize_if(n, lambda d: not self._is_simple_node(d))
392
393 def _is_simple_node(self, n):
394 """ Returns True for nodes that are "simple" - i.e. nodes that always
395 have higher precedence than operators.
396 """
397 return isinstance(n,( c_ast.Constant, c_ast.ID, c_ast.ArrayRef,
398 c_ast.StructRef, c_ast.FuncCall))
399
eli.benderskyd5ba3452011-02-18 21:32:47 +0200400
401def translate_to_c(filename):
402 ast = parse_file(filename, use_cpp=True)
403 generator = CGenerator()
eli.benderskyafcfaac2011-02-25 16:46:01 +0200404 print(generator.visit(ast))
eli.benderskyd5ba3452011-02-18 21:32:47 +0200405
406
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200407def zz_test_translate():
408 # internal use
eli.bendersky8e6c5862011-05-20 12:35:08 +0300409 src = r'''
eli.bendersky8348a9d2011-05-26 07:01:43 +0300410int main(void)
411{
eli.benderskyd4a99752011-05-26 07:04:19 +0300412 int a;
413 int b = a++;
414 int c = ++a;
415 int d = a--;
416 int e = --a;
eli.bendersky8348a9d2011-05-26 07:01:43 +0300417}
418'''
eli.bendersky8e6c5862011-05-20 12:35:08 +0300419 parser = c_parser.CParser()
420 ast = parser.parse(src)
421 ast.show()
422 generator = CGenerator()
423
424 print(generator.visit(ast))
425
426 # tracing the generator for debugging
427 #~ import trace
428 #~ tr = trace.Trace(countcallers=1)
429 #~ tr.runfunc(generator.visit, ast)
430 #~ tr.results().write_results()
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200431
432
433#------------------------------------------------------------------------------
434if __name__ == "__main__":
eli.bendersky6ed80a62011-05-09 18:22:33 +0300435 zz_test_translate()
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200436 if len(sys.argv) > 1:
437 translate_to_c(sys.argv[1])
438 else:
eli.benderskyd5ba3452011-02-18 21:32:47 +0200439 print("Please provide a filename as argument")
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200440