blob: a292c36a6f9e67bec9cceec3abeee8b15e6ca264 [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.benderskyae36e962011-02-27 08:25:05 +020073 else:
eli.benderskyb36f5bb2011-03-06 07:22:33 +020074 return '%s%s' % (n.op, operand)
eli.benderskyae36e962011-02-27 08:25:05 +020075
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020076 def visit_BinaryOp(self, n):
eli.benderskyb36f5bb2011-03-06 07:22:33 +020077 lval_str = self._parenthesize_if(n.left,
78 lambda d: not self._is_simple_node(d))
79 rval_str = self._parenthesize_if(n.right,
80 lambda d: not self._is_simple_node(d))
eli.bendersky724b1cc2011-03-05 10:45:08 +020081 return '%s %s %s' % (lval_str, n.op, rval_str)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020082
83 def visit_Assignment(self, n):
eli.bendersky724b1cc2011-03-05 10:45:08 +020084 rval_str = self._parenthesize_if(
85 n.rvalue,
86 lambda n: isinstance(n, c_ast.Assignment))
87 return '%s %s %s' % (self.visit(n.lvalue), n.op, rval_str)
eli.benderskyd5ba3452011-02-18 21:32:47 +020088
89 def visit_IdentifierType(self, n):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +020090 return ' '.join(n.names)
eli.benderskyd5ba3452011-02-18 21:32:47 +020091
eli.benderskyabb96e62011-03-05 16:30:05 +020092 def visit_Decl(self, n, no_type=False):
93 # no_type is used when a Decl is part of a DeclList, where the type is
94 # explicitly only for the first delaration in a list.
95 #
96 s = n.name if no_type else self._generate_decl(n)
eli.benderskyafcfaac2011-02-25 16:46:01 +020097 if n.bitsize: s += ' : ' + self.visit(n.bitsize)
eli.benderskydcb859c2011-03-05 17:02:43 +020098 if n.init:
99 if isinstance(n.init, c_ast.ExprList):
100 s += ' = {' + self.visit(n.init) + '}'
101 else:
102 s += ' = ' + self.visit(n.init)
eli.benderskyafcfaac2011-02-25 16:46:01 +0200103 return s
eli.benderskyd5ba3452011-02-18 21:32:47 +0200104
eli.benderskyabb96e62011-03-05 16:30:05 +0200105 def visit_DeclList(self, n):
106 s = self.visit(n.decls[0])
107 if len(n.decls) > 1:
108 s += ', ' + ', '.join(self.visit_Decl(decl, no_type=True)
109 for decl in n.decls[1:])
110 return s
111
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200112 def visit_Typedef(self, n):
113 s = ''
114 if n.storage: s += ' '.join(n.storage) + ' '
115 s += self._generate_type(n.type)
116 return s
117
118 def visit_Cast(self, n):
119 s = '(' + self.visit(n.to_type) + ')'
120 return s + ' ' + self.visit(n.expr)
121
eli.bendersky724b1cc2011-03-05 10:45:08 +0200122 def visit_ExprList(self, n):
123 return ', '.join(self.visit(expr) for expr in n.exprs)
124
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200125 def visit_Enum(self, n):
126 s = 'enum'
127 if n.name: s += ' ' + n.name
128 if n.values:
129 s += ' {'
130 for i, enumerator in enumerate(n.values.enumerators):
131 s += enumerator.name
132 if enumerator.value:
133 s += ' = ' + self.visit(enumerator.value)
134 if i != len(n.values.enumerators) - 1:
135 s += ', '
136 s += '}'
137 return s
138
139 def visit_Struct(self, n):
140 s = 'struct'
141 if n.name: s += ' ' + n.name
142 if n.decls:
143 s += ' { \n'
144 for decl in n.decls:
145 s += ' ' + self.visit(decl) + ';\n'
146 s += '}'
147 return s
148
eli.benderskyafcfaac2011-02-25 16:46:01 +0200149 def visit_FuncDef(self, n):
150 decl = self.visit(n.decl)
151 self.indent_level = 0
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200152 # The body is a Compound node
eli.benderskyafcfaac2011-02-25 16:46:01 +0200153 body = self.visit(n.body)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200154 return decl + '\n' + body + '\n'
155
156 def visit_FileAST(self, n):
157 s = ''
158 for ext in n.ext:
159 if isinstance(ext, c_ast.FuncDef):
160 s += self.visit(ext)
161 else:
162 s += self.visit(ext) + ';\n'
163 return s
164
165 def visit_Compound(self, n):
166 s = self._make_indent() + '{\n'
167 self.indent_level += 2
168 s += ''.join(self._generate_stmt(stmt) for stmt in n.block_items)
169 self.indent_level -= 2
170 s += self._make_indent() + '}\n'
171 return s
172
173 def visit_ParamList(self, n):
174 return ', '.join(self.visit(param) for param in n.params)
175
176 def visit_Return(self, n):
177 s = 'return'
178 if n.expr: s += ' ' + self.visit(n.expr)
179 return s + ';'
180
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200181 def visit_Break(self, n):
182 return 'break;'
183
184 def visit_Continue(self, n):
185 return 'continue;'
eli.bendersky724b1cc2011-03-05 10:45:08 +0200186
187 def visit_TernaryOp(self, n):
188 s = self.visit(n.cond) + ' ? '
189 s += self.visit(n.iftrue) + ' : '
190 s += self.visit(n.iffalse)
191 return s
192
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200193 def visit_If(self, n):
194 s = 'if ('
195 if n.cond: s += self.visit(n.cond)
196 s += ')\n'
197 s += self._generate_stmt(n.iftrue, add_indent=True)
198 if n.iffalse:
199 s += self._make_indent() + 'else\n'
200 s += self._generate_stmt(n.iffalse, add_indent=True)
201 return s
202
eli.benderskyae36e962011-02-27 08:25:05 +0200203 def visit_For(self, n):
204 s = 'for ('
205 if n.init: s += self.visit(n.init)
206 s += ';'
207 if n.cond: s += ' ' + self.visit(n.cond)
208 s += ';'
209 if n.next: s += ' ' + self.visit(n.next)
210 s += ')\n'
211 s += self._generate_stmt(n.stmt, add_indent=True)
212 return s
213
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200214 def visit_While(self, n):
215 s = 'while ('
216 if n.cond: s += self.visit(n.cond)
217 s += ')\n'
218 s += self._generate_stmt(n.stmt, add_indent=True)
219 return s
220
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200221 def visit_DoWhile(self, n):
222 s = 'do\n'
223 s += self._generate_stmt(n.stmt, add_indent=True)
224 s += self._make_indent() + 'while ('
225 if n.cond: s += self.visit(n.cond)
226 s += ');'
227 return s
228
eli.benderskyabb96e62011-03-05 16:30:05 +0200229 def visit_Switch(self, n):
230 s = 'switch (' + self.visit(n.cond) + ')\n'
231 s += self._generate_stmt(n.stmt, add_indent=True)
232 return s
233
234 def visit_Case(self, n):
235 s = 'case ' + self.visit(n.expr) + ':\n'
236 s += self._generate_stmt(n.stmt, add_indent=True)
237 return s
238
239 def visit_Default(self, n):
240 return 'default:\n' + self._generate_stmt(n.stmt, add_indent=True)
241
242 def visit_Label(self, n):
243 return n.name + ':\n' + self._generate_stmt(n.stmt)
244
245 def visit_Goto(self, n):
246 return 'goto ' + n.name + ';'
247
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200248 def visit_EllipsisParam(self, n):
249 return '...'
250
eli.benderskyabb96e62011-03-05 16:30:05 +0200251 def visit_Struct(self, n):
252 return self._generate_struct_union(n, 'struct')
253
254 def visit_Union(self, n):
255 return self._generate_struct_union(n, 'union')
256
eli.benderskydcb859c2011-03-05 17:02:43 +0200257 def visit_NamedInitializer(self, n):
258 s = ''
259 for name in n.name:
260 if isinstance(name, c_ast.ID):
261 s += '.' + name.name
262 elif isinstance(name, c_ast.Constant):
263 s += '[' + name.value + ']'
264 s += ' = ' + self.visit(n.expr)
265 return s
266
eli.benderskyabb96e62011-03-05 16:30:05 +0200267 def _generate_struct_union(self, n, name):
268 """ Generates code for structs and unions. name should be either
269 'struct' or union.
270 """
eli.benderskydcb859c2011-03-05 17:02:43 +0200271 s = name + ' ' + (n.name or '')
272 if n.decls:
273 s += '\n'
274 s += self._make_indent()
275 self.indent_level += 2
276 s += '{\n'
277 for decl in n.decls:
278 s += self._generate_stmt(decl)
279 self.indent_level -= 2
280 s += self._make_indent() + '}'
eli.benderskyabb96e62011-03-05 16:30:05 +0200281 return s
282
eli.benderskyae36e962011-02-27 08:25:05 +0200283 def _generate_stmt(self, n, add_indent=False):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200284 """ Generation from a statement node. This method exists as a wrapper
285 for individual visit_* methods to handle different treatment of
286 some statements in this context.
287 """
288 typ = type(n)
eli.benderskyae36e962011-02-27 08:25:05 +0200289 if add_indent: self.indent_level += 2
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200290 indent = self._make_indent()
eli.benderskyae36e962011-02-27 08:25:05 +0200291 if add_indent: self.indent_level -= 2
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200292
eli.bendersky724b1cc2011-03-05 10:45:08 +0200293 if typ in (
294 c_ast.Decl, c_ast.Assignment, c_ast.Cast, c_ast.UnaryOp,
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200295 c_ast.BinaryOp, c_ast.TernaryOp, c_ast.FuncCall, c_ast.ArrayRef,
296 c_ast.StructRef):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200297 # These can also appear in an expression context so no semicolon
298 # is added to them automatically
299 #
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200300 return indent + self.visit(n) + ';\n'
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200301 elif typ in (c_ast.Compound,):
302 # No extra indentation required before the opening brace of a
303 # compound - because it consists of multiple lines it has to
304 # compute its own indentation.
305 #
eli.benderskyae36e962011-02-27 08:25:05 +0200306 return self.visit(n)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200307 else:
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200308 return indent + self.visit(n) + '\n'
eli.benderskyafcfaac2011-02-25 16:46:01 +0200309
eli.benderskyd5ba3452011-02-18 21:32:47 +0200310 def _generate_decl(self, n):
311 """ Generation from a Decl node.
312 """
eli.benderskyafcfaac2011-02-25 16:46:01 +0200313 s = ''
314 if n.funcspec: s = ' '.join(n.funcspec) + ' '
315 if n.storage: s += ' '.join(n.storage) + ' '
316 s += self._generate_type(n.type)
317 return s
eli.benderskyd5ba3452011-02-18 21:32:47 +0200318
319 def _generate_type(self, n, modifiers=[]):
320 """ Recursive generation from a type node. n is the type node.
eli.benderskyafcfaac2011-02-25 16:46:01 +0200321 modifiers collects the PtrDecl, ArrayDecl and FuncDecl modifiers
322 encountered on the way down to a TypeDecl, to allow proper
323 generation from it.
eli.benderskyd5ba3452011-02-18 21:32:47 +0200324 """
325 typ = type(n)
326 #~ print(n, modifiers)
327
328 if typ == c_ast.TypeDecl:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200329 s = ''
330 if n.quals: s += ' '.join(n.quals) + ' '
331 s += self.visit(n.type)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200332
333 nstr = n.declname if n.declname else ''
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200334 # Resolve modifiers.
eli.benderskyafcfaac2011-02-25 16:46:01 +0200335 # Wrap in parens to distinguish pointer to array and pointer to
336 # function syntax.
eli.benderskyd5ba3452011-02-18 21:32:47 +0200337 #
338 for i, modifier in enumerate(modifiers):
339 if isinstance(modifier, c_ast.ArrayDecl):
340 if (i != 0 and isinstance(modifiers[i - 1], c_ast.PtrDecl)):
341 nstr = '(' + nstr + ')'
eli.benderskyafcfaac2011-02-25 16:46:01 +0200342 nstr += '[' + self.visit(modifier.dim) + ']'
343 elif isinstance(modifier, c_ast.FuncDecl):
344 if (i != 0 and isinstance(modifiers[i - 1], c_ast.PtrDecl)):
345 nstr = '(' + nstr + ')'
346 nstr += '(' + self.visit(modifier.args) + ')'
eli.benderskyd5ba3452011-02-18 21:32:47 +0200347 elif isinstance(modifier, c_ast.PtrDecl):
348 nstr = '*' + nstr
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200349 s += ' ' + nstr
eli.benderskyafcfaac2011-02-25 16:46:01 +0200350 return s
eli.benderskyd5ba3452011-02-18 21:32:47 +0200351 elif typ in (c_ast.Typename, c_ast.Decl):
eli.benderskyafcfaac2011-02-25 16:46:01 +0200352 return self._generate_decl(n.type)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200353 elif typ == c_ast.IdentifierType:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200354 return ' '.join(n.names) + ' '
355 elif typ in (c_ast.ArrayDecl, c_ast.PtrDecl, c_ast.FuncDecl):
356 return self._generate_type(n.type, modifiers + [n])
eli.benderskyabb96e62011-03-05 16:30:05 +0200357 else:
358 return self.visit(n)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200359
eli.bendersky724b1cc2011-03-05 10:45:08 +0200360 def _parenthesize_if(self, n, condition):
361 """ Visits 'n' and returns its string representation, parenthesized
362 if the condition function applied to the node returns True.
363 """
364 s = self.visit(n)
365 if condition(n):
366 return '(' + s + ')'
367 else:
368 return s
369
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200370 def _parenthesize_unless_simple(self, n):
371 """ Common use case for _parenthesize_if
372 """
373 return self._parenthesize_if(n, lambda d: not self._is_simple_node(d))
374
375 def _is_simple_node(self, n):
376 """ Returns True for nodes that are "simple" - i.e. nodes that always
377 have higher precedence than operators.
378 """
379 return isinstance(n,( c_ast.Constant, c_ast.ID, c_ast.ArrayRef,
380 c_ast.StructRef, c_ast.FuncCall))
381
eli.benderskyd5ba3452011-02-18 21:32:47 +0200382
383def translate_to_c(filename):
384 ast = parse_file(filename, use_cpp=True)
385 generator = CGenerator()
eli.benderskyafcfaac2011-02-25 16:46:01 +0200386 print(generator.visit(ast))
eli.benderskyd5ba3452011-02-18 21:32:47 +0200387
388
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200389def zz_test_translate():
390 # internal use
eli.benderskyd5ba3452011-02-18 21:32:47 +0200391 src = r'''
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200392 void v(void)
393 {
394 int poa = {5, 6};
395 int ka = {.ko[4] = 8, [5] = 3};
396 joker[2][4];
397 **ptr;
398 }
eli.benderskyd5ba3452011-02-18 21:32:47 +0200399 '''
400 parser = c_parser.CParser()
401 ast = parser.parse(src)
402 ast.show()
403 generator = CGenerator()
eli.benderskyafcfaac2011-02-25 16:46:01 +0200404 print(generator.visit(ast))
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200405
406
407#------------------------------------------------------------------------------
408if __name__ == "__main__":
409 if len(sys.argv) > 1:
410 translate_to_c(sys.argv[1])
411 else:
eli.benderskyd5ba3452011-02-18 21:32:47 +0200412 print("Please provide a filename as argument")
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200413