blob: ec2f04bd1190a6db3074b1a16ad1e27aea4278b8 [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):
eli.bendersky6ed80a62011-05-09 18:22:33 +0300119 s = '(' + self._generate_type(n.to_type) + ')'
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200120 return s + ' ' + self.visit(n.expr)
121
eli.bendersky724b1cc2011-03-05 10:45:08 +0200122 def visit_ExprList(self, n):
eli.bendersky8e6c5862011-05-20 12:35:08 +0300123 visited_subexprs = []
124 for expr in n.exprs:
125 if isinstance(expr, c_ast.ExprList):
126 visited_subexprs.append('{' + self.visit(expr) + '}')
127 else:
128 visited_subexprs.append(self.visit(expr))
129 return ', '.join(visited_subexprs)
eli.bendersky724b1cc2011-03-05 10:45:08 +0200130
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200131 def visit_Enum(self, n):
132 s = 'enum'
133 if n.name: s += ' ' + n.name
134 if n.values:
135 s += ' {'
136 for i, enumerator in enumerate(n.values.enumerators):
137 s += enumerator.name
138 if enumerator.value:
139 s += ' = ' + self.visit(enumerator.value)
140 if i != len(n.values.enumerators) - 1:
141 s += ', '
142 s += '}'
143 return s
144
145 def visit_Struct(self, n):
146 s = 'struct'
147 if n.name: s += ' ' + n.name
148 if n.decls:
149 s += ' { \n'
150 for decl in n.decls:
151 s += ' ' + self.visit(decl) + ';\n'
152 s += '}'
153 return s
154
eli.benderskyafcfaac2011-02-25 16:46:01 +0200155 def visit_FuncDef(self, n):
156 decl = self.visit(n.decl)
157 self.indent_level = 0
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200158 # The body is a Compound node
eli.benderskyafcfaac2011-02-25 16:46:01 +0200159 body = self.visit(n.body)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200160 return decl + '\n' + body + '\n'
161
162 def visit_FileAST(self, n):
163 s = ''
164 for ext in n.ext:
165 if isinstance(ext, c_ast.FuncDef):
166 s += self.visit(ext)
167 else:
168 s += self.visit(ext) + ';\n'
169 return s
170
171 def visit_Compound(self, n):
172 s = self._make_indent() + '{\n'
173 self.indent_level += 2
eli.benderskycad1cfd2011-05-26 06:56:27 +0300174 if n.block_items:
175 s += ''.join(self._generate_stmt(stmt) for stmt in n.block_items)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200176 self.indent_level -= 2
177 s += self._make_indent() + '}\n'
178 return s
179
180 def visit_ParamList(self, n):
181 return ', '.join(self.visit(param) for param in n.params)
182
183 def visit_Return(self, n):
184 s = 'return'
185 if n.expr: s += ' ' + self.visit(n.expr)
186 return s + ';'
187
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200188 def visit_Break(self, n):
189 return 'break;'
190
191 def visit_Continue(self, n):
192 return 'continue;'
eli.bendersky724b1cc2011-03-05 10:45:08 +0200193
194 def visit_TernaryOp(self, n):
195 s = self.visit(n.cond) + ' ? '
196 s += self.visit(n.iftrue) + ' : '
197 s += self.visit(n.iffalse)
198 return s
199
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200200 def visit_If(self, n):
201 s = 'if ('
202 if n.cond: s += self.visit(n.cond)
203 s += ')\n'
204 s += self._generate_stmt(n.iftrue, add_indent=True)
205 if n.iffalse:
206 s += self._make_indent() + 'else\n'
207 s += self._generate_stmt(n.iffalse, add_indent=True)
208 return s
209
eli.benderskyae36e962011-02-27 08:25:05 +0200210 def visit_For(self, n):
211 s = 'for ('
212 if n.init: s += self.visit(n.init)
213 s += ';'
214 if n.cond: s += ' ' + self.visit(n.cond)
215 s += ';'
216 if n.next: s += ' ' + self.visit(n.next)
217 s += ')\n'
218 s += self._generate_stmt(n.stmt, add_indent=True)
219 return s
220
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200221 def visit_While(self, n):
222 s = 'while ('
223 if n.cond: s += self.visit(n.cond)
224 s += ')\n'
225 s += self._generate_stmt(n.stmt, add_indent=True)
226 return s
227
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200228 def visit_DoWhile(self, n):
229 s = 'do\n'
230 s += self._generate_stmt(n.stmt, add_indent=True)
231 s += self._make_indent() + 'while ('
232 if n.cond: s += self.visit(n.cond)
233 s += ');'
234 return s
235
eli.benderskyabb96e62011-03-05 16:30:05 +0200236 def visit_Switch(self, n):
237 s = 'switch (' + self.visit(n.cond) + ')\n'
238 s += self._generate_stmt(n.stmt, add_indent=True)
239 return s
240
241 def visit_Case(self, n):
242 s = 'case ' + self.visit(n.expr) + ':\n'
243 s += self._generate_stmt(n.stmt, add_indent=True)
244 return s
245
246 def visit_Default(self, n):
247 return 'default:\n' + self._generate_stmt(n.stmt, add_indent=True)
248
249 def visit_Label(self, n):
250 return n.name + ':\n' + self._generate_stmt(n.stmt)
251
252 def visit_Goto(self, n):
253 return 'goto ' + n.name + ';'
254
eli.benderskye6ecb1a2011-03-05 10:58:46 +0200255 def visit_EllipsisParam(self, n):
256 return '...'
257
eli.benderskyabb96e62011-03-05 16:30:05 +0200258 def visit_Struct(self, n):
259 return self._generate_struct_union(n, 'struct')
260
eli.bendersky8e6c5862011-05-20 12:35:08 +0300261 def visit_Typename(self, n):
262 return self._generate_type(n.type)
263
eli.benderskyabb96e62011-03-05 16:30:05 +0200264 def visit_Union(self, n):
265 return self._generate_struct_union(n, 'union')
266
eli.benderskydcb859c2011-03-05 17:02:43 +0200267 def visit_NamedInitializer(self, n):
268 s = ''
269 for name in n.name:
270 if isinstance(name, c_ast.ID):
271 s += '.' + name.name
272 elif isinstance(name, c_ast.Constant):
273 s += '[' + name.value + ']'
274 s += ' = ' + self.visit(n.expr)
275 return s
276
eli.benderskyabb96e62011-03-05 16:30:05 +0200277 def _generate_struct_union(self, n, name):
278 """ Generates code for structs and unions. name should be either
279 'struct' or union.
280 """
eli.benderskydcb859c2011-03-05 17:02:43 +0200281 s = name + ' ' + (n.name or '')
282 if n.decls:
283 s += '\n'
284 s += self._make_indent()
285 self.indent_level += 2
286 s += '{\n'
287 for decl in n.decls:
288 s += self._generate_stmt(decl)
289 self.indent_level -= 2
290 s += self._make_indent() + '}'
eli.benderskyabb96e62011-03-05 16:30:05 +0200291 return s
292
eli.benderskyae36e962011-02-27 08:25:05 +0200293 def _generate_stmt(self, n, add_indent=False):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200294 """ Generation from a statement node. This method exists as a wrapper
295 for individual visit_* methods to handle different treatment of
296 some statements in this context.
297 """
298 typ = type(n)
eli.benderskyae36e962011-02-27 08:25:05 +0200299 if add_indent: self.indent_level += 2
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200300 indent = self._make_indent()
eli.benderskyae36e962011-02-27 08:25:05 +0200301 if add_indent: self.indent_level -= 2
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200302
eli.bendersky724b1cc2011-03-05 10:45:08 +0200303 if typ in (
304 c_ast.Decl, c_ast.Assignment, c_ast.Cast, c_ast.UnaryOp,
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200305 c_ast.BinaryOp, c_ast.TernaryOp, c_ast.FuncCall, c_ast.ArrayRef,
306 c_ast.StructRef):
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200307 # These can also appear in an expression context so no semicolon
308 # is added to them automatically
309 #
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200310 return indent + self.visit(n) + ';\n'
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200311 elif typ in (c_ast.Compound,):
312 # No extra indentation required before the opening brace of a
313 # compound - because it consists of multiple lines it has to
314 # compute its own indentation.
315 #
eli.benderskyae36e962011-02-27 08:25:05 +0200316 return self.visit(n)
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200317 else:
eli.benderskyfc96e5e2011-03-04 09:51:23 +0200318 return indent + self.visit(n) + '\n'
eli.benderskyafcfaac2011-02-25 16:46:01 +0200319
eli.benderskyd5ba3452011-02-18 21:32:47 +0200320 def _generate_decl(self, n):
321 """ Generation from a Decl node.
322 """
eli.benderskyafcfaac2011-02-25 16:46:01 +0200323 s = ''
324 if n.funcspec: s = ' '.join(n.funcspec) + ' '
325 if n.storage: s += ' '.join(n.storage) + ' '
326 s += self._generate_type(n.type)
327 return s
eli.benderskyd5ba3452011-02-18 21:32:47 +0200328
329 def _generate_type(self, n, modifiers=[]):
330 """ Recursive generation from a type node. n is the type node.
eli.benderskyafcfaac2011-02-25 16:46:01 +0200331 modifiers collects the PtrDecl, ArrayDecl and FuncDecl modifiers
332 encountered on the way down to a TypeDecl, to allow proper
333 generation from it.
eli.benderskyd5ba3452011-02-18 21:32:47 +0200334 """
335 typ = type(n)
336 #~ print(n, modifiers)
337
338 if typ == c_ast.TypeDecl:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200339 s = ''
340 if n.quals: s += ' '.join(n.quals) + ' '
341 s += self.visit(n.type)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200342
343 nstr = n.declname if n.declname else ''
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200344 # Resolve modifiers.
eli.benderskyafcfaac2011-02-25 16:46:01 +0200345 # Wrap in parens to distinguish pointer to array and pointer to
346 # function syntax.
eli.benderskyd5ba3452011-02-18 21:32:47 +0200347 #
348 for i, modifier in enumerate(modifiers):
349 if isinstance(modifier, c_ast.ArrayDecl):
350 if (i != 0 and isinstance(modifiers[i - 1], c_ast.PtrDecl)):
351 nstr = '(' + nstr + ')'
eli.benderskyafcfaac2011-02-25 16:46:01 +0200352 nstr += '[' + self.visit(modifier.dim) + ']'
353 elif isinstance(modifier, c_ast.FuncDecl):
354 if (i != 0 and isinstance(modifiers[i - 1], c_ast.PtrDecl)):
355 nstr = '(' + nstr + ')'
356 nstr += '(' + self.visit(modifier.args) + ')'
eli.benderskyd5ba3452011-02-18 21:32:47 +0200357 elif isinstance(modifier, c_ast.PtrDecl):
358 nstr = '*' + nstr
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200359 s += ' ' + nstr
eli.benderskyafcfaac2011-02-25 16:46:01 +0200360 return s
eli.bendersky6ed80a62011-05-09 18:22:33 +0300361 elif typ == c_ast.Decl:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200362 return self._generate_decl(n.type)
eli.bendersky6ed80a62011-05-09 18:22:33 +0300363 elif typ == c_ast.Typename:
364 return self._generate_type(n.type)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200365 elif typ == c_ast.IdentifierType:
eli.benderskyafcfaac2011-02-25 16:46:01 +0200366 return ' '.join(n.names) + ' '
367 elif typ in (c_ast.ArrayDecl, c_ast.PtrDecl, c_ast.FuncDecl):
368 return self._generate_type(n.type, modifiers + [n])
eli.benderskyabb96e62011-03-05 16:30:05 +0200369 else:
370 return self.visit(n)
eli.benderskyd5ba3452011-02-18 21:32:47 +0200371
eli.bendersky724b1cc2011-03-05 10:45:08 +0200372 def _parenthesize_if(self, n, condition):
373 """ Visits 'n' and returns its string representation, parenthesized
374 if the condition function applied to the node returns True.
375 """
376 s = self.visit(n)
377 if condition(n):
378 return '(' + s + ')'
379 else:
380 return s
381
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200382 def _parenthesize_unless_simple(self, n):
383 """ Common use case for _parenthesize_if
384 """
385 return self._parenthesize_if(n, lambda d: not self._is_simple_node(d))
386
387 def _is_simple_node(self, n):
388 """ Returns True for nodes that are "simple" - i.e. nodes that always
389 have higher precedence than operators.
390 """
391 return isinstance(n,( c_ast.Constant, c_ast.ID, c_ast.ArrayRef,
392 c_ast.StructRef, c_ast.FuncCall))
393
eli.benderskyd5ba3452011-02-18 21:32:47 +0200394
395def translate_to_c(filename):
396 ast = parse_file(filename, use_cpp=True)
397 generator = CGenerator()
eli.benderskyafcfaac2011-02-25 16:46:01 +0200398 print(generator.visit(ast))
eli.benderskyd5ba3452011-02-18 21:32:47 +0200399
400
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200401def zz_test_translate():
402 # internal use
eli.bendersky8e6c5862011-05-20 12:35:08 +0300403 src = r'''
eli.benderskycad1cfd2011-05-26 06:56:27 +0300404int main(){} '''
eli.bendersky8e6c5862011-05-20 12:35:08 +0300405 parser = c_parser.CParser()
406 ast = parser.parse(src)
407 ast.show()
408 generator = CGenerator()
409
410 print(generator.visit(ast))
411
412 # tracing the generator for debugging
413 #~ import trace
414 #~ tr = trace.Trace(countcallers=1)
415 #~ tr.runfunc(generator.visit, ast)
416 #~ tr.results().write_results()
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200417
418
419#------------------------------------------------------------------------------
420if __name__ == "__main__":
eli.bendersky6ed80a62011-05-09 18:22:33 +0300421 zz_test_translate()
eli.benderskyb36f5bb2011-03-06 07:22:33 +0200422 if len(sys.argv) > 1:
423 translate_to_c(sys.argv[1])
424 else:
eli.benderskyd5ba3452011-02-18 21:32:47 +0200425 print("Please provide a filename as argument")
eli.bendersky3ae9f7a2011-02-27 07:19:41 +0200426