| 1 | #!/usr/bin/env python2
|
| 2 | """Generate ast module from specification
|
| 3 |
|
| 4 | This script generates the ast module from a simple specification,
|
| 5 | which makes it easy to accommodate changes in the grammar. This
|
| 6 | approach would be quite reasonable if the grammar changed often.
|
| 7 | Instead, it is rather complex to generate the appropriate code. And
|
| 8 | the Node interface has changed more often than the grammar.
|
| 9 | """
|
| 10 | from __future__ import print_function
|
| 11 |
|
| 12 | import fileinput
|
| 13 | import re
|
| 14 | import sys
|
| 15 | import cStringIO
|
| 16 |
|
| 17 | COMMA = ", "
|
| 18 |
|
| 19 | def load_boilerplate(file):
|
| 20 | f = open(file)
|
| 21 | buf = f.read()
|
| 22 | f.close()
|
| 23 | i = buf.find('### ''PROLOGUE')
|
| 24 | j = buf.find('### ''EPILOGUE')
|
| 25 | pro = buf[i+12:j].strip()
|
| 26 | epi = buf[j+12:].strip()
|
| 27 | return pro, epi
|
| 28 |
|
| 29 | def strip_default(arg):
|
| 30 | """Return the argname from an 'arg = default' string"""
|
| 31 | i = arg.find('=')
|
| 32 | if i == -1:
|
| 33 | return arg
|
| 34 | t = arg[:i].strip()
|
| 35 | return t
|
| 36 |
|
| 37 | P_NODE = 1 # It's another node.
|
| 38 | P_OTHER = 2 # * means it's a Python value like string, int, tuple, etc.
|
| 39 | P_NESTED = 3 # ! means this is a list of more nodes.
|
| 40 | P_NONE = 4 # & means it's a node or None.
|
| 41 |
|
| 42 | class NodeInfo(object):
|
| 43 | """Each instance describes a specific AST node"""
|
| 44 | def __init__(self, name, args):
|
| 45 | self.name = name
|
| 46 | self.args = args.strip()
|
| 47 | self.argnames = self.get_argnames()
|
| 48 | self.argprops = self.get_argprops()
|
| 49 | self.nargs = len(self.argnames)
|
| 50 | self.init = []
|
| 51 |
|
| 52 | def get_argnames(self):
|
| 53 | if '(' in self.args:
|
| 54 | i = self.args.find('(')
|
| 55 | j = self.args.rfind(')')
|
| 56 | args = self.args[i+1:j]
|
| 57 | else:
|
| 58 | args = self.args
|
| 59 | return [strip_default(arg.strip())
|
| 60 | for arg in args.split(',') if arg]
|
| 61 |
|
| 62 | def get_argprops(self):
|
| 63 | """Each argument can have a property like '*' or '!'
|
| 64 |
|
| 65 | XXX This method modifies the argnames in place!
|
| 66 | """
|
| 67 | d = {}
|
| 68 | hardest_arg = P_NODE
|
| 69 | for i in range(len(self.argnames)):
|
| 70 | arg = self.argnames[i]
|
| 71 | if arg.endswith('*'):
|
| 72 | arg = self.argnames[i] = arg[:-1]
|
| 73 | d[arg] = P_OTHER
|
| 74 | hardest_arg = max(hardest_arg, P_OTHER)
|
| 75 | elif arg.endswith('!'):
|
| 76 | arg = self.argnames[i] = arg[:-1]
|
| 77 | d[arg] = P_NESTED
|
| 78 | hardest_arg = max(hardest_arg, P_NESTED)
|
| 79 | elif arg.endswith('&'):
|
| 80 | arg = self.argnames[i] = arg[:-1]
|
| 81 | d[arg] = P_NONE
|
| 82 | hardest_arg = max(hardest_arg, P_NONE)
|
| 83 | else:
|
| 84 | d[arg] = P_NODE
|
| 85 | self.hardest_arg = hardest_arg
|
| 86 |
|
| 87 | if hardest_arg > P_NODE:
|
| 88 | self.args = self.args.replace('*', '')
|
| 89 | self.args = self.args.replace('!', '')
|
| 90 | self.args = self.args.replace('&', '')
|
| 91 |
|
| 92 | return d
|
| 93 |
|
| 94 | def gen_source(self):
|
| 95 | print("class %s(Node):" % self.name)
|
| 96 | print(" ARGNAMES = %r" % self.argnames)
|
| 97 | self._gen_init(sys.stdout)
|
| 98 | print()
|
| 99 | self._gen_getChildren(sys.stdout)
|
| 100 | print()
|
| 101 | self._gen_getChildNodes(sys.stdout)
|
| 102 | print()
|
| 103 |
|
| 104 | def _gen_init(self, buf):
|
| 105 | if self.args:
|
| 106 | argtuple = '(' in self.args
|
| 107 | args = self.args if not argtuple else ''.join(self.argnames)
|
| 108 | print(" def __init__(self, %s, lineno=None):" % args, file=buf)
|
| 109 | else:
|
| 110 | print(" def __init__(self, lineno=None):", file=buf)
|
| 111 | if self.argnames:
|
| 112 | if argtuple:
|
| 113 | for idx, name in enumerate(self.argnames):
|
| 114 | print(" self.%s = %s[%s]" % (name, args, idx), file=buf)
|
| 115 | else:
|
| 116 | for name in self.argnames:
|
| 117 | print(" self.%s = %s" % (name, name), file=buf)
|
| 118 | print(" self.lineno = lineno", file=buf)
|
| 119 | # Copy the lines in self.init, indented four spaces. The rstrip()
|
| 120 | # business is to get rid of the four spaces if line happens to be
|
| 121 | # empty, so that reindent.py is happy with the output.
|
| 122 | for line in self.init:
|
| 123 | print((" " + line).rstrip(), file=buf)
|
| 124 |
|
| 125 | def _gen_getChildren(self, buf):
|
| 126 | print(" def getChildren(self):", file=buf)
|
| 127 | if len(self.argnames) == 0:
|
| 128 | print(" return ()", file=buf)
|
| 129 | else:
|
| 130 | if self.hardest_arg < P_NESTED:
|
| 131 | clist = COMMA.join(["self.%s" % c
|
| 132 | for c in self.argnames])
|
| 133 | if self.nargs == 1:
|
| 134 | print(" return %s," % clist, file=buf)
|
| 135 | else:
|
| 136 | print(" return %s" % clist, file=buf)
|
| 137 | else:
|
| 138 | if len(self.argnames) == 1:
|
| 139 | print(" return tuple(flatten(self.%s))" % self.argnames[0], file=buf)
|
| 140 | else:
|
| 141 | print(" children = []", file=buf)
|
| 142 | template = " children.%s(%sself.%s%s)"
|
| 143 | for name in self.argnames:
|
| 144 | if self.argprops[name] == P_NESTED:
|
| 145 | print(template % ("extend", "flatten(",
|
| 146 | name, ")"), file=buf)
|
| 147 | else:
|
| 148 | print(template % ("append", "", name, ""), file=buf)
|
| 149 | print(" return tuple(children)", file=buf)
|
| 150 |
|
| 151 | def _gen_getChildNodes(self, buf):
|
| 152 | print(" def getChildNodes(self):", file=buf)
|
| 153 | if len(self.argnames) == 0:
|
| 154 | print(" return ()", file=buf)
|
| 155 | else:
|
| 156 | if self.hardest_arg < P_NESTED:
|
| 157 | clist = ["self.%s" % c
|
| 158 | for c in self.argnames
|
| 159 | if self.argprops[c] == P_NODE]
|
| 160 | if len(clist) == 0:
|
| 161 | print(" return ()", file=buf)
|
| 162 | elif len(clist) == 1:
|
| 163 | print(" return %s," % clist[0], file=buf)
|
| 164 | else:
|
| 165 | print(" return %s" % COMMA.join(clist), file=buf)
|
| 166 | else:
|
| 167 | print(" nodelist = []", file=buf)
|
| 168 | template = " nodelist.%s(%sself.%s%s)"
|
| 169 | for name in self.argnames:
|
| 170 | if self.argprops[name] == P_NONE:
|
| 171 | tmp = (" if self.%s is not None:\n"
|
| 172 | " nodelist.append(self.%s)")
|
| 173 | print(tmp % (name, name), file=buf)
|
| 174 | elif self.argprops[name] == P_NESTED:
|
| 175 | print(template % ("extend", "flatten_nodes(",
|
| 176 | name, ")"), file=buf)
|
| 177 | elif self.argprops[name] == P_NODE:
|
| 178 | print(template % ("append", "", name, ""), file=buf)
|
| 179 | print(" return tuple(nodelist)", file=buf)
|
| 180 |
|
| 181 |
|
| 182 | rx_init = re.compile('init\((.*)\):')
|
| 183 |
|
| 184 | def parse_spec(file):
|
| 185 | classes = {}
|
| 186 | cur = None
|
| 187 | for line in fileinput.input(file):
|
| 188 | if line.strip().startswith('#'):
|
| 189 | continue
|
| 190 | mo = rx_init.search(line)
|
| 191 | if mo is None:
|
| 192 | if cur is None:
|
| 193 | # a normal entry
|
| 194 | try:
|
| 195 | name, args = line.split(':')
|
| 196 | except ValueError:
|
| 197 | continue
|
| 198 | classes[name] = NodeInfo(name, args)
|
| 199 | cur = None
|
| 200 | else:
|
| 201 | # some code for the __init__ method
|
| 202 | cur.init.append(line)
|
| 203 | else:
|
| 204 | # some extra code for a Node's __init__ method
|
| 205 | name = mo.group(1)
|
| 206 | cur = classes[name]
|
| 207 | return sorted(classes.values(), key=lambda n: n.name)
|
| 208 |
|
| 209 | def main():
|
| 210 | prologue, epilogue = load_boilerplate(sys.argv[0])
|
| 211 | print('from __future__ import print_function')
|
| 212 | print('import cStringIO')
|
| 213 | print(prologue)
|
| 214 | print()
|
| 215 | classes = parse_spec(sys.argv[1])
|
| 216 | for info in classes:
|
| 217 | info.gen_source()
|
| 218 | print(epilogue)
|
| 219 |
|
| 220 | if __name__ == "__main__":
|
| 221 | main()
|
| 222 | sys.exit(0)
|
| 223 |
|
| 224 | ### PROLOGUE
|
| 225 | """Python abstract syntax node definitions
|
| 226 |
|
| 227 | This file is automatically generated by Tools/compiler/astgen.py
|
| 228 | """
|
| 229 | from .consts import CO_VARARGS, CO_VARKEYWORDS
|
| 230 |
|
| 231 | # NOTE: Similar to pyassem.flatten().
|
| 232 | def flatten(seq):
|
| 233 | l = []
|
| 234 | for elt in seq:
|
| 235 | if isinstance(elt, (tuple, list)):
|
| 236 | l.extend(flatten(elt))
|
| 237 | else:
|
| 238 | l.append(elt)
|
| 239 | return l
|
| 240 |
|
| 241 | def flatten_nodes(seq):
|
| 242 | return [n for n in flatten(seq) if isinstance(n, Node)]
|
| 243 |
|
| 244 | nodes = {}
|
| 245 |
|
| 246 |
|
| 247 | # NOTE: after_equals is a hack to make the output prettier. You could copy
|
| 248 | # _TrySingleLine in asdl/format.py. That took a long time to get right!
|
| 249 | def _PrettyPrint(val, f, indent=0, after_equals=False):
|
| 250 | indent_str = ' ' * indent
|
| 251 |
|
| 252 | if isinstance(val, Node):
|
| 253 | val.PrettyPrint(f, indent=indent, after_equals=after_equals)
|
| 254 |
|
| 255 | elif isinstance(val, list):
|
| 256 | if not after_equals:
|
| 257 | print('%s' % indent_str, end='', file=f)
|
| 258 | print('[', file=f) # No indent here
|
| 259 | for item in val:
|
| 260 | _PrettyPrint(item, f, indent=indent+2)
|
| 261 | # Not indented as much
|
| 262 | print('%s]' % indent_str, file=f)
|
| 263 |
|
| 264 | elif isinstance(val, tuple):
|
| 265 | if not after_equals:
|
| 266 | print('%s' % indent_str, end='', file=f)
|
| 267 | print('(', file=f)
|
| 268 | for item in val:
|
| 269 | _PrettyPrint(item, f, indent=indent+2)
|
| 270 | print('%s)' % indent_str, file=f)
|
| 271 |
|
| 272 | else:
|
| 273 | if not after_equals:
|
| 274 | print('%s' % indent_str, end='', file=f)
|
| 275 | # String or int?
|
| 276 | print('%r' % val, file=f)
|
| 277 |
|
| 278 |
|
| 279 | class Node(object):
|
| 280 | """Abstract base class for ast nodes."""
|
| 281 |
|
| 282 | ARGNAMES = []
|
| 283 |
|
| 284 | def getChildren(self):
|
| 285 | pass # implemented by subclasses
|
| 286 |
|
| 287 | def __iter__(self):
|
| 288 | for n in self.getChildren():
|
| 289 | yield n
|
| 290 |
|
| 291 | def asList(self): # for backwards compatibility
|
| 292 | return self.getChildren()
|
| 293 |
|
| 294 | def getChildNodes(self):
|
| 295 | pass # implemented by subclasses
|
| 296 |
|
| 297 | def __repr__(self):
|
| 298 | f = cStringIO.StringIO()
|
| 299 | self.PrettyPrint(f)
|
| 300 | return f.getvalue()
|
| 301 |
|
| 302 | def PrettyPrint(self, f, indent=0, after_equals=False):
|
| 303 | indent_str = ' ' * indent
|
| 304 |
|
| 305 | if not after_equals:
|
| 306 | print('%s' % indent_str, end='', file=f)
|
| 307 | print('%s(' % self.__class__.__name__, file=f)
|
| 308 | for name in self.ARGNAMES:
|
| 309 | # Print the field name
|
| 310 | print('%s %s = ' % (indent_str, name), end='', file=f)
|
| 311 |
|
| 312 | # Print the value
|
| 313 | val = getattr(self, name)
|
| 314 |
|
| 315 | _PrettyPrint(val, f, indent=indent+2, after_equals=True)
|
| 316 |
|
| 317 | print('%s) # %s' % (indent_str, self.__class__.__name__), file=f)
|
| 318 |
|
| 319 |
|
| 320 | class EmptyNode(Node):
|
| 321 | pass
|
| 322 |
|
| 323 | class Expression(Node):
|
| 324 | # Expression is an artificial node class to support "eval"
|
| 325 | nodes["expression"] = "Expression"
|
| 326 | def __init__(self, node):
|
| 327 | self.node = node
|
| 328 |
|
| 329 | def getChildren(self):
|
| 330 | return self.node,
|
| 331 |
|
| 332 | def getChildNodes(self):
|
| 333 | return self.node,
|
| 334 |
|
| 335 | def __repr__(self):
|
| 336 | return "Expression(%s)" % (repr(self.node))
|
| 337 |
|
| 338 | ### EPILOGUE
|
| 339 | for name, obj in globals().items():
|
| 340 | if isinstance(obj, type) and issubclass(obj, Node):
|
| 341 | nodes[name.lower()] = obj
|