utils/lit/lit/ShUtil.py

   1 import itertools
   2
   3 import Util
   4 from ShCommands import Command, Pipeline, Seq
   5
   6 class ShLexer:
   7     def __init__(self, data, win32Escapes = False):
   8         self.data = data
   9         self.pos = 0
  10         self.end = len(data)
  11         self.win32Escapes = win32Escapes
  12
  13     def eat(self):
  14         c = self.data[self.pos]
  15         self.pos += 1
  16         return c
  17
  18     def look(self):
  19         return self.data[self.pos]
  20
  21     def maybe_eat(self, c):
  22         """
  23         maybe_eat(c) - Consume the character c if it is the next character,
  24         returning True if a character was consumed. """
  25         if self.data[self.pos] == c:
  26             self.pos += 1
  27             return True
  28         return False
  29
  30     def lex_arg_fast(self, c):
  31         # Get the leading whitespace free section.
  32         chunk = self.data[self.pos - 1:].split(None, 1)[0]
  33
  34         # If it has special characters, the fast path failed.
  35         if ('|' in chunk or '&' in chunk or
  36             '<' in chunk or '>' in chunk or
  37             "'" in chunk or '"' in chunk or
  38             ';' in chunk or '\\' in chunk):
  39             return None
  40
  41         self.pos = self.pos - 1 + len(chunk)
  42         return chunk
  43
  44     def lex_arg_slow(self, c):
  45         if c in "'\"":
  46             str = self.lex_arg_quoted(c)
  47         else:
  48             str = c
  49         while self.pos != self.end:
  50             c = self.look()
  51             if c.isspace() or c in "|&;":
  52                 break
  53             elif c in '><':
  54                 # This is an annoying case; we treat '2>' as a single token so
  55                 # we don't have to track whitespace tokens.
  56
  57                 # If the parse string isn't an integer, do the usual thing.
  58                 if not str.isdigit():
  59                     break
  60
  61                 # Otherwise, lex the operator and convert to a redirection
  62                 # token.
  63                 num = int(str)
  64                 tok = self.lex_one_token()
  65                 assert isinstance(tok, tuple) and len(tok) == 1
  66                 return (tok[0], num)
  67             elif c == '"':
  68                 self.eat()
  69                 str += self.lex_arg_quoted('"')
  70             elif c == "'":
  71                 self.eat()
  72                 str += self.lex_arg_quoted("'")
  73             elif not self.win32Escapes and c == '\\':
  74                 # Outside of a string, '\\' escapes everything.
  75                 self.eat()
  76                 if self.pos == self.end:
  77                     Util.warning("escape at end of quoted argument in: %r" %
  78                                  self.data)
  79                     return str
  80                 str += self.eat()
  81             else:
  82                 str += self.eat()
  83         return str
  84
  85     def lex_arg_quoted(self, delim):
  86         str = ''
  87         while self.pos != self.end:
  88             c = self.eat()
  89             if c == delim:
  90                 return str
  91             elif c == '\\' and delim == '"':
  92                 # Inside a '"' quoted string, '\\' only escapes the quote
  93                 # character and backslash, otherwise it is preserved.
  94                 if self.pos == self.end:
  95                     Util.warning("escape at end of quoted argument in: %r" %
  96                                  self.data)
  97                     return str
  98                 c = self.eat()
  99                 if c == '"': #
 100                     str += '"'
 101                 elif c == '\\':
 102                     str += '\\'
 103                 else:
 104                     str += '\\' + c
 105             else:
 106                 str += c
 107         Util.warning("missing quote character in %r" % self.data)
 108         return str
 109
 110     def lex_arg_checked(self, c):
 111         pos = self.pos
 112         res = self.lex_arg_fast(c)
 113         end = self.pos
 114
 115         self.pos = pos
 116         reference = self.lex_arg_slow(c)
 117         if res is not None:
 118             if res != reference:
 119                 raise ValueError("Fast path failure: %r != %r" % (
 120                         res, reference))
 121             if self.pos != end:
 122                 raise ValueError("Fast path failure: %r != %r" % (
 123                         self.pos, end))
 124         return reference
 125
 126     def lex_arg(self, c):
 127         return self.lex_arg_fast(c) or self.lex_arg_slow(c)
 128
 129     def lex_one_token(self):
 130         """
 131         lex_one_token - Lex a single 'sh' token. """
 132
 133         c = self.eat()
 134         if c == ';':
 135             return (c,)
 136         if c == '|':
 137             if self.maybe_eat('|'):
 138                 return ('||',)
 139             return (c,)
 140         if c == '&':
 141             if self.maybe_eat('&'):
 142                 return ('&&',)
 143             if self.maybe_eat('>'):
 144                 return ('&>',)
 145             return (c,)
 146         if c == '>':
 147             if self.maybe_eat('&'):
 148                 return ('>&',)
 149             if self.maybe_eat('>'):
 150                 return ('>>',)
 151             return (c,)
 152         if c == '<':
 153             if self.maybe_eat('&'):
 154                 return ('<&',)
 155             if self.maybe_eat('>'):
 156                 return ('<<',)
 157             return (c,)
 158
 159         return self.lex_arg(c)
 160
 161     def lex(self):
 162         while self.pos != self.end:
 163             if self.look().isspace():
 164                 self.eat()
 165             else:
 166                 yield self.lex_one_token()
 167
 168 ###
 169
 170 class ShParser:
 171     def __init__(self, data, win32Escapes = False, pipefail = False):
 172         self.data = data
 173         self.pipefail = pipefail
 174         self.tokens = ShLexer(data, win32Escapes = win32Escapes).lex()
 175
 176     def lex(self):
 177         try:
 178             return self.tokens.next()
 179         except StopIteration:
 180             return None
 181
 182     def look(self):
 183         next = self.lex()
 184         if next is not None:
 185             self.tokens = itertools.chain([next], self.tokens)
 186         return next
 187
 188     def parse_command(self):
 189         tok = self.lex()
 190         if not tok:
 191             raise ValueError("empty command!")
 192         if isinstance(tok, tuple):
 193             raise ValueError("syntax error near unexpected token %r" % tok[0])
 194
 195         args = [tok]
 196         redirects = []
 197         while 1:
 198             tok = self.look()
 199
 200             # EOF?
 201             if tok is None:
 202                 break
 203
 204             # If this is an argument, just add it to the current command.
 205             if isinstance(tok, str):
 206                 args.append(self.lex())
 207                 continue
 208
 209             # Otherwise see if it is a terminator.
 210             assert isinstance(tok, tuple)
 211             if tok[0] in ('|',';','&','||','&&'):
 212                 break
 213
 214             # Otherwise it must be a redirection.
 215             op = self.lex()
 216             arg = self.lex()
 217             if not arg:
 218                 raise ValueError("syntax error near token %r" % op[0])
 219             redirects.append((op, arg))
 220
 221         return Command(args, redirects)
 222
 223     def parse_pipeline(self):
 224         negate = False
 225
 226         commands = [self.parse_command()]
 227         while self.look() == ('|',):
 228             self.lex()
 229             commands.append(self.parse_command())
 230         return Pipeline(commands, negate, self.pipefail)
 231
 232     def parse(self):
 233         lhs = self.parse_pipeline()
 234
 235         while self.look():
 236             operator = self.lex()
 237             assert isinstance(operator, tuple) and len(operator) == 1
 238
 239             if not self.look():
 240                 raise ValueError(
 241                     "missing argument to operator %r" % operator[0])
 242
 243             # FIXME: Operator precedence!!
 244             lhs = Seq(lhs, operator[0], self.parse_pipeline())
 245
 246         return lhs
 247
 248 ###
 249
 250 import unittest
 251
 252 class TestShLexer(unittest.TestCase):
 253     def lex(self, str, *args, **kwargs):
 254         return list(ShLexer(str, *args, **kwargs).lex())
 255
 256     def test_basic(self):
 257         self.assertEqual(self.lex('a|b>c&d<e;f'),
 258                          ['a', ('|',), 'b', ('>',), 'c', ('&',), 'd',
 259                           ('<',), 'e', (';',), 'f'])
 260
 261     def test_redirection_tokens(self):
 262         self.assertEqual(self.lex('a2>c'),
 263                          ['a2', ('>',), 'c'])
 264         self.assertEqual(self.lex('a 2>c'),
 265                          ['a', ('>',2), 'c'])
 266
 267     def test_quoting(self):
 268         self.assertEqual(self.lex(""" 'a' """),
 269                          ['a'])
 270         self.assertEqual(self.lex(""" "hello\\"world" """),
 271                          ['hello"world'])
 272         self.assertEqual(self.lex(""" "hello\\'world" """),
 273                          ["hello\\'world"])
 274         self.assertEqual(self.lex(""" "hello\\\\world" """),
 275                          ["hello\\world"])
 276         self.assertEqual(self.lex(""" he"llo wo"rld """),
 277                          ["hello world"])
 278         self.assertEqual(self.lex(""" a\\ b a\\\\b """),
 279                          ["a b", "a\\b"])
 280         self.assertEqual(self.lex(""" "" "" """),
 281                          ["", ""])
 282         self.assertEqual(self.lex(""" a\\ b """, win32Escapes = True),
 283                          ['a\\', 'b'])
 284
 285 class TestShParse(unittest.TestCase):
 286     def parse(self, str):
 287         return ShParser(str).parse()
 288
 289     def test_basic(self):
 290         self.assertEqual(self.parse('echo hello'),
 291                          Pipeline([Command(['echo', 'hello'], [])], False))
 292         self.assertEqual(self.parse('echo ""'),
 293                          Pipeline([Command(['echo', ''], [])], False))
 294         self.assertEqual(self.parse("""echo -DFOO='a'"""),
 295                          Pipeline([Command(['echo', '-DFOO=a'], [])], False))
 296         self.assertEqual(self.parse('echo -DFOO="a"'),
 297                          Pipeline([Command(['echo', '-DFOO=a'], [])], False))
 298
 299     def test_redirection(self):
 300         self.assertEqual(self.parse('echo hello > c'),
 301                          Pipeline([Command(['echo', 'hello'],
 302                                            [((('>'),), 'c')])], False))
 303         self.assertEqual(self.parse('echo hello > c >> d'),
 304                          Pipeline([Command(['echo', 'hello'], [(('>',), 'c'),
 305                                                      (('>>',), 'd')])], False))
 306         self.assertEqual(self.parse('a 2>&1'),
 307                          Pipeline([Command(['a'], [(('>&',2), '1')])], False))
 308
 309     def test_pipeline(self):
 310         self.assertEqual(self.parse('a | b'),
 311                          Pipeline([Command(['a'], []),
 312                                    Command(['b'], [])],
 313                                   False))
 314
 315         self.assertEqual(self.parse('a | b | c'),
 316                          Pipeline([Command(['a'], []),
 317                                    Command(['b'], []),
 318                                    Command(['c'], [])],
 319                                   False))
 320
 321     def test_list(self):
 322         self.assertEqual(self.parse('a ; b'),
 323                          Seq(Pipeline([Command(['a'], [])], False),
 324                              ';',
 325                              Pipeline([Command(['b'], [])], False)))
 326
 327         self.assertEqual(self.parse('a & b'),
 328                          Seq(Pipeline([Command(['a'], [])], False),
 329                              '&',
 330                              Pipeline([Command(['b'], [])], False)))
 331
 332         self.assertEqual(self.parse('a && b'),
 333                          Seq(Pipeline([Command(['a'], [])], False),
 334                              '&&',
 335                              Pipeline([Command(['b'], [])], False)))
 336
 337         self.assertEqual(self.parse('a || b'),
 338                          Seq(Pipeline([Command(['a'], [])], False),
 339                              '||',
 340                              Pipeline([Command(['b'], [])], False)))
 341
 342         self.assertEqual(self.parse('a && b || c'),
 343                          Seq(Seq(Pipeline([Command(['a'], [])], False),
 344                                  '&&',
 345                                  Pipeline([Command(['b'], [])], False)),
 346                              '||',
 347                              Pipeline([Command(['c'], [])], False)))
 348
 349         self.assertEqual(self.parse('a; b'),
 350                          Seq(Pipeline([Command(['a'], [])], False),
 351                              ';',
 352                              Pipeline([Command(['b'], [])], False)))
 353
 354 if __name__ == '__main__':
 355     unittest.main()