4 from ShCommands import Command, Pipeline, Seq
7 def __init__(self, data, win32Escapes = False):
11 self.win32Escapes = win32Escapes
14 c = self.data[self.pos]
19 return self.data[self.pos]
21 def maybe_eat(self, c):
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:
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]
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):
41 self.pos = self.pos - 1 + len(chunk)
44 def lex_arg_slow(self, c):
46 str = self.lex_arg_quoted(c)
49 while self.pos != self.end:
51 if c.isspace() or 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.
57 # If the parse string isn't an integer, do the usual thing.
61 # Otherwise, lex the operator and convert to a redirection
64 tok = self.lex_one_token()
65 assert isinstance(tok, tuple) and len(tok) == 1
69 str += self.lex_arg_quoted('"')
72 str += self.lex_arg_quoted("'")
73 elif not self.win32Escapes and c == '\\':
74 # Outside of a string, '\\' escapes everything.
76 if self.pos == self.end:
77 Util.warning("escape at end of quoted argument in: %r" %
85 def lex_arg_quoted(self, delim):
87 while self.pos != self.end:
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" %
107 Util.warning("missing quote character in %r" % self.data)
110 def lex_arg_checked(self, c):
112 res = self.lex_arg_fast(c)
116 reference = self.lex_arg_slow(c)
119 raise ValueError,"Fast path failure: %r != %r" % (res, reference)
121 raise ValueError,"Fast path failure: %r != %r" % (self.pos, end)
124 def lex_arg(self, c):
125 return self.lex_arg_fast(c) or self.lex_arg_slow(c)
127 def lex_one_token(self):
129 lex_one_token - Lex a single 'sh' token. """
135 if self.maybe_eat('|'):
139 if self.maybe_eat('&'):
141 if self.maybe_eat('>'):
145 if self.maybe_eat('&'):
147 if self.maybe_eat('>'):
151 if self.maybe_eat('&'):
153 if self.maybe_eat('>'):
157 return self.lex_arg(c)
160 while self.pos != self.end:
161 if self.look().isspace():
164 yield self.lex_one_token()
169 def __init__(self, data, win32Escapes = False):
171 self.tokens = ShLexer(data, win32Escapes = win32Escapes).lex()
175 return self.tokens.next()
176 except StopIteration:
182 self.tokens = itertools.chain([next], self.tokens)
185 def parse_command(self):
188 raise ValueError,"empty command!"
189 if isinstance(tok, tuple):
190 raise ValueError,"syntax error near unexpected token %r" % tok[0]
201 # If this is an argument, just add it to the current command.
202 if isinstance(tok, str):
203 args.append(self.lex())
206 # Otherwise see if it is a terminator.
207 assert isinstance(tok, tuple)
208 if tok[0] in ('|',';','&','||','&&'):
211 # Otherwise it must be a redirection.
215 raise ValueError,"syntax error near token %r" % op[0]
216 redirects.append((op, arg))
218 return Command(args, redirects)
220 def parse_pipeline(self):
223 commands = [self.parse_command()]
224 while self.look() == ('|',):
226 commands.append(self.parse_command())
227 return Pipeline(commands, negate)
230 lhs = self.parse_pipeline()
233 operator = self.lex()
234 assert isinstance(operator, tuple) and len(operator) == 1
237 raise ValueError, "missing argument to operator %r" % operator[0]
239 # FIXME: Operator precedence!!
240 lhs = Seq(lhs, operator[0], self.parse_pipeline())
248 class TestShLexer(unittest.TestCase):
249 def lex(self, str, *args, **kwargs):
250 return list(ShLexer(str, *args, **kwargs).lex())
252 def test_basic(self):
253 self.assertEqual(self.lex('a|b>c&d<e;f'),
254 ['a', ('|',), 'b', ('>',), 'c', ('&',), 'd',
255 ('<',), 'e', (';',), 'f'])
257 def test_redirection_tokens(self):
258 self.assertEqual(self.lex('a2>c'),
260 self.assertEqual(self.lex('a 2>c'),
263 def test_quoting(self):
264 self.assertEqual(self.lex(""" 'a' """),
266 self.assertEqual(self.lex(""" "hello\\"world" """),
268 self.assertEqual(self.lex(""" "hello\\'world" """),
270 self.assertEqual(self.lex(""" "hello\\\\world" """),
272 self.assertEqual(self.lex(""" he"llo wo"rld """),
274 self.assertEqual(self.lex(""" a\\ b a\\\\b """),
276 self.assertEqual(self.lex(""" "" "" """),
278 self.assertEqual(self.lex(""" a\\ b """, win32Escapes = True),
281 class TestShParse(unittest.TestCase):
282 def parse(self, str):
283 return ShParser(str).parse()
285 def test_basic(self):
286 self.assertEqual(self.parse('echo hello'),
287 Pipeline([Command(['echo', 'hello'], [])], False))
288 self.assertEqual(self.parse('echo ""'),
289 Pipeline([Command(['echo', ''], [])], False))
290 self.assertEqual(self.parse("""echo -DFOO='a'"""),
291 Pipeline([Command(['echo', '-DFOO=a'], [])], False))
292 self.assertEqual(self.parse('echo -DFOO="a"'),
293 Pipeline([Command(['echo', '-DFOO=a'], [])], False))
295 def test_redirection(self):
296 self.assertEqual(self.parse('echo hello > c'),
297 Pipeline([Command(['echo', 'hello'],
298 [((('>'),), 'c')])], False))
299 self.assertEqual(self.parse('echo hello > c >> d'),
300 Pipeline([Command(['echo', 'hello'], [(('>',), 'c'),
301 (('>>',), 'd')])], False))
302 self.assertEqual(self.parse('a 2>&1'),
303 Pipeline([Command(['a'], [(('>&',2), '1')])], False))
305 def test_pipeline(self):
306 self.assertEqual(self.parse('a | b'),
307 Pipeline([Command(['a'], []),
311 self.assertEqual(self.parse('a | b | c'),
312 Pipeline([Command(['a'], []),
318 self.assertEqual(self.parse('a ; b'),
319 Seq(Pipeline([Command(['a'], [])], False),
321 Pipeline([Command(['b'], [])], False)))
323 self.assertEqual(self.parse('a & b'),
324 Seq(Pipeline([Command(['a'], [])], False),
326 Pipeline([Command(['b'], [])], False)))
328 self.assertEqual(self.parse('a && b'),
329 Seq(Pipeline([Command(['a'], [])], False),
331 Pipeline([Command(['b'], [])], False)))
333 self.assertEqual(self.parse('a || b'),
334 Seq(Pipeline([Command(['a'], [])], False),
336 Pipeline([Command(['b'], [])], False)))
338 self.assertEqual(self.parse('a && b || c'),
339 Seq(Seq(Pipeline([Command(['a'], [])], False),
341 Pipeline([Command(['b'], [])], False)),
343 Pipeline([Command(['c'], [])], False)))
345 self.assertEqual(self.parse('a; b'),
346 Seq(Pipeline([Command(['a'], [])], False),
348 Pipeline([Command(['b'], [])], False)))
350 if __name__ == '__main__':