blob: ca9c557fead17c39a96357ea45e0c2e7185bf557 [file] [log] [blame]
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +00001# -*- coding: iso-8859-1 -*-
Guido van Rossume7b146f2000-02-04 15:28:42 +00002"""A lexical analyzer class for simple shell-like syntaxes."""
3
Tim Peters70c43782001-01-17 08:48:39 +00004# Module and documentation by Eric S. Raymond, 21 Dec 1998
Guido van Rossumeb4e11a2000-05-01 20:08:46 +00005# Input stacking and error message cleanup added by ESR, March 2000
Tim Peters70c43782001-01-17 08:48:39 +00006# push_source() and pop_source() made explicit by ESR, January 2001.
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +00007# Posix compliance, split(), string arguments, and
8# iterator interface by Gustavo Niemeyer, April 2003.
Guido van Rossum9c30c241998-12-22 05:19:29 +00009
Fred Drake52dc76c2000-07-03 09:56:23 +000010import os.path
Guido van Rossum73898c71999-05-03 18:14:16 +000011import sys
Guido van Rossum9c30c241998-12-22 05:19:29 +000012
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000013from types import StringTypes
14
15try:
16 from cStringIO import StringIO
17except ImportError:
18 from StringIO import StringIO
19
20__all__ = ["shlex", "split"]
Skip Montanaro0de65802001-02-15 22:15:14 +000021
Guido van Rossum9c30c241998-12-22 05:19:29 +000022class shlex:
Tim Peters70c43782001-01-17 08:48:39 +000023 "A lexical analyzer class for simple shell-like syntaxes."
Fred Drake24315232003-04-17 22:01:17 +000024 def __init__(self, instream=None, infile=None, posix=False):
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000025 if type(instream) in StringTypes:
26 instream = StringIO(instream)
Raymond Hettingerf13eb552002-06-02 00:40:05 +000027 if instream is not None:
Guido van Rossum9c30c241998-12-22 05:19:29 +000028 self.instream = instream
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000029 self.infile = infile
Guido van Rossum9c30c241998-12-22 05:19:29 +000030 else:
31 self.instream = sys.stdin
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000032 self.infile = None
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000033 self.posix = posix
34 if posix:
35 self.eof = None
36 else:
37 self.eof = ''
Guido van Rossum9c30c241998-12-22 05:19:29 +000038 self.commenters = '#'
Fred Drakedbbf76b2000-07-09 16:44:26 +000039 self.wordchars = ('abcdfeghijklmnopqrstuvwxyz'
40 'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_')
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000041 if self.posix:
42 self.wordchars += ('ßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ'
43 'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ')
Guido van Rossum9c30c241998-12-22 05:19:29 +000044 self.whitespace = ' \t\r\n'
Fred Drake24315232003-04-17 22:01:17 +000045 self.whitespace_split = False
Guido van Rossum9c30c241998-12-22 05:19:29 +000046 self.quotes = '\'"'
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000047 self.escape = '\\'
48 self.escapedquotes = '"'
Guido van Rossum9c30c241998-12-22 05:19:29 +000049 self.state = ' '
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +000050 self.pushback = []
Guido van Rossum9c30c241998-12-22 05:19:29 +000051 self.lineno = 1
52 self.debug = 0
53 self.token = ''
Fred Drakedbbf76b2000-07-09 16:44:26 +000054 self.filestack = []
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000055 self.source = None
56 if self.debug:
Fred Drake52dc76c2000-07-03 09:56:23 +000057 print 'shlex: reading from %s, line %d' \
58 % (self.instream, self.lineno)
Guido van Rossum9c30c241998-12-22 05:19:29 +000059
60 def push_token(self, tok):
61 "Push a token onto the stack popped by the get_token method"
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000062 if self.debug >= 1:
63 print "shlex: pushing token " + `tok`
Fred Drake24315232003-04-17 22:01:17 +000064 self.pushback.insert(0, tok)
Guido van Rossum9c30c241998-12-22 05:19:29 +000065
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000066 def push_source(self, newstream, newfile=None):
67 "Push an input source onto the lexer's input source stack."
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000068 if type(newstream) in StringTypes:
69 newstream = StringIO(newstream)
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000070 self.filestack.insert(0, (self.infile, self.instream, self.lineno))
71 self.infile = newfile
72 self.instream = newstream
73 self.lineno = 1
74 if self.debug:
Raymond Hettingerf13eb552002-06-02 00:40:05 +000075 if newfile is not None:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000076 print 'shlex: pushing to file %s' % (self.infile,)
77 else:
78 print 'shlex: pushing to stream %s' % (self.instream,)
79
80 def pop_source(self):
81 "Pop the input source stack."
82 self.instream.close()
83 (self.infile, self.instream, self.lineno) = self.filestack[0]
84 self.filestack = self.filestack[1:]
85 if self.debug:
86 print 'shlex: popping to %s, line %d' \
87 % (self.instream, self.lineno)
88 self.state = ' '
89
Guido van Rossum9c30c241998-12-22 05:19:29 +000090 def get_token(self):
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000091 "Get a token from the input stream (or from stack if it's nonempty)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000092 if self.pushback:
Fred Drake24315232003-04-17 22:01:17 +000093 tok = self.pushback.pop(0)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000094 if self.debug >= 1:
95 print "shlex: popping token " + `tok`
Guido van Rossum9c30c241998-12-22 05:19:29 +000096 return tok
Fred Drakedbbf76b2000-07-09 16:44:26 +000097 # No pushback. Get a token.
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000098 raw = self.read_token()
99 # Handle inclusions
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000100 if self.source is not None:
101 while raw == self.source:
102 spec = self.sourcehook(self.read_token())
103 if spec:
104 (newfile, newstream) = spec
105 self.push_source(newstream, newfile)
106 raw = self.get_token()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000107 # Maybe we got EOF instead?
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000108 while raw == self.eof:
Fred Drake24315232003-04-17 22:01:17 +0000109 if not self.filestack:
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000110 return self.eof
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000111 else:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000112 self.pop_source()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000113 raw = self.get_token()
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000114 # Neither inclusion nor EOF
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000115 if self.debug >= 1:
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000116 if raw != self.eof:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000117 print "shlex: token=" + `raw`
118 else:
119 print "shlex: token=EOF"
120 return raw
121
122 def read_token(self):
Fred Drake24315232003-04-17 22:01:17 +0000123 quoted = False
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000124 escapedstate = ' '
Guido van Rossum9c30c241998-12-22 05:19:29 +0000125 while 1:
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +0000126 nextchar = self.instream.read(1)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000127 if nextchar == '\n':
128 self.lineno = self.lineno + 1
129 if self.debug >= 3:
Fred Drake52dc76c2000-07-03 09:56:23 +0000130 print "shlex: in state", repr(self.state), \
Tim Peters70c43782001-01-17 08:48:39 +0000131 "I see character:", repr(nextchar)
Fred Drakedbbf76b2000-07-09 16:44:26 +0000132 if self.state is None:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000133 self.token = '' # past end of file
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000134 break
Guido van Rossum9c30c241998-12-22 05:19:29 +0000135 elif self.state == ' ':
136 if not nextchar:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000137 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000138 break
139 elif nextchar in self.whitespace:
140 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000141 print "shlex: I see whitespace in whitespace state"
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000142 if self.token or (self.posix and quoted):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000143 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000144 else:
145 continue
146 elif nextchar in self.commenters:
147 self.instream.readline()
148 self.lineno = self.lineno + 1
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000149 elif self.posix and nextchar in self.escape:
150 escapedstate = 'a'
151 self.state = nextchar
Guido van Rossum9c30c241998-12-22 05:19:29 +0000152 elif nextchar in self.wordchars:
153 self.token = nextchar
154 self.state = 'a'
155 elif nextchar in self.quotes:
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000156 if not self.posix:
157 self.token = nextchar
Guido van Rossum9c30c241998-12-22 05:19:29 +0000158 self.state = nextchar
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000159 elif self.whitespace_split:
160 self.token = nextchar
161 self.state = 'a'
Guido van Rossum9c30c241998-12-22 05:19:29 +0000162 else:
163 self.token = nextchar
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000164 if self.token or (self.posix and quoted):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000165 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000166 else:
167 continue
168 elif self.state in self.quotes:
Fred Drake24315232003-04-17 22:01:17 +0000169 quoted = True
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000170 if not nextchar: # end of file
Andrew M. Kuchling9d56cd12001-01-09 03:01:15 +0000171 if self.debug >= 2:
172 print "shlex: I see EOF in quotes state"
173 # XXX what error should be raised here?
174 raise ValueError, "No closing quotation"
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000175 if nextchar == self.state:
176 if not self.posix:
177 self.token = self.token + nextchar
178 self.state = ' '
179 break
180 else:
181 self.state = 'a'
182 elif self.posix and nextchar in self.escape and \
183 self.state in self.escapedquotes:
184 escapedstate = self.state
185 self.state = nextchar
186 else:
187 self.token = self.token + nextchar
188 elif self.state in self.escape:
189 if not nextchar: # end of file
190 if self.debug >= 2:
191 print "shlex: I see EOF in escape state"
192 # XXX what error should be raised here?
193 raise ValueError, "No escaped character"
194 # In posix shells, only the quote itself or the escape
195 # character may be escaped within quotes.
196 if escapedstate in self.quotes and \
197 nextchar != self.state and nextchar != escapedstate:
198 self.token = self.token + self.state
199 self.token = self.token + nextchar
200 self.state = escapedstate
Guido van Rossum9c30c241998-12-22 05:19:29 +0000201 elif self.state == 'a':
202 if not nextchar:
Tim Peters70c43782001-01-17 08:48:39 +0000203 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000204 break
205 elif nextchar in self.whitespace:
206 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000207 print "shlex: I see whitespace in word state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000208 self.state = ' '
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000209 if self.token or (self.posix and quoted):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000210 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000211 else:
212 continue
213 elif nextchar in self.commenters:
214 self.instream.readline()
215 self.lineno = self.lineno + 1
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000216 if self.posix:
217 self.state = ' '
218 if self.token or (self.posix and quoted):
219 break # emit current token
220 else:
221 continue
222 elif self.posix and nextchar in self.quotes:
223 self.state = nextchar
224 elif self.posix and nextchar in self.escape:
225 escapedstate = 'a'
226 self.state = nextchar
227 elif nextchar in self.wordchars or nextchar in self.quotes \
228 or self.whitespace_split:
Guido van Rossum9c30c241998-12-22 05:19:29 +0000229 self.token = self.token + nextchar
230 else:
Fred Drake24315232003-04-17 22:01:17 +0000231 self.pushback.insert(0, nextchar)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000232 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000233 print "shlex: I see punctuation in word state"
Guido van Rossumf247d751999-03-22 15:28:08 +0000234 self.state = ' '
Guido van Rossum9c30c241998-12-22 05:19:29 +0000235 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000236 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000237 else:
238 continue
Guido van Rossum9c30c241998-12-22 05:19:29 +0000239 result = self.token
240 self.token = ''
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000241 if self.posix and not quoted and result == '':
242 result = None
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000243 if self.debug > 1:
244 if result:
245 print "shlex: raw token=" + `result`
246 else:
247 print "shlex: raw token=EOF"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000248 return result
249
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000250 def sourcehook(self, newfile):
251 "Hook called on a filename to be sourced."
252 if newfile[0] == '"':
253 newfile = newfile[1:-1]
Fred Drake52dc76c2000-07-03 09:56:23 +0000254 # This implements cpp-like semantics for relative-path inclusion.
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000255 if type(self.infile) in StringTypes and not os.path.isabs(newfile):
Fred Drake52dc76c2000-07-03 09:56:23 +0000256 newfile = os.path.join(os.path.dirname(self.infile), newfile)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000257 return (newfile, open(newfile, "r"))
258
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000259 def error_leader(self, infile=None, lineno=None):
260 "Emit a C-compiler-like, Emacs-friendly error-message leader."
Raymond Hettingerf13eb552002-06-02 00:40:05 +0000261 if infile is None:
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000262 infile = self.infile
Raymond Hettingerf13eb552002-06-02 00:40:05 +0000263 if lineno is None:
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000264 lineno = self.lineno
265 return "\"%s\", line %d: " % (infile, lineno)
266
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000267 def __iter__(self):
268 return self
269
270 def next(self):
271 token = self.get_token()
272 if token == self.eof:
273 raise StopIteration
274 return token
275
276def split(s, posix=1, spaces=1):
277 lex = shlex(s, posix=posix)
278 lex.whitespace_split = spaces
279 return list(lex)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000280
Tim Peters70c43782001-01-17 08:48:39 +0000281if __name__ == '__main__':
Fred Drake52dc76c2000-07-03 09:56:23 +0000282 if len(sys.argv) == 1:
283 lexer = shlex()
284 else:
285 file = sys.argv[1]
286 lexer = shlex(open(file), file)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000287 while 1:
288 tt = lexer.get_token()
Fred Drake52dc76c2000-07-03 09:56:23 +0000289 if tt:
290 print "Token: " + repr(tt)
291 else:
Guido van Rossum9c30c241998-12-22 05:19:29 +0000292 break