blob: 6632b87596047c55d75d1b2d739c355f17807c9f [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
Raymond Hettinger756b3f32004-01-29 06:37:52 +000012from collections import deque
Guido van Rossum9c30c241998-12-22 05:19:29 +000013
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000014try:
15 from cStringIO import StringIO
16except ImportError:
17 from StringIO import StringIO
18
19__all__ = ["shlex", "split"]
Skip Montanaro0de65802001-02-15 22:15:14 +000020
Guido van Rossum9c30c241998-12-22 05:19:29 +000021class shlex:
Tim Peters70c43782001-01-17 08:48:39 +000022 "A lexical analyzer class for simple shell-like syntaxes."
Fred Drake24315232003-04-17 22:01:17 +000023 def __init__(self, instream=None, infile=None, posix=False):
Neal Norwitz10cf2182003-04-17 23:09:08 +000024 if isinstance(instream, basestring):
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000025 instream = StringIO(instream)
Raymond Hettingerf13eb552002-06-02 00:40:05 +000026 if instream is not None:
Guido van Rossum9c30c241998-12-22 05:19:29 +000027 self.instream = instream
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000028 self.infile = infile
Guido van Rossum9c30c241998-12-22 05:19:29 +000029 else:
30 self.instream = sys.stdin
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000031 self.infile = None
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000032 self.posix = posix
33 if posix:
34 self.eof = None
35 else:
36 self.eof = ''
Guido van Rossum9c30c241998-12-22 05:19:29 +000037 self.commenters = '#'
Fred Drakedbbf76b2000-07-09 16:44:26 +000038 self.wordchars = ('abcdfeghijklmnopqrstuvwxyz'
39 'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_')
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000040 if self.posix:
41 self.wordchars += ('ßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ'
42 'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ')
Guido van Rossum9c30c241998-12-22 05:19:29 +000043 self.whitespace = ' \t\r\n'
Fred Drake24315232003-04-17 22:01:17 +000044 self.whitespace_split = False
Guido van Rossum9c30c241998-12-22 05:19:29 +000045 self.quotes = '\'"'
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000046 self.escape = '\\'
47 self.escapedquotes = '"'
Guido van Rossum9c30c241998-12-22 05:19:29 +000048 self.state = ' '
Raymond Hettinger756b3f32004-01-29 06:37:52 +000049 self.pushback = deque()
Guido van Rossum9c30c241998-12-22 05:19:29 +000050 self.lineno = 1
51 self.debug = 0
52 self.token = ''
Raymond Hettinger756b3f32004-01-29 06:37:52 +000053 self.filestack = deque()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000054 self.source = None
55 if self.debug:
Fred Drake52dc76c2000-07-03 09:56:23 +000056 print 'shlex: reading from %s, line %d' \
57 % (self.instream, self.lineno)
Guido van Rossum9c30c241998-12-22 05:19:29 +000058
59 def push_token(self, tok):
60 "Push a token onto the stack popped by the get_token method"
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000061 if self.debug >= 1:
Walter Dörwald70a6b492004-02-12 17:35:32 +000062 print "shlex: pushing token " + repr(tok)
Raymond Hettinger756b3f32004-01-29 06:37:52 +000063 self.pushback.appendleft(tok)
Guido van Rossum9c30c241998-12-22 05:19:29 +000064
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000065 def push_source(self, newstream, newfile=None):
66 "Push an input source onto the lexer's input source stack."
Neal Norwitz10cf2182003-04-17 23:09:08 +000067 if isinstance(newstream, basestring):
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000068 newstream = StringIO(newstream)
Raymond Hettinger756b3f32004-01-29 06:37:52 +000069 self.filestack.appendleft((self.infile, self.instream, self.lineno))
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000070 self.infile = newfile
71 self.instream = newstream
72 self.lineno = 1
73 if self.debug:
Raymond Hettingerf13eb552002-06-02 00:40:05 +000074 if newfile is not None:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000075 print 'shlex: pushing to file %s' % (self.infile,)
76 else:
77 print 'shlex: pushing to stream %s' % (self.instream,)
78
79 def pop_source(self):
80 "Pop the input source stack."
81 self.instream.close()
Raymond Hettinger756b3f32004-01-29 06:37:52 +000082 (self.infile, self.instream, self.lineno) = self.filestack.popleft()
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000083 if self.debug:
84 print 'shlex: popping to %s, line %d' \
85 % (self.instream, self.lineno)
86 self.state = ' '
87
Guido van Rossum9c30c241998-12-22 05:19:29 +000088 def get_token(self):
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000089 "Get a token from the input stream (or from stack if it's nonempty)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000090 if self.pushback:
Raymond Hettinger756b3f32004-01-29 06:37:52 +000091 tok = self.pushback.popleft()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000092 if self.debug >= 1:
Walter Dörwald70a6b492004-02-12 17:35:32 +000093 print "shlex: popping token " + repr(tok)
Guido van Rossum9c30c241998-12-22 05:19:29 +000094 return tok
Fred Drakedbbf76b2000-07-09 16:44:26 +000095 # No pushback. Get a token.
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000096 raw = self.read_token()
97 # Handle inclusions
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +000098 if self.source is not None:
99 while raw == self.source:
100 spec = self.sourcehook(self.read_token())
101 if spec:
102 (newfile, newstream) = spec
103 self.push_source(newstream, newfile)
104 raw = self.get_token()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000105 # Maybe we got EOF instead?
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000106 while raw == self.eof:
Fred Drake24315232003-04-17 22:01:17 +0000107 if not self.filestack:
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000108 return self.eof
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000109 else:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000110 self.pop_source()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000111 raw = self.get_token()
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000112 # Neither inclusion nor EOF
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000113 if self.debug >= 1:
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000114 if raw != self.eof:
Walter Dörwald70a6b492004-02-12 17:35:32 +0000115 print "shlex: token=" + repr(raw)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000116 else:
117 print "shlex: token=EOF"
118 return raw
119
120 def read_token(self):
Fred Drake24315232003-04-17 22:01:17 +0000121 quoted = False
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000122 escapedstate = ' '
Neal Norwitz10cf2182003-04-17 23:09:08 +0000123 while True:
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +0000124 nextchar = self.instream.read(1)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000125 if nextchar == '\n':
126 self.lineno = self.lineno + 1
127 if self.debug >= 3:
Fred Drake52dc76c2000-07-03 09:56:23 +0000128 print "shlex: in state", repr(self.state), \
Tim Peters70c43782001-01-17 08:48:39 +0000129 "I see character:", repr(nextchar)
Fred Drakedbbf76b2000-07-09 16:44:26 +0000130 if self.state is None:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000131 self.token = '' # past end of file
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000132 break
Guido van Rossum9c30c241998-12-22 05:19:29 +0000133 elif self.state == ' ':
134 if not nextchar:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000135 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000136 break
137 elif nextchar in self.whitespace:
138 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000139 print "shlex: I see whitespace in whitespace state"
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000140 if self.token or (self.posix and quoted):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000141 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000142 else:
143 continue
144 elif nextchar in self.commenters:
145 self.instream.readline()
146 self.lineno = self.lineno + 1
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000147 elif self.posix and nextchar in self.escape:
148 escapedstate = 'a'
149 self.state = nextchar
Guido van Rossum9c30c241998-12-22 05:19:29 +0000150 elif nextchar in self.wordchars:
151 self.token = nextchar
152 self.state = 'a'
153 elif nextchar in self.quotes:
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000154 if not self.posix:
155 self.token = nextchar
Guido van Rossum9c30c241998-12-22 05:19:29 +0000156 self.state = nextchar
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000157 elif self.whitespace_split:
158 self.token = nextchar
159 self.state = 'a'
Guido van Rossum9c30c241998-12-22 05:19:29 +0000160 else:
161 self.token = nextchar
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000162 if self.token or (self.posix and quoted):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000163 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000164 else:
165 continue
166 elif self.state in self.quotes:
Fred Drake24315232003-04-17 22:01:17 +0000167 quoted = True
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000168 if not nextchar: # end of file
Andrew M. Kuchling9d56cd12001-01-09 03:01:15 +0000169 if self.debug >= 2:
170 print "shlex: I see EOF in quotes state"
171 # XXX what error should be raised here?
172 raise ValueError, "No closing quotation"
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000173 if nextchar == self.state:
174 if not self.posix:
175 self.token = self.token + nextchar
176 self.state = ' '
177 break
178 else:
179 self.state = 'a'
180 elif self.posix and nextchar in self.escape and \
181 self.state in self.escapedquotes:
182 escapedstate = self.state
183 self.state = nextchar
184 else:
185 self.token = self.token + nextchar
186 elif self.state in self.escape:
187 if not nextchar: # end of file
188 if self.debug >= 2:
189 print "shlex: I see EOF in escape state"
190 # XXX what error should be raised here?
191 raise ValueError, "No escaped character"
192 # In posix shells, only the quote itself or the escape
193 # character may be escaped within quotes.
194 if escapedstate in self.quotes and \
195 nextchar != self.state and nextchar != escapedstate:
196 self.token = self.token + self.state
197 self.token = self.token + nextchar
198 self.state = escapedstate
Guido van Rossum9c30c241998-12-22 05:19:29 +0000199 elif self.state == 'a':
200 if not nextchar:
Tim Peters70c43782001-01-17 08:48:39 +0000201 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000202 break
203 elif nextchar in self.whitespace:
204 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000205 print "shlex: I see whitespace in word state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000206 self.state = ' '
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000207 if self.token or (self.posix and quoted):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000208 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000209 else:
210 continue
211 elif nextchar in self.commenters:
212 self.instream.readline()
213 self.lineno = self.lineno + 1
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000214 if self.posix:
215 self.state = ' '
216 if self.token or (self.posix and quoted):
217 break # emit current token
218 else:
219 continue
220 elif self.posix and nextchar in self.quotes:
221 self.state = nextchar
222 elif self.posix and nextchar in self.escape:
223 escapedstate = 'a'
224 self.state = nextchar
225 elif nextchar in self.wordchars or nextchar in self.quotes \
226 or self.whitespace_split:
Guido van Rossum9c30c241998-12-22 05:19:29 +0000227 self.token = self.token + nextchar
228 else:
Raymond Hettinger756b3f32004-01-29 06:37:52 +0000229 self.pushback.appendleft(nextchar)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000230 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000231 print "shlex: I see punctuation in word state"
Guido van Rossumf247d751999-03-22 15:28:08 +0000232 self.state = ' '
Guido van Rossum9c30c241998-12-22 05:19:29 +0000233 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000234 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000235 else:
236 continue
Guido van Rossum9c30c241998-12-22 05:19:29 +0000237 result = self.token
238 self.token = ''
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000239 if self.posix and not quoted and result == '':
240 result = None
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000241 if self.debug > 1:
242 if result:
Walter Dörwald70a6b492004-02-12 17:35:32 +0000243 print "shlex: raw token=" + repr(result)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000244 else:
245 print "shlex: raw token=EOF"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000246 return result
247
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000248 def sourcehook(self, newfile):
249 "Hook called on a filename to be sourced."
250 if newfile[0] == '"':
251 newfile = newfile[1:-1]
Fred Drake52dc76c2000-07-03 09:56:23 +0000252 # This implements cpp-like semantics for relative-path inclusion.
Neal Norwitz10cf2182003-04-17 23:09:08 +0000253 if isinstance(self.infile, basestring) and not os.path.isabs(newfile):
Fred Drake52dc76c2000-07-03 09:56:23 +0000254 newfile = os.path.join(os.path.dirname(self.infile), newfile)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000255 return (newfile, open(newfile, "r"))
256
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000257 def error_leader(self, infile=None, lineno=None):
258 "Emit a C-compiler-like, Emacs-friendly error-message leader."
Raymond Hettingerf13eb552002-06-02 00:40:05 +0000259 if infile is None:
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000260 infile = self.infile
Raymond Hettingerf13eb552002-06-02 00:40:05 +0000261 if lineno is None:
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000262 lineno = self.lineno
263 return "\"%s\", line %d: " % (infile, lineno)
264
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000265 def __iter__(self):
266 return self
267
268 def next(self):
269 token = self.get_token()
270 if token == self.eof:
271 raise StopIteration
272 return token
273
Gustavo Niemeyer48f3dcc2003-04-20 01:57:03 +0000274def split(s, comments=False):
275 lex = shlex(s, posix=True)
276 lex.whitespace_split = True
277 if not comments:
278 lex.commenters = ''
Gustavo Niemeyer68d8cef2003-04-17 21:31:33 +0000279 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