blob: d90a0b9dc47db34ad8ff483cb11e138924fa908e [file] [log] [blame]
Guido van Rossume7b146f2000-02-04 15:28:42 +00001"""A lexical analyzer class for simple shell-like syntaxes."""
2
Tim Peters70c43782001-01-17 08:48:39 +00003# Module and documentation by Eric S. Raymond, 21 Dec 1998
Guido van Rossumeb4e11a2000-05-01 20:08:46 +00004# Input stacking and error message cleanup added by ESR, March 2000
Tim Peters70c43782001-01-17 08:48:39 +00005# push_source() and pop_source() made explicit by ESR, January 2001.
Guido van Rossum9c30c241998-12-22 05:19:29 +00006
Fred Drake52dc76c2000-07-03 09:56:23 +00007import os.path
Guido van Rossum73898c71999-05-03 18:14:16 +00008import sys
Guido van Rossum9c30c241998-12-22 05:19:29 +00009
10class shlex:
Tim Peters70c43782001-01-17 08:48:39 +000011 "A lexical analyzer class for simple shell-like syntaxes."
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000012 def __init__(self, instream=None, infile=None):
Guido van Rossum9c30c241998-12-22 05:19:29 +000013 if instream:
14 self.instream = instream
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000015 self.infile = infile
Guido van Rossum9c30c241998-12-22 05:19:29 +000016 else:
17 self.instream = sys.stdin
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000018 self.infile = None
Guido van Rossum9c30c241998-12-22 05:19:29 +000019 self.commenters = '#'
Fred Drakedbbf76b2000-07-09 16:44:26 +000020 self.wordchars = ('abcdfeghijklmnopqrstuvwxyz'
21 'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_')
Guido van Rossum9c30c241998-12-22 05:19:29 +000022 self.whitespace = ' \t\r\n'
23 self.quotes = '\'"'
24 self.state = ' '
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +000025 self.pushback = []
Guido van Rossum9c30c241998-12-22 05:19:29 +000026 self.lineno = 1
27 self.debug = 0
28 self.token = ''
Fred Drakedbbf76b2000-07-09 16:44:26 +000029 self.filestack = []
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000030 self.source = None
31 if self.debug:
Fred Drake52dc76c2000-07-03 09:56:23 +000032 print 'shlex: reading from %s, line %d' \
33 % (self.instream, self.lineno)
Guido van Rossum9c30c241998-12-22 05:19:29 +000034
35 def push_token(self, tok):
36 "Push a token onto the stack popped by the get_token method"
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000037 if self.debug >= 1:
38 print "shlex: pushing token " + `tok`
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +000039 self.pushback = [tok] + self.pushback
Guido van Rossum9c30c241998-12-22 05:19:29 +000040
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000041 def push_source(self, newstream, newfile=None):
42 "Push an input source onto the lexer's input source stack."
43 self.filestack.insert(0, (self.infile, self.instream, self.lineno))
44 self.infile = newfile
45 self.instream = newstream
46 self.lineno = 1
47 if self.debug:
48 if newfile:
49 print 'shlex: pushing to file %s' % (self.infile,)
50 else:
51 print 'shlex: pushing to stream %s' % (self.instream,)
52
53 def pop_source(self):
54 "Pop the input source stack."
55 self.instream.close()
56 (self.infile, self.instream, self.lineno) = self.filestack[0]
57 self.filestack = self.filestack[1:]
58 if self.debug:
59 print 'shlex: popping to %s, line %d' \
60 % (self.instream, self.lineno)
61 self.state = ' '
62
Guido van Rossum9c30c241998-12-22 05:19:29 +000063 def get_token(self):
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000064 "Get a token from the input stream (or from stack if it's nonempty)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000065 if self.pushback:
66 tok = self.pushback[0]
67 self.pushback = self.pushback[1:]
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000068 if self.debug >= 1:
69 print "shlex: popping token " + `tok`
Guido van Rossum9c30c241998-12-22 05:19:29 +000070 return tok
Fred Drakedbbf76b2000-07-09 16:44:26 +000071 # No pushback. Get a token.
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000072 raw = self.read_token()
73 # Handle inclusions
74 while raw == self.source:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000075 spec = self.sourcehook(self.read_token())
76 if spec:
77 (newfile, newstream) = spec
78 self.push_source(newstream, newfile)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000079 raw = self.get_token()
80 # Maybe we got EOF instead?
81 while raw == "":
82 if len(self.filestack) == 0:
83 return ""
84 else:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000085 self.pop_source()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000086 raw = self.get_token()
87 # Neither inclusion nor EOF
88 if self.debug >= 1:
89 if raw:
90 print "shlex: token=" + `raw`
91 else:
92 print "shlex: token=EOF"
93 return raw
94
95 def read_token(self):
96 "Read a token from the input stream (no pushback or inclusions)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000097 tok = ''
98 while 1:
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +000099 nextchar = self.instream.read(1)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000100 if nextchar == '\n':
101 self.lineno = self.lineno + 1
102 if self.debug >= 3:
Fred Drake52dc76c2000-07-03 09:56:23 +0000103 print "shlex: in state", repr(self.state), \
Tim Peters70c43782001-01-17 08:48:39 +0000104 "I see character:", repr(nextchar)
Fred Drakedbbf76b2000-07-09 16:44:26 +0000105 if self.state is None:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000106 self.token = '' # past end of file
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000107 break
Guido van Rossum9c30c241998-12-22 05:19:29 +0000108 elif self.state == ' ':
109 if not nextchar:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000110 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000111 break
112 elif nextchar in self.whitespace:
113 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000114 print "shlex: I see whitespace in whitespace state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000115 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000116 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000117 else:
118 continue
119 elif nextchar in self.commenters:
120 self.instream.readline()
121 self.lineno = self.lineno + 1
122 elif nextchar in self.wordchars:
123 self.token = nextchar
124 self.state = 'a'
125 elif nextchar in self.quotes:
126 self.token = nextchar
127 self.state = nextchar
128 else:
129 self.token = nextchar
130 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000131 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000132 else:
133 continue
134 elif self.state in self.quotes:
135 self.token = self.token + nextchar
136 if nextchar == self.state:
137 self.state = ' '
138 break
Tim Peters70c43782001-01-17 08:48:39 +0000139 elif not nextchar: # end of file
Andrew M. Kuchling9d56cd12001-01-09 03:01:15 +0000140 if self.debug >= 2:
141 print "shlex: I see EOF in quotes state"
142 # XXX what error should be raised here?
143 raise ValueError, "No closing quotation"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000144 elif self.state == 'a':
145 if not nextchar:
Tim Peters70c43782001-01-17 08:48:39 +0000146 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000147 break
148 elif nextchar in self.whitespace:
149 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000150 print "shlex: I see whitespace in word state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000151 self.state = ' '
152 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000153 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000154 else:
155 continue
156 elif nextchar in self.commenters:
157 self.instream.readline()
158 self.lineno = self.lineno + 1
159 elif nextchar in self.wordchars or nextchar in self.quotes:
160 self.token = self.token + nextchar
161 else:
162 self.pushback = [nextchar] + self.pushback
163 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000164 print "shlex: I see punctuation in word state"
Guido van Rossumf247d751999-03-22 15:28:08 +0000165 self.state = ' '
Guido van Rossum9c30c241998-12-22 05:19:29 +0000166 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000167 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000168 else:
169 continue
Guido van Rossum9c30c241998-12-22 05:19:29 +0000170 result = self.token
171 self.token = ''
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000172 if self.debug > 1:
173 if result:
174 print "shlex: raw token=" + `result`
175 else:
176 print "shlex: raw token=EOF"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000177 return result
178
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000179 def sourcehook(self, newfile):
180 "Hook called on a filename to be sourced."
181 if newfile[0] == '"':
182 newfile = newfile[1:-1]
Fred Drake52dc76c2000-07-03 09:56:23 +0000183 # This implements cpp-like semantics for relative-path inclusion.
184 if type(self.infile) == type("") and not os.path.isabs(newfile):
185 newfile = os.path.join(os.path.dirname(self.infile), newfile)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000186 return (newfile, open(newfile, "r"))
187
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000188 def error_leader(self, infile=None, lineno=None):
189 "Emit a C-compiler-like, Emacs-friendly error-message leader."
190 if not infile:
191 infile = self.infile
192 if not lineno:
193 lineno = self.lineno
194 return "\"%s\", line %d: " % (infile, lineno)
195
Guido van Rossum9c30c241998-12-22 05:19:29 +0000196
Tim Peters70c43782001-01-17 08:48:39 +0000197if __name__ == '__main__':
Fred Drake52dc76c2000-07-03 09:56:23 +0000198 if len(sys.argv) == 1:
199 lexer = shlex()
200 else:
201 file = sys.argv[1]
202 lexer = shlex(open(file), file)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000203 while 1:
204 tt = lexer.get_token()
Fred Drake52dc76c2000-07-03 09:56:23 +0000205 if tt:
206 print "Token: " + repr(tt)
207 else:
Guido van Rossum9c30c241998-12-22 05:19:29 +0000208 break