blob: 97b8f2e5c041a678829bc1440b4eb1c55d7e1b62 [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
Skip Montanaro0de65802001-02-15 22:15:14 +000010__all__ = ["shlex"]
11
Guido van Rossum9c30c241998-12-22 05:19:29 +000012class shlex:
Tim Peters70c43782001-01-17 08:48:39 +000013 "A lexical analyzer class for simple shell-like syntaxes."
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000014 def __init__(self, instream=None, infile=None):
Guido van Rossum9c30c241998-12-22 05:19:29 +000015 if instream:
16 self.instream = instream
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000017 self.infile = infile
Guido van Rossum9c30c241998-12-22 05:19:29 +000018 else:
19 self.instream = sys.stdin
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000020 self.infile = None
Guido van Rossum9c30c241998-12-22 05:19:29 +000021 self.commenters = '#'
Fred Drakedbbf76b2000-07-09 16:44:26 +000022 self.wordchars = ('abcdfeghijklmnopqrstuvwxyz'
23 'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_')
Guido van Rossum9c30c241998-12-22 05:19:29 +000024 self.whitespace = ' \t\r\n'
25 self.quotes = '\'"'
26 self.state = ' '
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +000027 self.pushback = []
Guido van Rossum9c30c241998-12-22 05:19:29 +000028 self.lineno = 1
29 self.debug = 0
30 self.token = ''
Fred Drakedbbf76b2000-07-09 16:44:26 +000031 self.filestack = []
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000032 self.source = None
33 if self.debug:
Fred Drake52dc76c2000-07-03 09:56:23 +000034 print 'shlex: reading from %s, line %d' \
35 % (self.instream, self.lineno)
Guido van Rossum9c30c241998-12-22 05:19:29 +000036
37 def push_token(self, tok):
38 "Push a token onto the stack popped by the get_token method"
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000039 if self.debug >= 1:
40 print "shlex: pushing token " + `tok`
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +000041 self.pushback = [tok] + self.pushback
Guido van Rossum9c30c241998-12-22 05:19:29 +000042
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000043 def push_source(self, newstream, newfile=None):
44 "Push an input source onto the lexer's input source stack."
45 self.filestack.insert(0, (self.infile, self.instream, self.lineno))
46 self.infile = newfile
47 self.instream = newstream
48 self.lineno = 1
49 if self.debug:
50 if newfile:
51 print 'shlex: pushing to file %s' % (self.infile,)
52 else:
53 print 'shlex: pushing to stream %s' % (self.instream,)
54
55 def pop_source(self):
56 "Pop the input source stack."
57 self.instream.close()
58 (self.infile, self.instream, self.lineno) = self.filestack[0]
59 self.filestack = self.filestack[1:]
60 if self.debug:
61 print 'shlex: popping to %s, line %d' \
62 % (self.instream, self.lineno)
63 self.state = ' '
64
Guido van Rossum9c30c241998-12-22 05:19:29 +000065 def get_token(self):
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000066 "Get a token from the input stream (or from stack if it's nonempty)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000067 if self.pushback:
68 tok = self.pushback[0]
69 self.pushback = self.pushback[1:]
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000070 if self.debug >= 1:
71 print "shlex: popping token " + `tok`
Guido van Rossum9c30c241998-12-22 05:19:29 +000072 return tok
Fred Drakedbbf76b2000-07-09 16:44:26 +000073 # No pushback. Get a token.
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000074 raw = self.read_token()
75 # Handle inclusions
76 while raw == self.source:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000077 spec = self.sourcehook(self.read_token())
78 if spec:
79 (newfile, newstream) = spec
80 self.push_source(newstream, newfile)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000081 raw = self.get_token()
82 # Maybe we got EOF instead?
83 while raw == "":
84 if len(self.filestack) == 0:
85 return ""
86 else:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +000087 self.pop_source()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000088 raw = self.get_token()
89 # Neither inclusion nor EOF
90 if self.debug >= 1:
91 if raw:
92 print "shlex: token=" + `raw`
93 else:
94 print "shlex: token=EOF"
95 return raw
96
97 def read_token(self):
98 "Read a token from the input stream (no pushback or inclusions)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000099 tok = ''
100 while 1:
Andrew M. Kuchling49d27c82000-12-23 14:20:24 +0000101 nextchar = self.instream.read(1)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000102 if nextchar == '\n':
103 self.lineno = self.lineno + 1
104 if self.debug >= 3:
Fred Drake52dc76c2000-07-03 09:56:23 +0000105 print "shlex: in state", repr(self.state), \
Tim Peters70c43782001-01-17 08:48:39 +0000106 "I see character:", repr(nextchar)
Fred Drakedbbf76b2000-07-09 16:44:26 +0000107 if self.state is None:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000108 self.token = '' # past end of file
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000109 break
Guido van Rossum9c30c241998-12-22 05:19:29 +0000110 elif self.state == ' ':
111 if not nextchar:
Eric S. Raymondbddbaf72001-01-16 15:19:13 +0000112 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000113 break
114 elif nextchar in self.whitespace:
115 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000116 print "shlex: I see whitespace in whitespace state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000117 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000118 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000119 else:
120 continue
121 elif nextchar in self.commenters:
122 self.instream.readline()
123 self.lineno = self.lineno + 1
124 elif nextchar in self.wordchars:
125 self.token = nextchar
126 self.state = 'a'
127 elif nextchar in self.quotes:
128 self.token = nextchar
129 self.state = nextchar
130 else:
131 self.token = nextchar
132 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000133 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000134 else:
135 continue
136 elif self.state in self.quotes:
137 self.token = self.token + nextchar
138 if nextchar == self.state:
139 self.state = ' '
140 break
Tim Peters70c43782001-01-17 08:48:39 +0000141 elif not nextchar: # end of file
Andrew M. Kuchling9d56cd12001-01-09 03:01:15 +0000142 if self.debug >= 2:
143 print "shlex: I see EOF in quotes state"
144 # XXX what error should be raised here?
145 raise ValueError, "No closing quotation"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000146 elif self.state == 'a':
147 if not nextchar:
Tim Peters70c43782001-01-17 08:48:39 +0000148 self.state = None # end of file
Guido van Rossum9c30c241998-12-22 05:19:29 +0000149 break
150 elif nextchar in self.whitespace:
151 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000152 print "shlex: I see whitespace in word state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000153 self.state = ' '
154 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000155 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000156 else:
157 continue
158 elif nextchar in self.commenters:
159 self.instream.readline()
160 self.lineno = self.lineno + 1
161 elif nextchar in self.wordchars or nextchar in self.quotes:
162 self.token = self.token + nextchar
163 else:
164 self.pushback = [nextchar] + self.pushback
165 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000166 print "shlex: I see punctuation in word state"
Guido van Rossumf247d751999-03-22 15:28:08 +0000167 self.state = ' '
Guido van Rossum9c30c241998-12-22 05:19:29 +0000168 if self.token:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000169 break # emit current token
Guido van Rossum9c30c241998-12-22 05:19:29 +0000170 else:
171 continue
Guido van Rossum9c30c241998-12-22 05:19:29 +0000172 result = self.token
173 self.token = ''
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000174 if self.debug > 1:
175 if result:
176 print "shlex: raw token=" + `result`
177 else:
178 print "shlex: raw token=EOF"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000179 return result
180
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000181 def sourcehook(self, newfile):
182 "Hook called on a filename to be sourced."
183 if newfile[0] == '"':
184 newfile = newfile[1:-1]
Fred Drake52dc76c2000-07-03 09:56:23 +0000185 # This implements cpp-like semantics for relative-path inclusion.
186 if type(self.infile) == type("") and not os.path.isabs(newfile):
187 newfile = os.path.join(os.path.dirname(self.infile), newfile)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000188 return (newfile, open(newfile, "r"))
189
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000190 def error_leader(self, infile=None, lineno=None):
191 "Emit a C-compiler-like, Emacs-friendly error-message leader."
192 if not infile:
193 infile = self.infile
194 if not lineno:
195 lineno = self.lineno
196 return "\"%s\", line %d: " % (infile, lineno)
197
Guido van Rossum9c30c241998-12-22 05:19:29 +0000198
Tim Peters70c43782001-01-17 08:48:39 +0000199if __name__ == '__main__':
Fred Drake52dc76c2000-07-03 09:56:23 +0000200 if len(sys.argv) == 1:
201 lexer = shlex()
202 else:
203 file = sys.argv[1]
204 lexer = shlex(open(file), file)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000205 while 1:
206 tt = lexer.get_token()
Fred Drake52dc76c2000-07-03 09:56:23 +0000207 if tt:
208 print "Token: " + repr(tt)
209 else:
Guido van Rossum9c30c241998-12-22 05:19:29 +0000210 break