blob: 575fc35ee639d11ec89d437167fd0b643a2289c3 [file] [log] [blame]
Guido van Rossume7b146f2000-02-04 15:28:42 +00001"""A lexical analyzer class for simple shell-like syntaxes."""
2
Guido van Rossum9c30c241998-12-22 05:19:29 +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
Guido van Rossum9c30c241998-12-22 05:19:29 +00005
Guido van Rossum73898c71999-05-03 18:14:16 +00006import sys
Guido van Rossum9c30c241998-12-22 05:19:29 +00007
8class shlex:
9 "A lexical analyzer class for simple shell-like syntaxes."
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000010 def __init__(self, instream=None, infile=None):
Guido van Rossum9c30c241998-12-22 05:19:29 +000011 if instream:
12 self.instream = instream
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000013 self.infile = infile
Guido van Rossum9c30c241998-12-22 05:19:29 +000014 else:
15 self.instream = sys.stdin
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000016 self.infile = None
Guido van Rossum9c30c241998-12-22 05:19:29 +000017 self.commenters = '#'
18 self.wordchars = 'abcdfeghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_'
19 self.whitespace = ' \t\r\n'
20 self.quotes = '\'"'
21 self.state = ' '
22 self.pushback = [];
23 self.lineno = 1
24 self.debug = 0
25 self.token = ''
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000026 self.filestack = []
27 self.source = None
28 if self.debug:
29 print 'shlex: reading from %s, line %d' % (self.instream,self.lineno)
Guido van Rossum9c30c241998-12-22 05:19:29 +000030
31 def push_token(self, tok):
32 "Push a token onto the stack popped by the get_token method"
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000033 if self.debug >= 1:
34 print "shlex: pushing token " + `tok`
Guido van Rossum9c30c241998-12-22 05:19:29 +000035 self.pushback = [tok] + self.pushback;
36
37 def get_token(self):
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000038 "Get a token from the input stream (or from stack if it's nonempty)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000039 if self.pushback:
40 tok = self.pushback[0]
41 self.pushback = self.pushback[1:]
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000042 if self.debug >= 1:
43 print "shlex: popping token " + `tok`
Guido van Rossum9c30c241998-12-22 05:19:29 +000044 return tok
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000045 # No pushback. Get a token.
46 raw = self.read_token()
47 # Handle inclusions
48 while raw == self.source:
49 (newfile, newstream) = self.sourcehook(self.read_token())
50 self.filestack = [(self.infile,self.instream,self.lineno)] + self.filestack
51 self.infile = newfile
52 self.instream = newstream
53 self.lineno = 1
54 if self.debug:
55 print 'shlex: pushing to file %s' % (self.infile,)
56 raw = self.get_token()
57 # Maybe we got EOF instead?
58 while raw == "":
59 if len(self.filestack) == 0:
60 return ""
61 else:
62 self.instream.close()
63 (self.infile, self.instream, self.lineno) = self.filestack[0]
64 self.filestack = self.filestack[1:]
65 if self.debug:
66 print 'shlex: popping to %s, line %d' % (self.instream, self.lineno)
67 self.state = ' '
68 raw = self.get_token()
69 # Neither inclusion nor EOF
70 if self.debug >= 1:
71 if raw:
72 print "shlex: token=" + `raw`
73 else:
74 print "shlex: token=EOF"
75 return raw
76
77 def read_token(self):
78 "Read a token from the input stream (no pushback or inclusions)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000079 tok = ''
80 while 1:
81 nextchar = self.instream.read(1);
82 if nextchar == '\n':
83 self.lineno = self.lineno + 1
84 if self.debug >= 3:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000085 print "shlex: in state " + repr(self.state) + " I see character: " + repr(nextchar)
Guido van Rossum9c30c241998-12-22 05:19:29 +000086 if self.state == None:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000087 self.token = ''; # past end of file
88 break
Guido van Rossum9c30c241998-12-22 05:19:29 +000089 elif self.state == ' ':
90 if not nextchar:
91 self.state = None; # end of file
92 break
93 elif nextchar in self.whitespace:
94 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000095 print "shlex: I see whitespace in whitespace state"
Guido van Rossum9c30c241998-12-22 05:19:29 +000096 if self.token:
97 break # emit current token
98 else:
99 continue
100 elif nextchar in self.commenters:
101 self.instream.readline()
102 self.lineno = self.lineno + 1
103 elif nextchar in self.wordchars:
104 self.token = nextchar
105 self.state = 'a'
106 elif nextchar in self.quotes:
107 self.token = nextchar
108 self.state = nextchar
109 else:
110 self.token = nextchar
111 if self.token:
112 break # emit current token
113 else:
114 continue
115 elif self.state in self.quotes:
116 self.token = self.token + nextchar
117 if nextchar == self.state:
118 self.state = ' '
119 break
120 elif self.state == 'a':
121 if not nextchar:
122 self.state = None; # end of file
123 break
124 elif nextchar in self.whitespace:
125 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000126 print "shlex: I see whitespace in word state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000127 self.state = ' '
128 if self.token:
129 break # emit current token
130 else:
131 continue
132 elif nextchar in self.commenters:
133 self.instream.readline()
134 self.lineno = self.lineno + 1
135 elif nextchar in self.wordchars or nextchar in self.quotes:
136 self.token = self.token + nextchar
137 else:
138 self.pushback = [nextchar] + self.pushback
139 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000140 print "shlex: I see punctuation in word state"
Guido van Rossumf247d751999-03-22 15:28:08 +0000141 self.state = ' '
Guido van Rossum9c30c241998-12-22 05:19:29 +0000142 if self.token:
143 break # emit current token
144 else:
145 continue
Guido van Rossum9c30c241998-12-22 05:19:29 +0000146 result = self.token
147 self.token = ''
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000148 if self.debug > 1:
149 if result:
150 print "shlex: raw token=" + `result`
151 else:
152 print "shlex: raw token=EOF"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000153 return result
154
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000155 def sourcehook(self, newfile):
156 "Hook called on a filename to be sourced."
157 if newfile[0] == '"':
158 newfile = newfile[1:-1]
159 return (newfile, open(newfile, "r"))
160
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000161 def error_leader(self, infile=None, lineno=None):
162 "Emit a C-compiler-like, Emacs-friendly error-message leader."
163 if not infile:
164 infile = self.infile
165 if not lineno:
166 lineno = self.lineno
167 return "\"%s\", line %d: " % (infile, lineno)
168
Guido van Rossum9c30c241998-12-22 05:19:29 +0000169if __name__ == '__main__':
170
171 lexer = shlex()
172 while 1:
173 tt = lexer.get_token()
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000174 print "Token: " + repr(tt)
175 if not tt:
Guido van Rossum9c30c241998-12-22 05:19:29 +0000176 break
177