blob: 876dcdf48ba555906512e4d0e6201cdcf47e7dcd [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
Fred Drake52dc76c2000-07-03 09:56:23 +00006import os.path
Guido van Rossum73898c71999-05-03 18:14:16 +00007import sys
Guido van Rossum9c30c241998-12-22 05:19:29 +00008
Fred Drake52dc76c2000-07-03 09:56:23 +00009
Guido van Rossum9c30c241998-12-22 05:19:29 +000010class shlex:
11 "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 = '#'
20 self.wordchars = 'abcdfeghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_'
21 self.whitespace = ' \t\r\n'
22 self.quotes = '\'"'
23 self.state = ' '
24 self.pushback = [];
25 self.lineno = 1
26 self.debug = 0
27 self.token = ''
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000028 self.filestack = []
29 self.source = None
30 if self.debug:
Fred Drake52dc76c2000-07-03 09:56:23 +000031 print 'shlex: reading from %s, line %d' \
32 % (self.instream, self.lineno)
Guido van Rossum9c30c241998-12-22 05:19:29 +000033
34 def push_token(self, tok):
35 "Push a token onto the stack popped by the get_token method"
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000036 if self.debug >= 1:
37 print "shlex: pushing token " + `tok`
Guido van Rossum9c30c241998-12-22 05:19:29 +000038 self.pushback = [tok] + self.pushback;
39
40 def get_token(self):
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000041 "Get a token from the input stream (or from stack if it's nonempty)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000042 if self.pushback:
43 tok = self.pushback[0]
44 self.pushback = self.pushback[1:]
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000045 if self.debug >= 1:
46 print "shlex: popping token " + `tok`
Guido van Rossum9c30c241998-12-22 05:19:29 +000047 return tok
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000048 # No pushback. Get a token.
49 raw = self.read_token()
50 # Handle inclusions
51 while raw == self.source:
52 (newfile, newstream) = self.sourcehook(self.read_token())
Fred Drake52dc76c2000-07-03 09:56:23 +000053 self.filestack.insert(0, (self.infile, self.instream, self.lineno))
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000054 self.infile = newfile
55 self.instream = newstream
56 self.lineno = 1
57 if self.debug:
58 print 'shlex: pushing to file %s' % (self.infile,)
59 raw = self.get_token()
60 # Maybe we got EOF instead?
61 while raw == "":
62 if len(self.filestack) == 0:
63 return ""
64 else:
65 self.instream.close()
66 (self.infile, self.instream, self.lineno) = self.filestack[0]
67 self.filestack = self.filestack[1:]
68 if self.debug:
Fred Drake52dc76c2000-07-03 09:56:23 +000069 print 'shlex: popping to %s, line %d' \
70 % (self.instream, self.lineno)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000071 self.state = ' '
72 raw = self.get_token()
73 # Neither inclusion nor EOF
74 if self.debug >= 1:
75 if raw:
76 print "shlex: token=" + `raw`
77 else:
78 print "shlex: token=EOF"
79 return raw
80
81 def read_token(self):
82 "Read a token from the input stream (no pushback or inclusions)"
Guido van Rossum9c30c241998-12-22 05:19:29 +000083 tok = ''
84 while 1:
85 nextchar = self.instream.read(1);
86 if nextchar == '\n':
87 self.lineno = self.lineno + 1
88 if self.debug >= 3:
Fred Drake52dc76c2000-07-03 09:56:23 +000089 print "shlex: in state", repr(self.state), \
90 "I see character:", repr(nextchar)
Guido van Rossum9c30c241998-12-22 05:19:29 +000091 if self.state == None:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +000092 self.token = ''; # past end of file
93 break
Guido van Rossum9c30c241998-12-22 05:19:29 +000094 elif self.state == ' ':
95 if not nextchar:
96 self.state = None; # end of file
97 break
98 elif nextchar in self.whitespace:
99 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000100 print "shlex: I see whitespace in whitespace state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000101 if self.token:
102 break # emit current token
103 else:
104 continue
105 elif nextchar in self.commenters:
106 self.instream.readline()
107 self.lineno = self.lineno + 1
108 elif nextchar in self.wordchars:
109 self.token = nextchar
110 self.state = 'a'
111 elif nextchar in self.quotes:
112 self.token = nextchar
113 self.state = nextchar
114 else:
115 self.token = nextchar
116 if self.token:
117 break # emit current token
118 else:
119 continue
120 elif self.state in self.quotes:
121 self.token = self.token + nextchar
122 if nextchar == self.state:
123 self.state = ' '
124 break
125 elif self.state == 'a':
126 if not nextchar:
127 self.state = None; # end of file
128 break
129 elif nextchar in self.whitespace:
130 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000131 print "shlex: I see whitespace in word state"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000132 self.state = ' '
133 if self.token:
134 break # emit current token
135 else:
136 continue
137 elif nextchar in self.commenters:
138 self.instream.readline()
139 self.lineno = self.lineno + 1
140 elif nextchar in self.wordchars or nextchar in self.quotes:
141 self.token = self.token + nextchar
142 else:
143 self.pushback = [nextchar] + self.pushback
144 if self.debug >= 2:
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000145 print "shlex: I see punctuation in word state"
Guido van Rossumf247d751999-03-22 15:28:08 +0000146 self.state = ' '
Guido van Rossum9c30c241998-12-22 05:19:29 +0000147 if self.token:
148 break # emit current token
149 else:
150 continue
Guido van Rossum9c30c241998-12-22 05:19:29 +0000151 result = self.token
152 self.token = ''
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000153 if self.debug > 1:
154 if result:
155 print "shlex: raw token=" + `result`
156 else:
157 print "shlex: raw token=EOF"
Guido van Rossum9c30c241998-12-22 05:19:29 +0000158 return result
159
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000160 def sourcehook(self, newfile):
161 "Hook called on a filename to be sourced."
162 if newfile[0] == '"':
163 newfile = newfile[1:-1]
Fred Drake52dc76c2000-07-03 09:56:23 +0000164 # This implements cpp-like semantics for relative-path inclusion.
165 if type(self.infile) == type("") and not os.path.isabs(newfile):
166 newfile = os.path.join(os.path.dirname(self.infile), newfile)
Guido van Rossumeb4e11a2000-05-01 20:08:46 +0000167 return (newfile, open(newfile, "r"))
168
Guido van Rossum4b83ecb2000-05-01 20:14:12 +0000169 def error_leader(self, infile=None, lineno=None):
170 "Emit a C-compiler-like, Emacs-friendly error-message leader."
171 if not infile:
172 infile = self.infile
173 if not lineno:
174 lineno = self.lineno
175 return "\"%s\", line %d: " % (infile, lineno)
176
Guido van Rossum9c30c241998-12-22 05:19:29 +0000177
Fred Drake52dc76c2000-07-03 09:56:23 +0000178if __name__ == '__main__':
179 if len(sys.argv) == 1:
180 lexer = shlex()
181 else:
182 file = sys.argv[1]
183 lexer = shlex(open(file), file)
Guido van Rossum9c30c241998-12-22 05:19:29 +0000184 while 1:
185 tt = lexer.get_token()
Fred Drake52dc76c2000-07-03 09:56:23 +0000186 if tt:
187 print "Token: " + repr(tt)
188 else:
Guido van Rossum9c30c241998-12-22 05:19:29 +0000189 break