Lib/shlex.py

   1 """A lexical analyzer class for simple shell-like syntaxes."""
   2
   3 # Module and documentation by Eric S. Raymond, 21 Dec 1998
   4 # Input stacking and error message cleanup added by ESR, March 2000
   5 # push_source() and pop_source() made explicit by ESR, January 2001.
   6
   7 import os.path
   8 import sys
   9
  10 __all__ = ["shlex"]
  11
  12 class shlex:
  13     "A lexical analyzer class for simple shell-like syntaxes."
  14     def __init__(self, instream=None, infile=None):
  15         if instream:
  16             self.instream = instream
  17             self.infile = infile
  18         else:
  19             self.instream = sys.stdin
  20             self.infile = None
  21         self.commenters = '#'
  22         self.wordchars = ('abcdfeghijklmnopqrstuvwxyz'
  23                           'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_')
  24         self.whitespace = ' \t\r\n'
  25         self.quotes = '\'"'
  26         self.state = ' '
  27         self.pushback = []
  28         self.lineno = 1
  29         self.debug = 0
  30         self.token = ''
  31         self.filestack = []
  32         self.source = None
  33         if self.debug:
  34             print 'shlex: reading from %s, line %d' \
  35                   % (self.instream, self.lineno)
  36
  37     def push_token(self, tok):
  38         "Push a token onto the stack popped by the get_token method"
  39         if self.debug >= 1:
  40             print "shlex: pushing token " + `tok`
  41         self.pushback = [tok] + self.pushback
  42
  43     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
  65     def get_token(self):
  66         "Get a token from the input stream (or from stack if it's nonempty)"
  67         if self.pushback:
  68             tok = self.pushback[0]
  69             self.pushback = self.pushback[1:]
  70             if self.debug >= 1:
  71                 print "shlex: popping token " + `tok`
  72             return tok
  73         # No pushback.  Get a token.
  74         raw = self.read_token()
  75         # Handle inclusions
  76         while raw == self.source:
  77             spec = self.sourcehook(self.read_token())
  78             if spec:
  79                 (newfile, newstream) = spec
  80                 self.push_source(newstream, newfile)
  81             raw = self.get_token()
  82         # Maybe we got EOF instead?
  83         while raw == "":
  84             if len(self.filestack) == 0:
  85                 return ""
  86             else:
  87                 self.pop_source()
  88                 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)"
  99         tok = ''
 100         while 1:
 101             nextchar = self.instream.read(1)
 102             if nextchar == '\n':
 103                 self.lineno = self.lineno + 1
 104             if self.debug >= 3:
 105                 print "shlex: in state", repr(self.state), \
 106                       "I see character:", repr(nextchar)
 107             if self.state is None:
 108                 self.token = ''        # past end of file
 109                 break
 110             elif self.state == ' ':
 111                 if not nextchar:
 112                     self.state = None  # end of file
 113                     break
 114                 elif nextchar in self.whitespace:
 115                     if self.debug >= 2:
 116                         print "shlex: I see whitespace in whitespace state"
 117                     if self.token:
 118                         break   # emit current token
 119                     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:
 133                         break   # emit current token
 134                     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
 141                 elif not nextchar:      # end of file
 142                     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"
 146             elif self.state == 'a':
 147                 if not nextchar:
 148                     self.state = None   # end of file
 149                     break
 150                 elif nextchar in self.whitespace:
 151                     if self.debug >= 2:
 152                         print "shlex: I see whitespace in word state"
 153                     self.state = ' '
 154                     if self.token:
 155                         break   # emit current token
 156                     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:
 166                         print "shlex: I see punctuation in word state"
 167                     self.state = ' '
 168                     if self.token:
 169                         break   # emit current token
 170                     else:
 171                         continue
 172         result = self.token
 173         self.token = ''
 174         if self.debug > 1:
 175             if result:
 176                 print "shlex: raw token=" + `result`
 177             else:
 178                 print "shlex: raw token=EOF"
 179         return result
 180
 181     def sourcehook(self, newfile):
 182         "Hook called on a filename to be sourced."
 183         if newfile[0] == '"':
 184             newfile = newfile[1:-1]
 185         # 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)
 188         return (newfile, open(newfile, "r"))
 189
 190     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
 198
 199 if __name__ == '__main__':
 200     if len(sys.argv) == 1:
 201         lexer = shlex()
 202     else:
 203         file = sys.argv[1]
 204         lexer = shlex(open(file), file)
 205     while 1:
 206         tt = lexer.get_token()
 207         if tt:
 208             print "Token: " + repr(tt)
 209         else:
 210             break