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
   6 import os.path
   7 import sys
   8
   9
  10 class shlex:
  11     "A lexical analyzer class for simple shell-like syntaxes."
  12     def __init__(self, instream=None, infile=None):
  13         if instream:
  14             self.instream = instream
  15             self.infile = infile
  16         else:
  17             self.instream = sys.stdin
  18             self.infile = None
  19         self.commenters = '#'
  20         self.wordchars = ('abcdfeghijklmnopqrstuvwxyz'
  21                           'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_')
  22         self.whitespace = ' \t\r\n'
  23         self.quotes = '\'"'
  24         self.state = ' '
  25         self.pushback = [];
  26         self.lineno = 1
  27         self.debug = 0
  28         self.token = ''
  29         self.filestack = []
  30         self.source = None
  31         if self.debug:
  32             print 'shlex: reading from %s, line %d' \
  33                   % (self.instream, self.lineno)
  34
  35     def push_token(self, tok):
  36         "Push a token onto the stack popped by the get_token method"
  37         if self.debug >= 1:
  38             print "shlex: pushing token " + `tok`
  39         self.pushback = [tok] + self.pushback;
  40
  41     def get_token(self):
  42         "Get a token from the input stream (or from stack if it's nonempty)"
  43         if self.pushback:
  44             tok = self.pushback[0]
  45             self.pushback = self.pushback[1:]
  46             if self.debug >= 1:
  47                 print "shlex: popping token " + `tok`
  48             return tok
  49         # No pushback.  Get a token.
  50         raw = self.read_token()
  51         # Handle inclusions
  52         while raw == self.source:
  53             (newfile, newstream) = self.sourcehook(self.read_token())
  54             self.filestack.insert(0, (self.infile, self.instream, self.lineno))
  55             self.infile = newfile
  56             self.instream = newstream
  57             self.lineno = 1
  58             if self.debug:
  59                 print 'shlex: pushing to file %s' % (self.infile,)
  60             raw = self.get_token()
  61         # Maybe we got EOF instead?
  62         while raw == "":
  63             if len(self.filestack) == 0:
  64                 return ""
  65             else:
  66                 self.instream.close()
  67                 (self.infile, self.instream, self.lineno) = self.filestack[0]
  68                 self.filestack = self.filestack[1:]
  69                 if self.debug:
  70                     print 'shlex: popping to %s, line %d' \
  71                           % (self.instream, self.lineno)
  72                 self.state = ' '
  73                 raw = self.get_token()
  74          # Neither inclusion nor EOF
  75         if self.debug >= 1:
  76             if raw:
  77                 print "shlex: token=" + `raw`
  78             else:
  79                 print "shlex: token=EOF"
  80         return raw
  81
  82     def read_token(self):
  83         "Read a token from the input stream (no pushback or inclusions)"
  84         tok = ''
  85         while 1:
  86             nextchar = self.instream.read(1);
  87             if nextchar == '\n':
  88                 self.lineno = self.lineno + 1
  89             if self.debug >= 3:
  90                 print "shlex: in state", repr(self.state), \
  91                       "I see character:", repr(nextchar)
  92             if self.state is None:
  93                 self.token = '';        # past end of file
  94                 break
  95             elif self.state == ' ':
  96                 if not nextchar:
  97                     self.state = None;  # end of file
  98                     break
  99                 elif nextchar in self.whitespace:
 100                     if self.debug >= 2:
 101                         print "shlex: I see whitespace in whitespace state"
 102                     if self.token:
 103                         break   # emit current token
 104                     else:
 105                         continue
 106                 elif nextchar in self.commenters:
 107                     self.instream.readline()
 108                     self.lineno = self.lineno + 1
 109                 elif nextchar in self.wordchars:
 110                     self.token = nextchar
 111                     self.state = 'a'
 112                 elif nextchar in self.quotes:
 113                     self.token = nextchar
 114                     self.state = nextchar
 115                 else:
 116                     self.token = nextchar
 117                     if self.token:
 118                         break   # emit current token
 119                     else:
 120                         continue
 121             elif self.state in self.quotes:
 122                 self.token = self.token + nextchar
 123                 if nextchar == self.state:
 124                     self.state = ' '
 125                     break
 126             elif self.state == 'a':
 127                 if not nextchar:
 128                     self.state = None;  # end of file
 129                     break
 130                 elif nextchar in self.whitespace:
 131                     if self.debug >= 2:
 132                         print "shlex: I see whitespace in word state"
 133                     self.state = ' '
 134                     if self.token:
 135                         break   # emit current token
 136                     else:
 137                         continue
 138                 elif nextchar in self.commenters:
 139                     self.instream.readline()
 140                     self.lineno = self.lineno + 1
 141                 elif nextchar in self.wordchars or nextchar in self.quotes:
 142                     self.token = self.token + nextchar
 143                 else:
 144                     self.pushback = [nextchar] + self.pushback
 145                     if self.debug >= 2:
 146                         print "shlex: I see punctuation in word state"
 147                     self.state = ' '
 148                     if self.token:
 149                         break   # emit current token
 150                     else:
 151                         continue
 152         result = self.token
 153         self.token = ''
 154         if self.debug > 1:
 155             if result:
 156                 print "shlex: raw token=" + `result`
 157             else:
 158                 print "shlex: raw token=EOF"
 159         return result
 160
 161     def sourcehook(self, newfile):
 162         "Hook called on a filename to be sourced."
 163         if newfile[0] == '"':
 164             newfile = newfile[1:-1]
 165         # This implements cpp-like semantics for relative-path inclusion.
 166         if type(self.infile) == type("") and not os.path.isabs(newfile):
 167             newfile = os.path.join(os.path.dirname(self.infile), newfile)
 168         return (newfile, open(newfile, "r"))
 169
 170     def error_leader(self, infile=None, lineno=None):
 171         "Emit a C-compiler-like, Emacs-friendly error-message leader."
 172         if not infile:
 173             infile = self.infile
 174         if not lineno:
 175             lineno = self.lineno
 176         return "\"%s\", line %d: " % (infile, lineno)
 177
 178
 179 if __name__ == '__main__':
 180     if len(sys.argv) == 1:
 181         lexer = shlex()
 182     else:
 183         file = sys.argv[1]
 184         lexer = shlex(open(file), file)
 185     while 1:
 186         tt = lexer.get_token()
 187         if tt:
 188             print "Token: " + repr(tt)
 189         else:
 190             break