| Mark Dickinson | 45a26b9 | 2010-07-04 20:07:09 +0000 | [diff] [blame] | 1 | # -*- coding: iso-8859-1 -*- | 
| Guido van Rossum | e7b146f | 2000-02-04 15:28:42 +0000 | [diff] [blame] | 2 | """A lexical analyzer class for simple shell-like syntaxes.""" | 
 | 3 |  | 
| Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 4 | # Module and documentation by Eric S. Raymond, 21 Dec 1998 | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 5 | # Input stacking and error message cleanup added by ESR, March 2000 | 
| Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 6 | # push_source() and pop_source() made explicit by ESR, January 2001. | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 7 | # Posix compliance, split(), string arguments, and | 
 | 8 | # iterator interface by Gustavo Niemeyer, April 2003. | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 9 |  | 
| Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 10 | import os.path | 
| Guido van Rossum | 73898c7 | 1999-05-03 18:14:16 +0000 | [diff] [blame] | 11 | import sys | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 12 | from collections import deque | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 13 |  | 
| Guido van Rossum | 68937b4 | 2007-05-18 00:51:22 +0000 | [diff] [blame] | 14 | from io import StringIO | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 15 |  | 
 | 16 | __all__ = ["shlex", "split"] | 
| Skip Montanaro | 0de6580 | 2001-02-15 22:15:14 +0000 | [diff] [blame] | 17 |  | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 18 | class shlex: | 
| Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 19 |     "A lexical analyzer class for simple shell-like syntaxes." | 
| Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 20 |     def __init__(self, instream=None, infile=None, posix=False): | 
| Guido van Rossum | 3172c5d | 2007-10-16 18:12:55 +0000 | [diff] [blame] | 21 |         if isinstance(instream, str): | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 22 |             instream = StringIO(instream) | 
| Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 23 |         if instream is not None: | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 24 |             self.instream = instream | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 25 |             self.infile = infile | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 26 |         else: | 
 | 27 |             self.instream = sys.stdin | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 28 |             self.infile = None | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 29 |         self.posix = posix | 
 | 30 |         if posix: | 
 | 31 |             self.eof = None | 
 | 32 |         else: | 
 | 33 |             self.eof = '' | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 34 |         self.commenters = '#' | 
| Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 35 |         self.wordchars = ('abcdfeghijklmnopqrstuvwxyz' | 
 | 36 |                           'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_') | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 37 |         if self.posix: | 
| Mark Dickinson | 45a26b9 | 2010-07-04 20:07:09 +0000 | [diff] [blame] | 38 |             self.wordchars += ('ßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ' | 
 | 39 |                                'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ') | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 40 |         self.whitespace = ' \t\r\n' | 
| Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 41 |         self.whitespace_split = False | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 42 |         self.quotes = '\'"' | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 43 |         self.escape = '\\' | 
 | 44 |         self.escapedquotes = '"' | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 45 |         self.state = ' ' | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 46 |         self.pushback = deque() | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 47 |         self.lineno = 1 | 
 | 48 |         self.debug = 0 | 
 | 49 |         self.token = '' | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 50 |         self.filestack = deque() | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 51 |         self.source = None | 
 | 52 |         if self.debug: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 53 |             print('shlex: reading from %s, line %d' \ | 
 | 54 |                   % (self.instream, self.lineno)) | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 55 |  | 
 | 56 |     def push_token(self, tok): | 
 | 57 |         "Push a token onto the stack popped by the get_token method" | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 58 |         if self.debug >= 1: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 59 |             print("shlex: pushing token " + repr(tok)) | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 60 |         self.pushback.appendleft(tok) | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 61 |  | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 62 |     def push_source(self, newstream, newfile=None): | 
 | 63 |         "Push an input source onto the lexer's input source stack." | 
| Guido van Rossum | 3172c5d | 2007-10-16 18:12:55 +0000 | [diff] [blame] | 64 |         if isinstance(newstream, str): | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 65 |             newstream = StringIO(newstream) | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 66 |         self.filestack.appendleft((self.infile, self.instream, self.lineno)) | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 67 |         self.infile = newfile | 
 | 68 |         self.instream = newstream | 
 | 69 |         self.lineno = 1 | 
 | 70 |         if self.debug: | 
| Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 71 |             if newfile is not None: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 72 |                 print('shlex: pushing to file %s' % (self.infile,)) | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 73 |             else: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 74 |                 print('shlex: pushing to stream %s' % (self.instream,)) | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 75 |  | 
 | 76 |     def pop_source(self): | 
 | 77 |         "Pop the input source stack." | 
 | 78 |         self.instream.close() | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 79 |         (self.infile, self.instream, self.lineno) = self.filestack.popleft() | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 80 |         if self.debug: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 81 |             print('shlex: popping to %s, line %d' \ | 
 | 82 |                   % (self.instream, self.lineno)) | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 83 |         self.state = ' ' | 
 | 84 |  | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 85 |     def get_token(self): | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 86 |         "Get a token from the input stream (or from stack if it's nonempty)" | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 87 |         if self.pushback: | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 88 |             tok = self.pushback.popleft() | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 89 |             if self.debug >= 1: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 90 |                 print("shlex: popping token " + repr(tok)) | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 91 |             return tok | 
| Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 92 |         # No pushback.  Get a token. | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 93 |         raw = self.read_token() | 
 | 94 |         # Handle inclusions | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 95 |         if self.source is not None: | 
 | 96 |             while raw == self.source: | 
 | 97 |                 spec = self.sourcehook(self.read_token()) | 
 | 98 |                 if spec: | 
 | 99 |                     (newfile, newstream) = spec | 
 | 100 |                     self.push_source(newstream, newfile) | 
 | 101 |                 raw = self.get_token() | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 102 |         # Maybe we got EOF instead? | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 103 |         while raw == self.eof: | 
| Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 104 |             if not self.filestack: | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 105 |                 return self.eof | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 106 |             else: | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 107 |                 self.pop_source() | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 108 |                 raw = self.get_token() | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 109 |         # Neither inclusion nor EOF | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 110 |         if self.debug >= 1: | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 111 |             if raw != self.eof: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 112 |                 print("shlex: token=" + repr(raw)) | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 113 |             else: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 114 |                 print("shlex: token=EOF") | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 115 |         return raw | 
 | 116 |  | 
 | 117 |     def read_token(self): | 
| Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 118 |         quoted = False | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 119 |         escapedstate = ' ' | 
| Neal Norwitz | 10cf218 | 2003-04-17 23:09:08 +0000 | [diff] [blame] | 120 |         while True: | 
| Andrew M. Kuchling | 49d27c8 | 2000-12-23 14:20:24 +0000 | [diff] [blame] | 121 |             nextchar = self.instream.read(1) | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 122 |             if nextchar == '\n': | 
 | 123 |                 self.lineno = self.lineno + 1 | 
 | 124 |             if self.debug >= 3: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 125 |                 print("shlex: in state", repr(self.state), \ | 
 | 126 |                       "I see character:", repr(nextchar)) | 
| Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 127 |             if self.state is None: | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 128 |                 self.token = ''        # past end of file | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 129 |                 break | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 130 |             elif self.state == ' ': | 
 | 131 |                 if not nextchar: | 
| Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 132 |                     self.state = None  # end of file | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 133 |                     break | 
 | 134 |                 elif nextchar in self.whitespace: | 
 | 135 |                     if self.debug >= 2: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 136 |                         print("shlex: I see whitespace in whitespace state") | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 137 |                     if self.token or (self.posix and quoted): | 
| Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 138 |                         break   # emit current token | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 139 |                     else: | 
 | 140 |                         continue | 
 | 141 |                 elif nextchar in self.commenters: | 
 | 142 |                     self.instream.readline() | 
 | 143 |                     self.lineno = self.lineno + 1 | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 144 |                 elif self.posix and nextchar in self.escape: | 
 | 145 |                     escapedstate = 'a' | 
 | 146 |                     self.state = nextchar | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 147 |                 elif nextchar in self.wordchars: | 
 | 148 |                     self.token = nextchar | 
 | 149 |                     self.state = 'a' | 
 | 150 |                 elif nextchar in self.quotes: | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 151 |                     if not self.posix: | 
 | 152 |                         self.token = nextchar | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 153 |                     self.state = nextchar | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 154 |                 elif self.whitespace_split: | 
 | 155 |                     self.token = nextchar | 
 | 156 |                     self.state = 'a' | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 157 |                 else: | 
 | 158 |                     self.token = nextchar | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 159 |                     if self.token or (self.posix and quoted): | 
| Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 160 |                         break   # emit current token | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 161 |                     else: | 
 | 162 |                         continue | 
 | 163 |             elif self.state in self.quotes: | 
| Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 164 |                 quoted = True | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 165 |                 if not nextchar:      # end of file | 
| Andrew M. Kuchling | 9d56cd1 | 2001-01-09 03:01:15 +0000 | [diff] [blame] | 166 |                     if self.debug >= 2: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 167 |                         print("shlex: I see EOF in quotes state") | 
| Andrew M. Kuchling | 9d56cd1 | 2001-01-09 03:01:15 +0000 | [diff] [blame] | 168 |                     # XXX what error should be raised here? | 
| Collin Winter | ce36ad8 | 2007-08-30 01:19:48 +0000 | [diff] [blame] | 169 |                     raise ValueError("No closing quotation") | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 170 |                 if nextchar == self.state: | 
 | 171 |                     if not self.posix: | 
 | 172 |                         self.token = self.token + nextchar | 
 | 173 |                         self.state = ' ' | 
 | 174 |                         break | 
 | 175 |                     else: | 
 | 176 |                         self.state = 'a' | 
 | 177 |                 elif self.posix and nextchar in self.escape and \ | 
 | 178 |                      self.state in self.escapedquotes: | 
 | 179 |                     escapedstate = self.state | 
 | 180 |                     self.state = nextchar | 
 | 181 |                 else: | 
 | 182 |                     self.token = self.token + nextchar | 
 | 183 |             elif self.state in self.escape: | 
 | 184 |                 if not nextchar:      # end of file | 
 | 185 |                     if self.debug >= 2: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 186 |                         print("shlex: I see EOF in escape state") | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 187 |                     # XXX what error should be raised here? | 
| Collin Winter | ce36ad8 | 2007-08-30 01:19:48 +0000 | [diff] [blame] | 188 |                     raise ValueError("No escaped character") | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 189 |                 # In posix shells, only the quote itself or the escape | 
 | 190 |                 # character may be escaped within quotes. | 
 | 191 |                 if escapedstate in self.quotes and \ | 
 | 192 |                    nextchar != self.state and nextchar != escapedstate: | 
 | 193 |                     self.token = self.token + self.state | 
 | 194 |                 self.token = self.token + nextchar | 
 | 195 |                 self.state = escapedstate | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 196 |             elif self.state == 'a': | 
 | 197 |                 if not nextchar: | 
| Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 198 |                     self.state = None   # end of file | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 199 |                     break | 
 | 200 |                 elif nextchar in self.whitespace: | 
 | 201 |                     if self.debug >= 2: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 202 |                         print("shlex: I see whitespace in word state") | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 203 |                     self.state = ' ' | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 204 |                     if self.token or (self.posix and quoted): | 
| Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 205 |                         break   # emit current token | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 206 |                     else: | 
 | 207 |                         continue | 
 | 208 |                 elif nextchar in self.commenters: | 
 | 209 |                     self.instream.readline() | 
 | 210 |                     self.lineno = self.lineno + 1 | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 211 |                     if self.posix: | 
 | 212 |                         self.state = ' ' | 
 | 213 |                         if self.token or (self.posix and quoted): | 
 | 214 |                             break   # emit current token | 
 | 215 |                         else: | 
 | 216 |                             continue | 
 | 217 |                 elif self.posix and nextchar in self.quotes: | 
 | 218 |                     self.state = nextchar | 
 | 219 |                 elif self.posix and nextchar in self.escape: | 
 | 220 |                     escapedstate = 'a' | 
 | 221 |                     self.state = nextchar | 
 | 222 |                 elif nextchar in self.wordchars or nextchar in self.quotes \ | 
 | 223 |                     or self.whitespace_split: | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 224 |                     self.token = self.token + nextchar | 
 | 225 |                 else: | 
| Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 226 |                     self.pushback.appendleft(nextchar) | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 227 |                     if self.debug >= 2: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 228 |                         print("shlex: I see punctuation in word state") | 
| Guido van Rossum | f247d75 | 1999-03-22 15:28:08 +0000 | [diff] [blame] | 229 |                     self.state = ' ' | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 230 |                     if self.token: | 
| Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 231 |                         break   # emit current token | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 232 |                     else: | 
 | 233 |                         continue | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 234 |         result = self.token | 
 | 235 |         self.token = '' | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 236 |         if self.posix and not quoted and result == '': | 
 | 237 |             result = None | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 238 |         if self.debug > 1: | 
 | 239 |             if result: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 240 |                 print("shlex: raw token=" + repr(result)) | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 241 |             else: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 242 |                 print("shlex: raw token=EOF") | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 243 |         return result | 
 | 244 |  | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 245 |     def sourcehook(self, newfile): | 
 | 246 |         "Hook called on a filename to be sourced." | 
 | 247 |         if newfile[0] == '"': | 
 | 248 |             newfile = newfile[1:-1] | 
| Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 249 |         # This implements cpp-like semantics for relative-path inclusion. | 
| Guido van Rossum | 3172c5d | 2007-10-16 18:12:55 +0000 | [diff] [blame] | 250 |         if isinstance(self.infile, str) and not os.path.isabs(newfile): | 
| Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 251 |             newfile = os.path.join(os.path.dirname(self.infile), newfile) | 
| Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 252 |         return (newfile, open(newfile, "r")) | 
 | 253 |  | 
| Guido van Rossum | 4b83ecb | 2000-05-01 20:14:12 +0000 | [diff] [blame] | 254 |     def error_leader(self, infile=None, lineno=None): | 
 | 255 |         "Emit a C-compiler-like, Emacs-friendly error-message leader." | 
| Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 256 |         if infile is None: | 
| Guido van Rossum | 4b83ecb | 2000-05-01 20:14:12 +0000 | [diff] [blame] | 257 |             infile = self.infile | 
| Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 258 |         if lineno is None: | 
| Guido van Rossum | 4b83ecb | 2000-05-01 20:14:12 +0000 | [diff] [blame] | 259 |             lineno = self.lineno | 
 | 260 |         return "\"%s\", line %d: " % (infile, lineno) | 
 | 261 |  | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 262 |     def __iter__(self): | 
 | 263 |         return self | 
 | 264 |  | 
| Georg Brandl | a18af4e | 2007-04-21 15:47:16 +0000 | [diff] [blame] | 265 |     def __next__(self): | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 266 |         token = self.get_token() | 
 | 267 |         if token == self.eof: | 
 | 268 |             raise StopIteration | 
 | 269 |         return token | 
 | 270 |  | 
| Guido van Rossum | e7ba495 | 2007-06-06 23:52:48 +0000 | [diff] [blame] | 271 | def split(s, comments=False, posix=True): | 
 | 272 |     lex = shlex(s, posix=posix) | 
| Gustavo Niemeyer | 48f3dcc | 2003-04-20 01:57:03 +0000 | [diff] [blame] | 273 |     lex.whitespace_split = True | 
 | 274 |     if not comments: | 
 | 275 |         lex.commenters = '' | 
| Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 276 |     return list(lex) | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 277 |  | 
| Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 278 | if __name__ == '__main__': | 
| Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 279 |     if len(sys.argv) == 1: | 
 | 280 |         lexer = shlex() | 
 | 281 |     else: | 
 | 282 |         file = sys.argv[1] | 
 | 283 |         lexer = shlex(open(file), file) | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 284 |     while 1: | 
 | 285 |         tt = lexer.get_token() | 
| Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 286 |         if tt: | 
| Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 287 |             print("Token: " + repr(tt)) | 
| Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 288 |         else: | 
| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 289 |             break |