Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +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 | |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 14 | try: |
| 15 | from cStringIO import StringIO |
| 16 | except ImportError: |
| 17 | from StringIO import StringIO |
| 18 | |
| 19 | __all__ = ["shlex", "split"] |
Skip Montanaro | 0de6580 | 2001-02-15 22:15:14 +0000 | [diff] [blame] | 20 | |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 21 | class shlex: |
Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 22 | "A lexical analyzer class for simple shell-like syntaxes." |
Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 23 | def __init__(self, instream=None, infile=None, posix=False): |
Neal Norwitz | 10cf218 | 2003-04-17 23:09:08 +0000 | [diff] [blame] | 24 | if isinstance(instream, basestring): |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 25 | instream = StringIO(instream) |
Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 26 | if instream is not None: |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 27 | self.instream = instream |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 28 | self.infile = infile |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 29 | else: |
| 30 | self.instream = sys.stdin |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 31 | self.infile = None |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 32 | self.posix = posix |
| 33 | if posix: |
| 34 | self.eof = None |
| 35 | else: |
| 36 | self.eof = '' |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 37 | self.commenters = '#' |
Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 38 | self.wordchars = ('abcdfeghijklmnopqrstuvwxyz' |
| 39 | 'ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_') |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 40 | if self.posix: |
| 41 | self.wordchars += ('ßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ' |
| 42 | 'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ') |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 43 | self.whitespace = ' \t\r\n' |
Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 44 | self.whitespace_split = False |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 45 | self.quotes = '\'"' |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 46 | self.escape = '\\' |
| 47 | self.escapedquotes = '"' |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 48 | self.state = ' ' |
Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 49 | self.pushback = deque() |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 50 | self.lineno = 1 |
| 51 | self.debug = 0 |
| 52 | self.token = '' |
Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 53 | self.filestack = deque() |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 54 | self.source = None |
| 55 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 56 | print('shlex: reading from %s, line %d' \ |
| 57 | % (self.instream, self.lineno)) |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 58 | |
| 59 | def push_token(self, tok): |
| 60 | "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] | 61 | if self.debug >= 1: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 62 | print("shlex: pushing token " + repr(tok)) |
Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 63 | self.pushback.appendleft(tok) |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 64 | |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 65 | def push_source(self, newstream, newfile=None): |
| 66 | "Push an input source onto the lexer's input source stack." |
Neal Norwitz | 10cf218 | 2003-04-17 23:09:08 +0000 | [diff] [blame] | 67 | if isinstance(newstream, basestring): |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 68 | newstream = StringIO(newstream) |
Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 69 | self.filestack.appendleft((self.infile, self.instream, self.lineno)) |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 70 | self.infile = newfile |
| 71 | self.instream = newstream |
| 72 | self.lineno = 1 |
| 73 | if self.debug: |
Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 74 | if newfile is not None: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 75 | print('shlex: pushing to file %s' % (self.infile,)) |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 76 | else: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 77 | print('shlex: pushing to stream %s' % (self.instream,)) |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 78 | |
| 79 | def pop_source(self): |
| 80 | "Pop the input source stack." |
| 81 | self.instream.close() |
Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 82 | (self.infile, self.instream, self.lineno) = self.filestack.popleft() |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 83 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 84 | print('shlex: popping to %s, line %d' \ |
| 85 | % (self.instream, self.lineno)) |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 86 | self.state = ' ' |
| 87 | |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 88 | def get_token(self): |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 89 | "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] | 90 | if self.pushback: |
Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 91 | tok = self.pushback.popleft() |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 92 | if self.debug >= 1: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 93 | print("shlex: popping token " + repr(tok)) |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 94 | return tok |
Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 95 | # No pushback. Get a token. |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 96 | raw = self.read_token() |
| 97 | # Handle inclusions |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 98 | if self.source is not None: |
| 99 | while raw == self.source: |
| 100 | spec = self.sourcehook(self.read_token()) |
| 101 | if spec: |
| 102 | (newfile, newstream) = spec |
| 103 | self.push_source(newstream, newfile) |
| 104 | raw = self.get_token() |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 105 | # Maybe we got EOF instead? |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 106 | while raw == self.eof: |
Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 107 | if not self.filestack: |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 108 | return self.eof |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 109 | else: |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 110 | self.pop_source() |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 111 | raw = self.get_token() |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 112 | # Neither inclusion nor EOF |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 113 | if self.debug >= 1: |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 114 | if raw != self.eof: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 115 | print("shlex: token=" + repr(raw)) |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 116 | else: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 117 | print("shlex: token=EOF") |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 118 | return raw |
| 119 | |
| 120 | def read_token(self): |
Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 121 | quoted = False |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 122 | escapedstate = ' ' |
Neal Norwitz | 10cf218 | 2003-04-17 23:09:08 +0000 | [diff] [blame] | 123 | while True: |
Andrew M. Kuchling | 49d27c8 | 2000-12-23 14:20:24 +0000 | [diff] [blame] | 124 | nextchar = self.instream.read(1) |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 125 | if nextchar == '\n': |
| 126 | self.lineno = self.lineno + 1 |
| 127 | if self.debug >= 3: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 128 | print("shlex: in state", repr(self.state), \ |
| 129 | "I see character:", repr(nextchar)) |
Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 130 | if self.state is None: |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 131 | self.token = '' # past end of file |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 132 | break |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 133 | elif self.state == ' ': |
| 134 | if not nextchar: |
Eric S. Raymond | bddbaf7 | 2001-01-16 15:19:13 +0000 | [diff] [blame] | 135 | self.state = None # end of file |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 136 | break |
| 137 | elif nextchar in self.whitespace: |
| 138 | if self.debug >= 2: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 139 | print("shlex: I see whitespace in whitespace state") |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 140 | if self.token or (self.posix and quoted): |
Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 141 | break # emit current token |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 142 | else: |
| 143 | continue |
| 144 | elif nextchar in self.commenters: |
| 145 | self.instream.readline() |
| 146 | self.lineno = self.lineno + 1 |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 147 | elif self.posix and nextchar in self.escape: |
| 148 | escapedstate = 'a' |
| 149 | self.state = nextchar |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 150 | elif nextchar in self.wordchars: |
| 151 | self.token = nextchar |
| 152 | self.state = 'a' |
| 153 | elif nextchar in self.quotes: |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 154 | if not self.posix: |
| 155 | self.token = nextchar |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 156 | self.state = nextchar |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 157 | elif self.whitespace_split: |
| 158 | self.token = nextchar |
| 159 | self.state = 'a' |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 160 | else: |
| 161 | self.token = nextchar |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 162 | if self.token or (self.posix and quoted): |
Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 163 | break # emit current token |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 164 | else: |
| 165 | continue |
| 166 | elif self.state in self.quotes: |
Fred Drake | 2431523 | 2003-04-17 22:01:17 +0000 | [diff] [blame] | 167 | quoted = True |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 168 | if not nextchar: # end of file |
Andrew M. Kuchling | 9d56cd1 | 2001-01-09 03:01:15 +0000 | [diff] [blame] | 169 | if self.debug >= 2: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 170 | print("shlex: I see EOF in quotes state") |
Andrew M. Kuchling | 9d56cd1 | 2001-01-09 03:01:15 +0000 | [diff] [blame] | 171 | # XXX what error should be raised here? |
| 172 | raise ValueError, "No closing quotation" |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 173 | if nextchar == self.state: |
| 174 | if not self.posix: |
| 175 | self.token = self.token + nextchar |
| 176 | self.state = ' ' |
| 177 | break |
| 178 | else: |
| 179 | self.state = 'a' |
| 180 | elif self.posix and nextchar in self.escape and \ |
| 181 | self.state in self.escapedquotes: |
| 182 | escapedstate = self.state |
| 183 | self.state = nextchar |
| 184 | else: |
| 185 | self.token = self.token + nextchar |
| 186 | elif self.state in self.escape: |
| 187 | if not nextchar: # end of file |
| 188 | if self.debug >= 2: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 189 | print("shlex: I see EOF in escape state") |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 190 | # XXX what error should be raised here? |
| 191 | raise ValueError, "No escaped character" |
| 192 | # In posix shells, only the quote itself or the escape |
| 193 | # character may be escaped within quotes. |
| 194 | if escapedstate in self.quotes and \ |
| 195 | nextchar != self.state and nextchar != escapedstate: |
| 196 | self.token = self.token + self.state |
| 197 | self.token = self.token + nextchar |
| 198 | self.state = escapedstate |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 199 | elif self.state == 'a': |
| 200 | if not nextchar: |
Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 201 | self.state = None # end of file |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 202 | break |
| 203 | elif nextchar in self.whitespace: |
| 204 | if self.debug >= 2: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 205 | print("shlex: I see whitespace in word state") |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 206 | self.state = ' ' |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 207 | if self.token or (self.posix and quoted): |
Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 208 | break # emit current token |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 209 | else: |
| 210 | continue |
| 211 | elif nextchar in self.commenters: |
| 212 | self.instream.readline() |
| 213 | self.lineno = self.lineno + 1 |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 214 | if self.posix: |
| 215 | self.state = ' ' |
| 216 | if self.token or (self.posix and quoted): |
| 217 | break # emit current token |
| 218 | else: |
| 219 | continue |
| 220 | elif self.posix and nextchar in self.quotes: |
| 221 | self.state = nextchar |
| 222 | elif self.posix and nextchar in self.escape: |
| 223 | escapedstate = 'a' |
| 224 | self.state = nextchar |
| 225 | elif nextchar in self.wordchars or nextchar in self.quotes \ |
| 226 | or self.whitespace_split: |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 227 | self.token = self.token + nextchar |
| 228 | else: |
Raymond Hettinger | 756b3f3 | 2004-01-29 06:37:52 +0000 | [diff] [blame] | 229 | self.pushback.appendleft(nextchar) |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 230 | if self.debug >= 2: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 231 | print("shlex: I see punctuation in word state") |
Guido van Rossum | f247d75 | 1999-03-22 15:28:08 +0000 | [diff] [blame] | 232 | self.state = ' ' |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 233 | if self.token: |
Fred Drake | dbbf76b | 2000-07-09 16:44:26 +0000 | [diff] [blame] | 234 | break # emit current token |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 235 | else: |
| 236 | continue |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 237 | result = self.token |
| 238 | self.token = '' |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 239 | if self.posix and not quoted and result == '': |
| 240 | result = None |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 241 | if self.debug > 1: |
| 242 | if result: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 243 | print("shlex: raw token=" + repr(result)) |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 244 | else: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 245 | print("shlex: raw token=EOF") |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 246 | return result |
| 247 | |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 248 | def sourcehook(self, newfile): |
| 249 | "Hook called on a filename to be sourced." |
| 250 | if newfile[0] == '"': |
| 251 | newfile = newfile[1:-1] |
Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 252 | # This implements cpp-like semantics for relative-path inclusion. |
Neal Norwitz | 10cf218 | 2003-04-17 23:09:08 +0000 | [diff] [blame] | 253 | if isinstance(self.infile, basestring) and not os.path.isabs(newfile): |
Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 254 | newfile = os.path.join(os.path.dirname(self.infile), newfile) |
Guido van Rossum | eb4e11a | 2000-05-01 20:08:46 +0000 | [diff] [blame] | 255 | return (newfile, open(newfile, "r")) |
| 256 | |
Guido van Rossum | 4b83ecb | 2000-05-01 20:14:12 +0000 | [diff] [blame] | 257 | def error_leader(self, infile=None, lineno=None): |
| 258 | "Emit a C-compiler-like, Emacs-friendly error-message leader." |
Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 259 | if infile is None: |
Guido van Rossum | 4b83ecb | 2000-05-01 20:14:12 +0000 | [diff] [blame] | 260 | infile = self.infile |
Raymond Hettinger | f13eb55 | 2002-06-02 00:40:05 +0000 | [diff] [blame] | 261 | if lineno is None: |
Guido van Rossum | 4b83ecb | 2000-05-01 20:14:12 +0000 | [diff] [blame] | 262 | lineno = self.lineno |
| 263 | return "\"%s\", line %d: " % (infile, lineno) |
| 264 | |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 265 | def __iter__(self): |
| 266 | return self |
| 267 | |
Georg Brandl | a18af4e | 2007-04-21 15:47:16 +0000 | [diff] [blame] | 268 | def __next__(self): |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 269 | token = self.get_token() |
| 270 | if token == self.eof: |
| 271 | raise StopIteration |
| 272 | return token |
| 273 | |
Gustavo Niemeyer | 48f3dcc | 2003-04-20 01:57:03 +0000 | [diff] [blame] | 274 | def split(s, comments=False): |
| 275 | lex = shlex(s, posix=True) |
| 276 | lex.whitespace_split = True |
| 277 | if not comments: |
| 278 | lex.commenters = '' |
Gustavo Niemeyer | 68d8cef | 2003-04-17 21:31:33 +0000 | [diff] [blame] | 279 | return list(lex) |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 280 | |
Tim Peters | 70c4378 | 2001-01-17 08:48:39 +0000 | [diff] [blame] | 281 | if __name__ == '__main__': |
Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 282 | if len(sys.argv) == 1: |
| 283 | lexer = shlex() |
| 284 | else: |
| 285 | file = sys.argv[1] |
| 286 | lexer = shlex(open(file), file) |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 287 | while 1: |
| 288 | tt = lexer.get_token() |
Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 289 | if tt: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 290 | print("Token: " + repr(tt)) |
Fred Drake | 52dc76c | 2000-07-03 09:56:23 +0000 | [diff] [blame] | 291 | else: |
Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 292 | break |