Files for 2.1b1 distribution.
[python/dscho.git] / Lib / distutils / util.py
blobe596150422526e531d4fc97c122ec249755ecf9f
1 """distutils.util
3 Miscellaneous utility functions -- anything that doesn't fit into
4 one of the other *util.py modules.
5 """
7 # created 1999/03/08, Greg Ward
9 __revision__ = "$Id$"
11 import sys, os, string, re
12 from distutils.errors import DistutilsPlatformError
13 from distutils.dep_util import newer
14 from distutils.spawn import spawn
17 def get_platform ():
18 """Return a string that identifies the current platform. This is used
19 mainly to distinguish platform-specific build directories and
20 platform-specific built distributions. Typically includes the OS name
21 and version and the architecture (as supplied by 'os.uname()'),
22 although the exact information included depends on the OS; eg. for IRIX
23 the architecture isn't particularly important (IRIX only runs on SGI
24 hardware), but for Linux the kernel version isn't particularly
25 important.
27 Examples of returned values:
28 linux-i586
29 linux-alpha (?)
30 solaris-2.6-sun4u
31 irix-5.3
32 irix64-6.2
34 For non-POSIX platforms, currently just returns 'sys.platform'.
35 """
36 if os.name != "posix" or not hasattr(os, 'uname'):
37 # XXX what about the architecture? NT is Intel or Alpha,
38 # Mac OS is M68k or PPC, etc.
39 return sys.platform
41 # Try to distinguish various flavours of Unix
43 (osname, host, release, version, machine) = os.uname()
45 # Convert the OS name to lowercase and remove '/' characters
46 # (to accommodate BSD/OS)
47 osname = string.lower(osname)
48 osname = string.replace(osname, '/', '')
50 if osname[:5] == "linux":
51 # At least on Linux/Intel, 'machine' is the processor --
52 # i386, etc.
53 # XXX what about Alpha, SPARC, etc?
54 return "%s-%s" % (osname, machine)
55 elif osname[:5] == "sunos":
56 if release[0] >= "5": # SunOS 5 == Solaris 2
57 osname = "solaris"
58 release = "%d.%s" % (int(release[0]) - 3, release[2:])
59 # fall through to standard osname-release-machine representation
60 elif osname[:4] == "irix": # could be "irix64"!
61 return "%s-%s" % (osname, release)
62 elif osname[:3] == "aix":
63 return "%s-%s.%s" % (osname, version, release)
64 elif osname[:6] == "cygwin":
65 rel_re = re.compile (r'[\d.]+')
66 m = rel_re.match(release)
67 if m:
68 release = m.group()
70 return "%s-%s-%s" % (osname, release, machine)
72 # get_platform ()
75 def convert_path (pathname):
76 """Return 'pathname' as a name that will work on the native filesystem,
77 i.e. split it on '/' and put it back together again using the current
78 directory separator. Needed because filenames in the setup script are
79 always supplied in Unix style, and have to be converted to the local
80 convention before we can actually use them in the filesystem. Raises
81 ValueError on non-Unix-ish systems if 'pathname' either starts or
82 ends with a slash.
83 """
84 if os.sep == '/':
85 return pathname
86 if pathname[0] == '/':
87 raise ValueError, "path '%s' cannot be absolute" % pathname
88 if pathname[-1] == '/':
89 raise ValueError, "path '%s' cannot end with '/'" % pathname
91 paths = string.split(pathname, '/')
92 while '.' in paths:
93 paths.remove('.')
94 if not paths:
95 return os.curdir
96 return apply(os.path.join, paths)
98 # convert_path ()
101 def change_root (new_root, pathname):
102 """Return 'pathname' with 'new_root' prepended. If 'pathname' is
103 relative, this is equivalent to "os.path.join(new_root,pathname)".
104 Otherwise, it requires making 'pathname' relative and then joining the
105 two, which is tricky on DOS/Windows and Mac OS.
107 if os.name == 'posix':
108 if not os.path.isabs(pathname):
109 return os.path.join(new_root, pathname)
110 else:
111 return os.path.join(new_root, pathname[1:])
113 elif os.name == 'nt':
114 (drive, path) = os.path.splitdrive(pathname)
115 if path[0] == '\\':
116 path = path[1:]
117 return os.path.join(new_root, path)
119 elif os.name == 'mac':
120 if not os.path.isabs(pathname):
121 return os.path.join(new_root, pathname)
122 else:
123 # Chop off volume name from start of path
124 elements = string.split(pathname, ":", 1)
125 pathname = ":" + elements[1]
126 return os.path.join(new_root, pathname)
128 else:
129 raise DistutilsPlatformError, \
130 "nothing known about platform '%s'" % os.name
133 _environ_checked = 0
134 def check_environ ():
135 """Ensure that 'os.environ' has all the environment variables we
136 guarantee that users can use in config files, command-line options,
137 etc. Currently this includes:
138 HOME - user's home directory (Unix only)
139 PLAT - description of the current platform, including hardware
140 and OS (see 'get_platform()')
142 global _environ_checked
143 if _environ_checked:
144 return
146 if os.name == 'posix' and not os.environ.has_key('HOME'):
147 import pwd
148 os.environ['HOME'] = pwd.getpwuid(os.getuid())[5]
150 if not os.environ.has_key('PLAT'):
151 os.environ['PLAT'] = get_platform()
153 _environ_checked = 1
156 def subst_vars (s, local_vars):
157 """Perform shell/Perl-style variable substitution on 'string'. Every
158 occurrence of '$' followed by a name is considered a variable, and
159 variable is substituted by the value found in the 'local_vars'
160 dictionary, or in 'os.environ' if it's not in 'local_vars'.
161 'os.environ' is first checked/augmented to guarantee that it contains
162 certain values: see 'check_environ()'. Raise ValueError for any
163 variables not found in either 'local_vars' or 'os.environ'.
165 check_environ()
166 def _subst (match, local_vars=local_vars):
167 var_name = match.group(1)
168 if local_vars.has_key(var_name):
169 return str(local_vars[var_name])
170 else:
171 return os.environ[var_name]
173 try:
174 return re.sub(r'\$([a-zA-Z_][a-zA-Z_0-9]*)', _subst, s)
175 except KeyError, var:
176 raise ValueError, "invalid variable '$%s'" % var
178 # subst_vars ()
181 def grok_environment_error (exc, prefix="error: "):
182 """Generate a useful error message from an EnvironmentError (IOError or
183 OSError) exception object. Handles Python 1.5.1 and 1.5.2 styles, and
184 does what it can to deal with exception objects that don't have a
185 filename (which happens when the error is due to a two-file operation,
186 such as 'rename()' or 'link()'. Returns the error message as a string
187 prefixed with 'prefix'.
189 # check for Python 1.5.2-style {IO,OS}Error exception objects
190 if hasattr(exc, 'filename') and hasattr(exc, 'strerror'):
191 if exc.filename:
192 error = prefix + "%s: %s" % (exc.filename, exc.strerror)
193 else:
194 # two-argument functions in posix module don't
195 # include the filename in the exception object!
196 error = prefix + "%s" % exc.strerror
197 else:
198 error = prefix + str(exc[-1])
200 return error
203 # Needed by 'split_quoted()'
204 _wordchars_re = re.compile(r'[^\\\'\"%s ]*' % string.whitespace)
205 _squote_re = re.compile(r"'(?:[^'\\]|\\.)*'")
206 _dquote_re = re.compile(r'"(?:[^"\\]|\\.)*"')
208 def split_quoted (s):
209 """Split a string up according to Unix shell-like rules for quotes and
210 backslashes. In short: words are delimited by spaces, as long as those
211 spaces are not escaped by a backslash, or inside a quoted string.
212 Single and double quotes are equivalent, and the quote characters can
213 be backslash-escaped. The backslash is stripped from any two-character
214 escape sequence, leaving only the escaped character. The quote
215 characters are stripped from any quoted string. Returns a list of
216 words.
219 # This is a nice algorithm for splitting up a single string, since it
220 # doesn't require character-by-character examination. It was a little
221 # bit of a brain-bender to get it working right, though...
223 s = string.strip(s)
224 words = []
225 pos = 0
227 while s:
228 m = _wordchars_re.match(s, pos)
229 end = m.end()
230 if end == len(s):
231 words.append(s[:end])
232 break
234 if s[end] in string.whitespace: # unescaped, unquoted whitespace: now
235 words.append(s[:end]) # we definitely have a word delimiter
236 s = string.lstrip(s[end:])
237 pos = 0
239 elif s[end] == '\\': # preserve whatever is being escaped;
240 # will become part of the current word
241 s = s[:end] + s[end+1:]
242 pos = end+1
244 else:
245 if s[end] == "'": # slurp singly-quoted string
246 m = _squote_re.match(s, end)
247 elif s[end] == '"': # slurp doubly-quoted string
248 m = _dquote_re.match(s, end)
249 else:
250 raise RuntimeError, \
251 "this can't happen (bad char '%c')" % s[end]
253 if m is None:
254 raise ValueError, \
255 "bad string (mismatched %s quotes?)" % s[end]
257 (beg, end) = m.span()
258 s = s[:beg] + s[beg+1:end-1] + s[end:]
259 pos = m.end() - 2
261 if pos >= len(s):
262 words.append(s)
263 break
265 return words
267 # split_quoted ()
270 def execute (func, args, msg=None, verbose=0, dry_run=0):
271 """Perform some action that affects the outside world (eg. by writing
272 to the filesystem). Such actions are special because they are disabled
273 by the 'dry_run' flag, and announce themselves if 'verbose' is true.
274 This method takes care of all that bureaucracy for you; all you have to
275 do is supply the function to call and an argument tuple for it (to
276 embody the "external action" being performed), and an optional message
277 to print.
279 # Generate a message if we weren't passed one
280 if msg is None:
281 msg = "%s%s" % (func.__name__, `args`)
282 if msg[-2:] == ',)': # correct for singleton tuple
283 msg = msg[0:-2] + ')'
285 # Print it if verbosity level is high enough
286 if verbose:
287 print msg
289 # And do it, as long as we're not in dry-run mode
290 if not dry_run:
291 apply(func, args)
293 # execute()
296 def strtobool (val):
297 """Convert a string representation of truth to true (1) or false (0).
298 True values are 'y', 'yes', 't', 'true', 'on', and '1'; false values
299 are 'n', 'no', 'f', 'false', 'off', and '0'. Raises ValueError if
300 'val' is anything else.
302 val = string.lower(val)
303 if val in ('y', 'yes', 't', 'true', 'on', '1'):
304 return 1
305 elif val in ('n', 'no', 'f', 'false', 'off', '0'):
306 return 0
307 else:
308 raise ValueError, "invalid truth value %s" % `val`
311 def byte_compile (py_files,
312 optimize=0, force=0,
313 prefix=None, base_dir=None,
314 verbose=1, dry_run=0,
315 direct=None):
316 """Byte-compile a collection of Python source files to either .pyc
317 or .pyo files in the same directory. 'py_files' is a list of files
318 to compile; any files that don't end in ".py" are silently skipped.
319 'optimize' must be one of the following:
320 0 - don't optimize (generate .pyc)
321 1 - normal optimization (like "python -O")
322 2 - extra optimization (like "python -OO")
323 If 'force' is true, all files are recompiled regardless of
324 timestamps.
326 The source filename encoded in each bytecode file defaults to the
327 filenames listed in 'py_files'; you can modify these with 'prefix' and
328 'basedir'. 'prefix' is a string that will be stripped off of each
329 source filename, and 'base_dir' is a directory name that will be
330 prepended (after 'prefix' is stripped). You can supply either or both
331 (or neither) of 'prefix' and 'base_dir', as you wish.
333 If 'verbose' is true, prints out a report of each file. If 'dry_run'
334 is true, doesn't actually do anything that would affect the filesystem.
336 Byte-compilation is either done directly in this interpreter process
337 with the standard py_compile module, or indirectly by writing a
338 temporary script and executing it. Normally, you should let
339 'byte_compile()' figure out to use direct compilation or not (see
340 the source for details). The 'direct' flag is used by the script
341 generated in indirect mode; unless you know what you're doing, leave
342 it set to None.
345 # First, if the caller didn't force us into direct or indirect mode,
346 # figure out which mode we should be in. We take a conservative
347 # approach: choose direct mode *only* if the current interpreter is
348 # in debug mode and optimize is 0. If we're not in debug mode (-O
349 # or -OO), we don't know which level of optimization this
350 # interpreter is running with, so we can't do direct
351 # byte-compilation and be certain that it's the right thing. Thus,
352 # always compile indirectly if the current interpreter is in either
353 # optimize mode, or if either optimization level was requested by
354 # the caller.
355 if direct is None:
356 direct = (__debug__ and optimize == 0)
358 # "Indirect" byte-compilation: write a temporary script and then
359 # run it with the appropriate flags.
360 if not direct:
361 from tempfile import mktemp
362 script_name = mktemp(".py")
363 if verbose:
364 print "writing byte-compilation script '%s'" % script_name
365 if not dry_run:
366 script = open(script_name, "w")
368 script.write("""\
369 from distutils.util import byte_compile
370 files = [
371 """)
373 # XXX would be nice to write absolute filenames, just for
374 # safety's sake (script should be more robust in the face of
375 # chdir'ing before running it). But this requires abspath'ing
376 # 'prefix' as well, and that breaks the hack in build_lib's
377 # 'byte_compile()' method that carefully tacks on a trailing
378 # slash (os.sep really) to make sure the prefix here is "just
379 # right". This whole prefix business is rather delicate -- the
380 # problem is that it's really a directory, but I'm treating it
381 # as a dumb string, so trailing slashes and so forth matter.
383 #py_files = map(os.path.abspath, py_files)
384 #if prefix:
385 # prefix = os.path.abspath(prefix)
387 script.write(string.join(map(repr, py_files), ",\n") + "]\n")
388 script.write("""
389 byte_compile(files, optimize=%s, force=%s,
390 prefix=%s, base_dir=%s,
391 verbose=%s, dry_run=0,
392 direct=1)
393 """ % (`optimize`, `force`, `prefix`, `base_dir`, `verbose`))
395 script.close()
397 cmd = [sys.executable, script_name]
398 if optimize == 1:
399 cmd.insert(1, "-O")
400 elif optimize == 2:
401 cmd.insert(1, "-OO")
402 spawn(cmd, verbose=verbose, dry_run=dry_run)
403 execute(os.remove, (script_name,), "removing %s" % script_name,
404 verbose=verbose, dry_run=dry_run)
406 # "Direct" byte-compilation: use the py_compile module to compile
407 # right here, right now. Note that the script generated in indirect
408 # mode simply calls 'byte_compile()' in direct mode, a weird sort of
409 # cross-process recursion. Hey, it works!
410 else:
411 from py_compile import compile
413 for file in py_files:
414 if file[-3:] != ".py":
415 # This lets us be lazy and not filter filenames in
416 # the "install_lib" command.
417 continue
419 # Terminology from the py_compile module:
420 # cfile - byte-compiled file
421 # dfile - purported source filename (same as 'file' by default)
422 cfile = file + (__debug__ and "c" or "o")
423 dfile = file
424 if prefix:
425 if file[:len(prefix)] != prefix:
426 raise ValueError, \
427 ("invalid prefix: filename %s doesn't start with %s"
428 % (`file`, `prefix`))
429 dfile = dfile[len(prefix):]
430 if base_dir:
431 dfile = os.path.join(base_dir, dfile)
433 cfile_base = os.path.basename(cfile)
434 if direct:
435 if force or newer(file, cfile):
436 if verbose:
437 print "byte-compiling %s to %s" % (file, cfile_base)
438 if not dry_run:
439 compile(file, cfile, dfile)
440 else:
441 if verbose:
442 print "skipping byte-compilation of %s to %s" % \
443 (file, cfile_base)
445 # byte_compile ()