2 Functions used to convert inputs from whatever encoding used in the system to
9 from allmydata.util.assertutil import precondition
10 from twisted.python import usage
12 from allmydata.util import log
13 from allmydata.util.fileutil import abspath_expanduser_unicode
16 def canonical_encoding(encoding):
18 log.msg("Warning: falling back to UTF-8 encoding.", level=log.WEIRD)
20 encoding = encoding.lower()
21 if encoding == "cp65001":
23 elif encoding == "us-ascii" or encoding == "646" or encoding == "ansi_x3.4-1968":
28 def check_encoding(encoding):
29 # sometimes Python returns an encoding name that it doesn't support for conversion
30 # fail early if this happens
32 u"test".encode(encoding)
33 except (LookupError, AttributeError):
34 raise AssertionError("The character encoding '%s' is not supported for conversion." % (encoding,))
36 filesystem_encoding = None
38 is_unicode_platform = False
41 global filesystem_encoding, io_encoding, is_unicode_platform
43 filesystem_encoding = canonical_encoding(sys.getfilesystemencoding())
44 check_encoding(filesystem_encoding)
46 if sys.platform == 'win32':
47 # On Windows we install UTF-8 stream wrappers for sys.stdout and
48 # sys.stderr, and reencode the arguments as UTF-8 (see scripts/runner.py).
52 if hasattr(sys.stdout, 'encoding'):
53 ioenc = sys.stdout.encoding
56 ioenc = locale.getpreferredencoding()
58 pass # work around <http://bugs.python.org/issue1443504>
59 io_encoding = canonical_encoding(ioenc)
61 check_encoding(io_encoding)
63 is_unicode_platform = sys.platform in ["win32", "darwin"]
68 def get_filesystem_encoding():
70 Returns expected encoding for local filenames.
72 return filesystem_encoding
74 def get_io_encoding():
76 Returns expected encoding for writing to stdout or stderr, and for arguments in sys.argv.
80 def argv_to_unicode(s):
82 Decode given argv element to unicode. If this fails, raise a UsageError.
84 precondition(isinstance(s, str), s)
87 return unicode(s, io_encoding)
88 except UnicodeDecodeError:
89 raise usage.UsageError("Argument %s cannot be decoded as %s." %
90 (quote_output(s), io_encoding))
92 def argv_to_abspath(s):
94 Convenience function to decode an argv element to an absolute path, with ~ expanded.
95 If this fails, raise a UsageError.
97 return abspath_expanduser_unicode(argv_to_unicode(s))
99 def unicode_to_argv(s, mangle=False):
101 Encode the given Unicode argument as a bytestring.
102 If the argument is to be passed to a different process, then the 'mangle' argument
103 should be true; on Windows, this uses a mangled encoding that will be reversed by
106 precondition(isinstance(s, unicode), s)
108 if mangle and sys.platform == "win32":
109 # This must be the same as 'mangle' in bin/tahoe-script.template.
110 return str(re.sub(ur'[^\x20-\x7F]', lambda m: u'\x7F%x;' % (ord(m.group(0)),), s))
112 return s.encode(io_encoding)
114 def unicode_to_url(s):
116 Encode an unicode object used in an URL.
118 # According to RFC 2718, non-ascii characters in URLs must be UTF-8 encoded.
122 #precondition(isinstance(s, unicode), s)
123 #return s.encode('utf-8')
126 if s is None or isinstance(s, str):
128 return s.encode('utf-8')
130 PRINTABLE_ASCII = re.compile(r'^[\n\r\x20-\x7E]*$', re.DOTALL)
131 PRINTABLE_8BIT = re.compile(r'^[\n\r\x20-\x7E\x80-\xFF]*$', re.DOTALL)
133 def is_printable_ascii(s):
134 return PRINTABLE_ASCII.search(s) is not None
136 def unicode_to_output(s):
138 Encode an unicode object for representation on stdout or stderr.
140 precondition(isinstance(s, unicode), s)
143 out = s.encode(io_encoding)
144 except (UnicodeEncodeError, UnicodeDecodeError):
145 raise UnicodeEncodeError(io_encoding, s, 0, 0,
146 "A string could not be encoded as %s for output to the terminal:\n%r" %
147 (io_encoding, repr(s)))
149 if PRINTABLE_8BIT.search(out) is None:
150 raise UnicodeEncodeError(io_encoding, s, 0, 0,
151 "A string encoded as %s for output to the terminal contained unsafe bytes:\n%r" %
152 (io_encoding, repr(s)))
156 def _unicode_escape(m, quote_newlines):
158 if u == u'"' or u == u'$' or u == u'`' or u == u'\\':
160 elif u == u'\n' and not quote_newlines:
163 codepoint = (ord(u[0])-0xD800)*0x400 + ord(u[1])-0xDC00 + 0x10000
166 if codepoint > 0xFFFF:
167 return u'\\U%08x' % (codepoint,)
168 elif codepoint > 0xFF:
169 return u'\\u%04x' % (codepoint,)
171 return u'\\x%02x' % (codepoint,)
173 def _str_escape(m, quote_newlines):
175 if c == '"' or c == '$' or c == '`' or c == '\\':
177 elif c == '\n' and not quote_newlines:
180 return '\\x%02x' % (ord(c),)
182 MUST_DOUBLE_QUOTE_NL = re.compile(ur'[^\x20-\x26\x28-\x7E\u00A0-\uD7FF\uE000-\uFDCF\uFDF0-\uFFFC]', re.DOTALL)
183 MUST_DOUBLE_QUOTE = re.compile(ur'[^\n\x20-\x26\x28-\x7E\u00A0-\uD7FF\uE000-\uFDCF\uFDF0-\uFFFC]', re.DOTALL)
185 # if we must double-quote, then we have to escape ", $ and `, but need not escape '
186 ESCAPABLE_UNICODE = re.compile(ur'([\uD800-\uDBFF][\uDC00-\uDFFF])|' # valid surrogate pairs
187 ur'[^ !#\x25-\x5B\x5D-\x5F\x61-\x7E\u00A0-\uD7FF\uE000-\uFDCF\uFDF0-\uFFFC]',
190 ESCAPABLE_8BIT = re.compile( r'[^ !#\x25-\x5B\x5D-\x5F\x61-\x7E]', re.DOTALL)
192 def quote_output(s, quotemarks=True, quote_newlines=None, encoding=None):
194 Encode either a Unicode string or a UTF-8-encoded bytestring for representation
195 on stdout or stderr, tolerating errors. If 'quotemarks' is True, the string is
196 always quoted; otherwise, it is quoted only if necessary to avoid ambiguity or
197 control bytes in the output. (Newlines are counted as control bytes iff
198 quote_newlines is True.)
200 Quoting may use either single or double quotes. Within single quotes, all
201 characters stand for themselves, and ' will not appear. Within double quotes,
202 Python-compatible backslash escaping is used.
204 If not explicitly given, quote_newlines is True when quotemarks is True.
206 precondition(isinstance(s, (str, unicode)), s)
207 if quote_newlines is None:
208 quote_newlines = quotemarks
210 if isinstance(s, str):
212 s = s.decode('utf-8')
213 except UnicodeDecodeError:
214 return 'b"%s"' % (ESCAPABLE_8BIT.sub(lambda m: _str_escape(m, quote_newlines), s),)
216 must_double_quote = quote_newlines and MUST_DOUBLE_QUOTE_NL or MUST_DOUBLE_QUOTE
217 if must_double_quote.search(s) is None:
219 out = s.encode(encoding or io_encoding)
220 if quotemarks or out.startswith('"'):
221 return "'%s'" % (out,)
224 except (UnicodeDecodeError, UnicodeEncodeError):
227 escaped = ESCAPABLE_UNICODE.sub(lambda m: _unicode_escape(m, quote_newlines), s)
228 return '"%s"' % (escaped.encode(encoding or io_encoding, 'backslashreplace'),)
230 def quote_path(path, quotemarks=True):
231 return quote_output("/".join(map(to_str, path)), quotemarks=quotemarks, quote_newlines=True)
233 def quote_local_unicode_path(path, quotemarks=True):
234 precondition(isinstance(path, unicode), path)
236 if sys.platform == "win32" and path.startswith(u"\\\\?\\"):
238 if path.startswith(u"UNC\\"):
239 path = u"\\\\" + path[4 :]
241 return quote_output(path, quotemarks=quotemarks, quote_newlines=True)
244 def unicode_platform():
246 Does the current platform handle Unicode filenames natively?
248 return is_unicode_platform
250 class FilenameEncodingError(Exception):
252 Filename cannot be encoded using the current encoding of your filesystem
253 (%s). Please configure your locale correctly or rename this file.
257 def listdir_unicode_fallback(path):
259 This function emulates a fallback Unicode API similar to one available
260 under Windows or MacOS X.
262 If badly encoded filenames are encountered, an exception is raised.
264 precondition(isinstance(path, unicode), path)
267 byte_path = path.encode(filesystem_encoding)
268 except (UnicodeEncodeError, UnicodeDecodeError):
269 raise FilenameEncodingError(path)
272 return [unicode(fn, filesystem_encoding) for fn in os.listdir(byte_path)]
273 except UnicodeDecodeError:
274 raise FilenameEncodingError(fn)
276 def listdir_unicode(path):
278 Wrapper around listdir() which provides safe access to the convenient
279 Unicode API even under platforms that don't provide one natively.
281 precondition(isinstance(path, unicode), path)
283 # On Windows and MacOS X, the Unicode API is used
284 # On other platforms (ie. Unix systems), the byte-level API is used
286 if is_unicode_platform:
287 return os.listdir(path)
289 return listdir_unicode_fallback(path)