Add FilePath support functions in encodingutil.py.

Signed-off-by: Daira Hopwood <daira@jacaranda.org>

diff --git a/src/allmydata/test/test_encodingutil.py b/src/allmydata/test/test_encodingutil.py
index 9d6fe8f33f2e7d9bd8024366dc05d60ef14f9a24..0632bea031a4a32afd325f678aefb7992a272980 100644 (file)
--- a/src/allmydata/test/test_encodingutil.py
+++ b/src/allmydata/test/test_encodingutil.py
@@ -61,12 +61,15 @@ import os, sys, locale
 
 from twisted.trial import unittest
 
+from twisted.python.filepath import FilePath
+
 from allmydata.test.common_util import ReallyEqualMixin
 from allmydata.util import encodingutil, fileutil
 from allmydata.util.encodingutil import argv_to_unicode, unicode_to_url, \
     unicode_to_output, quote_output, quote_path, quote_local_unicode_path, \
-    unicode_platform, listdir_unicode, FilenameEncodingError, get_io_encoding, \
-    get_filesystem_encoding, to_str, from_utf8_or_none, _reload
+    quote_filepath, unicode_platform, listdir_unicode, FilenameEncodingError, \
+    get_io_encoding, get_filesystem_encoding, to_str, from_utf8_or_none, _reload, \
+    to_filepath, extend_filepath, unicode_from_filepath, unicode_segments_from
 from allmydata.dirnode import normalize
 
 from twisted.python import usage
@@ -410,6 +413,9 @@ class QuoteOutput(ReallyEqualMixin, unittest.TestCase):
         self.test_quote_output_utf8(None)
 
 
+def win32_other(win32, other):
+    return win32 if sys.platform == "win32" else other
+
 class QuotePaths(ReallyEqualMixin, unittest.TestCase):
     def test_quote_path(self):
         self.failUnlessReallyEqual(quote_path([u'foo', u'bar']), "'foo/bar'")
@@ -419,9 +425,6 @@ class QuotePaths(ReallyEqualMixin, unittest.TestCase):
         self.failUnlessReallyEqual(quote_path([u'foo', u'\nbar'], quotemarks=True), '"foo/\\x0abar"')
         self.failUnlessReallyEqual(quote_path([u'foo', u'\nbar'], quotemarks=False), '"foo/\\x0abar"')
 
-        def win32_other(win32, other):
-            return win32 if sys.platform == "win32" else other
-
         self.failUnlessReallyEqual(quote_local_unicode_path(u"\\\\?\\C:\\foo"),
                                    win32_other("'C:\\foo'", "'\\\\?\\C:\\foo'"))
         self.failUnlessReallyEqual(quote_local_unicode_path(u"\\\\?\\C:\\foo", quotemarks=True),
@@ -435,6 +438,73 @@ class QuotePaths(ReallyEqualMixin, unittest.TestCase):
         self.failUnlessReallyEqual(quote_local_unicode_path(u"\\\\?\\UNC\\foo\\bar", quotemarks=False),
                                    win32_other("\\\\foo\\bar", "\\\\?\\UNC\\foo\\bar"))
 
+    def test_quote_filepath(self):
+        foo_bar_fp = FilePath(win32_other(u'C:\\foo\\bar', u'/foo/bar'))
+        self.failUnlessReallyEqual(quote_filepath(foo_bar_fp),
+                                   win32_other("'C:\\foo\\bar'", "'/foo/bar'"))
+        self.failUnlessReallyEqual(quote_filepath(foo_bar_fp, quotemarks=True),
+                                   win32_other("'C:\\foo\\bar'", "'/foo/bar'"))
+        self.failUnlessReallyEqual(quote_filepath(foo_bar_fp, quotemarks=False),
+                                   win32_other("C:\\foo\\bar", "/foo/bar"))
+
+        if sys.platform == "win32":
+            foo_longfp = FilePath(u'\\\\?\\C:\\foo')
+            self.failUnlessReallyEqual(quote_filepath(foo_longfp),
+                                       "'C:\\foo'")
+            self.failUnlessReallyEqual(quote_filepath(foo_longfp, quotemarks=True),
+                                       "'C:\\foo'")
+            self.failUnlessReallyEqual(quote_filepath(foo_longfp, quotemarks=False),
+                                       "C:\\foo")
+
+
+class FilePaths(ReallyEqualMixin, unittest.TestCase):
+    def test_to_filepath(self):
+        foo_u = win32_other(u'C:\\foo', u'/foo')
+
+        nosep_fp = to_filepath(foo_u)
+        sep_fp = to_filepath(foo_u + os.path.sep)
+
+        for fp in (nosep_fp, sep_fp):
+            self.failUnlessReallyEqual(fp, FilePath(foo_u))
+            if encodingutil.use_unicode_filepath:
+                self.failUnlessReallyEqual(fp.path, foo_u)
+
+        if sys.platform == "win32":
+            long_u = u'\\\\?\\C:\\foo'
+            longfp = to_filepath(long_u + u'\\')
+            self.failUnlessReallyEqual(longfp, FilePath(long_u))
+            self.failUnlessReallyEqual(longfp.path, long_u)
+
+    def test_extend_filepath(self):
+        foo_bfp = FilePath(win32_other(b'C:\\foo', b'/foo'))
+        foo_ufp = FilePath(win32_other(u'C:\\foo', u'/foo'))
+        foo_bar_baz_u = win32_other(u'C:\\foo\\bar\\baz', u'/foo/bar/baz')
+
+        for foo_fp in (foo_bfp, foo_ufp):
+            fp = extend_filepath(foo_fp, [u'bar', u'baz'])
+            self.failUnlessReallyEqual(fp, FilePath(foo_bar_baz_u))
+            if encodingutil.use_unicode_filepath:
+                self.failUnlessReallyEqual(fp.path, foo_bar_baz_u)
+
+    def test_unicode_from_filepath(self):
+        foo_bfp = FilePath(win32_other(b'C:\\foo', b'/foo'))
+        foo_ufp = FilePath(win32_other(u'C:\\foo', u'/foo'))
+        foo_u = win32_other(u'C:\\foo', u'/foo')
+
+        for foo_fp in (foo_bfp, foo_ufp):
+            self.failUnlessReallyEqual(unicode_from_filepath(foo_fp), foo_u)
+
+    def test_unicode_segments_from(self):
+        foo_bfp = FilePath(win32_other(b'C:\\foo', b'/foo'))
+        foo_ufp = FilePath(win32_other(u'C:\\foo', u'/foo'))
+        foo_bar_baz_bfp = FilePath(win32_other(b'C:\\foo\\bar\\baz', b'/foo/bar/baz'))
+        foo_bar_baz_ufp = FilePath(win32_other(u'C:\\foo\\bar\\baz', u'/foo/bar/baz'))
+
+        for foo_fp in (foo_bfp, foo_ufp):
+            for foo_bar_baz_fp in (foo_bar_baz_bfp, foo_bar_baz_ufp):
+                self.failUnlessReallyEqual(unicode_segments_from(foo_bar_baz_fp, foo_fp),
+                                           [u'bar', u'baz'])
+
 
 class UbuntuKarmicUTF8(EncodingUtil, unittest.TestCase):
     uname = 'Linux korn 2.6.31-14-generic #48-Ubuntu SMP Fri Oct 16 14:05:01 UTC 2009 x86_64'

diff --git a/src/allmydata/util/encodingutil.py b/src/allmydata/util/encodingutil.py
index 61ce23ecbe60bb61ea3b3095b453eb190a981d8f..65f5911a147515d786909aa03c05820caaba885f 100644 (file)
--- a/src/allmydata/util/encodingutil.py
+++ b/src/allmydata/util/encodingutil.py
@@ -6,8 +6,9 @@ unicode and back.
 import sys, os, re, locale
 from types import NoneType
 
-from allmydata.util.assertutil import precondition
+from allmydata.util.assertutil import precondition, _assert
 from twisted.python import usage
+from twisted.python.filepath import FilePath
 from allmydata.util import log
 from allmydata.util.fileutil import abspath_expanduser_unicode
 
@@ -35,9 +36,10 @@ def check_encoding(encoding):
 filesystem_encoding = None
 io_encoding = None
 is_unicode_platform = False
+use_unicode_filepath = False
 
 def _reload():
-    global filesystem_encoding, io_encoding, is_unicode_platform
+    global filesystem_encoding, io_encoding, is_unicode_platform, use_unicode_filepath
 
     filesystem_encoding = canonical_encoding(sys.getfilesystemencoding())
     check_encoding(filesystem_encoding)
@@ -61,6 +63,12 @@ def _reload():
 
     is_unicode_platform = sys.platform in ["win32", "darwin"]
 
+    # Despite the Unicode-mode FilePath support added to Twisted in
+    # <https://twistedmatrix.com/trac/ticket/7805>, we can't yet use
+    # Unicode-mode FilePaths with INotify on non-Windows platforms
+    # due to <https://twistedmatrix.com/trac/ticket/7928>.
+    use_unicode_filepath = sys.platform == "win32"
+
 _reload()
 
 
@@ -249,6 +257,54 @@ def quote_local_unicode_path(path, quotemarks=True):
 
     return quote_output(path, quotemarks=quotemarks, quote_newlines=True)
 
+def quote_filepath(path, quotemarks=True):
+    return quote_local_unicode_path(unicode_from_filepath(path), quotemarks=quotemarks)
+
+def extend_filepath(fp, segments):
+    # We cannot use FilePath.preauthChild, because
+    # * it has the security flaw described in <https://twistedmatrix.com/trac/ticket/6527>;
+    # * it may return a FilePath in the wrong mode.
+
+    for segment in segments:
+        fp = fp.child(segment)
+
+    if isinstance(fp.path, unicode) and not use_unicode_filepath:
+        return FilePath(fp.path.encode(filesystem_encoding))
+    else:
+        return fp
+
+def to_filepath(path):
+    precondition(isinstance(path, unicode if use_unicode_filepath else basestring),
+                 path=path)
+
+    if isinstance(path, unicode) and not use_unicode_filepath:
+        path = path.encode(filesystem_encoding)
+
+    if sys.platform == "win32":
+        _assert(isinstance(path, unicode), path=path)
+        if path.startswith(u"\\\\?\\") and len(path) > 4:
+            # FilePath normally strips trailing path separators, but not in this case.
+            path = path.rstrip(u"\\")
+
+    return FilePath(path)
+
+def _decode(s):
+    precondition(isinstance(s, basestring), s=s)
+
+    if isinstance(s, bytes):
+        return s.decode(filesystem_encoding)
+    else:
+        return s
+
+def unicode_from_filepath(fp):
+    precondition(isinstance(fp, FilePath), fp=fp)
+    return _decode(fp.path)
+
+def unicode_segments_from(base_fp, ancestor_fp):
+    precondition(isinstance(base_fp, FilePath), base_fp=base_fp)
+    precondition(isinstance(ancestor_fp, FilePath), ancestor_fp=ancestor_fp)
+
+    return base_fp.asTextMode().segmentsFrom(ancestor_fp.asTextMode())
 
 def unicode_platform():
     """
@@ -296,3 +352,6 @@ def listdir_unicode(path):
         return os.listdir(path)
     else:
         return listdir_unicode_fallback(path)
+
+def listdir_filepath(fp):
+    return listdir_unicode(unicode_from_filepath(fp))