#9411: allow selecting an encoding for configparser files. Also adds a new test config file to test special cases.
diff --git a/Lib/configparser.py b/Lib/configparser.py
index 8c0546a..7ad24d8 100644
--- a/Lib/configparser.py
+++ b/Lib/configparser.py
@@ -61,7 +61,7 @@
options(section)
Return list of configuration options for the named section.
- read(filenames)
+ read(filenames, encoding=None)
Read and parse the list of named configuration files, given by
name. A single filename is also allowed. Non-existing files
are ignored. Return list of successfully read files.
@@ -369,7 +369,7 @@
del opts['__name__']
return list(opts.keys())
- def read(self, filenames):
+ def read(self, filenames, encoding=None):
"""Read and parse a filename or a list of filenames.
Files that cannot be opened are silently ignored; this is
@@ -386,7 +386,7 @@
read_ok = []
for filename in filenames:
try:
- fp = open(filename)
+ fp = open(filename, encoding=encoding)
except IOError:
continue
self._read(fp, filename)
diff --git a/Lib/test/cfgparser.3 b/Lib/test/cfgparser.3
new file mode 100644
index 0000000..c182cd7
--- /dev/null
+++ b/Lib/test/cfgparser.3
@@ -0,0 +1,69 @@
+ # INI with as many tricky parts as possible
+ # Most of them could not be used before 3.2
+
+ # This will be parsed with the following options
+ # delimiters = {'='}
+ # comment_prefixes = {'#'}
+ # allow_no_value = True
+
+[DEFAULT]
+go = %(interpolate)s
+
+[strange]
+ values = that are indented # and end with hash comments
+ other = that do continue
+ in # and still have
+ other # comments mixed
+ lines # with the values
+
+
+
+
+
+[corruption]
+ value = that is
+
+
+ actually still here
+
+
+ and holds all these weird newlines
+
+
+ # but not for the lines that are comments
+ nor the indentation
+
+ another value = # empty string
+ yet another # None!
+
+ [yeah, sections can be indented as well]
+ and that does not mean = anything
+ are they subsections = False
+ if you want subsections = use XML
+ lets use some Unicode = 片仮名
+
+ [another one!]
+ even if values are indented like this = seriously
+yes, this still applies to = section "another one!"
+this too = are there people with configurations broken as this?
+ beware, this is going to be a continuation
+ of the value for
+ key "this too"
+ even if it has a = character
+ this is still the continuation
+ your editor probably highlights it wrong
+ but that's life
+# let's set this value so there is no error
+# when getting all items for this section:
+interpolate = anything will do
+
+[no values here]
+# but there's this `go` in DEFAULT
+
+ [tricky interpolation]
+ interpolate = do this
+ lets = %(go)s
+
+ [more interpolation]
+ interpolate = go shopping
+ lets = %(go)s
diff --git a/Lib/test/test_cfgparser.py b/Lib/test/test_cfgparser.py
index e00a51a..5a77bfd 100644
--- a/Lib/test/test_cfgparser.py
+++ b/Lib/test/test_cfgparser.py
@@ -533,7 +533,7 @@
smbconf = support.findfile("cfgparser.2")
# check when we pass a mix of readable and non-readable files:
cf = self.newconfig()
- parsed_files = cf.read([smbconf, "nonexistent-file"])
+ parsed_files = cf.read([smbconf, "nonexistent-file"], encoding='utf-8')
self.assertEqual(parsed_files, [smbconf])
sections = ['global', 'homes', 'printers',
'print$', 'pdf-generator', 'tmp', 'Agustin']
@@ -600,6 +600,46 @@
class SafeConfigParserTestCaseNoValue(SafeConfigParserTestCase):
allow_no_value = True
+class SafeConfigParserTestCaseTrickyFile(CfgParserTestCaseClass):
+ config_class = configparser.SafeConfigParser
+ delimiters = {'='}
+ comment_prefixes = {'#'}
+ allow_no_value = True
+
+ def test_cfgparser_dot_3(self):
+ tricky = support.findfile("cfgparser.3")
+ cf = self.newconfig()
+ self.assertEqual(len(cf.read(tricky, encoding='utf-8')), 1)
+ self.assertEqual(cf.sections(), ['strange',
+ 'corruption',
+ 'yeah, sections can be '
+ 'indented as well',
+ 'another one!',
+ 'no values here',
+ 'tricky interpolation',
+ 'more interpolation'])
+ #self.assertEqual(cf.getint('DEFAULT', 'go', vars={'interpolate': '-1'}),
+ # -1)
+ self.assertEqual(len(cf.get('strange', 'other').split('\n')), 4)
+ self.assertEqual(len(cf.get('corruption', 'value').split('\n')), 10)
+ longname = 'yeah, sections can be indented as well'
+ self.assertFalse(cf.getboolean(longname, 'are they subsections'))
+ self.assertEquals(cf.get(longname, 'lets use some Unicode'),
+ '片仮名')
+ self.assertEqual(len(cf.items('another one!')), 5) # 4 in section and
+ # `go` from DEFAULT
+ with self.assertRaises(configparser.InterpolationMissingOptionError):
+ cf.items('no values here')
+ self.assertEqual(cf.get('tricky interpolation', 'lets'), 'do this')
+ self.assertEqual(cf.get('tricky interpolation', 'lets'),
+ cf.get('tricky interpolation', 'go'))
+ self.assertEqual(cf.get('more interpolation', 'lets'), 'go shopping')
+
+ def test_unicode_failure(self):
+ tricky = support.findfile("cfgparser.3")
+ cf = self.newconfig()
+ with self.assertRaises(UnicodeDecodeError):
+ cf.read(tricky, encoding='ascii')
class SortedTestCase(RawConfigParserTestCase):
dict_type = SortedDict
@@ -635,10 +675,13 @@
foo: bar # not a comment!
# but this is a comment
; another comment
+ quirk: this;is not a comment
+ ; a space must precede a comment character
""")
cf = self.fromstring(config_string)
self.assertEqual(cf.get('Commented Bar', 'foo'), 'bar # not a comment!')
self.assertEqual(cf.get('Commented Bar', 'baz'), 'qwe')
+ self.assertEqual(cf.get('Commented Bar', 'quirk'), 'this;is not a comment')
def test_main():
@@ -652,6 +695,7 @@
SafeConfigParserTestCase,
SafeConfigParserTestCaseNonStandardDelimiters,
SafeConfigParserTestCaseNoValue,
+ SafeConfigParserTestCaseTrickyFile,
SortedTestCase,
CompatibleTestCase,
)