Unicode assumption fails

[jaraco]

Original report by Klaus Alexander Seistrup (Bitbucket: kseistrup, GitHub: kseistrup).

---

Standing in a directory with a setup.cfg file, using python2.7:

#!python

>>> import configparser
>>> parser=configparser.ConfigParser()
>>> parser.read('setup.cfg')
[]
>>> parser.read(u'setup.cfg')
[u'setup.cfg']
>>>

So if there accidentally is a file byt the name of s, e, t, u, p, c, f, or g, that file will be read:

#!python

$ ln -s setup.cfg s
$ python2.7
>>> import configparser
>>> parser=configparser.ConfigParser()
>>> parser.read('setup.cfg')
['s']
>>>

Hint: ConfigParser()'s .read() method says that the filename should only be included in the file list if the name is a unicode instance, and thus a plain string object gets treated as a list.

[jaraco]

Original comment by Klaus Alexander Seistrup (Bitbucket: kseistrup, GitHub: kseistrup).

---

Possible fix

#!diff

--- a/configparser.py
+++ b/configparser.py
@@ -701,7 +701,7 @@ class RawConfigParser(MutableMapping):
 
         Return list of successfully read files.
         """
-        if isinstance(filenames, unicode):
+        if isinstance(filenames, unicode) or isinstance(filenames, str):
             filenames = [filenames]
         read_ok = []
         for filename in filenames:

[jaraco]

Original comment by Anonymous.

---

I am running into this bug also, please fix it.

[jaraco]

Original comment by Łukasz Langa (Bitbucket: ambv, GitHub: ambv).

---

Sorry to have kept you waiting for so long. I somehow missed notifications about that issue.

This is now fixed in version 3.3.0r1. Thank you for your feedback.

[jaraco]

Original comment by Klaus Alexander Seistrup (Bitbucket: kseistrup, GitHub: kseistrup).

---

Cool, thanks — and happy new year! :)