pydoc info for a package doesn't list all package contents

[ngrover]

BPO	1613479
Nosy	@merwok
Files	pydoc.py

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2011-02-04.00:03:51.793>
created_at = <Date 2006-12-11.20:40:12.000>
labels = ['type-bug', 'library']
title = "pydoc info for a package doesn't list all package contents"
updated_at = <Date 2011-02-04.00:03:51.791>
user = 'https://bugs.python.org/ngrover'

bugs.python.org fields:

activity = <Date 2011-02-04.00:03:51.791>
actor = 'eric.araujo'
assignee = 'none'
closed = True
closed_date = <Date 2011-02-04.00:03:51.793>
closer = 'eric.araujo'
components = ['Library (Lib)']
creation = <Date 2006-12-11.20:40:12.000>
creator = 'ngrover'
dependencies = []
files = ['2252']
hgrepos = []
issue_num = 1613479
keywords = ['patch']
message_count = 3.0
messages = ['30787', '109838', '127831']
nosy_count = 2.0
nosy_names = ['ngrover', 'eric.araujo']
pr_nums = []
priority = 'normal'
resolution = 'out of date'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue1613479'
versions = []

[ngrover]

When using pydoc to query a package, a "PACKAGE CONTENTS" list is provided to show the modules and packages that are in that package. That list will be incomplete if we are querying a package that has been split across multiple directories.

Suppose you have the following:

/first/path/foo/init.py
/first/path/foo/one.py
/second/path/foo/init.py
/second/path/foo/two.py

and sys.path includes /first/path/ and /second/path/. If both of the foo/init.py files are empty, then "import foo" will only allow you to import modules from one of those two foo/ directories (whichever is found first in sys.path). However, if we add the following to both foo/init.py files, then we can import foo.one and foo.two because "foo" is considered to be a single package split across two directories:

from pkgutil import extend_path
__path__ = extend_path(__path__, __name__)

Please see http://www.python.org/doc/2.4.2/lib/module-pkgutil.html for some related information.

On line 1052 of pydoc.py, we have the following:

for file in os.listdir(object.__path__[0]):

and in that loop we only read the contents of the FIRST directory in the package's __path__. That should be updated to read the contents of ALL directories in the package's __path__. The following change will do that:

% diff -w pydoc.py pydoc.py.orig
1052,1054c1052,1053
< for objectDir in object.__path__:
< for file in os.listdir(objectDir):
< path = os.path.join(objectDir, file)
---

        for file in os.listdir(object.\_\_path__[0]):
            path = os.path.join(object.\_\_path__[0], file)

I've attached that updated pydoc.py file to this submission. Please consider that as a replacement for the existing pydoc.py module that's currently being distributed.

[BreamoreBoy]

Surely this is a bug?

[merwok]

I’ve just tested that the behavior is now correct. I reproduced the tree structure, set PYTHONPATH, ran “pydoc foo” to get only “one” listed, added the pkgutil call in first/path/foo/init.py, re-ran “pydoc foo”, got “one” and “two” listed. I can also import both submodules.

The pydoc code now uses pkgutil.iter_modules. Someone must have fixed that before 2.6, closing. Thanks for the report nonetheless!