dmdm / feedgenerator-py3k Goto Github PK

Though theoretically this code is new-style BSD like Django, there is no COPYING file, so this is effectively nonfree. I would really like to package this for Guix, but this is a showstopper at the moment for us...

@dmdm
Thanks for the standalone package.

Would you be willing to upgrade to the 1.9 version?

Specifically there is a missing feature in the current version which is related to the differences in between published and updated. And I need it for getpelican/pelican#1937

Or do you prefer to have a pull request?

I'm upgrading the FreeBSD port for py-feedgenerator from 1.2.1 -> 1.5, and it seems setup.py currently points test_suite to a nonexistent tests module. The following patch points it to the existing tests_feedgenerator module to make it function via python setup.py test.

Also, zip_safe=False helps us OS packagers and porters by ensuring the egg gets installed unzipped, so i've included that too.

Feel free to rename the module if you prefer that approach (A module named 'tests' is pretty common for ease of discovery)

--- ./setup.py.orig     2013-01-11 11:56:38.000000000 +1100
+++ ./setup.py  2013-05-19 14:41:39.961240204 +1000
@@ -32,7 +32,7 @@
 KEYWORDS = "feed atom rss".split(' ')
 VERSION = '1.5'

-TEST_SUITE = 'tests'
+TEST_SUITE = 'tests_feedgenerator'

 REQUIRES = ['pytz', 'six']

@@ -52,4 +52,5 @@
     keywords=KEYWORDS,
     url=URL,
     classifiers=CLASSIFIERS,
+    zip_safe=False,
 )

I came to this error through pelican but I can reproduce it by the test suite:

 $ nosetests                                                                               
ERROR: test_001_string_results (tests_feedgenerator.test_feedgenerator.TestFeedGenerator)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/fabian/dev/feedgenerator-py3k/tests_feedgenerator/test_feedgenerator.py", line 69, in test_001_string_results
    result = feed.writeString(ENCODING)
  File "/Users/fabian/dev/feedgenerator-py3k/feedgenerator/django/utils/feedgenerator.py", line 188, in writeString
    self.write(s, encoding)
  File "/Users/fabian/dev/feedgenerator-py3k/feedgenerator/django/utils/feedgenerator.py", line 215, in write
    handler.startElement("rss", self.rss_attributes())
  File "/Users/fabian/dev/feedgenerator-py3k/feedgenerator/django/utils/xmlutils.py", line 19, in startElement
    self._write('<' + name)
AttributeError: SimplerXMLGenerator instance has no attribute '_write'

----------------------------------------------------------------------
Ran 5 tests in 0.052s

FAILED (errors=1)
FAIL

Hi there,

Firstly, thanks for writing this library -- I see that it's used in a ton of places in the Python world, so +1 for that.

Secondly, I'd like to see if you'd consider renaming the lastBuildDate RSS field to pubDate.

I've been looking around the internet at various RSS clients / services, and it appears that almost none of them recognize the lastBuildDate field that is outputted by feedgenerator.

MailChimp (the largest email campaign site) doesn't recognize lastBuildDate, which causes numerous problems setting up RSS -> Email campaigns (see this support page for examples: http://kb.mailchimp.com/article/ive-posted-new-information-to-my-rss-feed-but-no-campaigns-have-been-sent/

I propose renaming the lastBuildDate field to pubDate, this way support will be largely fixed in a number of RSS clients / RSS services out there.

Thoughts?

Tags such as <script> and <video> may not be appropriate for post content in feeds, potentially causing validation problems for feed readers. Perhaps we should use the lxml Cleaner class (see: kill_tags attribute) to remove these tags — and the content within — before adding the summary/description element to the feed entry.

@dmdm: What do you think?