Failure to download episodes from "Hello Internet" podcast

[githuq]

I have been trying to download episodes from the "Hello Internet" podcast, but greg fails with an error message:

$ greg add HelloInternet http://www.hellointernet.fm/podcast?format=rss
$ greg sync HelloInternet
Checking HelloInternet...
Traceback (most recent call last):
  File "/usr/bin/greg", line 9, in <module>
    load_entry_point('Greg==0.4.4.3', 'console_scripts', 'greg')()
  File "/usr/lib/python3/dist-packages/greg/gregparser.py", line 112, in main
    args.func(vars(args))
  File "/usr/lib/python3/dist-packages/greg/greg.py", line 746, in sync
    feed.download_entry(entry)
  File "/usr/lib/python3/dist-packages/greg/greg.py", line 286, in download_entry
    for mimetype in self.mime]):
  File "/usr/lib/python3/dist-packages/greg/greg.py", line 286, in <listcomp>
    for mimetype in self.mime]):
  File "/usr/lib/python3/dist-packages/feedparser.py", line 383, in __getitem__
    return dict.__getitem__(self, key)
KeyError: 'type'

Other podcasts work fine. I'm running Debian Jessie with stock packages (python 3.4.2 and python3-feedparser 5.1.3).

[manolomartinez]

Hi, yes, that podcast is defective in that it doesn't carry information about enclosure types, which greg relies on.

In the development-main branch, I have introduced a new notype option for greg.conf. The idea would be to have a section in this file with

[HelloInternet]

notype = yes

This would instruct greg to skip checking for type and simply download all enclosures. Would you mind installing greg from the development branch, and let me know if that works for you?

Manolo

[FilipBB]

Wouldn't the ignoreenclosures flag work?

Sent with AquaMail for Android http://www.aqua-mail.com

On July 16, 2015 7:44:04 PM manolomartinez notifications@github.com wrote:

Hi, yes, that podcast is defective in that it doesn't carry information about enclosure types, which greg relies on.

In the development-main branch, I have introduced a new notype option for greg.conf. The idea would be to have a section in this file with

[HelloInternet]

notype = yes

This would instruct greg to skip checking for type and simply download all enclosures. Would you mind installing greg from the development branch, and let me know if that works for you?

Manolo

---

Reply to this email directly or view it on GitHub: https://github.com/manolomartinez/greg/issues/31#issuecomment-122131943

[manolomartinez]

Wouldn't the ignoreenclosures flag work?

Ha, I hoped it might :) In fact it's a different way for a feed to be defective. The ignoreenclosures option is there to deal with feeds with no enclosures. This new issue is about feeds which enclosures, but no enclosure types.

I think the new notype option will help greg overcome this problem, but defective feeds are, admittedly, a bit of a wild goose chase. Perhaps at some point I will need to figure out a way for greg to expose all of the info parsed by feedparser, so that corner cases can be dealt with.

[githuq]

manolomartinez notifications@github.com writes:

I think the new notype option will help greg overcome this problem, but defective feeds are, admittedly, a bit of a wild goose chase. Perhaps at some point I will need to figure out a way for greg to expose all of the info parsed by feedparser, so that corner cases can be dealt with.

The 'notype' options sounds like it would do the trick. I'll try it out and get back to you.

Maybe you could introduce an option 'workarounds' or 'handledefects' that is list-valued? The two possible entries for the list as of now would be 'ignoreenclosures' and 'notype'. More can be added in the future to follow the goose whereever it goes.

My proposal is not fundamentally different from what is in greg now, but it allows one to call these things what they are: defects.

[manolomartinez]

Maybe you could introduce an option 'workarounds' or 'handledefects' that is list-valued? The two possible entries for the list as of now would be 'ignoreenclosures' and 'notype'. More can be added in the future to follow the goose whereever it goes.

My proposal is not fundamentally different from what is in greg now, but it allows one to call these things what they are: defects.

:) yep, defective feeds are frustrating. I don't know about adding a workarounds option, but perhaps it would be good to clearly separate bona-fide options and workaround options in greg.conf. I'll think about that, thanks!

M

[githuq]

I just had a look at the 'notype' option in the development-main branch. It does the trick and I can download the episodes from Hello Internet. Thanks for the quick patch.

[manolomartinez]

Sorry, I somehow missed your last message. Thanks for testing it. Closing for now.

[ghost]

FYI, it's not just Hello Internet, but it's largely any podcast hosted on SquareSpace. They have malformed podcast feeds and fixing them doesn't seem to be of high priority. Which is a pity.

[manolomartinez]

Ugh. Thanks for the heads up. That reminds me that, at least, I should make greg recommend adding the notype option to the feed conf when this problem arises.

[ghost]

Sounds like a good plan.