Broken Link Checker

[svenseeberg]

Verwalter should be able to automatically check all links in the content. All links that return a 200 or 301/302 are ok.

[benginoe]

What should be replied if we get any connection error which appears in try-except-manner (ConnectionError - at the moment i give back a HTTP: 503)?

I am giving back a list with link, http_code, datetime, is this enough for you @sascha11110 to display the status?

And another question: to which folder in the cms-subdirectory should i commit this class?

[benginoe]

Verwalter should have the possibility to download a excel table with all broken links

[benginoe]

There should be the possibility to inform more then one person via mail.

[benginoe]

There should also be the possibility to let the algorithm check on image tags and if the media file is still available, it should be visible in the view which type of content was checked.

[melegiul]

I found a django app for this feature, that could be helpful: django-linkcheck.
This app can crawl through the specified models to check for broken links.

Pros:

• Saves us some work
• Logic seems to be very efficient: uses post_save signals and assigns tasks to threads.
• I like the nested lists feature.

Cons:

• One model of the app uses a NullBooleanField, which is deprecated. There is a open PR, but it couldn't be merged yet.
• The default location is the admin site and the app does not use class based views, so customization will get more difficult. I would need to replace the urls, views and templates, but still use the linkcheck logic under the hood.

Do you want to give the app a try or should we re-engineer something similar?

[timobrembeck]

I'ts a pity that this is optimized for Django admin, but I guess the complicated part is the link-checking itself which we could definitely use. The link checking seems a sufficient complex topic to me that we are willing to make some concessions. If we want a 100% perfectly customized solution for us, we can also implement this in the distant future when there are no other important issues anymore 🙂

I think the NullBooleanField isn't a problem, since "deprecated" just means it will be removed in the next major version (Django 4.0)... but we won't upgrade to Django 4 until there is the LTS release 4.2 which is planned to be released in April 2023... so hopefully until then, the PR gets merged :D

One thing to test before we get too invested: Do the internals of this app also work when Django admin is not in INSTALLED_APPS? (Because Django admin is only activated in DEBUG mode)

[timobrembeck]

Regarding a mock-up of the broken link checker UI: This is the current wordpress UI:

Working Links

Broken Links

Obviously, you don't have to copy the design 1:1 since it's not perfect, but use it as a rough guideline.

[melegiul]

One thing to test before we get too invested: Do the internals of this app also work when Django admin is not in INSTALLED_APPS? (Because Django admin is only activated in DEBUG mode)

I think it should work. I just remove admin from installed apps and then it looks like this:

Seems like the only dependency to admin was {% extends "admin/change_list.html" %} in this file
But we need to remove the app filer as well when the admin app is not set.

[timobrembeck]

I think it should work.

Cool! 👍

But we need to remove the app filer as well when the admin app is not set.

Yes, I think the new media library which is currently being developed by @jnugh and @ulliholtgrave works independently from django-filer, right?

[ulliholtgrave]

Yep, never used it