Patch T277832

[theresnotime]

Bug: T277832

[jsnshrmn]

Thanks for your work on this! The redaction logic looks good, but it also looks like a couple of paths got missed. Also, please separate the dependency work from the redaction work. short commit messages are the norm here, but it makes it even more important to try to keep each commit's scope tight.

[theresnotime]

Thanks for your work on this! The redaction logic looks good, but it also looks like a couple of paths got missed.

The download pages now check the rows too, but they refuse a search with more than 5000 rows, because (e.g.) ?query=WPWP has 546,166 rows and one API call checks only 50. To make large downloads work again we must keep the visibility state in our own database, from the mediawiki.revision-visibility-change stream.

[jsnshrmn]

Thanks for the changes! One inline question/suggestion.

[jsnshrmn]

Thanks for working on this!

I noticed that hashtagsv2/graphs/views.py's top_user_statistics_data, users_csv, All_users_view, and projects_csv all call hashtag_queryset directly with no redact() call.

[jsnshrmn]

lgtm; Note that I rebased, then munged the the commits related to django nose for cleanliness and speed.