New feature: Deduplicator

[mesaprotector]

Description

Some of my mutuals have taken to reblogging the same post 3, 5, or even 15 times in a row, I imagine to express how much they love it or whatever.

I don't know why this has become a thing but I wish it weren't, and I'd love a feature that hides duplicate reblogs from the same person. Since this tends to come in the form of spam over a short period of time an implementation wouldn't need to keep a long-term track, just hiding duplicates within the span of an hour would be enough.

[AprilSylph]

I've been thinking about this too, but I guess I never made an issue for it. I might as well repurpose this issue!

[AprilSylph]

I'll share the details of what I have planned, too:

• Name: Deduplicator
• Description: Remove duplicate posts from timelines
• Preferences:
  • Selection: Mode
    • Option: Exact match (default)
    • Option: Match subset threads
    • Option: Match root posts

Activates on all timelines, and processes every post. (Note: we gather posts via querySelector, so they should be processed on a first-visible basis, right? Should double-check this.) The feature will need a util for parsing the IDs of the visible content in a post, consisting of the reblog trail's post IDs and, if the post has any original/added content, the ID of the post itself. What happens to a post when processed depends on the mode:

In "exact match" mode, all the visible content is considered. If it's seen a post in that timeline element before that has the exact same content, the post is hidden; if not, it saves the post IDs combination for later checking. (Probably would work via WeakMap using the timeline element as keys, and the post ID combination as just a big string of comma-separated post IDs?) This would mean spam reblogs would be hidden, as well as any original posts that Tumblr erroneously duplicates in the same feed (which I've seen happen on the "Your tags" dashboard tab, not sure if that was ever fixed).

(Actually, we'd need another level of weak mapping, I think... we would need to wipe the feature's effective memory of a timeline upon refreshing it, so we would actually have to keep track of post containers as well. Well, that's an architecture problem for future April.)

In "subset" mode, all the visible content is considered, but the post's content IDs don't have to match the saved seen IDs exactly; if only a subset of the current post's IDs match a previously-seen post's content IDs, the post is hidden. This would mean spam reblogs, and less-complete versions of any reblog conversations, would be hidden.

In "root" mode, only the root post is considered, like with PostBlock. For every post in the timeline, subsequent posts sharing the same root ID would be hidden, even if their reblog threads were wildly different. This would mean spam reblogs, less-complete versions of any reblog conversations, or completely different threads sharing the same original post, would all be hidden.

(Maybe I should reduce the scope for the initial implementation... the default mode would satisfy the original issue request, after all.)

[mesaprotector]

I just remembered I made this issue but: yes, the default mode is pretty much what I'm after, and I'd be happy with just that. (While it's not as much how people use Tumblr nowadays, I'd rather not cut off when people have reblog-conversations adding to a post. Maybe if in 2031 that's become a new and fashionable form of spam, I'll reopen the issue >:)