Add cleanup pattern from track names in scrobbling operations

[silversonicaxel]

open todos

• prevent scrobbling of empty song
• add a visual preview in the tracklist of the part of the song affected by the cleanup, something as a strikethrough, for instance, to help see potential user manual issue:
  • song name: live forever - live
  • cleanup pattern: live
  • song displayed: live forever - live
• remove just last occurrence of the cleanup pattern, and not all occurrences
• create <TracklistCleanupForm />
• maybe anticipate, and fix, user manual mistake if - or / are missing from the cleanup pattern

description

This feature helps to clean up some album track names before the scrobbling operations.
The cleanup is intended to help those lastfm users who prefer not to see among their scrobbles patterns like - remastered, demo, live ....

So the PR add an extra field to highlight a pattern to remove from track names in scrobbling operations.

I hope I covered all scrobbling possibilities, in case, let me know where to fix.

[silversonicaxel]

Or, if you have a different or better ideas, let's discuss about it.

[silversonicaxel]

I've just a question that came to my mind now, shall I add the new translation ids for all locales? Or just for the one I know and the other I do not know, simply add them in English?

[elamperti]

A lot of users will like this feature. Thanks for your contribution!

A few UX concerns:

• What happens if the user decides to filter things like the whole title of a song? (the application would probably crash or the request to lastfm will fail)

• Given a title like She's Electric - Live at Knebworth, the user may want to use just Live at Knebworth, which would end up as She's electric -. Maybe there should be a preview below of how a random track from the list would look like. ScrobbleItem is already a huge component that I don't want to make more complex, but ideally I'd love to see the matching text striked-through and grayed in the tracklist. Like:

  She's Electric - Live

  Meanwhile, we can show a preview below the filter so the users don't have to scrobble to check if their pattern is correct.

• Another situation: what if the track is Live Forever - Live and the user wants to remove Live? (yes, they should add the dash, but people usually don't read instructions 😅 )

Overall, I like this feature! Let's try to add it, keeping it as simple as possible, then we can improve it from there.

[silversonicaxel]

And I'll check your remarks.

[elamperti]

Thanks for your work!

I took the opportunity to make a number of adjustments after my first attempts to use it as user and what I saw fit for the codebase. Let me know if you have any questions or thoughts on them! Always happy to discuss improvements together :)

Seeing that it was becoming difficult to catch edge cases such as empty parenthesis, astray dashes and/or extra spaces, and considering that it could be expanded to accept wildcards, I moved from using text-based matching to crafting a regular expression. Starting with the initial user string (e.g. live), and based on a series of criteria, it's then expanded into a regexp that accounts from partial words and other nuances (check the test file).

The resulting pattern is then used to uniformly break down the track title into pieces ("crumbles") which generate an uniform output to be used at two different places: in the tracklist seen by the user (where the full text is shown, with the parts to be removed highlighted in red), and also at the moment the user scrobbles (through the function cleanTitleWithPattern()), which results in a consistent result between what the user sees and what is finally scrobbled.

This filter allows you to just enter live and filter tracks with (Live) (including parenthesis, disregarding capitalization, and removing extra spaces).

At the moment the filter will remove more than one occurrence in the text, which may be a problem with songs like Live Forever (live), but it can be circumvented by explicitly filtering the parenthesis (live). If this becomes an issue, breakStringUsingPattern() could be modified to return only one matching crumb (the last one).

From the UI perspective, I moved the filter to the top-right of the list, hidden under a "filter" icon. This keeps it easy to access while being out of the way for users who don't need it, and hopefully the icon + label are descriptive enough. This simplifies translations as well: I think that if someone wants this function, they know how it works. If they don't want it, they won't look for it. And curious users can experiment what it does without risk as well.

I'll be pushing a few more changes to track the used patterns server-side. Based on user reception and experience it can be improved and expanded (e.g. to support wildcard characters, which is something I didn't add yet because I want to approach it more carefully). Let's see how it goes! I'll include this in the next release of the scrobbler (2.9.0), as soon as I merge and stabilize the rest of the pending changes (over the following days, if everything stays on track).