And that's the problem.

I can either just limit listens to a single configurable tag and discard the others or deduplicate all values with some kind of heuristics.
I will see. I will need you to test it, since I don't have any track with lastfm listens...

Ok, current algorithm will consider a duplicate any listen equal to another one within a 3 secs window.

It should only try to deduplicate when there are last.fm listens, otherwise it just considers all of them (and does not affect performance for other users).

Check the repository. Seems to work fine with an array of dates, check it with your library. (probably all your listening times will be halved)

Since historical last.fm times are most likely to predate listens generated by foo_enhanced_playcount, I simply count all last.fm data up until the very first play recorded by the component. After that, I'd count component listens only.

Quote from: marc2k3 on 2024-02-21 20:27:59

Since historical last.fm times are most likely to predate listens generated by foo_enhanced_playcount, I simply count all last.fm data up until the very first play recorded by the component. After that, I'd count component listens only.

That could do too, but then there are also situations where people is playing things on the mobile and therefore there would be also valid last.fm times posterior to first play on component.

It's fine for me to allow a bigger time window, but anyway this is just starting to be a last.fm quirk and how listens are scrobbled which is not something on my side. Will put a 30 secs window and it's done for me.

Played times comes from enhanced playcount. I have zero idea what reset statistics does for specific tags or where it comes from.

but then there could be situations where people play things on the phone and there would also be valid last.fm times posterior to first play by the component (if they are imported)

Only remains the day with a lot of hours that belong to the previous one.

I mean that the repot take a month forward. For example In the 2024 it gives values to march, for example it says special day March 17 and the artist graphs have an offset month too.

That was already fixed ?

Nevermind, dates were not being parsed right at tag values due to months being 0-11 in js and 1-12 in tags...

Error: Spider Monkey Panel v1.6.1-mod ({341046C8-56E8-4AC9-8E11-E25A8789D068}: Playlist Tools: Buttons Bar v1.0.0 by regorxxx)
Wrapped.getGenreImg: group not recognized downtempo_cluster

File: wrapped.js
Line: 1362, Column: 11
Stack trace:
  getGenreGroupImg@wrapped.js:1362:11
  getMainGenreGroupImg@wrapped.js:1377:15
  getMainGenreImg@wrapped.js:1623:42
  formatLatexReport@wrapped.js:1770:53
  createPdfReport/<@wrapped.js:1577:17

Quote from: regor on 2024-02-23 00:53:21

That was already fixed ?

Nevermind, dates were not being parsed right at tag values due to months being 0-11 in js and 1-12 in tags...

Fixed now.

I noticed that in the country report Los Deltonos have 92 listens, they are from Spain. They have more listens than the United Kingdom's combine, but Spain doesn't show in the counties report.

All Los Deltonos' songs have LOCALE LAST.FM tag, multivalue "Cantabria; Spain" from the biography panel

Edit: The tag have a ; but it is not multivalue.

Locale tags are not used. You must have the track on the world map library.

Yes. write to json. It will be the default option on new installations from now on and document it on wrapped installation instructions.

If your files have the tag, you don't really need to start playback, just scroll through all tracks. Or use the merge tags to JSON as you did.

If tag is retrieved from bio, preview all tracks for 1 o 2 secs and done.

Note you don't need to preview "all" tracks, just 1 track per artist. Such playlist is created with the "find artist without locale..."

There are multiple scripts using the JSON database, so it's useful to fill it. You have multiple scoring and filter options per country in search by distance with it for ex.

I may add more code to merge file tags (if present) with he JSON for wrapped. But it has zero priority for me.

EDIT: Updated readme, let me know if it's clear enough.
Spoiler (click to show/hide)

Quote

Creates a report of user listening habits during the selected year
as a PDF and outputs some playlists recommendations.
   - Computes statistics from the library by:
      * Listening behavior
      * Listens
         + Total (minutes/days)
         + Day with most listens
      * Genres/styles
         + Total
         + Top 5
         + Suggested genres
      * Musical categories
      * Tracks
         + Total
         + Most played track
         + Top 5
      * Artists
         + Total
         + Top 5
         + Months reels
         + Most played artist (and track)
      * Regions
         + Top 5
         + Artist/Country with more listens
      * Sound town
         + Top 3 artists from that city
   - Downloads artist and city images.
     (Offline mode may be set to skip image downloading)
   - Retrieves album covers.
   - Generates random backgrounds (which can be customized by user).
   - Outputs a report in Latex format. Compilation in PDF.
   - Playlists recommendations:
      * Top Favourite Songs
      * Discovered Songs (during the year)
      * Top Artists
      * Top Genres
      * Top Countries
      * Suggested Genres (you may like) (ListenBrainz)
      * Suggested Artists (you make like) (ListenBrainz)
        (0ffline mode may be set to use only queries)

Requirements:
-------------
   - Playback Statistics:
   https://www.foobar2000.org/components/view/foo_playcount

   - Enhanced Playback Statistics:
   https://www.foobar2000.org/components/view/foo_enhanced_playcount

   - LaTeX: Recommended LuaLaTeX via MiKTeX. Any version works.
      * Win 10+: any version
      https://miktex.org
      * Win 7 / Wine (64 bit): 22.7 version
      https://www.npackd.org/p/org.miktex.MiKTeX64/22.7
      * Win 7 / Wine (32 bit): 21.6 version
      https://www.npackd.org/p/org.miktex.MiKTeX/21.6

Optional:
---------
   - World-Map-SMP: To get country stats. Tags must be written
   to the JSON database. If locale tags are already present
   on tracks, use 'Database...\Merge file tags with JSON...';
   otherwise you will need to retrieve tags using the Bio panel.
   Note artists without a country on the JSON database will
   not be used on the country/city stats, so be sure to get
   the tags for all artists in your library (use 'Database...\
   Find artists without locale tags...').
   https://github.com/regorxxx/World-Map-SMP

   - Biography (Wil-B):
   https://github.com/Wil-B/Biography

   - ListenBrainz token: It's not a requisite to get
   playlists suggestions, since it will use the public API,
   but it will be less probably to reach rate limits with it.
   https://listenbrainz.org/settings/import/

   - TeXstudio: To compile manually (if desired), can also be
   configured to use LuaLaTeX engine, giving identical results
   to the automated CMD build.
   https://github.com/texstudio-org/texstudio
      * Win 10+: any version
      * Win 7: 4.3.0 (or lower)

   - ExifTool: To clean all metadata from images used on report.
   Binaries must be put at: .\helpers_external\exiftool\exiftool.exe
      * 64 bit: https://exiftool.org/
      * 32 bit: https://oliverbetz.de/pages/Artikel/ExifTool-for-Windows

   - Pingo: To optimize images used on report (will greatly reduce file size).
   Binaries must be put at: .\helpers_external\pingo\pingo.exe
      * 64 bit: https://css-ig.net/pingo
      * 32 bit: not available

LaTeX compilation:
------------------
The default installation uses LuaLaTeX via CMD:
lualatex --enable-installer --interaction=nonstopmode --jobname=Wrapped_%4 --output-directory=%3 %1

This command may be changed at the settings menu, in case other
other engine or format is preferred (any layout problems is your
responsibility though).

The CMD expression allows these numbered arguments which
are replaced at execution:
   - %1: input .tex file path
   - %2: output .pdf file path
   - %3: output directory path
   - %4: year

Ex:
lualatex --enable-installer --interaction=nonstopmode --jobname=Wrapped_2023 --output-directory="B:\foobar2000\profile\js_data\temp\wrapped" "B:\foobar2000\profile\js_data\temp\wrapped\Wrapped_2023.tex"

getGenreGroupImg: function (group, root = this.basePath, bRelative = true) {
		group = group.toLowerCase();
		let fileName = '';
		switch (group) {
			case 'industrial_cluster':
			case 'metal_cluster':
			case 'rock_cluster':
			case 'pop_cluster':
			case 'country_cluster':
			case 'r&b_cluster':
			case 'blue_note_cluster':
			case 'jamaican_cluster':
			case 'rap_cluster':
			case 'downtempo_cluster':
			case 'folk_cluster':
				fileName = group.replace(/([_&])|(cluster)/g, '');
				break;
			case 'breakbeat dance_cluster':
				fileName = 'breakbeat';
				break;
			case 'four-to-the-floor dance_cluster':
				fileName = 'fourfloor';
				break;
			case 'classical music_cluster':
				fileName = 'classical';
				break;
			default:
				throw new Error('Wrapped.getGenreImg: group not recognized ' + group);
		}
		return (bRelative ? '' : root) + 'img\\genres\\' + fileName.replace(/ /g, '') + '.png';
	}

Nope, thanks.
Fixed, was a upper case typo

Can not update github but if you want to fix it locally, just substitute the 'getGenreGroupImg' function around line 1362 at wrapped.js with this one (only changes 2 values to lower case):

Code: [Select]

getGenreGroupImg: function (group, root = this.basePath, bRelative = true) {
		group = group.toLowerCase();
		let fileName = '';
		switch (group) {
			case 'industrial_cluster':
			case 'metal_cluster':
			case 'rock_cluster':
			case 'pop_cluster':
			case 'country_cluster':
			case 'r&b_cluster':
			case 'blue_note_cluster':
			case 'jamaican_cluster':
			case 'rap_cluster':
			case 'downtempo_cluster':
			case 'folk_cluster':
				fileName = group.replace(/([_&])|(cluster)/g, '');
				break;
			case 'breakbeat dance_cluster':
				fileName = 'breakbeat';
				break;
			case 'four-to-the-floor dance_cluster':
				fileName = 'fourfloor';
				break;
			case 'classical music_cluster':
				fileName = 'classical';
				break;
			default:
				throw new Error('Wrapped.getGenreImg: group not recognized ' + group);
		}
		return (bRelative ? '' : root) + 'img\\genres\\' + fileName.replace(/ /g, '') + '.png';
	}