Language stats incorrectly show 99% JavaScript (should be different)

[catalinacatalinacatalina]

Hi, I’ve noticed that the language breakdown for my repository is showing ~99% JavaScript, but this is not accurate.
MY REPOSITORY

The project is primarily written in Processing, but it seems GitHub Linguist is incorrectly counting the files. I would like to see a more accurate language distribution that reflects the actual codebase.

What can I do to fix the language statistics?

Thanks in advance for any guidance!

[lildude]

The language stats are 100% accurate. Whilst you think the project it primarily written in Processing, the total bytes of JavaScript far outweighs everything else:

$ github-linguist GenerativeArtworks/
99.87%  31315869   JavaScript
0.12%   36865      Processing
0.01%   2506       HTML
0.00%   426        CSS
$

As you can see, you have over 30mb of JavaScript but only 36kb of Processing.

Why? Because you have six 5mb JavaScript files:

$ ls -lSh GenerativeArtworks/arte_generativo/**/*.js
-rw-r--r--@ 1 lildude  staff   5.0M 29 Aug 14:57 GenerativeArtworks/arte_generativo/p5js- tipografias/static/p5.js
-rw-r--r--@ 1 lildude  staff   5.0M 29 Aug 14:57 GenerativeArtworks/arte_generativo/p5js-ondas/p5.js
-rw-r--r--@ 1 lildude  staff   5.0M 29 Aug 14:57 GenerativeArtworks/arte_generativo/p5js-ruido perlin/p5.js
-rw-r--r--@ 1 lildude  staff   5.0M 29 Aug 14:57 GenerativeArtworks/arte_generativo/p5js-sistema de particulas/p5.js
-rw-r--r--@ 1 lildude  staff   5.0M 29 Aug 14:57 GenerativeArtworks/arte_generativo/p5js-vera molnar lineas/p5.js
-rw-r--r--@ 1 lildude  staff   5.0M 29 Aug 14:57 GenerativeArtworks/arte_generativo/p5js-vera molnar triangulos/p5.js
-rw-r--r--@ 1 lildude  staff   195K 29 Aug 14:57 GenerativeArtworks/arte_generativo/p5js- tipografias/static/p5.sound.min.js
[...]

… and they're all identical:

$ find ./ -name p5.js -exec cksum {} \;
1271484993 5217711 ./GenerativeArtworks/arte_generativo/p5js-ondas/p5.js
1271484993 5217711 ./GenerativeArtworks/arte_generativo/p5js-vera molnar lineas/p5.js
1271484993 5217711 ./GenerativeArtworks/arte_generativo/p5js- tipografias/static/p5.js
1271484993 5217711 ./GenerativeArtworks/arte_generativo/p5js-ruido perlin/p5.js
1271484993 5217711 ./GenerativeArtworks/arte_generativo/p5js-sistema de particulas/p5.js
1271484993 5217711 ./GenerativeArtworks/arte_generativo/p5js-vera molnar triangulos/p5.js
$

So to fix this, you will need to use an override to exclude those files. Marking the files as vendored would probably be the best option.

[catalinacatalinacatalina]

I didn’t know that the language statistics were based on file size (bytes) instead of lines of code — I thought it worked by lines. It would be great if there was an option to change the weight calculation.

I applied the suggested .gitattributes configuration, and now everything looks perfect. Problem solved! Thank you!

[lildude]

I didn’t know that the language statistics were based on file size (bytes) instead of lines of code — I thought it worked by lines. It would be great if there was an option to change the weight calculation.

Why? It wouldn't make any difference to this situation as each of those files have 147330 lines each, with only the first line being a comment. One file alone has more lines of code than all of your Processing code. It's worth pointing out that those files aren't minified either. Your problems would be resolved without an override if you used the minified versions of those files as Linguist ignores minified files.