The PKGBUILD for atom is designed to delete a lot of frankly useless content shipped in node_modules in order to keep down the size. I've observed that currently, the largest node_modules directory is /usr/lib/atom/node_modules/tree-sitter-ruby, coming in at a whopping 36M.

This is almost entirely due to the 29M file /usr/lib/atom/node_modules/tree-sitter-ruby/src/parser.c which is a C source file and should not be getting shipped.

In fact, there are a lot of C sources being shipped, but the largest ones are all from tree-sitter*.

$ find /usr/lib/atom/node_modules/ -name \*.c -exec du -sh {} + | sort --human-numeric-sort
[...]
1.6M /usr/lib/atom/node_modules/tree-sitter/vendor/tree-sitter/externals/utf8proc/utf8proc_data.c
2.0M /usr/lib/atom/node_modules/tree-sitter-go/src/parser.c
2.2M /usr/lib/atom/node_modules/tree-sitter-c/src/parser.c
2.4M /usr/lib/atom/node_modules/tree-sitter-python/src/parser.c
3.6M /usr/lib/atom/node_modules/tree-sitter-javascript/src/parser.c
5.2M /usr/lib/atom/node_modules/tree-sitter-bash/src/parser.c
6.2M /usr/lib/atom/node_modules/tree-sitter-rust/src/parser.c
6.5M /usr/lib/atom/node_modules/tree-sitter-cpp/src/parser.c
8.0M /usr/lib/atom/node_modules/tree-sitter-typescript/src/parser.c
29M /usr/lib/atom/node_modules/tree-sitter-ruby/src/parser.c

BTW all C sources being shipped in the node_modules directory come either from tree-sitter* or from /usr/lib/atom/node_modules/superstring/vendor/pcre/10.23/

There are a total of 70MB worth of them all told.