The number of f**ks and sh**ts in the linux kernel source

I like the experiment. I'd be curious to see what the results would be in enterprise software.

One small optimization I can think of: if the first character isn't an 'f' or 's', you can skip 4 characters ( fseek(file, 4, CURR_SET) ). Similarly for ({'u','h'}, 2), etc. You'll save traversals, lookups, and comparisons, which would be a boon in larger directories.

Would have been better if it didn't count words like "pushit" as a shit.