Most used commands in history

Most of the “most used commands” approaches does not consider pipes and other complexities.

This approach considers pipes, process substitution by backticks or $() and multiple commands separated by ;

Perl regular expression breaks up each line using | or <( or ; or ` or $( and picks the first word (excluding “do” in case of for loops)

history | perl -F"\||<\(|;|\`|\\$\(" -alne 'foreach (@F) { print $1 if /\b((?!do)[a-z]+)\b/i }' | sort | uniq -c | sort -nr | head

Let’s generate a fake history file which looks like this:

1  command file | command file | command | command
2  command <(command file) <(command file)
3  command file > file
4  for i in `command file`; do command file; command file; done | command
5  for i in $(command file); do command file; command file | command; done

This approach successfully counts 16 occurrences of “command” and 2 occurrences of “for”.

Note: if you are using lots of perl one-liners, the perl commands/functions will be counted as well in this approach, since semicolon is used as a separator