Count and rank: sort, uniq and wc

votes.txt has one line per vote for a favourite fruit. Which fruit won? Three filters answer it.

uniq only sees neighbours

uniq collapses repeated lines into one, but only when the repeats are next to each other. Run it on the raw votes and you'll see mango several times:

So sort first, which puts identical lines side by side, then uniq:

Counting

uniq -c puts a count in front of each line:

Ranking

The counts are now at the start of each line, so a numeric sort ranks them. -n sorts by number (so 10 comes after 9, not after 1) and -r puts the biggest first:

Add | head -n 3 for a top three. This one pipeline, sort | uniq -c | sort -nr, answers lots of everyday questions: the most common error in a log, the busiest visitor, the most used word.

wc -l completes the set: wc -l votes.txt is the number of votes, and sort -u votes.txt | wc -l the number of different fruits.