Just like uniq, but works when there are non-adjacent unique lines.

Suppose you have fruits.txt and you want the unique fruits:

$ cat fruits.txt
blueberry
apple
blueberry
apple
cherry
cherry

You could try passing it through uniq, but uniq only gets adjacent lines.

$ cat fruits.txt | uniq
blueberry
apple
blueberry
apple
cherry

To get the actually unique lines, use uniqna.

$ cat fruits.txt | uniqna
blueberry
apple
cherry

Another useful feature is verbose mode, which prints out summary statistics to stderr every few lines. This is great for parsing very long streams with lots of repeated lines, because you can see how much progress has been made and get a sense for how much uniqueness there is.

For example, yes prints out an infinite sequence of lines containing y. Using head to get the first million lines, uniqna only prints out the first line, but continues to output statistics to stderr.

$ yes | head -n 1M | uniqna -v 100000
y
lines: 100000, uniques: 1, 0.00100% unique
lines: 200000, uniques: 1, 0.00050% unique
lines: 300000, uniques: 1, 0.00033% unique
lines: 400000, uniques: 1, 0.00025% unique
lines: 500000, uniques: 1, 0.00020% unique
lines: 600000, uniques: 1, 0.00017% unique
lines: 700000, uniques: 1, 0.00014% unique
lines: 800000, uniques: 1, 0.00012% unique
lines: 900000, uniques: 1, 0.00011% unique
lines: 1000000, uniques: 1, 0.00010% unique

Unlike uniq, uniqna has to keep track of every line it has seen so far. If every line is unique, that means keeping all of stdin in memory. This means that large streams with lots of uniqueness can cause issues. Small datasets and large datasets with lots of repetition work well.

Two reasons:

First, speed. uniqna only keeps track of unique lines. If we get N lines, in the best case (every line is identical) that's just one line to track, but in the worst case (every line is unique) that's N lines to track. However, sort | uniq has to track every line no matter what -- every case is the worst case. If you have a multi-gigabyte stream with lots of repetition, uniqna will probably handle it fine, but sort | uniq will definitely be slower and might even OOM.

Second, semantics. uniqna preserves the order of the first time the line is seen. If that's meaningful for your use-case, sort | uniq won't work for you because it produces the lines in sorted order. If you want the sort | uniq semantics with the speed advantages of uniqna, consider uniqna | sort.