freq

freq

Compute frequency distributions from stdin.

Install or update it:

go install github.com/clfs/freq@latest

Uninstall it:

rm -i $(which freq)

Examples

Usage:

$ freq -h
Usage of freq:
  -by string
        token type: line, byte, or rune (default "line")

By lines:

$ awk '{print $1}' go.sum | freq
2	golang.org/x/exp
2	golang.org/x/text

By bytes:

$ head -c 10 /bin/ls | freq -by byte
4	00	�	<control>
1	01	�	<control>
1	02	�	<control>
1	ba	º	MASCULINE ORDINAL INDICATOR
1	be	¾	VULGAR FRACTION THREE QUARTERS
1	ca	Ê	LATIN CAPITAL LETTER E WITH CIRCUMFLEX
1	fe	þ	LATIN SMALL LETTER THORN

By runes:

$ echo -n "Hello? Привет? Hello?" | freq -by rune
4	U+006C	l	LATIN SMALL LETTER L
3	U+003F	?	QUESTION MARK
2	U+0020	 	SPACE
2	U+0048	H	LATIN CAPITAL LETTER H
2	U+0065	e	LATIN SMALL LETTER E
2	U+006F	o	LATIN SMALL LETTER O
1	U+041F	П	CYRILLIC CAPITAL LETTER PE
1	U+0432	в	CYRILLIC SMALL LETTER VE
1	U+0435	е	CYRILLIC SMALL LETTER IE
1	U+0438	и	CYRILLIC SMALL LETTER I
1	U+0440	р	CYRILLIC SMALL LETTER ER
1	U+0442	т	CYRILLIC SMALL LETTER TE

Why write this?

This is the usual recommendation for computing frequency distributions:

cat example.txt | sort | uniq -c | sort -nr

However, it's not necessary to sort the input. Sorting is slow and uses a lot of memory. In comparison, freq is faster and has more features.

Some people also recommend using awk, like this:

awk ' { tot[$0]++ } END { for (i in tot) print tot[i],i } ' example.txt | sort

It doesn't use any sorting, but it seems hard to remember.