sort Command in Linux: Sort Text, Numbers, Columns, and More

The sort command in Linux reads lines from files or standard input and writes them to standard output in sorted order. By default, it sorts alphabetically in ascending order and never modifies the original file.

How to Use the sort Command in Linux

The syntax is:

bashsort [OPTIONS] [FILE]...

To sort a file alphabetically:

bashsort file.txt

sort writes the result to standard output. To save it, redirect the output:

bashsort file.txt > sorted.txt

Never redirect sort output back to the same file. The shell opens the output file for writing, truncating it to zero bytes, before sort even reads the input. sort file.txt > file.txt will empty your file. Use the -o option instead, which is safe for the same input and output file:

bashsort -o file.txt file.txt

To sort in reverse (descending) order:

bashsort -r file.txt

When multiple files are passed, their contents are merged and sorted together as a single stream, not sorted separately.

Sort Numerically, by Column, and Remove Duplicates

Alphabetical sort fails for numbers. The string 10 sorts before 2 because the character 1 comes before 2 in ASCII. sort compares characters one at a time, not values. Use -n to sort by numeric value:

bashsort -n numbers.txt

To sort numerically in reverse (largest first):

bashsort -nr numbers.txt

Sort by a specific column with -k. Fields are whitespace-separated by default. To sort by the second column numerically:

bashsort -k2 -n files.txt

To use a custom field delimiter, add -t. This example sorts /etc/passwd by the third field (UID) using : as the separator:

bashsort -t: -k3 -n /etc/passwd

Remove duplicate lines with -u. After sorting, any adjacent duplicate lines are dropped from the output:

bashsort -u file.txt

This produces the same result as sort file.txt | uniq, but -u is slightly more efficient because it removes duplicates during the sort pass. Piping to uniq separately is more flexible: uniq -c counts occurrences and uniq -d shows only the lines that appear more than once.

Check if Sorted, Human-Readable Sizes, and Pipelines

Check if a file is already sorted with -c. It exits silently if the file is in order, or prints an error and exits with a non-zero status if any line is out of place:

bashsort -c file.txt

sort: file.txt:2: disorder: banana

This is useful in scripts to validate input before processing it further.

Sort human-readable sizes with -h. Standard numeric sort (-n) only parses plain integers and cannot interpret SI suffixes like K, M, or G. Use -h when working with output from du:

bashdu -sh /var/* | sort -h

-h correctly orders 4.0K, 16K, 1.2M, and 3.4G by magnitude rather than alphabetically.

Pipeline examples. sort is most powerful when combined with other tools:

Find the ten largest directories:

bashdu -sh /var/* | sort -rh | head -10

Count and rank the most frequent words in a file:

bashgrep -Eo '[[:alnum:]_]+' file.txt | sort | uniq -c | sort -rn

Sort a specific column from a CSV file:

bashcut -d',' -f2 data.csv | sort

The sort command handles most ordering tasks on its own, but it becomes a core building block for data processing when combined with uniq, cut, head, and grep in pipelines. Leave a comment below if you run into any issues.