More sed Examples on Debian 12: Creative and Useful Use Cases

Once you’ve mastered the basics and a few advanced patterns, sed starts to feel less like a tool and more like a language. Its syntax may be terse, but it allows for elegant and powerful transformations. This article explores even more examples of how sed can be used effectively on Debian 12, especially for creative or specialized tasks.

Let’s begin with extracting specific parts of a line. Say you have a CSV file and want only the second column. Assuming comma-separated values:

sed 's/^[^,]*,\([^,]*\).*/\1/' file.csv

Here’s the breakdown:

• ^[^,]*, matches the first column and comma. [^,]* matches any character except commas.
• \([^,]*\) captures the second column (everything up to the next comma).
• .* matches the rest of the line.
• \1 replaces the entire line with the second column.

This is useful when cut isn’t flexible enough—for example, if delimiters are inconsistent.

Another handy example is wrapping each line with quotes. If you want to convert:

apple
banana
cherry

into:

"apple",
"banana",
"cherry",

You can use:

sed 's/.*/"&",/' list.txt

• .* matches the whole line.
• "&”` adds quotes around the entire match.
• The trailing comma is appended as part of the replacement.

Suppose you want to insert a line after every 3 lines—for example, adding a divider line. Use a counter with ~:

sed 'n;n;a---' file.txt

This prints the first line normally, skips the next two with n, and then appends ---. It then repeats. This works when line structure is predictable.

To replace the last word in a line:

sed 's/\s\+\([^ \t]*\)$/ [\1]/' file.txt

This finds the final whitespace-separated word and puts it in brackets.

• \s\+ matches one or more spaces or tabs.
• \([^ \t]*\)$ captures the final non-whitespace sequence at the end of the line.
• The replacement adds brackets around the last word.

Let’s say you want to number each line of a file:

sed = file.txt | sed 'N;s/\n/ /'

• The first sed = prints line numbers followed by the lines.
• The second sed joins each number with its corresponding line and replaces the newline with a space.

Sometimes you want to make complex changes but only after a pattern has been matched. For instance, changing the format of all lines after a certain marker:

sed '/^START$/,$s/foo/bar/g' file.txt

• /^START$/,$ limits the substitution range to all lines from the line that says “START” to the end.
• s/foo/bar/g applies globally on those lines.

For working with JSON-like structures, you might want to collapse multiple lines into one:

sed ':a;N;$!ba;s/\n/ /g' file.json

• :a marks a label.
• N appends the next line.
• $!ba repeats this process until the last line.
• The s/\n/ /g replaces all newlines with spaces.

These examples show that sed can serve as a mini text processor, capable of restructuring, sanitizing, and formatting content in countless ways. While sed has limitations (especially with deeply nested or hierarchical data), for linear and pattern-based tasks, it’s fast, scriptable, and available out-of-the-box on Debian 12.