Sed Command Practice
sed reads a stream, applies an address plus a command, and writes the result. This track drills s///g, range delete, -n print, and -i in-place edits — the scripts you type in a pipeline, not a GUI find-and-replace. Grep selects lines; sed changes them. Awk can wait for its own field track.
Command Reference
delete
| Command | Description |
|---|---|
sed '<num>,<num>d' <file> | Delete a start-end line range and print the rest. |
sed '$d' <file> | Delete the last line and print the rest. |
sed '/^$/d' <file> | Delete empty lines and print the rest. |
sed '/^#/d' <file> | Delete lines that start with # and print the rest. |
sed '/^<pattern>/d' <file> | Delete lines whose start matches a pattern and print the rest. |
| Command | Description |
|---|---|
sed -n '<num>,<num>p' <file> | Print only a start-end line range. |
sed -n '/<pattern>/p' <file> | Print only lines that match a pattern. |
sed -n '<num>p' <file> | Print only one numbered line; suppress the default print of every line. |
sed -n '/<pattern>/!p' <file> | Print only lines that do not match a pattern. |
sed -n '/<pattern>/Ip' <file> | Print matching lines, ignoring case in the search. |
sed -n '$=' <file> | Print the last line number (a line count) and nothing else. |
replace
| Command | Description |
|---|---|
sed -i 's/<old>/<new>/g' <file> | Edit the named file in place, replacing every match on each line. |
sed 's/<old>/<new>/g' <file> | Replace every match of a string on each line and print the result. |
sed 's/<old>/<new>/2' <file> | Replace only the second match of a string on each line. |
sed '<num>s/<old>/<new>/' <file> | Replace the first match on one numbered line only. |
sed '/<pattern>/s/<old>/<new>/' <file> | Replace the first match only on lines that match a pattern. |
sed -i.bak 's/<old>/<new>/g' <file> | Edit the file in place after writing a .bak backup, replacing every match. |
sed -E 's/<pattern>//g' <file> | Delete every extended-regex match on each line (for example a digit run). |
script
| Command | Description |
|---|---|
sed -e 's/<old>/<new>/g' <file> | Run one substitute script passed with -e. |
sed -f <script> <file> | Run commands from a sed script file against an input file. |
sed -e 's/<old>/<new>/g' -e 's/<pattern>/<text>/g' <file> | Run two substitute scripts in order on the same input. |
sed -n -e '<num>,<num>p' -e '<num>,<num>p' <file> | Print two line ranges and suppress the default print of every line. |
insert
| Command | Description |
|---|---|
sed '<num>i <text>' <file> | Insert a text line before a numbered line. |
sed '$a <text>' <file> | Append a text line after the last line of the file. |
sed '<num>a <text>' <file> | Append a text line after a numbered line. |
sed '/<pattern>/i <text>' <file> | Insert a text line before every line that matches a pattern. |
Key Use Cases
- Replace every occurrence of a token in a file and print the result
- Edit a config in place and keep a .bak copy
- Delete blank lines or comment lines from a stream
- Print only lines that match a pattern, or only a line range
- Insert a banner before matching lines or append text at EOF
- Strip digits with an extended regex (-E)
Frequently Asked Questions
How is sed different from grep?
grep prints lines that match. sed transforms the stream: substitute, delete, insert, or print a subset. sed -n '/pat/p' can look like grep, but s/// and -i do not.
What does the g in s/old/new/g do?
Without g, only the first match on each line is replaced. g replaces every match on the line. A number like 2 replaces only the second match.
Does sed -i change the file I named?
Yes. GNU -i edits in place. -i.bak writes a backup first. Without -i, sed prints to stdout and leaves the file alone.
When do I need sed -E?
Basic sed regex treats + and grouping differently. -E (--regexp-extended) lets you write [0-9]+ without extra backslashes.
Ready to master Sed commands?
Test your muscle memory with our spaced-repetition quiz system. Free forever.