Perl One-Liner Cheat Sheet (for ex-`sed` users)

perl-not-sed.md

Perl One-Liner Cheat Sheet (for ex-sed users)

A pocket reference for replacing sed with perl. The big win: perl -i behaves identically on macOS and Linux, unlike sed -i (BSD vs GNU).

---

The flags you actually need

Flag	Meaning
-e	"the program is this string" (the code lives on the command line)
-p	loop over every line, run the code, then print the line (like sed)
-n	loop over every line, run the code, don't auto-print (you print)
-i	edit file(s) in place
-i.bak	edit in place, but save the original as file.bak first
-l	auto-chomp input newlines and re-add them on print (tidy line handling)
-a	autosplit each line into @F on whitespace (awk-like)
-F	set the autosplit delimiter, e.g. -F: or -F','

Mnemonic: -p prints for you, -n does not.

---

sed → perl, side by side

Task	sed	perl
Substitute (stream to stdout)	sed 's/old/new/'	perl -pe 's/old/new/'
Substitute, all on line	sed 's/old/new/g'	perl -pe 's/old/new/g'
Substitute in place	sed -i 's/a/b/' f (GNU)	perl -i -pe 's/a/b/' f
In place, keep backup	sed -i.bak 's/a/b/' f	perl -i.bak -pe 's/a/b/' f
Case-insensitive	sed 's/old/new/gI'	perl -pe 's/old/new/gi'
Print only matching lines	sed -n '/foo/p'	perl -ne 'print if /foo/'
Delete matching lines	sed '/foo/d'	perl -ne 'print unless /foo/'
Print a specific line (#5)	sed -n '5p'	perl -ne 'print if $. == 5'
Print a line range (5–10)	sed -n '5,10p'	perl -ne 'print if 5..10'

The macOS gotcha: BSD sed -i requires an argument (sed -i '' 's/a/b/' f), GNU sed -i forbids one. perl -i just works the same everywhere.

---

Substitution tricks

# Pick a different delimiter when the pattern has slashes — no escaping needed
perl -pe 's{/usr/local}{/opt}g'        # braces
perl -pe 's|/usr/local|/opt|g'         # pipes
 
# Capture groups + backreferences ($1, $2, ...)
perl -pe 's/(\w+)\s+(\w+)/$2 $1/'      # swap two words
 
# Anchors
perl -pe 's/^/    /'                   # indent every line 4 spaces
perl -pe 's/\s+$//'                    # strip trailing whitespace
 
# Case-changing in the replacement
perl -pe 's/(\w+)/\U$1/'               # uppercase   (\L lower, \u/\l one char)
 
# Only substitute on lines matching a condition
perl -pe 's/foo/bar/ if /^config/'     # only on lines starting with "config"

---

Editing files in place

# One file
perl -i -pe 's/old/new/g' file.txt
 
# Many files at once
perl -i -pe 's/2025/2026/g' *.md
 
# Safest habit: keep a backup until you trust the result
perl -i.bak -pe 's/old/new/g' file.txt

Want to preview first? Drop the -i — it streams the result to your terminal without touching the file. Confirm, then re-run with -i.

---

Beyond sed: small awk-style jobs

# Print the 2nd whitespace-separated field of each line
perl -ane 'print "$F[1]\n"'
 
# Sum the first column of numbers
perl -ane '$sum += $F[0]; END { print "$sum\n" }'
 
# CSV: print column 3 (naive split — fine for simple, comma-free data)
perl -F, -ane 'print "$F[2]\n"'
 
# Grep with a regex, but also rewrite as you go
perl -ne 'print "$1\n" if /id=(\d+)/'

---

Multi-line / whole-file mode

# Slurp the entire file as one string (-0777), then operate across newlines
perl -0777 -pe 's/\n{3,}/\n\n/g' file   # collapse 3+ blank lines into one
 
# This is something plain sed can't do cleanly — a real reason to switch.

---

Quick gotchas

• Quote with single quotes in the shell so $1, @F, etc. aren't eaten by bash.
• @ inside a double-quoted Perl context looks like an array — escape it: \@.
• $ means "end of line" in a regex but "variable" elsewhere; mind the context.
• -p already prints; adding your own print will duplicate lines. Use -n when you print manually.
• Test destructive edits on a copy, or use -i.bak, before running on *.

---

The 5 you'll use 90% of the time

perl -pe 's/old/new/g'            # find & replace, preview to stdout
perl -i -pe 's/old/new/g' file    # ...same, in place
perl -i.bak -pe 's/old/new/g' f   # ...in place with a safety backup
perl -ne 'print if /pattern/'     # grep-ish
perl -ne 'print unless /pattern/' # delete matching lines