Why does my regex match in the preview but not in real sed?

Because the preview uses **JavaScript regex** which is closest to **POSIX extended regex (ERE)**, and **real sed defaults to BRE (basic regex)** where many metacharacters need a backslash. In BRE: `(`, `)`, `{`, `}`, `+`, `?` and `|` are **literal characters** unless escaped. So `sed 's/foo+/bar/'` does **not** mean "one or more foo"; it means "literal string `foo+`". To get the JavaScript behaviour, **use `sed -E`** (or `sed -r` on Linux), which enables extended regex. **In this helper**, the live preview assumes extended regex semantics. When you copy the command, add `-E` to your real sed call and you will get the same result.

When should I use sed and when awk?

Rule of thumb: **sed for line-level edits, awk for field-level work**. If you want to **replace a string, delete blank lines, insert a header, print lines 100 to 200**, that is sed. If you want to **print column 3, sum a column, filter rows where column 5 is greater than 100, group by column 2 and count**, that is awk. They overlap in the middle (both can match a regex and print, both can filter), but awk is **way easier** when fields matter. **awk** also has variables, arithmetic and arrays; sed only has pattern space and hold space, which are powerful but read-only-syntax-twice obscure.

How do I do "case-insensitive replace" portably?

The naive answer is `sed 's/foo/bar/I'`, **but the `I` flag is GNU-only**, so it will not work on macOS or BSD. **Portable answer**: write the character class manually, `sed 's/[fF][oO][oO]/bar/g'`. **Faster typing for long strings**: pipe through `tr` or use `awk 'tolower($0) ~ /foo/ { gsub(/[fF][oO][oO]/, "bar"); print }'` which works everywhere awk is installed. **Modern alternative**: many people just use `perl -pe 's/foo/bar/gi'` for cross-platform case-insensitive substitute, since Perl is also on every Unix and its regex is the closest to JavaScript regex.

What does the `-n` flag do in sed?

**It turns off automatic printing.** By default sed reads each line, applies your script, then prints the result. With `-n`, sed reads each line and **prints nothing** unless you explicitly ask it to with the `p` command. That is why `sed -n '/pattern/p'` prints **only the matching lines**: without `-n` it would print every line and print matching lines a second time. Whenever you only want to see a subset (matches, a range, a single line), reach for `-n` first.

How do I escape a slash in the substitute pattern?

**Two ways.** Escape each slash with a backslash: `sed 's/\//SLASH/g'`. Or, much cleaner, **change the delimiter**: sed accepts **any single character** after the `s` as the delimiter. So you can write `sed 's|/|SLASH|g'` using a pipe, or `sed 's#/#SLASH#g'` using a hash. This is the **universal trick** for replacing paths and URLs: `sed 's|/home/old/|/home/new/|g'` is far more readable than a wall of escapes.

What is FS, OFS, NR, NF in awk?

Four built-in variables you will see everywhere: - **FS** (field separator): the character awk uses to split each line into fields. Default is **whitespace**. Set it with `-F','` on the command line or `BEGIN { FS = "," }` inside the program. - **OFS** (output field separator): the character awk uses to join fields when you `print $1, $2`. Default is a **single space**. - **NR** (number of records): the current line number, starting at 1. - **NF** (number of fields): how many fields the current line has. `$NF` is the **last field**. Common pattern: `awk -F',' '{ print $1, $NF }'` prints the first and last column of a CSV.

Why is my awk sum showing 0 or NaN?

**Two usual reasons.** First, your **separator is wrong**: if your file is CSV but you did not pass `-F','`, awk splits on whitespace and `$3` is not what you think. Add `-F','` (or whatever your real separator is) and the column lines up. Second, the column **contains non-numeric prefixes** like a currency symbol: `$100` or `100 USD`. awk coerces leading digits to a number, so `$100` becomes `100`, but `USD 100` becomes `0`. **Fix**: strip the prefix first with `sed`, or use `gsub(/[^0-9.]/, "", $3)` inside the awk program before the sum.

How do I print only the last N lines or only the first line?

For the **first line**: `sed -n '1p' file` or `head -1 file`, both work. For the **first N lines**: `sed -n '1,Np' file` or `head -N file`. For the **last line**: `sed -n '$p' file` (`$` means the last address in sed). For the **last N lines**, sed cannot do it efficiently because it streams forward; **use `tail -N`** instead, which jumps to the end of the file. **Mixing them**: `tail -N | head -M` gives you a slice from the end, very handy for log analysis.

Can I use sed and awk on Windows?

**Yes, three ways.** First, **Windows Subsystem for Linux (WSL)**: a real Ubuntu inside Windows, with the real GNU sed and awk. Best option if you do this often. Second, **Git Bash**: comes with `sed` and `awk` from MSYS2 / busybox, slightly older versions but they work for everyday tasks. Third, **PowerShell equivalents**: `Select-String` is roughly sed-print, `ForEach-Object` plus `-split` is roughly awk. PowerShell is more verbose but native to Windows. **In production**: if you target Windows, write your text pipelines in **Python** or **PowerShell**, they are easier to maintain than chained sed and awk calls on a Windows build agent.

sed / awk Helper - free

sed and awk without the man-page anxiety

sed and awk are two tiny Unix programs that move text around. sed edits files line by line (replace, delete, insert). awk reads tables and does math or filtering on columns. They are on every Mac, every Linux box, every server you SSH into. They do not need installation. They are also famously cryptic: the syntax was designed in 1977 and never updated.

This helper builds the command for you. Pick an action (substitute, delete a line, print columns, sum a column...), fill the inputs, and the right command appears on the right. Below that, a live preview runs the same logic in your browser on sample text you can edit, so you see the result before you copy and paste.

We cover the most useful subset of both tools (substitute, delete, insert / append / change lines, print ranges for sed; print columns, conditional print, count, sum, average, filter, custom program for awk), with a cookbook of 20 ready recipes: delete blank lines, squeeze spaces, sum a column, group by column and count, convert CSV to TSV, and more. Click any recipe to load it into the builder.

Caveat: the preview uses JavaScript regex, which is close to POSIX extended regex but not identical to GNU sed or BSD sed. For everyday substitutions and column work it gives the right answer. For corner cases (look-arounds, locale-aware word boundaries, complex address ranges) consult the FAQ.

How to use it

Pick the tool at the top: sed for whole-line edits (replace, delete, print, insert) and awk for columns (print a column, sum, filter by value).
In sed mode, choose the action (Substitute, Delete, Print only matching, Insert before line N, Append after line N, Replace line N, Print lines N to M). The right inputs appear below.
Type your pattern and replacement for substitute, or a line number for insert / append / change. Flags (g, i, p, Nth) are chips you click on or off.
Edit the sample input on the right to match your real data shape. The output pane updates live: you see exactly what the command does before you copy it.
Switch on In-place edit if you want the file modified directly (`-i`). Pick GNU for Linux or BSD for macOS, the helper shows the right variant (`-i` versus `-i ""`).
In awk mode, pick a field separator (whitespace by default, comma for CSV, semicolon, tab, or a custom character). Then pick an action: print column N, conditional print, count matches, sum, average, filter, or write a custom program.
Open the cookbook at the bottom for 20 ready recipes (10 for sed, 10 for awk). Click any recipe to load it into the builder, edit it, and copy the final command.
Copy the command from the right side. Paste into your terminal with your real filename instead of `file.txt`. The output pane is for verification, not for production data: run the actual command on your file when you trust the result.

When this is useful

Six real moments when sed or awk saves you minutes (or hours):

Bulk-replacing a string across many files. You renamed a variable from `userId` to `accountId` in a project. `sed -i 's/userId/accountId/g' src/**/*.ts` does the entire codebase in one command, faster than your IDE's "find in files" and works over SSH on a server.
Cleaning a log file before analysis. The log has blank lines, debug noise and 50 columns when you only need 3. `awk '/ERROR/ { print $1, $5, $9 }'` keeps only the error rows and only the timestamp, status and message columns. One line, no parsing library.
Converting between CSV and TSV. A tool exports CSV, your other tool wants TSV. `awk -F',' '{$1=$1; print}' OFS='\t'` swaps separators in one go, no Python, no Excel, no temporary files.
Pulling a date range out of a file. You need lines 1000 to 2000 of a 50,000-line log. `sed -n '1000,2000p' file.log` is instantly the answer, and works on a remote server with no extra tools.
Summing a column from a report. The accountant sent a CSV of invoices. You want the total. `awk -F',' '{ sum += $5 } END { print sum }' invoices.csv` adds column 5 across every row in milliseconds.
Renaming files in a folder. You have hundreds of files like `IMG_001.jpg` and want them lower-cased and dated. A loop with `sed` building each new name does the rename without a single line of code: just shell glue.

Questions and answers

Three big ones. First, in-place edit: GNU sed accepts `-i` alone; BSD sed (the default on macOS) requires an argument, even if empty, so the macOS form is `sed -i '' 's/foo/bar/' file`. Forget the empty quotes and macOS sed will use your next argument as a backup extension. Second, flag I (case-insensitive substitute) is GNU only, BSD does not understand it. To stay portable, write `[fF][oO][oO]` instead of `foo/I`. Third, alternation `|` works inside `[...]` differently and extended regex needs `-E` on both, but the anchor and word-boundary syntax still diverges in edge cases. Practical advice: write your scripts to target the platform you are on, and if you need portability, stick to the boring subset (basic substitute with `g`, plain character classes, no lookarounds).

sed and awk without the man-page anxiety

How to use it

Pick the tool at the top: sed for whole-line edits (replace, delete, print, insert) and awk for columns (print a column, sum, filter by value).

In sed mode, choose the action (Substitute, Delete, Print only matching, Insert before line N, Append after line N, Replace line N, Print lines N to M). The right inputs appear below.

Type your pattern and replacement for substitute, or a line number for insert / append / change. Flags (g, i, p, Nth) are chips you click on or off.

Edit the sample input on the right to match your real data shape. The output pane updates live: you see exactly what the command does before you copy it.

Switch on In-place edit if you want the file modified directly (`-i`). Pick GNU for Linux or BSD for macOS, the helper shows the right variant (`-i` versus `-i ""`).

In awk mode, pick a field separator (whitespace by default, comma for CSV, semicolon, tab, or a custom character). Then pick an action: print column N, conditional print, count matches, sum, average, filter, or write a custom program.

Open the cookbook at the bottom for 20 ready recipes (10 for sed, 10 for awk). Click any recipe to load it into the builder, edit it, and copy the final command.

Copy the command from the right side. Paste into your terminal with your real filename instead of `file.txt`. The output pane is for verification, not for production data: run the actual command on your file when you trust the result.

When this is useful

Six real moments when sed or awk saves you minutes (or hours):

Bulk-replacing a string across many files. You renamed a variable from `userId` to `accountId` in a project. `sed -i 's/userId/accountId/g' src/**/*.ts` does the entire codebase in one command, faster than your IDE's "find in files" and works over SSH on a server.
Cleaning a log file before analysis. The log has blank lines, debug noise and 50 columns when you only need 3. `awk '/ERROR/ { print $1, $5, $9 }'` keeps only the error rows and only the timestamp, status and message columns. One line, no parsing library.
Converting between CSV and TSV. A tool exports CSV, your other tool wants TSV. `awk -F',' '{$1=$1; print}' OFS='\t'` swaps separators in one go, no Python, no Excel, no temporary files.
Pulling a date range out of a file. You need lines 1000 to 2000 of a 50,000-line log. `sed -n '1000,2000p' file.log` is instantly the answer, and works on a remote server with no extra tools.
Summing a column from a report. The accountant sent a CSV of invoices. You want the total. `awk -F',' '{ sum += $5 } END { print sum }' invoices.csv` adds column 5 across every row in milliseconds.
Renaming files in a folder. You have hundreds of files like `IMG_001.jpg` and want them lower-cased and dated. A loop with `sed` building each new name does the rename without a single line of code: just shell glue.

Questions and answers

sed / awk Helper

sed and awk without the man-page anxiety

How to use it

When this is useful

Questions and answers

Related tools

Regex tester

Find and Replace

cURL to Code Converter

Whitespace Cleaner

sed / awk Helper

sed and awk without the man-page anxiety

How to use it

When this is useful

Questions and answers

Related tools

Regex tester

Find and Replace

cURL to Code Converter

Whitespace Cleaner