grep command: find text inside files

Syntax

grep [options] pattern file...

Reads a file line by line and prints only the lines that match. Lines that do not match are simply not shown.

Try it first

The terminal on this page has a log file and some source code.

$ grep ERROR app.log
ERROR connection refused
ERROR timeout

Out of five lines, the two containing ERROR came back. That is the whole idea.

Showing line numbers

-n tells you where in the file each match is.

$ grep -n ERROR app.log
3:ERROR connection refused
5:ERROR timeout

Those numbers are the ones your editor uses, so you can jump straight there.

Ignoring capitalisation

-i matches regardless of case.

$ grep -i todo README.md
Todo: write the setup guide.

Useful whenever a word is written inconsistently.

Searching a whole directory

-r walks everything below a directory, and each match tells you which file it came from.

$ grep -r TODO .
./src/index.js:// TODO: handle errors

Options combine.

$ grep -ri todo src
src/index.js:// TODO: handle errors
src/utils.js:// todo: add tests

Printing what does not match

-v inverts the result and prints the lines that did not match.

$ grep -v INFO app.log
WARN  slow query
ERROR connection refused
ERROR timeout

Strip the ordinary lines out of a log and what remains is the interesting part.

When you actually reach for it

grep is how you pull the handful of relevant lines out of a mountain of them.

-c gives you the count on its own.

$ grep -c ERROR app.log
2

Things that trip people up

No output means no match, not a failure.

$ grep nope app.log

Before assuming a typo, check the capitalisation (-i) and whether you are searching the right place.

Quote patterns that contain spaces. grep connection refused app.log treats refused as a filename. grep "connection refused" app.log keeps it as one pattern.

It searches contents, not names. When you want to find files by name, that is find.