Delete all lines containing a string from a file in Bash

David Y.

The Problem

How can I delete all lines containing a particular string from a text file using Bash?

The Solution

We can do this using sed. For example, the following command will delete all lines containing the text “NOT FOR RELEASE” from the file info.txt:

sed -i.bak '/NOT FOR RELEASE/d' ./info.txt
Click to Copy

sed supports a large number of commands for editing text files. Here, we use the d command, which deletes lines. For example, 1d would delete the first line in the file, and 2d would delete the second line. If we provide a regular expression between /s instead of a line number, sed will delete every line that matches the expression.

The -i flag indicates that the change should be made in place. .bak is the suffix to use when creating a backup file. Before making any changes, sed will create a copy of info.txt in its original state called info.txt.bak. If we’re happy with the modified state of info.txt, we can delete this file. Otherwise, we can delete info.txt and rename info.txt.bak to info.txt to return to our initial state.

Remember that the text between /s is a regular expression, and will thus interpret certain characters as part of that syntax. If we merely want to search for a literal string, we will need to escape the characters $.*/[\]^ by preceding them with a backslash (\). For example:

sed -i.bak '/NOT FOR RELEASE\./d' ./info.txt # will delete lines containing "NOT FOR RELEASE."
Click to Copy

To learn more about sed, consult its manual page by typing man sed into your terminal.