GNU Grep

https://www.gnu.org/software/grep/manual/html_node/
Admin
GNU Grep is a command-line utility that searches input files or standard input for lines matching...
Tokens:22,197
Snippets:233
Trust Score:10
Update:5 days ago
Show doc for...
Tokens:
Raw
### Example script to replace GREP_OPTIONS

Source: https://www.gnu.org/software/grep/manual/html_node/Environment-Variables.html

This shell script is an example of how to set grep options via a script instead of the deprecated GREP_OPTIONS environment variable. It prepends $HOME/bin to PATH and creates an executable script that calls grep with --color=auto and --devices=skip.

```shell
#! /bin/sh
export PATH=/usr/bin
exec grep --color=auto --devices=skip "$@"

```

--------------------------------

### Example of early exit with grep -q

Source: https://www.gnu.org/software/grep/manual/html_node/Performance.html

Using grep -q allows the command to exit immediately after finding a match, which may cause the preceding process in a pipe to exit with a nonzero status.

```shell
sed '/^...$/d' | grep -q X
```

--------------------------------

### Use --label with grep and gzip

Source: https://www.gnu.org/software/grep/manual/html_node/Output-Line-Prefix-Control.html

Example of using the --label option to treat standard input as a file named 'foo' when searching through a gzipped file.

```bash
gzip -cd foo.gz | grep --label=foo -H 'some pattern'
```

--------------------------------

### Shell script using grep -m 1 with a regular file

Source: https://www.gnu.org/software/grep/manual/html_node/General-Output-Control.html

This example shows a while loop that repeatedly runs grep -m 1 'PATTERN' on a regular file, printing 'xxxx' after each match. It works because grep positions the input just after the last selected line, allowing the loop to resume.

```shell
while grep -m 1 'PATTERN'
do
  echo xxxx
done < FILE

```

--------------------------------

### grep with pattern variable and *

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Shows a potential problem when pattern or file names start with '-'. The command may behave unexpectedly.

```shell
grep "$pattern" *

```

--------------------------------

### Shell script using grep -m 1 with a pipe (probably will not work)

Source: https://www.gnu.org/software/grep/manual/html_node/General-Output-Control.html

This example shows the same loop but reading from a pipe instead of a regular file. It probably will not work because a pipe is not a regular file, so grep cannot reposition the input.

```shell
# This probably will not work.
cat FILE |
while grep -m 1 'PATTERN'
do
  echo xxxx
done

```

--------------------------------

### Search standard input and files using '-'

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Uses the special file name '-' to tell grep to read from standard input in addition to the named files. The example pipes the output of cat into grep and also searches /etc/motd.

```shell
cat /etc/passwd | grep 'alain' - /etc/motd

```

--------------------------------

### Example of back-reference performance impact

Source: https://www.gnu.org/software/grep/manual/html_node/Performance.html

This pattern uses back-references which trigger an inefficient backtracking algorithm, resulting in exponential worst-case performance.

```regex
grep -E "^(.*)\1{14}(.*)\2{13}$"
```

--------------------------------

### Using grep -zP with (?s) to match across newlines

Source: https://www.gnu.org/software/grep/manual/html_node/grep-Programs.html

This example shows how to use the -z (--null-data) option with -P (--perl-regexp) and the (?s) directive to make the dot match line breaks. The command prints the input lines 'a' and 'b' as output. Note that with -z and a file containing no NUL byte, grep must read the entire file into memory, which can exhaust memory for large files.

```shell
$ printf 'a\nb\n' |grep -zP '(?s)a.b'
a
b

```

--------------------------------

### Example of cascaded repetition

Source: https://www.gnu.org/software/grep/manual/html_node/Performance.html

This pattern demonstrates a cascaded repetition that can cause stack overflow and is not portable under POSIX.

```regex
grep -E ".*{10,}{10,}{10,}{10,}{10,}"
```

--------------------------------

### Basic grep with -i and pattern 'hello.*world'

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Lists lines in menu.h and main.c containing 'hello' followed by 'world', ignoring case with -i.

```shell
grep -i 'hello.*world' menu.h main.c

```

--------------------------------

### grep -r with --include for C files

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Searches recursively for 'hello' in files matching '*.c' under /home/gigi, similar to the find command.

```shell
grep -r --include='*.c' 'hello' /home/gigi

```

--------------------------------

### grep with -- before pattern and files

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Alternative to -e and './' to handle leading '-'. Note that a file named '-' is misinterpreted as standard input.

```shell
grep -- "$pattern" *

```

--------------------------------

### Bash ANSI-C quoting with ASCII escape

Source: https://www.gnu.org/software/grep/manual/html_node/Matching-Non_002dASCII.html

Shows that using ANSI-C quoting for printable ASCII characters is equivalent to the literal character, which can lead to unintended matching. For example, this matches any line, not just lines containing '^'.

```shell
grep $'\u005E'
```

--------------------------------

### Output of grep 'eli' /etc/passwd /dev/null

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Shows the output of the previous command, including the file name and matching line.

```shell
/etc/passwd:eli:x:2098:1000:Eli Smith:/home/eli:/bin/bash

```

--------------------------------

### grep command line synopsis

Source: https://www.gnu.org/software/grep/manual/html_node/Invoking.html

Shows the general syntax for invoking grep. Options and file arguments are optional; patterns can be provided directly or via -e or -f options. Patterns should typically be quoted in shell commands.

```bash
grep [option]... [patterns] [file]...

```

--------------------------------

### List matching file names with grep -l

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Lists names of test-*.c files in the current directory whose contents mention 'main'.

```shell
grep -l 'main' test-*.c

```

--------------------------------

### Context output with grep -C 2

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Prints two lines of context around each matching line.

```shell
grep -C 2 'hello' test*.log

```

--------------------------------

### Recursive search with grep -r

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Searches for 'hello' in all files under /home/gigi.

```shell
grep -r 'hello' /home/gigi

```

--------------------------------

### Search only C files using find and grep

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Uses find to locate C files and grep -H to search for 'hello' in them. The -exec ... + form passes multiple files to grep.

```shell
find /home/gigi -name '*.c' ! -type d \
  -exec grep -H 'hello' '{}' +

```

--------------------------------

### Word-ending search with grep 'hello\>'

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Searches for words ending in 'hello', matching 'Othello'.

```shell
grep 'hello\>' test*.log

```

--------------------------------

### grep with -e and leading ./ for safety

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Avoids problems with leading '-' in patterns and file names by using -e for the pattern and './' for files.

```shell
grep -e "$pattern" ./*

```

--------------------------------

### Back-reference with no referent

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

This command produces an error because the second '\1' has nothing to refer back to, so it will never match anything.

```shell
echo 'ba' | grep -E '(a)\1|b\1'

```

--------------------------------

### grep with -n, --, and /dev/null for file names

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Outputs line numbers and file names for lines containing 'f' and ending in '.c' in files matching *g*.h. The -- treats later arguments as file names, and /dev/null forces file name output.

```shell
grep -n -- 'f.*\.c$' *g*.h /dev/null

```

--------------------------------

### Whole-word search with grep -w

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Searches for 'hello' as a whole word, not matching 'Othello'.

```shell
grep -w 'hello' test*.log

```

--------------------------------

### Match across lines with -z and -q

Source: https://www.gnu.org/software/grep/manual/html_node/Usage.html

Uses the -z (--null-data) option to treat input as null-terminated lines, allowing newlines to be matched. The -q option suppresses output, which is often combined with -z because a match typically outputs the entire input.

```shell
printf 'foo\nbar\n' | grep -z -q 'foo[[:space:]]\+bar'

```
Tokens:
Raw