I've been trying to find a way to filter a line that has the word "lemon" and "rice" in it. I know how to find "lemon" or "rice" but not the two of them. They don't need to be next to the other, just one the same line of text.
7 Answers
"Both on the same line" means "'rice' followed by random characters followed by 'lemon' or the other way around".
In regex that is rice.*lemon
or lemon.*rice
. You can combine that using a |
:
grep -E 'rice.*lemon|lemon.*rice' some_file
If you want to use normal regex instead of extended ones (-E
) you need a backslash before the |
:
grep 'rice.*lemon\|lemon.*rice' some_file
For more words that quickly gets a bit lengthy and it's usually easier to use multiple calls of grep
, for example:
grep rice some_file | grep lemon | grep chicken
-
Your last line is a conjunction not disjunction no? To wit: the
grep rice
finds lines containingrice
. It is fed intogrep lemon
which will only find lines containing lemon .. and so on. Whereas the OP - as well as your prior answers - are allowing any of [rice|lemon|chicken] Jan 4, 2017 at 22:16 -
-
@Florian Diesch - Mind explaining why
|
needs to be escaped ingrep
? Thanks!– fugitiveFeb 3, 2017 at 21:25 -
1@fugitive
egrep
uses extended regex where|
is understood as OR logic.grep
defaults to basic regex, where\|
is OR Jul 24, 2017 at 15:06 -
1As stated in
grep
's manpage,egrep
is deprecated and should be replaced bygrep -E
. I took the freedom to edit the answer accordingly.– dessertSep 27, 2017 at 17:19
You can pipe the output of first grep command to another grep command and that would match both the patterns. So, you can do something like:
grep <first_pattern> <file_name> | grep <second_pattern>
or,
cat <file_name> | grep <first_pattern> | grep <second_pattern>
Example:
Let's add some contents to our file:
$ echo "This line contains lemon." > test_grep.txt
$ echo "This line contains rice." >> test_grep.txt
$ echo "This line contains both lemon and rice." >> test_grep.txt
$ echo "This line doesn't contain any of them." >> test_grep.txt
$ echo "This line also contains both rice and lemon." >> test_grep.txt
What does the file contain:
$ cat test_grep.txt
This line contains lemon.
This line contains rice.
This line contains both lemon and rice.
This line doesn't contain any of them.
This line also contains both rice and lemon.
Now, let's grep what we want:
$ grep rice test_grep.txt | grep lemon
This line contains both lemon and rice.
This line also contains both rice and lemon.
We only get the lines where both the patterns match. You can extend this and pipe the output to another grep command for further "AND" matches.
Though the question asks for 'grep', I thought it might be helpful to post a simple 'awk' solution:
awk '/lemon/ && /rice/'
This can easily be extended with more words, or other boolean expressions besides 'and'.
Another idea to finding the matches in any order is using:
grep with -P
(Perl-Compatibility) option and positive lookahead regex (?=(regex))
:
grep -P '(?=.*?lemon)(?=.*?rice)' infile
or you can use below, instead:
grep -P '(?=.*?rice)(?=.*?lemon)' infile
- The
.*?
means matching any characters.
that occurrences zero or more times*
while they are optional followed by a pattern(rice
orlemon
). The?
makes everything optional before it (means zero or one time of everything matched.*
)
(?=pattern)
: Positive Lookahead: The positive lookahead construct is a pair of parentheses, with the opening parenthesis followed by a question mark and an equals sign.
So this will return all lines with contains both lemon
and rice
in random order. Also this will avoid of using |
s and doubled grep
s.
External links:
Advanced Grep Topics
Positive Lookahead – GREP for Designers
This command returns matches for a line which has either foo or goo.
grep -e foo -e goo
This command return matches for lines which have both foo and goo in any order.
grep -e foo.*goo -e goo.*foo
-
This is mostly good, but also returns lines with only "foo" or "goo" in addition to "foo" and "goo".– BrianOct 3, 2022 at 11:42
-
Indeed. "either foo or goo" And, upon rereading, I see that the OP said ~"I know how to do OR I want to know how to do AND". I will update the answer.– netskinkOct 3, 2022 at 13:51
If we admit that providing an answer that is not grep
based is acceptable, like the above answer based on awk
, I would propose a simple perl
line like:
$ perl -ne 'print if /lemon/ and /rice/' my_text_file
The search can be ignoring case with some/all of the words like /lemon/i and /rice/i
.
On most Unix/Linux machines perl is installed as well as awk anyway.
Here's a script to automate the grep piping solution:
#!/bin/bash
# Use filename if provided as environment variable, or "foo" as default
filename=${filename-foo}
grepand () {
# disable word splitting and globbing
IFS=
set -f
if [[ -n $1 ]]
then
grep -i "$1" ${filename} | filename="" grepand "${@:2}"
else
# If there are no arguments, assume last command in pipe and print everything
cat
fi
}
grepand "$@"
-
1This should probably be implemented using a recursive function, instead of building a command string and
eval
ing it, which breaks easily– muruFeb 3, 2017 at 1:42 -
-
1Editing it do that will be too much of a rewrite, so I won't do that. If you want to add it, here's what I imagine it should look like: paste.ubuntu.com/23915379– muruFeb 3, 2017 at 2:36