Search Multiple Words / String Pattern Using grep Command on Bash shell

Posted on in Categories , , , , , , last updated May 3, 2017

How do I search multiple strings or words using the grep command? For example I’d like to search word1, word2, word3 and so on within /path/to/file. How do I force grep to search multiple words? How can I grep for multiple patterns on Linux, OS X, FreeBSD, or Unix-like system?

The grep command supports regular expression pattern. To search multiple words, use following syntax:

How do I grep for multiple patterns?

The syntax is:

  1. Use single quotes in the pattern: grep 'pattern*' file1 file2
  2. Use extended regular expressions: egrep 'pattern1|pattern2' *.py
  3. Use this syntax on older Unix shells: grep -e pattern1 -e pattern2 *.pl

Here are all other possibilities:
grep 'word1\|word2\|word3' /path/to/file
### Search all text files ###
grep 'word*' *.txt
### Search all python files for 'wordA' or 'wordB' ###
grep 'wordA*'\''wordB' *.py
grep -E 'word1|word2' *.doc
grep -e string1 -e string2 *.pl
egrep "word1|word2" *.c

Examples

In this example, search warning, error, and critical words in a text log file called /var/log/messages, enter:
$ grep 'warning\|error\|critical' /var/log/messages
To just match words, add the -w option:
$ grep -w 'warning\|error\|critical' /var/log/messages
Use the egrep command and you can skip the above syntax to search three words:
$ egrep -w 'warning|error|critical' /var/log/messages
OR
$ grep -e 'warning|error|critical' /var/log/messages
I recommend that you pass the -i (ignore case) and --color option as follows too:
$ egrep -wi --color 'warning|error|critical' /var/log/messages
Sample outputs:

Fig.01: Linux / Unix egrep Command Search Multiple Words Demo Output
Fig.01: Linux / Unix egrep Command Search Multiple Words Demo Output

To search all *.conf files under /etc/, enter:
# egrep -wi --color 'foo|bar' /etc/*.conf
To search recursively (including sub-directories) listed, run:
# egrep -Rwi --color 'foo|bar' /etc/
This entry is 3 of 7 in the Linux / UNIX grep Command Tutorial series. Keep reading the rest of the series:
  1. How To Use grep Command In Linux / UNIX
  2. Regular Expressions In grep
  3. Search Multiple Words / String Pattern Using grep Command
  4. Grep Count Lines If a String / Word Matches
  5. Grep From Files and Display the File Name
  6. How To Find Files by Content Under UNIX
  7. grep command: View Only Configuration File Directives

Posted by: Vivek Gite

The author is the creator of nixCraft and a seasoned sysadmin and a trainer for the Linux operating system/Unix shell scripting. He has worked with global clients and in various industries, including IT, education, defense and space research, and the nonprofit sector. Follow him on Twitter, Facebook, Google+.

Share this on:

104 comment

  3. Thanks for the article. Is there a way to supply the words in a file? So in your example I could create a simple file containing:

    warning
    error
    critical

    I have over 200 words that I need to search – which is tiresome to put onto the command line.

    Thanks,

    Reply Report comment

    1. Fantastic !! I had a scenario where in I had to pick up lines from a log file based on a word. Unfortunately I had 1000 such words and so it would have been difficult to find 1000 lines. But this solution really helped. Thanks.

      Reply Report comment

      1. Hi,

        I have a requirement in which I would need to grep/find a line based on matching 3 different patterns. I know we can grep with -E multiple parameters seprated by pipe but this work Pipe (|) as OR condition. My requirement is I want to use AND condition. It should show line where it satisfy both the parameter.
        e.g. line in a file as —
        10-Aug-2010 Hello, this is a new example for unix.
        I need to show this line only when my grep command matches all 3 words
        this, new, unix How to right such grep command ?

        Reply Report comment

        1. grep -n “$search1” . | grep -n “$search2” | grep -n “$search3”

          after the first grep only statements “search1” will come. this is filtered by the next grep creating and AND condition.

          this is a round about solution coming out the top of my mind. if i find a better solution i’ll post it.

          Reply Report comment

          1. this dint work for me:(

            i used the cmd like grep -il SDI *.sql | grep -il Account *.sql | grep -il Customer *.sql
            My requirement is as follows :

            in the current folder, inside the files with extension .sql, i need to get the files hwich have SDI,Account and Costomer words in it.

            Reply Report comment

            1. -l, –files-with-matches
              Suppress normal output; instead print the name of each input file from which output would normally have been printed. The scanning
              will stop on the first match.

              did -l work for u??

              Reply Report comment

    1. The problem is that you need to ‘escape’ the or operator ‘|’ as shown above AND use the ‘i’ flag to be case insensitive:

      grep -i 'warning\|critical\|error'  /var/adm/messages
      Reply Report comment

  11. I figured I should be more specific in my request.

    I have the following line;

    /usr/bin/dsmc archive -des=”DAILY” -archm=FS_DBARCH_DAILY “/brlog1/BIDW/redo/*”

    I want a grep command to return the line number of this line

    grep archive filename | grep -n redo doesn’t work because it returns the number of the line in the results from the first grep.

    Reply Report comment

  16. “grep -Rl word1 *| xargs grep -l word2… how this work?”

    It just doesn’t work.
    It will return the filename where exists at least one line where the two words are present. Which is no information at all.

    Reply Report comment

  17. @Robert Guest:
    Attach the ‘-n’ parameter to the first grep, not to the second:

    grep -n archive filename | grep redo

    that way the second grep will get a line that contains the line number.

    Reply Report comment

    1. That won’t work if you need to check if word 1 and word2 are anywhere in the file, because the first grep returns the matching line, so the second grep would only match if both words were on the same line.

      Reply Report comment

  21. Hi,

    I want to search multiple string using grep and want to display only matching string.

    can any one pl guide me in this regards.
    eg.
    cat abc.txt |grep -e ‘ab’ -e ‘bc’ -e ‘cd’

    If ‘bc’ is there in the file abc.txt then output should display only ‘bc’ rather than displaying the entire line.

    Thanks & Regards,
    Omprakash

    Google for: grep multiple string and returning matching string

    Reply Report comment

  22. How to search multiple words in separate lines, inside a directory including sub-directory? Pls. give easy example.
    I tried $grep -r “word1” |grep -r “word2” /Folder/subfolder/ > search.log

    Reply Report comment

  24. hi

    I have to grep exact line in file
    for eg : file name test.txt includes
    ram
    5) 1,2,3
    sohan
    5) 6,7,8

    so i want to grep ram and 5 so the output shud be
    ram
    1,2,3

    should not be
    ram
    1,2,3
    6,7,8

    Reply Report comment

    1. First: grep is a single line utility. And because “5)” is on two lines, grep will find them, because it walks down the lines, and matches them to your rules, period.
      Second: what grep finds, that ‘entire’ line will be displayed. So it is not possible to cut the “5)” from the beginning of the line, and display the rest.
      If you want to find “ram” and after that the next “5)”, you need to use some utility that allows you to implement some logic. Like awk.
      However if you can ensure that “5)…” will be after the found line – like “ram” – then you can use the ‘after context’ feature of grep, whic displayes the matching lines PLUS some line after that.

      Reply Report comment

    1. @Kumar and others

      To search for multiple strings in a file try doing this :

      grep -il "String1" "PATH-OF-FILE" | xargs -I % grep -H "String2" % | xargs -I % grep -H "String3" %

      For eg:
      Let’s say I want to search for all those log4j.xml files which have the words CONSOLE and ASYNC in them .. then this is what I would do :

      find "/cygdrive/e/MyDocs/Downloads/work/OATS Domain related/" -iname "log4j*.xml" | xargs -I % grep -ilr "CONSOLE" "%" | xargs -I % grep -H "ASYNC" %

      Cheers!

      Reply Report comment

  26. hi.
    how to search a single line using grep command..
    for example a file having 100 lines in that 100 line only one error line is there.

    how do i retrive that single line using grep command.. i don’t know in which line in that error msg and like that error msg many of the lines in that file.. how do i find using ‘Grep’ command..

    Reply Report comment

  27. echo year, month, Quarter,productTOT, RegionTOT,Per_type,Data_type,Bucket_noTOT,Curr_dateTOT

    from this output, i want only the string that has “TOT”

    so the output,

    productTOT, RegionTOT,Bucket_noTOT,Curr_dateTOT

    could anybody please help me???

    Reply Report comment

  31. Could you please help me regarding while connect putty Linux based logs like collect grep is working

    tail -f test_log | grep ‘\” 50[0234]‘ – working

    tail -f test_log | grep ‘\” 50[0234]‘ | grep “404″ – not working

    Any one help me on this regard? how to collect 1.500 to 504 2.404 alone.

    Reply Report comment

  32. Great info site.
    I have a need to search a file looking for dates and a string, For example:
    (Dec 2 13:25:27 name local5:warn|warning vmdaemon[180412]: #415 Moved volume tape_1 #0055 (12345678) (abc123 from online to offline.). I need to search for a date and the “online to offline” string together. Also, with grep is it possible to do a date range in the search as oppose to a single date?

    Reply Report comment

  34. I also wanted to find all files that contained two separate words. The words could be anywhere in the file.

    One way that I found to do this is:
    grep -l word1 `grep -l word2 *`

    The backtick quotes execute what is inside the quotes and replace it with the results. -l returns file names only.

    More commonly I would also include -ir for case-ignorant recursive search:
    grep -irl word1 `grep -irl word2 *`

    Reply Report comment

  35. Here’s an interesting one that I have been wracking my brain on…

    I have some fairly large data files I need to search for certain conditions. Easy enough because I know the string I want to search for. However, the problem arises when someone duplicates the string in the field that I am searching for which expands the data criteria past my grep. here’s example:

    String = “000/C///” Field delimiter is pipe so I use: grep -i “|000/C///|” to match the condition. Now what I have found some people doing is using multiple strings in that field where only one exists. Is there a way to search for multiples of the same string without having to do a ton of greps in a row?

    Reply Report comment

  36. Trying to do things with the -A -B options –

    To display every line mentioning an astrological element:

    grep “earth|air|fire|water” astro.log

    How do I accomplish this new result with grep – I cannot figure out how to use the options -A and -B more than once within the same command.

    earth and the 2 lines following
    air – only the line containing air
    fire and the 6 lines following
    water and the 3 lines preceding

    This would result in a 15-line capture if the command works.

    Reply Report comment

  38. I am trying to pass the word/pattern from command line which I want to find in array but its not working. Say I want to find Jacob but its not returning anything.

    #! usr/bin/perl
 
use Fcntl; 
print "content-type: text/html nn"; 
 
print "Enter word : ";
$word = ;
 
@myNames = ('Jacob', 'Michael', 'Joshua', 'Matthew', 'Alexander', 'Andrew');
@grepNames = grep(/$word/, @myNames); 
 
foreach $Names(@grepNames)
{
	print $Names;
	print "n";
}
    Reply Report comment

  39. For me it works only if grep is set to use extended regular expressions with the -E flag

    grep -E One\|Two\|”Thirty five”

    Versions
    GNU bash, version 4.1.2(1)-release (x86_64-koji-linux-gnu)
    GNU grep 2.6.3

    Reply Report comment

  41. I have a file “nynorsk-utf-8.txt” in UTF-8 that is supposed to be in Nynorsk, one of the two official dialects of Norwegian. But it contains some “contamination” of Bokmal, the other official dialect. To try to identify the sections of Bokmal, I want to find all lines in nynorsk-utf-8.txt that contain any one of a set of 13 short words that are exclusive to Bokmal:

    ikke|jeg|fra|en|et|de|mye|hun|noen|se|selv|særlig|uke

    I’m on OS X, using egrep (GNU grep) 2.5.1
    I’ve tried the following:

    egrep -in “\” nynorsk-utf_8.txt > out.txt

    using \ to enforce word boundaries. Yet the output contains lines that do not contain any of these 13 words (as stand-alone words). I also tried

    egrep -inw ‘ikke|jeg|fra|en|et|de|mye|hun|noen|se|selv|særlig|uke’ nynorsk-utf_8.txt > outw.txt

    with the exact same results. Then I tried listing the 13 words, one per line, in a separate file ‘wordlist’, and launched

    grep -inw -f wordlist nynorsk-utf_8.txt > outf.txt

    again with the same results. E.g. each output file contains line 16

    16:– Dei fleste meiner at rota til den nye terrorismen er Ã¥ finne i ein tilnærma samfunnsstrukturell stillstand – ein tilstand som pregar bÃ¥de dei rike og fattige arabiske landa.

    that does not contain any of the 13 words (as a stand-alone word).

    Can anyone tell me what I’m doing wrong? or is there, perchance, a big in (e)grep?

    Reply Report comment

  42. Sorry–I’m new to this list, and my first example got corrupted. Let me try to mark it up to display correctly

    egrep -in “\<(ikke|jeg|fra|en|et|de|mye|hun|noen|se|selv|særlig|uke)\>” nynorsk-utf_8.txt > out.txt

    i.e. I used \< and \> as explicit word boundaries. The results were equivalent to launching

    egrep -inw “ikke|jeg|fra|en|et|de|mye|hun|noen|se|selv|særlig|uke” nynorsk-utf_8.txt > out.txt
    Reply Report comment

  43. Hi im facing a prob in matching multiple patterns.

    CODE:

    Variable=”abc* def*” # two patterns abc* and def* are stored in variable.

    I should use variable to find the matching lines. How can i achieve it.

    We use -f option if patterns are in a file. How to use when patterns are in a variable???

    Reply Report comment

  44. Hi All,

    I have two text files
    file one contains many words (single word per line) and file two contains many strings per line .
    Now I want to Grep every line from file2 which contains file1 word.
    Could any one help me out ?

    This is like extracting from file2 matches from file1 !!

    Very appreciated to the one who give solution here !

    Reply Report comment

  45. I want the code to Grep the row from file2 containing file1 strings !!!

    Eg :
    1.file2.txt
    Happy Diwali
    All is well
    One day
    2.file1.txt
    Happy
    One

    Now I want to Grep 1st and 3rd line from the file2.txt by file1.txt

    Appreciated answer !

    Reply Report comment

  48. Best answer would be to use the egrep command to get this worked:-
    egrep ‘(str1|str2|str3)’

    For example:-
    bash-3.00$ cat triana
    Man is the most beautiful creator of god.
    Man is grate among all creators.
    Awsome one…

    Solution:-
    bash-3.00$ egrep ‘(grate|god)’ triana

    Man is the most beautiful creator of god.
    Man is grate among all creators.

    Reply Report comment

  50. FILENAME=’trade.tmp’

    cat $FILENAME | while read line
    do
    #echo “$line”
    Keyword=`cut -d “)” -f 2 $line| cut -d “(” -f 2`
    Keyword_Count=`grep $KewKeyword $FILENAME`
    echo “$Keyword_Count,$Keyword” > trade1.log

    done

    HI ,

    In the above script i want to get the keyword and once i get the keyword I need to get the count of keyword.So i tried with the above script .it is showing the count first keyword only.Can anyone suggest me to resolve the problem.

    Thanks In advance

    Reply Report comment

  52. Thanks for ur reply but I think i Did not tell my problem properly.

    My qustion is :-

    I need to search a pattern /someword*/ in a property file

    here someword can be any thing eg.A-Z or a-z or 0-9
    eg:-
    /goingTo/FollowingPattern*/=Yes
    /WhatASite/FloodsIn*/=No
    /dummyData/FinallyHere*/=QT

    here I have to search for patterns
    /FollowingPattern*/
    /FloodsIn*/
    /FinallyHere*/

    Hence someword is variable in property file.

    Any help is appreciated

    Reply Report comment

  55. I need to calculate the received packets from .tr file.
    Problem is that one string is necessary for me but some unnecessary events are also counted that are not needed.

    So I want a solution.

    line1: r 0.500000000 _1_ RTR — 0 cbr 210 [0 0 0 0] ——- [1:0 5:0 32 0] [0] 0 0
    line 2: r 0.501408175 _3_ RTR — 0 AODV 48 [0 ffffffff 1 800] ——- [1:255 -1:255 30 0] [0x2 1 1 [5 0] [1 4]] (REQUEST)

    I want only line 1 but as I am searching for ‘^r’ only so both files are counting. Pls help me how can I search the line where 2 patterns are needed?

    Reply Report comment

  56. I want to use grep command for multiple patterns in a file and print a number of lines after each match into a new file. Say in the command
    $grep ‘Well\|MultiLayer\|Prediction’ PredResult.prn >List.txt

    – after ‘Well’ match I want to print 100 lines starting with pattern ’01/’,
    – after ‘MultiLayer’ match I do not want to print any line,
    – after ‘Prediction’ match I want to print 100 lines starting with pattern ’01/’.

    Please suggest a workaround.

    Reply Report comment

  61. Maybe stupid question but I can’t understand WHY this one works at all:

    $ grep 'warning\|error\|critical' /var/log/messages

    Shouldn’t the \ mean it will escape the | and treat it as a string value of ‘|’? It works and treats | as the regular expression OR (alternation) function, yes, but it’s driving me crazy. I am thinking I should just egrep everything if I plan to use regular expressions because that usage makes no sense to me and the egrep usage does.

    Reply Report comment

  64. Hi,
    Need help!
    I need to grep “timeout” with sourrending 2 lines. (grep -C 2 “timeout” filename) is not working since using HP-UX.

    Please let me know if there are any other commands for HP-UX.

    Thanks,
    Nagaraj

    Reply Report comment

  65. Is there a way where i can get individual count using multiple strings..
    say i want to get the individual count of each word using
    grep -i warning|error|critical’ /var/log/messages

    Reply Report comment

  71. Enter the command grep ‘if ’ bashrc. This will find every line in the file that contains an if followed by a blank space. How many lines did the statement find? Were any of the lines not actually if statements? (comments start with a #). You can obtain just the number of matches using grep –c. Repeat the command using –c to make sure you counted correctly!

    Reply Report comment

  72. Hi ,

    The below script is not working
    $ grep -e 'warning|error|critical' /var/log/messages

    so I tried to use ‘\|’ instead of ‘|’ alone
    $ grep -e 'warning\|error\|critical' /var/log/messages

    Then I got desired output… so does the above script varies based on OS ?? I am using
    Linux 3.0.101-0.47.93-default

    Reply Report comment

Leave a Comment

Tagged as: Tags , , , , , , , ,