Take the 2-minute tour ×
Ask Ubuntu is a question and answer site for Ubuntu users and developers. It's 100% free, no registration required.

I want to find all files which contain a specific string of text. The grep command works, but I don't know how to use it for every directory (I can only do it for my current directory). I tried reading man grep, but it didn't yield any help.

share|improve this question
    
grep -RIn <yor pattern> * Will search from current directories down in all text files. Not sure how to do my search recursively in file patterns like *.C with only grep –  user311346 Aug 1 '14 at 17:13
    
Wildcard with --include="*.C" option, @user311346, thanks to @Lekensteyn. –  BobStein-VisiBone Feb 6 at 20:36

9 Answers 9

up vote 132 down vote accepted

It would be better to use

grep -rl "string" /path

where

  • -r (or --recursive) option is used to traverse also all sub-directories of /path, whereas
  • -l (or --files-with-matches) option is used to only print filenames of matching files, and not the matching lines (this could also improve the speed, given that grep stop reading a file at first match with this option).
share|improve this answer
6  
Actually if "string" is a text pattern to find, it's better to use that functionality, otherwise someone can face problems when the string contains dot or special character which has meaning in regular expressions and not just a dot which should be found as a string, as-is. Then I would use -rlF switches, -F for "fixed string" (and not regexp - for example). Of course, if the task was using regexps, then excuse me. Sure, the same theory without -r too, I often see that people assumes grep searches "text" and it can cause problems which special ones which mean something as a regexp. –  LGB Aug 1 '11 at 12:19
    
There is also the -i flag which ignores case. –  Marco Ceppi Aug 1 '11 at 13:15
    
I would only to show the --recursive option, there are ton of options and usage scenarios one can talk about. I started from the @dmityugov accepted answer and modified to work without find. –  enzotib Aug 1 '11 at 13:42
    
I would appreciate if you include the meaning of the command in your answer, e.g. the meaning of the options -rl. –  N.N. Aug 8 '11 at 15:54
1  
@N.N.: done :-) –  enzotib Aug 8 '11 at 16:04

If you're looking for lines matching in files, my favorite command is:

grep -Hrn 'search term' path/to/files
  • -H causes the filename to be printed (implied when multiple files are searched)
  • -r does a recursive search
  • -n causes the line number to be printed

path/to/files can be . to search in the current directory

Further options that I find very useful:

  • -I ignore binary files (complement: -a treat all files as text)
  • -F treat search term as a literal, not a regular expression
  • -i do a case-insensitive search
  • --color=always to force colors even when piping through less. To make less support colors, you need to use the -r option:

    grep -Hrn search . | less -r
    
  • --exclude-dir=dir useful for excluding directories like .svn and .git.

Example output

share|improve this answer
9  
-H on a folder is redundant, if there is more then one file, as is probable. In fact, the man page says -H, --with-filename: Print the file name for each match. This is the default when there is more than one file to search. –  enzotib Aug 1 '11 at 12:31
    
I did not know that, it always worked like I expected. It's my default command when looking for files. –  Lekensteyn Aug 1 '11 at 12:34
    
Is there a way to only consider files with a, say, .a extention (and to combine this with -r)? –  user2413 Nov 4 '12 at 22:54
3  
@user2413 Try --include '*.*' –  Lekensteyn Nov 5 '12 at 15:24
    
@enzotib Just a FYI: I don't get the nice clean output of every line in every file where my text appears with -rln nor -rl, but I do with -Hrn. Redundant or not, it looks like the screenshot above with the -Hrn options. Also, it doesn't seem like -n works without -H. –  SgtPooki Mar 25 '14 at 21:59

I believe you can use something like this:

find /path -type f -exec grep -l "string" {} \;

Explanation from comments

find is a command that lets you find files and other objects like directories and links in subdirectories of a given path. If you don't specify a mask that filesnames should meet, it enumerates all directory objects.

  • -type f specifies that it should proceed only files, not directories etc.
  • -exec grep specifies that for every found file, it should run grep command, passing its filename as an argument to it, by replacing {} with the filename
share|improve this answer
1  
Just for those who don't know, adding -name '*.py' restricts the matches to files ending in '.py'. –  Daniel F Mar 23 '14 at 1:10
    
I like that this covers clients that don't have -R implemented in their grep command. –  Aviose Aug 11 '14 at 20:29

If you’re willing to try something new, give ack a shot. The command to recursively search the current directory for string is:

ack string

Installation is quite simple:

curl http://betterthangrep.com/ack-standalone > ~/bin/ack && chmod 0755 !#:3

(Provided you’ve already got the directory ~/bin and it’s preferably in your PATH.)

share|improve this answer
2  
Or just apt-get install ack-grep (, and add alias ack=ack-grep to your .bashrc) –  markijbema Aug 1 '11 at 20:37
    
What do the last parameters to the chmod command do? Are they specific for chmod or are they bash related (the !#:3 part)? –  Elliott Darfink Jan 21 '14 at 10:06
    
@ElliottDarfink That’s using Bash’s history feature! is an event designator. They’re pretty powerful to avoid repetitions. !#:3 references the third token of the command line so far, i.e. ~/bin/ack in this case. –  Konrad Rudolph Jan 21 '14 at 16:59

My default command is

grep -Rin string *

I use a capitol 'R' because ls uses it for recursive. Since grep accepts both, no reason to not use it.

EDIT: per HVNSweeting, apparently -R will follow symlinks where as -r will not.

share|improve this answer
1  
To search in hidden files too, run shopt -s dotglob (remember -s as "set"). Be careful when removing files then. If you've dotglob enabled, rm -r * removes everything in the current dir, but also the directory above it because .. matches. To disable dotglob, use shopt -u dotglob ("unset"). The changes are temporary though, it only applies to the current shell. –  Lekensteyn Aug 1 '11 at 19:09
    
I forgot about this. Is there a way to set it for one the one line? something like shopt -s dotglob & <grep cmd> & shopt -y dotglob only more convenient? This way we won't have to worry about resetting it –  user606723 Aug 1 '11 at 19:42
    
Also, it's probably easier to use grep -Rin string . in most of these cases. I just use * because it seems to come more naturally. –  user606723 Aug 1 '11 at 19:44
    
shops -s dotglob;grep -Rin string *;shopt -u dotglob works. –  Lekensteyn Aug 1 '11 at 19:56
    
yeah... but.. it'd be nice if there was a command that only worked for the single line. instead of turning it on and off in the same line, whjich is silly –  user606723 Aug 1 '11 at 20:13

The command rgrep is dedicated for such need

If not available, you can get it like this

mkdir -p ~/bin
cd ~/bin
wget http://sdjf.esmartdesign.com/files/rgrep
chmod +x rgrep

You can directly set into your default grep options as described above.

I personnaly use

[[  ${#args} -lt 5 && "${args//[[:space:]]/}" == "-i" ]] && args="-Hin"
args="${args:--Hns} --color=auto"

related topic : how to always use rgrep with color

share|improve this answer
    
rgrep is provided by the grep package which is installed by default in Ubuntu. –  karel Jan 2 at 14:16

Update 2:

This line of commands using find and grep fixes the problem:

$ find path_to_search_in -type f -exec grep -in searchString {} 2> /dev/null +

--color=<always or auto> for colored output:

$ find path_to_search_in -type f \
            -exec grep --color=always -in searchString {} 2>/dev/null +

Example:

$ find /tmp/test/ -type f -exec grep --color=auto -in "Search string" {} 2>/dev/null +

An example run in the snapshot below: snap1


Update 1:

You can try following code; as a function in your .bashrc or .bash_aliases or in a script:

wherein () 
{ 
    for i in $(find "$1" -type f 2> /dev/null);
    do
        if grep --color=auto -i "$2" "$i" 2> /dev/null; then
            echo -e "\033[0;32mFound in: $i \033[0m\n";
        fi;
    done
}

Usage: wherein /path/to/search/in/ searchkeyword

example:

$ wherein ~/Documents/ "hello world"

(Note: As suggested in the comments below by @enzotib, this doesn't work with file/directories including spaces in their names.)


Original post

To search for the string and output just that line with the search string:

$ for i in $(find /path/of/target/directory -type f); do \
    grep -i "the string to look for" "$i"; done

e.g.:

$ for i in $(find /usr/share/applications -type f); \
    do grep -i "web browser" "$i"; done

To display filename containing the search string:

$ for i in $(find /path/of/target/directory -type f); do \
    if grep -i "the string to look for" "$i" > /dev/null; then echo "$i"; fi; done;

e.g.:

$ for i in $(find /usr/share/applications -type f); \
    do if grep -i "web browser" "$i" > /dev/null; then echo "$i"; \
    fi; done;
share|improve this answer
    
Fails on filenames containing spaces. The failing is hided by the fact that stderr is not shown. –  enzotib Jan 13 at 9:38
    
@enzotib thanks for pointing that out.. that's still unresolved for the function specified.. I've added another one-liner though.. –  precise Jan 13 at 20:38
    
Now the answer is similar to that of @dmityugov. –  enzotib Jan 13 at 21:28
    
yes, but in that sense most of the answers in this page if you check are similar in respect that they use grep, next to that being the subset using find with grep... but if you are to accept different switches and tweaks as a distinct answer, probably mine will fit in here too.. or do you differ? the last update does what I'd like in my search: file names with lines with search key and line no. too :) and colored output and error filter for better readability.. –  precise Jan 14 at 6:44

I do this using xargs, a very underrated command

find ./ -type f -print0 | xargs -0 grep 'string_you_are_looking_for'

find ./ gives you a recursive list of all the files in a current folder then you pipe it to xargs that executes the grep command on each one of those files

share|improve this answer
4  
Using xargs without -print0 option to find and -0 option to xargs is deprecated, it will fail on filenames containing spaces. –  enzotib Aug 1 '11 at 15:45
    
@enzotib I have edited the answer as you suggested.- please review and if needs to editing and correcting, I will happy re-editing by you. thank you –  KasiyA Jan 13 at 9:00
1  
@KasiyA: it is ok now, removed my downvote. –  enzotib Jan 13 at 9:32

I know there are plenty of answers here, but here's an alternative if you'd like to add other restrictions when searching the files:

find . -type f -exec grep --quiet string_to_look_for {} ';' -print

This works because grep will return 0 if it found a result, 1 otherwise. For example you can find files 1 MB large and containing something:

find . -type f -exec grep --quiet string_to_look_for {} ';' -size 1M -print

For multiple requirements you probably want to use the optimizer flag -O that exists in GNU grep.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.