Search String: Find and Grep

Question

There must be a better / shorter way to do this:

# Find files that contain <string-to-find> in current directory
#   (including sub directories) 
$ find . | xargs grep <string-to-find>

Also, to search only e.g. HTML files:

 # find . | grep html$ | xargs grep <string-to-find>

Thanks beforehand!

Answer 1

find . -name \*.html

or, if you want to find files with names matching a regular expression:

find . -regex filename-regex.\*\.html 

or, if you want to search for a regular expression in files with names matching a regular expression

find . -regex filename-regex.\*\.html -exec grep -H string-to-find {} \;

The grep argument -H outputs the name of the file, if that's of interest. If not, you can safely remove it and simply use grep. This will instruct find to execute grep string-to-find filename for each file name it finds, thus avoiding the possibility of the list of arguments being too long, and the need for find to finish executing before it can pass its results to xargs.

---

To address your examples:

find . | xargs grep <string-to-find>

could be replaced with

find . -exec grep -H string-to-find {} \;

and

find . | grep html$ | xargs grep <string-to-find>

could be replaced with

find . -name \*.html -exec grep -H string-to-find {} \;

Answer 2

Not sure what do you mean by better, my first thought was something like this:

grep <string-to-find> $(find -regex .*\.html)

But that's worse because result of the find would be accumulated somewhere in shells memory and then sent as a huge chunk of input arguments

The only imporvement I see too your suggestion is

find -regex .*\.html | xargs grep <string-to-find>

That way find performs all the filtering and you still retain piped processing

Answer 3

If this is going to be a common search utility you're going to utilize, you may want to take a look at ack, which combines both the find and the grep together into this functionality that you're looking for. It has fewer features than grep, though 99% of my searches are suited perfectly by replacing all instances of grep with ack.

Besides the other answers given, I also suggest this construct:

find . -type f -name "*.html" -print|xargs -I FILENAME grep "< string-to-find>" FILENAME

Even better, if the filenames have spaces in them, you can either quote "FILENAME" or pass a null-terminated (instead of newline-terminated) result from find to xargs, and then have xargs strip those out itself:

find . -type f -name "*.html" -print0|xargs -0 -I FILENAME grep "< string-to-find>" FILENAME
                             here --^ and --^

Here, the name FILENAME can actually be anything, but it needs to match both

find . -type f -name "*.html" -print0|xargs -0 -I FILENAME grep "< string-to-find>" FILENAME
                                           here --^                           and --^

Like this:

find . -type f -name "*.html" -print0|xargs -0 -I GRRRR grep "< string-to-find>" GRRR
                                           this --^                       this --^

It's essentially doing the same thing as the {} used within the find statement itself to state "the line of text that this returned". Otherwise, xargs just tacks the results of find to the END of all the rest of the commands you give it (which doesn't help much if you want grep to search inside a file, which is usually specified last on the command-line).