How to search through all files in a folder, recursively...

MJBoa · 09-13-2008, 07:24 PM

So I have a folder with files in it and these have files and folders which have folders with files... I have a complicated directory structure but I want to search through every file ending with a certain extension, they are text files and I want to find a certain word. What's the fastest way to do this through the command line?

Mr. C. · 09-13-2008, 07:25 PM

Code:

find /path/to/folder -name "*.ext" -print0 | xargs -0 grep word

Replace ext and word appropriately.

MJBoa · 09-13-2008, 08:04 PM

find `pwd`/myproject -name "*.py" -print 0 | xargs -0 grep down
find: paths must precede expression
Usage: find [-H] [-L] [-P] [path...] [expression]

down being the word

jschiwal · 09-13-2008, 08:08 PM

Use a dot instead of `pwd`.

MJBoa · 09-13-2008, 08:15 PM

Yeah that's not the problem, as indicated by the error message.

billymayday · 09-13-2008, 08:29 PM

what is `pwd` meant to be?

MJBoa · 09-13-2008, 08:48 PM

pwd = print working directory = the current directory, it was so I didn't have to type it. I'm sure I could have omitted it but I felt like adding it.

billymayday · 09-13-2008, 08:51 PM

Lose the space between print and 0

Mr. C. · 09-13-2008, 09:06 PM

Oops, the typo space in between the -print and the 0 was my fault. Sorry. All fixed. Thanks for the correction billymayday.

MJBoa · 09-13-2008, 09:13 PM

Alright, thanks a lot guys.

Foldarn · 09-13-2008, 10:05 PM

What is the xargs command for? I usually just use:

Code:

find . -name "*.txt" -exec grep {} text_to_search_for \;

and that just returns the stuff I'm looking for.

Mr. C. · 09-14-2008, 12:12 AM

With -exec, you are forking/execing as many processes as items found. This could be 10's or 100's or 1000's, and creates a substantially larger load on the system. Here's some proof that should convince you:

Code:

$ find . -type f | wc   
    4487    4487  152335

$ time find . -type f -exec grep -q fubar {} \;

real    0m11.974s
user    0m3.233s
sys     0m3.913s

$ time find . -type f -print0 | xargs -0  grep -q fubar 

real    0m0.091s
user    0m0.000s
sys     0m0.097s

So, that's almost 12 seconds vs. about .1 second. Your method proves to be 120 times slower.

This is on a relatively unloaded system. With a system actually performing some work, the difference will be much greater.

lwasserm · 09-14-2008, 10:14 AM

.........oops!

Foldarn · 09-14-2008, 11:25 PM

Quote:

Originally Posted by Mr. C.

This is on a relatively unloaded system. With a system actually performing some work, the difference will be much greater.

Oh, snap! Thanks, that's incredible! You learn something new every day... I have a few scripts I need to rewrite using this.

Come to think of it, I hate you! You're making me do more work! *sigh* And to think tomorrow's Monday. Another reason to hate it.