How to find files with specific text inside and specific extension?

Question

In UNIX: in folders there are files with .sas (programs) extension and files with .sas7bdat (tables). I need to find programs (and NOT tables) which contain the text "liasse". I tried by launching this command:

grep -rli liasse *.sas ./

In the result I can see a table name. What should be the command to retrieve only programs?

Answer 1

This is a POSIX-compliant command that launches grep for many files at once:

find . -type f -name '*.sas' -exec grep -li liasse {} +

---

• POSIX 1003.1 - man page for find
• POSIX 1003.1 - man page for grep

---

Your command grep -rli liasse *.sas ./ doesn't work as you expected. Let's analyze what happens:

1. *.sas undergoes shell globbing.
   1. If there is at least one object in the current directory that matches *.sas, all of them are placed after liasse.
   2. If there's no such object, grep may get literal *.sas as a path to be inspected (it depends; investigate e.g. shopt -s nullglob, shopt -s failglob in bash) and throw warning.
2. Either way grep does get ./ and process it recursively in a search for liasse inside files. This processes all files under ./: programs, tables, whatever.

That's why you could get a table name in your result.

---

My command find . -type f -name '*.sas' -exec grep -li liasse {} + runs grep only for files that match *.sas pattern. Important things:

• -type f chooses files only. This way I prevent directory names matching *.sas (if any) from being passed to grep (grep without -r should reject them anyway but it would be less elegant).
• Quoting the pattern prevents shell from globbing; find gets literal *.sas as an argument to the -name operand; it knows how to interpret patterns like this.
• find … -exec … {} + syntax substitutes multiple objects in place of {}. This way fewer (maybe just one) grep processes are created, compared to find … -exec … {} \;.

Answer 2

You can use find command for recursively finding all files by name, then run grep on each file. The simplest solution is to the -exec option of find:

find . -name '*.sas' -exec grep -li liasse {} \;

Alternatively, you can combine find with xargs:

find . -name '*.sas' -print0 | xargs -0 grep -li liasse

This has slightly better performace, because it executes grep for a large batch of files, not for every single file.

Using the -P option of xargs, you can even run multiple grep invocations in parallel.

Answer 3

This can be achieved with

find . -type f -name "*.sas" -print0 | "xargs" -0 -e grep -liH -e liasse

If you have no xargs implementation with -0 option, which is about communicating a set of paths between find and xargs, you might try

 find . -type f -name "*.sas" -print | "xargs" -e grep -liH -e liasse