October 30, 2001, 4:59 PM — The grep utility, which allows files to be searched for strings of words, uses a syntax similar to the regular expression syntax of the vi, ex, ed, and sed editors. grep comes in three flavors, grep, fgrep, and egrep, all of which I'll cover in this article.
The name grep is derived from the editor command g/re/p, which literally translates to "globally search for a regular wxpression and print what you find." Regular expressions are at the core of grep, and I'll cover them after a brief description of some of the utility's command options.
The simplest grep command is grep (search pattern) (files list), as in:
grep hello *
The output of this command might be something like this:
$ grep hello * story.txt: so I said hello and she smiled back intro.txt: use the hello.c program as an example of C programming $
grep is case sensitive, so in order to change the search to include "hello," "Hello," or "HELLO," use the -y or -i option. Earlier versions of grep used -y, and later versions use -i. -y is now considered obsolete, although some versions of grep do support both. In the following example, more hellos show up because the search is case independent.
$ grep -i hello *
story.txt: so I said hello and she smiled back
story.txt: I could hear my echo, "HELLO."
intro.txt: use the hello.c program as an example of C programming
hello.c: printf("Hello, world. \n");
$
This command searches all files in the current directory and prints the file name and the line containing the string "hello" for any files that contain that string.
The output of grep varies depending on whether you're searching one or several files. If only one file is named on the command line, the output doesn't include the file name, as in the following example:
$ grep -i hello hello.c
printf("Hello, world. \n");
$
The one-file rule applies whether you use a wild card in your file list or not. If hello.c were the only file in the current directory, using a wild card to locate the file would still produce an unnamed file output. In the following example, the user is searching for any C files containing "hello." There is only one C file in the directory, so the output is identical to the previous example.
$ grep -i hello *.c
printf("Hello, world. \n");
$
I don't know of a grep that has a work-around for this behavior, but you could use the -l option instead, which prints the file name only and not the line containing the string. At least you would know the name of the file that contained the string.
$ grep -il hello *.c hello.c: $
The -l option can be used to extract a list of files containing the string. The file name is printed only once, even though the string may appear in multiple lines within that file. In the following example, story.txt appears only once, even though it contains more than one "hello."
$ grep -il hello * hello.c: intro.txt: story.txt: $
The -l option suppresses most of the other output options from grep. On the other hand, the -n option will print a line number as well as the text, as in the following example:
