Brackets allow you to specify a single character from a group . For example, if you wanted any single vowel, you can use [aeiou].
$ ls /usr/bin | grep 'b[aeiou]t'
batch bitesize.d smbutil
To negate all characters within brackets, precede the characters within the brackets with a caret (^)
$ ls /usr/bin | grep 'b[^aeiou]t'
This would specify some text pattern that has a single character that is not [aeiou] between b and t.
We can specify a range if we want a range of characters or numbers.
$ ls /usr/bin | grep '[a-d][e-g][h-l]' afhash afida afinfo cancel git-receive-pack ldapdelete mdfind snmpdelta
With this command, we selected words that contains a first letter from a, b, c, or d, a second letter from e, f, or g, and a third letter from h, i, j, k, or l. Notice how this sequence of three letters can appear anywhere in the word.
A severe downside to using the - metacharacter for range is that it's not portable due to different character collation orders. To explain this, we need to learn a bit of history.
Unix was first developed with just ASCII characters. These were the canonical English characters which had order from 0 to 127, including characters such as control codes, printable characters, and upper/lowercase letters with numbers and punctuation marks. For letters, we had an ordering for characters like:
As other countries began adopting Unix, they had to make room for more characters. They had to include special characters such as an e with an accent over it, or a c with a squiggly line beneath. Thus, some collations arose with an ordering like this:
You could probably imagine the problem already. An expression such as A-Z would capture all uppercase letters in the first example, but all letters except a in the second.
Thus, try not to use the range character too much. You can instead rely on Character Classes (below), which are POSIX standard.
To check your current locale, print the
$ echo $LANG en_US.UTF-8
Because of the discrepancies in collation ordering, Unix provides several character classes in order to make shell scripts more portable. Here is a list of the character classes:
When using character classes you must place them within brackets.
$ ls /usr/bin/ | grep '[[:digit:]][[:alpha:]][[:digit:]]' a2p5.16 a2p5.18 s2p5.16 s2p5.18
This matched any files that had a sequence containing a digit, an alphabet character, followed by another digit.
When metacharacters are placed within brackets, they lose their special meaning.
The following code would match any listings with a minus symbol (-), a period (.) or the letter x.
$ ls /usr/bin/ | grep '[-.x]'
... weblatency.d wish8.4 wish8.5 xar xargs xattr xattr-2.6 xattr-2.7 xcode-select ...
If you want to specify the bracket (]) or the minus symbol (-), place them first in the list.
In some languages, two letters in sequence may identify itself as a one unit.
For example, if we were to consider the characters 'ts' as one unit, we could do so by placing them in brackets and periods [.ts.].
Furthermore, we can specify characters that have some variations such as an accent mark or tilde. By having the expression [=a=], we can specify all variations of the letter a. This includes à, á, â, and ã.
Ever feel achy from sitting crunched up on your computer table? Try lying down with these optical glasses that allow you to work on your laptop while lying flat on your back. This is the perfect solution with those with limited mobility or those who wish to prevent neck cramps and back strains.$ Check price
In this completely revised second edition of the perennial best seller How Linux Works, author Brian Ward makes the concepts behind Linux internals accessible to anyone curious about the inner workings of the operating system. Inside, you'll find the kind of knowledge that normally comes from years of experience doing things the hard way.$ Check price