Shell Programming and Scripting

hey,

i want to use tr to remove all the lowercase characters from a string and replace them with a '.', i can't seem to get it to work, heres an example of what im trying, someone please correct me

: tr [a-z] .

Use sed instead...

Code:

$ echo 123abcABC456aaaaAAAAAZxCvBnM | sed 's/[a-z]/./g'
123...ABC456....AAAAAZ.C.B.M
$

thanks for that mate, that was giving me retard attacks

I am not sure why you were having problem with your tr statement.

It looks perfectly ok to me, have a look

----------------------------------------------------
$ echo AmiT|tr [a-z] .
A..T
----------------------------------------------------

and this is what you were doing. Right ??

Thanks,
Amit

If the current directory contains file(s) with a name formed by a signle lowercase character the shell will substitute [a-z] by the file list.

Code:

tr '[a-z]' .

Jean-Pierre.

Yeah, but even with those quotes, there are still problems. tr varies quite a bit from system to system. And, as per tradition, the orignal poster provided no clues about his version of unix. According to the Posix Standard So if the tr program is following the standard, you will convert [ and ] into periods for sure. And if the locale is set to Posix, the lower case letters will also be converted into periods as well. There are still some non-standard version of tr that require '[a-z]' for a range and treat 'a-z' as 3 characters. Among the tr programs that accept a-z, there are two behaviors for '[a-z]':
[a-z] == a-z
[a-z] == ]a-z[
with the second being what the standard requires. With a non-standard locale, the author of the locale determines what a-z means. The confusion over a-z verses [a-z] arises from the fact that BSD used a-z and System V used [a-z]. Posix had to pick one and the went with BSD. Apparently not everyone was happy with choice and compliance with this requirement is not universal.

Now that we have discussed the difficulty with the first parameter, let's look at the second parameter. We have a second System V/BSD clash and this time Posix wimped out. Whatever the first parameter means it is more than one character but the second parameter is guaranteed to be only one character. So the strings are not equal in length. According to the standard: So, to do exactly what the author intended,
tr '[a-z]' .
requires both the System V behavior or ignoring (or requiring) brackets around a range (a Posix violation) and BSD's "extend the last character of a shorter string" behavior (not guaranteed by Posix). A more correct solution is:
tr '[:lower:]' '[.*]'
but now you probably see why I prefer sed for something like this...

Good thing this question was "ultra simple"!

hey, well i didn't expect my post to ellicit this much discussion i was just tearing my hair out because i thought that my command:

should have been converting all of the lowercase chars to . What was happening which was confusing me so much, was that it was only converting the first lowercase char in the string, so it seemed like i was missing some obvious command, i tried all combinations of quotes etc and even misinformedly tried putting brackets around the . -> [.] :P the version of unix i was working on was solaris and the bourne shell, don't know if that helps explain anything, but im grateful for the help and have the solution already in place thanks again