purge a string for "whitespace and dashes"

matt007 · 12-11-2009, 04:37 PM

So is there a command i can use to remove dashes ("-") and whitespaces from a string and get new string without the whitespace and dashes...

Komakino · 12-11-2009, 04:38 PM

In which language?

pixellany · 12-11-2009, 04:46 PM

Sed, tr

ghostdog74 · 12-11-2009, 07:06 PM

using bash shell, no need external commands

Code:

$ string="my string with spaces and - dashes - dashes"
$ echo ${string//[ -]/}
mystringwithspacesanddashesdashes

syg00 · 12-11-2009, 07:28 PM

The request was to remove whitespace - better to use a class rather than a single space character.

ghostdog74 · 12-11-2009, 07:33 PM

sure thing, if that's what OP want, i refer him/her to bash reference: "3.5.8.1 Pattern Matching"
or just set it first

Code:

set -- $string
echo ${string//[ -]/}

catkin · 12-12-2009, 05:19 AM

Quote:

Originally Posted by ghostdog74

sure thing, if that's what OP want, i refer him/her to bash reference: "3.5.8.1 Pattern Matching"
or just set it first

Code:

set -- $string
echo ${string//[ -]/}

Ah! Bash pattern matching -- one of its most under-used features, perhaps because it is so un-intuitive and requires understanding filename expansion too.

But what does set -- $ string do that's useful in this case? AFAIK it just parses $string into $IFS-separated words and assigns them to $1, $2 ... ???

The requirement to remove whitespace and - can be met by including $IFS in the character range to be substituted

Code:

c:~$ x=$'a\t \n-b'
c:~$ echo "'${x//[$IFS-]}'"
'ab'

ghostdog74 · 12-12-2009, 06:16 AM

@catkin, what i meant is unsetting of IFS to become default...because by default IFS is whitespace, (ie space, tabs, newline). So setting it first using the default IFS make the string all become

Code:

$ s=$(printf "%s\t\t%s %s    %s" one two three four)
$ echo $s | od -c
0000000   o   n   e  \t  \t   t   w   o       t   h   r   e   e
0000020           f   o   u   r  \n
0000027

$ unset IFS
$ echo $s
one two three four

so the use of set -- $string is redundant

catkin · 12-12-2009, 07:24 AM

Quote:

Originally Posted by ghostdog74

@catkin, what i meant is unsetting of IFS to become default...because by default IFS is whitespace, (ie space, tabs, newline). So setting it first using the default IFS make the string all become

Code:

$ s=$(printf "%s\t\t%s %s    %s" one two three four)
$ echo $s | od -c
0000000   o   n   e  \t  \t   t   w   o       t   h   r   e   e
0000020           f   o   u   r  \n
0000027

$ unset IFS
$ echo $s
one two three four

so the use of set -- $string is redundant

Thanks ghostdog74

but I'm being dense and don't understand.

AIUI set -- $string does not unset IFS and unsettting IFS does not not set it to the default value; what it does do is make bash behave as if it had the default value.

Code:

c:~$ echo -n "$IFS" | od -c -to1
0000000      \t  \n
        040 011 012
0000003
c:~$ x=$'a\t \n-b'
c:~$ echo "'${x//[$IFS-]}'"
'ab'
c:~$ unset IFS
c:~$ echo "'${x//[$IFS-]}'"
'a
b'

ghostdog74 · 12-12-2009, 07:37 AM

Quote:

Originally Posted by catkin

Thanks ghostdog74

but I'm being dense and don't understand.
AIUI set -- $string does not unset IFS

first, yes you are right. secondly, in the last post, i said set -- $string is redundant. Maybe i chose the wrong words. I should have just said set -- $string is wrong.

Quote:

and unsetting IFS does not not set it to the default value;

well, maybe default value is the "wrong" phrase? anyway, i will refer you to the trusted bash ref (man bash) under special parameters

Quote:

Special Parameters
The shell treats several parameters specially. These parameters may only be referenced; assignment to them is not allowed.
* Expands to the positional parameters, starting from one. When the expansion occurs within double quotes, it expands to a sin-
gle word with the value of each parameter separated by the first character of the IFS special variable. That is, "$*" is
equivalent to "$1c$2c...", where c is the first character of the value of the IFS variable. If IFS is unset, the parameters
are separated by spaces. If IFS is null, the parameters are joined without intervening separators.

so the supposedly correct version

Code:

string="my string with spaces and - dashes - dashes"
unset IFS
string=${string//-/}
set -- $string
for i in $@
do
    echo $i
done

catkin · 12-12-2009, 08:33 AM

Quote:

Originally Posted by ghostdog74

first, yes you are right. secondly, in the last post, i said set -- $string is redundant. Maybe i chose the wrong words. I should have just said set -- $string is wrong.

well, maybe default value is the "wrong" phrase? anyway, i will refer you to the trusted bash ref (man bash) under special parameters

so the supposedly correct version

Code:

string="my string with spaces and - dashes - dashes"
unset IFS
string=${string//-/}
set -- $string
for i in $@
do
    echo $i
done

OK -- I'm getting there. Thanks for bearing with me.

I tried that code and it printed the space-separated words of "my string with spaces and dashes dashes", one per line.

Using the information you showed from the man page, is this the technique you had in mind to do what matt007 asked for

Code:

c:~$ string="my string with spaces and - dashes - dashes"
c:~$ string=${string//-/}
c:~$ set -- $string
c:~$ IFS=''
c:~$ echo -n "$IFS" | od -c -to1  # For illustration, not required
0000000
c:~$ echo "$*"
my string with spaces and  dashes  dashes

Ugh! As I understand "If IFS is null, the parameters are joined without intervening separators" there should not have been any spaces in the echo "$*" output

ghostdog74 · 12-12-2009, 09:00 AM

i think i have confused you, sorry, this part

Code:

string="my<tabs tabs> string with spaces and - dashes - dashes"
unset IFS
string=${string//[ -]/}

is the alternative to solve OP's problem. the rest of the code, is not. they are there to confirm the spaces and dashes are removed. that's all.

catkin · 12-12-2009, 09:34 AM

Quote:

Originally Posted by ghostdog74

i think i have confused you, sorry,

Don't worry -- it's not hard!

Quote:

Originally Posted by ghostdog74

this part

Code:

string="my<tabs tabs> string with spaces and - dashes - dashes"
unset IFS
string=${string//-/}

is the alternative to solve OP's problem. the rest of the code, is not. they are there to confirm the spaces and dashes are removed. that's all.

Sorry -- it doesn't work for me and I still don't understand what effect unsetting IFS is intended to have

Code:

c:~$ string=$'my\t  string with spaces and - dashes - dashes'
c:~$ unset IFS
c:~$ string=${string//-/}
c:~$ echo $string
my string with spaces and dashes dashes

ghostdog74 · 12-12-2009, 09:40 AM

well, as the ref says: "If IFS is unset, the parameters are separated by spaces".
I am just using this fact to change all the multiple tabs or multiple spaces that may be in the strings to just spaces...
then do the substitution...

Code:

string=${string//[ -]/}

I edited because i forget OP wants to remove spaces as well..
Of course, this is just an alternative. If you may like, there's also character class you can use

Code:

echo ${string//[[:space:]]/}

this is more easier...

catkin · 12-12-2009, 09:53 AM

Quote:

Originally Posted by ghostdog74

well, as the ref says: "If IFS is unset, the parameters are separated by spaces".

Wasn't that only for expanding "$*", not for any of the other special parameters and not for parameters in general and so not applicable to $string?