Hi,

suppose I have a string composed by a number and some characters (e.g. a
BASIC statement like

10 PRINT I

that could be also written as

10PRINT I
10 <spaces> PRINT I
10 <tabs> PRINT I
10 <spaces and tabs> PRINT I

or something like this); how can I separate in bash the number from the
characters, splitting the string so that, for instance,

ln=10
pr="PRINT I" or whatever, discarding also all the trailing blanks after
last useful character of "PRINT I"? ln should be any integer number, and
it could be 1 (one) or 15678, for example. Can you me?

I use bash 3.x, and I'm not really a guru (though not a newbie).

Thanks in advance

-- Antonio

Antonio Maschio wrote:
> Hi,
>
> suppose I have a string composed by a number and some characters (e.g. a
> BASIC statement like
>
> 10 PRINT I
>
> that could be also written as
>
> 10PRINT I
> 10 <spaces> PRINT I
> 10 <tabs> PRINT I
> 10 <spaces and tabs> PRINT I
>
> or something like this); how can I separate in bash the number from the
> characters, splitting the string so that, for instance,
>
> ln=10
> pr="PRINT I" or whatever, discarding also all the trailing blanks after
> last useful character of "PRINT I"? ln should be any integer number, and
> it could be 1 (one) or 15678, for example. Can you me?
>
> I use bash 3.x, and I'm not really a guru (though not a newbie).
>
> Thanks in advance
>
> -- Antonio

Is the Basic command in a file or in a variable?
If you read the line from file...

read -r line
ln=${line%%[!0-9]*}
cmd=${line##*[0-9]}
echo "$ln"
echo "$cmd"


Janis

On Thu, 07 Dec 2006 08:05:11 -0800, Janis wrote:

> Antonio Maschio wrote:
>> Hi,
>>
>> suppose I have a string composed by a number and some characters (e.g. a
>> BASIC statement like
>>
>> 10 PRINT I
>>
>> that could be also written as
>>
>> 10PRINT I
>> 10 <spaces> PRINT I
>> 10 <tabs> PRINT I
>> 10 <spaces and tabs> PRINT I
>>
>> or something like this); how can I separate in bash the number from the
>> characters, splitting the string so that, for instance,
>>
>> ln=10
>> pr="PRINT I" or whatever, discarding also all the trailing blanks after
>> last useful character of "PRINT I"? ln should be any integer number, and
>> it could be 1 (one) or 15678, for example. Can you me?
>>
>> I use bash 3.x, and I'm not really a guru (though not a newbie).
>>
>> Thanks in advance
>>
>> -- Antonio
>
> Is the Basic command in a file or in a variable?
> If you read the line from file...
>
> read -r line
> ln=${line%%[!0-9]*}
> cmd=${line##*[0-9]}
> echo "$ln"
> echo "$cmd"
>
>
> Janis

The line number extraction is fine provided there are no initial spaces,
but the cmd extraction fails say for the example

10 print 42

Try

cmd=${line#${ln}}
if there are no spaces at the start of the line. cmd will contain initial
spaces if there are any after the line number. Since you are using bash 3,
you can also use extended glob patterns. These are

# Enable extended patterns
shopt -s extglob
# like sed 's/^ *[0-9][0-9]* *//'
cmd=${line/#*( )+([0-9])*( )/}
# like sed 's/^ *//'
ln=${line##*( )}
# like sed 's/[!0-9].*//'
ln=${ln%%[!0-9]*}

sed 's/\([0-9]*\)[a-zA-Z ]/\1/g' < inputfile

or

cat inputfile | sed 's/\([0-9]*\)[a-zA-Z ]/\1/g'
|
this is a space followed
by a tab

where there is a space and a tab after the uppercase Z in A-Z

hth

Scott Gillespie

Antonio Maschio wrote:
> Hi,
>
> suppose I have a string composed by a number and some characters (e.g. a
> BASIC statement like
>
> 10 PRINT I
>
> that could be also written as
>
> 10PRINT I
> 10 <spaces> PRINT I
> 10 <tabs> PRINT I
> 10 <spaces and tabs> PRINT I
>
> or something like this); how can I separate in bash the number from the
> characters, splitting the string so that, for instance,
>
> ln=10
> pr="PRINT I" or whatever, discarding also all the trailing blanks after
> last useful character of "PRINT I"? ln should be any integer number, and
> it could be 1 (one) or 15678, for example. Can you me?
>
> I use bash 3.x, and I'm not really a guru (though not a newbie).
>
> Thanks in advance
>
> -- Antonio

Icarus Sparry wrote:
> On Thu, 07 Dec 2006 08:05:11 -0800, Janis wrote:
>>
>>read -r line
>>ln=${line%%[!0-9]*}
>>cmd=${line##*[0-9]}
>>echo "$ln"
>>echo "$cmd"
>>
>
> The line number extraction is fine provided there are no initial spaces,
> but the cmd extraction fails say for the example

Yes, you're right. I had the ksh syntax

cmd=${line##*([0-9])}

in mind (which wouldn't be available in bash anyway).
I shouldn't post while in a hurry.
Thank's for the correction.

Janis

2006-12-08, 00:53(+01), Janis Papanagnou:
[...]
>> The line number extraction is fine provided there are no initial spaces,
>> but the cmd extraction fails say for the example
>
> Yes, you're right. I had the ksh syntax
>
> cmd=${line##*([0-9])}
>
> in mind (which wouldn't be available in bash anyway).
[...]

The ksh88 syntax is available in bash after a:

shopt -s extglob

and in zsh after

setopt kshglob

In zsh, you may prefer zsh's extendedglob

cmd=${line##[0-9]#}

or

cmd=${line##<0->}

(<0-> is any decimal number starting from 0).

Newer ksh93 operators such as {1,2}(...), %(...) are not
available yet in bash or zsh.

--
Stéphane

Stephane CHAZELAS wrote:
> 2006-12-08, 00:53(+01), Janis Papanagnou:
> >
> > cmd=${line##*([0-9])}
>
> The ksh88 syntax is available in bash after a:
>
> shopt -s extglob

Ah, yes; I had tried it just in plain bash.
Why don't they activate that feature by default?

Janis