topic Re: script help... in Operating System - HP-UX

script help...

Keith Floyd — Tue, 02 Jan 2001 15:05:15 GMT

Hi

I want to parse a text file for a particular line - then when found read the
following lines until another line found

ie

find this line PV Name LE on PV PE on PV

push these to a new file /dev/dsk/c2t4d0 2048 2048

finsh here --- Logical extents ---

Hope someone can help

Thanks

Keith

Re: script help...

Dan Hetzel — Tue, 02 Jan 2001 15:11:27 GMT

Hi Keith,

AGREP is the best tool for you.
The main difference with grep, is that agrep has definabla record delimiters (could be a string) while grep only has the End-Of-line.

You'll find it at the porting center:
http://hpux.cae.wisc.edu/hppd/hpux/Misc/agrep-2.04/

Best regards,

Dan

Re: script help...

Dan Hetzel — Tue, 02 Jan 2001 15:12:50 GMT

Hi again,

Here is the small abstract for agrep:
Searches text for strings or approximations thereof. Agrep is similar to the standard UNIX grep utility, but is more general, and according to its authors, usually faster. Its most notable feature is the approximate pattern matching facility. Agrep has other features not found in the grep family: It is record rather than line orientated; pattern matches can be output in a larger context than a single line. Multiple search patterns can be specified with AND/OR logic.

Dan

Re: script help...

Andreas Voss — Tue, 02 Jan 2001 15:14:45 GMT

Hi,

looks like you want to get the devices for a log. volume.
Here my code:
lvdisplay -v /dev/vgXX/lvolX |awk 'BEGIN{f=0}
{
if($0 ~ "PV Name")
{
f=1
continue;
}
if($0 ~ "Logical extents")
{
f=2;
exit;
}
if(f==1)
{
print $0
}
}'

Regards

Re: script help...

Curtis Larson — Tue, 02 Jan 2001 15:25:45 GMT

cat your_file|
awk ' /PV Name/ {
getline;
while ( $0 !~ "Logical extends" ) {
print $0;
getline;
}
exit;
}'

Re: script help...

Carlos Fernandez Riera — Tue, 02 Jan 2001 15:26:41 GMT

Are you trying to something like this: http://my1.itrc.hp.com/cm/QuestionAnswer/1,1150,0x5cfe6c96588ad4118fef0090279cd0f9!0,00.html

Re: script help...

Tommy Palo — Tue, 02 Jan 2001 15:42:31 GMT

Try:
cat FILE | awk '/PV Name LE/,/Logical extents/' > NEW_FILE

Re: script help...

Curtis Larson — Tue, 02 Jan 2001 16:28:45 GMT

and if you want something more generic put this into an executable file, such as print_from_s1_to_s2.

#!/usr/bin/ksh

string1=$1
string2=$2

awk ' {
if ( $0 ~ "'"$string1"'" ) {
getline;
while ( $0 !~ "'"$string2"'" ) {
# remove blanklines
if ( $0 !~ /^$/ ) print $0;
getline;
}
exit;
}
}'

then use it where ever you like

lvdisplay | print_from_s1_to_s2 "PV Name" "Logical extents" > yourfile

Re: script help...

Keith Floyd — Tue, 02 Jan 2001 17:17:08 GMT

Thanks
At least one works - I will check the various methods - and assign points -

Again thanks for all your help

Re: script help...

David Totsch — Thu, 04 Jan 2001 18:46:56 GMT

Keith:

lvdisplay -v /dev/vg00/lvol1 | sed -n "/-- Distribution/,/^$/p"

is close to what you want.

This is a sed(1) "ranging". The first /RE/ instructs sed(1) to find the line with the RE (where RE is a regular expression). The second /RE/, /^$/ is looking for a blank line that follows the first RE. 'p' prints lines (that is why sed(1) was called with -n, don't print unless I tell you to).

Now, using sed(1), lets strip out what you don't want:

lvdisplay...| sed -n "/--Distribution/,/^$/{
/--Distribution/d
/PV Name/d
/^$/d
p
}"

The stuff between the curly braces is a subroutine. The statements delete what you don't want (including the trailing blank line) and explicity print what remains.

Enjoy,
-dlt-

Re: script help...

Tommy Palo — Fri, 05 Jan 2001 07:40:00 GMT

I realized that you wanted to get rid of the target lines also so try:

cat FILE | awk '/PV Name LE/,/Logical extents/' | sed -e '1d' -e '$d' > NEW_FILE