extract text b/w two delimiters

UNIX for Advanced & Expert Users
1

I have an input file which looks like

"
@$SCRIPT/atp_asrmt_adj.sql
$SCRIPT/dba2000.scr -s / @$SCRIPT/cim1005w.pls
$SCRIPT/dba2000.scr -s / @$SCRIPT/cim1006w.pls
start $SCRIPT/cim1020d.sql;^M
spool $DATA/cim1021m.sql
@$DATA/cim1021m.sql
! rm $DATA/cim1021m.sql
spool $DATA/cim1021m.sql
@$DATA/cim1021m.sql
! rm $DATA/cim1021m.sql
start $SCRIPT/art_matrix_audit_purge.pls
start $SCRIPT/event_sku_price_log_purge.pls
start $SCRIPT/event_sku_price_audit_purge.pls
start $SCRIPT/event_sku_rscn_audit_purge.pls
@$SCRIPT/in_process_audit_purge.sql
$SCRIPT/dba2000.scr /NOLOG @$SCRIPT/mphdsum.pls
ERROR_SCRIPT "Error with deduct.pls"
$SCRIPT/dba2000.scr /NOLOG @$SCRIPT/pastdue.pls
ERROR_SCRIPT "Error with pastdue.pls"
$SCRIPT/dba2000.scr /NOLOG @$SCRIPT/ordpos1.pls
ERROR_SCRIPT "Error with ordpos1.pls"
$SCRIPT/dba2000.scr /NOLOG @$SCRIPT/ordpos2.pls
ERROR_SCRIPT "Error with ordpos2.pls"
ERROR_SCRIPT "Error Code:'$RCSUM!' in sum1so.sql"
ERROR_SCRIPT "Error Code:'$RCSUM2' in sum2so.sql"
ERROR_SCRIPT "Error Code:'$RCSUM3' in sum3so.sql"
"

My requirement is to extract all the text b/w :
1). "/" and ".sql"
2)."/" and ".pls"
3)." " and ".sql"
4)." " and ".pls"

It is really urgent, any suggestion would be highly appreciated.
Thanks
NB: there might be a number of "/" in a line

2

You will need to do something more than just copy/paste as I am not going to write everything by you (you should learn it and code it), but here it goes:

  1. You will fall into deep s**t whenever a filename contain special characters/new line.

  2. Code:
    while read -p line; do
    [[ -z "${line}" ]] && continue
    line="${line##/|\ }"
    line="${line%.pls}"
    [[ -n "${line}" ]] && print -- "${line}"
    done

EDIT: This is a korn shell's code to be clear.
I believe that you could do it with one command using sed

3

If you provide an example of the output you expect, somebody may be able to help you.

4

It is really tough to understand. Could you please explain me How didi you do that. I was applying a logic :
1). cut the line with the delimiter ".sql" and take the first field.(e.g spool $DATA/cim1021m)
2). convert all "/" to spaces and then (spool $DATA cim1021m)
3).delete everything till the last space(cim1021m)

5

If the input file is
"
@$SCRIPT/atp_asrmt_adj.sql
$SCRIPT/dba2000.scr -s / @$SCRIPT/cim1005w.pls
$SCRIPT/dba2000.scr -s / @$SCRIPT/cim1006w.pls
start $SCRIPT/cim1020d.sql;^M
spool $DATA/cim1021m.sql
@$DATA/cim1021m.sql
! rm $DATA/cim1021m.sql
spool $DATA/cim1021m.sql
@$DATA/cim1021m.sql
! rm $DATA/cim1021m.sql
"
Then the output file must be

"
atp_asrmt_adj
cim1005w
cim1006w
cim1020d
cim1021m
cim1021m
cim1021m
cim1021m
cim1021m
cim1021m

"

6 
awk 'BEGIN{ FS="[/ ]"}{print $NF} ' file
7

You can do it using sed, awk and other.
The idea is to parse this line by line and to remove prefix and postfix.
In ksh following syntax:
line="${line#xxx}"
is "remove trailing string "xxx" from the variable line. if not possible to remove then return me the string unmodified. then assign result to line".
# : remove trailing part
## : remove trailing part (as much as possible)
% : similar but for ending part
%% : similar

Example:
x='aaabbb'
"${x#a}" is 'aabbb'
"${x##a}" is 'bbb'

sed should handle this best (however I had some bad experience with lines longer than 255 chars in sed). Just tell him to remove replace {prefix_pattern}{?}{postfix_pattern} and replace it with {?}.
Sory, but I don't remember exactly what was the syntax in sed.

In example given by me you can see:

line="${line##/|\ }"
This is "remove prefix". And prefix is a pattern: '/' or ' '

line="${line%.pls}"
This is "remove postfix". And postfix is '.pls'. Just replace this with a pattern and you have it.
To pass the data for 'read -p' you could use
cat file |&