UNIX Desktop Questions & Answers

I'm not totally sure what you want to do in the end, so maybe I'm way off here. If so, sorry.

Given your snippet,
Assume that is in a file called "bla.txt".
This:
Will output:
PROCNAME1 VEH_INC_PART_UPDTBASE_M.SQL

If you had more scripts in it it would pick up each one (I think), unless the var/proc pair was broken up by a carriage return.

You can probably tune it quite a bit for your situation, whatever that is.
Brief explanation:
-i case insensitive match
-o just print the matched string, not the whole line like usual
-P use Perl regex's (for the lookbehind)
(<?<= ) lookbehind matches the space before your "proc" name but keeps it out of the -o print
[[:alnum:]_]+=[[:alnum:]_]+\.SQL match one or more alphanumeric or _ characters followed by "=" followed by one or more alphanumerics or _ followed by ".sql"

Then pipe it to awk to print the first and second fields delimited by "=".

You could do other things too like add a -r to the grep to search a whole directory structure recursively, or pipe everything to "sort -u" to eliminate duplicate results, even "sort | uniq -c | sort -n" to get numeric values for how often the same proc/var name pairs occur.

---------- Post updated Jan 20th, 2011 at 03:58 PM ---------- Previous update was Jan 19th, 2011 at 10:58 PM ----------

I just realized I should have posted that reply to owenian rather than methyl to be in the proper place in this thread....sorry...it's my first post @ unix.com.

grepeeverything thank you for responding. I can't get it to work but I will continue to play with it.

I did learn from your response as you provided a break down of the code and what each bit does.

If you describe what "can't get it to work" means I might be able to help. I could imagine maybe your version of grep is different, e.g. maybe doesn't support the -P option.

correct it does not support -P or -o
ie

Code:

[ku2q@dwsdv1]/proj/dw/devl/script_job/controlm/map_tables/work
$ grep -ioP '(?<= )[[:alnum:]_]+=[[:alnum:]_]+\.SQL ' cccda331z_ld_ccc_bse_tbls_m.sh | awk -F= '{print $1,$2}'
grep: illegal option -- o
grep: illegal option -- P
Usage: grep -hblcnsviw pattern file . . .
[ku2q@dwsdv1]/proj/dw/devl/script_job/controlm/map_tables/work

Thanks for your time grepeverything.

ps I thought I had hacked out a fix but when i ran a bigger test volume it was not the case ; )

hm, well you could just download (& build & install) gnu grep, which does have the -o and -P switches:http: / / directory.fsf.org / project / grep /
(remove the spaces)

Otherwise, this sort of works but is quite ugly:
Getting a fuller-featured version of grep would be much cleaner.

can't download as the company has policies

not working.

Code:

egrep -i '[[:alnum:]_]+=[[:alnum]_]+\.SQL ' cccda331z_ld_ccc_bse_tbls_m.sh

- does not pipe anything into awk

I will play around with it.
thanks

You have a space missing at the beginning of the pattern and the ":" missing the named character class [:alnum:]

egrep -i '***SPACE***[[:alnum:]_]+=[[:alnum**COLON**]_]+\.SQL '
The space is somewhat optional but the colon not.