@cokedude , you can edit your post now , though its (probably) preferable that you simply respond with an updated post on the same thread, that way context is maintained.
Please mind, that you're searching for a matching pattern across multiple files. So given that the files will be searched through in a particular order (filename sorting order may depend on your locale settings), do you want to get all files with two or more matches (across all files), or do you only want the first found matching file (if more than one file contains 2 or more matches)?
#!/usr/bin/env bash
for filename in ./*; do
[[ ! -f "$filename" ]] && continue;
if (( $(grep -i -c 'EXEC .* CURSOR for' "$filename") >= 2 )); then
grep -iHm 2 'EXEC .* CURSOR for' "$filename";
# this will stop the execution after first matching file
break;
fi;
done
Also, is it first two occurrences you're interested in? grep -iHm 2 'EXEC .* CURSOR for' "$filename"
or last two occurrences? grep -iH 'EXEC .* CURSOR for' "$filename" | tail -n 2
or randomly selected two occurrences (if there are more than two)? grep -iH 'EXEC .* CURSOR for' "$filename" | shuf -n 2
or do you want to see all "2 or more" occurrences?? grep -iH 'EXEC .* CURSOR for' "$filename"
It would be helpful to know which flavour of awk you have available. GNU/awk has patterns for BEGINFILE and ENDFILE, which would considerably simplify a solution here.
I really like the first one :). Its really easy to read. With what I am looking being code it is tabbed so it easy see.
How do you come up with this? When grep does not work I am not sure where I should go. Always have trouble deciding between awk, sed, and perl. awk does usually work best with sunos. Then I have no idea what I am trying to do or what I need to do to accomplish my goal. Did this take you long to come up with?
I do not have the GNU/awk. Not sure how to check the version. I am using some old school sunos version. The man pages mention both oawk or nawk.
HISTORY
The oawk command was installed as /usr/bin/awk in all Sun and Oracle
releases of Solaris up through Oracle Solaris 11.4.32. Starting in
11.4.33, that version of awk was moved to /usr/bin/oawk, and the awk
pkg(7) mediator was created to allow sites to choose whether
/usr/bin/awk links to oawk or nawk.
Oracle Solaris 11.4 29 Nov 2022 awk(1)
awk --version
awk: unknown option --version ignored
awk: no program given
awk -v
awk: no program given
Good, after decades Oracle finally allows to switch /usr/bin/awk to nawk
(While other Unix vendors switched with their next major OS release.)
Doing so there is no risk, but a big advantage.
Of course, you could also run the POSIX versions in /usr/xpg4/bin/ - not only awk but also grep and sed.
One reading of each file with awk is efficient, only half of the I/O, compared to two greps.
How did I make the awk code? First I realized that it runs file by file, so it is sufficient to count and store the matches in simple variables. And at the end of a file print the store s if the count n is >= 2. Missing an ENDFILE (only GNU awk) I use two conditions: when the FILENAME has changed and at the very END (after the last file).
When the FILENAME has changed then also reset the count and store, and change the pFN (previous FILENAME), so another change of the FILENAME can be detected.
To be fair, nawk was available on SunOs/Solaris much earlier -- somewhere around 1995, I think. Certainly before Oracle acquired Solaris. It was annoying that you needed to explicitly invoke nawk (or set up your own links), but it worked.
The only bug I remember was that it would SegViol if you passed it an untyped argument and treated it as an array. My code had a lot of split ("", X, ""); calls to fix this.
Versioning was not available. I think the acid test to find out if you had awk or nawk was to have a tiny pre-test that invoked asort() on a single array entry. If that worked, you had nawk. If it wrote to stderr, you only had old awk.
I like the boolean test. My bad: there was something that I used, but not asort(). It might have been the $99 limit on field numbering. Possibly even the function keyword: seems to me that if you wanted a control break on change of FILENAME and on END, you had to repeat (and maintain) all of the associated code.
I have a lot of archives, but nothing before about 2000. I wrote a HeapSort for small internal sorts in nawk, and an external sort command plus readback with getline() for larger sorts.
the Solaris awk=oawk does not have sub(), gsub(), toupper(), tolower(), close(). And doesn't know the function keyword for custom functions.
BTW my script with a function that avoids the repeated printing code, plus a parameter for the search string:
