awk/sed line parsing

Shell Programming and Scripting
1

I'm new to shell programming, but I think I learn best by following an example. I'm trying to cook up an awk/sed script, but I obviously lack the required syntax skills to achieve it. The output that I get from running my ksh script looks like this:

I need to search each numbered line for certain keywords. If I have no match, then simply continue with the rest of my script. If there are matches, then I want to be able to capture matched lines (to a temp file?) then report the matched lines as the command's output ('cat' temp file?) and exit my script. Appreciate any suggestions.

---------- Post updated at 06:38 AM ---------- Previous update was at 06:20 AM ----------

A bit more details of my framework:

( ksh myscript.ksh 2>&1 ) | tee -a $TEMP_LOG
            perl -ne "exit 1 if ( \
                /keyword1/ || \
                /keyword2/ || \
                /keyword3/ \
            );" $TEMP_LOG
            if [ $? -eq 1 ]
            then
                echo ""
                echo "Error while running myscript.sh!"
                EXITCODE=1
            else
                echo "Done running myscript.sh."
            fi
            rm -f $TEMP_LOG

So I basically want to be able to capture the keywords from the output of another command shown above and then error out along with reporting the lines with found keywords.

2

You should code the error handling and output generation directly inside myscript.ksh insead of creating another script that parse its output afterward.

3

Agreed. I will incorporate my error checking within myscript.sh. Meanwhile any suggestions on writing the awk code to filter out the keywords? In the output example above I want to be able to read between the CHANGED FILES: line and the next blank line. Then again between the NEW FILES: line and the next blank line, etc.

4

Give more clue about how does your initial input file look like (upload it if necessary), and how should the output look like.
.

5

This is the output of a tool command that I run :

The flow should be like this: I run an external tool command and I parse its output for certain keywords. If there's a match of keywords, then I want to spit out a message saying that the script cannot proceed any further due to the keyword matches and then spit out the matched lines. If there are no matches, however, then I want my external script to continue on.
I want to be able to parse the output like the one above and catch certain words, such as 'bar.txt' or 'foo/'. If I do have the matches, I'd like to say something like this:

Your request cannot proceed because you had the following critical changes:

... then list the captured matches as a whole line ...

6

Is "grep" not enough for your requirement?

7

I need to capture and list each line that contains a matching keyword. I think grep will handle only one case form the output, no?

8

from where you will get the matching keyword? or those static?

 
grep -e "pattern1|pattern2" will search for multiple patterns!
9

The keywords will be static and defined in the script. The output I've quoted earlier will be generated by some command which I call in my script and I want to be able to capture and parse the output for certain keywords. Then if I do have a keyword match, I want the script to quit and present the matched lines. I think grep only does one keyword at a time? I want my output to look like this (assuming there's a match for all instances of bar1.txt files and directories named 'foo'):

10

I did not get you fully!!!

Please post a sample input and output you are expecting along with the sample keywords.

11

Sorry for not being clear! Here's the sample run of an external command which I need to call within my script and its output:

When I run this command in my script, I want to be able to parse the output above for keywords which I also define in the script, such as "bar.txt", "foo/", etc. If there's a match, then I want my script to say something like this:

---------- Post updated at 03:48 AM ---------- Previous update was at 03:12 AM ----------

I tried to use this perl command as a parsing mechanism, but if I have a keyword that looks like "directory/file" then perl complains and I'm not sure how to escape the slash.

cat command_output.txt | perl -ne "exit 1 if ( /keyword1/ || /directory/keyword2/ || /XXXXXXXXXXXXXXXXX/ );" ; echo $?

---------- Post updated at 04:13 AM ---------- Previous update was at 03:48 AM ----------

Basically I want my script to look something like this:

external command | awk 'parse output for keywords'
if matches found
then
     print matched lines
     exit 1
else
     echo "no matches found"
fi