Shell Programming and Scripting

So what changed between when it was working and when it stopped working? Did the script change? Or, did the data change? It has to be one of these (unless it is both)!

As I said before, there is no place in your data change contains the string C00 that appears as a word by itself in your sample data.

maybe this command can be altered to take into account the searching of multiple patterns and have it be efficient

Code:

awk '/C00/ {++c} END{print "C00="c}' test.log

here we're only searching for one pattern.

suppose i want to search a 70MB file for multiple different patterns.

i would like to run a version of this command so the output i get is something similar to this:

Code:

awk '/C00|NHHA|MATKA|NORI|ERROR|FAILURE|SEGFAULT/ {++c} END{print "C00="c}' test.log

C00=33 NHHA=3 MATKA=45 NORI=3 ERROR=11 FAILURE=2 SEGFAULT=1 ...and so on

---------- Post updated at 03:02 AM ---------- Previous update was at 03:01 AM ----------

i dont know what happened. still trying to figure that out.

You can't count the number of lines that contain one or more of seven patterns (incrementing one variable and using a single ERE match) and then magically determine how many lines contained any particular one of those seven patterns.

The script I suggested with the printf feeding data into awk changed to:

Code:

printf '%s\n' C00 NHHA MATKA NORI ERROR FAILURE SEGFAULT

should give you what you want pretty efficiently. The following is probably slightly faster:

Code:

awk '
/C00/ {++c}
/NHHA/ {++nh}
/MATKA/ {++m}
/NORI/ {++no}
/ERROR/ {++e}
/FAILURE/ {++f}
/SEGFAULT/ {++s}
END {print "C00="c, "NHHA="nh, "MATKA="m, "NORI="no, "ERROR="e, "FAILURE="f, "SEGFAULT="s}
' test.log

Again, this is assuming that you want to search every line (while the code you showed us before ignored the first input line). Is there a header line in your input file that needs to be skipped?

And, are you trying to count the number of lines containing these strings? (That is what the code above does.) Or, are you trying to count the number of appearances of these strings? (Which would need to use calls to gsub() as was done in the script you showed us before.)

Assign zero to each search at the beginning ...

Code:

awk '
BEGIN{c=nh=m=no=e=f=s=0}
/C00/ {++c}
/NHHA/ {++nh}
/MATKA/ {++m}
/NORI/ {++no}
/ERROR/ {++e}
/FAILURE/ {++f}
/SEGFAULT/ {++s}
END {print "C00="c, "NHHA="nh, "MATKA="m, "NORI="no, "ERROR="e, "FAILURE="f, "SEGFAULT="s}
' test.log

... so you can get an output with zero for each entry with no match in the file, if you want.

Code:

C00=4 NHHA=0 MATKA=0 NORI=0 ERROR=0 FAILURE=0 SEGFAULT=0

Ouch. Yes. Or use printf instead of print to display the results (as I did in my earlier suggestion):

Code:

printf("c00=%d NHHA=%d MATKA=%d NORI=%d ERROR=%d FAILURE=%d SEGFAULT=%d\n",c,nh,m,no,e,f,s)