UNIX for Dummies Questions & Answers

I want to build an array using awk, consisting only of a subset of lines of a file. I know how to have awk assess whether a key phrase is in a particular line, but I can't find anywhere how to then append the line containing that phrase to an array that has previously-found lines also containing that phrase (that I will eventually print out as a separate text file). I can't just print out the line itself each time I find one matching the phrase, because later lines indicate whether the previous lines were "good" or not. In need to save the lines, and only print them out if I at some point hit a line telling me they were good.

Hope this makes sense? Basically, I can see how to append awk variables to print, but how to append them to a variable/array?? I need each to be on a new line in the array, so they ultimately print out correctly into a file.

-Tom

I am not sure I understood - posting an example of input and expected output is often helpful; when doing so use code tags please. Nevertheless:

Code:

$> cat infile
this bla one
this two
this three
this bla four
this bla five
$> awk '/bla/ {arr[c++]=$0} END{for(i=0; i<c; i++){print i,arr[i]}}' infile
0 this bla one
1 this bla four
2 this bla five

Filling an array with the lines matching /bla/ and using c as counter. At the END cycling through the filled array, using a new counter i, while printing index number and array elements.

I have some awk code that, at one point, does this:

Code:

{
    if ($4 == "good_trial") {
        goodTrialCount++  # increment the goodTrial counter
	stuffImKeeping=$0
}

I'm assuming that

is wrong, if I want to add the current line ($0) to stuffImKeeping. If it is wrong, how do I add $0 as a new line to the end of stuffImKeeping?

Do you just want to append? I ask because you talk of an array and the code you wrote has no sign of an array. It incerements some variable, but that variable is not being used as counter when assigning values. An array is more than just a list of elements as you can address elements either by it's index or keys.

Regarding my former example just appending without using an array:

Code:

$> awk '/bla/ {l ? l=l FS $0 : l=$0} END{print l}' infile
this bla one this bla four this bla five

Change FS to RS if you want each row on it's own line.


For your code it might look like:

Code:

{
    if ($4 == "good_trial") {
        stuffImKeeping[goodTrialCount++]=$0
}

That's the short form - you could just use the variable and increment it (the index) in the next line:

Code:

{
    if ($4 == "good_trial") {
        stuffImKeeping[goodTrialCount]=$0
        goodTrialCount++
}

Hmmm.. not sure what is best for my purposes. I think I just need to append.

Basically, I want to go through a log file looking for certain items in specific fields. In column 4, I'm looking for "good_trial". If I find it, I want to append that entire line into a temporary variable/array/whatever-is-best-for-awk that I will keep appending to, and ultimately print out to a file.

I can't print out the line immediately each time I come to one that has "good_trial" in it, because the log has 2 types of "good_trials": type "A" and type "B". I don't know if a set of "good_trials" is type A or B until after a series of "good_trials" have passed. So I need to save these lines in a temporary file until a later entry in the log tells me what to do with them (put them in type "A" category or type "B" category).

So I think that appending is all I need right now.

---------- Post updated at 03:28 AM ---------- Previous update was at 03:20 AM ----------

Did you mean:

Code:

{
    if ($4 == "good_trial") {
        stuffImKeeping[goodTrialCount]=$0
        stuffImKeeping++
}

or did you mean:

Got me, meant

Code:

{
    if ($4 == "good_trial") {
        stuffImKeeping[goodTrialCount]=$0
        goodTrialCount++
}

I think you do not need an array, as you don't use any of the advantages to address an element directly for example. So it sounds like

Code:

awk '/good_trial/ {l ? l=l RS $0 : l=$0} END{print l}' infile > newfile

might be suitable for you. An array might be just overhead/overpowered in this case.

Thanks much for your help.

Can you unpack this part for me and explain what it is doing?

Code:

{l ? l=l RS $0 : l=$0}

my brain hurts trying to figure out what it does, and why it does it.