awk blues ..怎么了？

Question

I have to read from a file 'main.csv' and print the output as below:

Set 4,1095,Set 1 or 4
Set 2,18277,Set 2 or 3
Set 1 or 4,Set 4,944
Set 2,1373,Set 2 or 3
Set 4,83,Set 1 or 4
Set 2,14356,Set 2 or 3
Set 2,14800,Set 2 or 3

I tried below with respective outputs:

awk 'BEGIN {FS=","}{OFS=","}{print $12,$15}' main.csv
Set 4,1095
Set 2,18277
Set 4,944
Set 2,1373
Set 4,83

awk 'BEGIN {FS=","}{OFS=","}{if($12 == "Set 1" || $12 == "Set 4"){print "Set 1 or 4",$12,$15}else{print "Set 2 or 3",$12,$15}}' main.csv
Set 1 or 4,Set 4,1095
Set 2 or 3,Set 2,18277
Set 1 or 4,Set 4,944
Set 2 or 3,Set 2,1373
Set 1 or 4,Set 4,83
Set 2 or 3,Set 2,14356
Set 2 or 3,Set 2,14800

awk 'BEGIN {FS=","}{OFS=","}{if($12 == "Set 1" || $12 == "Set 4"){print $12,$15,"Set 1 or 4"}else{print $12,$15,"Set 2 or 3"}}' main.csv
,Set 1 or 4
,Set 2 or 3
,Set 1 or 4
,Set 2 or 3
,Set 1 or 4
,Set 2 or 3
,Set 2 or 3

First 2 "awk"s run fine, then why not the 3rd one ?

main.csv has 15 comma separated values per line with 12th value as either of below:

Set 1
Set 2
Set 3
Set 4

PS: I'm using bash on Solaris 10

I'm still high on "awk" issues, below is what I noticed... not sure why it's happening (I'm getting this on both Solaris 10 and RHEL 5)

awk -F, '{print "Var13->"$13,"\t","Var12->"$12,"\t","Var15 ->"$15}' < main.csv
Var13->0         Var12->Set 4    Var15 ->1095
Var13->1631      Var12->Set 2    Var15 ->18277
Var13->0         Var12->Set 4    Var15 ->944
Var13->2832      Var12->Set 2    Var15 ->1373
Var13->0         Var12->Set 4    Var15 ->83

awk -F, '{print "Var13 & Var15->"$13,"\t","Var12->"$12,"\t","Var15 ->"$15}' < main.csv
Var13 & Var15->0         Var12->Set 4    Var15 ->1095
Var13 & Var15->1631      Var12->Set 2    Var15 ->18277
Var13 & Var15->0         Var12->Set 4    Var15 ->944
Var13 & Var15->2832      Var12->Set 2    Var15 ->1373
Var13 & Var15->0         Var12->Set 4    Var15 ->83


awk -F, '{print "Var13 & Var15->"$13,$15"\t","Var12->"$12,"\t","Var15 ->"$15}' < main.csv
or
awk -F, '{print "Var13 & Var15->"$13,$15,"\t","Var12->"$12,"\t","Var15 ->"$15}' < main.csv
or
awk -F, '{print "Var13 & Var15->"$13,"\t",$15,"\t","Var12->"$12,"\t","Var15 ->"$15}' < main.csv

ar13 &  Var12->Set 4    Var15 ->1095
ar13 &  Var12->Set 2 27 Var15 ->18277
ar13 &  Var12->Set 4    Var15 ->944
ar13 &  Var12->Set 2 73 Var15 ->1373
ar13 &  Var12->Set 4    Var15 ->83
ar13 &  Var12->Set 2 56 Var15 ->14356

As you can see last 3 runs had eaten up the letter from "Var13" output and added some junk info.

Answer 1

No need to vote for this, I just wanted to paste in some formatted code.

Your program is getting big enough to deserve its own code file where it can be formatted in a civilized fashion and compared to other versions. When we do this we can easily see that you are assigning OFS in an action on every input line. No harm done, but it's obviously unintended so you lose 1337 points. With that fixed we move on.

You can't diff a one-liner and learn anything. Running diff(1) on the nicely formatted code shows that only the output lines have changed, and so both programs are equally valid. You must have made the third run under different conditions. Perhaps a different input file?

BEGIN  {
  FS  = ","
  OFS = ","
}
{ if($12 == "Set 1" || $12 == "Set 4") {
    print "Set 1 or 4",$12,$15
  } else {
    print "Set 2 or 3",$12,$15
  }
}

---

BEGIN {
  FS  = ","
  OFS = ","
}
{ if($12 == "Set 1" || $12 == "Set 4") {
    print $12,$15,"Set 1 or 4"
  } else {
    print $12,$15,"Set 2 or 3"
  }
}

Answer 2

My money's on you having control-Ms at the end of your ".csv" file lines causing the first part of the displayed lines to get hidden.

This corruption is brought to you at no extra cost by the people who brought you Windows Vista, etc.

Run "dos2unix" on your ".csv" file and then try again.

If that doesn't work, run "where awk" and "awk --version" and tell us the result. On Solaris you must use /usr/xpg4/bin/awk or nawk, do not use /bin/awk or /usr/bin/awk as those are both old, broken awk.