Question

Have been trying to write an awk script which processes a log file, but am struggling.

I have a file which has lines which look like this:

[2011-07-29 04:44:37.100 INFO] AU/SUB1/Server:WebHits : Hits[ABC]=0; Hits[DEF]=876; Hits[THY]=0; Hits[SG]=891; Hits[XFRR]=1386
[2011-07-29 04:44:37.599 INFO] AU/SUB2/Server:WebHits : Hits[XARR]=0; Hits[XXX]=0; Hits[ABS]=0; Hits[SM]=0
[2011-07-29 04:44:37.699 INFO] AU/MAIN/Server:Main : Hits=254
[2011-07-29 04:44:38.100 INFO] AU/SUB1/Server:WebHits : Hits[ABC]=0; Hits[DEF]=1134; Hits[THY]=0; Hits[SG]=1153; Hits[XFRR]=426
[2011-07-29 04:44:38.599 INFO] AU/SUB2/Server:WebHits : Hits[XARR]=0; Hits[XXX]=0; Hits[ABS]=0; Hits[SM]=22
[2011-07-29 04:44:38.699 INFO] AU/MAIN/Server:Main : Hits=436

As you can see there are three lines per second (there will always be three lines per second). I would like to combine these lines (one line per second), so that a summary would look like this, so one line of summary correlators to three lines from the input log file:

[2011-07-29 04:44:37 INFO] MainHits=254,ABC=0,DEF=876,THY=0,SG=891,XFRR=1386,XARR=0,XXX=0,ABS=0,SM=0
[2011-07-29 04:44:38 INFO] MainHits=436,ABC=0,DEF=1134,THY=0,SG=1153,XFRR=426,XARR=0,XXX=0,ABS=0,SM=22

Please note the venue names e.g. “ABC,DEF,THY,SG etc” can change dynamically.

Any help would be appreciated.

Answer 1

The solution relies heavily on the input format you’ve posted:

awk '/Main :/ {
  sub(/\.[0-9]*/, x, $2)
  print $1, $2, $3, "Main" $NF, r
  r = x; next
  }
{
  gsub(/Hits\[/, x)
  gsub(/[];]/, x)
  for (i = 5; ++i <= NF;)
    r = r ? r "," $i : $i
  }' infile