CODEWITHSUNDEEP
linuxbashshellawk
gletscher
gletscher

Reputation: 1132

awk strftime to return timest

I'm using this nifty one-liner to add a timestamp to each line of my logfile:

tail -f log  | grep -a --line-buffered "." | awk '{ print strftime("%s\t"), $0; fflush() }'

Unfortunately it only gives full seconds while I need millisecond resolution between the datapoints. Is there an equally elegant solution to get a ms timestamp? I don't care about the time since the epoch, only about the difference between the lines.

Thanks!

Upvotes: 1

Views: 1545

Answers (3)

Jotne
Jotne

Reputation: 41456

You can use your awk but need to change it some:

tail -f log  | grep -a --line-buffered "." | awk '{ print d, $0; fflush() }' d=$(date -Ins)

Upvotes: 2

w0ng
w0ng

Reputation: 106

Replace awk with sed, then use $(date -Ins) to add an ISO 8601 timestamp with nanosecond precision.

tail -f infile | grep -a --line-buffered "." | sed 's/^/'"$(date -Ins)\t"'/'

2014-01-18T17:24:08,110459605+1100  one
2014-01-18T17:24:08,110459605+1100  two
2014-01-18T17:24:08,110459605+1100  three

or $(date --rfc-3339=ns) for an alternate format:

tail -f infile | grep -a --line-buffered "." | sed 's/^/'"$(date --rfc-3339=ns)\t"'/'

2014-01-18 17:24:51.985434198+11:00 one
2014-01-18 17:24:51.985434198+11:00 two
2014-01-18 17:24:51.985434198+11:00 three

Upvotes: 1

Birei
Birei

Reputation: 36262

You can try to replace the and the with and its Time::HiRes built-in module, like:

tail -f infile | perl -MTime::HiRes=time -ne 'printf "%.3f\t%s", time(), $_'

It yields something like:

1390014680.197  one
1390014680.197  two
1390014680.197  three
1390014680.197  four
1390014680.197  five
1390014680.197  six
1390014689.414  seven
1390014693.542  eight

Upvotes: 3

Related Questions