Parsing the output of date with sed

Question

I am trying to replace the whitespaces in the output of date with '_' with no success.

$date
Fri Sep 14 14:10:04 EDT 2012

$ date | sed 's/ /_/'
Fri_Sep 14 14:10:24 EDT 2012

As you can see, the last command only replaced the first whitespace by a _'. Why did it skip the other whitespaces?

Answer 1

Try:

$ date | tr ' ' '_'

or

$ date | sed 's/ /_/g'

Your command only replaced the first matching instance from the date input because that is the default behaviour of sed. By adding the g (global replacement) option to the end of the command, sed will instead match and replace all occurrences of the expression.

Answer 2

Just my 2 cents. If you want a date as a single word, a good format is the ISO 8601 standard 2012-09-14T21:08:12. It is terse, unambiguous, and its chronological order is the same as its alphabetical order.

date +%Y-%m-%dT%T

which on some systems you may even write as

date +%FT%T