Line by line parsing about jc

[kellyjonbrazil]

Hi there!

Does this provide the expected output?

while true; do
  date | jc —date
  sleep 1
done

[mitar]

Suure. But I do not control how is the stream of dates generates. That was just an example/reproduction to motivate this issue.

[kellyjonbrazil]

Gotcha. I’d have to play around with it but can’t get to it for a few days - not sure if it’s possible in a stream you don’t control. There are a few parsers that support streams (like ping-s) but the date parser is not one of them.

You could implement a plugin parser (e.g. date-s. You could take a look at how the other streaming parsers are done and do something similar.

[mitar]

No hurry. I think it would be nice if any plugin could be converted to a stream one by simply reapplying it again and again on the stream. But probably that would require from each plugin to have a concept of chunks.

[kellyjonbrazil]

That would definitely work for the parsers that work on a single line of text like date. Like you mentioned, would require more thought for other parsers.

[mitar]

I got some inspiration from your blog post and made regex2json tool which allows conversion of test-based data into JSON by just providing a regex. So similar to how jc works, but for more ad-hoc use cases where you write the pattern as CLI argument. And it also operates on line-to-line. This I think nicely works together with the jc tool and other tools operating on JSON. I think piping JSON lines instead of text lines between processes is really powerful.

So the example above could then be:

$ while true; do LC_TIME=C date; sleep 1; done | regex2json "(?P<date___time__UnixDate__RFC3339>.+)"
{"date":"2023-06-13T11:26:45Z"}
{"date":"2023-06-13T11:26:46Z"}
{"date":"2023-06-13T11:26:47Z"}

[kellyjonbrazil]

This is fantastic! Very cool tool. I’ll link to it in the readme in the next release.