jq null unix timestamps parsing issue

Question

I'm trying to parse a big json file which I receive using curl.

By following this answer I could parse the next file:

$ cat test.json 
{"items": [{"id": 110, "date1": 1590590723, "date2": 1590110000, "name": "somename"}]}

using the next command:

TZ=Europe/Kyiv jq -r '.[] | .[] | .name + "; " + (.date1|strftime("%B %d %Y %I:%M%p")) + "; " + (.date2|strftime("%B %d %Y %I:%M%p"))' test.json

Output is:

somename; May 27 2020 02:45PM; May 22 2020 01:13AM

But when I try to parse the next file using the same command:

$ cat test2.json 
{"items": [{"id": 110, "date1": 1590590723, "date2": null, "name": "somename"}]}

Output is:

jq: error (at test2.json:1): strftime/1 requires parsed datetime inputs

I could replace those null values using sed by some valid values before parsing. But maybe there is a better way to skip (ignore) those values, leaving nulls in output:

somename; May 27 2020 02:45PM; null

Answer 1

You could tweak your jq program so that it reads:

def tod: if type=="number" then strftime("%B %d %Y %I:%M%p") else tostring end;

.[] | .[] | .name + "; " + (.date1|tod) + "; " + (.date2|tod)

An alternative would be:

def tod: (tonumber? | strftime("%B %d %Y %I:%M%p")) // null;

.[] | .[] | "\(.name); \(.date1|tod); \(.date2|tod)"