Getting the metadata via CLI from tika with output format set to JSON gives sometimes invalid JSON. I only found float/array errors here in jira and thus created this ticket with a new case.
In my case the file that lead to invalid JSON output was a PNG file (that I unfortunately can't provide for testing):
is invalid JSON.
It would be nice if there's always valid json output from tika. For other cases that might not be catched via fixes by this ticket it would be nice to have a CLI argument/option that disables the output of certain (unknown?) fields or allows giving a whitelist of fieldnames to output. That way users can bridge the time until new releases of tika by being more specific on the shell. If that feature already exists I apology for not having found it directly and a hint to the CLI option would be nice.