ijson.kvitems(prefix) iterates through items even after the prefix was found and closed about ijson HOT 3 CLOSED

@nisace thanks for the very thorough report.

I think there's a misalignment between the expectations you have for your particular JSON document and prefix, and how kvitems works for arbitrary JSON documents and prefixes. Consider the following two cases:

$> echo '[{"b": 1}, {"b": 1}]' | python -m ijson.dump -m kvitems -p item
#: key, value
-------------
0: b, 1
1: b, 1

$> echo '{"a": {"b": 1}, "a": {"b": 1}}' | python -m ijson.dump -m kvitems -p a
#: key, value
-------------
0: b, 1
1: b, 1

The first case is not uncommon, and the second is a bit weirder, in the sense that the input JSON document has duplicate keys -- but possible, as the JSON specification doesn't disallow duplicate keys.

These two examples should hopefully make it clear that kvitems's behavior should be to traverse the input document until exhaustion, as there's no way for it to know in advance when all matches for a given prefix have been found. Hence this is not really a bug, but a case of wrong expectations.

However....

I do see the opportunity to implement a new option for kvitems (e.g. stop_after_first or similar). Using this option, users could indicate that they know the prefix appears only once in the document. The option would default to False for the reasons explained. When set to True, kvitems would behave basically as you describe, yielding the key/value pairs off the prefix the first time around and then exiting.

Thoughts?

from ijson.

@rtobar thanks for your answer. Indeed, I understand your point and agree with the stop_after_first option as a solution. That would be great !

Not sure how to implement though :)

from ijson.

@rtobar thanks for your answer. Indeed, I understand your point and agree with the stop_after_first option as a solution. That would be great !

Not sure how to implement though :)

Yeah, I have some ideas but it could take a while. I'll close this ticket and create a new one to keep track of the potential new feature.

from ijson.