JSON to Table Can't handle large floats

KNIME Analytics Platform
1

I am running into a problem using ‘JSON to table’ to parse large numbers. If I try to parse a number greater than 1E18 (for example 1E20, 1E30, etc.), I get an error:

ERROR JSON to Table 0:623 Execute failed: For input string: “10000000000000000000”

You can try to reproduce this by creating the workflow shown in the image below and putting the following in cell A1
{"num": 1E30}

image

I’m not sure if this is a bug or a natural limitation of KNIME. In either case, does anyone have a workaround?

2

Hi @joshuahoran,

The numbers parsed by the JSON to Table node are saved into a long, the maximum number presentable by a long in java is 9,223,372,036,854,775,807 which is just short of 1E19.

One workaround I found, but I am unsure if it will help you is the following: By wrapping the number in quotation marks you can force it to be interpreted as a string: {"num": "1E30"} then it can be transformed into a table without any issue.

best,
Gabriel

3

Thanks for the details Gabriel. It looks like the solution is to use String Replacer to modify the JSON. I will give this a try. Given that it seems unsavory to have to hack our way around valid JSON, is this worth promoting as a new feature request for the JSON node set?

4

Hi guys, I have the same issue regarding “trimming” the value that comes from JSON. My question is, is it possible to have this implemented or fixed in KNIME JSON to Table node?

5

I just made a test using JSON Path, the problem is the same. It trims the value.

6

Hi there!

Tnx for pointing this out. It is reported and should be taken care of. Stay tuned :wink:

Br,
Ivan

1 Like
7

I got the point. It is a matter of selecting the right renderer on node column. Right click over the column and select the “full precision” renderer.
This kind of useless functionality should be documented anywhere …