getSourceAsString() for doc inserted as SMILE fails, auto convert to JSON

[andrewclegg]

In a unit test, I have a document containing one field "val" which is a list of floats:

{ Float.MIN_VALUE, Float.MIN_VALUE - 1, Float.MAX_VALUE, -Float.MIN_VALUE }

The JSON representation of this is:

{"val":[1.4E-45,-1.0,3.4028235E38,-1.4E-45]}

getSourceAsString() on the corresponding GetResponse object returns this successfully when I inserted the document using XContentFactory.jsonBuilder(), but when I inserted the document using XContentFactory.smileBuilder(), I got:

java.lang.ArrayIndexOutOfBoundsException: 87
at org.elasticsearch.common.Unicode.UTF8toUTF16(Unicode.java:190)
at org.elasticsearch.common.Unicode.unsafeFromBytesAsUtf16(Unicode.java:106)
at org.elasticsearch.common.Unicode.fromBytes(Unicode.java:80)
at org.elasticsearch.index.get.GetResult.sourceAsString(GetResult.java:203)
at org.elasticsearch.action.get.GetResponse.sourceAsString(GetResponse.java:150)
...

Will attach a demonstration case when I have a chance.

[kimchy]

Yea, that will happen..., wondering what should be done here, obviously, smile can't be converted to string, maybe we should do auto conversion to json and then return it as string?

[andrewclegg]

Aaah okay. Won't need a demo case if it's known behaviour :-)

Auto conversion would work (if it's not super heavyweight?), but even some sort of NotSupportedException that told you you couldn't do this with SMILE docs would be useful.

[kimchy]

I think I will add auto conversion, in any case, you are usually better off using the bytes and not the string itself in your app.

[andrewclegg]

We're not actually even using the source, this was just in a defunct test that died when I switched to SMILE. Thanks for quick feedback!