Apparently .j.k loses precision with large enough numbers:
q)`long$.j.k"1471220573128024107"
1471220573128024064
However I’m trying to interface with a program that expects the number to round-trip, i.e. if I export the JSON again, it should contain exactly 1471220573128024107, so I can’t use .j.k to import the file as it loses precision. There are parsers for other languages that can convert to long integers when a number is encountered and only force a conversion to float when there is a decimal point or exponent in the JSON.
Is there any existing library that can do the same for q?
A user has written qrapidjson which implements .j.j but not .j.k
The underlying rapidjson library does allow a parse flag kParseNumbersAsStringsFlag
http://rapidjson.org/namespacerapidjson.html#a81379eb4e94a0386d71d15fda882ebc9
Although in a quick test the library looks to already parse numbers without decimal places as longs by default:
q)rapidjson:.p.importrapidjson q)rapidjson[:loads][“1471220573128024107”] 1471220573128024107 q)type rapidjson[:loads][“1471220573128024107”] -7h q)type each rapidjson[:loads][“[1471220573128024107, 1471220573128024107.0]”] -7 -9h //Same for default json library in python q)json:.p.importjson q)json[:loads]["1471220573128024107"] 1471220573128024107 q)type json[:loads]["1471220573128024107"] -7h