How to convert R dataframe to Json in name/value pair?

Question

I have the following R data frame:

        Values
Type1   123
Type2   4565
Type3   7812

I expect the JSON output to be

{"Type1":123, "Type2":4565, "Type3":7812}

The number field can w/wo quote

I used jsonlite toJSON, the output is:

[{"Values":123,"_row":"Type1"}, 
 {"Values": 4565,"_row":"Type2"}, 
 {"Values": 7812,"_row":"Type3"}]

Try `toJSON(as.list(setNames(df$Values,rownames(df))),auto_unbox=TRUE)`. — nicola, Sep 21 '17 at 18:53

score 7 · Accepted Answer · answered Sep 21 '17 at 21:00

7

Solution using rjson:

df <- data.frame(Values = c(123, 4565, 7812))
rownames(df) <- paste0("Type", 1:3)

library(rjson)
toJSON(setNames(df$Values, rownames(df)))

[1] "{\"Type1\":123,\"Type2\":4565,\"Type3\":7812}"

answered Sep 21 '17 at 21:00

pogibas

27,303
19
84
117

Seems jsonlite drop the name in the same "toJSON" command. I will try to use rjson then. – BlueDolphin Sep 22 '17 at 17:27
1

BTW, the string can convert back to data frame using t(fromJSON("{\"Type1\":123,\"Type2\":4565,\"Type3\":7812}")) – BlueDolphin Sep 22 '17 at 17:31

score 2 · Answer 2 · answered Sep 25 '17 at 19:19

jsonlite is actually preserving your data structure, i.e., keeping your rownames (Type1, Type2, Type3).

Anyway, using jsonlite you could get the same behaviour with:

> jsonlite::toJSON(df %>% t() %>% tibble::as_data_frame())
[{"Type1":123,"Type2":4565,"Type3":7812}]

Please realize that with this solution you will lose the original column name Values. If the row names are important but not the column name, you should think about defining your data in a different way, as dealing with rownames can get messy. You can add the Type as a second column or transpose your data -- one row, as many columns as types.

How to convert R dataframe to Json in name/value pair?

2 Answers2