"records are values not objects" on valid JSON

[smithjd]

This snippet of valid JSON results in "records are value not objects":

library(dplyr)
library(tidyjson)

json <- '
    [{"country":"us","city":"Portland","topics":[{"urlkey":"videogame","name":"Video Games","id":4471},{"urlkey":"board-games","name":"Board Games","id":19585},{"urlkey":"computer-programming","name":"Computer programming","id":48471},{"urlkey":"opensource","name":"Open Source","id":563}],"joined":1416349237000,"link":"http://www.meetup.com/members/156440062","bio":"Analytics engineer.  Primarily work in the Hadoop space.","lon":-122.65,"other_services":{},"name":"Aaron Wirick","visited":1443078098000,"self":{"common":{}},"id":156440062,"state":"OR","lat":45.56,"status":"active"}]
    '
    json %>% as.tbl_json %>% gather_keys

The issue is posted on Sack Overflow: http://stackoverflow.com/questions/34624172/what-does-records-are-values-not-objects-mean-in-tidyjson

According to http://jsonlint.com/ the JSON is valid.

[colearendt]

Posted answer on SO:

As mentioned in one of the comments, gather_keys is looking for objects, where you have an array. What you should probably be using here is gather_array.

Further, the other answer (on SO) uses a more brute-force approach to parsing the JSON attribute that the tidyjson package creates. tidyjson provides methods for dealing with this in a bit cleaner pipeline if desired:

library(dplyr)
library(tidyjson)

json <- '
[{"country":"us","city":"Portland"
,"topics":[
 {"urlkey":"videogame","name":"Video Games","id":4471}
 ,{"urlkey":"board-games","name":"Board Games","id":19585}
 ,{"urlkey":"computer-programming","name":"Computer programming","id":48471}
 ,{"urlkey":"opensource","name":"Open Source","id":563}
]
,"joined":1416349237000
,"link":"http://www.meetup.com/members/156440062"
,"bio":"Analytics engineer.  Primarily work in the Hadoop space."
,"lon":-122.65,"other_services":{}
,"name":"Aaron Wirick","visited":1443078098000
,"self":{"common":{}}
,"id":156440062,"state":"OR","lat":45.56,"status":"active"
}]
'

mydf <- json %>% as.tbl_json %>% gather_array %>% 
spread_values(
 country=jstring('country')
 , city=jstring('city')
 , joined=jnumber('joined')
 , bio=jstring('bio')
) %>% 
enter_object('topics') %>% 
gather_array %>%
spread_values(urlkey=jstring('urlkey'))

This pipeline really shines if there are multiple such objects in the array. Hope that is helpful, even if very long after the fact!