Problem with Nest() 1.0.0. and subsequent jsonlite::toJSON

[skr5k]

Hey,

I use nest() a lot for creating JSONs from dataframes to interact with an API. Here's an example:

require(tidyverse)
require(lubridate)
require(tidyr)
require(jsonlite)

df = data.frame(id = "xyz",
                time =  seq(from = ymd_hms("2020-10-10 00:00:00", tz = "GMT"),
                            to = ymd_hms("2020-10-10 23:00:00", tz = "GMT"),
                            by="hours") %>% as.integer(),
                value = 1:24)

df %>% nest_legacy(time, value) %>% toJSON(pretty=FALSE, auto_unbox = FALSE)

This returns:

[{"id":"xyz","data":[{"time":1602288000,"value":1},{"time":1602291600,"value":2},{"time":1602295200,"value":3},{"time":1602298800,"value":4},{"time":1602302400,"value":5},{"time":1602306000,"value":6},{"time":1602309600,"value":7},{"time":1602313200,"value":8},{"time":1602316800,"value":9},{"time":1602320400,"value":10},{"time":1602324000,"value":11},{"time":1602327600,"value":12},{"time":1602331200,"value":13},{"time":1602334800,"value":14},{"time":1602338400,"value":15},{"time":1602342000,"value":16},{"time":1602345600,"value":17},{"time":1602349200,"value":18},{"time":1602352800,"value":19},{"time":1602356400,"value":20},{"time":1602360000,"value":21},{"time":1602363600,"value":22},{"time":1602367200,"value":23},{"time":1602370800,"value":24}]}] 

When I run instead:

df %>% nest(data = c(time, value)) %>% toJSON(pretty=FALSE, auto_unbox = FALSE)

I get an error saying No method asJSON S3 class: vctrs_vctr.

It seems to me that the new nest creates an object of class vctrs_vctr (which I dont know) inside the data.frame instead of using a list as a container for the nested elements. In case I got that right, can I make nest create a list or alternatively, how can I have the nest() function output something that's digestable for toJSON?

Thanks a lot in advance!
skr5k

[hadley]

Same as jeroen/jsonlite#292 (although it's not clear where this will need to be fixed yet)

[SymbolixAU]

A work-around for jsonlite is to use force = TRUE when it doesn't recognise the class

df %>% 
  nest(data = c(time, value)) %>% 
  toJSON(pretty=FALSE, auto_unbox = FALSE, force = TRUE)

# [{"id":"xyz","data":[{"time":1602288000,"value":1} ... etc

And also jsonify works too

df %>% 
  nest(data = c(time, value)) %>% 
  jsonify::to_json( unbox = T )

# [{"id":"xyz","data":[{"time":1602288000,"value":1} ... etc

[hadley]

Should now be resolved in dev tidyr via changes to dev vctrs.