How to unnest a dictionary from XML in R?

I am attempting to convert this xml to a dataframe in R:

https://d396qusza40orc.cloudfront.net/getdata%2Fdata%2Frestaurants.xml

library(xml2)
library(tidyverse)

fileurl <- "https://d396qusza40orc.cloudfront.net/getdata%2Fdata%2Frestaurants.xml"
xmllist <- as_list(read_xml(fileurl)) 
xml_df = tibble::as_tibble(xmllist) %>%
        unnest_longer(response)

row_wider = xml_df %>%
        unnest_wider(response)

row_df = row_wider %>%
        unnest(cols = names(.)) %>%
        unnest(cols = names(.)) %>%
        readr::type_convert()

The issue is that the 'location_1' column is a dictionary and shows up as NA when I unnest. How can I get each of the values of this dictionary into this column? Any help is much appreciated, thanks.

Solution

The location_1 column is an empty list (hence you get NAs) with two attributes human_address which is a JSON string and a logical needs_recoding. One option to get your desired result would be to first extract the content of these attributes and store them in a list. Afterwards you could use two unnest_wider to unnest the list column.

library(xml2)
library(tidyverse)

parse_location_1 <- function(x) {
  x$location_1 <- list(
    human_address = jsonlite::fromJSON(attr(x$location_1, "human_address")),
    needs_recoding = attr(x$location_1, "needs_recoding")
  )
  x
}

fileurl <- "https://d396qusza40orc.cloudfront.net/getdata%2Fdata%2Frestaurants.xml"
xmllist <- as_list(read_xml(fileurl))
xml_df <- tibble::as_tibble(xmllist) %>%
  unnest_longer(response) |>
  mutate(response = map(
    response, parse_location_1
  ))

row_wider <- xml_df %>%
  unnest_wider(response) |>
  unnest_wider(location_1) |> 
  unnest_wider(human_address)

row_df <- row_wider %>%
  unnest(cols = where(is.list)) %>%
  unnest(cols = where(is.list)) %>%
  readr::type_convert()
#> 
#> ── Column specification ────────────────────────────────────────────────────────
#> cols(
#>   name = col_character(),
#>   zipcode = col_double(),
#>   neighborhood = col_character(),
#>   councildistrict = col_double(),
#>   policedistrict = col_character(),
#>   address = col_character(),
#>   city = col_character(),
#>   state = col_character(),
#>   zip = col_logical(),
#>   needs_recoding = col_logical(),
#>   response_id = col_character()
#> )

head(row_df)
#> # A tibble: 6 × 11
#>   name   zipcode neighborhood councildistrict policedistrict address city  state
#>   <chr>    <dbl> <chr>                  <dbl> <chr>          <chr>   <chr> <chr>
#> 1 410      21206 Frankford                  2 NORTHEASTERN   4509 B… Balt… MD   
#> 2 1919     21231 Fells Point                1 SOUTHEASTERN   1919 F… Balt… MD   
#> 3 SAUTE    21224 Canton                     1 SOUTHEASTERN   2844 H… Balt… MD   
#> 4 #1 CH…   21211 Hampden                   14 NORTHERN       3998 R… Balt… MD   
#> 5 #1 ch…   21223 Millhill                   9 SOUTHWESTERN   2481 f… Balt… MD   
#> 6 19TH …   21218 Clifton Park              14 NORTHEASTERN   2722 H… Balt… MD   
#> # ℹ 3 more variables: zip <lgl>, needs_recoding <lgl>, response_id <chr>