CODEWITHSUNDEEP
rjsondata.table
Artem Klevtsov
Artem Klevtsov

Reputation: 9423

Write and load JSON string with data.table

I tried to save JSON string to CSV with fwrite. But I can't load it back with fread.

Question: What the quotes settings should I use here?

library(data.table)
library(jsonlite)

r <- list(a = "text")
d <- data.table(
  a = toJSON(r, auto_unbox = TRUE)
)

tmp <- tempfile()
fwrite(d, tmp, quote = TRUE, qmethod = "escape")
fromJSON(fread(tmp)[["a"]][1])

Code above produce error:

Error: lexical error: invalid char in json text.
                                      {\"a\":\"text\"}
                     (right here) ------^

Upvotes: 5

Views: 885

Answers (2)

Artem Klevtsov
Artem Klevtsov

Reputation: 9423

Solution with qmethod = "double" for the current version (1.12.3):

library(data.table)
library(jsonlite)

r <- list(a = "text")
d <- data.table(
  a = toJSON(r, auto_unbox = TRUE)
)

tmp <- tempfile()
fwrite(d, tmp, quote = TRUE, qmethod = "double")
fromJSON(gsub("\"\"", "\"", fread(tmp)[["a"]], fixed = TRUE)[1])

Upvotes: 4

Mikko Marttila
Mikko Marttila

Reputation: 11898

It seems you can't.

The docs for fread() clearly state that slash-escaped quotes should be handled in quoted columns:

Unescaped quotes may be present in a quoted field, e.g., ...,2,"Joe, "Bloggs"",3.14,..., as well as escaped quotes, e.g., ...,2,"Joe \",Bloggs\"",3.14,....

If this were the case, the method you are using should work. In fact, a search on the data.table GitHub repository reveals an open issue about this. In the mean time, you'll need to either fix the data after fread(), or use some other function to read the data.

Upvotes: 2

Related Questions