tidyjson 0.3.1

Bug fixes

Add tbl_json methods for join to drop the tbl_json class early. If you need the ..JSON column, ensure to save it with json_get_column() before joining.
Work around an issue with dplyr 1.0.0 and [ not subsetting properly with transmute

Please let us know if you run into any errors of the form

The `[` method for class <tbl_json/tbl_df/tbl/data.frame> must return a data frame with 1 column

We will need to do some more work on how we manage the ..JSON column.

Instead of attr(., "JSON"), the JSON object is now a hidden column (..JSON). To prevent future backwards incompatibilities of this nature, there is now an “extractor” function to pull the raw JSON object off of the tbl_json: json_get(). You can also use json_get_column() to add the raw json onto your tbl_json as a dedicated column
Related to this, selecting a ..JSON column with dplyr::select() will mostly ignore you for complicated reasons. Use json_get_column() if you want to access the raw ..JSON data.
As always, if you want to remove tidyjson “magic”, tibble::as_tibble() will drop the tbl_json class and you are back to normal!

Add a few generics to make behavior generally more consistent: $<-.tbl_json, etc.
Add as_tbl_json as a future replacement for as.tbl_json
Add as_tbl_json.list so that you can more easily parse the JSON outside of tbl_json if you like. Further, this allows tbl_json to work with any arbitrary nested list. (#119)

Add bind_rows() support. Though currently not an S3 implementation, it behaves as much like the dplyr variant as possible, preserving the attr(.,'JSON') components if all components are tbl_json objects. (#58)

“Using Multiple APIs” vignette added to show support for using tidyjson with multiple APIs (#85)
Updated README.md to better explain spread_all() (#92)

Improve compatibility with newer dplyr and tidyr
DROP=TRUE caused an error. Altered behavior to be consistent with tbl_df
Fix spread_all(recursive=FALSE) bug that caused an error (#65)
Alter spread_all() behavior to recursively check for deduplication of names (and thus avoid an error) (#76)
Add named support for the NSE versions of dplyr functions (filter(),mutate(),slice(), etc.) since the SE variants are no longer called behind-the-scenes since dplyr 0.6.0. (#97)
Fix errors with print.tbl_json() when the JSON attribute is missing
Fix json_structure() failure if document.id missing by imputing the missing document.id. (#86)

json_complexity() computes the “complexity” (recursively unlisted length) of JSON data (#5)
json_structure() recursively structures arbitrary JSON data into a single data frame (#2)
json_schema() creates a schema for a JSON document or collection (#12)
is_json functions for testing JSON types, such as is_json_string(), is_json_null() or is_json_object() (#39)
spread_all() spreads all scalar values of a JSON object into new columns (#56)
as.character.tbl_json() converts tbl_json objects back into JSON character strings (#62)
gather_object() replaces gather_keys(), with default column.name of name instead of key (#66). This more closely matches the JSON standard, which refers to objects as name-value pairs, and is now consistent with gather_array().

“Using Multiple APIs” vignette added to show support for using tidyjson with multiple APIs (#85)
Updated README.md to better explain spread_all() (#92)
“Visualizing JSON” vignette for visualizing the structure of complex JSON data, like the companies example (#4)
Significant updates to all documentation and examples for clarity (#42)
Updated “Introduction to tidyjson” vignette to be more concise and use new functionality (#74)

enter_object and the jstring, jnumber and jlogical functions now accept unqoted strings to specify their path (#26)
tbl_json objects now print with a tidy character representation of the JSON attribute (#61)
Use purrr for most list based internal operations (#1)
Use tidyr for gather_array and gather_object functions (#28)
Imported the magrittr %>% operator (#17)
Fixed dplyr::slice() not working correctly with tbl_json objects (#18)
First argument to verbs is .x rather than x to avoid name conflicts in NSE (#23)
Fixed spread_values() to not coerce types (#24)
gather_array() and gather_object() can be called repeatedly in the same pipeline with the same column.name argument, and will simply append an integer identifer to the new columns (#38)