- fix link with 302 code for CRAN checks
- fix documentation to avoid CRAN NOTEs
- handle invalid JSON in old CZSO API
- hard-deprecate old functions; they will be removed in the next version
- update Roxygen2 version and rebuild documentation to handle CRAN validation of HTML manual
- removed usethis dependency; messages now done by {cli}
- removed stringr dependency
- readded stringi dependency
- On MacOS Monterey, native
curl cannot reach https://www.czso.cz/, making it impossible for R to reach it. So on Monterey machines with the default curl configuration, all functions accessing the server will fail with a message instructing the user to set an option in .Renviron to resolve this issue. This is a temporary situation until the bug is fixed in MacOS or a fix is found in R that does not require the user to set environment variables.
- remove the unnecessary {stringi} dependency
- update tests to fix CRAN failure due to unreachable provider server
- fixed minor bug in how multi-file archives are handled
NOTE: there has been an unexpected change to the National Open Data Catalogue which caused problems with the dataset IDs returned by czso_get_catalogue(). This release provides a temporary fix based on a patch to the API kindly provided by the Czech Statistical Office.
Future releases will handle the change in a more robust way once the final form of the catalogue API is determined by the CZSO.
Fixes to code that is currently not used but may be again as the dust settles:
- fix SPARQL query to return all datasets, incl. 2 that were not showing up compared to the official data catalogue web interface
- return label, rather than IRI, for temporal resolution, in the output of
czso_get_catalogue()
- minor fixes for CRAN checks
- minor README edit for CRAN resubmission
- minor README edits for CRAN resubmission
New features
- codelists (číselníky) can now be retrieved with
czso_get_codelist(). This includes hierarchies between codelist items, and English versions where available
Improvements
- new
dest_dir parameter in czso_get_[table|codelist]() allows you to store downloaded data files in a custom location. This location can be set at script or user (.Rprofile) level by setting the czso.dest_dir option.
- improved UI: more informative and better formatted messages, hints and reminders about common mishaps
- documentation added in
czso_get_table() on where to get definitions of indicators and variables used by CZSO.
- more date-type columns are parsed and typed correctly
Bug fixes
- deprecated functions no longer used internally
- encoding guess no longer breaks on Linux (@jlacko)
Deprecations of functions with old names
- functions named
get_czso_* are now deprecated and will emit a warning if used. Use czso_*() instead.
Fixes responding to CRAN feedback
- Description field of DESCRIPTION now contains a link to the data provider
- documentation in all functions now provide more detail on what the functions return
Prep for CRAN
- add cran-comments.md
- fixed dplyr-related CHECK NOTE
- updated LICENSE for CRAN
- update URL in README
Bug fixes and minor improvements
- all functions accessing CZSO data now return helpful error if the dataset cannot be found
- fixed deprecation warnings to display correct package name
- added helpful error message for no access to the internet
- spelling corrections
- better error output when CZSO server returns error
New function names and deprecations
- all user-facing functions are now
czso_* to avoid conflicts and aid discovery via auto-complete. Original functions are soft-deprecated.
Improvements
- improvements to documentation
- added code of conduct and contributing guide
New features
- new
get_dataset_doc() function for accessing documentation
- new
get_table_schema() function for retrieving table schema
- exported
get_dataset_metadata() function for accessing detailed metadata
Improvements
- get_czso_catalogue() is now much faster as it uses the open data catalogue’s API instead of downloading a huge CSV list of all datasets. It is less flexible as it does not allow direct filtering.
- handle encoding of some older datasets, which may not be UTF-8
- relaxed stringi version requirement to make Win build work
Deprecated functions
- both exported functions are renamed to
get_czso_catalogue() and get_czso_table() to avoid clashes with other packages; original functions are soft-deprecated and will be removed in future versions.
Bug fixes
- fix bug in
get_czso_catalogue() where an empty tibble was returned because the source CSV was read incorrectly (due to vroom handling of newlines inside fields)
- first version with functioning core workflow
- Added a
NEWS.md file to track changes to the package.