- NOTE: this is a major release and removes some functionality
- Update for new corpuses
- Remove the add_count option (no data provided by Google)
- Remove drop_corpus (Google no longer supports the :corpus operator for the current corpus)
- Fix package after more changes to the Google Ngram Viewer website
- Fix package after latest changes to the Google Ngram Viewer website
- Improved error handling
- Suppress testing of all examples that make internet calls (fail gracefully)
- Handle offline state
- Skip testing if offline
- Incremented version to reflect that 1.7.7 was a major release
- Rolled back use of |> for compatibility with earlier versions of R
- Update for changes in ngram viewer website
- New corpus names (e.g. eng_2019 changed to en_2019)
- Drop use of lifecycle badges
- Add markdown format NEWS file
- Tidied fromJSON call
- Started to use lifecycle in documentation (ngrami)
- Imposed version dependency for dplyr to ensure relocate available
- Updated documentation to provide details of return values
- Change download code to use 'url' to ensure code works behind a proxy server
- Addressed CRAN submission requirements
- Change year_start default to 1800 in documentation
- Comprehensive refactor of underlying code
- More robust error/warning handling
- Dropped the "tag" argument from ngram functions
- Fix case_sensitive attribute
- Fix error in corpus count dataset
- Update to address issue (#26) resulting from change in the format of Google Ngram Viewer webpage
- Incorporated pull changes #22, @seancarmody
- Make wildcard searches expand to all terms
- Error out on server answer "Please try again later." 1a655f3
- Fix setting default corpus. 0b22dc4
- scale functions: do not explicitly set name, allow overwrite. 3a21061
- Allow passing through additional parameters to ngram_single. ac6b1cc
- For wildcard searches, drop the cumulated (All) column
- Added travis-ci testing
- Fixed problems with (some) advanced operators
- Removed debugging from ngrami
- Fixed the Pulser bug
- Fix accented character encoding problem on Windows
- Improve ssl handling (refer Hadley's comment here: http://www.statsravingmad.com/blog/statistics/a-tiny-rcurl-headache/)
Google has switched to SSL for the N-gram viewer and the format of the web-pages has changed. This means that earlier versions of the package are completely broken. This release fixes this major problem.
- Add README.md to .Rbuildignore to remove from CRAN
- Fix count for n-grams with n>1, including a "fudge" for 2012 corpuses
- Add option to display long-form corpus name
- Warn about smoothing >0 for geoms other than "line"
- Tidy documentation for print.ngram
- ngram and ngrami return S3 class "ngram"
- Format print for ngram objects
- ggram can take either a list of phrases or an ngram object
- Add option to relabel y-axis
- Add word counts option to ngram
- Change ggplot2 and scales from Requires to Suggests
- Prevent use of complex operators in case insensitive searches
- Warn about character substitution
- CRAN release version
- More efficient handling of escaped Unicode (thanks Hadley http://stackoverflow.com/a/17787736/1543437)
- Fix package checking problems associated with plyr
- Tidy Google theme
- First semi-official release. All future development moved to the 'develop' branch.
- Allow case insensitive plotting with ggram
- Avoid reshape/reshape2 conflicts (thanks to Francois Briatte)
- Pass arbitrary geoms to
ggram - New function
ngramwto return results in "wide" format - Removed
wideoption fromggramandggrami - Better handling of legends when
ignore_case = TRUE - Error trapping long phrase lists
- Google theme option
- Added plot wrapper ggram
- Detect invalid corpus names
- Initial release of the ngramr package