- Added a
NEWS.md
file to track changes to the package. - bind_* functions no longer run clinspacy_init() -- this should speed up load times
- Added
NA
tosemantic_types
argument for bothclinspacy()
andclinspacy_single()
to prevent tokens from being discarded if they do not match a listed semantic type. - Moved
clinspacy_single()
logic intolapply()
instead of gradually building a list using afor
loop for boost in speed - Moved progress bar to
clinspacy()
so that it iterates over documents rather than tokens
- Changed lifecycle badge to stable
- Bumped version number to 1.0.0 since it's ready for CRAN submission
- Bug fix: removed unnecessary arguments and some clean up in prep for CRAN submission
- Fixed documentation prior to CRAN submission based on feedback
- Bug fix: Specified version numbers for spaCy (2.3.0), scispaCy (0.2.5), and medspaCy (0.1.0.2) to ensure that the versions are compatible with one another
- Bug fix: spaCy 2.3.0 must be installed from conda-forge (
pip
set toFALSE
) because the source fails to build properly on Windows even with Visual C++ build tools installed - Update: Switched to using medspaCy instead of its individual components because medspaCy 0.1.0.2 is compatible with spaCy 2.3.0 (an older version was not).
- Bug fix: changed
section_title
tosection_category
due to updates in medspaCy sectionizer API - Known issue: After first running
clinspacy_init()
on Windows, sometimes it cannot findnumpy
. This is a known issue withreticulate
rstudio/reticulate#367. Restarting the R session and re-runningclinspacy_init()
appears to fix the issue