- Updated format of documentation to pass CRAN checks
- Updated
getRetrosheet()
to account for new paths on retrosheet.org - Updated two tests to account for changed location of All Star Team rosters
- Updated tests to account for addition of All Star teams to Retrosheet data
- Added checks to testing so it 'fails gracefully' when
retrosheet.org
is not available - Improved documentation in tests
- Updated tests to skip Rmarkdown tests on CRAN. Avoids pandoc dependency issues on Solaris and MacOS CRAN package checks.
- Updated tests to account for
stringsAsFactors = FALSE
- Switched to making requests via https to align with retrosheet.org changeover
- Updated
README.md
to account formaster
tomain
change in repo
- Replaced kludge-y
closeAllConnections()
withclose()
calls to specific connections (#1) - Added some rudimentary data cleanup to the
get_retrosheet()
wrapper function- Dates in retrosheet data now return date-types, everything else is converted from
chr
via the relatively aggressivetype.convert()
- Dates in retrosheet data now return date-types, everything else is converted from
- Removed
data.table
dependency, replaceddata.table::fread()
withbase::read.csv()
calls get_retrosheet()
now returns regular old dataframes now, instead of tibbles.- Added better handling of HTTP requests. Failed requests are now retried up to three times via
httr::RETRY()
- Added
get_retrosheet()
, a wrapper forgetRetrosheet()
that takes the same arguments and returns the same data, except coercing matrices into tibbles. - Updated tests to account for more data being added to retrosheet.org (whoops)
- Schedule scraping now expects a
*.zip
file instead of*.txt
file, to accommodate change to retrosheet.org file structure - Added
cache
parameter to getRetrosheet(). This defaults to NA. If set to a local directory path, retrosheet data will be cached locally and re-used so as not to constantly download the same file from the retrosheet.org servers - Fixed parsing of substitutions in games where no substitutions occurred. This is a rare edge case.
- Removed
RCurl
dependency (due to incompatibility with TLS > 1.0). Web data is now downloaded usinghttr
functions - Removed
XML
dependency, replaced withxml2
andrvest