Adds argument check functions, chk_*(). The family
covers the atomic types in both scalar and vector form
(chk_string(), chk_character(),
chk_number(), and so on), lists and composite objects
(chk_list(), chk_data_frame(),
chk_tibble(), …), classes and arbitrary conditions
(chk_class(), chk_true(),
chk_that()), and chk_any() for a requirement
that any one of several checks may satisfy. See
vignette("chk"), and ?chk_atomic,
?chk_composite and ?chk_other.
Adds glue_vector(), which applies a
glue template element-wise to a vector. glue()
and glue_data() are re-exported alongside it.
Adds asciify(), which transliterates accented
characters to ASCII.
Adds yencode() and ydecode(), which
encode strings into a restricted character set and back.
Fixes NOTE regarding pipes and dependency on R 4.1.
In lookup() and lookuper(), the
default argument is renamed to .default and
follows ..., so it must be named. Calls that pass
default by name or by position now raise an error.
lookup() no longer accepts factors, either as
x or as columns of the lookup table. Previously, a factor
x had its levels looked up. Use
levels(x) <- lookup(levels(x), lookup_table) for that
behavior.
zingle() no longer has an na.rm
argument. Missing values are handled by na.ok.partial,
na.ok.all and empty.ok instead. The closest
equivalent of na.rm = TRUE is
na.ok.partial = TRUE, na.ok.all = TRUE, empty.ok = TRUE,
except that NaN now counts as a value rather than as
missing. An empty vector is now an error unless
empty.ok = TRUE; previously it returned
NA.
ll_assert_labelled() is renamed to
ll_chk_labelled(), following the chk_ naming
used for the new check functions. The old name is removed.
Various fixes and improvements to zeq(), zample() and zingle().
wrap_error() now honors its wrap
argument correctly.
Adds the notate() function.
Adds a default parameter to the lookup() and lookuper() functions.
lookup() and
lookuper(), designed to look up values from a lookup table
in a way that works smoothly in an R workflow, in particular within
pipes.zample(), zeq(), and
zingle(), for safer sampling, sequencing and subsetting in
a data processing workflow