The hash package provides a fully-functional hash/dictionary for R. It has richer features and finer control than using native R structures like lists or environments and has as a user-friendly interface.
Performance-wise it has similar and sometimes better performance than these structures.
Latest Release:
install.packages('hash')Development Version:
install.packages('devtools')
devtools::install_github('decisionpatterns/hash')# Create a hash
h <- hash(a=1, b=2, c=3)
h <- hash( letters[1:3], 1:3 )
h <- hash( list(a=1,b=2,c=3) )
# Keys
keys(h)
# Values (named list)
values(h)
# Assign to single key hash
h$a <- "foo"
h[['a']] <- "bar"
# Slice
h[ c('a','c') ]## NAMED ACCESS/REPLACEMENT:
h$x : returns value of key `x`;
h$x <- value : sets key `x` to value;
h$x <- NULL : deletes key-value pair `x`
## INTERPRETED ACCES/REPLACEMENT:
h[[x]] : returns value of key x; `x` is interpreted.
h[[x]] <- value : sets the values of key `x`; `x` is interpreted.
h[[x]] <- NULL : deletes key-value pair `x`; `x` is interpreted.
## HASH SLICING:
h[] : returns a copy of h, same as `copy(h)`
h[x] : a hash slice of keys
h[] <- value : error, undefined key
h[x] <- value : set values for keys `x` to `value`(s)
h[x] <- NULL : delete keys `x`
KEYS must be a valid character value and may not be the empty string (““). Keys must be unique.
VALUES can be any R value, vector, object, etc.
Hashes probably work about how you would expect, but since there are built from R’s native environments. There are three things to Remember:
PASS-BY REFERENCE. hashes are environments, special objects in R where only one copy exists globally. When passed as an argument to a function, no local copy is made and any changes to the hash in the functions are reflected globally, i.e. in the caller’s namespace.
PERFORMANCE. Hashes are designed to be exceedingly fast using R environment’s internal hash table. The hash function is not without its cost. For small data structures, a named lists and vectors will out-perform a hash in nearly every case. After approximately 500+ elements, the performance of the hash becomes faster than native lists and vectors.