Persistent Weighted Dictionaries

Air supports, in additon to the standard persistent dictionary type PDict, a weighted dictionary PWDict (and PWIdDict). These types are broadly similar to PDict and PIdDict except that their entries additionally have non-negative numeric weights. These dictoinaries act as (1) dictionaries, (2) priority queues, and (3) discrete random variables.

Examples

Weights are required whenever an entry is inserted, so they are given as the second half of a nested pair, (key => value) => weight:

julia> using Air, Random

julia> d = PWDict{Symbol,Int,Float64}();

julia> d = push(d, (:a => 1) => 1.0)
PWDict(:a => 1)

julia> d = push(d, (:b => 2) => 5.0)
PWDict(:b => 2, :a => 1)

julia> d = push(d, (:c => 3) => 3.0)
PWDict(:b => 2, :c => 3, :a => 1)

A weighted dictionary behaves as an ordinary dictionary:

julia> d[:a]
1

julia> d[:missing]
ERROR: KeyError: key :missing not found

julia> length(d)
3

It also behaves as a priority queue, ordered by weight:

julia> first(d)          # the key-value pair with the highest weight
:b => 2

julia> pop(d)            # a copy without that entry
PWDict(:c => 3, :a => 1)

julia> d                 # the original is unchanged
PWDict(:b => 2, :c => 3, :a => 1)

julia> collect(d)        # iterates from highest to lowest weight
3-element Vector{Pair{Symbol, Int64}}:
 :b => 2
 :c => 3
 :a => 1

Weights can be inspected and changed, and they determine the distribution used by rand:

julia> getweight(d, :b)
5.0

julia> d = setweight(d, :a, 10.0)
PWDict(:a => 1, :b => 2, :c => 3)

julia> first(d)
:a => 1

julia> rand(d)           # :a is now ten times as likely as :c
:a => 1