Actor centrality in Julia

actor_centrality.jl

type Node
    name::UTF8String
    sn::Set{Node}
    n::Vector{Node} #this is assigned later, as an efficiently iterable copy of sn

    Node(name) = new(name, Set{Node}())
end

typealias Graph Dict{UTF8String, Node}

function get(G::Graph, name)
    if has(G, name)
        return G[name]
    end
    G[name] = Node(name)
end

function centrality_mean(G::Graph, start_node, next, nnext)
    dists = Dict{Node,Uint64}()
    del_all(next)
    del_all(nnext)
    push(next, G[start_node])

    cdist = 0
    while !isempty(next)
        del_all(nnext)
        for n in next
            if !has(dists, n) 
                dists[n] = cdist
                for neigh in n.n
                    push(nnext, neigh)
                end
            end
        end
        cdist += 1
        next,nnext = nnext,next
    end
    mean([ v for (k,v) in dists ])
end

function read_graph()
    G = Graph()
    actors = Set()

    open("imdb-1.tsv", "r") do io
        while !eof(io) 
            k = split(strip(readline(io)), "\t")
            actor, movie = k[1], join(k[2:3], "_")
            ac, mn = get(G, actor), get(G, movie)
            add(actors, actor)
            add(ac.sn, mn)
            add(mn.sn, ac)
        end
    end
    for (k,n) = G
        n.n = elements(n.sn)
        del_all(n.sn)
    end
    G, sort!([ a for a in actors])
end

function main()
    G, actors = read_graph()
    d = Dict{UTF8String, Float64}()

    n = Node[]
    nn = Node[]
    for a in actors[1:100]
        d[a] = centrality_mean(G, a, n,nn)
        print("$a: ", d[a], "\n")
    end

    vals = sort!([(v,k) for (k,v) in d])
    for i=1:20 
        print("$i: ", vals[i], "\n")
    end

    # print(centrality_mean(G, "Hoffman, Dustin"), "\n")
end


print("Elapsed: ", @elapsed main(), "\n")