arbenson

using SparseArrays

function find_triangles(A::SparseMatrixCSC{Int64,Int64})
    n = size(A, 1)  # number of nodes
    d = vec(sum(A, dims=2))  # degree vector
    deg_order = zeros(Int64, n)
    deg_order[sortperm(d)] = 1:n
    triangles = []
    for i = 1:n
        N_i = findnz(A[:, i])[1]  # neighbors of node i

arbenson

using SparseArrays, Random
function UMVC(A::SparseMatrixCSC{Int64,Int64},
              ncovers::Int64=300)
  edges = filter(e->e[1] < e[2],
                 collect(zip(findnz(A)[1:2]...)))
  umvc = zeros(Int64,size(A,1))
  for _ in 1:ncovers
    # Run 2-approximation with random edge ordering                                                                                                                                       
    vc = zeros(Int64, size(A,1))
    for (i, j) in shuffle(edges)

arbenson

using LightXML

function parse_tags(filename::AbstractString)
    tag_map = Dict{String, Int64}()
    function get_tag_key(tag::AbstractString)
        if !haskey(tag_map, tag)
            n = length(tag_map) + 1
            tag_map[tag] = n
            return n
        end

arbenson

using LinearAlgebra
function tensor_apply(T::Array{Float64,3}, x::Vector{Float64})
    n = length(x)
    y = zeros(Float64, n)
    for k in 1:n; y += T[:, :, k] * x * x[k]; end
    return y
end
function tensor_collapse(T::Array{Float64,3}, x::Vector{Float64})
    n = length(x)
    Y = zeros(Float64, n, n)

arbenson

function UMVC(A::SparseMatrixCSC{Int64,Int64}, ncovers=300)
  edgevec = filter(e -> e[1] < e[2],
                   collect(zip(findnz(A)[1:2]...)))
  umvc = zeros(Int64, size(A,1))
  for _ in 1:ncovers
    # Run 2-approximation with random edge ordering                                                                                                                                  
    cover = zeros(Int64, size(A,1))
    edge_queue = shuffle(edgevec)
    while length(edge_queue) > 0
      i, j = pop!(edge_queue)

arbenson

using TimeZones

function parse_timestamp(timestamp::String)
    ts = strip(timestamp)
    if length(ts) < 10; return -1; end
    if ts[1] == '"' && ts[end] == '"'; ts = ts[2:(end - 1)]; end
    try @show ts; catch; return -1; end

    # Truly terrible cases that we give up on
    if contains(ts, "???"); return -1; end

arbenson

# Find a motif-based cluster for any directed triangle motif.                                                                                                                  
function MotifSpectralClust(A::SparseMatrixCSC{Int64,Int64}, motif::AbstractString)
    # Form motif adjacency matrix                                                                                                                                              
    B = min.(A, A')  # bidirectional links                                                                                                                                     
    U = A - B        # unidirectional links                                                                                                                                    
    if     motif == "M1"
        C = (U * U) .* U'
        W = C + C'
    elseif motif == "M2"
        C =  (B * U) .* U' + (U * B) .* U' + (U * U) .* B

arbenson

# Find a motif-based cluster for any directed triangle motif.                                                                                                                        
function MotifSpectralClust(A::SparseMatrixCSC{Int64,Int64}, motif::AbstractString)
    # Form motif adjacency matrix                                                                                                                                                    
    B = min.(A, A')  # bidirectional links                                                                                                                                           
    U = A - B        # unidirectional links                                                                                                                                          
    if     motif == "M1"
        C = (U * U) .* U'
        W = C + C'
    elseif motif == "M2"
        C =  (B * U) .* U' + (U * B) .* U' + (U * U) .* B