#consensus model with group to individual distance as a consensus measure
using JuMP
using MathOptInterface
using GLPK
using DelimitedFiles

function linear_MCC_distance_collective(opinion, n, p, l, epsilon, consensus, cost, weight)
    
    #solver declaration
    m = Model(GLPK.Optimizer)

    # decision variables

    # changed opinions
    @variable(m, o[1:n,1:p] >= 0)

    # variable for converting the modulus
    @variable(m, d[1:n,1:p] >= 0)
    @variable(m, x[1:n,1:p])

    #variable for the group opinion
    @variable(m, g[1:p] >= 0)

    #variable for consensus constraint
    @variable(m, y[1:n,1:p])
    @variable(m, z[1:n,1:p] >= 0)

    #add constraints for the model one by one

    # epsilon constraints
    for i = 1:n
        for j = 1:p
            @constraint(m, z[i,j] <= epsilon)
        end
    end

    #add equality constraint
    for j=1:p
        @constraint(m, g[j] - sum(o[i,j]*weight[i] for i = 1:n) == 0)
    end

    # equality constraints on the x
    for i = 1:n
        for j = 1:p
            @constraint(m, opinion[i,j] - o[i,j] == x[i,j])
            @constraint(m, x[i,j] <= d[i,j])
            @constraint(m,-x[i,j] <= d[i,j])
        end
    end

    # consensus constraints
    for i = 1:n
        for j = 1:p
            @constraint(m, o[i,j] - g[j] == y[i,j])
            @constraint(m, y[i,j] <= z[i,j])
            @constraint(m,-y[i,j] <= z[i,j])
        end
    end

    #consensus constraints
    @constraint(m, (2/(p*(p-1)))*sum(sum(weight[i]*(z[i,j]) for j = 1:p) for i = 1:n) <= consensus)

    #objective function
    @objective(m, Min, sum(sum(cost[i]*(d[i,j]) for j = 1:p) for i = 1:n))

    #solve the model
    optimize!(m)
end

s = read("linear_MCC_data.txt", String)
data = split(s, "&")

opinion = readdlm(IOBuffer(data[1]),',',';')
dimension = size(opinion)
n = dimension[1]
p = dimension[2]
l = parse(Int64, ARGS[1])

epsilon = parse(Float64, ARGS[2])
consensus = parse(Float64, ARGS[3])

cost = readdlm(IOBuffer(data[2]),',')
weight = readdlm(IOBuffer(data[3]),',')

linear_MCC_distance_collective(opinion, n, p, l, epsilon, consensus, cost, weight)