require 'statsample'
require "rinruby"
R.eval("library(\"Kendall\")")
R.eval("library(\"irr\")")
#test = ARGV[0]
#auto = ARGV[1]

def stats(input, type)
    if type == "hash"
        sent_array = input.values
    elsif type == "array"
        sent_array = input
    end
    sent_sum = 0.0
    sent_array.each do |sent|
        sent_sum += sent
    end
    mean = sent_sum/sent_array.length
    
    sumsq = 0.0
    sent_array.each do |sent|
        sumsq += (mean - sent)*(mean - sent)
    end
    sd = Math.sqrt(sumsq/(sent_array.length - 1))
    se = sd/Math.sqrt(sent_array.length)
    return mean, sd, se
end


def readfromfile(filename,sentindex,testindex)
    file = File.open(filename,"r:utf-8")
    if filename[-3..-1] == "csv"
        separator = ","
    elsif filename[-3..-1] == "tsv"
        separator = "\t"
    end
    array = []
    file.each_line.with_index do |line, index|
        if index > 0
            #STDERR.puts line
            #STDERR.puts line.strip.split(separator)[1]
            #break
            array << [line.strip.split(separator)[sentindex].to_f, line.strip.split(separator)[testindex].to_f]
        end
    end
    file.close
    return array
end

def readfromfile2(filename,goldindex,testindex,method)
    file = File.open(filename,"r:utf-8")
    if filename[-3..-1] == "csv"
        separator = ","
    elsif filename[-3..-1] == "tsv"
        separator = "\t"
    end
    test = []
    gold = []
    file.each_line.with_index do |line, index|
        if index > 0
            #STDERR.puts line
            #STDERR.puts line.strip.split(separator)[1]
            #break
            test << line.strip.split(separator)[testindex].to_f
            gold << line.strip.split(separator)[goldindex].to_f
        end
    end
    file.close
    a = Daru::Vector.new(test)
    b = Daru::Vector.new(gold)
    if method == "pearson"
        pearson = Statsample::Bivariate::Pearson.new(a,b)
        rho = pearson.r
    elsif method == "spearman"
        rho = Statsample::Bivariate.spearman(a, b)
    elsif method == "kendall"
        R.assign "test", test
        R.assign "gold", gold
        R.eval "out <- Kendall(test,gold)"
        rho = R.pull "as.numeric(out$tau)"
        #p = R.pull "Kendall(test,gold)$s1"
    elsif method == "krippendorff"
        R.assign "test", test
        R.assign "gold", gold
        R.eval "m1 <- matrix(c(test,gold),nrow=2,byrow=TRUE)"
        rho = R.pull "kripp.alpha(m1,\"interval\")$value"
    end

    return rho
end



dirs = ["results5_ca", "results6_ca", "results7_ca", "results8_ca", "results9_ca", "results20_ca"]
mode = "usual"
method = "krippendorff"
o2 = File.open("results_new.tsv","w:utf-8")
annot_hash2 = {1 => "a", 3 => "b", 6 => "e", 7 => "f", 8 => "g", 9 => "h", 10 => "i", 11 => "j"}
o2.puts "annot_id	annot_r	ra_r_mean	ra_r_se	diff\tdiff_adj"
dirs.each do |dir|
    STDERR.puts dir
    if mode == "ttest"

    else
        o = File.open("#{dir}_#{method}.tsv","w:utf-8")
        folds_rhos = []
        annot_rhos = {}
        filelist = Dir.children(dir)
        filelist.each do |filename|
            rho = readfromfile2("#{dir}\\#{filename}",3,2,method)
            if dir != "results20_ca"
                id = filename.split(".")[0][-1]
            else
                id = filename.split(".")[0][-2..-1]
            end
            
            o.puts "#{id}\t#{rho}"
            if ["0","1","2","3","4","5","6","7","8","9"].include?(id[0])
                folds_rhos << rho
            else
                annot_rhos[id] = rho
            end
        end
        annot_rhos.each_pair do |annot, rho|
            annot_id = annot_hash2.key(annot[0])
            mean = stats(folds_rhos, "array")[0]
            mean_adj = stats(folds_rhos[0..-2], "array")[0]
            se = stats(folds_rhos, "array")[2]
            diff = mean - rho
            diff_adj = mean_adj - rho
            o2.puts "#{annot_id}\t#{rho}\t#{mean}\t#{se}\t#{diff}\t#{diff_adj}"
        end
    end
end