module DataMining

# File lib/log_analysis/data_mining.rb, line 8
def self.execute(transform_data, min_conf, min_sup)
  File.open(LogAnalysis::TRANSFORM_DATA_PATH, 'w+') { |f| transform_data.keys.each { |e| f.puts(transform_data[e].map { |i| i.is_a?(Array) ? i.join(' ') : i }.join(' -1 ').concat(' -1 -2')) } }
  system("java -jar #{LogAnalysis::JAR_FILE_PATH} run SPADE #{LogAnalysis::TRANSFORM_DATA_PATH} #{LogAnalysis::RULE_FILE_PATH} #{min_sup}%")
  rule_gen(get_seq(File.read(LogAnalysis::RULE_FILE_PATH)), min_conf)
end

# File lib/log_analysis/data_mining.rb, line 41
def self.get_seq(seq_str)
  seq = seq_str.split("\n")
  seq.each_with_object([]) do |s, arr|
    split_seq = s.split('-1')
    arr.push([split_seq[0..-2], split_seq[-1][-1].to_f])
  end
end

# File lib/log_analysis/data_mining.rb, line 14
def self.rule_gen(seqs, min_conf)
  seqs.each_with_object([]) { |seq, arr| seqs.each { |sub| arr.push([seq[0], sub[0], seq[1] / sub[1]]) if sub[0] != seq[0] && sub_seq?(sub[0], seq[0]) && seq[1] / sub[1] >= min_conf } }
end

# File lib/log_analysis/data_mining.rb, line 34
def self.sub?(str, sub)
  mark_sub = 0
  sub.split(' ').each { |char| mark_sub += 1 if str.include?(char) }

  mark_sub == sub.split(' ').size
end

# File lib/log_analysis/data_mining.rb, line 18
def self.sub_seq?(first, second)
  ptr = 0
  first.each do |sub|
    return false if ptr >= second.size

    (ptr..second.size - 1).each do |n|
      if sub?(second[n], sub)
        ptr = n + 1
        break
      end
      return false if ptr == second.size - 1
    end
  end
  true
end