convert indented text to graphml

txt2graphml.rb

class File
  def to_string
    string = ''
    self.each {|line|
      string << line
    }
    return string
  end
end

class Graph
  attr_accessor :name
  def initialize(labels=[])
    @name = "graph"
    @edges = []
    @labels = labels
  end

  def add(label)
    @labels << label
  end

  def connect(source, target)
    s = @labels.index(source)
    t = @labels.index(target)
    if s and t and [email protected]?([s,t])
      @edges << [s, t] 
    end
  end

  def connect_by_id(s,t)
    @edges << [s,t] if @edges[s] and @edges[t]
  end

  def to_file(name)
    out = File.open(name, 'w')
    out << @labels.join("\n")
  end

  def to_xml(name=(@name+".xml"))
    xml = <<EOF
<?xml version="1.0" encoding="UTF-8"?>
<graphml xmlns="http://graphml.graphdrawing.org/xmlns">
<graph edgedefault="directed">
<key id="label" for="node" attr.name="label" attr.type="string"/>
EOF
    footer = "</graph></graphml>"

    @labels.each_with_index{|l, i| xml +="<node id=\"#{i}\"><data key=\"label\">#{l}</data></node>\n"}
    @edges.each{|e| xml += "<edge source=\"#{e[0]}\" target=\"#{e[1]}\"></edge>\n"}
    xml += footer
    File.open(name, 'w') << xml
  end

  def self.read(file)
    string = File.open(file, 'r').to_string
    graph = case File.extname(file) 
            when ".graphml" then read_graphML(string)
            when ".txt" then read_ws(string) #whitespaced
            end
    graph.name = File.basename(file, File.extname(file))
    graph
  end

  def self.extract_keywords(file)
    str = File.open(file, 'r').to_string
    str.chop!; str[0,1]=' '; str.lstrip! #remove first and last brace
    array = str.split(/\}\{/).uniq.reject{|w| w.size == 1} #treat '}{' as separator and reject one symbol words
    out = File.open("out", 'w')
    out << array.join("\n")
  end

  def self.normalize(file)
    `mystem.exe -l #{file} out`
  end

  private

  def self.read_graphML(string)
    ids = string.scan(/<node id\="((\w+|\:{2})*)"/).map{|id| id[0]}
    labels = string.scan(/>(.*)<\/y:NodeLabel/).map{|l| l[0] unless l[0].match(/Folder/)}.compact
    edges = string.scan(/<edge id=".+" source="((\w+|:{2})+)" target="((\w+|:{2})+)"/).map{|edge| [edge[0], edge[2]]}

    graph = Graph.new(labels)
    edges.each{|edge| graph.connect(labels[ids.index(edge[0])], labels[ids.index(edge[1])])}
    ids.each_with_index{|id, i| 
      group = id.split(/::/)
      if group.size > 1
        group.pop
        graph.connect_by_id(ids.index(group.join('::')), i)
      end
    }
    return graph
  end

  def self.read_ws(string)
    graph = Graph.new()
    parent = ""
    string.split("\n").each_with_index{|node,index|
      graph.add node
      if node[0,2] == "  "
        node[0,2] = ""
        graph.connect(parent, node)
        puts "#{parent} - #{index}"
      else
        parent = node
        puts "#{parent} ->"
      end
    }
    return graph
  end
end

Graph.read("input.txt").to_xml
#graph.to_file('out.graphml')