# Generates header -> ids in the given doc, calculating the ids from the heading
# text and ensuring that there are no duplicated ids
class HeaderIdGenerator

  def self.apply!(doc)
    self.new(doc)
  end

  def initialize(doc)
    @doc = doc
    @header_nodes = @doc.css("h2,h3,h4,h5,h6").to_a

    # { node -> id } hash
    @nodes_ids = @header_nodes.inject({}){ |hash, node| hash[node] = nil; hash }

    add_default_ids
    resolve_conflicts
    @nodes_ids.each{ |node, id| node['id'] = id }
  end

  private

  def resolve_conflicts
    return if conflicts(@nodes_ids).empty?

    # { h4-node -> [h3-node, h2-node] }
    @parents = @nodes_ids.keys.inject({}){ |hash, node| hash[node] = parent_header_nodes(node); hash }

    prepend_parents_on_conflicts
    append_numbers_on_conflicts
  end

  # Prepend parents recursively, one level at a time, until there are no conflicts or no changes in the result
  def prepend_parents_on_conflicts(parent_index = 0)
    original = @nodes_ids.dup
    conflicts(@nodes_ids).each do |node, id|
      if parent = @parents[node][parent_index]
        new_id = subheader_id("#{subheader_id(parent.content)} #{@nodes_ids[node]}")
        @nodes_ids[node] = new_id
      end
    end
    prepend_parents_on_conflicts(parent_index + 1) if original != @nodes_ids
  end

  # Parent != DOM nesting, but in the context of the content <h2></h2> ... <h3></h3>

  def parent_header_nodes(node)
    parent_tags(node.name).map do |parent_tag|
      @header_nodes[0..@header_nodes.index(node)-1].select{ |sibling| sibling.name == parent_tag }.last
    end.compact
  end

  # "h4" -> ["h2", "h3"]
  def parent_tags(tag)
    level = tag.gsub('h','').to_i
    (2..level-1).map{ |n| "h#{n}" }.reverse
  end


  def append_numbers_on_conflicts
    conflicts(@nodes_ids).group_by{ |node, id| id }.each do |id, id_conflicts|
      id_conflicts.each_with_index do |conflict, n|
        node = conflict[0]
        new_id = "#{id}-#{n+1}"
        @nodes_ids[node] = new_id
      end
    end
  end

  def conflicts(hash)
    hash.select{ |node1, id1| hash.select{ |_node1, id2| id1 == id2 }.size > 1 }
  end

  def add_default_ids
    @nodes_ids.each{ |node, id| @nodes_ids[node] = subheader_id(node.content) }
  end

  def subheader_id(content)
    content.to_s.downcase.gsub(/\W+/, '-').gsub(/\A-+|-+\Z/, '')
  end
end