[ruby-list:22945] rbnamazu (Re: [Q] blade internal design)

咳といいます。

> 拡張モジュールではありませんが、Rubyのみでnamazuのインデックスの内容
> を読み出し、スコアの計算をおこなうrbnamazuを書いています。
> 
> 以下のURLに、といあえず、おいておきます。
> http://shinoda-www.jaist.ac.jp/Projects/hige/src/namazu.rb

rbnamazuをrequireして動く分散namazuを書いてみました。
rbnamazuをライブラリ風にしているのでちょっと大きくなってしまいました。
今はスコアをクライアントに返さないのですが、もうちょっと頑張ってみます。

こんな風に使います。もうちょっといい例をそのうち書きたいです。

-- client --
require 'drb/drb'

there = ARGV.shift
unless there
  $stderr.puts("usage: #{$0} <server_uri> <keyword...>")
  exit 1
end

DRb.start_service()
ro = DRbObject.new(nil, there)

docs, hits, ref_hits = ro.query(ARGV)
p docs						# Array of Hash

ro.search(ARGV, "late") do |r|			# GC safe!
  p r.result(%w(uri))
  r.result(%w(uri summary), nil).each do |rec|	# Array of Array
    puts rec[0]
    puts rec[1]
  end
  nil
end


# namadu-0.0 - distributed namazu
#
# $Id: namadu.rb,v 1.2 2000/05/26 17:48:59 mas Exp $
# Copyright (c) 2000 Masatoshi SEKI
#
# namadu.rb is copyrighted free software by Masatoshi SEKI.
# You can redistribute it and/or modify it under the same term as Ruby.

require 'rbnamazu'

class NamazuFieldAll
  def gets_as_array(docid, fieldnames = nil)
    unless fieldnames
      fieldnames = @fieldnames
    end
    fieldnames.collect do |fieldname|
      field = @nmzfieldall[fieldname].gets(docid)
      field.chomp!()
      field
    end
  end
end

class Namadu
  class Result
    def initialize(namadu, order, words)
      @namadu = namadu

      result, times, references = @namadu.finder.get_scores(words)

      @references = ref_hits(words, references)
      @docids = sort(result, times, order)
      @docids.freeze
      @num_docs = @docids.size
    end
    attr_reader(:references, :num_docs)
    
    private
    def ref_hits(words, counts)
      hash = {}
      words.each_with_index do |w, i|
	hash[w] = counts[i]	
      end
      hash.freeze
    end

    private
    def sort(scores, times, order='score')
      docids = scores.keys
      case order
      when 'score'
	docids.sort { |a, b| scores[a] <=> scores[b] }
      when 'early'
	docids.sort { |a, b| times[a] <=> times[b] }	
      when 'late'
	docids.sort { |a, b| times[b] <=> times[a] }	
      else
	docids
      end
    end

    private
    def check_fields(fields)
      return unless fields
      all = @namadu.fieldnames
      for f in fields
	raise "Unknown field: #{f}" unless all.include? f
      end
    end
    
    public
    def result(fields=nil, max=20, whence=0)
      check_fields(fields)
      docids = @docids
      docids = docids[whence, max] if max

      if fields
	docids.collect do |docid|
	  @namadu.fields_loader.gets_as_array(docid, fields)
	end
      else
	docids.collect do |docid|
	  @namadu.fields_loader.gets(docid, fields)
	end
      end
    end
  end

  def initialize(index)
    @index = index.dup.freeze
    @finder = NamazuScores.new(@index)
    @fields_loader = NamazuFieldAll.new(@index)
  end
  attr_reader(:finder, :fields_loader)

  public
  def fieldnames
    @fields_loader.fieldnames
  end

  public
  def search(words, order='score')
    words = words.collect { |w| w.downcase }
    r = Result.new(self, order, words)
    yield(r) if iterator? 		# GC trick
    r
  end

  public
  def query(words, order='score', fields=nil, max=20, whence=0)
    r = search(words, order)
    list = r.result(fields, max, whence)
    [list, r.num_docs, r.references]
  end
end

require 'drb/drb'

class Namadu
  include DRbUndumped
  class Result
    include DRbUndumped
  end
end

if __FILE__ == $0
  index = ARGV.shift
  nmd = Namadu.new(index)
  DRb.start_service('druby://:7952', nmd)
  puts DRb.uri
  if $DEBUG
    gets
  else
    DRb.thread.join
  end
end