1

我正在使用 Neo4j 进行我的第一个项目。我正在解析维基百科的页面和页面链接转储以创建一个图形，其中节点是页面，边缘是链接。我已经定义了一些 rake 任务来下载转储、解析数据并将其保存在 Neo4j 数据库中。在 rake 任务结束时，我打印创建的页面和链接的数量，以及一些链接最多的页面。这是 zawiki 的 raks 任务的输出。

$ rake wiki[zawiki]
[ omitted ] 
...
:: Done parsing zawiki
:: 1984 pages
:: 2144 links
:: The pages with the most links are:
9625.0 - Emijrp/List_of_Wikipedians_by_number_of_edits_(bots_included): 40
1363.0 - Gvangjsih_Bouxcuengh_Swcigih: 30
9112.0 - Fuzsuih: 27
1367.0 - Cungzcoj: 26
9279.0 - Vangz_Yenfanh: 19

看起来正在创建页面和链接，但是当我启动 rails 控制台或服务器时，找不到链接。

$ rails c 
jruby-1.7.5 :013 > Pages.all.count
=> 1984 
jruby-1.7.5 :003 > Pages.all.reduce(0) { |count, page| count + page.links.count}
=> 0 
jruby-1.7.5 :012 > Pages.all.sort_by { |p| p.links.count }.reverse[0...5].map { |p|  p.links.count }
=> [0, 0, 0, 0, 0]

这是 rake 任务，这是项目 github 页面。谁能告诉我为什么链接没有保存？

DUMP_DIR = Rails.root.join('lib','assets')


desc "Download wiki dumps and parse them"
task :wiki, [:wiki] => 'wiki:all'

namespace :wiki do
  task :all, [:wiki] => [:get, :parse] do |t, args|
    # Print info about the newly created pages and links.
    link_count = 0
    Pages.all.each do |page|
      link_count += page.links.count
    end
    indent "Done parsing #{args[:wiki]}"
    indent "#{Pages.count} pages"
    indent "#{link_count} links"

    indent "The pages with the most links are:"
    Pages.all.sort_by { |a| a.links.count }.reverse[0...5].each do |page|
      puts "#{page.page_id} - #{page.title}: #{page.links.count}"
    end
  end


  desc "Download wiki page and page links database dumps to /lib/assets"
  task :get, :wiki do |t, args|
    indent "Downloading dumps"
    sh "#{Rails.root.join('lib', "get_wiki").to_s} #{args[:wiki]}"
    indent "Done"
  end


  desc "Parse all dumps"
  task :parse, [:wiki] => 'parse:all'

  namespace :parse do
    task :all, [:wiki] => [:pages, :pagelinks]

    desc "Read wiki page dumps from lib/assests into the database"
    task :pages, [:wiki] => :environment do |t, args|
      parse_dumps('page', args[:wiki]) do |obj|
        page = Pages.create_from_dump(obj)
      end
      indent = "Created #{Pages.count} pages"
    end

    desc "Read wiki pagelink dumps from lib/assests into the database"
    task :pagelinks, [:wiki] => :environment do |t, args|
      errors = 0
      parse_dumps('pagelinks', args[:wiki]) do |from_id, namespace, to_title|
        from = Pages.find(:page_id => from_id)
        to = Pages.find(:title => to_title)
        if to.nil? || from.nil?
          errors = errors.succ
        else
          from.links << to
          from.save
        end
      end
    end
  end
end

def indent *args
  print ":: "
  puts args
end

def parse_dumps(dump, wiki_match, &block)
  wiki_match ||= /\w+/
  DUMP_DIR.entries.each do |file|
    file, wiki = *(file.to_s.match(Regexp.new "(#{wiki_match})-#{dump}.sql"))
    if file
      indent "Parsing #{wiki} #{dump.pluralize} from #{file}"
      each_value(DUMP_DIR.join(file), &block)
    end
  end
end

def each_value(filename)
  f = File.open(filename)
  num_read = 0

  begin                               # read file until line starting with INSERT INTO
    line = f.gets
  end until line.match /^INSERT INTO/

  begin
    line = line.match(/\(.*\)[,;]/)[0]  # ignore begining of line until (...) object
    begin
      yield line[1..-3].split(',').map { |e| e.match(/^['"].*['"]$/) ?  e[1..-2] : e.to_f }
      num_read = num_read.succ

      line = f.gets.chomp
    end while(line[0] == '(')          # until next insert block, or end of file
  end while  line.match /^INSERT INTO/ # Until line doesn't start with (...

  f.close
end

应用程序/模型/pages.rb

class Pages < Neo4j::Rails::Model
  include Neo4j::NodeMixin
  has_n(:links).to(Pages)

  property :page_id
  property :namespace,        :type => Fixnum
  property :title,            :type => String
  property :restrictions,     :type => String
  property :counter,          :type => Fixnum
  property :is_redirect,      :type => Fixnum
  property :is_new,           :type => Fixnum
  property :random,           :type => Float
  property :touched,          :type => String
  property :latest,           :type => Fixnum
  property :length,           :type => Fixnum
  property :no_title_convert, :type => Fixnum


  def self.create_from_dump(obj)
    # TODO: I wonder if there is a way to compine these calls
    page = {}

    # order of this array is important, it corresponds to the data in obj
    attrs = [:page_id, :namespace, :title, :restrictions, :counter, :is_redirect,
             :is_new, :random, :touched, :latest, :length, :no_title_convert]

    attrs.each_index { |i| page[attrs[i]] = obj[i] }
    page = Pages.create(page)
    return page
  end
end
4

1 回答 1

0

我必须承认我不知道 Neo4j 是如何工作的。

但是，从其他数据库转移，我也假设某些验证是错误的，或者在您使用数据库时甚至可能配置错误。后者我无法给出任何建议，但如果是关于验证，你可以查看Page#errors或尝试Page#save!跟注，看看它会加注什么。

看到这个例子时突然想到的一个疯狂的想法是，也许要正确配置该关系，您也需要一个反向引用。

也许has_n(:links).to(Page, :links)会帮助你。或者，如果这不起作用：

has_n(:links_left).to(Page, :links_right)
has_n(:links_right).from(Page, :links_left)

我越看这个，我越认为对同一个表的反向引用配置不正确，因此不会验证。

于 2013-10-25T08:01:00.537 回答