某件拖了一年的任务

# 第一版
$redis.sadd "stock:chinese_name" , Regexp.escape(Stock.only(:chinese_name).map(:chinese_name))
$redis.smembers("stock:chinese_name").each do |n|
  content.gsub(/n/).each do |x|
    st = Stock.where(chinese_name: n).first
    st ? "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>" : x
  end
end

# 第二版
$redis.sadd "stocks:chinese_name" , Stock.only(:chinese_name).map { |x| Regexp.escape(x.chinese_name) }
content.gsub(/[$redis.smembers("stocks:chinese_name").join("|")]/).each do |x|
  st = Stock.where(chinese_name: x).first
  st ? "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>" : x
end

# 输出失败, 打印了一下，看到转换成功了，看来是自己被检测输出给坑了，应该打印出来看看

# 重试正则表达式:

re = Regexp.new($redis.smembers("stocks:chinese_name").join("|"))
content.gsub(re).each do |x|
  p x
  st = Stock.where(chinese_name: x).first
  p "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>"
  st ? "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>" : x
end

content.gsub(/上汽集团|新华保险/).each do |x|
  st = Stock.where(chinese_name: x).first
  p "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>"
  st ? "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>" : x
end

# 最终版, views的辅助方法
def convert_stock_to_link(content)
  re = Regexp.new($redis.smembers("stocks:chinese_name").join("|"))
  content.gsub(re).each do |x|
    st = Stock.where(chinese_name: x).first
    # p "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>"
    st ? "<strong><a href='/asset_class/stocks/#{st.slug}' target='_blank'>#{x}</a></strong>" : x
  end
end

# 改变思路了，只要存个正则表达式字符串就行了，表达式构建如下: 
$redis.set "stocks:chinese_name:regexp" , Stock.only(:chinese_name).map { |x| Regexp.escape(x.chinese_name) }.uniq.join('|')

# 新的view辅助方法
def convert_stock_to_link(article)
  return article.content if article.stocks.nil?
  re = Regexp.new(article.stocks.only(:chinese_name).map { |x| Regexp.escape(x.chinese_name) }.join('|'))
  article.content.gsub(re).each do |x|
    st = Stock.where(chinese_name: x).first
    # p "<strong><a href='/asset_class/stocks/#{st.slug}'>#{x}</a></strong>"
    st ? "<a href='/asset_class/stocks/#{st.slug}' target='_blank' >#{x}</a>" : x
  end
end

# Article模型的before_save回调
class Article
  ...
  include Mongoid::Document
  has_and_belongs_to_many :stocks

  before_save :add_stocks

  private 
  
  def add_stocks
    if content_changed?
      re = Regexp.new($redis.smembers("stocks:chinese_name").join("|"))
      self.content.scan(re).each do |x|
        st = Stock.where(chinese_name: x).first
        self.stocks << st 
      end
    end
  end
  ....
end

# 创建的Rake任务，7千个文章，执行了半个小时，29万的话，大概要一天，我擦啊！！！
$redis.sadd "stocks:chinese_name" , Stock.only(:chinese_name).map { |x| Regexp.escape(x.chinese_name) }
re = Regexp.new($redis.smembers("stocks:chinese_name").join("|"))
Article.each do |article|
  article.content.scan(re).each do |x|
    st = Stock.where(chinese_name: x).first
    article.stocks << st # 等价于 => articles.stocks.push st
  end
  article.save
end

# 沿着之前的sadd的思路折腾了好久，最后，换了个思路，干吗要一次搞定?? 循环遍历
Stock.only(:chinese_name).each { $redis.hset "stocks" , x.id , x.chinese_name  }

# 循环遍历的方案被否决了，
$redis.mapped_hmset("stocks",Stock.only(:chinese_name).map { |x| [ x.id, Regexp.escape(x.chinese_name) ] })
$redis.mapped_hmset("stocks",Stock.only(:chinese_name).map { |x| [ x.id ,  { id:  x.id,  chinese_name: x.chinese_name} ] })
# 取出来方法
$redis.hgetall("stocks").values

# 初版 & 终版
$redis.mapped_hmset("stocks",Stock.only(:chinese_name).map { |x| [ x.id , Regexp.escape(x.chinese_name) ] })
re = Regexp.new($redis.hgetall("stocks").values.join("|"))
Article.each do |article|
  article.content.scan(re).each do |x|
    #改进的话，需要使用chinese_name取到id, 怎么才能根据值去取key，hash结构设计有问题。
    st = Stock.where(chinese_name: x).first
    article.stocks << st
  end
  article.save
end

# 某中间版
# 放弃的原因是: article和stock是多对多的关系，单方面操作id会出问题的
$redis.mapped_hmset("stocks",Stock.only(:chinese_name).map { |x| [ x.id , x.chinese_name ] })
s = $redis.hgetall("stocks").invert # 反转hash，有时很很有用处
re = Regexp.new($redis.hgetall("stocks").values.map { |x| Regexp.escape(x) }.join("|"))
Article.each do |article|
  article.content.scan(re).each do |x|
    article.stock_ids << s[x] 
  end
  article.save
end

before_save :update_stock_cache
after_create :create_stock_cache
after_destory :remove_stock_cahce

protected

def create_stock_cache
  $redis.hset "stocks", self.id, Regexp.escape(self.chinese_name)
end

def remove_stock_cache
  $redis.hdel "stocks", self.id
end

def update_stock_cache
  $redis.hset("stocks", self.id, Regexp.escape(self.chinese_name)) if self.chinese_name_changed?
end

def convert_stock_to_link(article)
  content = article.content
  article.stocks.only(:chinese_name, :slug).each do |stock|
    content.gsub!(Regexp.escape(stock.chinese_name), "<a href='/asset_class/stocks/#{stock.slug}' target='_blank' class='blue_txt' >#{stock.chinese_name}</a>" )
  end
  content
end

task :create_stock_ids => :environment do
  Stock.check_key "stocks"
  re = Regexp.new($redis.hgetall("stocks").values.join("|"))
  Article.each do |article|
    next if article.content.scan(re).blank?
    article.stocks = Stock.where(:chinese_name.in => article.content.scan(re))
    article.save
  end
end