make-test-data

convert_to_records.rb

def convert_to_records(raw_records, data_column)

  if raw_records.first.length != data_column.length
    puts '取得したデータの列の数が必要なデータの列の数と違います'
    return false
  end

  records = []
  raw_records.each do |raw_record|
    record = {}
    data_column.each_with_index do |data_column, i|
      record[data_column] = raw_record[i]
    end
    records << record
  end

  records
end

get_file_contents.rb

def get_file_contents(html_path)
  file = open(html_path)
  contents = file.read
  file.close
  contents
end

get_raw_records.rb

def get_raw_records(contents)

  names = contents.scan(/h[\d].+nme[\d][^\/]+\//)
  introductions = contents.scan(/pft[^v]+v/)

  records = []
  names.each_with_index do |name, i|
    name = name.split(/(\w|\/)+/)[2]

    words = introductions[i].split(/(\w|\/|\n|;|&)+/)

    introduction = ''
    words.each do |word|
      if /^([ぁ-ん]|[ァ-ヴ]|\d)/ =~ word
        introduction += word + ' '
      end
    end
    record = [name, introduction]
    records << record
  end

  records
end

guide.rake

namespace :db do
 task guide: :environment do
    Page.create!(name: "フシギダネ", introduction: "うまれたときから せなかに しょくぶつの タネが あって すこしずつ おおきく そだつ。 ")
    Page.create!(name: "フシギソウ", introduction: "つぼみが せなかに ついていて ようぶんを きゅうしゅうしていくと おおきな はなが さくという。 ")
    Page.create!(name: "フシギバナ", introduction: "はなから うっとりする かおりが ただよい たたかうものの きもちを なだめてしまう。 ")
    .
    .  
    .
  end
end

make_rake_task.rb

def make_rake_task(records, object_name, data_column, task_name)

  start_pattern = "namespace :db do\n task #{task_name}: :environment do\n"
  end_pattern = "  end\nend"

  create_pattern = ""
  records.each do |record|
    hash = ""
    command = "    #{object_name}.create!("

    data_column.each do |column|
      value = record[column]
      hash += ", #{column}: \"#{value}\""
    end
    hash.slice!(0,2)
    command += "#{hash})"
    create_pattern += "#{command}\n"
  end

  create_pattern
  rake_task = start_pattern + create_pattern + end_pattern
end

make_test_data.rb

def make_test_data(html_path, object_name, data_column, task_name)
  html_file =  get_file_contents(html_path)
  contents = trim_file(html_file)
  raw_records = get_raw_records(contents)
  records = convert_to_records(raw_records, data_column)
  rake_task = make_rake_task(records, object_name, data_column, task_name)

  rake_file = File.open("#{task_name}.rake","w")
  rake_file.write(rake_task)
  rake_file.close
end

trim_file.rb

def trim_file(html_file)
  contents_start_pattern = "<script src=\"https:\/\/pagead2.googlesyndication.com\/pagead\/show_ads.js\"><\/script>"
  contents_end_pattern = "<div\sclass=\"ad\">"
  contents = html_file.scan(/#{contents_start_pattern}.+#{contents_end_pattern}/m)[0]
  contents.delete!(contents_start_pattern, contents_end_pattern)
  contents
end