massive_record table creations

massive_record_migrations.rb

# A workaround (hackish) way to perform migrations using ORM style models
# and massive_record. Ideally, you shouldn't be using this because
# in massive_record ORM, tables are created (with column families)
# whenever you perform an action toward the database. However if you are unlucky
# and get stuck with issues like
# https://github.com/CompanyBook/massive_record/issues/95#issuecomment-246317908
# this script provides a useful way to deal with it.

require 'activerecord'

def migrate
  watch_stack = ActiveSupport::Dependencies::WatchStack.new
  
  # If your classes exist inside a namespace like MyGem::Employee, you should
  # add the top level namespace here. If there is none, and your class name
  # is simply Employee, you can add top level namespace as "Object"
  watch_namespaces = ['MyTopLevelNamespace']
  
  watch_stack.watch_namespaces(watch_namespaces)

  project_root = Dir.pwd
  
  # This is the directory in which all your model files reside.
  models_dir_path = "#{File.expand_path('lib/hbase/models', project_root)}/"
  model_files = Dir.glob(File.join(models_dir_path, '**', '*.rb'))

  # Data struct to store tables with their column families
  tables = {}

  model_files.sort.each do |file|
    watch_stack.watch_namespaces(watch_namespaces)
    require_dependency(file)
    new_constants = watch_stack.new_constants
    new_constants.each do |konst|
      if MyTopLevelNamespace.const_get(konst).ancestors.include?(MassiveRecord::ORM::Table)
        clazz = konst.constantize
        if clazz.send(:table_exists?)
          puts "Table Exists: #{ clazz.send(:table_name) }"
        else
          tables[clazz.table_name] = clazz.column_families.collect { |x| x.name }
        end
      end
    end
  end

  # We are creating HBase shell commands and passing them to hbase shell
  # These hbase shell commands are create commands.
  tables.each do |name, col_families|
    puts "Creating Table: #{ name }"
    
    # This is important. The HBASE_PATH_BIN is the dir path of hbase's bin dir
    # This is where the `hbase` resides.
    hbase_path = ENV['HBASE_PATH_BIN']
    cmd = "echo \"create '#{ name }', '#{ col_families.join(',') }' \" |  #{ hbase_path }/hbase shell "
    `#{ cmd }`
  end

end