README.md in data_seeder-0.0.5 vs README.md in data_seeder-1.0.0

@@ -36,34 +36,33 @@
       t.string :name, null: false
     end
 
 And you have a corresponding db/seed/countries.txt file as follows:
 
-    # config: { key_attribute: 'code', line: ->(line) { { code: line[0,2], name: line[3...-1] } } }
     AD Andorra
     AE United Arab Emirates
     AF Arghanistan
 
-The first line in a file can define the config attributes associated with the file.  For this seed file,
-the key_attribute says that it will use the 'code' attribute to lookup existing records (defaults to 'id')
-and the line function
-defines how the line is converted to an attribute hash defining the instance.
+And a db/seed/countries.cfg file as follows:
 
-Since the first line can get a little busy with config information, you can also store your config in a
-separate .cfg file with the same name.  This contents of this file should eval to a hash.  The above config line would be
-equivalent to a db/seed/countries.cfg file with the following:
-
     {
-      key_attribute: :'code',
-      line: ->(attr) {
+      key_attribute: 'code',
+      line: ->(line) {
         {
           code: line[0,2],
           name: line[3...-1]
         }
       }
     }
 
+The cfg file defines the config attributes associated with the file.  This contents of this file
+should eval to a hash.  For this seed file,
+the key_attribute says that it will use the 'code' attribute to lookup existing records (defaults to 'id')
+and the line function
+defines how the line is converted to an attribute hash defining the instance.  
+
+
 Running rake db:seed will result in the following output:
 
     # rake db:seed
     Loading countries
       Saving #<Country id: 1, code: "AD", name: "Andorra">
@@ -95,15 +94,15 @@
 data_seeder has default loaders for txt, csv, json and yml extensions but you can also create
 your own custom loaders.
 For instance, suppose you had the following tables:
 
 ```ruby
-create_table "apps", force: :cascade do |t|
+create_table "apps" do |t|
   t.string "name"
 end
 
-create_table "app_errors", force: :cascade do |t|
+create_table "app_errors" do |t|
   t.integer "app_id"
   t.string  "code"
   t.string  "message"
 end
 add_index "app_errors", ["app_id"], name: "index_app_errors_on_app_id"
@@ -120,73 +119,18 @@
     A1 Error message for A1
     A2 Error message for A2
     B1 Error message for B1
 
 
-You could create your own custom loader that might look as follows:
+Look [here](test/dummy/app/models/app_error_data_seeder.rb) for an example of creating your own custom loader.
 
-```ruby
-require 'data_seeder'
+To add this seeder, you would create the following config/initializers/data_seeder.rb:
 
-class AppErrorDataSeeder
-  include ::DataSeeder::Loader
-
-  def setup
-    @app = App.find_or_initialize_by(name: self.path_minus_ext)
-    @existing_errors = {}
-    if @app.new_record?
-      logger.info "Loading errors for new App: #{@app.name}"
-      @app.save!
-    else
-      logger.info "Loading errors for existing App: #{@app.name}"
-      @app.app_errors.each do |app_error|
-        @existing_errors[app_error.code] = app_error
-      end
-    end
-  end
-
-  def teardown
-    unless @existing_errors.empty?
-      logger.info { "  The following are begin removed:" }
-      @existing_errors.each do |code, app_error|
-        logger.info "    #{code}: #{app_error.message}"
-        app_error.destroy
-      end
-    end
-  end
-
-  def load(io)
-    io.each_line do |line|
-      line.strip!
-      next if line.blank? || line[0] == ?#
-      space_i   = line.index(' ')
-      raise "Invalid line: #{line}" unless space_i
-      code      = line[0,space_i].strip
-      message   = line[space_i+1..-1].strip
-      app_error = @existing_errors[code]
-      if app_error
-        @existing_errors.delete(code)
-        app_error.message = message
-        unless app_error.changes.empty?
-          logger.info { "  Changing #{code}: #{app_error.changes}" }
-          app_error.save!
-        end
-      else
-        logger.info { "  Creating #{code}: #{message}" }
-        @app.app_errors.create!(code: code, message: message)
-      end
-    end
-  end
-end
-```  
-
-To add the seeder, you would create the following config/initializers/data_seeder.rb:
-
 ```ruby
 MyApp::Application.config.after_initialize do
   DataSeeder.configure do |config|
-    config.add_loader('err', AppErrorDataSeeder.new)
+    config.add_loader('err', AppErrorDataSeeder)
   end
 end
 ```
 
 Executing DataSeeder.run would result in the following:
@@ -198,48 +142,77 @@
     Loading errors for new App: foo
       Creating 1: Something went wrong
       Creating 2: We are seriously foobared
       Creating 3: We are less seriously foobared
 
-TODO
-----
+## Configurable values
 
-Ability to specify more than 1 directory for Rails.env overrides.  Could potentially be used if you have that
-x Gigabyte seed file that you don't want to check into source control and only want run on production?
+#### depends
 
-YAML should allow loading as either array or hash. (currently only does hash)
+Value or array that this model depends on such that they must be seeded first.  Examples:
 
-CSV should have options such as only: and except: for using/skipping the specified header columns.
+    {
+      depends: ['countries','states']
+    }
 
-The structure.sql caching within rails uses the file timestamp to determine whether to prepare the test database.  This
-is error prone and forces you to do a 'touch db/structure.sql' to get around the not getting reloaded problem.  Should
-I add a utility to override this rails implementation with a sha-based one like the seed files use?  (or am I the only
-one who has to 'touch db/structure.sql' everytime I switch branches?)
+#### key_attribute
 
-Add 'sql' loader (with disclaimer that it will temporarily truncate the table)
+The attribute used to define uniqueness within the model.  Can be a single attribute or an array. Defaults to 'id'
 
-Ability to stop early when loading up a large seed file for a given environment, i.e., stop after processing the
-first 10 lines when Rails.env.test?
+#### klass
 
-I want to allow different seeding for different environments.  For instance development might have a bunch of dummy
-data useful for getting an environment up and running.  I'm thinking either the seed_dir similar to like a PATH
-environment variable where the first one found would override the others, or maybe make it automatic based on the
-directory names and the environment (seed.development/state.yml would override seed/state.yml).
+Defines the ActiveRecord Class if it can't be inferred from the seed file.
 
-The test environment will be the one that will constantly being seeded after migrations or branch changes.  Some of
-the seed files might be large and take a long time to seed.  The above
-strategy using seed.test might be useful but it might also be useful to have a preprocessor type such as .sh so for
-instance you might have seed.test/table_with_lotsa_rows.csv.sh which might consist of the line
-'head -20 ../seed/table_with_lotsa_rows.csv'
+#### line
 
-Caching of long-running stuff via pg_dump, mysqldump, or other?  This belongs with discussion of the environment-specific
-seeding above.
+Proc used for converting a line to attributes (txt files only).
 
-Allow config-driven initialization so that we could require: false in the Gemfile and only load as needed.
+#### postprocess
 
-Add depends_on option.
+Modify the attributes from the seed file before applying them to the model.
 
+Example:
+
+    {
+      key_attribute: 'code',
+      postprocess: ->(attrs) {
+        {
+          code: attrs['country_code'],
+          name: attrs['country']
+        }
+      }
+    }
+
+#### purge
+
+Destroys rows that no longer exist in the seed file.
+
+#### update_display_method
+
+Model method used for displaying updates to a model.
+
+#### use_line_number_as_id
+
+Use the line number of the seed file as the id
+
+## Incompatibilities from 0.0.x version
+
+Custom seeders should now be specified as a Class and not an instance (MySeeder instead of MySeeder.new)
+
+data_seeder_<config-item> methods within the models are no longer supported.
+
+Using the first line of txt, json, and yaml files as the config is no longer supported.  Move them to
+a separate .cfg file.
+
+
+TODO
+----
+
+Add 'sql' loader (with disclaimer that it will temporarily truncate the table)
+
+Caching of long-running stuff via pg_dump, mysqldump, or other?  
+
 Document options (key_attribute, line, postprocess, etc)
 
 Meta
 ----
 
@@ -256,10 +229,10 @@
 [Brad Pardee](https://github.com/bpardee)
 
 License
 -------
 
-Copyright 2015 Brad Pardee
+Copyright 2015-2016 Brad Pardee
 
 Licensed under the Apache License, Version 2.0 (the "License");
 you may not use this file except in compliance with the License.
 You may obtain a copy of the License at