Configurable namespace separator (#106)

* make registry namespace separator configurable
* fix settings and reorg `Kiba::Extend` module so inter-app require statements can be removed
* require the show me extension so using this option does not fail!
* add doc section on using transforms in job definitions

CHANGELOG.adoc

@@ -41,6 +41,17 @@ These changes are merged into the `main` branch but have not yet been tagged as
 
 == Releases
 
+=== 3.2.1 - 2022-09-21
+
+==== Added
+* Config setting to control string used as registry namespace separator
+
+==== Bugfixes
+* Require the kiba-common ShowMe extension so that option actually works when running jobs
+
+==== Changed
+* Refactoring `lib/kiba/extend.rb` so inter-application `require` statements can be removed
+
 === 3.2.0 - 2022-09-20
 
 ==== Added

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    kiba-extend (3.2.0)
+    kiba-extend (3.2.1)
       activesupport (~> 6)
       amazing_print (~> 1.4)
       csv (~> 3)

doc/common_patterns_tips_tricks.md

@@ -43,8 +43,7 @@ class Transformer
 end
 ```
 
-It can also be useful in other transforms as shown below:
-
+It can also be used in order to compose additional behavior in another transform as shown below:
 
 ```
 class Transformer
@@ -70,6 +69,29 @@ end
 
 See the code for {Kiba::Extend::Transforms::Rename::Fields} for an example of embedding another transform to compose transformation logic.
 
+## Using transforms in job definitions
+
+The following code snippets are equivalent.
+
+This one relies on the domain specific language (DSL) "magic" defined in kiba:
+
+```
+Kiba.job_segment do
+  transform Merge::ConstantValue, target: :data_source, value: 'source system'
+end
+```
+
+This one uses plain Ruby to set up the transform class and calls its `:process` method on each row: 
+
+```
+Kiba.job_segment do
+  xform = Merge::ConstantValue.new(target: :data_source, value: 'source system')
+  transform{ |row| xform.process(row) }
+end
+```
+
+The second one might be useful in situations when you are trying to set things up more flexibly.
+
 ## Calling a job with parameters
 
 No need to write repetitive jobs with the exact same logic to handle variable values that differ according to a pattern. See [File registry documentation on Hash creator](https://lyrasis.github.io/kiba-extend/file.file_registry_entry.html#hash-creator-example-since-2-7-2) for a full example of how to do this.

lib/kiba/extend.rb

@@ -13,13 +13,6 @@
 require 'xxhash'
 require 'zeitwerk'
 
-require 'kiba/extend/error'
-require 'kiba/extend/registry/file_registry'
-require 'kiba/extend/jobs'
-require 'kiba/extend/jobs/job_segmenter'
-require 'kiba/extend/destinations'
-require 'kiba/extend/destinations/csv'
-
 # The Kiba ETL framework for Ruby.
 # `kiba-extend` extends only Kiba OSS. Kiba Pro features are not used.
 #
@@ -70,9 +63,6 @@ def reload!
       @loader.reload
     end
 
-    # So we can call Kiba.job_segment
-    Kiba.extend(Kiba::Extend::Jobs::JobSegmenter)
-
     # @return [Hash] default options used for CSV sources/destinations
     setting :csvopts, default: { headers: true, header_converters: %i[symbol downcase] }, reader: true
 
@@ -91,14 +81,21 @@ def reload!
 
     # @return [String] default string to be treated as though it were a null/empty value.
     setting :nullvalue, default: '%NULLVALUE%', reader: true
+
+    # @return [String] used to join nested namespaces and registered keys in FileRegistry
+    # @example With namespace 'ns' and registered key 'foo'
+    #   'ns__foo'
+    # @example With parent namespace 'ns', child namespace 'child', and registered key 'foo'
+    #   'ns__child__foo'
+    setting :registry_namespace_separator, default: '__', reader: true
 
     # @!method source
     # Default source class for jobs. Must meet implementation criteria in [Kiba wiki](https://github.com/thbar/kiba/wiki/Implementing-ETL-sources)
     setting :source, default: Kiba::Common::Sources::CSV, reader: true
 
     # @!method destination
     # Default destination class for jobs. Must meet implementation criteria in [Kiba wiki](https://github.com/thbar/kiba/wiki/Implementing-ETL-destinations)
-    setting :destination, default: Kiba::Extend::Destinations::CSV, reader: true
+    setting :destination, constructor: proc{ Kiba::Extend::Destinations::CSV }, reader: true
 
     # @return [String] prefix for warnings from the ETL
     setting :warning_label, default: 'KIBA WARNING', reader: true
@@ -107,8 +104,7 @@ def reload!
     #   [dry-container](https://dry-rb.org/gems/dry-container/main/) for registering and resolving
     #   jobs
     setting :registry,
-      default: Kiba::Extend::Registry::FileRegistry,
-      constructor: proc { |value| value.new },
+      constructor: proc{ Kiba::Extend::Registry::FileRegistry.new },
       reader: true
 
     # @return [Symbol] the job definition module method expected to be present if you [define a registry
@@ -237,3 +233,5 @@ def warn_unnested(name, value)
 
 
 Kiba::Extend.loader
+# So we can call Kiba.job_segment
+Kiba.extend(Kiba::Extend::Jobs::JobSegmenter)

lib/kiba/extend/jobs/show_me_job.rb

@@ -1,6 +1,7 @@
 # frozen_string_literal: true
 
-require 'kiba/extend'
+require 'kiba-common/dsl_extensions/show_me'
+
 module Kiba
   module Extend
     module Jobs

lib/kiba/extend/registry/file_registry.rb

@@ -19,7 +19,7 @@ module Registry
       class FileRegistry
         include Dry::Container::Mixin
 
-        config.namespace_separator = '__'
+        config.namespace_separator = Kiba::Extend.registry_namespace_separator
 
         # Exception raised if the file key is not registered
         class KeyNotRegisteredError < Kiba::Extend::Error

lib/kiba/extend/version.rb

@@ -2,6 +2,6 @@
 
 module Kiba
   module Extend
-    VERSION = '3.2.0'
+    VERSION = '3.2.1'
   end
 end