fedora_2_to_3_pid_renamer 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 32a1feca267687e4004907cff55c5600d2f91567
-  data.tar.gz: 9db15e938991fae0bf87d1027ef5d5a38efad5b6
+  metadata.gz: a0abd69d649aee20584ec54a1efd3d72f204204a
+  data.tar.gz: c15acd2a1a50a7e985568dda8bb575165ae72cd9
 SHA512:
-  metadata.gz: 5ceb7ef9e389863a7efba063a63b45de307a38ab94d4de64b0df837448fd52efb14df4bb99fb5b27179acb0982a35bd590b2ef3783859775ddd67e7a5a67de1e
-  data.tar.gz: e4d6212a162251f325bc47d8ed4a5b2a103cd30abbb257e0ef82b63a73aa90c747f990af7422e1be68f73bfa4ae7d6d80b0028a081d9624c3c79082ecf809cae
+  metadata.gz: 47470889083bc18f1aa7b543f042af0f629d75930c1bc286fb64cd23351c40064eff2f25e909db3da6fe21a7d998e74908e74621534cfdc30cb448f48e130977
+  data.tar.gz: df95f326d5b43825ec06787381b9e209a97606d3fb0a1619d1b3d34303a9b746692c019c48d77e7379f41817bb9dea2d4dfef3c2d8eec24307fb7a980f4cff77

data/README.md

@@ -12,10 +12,10 @@ the process involves running an Analyzer, that outputs a series of files.
 Among these files are a some XML files that describe the data objects that 
 are in the existing Fedora 2 database. These files need to be manipulated
 so as to modify the Fedora 2 objects so that they will suitable to be inserted
-into a Fedora 3 database.
+into a Fedora 3 database. Analyser also creates a set of cmodel-n-deployments.txt 
+files that also need to be manipulated
 
-This app will carry out the manipulation of the XML files generated by the 
-Analyser
+This app will carry out the manipulation of the files generated by the Analyser
 
 ## Installation
 
@@ -29,8 +29,10 @@ To use this app, you must first create a config.yml file. For example:
 
 ```
 changes:
-  CModel1: thesis
-  CModel1-SDep1: thesis-SDep1
+  CModel1: book
+  CModel1-SDep1: book-SDep1
+  CModel2: thesis
+  CModel2-SDep1: thesis-SDep1
 
 changeme: foo
 
@@ -67,12 +69,14 @@ namespace.
 
 Defines the paths to the input and output folders. 
 
-The input folder should contain the XML files generated by the Analyser
+The input folder should contain the files generated by the Analyser
 
 An output folder is used so that Analyser files are not directly modified by
 this process. The files can be deleted and the process repeated any number
 of times until the output is satisfactory.
 
+Only files that are changed by the process, will be copied to the output folder.
+
 ### locations
 
 The locations in the XML files, where the content should be modified. Each 

data/lib/fedora_2_to_3_pid_renamer/config.rb

@@ -1,3 +1,6 @@
+# The Config object is used to read a config file and pass around the settings
+# defined in that file.
+# See the README.md and test/data/config.xml for sample config.yml files.
 require 'yaml'
 module Fedora2To3PidRenamer
   class Config

data/lib/fedora_2_to_3_pid_renamer/manager.rb

@@ -1,4 +1,5 @@
-
+# Manager manages the process of modify a group of XML file as defined via
+# a config object
 module Fedora2To3PidRenamer
   class Manager
     
@@ -17,15 +18,29 @@ module Fedora2To3PidRenamer
       create_output_folder
       Dir.glob(input_file_pattern).each do |input_file|
         file_name = File.basename input_file
+        manipulator = manipulator_for(file_name)
+        next unless manipulator
+        
         input = File.read input_file        
         output_file = File.join config.output_folder, file_name
-        File.write output_file, Manipulator.output_for(input, config)
-        puts output_file
+        
+        File.write output_file, manipulator.output_for(input, config)
       end
     end
     
     def input_file_pattern
-      File.join config.input_folder, '*.xml'
+      File.join config.input_folder, '*'
+    end
+    
+    def manipulator_for(file_name)
+      case file_name
+      when /\.xml$/
+        Manipulator::Xml
+      when /cmodel-\d+\.deployments.txt/
+        Manipulator::Text
+      else
+        nil
+      end
     end
   end
 end

data/lib/fedora_2_to_3_pid_renamer/manipulator/base.rb

@@ -0,0 +1,45 @@
+
+
+module Fedora2To3PidRenamer::Manipulator
+  class Base
+    attr_reader :source, :config
+    
+    def self.output_for(*args)
+      manipulator = new(*args)
+      manipulator.run
+      manipulator.output
+    end
+    
+    def initialize(source, config)
+      @source = source
+      @config = config
+    end
+    
+    def run
+      raise 'run must be defined in sub-classes'
+    end
+    
+    def output
+      raise 'output must be defined in sub-classes'
+    end
+    
+    private
+    def change(text, before, after)
+      pattern = Regexp.new("changeme:#{before}(#{end_of_string_or_non_word})")      
+      replacement = "#{changeme_replacement}:#{after}#{first_capture_group}"     
+      text.gsub! pattern, replacement
+    end
+    
+    def end_of_string_or_non_word 
+      '$|\W'
+    end
+    
+    def first_capture_group
+      '\1'
+    end
+    
+    def changeme_replacement
+      config.changeme_replacement
+    end
+  end
+end

data/lib/fedora_2_to_3_pid_renamer/manipulator/text.rb

@@ -0,0 +1,19 @@
+
+module Fedora2To3PidRenamer::Manipulator
+  class Text < Base
+    
+    alias_method :text, :source
+    
+    def run
+      config.changes.each do |before, after|
+        change(text, before, after)
+      end
+    end
+    
+    def output
+      text
+    end
+    
+    
+  end
+end

data/lib/fedora_2_to_3_pid_renamer/manipulator/xml.rb

@@ -0,0 +1,38 @@
+# Manipulator is the tool used to modify the XML files passed to it based on
+# the settings defined in the config.
+require 'nokogiri'
+module Fedora2To3PidRenamer::Manipulator
+  class Xml < Base
+
+    alias_method :raw_xml, :source
+
+    def run
+      modify_text_at_each_location_in_config
+    end
+
+    def xml
+      @xml ||= Nokogiri::XML(raw_xml)
+    end
+    
+    def output
+      xml.to_s
+    end
+    
+    private
+    def modify_text_at_each_location_in_config
+      config.locations.each do |location|
+
+        node = xml.xpath(location, config.namespaces).first
+        next unless node
+        
+        text = node.value
+        config.changes.each do |before, after|
+          change(text, before, after)
+        end       
+        node.value = text
+
+      end
+    end
+
+  end
+end

data/lib/fedora_2_to_3_pid_renamer/manipulator.rb

@@ -1,50 +1,8 @@
-require 'nokogiri'
+require_relative 'manipulator/base'
+require_relative 'manipulator/text'
+require_relative 'manipulator/xml'
 module Fedora2To3PidRenamer
-  class Manipulator
-
-    attr_accessor :raw_xml, :config
-    
-    def self.output_for(raw_xml, config)
-      manipulator = new(raw_xml, config)
-      manipulator.run
-      manipulator.output
-    end
-
-    def initialize(raw_xml, config)
-      @raw_xml = raw_xml
-      @config = config
-    end
-
-    def run
-      modify_text_at_each_location_in_config
-    end
-
-    def xml
-      @xml ||= Nokogiri::XML(raw_xml)
-    end
-    
-    def output
-      xml.to_s
-    end
-    
-    private
-    def modify_text_at_each_location_in_config
-      config.locations.each do |location|
-
-        node = xml.xpath(location, config.namespaces).first
-        next unless node
-        
-        text = node.value
-        config.changes.each do |before, after|
-          text.gsub! before, after
-        end
-
-        text.gsub! 'changeme:', (config.changeme_replacement + ':')
-        
-        node.value = text
-
-      end
-    end
-
+  module Manipulator
+      
   end
 end

data/lib/fedora_2_to_3_pid_renamer/version.rb

@@ -1,3 +1,11 @@
 module Fedora2To3PidRenamer
-  VERSION = '0.0.1'
+  VERSION = '0.0.2'
 end
+
+# History
+# =======
+# 
+# 0.0.2 - Adds modification of cmodel-n-deployments.txt 
+#  
+# 0.0.1 - Initial working app
+#

data/lib/fedora_2_to_3_pid_renamer.rb

@@ -3,5 +3,5 @@ require_relative 'fedora_2_to_3_pid_renamer/manipulator'
 require_relative 'fedora_2_to_3_pid_renamer/manager'
 
 module Fedora2To3PidRenamer
-  # TODO Put your code here
+  
 end

data/test/data/cmodel-1.deployments.txt

@@ -0,0 +1,7 @@
+# The following BMechs will be copied and written as FOXML1.1 service deployments
+# with new PIDs and part names as given below
+OLD_BMECH demo:9
+NEW_DEPLOYMENTS changeme:CModel1-SDep1
+NEW_PARTS MRSID=DS1
+
+

data/test/data/cmodel-1.members.txt

@@ -0,0 +1,30 @@
+# The following objects will be assigned to cmodel-1
+# <class-description>
+# OrigContentModel
+#   'UVA_MRSID_IMAGE'
+# BDefPIDs
+#   'demo:8'
+# BMechPIDs
+#   'demo:9'
+# BindingKeyAssignments
+#   for demo:9
+#     'MRSID=DS1'
+# DatastreamIDs
+#   'DIGIPROV1', 'DS1', 'SOURCE1', 'DESC1', 'RIGHTS1', 'TECH1'
+# MIMETypes
+#   for DIGIPROV1, 'text/xml'
+#   for DS1, 'image/x-mrsid-image'
+#   for SOURCE1, 'text/xml'
+#   for DESC1, 'text/xml'
+#   for RIGHTS1, 'text/xml'
+#   for TECH1, 'text/xml'
+# FormatURIs
+#   for DIGIPROV1, none
+#   for DS1, none
+#   for SOURCE1, none
+#   for DESC1, none
+#   for RIGHTS1, none
+#   for TECH1, none
+# </class-description>
+demo:11
+demo:10

data/test/units/fedora_2_to_3_pid_renamer/manager_test.rb

@@ -4,6 +4,8 @@ require 'fileutils'
 module Fedora2To3PidRenamer
   class ManagerTest < Minitest::Test
     
+    # Creates data/input and data/output and copies test files into 
+    # data/input for start of test
     def setup
       FileUtils.mkdir data_file_path('input')
       FileUtils.mkdir data_file_path('output')
@@ -14,8 +16,9 @@ module Fedora2To3PidRenamer
     end
 
     def teardown
-      FileUtils.remove_dir data_file_path('input'), true
-      FileUtils.remove_dir data_file_path('output'), true
+      force = true
+      FileUtils.remove_dir data_file_path('input'), force
+      FileUtils.remove_dir data_file_path('output'), force
     end
     
     def test_before
@@ -36,7 +39,7 @@ module Fedora2To3PidRenamer
     
     def test_run_manipulation
       manager.run_manipulation
-      target_files.each do |file|
+      target_file_to_be_modified.each do |file|
         output_file = data_file_path(File.join('output', file))
         assert File.exist?(output_file), "#{output_file} should exist"
       end
@@ -44,7 +47,7 @@ module Fedora2To3PidRenamer
     
     def test_run_manipulation_alters_content
       manager.run_manipulation
-      xml = xml_load File.join('output', target_files.first)
+      xml = xml_load File.join('output', xml_files_to_be_modified.first)
       xpath = config.locations.first
       text = xml.xpath(xpath).text
       assert_match config.changeme_replacement, text
@@ -55,9 +58,29 @@ module Fedora2To3PidRenamer
     end
     
     def target_files
+      target_file_to_be_modified + target_file_to_be_skipped
+    end
+    
+    def xml_files_to_be_modified
       [
         'cmodel-1.xml', 
-        'cmodel-1.deployment1.xml'
+        'cmodel-1.deployment1.xml'        
+      ]
+    end
+    
+    def text_file_to_be_modified
+      [
+        'cmodel-1.deployments.txt'
+      ]
+    end
+    
+    def target_file_to_be_modified
+      text_file_to_be_modified + xml_files_to_be_modified
+    end
+    
+    def target_file_to_be_skipped
+      [
+        'cmodel-1.members.txt'
       ]
     end
 

data/test/units/fedora_2_to_3_pid_renamer/manipulator/text_test.rb

@@ -0,0 +1,37 @@
+require_relative '../../../test_helper'
+
+module Fedora2To3PidRenamer::Manipulator
+  class TextTest < Minitest::Test
+
+    def test_run
+      before = 'changeme:CModel1-SDep1'
+      after = [
+        config.changeme_replacement,
+        config.change_for('CModel1-SDep1')
+      ].join(':')
+      
+      
+      assert_match before, text_manipulator.text
+      refute_match after, text_manipulator.text
+      
+      text_manipulator.run
+      
+      refute_match before, text_manipulator.text
+      assert_match after, text_manipulator.text
+    end
+    
+    def test_text
+      assert_equal file_text, text_manipulator.text
+    end
+    
+    private
+    def text_manipulator
+      @text_manipulator ||= Text.new(file_text, config)
+    end
+    
+    def file_text
+      @file_text ||= file_read 'cmodel-1.deployments.txt'
+    end
+    
+  end
+end

data/test/units/fedora_2_to_3_pid_renamer/{manipulator_test.rb → manipulator/xml_test.rb}

@@ -1,57 +1,57 @@
-require_relative '../../test_helper'
+require_relative '../../../test_helper'
 
-module Fedora2To3PidRenamer
-  class ManipulatorTest < Minitest::Test
+module Fedora2To3PidRenamer::Manipulator
+  class XmlTest < Minitest::Test
 
     def test_run
       before, after = config.changes.first      
-      assert_at_xpath_text_changes(before, after) { manipulator.run }
+      assert_at_xpath_text_changes(before, after) { xml_manipulator.run }
     end
     
     def test_run_modifies_changeme_namespace
       before = 'changeme:'
       after = config.changeme_replacement + ':'      
-      assert_at_xpath_text_changes(before, after) { manipulator.run }
+      assert_at_xpath_text_changes(before, after) { xml_manipulator.run }
     end
     
     def test_output
-      assert_equal Nokogiri::XML(xml_raw).to_s, manipulator.output
+      assert_equal Nokogiri::XML(xml_raw).to_s, xml_manipulator.output
     end
     
     def test_run_alters_output
-      manipulator.run
-      refute_equal Nokogiri::XML(xml_raw).to_s, manipulator.output
+      xml_manipulator.run
+      refute_equal Nokogiri::XML(xml_raw).to_s, xml_manipulator.output
     end
     
     def test_run_produces_well_formed_xml
-      manipulator.run
-      doc = Nokogiri::XML manipulator.output
+      xml_manipulator.run
+      doc = Nokogiri::XML xml_manipulator.output
       assert doc.errors.empty?, "Nokogiri should not find errors: #{doc.errors}"
     end
     
     def test_output_for
-      manipulator.run
-      assert_equal manipulator.output, Manipulator.output_for(xml_raw, config)
+      xml_manipulator.run
+      assert_equal xml_manipulator.output, Xml.output_for(xml_raw, config)
     end
     
     private
     def assert_at_xpath_text_changes(before, after)
       xpath = config.locations.first
       
-      text_before = manipulator.xml.xpath(xpath, config.namespaces).text
+      text_before = xml_manipulator.xml.xpath(xpath, config.namespaces).text
       assert_match before, text_before
       refute_match after, text_before
       
       yield
       
-      text_after = manipulator.xml.xpath(xpath, config.namespaces).text
+      text_after = xml_manipulator.xml.xpath(xpath, config.namespaces).text
       refute_match before, text_after
       assert_match after, text_after
     end
     
 
-    def manipulator
-      @manipulator ||= Manipulator.new(xml_raw, config)
+    def xml_manipulator
+      @xml_manipulator ||= Xml.new(xml_raw, config)
     end
 
     def xml_raw

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: fedora_2_to_3_pid_renamer
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors:
 - Rob Nichols
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-09-24 00:00:00.000000000 Z
+date: 2015-09-25 00:00:00.000000000 Z
 dependencies: []
 description: Interates through a series of xml files and modifies their content based
   on the setting defined in a configuration file
@@ -30,15 +30,21 @@ files:
 - lib/fedora_2_to_3_pid_renamer/config.rb
 - lib/fedora_2_to_3_pid_renamer/manager.rb
 - lib/fedora_2_to_3_pid_renamer/manipulator.rb
+- lib/fedora_2_to_3_pid_renamer/manipulator/base.rb
+- lib/fedora_2_to_3_pid_renamer/manipulator/text.rb
+- lib/fedora_2_to_3_pid_renamer/manipulator/xml.rb
 - lib/fedora_2_to_3_pid_renamer/version.rb
 - test/data/cmodel-1.deployment1.xml
+- test/data/cmodel-1.deployments.txt
+- test/data/cmodel-1.members.txt
 - test/data/cmodel-1.xml
 - test/data/config.yml
 - test/data/simple.xml
 - test/test_helper.rb
 - test/units/fedora_2_to_3_pid_renamer/config_test.rb
 - test/units/fedora_2_to_3_pid_renamer/manager_test.rb
-- test/units/fedora_2_to_3_pid_renamer/manipulator_test.rb
+- test/units/fedora_2_to_3_pid_renamer/manipulator/text_test.rb
+- test/units/fedora_2_to_3_pid_renamer/manipulator/xml_test.rb
 - test/units/fedora_2_to_3_pid_renamer_test.rb
 homepage: https://github.com/reggieb/fedora_2_to_3_pid_renamer
 licenses: []
@@ -66,11 +72,14 @@ summary: A small Ruby app used to apply alternative pid names to configuration f
   created during the migration of fedora 2 data to fedora 3
 test_files:
 - test/data/cmodel-1.deployment1.xml
+- test/data/cmodel-1.deployments.txt
+- test/data/cmodel-1.members.txt
 - test/data/cmodel-1.xml
 - test/data/config.yml
 - test/data/simple.xml
 - test/test_helper.rb
 - test/units/fedora_2_to_3_pid_renamer/config_test.rb
 - test/units/fedora_2_to_3_pid_renamer/manager_test.rb
-- test/units/fedora_2_to_3_pid_renamer/manipulator_test.rb
+- test/units/fedora_2_to_3_pid_renamer/manipulator/text_test.rb
+- test/units/fedora_2_to_3_pid_renamer/manipulator/xml_test.rb
 - test/units/fedora_2_to_3_pid_renamer_test.rb