wordpress-exporter 0.0.1

data/spec/lib/configuration_spec.rb

@@ -0,0 +1,23 @@
+require 'spec_helper'
+require 'yaml'
+
+module Contentful
+  describe Configuration do
+    it 'initialize' do
+      yaml_text = <<-EOF
+      data_dir: path_to_data_dir
+      wordpress_xml_path: path_to_xml_file.xml
+      contentful_structure_dir: spec/fixtures/default_contentful_structure.json
+      EOF
+
+      yaml = YAML.load(yaml_text)
+      configuration = Contentful::Configuration.new(yaml)
+
+      expect(configuration.assets_dir).to eq 'path_to_data_dir/assets'
+      expect(configuration.collections_dir).to eq 'path_to_data_dir/collections'
+      expect(configuration.data_dir).to eq 'path_to_data_dir'
+      expect(configuration.entries_dir).to eq 'path_to_data_dir/entries'
+      expect(configuration.wordpress_xml).to eq 'path_to_xml_file.xml'
+    end
+  end
+end

data/spec/lib/converters/markup_converter_spec.rb

@@ -0,0 +1,27 @@
+require 'spec_helper'
+require './lib/converters/markup_converter'
+require 'logger'
+
+module Contentful
+  module Converter
+    describe MarkupConverter do
+      include_context 'shared_configuration'
+
+      before do
+        @converter = MarkupConverter.new(@settings)
+      end
+
+      it 'convert_markup_to_markdown' do
+        expect_any_instance_of(MarkupConverter).to receive(:convert_post_content).exactly(7).times
+        @converter.convert_markup_to_markdown
+      end
+
+      it 'convert post content' do
+        allow(File).to receive(:open)
+        allow(File).to receive(:read).with('post_file_path')
+        allow(JSON).to receive(:parse) { { 'content' => '<strong>TEST</strong>' } }
+        @converter.convert_post_content('post_file_path')
+      end
+    end
+  end
+end

data/spec/lib/wordpress/blog_spec.rb

@@ -0,0 +1,64 @@
+require 'spec_helper'
+require './lib/wordpress/blog'
+require './lib/wordpress/post'
+require './lib/wordpress/post_attachment'
+require './lib/wordpress/post_category_domain'
+require './lib/wordpress/category'
+require './lib/wordpress/tag'
+
+module Contentful
+  module Exporter
+    module Wordpress
+      describe Blog do
+        include_context 'shared_configuration'
+
+        before do
+          xml_doc = Nokogiri::XML(File.open('spec/fixtures/wordpress.xml'))
+          @blog = Blog.new(xml_doc, @settings)
+        end
+
+        it 'initialize' do
+          expect(@blog.xml).to be_kind_of Nokogiri::XML::Document
+          expect(@blog.settings).to be_kind_of Contentful::Configuration
+        end
+
+        it 'blog_extractor' do
+          allow(Blog).to receive(:extract_blog)
+        end
+
+        it 'link_entry' do
+          entry = [{ test: 'remove', id: 'entry_id' }, { test: 'remove', id: 'entry_id_2' }]
+          link_entry = @blog.link_entry(entry)
+          expect(link_entry.count).to eq 2
+          expect(link_entry).to include(id: 'entry_id', type: 'Entry')
+        end
+
+        it 'link_asset' do
+          asset = { test: 'remove', id: 'asset_id' }
+          link_asset = @blog.link_asset(asset)
+          expect(link_asset.count).to eq 2
+          expect(link_asset).to include(id: 'asset_id', type: 'File')
+        end
+
+        it 'create_directory' do
+          allow(File).to receive(:directory?).and_return(true)
+          allow(File).to receive(:mkdir_p).with('/test/test.json').and_return(true)
+          @blog.create_directory('/test/test.json')
+        end
+
+        it 'valid blog entry' do
+          blog = JSON.parse(File.read('spec/fixtures/blog/entries/blog/blog_1.json'))
+          expect(blog.count).to eq 5
+          expect(blog['id']).to eq 'blog_id'
+          expect(blog['title']).to eq 'Moj blog 2'
+          expect(blog['posts'].count).to eq 7
+          expect(blog['posts'].first).to include('id' => 'post_1', 'type' => 'Entry')
+          expect(blog['categories'].count).to eq 4
+          expect(blog['categories'].first).to include('id' => 'category_14786', 'type' => 'Entry')
+          expect(blog['tags'].count).to eq 2
+          expect(blog['tags'].first).to include('id' => 'tag_2656354', 'type' => 'Entry')
+        end
+      end
+    end
+  end
+end

data/spec/lib/wordpress/category_spec.rb

@@ -0,0 +1,39 @@
+require 'spec_helper'
+require './lib/wordpress/category'
+require './lib/wordpress/blog'
+
+module Contentful
+  module Exporter
+    module Wordpress
+      describe Category do
+        include_context 'shared_configuration'
+
+        before do
+          xml_doc = Nokogiri::XML(File.open('spec/fixtures/wordpress.xml'))
+          @category = Category.new(xml_doc, @settings)
+        end
+
+        it 'initialize' do
+          expect(@category.settings).to be_kind_of Contentful::Configuration
+          expect(@category.xml).to be_kind_of Nokogiri::XML::Document
+        end
+
+        it 'extract_categories' do
+          categories = @category.send(:extract_categories)
+          expect(categories).to be_kind_of Array
+          expect(categories.count).to eq 4
+          expect(categories.first).to include(id: 'category_14786', nicename: 'bez-kategorii', name: 'Bez kategorii')
+          expect(categories.last).to include(id: 'category_11599757', nicename: 'puchatka', name: 'puchatka')
+        end
+
+        it 'extracted_category' do
+          categories_xml = @category.xml.xpath('//wp:category').to_a
+          category = @category.send(:extracted_category, categories_xml.first)
+          expect(category).to be_kind_of Hash
+          expect(category.count).to eq 3
+          expect(category).to include(id: 'category_14786', nicename: 'bez-kategorii', name: 'Bez kategorii')
+        end
+      end
+    end
+  end
+end

data/spec/lib/wordpress/export_spec.rb

@@ -0,0 +1,27 @@
+require 'spec_helper'
+require 'yaml'
+require './lib/wordpress/export'
+
+module Contentful
+  module Exporter
+    module Wordpress
+      describe Export do
+        include_context 'shared_configuration'
+
+        before do
+          @exporter = Export.new(@settings)
+        end
+
+        it 'initialize' do
+          expect(@exporter.settings).to be_kind_of Contentful::Configuration
+          expect(@exporter.wordpress_xml).to be_kind_of Nokogiri::XML::Document
+        end
+
+        it 'export_blog' do
+          @exporter.export_blog
+          expect(Dir.glob('spec/fixtures/blog/**/*').count).to eq 23
+        end
+      end
+    end
+  end
+end

data/spec/lib/wordpress/post_category_domain_spec.rb

@@ -0,0 +1,41 @@
+require 'spec_helper'
+require './lib/wordpress/post_category_domain'
+require './lib/wordpress/post'
+require './lib/wordpress/blog'
+
+module Contentful
+  module Exporter
+    module Wordpress
+      describe PostCategoryDomain do
+        include_context 'shared_configuration'
+
+        before do
+          xml_doc = Nokogiri::XML(File.open('spec/fixtures/wordpress.xml'))
+          post = xml_doc.xpath('//item').to_a[4]
+          @pc_domain = PostCategoryDomain.new(xml_doc, post, @settings)
+        end
+
+        it 'initialize' do
+          expect(@pc_domain.settings).to be_kind_of Contentful::Configuration
+          expect(@pc_domain.xml).to be_kind_of Nokogiri::XML::Document
+        end
+
+        it 'extract_tags' do
+          tags = @pc_domain.send(:extract_tags)
+          expect(tags).to be_kind_of Array
+          expect(tags.count).to eq 2
+          expect(tags.first).to include(id: 'tag_2656354')
+          expect(tags.last).to include(id: 'tag_306830130')
+        end
+
+        it 'extract_categories' do
+          categories = @pc_domain.send(:extract_categories)
+          expect(categories).to be_kind_of Array
+          expect(categories.count).to eq 3
+          expect(categories.first).to include(id: 'category_2214351')
+          expect(categories.last).to include(id: 'category_11599757')
+        end
+      end
+    end
+  end
+end

data/spec/lib/wordpress/post_spec.rb

@@ -0,0 +1,41 @@
+require 'spec_helper'
+require './lib/wordpress/post'
+require './lib/wordpress/blog'
+require './lib/wordpress/post_attachment'
+require './lib/wordpress/post_category_domain'
+
+module Contentful
+  module Exporter
+    module Wordpress
+      describe Post do
+        include_context 'shared_configuration'
+
+        before do
+          @xml_doc = Nokogiri::XML(File.open('spec/fixtures/wordpress.xml'))
+          @post = Post.new(@xml_doc, @settings)
+        end
+
+        it 'initialize' do
+          expect(@post.settings).to be_kind_of Contentful::Configuration
+          expect(@post.xml).to be_kind_of Nokogiri::XML::Document
+        end
+
+        it 'extract_data' do
+          post_xml = @xml_doc.xpath('//item').to_a.first
+          post = @post.send(:extract_data, post_xml)
+          expect(post.count).to eq 5
+          expect(post[:id]).to eq 'post_1'
+          expect(post[:title]).to eq 'Informacje'
+          expect(post[:wordpress_url]).to eq 'http://szpryc.wordpress.com/informacje/'
+          expect(post[:created_at]).to eq Date.parse('2014-11-26')
+        end
+
+        it 'post_id(post)' do
+          post = @xml_doc.xpath('//item').to_a.first
+          post_id = @post.post_id(post)
+          expect(post_id).to eq 'post_1'
+        end
+      end
+    end
+  end
+end

data/spec/lib/wordpress/tag_spec.rb

@@ -0,0 +1,39 @@
+require 'spec_helper'
+require './lib/wordpress/tag'
+require './lib/wordpress/blog'
+
+module Contentful
+  module Exporter
+    module Wordpress
+      describe Tag do
+        include_context 'shared_configuration'
+
+        before do
+          xml_doc = Nokogiri::XML(File.open('spec/fixtures/wordpress.xml'))
+          @tag = Tag.new(xml_doc, @settings)
+        end
+
+        it 'initialize' do
+          expect(@tag.settings).to be_kind_of Contentful::Configuration
+          expect(@tag.xml).to be_kind_of Nokogiri::XML::Document
+        end
+
+        it 'extract_tags' do
+          tags = @tag.send(:extract_tags)
+          expect(tags).to be_kind_of Array
+          expect(tags.count).to eq 2
+          expect(tags.first).to include(id: 'tag_2656354', nicename: 'testowy', name: 'testowy')
+          expect(tags.last).to include(id: 'tag_306830130', nicename: 'testowy2', name: 'testowy2')
+        end
+
+        it 'extracted_data' do
+          tags_xml = @tag.xml.xpath('//wp:tag').to_a
+          tag = @tag.send(:extracted_data, tags_xml.first)
+          expect(tag).to be_kind_of Hash
+          expect(tag.count).to eq 3
+          expect(tag).to include(id: 'tag_2656354', nicename: 'testowy', name: 'testowy')
+        end
+      end
+    end
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,13 @@
+require 'simplecov'
+SimpleCov.start 'rails'
+
+require 'bundler/setup'
+Bundler.setup
+
+require 'rspec'
+require 'rspec/its'
+require 'yaml'
+require 'json'
+require 'nokogiri'
+
+Dir[File.dirname(__FILE__) + '/support/*.rb'].each { |f| require f }

data/spec/support/db_rows_json.rb

@@ -0,0 +1,5 @@
+require 'multi_json'
+
+def load_json(file, _as_json = false)
+  JSON.parse(File.read(File.dirname(__FILE__) + "/../fixtures/#{file}.json"))
+end

data/spec/support/shared_configuration.rb

@@ -0,0 +1,13 @@
+require_relative '../../lib/configuration'
+
+RSpec.shared_context 'shared_configuration', a: :b do
+  before do
+    yaml_text = <<-EOF
+          data_dir: spec/fixtures/blog
+          wordpress_xml_path: spec/fixtures/wordpress.xml
+          contentful_structure_dir: spec/fixtures//default_contentful_structure.json
+    EOF
+    yaml = YAML.load(yaml_text)
+    @settings = Contentful::Configuration.new(yaml)
+  end
+end

data/wordpress_exporter.gemspec

@@ -0,0 +1,33 @@
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+
+require File.expand_path('../lib/version', __FILE__)
+
+Gem::Specification.new do |spec|
+  spec.name          = 'wordpress-exporter'
+  spec.version       = Version::VERSION
+  spec.authors       = ['Contentful GmbH (Andreas Tiefenthaler)']
+  spec.email         = ['rubygems@contentful.com']
+  spec.description   = 'WordPress exporter that prepares content to be imported'
+  spec.summary       = 'Exporter for WordPress based blogs'
+  spec.homepage      = 'https://github.com/contentful/wordpress-exporter.rb'
+  spec.license       = 'MIT'
+
+  spec.files         = `git ls-files -z`.split("\x0")
+  spec.executables    << 'wordpress-exporter'
+  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
+  spec.require_paths = ['lib']
+
+  spec.add_dependency 'http', '~> 0.6'
+  spec.add_dependency 'multi_json', '~> 1'
+  spec.add_dependency 'activesupport', '~> 4.1'
+  spec.add_dependency 'escort', '~> 0.4.0'
+  spec.add_dependency 'i18n', '~> 0.6'
+  spec.add_dependency 'nokogiri', '~> 1.6.3.1'
+  spec.add_dependency 'reverse_markdown', '~> 0.6.0'
+
+  spec.add_development_dependency 'rspec', '~> 3'
+  spec.add_development_dependency 'rspec-its', '~> 1.1.0'
+  spec.add_development_dependency 'bundler', '~> 1.6'
+  spec.add_development_dependency 'rake'
+end

data/wordpress_settings/contentful_model.json

@@ -0,0 +1,288 @@
+{
+  "sys": {
+    "type": "Array"
+  },
+  "total": 4,
+  "skip": 0,
+  "limit": 100,
+  "items": [
+    {
+      "fields": [
+        {
+          "name": "posts",
+          "id": "posts",
+          "type": "Array",
+          "items": {
+            "type": "Link",
+            "linkType": "Entry"
+          }
+        },
+        {
+          "name": "title",
+          "id": "title",
+          "type": "Text",
+          "localized": true
+        },
+        {
+          "name": "categories",
+          "id": "categories",
+          "type": "Array",
+          "items": {
+            "type": "Link",
+            "linkType": "Entry"
+          }
+        },
+        {
+          "name": "tags",
+          "id": "tags",
+          "type": "Array",
+          "items": {
+            "type": "Link",
+            "linkType": "Entry"
+          }
+        }
+      ],
+      "name": "Blog",
+      "sys": {
+        "id": "3gldgGyAyci8ak0cykKeou",
+        "type": "ContentType",
+        "createdAt": "2014-11-26T13:49:24.203Z",
+        "createdBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "space": {
+          "sys": {
+            "type": "Link",
+            "linkType": "Space",
+            "id": "1j2osuxmfs25"
+          }
+        },
+        "firstPublishedAt": "2014-11-26T13:49:45.142Z",
+        "publishedCounter": 4,
+        "publishedAt": "2014-12-01T09:59:19.695Z",
+        "publishedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "publishedVersion": 46,
+        "version": 47,
+        "updatedAt": "2014-12-01T09:59:19.743Z",
+        "updatedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        }
+      },
+      "displayField": "title"
+    },
+    {
+      "fields": [
+        {
+          "name": "nicename",
+          "id": "nicename",
+          "type": "Text"
+        },
+        {
+          "name": "name",
+          "id": "name",
+          "type": "Text"
+        }
+      ],
+      "name": "Tag",
+      "sys": {
+        "id": "wO1krKenNQcWMSqaeysWo",
+        "type": "ContentType",
+        "createdAt": "2014-12-01T09:25:56.175Z",
+        "createdBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "space": {
+          "sys": {
+            "type": "Link",
+            "linkType": "Space",
+            "id": "1j2osuxmfs25"
+          }
+        },
+        "firstPublishedAt": "2014-12-01T09:26:45.362Z",
+        "publishedCounter": 2,
+        "publishedAt": "2014-12-01T09:27:45.169Z",
+        "publishedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "publishedVersion": 44,
+        "version": 45,
+        "updatedAt": "2014-12-01T09:27:45.216Z",
+        "updatedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        }
+      },
+      "description": null,
+      "displayField": "nicename"
+    },
+    {
+      "fields": [
+        {
+          "name": "nicename",
+          "id": "nicename",
+          "type": "Text"
+        },
+        {
+          "name": "name",
+          "id": "name",
+          "type": "Text"
+        }
+      ],
+      "name": "Category",
+      "sys": {
+        "id": "5PIooPqWmAcYyQ04c2KUcQ",
+        "type": "ContentType",
+        "createdAt": "2014-12-01T09:26:50.679Z",
+        "createdBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "space": {
+          "sys": {
+            "type": "Link",
+            "linkType": "Space",
+            "id": "1j2osuxmfs25"
+          }
+        },
+        "firstPublishedAt": "2014-12-01T09:27:16.440Z",
+        "publishedCounter": 3,
+        "publishedAt": "2014-12-01T10:44:56.239Z",
+        "publishedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "publishedVersion": 31,
+        "version": 32,
+        "updatedAt": "2014-12-01T10:44:56.283Z",
+        "updatedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        }
+      },
+      "displayField": "name"
+    },
+    {
+      "fields": [
+        {
+          "name": "title",
+          "id": "title",
+          "type": "Text"
+        },
+        {
+          "name": "content",
+          "id": "content",
+          "type": "Text"
+        },
+        {
+          "name": "tags",
+          "id": "tags",
+          "type": "Array",
+          "items": {
+            "type": "Link",
+            "linkType": "Entry"
+          }
+        },
+        {
+          "name": "categories",
+          "id": "categories",
+          "type": "Array",
+          "items": {
+            "type": "Link",
+            "linkType": "Entry"
+          }
+        },
+        {
+          "name": "wordpress_url",
+          "id": "wordpress_url",
+          "type": "Text"
+        },
+        {
+          "name": "attachment",
+          "id": "attachment",
+          "type": "Link",
+          "linkType": "Asset"
+        },
+        {
+          "name": "created_at",
+          "id": "created_at",
+          "type": "Date"
+        }
+      ],
+      "name": "Post",
+      "sys": {
+        "id": "KHqITFZXeCIokyOEc4k0k",
+        "type": "ContentType",
+        "createdAt": "2014-11-26T13:15:08.574Z",
+        "createdBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "space": {
+          "sys": {
+            "type": "Link",
+            "linkType": "Space",
+            "id": "1j2osuxmfs25"
+          }
+        },
+        "firstPublishedAt": "2014-11-26T13:16:17.537Z",
+        "publishedCounter": 6,
+        "publishedAt": "2015-01-07T10:25:09.043Z",
+        "publishedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        },
+        "publishedVersion": 92,
+        "version": 93,
+        "updatedAt": "2015-01-07T10:25:09.102Z",
+        "updatedBy": {
+          "sys": {
+            "type": "Link",
+            "linkType": "User",
+            "id": "1E7acJL8I5XUXAMHQt9Grs"
+          }
+        }
+      },
+      "displayField": "title"
+    }
+  ]
+}