wiki-api 0.0.2 → 0.1.0

data/test/unit/wiki_page_offline.rb

@@ -0,0 +1,262 @@
+# encoding: utf-8
+
+require 'rubygems'
+require 'test/unit'
+require File.expand_path(File.dirname(__FILE__) + "/../../lib/wiki/api")
+
+#
+# Testing the parsing of URI (with a predownloaded HTML file):
+#   /files/Wiktionary_Welcome,_newcomers.html (2013-03-27)
+#
+# Online equivalent:
+#   https://en.wiktionary.org/wiki/Wiktionary:Welcome,_newcomers
+#
+
+class WikiPageOfflinePage < Test::Unit::TestCase
+
+  # this global is required to resolve URIs (MediaWiki uses relative paths in their links)
+  URI_CONFIG = { uri: "http://en.wiktionary.org" }
+  # use local file for test loading
+  PAGE_CONFIG = { file: File.expand_path(File.dirname(__FILE__) + "/files/Wiktionary_program.html") }
+
+  def setup
+    # NOTE: comment Page.config, to use the online MediaWiki instance
+    Wiki::Api::Connect.config = URI_CONFIG
+    Wiki::Api::Connect.config.merge! PAGE_CONFIG
+  end
+
+  def teardown
+  end
+
+  # test simple page invocation
+  def test_headlines_nested
+
+    # load page
+    page = Wiki::Api::Page.new name: "program"
+    assert page.is_a?(Wiki::Api::Page), "expected Page object"
+    headline = page.root_headline
+    assert headline.is_a?(Wiki::Api::PageHeadline), "expected PageHeadline object"
+    assert headline.name == "program", "expected developer headline"
+
+    # search nested headline: english   
+    english = headline.headline("english").first
+    assert english.is_a?(Wiki::Api::PageHeadline), "expected PageHeadline object"
+
+    # search nested headline: noun
+    noun = english.headline("noun").first
+    assert noun.is_a?(Wiki::Api::PageHeadline), "expected PageHeadline object"
+
+    # get block
+    block = noun.block
+    assert block.is_a?(Wiki::Api::PageBlock), "expected PageBlock object"
+
+    # list items
+    block.list_items.each do |list_item|
+      assert list_item.is_a?(Wiki::Api::PageListItem), "expected PageListItem object"
+      # links
+      list_item.links.each do |link|
+        assert link.is_a?(Wiki::Api::PageLink), "expected PageListItem object"
+      end
+    end
+
+    # links
+    block.links.each do |link|
+      assert link.is_a?(Wiki::Api::PageLink), "expected PageListItem object"
+    end
+  end
+
+
+  # def test_headlines_verify_tree_objects
+  #   page = Wiki::Api::Page.new name: "program"
+  #   assert page.is_a?(Wiki::Api::Page), "expected Page object"
+
+  #   # get root headline
+  #   headline = page.root_headline
+  #   assert headline.is_a?(Wiki::Api::PageHeadline), "expected PageHeadline object"
+  #   assert headline.name == "program", "expected program headline"
+
+  #   # search in depth on headline noun
+  #   nouns = headline.headline_by_name "noun", 1
+
+  #   nouns.each do |noun|
+  #     assert noun.is_a?(Wiki::Api::PageHeadline), "expected PageHeadline object"
+
+  #     # get block
+  #     block = noun.block
+  #     assert block.is_a?(Wiki::Api::PageBlock), "expected PageBlock object"
+
+  #     # list items
+  #     block.list_items.each do |list_item|
+  #       assert list_item.is_a?(Wiki::Api::PageListItem), "expected PageListItem object"
+  #       # links
+  #       list_item.links.each do |link|
+  #         assert link.is_a?(Wiki::Api::PageLink), "expected PageListItem object"
+  #       end
+  #     end
+
+  #     # links
+  #     block.links.each do |link|
+  #       assert link.is_a?(Wiki::Api::PageLink), "expected PageListItem object"
+  #     end
+  #   end
+  # end
+
+
+  def test_headlines_verify_tree_data
+    page = Wiki::Api::Page.new name: "program"
+    headline = page.root_headline
+    # get root
+    assert headline.name == "program", "expected program name"
+    
+    # iterate one deep and verify headline index names
+    headline.headlines.each do |name, headline|
+      assert headline.name == name
+    end
+    headlines = headline.headlines.values
+
+    # English
+    english_headlines = headlines[0]
+    assert english_headlines.name == "English"
+    english_headlines = english_headlines.headlines.values
+    assert english_headlines[0].name == "Alternative_forms"
+    assert english_headlines[1].name == "Etymology"
+    assert english_headlines[2].name == "Pronunciation"
+    assert english_headlines[3].name == "Noun"
+    noun_english_headlines = english_headlines[3].headlines.values
+    assert noun_english_headlines[0].name == "Usage_notes"
+    assert noun_english_headlines[1].name == "Synonyms"
+    assert noun_english_headlines[2].name == "Translations"
+    assert english_headlines[4].name == "Verb"
+    verb_english_headlines = english_headlines[4].headlines.values
+    assert verb_english_headlines[0].name == "Related_terms"
+    assert verb_english_headlines[1].name == "Translations_2"
+    assert english_headlines[5].name == "External_links"
+
+    # Czech
+    czech_headlines = headlines[1]
+    assert czech_headlines.name == "Czech"
+    czech_headlines = czech_headlines.headlines.values
+    assert czech_headlines[0].name == "Pronunciation_2"
+    assert czech_headlines[1].name == "Noun_2"
+
+    # Hungarian
+    hungarian_headlines = headlines[2]
+    assert hungarian_headlines.name == "Hungarian"
+    hungarian_headlines = hungarian_headlines.headlines.values
+    assert hungarian_headlines[0].name == "Pronunciation_3"
+    assert hungarian_headlines[1].name == "Noun_3"
+    noun_hungarian_headlines = hungarian_headlines[1].headlines.values
+    assert noun_hungarian_headlines[0].name == "Declension"
+    assert noun_hungarian_headlines[1].name == "Derived_terms"
+
+    # Norwegian
+    norwegian_headlines = headlines[3]
+    assert norwegian_headlines.name == "Norwegian_Bokm.C3.A5l"
+    norwegian_headlines = norwegian_headlines.headlines.values
+    assert norwegian_headlines[0].name == "Noun_4"
+
+    # Romanian
+    romanian_headlines = headlines[4]
+    assert romanian_headlines.name == "Romanian"
+    romanian_headlines = romanian_headlines.headlines.values
+    assert romanian_headlines[0].name == "Etymology_2"
+    assert romanian_headlines[1].name == "Noun_5"
+    noun_romanian_headlines = romanian_headlines[1].headlines.values
+    assert noun_romanian_headlines[0].name == "Declension_2"
+    assert noun_romanian_headlines[1].name == "Related_terms_2"
+
+    # Serbo-Croatian
+    serb_croatian_headlines = headlines[5]
+    assert serb_croatian_headlines.name == "Serbo-Croatian"
+    serb_croatian_headlines = serb_croatian_headlines.headlines.values
+    assert serb_croatian_headlines[0].name == "Noun_6"
+    noun_serb_croatian_headlines = serb_croatian_headlines[0].headlines.values
+    assert noun_serb_croatian_headlines[0].name == "Declension_3"
+
+    # Slovak
+    slovak_headlines = headlines[6]
+    assert slovak_headlines.name == "Slovak"
+    slovak_headlines = slovak_headlines.headlines.values
+    assert slovak_headlines[0].name == "Noun_7"
+
+    # Swedish
+    swedish_headlines = headlines[7]
+    assert swedish_headlines.name == "Swedish"
+    swedish_headlines = swedish_headlines.headlines.values
+    assert swedish_headlines[0].name == "Etymology_3"
+    assert swedish_headlines[1].name == "Noun_8"
+    noun_swedish_headlines = swedish_headlines[1].headlines.values
+    assert noun_swedish_headlines[0].name == "Declension_4"
+
+    # Turkish
+    turkish_headlines = headlines[8]
+    assert turkish_headlines.name == "Turkish"
+    turkish_headlines = turkish_headlines.headlines.values
+    assert turkish_headlines[0].name == "Etymology_4"
+    assert turkish_headlines[1].name == "Noun_9"
+    noun_turkish_headlines = turkish_headlines[1].headlines.values
+    assert noun_turkish_headlines[0].name == "Declension_5"
+
+  end
+
+
+  def test_headlines_search
+    page = Wiki::Api::Page.new name: "program"
+    headline = page.root_headline
+    # get root
+    assert headline.name == "program", "expected program name"
+
+    headlines = headline.headline "english"
+    assert !headlines.empty?, "expected a headline"
+
+    # iterate english
+    headlines.each do |headline|
+      assert headline.is_a?(Wiki::Api::PageHeadline), "expected PageHeadline object"
+      assert headline.name == "English", "expected English"
+      noun_headlines = headline.headline "noun"
+      assert !noun_headlines.empty?, "exptected a headline"
+      # iterate nouns
+      noun_headlines.each do |noun_headline|
+        assert noun_headline.is_a?(Wiki::Api::PageHeadline), "expected PageHeadline object"
+        assert noun_headline.name == "Noun", "expected Noun"
+        
+        #list_item.links.each do |link|  
+      end
+    end
+  end
+
+  def test_headlines_has_headline
+    # load page
+    page = Wiki::Api::Page.new name: "program"
+
+    # get root
+    root_headline = page.root_headline
+    assert root_headline.name == "program", "expected program name"
+    headline = root_headline.headline("english").first
+
+    # verify existence english headlines
+    assert headline.has_headline?("alternative forms")
+    assert headline.has_headline?("etymology")
+    assert headline.has_headline?("pronunciation")
+    assert headline.has_headline?("noun")
+    assert headline.has_headline?("verb")
+    assert headline.has_headline?("external links")
+    assert !headline.has_headline?("alternative12 forms")
+    assert !headline.has_headline?("etymolfaogy")
+    assert !headline.has_headline?("pronunciafaation")
+    assert !headline.has_headline?("nounia")
+    assert !headline.has_headline?("verbaf")
+    assert !headline.has_headline?("externalaf links")
+
+    # verify existence czech headlines
+    headline = root_headline.headline("czech").first
+    assert headline.has_headline?("pronunciation")
+    assert headline.has_headline?("noun")
+    assert !headline.has_headline?("pronunciation82")
+    assert !headline.has_headline?("nouan")
+
+    #TODO: add other headlines here as well!
+
+  end
+
+end

data/wiki-api.gemspec

@@ -8,8 +8,8 @@ Gem::Specification.new do |spec|
   spec.version       = Wiki::Api::VERSION
   spec.authors       = ["Dennis Blommesteijn"]
   spec.email         = ["dennis@blommesteijn.com"]
-  spec.description   = %q{MediaWiki API and Page content parser for Headlines, Blocks, Elements, ListItems, and Links.}
-  spec.summary       = %q{MediaWiki API and Page content parser for Headlines, Blocks, Elements, ListItems, and Links.}
+  spec.description   = %q{MediaWiki API and Page content parser for Headlines (nested), TextBlocks, ListItems, and Links.}
+  spec.summary       = %q{MediaWiki API and Page content parser for Headlines (nested), TextBlocks, ListItems, and Links.}
   spec.homepage      = "https://github.com/dblommesteijn/wiki-api"
   spec.license       = "MIT"
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: wiki-api
 version: !ruby/object:Gem::Version
-  version: 0.0.2
+  version: 0.1.0
 platform: ruby
 authors:
 - Dennis Blommesteijn
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2013-04-03 00:00:00.000000000 Z
+date: 2013-04-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -80,7 +80,7 @@ dependencies:
         version: 2.0.0
   prerelease: false
   type: :development
-description: MediaWiki API and Page content parser for Headlines, Blocks, Elements, ListItems, and Links.
+description: MediaWiki API and Page content parser for Headlines (nested), TextBlocks, ListItems, and Links.
 email:
 - dennis@blommesteijn.com
 executables: []
@@ -103,9 +103,9 @@ files:
 - lib/wiki/api/version.rb
 - test/test_helper.rb
 - test/unit/files/Wiktionary_Welcome,_newcomers.html
+- test/unit/files/Wiktionary_program.html
 - test/unit/wiki_connect.rb
-- test/unit/wiki_page_config.rb
-- test/unit/wiki_page_object.rb
+- test/unit/wiki_page_offline.rb
 - wiki-api.gemspec
 homepage: https://github.com/dblommesteijn/wiki-api
 licenses:
@@ -130,10 +130,10 @@ rubyforge_project:
 rubygems_version: 2.0.3
 signing_key:
 specification_version: 4
-summary: MediaWiki API and Page content parser for Headlines, Blocks, Elements, ListItems, and Links.
+summary: MediaWiki API and Page content parser for Headlines (nested), TextBlocks, ListItems, and Links.
 test_files:
 - test/test_helper.rb
 - test/unit/files/Wiktionary_Welcome,_newcomers.html
+- test/unit/files/Wiktionary_program.html
 - test/unit/wiki_connect.rb
-- test/unit/wiki_page_config.rb
-- test/unit/wiki_page_object.rb
+- test/unit/wiki_page_offline.rb

data/test/unit/wiki_page_config.rb

@@ -1,45 +0,0 @@
-# encoding: utf-8
-
-require 'rubygems'
-require 'test/unit'
-require File.expand_path(File.dirname(__FILE__) + "/../../lib/wiki/api")
-
-#
-# Testing the parsing of URI by passing a uri variable to Page:
-#   https://en.wiktionary.org/wiki/Wiktionary:Welcome,_newcomers
-#
-
-class WikiPageConfig < Test::Unit::TestCase
-
-  def setup
-    # NOTE: comment Page.config, to use the online MediaWiki instance
-    # Wiki::Api::Page.config = PAGE_CONFIG
-    # Wiki::Api::Connect.config = GLB_CONFIG
-    # @page_name = "Wiktionary:Welcome,_newcomers"
-  end
-
-  def teardown
-  end
-
-  # test simple page invocation
-  def test_page_invocation_with_uri
-    page = Wiki::Api::Page.new name: "Wiktionary:Welcome,_newcomers", uri: "http://en.wiktionary.org"
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size < 1, "expected more than one headline"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-    end
-  end
-
-  def test_wrong_page_invocation_with_uri
-    begin
-      page = Wiki::Api::Page.new name: "A_Wrong_Page_Name", uri: "http://en.wiktionary.org"
-      assert false, "expected a failiure"
-    rescue Exception => e
-      assert true, "expected a failiure"
-    end
-  end
-
-end
-

data/test/unit/wiki_page_object.rb

@@ -1,229 +0,0 @@
-# encoding: utf-8
-
-require 'rubygems'
-require 'test/unit'
-require File.expand_path(File.dirname(__FILE__) + "/../../lib/wiki/api")
-
-#
-# Testing the parsing of URI (with a predownloaded HTML file):
-#   /files/Wiktionary_Welcome,_newcomers.html (2013-03-27)
-#
-# Online equivalent:
-#   https://en.wiktionary.org/wiki/Wiktionary:Welcome,_newcomers
-#
-
-class WikiPageObject < Test::Unit::TestCase
-
-  # this global is required to resolve URIs (MediaWiki uses relative paths in their links)
-  GLB_CONFIG = { uri: "http://en.wiktionary.org" }
-
-  # use local file for test loading
-  PAGE_CONFIG = { file: File.expand_path(File.dirname(__FILE__) + "/files/Wiktionary_Welcome,_newcomers.html") }
-
-  def setup
-    # NOTE: comment Page.config, to use the online MediaWiki instance
-    Wiki::Api::Page.config = PAGE_CONFIG
-    Wiki::Api::Connect.config = GLB_CONFIG
-    @page_name = "Wiktionary:Welcome,_newcomers"
-  end
-
-  def teardown
-  end
-
-  # test simple page invocation
-  def test_page_invocation
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-    end
-  end
-
-  # test nokogiri elements per headline
-  def test_page_elements
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size > 1, "expected more than one headline"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      elements = headline.elements.flatten
-      assert !elements.empty?, "expected elements"
-      elements.each do |element|
-        assert element.is_a?(Nokogiri::XML::Element) || 
-          element.is_a?(Nokogiri::XML::Text) ||
-          element.is_a?(Nokogiri::XML::Comment), "expected nokogiri internals"
-      end
-    end
-  end
-
-  # test pageblocks for each headline
-  def test_page_blocks
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size > 1, "expected more than one headline"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      block = headline.block
-      assert block.is_a?(Wiki::Api::PageBlock), "expected block object"
-    end
-  end
-
-  # test string text from page block
-  def test_page_block_string_text
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size > 1, "expected more than one headline"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      block = headline.block
-      assert block.is_a?(Wiki::Api::PageBlock), "expected block object"
-      texts = block.to_texts
-      assert texts.is_a?(Array) && !texts.empty?, "expected array"
-      texts.each do |text|
-        assert text.is_a?(String), "expected string"
-      end
-    end
-  end
-
-  # test list items from page blocks
-  def test_page_block_list_items
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size > 1, "expected more than one headline"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      block = headline.block
-      assert block.is_a?(Wiki::Api::PageBlock), "expected block object"
-      list_items = block.list_items
-      assert list_items.is_a?(Array), "expected array"
-      list_items.each do |list_item|
-        assert list_item.is_a?(Wiki::Api::PageListItem), "expected list item object"
-      end
-    end
-  end
-
-  # test links within page blocks
-  def test_page_block_links
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size > 1, "expected more than one headline"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      block = headline.block
-      assert block.is_a?(Wiki::Api::PageBlock), "expected block object"
-      links = block.links
-      assert links.is_a?(Array), "expected array"
-      links.each do |link|
-        assert link.is_a?(Wiki::Api::PageLink), "expected link object"
-        assert link.uri.is_a?(URI), "expected uri object"
-      end
-    end
-  end
-
-  # test links within list items
-  def test_page_block_list_inner_links
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size > 1, "expected more than one headline"
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      block = headline.block
-      assert block.is_a?(Wiki::Api::PageBlock), "expected block object"
-      list_items = block.list_items
-      assert list_items.is_a?(Array), "expected array"
-      list_items.each do |list_item|
-        assert list_item.is_a?(Wiki::Api::PageListItem), "expected list item object"
-        links = list_item.links
-        links.each do |link|
-          assert link.is_a?(Wiki::Api::PageLink), "expected link object"
-          assert link.uri.is_a?(URI), "expected uri object"
-        end
-      end
-    end
-  end
-
-  # test single headline invocation
-  def test_page_invocation_single
-    page = Wiki::Api::Page.new name: @page_name
-    headlines = page.headlines
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size > 1, "expected more than one headline"
-
-    # collect headline names
-    hs = []
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      hs << headline.name
-    end
-
-    # query every headline manually
-    hs.each do |h|
-      # test headline query
-      headlines = page.headline h
-      # test for at least one (many indicates multiple headlines with the same name)
-      assert !headlines.empty?, "expected a list of headlines"
-      assert headlines.size == 1, "expected one headline"
-      headlines.each do |headline|
-        assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-      end
-    end
-  end
-
-  def test_page_headline_search_downcase
-    page = Wiki::Api::Page.new name: @page_name
-
-    headlines = page.headline "Editing_Wiktionary"
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size == 1, "expected one headline"
-
-
-    # iterate headlines
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-    end
-
-    # search downcase
-    headlines = page.headline "editing_wiktionary"
-    assert !headlines.empty?, "expected headlines"
-
-    # iterate headlines
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-    end
-
-  end
-
-  def test_page_headline_search_regular
-    page = Wiki::Api::Page.new name: @page_name
-
-    headlines = page.headline "Editing_Wiktionary"
-    assert !headlines.empty?, "expected headlines"
-    assert headlines.size == 1, "expected one headline"
-
-    # iterate headlines
-    headlines.each do |headline|
-      assert headline.is_a?(Wiki::Api::PageHeadline), "expected headline object"
-    end
-
-    # search downcase
-    headlines = page.headline "editing_wiktionary"
-    assert headlines.size == 1, "expected one headline"
-
-    # search downcase with spaces
-    headlines = page.headline "editing wiktionary"
-    assert headlines.size == 1, "expected one headline"
-
-    # search idiot case with spaces
-    headlines = page.headline "eDiTinG wiKtiOnarY"
-    assert headlines.size == 1, "expected one headline"
-
-  end
-
-end