add more robust tagging [Moritz Angermann]

git-svn-id: http://svn.techno-weenie.net/projects/mephisto/trunk@2434 567b1171-46fb-0310-a4c9-b4bef9110e78

CHANGELOG

@@ -1,5 +1,9 @@
 * SVN *
 
+* 0.7.1 PRE-RELEASE *
+
+* add more robust tagging [Moritz Angermann]
+
 * Fixed duplicate body bug in default Simpla theme when article has no excerpt
 
 * Show excerpt by default on admin edit if the article has one

app/apis/meta_weblog_service.rb

@@ -52,7 +52,7 @@ def article_dto_from(article)
       :categories        => article.sections.collect { |c| c.name },
       :mt_text_more      => article.body.to_s,
       :mt_excerpt        => article.excerpt.to_s,
-      # :mt_keywords       => article.keywords.to_s,
+      :mt_keywords       => article.tag,
       # :mt_allow_comments => article.allow_comments? ? 1 : 0,
       # :mt_allow_pings    => article.allow_pings? ? 1 : 0,
       # :mt_convert_breaks => (article.text_filter.name.to_s rescue ''),
@@ -72,6 +72,9 @@ def post_it(article, user, password, struct, publish)
       # if no categories are supplied do not attempt to set any.
       article.section_ids = Section.find(:all, :conditions => ['name IN (?)', struct['categories']]).collect(&:id) if struct['categories']
       article.attributes = {:updater => @user, :body => struct['description'].to_s, :title => struct['title'].to_s, :excerpt => struct['mt_excerpt'].to_s}
+      # Keywords/Tags support
+      Tagging.set_on article, struct['mt_keywords'] if struct['mt_keywords'] # set/modify keywords _only_ if they are supplied. mt_keywords _overwrite_ not alter the ``tags''
+
       utc_date = Time.utc(struct['dateCreated'].year, struct['dateCreated'].month, struct['dateCreated'].day, struct['dateCreated'].hour, struct['dateCreated'].sec, struct['dateCreated'].min) rescue article.published_at || Time.now.utc
       article.published_at = publish == true ? utc_date : nil
       article.save!

app/models/tag.rb

@@ -12,19 +12,27 @@ def [](tag)
     #   # => ['a', 'b', 'c']
     def parse(list)
       return list if list.is_a?(Array)
-      list.split(',').collect! { |s| s.gsub(/[^\w\ ]+/, '').downcase.strip }.delete_if { |s| s.blank? }
+      # more robust does handle all kinds of different tags. (comma seperated, space seperated (in quotation marks))
+      # should handle most the common keyword formats.
+      # e.g.: b'log, emacs fun, rails, ruby => "b'log", "emacs fun", "rails", "ruby"
+      #       "b'log" "emacs fun" "rails" "ruby" => "b'log", "emacs fun", "rails", "ruby"
+      #       'b\'log' 'emacs fun' 'rails' 'ruby' => "b'log", "emacs fun", "rails", "ruby"
+      #
+      list.scan(/((?: |)['"]{0,1})['"]?(.*?)(?:[,'"]|$)(?:\1(?: |$))/).collect{ |tag| tag.last }.uniq.delete_if{ |tag| tag == "" }
+      # the old version for legacy comparison.
+      #list.split(',').collect! { |s| s.gsub(/[^\w\ ]+/, '').downcase.strip }.delete_if { |s| s.blank? }
     end
-    
+
     # Parses comma separated tag list and returns tags for them.
     #
     #   Tag.parse_to_tags('a, b, c')
     #   # => [Tag, Tag, Tag]
     def parse_to_tags(list)
       find_or_create(parse(list))
     end
-    
+
     # Returns Tags from an array of tag names
-    # 
+    #
     #   Tag.find_or_create(['a', 'b', 'c'])
     #   # => [Tag, Tag, Tag]
     def find_or_create(tag_names)
@@ -38,8 +46,8 @@ def find_or_create(tag_names)
   def ==(comparison_object)
     super || name == comparison_object.to_s
   end
-  
+
   def to_s()     name end
   alias to_param  to_s
   alias to_liquid to_s
-end
+end

test/functional/backend_controller_test.rb

@@ -151,6 +151,54 @@ def test_meta_weblog_new_post
     end
   end
 
+  def test_meta_weblog_new_post_tags
+    # test weather we can create a new post, set tags and get away with it...
+    tags = 'blog, emacs, fun, rails, ruby'
+    article = MetaWeblogStructs::Article.new(:title => 'This is a title', :description => 'This is a post', :mt_keywords => tags)
+
+    args = [ 1, 'quentin', 'test', article, 1 ]
+
+    result = invoke_layered :metaWeblog, :newPost, *args
+    # assert result
+    new_post = Article.find(result)
+    assert_equal Tag.parse_to_tags(tags).collect(&:name).sort, new_post.tags.collect(&:name).sort
+  end
+
+  def test_meta_weblog_edit_post_tags
+    # changing tags
+    tags = 'blog, emacs, fun, rails, ruby'
+    article = MetaWeblogStructs::Article.new(:title => 'This is a title', :description => 'This is a post', :mt_keywords => tags)
+    args = [ 1, 'quentin', 'test', article, 1 ]
+    post_id = invoke_layered :metaWeblog, :newPost, *args
+
+    new_post = Article.find(post_id)
+    assert_equal Tag.parse_to_tags(tags).collect(&:name).sort, new_post.tags.collect(&:name).sort
+
+    tags = 'bar, baz, foo'
+    article = MetaWeblogStructs::Article.new(:title => 'This is a title', :description => 'This is a post', :mt_keywords => tags)
+    args = [ post_id, 'quentin', 'test', article, 1]
+    edit_result = invoke_layered :metaWeblog, :editPost, *args
+    assert_equal true, edit_result
+
+    new_post = Article.find(post_id)
+    assert_equal Tag.parse_to_tags(tags).collect(&:name).sort, new_post.tags.collect(&:name).sort
+  end
+
+  def test_meta_weblog_get_post_with_tags
+    # set tags and see if we recive them!
+    tags = 'blog, emacs, fun, rails, ruby'
+    article = MetaWeblogStructs::Article.new(:title => 'This is a title', :description => 'This is a post', :mt_keywords => tags)
+
+    args = [ 1, 'quentin', 'test', article, 1 ]
+    post_id = invoke_layered :metaWeblog, :newPost, *args
+
+    args = [ post_id, 'quentin', 'test' ]
+    result = invoke_layered :metaWeblog, :getPost, *args
+
+    assert_equal tags, result['mt_keywords']
+  end
+
+
   def test_meta_weblog_new_post_min
     # This is going to test weather a post is correctly submited or not without the published_at field!
     # See http://www.xmlrpc.com/metaWeblogApi#theStruct or
@@ -195,7 +243,6 @@ def test_meta_weblog_new_post_publish
         assert_equal 'This is a title', new_post.title
         assert_equal c[:expect], new_post.status
       end
-
     end
   end
 
@@ -253,5 +300,4 @@ def new_media_object(options = {})
         'bits' => Base64.encode64(File.open(File.expand_path(RAILS_ROOT) + '/public/images/mephisto/shadow.png', 'rb') { |f| f.read })
       ))
     end
-
 end

test/unit/tag_test.rb

@@ -5,6 +5,23 @@ class TagTest < Test::Unit::TestCase
 
   def test_should_parse_comma_separated_tags
     assert_equal %w(a b c), Tag.parse('a, b, c')
+    assert_equal %w(a b\ c), Tag.parse('a, b c')
+  end
+
+  def test_should_parse_simple_tags
+    assert_equal %w(a b c), Tag.parse("'a' 'b' 'c'")
+    assert_equal %w(a b c), Tag.parse('"a" "b" "c"')
+  end
+
+  def test_should_parse_more_complicated_tags
+    # with quotation marks _in_ the string.
+    assert_equal %w(tagging it's weirdness), Tag.parse('tagging, it\'s, weirdness')
+    assert_equal %w(tagging it"s weirdness), Tag.parse('tagging, it"s, weirdness')
+    assert_equal %w(tagging it's weirdness), Tag.parse('"tagging" "it\'s" "weirdness"')
+    assert_equal %w(tagging it's weirdness), Tag.parse("'tagging' 'it's' 'weirdness'")
+    # with spaces...
+    assert_equal %w(tagging it's\ weirdness), Tag.parse("'tagging' 'it's weirdness'")
+    assert_equal %w(tagging it's\ weirdness), Tag.parse('"tagging" "it\'s weirdness"')
   end
 
   def test_should_return_tag_array
@@ -22,12 +39,12 @@ def test_should_create_tags_from_comma_separated_list
       Tag.parse_to_tags 'ruby, a, b, rails, c'
     end
   end
-  
+
   def test_tag_equality
     assert_equal tags(:ruby), 'ruby'
     assert_equal Tag.find_by_name('ruby'), tags(:ruby)
   end
-  
+
   def test_should_select_tags_by_name
     assert_equal tags(:ruby), Tag[:ruby]
   end

test/unit/tagging_test.rb

@@ -24,7 +24,7 @@
     end
     assert_equal [], assets(:gif).reload.tags
   end
-  
+
   specify "should change tags" do
     assert_difference Tagging, :count, 2 do
       assert_difference Tag, :count do
@@ -63,7 +63,7 @@
     end
     assert_equal [], contents(:another).reload.tags
   end
-  
+
   specify "should change tags" do
     assert_difference Tagging, :count, 2 do
       assert_difference Tag, :count do