remove extra space from tokens

Author: gregors

lib/boilerpipe/sax/tag_actions/anchor_text.rb

@@ -29,13 +29,11 @@ def changes_tag_level?
     def append_anchor_text_start(handler)
       handler.append_space
       handler.append_token(Boilerpipe::SAX::HTMLContentHandler::ANCHOR_TEXT_START)
-      handler.append_token(' ')
     end
 
     def append_anchor_text_end(handler)
       handler.append_space
       handler.append_token(Boilerpipe::SAX::HTMLContentHandler::ANCHOR_TEXT_END)
-      handler.append_token(' ')
     end
 
     def nested_achor_tag_error_recovering(handler, name)

spec/sax/tag_actions/anchor_text_spec.rb

@@ -10,7 +10,7 @@ module Boilerpipe::SAX::TagActions
 
       it 'adds anchor text start' do
         handler = Boilerpipe::SAX::HTMLContentHandler.new
-        expect { subject.start(handler, nil, nil) }.to change { handler.token_buffer_size }.from(0).to(5)
+        expect { subject.start(handler, nil, nil) }.to change { handler.token_buffer_size }.from(0).to(4)
       end
 
       it 'returns false' do
@@ -24,15 +24,16 @@ module Boilerpipe::SAX::TagActions
           subject.start(handler, nil, nil)
           expect(handler.in_anchor_tag).to eq(1)
         end
+
         it 'doesnt append end anchor text' do
           handler = Boilerpipe::SAX::HTMLContentHandler.new
           expect { subject.start(handler, nil, nil) }.to change { handler.in_anchor_tag }.from(0).to(1)
           # puts handler.token_buffer
-          expect(handler.token_buffer_size).to eq(5)
+          expect(handler.token_buffer_size).to eq(4)
 
           subject.start(handler, nil, nil)
           # puts handler.token_buffer
-          expect(handler.token_buffer_size).to eq(5)
+          expect(handler.token_buffer_size).to eq(4)
         end
       end
     end
@@ -47,7 +48,8 @@ module Boilerpipe::SAX::TagActions
       it 'adds end anchor text' do
         handler = Boilerpipe::SAX::HTMLContentHandler.new
         handler.in_anchor_tag = 1
-        expect { subject.end_tag(handler, nil) }.to change { handler.token_buffer_size }.from(0).to(5)
+        expect { subject.end_tag(handler, nil) }.to change { handler.token_buffer_size }.from(0).to(4)
+        puts Boilerpipe::SAX::HTMLContentHandler::ANCHOR_TEXT_END.size
       end
 
       context 'if in nested anchor tag' do