num_wrapped_lines needs to default to 1

Author: gregors

lib/boilerpipe/document/text_block.rb

@@ -25,7 +25,7 @@ def initialize(text, num_words = 0, num_words_in_anchor_text = 0, num_words_in_w
       end
 
       def self.empty_start
-        new('', 0, 0, 0, 0, -1)
+        new('', 0, 0, 0, 1, -1)
       end
 
       def set_tag_level(level)
@@ -98,7 +98,6 @@ def clone
       def init_densities
         if @num_words_in_wrapped_lines == 0
           @num_words_in_wrapped_lines = @num_words
-          @num_wrapped_lines = 1
         end
         @text_density = @num_words_in_wrapped_lines / @num_wrapped_lines.to_f
         @link_density = @num_words == 0 ? 0.0 : @num_words_in_anchor_text / @num_words.to_f

spec/document/text_block_spec.rb

@@ -99,6 +99,16 @@ module Boilerpipe
         expect(block.text_density).to eq 4.0
         expect(block.link_density).to eq 0.5
       end
+
+      it 'resets wrapped lines' do
+        block = Document::TextBlock.new('one', 10)
+        another_block = Document::TextBlock.new('two', 10)
+
+        block.merge_next(another_block)
+
+        expect(block.num_words_in_wrapped_lines).to eq 20
+        expect(block.num_wrapped_lines).to eq 2
+      end
     end
 
     describe '#add_label' do

spec/filters/simple_block_fusion_processor_spec.rb

@@ -6,10 +6,10 @@ module Boilerpipe::Filters
     let!(:doc) { Boilerpipe::Document::TextDocument.new('', text_blocks) }
 
     context 'where blocks have same text density' do
-      let(:text_block1) { Boilerpipe::Document::TextBlock.new('one',   0, 0, 0, 0, 0) }
-      let(:text_block2) { Boilerpipe::Document::TextBlock.new('two',   0, 0, 0, 0, 1) }
-      let(:text_block3) { Boilerpipe::Document::TextBlock.new('three', 0, 0, 0, 0, 2) }
-      let(:text_block4) { Boilerpipe::Document::TextBlock.new('four',  0, 0, 0, 0, 3) }
+      let(:text_block1) { Boilerpipe::Document::TextBlock.new('one',   0, 0, 0, 1, 0) }
+      let(:text_block2) { Boilerpipe::Document::TextBlock.new('two',   0, 0, 0, 1, 1) }
+      let(:text_block3) { Boilerpipe::Document::TextBlock.new('three', 0, 0, 0, 1, 2) }
+      let(:text_block4) { Boilerpipe::Document::TextBlock.new('four',  0, 0, 0, 1, 3) }
       it 'the blocks are merged' do
         expect(doc.text_blocks.size).to eq 4
         SimpleBlockFusionProcessor.process(doc)