Merge remote branch 'bcherry/master' into hit_highlighting

README.rdoc

@@ -57,6 +57,11 @@ Special care has been taken to be sure that auto-linking and extraction work
 in Tweets of all languages. This means that languages without spaces between
 words should work equally well.
 
+=== Hit Highlighting
+
+Use to provide emphasis around the "hits" returned from the Search API, built
+to work against text that has been auto-linked already.
+
 === Conformance
 
 To run the Conformance suite, you'll need to add that project as a git submodule.  From the root twitter-text-rb directory, run:

lib/autolink.rb

@@ -38,7 +38,7 @@ def auto_link(text, options = {})
         options),
       options)
     end
-
+    
     # Add <tt><a></a></tt> tags around the usernames and lists in the provided <tt>text</tt>. The
     # <tt><a></tt> tags can be controlled with the following entries in the <tt>options</tt>
     # hash:

lib/hithighlighter.rb

@@ -0,0 +1,66 @@
+
+module Twitter
+  # Module for doing "hit highlighting" on tweets that have been auto-linked already.  
+  # Useful with the results returned from the Search API.
+  module HitHighlighter
+    # Default Tag used for hit highlighting
+    DEFAULT_HIGHLIGHT_TAG = "em"
+
+    # Add <tt><em></em></tt> tags around the <tt>hits</tt> provided in the <tt>text</tt>. The
+    # <tt>hits</tt> should be an array of (start, end) index pairs, relative to the original
+    # text, before auto-linking (but the <tt>text</tt> may already be auto-linked if desired)
+    def hit_highlight(text, hits = [])
+      if hits.empty?
+        return text
+      end
+
+      chunks = text.split("<").map do |item|
+        item.blank? ? item : item.split(">")
+      end.flatten
+
+      tags = ["<" + DEFAULT_HIGHLIGHT_TAG + ">", "</" + DEFAULT_HIGHLIGHT_TAG + ">"]
+
+      result = ""
+      chunk_index, chunk = 0, chunks[0]
+      prev_chunks_len = 0
+      chunk_cursor = 0
+      start_in_chunk = false
+      for hit, index in hits.flatten.each_with_index do
+        tag = tags[index % 2]
+
+        placed = false
+        until chunk.nil? || hit < prev_chunks_len + chunk.length do
+          result << chunk[chunk_cursor..-1] 
+          if start_in_chunk && hit == prev_chunks_len + chunk.length
+            result << tag
+            placed = true
+          end
+          result << "<#{chunks[chunk_index+1]}>"
+          prev_chunks_len += chunk.length
+          chunk_cursor = 0
+          chunk_index += 2
+          chunk = chunks[chunk_index]
+          start_in_chunk = false
+        end
+
+        if !placed && !chunk.nil?
+          hit_spot = hit - prev_chunks_len
+          result << chunk[chunk_cursor...hit_spot] + tag
+          chunk_cursor = hit_spot
+          if index % 2 == 0
+            start_in_chunk = true
+          end
+        end
+      end
+
+      if !chunk.nil?
+        result << chunk[chunk_cursor..-1]
+        for index in chunk_index+1..chunks.length-1
+          result << (index.even? ? chunks[index] : "<#{chunks[index]}>")
+        end
+      end
+
+      result
+    end
+  end
+end

lib/twitter-text.rb

@@ -10,3 +10,4 @@
 require File.join(File.dirname(__FILE__), 'extractor')
 require File.join(File.dirname(__FILE__), 'unicode')
 require File.join(File.dirname(__FILE__), 'validation')
+require File.join(File.dirname(__FILE__), 'hithighlighter')

spec/autolinking_spec.rb

@@ -430,6 +430,7 @@ def original_text; "I like www.foobar.com dudes"; end
       end
 
     end
+
   end
 
 end

spec/hithighlighter_spec.rb

@@ -0,0 +1,61 @@
+require File.dirname(__FILE__) + '/spec_helper'
+
+class TestHitHighlighter
+  include Twitter::HitHighlighter
+end
+
+describe Twitter::HitHighlighter do
+  describe "highlight" do
+    before do
+      @highlighter = TestHitHighlighter.new
+    end
+
+    context "without links" do
+      before do
+        @original = "Hey! this is a test tweet"
+      end
+
+      it "should return original when no hits are provided" do
+        @highlighter.hit_highlight(@original).should == @original
+      end
+
+      it "should highlight one hit" do
+        @highlighter.hit_highlight(@original, hits = [[5, 9]]).should == "Hey! <em>this</em> is a test tweet"
+      end
+
+      it "should highlight two hits" do
+        @highlighter.hit_highlight(@original, hits = [[5, 9], [15, 19]]).should == "Hey! <em>this</em> is a <em>test</em> tweet"
+      end
+
+    end
+
+    context "with links" do
+      it "should highlight with a single link" do
+        @highlighter.hit_highlight("@<a>bcherry</a> this was a test tweet", [[9, 13]]).should == "@<a>bcherry</a> <em>this</em> was a test tweet"
+      end
+
+      it "should highlight with link at the end" do
+        @highlighter.hit_highlight("test test <a>test</a>", [[5, 9]]).should == "test <em>test</em> <a>test</a>"
+      end
+
+      it "should highlight with a link at the beginning" do
+        @highlighter.hit_highlight("<a>test</a> test test", [[5, 9]]).should == "<a>test</a> <em>test</em> test"
+      end
+
+      it "should highlight an entire link" do
+        @highlighter.hit_highlight("test <a>test</a> test", [[5, 9]]).should == "test <a><em>test</em></a> test"
+      end
+
+      it "should highlight within a link" do
+        @highlighter.hit_highlight("test <a>test</a> test", [[6, 8]]).should == "test <a>t<em>es</em>t</a> test"
+      end
+
+      it "should highlight around a link" do
+        @highlighter.hit_highlight("test <a>test</a> test", [[3, 11]]).should == "tes<em>t <a>test</a> t</em>est"
+      end
+
+    end
+
+  end
+
+end