0
@@ -101,14 +101,15 @@ std::vector<std::string> NWordTagger::execute( const char *text, short max )cons
0
     for( short j = 1; (j <= this->nwords) && ((i+j) < words.size()); ++j ) {
0
       matched = this->tags.find( match_word );
0
       if( matched != this->tags.end() ){
0
-        //printf( "word: %d:(%s->%s)\n", i, match_word.c_str(), matched->second.c_str() );
0
         std::map<std::string, int>::iterator mloc = matched_tags.find( matched->second );
0
         if( mloc == matched_tags.end() ) {
0
           matched_tags[matched->second] = 1; // count 1
0
+          //printf( "word: %d:(%s->%s) %d, hits: 1\n", i, match_word.c_str(), matched->second.c_str(), j );
0
         }
0
         else {
0
           mloc->second++;
0
           if( max_count < mloc->second ) { max_count = mloc->second; }
0
+          //printf( "word: %d:(%s->%s) %d, hits: %d\n", i, match_word.c_str(), matched->second.c_str(), j, mloc->second );
0
         }
0
       }
0
       // stem each word and compare against our tag bank
0
@@ -129,11 +130,22 @@ std::vector<std::string> NWordTagger::execute( const char *text, short max )cons
0
       }
0
     }
0
   }
0
+ 
0
+  std::vector< std::string > reduced_tags;
0
+
0
+  // now we have a list of tags that match within the document text, check if we need to reduce the tags
0
+  if( matched_tags.size() < max ) {
0
+    // prepare the return vector
0
+    for( std::map<std::string, int>::iterator mloc = matched_tags.begin(); mloc != matched_tags.end(); ++mloc ){
0
+      reduced_tags.push_back( mloc->first );
0
+    }
0
+    return reduced_tags;
0
+  }
0
 
0
   // now that we have all the matched tags reduce to max using the tag frequency as a reduction measure
0
   std::vector< std::pair<std::string,int> > sorted_tags;
0
 
0
-  //printf( "max frequency: %d\n", max_count );
0
+  //printf( "max frequency: %d, total tagged: %d, reducing to %d\n", max_count, matched_tags.size(), max );
0
   for( std::map<std::string, int>::iterator mloc = matched_tags.begin(); mloc != matched_tags.end(); ++mloc ){
0
     //printf( "word: %s, frequency: %d\n", mloc->first.c_str(), mloc->second ); 
0
     sorted_tags.push_back(*mloc);
0
@@ -142,14 +154,13 @@ std::vector<std::string> NWordTagger::execute( const char *text, short max )cons
0
   // sort the tags in frequency order
0
   std::sort( sorted_tags.begin(), sorted_tags.end(), WordComparitor() );
0
 
0
-  std::vector< std::string > reduced_tags;
0
 
0
   std::vector< std::pair<std::string, int> >::iterator mloc;
0
   do {
0
     for(mloc = sorted_tags.begin(); mloc != sorted_tags.end(); ++mloc ) {
0
       std::pair< std::string, int > word_freq = *mloc;
0
-     // printf( "word: %s, frequency: %d\n", word_freq.first.c_str(), word_freq.second );
0
-      //printf( "word: %s, frequency: %d\n", mloc->first.c_str(), mloc->second ); 
0
+      printf( "word: %s, frequency: %d\n", word_freq.first.c_str(), word_freq.second );
0
+      printf( "word: %s, frequency: %d\n", mloc->first.c_str(), mloc->second ); 
0
       if( word_freq.second < max_count ) { 
0
         sorted_tags.erase( mloc );
0
         break;

0
@@ -16,6 +16,14 @@ class TestWordTagger < Test::Unit::TestCase
0
     puts "Duration: #{Time.now - timer} sec"
0
   end
0
 
0
+  def test_sample_bug
0
+    tags = ["foo", "bar", "baz", "squishy", "yummy"]
0
+    txt = 'This is some sample text. Foo walked into a bar. The bartender said "What can I get you?" Foo said he wanted something yummy - like a baz.'
0
+    tagger = Word::Tagger.new tags, :words => 4
0
+    result_tags = tagger.execute( txt )
0
+    assert_equal ["bar", "baz", "foo", "yummy"], result_tags
0
+  end
0
+
0
   def test_ngram_size3
0
     timer = Time.now
0
     text = "This body of text contains something like ventricular septal defect"