gistfile1.diff

commit 95101dc9d5a1cd201db25cca25b3847710759930
Author: Jesse Newland <[email protected]>
Date:   Thu Oct 23 19:23:28 2008 -0400

    don't negate dash-preceded parts of words

diff --git a/lib/scoped_search/query_language_parser.rb b/lib/scoped_search/query_language_parser.rb
index d68e22b..2eb3973 100644
--- a/lib/scoped_search/query_language_parser.rb
+++ b/lib/scoped_search/query_language_parser.rb
@@ -87,10 +87,12 @@ module ScopedSearch
       tokens = []
       matches = query.scan(pattern).flatten.compact
       matches.each { |match|
-        tokens << :not unless match.index('-').nil?
-        # Remove any escaped quotes and any dashes - the dash usually the first character.
+        tokens << :not if match.index('-') == 0
+        # Remove any escaped quotes
+        # Remove any dashes preceded by a space or at the beginning of a token
         # Remove any additional spaces - more that one.
-        tokens << match.gsub(/[-"]/,'').gsub(/[ ]{2,}/, ' ')
+        cleaned_token = match.gsub(/"/,'').gsub(/^-| -/,'').gsub(/[ ]{2,}/, ' ')
+        tokens << cleaned_token if cleaned_token.length > 0
       }
       return tokens
     end
diff --git a/lib/scoped_search/reg_tokens.rb b/lib/scoped_search/reg_tokens.rb
index 6755920..d10e94a 100644
--- a/lib/scoped_search/reg_tokens.rb
+++ b/lib/scoped_search/reg_tokens.rb
@@ -1,7 +1,7 @@
 # Regular expression tokens to be used for parsing.
 module RegTokens
    
-  WORD = '[\w]+'
+  WORD = '[\w-]+'
   SPACE = '[ ]'
   STRING = '["][\w ]+["]'
   OR = 'OR'
diff --git a/test/query_language_test.rb b/test/query_language_test.rb
index 86b2d56..69ef346 100644
--- a/test/query_language_test.rb
+++ b/test/query_language_test.rb
@@ -76,7 +76,17 @@ class QueryLanguageTest < Test::Unit::TestCase
     assert_equal 1, parsed.length
     assert_equal 'willem', parsed[0].first
     assert_equal :not,     parsed[0].last
-  
+
+    parsed = parse_query('hallo-world')
+    assert_equal 1, parsed.length
+    assert_equal 'hallo-world', parsed.first.first
+
+    parsed = parse_query('hallo -world')
+    assert_equal 2, parsed.length
+    assert_equal 'hallo', parsed.first.first
+    assert_equal 'world', parsed.last.first
+    assert_equal :not,     parsed.last.last
+
     parsed = parse_query('123 -"456 789"')
     assert_equal 2, parsed.length
     assert_equal '123', parsed[0].first