reorganize keyword creation for legacy tokenizer

- only save partial words without internal spaces
- consider comma and semicolon a separator of full words
- consider parts before an opening bracket a full word
  (but not the part after the bracket)

Fixes #244.

test/bdd/db/import/search_name.feature

@@ -2,6 +2,29 @@
 Feature: Creation of search terms
     Tests that search_name table is filled correctly
 
+    Scenario Outline: Comma- and semicolon separated names appear as full names
+        Given the places
+         | osm | class   | type | name+alt_name |
+         | N1  | place   | city | New York<sep>Big Apple |
+        When importing
+        Then search_name contains
+         | object | name_vector |
+         | N1     | #New York, #Big Apple |
+
+    Examples:
+         | sep |
+         | ,   |
+         | ;   |
+
+    Scenario Outline: Name parts before brackets appear as full names
+        Given the places
+         | osm | class   | type | name+name |
+         | N1  | place   | city | Halle (Saale) |
+        When importing
+        Then search_name contains
+         | object | name_vector |
+         | N1     | #Halle Saale, #Halle |
+
     Scenario: Unnamed POIs have no search entry
         Given the scene roads-with-pois
         And the places
@@ -49,7 +72,7 @@ Feature: Creation of search terms
         When importing
         Then search_name contains
          | object | nameaddress_vector |
-         | N1     | Rose Street, Little, Big, Town |
+         | N1     | #Rose Street, rose, Little, Big, Town |
         When searching for "23 Rose Street, Little Big Town"
         Then results contain
          | osm_type | osm_id | name |

test/python/test_tokenizer_legacy_icu.py

@@ -223,11 +223,32 @@ def test_update_special_phrase_modify(analyzer, word_table):
 
 
 def test_process_place_names(analyzer, getorcreate_term_id):
-
     with analyzer() as anl:
         info = anl.process_place({'name' : {'name' : 'Soft bAr', 'ref': '34'}})
 
-    assert info['names'] == '{1,2,3,4,5,6}'
+    assert info['names'] == '{1,2,3,4,5}'
+
+
+@pytest.mark.parametrize('sep', [',' , ';'])
+def test_full_names_with_separator(analyzer, getorcreate_term_id, sep):
+    with analyzer() as anl:
+        full_names =
+            anl._compute_full_names({'name' : sep.join(('New York', 'Big Apple'))})
+
+        expect = set((anl.make_standard_word(w) for w in ('New York', 'Big Apple')))
+
+    assert full_names == expect
+
+
+def test_process_place_names_with_bracket(analyzer, getorcreate_term_id):
+    with analyzer() as anl:
+        info = anl.process_place({'name' :
+                                   {'name' : 'Houseboat (left)'}})
+
+        expect = set((anl.make_standard_word(w) for w in
+                       (' houseboat', ' houseboat left', 'houseboat', 'left')))
+
+    assert eval(info['names']) == expect
 
 
 @pytest.mark.parametrize('pcode', ['12345', 'AB 123', '34-345'])