Merge pull request #461 from drocco007/master

Use lexical order to break ties when fuzzy matching

pgcli/pgcompleter.py

@@ -212,7 +212,7 @@ class PGCompleter(Completer):
         else:
             fuzzy = False
             priority_func = self.prioritizer.keyword_count
-            
+
         # Construct a `_match` function for either fuzzy or non-fuzzy matching
         # The match function returns a 2-tuple used for sorting the matches,
         # or None if the item doesn't match
@@ -236,6 +236,14 @@ class PGCompleter(Completer):
                     # fuzzy matches
                     return -float('Infinity'), -match_point
 
+        # Lexical order of items in the collection, used for tiebreaking items
+        # with the same match group length and start position. In Python,
+        # 'user' < 'users', i.e. the "lower" value comes first. Since we use
+        # *higher* priority to mean "more important," we need to flip Python's
+        # usual position ranking, hence -position.
+        lexical_order = dict([(name, -position) for position, name in
+                              enumerate(sorted(collection))])
+
         if meta_collection:
             # Each possible completion in the collection has a corresponding
             # meta-display string
@@ -253,9 +261,11 @@ class PGCompleter(Completer):
                     # Truncate meta-text to 50 characters, if necessary
                     meta = meta[:47] + u'...'
 
+                priority = sort_key, priority_func(item), lexical_order[item]
+
                 matches.append(Match(
                     completion=Completion(item, -text_len, display_meta=meta),
-                    priority=(sort_key, priority_func(item))))
+                    priority=priority))
 
         return matches
 

tests/test_fuzzy_completion.py

@@ -49,3 +49,28 @@ def test_ranking_based_on_shortest_match(completer):
     matches = completer.find_matches(text, collection)
 
     assert matches[1].priority > matches[0].priority
+
+
+@pytest.mark.parametrize('collection', [
+    ['user_action', 'user'],
+    ['user_group', 'user'],
+    ['user_group', 'user_action'],
+])
+def test_should_break_ties_using_lexical_order(completer, collection):
+    """Fuzzy result rank should use lexical order to break ties.
+
+    When fuzzy matching, if multiple matches have the same match length and
+    start position, present them in lexical (rather than arbitrary) order. For
+    example, if we have tables 'user', 'user_action', and 'user_group', a
+    search for the text 'user' should present these tables in this order.
+
+    The input collections to this test are out of order; each run checks that
+    the search text 'user' results in the input tables being reordered
+    lexically.
+
+    """
+
+    text = 'user'
+    matches = completer.find_matches(text, collection)
+
+    assert matches[1].priority > matches[0].priority