Fix sort issue #68 that not places super-catch-all last and add some sorting tests.

rivescript/sorting.py

@@ -9,7 +9,7 @@
 from .regexp import RE
 from . import utils
 import re
-from operator import itemgetter, attrgetter
+from operator import attrgetter
 import sys
 
 
@@ -31,20 +31,30 @@ class TriggerObj(object):
             star: Number of wildcards (``*``), excluding alphabetical wildcards, and numeric wildcards
             pound: Number of numeric wildcards (``#``)
             under: Number of alphabetical wildcards (``_``)
-            option: Number of optional tags ("[man]" in "hey [man]")
+            option: Number of optional tags ("[man]" in "hey [man]"), assume that the template is properly formatted
         """
 
     def __init__(self, pattern, index, weight, inherit = sys.maxsize):
         self.alphabet = pattern  # Sort according to alphabet order i.e. haha < hihi
         self.index = index  # For rearrange items in the sorted array
         self.weight = - weight  # Negative weight to place i.e. -100 < 0
         self.inherit = inherit  # Low inherit takes precedence i.e. 0 < 1
-        self.wordcount = - utils.word_count(pattern)  # Length -2 < -1
-        self.len = -len(self.alphabet)  # Length -10 < -5
-        self.star =  self.alphabet.count('*')  # Number of wildcards 0 < 1
-        self.pound = self.alphabet.count('#')  # Number of numeric wildcards 0 < 1
-        self.under = self.alphabet.count('_') # Number of alphabetical wildcards 0 < 1
-        self.option = self.alphabet.count('[')  # Assume that the template is properly formatted 0 < 1
+        self.wordcount = - utils.word_count(pattern)  # Length -2 < -1. Use `utils` for counting choice of wildcards
+        self.len    = -len(self.alphabet)  # Length -10 < -5
+        self.star   = self.alphabet.count('*')  # Number of wildcards 0 < 1
+        self.pound  = self.alphabet.count('#')  # Number of numeric wildcards 0 < 1
+        self.under  = self.alphabet.count('_')  # Number of alphabetical wildcards 0 < 1
+        self.option = self.alphabet.count('[') + self.alphabet.count('(')  # Number of option 0 < 1
+
+        if self.star > 0:
+            if self.pound == 0 & self.under == 0 & self.option == 0:  # Place single star last in the rank
+                self.pound = sys.maxsize
+                self.under = sys.maxsize
+                self.option = sys.maxsize
+                if self.wordcount == 0:  # The special case for single star "*", or a grey case "* *"
+                    self.wordcount = sys.maxsize  # Make sure template "hello *" > "*"
+                    # Without any words number of stars does not matter, they all mean match any.
+                    self.star = sys.maxsize  # Make sure "*" is last in the list, "* love *" > "*"
 
 
 def sort_trigger_set(triggers, exclude_previous=True, say=None):

tests/test_sorting.py

@@ -0,0 +1,73 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+from __future__ import unicode_literals, absolute_import
+
+from rivescript.exceptions import RS_ERR_MATCH
+from .config import RiveScriptTestCase
+
+class SortingTriggersTest(RiveScriptTestCase):
+    """Topic tests."""
+
+    def test_sorting_triggers(self):
+        self.new("""
+            + * you *
+            - 1
+
+            + * are *
+            - 2
+
+            + * how are you *
+            - 3
+
+            + * hallo ween *
+            - 4
+
+            + (hi|hii)
+            - 5
+
+            + hello
+            - 6 
+
+            + hey [man]
+            - 7
+
+            + good morning
+            - 8
+
+            + *
+            - 9
+
+            + hel lo
+            - 10
+
+            + hi *
+            - 11
+
+            + hi [*]
+            - 12
+        """)
+
+        sorted_triggers =  {trig[0]:position for position, trig in enumerate(self.rs._brain.master._sorted["topics"]['random'])}
+
+        # 1) Atomic is first matched. 
+        self.assertLess(sorted_triggers['hello'],sorted_triggers['hey [man]'])
+
+        # 2) Sorted by number of words 
+        self.assertLess(sorted_triggers['hel lo'],sorted_triggers['hello'])
+        self.assertLess(sorted_triggers['* hallo ween *'],sorted_triggers['* are *'])
+
+        # 3) Sorted by length by characters 
+        self.assertLess(sorted_triggers['good morning'],sorted_triggers['hel lo'])
+
+        # 4) Sorted by alphabetical order
+        self.assertLess(sorted_triggers['* are *'],sorted_triggers['* you *'])
+
+        # 5) Sorted by number of wildcard triggers 
+        self.assertLess(sorted_triggers['hi *'],sorted_triggers['* you *'])
+
+        # 6) The `super catch all` (only single star `*`) should be least priority
+        self.assertEqual(sorted_triggers['*'],max(sorted_triggers.values()))
+        self.assertLess(sorted_triggers['hi [*]'],sorted_triggers['*']) # another check but will be covered by max check above
+
+