Add min_length, max_permutations to get_best_permutations().

data_capture/templatetags/analyze_contract.py

@@ -32,7 +32,7 @@ def powerset(iterable):
     return chain.from_iterable(combinations(s, r) for r in range(len(s)+1))
 
 
-def get_best_permutations(vocab, lexemes):
+def get_best_permutations(vocab, lexemes, min_length=4, max_permutations=8):
     def compare(a, b):
         a_len = len(a)
         b_len = len(b)
@@ -45,11 +45,16 @@ def compare(a, b):
     def vocab_val(iterable):
         return sum([vocab[i] for i in iterable])
 
-    permutations = list(powerset(lexemes))
+    # Remove the first element, as it's the empty set.
+    permutations = list(powerset(lexemes))[1:]
+
+    permutations = list(filter(
+        lambda x: len(' '.join(x)) >= min_length,
+        permutations
+    ))
     permutations.sort(key=cmp_to_key(compare), reverse=True)
 
-    # Remove the last element, as it's the empty set.
-    return permutations[:-1]
+    return permutations[:max_permutations]
 
 
 def get_vocab(cursor, model=Contract, field='search_index', min_ndoc=100):

data_capture/tests/test_analyze_contract.py

@@ -3,16 +3,33 @@
 from ..templatetags.analyze_contract import get_best_permutations
 
 
-class AnalyzeContractTests(TestCase):
-    def test_get_best_permutations_works(self):
-        vocab = {
-            'junior': 1,
-            'administrative': 2,
-            'engineer': 3,
-        }
+VOCAB = {
+    'junior': 1,
+    'administrative': 2,
+    'engineer': 3,
+    'ii': 4,
+}
 
+
+class GetBestPermutationsTests(TestCase):
+    def test_min_length_works(self):
+        self.assertEqual(
+            get_best_permutations(VOCAB, ['engineer', 'ii'], min_length=3),
+            [('engineer', 'ii'),
+             ('engineer',)],
+        )
+
+    def test_max_permutations_works(self):
+        self.assertEqual(
+            get_best_permutations(VOCAB, ['junior', 'administrative',
+                                          'engineer'], max_permutations=2),
+            [('junior', 'administrative', 'engineer'),
+             ('administrative', 'engineer')]
+        )
+
+    def test_it_works(self):
         self.assertEqual(
-            get_best_permutations(vocab, ['junior', 'administrative',
+            get_best_permutations(VOCAB, ['junior', 'administrative',
                                           'engineer']),
             [('junior', 'administrative', 'engineer'),
              ('administrative', 'engineer'),