Merge pull request #131 from kkamkou/issue-130

status_detect is not triggered in some cases, closes #130

gitmostwanted/tasks/repo_status.py

@@ -3,6 +3,7 @@
 from sqlalchemy.sql import expression
 from statistics import variance, mean
 from datetime import datetime, timedelta
+from types import GeneratorType
 
 
 @celery.task()
@@ -13,19 +14,14 @@ def status_detect(num_days, num_segments):
             .filter(RepoStars.repo_id == repo.id)\
             .order_by(expression.asc(RepoStars.day))\
             .all()
-        if not result:
-            continue
 
-        means = []
-        chunks = result_split(list(result_normalize(result, num_days)), num_segments)
-        for chunk in chunks:
-            means.append(1 if variance(chunk) >= 1000 else mean(chunk))
+        val = 0 if not result else result_mean(
+            result_split(list(result_normalize(result, num_days)), num_segments)
+        )
 
-        mean_val = mean(means)
+        repo.status = 'hopeless' if val < 1 else 'promising'
 
-        repo.status = 'hopeless' if mean_val < 1 else 'promising'
-
-        db.session.add(RepoMean(repo=repo, value=mean_val))
+        db.session.add(RepoMean(repo=repo, value=val))
         db.session.commit()
 
 
@@ -43,6 +39,10 @@ def status_refresh(num_days):
         db.session.commit()
 
 
+def result_mean(chunks: GeneratorType):
+    return mean([1 if variance(chunk) >= 1000 else mean(chunk) for chunk in chunks])
+
+
 def result_normalize(lst: list, num_days: int):
     fst = lst[0][0]
     lst = dict(lst)

tests/unit/tasks/test_repo_status.py

@@ -1,6 +1,7 @@
-from gitmostwanted.tasks.repo_status import result_normalize, result_split
+from gitmostwanted.tasks.repo_status import result_normalize, result_split, result_mean
 from types import GeneratorType
 from unittest import TestCase
+from itertools import chain
 
 
 class TasksRepoStatusTestCase(TestCase):
@@ -20,3 +21,14 @@ def test_split_result(self):
         lst = list(result)
         self.assertEquals(len(lst), 4)
         self.assertEquals(len(lst[0]), 7)
+
+    def test_calculate_mean(self):
+        def gn_normal():
+            yield [10, 20]
+
+        def gn_mixed():
+            yield [10, 99999, 10]
+
+        self.assertEquals(result_mean(gn_normal()), 15)
+        self.assertEquals(result_mean(gn_mixed()), 1)
+        self.assertEquals(result_mean(chain(gn_mixed(), gn_normal())), 8)