Adding methods in preparation to expose new function in chrome plugin

* Ability to trigger a model update (forget old and retrain clf)
* Ability to record selected links from the plugin page in the model

Prediction/Linker.py

@@ -270,6 +270,9 @@ def predict(self, prediction_object):
                 )
             return prediction_object.id_, predictions
 
+    def most_recent_sha(self):
+        return sorted(self.repository_obj.commits, key=lambda c: c.timestamp)[-1].c_hash
+
     def update_and_predict(self, event):
         if isinstance(event[1], Commit):
             if event[0] not in self.repository_obj.commits:
@@ -304,9 +307,10 @@ def update_from_github(self, since):
         links = list()
         # TODO: Record links as we add new entities
         for pr_ref in repo.get_pulls(state='all'):
-            if pr_ref.number in pr_numbers:
-                break
             pr = parse_pr_ref(pr_ref, self.repository_obj.name)
+            if pr_ref.number in pr_numbers:
+                old_pr = [pr for pr in self.repository_obj.prs if pr.number == pr_ref.number][0]
+                self.repository_obj.prs.remove(old_pr)
             self.repository_obj.prs.append(pr)
         for issue_ref in repo.get_issues(state='all', since=since):
             issue = parse_issue_ref(issue_ref)

backend/backend.py

@@ -42,12 +42,11 @@ class RequestHandler(SimpleXMLRPCRequestHandler):
             locations[os.path.basename(location)] = location
             git_locations[os.path.basename(location)] = config['git_locations'][config['locations'].index(location)]
             last_update[os.path.basename(location)] = datetime.now
-            # TODO: Extract most recent sha from git
-            most_recent_sha[os.path.basename(location)] = None
         max_age_to_keep = config['max_age_to_keep']
 
         for model in models:
             linkers[model] = Linker.load_from_disk(locations[model])
+            most_recent_sha[model] = linkers[model].most_recent_sha()
 
         server.register_introspection_functions()
 
@@ -76,6 +75,16 @@ def predict_pr(self, project, pr_id):
                 except KeyError:
                     return None
 
+            def trigger_model_updates(self):
+                for model in models:
+                    linkers[model].update_from_github(last_update[model])
+                    last_update[model] = datetime.now
+                    linkers[model].update_from_local_git(git_locations[model], most_recent_sha[model])
+                    linkers[model].forget_older_than(max_age_to_keep)
+
+            def record_link(self, project, issue_id, pr_id):
+                linkers[project].update_truth((issue_id, pr_id))
+
 
         server.register_instance(PredictionFunctions())
         print('Loading done, entering serve loop')