elections/tools: Handle unknown logins better

We have some users (maybe as they have changed email, or
don't have a public email on github), that we can't find gh logins
for, so we've just used the email, but this could result in us double
counting emails in the electorate list, so instead provide a mapping
of these problematic emails to the correct github username and
add warning messages to let users know to keep these up-to-date.

Signed-off-by: stevenhorsman <[email protected]>

Author: stevenhorsman

elections/tools/generate_electorate.py

@@ -108,6 +108,25 @@ def find_authors_by_project(start_time, end_time):
         'tests',
     ]
 
+    # Let's build a list of the users we can't get the logins for so
+    # we can prompt the runner to update the following map
+    unknown_logins = {}
+
+    # Some committers have emails that don't map to their
+    # userid properly, so maintain a map of these so the data is consistent
+    email_id_map = {
+        "[email protected]":"lifupan",
+        "[email protected]": "Ankita13-code",
+        "[email protected]": "ChengyuZhu6",
+        "[email protected]": "lima-emanuel",
+        "[email protected]": "l8huang",
+        "[email protected]": "Bickor",
+        "[email protected]": "huoqifeng",
+        "[email protected]": "niteeshkd",
+        "[email protected]": "gpyrros",
+        "[email protected]": "cmaf",
+        "[email protected]": "amshinde",
+        }
 
     author_cache = {}
     for repo in org.repositories():
@@ -129,12 +148,17 @@ def find_authors_by_project(start_time, end_time):
             if commit.author is None:
                 if commit.commit.author is None:
                     print('Skipping %s in %s as it has no author. Did this merge via GitHub?' %
-                            (commit, repo))
+                        (commit, repo))
                     continue
 
                 author_id = commit.commit.author.get('email')
-                print('%s in %s as has no author. Using email (%s) as the author id' %
-                        (commit, repo, author_id))
+                if author_id in email_id_map:
+                    author_id = email_id_map[author_id]
+                else:
+                    if not author_id in unknown_logins:
+                        unknown_logins[author_id] = commit.html_url
+                    print('%s in %s as has no author. Using email (%s) as the author id' %
+                (commit, repo, author_id))
             else:
                 author_id = commit.author.login
 
@@ -171,6 +195,12 @@ def find_authors_by_project(start_time, end_time):
                     author.name = match.group('name')
             authors.add(author)
         projects.append({str(repo): authors})
+
+    if len(unknown_logins) > 0:
+        print("Warning: failed to match some emails. Please follow the commit link and add the email -> id " \
+        "map to `email_id_map` in generate_electorate.py:")
+        for email, commit_html in unknown_logins.items():
+            print("Email", email, "who committed", commit_html)
     return projects
 
 def main():