From faf42082950582ae3e33bf2446982b5434ecac66 Mon Sep 17 00:00:00 2001
From: Stephanos Ioannidis <root@stephanos.io>
Date: Thu, 20 Oct 2022 21:51:03 +0900
Subject: [PATCH] scripts: set_assignee: Make review requests more
 deterministic

When there are more than 15 reviewers, the `set_assignee` script was
adding the reviewers more or less on a random basis because the input
set was arbitrarily ordered (thanks to how Python "set" works), and the
attempt to add any reviewers beyond the count of 15 results in the
previoulsy added reviewer being removed.

This commit updates the `set_assignee` script such that:

1. The collaborator list (input for generating the reviewer list) is
   ordered by the area match, such that the collaborators of the most
   relevant area come first.

2. The reviewers of the relevant areas are added first, until the total
   reviewer count is 15. The script does not attempt to add more than
   15 reviewers because that can result in the previously added
   reviewers being removed from the list.

Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
---
 scripts/set_assignees.py | 33 ++++++++++++++++++++++-----------
 1 file changed, 22 insertions(+), 11 deletions(-)

diff --git a/scripts/set_assignees.py b/scripts/set_assignees.py
index 42537737d26..15564f06fd1 100755
--- a/scripts/set_assignees.py
+++ b/scripts/set_assignees.py
@@ -55,7 +55,6 @@ def process_pr(gh, maintainer_file, number):
     log(f"working on https://github.com/{args.org}/{args.repo}/pull/{pr.number} : {pr.title}")
 
     labels = set()
-    collab = set()
     area_counter = defaultdict(int)
     maint = defaultdict(int)
 
@@ -75,19 +74,24 @@ def process_pr(gh, maintainer_file, number):
             for a in areas:
                 area_counter[a.name] += 1
                 labels.update(a.labels)
-                collab.update(a.collaborators)
-                collab.update(a.maintainers)
                 for p in a.maintainers:
                     maint[p] += 1
 
     ac = dict(sorted(area_counter.items(), key=lambda item: item[1], reverse=True))
     log(f"Area matches: {ac}")
     log(f"labels: {labels}")
-    log(f"collab: {collab}")
     if len(labels) > 10:
         log(f"Too many labels to be applied")
         return
 
+    # Create a list of collaborators ordered by the area match
+    collab = list()
+    for a in ac:
+        collab += maintainer_file.areas[a].maintainers
+        collab += maintainer_file.areas[a].collaborators
+    collab = list(dict.fromkeys(collab))
+    log(f"collab: {collab}")
+
     sm = dict(sorted(maint.items(), key=lambda item: item[1], reverse=True))
 
     log(f"Submitted by: {pr.user.login}")
@@ -168,13 +172,20 @@ def process_pr(gh, maintainer_file, number):
             except UnknownObjectException as e:
                 log(f"Can't get user '{c}', account does not exist anymore? ({e})")
 
-        if reviewers:
-            try:
-                log(f"adding reviewers {reviewers}...")
-                if not args.dry_run:
-                    pr.create_review_request(reviewers=reviewers)
-            except GithubException:
-                log("cant add reviewer")
+        if len(existing_reviewers) < 15:
+            reviewer_vacancy = 15 - len(existing_reviewers)
+            reviewers = reviewers[:reviewer_vacancy]
+
+            if reviewers:
+                try:
+                    log(f"adding reviewers {reviewers}...")
+                    if not args.dry_run:
+                        pr.create_review_request(reviewers=reviewers)
+                except GithubException:
+                    log("cant add reviewer")
+        else:
+            log("not adding reviewers because the existing reviewer count is greater than or "
+                "equal to 15")
 
     ms = []
     # assignees