From 6de1476cc6e2b384e125163cd93d236a935745cd Mon Sep 17 00:00:00 2001
From: David Lawrence <dkl@mozilla.com>
Date: Mon, 16 Mar 2026 15:15:47 -0400
Subject: [PATCH 1/2] Bug 2023624 - Some older repos have comments that have
 empty comment body and user object is null. The ETL script needs to handle
 these better instead of crashing

---
 main.py | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)

diff --git a/main.py b/main.py
index 884655f..3c2af83 100755
--- a/main.py
+++ b/main.py
@@ -301,8 +301,15 @@ def extract_reviewers(
 
     reviewers = github_get(session, reviewers_url).json()
 
-    logger.info(f"Extracted {len(reviewers)} reviewers for PR #{pr_number}")
-    return reviewers
+    filtered = [r for r in reviewers if r.get("user") is not None]
+    skipped = len(reviewers) - len(filtered)
+    if skipped:
+        logger.info(
+            f"Skipped {skipped} reviewer(s) with null user for PR #{pr_number}"
+        )
+
+    logger.info(f"Extracted {len(filtered)} reviewers for PR #{pr_number}")
+    return filtered
 
 
 def extract_comments(
@@ -329,8 +336,16 @@ def extract_comments(
     logger.info(f"Comments URL: {comments_url}")
 
     comments = github_get(session, comments_url).json()
-    logger.info(f"Extracted {len(comments)} comments for PR #{pr_number}")
-    return comments
+
+    filtered = [c for c in comments if c.get("user") is not None and c.get("body")]
+    skipped = len(comments) - len(filtered)
+    if skipped:
+        logger.info(
+            f"Skipped {skipped} comment(s) with null user or empty body for PR #{pr_number}"
+        )
+
+    logger.info(f"Extracted {len(filtered)} comments for PR #{pr_number}")
+    return filtered
 
 
 def sleep_for_rate_limit(resp: requests.Response) -> None:
@@ -459,7 +474,7 @@ def transform_data(raw_data: list[dict], repo: str) -> dict:
                 "target_repository": repo,
                 "date_reviewed": review.get("submitted_at"),
                 "reviewer_email": None,  # TODO Placeholder for reviewer email extraction logic
-                "reviewer_username": review.get("user", {}).get("login", "None"),
+                "reviewer_username": (review.get("user") or {}).get("login"),
                 "status": review.get("state"),
             }
             transformed_data["reviewers"].append(transformed_reviewer)

From 93e248666dabb4c019e23b6f06abaeef9d06e191 Mon Sep 17 00:00:00 2001
From: David Lawrence <dkl@mozilla.com>
Date: Tue, 17 Mar 2026 10:40:08 -0400
Subject: [PATCH 2/2] Black formatted

---
 main.py | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/main.py b/main.py
index 3c2af83..449e4ad 100755
--- a/main.py
+++ b/main.py
@@ -304,9 +304,7 @@ def extract_reviewers(
     filtered = [r for r in reviewers if r.get("user") is not None]
     skipped = len(reviewers) - len(filtered)
     if skipped:
-        logger.info(
-            f"Skipped {skipped} reviewer(s) with null user for PR #{pr_number}"
-        )
+        logger.info(f"Skipped {skipped} reviewer(s) with null user for PR #{pr_number}")
 
     logger.info(f"Extracted {len(filtered)} reviewers for PR #{pr_number}")
     return filtered