Adds mentor counting to json output and adds missing config.

This adds mentor counting output to json format. In addition this change makes
max number of comments to evaluate configurable as well as the cutoff for
heavily involved mentors.

README.md

@@ -119,6 +119,9 @@ Below are the allowed configuration options:
 | `IGNORE_USERS` | False |         | A comma separated list of users to ignore when calculating metrics. (ie. `IGNORE_USERS: 'user1,user2'`). To ignore bots, append `[bot]` to the user (ie. `IGNORE_USERS: 'github-actions[bot]'`)  |
 | `ENABLE_MENTOR_COUNT` | False | False | If set to 'TRUE' count number of comments users left on discussions, issues and PRs and display number of active mentors |
 | `MIN_MENTOR_COMMENTS` | False | 10 | Minimum number of comments to count as a mentor |
+| `MAX_COMMENTS_EVAL` | False | 20 | Maximum number of comments per thread to evaluate for mentor stats |
+| `HEAVILY_INVOLVED_CUTOFF` | False | 3 | Cutoff after which a mentor's comments in one issue are no longer counted against their total score |
+
 ## Further Documentation
 
 - [Example workflows](./docs/example-workflows.md)

config.py

@@ -30,12 +30,16 @@ class EnvVars:
         hide_label_metrics (str): If set, the label metrics are hidden in the output
         enable_mentor_count (str): If set to TRUE, compute number of mentors
         min_mentor_comments (str): If set, defines the minimum number of comments for mentors
+        max_comments_eval (str): If set, defines the maximum number of comments to look at for mentor evaluation
+        heavily_involved_cutoff (str): If set, defines the cutoff after which heavily involved commentors in
+            one thread are no longer counted.
     """
     def __init__(self, search_query: str, gh_token: str, labels_to_measure: List[str], ignore_user: List[str],
                  github_server_url: str, hide_author: str, hide_time_to_first_response: str,
                  hide_time_to_close: str, hide_time_to_answer: str,
                  hide_label_metrics: str, enable_mentor_count: str,
-                 min_mentor_comments: str):
+                 min_mentor_comments: str, max_comments_eval: str,
+                 heavily_involved_cutoff: str):
         self.search_query = search_query
         self.gh_token = gh_token
         self.labels_to_measure = labels_to_measure
@@ -48,6 +52,8 @@ def __init__(self, search_query: str, gh_token: str, labels_to_measure: List[str
         self.hide_label_metrics = hide_label_metrics
         self.enable_mentor_count = enable_mentor_count
         self.min_mentor_comments = min_mentor_comments
+        self.max_comments_eval = max_comments_eval
+        self.heavily_involved_cutoff = heavily_involved_cutoff
 
 
 def get_env_vars() -> EnvVars:
@@ -86,6 +92,8 @@ def get_env_vars() -> EnvVars:
     hide_label_metrics = os.getenv("HIDE_LABEL_METRICS")
     enable_mentor_count = os.getenv("ENABLE_MENTOR_COUNT", "FALSE")
     min_mentor_comments = os.getenv("MIN_MENTOR_COMMENTS", "10")
+    max_comments_eval = os.getenv("MAX_COMMENTS_EVAL", "20")
+    heavily_involved_cutoff = os.getenv("HEAVILY_INVOLVED_CUTOFF", "3")
 
     return EnvVars(
         search_query,
@@ -99,5 +107,7 @@ def get_env_vars() -> EnvVars:
         hide_time_to_answer,
         hide_label_metrics,
         enable_mentor_count,
-        min_mentor_comments
+        min_mentor_comments,
+        max_comments_eval,
+        heavily_involved_cutoff
     )

issue_metrics.py

@@ -120,6 +120,8 @@ def get_per_issue_metrics(
     discussions: bool = False,
     labels: Union[List[str], None] = None,
     ignore_users: List[str] = None,
+    max_comments_to_eval: int = 20,
+    heavily_involved: int = 3
 ) -> tuple[List, int, int]:
     """
     Calculate the metrics for each issue/pr/discussion in a list provided.
@@ -158,8 +160,8 @@ def get_per_issue_metrics(
                 None, issue, ignore_users
             )
             issue_with_metrics.mentor_activity = count_comments_per_user(
-                None, issue, ignore_users
-                # TODO review arguments max_comments_to_eval, heavily_involved
+                None, issue, ignore_users, None, None,
+                max_comments_to_eval, heavily_involved
             )
             issue_with_metrics.time_to_answer = measure_time_to_answer(issue)
             if issue["closedAt"]:
@@ -188,9 +190,9 @@ def get_per_issue_metrics(
                 issue, None, pull_request, ready_for_review_at, ignore_users
             )
             issue_with_metrics.mentor_activity = count_comments_per_user(
-                issue, None, pull_request, ready_for_review_at, ignore_users
+                issue, None, pull_request, ready_for_review_at, ignore_users,
+                max_comments_to_eval, heavily_involved
             )
-            # TODO review arguments max_comments_to_eval, heavily_involved
             if labels:
                 issue_with_metrics.label_metrics = get_label_metrics(issue, labels)
             if issue.state == "closed":  # type: ignore
@@ -256,9 +258,11 @@ def main():
     search_query = env_vars.search_query
     token = env_vars.gh_token
     ignore_users = env_vars.ignore_users
-    enable_mentor_count = env_vars.enable_mentor_count
 
+    enable_mentor_count = env_vars.enable_mentor_count
     min_mentor_count = int(env_vars.min_mentor_comments)
+    max_comments_eval = int(env_vars.max_comments_eval)
+    heavily_involved_cutoff = int(env_vars.heavily_involved_cutoff)
 
     # Get the repository owner and name from the search query
     owner = get_owner(search_query)
@@ -298,6 +302,8 @@ def main():
         discussions="type:discussions" in search_query,
         labels=labels,
         ignore_users=ignore_users,
+        max_comments_to_eval=max_comments_eval,
+        heavily_involved=heavily_involved_cutoff,
     )
 
     stats_time_to_first_response = get_stats_time_to_first_response(
@@ -311,7 +317,7 @@ def main():
 
     num_mentor_count = 0
     if enable_mentor_count == "TRUE":
-        num_mentor_count = get_mentor_count(issues_with_metrics, min_mentor_comments)
+        num_mentor_count = get_mentor_count(issues_with_metrics, min_mentor_count)
 
     # Get stats describing the time in label for each label and store it in a dictionary
     # where the key is the label and the value is the average time

json_writer.py

@@ -31,6 +31,7 @@ def write_to_json(
     stats_time_in_labels: Union[dict[str, dict[str, timedelta]], None],
     num_issues_opened: Union[int, None],
     num_issues_closed: Union[int, None],
+    num_mentor_count: Union[int, None],
     search_query: str,
 ) -> str:
     """
@@ -43,6 +44,7 @@ def write_to_json(
         "average_time_to_answer": "1 day, 0:00:00",
         "num_items_opened": 2,
         "num_items_closed": 1,
+        "num_mentor_count": 5,
         "total_item_count": 2,
         "issues": [
             {
@@ -129,6 +131,7 @@ def write_to_json(
         "90_percentile_time_in_labels": p90_time_in_labels,
         "num_items_opened": num_issues_opened,
         "num_items_closed": num_issues_closed,
+        "num_mentor_count": num_mentor_count,
         "total_item_count": len(issues_with_metrics),
     }
 

most_active_mentors.py

@@ -42,6 +42,7 @@
 
 from classes import IssueWithMetrics
 
+
 def count_comments_per_user(
     issue: Union[github3.issues.Issue, None],  # type: ignore
     discussion: Union[dict, None] = None,
@@ -112,6 +113,23 @@ def count_comments_per_user(
                 else:
                     mentor_count[review_comment.user.login] = 1
 
+        if discussion and len(discussion["comments"]["nodes"]) > 0:
+            for comment in discussion["comments"]["nodes"]:
+                if ignore_comment(
+                    comment.user,
+                    comment.user,
+                    ignore_users,
+                    comment.submitted_at,
+                    comment.ready_for_review_at
+                ):
+                    continue
+
+                # increase the number of comments left by current user by 1
+                if comment.user.login in mentor_count:
+                    mentor_count[comment.user.login] += 1
+                else:
+                    mentor_count[comment.user.login] = 1
+
     return mentor_count
 
 
@@ -153,8 +171,8 @@ def get_mentor_count(
     """
 
     mentor_count = Counter({})
-    for issueWithMetrics in issues_with_metrics:
-        current_counter = Counter(issueWithMetrics.mentor_activity)
+    for issue_with_metrics in issues_with_metrics:
+        current_counter = Counter(issue_with_metrics.mentor_activity)
         mentor_count = mentor_count + current_counter
 
     active_mentor_count = 0

test_json_writer.py

@@ -60,6 +60,7 @@ def test_write_to_json(self):
         }
         num_issues_opened = 2
         num_issues_closed = 1
+        num_mentor_count = 5
 
         expected_output = {
             "average_time_to_first_response": "2 days, 12:00:00",
@@ -76,6 +77,7 @@ def test_write_to_json(self):
             "90_percentile_time_in_labels": {"bug": "1 day, 16:24:12"},
             "num_items_opened": 2,
             "num_items_closed": 1,
+            "num_mentor_count": 5,
             "total_item_count": 2,
             "issues": [
                 {
@@ -110,6 +112,7 @@ def test_write_to_json(self):
                 stats_time_in_labels=stats_time_in_labels,
                 num_issues_opened=num_issues_opened,
                 num_issues_closed=num_issues_closed,
+                num_mentor_count=num_mentor_count,
                 search_query="is:issue repo:owner/repo",
             ),
             json.dumps(expected_output),
@@ -148,6 +151,7 @@ def test_write_to_json_with_no_response(self):
         }
         num_issues_opened = 2
         num_issues_closed = 0
+        num_mentor_count = 5
 
         expected_output = {
             "average_time_to_first_response": "None",
@@ -164,6 +168,7 @@ def test_write_to_json_with_no_response(self):
             "90_percentile_time_in_labels": {},
             "num_items_opened": 2,
             "num_items_closed": 0,
+            "num_mentor_count": 5,
             "total_item_count": 2,
             "issues": [
                 {
@@ -198,6 +203,7 @@ def test_write_to_json_with_no_response(self):
                 stats_time_in_labels=stats_time_in_labels,
                 num_issues_opened=num_issues_opened,
                 num_issues_closed=num_issues_closed,
+                num_mentor_count=num_mentor_count,
                 search_query="is:issue repo:owner/repo",
             ),
             json.dumps(expected_output),

test_most_active_mentors.py

@@ -61,10 +61,12 @@ def test_get_mentor_count(self):
 
         # Create moc data
         issues_with_metrics = [
-            IssueWithMetrics("Issue 1", "https://github.com/user/repo/issues/1",
-            "alice", None, mentor_activity=mentor_activity),
-            IssueWithMetrics("Issue 2", "https://github.com/user/repo/issues/2",
-            "bob", None, mentor_activity=mentor_activity),
+            IssueWithMetrics(
+                "Issue 1", "https://github.com/user/repo/issues/1",
+                "alice", None, mentor_activity=mentor_activity),
+            IssueWithMetrics(
+                "Issue 2", "https://github.com/user/repo/issues/2",
+                "bob", None, mentor_activity=mentor_activity),
         ]
 
         # Call the function and check the result