ci: update elasticsearch index with merged PR data

A simple workflow that runs when a PR is merged and updates the
elasticsearch index with merged PR info.
The dashboard for displaying the information can be found here:

https://kibana.zephyrproject.io/

Signed-off-by: Anas Nashif <[email protected]>

.github/workflows/stats_merged_prs.yml

@@ -0,0 +1,24 @@
+name: Merged PR stats
+
+on:
+  pull_request_target:
+    branches:
+      - main
+      - v*-branch
+    types: [closed]
+jobs:
+  record_merged:
+    if: github.event.pull_request.merged == true && github.repository == 'zephyrproject-rtos/zephyr'
+    runs-on: ubuntu-22.04
+    steps:
+      - name: checkout
+        uses: actions/checkout@v3
+      - name: PR event
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          ELASTICSEARCH_KEY: ${{ secrets.ELASTICSEARCH_KEY }}
+          ELASTICSEARCH_SERVER: "https://elasticsearch.zephyrproject.io:443"
+          PR_STAT_INDEX: "pr-test-7"
+        run: |
+          pip3 install pygithub elasticsearch
+          python3 ./scripts/ci/stats/merged_prs.py --pull-request ${{ github.event.pull_request.number }} --repo ${{ github.repository }}

scripts/ci/stats/merged_prs.py

@@ -0,0 +1,157 @@
+#!/usr/bin/env python3
+# Copyright (c) 2024 Intel Corp.
+# SPDX-License-Identifier: Apache-2.0
+
+# Script that operates on a merged PR and sends data to elasticsearch for
+# further insepctions using the PR dashboard at
+# https://kibana.zephyrproject.io/
+
+import sys
+import os
+from github import Github
+import argparse
+from elasticsearch import Elasticsearch
+from elasticsearch.helpers import bulk
+from datetime import timedelta
+
+
+date_format = '%Y-%m-%d %H:%M:%S'
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        formatter_class=argparse.RawDescriptionHelpFormatter, allow_abbrev=False)
+
+    parser.add_argument('--pull-request', required=True, help='pull request number', type=int)
+    parser.add_argument('--repo', required=True, help='github repo')
+
+    return parser.parse_args()
+
+def gendata(data, index):
+    for t in data:
+        yield {
+                "_index": index,
+                "_source": t
+                }
+
+def main():
+    args = parse_args()
+    token = os.environ.get('GITHUB_TOKEN')
+    if not token:
+        sys.exit('Github token not set in environment, please set the '
+                 'GITHUB_TOKEN environment variable and retry.')
+
+    gh = Github(token)
+    json_list = []
+    gh_repo = gh.get_repo(args.repo)
+
+    if args.pull_request:
+        pr = gh_repo.get_pull(args.pull_request)
+
+        reviews = pr.get_reviews()
+        print(f'#{pr.number}: {pr.title} - {pr.comments} Comments, reviews: {reviews.totalCount}, {len(pr.assignees)} Assignees (Updated {pr.updated_at})')
+        assignee_reviews = 0
+        reviewers = set()
+        prj = {}
+        for r in reviews:
+            if r.user and r.state == 'APPROVED':
+                reviewers.add(r.user.login)
+            if pr.assignees and r.user:
+                for assignee in pr.assignees:
+                    if r.user.login == assignee.login:
+                        assignee_reviews = assignee_reviews + 1
+                        # was reviewed at least by one assignee
+                        prj['reviewed_by_assignee'] = "yes"
+
+        # This is all data we get easily though the Github API and serves as the basis
+        # for displaying some trends and metrics.
+        # Data can be extended in the future if we find more information that
+        # is useful through the API
+
+        prj['nr'] = pr.number
+        prj['url'] = pr.url
+        prj['title'] = pr.title
+        prj['comments'] = pr.comments
+        prj['reviews'] = reviews.totalCount
+        prj['assignees'] = len(pr.assignees)
+        prj['updated'] = pr.updated_at.strftime("%Y-%m-%d %H:%M:%S")
+        prj['created'] = pr.created_at.strftime("%Y-%m-%d %H:%M:%S")
+        prj['closed'] = pr.closed_at.strftime("%Y-%m-%d %H:%M:%S")
+        prj['merged_by'] = pr.merged_by.login
+        prj['submitted_by'] = pr.user.login
+        prj['changed_files'] = pr.changed_files
+        prj['additions'] = pr.additions
+        prj['deletions'] = pr.deletions
+        prj['commits'] = pr.commits
+        # The branch we are targeting. main vs release branches.
+        prj['base'] = pr.base.ref
+
+        ll = []
+        for l in pr.labels:
+            ll.append(l.name)
+        prj['labels'] = ll
+
+        # take first assignee, otherwise we have no assignees and this rule is not applicable
+        if pr.assignee:
+            prj['assignee'] = pr.assignee.login
+        else:
+            prj['assignee'] = "none"
+            prj['reviewed_by_assignee'] = "na"
+            prj['review_rule'] = "na"
+
+        # go through all assignees and check if anyone has approved and reset assignee to the one who approved
+        for assignee in pr.assignees:
+            if assignee.login in reviewers:
+                prj['assignee'] = assignee.login
+            elif assignee.login == pr.user.login:
+                prj['reviewed_by_assignee'] = "yes"
+
+
+        # list assignees for later checks
+        assignees = [a.login for a in pr.assignees]
+
+        # Deal with exceptions when assignee approval is not needed.
+        if 'Trivial' in ll or 'Hotfix' in ll:
+            prj['review_rule'] = "yes"
+        elif pr.merged_by.login in assignees:
+            prj['review_rule'] = "yes"
+        else:
+            prj['review_rule'] = "no"
+
+        prj['assignee_reviews'] = assignee_reviews
+
+        delta = pr.closed_at - pr.created_at
+        deltah = delta.total_seconds() / 3600
+        prj['hours_open'] = deltah
+
+        dates = (pr.created_at + timedelta(idx + 1) for idx in range((pr.closed_at - pr.created_at).days))
+
+        # Get number of business days per the guidelines, we need at least 2.
+        res = sum(1 for day in dates if day.weekday() < 5)
+
+        if res < 2 and not ('Trivial' in ll or 'Hotfix' in ll):
+            prj['time_rule'] = False
+        elif deltah < 4 and 'Trivial' in ll:
+            prj['time_rule'] = False
+        else:
+            prj['time_rule'] = True
+        prj['reviewers'] = list(reviewers)
+
+        json_list.append(prj)
+
+
+    # Send data over to elasticsearch.
+    es = Elasticsearch(
+            [os.environ['ELASTICSEARCH_SERVER']],
+            api_key=os.environ['ELASTICSEARCH_KEY'],
+            verify_certs=False
+            )
+
+    try:
+        index = os.environ['PR_STAT_INDEX']
+        bulk(es, gendata(json_list, index))
+    except KeyError as e:
+        print(f"Error: {e} not set.")
+        print(json_list)
+
+if __name__ == "__main__":
+    main()