Add yaml file and script to check whether PRs are too long or have to…

…o many comments (snowflakedb#1385)

.github/workflows/checkprs.yml

@@ -0,0 +1,38 @@
+name: PR Checker
+
+on:
+  schedule:
+    - cron: "0 0 * * *" # Runs 0:0 UTC on every day
+  workflow_dispatch:
+    inputs:
+      logLevel:
+        default: warning
+        description: "Log level"
+        required: true
+      tags:
+        description: "Check PRs"
+
+permissions:
+  contents: read
+  pull-requests: read
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+    - name: Checkout Code
+      uses: actions/checkout@v4
+    - name: Setup Python 3.10
+      uses: actions/setup-python@v3
+      with:
+        python-version: "3.10"
+
+    - name: Dependencies
+      run: |
+        python3 -m pip install --upgrade pip
+        pip3 install requests
+    - name: Run script
+      env:
+        TOKEN: ${{secrets.GITHUB_TOKEN}}
+      run:
+        python3 ./scripts/longprs.py

scripts/longprs.py

@@ -0,0 +1,136 @@
+import os
+from datetime import datetime, timedelta
+
+import requests
+
+use_headers = {
+    "Accept": "application/vnd.github+json",
+    "X-GitHub-Api-Version": "2022-11-28",
+    "Authorization": "Bearer " + os.environ["TOKEN"],
+}
+
+
+def get_all(url, headers=use_headers, params=None):
+    if params is None:
+        params = {}
+    all_issues = requests.get(url, headers=headers, params=params)
+    issues_data = all_issues.json()
+    while "next" in all_issues.links.keys():
+        all_issues = requests.get(all_issues.links["next"]["url"], headers=headers)
+        issues_data = issues_data + all_issues.json()
+    return issues_data
+
+
+def no_api(url_string):
+    return (
+        url_string.replace("https://api.", "https://www.")
+        .replace("/repos/", "/")
+        .replace("pulls", "pull")
+    )
+
+
+def process_diff(diff_lines):
+    current_path = ""
+    adds = 0
+    dels = 0
+    del_run = 0
+    comment_run = 0
+    diff_by_file = dict()
+    for line in diff_lines.splitlines():
+        if line.startswith("diff"):
+            if len(current_path) > 0:
+                diff_by_file[current_path] = (adds, dels)
+            current_path = line.split()[-1]
+            adds = 0
+            dels = 0
+            del_run = 0
+            comment_run = 0
+        elif line.startswith("+") and not line.startswith("+++"):
+            del_run = 0
+
+            # Ignore comment lines
+            if line[1:].lstrip().startswith("#"):
+                continue
+
+            adds = adds + 1
+            # Count runs of multi-line comments and subtract from total adds
+            # We do not want to penalize comments.
+            if comment_run > 0:
+                comment_run = comment_run + 1
+            if '"""' in line:
+                if comment_run > 0:
+                    adds = adds - comment_run
+                    comment_run = 0
+                else:
+                    comment_run = 1
+        elif line.startswith("-") and not line.startswith("---"):
+            comment_run = 0
+            del_run = del_run + 1
+            # If we're seeing a large block of deletes, discount it
+            # We do not want to penalize diffs of large blocks of code.
+            if del_run < 30:
+                dels = dels + 1
+        else:
+            del_run = 0
+            comment_run = 0
+    if adds > 0 or dels > 0:
+        diff_by_file[current_path] = (adds, dels)
+
+    return diff_by_file
+
+
+stale = datetime.today() - timedelta(days=10)
+comment_limit = 50
+diff_limit = 450
+prs = get_all("https://github.com/snowflakedb/snowpark-python/pulls")
+
+# Flag PRS that are older than stale, or have more than comment_limit comments
+# or have a diff more than diff_limit lines
+# Find all comments, and all comments more recent than 6 days.
+summaries = []
+diff_fetch_header = use_headers.copy()
+diff_fetch_header["Accept"] = "application/vnd.github.v3.diff"
+
+for pr in prs:
+    url = pr["url"]
+    pr_detail = get_all(url)
+    data = requests.get(url, headers=diff_fetch_header, params={})
+    diff_by_file = process_diff(data.text)
+    num_delta = 0
+    for file_name, diff in diff_by_file.items():
+        if "test" not in file_name:
+            num_delta += diff[0] + diff[1]
+
+    num_comments = pr_detail["review_comments"]
+    created_at = datetime.strptime(pr["created_at"], "%Y-%m-%dT%H:%M:%SZ")
+    summaries.append(
+        (
+            no_api(url),
+            num_delta,
+            num_comments,
+            created_at,
+        )
+    )
+
+
+report = ""
+for summary in summaries:
+    url = summary[0]
+    num_delta = summary[1]
+    num_comments = summary[2]
+    created_at = summary[3]
+    if num_comments > comment_limit:
+        report += (
+            f"{url} has a large number ({num_comments}) of comments. Consider an offline discussion to "
+            f"resolve quickly, or additional design reviews.\n"
+        )
+    elif num_delta > diff_limit:
+        report += (
+            f"{url} has a large diff ({num_delta} lines). Consider breaking it up\n"
+        )
+    elif created_at < stale:
+        report += f"{url} has been open since {created_at}. Should this be closed or revised?\n"
+
+if len(report) > 0:
+    with open(os.environ["GITHUB_STEP_SUMMARY"], "a") as fh:
+        print(report, file=fh)  # noqa: T201