Add a script that generates a changelog from recent PRs and their labels

CHANGELOG.md

@@ -147,7 +147,7 @@ Meanwhile, we did a bunch of improvements to our manual. If you had trouble runn
   * Switch pre-release action to ncipollo/release-action [#1466](https://github.com/rerun-io/rerun/pull/1466)
 * Disallow some methods and types via Clippy[#1411](https://github.com/rerun-io/rerun/pull/1411)
 
-#### Other not user facing refactors
+#### Other non-user-facing refactors
 * Fix: don't create a dummy LogDb when opening the Rerun Menu [#1440](https://github.com/rerun-io/rerun/pull/1440)
 * `re_renderer`
   * `Draw Phases` in preparation of executing `Renderer` several times on different targets [#1419](https://github.com/rerun-io/rerun/pull/1419)

RELEASES.md

@@ -53,9 +53,11 @@ Copy this checklist to the the PR description, go through it from top to bottom,
 * [ ] `./scripts/publish_crates.sh --dry-run`
 * [ ] Bump version number in root `Cargo.toml`.
 * [ ] Update `CHANGELOG.md` with the new version number and the summary and the gif
-    * Go through https://github.com/rerun-io/rerun/compare/latest...HEAD and manually add any important PR descriptions to the `CHANGELOG.md`, with a link to the PR (which should have a deeper explanation).
-      * You can use git log to quickly generate a list of commit headlines, use `git fetch --tags --force && git log --pretty=format:%s latest..HEAD` (fetch with `--force` is necessary to update the `latest` tag)
-    * [ ] Make sure to it includes instructions for handling any breaking changes
+    * [ ] Run `pip install GitPython && scripts/generate_changelog.py`
+    * [ ] Edit PR descriptions/labels to improve the generated changelog
+    * [ ] Copy-paste the results into `CHANGELOG.md`.
+    * [ ] Editorialize the changelog if necessary
+    * [ ] Make sure the changelog  includes instructions for handling any breaking changes
 * [ ] Get the PR reviewed
 * [ ] Check that CI is green
 * [ ] Publish the crates (see below)

scripts/generate_changelog.py

@@ -0,0 +1,172 @@
+#!/usr/bin/env python3
+
+"""
+Summarizes recent PRs based on their GitHub labels.
+
+The result can be copy-pasted into CHANGELOG.md, though it often needs some manual editing too.
+"""
+
+import re
+import sys
+from typing import Any, List, Optional, Tuple
+
+import requests
+from git import Repo  # pip install GitPython
+from tqdm import tqdm
+
+
+def get_github_token() -> str:
+    import os
+
+    token = os.environ.get("GH_ACCESS_TOKEN", "")
+    if token != "":
+        return token
+
+    home_dir = os.path.expanduser("~")
+    token_file = os.path.join(home_dir, ".githubtoken")
+
+    try:
+        with open(token_file, "r") as f:
+            token = f.read().strip()
+        return token
+    except Exception:
+        pass
+
+    print("ERROR: expected a GitHub token in the environment variable GH_ACCESS_TOKEN or in ~/.githubtoken")
+    sys.exit(1)
+
+
+OWNER = "rerun-io"
+REPO = "rerun"
+COMMIT_RANGE = "latest..HEAD"
+INCLUDE_LABELS = False  # It adds quite a bit of visual noise
+
+
+def pr_title_labels(pr_number: int) -> Tuple[Optional[str], List[str]]:
+    url = f"https://api.github.com/repos/{OWNER}/{REPO}/pulls/{pr_number}"
+    gh_access_token = get_github_token()
+    headers = {"Authorization": f"Token {gh_access_token}"}
+    response = requests.get(url, headers=headers)
+    json = response.json()
+
+    # Check if the request was successful (status code 200)
+    if response.status_code == 200:
+        labels = [label["name"] for label in json["labels"]]
+        return (json["title"], labels)
+    else:
+        print(f"ERROR: {response.status_code} - {json['message']}")
+        return (None, [])
+
+
+def commit_title_pr_number(commit: Any) -> Tuple[str, Optional[int]]:
+    match = re.match(r"(.*) \(#(\d+)\)", commit.summary)
+    if match:
+        return (str(match.group(1)), int(match.group(2)))
+    else:
+        return (commit.summary, None)
+
+
+def print_section(title: str, items: List[str]) -> None:
+    if 0 < len(items):
+        print(f"#### {title}")
+        for line in items:
+            print(f"- {line}")
+    print()
+
+
+repo = Repo(".")
+commits = list(repo.iter_commits(COMMIT_RANGE))
+commits.reverse()  # Most recent last
+
+# Sections:
+analytics = []
+enhancement = []
+bugs = []
+dev_experience = []
+docs = []
+examples = []
+misc = []
+performance = []
+python = []
+renderer = []
+rfc = []
+rust = []
+ui = []
+viewer = []
+web = []
+
+for commit in tqdm(commits, desc="Processing commits"):
+    (title, pr_number) = commit_title_pr_number(commit)
+    if pr_number is None:
+        # Someone committed straight to main:
+        summary = f"{title} [{commit.hexsha}](https://github.com/{OWNER}/{REPO}/commit/{commit.hexsha})"
+        misc.append(summary)
+    else:
+        (pr_title, labels) = pr_title_labels(pr_number)
+        title = pr_title or title  # We prefer the PR title if available
+        summary = f"{title} [#{pr_number}](https://github.com/{OWNER}/{REPO}/pull/{pr_number})"
+
+        if INCLUDE_LABELS and 0 < len(labels):
+            summary += f" ({', '.join(labels)})"
+
+        added = False
+
+        if labels == ["⛴ release"]:
+            # Ignore release PRs
+            continue
+
+        # Some PRs can show up underm multiple sections:
+        if "🐍 python API" in labels:
+            python.append(summary)
+            added = True
+        if "🦀 rust SDK" in labels:
+            rust.append(summary)
+            added = True
+        if "📊 analytics" in labels:
+            analytics.append(summary)
+            added = True
+
+        if not added:
+            # Put the remaining PRs under just one section:
+            if "🪳 bug" in labels or "💣 crash" in labels:
+                bugs.append(summary)
+            elif "📉 performance" in labels:
+                performance.append(summary)
+            elif "examples" in labels:
+                examples.append(summary)
+            elif "📖 documentation" in labels:
+                docs.append(summary)
+            elif "ui" in labels:
+                ui.append(summary)
+            elif "📺 re_viewer" in labels:
+                viewer.append(summary)
+            elif "🔺 re_renderer" in labels:
+                renderer.append(summary)
+            elif "🕸️ web" in labels:
+                web.append(summary)
+            elif "enhancement" in labels:
+                enhancement.append(summary)
+            elif "🧑‍💻 dev experience" in labels:
+                dev_experience.append(summary)
+            elif "💬 discussion" in labels:
+                rfc.append(summary)
+            elif not added:
+                misc.append(summary)
+
+print()
+# Most interesting first:
+print_section("🐍 Python SDK", python)
+print_section("🦀 Rust SDK", rust)
+print_section("🪳 Bug Fixes", bugs)
+print_section("🚀 Performance Improvements", performance)
+print_section("🧑‍🏫 Examples", examples)
+print_section("📚 Docs", docs)
+print_section("🖼 UI Improvements", ui)
+print_section("🤷‍♂️ Other Viewer Improvements", viewer)
+print_section("🕸️ Web", web)
+print_section("🎨 Renderer Improvements", renderer)
+print_section("✨ Other Enhancement", enhancement)
+print_section("📈 Analytics", analytics)
+print_section("🗣 Merged RFCs", rfc)
+print_section("🧑‍💻 Dev-experience", dev_experience)
+print_section("🤷‍♂️ Other", misc)