`generate_changelog.py`: divide into "fixed/added/changed" sections (#4712)

Make each release a little bit easier to do

scripts/generate_changelog.py

@@ -12,6 +12,8 @@ import multiprocessing
 import os
 import re
 import sys
+
+from collections import defaultdict
 from datetime import date
 from dataclasses import dataclass
 from typing import Any, List, Optional
@@ -23,15 +25,13 @@ from tqdm import tqdm
 OWNER = "emilk"
 REPO = "egui"
 INCLUDE_LABELS = False  # It adds quite a bit of visual noise
-OFFICIAL_DEVS = [
-    "emilk",
-]
 
 
 @dataclass
 class PrInfo:
+    pr_number: int
     gh_user_name: str
-    pr_title: str
+    title: str
     labels: List[str]
 
 
@@ -85,7 +85,7 @@ def fetch_pr_info(pr_number: int) -> Optional[PrInfo]:
     if response.status_code == 200:
         labels = [label["name"] for label in json["labels"]]
         gh_user_name = json["user"]["login"]
-        return PrInfo(gh_user_name=gh_user_name, pr_title=json["title"], labels=labels)
+        return PrInfo(pr_number=pr_number, gh_user_name=gh_user_name, title=json["title"], labels=labels)
     else:
         print(f"ERROR {url}: {response.status_code} - {json['message']}")
         return None
@@ -101,18 +101,83 @@ def get_commit_info(commit: Any) -> CommitInfo:
         return CommitInfo(hexsha=commit.hexsha, title=commit.summary, pr_number=None)
 
 
+def pr_summary(pr: PrInfo, crate_name: Optional[str] = None) -> str:
+    title = pr.title
+
+    if crate_name is not None:
+        # Remove crate name prefix (common in PR titles):
+        title = remove_prefix(title, f"[{crate_name}] ")
+        title = remove_prefix(title, f"{crate_name}: ")
+        title = remove_prefix(title, f"`{crate_name}`: ")
+
+    # Upper-case first letter:
+    title = title[0].upper() + title[1:]
+
+    # Remove trailing periods:
+    title = title.rstrip(".")
+
+    summary = f"{title} [#{pr.pr_number}](https://github.com/{OWNER}/{REPO}/pull/{pr.pr_number})"
+
+    if INCLUDE_LABELS and 0 < len(pr.labels):
+        summary += f" ({', '.join(pr.labels)})"
+
+    summary += f" by [@{pr.gh_user_name}](https://github.com/{pr.gh_user_name})"
+
+    return summary
+
+
+def pr_info_section(prs: List[PrInfo], *, crate_name: str, heading: Optional[str] = None) -> str:
+    result = ""
+    if 0 < len(prs):
+        if heading is not None:
+            result += f"### {heading}\n"
+        for pr in prs:
+            result += f"* {pr_summary(pr, crate_name)}\n"
+        result += "\n"
+    return result
+
+
+def changelog_from_prs(pr_infos: List[PrInfo], crate_name: str) -> str:
+    if len(pr_infos) == 0:
+        return "Nothing new"
+
+    if len(pr_infos) <= 5:
+        # For small crates, or small releases
+        return pr_info_section(pr_infos, crate_name=crate_name)
+
+
+    fixed = []
+    added = []
+    rest = []
+    for pr in pr_infos:
+        summary = pr_summary(pr, crate_name)
+        if "bug" in pr.labels:
+            fixed.append(pr)
+        elif summary.startswith("Add") or "feature" in pr.labels:
+            added.append(pr)
+        else:
+            rest.append(pr)
+
+    result = ""
+
+    result += pr_info_section(added, crate_name=crate_name, heading="⭐ Added")
+    result += pr_info_section(rest, crate_name=crate_name, heading="🔧 Changed")
+    result += pr_info_section(fixed, crate_name=crate_name, heading="🐛 Fixed")
+
+    return result.rstrip()
+
+
 def remove_prefix(text, prefix):
     if text.startswith(prefix):
         return text[len(prefix) :]
     return text  # or whatever
 
 
-def print_section(crate: str, items: List[str]) -> None:
+def print_section(heading: str, content: str) -> None:
-    if 0 < len(items):
+    if content != "":
-        print(f"#### {crate}")
+        print(f"## {heading}")
-        for line in items:
+        print(content)
-            print(f"* {line}")
+        print()
-    print()
 
 
 def changelog_filepath(crate: str) -> str:
@@ -124,10 +189,9 @@ def changelog_filepath(crate: str) -> str:
     return os.path.normpath(file_path)
 
 
-def add_to_changelog_file(crate: str, items: List[str], version: str) -> None:
+def add_to_changelog_file(crate: str, content: str, version: str) -> None:
     insert_text = f"\n## {version} - {date.today()}\n"
-    for item in items:
+    insert_text += content
-        insert_text += f"* {item}\n"
     insert_text += "\n"
 
     file_path = changelog_filepath(crate)
@@ -193,7 +257,7 @@ def main() -> None:
 
     ignore_labels = ["CI", "dependencies"]
 
-    sections = {}
+    crate_sections = defaultdict(list)
     unsorted_prs = []
     unsorted_commits = []
 
@@ -209,66 +273,42 @@ def main() -> None:
             unsorted_commits.append(summary)
         else:
             if f"[#{pr_number}]" in all_changelogs:
-                print(f"Ignoring PR that is already in the changelog: #{pr_number}")
+                print(f"* Ignoring PR that is already in the changelog: [#{pr_number}](https://github.com/{OWNER}/{REPO}/pull/{pr_number})")
                 continue
 
-            # We prefer the PR title if available
+            assert pr_info is not None
-            title = pr_info.pr_title if pr_info else title
-            labels = pr_info.labels if pr_info else []
 
-            if "exclude from changelog" in labels:
+            if "exclude from changelog" in pr_info.labels:
                 continue
-            if "typo" in labels:
+            if "typo" in pr_info.labels:
                 # We get so many typo PRs. Let's not flood the changelog with them.
                 continue
 
-            summary = f"{title} [#{pr_number}](https://github.com/{OWNER}/{REPO}/pull/{pr_number})"
-
-            if INCLUDE_LABELS and 0 < len(labels):
-                summary += f" ({', '.join(labels)})"
-
-            if pr_info is not None:
-                gh_user_name = pr_info.gh_user_name
-                if gh_user_name not in OFFICIAL_DEVS:
-                    summary += f" (thanks [@{gh_user_name}](https://github.com/{gh_user_name})!)"
-
             added = False
 
             for crate in crate_names:
-                if crate in labels:
+                if crate in pr_info.labels:
-                    sections.setdefault(crate, []).append(summary)
+                    crate_sections[crate].append(pr_info)
                     added = True
 
             if not added:
-                if not any(label in labels for label in ignore_labels):
+                if not any(label in pr_info.labels for label in ignore_labels):
-                    unsorted_prs.append(summary)
+                    unsorted_prs.append(pr_summary(pr_info))
-
-    # Clean up:
-    for crate in crate_names:
-        if crate in sections:
-            items = sections[crate]
-            for i in range(len(items)):
-                line = items[i]
-                line = remove_prefix(line, f"[{crate}] ")
-                line = remove_prefix(line, f"{crate}: ")
-                line = remove_prefix(line, f"`{crate}`: ")
-                line = line[0].upper() + line[1:]  # Upper-case first letter
-                items[i] = line
 
 
     print()
     print(f"Full diff at https://github.com/emilk/egui/compare/{args.commit_range}")
     print()
     for crate in crate_names:
-        if crate in sections:
+        if crate in crate_sections:
-            items = sections[crate]
+            prs = crate_sections[crate]
-            print_section(crate, items)
+            print_section(crate, changelog_from_prs(prs, crate))
-    print_section("Unsorted PRs", unsorted_prs)
+    print_section("Unsorted PRs", "\n".join([f"* {item}" for item in unsorted_prs]))
-    print_section("Unsorted commits", unsorted_commits)
+    print_section("Unsorted commits", "\n".join([f"* {item}" for item in unsorted_commits]))
 
     if args.write:
         for crate in crate_names:
-            items = sections[crate] if crate in sections else ["Nothing new"]
+            items = changelog_from_prs(crate_sections[crate], crate)
             add_to_changelog_file(crate, items, args.version)