Add action to automatically check for updates and create a PR

.github/workflows/check-for-updates.yml

@@ -0,0 +1,111 @@
+name: Check for tzdata updates
+
+on:
+  schedule:
+    - cron: '0 9 * * *'  # Runs daily at 9AM UTC
+  workflow_dispatch:
+
+jobs:
+  check-pr-exists:
+    runs-on: ubuntu-latest
+    outputs:
+      pr_exists: ${{ steps.check_pr_exists.outputs.pr_exists }}
+    steps:
+      - name: Check if PR already exists
+        id: check_pr_exists
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          PR_EXISTS=$(gh pr --repo $GITHUB_REPOSITORY \
+                      list --search "Update tzdata to version" \
+                      --json number --jq '.[] | .number')
+          if [ -n "$PR_EXISTS" ]; then
+            echo "A PR updating the tzdata version already exists: https://github.com/python/tzdata/pulls/${PR_EXISTS}"
+            echo "pr_exists=true" >> $GITHUB_OUTPUT
+            exit 0
+          else
+            echo "pr_exists=false" >> $GITHUB_OUTPUT
+          fi
+
+  check-for-updates:
+    runs-on: ubuntu-latest
+    needs: check-pr-exists
+    permissions:
+      pull-requests: write
+      contents: write
+    if: needs.check-pr-exists.outputs.pr_exists == 'false'  # Run only if no PR exists
+    steps:
+      - name: Check out repository (shallow)
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 1  # Shallow clone to save time
+
+      - name: Set up Python 3.12
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.12'
+
+      - name: Install dependencies
+        run: |
+          python -m pip install tox
+          sudo apt-get install gh
+
+      - name: Run tox update
+        run: tox -e update
+
+      - name: Check for repository changes and commit
+        id: check_changes
+        run: |
+          git config --global user.email "action@github.com"
+          git config --global user.name "GitHub Action"
+
+          # Check for changes
+          if git diff --quiet; then
+            echo "No changes detected."
+            echo "CHANGES_DETECTED=false" >> $GITHUB_ENV
+            exit 0
+          fi
+
+          # Check for changes in the news.d directory
+          git add -A
+          news_files=$(git diff --cached --name-only --diff-filter=A | grep '^news.d/.*\.md' || true)
+
+          if [ -z "$news_files" ]; then
+            echo "No new file in news.d, failing the job."
+            exit 1
+          fi
+
+          if [ $(echo "$news_files" | wc -l) -ne 1 ]; then
+            echo "More than one new file added in news.d, failing the job."
+            exit 1
+          fi
+          echo "CHANGES_DETECTED=true" >> $GITHUB_ENV
+
+          # Extract TZDATA_VERSION from filename
+          TZDATA_VERSION=$(basename "$news_files" .md)
+
+          # Extract TZDATA_NEWS from file content
+          TZDATA_NEWS=$(cat "$news_files")
+
+          echo "TZDATA_VERSION=$TZDATA_VERSION" >> $GITHUB_ENV
+          echo "TZDATA_NEWS=$TZDATA_NEWS" >> $GITHUB_ENV
+
+      - name: Commit changes
+        id: commit_changes
+        if: env.CHANGES_DETECTED == 'true'
+        run: |
+          git checkout -b "updates/update_${TZDATA_VERSION}"
+          git commit -m "Update tzdata to version $TZDATA_VERSION" \
+                     -m "$TZDATA_NEWS"
+          git push --force origin "updates/update_${TZDATA_VERSION}"
+
+      - name: Create pull request
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        if: env.CHANGES_DETECTED == 'true'
+        run: |
+          gh pr create --title "Update tzdata to version $TZDATA_VERSION" \
+            --body "$TZDATA_NEWS" \
+            --base master \
+            --head $(git rev-parse --abbrev-ref HEAD) \
+            --label "automatic-updates"

update.py

@@ -186,6 +186,15 @@ def create_package(version: str, zonenames: Sequence[str], zoneinfo_dir: pathlib
             init_file.touch()
 
 
+def get_current_package_version() -> str:
+    with open(PKG_BASE / "tzdata/__init__.py", "rt") as f:
+        for line in f:
+            if line.startswith("IANA_VERSION"):
+                return line.split("=", 1)[1].strip(' "\n')
+
+    raise ValueError("IANA version not found!")
+
+
 def find_latest_version() -> str:
     r = requests.get(IANA_LATEST_LOCATION)
     fobj = io.BytesIO(r.content)
@@ -426,27 +435,54 @@ def update_news(news_entry: NewsEntry):
     "--news-only/--no-news-only",
     help="Flag to disable data updates and only update the news entry",
 )
+@click.option(
+    "--skip-existing/--no-skip-existing",
+    default=True,
+    help="Whether to skip the update if we're already at the current value.",
+)
 def main(
     version: str | None,
     news_only: bool,
+    skip_existing: bool,
     source_dir: pathlib.Path | None,
 ):
     logging.basicConfig(level=logging.INFO)
 
-    if source_dir is not None:
-        if version is None:
-            logging.error(
-                "--source-dir specified without --version: "
-                "If using --source-dir, --version must also be used."
+    if skip_existing:
+        existing_version: str | None = get_current_package_version()
+    else:
+        existing_version = None
+
+    if version is None or version != existing_version:
+        if source_dir is not None:
+            if version is None:
+                logging.error(
+                    "--source-dir specified without --version: "
+                    "If using --source-dir, --version must also be used."
+                )
+                sys.exit(-1)
+            download_locations: Sequence[pathlib.Path] | None = retrieve_local_tarballs(
+                version, source_dir
             )
-            sys.exit(-1)
-        download_locations = retrieve_local_tarballs(version, source_dir)
+        else:
+            if version is None:
+                version = find_latest_version()
+
+            if version != existing_version or not skip_existing:
+                download_locations = download_tzdb_tarballs(version)
+            else:
+                download_locations = None
     else:
-        if version is None:
-            version = find_latest_version()
+        download_locations = None
 
-        download_locations = download_tzdb_tarballs(version)
+    if skip_existing and version == existing_version:
+        logging.info(
+            f"Selected version {version} is identical "
+            f"to existing version {existing_version}; nothing to do!"
+        )
+        sys.exit(0)
 
+    assert download_locations is not None
     tzdb_location = unpack_tzdb_tarballs(download_locations)
 
     # Update the news entry