feat(changelog)!: Merge similar commits within group (#197)

When changelog contain commits like `Text (#123)` and `Text (#456)` in
some group (Features, Fixes, Refactoring, Other) _badabump_ now will
merge those commits into `Text (#123, #456)` line.

Fixes: #189

.github/actions/install_python_and_poetry/action.yml

@@ -43,7 +43,7 @@ runs:
       run: "pipx install --python='${{ steps.python.outputs.python-path }}' poetry==${{ inputs.poetry-version }}"
 
     - name: "Cache venv"
-      uses: "actions/[email protected].10"
+      uses: "actions/[email protected].11"
       with:
         path: "./.venv/"
         key: "venv-${{ runner.os }}-${{ steps.python.outputs.python-version }}-${{ hashFiles('poetry.lock') }}${{ inputs.cache-key-suffix }}"

.github/actions/run_pre_commit/action.yml

@@ -14,7 +14,7 @@ runs:
       shell: "bash"
 
     - name: "Cache mypy"
-      uses: "actions/[email protected].10"
+      uses: "actions/[email protected].11"
       with:
         path: "./.mypy_cache/"
         key: "mypy-${{ runner.os }}-${{ inputs.python-version }}"

.github/actions/run_tox/action.yml

@@ -36,7 +36,7 @@ runs:
       shell: "bash"
 
     - name: "Cache tox"
-      uses: "actions/[email protected].10"
+      uses: "actions/[email protected].11"
       with:
         path: "./.tox/"
         key: "tox-${{ inputs.python-version }}"

.github/workflows/release_pr.yml

@@ -46,7 +46,7 @@ jobs:
         run: "poetry run python3 -m badabump --ci ${{ github.event.inputs.args }}"
 
       - id: "token"
-        uses: "tibdex/[email protected]"
+        uses: "tibdex/[email protected].0"
         with:
           app_id: "${{ secrets.BADABUMP_APP_ID }}"
           private_key: "${{ secrets.BADABUMP_APP_PRIVATE_KEY }}"

.github/workflows/release_tag.yml

@@ -20,7 +20,7 @@ jobs:
 
     steps:
       - id: "token"
-        uses: "tibdex/[email protected]"
+        uses: "tibdex/[email protected].0"
         with:
           app_id: "${{ secrets.BADABUMP_APP_ID }}"
           private_key: "${{ secrets.BADABUMP_APP_PRIVATE_KEY }}"

src/badabump/changelog.py

@@ -29,6 +29,10 @@
     r"^(Closes|Fixes|Issue|Ref|Relates): (?P<issue>.+)$", re.M
 )
 
+FORMATTED_COMMIT_WITH_PR_RE = re.compile(
+    r"^(?P<formatted_commit>.*) \(\#(?P<pr_number>\d+)\)$"
+)
+
 logger = logging.getLogger(__name__)
 
 
@@ -202,16 +206,12 @@ def format_block(
             return "\n\n".join((header, items))
 
         def format_commits(commits: Iterator[ConventionalCommit]) -> str:
-            formatted_commits = []
-            for commit in commits:
-                formatted_commit = commit.format(
-                    format_type, ignore_footer_urls=ignore_footer_urls
+            return "\n".join(
+                ul_li(item)
+                for item in prepare_formatted_commits(
+                    commits, format_type, ignore_footer_urls=ignore_footer_urls
                 )
-                if formatted_commit in formatted_commits:
-                    continue
-                formatted_commits.append(formatted_commit)
-
-            return "\n".join(ul_li(item) for item in formatted_commits)
+            )
 
         features = format_block("Features:", self.feature_commits)
         fixes = format_block("Fixes:", self.fix_commits)
@@ -279,6 +279,55 @@ def markdown_header(value: str, *, level: int) -> str:
     return f'{"#" * level} {value}'
 
 
+def prepare_formatted_commits(
+    commits: Iterator[ConventionalCommit],
+    format_type: FormatTypeEnum,
+    *,
+    ignore_footer_urls: bool,
+) -> List[str]:
+    storage = []
+
+    for commit in commits:
+        # First, format commit as asked
+        formatted_commit = commit.format(
+            format_type, ignore_footer_urls=ignore_footer_urls
+        )
+
+        # Next, check whether it is already added to the storage. If previously
+        # added - do nothing
+        if formatted_commit in storage:
+            continue
+
+        # If not yet added, check if commit match regex "with PR"
+        with_pr_matched = FORMATTED_COMMIT_WITH_PR_RE.match(formatted_commit)
+        # If not matched - just append formatted commit to the storage
+        if not with_pr_matched:
+            storage.append(formatted_commit)
+            continue
+
+        # Now, attempt to match other commit in storage
+        with_pr_data = with_pr_matched.groupdict()
+        with_pr_prefix = with_pr_data["formatted_commit"]
+        other_idx = -1
+
+        for idx, other_commit in enumerate(storage):
+            if other_commit.startswith(with_pr_prefix):
+                other_idx = idx
+                break
+
+        # If such commit does not exist - just append formatted commit to the
+        # storage
+        if other_idx == -1:
+            storage.append(formatted_commit)
+            continue
+
+        # Otherwise, merge commit message of other commit and current commit
+        pr_number = with_pr_data["pr_number"]
+        storage[other_idx] = f"{storage[other_idx][:-1]}, #{pr_number})"
+
+    return storage
+
+
 def rst_h1(value: str) -> str:
     return rst_header(value, symbol="=")
 

tests/test_changelog.py

@@ -44,6 +44,18 @@
 Fixes: DEV-1010
 """
 
+DEFAULT_GIT_COMMITS = [
+    FEATURE_COMMIT,
+    FIX_COMMIT,
+    CI_BREAKING_COMMIT,
+    REFACTOR_COMMIT,
+    DOCS_SCOPE_COMMIT,
+    REFACTOR_COMMIT,
+    CI_BREAKING_COMMIT,
+    REFACTOR_COMMIT,
+]
+
+
 CHANGELOG_EMPTY = "No changes since last pre-release"
 
 CHANGELOG_FILE_MD = """## Features:
@@ -121,10 +133,6 @@
 UTCNOW = datetime.datetime.utcnow()
 
 
-def test_changelog_duplicate_commits_with_prs():
-    ...
-
-
 @pytest.mark.parametrize(
     "changelog_type, format_type, expected",
     (
@@ -162,18 +170,7 @@ def test_changelog_empty(changelog_type, format_type, expected):
     ),
 )
 def test_changelog_format_file(format_type, is_pre_release, expected):
-    changelog = ChangeLog.from_git_commits(
-        [
-            FEATURE_COMMIT,
-            FIX_COMMIT,
-            CI_BREAKING_COMMIT,
-            REFACTOR_COMMIT,
-            DOCS_SCOPE_COMMIT,
-            REFACTOR_COMMIT,
-            CI_BREAKING_COMMIT,
-            REFACTOR_COMMIT,
-        ]
-    )
+    changelog = ChangeLog.from_git_commits(DEFAULT_GIT_COMMITS)
     content = changelog.format(
         ChangeLogTypeEnum.changelog_file,
         format_type,
@@ -192,15 +189,7 @@ def test_changelog_format_file(format_type, is_pre_release, expected):
     ),
 )
 def test_changelog_format_git(format_type, is_pre_release, expected):
-    changelog = ChangeLog.from_git_commits(
-        [
-            FEATURE_COMMIT,
-            FIX_COMMIT,
-            CI_BREAKING_COMMIT,
-            DOCS_SCOPE_COMMIT,
-            REFACTOR_COMMIT,
-        ]
-    )
+    changelog = ChangeLog.from_git_commits(DEFAULT_GIT_COMMITS)
     content = changelog.format(
         ChangeLogTypeEnum.git_commit,
         format_type,
@@ -247,6 +236,29 @@ def test_changelog_invalid_commit_non_strict_mode():
     assert changelog.has_micro_change is True
 
 
+def test_changelog_merge_similar_commits():
+    changelog = ChangeLog.from_git_commits(
+        [
+            "fix: Does not matter (#9999)",
+            f"{FIX_COMMIT} (#9000)",
+            f"{FIX_COMMIT} (#69)",
+            f"{FIX_COMMIT} (#42)",
+        ]
+    )
+    content = changelog.format(
+        ChangeLogTypeEnum.changelog_file,
+        FormatTypeEnum.markdown,
+        is_pre_release=False,
+    )
+    assert (
+        content
+        == f"""## Fixes:
+
+- {FIX_COMMIT[5:]} (#42, #69, #9000)
+- Does not matter (#9999)"""
+    )
+
+
 @pytest.mark.parametrize(
     "fix_commit",
     (