Better checks for Changelog entries

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## 0.7.5dev
 * [Feature] Added `--version` option to `pkgmt setup` to specify Python version
+* [Feature] Better checks for new entries added to CHANGELOG.md
 
 ## 0.7.4 (2023-09-08)
 

src/pkgmt/changelog.py

@@ -197,6 +197,10 @@ def __init__(self, text, project_root=".") -> None:
     def from_path(cls, path, project_root="."):
         return cls(text=Path(path).read_text(), project_root=project_root)
 
+    def extract_text_from_entry(self):
+        """Extract text from a single Changelog entry"""
+        return _extract_text_from_items(self.tree[0])
+
     def sort_last_section(self):
         """Sorts last section depending on the prefix"""
         self.check_latest_changelog_entries()

src/pkgmt/fail_if_invalid_changelog.py

@@ -0,0 +1,100 @@
+import subprocess
+import argparse
+import sys
+from pathlib import Path
+
+from pkgmt import changelog
+
+
+def latest_changelog_header(base_branch):
+    changelog_contents = (
+        subprocess.check_output(f"git show {base_branch}:CHANGELOG.md", shell=True)
+        .decode("utf-8")
+        .split()
+    )
+    for line in changelog_contents:
+        if "dev" in line:
+            return line.strip()
+
+
+def check_modified(base_branch, debug=False):
+    latest_section_main = latest_changelog_header(base_branch)
+    text = Path("CHANGELOG.md").read_text()
+    changelog_parser = changelog.CHANGELOG(text)
+    latest_section_current = changelog_parser.get_first_subheading()[1].strip()
+    if latest_section_main != latest_section_current:
+        print(
+            f"Latest section in CHANGELOG.md "
+            f"not up-to-date. Latest section in "
+            f"{base_branch}: {latest_section_main}"
+        )
+        return 1
+    cmd = f"git diff -U0 {base_branch}... -- CHANGELOG.md"
+    try:
+        out = subprocess.check_output(cmd, shell=True).decode("utf-8")
+        all_diff = []
+        for line in out.split("\n"):
+            if not (line.startswith("+++") or line.startswith("---")):
+                all_diff.append(line.strip())
+        git_removals = [
+            line[1:].strip()
+            for line in all_diff
+            if line.startswith("-")
+            if line[1:].strip() != ""
+        ]
+        git_additions = [
+            line[1:].strip()
+            for line in all_diff
+            if line.startswith("+")
+            if line[1:].strip() != ""
+        ]
+
+        if len(git_additions) == 0 or out == "" and debug:
+            print(f"CHANGELOG.md has not been modified with respect to '{base_branch}'")
+            return 1
+
+        if git_removals:
+            print(
+                f"These entries have been removed: "
+                f"{'; '.join(git_removals)}. Please revert the changes."
+            )
+            return 1
+
+        if git_additions:
+            latest_entries = changelog_parser.get_latest_changelog_section()
+            for line in git_additions:
+                if line:
+                    try:
+                        extracted_text = changelog.CHANGELOG(
+                            line
+                        ).extract_text_from_entry()
+                    except KeyError:
+                        continue
+                    if extracted_text[0] not in latest_entries:
+                        print(
+                            f"Entry '{line}' should be added "
+                            f"to section {latest_section_main}"
+                        )
+                        return 1
+
+                    elif extracted_text[0].strip() == "":
+                        print(f"You have added an empty entry: {line}")
+                        return 1
+
+    except subprocess.CalledProcessError:
+        pass
+    return 0
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(
+        description="Check if proper modifications has been done to CHANGELOG.md"
+    )
+    parser.add_argument(
+        "-b", "--base-branch", default="main", help="Base branch to compare against"
+    )
+    parser.add_argument("--debug", action="store_true", help="Print debug info")
+
+    args = parser.parse_args()
+    return_code = check_modified(args.base_branch, debug=args.debug)
+    sys.exit(return_code)

tests/conftest.py

@@ -124,3 +124,27 @@ def tmp_package_modi_2(root, tmp_empty):
     yield tmp_empty
 
     os.chdir(old)
+
+
+@pytest.fixture
+def tmp_package_changelog(root, tmp_empty):
+    old = Path.cwd()
+    path_to_templates = root / "tests" / "assets" / "package_name"
+    shutil.copytree(str(path_to_templates), "copy")
+    os.chdir("copy")
+
+    subprocess.run(["git", "init"])
+    subprocess.check_call(["git", "config", "commit.gpgsign", "false"])
+    subprocess.check_call(["git", "config", "user.email", "ci@ploomberio"])
+    subprocess.check_call(["git", "config", "user.name", "Ploomber"])
+    subprocess.run(["git", "checkout", "-b", "main"])
+    with open("CHANGELOG.md", "a") as f:
+        f.write("\n")
+    subprocess.run(["git", "add", "--all"])
+    subprocess.run(["git", "commit", "-m", "init-commit-message"])
+
+    subprocess.run(["git", "checkout", "-b", "test_modified_changelog"])
+
+    yield tmp_empty
+
+    os.chdir(old)

tests/test_links.py

@@ -90,12 +90,6 @@ def test_check_if_broken(url, code, broken):
     assert response.broken == broken
 
 
-def test_check_if_broken_doesnt_accept_head_request():
-    response = links.LinkChecker().check_if_broken("https://binder.ploomber.io")
-    assert response.code == 405
-    assert not response.broken
-
-
 @pytest.mark.parametrize(
     "extensions, expected",
     [

tests/test_modified.py

@@ -1,5 +1,7 @@
 import pytest
-from pkgmt import fail_if_modified, fail_if_not_modified
+import subprocess
+
+from pkgmt import fail_if_modified, fail_if_not_modified, fail_if_invalid_changelog
 from pathlib import Path
 
 
@@ -81,3 +83,115 @@ def test_check_modified_in_2(tmp_package_modi_2, base_branch, include_path, retu
         fail_if_not_modified.check_modified(base_branch, include_path, debug=True)
         == returncode
     )
+
+
+# @pytest.mark.parametrize("contents", ["""
+# # CHANGELOG
+#
+# ## 0.1dev
+#
+# * [Fix] Fixes #1
+# * [Feature] Added feature 1
+# """
+# ])
+def test_check_modified_changelog(tmp_package_changelog):
+    Path("CHANGELOG.md").write_text(
+        """
+# CHANGELOG
+
+## 0.1dev
+
+* [Fix] Fixes #1
+* [Feature] Added feature 1
+"""
+    )
+    subprocess.run(["git", "add", "CHANGELOG.md"])
+    subprocess.run(["git", "commit", "-m", "changelog_modified"])
+    assert fail_if_invalid_changelog.check_modified("main", debug=True) == 0
+
+
+@pytest.mark.parametrize(
+    "contents, message",
+    [
+        (
+            """
+# CHANGELOG
+
+## 0.1dev
+
+* [Fix] Fixes #1
+
+
+""",
+            "CHANGELOG.md has not been modified with respect to 'main'",
+        ),
+        (
+            """
+# CHANGELOG
+
+## 0.1dev
+
+* [Fix] Fixes #1
+
+""",
+            "CHANGELOG.md has not been modified with respect to 'main'",
+        ),
+        (
+            """
+# CHANGELOG
+
+## 0.0.1dev
+
+* [Fix] Fixes #1
+
+""",
+            "Latest section in CHANGELOG.md not up-to-date. "
+            "Latest section in main: 0.1dev",
+        ),
+        (
+            """
+# CHANGELOG
+
+## 0.1dev
+
+* [Fix] Fixes #1
+
+## 0.0.1
+
+* [Feature] Some feature
+
+""",
+            "Entry '* [Feature] Some feature' should be added to section 0.1dev",
+        ),
+        (
+            """
+# CHANGELOG
+
+## 0.1dev
+
+* [Fix] Fixes #2
+
+""",
+            "These entries have been removed: * [Fix] Fixes #1. "
+            "Please revert the changes.",
+        ),
+    ],
+    ids=["blanks", "no-change", "outdated-header", "incorrect-section", "removal"],
+)
+def test_check_modified_changelog_error(
+    tmp_package_changelog, contents, message, capsys
+):
+    Path("CHANGELOG.md").write_text(contents)
+    subprocess.run(["git", "add", "CHANGELOG.md"])
+    subprocess.run(["git", "commit", "-m", "changelog_modified"])
+    assert fail_if_invalid_changelog.check_modified("main", debug=True) == 1
+    assert message in capsys.readouterr().out
+
+
+def test_check_modified_changelog_error_file_added(tmp_package_changelog, capsys):
+    Path("some_file.txt").write_text("content")
+    subprocess.run(["git", "add", "some_file.txt"])
+    subprocess.run(["git", "commit", "-m", "added file"])
+    assert fail_if_invalid_changelog.check_modified("main", debug=True) == 1
+    out = capsys.readouterr().out
+    assert "CHANGELOG.md has not been modified with respect to 'main'" in out