Review fixes v2

Author: MaciejKaras

scripts/release/changelog.py

@@ -10,6 +10,7 @@
 DEFAULT_INITIAL_GIT_TAG_VERSION = "1.0.0"
 FILENAME_DATE_FORMAT = "%Y%m%d"
 FRONTMATTER_DATE_FORMAT = "%Y-%m-%d"
+MAX_TITLE_LENGTH = 50
 
 PRELUDE_ENTRIES = ["prelude"]
 BREAKING_CHANGE_ENTRIES = ["breaking", "major"]
@@ -130,3 +131,33 @@ def strip_changelog_entry_frontmatter(file_contents: str) -> (ChangeMeta, str):
     contents = data.content + "\n"
 
     return meta, contents
+
+
+def get_changelog_filename(title: str, kind: ChangeKind, date: datetime) -> str:
+    sanitized_title = sanitize_title(title)
+    filename_date = datetime.datetime.strftime(date, FILENAME_DATE_FORMAT)
+
+    return f"{filename_date}_{kind}_{sanitized_title}.md"
+
+
+def sanitize_title(title: str) -> str:
+    # Only keep alphanumeric characters, dashes, underscores and spaces
+    regex = re.compile("[^a-zA-Z0-9-_ ]+")
+    title = regex.sub("", title)
+
+    # Replace multiple dashes, underscores and spaces with underscores
+    regex_underscore = re.compile("[-_ ]+")
+    title = regex_underscore.sub(" ", title).strip()
+
+    # Lowercase and split by space
+    words = [word.lower() for word in title.split(" ")]
+
+    result = words[0]
+
+    for word in words[1:]:
+        if len(result) + len("_") + len(word) <= MAX_TITLE_LENGTH:
+            result = result + "_" + word
+        else:
+            break
+
+    return result

scripts/release/changelog_test.py

@@ -1,86 +1,85 @@
 import datetime
+import unittest
 
-import pytest
 from changelog import (
+    MAX_TITLE_LENGTH,
     ChangeKind,
     extract_date_and_kind_from_file_name,
+    sanitize_title,
     strip_changelog_entry_frontmatter,
 )
 
 
-def test_extract_changelog_data_from_file_name():
-    # Test prelude
-    assert extract_date_and_kind_from_file_name("20250502_prelude_release_notes.md") == (
-        datetime.date(2025, 5, 2),
-        ChangeKind.PRELUDE,
-    )
+class TestExtractChangelogDataFromFileName(unittest.TestCase):
+    def test_prelude(self):
+        date, kind = extract_date_and_kind_from_file_name("20250502_prelude_release_notes.md")
+        self.assertEqual(date, datetime.date(2025, 5, 2))
+        self.assertEqual(kind, ChangeKind.PRELUDE)
 
-    # Test breaking changes
-    assert extract_date_and_kind_from_file_name("20250101_breaking_api_update.md") == (
-        datetime.date(2025, 1, 1),
-        ChangeKind.BREAKING,
-    )
-    assert extract_date_and_kind_from_file_name("20250508_breaking_remove_deprecated.md") == (
-        datetime.date(2025, 5, 8),
-        ChangeKind.BREAKING,
-    )
-    assert extract_date_and_kind_from_file_name("20250509_major_schema_change.md") == (
-        datetime.date(2025, 5, 9),
-        ChangeKind.BREAKING,
-    )
+    def test_breaking_changes(self):
+        date, kind = extract_date_and_kind_from_file_name("20250101_breaking_api_update.md")
+        self.assertEqual(date, datetime.date(2025, 1, 1))
+        self.assertEqual(kind, ChangeKind.BREAKING)
 
-    # Test features
-    assert extract_date_and_kind_from_file_name("20250509_feature_new_dashboard.md") == (
-        datetime.date(2025, 5, 9),
-        ChangeKind.FEATURE,
-    )
-    assert extract_date_and_kind_from_file_name("20250511_feat_add_metrics.md") == (
-        datetime.date(2025, 5, 11),
-        ChangeKind.FEATURE,
-    )
+        date, kind = extract_date_and_kind_from_file_name("20250508_breaking_remove_deprecated.md")
+        self.assertEqual(date, datetime.date(2025, 5, 8))
+        self.assertEqual(kind, ChangeKind.BREAKING)
 
-    # Test fixes
-    assert extract_date_and_kind_from_file_name("20251210_fix_olm_missing_images.md") == (
-        datetime.date(2025, 12, 10),
-        ChangeKind.FIX,
-    )
-    assert extract_date_and_kind_from_file_name("20251010_bugfix_memory_leak.md") == (
-        datetime.date(2025, 10, 10),
-        ChangeKind.FIX,
-    )
-    assert extract_date_and_kind_from_file_name("20250302_hotfix_security_issue.md") == (
-        datetime.date(2025, 3, 2),
-        ChangeKind.FIX,
-    )
-    assert extract_date_and_kind_from_file_name("20250301_patch_typo_correction.md") == (
-        datetime.date(2025, 3, 1),
-        ChangeKind.FIX,
-    )
+        date, kind = extract_date_and_kind_from_file_name("20250509_major_schema_change.md")
+        self.assertEqual(date, datetime.date(2025, 5, 9))
+        self.assertEqual(kind, ChangeKind.BREAKING)
 
-    # Test other
-    assert extract_date_and_kind_from_file_name("20250520_docs_update_readme.md") == (
-        datetime.date(2025, 5, 20),
-        ChangeKind.OTHER,
-    )
-    assert extract_date_and_kind_from_file_name("20250610_refactor_codebase.md") == (
-        datetime.date(2025, 6, 10),
-        ChangeKind.OTHER,
-    )
+    def test_features(self):
+        date, kind = extract_date_and_kind_from_file_name("20250509_feature_new_dashboard.md")
+        self.assertEqual(date, datetime.date(2025, 5, 9))
+        self.assertEqual(kind, ChangeKind.FEATURE)
+
+        date, kind = extract_date_and_kind_from_file_name("20250511_feat_add_metrics.md")
+        self.assertEqual(date, datetime.date(2025, 5, 11))
+        self.assertEqual(kind, ChangeKind.FEATURE)
+
+    def test_fixes(self):
+        date, kind = extract_date_and_kind_from_file_name("20251210_fix_olm_missing_images.md")
+        self.assertEqual(date, datetime.date(2025, 12, 10))
+        self.assertEqual(kind, ChangeKind.FIX)
 
-    # Invalid date part (day 40 does not exist)
-    with pytest.raises(Exception) as e:
-        extract_date_and_kind_from_file_name("20250640_refactor_codebase.md")
-    assert str(e.value) == "20250640_refactor_codebase.md - date 20250640 is not in the expected format YYYYMMDD"
+        date, kind = extract_date_and_kind_from_file_name("20251010_bugfix_memory_leak.md")
+        self.assertEqual(date, datetime.date(2025, 10, 10))
+        self.assertEqual(kind, ChangeKind.FIX)
 
-    # Wrong file name format (date part)
-    with pytest.raises(Exception) as e:
-        extract_date_and_kind_from_file_name("202yas_refactor_codebase.md")
-    assert str(e.value) == "202yas_refactor_codebase.md - doesn't match expected pattern"
+        date, kind = extract_date_and_kind_from_file_name("20250302_hotfix_security_issue.md")
+        self.assertEqual(date, datetime.date(2025, 3, 2))
+        self.assertEqual(kind, ChangeKind.FIX)
 
-    # Wrong file name format (missing title part)
-    with pytest.raises(Exception) as e:
-        extract_date_and_kind_from_file_name("20250620_change.md")
-    assert str(e.value) == "20250620_change.md - doesn't match expected pattern"
+        date, kind = extract_date_and_kind_from_file_name("20250301_patch_typo_correction.md")
+        self.assertEqual(date, datetime.date(2025, 3, 1))
+        self.assertEqual(kind, ChangeKind.FIX)
+
+    def test_other(self):
+        date, kind = extract_date_and_kind_from_file_name("20250520_docs_update_readme.md")
+        self.assertEqual(date, datetime.date(2025, 5, 20))
+        self.assertEqual(kind, ChangeKind.OTHER)
+
+        date, kind = extract_date_and_kind_from_file_name("20250610_refactor_codebase.md")
+        self.assertEqual(date, datetime.date(2025, 6, 10))
+        self.assertEqual(kind, ChangeKind.OTHER)
+
+    def test_invalid_date(self):
+        with self.assertRaises(Exception) as context:
+            extract_date_and_kind_from_file_name("20250640_refactor_codebase.md")
+        self.assertEqual(
+            str(context.exception), "20250640_refactor_codebase.md - date 20250640 is not in the expected format %Y%m%d"
+        )
+
+    def test_wrong_file_name_format_date(self):
+        with self.assertRaises(Exception) as context:
+            extract_date_and_kind_from_file_name("202yas_refactor_codebase.md")
+        self.assertEqual(str(context.exception), "202yas_refactor_codebase.md - doesn't match expected pattern")
+
+    def test_wrong_file_name_format_missing_title(self):
+        with self.assertRaises(Exception) as context:
+            extract_date_and_kind_from_file_name("20250620_change.md")
+        self.assertEqual(str(context.exception), "20250620_change.md - doesn't match expected pattern")
 
 
 def test_strip_changelog_entry_frontmatter():
@@ -107,3 +106,33 @@ def test_strip_changelog_entry_frontmatter():
   * Added new property [spec.search](https://www.mongodb.com/docs/kubernetes/current/mongodb/specification/#spec-search) to enable MongoDB Search.
 """
     )
+
+
+class TestSanitizeTitle(unittest.TestCase):
+    def test_basic_case(self):
+        self.assertEqual(sanitize_title("Simple Title"), "simple_title")
+
+    def test_non_alphabetic_chars(self):
+        self.assertEqual(sanitize_title("Title tha@t-_ contain's strange char&s!"), "title_that_contains_strange_chars")
+
+    def test_with_numbers_and_dashes(self):
+        self.assertEqual(sanitize_title("Title with 123 numbers to-go!"), "title_with_123_numbers_to_go")
+
+    def test_mixed_case(self):
+        self.assertEqual(sanitize_title("MiXeD CaSe TiTlE"), "mixed_case_title")
+
+    def test_length_limit(self):
+        long_title = "This is a very long title that should be truncated because it exceeds the maximum length"
+        sanitized_title = sanitize_title(long_title)
+        self.assertTrue(len(sanitized_title) <= MAX_TITLE_LENGTH)
+        self.assertEqual(sanitized_title, "this_is_a_very_long_title_that_should_be_truncated")
+
+    def test_leading_trailing_spaces(self):
+        sanitized_title = sanitize_title("  Title with spaces  ")
+        self.assertEqual(sanitized_title, "title_with_spaces")
+
+    def test_empty_title(self):
+        self.assertEqual(sanitize_title(""), "")
+
+    def test_only_non_alphabetic(self):
+        self.assertEqual(sanitize_title("!@#"), "")

scripts/release/create_changelog.py

@@ -1,41 +1,19 @@
 import argparse
 import datetime
 import os
-import re
 
 from scripts.release.changelog import (
     BREAKING_CHANGE_ENTRIES,
     BUGFIX_ENTRIES,
     DEFAULT_CHANGELOG_PATH,
     FEATURE_ENTRIES,
-    FILENAME_DATE_FORMAT,
     FRONTMATTER_DATE_FORMAT,
     PRELUDE_ENTRIES,
+    get_change_kind,
+    get_changelog_filename,
     parse_change_date,
 )
 
-MAX_TITLE_LENGTH = 50
-
-
-def sanitize_title(title: str) -> str:
-    # Remove non-alphabetic and space characters
-    regex = re.compile("[^a-zA-Z ]+")
-    title = regex.sub("", title)
-
-    # Lowercase and split by space
-    words = [word.lower() for word in title.split(" ")]
-
-    result = words[0]
-
-    for word in words[1:]:
-        if len(result) + len("_") + len(word) <= MAX_TITLE_LENGTH:
-            result = result + "_" + word
-        else:
-            break
-
-    return result
-
-
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter)
     parser.add_argument(
@@ -79,9 +57,11 @@ def sanitize_title(title: str) -> str:
     parser.add_argument("title", type=str, help="Title for the changelog entry")
     args = parser.parse_args()
 
+    title = args.title
+    date_str = args.date
     date = parse_change_date(args.date, FRONTMATTER_DATE_FORMAT)
-    sanitized_title = sanitize_title(args.title)
-    filename = f"{datetime.datetime.strftime(date, FILENAME_DATE_FORMAT)}_{args.kind}_{sanitized_title}.md"
+    kind = get_change_kind(args.kind)
+    filename = get_changelog_filename(title, kind, date)
 
     working_dir = os.getcwd()
     changelog_path = os.path.join(working_dir, args.changelog_path, filename)
@@ -93,9 +73,9 @@ def sanitize_title(title: str) -> str:
     with open(changelog_path, "w") as f:
         # Add frontmatter based on args
         f.write("---\n")
-        f.write(f"title: {args.title}\n")
-        f.write(f"kind: {args.kind}\n")
-        f.write(f"date: {date}\n")
+        f.write(f"title: {title}\n")
+        f.write(f"kind: {str(kind)}\n")
+        f.write(f"date: {date_str}\n")
         f.write("---\n\n")
 
     if args.editor: