Merge pull request #12 from SFTtech/milo/extend-debian-metadata-parsing

feat(common): add datamodels and parsing for package changelogs

Author: mikonse

packages/debmagic-common/src/debmagic/common/errors.py

@@ -0,0 +1,2 @@
+class DebmagicError(RuntimeError):
+    pass

packages/debmagic-common/src/debmagic/common/models/__init__.py

@@ -0,0 +1,85 @@
+from dataclasses import dataclass
+from datetime import datetime
+from pathlib import Path
+from typing import Self
+
+from debian.changelog import Changelog as DebianChangelog
+
+from ..errors import DebmagicError
+from ..type_utils import IterableDataSource
+from .package_version import PackageVersion
+
+
+class ChangelogFormatError(DebmagicError):
+    pass
+
+
+def _parse_changelog_date(date: str) -> datetime:
+    return datetime.strptime(date, "%a, %d %b %Y %H:%M:%S %z")
+
+
+def _parse_author(author: str | None) -> tuple[str | None, str | None]:
+    if author is None:
+        return None, None
+    if "<" not in author:
+        raise ChangelogFormatError()
+    name, email = author.split("<")
+    return name, email
+
+
+def _parse_distributions(distributions: str | None) -> list[str]:
+    parsed: list[str] = [] if distributions is None else distributions.split(",")
+    parsed = list(map(str.strip, parsed))
+    return parsed
+
+
+@dataclass
+class ChangelogMetadata:
+    urgency: str | None
+    binary_only: bool = False
+
+
+@dataclass
+class ChangelogEntry:
+    package: str | None
+    version: PackageVersion
+    distributions: list[str]
+    metadata: ChangelogMetadata
+    changes: list[str]
+    author_name: str | None
+    author_email: str | None
+    date: datetime | None
+
+
+@dataclass
+class Changelog:
+    entries: list[ChangelogEntry]
+
+    @classmethod
+    def from_file(cls, file: IterableDataSource) -> Self:
+        changelog = DebianChangelog()
+        changelog.parse_changelog(file)
+        entries = []
+        for block in changelog:
+            author_name, author_email = _parse_author(block.author)
+            entries.append(
+                ChangelogEntry(
+                    package=block.package,
+                    distributions=_parse_distributions(block.distributions),
+                    version=block.version,
+                    author_name=author_name,
+                    author_email=author_email,
+                    changes=block.changes(),
+                    date=_parse_changelog_date(block.date) if block.date is not None else None,
+                    metadata=ChangelogMetadata(
+                        urgency=block.urgency,
+                    ),
+                )
+            )
+
+        return cls(entries=entries)
+
+    @classmethod
+    def from_changelog_file(cls, changelog_file_path: Path) -> Self:
+        with changelog_file_path.open() as f:
+            return cls.from_file(f)

packages/debmagic-common/src/debmagic/common/models/changelog.py

@@ -4,6 +4,8 @@
 
 from debian import deb822
 
+from .models.changelog import Changelog
+
 
 @dataclass
 class BinaryPackage:
@@ -16,13 +18,17 @@ class BinaryPackage:
 class SourcePackage:
     name: str
     binary_packages: list[BinaryPackage]
+    changelog: Changelog
 
     @classmethod
-    def from_control_file(cls, control_file_path: Path) -> Self:
+    def from_debian_directory(cls, debian_dir_path: Path) -> Self:
+        control_file_path = debian_dir_path / "control"
         src_pkg: Self | None = None
         # which binary packages should be produced?
         bin_pkgs: list[BinaryPackage] = []
 
+        changelog = Changelog.from_changelog_file(debian_dir_path / "changelog")
+
         for block in deb822.DebControl.iter_paragraphs(
             control_file_path.open(),
             use_apt_pkg=False,  # don't depend on python3-apt for now.
@@ -31,10 +37,7 @@ def from_control_file(cls, control_file_path: Path) -> Self:
                 if src_pkg is not None:
                     raise RuntimeError("encountered multiple Source: blocks in control file")
                 src_name = block["Source"]
-                src_pkg = cls(
-                    src_name,
-                    bin_pkgs,
-                )
+                src_pkg = cls(src_name, bin_pkgs, changelog=changelog)
 
             if "Package" in block:
                 bin_pkg = BinaryPackage(

…n/src/debmagic/common/package_version.py …ebmagic/common/models/package_version.pypackages/debmagic-common/src/debmagic/common/package_version.py renamed to packages/debmagic-common/src/debmagic/common/models/package_version.py packages/debmagic-common/src/debmagic/common/package_version.py renamed to packages/debmagic-common/src/debmagic/common/models/package_version.py

@@ -0,0 +1,3 @@
+from typing import IO, Iterable, Text
+
+IterableDataSource = bytes | Text | IO[Text] | Iterable[Text] | Iterable[bytes]

packages/debmagic-common/src/debmagic/common/package.py

@@ -0,0 +1,8 @@
+pkg1 (0.1.0) trixie; urgency=medium
+
+  * dummy changelog entry.
+    multiline description
+  
+  * another entry.
+
+ -- Some Author <[email protected]>  Sat, 27 Sep 2025 20:55:33 +0200

packages/debmagic-common/src/debmagic/common/type_utils.py

@@ -1,8 +1,8 @@
 Source: pkg1
 Section: devel
-Maintainer: Some Maintainers <maintainers@foo.com>
+Maintainer: Some Maintainers <maintainers@debian.org>
 Uploaders:
- Some Uploader <uploader@foo.com>,
+ Some Uploader <uploader@debian.org>,
 Build-Depends:
  debhelper-compat (= 13),
 Rules-Requires-Root: no

packages/debmagic-common/tests/assets/pkg1/debian/changelog

@@ -0,0 +1,23 @@
+from datetime import datetime, timedelta, timezone
+from pathlib import Path
+
+import pytest
+from debmagic.common.models.changelog import Changelog
+
+asset_base = Path(__file__).parent / "assets"
+
+
+@pytest.mark.parametrize(
+    "debian_folder_path, name",
+    [
+        (asset_base / "pkg1/debian", "pkg1"),
+    ],
+)
+def test_source_package_parsing(debian_folder_path: Path, name: str):
+    changelog = Changelog.from_changelog_file(debian_folder_path / "changelog")
+
+    assert changelog.entries[0].package == name
+    assert changelog.entries[0].date == datetime(
+        year=2025, month=9, day=27, hour=20, minute=55, second=33, tzinfo=timezone(timedelta(hours=2))
+    )
+    assert changelog.entries[0].distributions == ["trixie"]

…n/tests/assets/pkg1_minimal_control_file …-common/tests/assets/pkg1/debian/controlpackages/debmagic-common/tests/assets/pkg1_minimal_control_file renamed to packages/debmagic-common/tests/assets/pkg1/debian/control packages/debmagic-common/tests/assets/pkg1_minimal_control_file renamed to packages/debmagic-common/tests/assets/pkg1/debian/control

@@ -7,12 +7,12 @@
 
 
 @pytest.mark.parametrize(
-    "control_file_path, name",
+    "debian_folder_path, name",
     [
-        (asset_base / "pkg1_minimal_control_file", "pkg1"),
+        (asset_base / "pkg1/debian", "pkg1"),
     ],
 )
-def test_source_package_parsing(control_file_path: Path, name: str):
-    package = SourcePackage.from_control_file(control_file_path)
+def test_source_package_parsing(debian_folder_path: Path, name: str):
+    package = SourcePackage.from_debian_directory(debian_folder_path)
 
     assert package.name == name