persist repository metadata in artifacts.yaml

Author: AthreyVinay

tests/unit/artifact/test_copr_repository.py

@@ -48,3 +48,31 @@ def test_fetch_contents_enables_repository(
     mock_from_file.assert_called_once()
     assert result == []
     assert provider.repository is mock_repo
+
+
+def test_enumerate_artifacts(mock_copr, mock_package_manager, artifact_provider, tmppath):
+    mock_repo = MagicMock()
+    mock_repo.repo_ids = ['group_teemtee-stable-fedora-42-x86_64']
+    mock_guest = MagicMock()
+    mock_guest.package_manager = mock_package_manager
+
+    mock_package_manager.list_packages.return_value = [
+        'tmt-1.69.0-1.fc42.noarch',
+        'tmt-all-0:1.69.0-1.fc42.noarch',
+    ]
+
+    provider = artifact_provider("copr.repository:@teemtee/stable")
+
+    with patch(
+        'tmt.steps.prepare.artifact.providers.copr_repository.Repository.from_file_path',
+        return_value=mock_repo,
+    ):
+        provider.fetch_contents(mock_guest, tmppath / "artifacts")
+
+    provider.enumerate_artifacts(mock_guest)
+
+    assert len(provider.artifacts) == 2
+    assert provider.artifacts[0].version.name == 'tmt'
+    assert provider.artifacts[0].version.version == '1.69.0'
+    assert provider.artifacts[1].version.name == 'tmt-all'
+    assert provider.artifacts[1].version.epoch == 0

tmt/steps/prepare/artifact/__init__.py

@@ -265,6 +265,10 @@ def go(
             guest.package_manager.install_repository(repo)
             logger.debug(f"Installed repository '{repo.name}'.")
 
+        # Enumerate artifacts from installed repositories.
+        for provider in providers:
+            provider.enumerate_artifacts(guest)
+
         # Persist artifact metadata to YAML
         self._save_artifacts_metadata(providers)
 

tmt/steps/prepare/artifact/providers/__init__.py

@@ -5,7 +5,6 @@
 import configparser
 from abc import ABC, abstractmethod
 from collections.abc import Iterator, Sequence
-from functools import cached_property
 from re import Pattern
 from shlex import quote
 from typing import Any, Optional
@@ -86,6 +85,32 @@ def from_rpm_meta(cls, rpm_meta: dict[str, Any]) -> Self:
             epoch=rpm_meta.get("epoch", 0),
         )
 
+    @classmethod
+    def from_nevra(
+        cls, nevra: str
+    ) -> Self:  # TODO: move this to `tmt.package_managers.PackageManager.list_packages`
+        """
+        Version constructed from a NEVRA string as returned by ``dnf repoquery``.
+
+        Example usage:
+
+        .. code-block:: python
+
+            version_info = RpmVersion.from_nevra("curl-0:8.11.1-7.fc42.x86_64")
+            version_info = RpmVersion.from_nevra("curl-8.11.1-7.fc42.x86_64")
+        """
+        nvr_epoch, sep, arch = nevra.rpartition('.')
+        if not sep:
+            raise ValueError(f"Cannot parse arch from NEVRA '{nevra}'.")
+        parts = nvr_epoch.rsplit('-', 2)
+        if len(parts) != 3:
+            raise ValueError(f"Cannot parse NVR from NEVRA '{nevra}'.")
+        name, ev, release = parts
+        epoch_str, sep, version = ev.partition(':')
+        epoch = int(epoch_str) if sep else 0
+        version = version if sep else epoch_str
+        return cls(name=name, version=version, release=release, arch=arch, epoch=epoch)
+
     @classmethod
     def from_filename(cls, filename: str) -> Self:
         """
@@ -177,6 +202,7 @@ def __init__(self, raw_id: str, repository_priority: int, logger: tmt.log.Logger
         self.sanitized_id = tmt.utils.sanitize_name(raw_id, allow_slash=False)
 
         self.id = self._extract_provider_id(raw_id)
+        self._artifacts: list[ArtifactInfo] = []
 
     @classmethod
     @abstractmethod
@@ -191,16 +217,12 @@ def _extract_provider_id(cls, raw_id: str) -> ArtifactProviderId:
 
         raise NotImplementedError
 
-    @cached_property
+    @property
     @abstractmethod
     def artifacts(self) -> Sequence[ArtifactInfo]:
         """
         Collect all artifacts available from this provider.
 
-        The method is left for derived classes to implement with respect
-        to the actual artifact provider they implement. The list of
-        artifacts will be cached, and is treated as read-only.
-
         :returns: a list of provided artifacts.
         """
 
@@ -300,6 +322,39 @@ def get_repositories(self) -> list['Repository']:
         """
         return []
 
+    def enumerate_artifacts(
+        self, guest: Guest
+    ) -> None:  # TODO: refactor this once the NEVRA parsing is centralized.
+        """
+        Enumerate artifacts available from this provider after its repositories
+        have been installed on the guest.
+        """
+        for repository in self.get_repositories():
+            try:
+                nevras = guest.package_manager.list_packages(repository)
+            except Exception:
+                self.logger.warning(
+                    f"Failed to enumerate packages from repository '{repository.name}'."
+                )
+                continue
+            for nevra in nevras:
+                nevra = nevra.strip()
+                if not nevra:
+                    continue
+                try:
+                    self._artifacts.append(
+                        ArtifactInfo(
+                            version=RpmVersion.from_nevra(nevra),
+                            provider=self,
+                            location=repository.name,
+                        )
+                    )
+                except ValueError:
+                    self.logger.warning(f"Could not parse NEVRA '{nevra}'.")
+            self.logger.debug(
+                f"Enumerated {len(self._artifacts)} packages from repository '{repository.name}'."
+            )
+
     # B027: "... is an empty method in an abstract base class, but has
     # no abstract decorator" - expected, it's a default implementation
     # provided for subclasses. It is acceptable to do nothing.

tmt/steps/prepare/artifact/providers/copr_repository.py

@@ -4,7 +4,6 @@
 
 import re
 from collections.abc import Sequence
-from functools import cached_property
 from re import Pattern
 from typing import Optional
 
@@ -91,11 +90,9 @@ def _extract_provider_id(cls, raw_id: str) -> ArtifactProviderId:
 
         return value
 
-    @cached_property
+    @property
     def artifacts(self) -> Sequence[ArtifactInfo]:
-        # Copr repository provider does not enumerate individual artifacts.
-        # The repository is enabled and packages are available through the package manager.
-        return []
+        return self._artifacts
 
     def _download_artifact(self, artifact: ArtifactInfo, guest: Guest, destination: Path) -> None:
         """This provider only enables repositories; it does not download individual RPMs."""

tmt/steps/prepare/artifact/providers/repository.py

@@ -3,7 +3,6 @@
 """
 
 from collections.abc import Sequence
-from functools import cached_property
 from re import Pattern
 from typing import Optional
 from urllib.parse import urlparse
@@ -57,12 +56,9 @@ def _extract_provider_id(cls, raw_id: str) -> ArtifactProviderId:
             raise ValueError("Missing repository URL.")
         return value
 
-    @cached_property
+    @property
     def artifacts(self) -> Sequence[ArtifactInfo]:
-        # Repository provider does not enumerate individual artifacts.
-        # The repository is installed and packages are available through the package manager.
-        # There is no need to download individual artifact files.
-        return []
+        return self._artifacts
 
     def _download_artifact(self, artifact: ArtifactInfo, guest: Guest, destination: Path) -> None:
         """This provider only discovers repos; it does not download individual RPMs."""

tmt/steps/prepare/artifact/providers/repository_url.py

@@ -3,7 +3,6 @@
 """
 
 from collections.abc import Sequence
-from functools import cached_property
 from re import Pattern
 from typing import Optional
 
@@ -57,12 +56,9 @@ def _extract_provider_id(cls, raw_id: str) -> ArtifactProviderId:
             raise ValueError("Missing repository baseurl.")
         return value
 
-    @cached_property
+    @property
     def artifacts(self) -> Sequence[ArtifactInfo]:
-        # Repository provider does not enumerate individual artifacts.
-        # The repository is installed and packages are available through the package manager.
-        # There is no need to download individual artifact files.
-        return []
+        return self._artifacts
 
     def _download_artifact(
         self, artifact: ArtifactInfo, guest: Guest, destination: tmt.utils.Path