Merge pull request #31 from simple-repository/feature/file-enrichment-base-class

Improve invalid version handling

Author: pelson

simple_repository_browser/controller.py

@@ -9,9 +9,11 @@
 import fastapi
 from fastapi.responses import StreamingResponse
 from markupsafe import Markup
-from packaging.version import InvalidVersion, Version
+from packaging.version import InvalidVersion as InvalidVersionError
+from packaging.version import Version
 
 from . import errors, model, view
+from .short_release_info import InvalidVersion
 from .static_files import HashedStaticFileHandler, StaticFilesManifest
 
 
@@ -137,14 +139,14 @@ async def project(
         recache: bool = False,
     ) -> str | StreamingResponse:
         _ = page_section  # Handled in javascript.
-        _version = None
+        _version: Version | InvalidVersion | None = None
         if version:
             try:
                 _version = Version(version)
-            except InvalidVersion:
-                raise errors.RequestError(
-                    status_code=404, detail=f"Invalid version {version}."
-                )
+            except InvalidVersionError:
+                # Version string doesn't conform to PEP 440.
+                # Try to find it as an InvalidVersion in the releases.
+                _version = InvalidVersion(version)
 
         t = asyncio.create_task(
             self.model.project_page(project_name, _version, recache)

simple_repository_browser/crawler.py

@@ -15,7 +15,7 @@
 
 from . import fetch_projects
 from .fetch_description import PackageInfo, package_info
-from .short_release_info import ReleaseInfoModel, ShortReleaseInfo
+from .short_release_info import InvalidVersion, ReleaseInfoModel, ShortReleaseInfo
 
 
 class Crawler:
@@ -146,8 +146,8 @@ async def run_reindex_periodically(self) -> None:
     async def fetch_pkg_info(
         self,
         prj: model.ProjectDetail,
-        version: Version,
-        releases: dict[Version, ShortReleaseInfo],
+        version: Version | InvalidVersion,
+        releases: dict[Version | InvalidVersion, ShortReleaseInfo],
         force_recache: bool,
     ) -> tuple[model.File, PackageInfo]:
         key = ("pkg-info", prj.name, str(version))

simple_repository_browser/model.py

@@ -14,7 +14,7 @@
 
 from . import _search, compatibility_matrix, crawler, errors, fetch_projects
 from .fetch_description import PackageInfo
-from .short_release_info import ReleaseInfoModel, ShortReleaseInfo
+from .short_release_info import InvalidVersion, ReleaseInfoModel, ShortReleaseInfo
 
 
 @dataclasses.dataclass(frozen=True)
@@ -189,7 +189,7 @@ async def project_query(
     async def project_page(
         self,
         project_name: str,
-        version: Version | None,
+        version: Version | InvalidVersion | None,
         recache: bool,
     ) -> ProjectPageModel:
         canonical_name = canonicalize_name(project_name)

simple_repository_browser/short_release_info.py

@@ -1,19 +1,59 @@
 import dataclasses
 from datetime import datetime
+import functools
 import types
 import typing
 
 from packaging.utils import canonicalize_name
-from packaging.version import InvalidVersion, Version
+from packaging.version import InvalidVersion as InvalidVersionError
+from packaging.version import Version
 from simple_repository import model
 from simple_repository.packaging import extract_package_version
 
 
+@functools.total_ordering
+class InvalidVersion:
+    """Represents a version string that doesn't conform to PEP 440."""
+
+    def __init__(self, version_string: str = "unknown"):
+        self._version_string = version_string
+
+    def __str__(self):
+        return self._version_string
+
+    def __repr__(self):
+        return f"InvalidVersion({self._version_string!r})"
+
+    def __hash__(self):
+        return hash(("invalid-version", self._version_string))
+
+    def __eq__(self, other):
+        return (
+            isinstance(other, InvalidVersion)
+            and self._version_string == other._version_string
+        )
+
+    def __lt__(self, other):
+        # Sort invalid versions to the beginning (before all real versions)
+        # so they won't be selected as the latest version
+        if isinstance(other, InvalidVersion):
+            return self._version_string < other._version_string
+        return True
+
+    @property
+    def is_prerelease(self):
+        return False
+
+    @property
+    def is_devrelease(self):
+        return False
+
+
 @dataclasses.dataclass(frozen=True)
 class ShortReleaseInfo:
     # A short representation of a release. Intended to be lightweight to compute,
     # such that many ShortReleaseInfo instances can be provided to a view.
-    version: Version
+    version: Version | InvalidVersion
     files: tuple[model.File, ...]
     release_date: datetime | None
     labels: typing.Mapping[
@@ -25,30 +65,39 @@ class ReleaseInfoModel:
     @classmethod
     def release_infos(
         cls, project_detail: model.ProjectDetail
-    ) -> tuple[dict[Version, ShortReleaseInfo], Version]:
-        files_grouped_by_version: dict[Version, list[model.File]] = {}
+    ) -> tuple[
+        dict[Version | InvalidVersion, ShortReleaseInfo], Version | InvalidVersion
+    ]:
+        files_grouped_by_version: dict[Version | InvalidVersion, list[model.File]] = {}
 
         if not project_detail.files:
             raise ValueError("No files for the release")
 
         canonical_name = canonicalize_name(project_detail.name)
+        release: Version | InvalidVersion
         for file in project_detail.files:
+            version_str = None
             try:
-                release = Version(
-                    version=extract_package_version(
-                        filename=file.filename,
-                        project_name=canonical_name,
-                    ),
+                version_str = extract_package_version(
+                    filename=file.filename,
+                    project_name=canonical_name,
                 )
-            except (ValueError, InvalidVersion):
-                release = Version("0.0rc0")
+                release = Version(version=version_str)
+            except (ValueError, InvalidVersionError):
+                # Use the extracted version_str if available, otherwise the filename
+                release = InvalidVersion(version_str or file.filename)
             files_grouped_by_version.setdefault(release, []).append(file)
 
         # Ensure there is a release for each version, even if there is no files for it.
+        version: Version | InvalidVersion
         for version_str in project_detail.versions or []:
-            files_grouped_by_version.setdefault(Version(version_str), [])
+            try:
+                version = Version(version_str)
+            except (ValueError, InvalidVersionError):
+                version = InvalidVersion(version_str)
+            files_grouped_by_version.setdefault(version, [])
 
-        result: dict[Version, ShortReleaseInfo] = {}
+        result: dict[Version | InvalidVersion, ShortReleaseInfo] = {}
 
         latest_version = cls.compute_latest_version(files_grouped_by_version)
 
@@ -77,7 +126,9 @@ def release_infos(
             or []
         )
 
-        quarantined_files_by_release: dict[Version, list[Quarantinefile]] = {}
+        quarantined_files_by_release: dict[
+            Version | InvalidVersion, list[Quarantinefile]
+        ] = {}
 
         date_format = "%Y-%m-%dT%H:%M:%SZ"
         for file_info in quarantined_files:
@@ -88,12 +139,16 @@ def release_infos(
                 ),
                 "upload_time": datetime.strptime(file_info["upload_time"], date_format),
             }
-            release = Version(
-                extract_package_version(
+            version_str = None
+            try:
+                version_str = extract_package_version(
                     filename=quarantined_file["filename"],
                     project_name=canonical_name,
-                ),
-            )
+                )
+                release = Version(version_str)
+            except (ValueError, InvalidVersionError):
+                # Use the extracted version_str if available, otherwise the filename
+                release = InvalidVersion(version_str or quarantined_file["filename"])
             quarantined_files_by_release.setdefault(release, []).append(
                 quarantined_file
             )
@@ -160,8 +215,8 @@ def release_infos(
 
     @classmethod
     def compute_latest_version(
-        cls, versions: dict[Version, list[typing.Any]]
-    ) -> Version:
+        cls, versions: dict[Version | InvalidVersion, list[typing.Any]]
+    ) -> Version | InvalidVersion:
         # Use the pip logic to determine the latest release. First, pick the greatest non-dev version,
         # and if nothing, fall back to the greatest dev version. If no release is available return None.
         sorted_versions = sorted(

simple_repository_browser/templates/base/project.html

@@ -225,6 +225,13 @@ <h1> {{ project.name }} {{ this_release.version }}</h1>
                        </button>
                     </span>
                     {% endif %}
+                    {% if release_info.version.__class__.__name__ == 'InvalidVersion' %}
+                    <span style="float: right; text-decoration: none; font-size: smaller; color: gray;">
+                       <button class="btn btn-danger position-relative me-2 mb-1 btn-sm active" data-bs-toggle="tooltip" data-bs-placement="right" title="Version string does not conform to PEP 440">
+                           Invalid version
+                       </button>
+                    </span>
+                    {% endif %}
                   </div>
                 </div>
             {% if 'quarantined' not in release_info.labels %}