fix: remove deprecated pkg_resources usage ahead of planned removal in Setuptools ≥81 (#1058)

Replaces all uses of the deprecated pkg_resources API in `utils.py`,
`docker_utils.py`, and `autobump.py` with `importlib.resources` and
`packaging.version` equivalents.

Adds a smoke test for `validate_config` to ensure schema validation
works without pkg_resources.

Closes #1056

---------

Co-authored-by: Joshua Zhuang <71105179+mencian@users.noreply.github.com>
Co-authored-by: Alicia A. Evans <108547992+aliciaaevans@users.noreply.github.com>

Author: 3 people

bioconda_utils/autobump.py

@@ -64,7 +64,7 @@
 from conda.exports import MatchSpec, VersionOrder
 import conda.exceptions
 
-from pkg_resources import parse_version
+from packaging.version import parse as _pep440_parse, InvalidVersion
 
 from . import __version__
 from . import utils
@@ -78,11 +78,28 @@
 from .githandler import GitHandler
 from .githubhandler import GitHubHandler
 
-# pkg_resources.parse_version returns a Version or LegacyVersion object
-# as defined in packaging.version. Since it's bundling it's own copy of
-# packaging, the class it returns is not the same as the one we can import.
-# So we cheat by having it create a LegacyVersion delibarately.
-LegacyVersion = parse_version("1.1").__class__  # pylint: disable=invalid-name
+def _parse_or_legacy(s: str):
+    """
+    Attempt to parse a version string as a PEP 440-compliant Version.
+
+    Returns
+    -------
+    tuple
+        (parsed_version, is_legacy)
+        - parsed_version : packaging.version.Version instance if parseable,
+          otherwise the original string for legacy (non-PEP 440) versions.
+        - is_legacy : bool indicating whether the string is a legacy version.
+
+    Notes
+    -----
+    This replaces the deprecated ``pkg_resources.parse_version`` + LegacyVersion
+    check. ``is_legacy`` allows downstream logic to gate acceptance of
+    unparseable versions.
+    """
+    try:
+        return _pep440_parse(s), False   # Version, not legacy
+    except InvalidVersion:
+        return s, True                   # raw string, legacy
 
 
 logger = logging.getLogger(__name__)  # pylint: disable=invalid-name
@@ -680,21 +697,22 @@ def select_version(current: str, versions: Sequence[str]) -> str:
           defined by parse_version)
 
         - may only be "Legacy" (=strange) if current is Legacy (as
-          defined by parse_version)
+          identified by PEP 440 parsing)
         """
-        current_version = parse_version(current)
-        current_is_legacy = isinstance(current_version, LegacyVersion)
+        current_version, current_is_legacy = _parse_or_legacy(current)
         latest_vo = VersionOrder(current)
         latest = current
         for vers in versions:
             if "-" in vers:  # ignore versions with local (FIXME)
                 continue
-            vers_version = parse_version(vers)
+            vers_version, vers_is_legacy = _parse_or_legacy(vers)
             # allow prerelease only if current is prerelease
-            if vers_version.is_prerelease and not current_version.is_prerelease:
+            if (
+                getattr(vers_version, "is_prerelease", False)
+                and not getattr(current_version, "is_prerelease", False)
+            ):
                 continue
             # allow legacy only if current is legacy
-            vers_is_legacy = isinstance(vers_version, LegacyVersion)
             if vers_is_legacy and not current_is_legacy:
                 continue
             # using conda version order here as that's what will be

bioconda_utils/docker_utils.py

@@ -53,7 +53,7 @@
 import pwd
 import grp
 from textwrap import dedent
-import pkg_resources
+from importlib.resources import files, as_file
 import re
 from distutils.version import LooseVersion
 
@@ -341,10 +341,10 @@ def _build_image(self):
             if self.requirements:
                 fout.write(open(self.requirements).read())
             else:
-                fout.write(open(pkg_resources.resource_filename(
-                    'bioconda_utils',
-                    'bioconda_utils-requirements.txt')
-                ).read())
+                # pkg_resources (deprecated) is replaced with importlib.resources
+                with as_file(files('bioconda_utils') / 'bioconda_utils-requirements.txt') as req_path:
+                    with open(req_path, 'r', encoding='utf-8') as fh:
+                        fout.write(fh.read())
 
         proxies = "\n".join("ENV {} {}".format(k, v)
                             for k, v in self._find_proxy_settings())

bioconda_utils/lint/check_build_help.py

@@ -66,9 +66,9 @@ def check_deps(self, deps):
 class uses_setuptools(LintCheck):
     """The recipe uses setuptools in run depends
 
-    Most Python packages only need setuptools during installation.
-    Check if the package really needs setuptools (e.g. because it uses
-    pkg_resources or setuptools console scripts).
+    Most Python packages only need setuptools during installation. Check if the
+    package really needs setuptools (e.g. because it uses pkg_resources
+    (deprecated; slated for removal) or setuptools console scripts).
 
     """
 

bioconda_utils/utils.py

@@ -40,7 +40,7 @@
 
 from github import Github
 
-import pkg_resources
+from importlib.resources import files, as_file
 import pandas as pd
 import tqdm as _tqdm
 import aiohttp
@@ -1170,10 +1170,13 @@ def validate_config(config):
     """
     if not isinstance(config, dict):
         config = yaml.safe_load(open(config))
-    fn = pkg_resources.resource_filename(
-        'bioconda_utils', 'config.schema.yaml'
-    )
-    schema = yaml.safe_load(open(fn))
+
+    # Load packaged schema without pkg_resources (deprecated)
+    # files('bioconda_utils') returns a Traversable to the package contents
+    with as_file(files('bioconda_utils') / 'config.schema.yaml') as schema_path:
+        with open(schema_path, 'r', encoding='utf-8') as fh:
+            schema = yaml.safe_load(fh)
+
     validate(config, schema)
 
 
@@ -1650,7 +1653,7 @@ def extract_stable_version(version):
 
 
 def yaml_remove_invalid_chars(text: str, valid_chars_re=re.compile(r"[^ \t\n\w\d:\{\}\[\]\(\);&|\$§\"'\?\!%#\\~*\.,-\^°]+")) -> str:
-    """Remove chars that are invalid in yaml literal strings. 
+    """Remove chars that are invalid in yaml literal strings.
 
     E.g. we do not want them to contain carriage return chars or delete chars.
     """

test/test_utils.py

@@ -22,6 +22,7 @@
 from bioconda_utils import docker_utils
 from bioconda_utils import build
 from bioconda_utils import upload
+from bioconda_utils.utils import validate_config
 from helpers import ensure_missing, Recipes
 
 
@@ -1342,3 +1343,16 @@ def test_pkg_test_conda_package_format(
             assert pkg_file.endswith({"1": ".tar.bz2", "2": ".conda"}[pkg_format])
             assert os.path.exists(pkg_file)
             ensure_missing(pkg_file)
+
+
+def test_validate_config_smoke():
+    """Minimal config that satisfies schema types should validate without error."""
+    cfg = {
+        "channels": ["conda-forge", "bioconda"],
+        "docker_image": "quay.io/bioconda/bioconda-utils",
+        "upload_channel": "bioconda",
+        "conda_build_version": "3.28.4",
+        "blacklists": [],
+    }
+    # Should not raise
+    validate_config(cfg)