Fix submodule_search_locations with module_dir setting

test/BUILD

@@ -185,6 +185,17 @@ python_test(
     srcs = ["interpreter_not_included_test.py"],
 )
 
+# Regression test for ModuleDirImport not setting submodule_search_locations on the specs it
+# creates for module_dir packages, which breaks importlib.resources.
+python_test(
+    name = "module_dir_import_resources_test",
+    srcs = ["module_dir_import_resources_test.py"],
+    deps = [
+        "//third_party/python:pygments",
+    ],
+    labels = ["manual"],    
+)
+
 python_test(
     name = "module_dir_import_test",
     srcs = ["module_dir_import_test.py"],
@@ -199,7 +210,7 @@ python_test(
     deps = [
         "//third_party/python:confluent-kafka",
     ],
-    labels = ["manual"],
+    labels = ["manual"] if CONFIG.OS == "freebsd" else [],
 )
 
 plugin_e2e_test(

test/module_dir_import_resources_test.py

@@ -0,0 +1,26 @@
+import importlib.resources
+import unittest
+
+
+class ModuleDirImportResourcesTest(unittest.TestCase):
+    """Tests that importlib.resources works for a package imported via its Please
+    module_dir alias (e.g. `third_party.python.pygments`), not just its plain top-level
+    name (`pygments`).
+
+    ModuleDirImport aliases third_party.python.X to the same module object as X, but the
+    ModuleSpec it hands back for the aliased name doesn't carry over submodule_search_locations
+    from the real module's spec. Because both names share the same module object,
+    importlib._bootstrap unconditionally overwrites module.__spec__ with that incomplete spec
+    as soon as the aliased name is imported, which then breaks importlib.resources for both
+    names (see https://docs.python.org/3/library/importlib.html#importlib.machinery.ModuleSpec).
+    """
+
+    def test_read_resource_text_file_via_module_dir_alias(self):
+        init = importlib.resources.files("third_party.python.pygments") / "__init__.py"
+        self.assertIn("__version__ = '2.19.2'", init.read_text())
+
+    def test_read_resource_text_file_via_plain_name_after_alias_import(self):
+        import third_party.python.pygments  # noqa: F401  (import triggers the aliasing)
+
+        init = importlib.resources.files("pygments") / "__init__.py"
+        self.assertIn("__version__ = '2.19.2'", init.read_text())

third_party/python/BUILD

@@ -505,7 +505,7 @@ python_wheel(
 pip_library(
     name = "confluent-kafka",
     licences = ["Apache-2.0"],
-    version = "2.6.1",
+    version = "2.15.1",
     test_only = True,
     zip_safe = False,
 )

tools/please_pex/ChangeLog

@@ -1,3 +1,7 @@
+Version 3.0.3
+-------------
+    * Fix `importlib.resources` for some usages relating to packages in module_dir (#324)
+
 Version 3.0.2
 -------------
     * Search `PATH` when .pex is invoked at a relative path with no `/` (#299)

tools/please_pex/VERSION

@@ -1 +1 @@
-3.0.2
+3.0.3

tools/please_pex/pex/plz.py

@@ -123,6 +123,8 @@ class ModuleDirImport(MetaPathFinder):
     def __init__(self, module_dir):
         self.prefix = module_dir.replace("/", ".") + "."
         self._distributions = self._find_all_distributions(module_dir)
+        # Real (unaliased) specs of modules we've imported, keyed by their aliased name.
+        self._real_specs = {}
 
     def _find_all_distributions(self, module_dir):
         pex_file = sys.argv[0]
@@ -149,12 +151,26 @@ def find_spec(self, name, path, target=None):
     def create_module(self, spec):
         """Actually load a module that we said we'd handle in find_module."""
         module = import_module(spec.name.removeprefix(self.prefix))
+        # Carry over submodule_search_locations from the real module's spec, otherwise
+        # importlib.resources (which relies on it to find a package's files) breaks once
+        # the module's __spec__ gets overwritten with this one.
+        real_spec = module.__spec__
+        spec.submodule_search_locations = real_spec.submodule_search_locations
+        self._real_specs[spec.name] = real_spec
         sys.modules[spec.name] = module
         return module
 
     def exec_module(self, mod):
         """Nothing to do, create_module already did the work."""
 
+    def get_resource_reader(self, fullname):
+        """Delegates to the real module's loader so importlib.resources can find its files."""
+        real_spec = self._real_specs.get(fullname)
+        if real_spec is not None:
+            reader = getattr(real_spec.loader, "get_resource_reader", None)
+            if reader is not None:
+                return reader(real_spec.name)
+
     def find_distributions(self, context):
         """Return an iterable of all Distribution instances capable of
         loading the metadata for packages for the indicated ``context``.