From 8ddd9b07a51b0a5864d3b0381dcef9a583fd89f8 Mon Sep 17 00:00:00 2001
From: Taneli Hukkinen <3275109+hukkin@users.noreply.github.com>
Date: Thu, 24 Oct 2024 14:54:07 +0300
Subject: [PATCH] Improve: Use distribution package names in version strings
 (#462)

---
 src/mdformat/_cli.py | 54 ++++++++++++++++++++++++++++----------------
 tests/test_cli.py    | 14 +++++++++++-
 2 files changed, 48 insertions(+), 20 deletions(-)

diff --git a/src/mdformat/_cli.py b/src/mdformat/_cli.py
index a4f3aca..0230ad8 100644
--- a/src/mdformat/_cli.py
+++ b/src/mdformat/_cli.py
@@ -3,6 +3,7 @@
 import argparse
 from collections.abc import Callable, Generator, Iterable, Mapping, Sequence
 import contextlib
+import inspect
 import itertools
 import logging
 import os.path
@@ -303,30 +304,45 @@ def log_handler_applied(
         logger.removeHandler(handler)
 
 
-def get_package_name(obj: object) -> str:
-    # Packages and modules should have `__package__`
-    if hasattr(obj, "__package__"):
-        package_name = obj.__package__
-    else:  # class or function
-        module_name = obj.__module__
-        package_name = module_name.split(".", maxsplit=1)[0]
-    return package_name
+def get_package_name(obj: object) -> str | None:
+    """Return top level module name, or None if not found."""
+    module = inspect.getmodule(obj)
+    return module.__name__.split(".", maxsplit=1)[0] if module else None
 
 
 def get_plugin_versions(
     parser_extensions: Mapping[str, mdformat.plugins.ParserExtensionInterface],
     codeformatters: Mapping[str, Callable[[str, str], str]],
-) -> dict[str, str]:
-    versions = {}
+) -> list[tuple[str, str]]:
+    """Return a list of (plugin_distro, plugin_version) tuples.
+
+    If many plugins come from the same distribution package, only return
+    the version of that distribution once. If we have no reliable way to
+    one-to-one map a plugin to a distribution package, use the top level
+    module name and set version to "unknown". If we don't even know the
+    top level module name, return the tuple ("unknown", "unknown").
+    """
+    problematic_versions = []
+    # Use a dict for successful version lookups so that if more than one plugin
+    # originates from the same distribution, it only shows up once in a version
+    # string.
+    success_versions = {}
+    import_package_to_distro = importlib_metadata.packages_distributions()
     for iface in itertools.chain(parser_extensions.values(), codeformatters.values()):
-        package_name = get_package_name(iface)
-        try:
-            package_version = importlib_metadata.version(package_name)
-        except importlib_metadata.PackageNotFoundError:
-            # In test scenarios the package may not exist
-            package_version = "unknown"
-        versions[package_name] = package_version
-    return versions
+        import_package = get_package_name(iface)
+        if import_package is None:
+            problematic_versions.append(("unknown", "unknown"))
+            continue
+        distro_list = import_package_to_distro.get(import_package)
+        if (
+            not distro_list  # No distribution package found
+            or len(distro_list) > 1  # Don't make any guesses with namespace packages
+        ):
+            problematic_versions.append((import_package, "unknown"))
+            continue
+        distro_name = distro_list[0]
+        success_versions[distro_name] = importlib_metadata.version(distro_name)
+    return [(k, v) for k, v in success_versions.items()] + problematic_versions
 
 
 def get_plugin_versions_str(
@@ -334,4 +350,4 @@ def get_plugin_versions_str(
     codeformatters: Mapping[str, Callable[[str, str], str]],
 ) -> str:
     plugin_versions = get_plugin_versions(parser_extensions, codeformatters)
-    return ", ".join(f"{name}: {version}" for name, version in plugin_versions.items())
+    return ", ".join(f"{name}: {version}" for name, version in plugin_versions)
diff --git a/tests/test_cli.py b/tests/test_cli.py
index 2bdf668..4df9140 100644
--- a/tests/test_cli.py
+++ b/tests/test_cli.py
@@ -6,7 +6,7 @@
 import pytest
 
 import mdformat
-from mdformat._cli import get_package_name, run, wrap_paragraphs
+from mdformat._cli import get_package_name, get_plugin_versions, run, wrap_paragraphs
 from mdformat.plugins import CODEFORMATTERS
 
 UNFORMATTED_MARKDOWN = "\n\n# A header\n\n"
@@ -350,6 +350,18 @@ def test_get_package_name():
     assert get_package_name(mdformat) == "mdformat"
 
 
+def test_get_plugin_versions():
+    # Pretend that "pytest" and "unittest.mock.patch" are plugins
+    versions = get_plugin_versions({"p1": pytest}, {"f1": patch})  # type: ignore[dict-item] # noqa: E501
+    assert versions[0][0] == "pytest"
+    assert versions[0][1] != "unknown"
+    assert versions[1] == ("unittest", "unknown")
+
+    with patch("mdformat._cli.inspect.getmodule", return_value=None):
+        versions = get_plugin_versions({"p1": pytest}, {})  # type: ignore[dict-item]
+    assert versions[0] == ("unknown", "unknown")
+
+
 def test_no_timestamp_modify(tmp_path):
     file_path = tmp_path / "test.md"