add ignore_system_package option and canonical names

ruff.toml

@@ -7,6 +7,9 @@ select = [
 ignore = ["COM812", "D203", "D213", "EM101", "EM102", "TRY003", "RUF012"]
 per-file-ignores = {"tests/*.py" = ["D100", "D101", "D102", "D103", "D104", "D105", "D106", "D107", "UP", "ANN", "S101", "S603"]}
 
+[lint.mccabe]
+max-complexity = 12
+
 [format]
 quote-style = "double"
 indent-style = "space"

src/picopip.py

@@ -26,7 +26,9 @@
 log = logging.getLogger(__name__)
 
 
-def get_site_package_paths(venv_path: str) -> List[Path]:
+def get_site_package_paths(
+    venv_path: str, *, include_system_packages: bool = True
+) -> List[Path]:
     """Return all directories where packages might be installed for the given venv."""
     for version_dir in (Path(venv_path) / "lib").iterdir():
         if version_dir.name.startswith("python"):
@@ -59,20 +61,30 @@ def get_site_package_paths(venv_path: str) -> List[Path]:
             )
             continue
 
-    # Append system packages at the end, so that venv site-packages take precedence
-    for system_path in _find_system_packages(venv_path):
-        if system_path not in seen:
-            scan_paths.append(system_path)
-            seen.add(system_path)
+    if include_system_packages:
+        # Append system packages at the end, so that venv site-packages take precedence
+        for system_path in _find_system_packages(venv_path):
+            if system_path not in seen:
+                scan_paths.append(system_path)
+                seen.add(system_path)
 
     return scan_paths
 
 
-def get_packages_from_env(venv_path: str) -> List[Tuple[str, str]]:
+def get_packages_from_env(
+    venv_path: str, *, ignore_system_packages: bool = False
+) -> List[Tuple[str, str]]:
     """Return a list of (name, version) for all installed packages in the given venv."""
+
+    def _canonical_name(name: str) -> str:
+        """PEP 503 normalization plus dashes as underscores."""
+        return re.sub(r"[-_.]+", "-", name).lower().replace("-", "_")
+
     seen = set()
     packages = []
-    for path in get_site_package_paths(venv_path):
+    for path in get_site_package_paths(
+        venv_path, include_system_packages=not ignore_system_packages
+    ):
         log.debug(f"Scanning {path} for installed packages...")
         for dist_info in itertools.chain(
             path.glob("*.dist-info"), path.glob("*.egg-info")
@@ -88,7 +100,7 @@ def get_packages_from_env(venv_path: str) -> List[Tuple[str, str]]:
                         dist_info,
                     )
                     continue
-                name = raw_name.lower()
+                name = _canonical_name(raw_name)
                 if name not in seen:
                     seen.add(name)
                     packages.append((raw_name, version))

tests/test_packages_discovery.py

@@ -153,6 +153,18 @@ def test_get_packages_from_env_egg_info(fake_venv):
     assert ("legacy-pkg", "2.1.0") in pkgs
 
 
+def test_get_packages_from_env_canonical_dedup(fake_venv):
+    """Packages differing only by dashes/underscores should be deduped."""
+    venv, site = fake_venv
+    make_dist_info(site, "pure-eval", "0.2.2")
+    extra = (site / "../dupe_extra").absolute()
+    extra.mkdir(parents=True)
+    make_dist_info(extra, "pure_eval", "0.2.3")
+    (site / "dupe_extra.pth").write_text("../dupe_extra\n")
+    pkgs = get_packages_from_env(str(venv))
+    assert pkgs == [("pure-eval", "0.2.2")]
+
+
 def test_get_packages_from_env_mixed_formats(fake_venv):
     """Test that both dist-info and egg-info packages are discovered together."""
     venv, site = fake_venv
@@ -176,6 +188,26 @@ def test_get_packages_from_env_egg_info_with_pth(fake_venv):
     assert ("external-legacy", "0.9.5") in pkgs
 
 
+def test_get_packages_from_env_ignore_system_packages(tmp_path, monkeypatch):
+    """System packages should be omitted when ignore_system_packages is True."""
+    venv_dir = tmp_path / "venv"
+    site_dir = venv_dir / "lib" / "python3.11" / "site-packages"
+    site_dir.mkdir(parents=True)
+    (venv_dir / "pyvenv.cfg").write_text("include-system-site-packages = true\n")
+
+    system_site = tmp_path / "system_site"
+    system_site.mkdir()
+    make_dist_info(system_site, "system-pkg", "0.1.0")
+
+    monkeypatch.setattr("picopip.site.getsitepackages", lambda: [str(system_site)])
+
+    pkgs = get_packages_from_env(str(venv_dir))
+    assert ("system-pkg", "0.1.0") in pkgs
+
+    pkgs = get_packages_from_env(str(venv_dir), ignore_system_packages=True)
+    assert ("system-pkg", "0.1.0") not in pkgs
+
+
 def test_e2e_readme_example():
     with tempfile.TemporaryDirectory() as tmpdir:
         venv.create(tmpdir, with_pip=True)