Add package name deduplication by stemming

pipcompilemulti/deduplicate.py

@@ -28,10 +28,39 @@ def ignored_packages(self, in_path):
         if self.env_confs is None:
             return {}
         rrefs = recursive_refs(self.env_confs, in_path)
-        return merged_packages(self.env_packages, rrefs)
+        return IgnoredPackages(merged_packages(self.env_packages, rrefs))
 
     def recursive_refs(self, in_path):
         """Return recursive list of environment names referenced by in_path."""
         if self.env_confs is None:
             return {}
         return recursive_refs(self.env_confs, in_path)
+
+
+class IgnoredPackages:
+    """Mapping from package name to version.
+
+    Handles name normalization for packages like:
+    zope.interface, zope-interface, zope_interface.
+    """
+    _DELIMITERS = ('_', '-', '.')
+
+    def __init__(self, package_versions):
+        self._package_versions = package_versions
+        self._stems = {
+            self._make_stem(name): name
+            for name in self._package_versions
+        }
+
+    def __getitem__(self, key):
+        canonical_key = self._stems[self._make_stem(key)]
+        return self._package_versions[canonical_key]
+
+    def __contains__(self, key):
+        return self._make_stem(key) in self._stems
+
+    @classmethod
+    def _make_stem(cls, name):
+        for delim in cls._DELIMITERS:
+            name = name.replace(delim, '')
+        return name.lower()

tests/test_deduplicate.py

@@ -0,0 +1,15 @@
+"""Package name deduplication tests"""
+from pipcompilemulti.deduplicate import PackageDeduplicator
+
+
+def test_package_deduplicator_handles_delimiters_normalization():
+    """Tests minor package name variations are handled."""
+    package_deduplicator = PackageDeduplicator()
+    package_deduplicator.on_discover([
+        {'in_path': 'a', 'refs': ['b']},
+        {'in_path': 'b', 'refs': []}
+    ])
+    package_deduplicator.register_packages_for_env('b', {'pkg.name': '1.0'})
+    ignored_packages = package_deduplicator.ignored_packages('a')
+    assert 'pkg-name' in ignored_packages
+    assert ignored_packages['Pkg_Name'] == '1.0'