fix(gazelle): handle purelib and platlib packages that don't set 'Root-Is-Purelib: true' (#768)

Co-authored-by: Alex Eagle <[email protected]>

Author: mattem

gazelle/modules_mapping/generator.py

@@ -17,26 +17,15 @@ def __init__(self, stderr, output_file):
     # by looking at the directory structure.
     def dig_wheel(self, whl):
         mapping = {}
-        wheel_name = get_wheel_name(whl)
         with zipfile.ZipFile(whl, "r") as zip_file:
             for path in zip_file.namelist():
                 if is_metadata(path):
-                    continue
-                ext = pathlib.Path(path).suffix
-                if ext == ".py" or ext == ".so":
-                    # Note the '/' here means that the __init__.py is not in the
-                    # root of the wheel, therefore we can index the directory
-                    # where this file is as an importable package.
-                    if path.endswith("/__init__.py"):
-                        module = path[: -len("/__init__.py")].replace("/", ".")
-                        mapping[module] = wheel_name
-                    # Always index the module file.
-                    if ext == ".so":
-                        # Also remove extra metadata that is embeded as part of
-                        # the file name as an extra extension.
-                        ext = "".join(pathlib.Path(path).suffixes)
-                    module = path[: -len(ext)].replace("/", ".")
-                    mapping[module] = wheel_name
+                    if data_has_purelib_or_platlib(path):
+                        module_for_path(path, whl, mapping)
+                    else:
+                        continue
+                else:
+                    module_for_path(path, whl, mapping)
         return mapping
 
     # run is the entrypoint for the generator.
@@ -73,6 +62,44 @@ def is_metadata(path):
     return top_level.endswith(".dist-info") or top_level.endswith(".data")
 
 
+# The .data is allowed to contain a full purelib or platlib directory
+# These get unpacked into site-packages, so require indexing too.
+# This is the same if "Root-Is-Purelib: true" is set and the files are at the root.
+# Ref: https://peps.python.org/pep-0427/#what-s-the-deal-with-purelib-vs-platlib
+def data_has_purelib_or_platlib(path):
+    maybe_lib = path.split("/")[1].lower()
+    return is_metadata(path) and (
+        maybe_lib == "purelib" or maybe_lib == "platlib"
+    )
+
+
+
+def module_for_path(path, whl, mapping):
+    ext = pathlib.Path(path).suffix
+    if ext == ".py" or ext == ".so":
+        if "purelib" in path or "platlib" in path:
+            root = "/".join(path.split("/")[2:])
+        else:
+            root = path
+
+        wheel_name = get_wheel_name(whl)
+
+        if root.endswith("/__init__.py"):
+            # Note the '/' here means that the __init__.py is not in the
+            # root of the wheel, therefore we can index the directory
+            # where this file is as an importable package.
+            module = root[: -len("/__init__.py")].replace("/", ".")
+            mapping[module] = wheel_name
+
+        # Always index the module file.
+        if ext == ".so":
+            # Also remove extra metadata that is embeded as part of
+            # the file name as an extra extension.
+            ext = ''.join(pathlib.Path(root).suffixes)
+        module = root[: -len(ext)].replace("/", ".")
+        mapping[module] = wheel_name
+
+
 if __name__ == "__main__":
     output_file = sys.argv[1]
     wheels = sys.argv[2:]