From 7935487dcfa3548d6bab497cfe3bc4649df44a94 Mon Sep 17 00:00:00 2001
From: Niels Pardon <par@zurich.ibm.com>
Date: Mon, 8 Jun 2026 11:07:14 +0200
Subject: [PATCH] fix(extensions): expose top-level metadata in YAML extension
 files

build_simple_extensions populated metadata for functions and types but
silently dropped the file-level metadata field, so it was not available
at runtime. Pass it through and add tests covering metadata exposure at
the file, function, and type levels.

Fixes #149

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---
 src/substrait/simple_extension_utils.py |  1 +
 tests/test_simple_extension_utils.py    | 51 +++++++++++++++++++++++++
 2 files changed, 52 insertions(+)
 create mode 100644 tests/test_simple_extension_utils.py

diff --git a/src/substrait/simple_extension_utils.py b/src/substrait/simple_extension_utils.py
index 7baa9aa..738bede 100644
--- a/src/substrait/simple_extension_utils.py
+++ b/src/substrait/simple_extension_utils.py
@@ -172,6 +172,7 @@ def build_type_variation(d: dict) -> se.TypeVariation:
 def build_simple_extensions(d: dict) -> se.SimpleExtensions:
     return se.SimpleExtensions(
         urn=d["urn"],
+        metadata=d.get("metadata"),
         dependencies=d.get("dependencies"),
         types=[build_type_model(t) for t in d["types"]] if "types" in d else None,
         type_variations=[build_type_variation(t) for t in d["type_variations"]]
diff --git a/tests/test_simple_extension_utils.py b/tests/test_simple_extension_utils.py
new file mode 100644
index 0000000..9a9ae0a
--- /dev/null
+++ b/tests/test_simple_extension_utils.py
@@ -0,0 +1,51 @@
+"""Tests for parsing simple extension YAML dicts into schema objects."""
+
+from substrait.simple_extension_utils import build_simple_extensions
+
+
+def test_metadata_is_exposed_at_all_levels():
+    """Metadata should be carried through from the YAML dict at every level
+    where the simple-extension schema supports it (file, function, type)."""
+    definitions = {
+        "urn": "extension:test:metadata",
+        "metadata": {"author": "me", "version": 1},
+        "types": [
+            {"name": "point", "metadata": {"kind": "geometry"}},
+        ],
+        "scalar_functions": [
+            {
+                "name": "f",
+                "impls": [{"return": "i64"}],
+                "metadata": {"category": "math"},
+            },
+        ],
+        "aggregate_functions": [
+            {
+                "name": "g",
+                "impls": [{"return": "i64"}],
+                "metadata": {"category": "stats"},
+            },
+        ],
+        "window_functions": [
+            {
+                "name": "h",
+                "impls": [{"return": "i64"}],
+                "metadata": {"category": "window"},
+            },
+        ],
+    }
+
+    extensions = build_simple_extensions(definitions)
+
+    assert extensions.metadata == {"author": "me", "version": 1}
+    assert extensions.types[0].metadata == {"kind": "geometry"}
+    assert extensions.scalar_functions[0].metadata == {"category": "math"}
+    assert extensions.aggregate_functions[0].metadata == {"category": "stats"}
+    assert extensions.window_functions[0].metadata == {"category": "window"}
+
+
+def test_metadata_defaults_to_none_when_absent():
+    """Metadata is optional and should default to None when not present."""
+    extensions = build_simple_extensions({"urn": "extension:test:no-metadata"})
+
+    assert extensions.metadata is None