first steps to better scanning of repos

tests/test_content_indexing.py

@@ -2,10 +2,12 @@ from repo_registry.content_indexing.extractor import ContentExtractor
 from repo_registry.core.models import ObservedFact
 
 
-def fact(id, kind, name, path="", line=None):
+def fact(id, kind, name, path="", line=None, source_role=""):
     metadata = {}
     if line is not None:
         metadata["line"] = line
+    if source_role:
+        metadata["source_role"] = source_role
     return ObservedFact(
         id=id,
         repository_id=1,
@@ -82,3 +84,20 @@ def test_content_extractor_chunks_provider_related_config(tmp_path):
     assert len(chunks) == 1
     assert chunks[0].path == ".env.example"
     assert "OPENROUTER_API_KEY" in chunks[0].text
+
+
+def test_content_extractor_preserves_source_role_metadata(tmp_path):
+    repo = tmp_path / "repo"
+    repo.mkdir()
+    (repo / "SCOPE.md").write_text("# SCOPE\n\nProvides OIDC.\n", encoding="utf-8")
+
+    chunks = ContentExtractor().extract(
+        repo,
+        [
+            fact(1, "scope", "SCOPE", "SCOPE.md", source_role="scope_summary"),
+        ],
+    )
+
+    assert len(chunks) == 1
+    assert chunks[0].kind == "scope"
+    assert chunks[0].metadata["source_role"] == "scope_summary"