Add codegen regression tests for determinism and README markers

Author: gerwyn-ng

tests/test_codegen.py

@@ -0,0 +1,81 @@
+"""Tests for codegen determinism and README marker replacement."""
+
+import sys
+from pathlib import Path
+
+import pytest
+
+# codegen/ is only present in the internal repo
+codegen_dir = Path(__file__).parent.parent / "codegen"
+if not codegen_dir.is_dir():
+    pytest.skip("codegen/ not available (public repo)", allow_module_level=True)
+
+sys.path.insert(0, str(codegen_dir))
+
+from schema_to_pydantic import PydanticModelGenerator  # type: ignore[import-not-found]
+
+
+@pytest.fixture
+def generator() -> PydanticModelGenerator:
+    schema_dir = codegen_dir / "schemas"
+    schema_files = sorted(schema_dir.glob("*.json"))
+    return PydanticModelGenerator([str(f) for f in schema_files])
+
+
+class TestApiReferenceDeterminism:
+    def test_identical_output_on_consecutive_calls(
+        self, generator: PydanticModelGenerator
+    ) -> None:
+        first = generator.generate_api_reference()
+        second = generator.generate_api_reference()
+        assert first == second
+
+    def test_identical_method_index_on_consecutive_calls(
+        self, generator: PydanticModelGenerator
+    ) -> None:
+        first = generator.generate_readme_method_index()
+        second = generator.generate_readme_method_index()
+        assert first == second
+
+
+class TestReadmeMarkerReplacement:
+    def test_replaces_content_between_markers(self) -> None:
+        readme = (
+            "# Title\n"
+            "<!-- METHOD_INDEX_START -->\n"
+            "old content\n"
+            "<!-- METHOD_INDEX_END -->\n"
+            "## Footer\n"
+        )
+        start_marker = "<!-- METHOD_INDEX_START -->"
+        end_marker = "<!-- METHOD_INDEX_END -->"
+        new_index = "new content\n"
+
+        before = readme[: readme.index(start_marker) + len(start_marker)]
+        after = readme[readme.index(end_marker) :]
+        result = f"{before}\n{new_index}\n{after}"
+
+        assert "old content" not in result
+        assert "new content" in result
+        assert start_marker in result
+        assert end_marker in result
+        assert "## Footer" in result
+
+    def test_preserves_content_outside_markers(self) -> None:
+        readme = (
+            "# Title\n"
+            "intro text\n"
+            "<!-- METHOD_INDEX_START -->\n"
+            "old\n"
+            "<!-- METHOD_INDEX_END -->\n"
+            "## Footer\n"
+        )
+        start_marker = "<!-- METHOD_INDEX_START -->"
+        end_marker = "<!-- METHOD_INDEX_END -->"
+
+        before = readme[: readme.index(start_marker) + len(start_marker)]
+        after = readme[readme.index(end_marker) :]
+        result = f"{before}\nnew\n{after}"
+
+        assert result.startswith("# Title\nintro text\n")
+        assert result.endswith("## Footer\n")