From 800e4544fca1d9f2978a0059db240c54329a130d Mon Sep 17 00:00:00 2001
From: Weaselbot <weaselbot@weaselab.dev>
Date: Thu, 18 Jun 2026 10:22:19 -0400
Subject: [PATCH] schemagen: reject enum values that collide after C++
 sanitization

When two distinct JSON enum values sanitized to the same C++ identifier,
weaseljson_schemagen.py produced an enum class with duplicate constants,
generating invalid C++. Detect this at generation time and emit a clear
GenError instead.

Also add contrib/schemagen/test_schemagen.py with unit tests for the
collision case and the normal generation path, and wire both the Python
tests and the existing example.schema.json/test_gen.cpp example into
ctest via CMakeLists.txt.

Closes #4
---
 CMakeLists.txt                            | 32 ++++++++++++++
 contrib/schemagen/test_schemagen.py       | 51 +++++++++++++++++++++++
 contrib/schemagen/weaseljson_schemagen.py |  9 ++++
 3 files changed, 92 insertions(+)
 create mode 100644 contrib/schemagen/test_schemagen.py

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 64a6ccd..5f04b6b 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -163,6 +163,38 @@ target_link_libraries(mytest PRIVATE ${PROJECT_NAME} doctest nanobench simdjson)
 target_compile_options(mytest PRIVATE ${TEST_FLAGS})
 doctest_discover_tests(mytest WORKING_DIRECTORY ${CMAKE_SOURCE_DIR})
 
+# schemagen tests
+add_test(
+  NAME schemagen_python_tests
+  COMMAND python3 contrib/schemagen/test_schemagen.py
+  WORKING_DIRECTORY ${CMAKE_SOURCE_DIR})
+
+set(SCHEMAGEN_DIR ${CMAKE_SOURCE_DIR}/contrib/schemagen)
+set(SCHEMAGEN_SCRIPT ${SCHEMAGEN_DIR}/weaseljson_schemagen.py)
+set(EXAMPLE_SCHEMA ${SCHEMAGEN_DIR}/example.schema.json)
+set(GEN_H ${CMAKE_BINARY_DIR}/gen.h)
+
+add_custom_command(
+  OUTPUT ${GEN_H}
+  COMMAND python3 ${SCHEMAGEN_SCRIPT} ${EXAMPLE_SCHEMA} -o ${GEN_H} --namespace
+          weasel_schema
+  DEPENDS ${SCHEMAGEN_SCRIPT} ${EXAMPLE_SCHEMA}
+  COMMENT "Generating gen.h from example.schema.json")
+
+add_custom_target(schemagen_gen_h DEPENDS ${GEN_H})
+
+add_executable(schemagen_example ${SCHEMAGEN_DIR}/test_gen.cpp)
+target_include_directories(schemagen_example PRIVATE include
+                                                     ${CMAKE_BINARY_DIR})
+target_link_libraries(schemagen_example PRIVATE ${PROJECT_NAME})
+target_compile_options(schemagen_example PRIVATE -Wno-switch-enum)
+add_dependencies(schemagen_example schemagen_gen_h)
+
+add_test(
+  NAME schemagen_example
+  COMMAND schemagen_example
+  WORKING_DIRECTORY ${CMAKE_BINARY_DIR})
+
 include(CMakePushCheckState)
 include(CheckCXXCompilerFlag)
 cmake_push_check_state()
diff --git a/contrib/schemagen/test_schemagen.py b/contrib/schemagen/test_schemagen.py
new file mode 100644
index 0000000..c5dba60
--- /dev/null
+++ b/contrib/schemagen/test_schemagen.py
@@ -0,0 +1,51 @@
+#!/usr/bin/env python3
+"""Tests for weaseljson_schemagen.py."""
+
+import json
+import os
+import subprocess
+import sys
+import tempfile
+import unittest
+
+SCRIPT = os.path.join(os.path.dirname(__file__), "weaseljson_schemagen.py")
+
+
+class SchemagenEnumTest(unittest.TestCase):
+    def run_schemagen(self, schema, args=None):
+        """Run schemagen on a schema dict. Returns (returncode, stdout, stderr)."""
+        with tempfile.NamedTemporaryFile("w", suffix=".json", delete=False) as fp:
+            json.dump(schema, fp)
+            schema_path = fp.name
+        try:
+            cmd = [sys.executable, SCRIPT, schema_path]
+            if args:
+                cmd.extend(args)
+            result = subprocess.run(cmd, capture_output=True, text=True, check=False)
+            return result.returncode, result.stdout, result.stderr
+        finally:
+            os.unlink(schema_path)
+
+    def test_colliding_enum_values_raise_error(self):
+        schema = {
+            "type": "object",
+            "properties": {"role": {"enum": ["foo-bar", "foo_bar"]}},
+        }
+        rc, stdout, stderr = self.run_schemagen(schema)
+        self.assertNotEqual(rc, 0)
+        self.assertIn("foo-bar", stderr)
+        self.assertIn("foo_bar", stderr)
+        self.assertIn("C++ identifier", stderr)
+
+    def test_distinct_enum_values_generate(self):
+        schema = {
+            "type": "object",
+            "properties": {"role": {"enum": ["admin", "user", "guest"]}},
+        }
+        rc, stdout, stderr = self.run_schemagen(schema)
+        self.assertEqual(rc, 0, msg=stderr)
+        self.assertIn("enum class Role : int { admin, user, guest };", stdout)
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/contrib/schemagen/weaseljson_schemagen.py b/contrib/schemagen/weaseljson_schemagen.py
index e7a1a6c..289a254 100644
--- a/contrib/schemagen/weaseljson_schemagen.py
+++ b/contrib/schemagen/weaseljson_schemagen.py
@@ -508,6 +508,15 @@ namespace {ns} {{"""
             return ""
         out = []
         for e in self.b.enums.values():
+            seen = {}
+            for v in e.values:
+                ident = sanitize(v)
+                if ident in seen:
+                    raise GenError(
+                        f"enum {e.name!r} values {seen[ident]!r} and {v!r} "
+                        f"both sanitize to C++ identifier {ident!r}"
+                    )
+                seen[ident] = v
             vals = ", ".join(sanitize(v) for v in e.values)
             out.append(f"enum class {e.name} : int {{ {vals} }};")
         return "\n".join(out) + "\n"