Merge pull request 'schemagen: deduplicate enum constants that collide after sanitization' (#11) from weaselbot/weaseljson:weaselbot/issue-4 into main

Reviewed-on: weaselab/weaseljson#11

CMakeLists.txt

@@ -170,6 +170,8 @@ if(Python3_Interpreter_FOUND)
                    ${CMAKE_CURRENT_SOURCE_DIR}/test_python_bindings.py)
 endif()
 
+add_subdirectory(contrib/schemagen)
+
 include(CMakePushCheckState)
 include(CheckCXXCompilerFlag)
 cmake_push_check_state()

contrib/schemagen/CMakeLists.txt

@@ -0,0 +1,35 @@
+# Tests for contrib/schemagen
+
+if(NOT Python3_Interpreter_FOUND)
+  return()
+endif()
+
+add_test(
+  NAME schemagen_python_tests
+  COMMAND ${Python3_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test_schemagen.py
+  WORKING_DIRECTORY ${CMAKE_SOURCE_DIR})
+
+set(SCHEMAGEN_SCRIPT ${CMAKE_CURRENT_SOURCE_DIR}/weaseljson_schemagen.py)
+set(EXAMPLE_SCHEMA ${CMAKE_CURRENT_SOURCE_DIR}/example.schema.json)
+set(GEN_H ${CMAKE_CURRENT_BINARY_DIR}/gen.h)
+
+add_custom_command(
+  OUTPUT ${GEN_H}
+  COMMAND ${Python3_EXECUTABLE} ${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 ${CMAKE_CURRENT_SOURCE_DIR}/test_gen.cpp)
+target_include_directories(schemagen_example
+                           PRIVATE include ${CMAKE_CURRENT_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_CURRENT_BINARY_DIR})

contrib/schemagen/test_schemagen.py

@@ -0,0 +1,53 @@
+#!/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_deduplicate(self):
+        schema = {
+            "type": "object",
+            "properties": {"role": {"enum": ["foo-bar", "foo_bar"]}},
+        }
+        rc, stdout, stderr = self.run_schemagen(schema)
+        self.assertEqual(rc, 0, msg=stderr)
+        self.assertIn("enum class Role : int { foo_bar, foo_bar_1 };", stdout)
+        self.assertIn(
+            'static constexpr const char *Role_names[] = { "foo-bar", "foo_bar" };',
+            stdout,
+        )
+
+    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()

contrib/schemagen/weaseljson_schemagen.py

@@ -86,6 +86,34 @@ def sanitize(name, fallback="x"):
     return s
 
 
+def unique_enum_identifiers(values):
+    """Return a list of sanitized C++ identifiers, one per input value.
+
+    Distinct JSON enum values may sanitize to the same C++ token (e.g.
+    "foo-bar" and "foo_bar" both become "foo_bar").  This helper appends a
+    numeric suffix to later collisions so the generated `enum class` stays
+    valid while preserving the original order and therefore the index-to-value
+    mapping used at parse time.
+    """
+    used = set()
+    out = []
+    for v in values:
+        base = sanitize(v)
+        if base not in used:
+            used.add(base)
+            out.append(base)
+            continue
+        n = 1
+        while True:
+            candidate = f"{base}_{n}"
+            if candidate not in used:
+                used.add(candidate)
+                out.append(candidate)
+                break
+            n += 1
+    return out
+
+
 def camel(name):
     parts = [p for p in name.replace("-", " ").replace("_", " ").split(" ") if p]
     if not parts:
@@ -508,7 +536,7 @@ namespace {ns} {{"""
             return ""
         out = []
         for e in self.b.enums.values():
-            vals = ", ".join(sanitize(v) for v in e.values)
+            vals = ", ".join(unique_enum_identifiers(e.values))
             out.append(f"enum class {e.name} : int {{ {vals} }};")
         return "\n".join(out) + "\n"