chore(codegen): stop on spec features it can't map

CONTRIBUTING.md

@@ -73,6 +73,11 @@ in `tools/codegen/`, or a spec, run `make codegen`, and commit the result.
 `make lint` fails on stale output. [docs/research.md](docs/research.md)
 explains how to refresh the specs.
 
+The script stops with an error on spec features it can't map yet, such as
+`oneOf`, PATCH, header parameters, and inline response schemas, rather than
+dropping them. `tools/codegen/test_codegen.py` covers each of these checks, and
+`make lint` runs it.
+
 ## Code style
 
 - Public functions return `kalshi::Result<T>` (`std::expected<T, Error>`) and

Makefile

@@ -67,6 +67,7 @@ lint:
 	$(CPP_SOURCES) | xargs -0 $(CLANG_FORMAT) --dry-run --Werror
 	$(PYTHON) tools/cpp_auto_audit.py
 	CLANG_FORMAT="$(CLANG_FORMAT)" $(PYTHON) tools/codegen/generate.py --check
+	$(PYTHON) -m unittest discover -q -s tools/codegen
 
 lint-docs:
 	markdownlint-cli2

docs/research.md

@@ -29,6 +29,8 @@ shasum -a 256 spec/*.yaml
 make codegen test
 ```
 
-Review the diff of the generated files, update the table above, and note
-user-visible changes in `CHANGELOG.md`. Kalshi's changelog at
+If `make codegen` stops on a spec feature it can't map, extend
+`tools/codegen/` to support it before committing the new spec. Review the diff
+of the generated files, update the table above, and note user-visible changes
+in `CHANGELOG.md`. Kalshi's changelog at
 <https://docs.kalshi.com/changelog> explains most changes.

tools/codegen/common.py

@@ -16,15 +16,23 @@
 ACRONYMS = {"RFQs": "Rfqs", "RFQ": "Rfq", "FCM": "Fcm", "ID": "Id", "API": "Api", "MVE": "Mve"}
 
 CPP_RESERVED = {
-    "and", "auto", "bool", "break", "case", "catch", "char", "class", "const", "continue",
-    "default", "delete", "do", "double", "else", "enum", "explicit", "export", "extern",
-    "false", "float", "for", "friend", "goto", "if", "inline", "int", "long", "mutable",
-    "namespace", "new", "not", "operator", "or", "private", "protected", "public", "return",
-    "short", "signed", "sizeof", "static", "struct", "switch", "template", "this", "throw",
-    "true", "try", "typedef", "typename", "union", "unsigned", "using", "virtual", "void",
-    "volatile", "while", "xor",
+    "alignas", "alignof", "and", "and_eq", "asm", "auto", "bitand", "bitor", "bool", "break",
+    "case", "catch", "char", "char8_t", "char16_t", "char32_t", "class", "compl", "concept",
+    "const", "const_cast", "consteval", "constexpr", "constinit", "continue", "co_await",
+    "co_return", "co_yield", "decltype", "default", "delete", "do", "double", "dynamic_cast",
+    "else", "enum", "explicit", "export", "extern", "false", "float", "for", "friend", "goto",
+    "if", "inline", "int", "long", "mutable", "namespace", "new", "noexcept", "not", "not_eq",
+    "nullptr", "operator", "or", "or_eq", "private", "protected", "public", "register",
+    "reinterpret_cast", "requires", "return", "short", "signed", "sizeof", "static",
+    "static_assert", "static_cast", "struct", "switch", "template", "this", "thread_local",
+    "throw", "true", "try", "typedef", "typeid", "typename", "union", "unsigned", "using",
+    "virtual", "void", "volatile", "wchar_t", "while", "xor", "xor_eq",
 }
 
+# Schema keywords with no C++ mapping yet. Without this check they would
+# become empty structs or RawJson, and no test would notice.
+UNSUPPORTED_KEYWORDS = ("oneOf", "anyOf", "not", "discriminator")
+
 # A server value spelled "unknown" shares the Unknown enumerator every enum has.
 UNKNOWN_VALUE = "unknown"
 
@@ -99,6 +107,7 @@ class Member:
     required: bool = False
     base_kind: str = ""  # "string", "enum", "struct", "vector", ...
     element: str = ""  # struct or enum name for validation
+    read_only: bool = False
 
 
 @dataclass
@@ -108,6 +117,24 @@ class StructType:
     doc: str = ""
 
 
+def check_schema(schema: dict, where: str, max_all_of: int) -> None:
+    """Stops on schema features that would otherwise map to a wrong or empty type."""
+    for keyword in UNSUPPORTED_KEYWORDS:
+        if keyword in schema:
+            raise SystemExit(f"{where}: `{keyword}` is not supported; extend tools/codegen first")
+    if len(schema.get("allOf", [])) > max_all_of:
+        raise SystemExit(f"{where}: `allOf` with {len(schema['allOf'])} members is not supported")
+    ref = schema.get("$ref")
+    if ref is not None and not re.fullmatch(r"#/components/schemas/[^/]+", ref):
+        raise SystemExit(f"{where}: {ref} is not a component schema")
+
+
+def check_identifier(owner: str, json_name: str) -> None:
+    """Members keep their JSON names so Glaze can reflect them; each must be a C++ name."""
+    if not re.fullmatch(r"[A-Za-z_][A-Za-z0-9_]*", json_name) or json_name in CPP_RESERVED:
+        raise SystemExit(f"{owner}.{json_name} needs a rename; add glz::meta support first")
+
+
 def check_enums(enums: dict[str, EnumType]) -> None:
     for enum in enums.values():
         names = [enumerator(v) for v in enum.known]