feat chaotic: escape emoji as 'U' and code point

Halfdarkangel · apolukhin · halfdarkangel · robot-piglet · commit ca356ebd8468 · 2026-04-24T18:41:48.000+03:00
Tests: протестировано CI --- Pull Request resolved: #1181 Co-authored-by: antoshkka <antoshkka@userver.tech> Co-authored-by: antoshkka <antoshkka@userver.tech> Co-authored-by: antoshkka <antoshkka@userver.tech> Co-authored-by: antoshkka <antoshkka@userver.tech> Co-authored-by: halfdarkangel <userty10012@gmail.com> commit_hash:0f7bf45d8652d824ab3680e52c4fc668d2121324
diff --git a/.mapping.json b/.mapping.json
@@ -319,6 +319,7 @@
   "chaotic/integration_tests/schemas/container_format.yaml":"taxi/uservices/userver/chaotic/integration_tests/schemas/container_format.yaml",
   "chaotic/integration_tests/schemas/custom_cpp_type.yaml":"taxi/uservices/userver/chaotic/integration_tests/schemas/custom_cpp_type.yaml",
   "chaotic/integration_tests/schemas/date.yaml":"taxi/uservices/userver/chaotic/integration_tests/schemas/date.yaml",
+  "chaotic/integration_tests/schemas/emoji_enum.yaml":"taxi/uservices/userver/chaotic/integration_tests/schemas/emoji_enum.yaml",
   "chaotic/integration_tests/schemas/external_ref.yaml":"taxi/uservices/userver/chaotic/integration_tests/schemas/external_ref.yaml",
   "chaotic/integration_tests/schemas/extra_container.yaml":"taxi/uservices/userver/chaotic/integration_tests/schemas/extra_container.yaml",
   "chaotic/integration_tests/schemas/indirect.yaml":"taxi/uservices/userver/chaotic/integration_tests/schemas/indirect.yaml",
@@ -347,6 +348,7 @@
   "chaotic/integration_tests/tests/my/custom_object.cpp":"taxi/uservices/userver/chaotic/integration_tests/tests/my/custom_object.cpp",
   "chaotic/integration_tests/tests/my/custom_one_of_with_discriminator.cpp":"taxi/uservices/userver/chaotic/integration_tests/tests/my/custom_one_of_with_discriminator.cpp",
   "chaotic/integration_tests/tests/render/custom.cpp":"taxi/uservices/userver/chaotic/integration_tests/tests/render/custom.cpp",
+  "chaotic/integration_tests/tests/render/emoji_enum.cpp":"taxi/uservices/userver/chaotic/integration_tests/tests/render/emoji_enum.cpp",
   "chaotic/integration_tests/tests/render/formatter.cpp":"taxi/uservices/userver/chaotic/integration_tests/tests/render/formatter.cpp",
   "chaotic/integration_tests/tests/render/fwd.cpp":"taxi/uservices/userver/chaotic/integration_tests/tests/render/fwd.cpp",
   "chaotic/integration_tests/tests/render/logging.cpp":"taxi/uservices/userver/chaotic/integration_tests/tests/render/logging.cpp",
diff --git a/chaotic/chaotic/back/cpp/translator.py b/chaotic/chaotic/back/cpp/translator.py
@@ -484,13 +484,38 @@ def _gen_string(
             default=schema.default,
         )
 
+    @staticmethod
+    def _is_emoji(code_point: int) -> bool:
+        return (
+            # Box Drawing, Block Elements, Geometric Shapes, Miscellaneous Symbols, Dingbats
+            (0x2500 <= code_point and code_point <= 0x27BF)
+            # Miscellaneous Symbols and Pictographs, Ornamental Dingbats, Transport and Map Symbols, Alchemical Symbols
+            or (0x1F300 <= code_point and code_point <= 0x1F77F)
+        )
+
     @staticmethod
     def _normalize_name(name: str) -> str:
         if re.search(NON_NAME_SYMBOL_RE, name):
+            initially_with_leading_underscore = name.startswith('_')
+
             lang = transliterate.detect_language(name, heavy_check=True)
             if lang:
                 name = transliterate.translit(name, lang, reversed=True)
+
+            replacements = {}
+            for symbol in name:
+                code_point = ord(symbol)
+                if Generator._is_emoji(code_point):
+                    replacements[symbol] = f'_u{code_point:X}'
+
+            for symbol, replacement in replacements.items():
+                name = name.replace(symbol, replacement)
+
             name = re.sub(NON_NAME_SYMBOL_RE, '_', name)
+
+            if not initially_with_leading_underscore and name.startswith('_'):
+                name = name[1:]
+
         return name
 
     def _gen_field(
diff --git a/chaotic/integration_tests/schemas/emoji_enum.yaml b/chaotic/integration_tests/schemas/emoji_enum.yaml
@@ -0,0 +1,97 @@
+definitions:
+    EnumEmoji:
+        oneOf:
+          - $ref: '#/definitions/EnumEmojiThe'
+          - $ref: '#/definitions/EnumEmojiQuick'
+          - $ref: '#/definitions/EnumEmojiBrown'
+          - $ref: '#/definitions/EnumEmojiFox'
+          - $ref: '#/definitions/EnumEmojiJumps'
+          - $ref: '#/definitions/EnumEmojiOver'
+          - $ref: '#/definitions/EnumEmojiLazyDog'
+        discriminator:
+            propertyName: emoji
+    EnumEmojiThe:
+        type: object
+        required:
+          - emoji
+        properties:
+            emoji:
+                type: string
+                enum: ["👉"]
+            text:
+                type: string
+                default: "the "
+        additionalProperties: false
+    EnumEmojiQuick:
+        type: object
+        required:
+          - emoji
+        properties:
+            emoji:
+                type: string
+                enum: ["⚡"]
+            text:
+                type: string
+                default: "quick "
+        additionalProperties: false
+    EnumEmojiBrown:
+        type: object
+        required:
+          - emoji
+        properties:
+            emoji:
+                type: string
+                enum: ["🟤"]
+            text:
+                type: string
+                default: "brown "
+        additionalProperties: false
+    EnumEmojiFox:
+        type: object
+        required:
+          - emoji
+        properties:
+            emoji:
+                type: string
+                enum: ["🦊"]
+            text:
+                type: string
+                default: "fox "
+        additionalProperties: false
+    EnumEmojiJumps:
+        type: object
+        required:
+          - emoji
+        properties:
+            emoji:
+                type: string
+                enum: ["🦘"]
+            text:
+                type: string
+                default: "jumps "
+        additionalProperties: false
+    EnumEmojiOver:
+        type: object
+        required:
+          - emoji
+        properties:
+            emoji:
+                type: string
+                enum: ["over"]
+            text:
+                type: string
+                default: "over "
+        additionalProperties: false
+    EnumEmojiLazyDog:
+        type: object
+        required:
+          - emoji
+        properties:
+            emoji:
+                type: string
+                enum: ["lazy🐶"]
+            text:
+                type: string
+                default: "lazy dog"
+        additionalProperties: false
+
diff --git a/chaotic/integration_tests/tests/render/emoji_enum.cpp b/chaotic/integration_tests/tests/render/emoji_enum.cpp
@@ -0,0 +1,33 @@
+#include <userver/formats/json/inline.hpp>
+#include <userver/utest/assert_macros.hpp>
+
+#include <userver/formats/parse/variant.hpp>  //  TODO: should work without it
+
+#include <gmock/gmock.h>
+
+#include <string>
+#include <vector>
+
+#include <schemas/emoji_enum.hpp>
+
+USERVER_NAMESPACE_BEGIN
+
+TEST(Formatter, EmojiEnum) {
+    auto obj1 = formats::json::MakeObject("emoji", "👉").As<ns::EnumEmojiThe>();
+    auto obj2 = formats::json::MakeObject("emoji", "⚡").As<ns::EnumEmoji>();
+    auto obj3 = formats::json::MakeObject("emoji", "🟤").As<ns::EnumEmoji>();
+    auto obj4 = formats::json::MakeObject("emoji", "🦊").As<ns::EnumEmoji>();
+    auto obj5 = formats::json::MakeObject("emoji", "🦘").As<ns::EnumEmoji>();
+    auto obj6 = formats::json::MakeObject("emoji", "over").As<ns::EnumEmoji>();
+    auto obj7 = formats::json::MakeObject("emoji", "lazy🐶").As<ns::EnumEmoji>();
+    std::vector<ns::EnumEmoji> objects{obj1, obj2, obj3, obj4, obj5, obj6, obj7};
+
+    std::string phrase;
+
+    for (const auto& obj : objects) {
+        phrase += std::visit([](const auto& item) -> std::string { return item.text; }, obj);
+    }
+    EXPECT_EQ(phrase, "the quick brown fox jumps over lazy dog");
+}
+
+USERVER_NAMESPACE_END
diff --git a/chaotic/tests/back/cpp/test_tr_object.py b/chaotic/tests/back/cpp/test_tr_object.py
@@ -125,6 +125,37 @@ def test_field_with_default(simple_gen, cpp_primitive_type):
     }
 
 
+def test_field_escaping(simple_gen, cpp_primitive_type):
+    schemas = simple_gen({
+        'type': 'object',
+        'properties': {'🙂🔥': {'type': 'integer', 'default': 1}},
+        'additionalProperties': False,
+    })
+    assert schemas == {
+        '::type': cpp_types.CppStruct(
+            raw_cpp_type=type_name.TypeName('::type'),
+            json_schema=front_types.Schema(),
+            nullable=False,
+            user_cpp_type=None,
+            # name='vfull#/definitions/type',
+            fields={
+                '🙂🔥': cpp_types.CppStructField(
+                    name='u1F642_u1F525',
+                    required=False,
+                    schema=cpp_primitive_type(
+                        validators=cpp_types.CppPrimitiveValidator(
+                            namespace='::type',
+                            prefix='u1F642_u1F525_',
+                        ),
+                        raw_cpp_type_str='int',
+                        default=1,
+                    ),
+                ),
+            },
+        ),
+    }
+
+
 def test_field_inplace(simple_gen, cpp_primitive_type):
     schemas = simple_gen({
         'type': 'object',
diff --git a/chaotic/tests/back/cpp/test_tr_string.py b/chaotic/tests/back/cpp/test_tr_string.py
@@ -34,6 +34,40 @@ def test_enum(simple_gen):
     }
 
 
+def test_enum_emoji(simple_gen):
+    types = simple_gen({'type': 'string', 'enum': ['🙂', '🔥'], 'default': '🔥'})
+    assert types == {
+        '::type': cpp_types.CppStringEnum(
+            raw_cpp_type=type_name.TypeName('::type'),
+            user_cpp_type=None,
+            json_schema=front_types.Schema(),
+            nullable=False,
+            name='::type',
+            default='::type::kU1f525',
+            enums=[
+                cpp_types.CppStringEnumItem(raw_name='🙂', cpp_name='kU1f642'),
+                cpp_types.CppStringEnumItem(raw_name='🔥', cpp_name='kU1f525'),
+            ],
+        ),
+    }
+
+    types = simple_gen({'type': 'string', 'enum': ['❗️ok🜈🙂', '❌ok🜈🔥']})
+    assert types == {
+        '::type': cpp_types.CppStringEnum(
+            raw_cpp_type=type_name.TypeName('::type'),
+            user_cpp_type=None,
+            json_schema=front_types.Schema(),
+            nullable=False,
+            name='::type',
+            default=None,
+            enums=[
+                cpp_types.CppStringEnumItem(raw_name='❗️ok🜈🙂', cpp_name='kU2757OkU1f708U1f642'),
+                cpp_types.CppStringEnumItem(raw_name='❌ok🜈🔥', cpp_name='kU274cokU1f708U1f525'),
+            ],
+        ),
+    }
+
+
 def test_datetime(simple_gen):
     types = simple_gen({'type': 'string', 'format': 'date-time'})
     assert types == {
