refactor(text): unify Chinese conversion on Rust (#6481)

tests/fixtures/architecture/dependency-baseline.json

@@ -1441,8 +1441,8 @@
       "runtime_only": true
     }
   },
-  "edge_count": 6836,
-  "edge_sha256": "3709e09a49257075f48a44aba353be1f25db610471ec756e09cbbfd7070acb9a",
+  "edge_count": 6834,
+  "edge_sha256": "07c0b6f24ef3ce3e7ea0fa7c1b617388e11e2453ce2b0b35bddb38ed2d7f5835",
   "edges": [
     "app -> app.runtime",
     "app -> app.runtime.compat",
@@ -5526,8 +5526,6 @@
     "app.factory -> app.schemas.response",
     "app.factory -> app.startup",
     "app.factory -> app.startup.lifecycle",
-    "app.foundation.text -> app.foundation",
-    "app.foundation.text -> app.foundation.environment",
     "app.main -> app.adapters",
     "app.main -> app.adapters.system",
     "app.main -> app.adapters.system.host",

tests/test_free_threaded_ab.py

@@ -54,7 +54,7 @@ def valid_preflight(variant: str) -> dict:
                 "gil_disabled": False,
                 "gil_enabled": True,
                 "thread_inherit_context": 0,
-                "has_zhconv_fast": False,
+                "has_zhconv_fast": True,
                 "gil_enabled_after_imports": True,
                 "packages": {
                     "bcrypt": "4.3.0",
@@ -65,7 +65,7 @@ def valid_preflight(variant: str) -> dict:
                     "orjson": "3.12.0",
                     "psycopg": None,
                     "psycopg2-binary": "2.9.12",
-                    "zhconv-rs": "0.4.1",
+                    "zhconv-rs": None,
                 },
                 "native": {
                     "crcmod_extension": True,
@@ -123,7 +123,7 @@ def valid_preflight(variant: str) -> dict:
         "zstandard",
     }
     expected_imports.update(
-        {"psycopg2-binary", "zhconv-rs"} if variant == "v3" else {"psycopg"}
+        {"psycopg2-binary"} if variant == "v3" else {"psycopg"}
     )
     common["imports"] = {
         name: {"imported": True, "gil_after": variant == "v3"}
@@ -341,9 +341,12 @@ def test_preflight_enforces_runtime_and_native_profiles() -> None:
     assert harness.validate_preflight("v3", valid_preflight("v3")) == []
     assert harness.validate_preflight("v3t", valid_preflight("v3t")) == []
 
-    leaked = valid_preflight("v3")
-    leaked["has_zhconv_fast"] = True
-    assert any("has_zhconv_fast" in item for item in harness.validate_preflight("v3", leaked))
+    incomplete_text = valid_preflight("v3")
+    incomplete_text["has_zhconv_fast"] = False
+    assert any(
+        "has_zhconv_fast" in item
+        for item in harness.validate_preflight("v3", incomplete_text)
+    )
 
     unsafe = valid_preflight("v3t")
     unsafe["gil_enabled_after_imports"] = True

tests/test_jieba_utils.py

@@ -1,4 +1,4 @@
-from app.foundation.text import cut
+from app.foundation.text import convert, cut
 
 
 def test_cut_accepts_hmm_argument():
@@ -19,3 +19,8 @@ def test_cut_preserves_full_mode_contract():
         "长江大桥",
         "大桥",
     ]
+
+
+def test_convert_uses_rust_mediawiki_contract():
+    """标准与 free-threaded 运行时应共享同一中文转换语义。"""
+    assert convert("后台", "zh-hant") == "後台"

tests/test_runtime_dependencies.py

@@ -79,6 +79,27 @@ def test_runtime_profiles_share_gil_safe_crcmod_distribution():
     } in document["tool"]["uv"]["exclude-dependencies"]
 
 
+def test_standard_runtime_uses_shared_rust_text_capabilities(monkeypatch):
+    """标准镜像的文本能力不得重新引入 profile 专属实现。"""
+    imported = []
+    moviepilot_rust = SimpleNamespace(
+        is_available=lambda: True,
+        jieba_cut=lambda _value: ["中文", "分词"],
+        zhconv_fast=lambda value, _target: value,
+    )
+
+    def import_module(name):
+        imported.append(name)
+        return moviepilot_rust if name == "moviepilot_rust" else SimpleNamespace()
+
+    monkeypatch.setattr(dependency_doctor, "import_module", import_module)
+    monkeypatch.setattr(dependency_doctor.sysconfig, "get_config_var", lambda _name: None)
+
+    dependency_doctor.main()
+
+    assert "zhconv_rs" not in imported
+
+
 def test_runtime_excluded_dependency_pairs_reads_uv_policy(tmp_path: Path):
     """运行时诊断应复用 uv 排除配置，不维护第二份包名特判。"""
     project_file = tmp_path / "pyproject.toml"