Filter helper schemas from catalog (#40)

* Filter helper schemas from site manifest

* Apply local site updates

Author: laubonghaudoi

.codex/environments/environment.toml

@@ -0,0 +1,14 @@
+# THIS IS AUTOGENERATED. DO NOT EDIT MANUALLY
+version = 1
+name = "Chinese_Rime"
+
+[setup]
+script = ""
+
+[[actions]]
+name = "Run"
+icon = "run"
+command = '''
+cd site
+corepack pnpm dev
+'''

script/manifest/walk_submodules.py

@@ -7,6 +7,9 @@
 import yaml
 
 
+HELPER_SCHEMA_DIRS = {"ext-dict"}
+
+
 def walk_submodules(sources_dir: Path) -> list[dict]:
     """Return one normalized dict per unique schema_id under sources_dir.
 
@@ -35,7 +38,16 @@ def walk_submodules(sources_dir: Path) -> list[dict]:
 def _iter_schema_yamls(root: Path) -> Iterable[Path]:
     # Walk recursively but emit shallower paths first so the de-dup in
     # walk_submodules keeps the primary file rather than a nested variant.
-    yield from sorted(root.rglob("*.schema.yaml"), key=lambda p: (len(p.parts), str(p)))
+    schema_paths = (
+        path
+        for path in root.rglob("*.schema.yaml")
+        if not _is_helper_schema(path.relative_to(root))
+    )
+    yield from sorted(schema_paths, key=lambda p: (len(p.parts), str(p)))
+
+
+def _is_helper_schema(relative_path: Path) -> bool:
+    return any(part in HELPER_SCHEMA_DIRS for part in relative_path.parts[:-1])
 
 
 def _load_one(path: Path) -> dict:

script/manifest_overrides.yaml

@@ -591,31 +591,91 @@ schema_assignments:
     dialect: 潮州話
     recipe: kahaani/dieghv
     upstream_url: https://github.com/kahaani/dieghv
+  dieziu_60:
+    branch: nan
+    dialect: 潮州話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
+  dieziu_mtr:
+    branch: nan
+    dialect: 潮州話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
   dioion:
     branch: nan
     dialect: 潮陽話
     recipe: kahaani/dieghv
     upstream_url: https://github.com/kahaani/dieghv
+  dioion_60:
+    branch: nan
+    dialect: 潮陽話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
+  dioion_mtr:
+    branch: nan
+    dialect: 潮陽話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
   gekion:
     branch: nan
     dialect: 揭陽話
     recipe: kahaani/dieghv
     upstream_url: https://github.com/kahaani/dieghv
+  gekion_60:
+    branch: nan
+    dialect: 揭陽話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
+  gekion_mtr:
+    branch: nan
+    dialect: 揭陽話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
   suantau:
     branch: nan
     dialect: 汕頭話
     recipe: kahaani/dieghv
     upstream_url: https://github.com/kahaani/dieghv
+  suantau_60:
+    branch: nan
+    dialect: 汕頭話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
+  suantau_mtr:
+    branch: nan
+    dialect: 汕頭話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
   tenghai:
     branch: nan
     dialect: 澄海話
     recipe: kahaani/dieghv
     upstream_url: https://github.com/kahaani/dieghv
+  tenghai_60:
+    branch: nan
+    dialect: 澄海話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
+  tenghai_mtr:
+    branch: nan
+    dialect: 澄海話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
   riaupeng:
     branch: nan
     dialect: 饒平話
     recipe: kahaani/dieghv
     upstream_url: https://github.com/kahaani/dieghv
+  riaupeng_60:
+    branch: nan
+    dialect: 饒平話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
+  riaupeng_mtr:
+    branch: nan
+    dialect: 饒平話
+    recipe: kahaani/dieghv
+    upstream_url: https://github.com/kahaani/dieghv
   diege_j:
     branch: nan
     dialect: 潮州話

script/tests/test_walk_submodules.py

@@ -70,3 +70,15 @@ def test_walk_submodules_keeps_shallow_duplicate_schema(tmp_path: Path):
     result = walk_submodules(tmp_path / "sources")
     assert len(result) == 1
     assert result[0]["display_name"] == "Primary"
+
+
+def test_walk_submodules_skips_extension_dictionary_schemas(tmp_path: Path):
+    primary = tmp_path / "sources" / "晉語" / "sample" / "jieny.schema.yaml"
+    helper = tmp_path / "sources" / "晉語" / "sample" / "ext-dict" / "easy_en.schema.yaml"
+    primary.parent.mkdir(parents=True)
+    helper.parent.mkdir(parents=True)
+    primary.write_text("schema:\n  schema_id: jieny\n  name: 嘉樂泉話\n", encoding="utf-8")
+    helper.write_text("schema:\n  schema_id: easy_en\n  name: Easy English\n", encoding="utf-8")
+
+    result = walk_submodules(tmp_path / "sources")
+    assert [schema["schema_id"] for schema in result] == ["jieny"]