[core] Load component aliases from a generated registry (#18335)

.github/workflows/ci.yml

@@ -179,6 +179,7 @@ jobs:
           . venv/bin/activate
           script/ci-custom.py
           script/build_codeowners.py --check
+          script/build_alias_registry.py --check
           script/build_language_schema.py --check
           script/generate-esp32-boards.py --check
           script/generate-rp2-boards.py --check

esphome/component_aliases.py

@@ -0,0 +1,10 @@
+"""Component alias registry.
+
+Generated by script/build_alias_registry.py - do not edit manually.
+See the component-alias section of esphome/loader.py.
+"""
+
+# alias -> (canonical component, removal version or None)
+COMPONENT_ALIASES: dict[str, tuple[str, str | None]] = {
+    "rp2040": ("rp2", "2027.7.0"),
+}

esphome/loader.py

@@ -269,10 +269,9 @@ def _lookup_module(domain: str, exception: bool) -> ComponentManifest | None:
     # If `domain` is the legacy name of a renamed component, redirect to the
     # canonical module so the rest of the loader (and every caller of
     # `get_component(legacy)`) transparently sees the new component.
-    alias_map = _get_alias_map()
-    if domain in alias_map:
-        canonical = alias_map[domain]
-        manif = _lookup_module(canonical, exception)
+    alias_meta = get_alias_metadata().get(domain)
+    if alias_meta is not None:
+        manif = _lookup_module(alias_meta.canonical, exception)
         if manif is not None:
             _COMPONENT_CACHE[domain] = manif
         return manif
@@ -329,8 +328,10 @@ def _replace_component_manifest(domain: str, manifest: ComponentManifest) -> Non
 # ---------------------------------------------------------------------------
 #
 # A component can declare ``ALIASES = ["legacy_name"]`` (and optionally
-# ``ALIAS_REMOVAL_VERSION = "YYYY.M.0"``) in its ``__init__.py``. Two
-# integrations are then wired up automatically:
+# ``ALIAS_REMOVAL_VERSION = "YYYY.M.0"``) in its ``__init__.py``, then run
+# ``script/build_alias_registry.py`` to regenerate
+# ``esphome/component_aliases.py`` (CI and a unit test fail if the registry
+# is stale). Two integrations are then wired up automatically:
 #
 #   1. **Python imports** — a ``sys.meta_path`` finder (``_AliasFinder``)
 #      intercepts ``esphome.components.<legacy>``/``...<legacy>.<sub>``
@@ -344,13 +345,13 @@ def _replace_component_manifest(domain: str, manifest: ComponentManifest) -> Non
 #      dependency checks, schema validation and codegen all see only the
 #      canonical name.
 #
-# Both lookups are populated by ``_build_alias_map``, which **AST-parses**
-# every component's ``__init__.py`` rather than importing it. That keeps the
-# cost low: scanning ~400 components on disk takes ~5 ms instead of the
-# multi-second cost of executing every component's import side-effects.
+# Both lookups read the checked-in registry in ``esphome.component_aliases``
+# (generated by ``script/build_alias_registry.py``, verified in CI), so no
+# component-directory scan happens at runtime. ``_build_alias_map`` below is
+# the generator's scan implementation; it **AST-parses** each component's
+# ``__init__.py`` rather than importing it.
 
 
-_ALIAS_MAP_CACHE: dict[str, str] | None = None
 _ALIAS_META_CACHE: dict[str, "AliasMeta"] | None = None
 
 
@@ -367,31 +368,17 @@ class AliasMeta:
     removal_version: str | None
 
 
-def _ensure_alias_caches() -> None:
-    """Populate both alias caches from a single directory scan.
-
-    ``_build_alias_map`` returns both maps together, so building them in one
-    shot avoids scanning every component's ``__init__.py`` twice when a run
-    needs both the canonical map (loader) and the metadata map (config
-    pre-pass).
-    """
-    global _ALIAS_MAP_CACHE, _ALIAS_META_CACHE
-    if _ALIAS_MAP_CACHE is None or _ALIAS_META_CACHE is None:
-        _ALIAS_MAP_CACHE, _ALIAS_META_CACHE = _build_alias_map()
-
-
-def _get_alias_map() -> dict[str, str]:
-    """Return the legacy-name → canonical-name map, building it lazily."""
-    _ensure_alias_caches()
-    return _ALIAS_MAP_CACHE
-
-
 def get_alias_metadata() -> dict[str, AliasMeta]:
-    """Return the legacy-name → :class:`AliasMeta` map (cached).
+    """Return the legacy-name → :class:`AliasMeta` map, built lazily from
+    the generated registry."""
+    global _ALIAS_META_CACHE  # noqa: PLW0603
+    if _ALIAS_META_CACHE is None:
+        from esphome.component_aliases import COMPONENT_ALIASES
 
-    Used by the YAML pre-pass to format a per-alias deprecation warning.
-    """
-    _ensure_alias_caches()
+        _ALIAS_META_CACHE = {
+            alias: AliasMeta(canonical=canonical, removal_version=removal_version)
+            for alias, (canonical, removal_version) in COMPONENT_ALIASES.items()
+        }
     return _ALIAS_META_CACHE
 
 
@@ -537,11 +524,11 @@ class _AliasFinder(importlib.abc.MetaPathFinder):
         # least three parts, so ``parts[2]`` (the domain) always exists.
         parts = fullname.split(".")
         domain = parts[2]
-        alias_map = _get_alias_map()
-        if domain not in alias_map:
+        alias_meta = get_alias_metadata().get(domain)
+        if alias_meta is None:
             return None
 
-        parts[2] = alias_map[domain]
+        parts[2] = alias_meta.canonical
         canonical_fullname = ".".join(parts)
         try:
             canonical_module = importlib.import_module(canonical_fullname)

script/build_alias_registry.py

@@ -0,0 +1,59 @@
+#!/usr/bin/env python3
+"""Generate esphome/component_aliases.py from component ALIASES declarations.
+
+Run without arguments to regenerate the registry; ``--check`` (run in CI)
+verifies it is up to date.
+"""
+
+import argparse
+from pathlib import Path
+import sys
+
+# The root directory of the repo
+root = Path(__file__).parent.parent
+# Make the repo's esphome package win over any installed copy
+sys.path.insert(0, str(root))
+
+from esphome.helpers import write_file_if_changed  # noqa: E402
+from esphome.loader import _build_alias_map  # noqa: E402
+
+parser = argparse.ArgumentParser()
+parser.add_argument(
+    "--check",
+    help="Check if the alias registry is up to date.",
+    action="store_true",
+)
+args = parser.parse_args()
+
+registry_file = root / "esphome" / "component_aliases.py"
+
+HEADER = '''"""Component alias registry.
+
+Generated by script/build_alias_registry.py - do not edit manually.
+See the component-alias section of esphome/loader.py.
+"""
+
+# alias -> (canonical component, removal version or None)
+COMPONENT_ALIASES: dict[str, tuple[str, str | None]] = {
+'''
+
+# _build_alias_map scans the real component tree and already rejects
+# duplicate and shadowing aliases with an EsphomeError.
+_, alias_meta = _build_alias_map()
+
+lines = [HEADER]
+for alias, meta in sorted(alias_meta.items()):
+    removal = f'"{meta.removal_version}"' if meta.removal_version else "None"
+    lines.append(f'    "{alias}": ("{meta.canonical}", {removal}),\n')
+lines.append("}\n")
+content = "".join(lines)
+
+if args.check:
+    if registry_file.read_text(encoding="utf-8") != content:
+        print("Component alias registry is not up to date.")
+        print("Please run `script/build_alias_registry.py`")
+        sys.exit(1)
+    print("Component alias registry is up to date")
+else:
+    write_file_if_changed(registry_file, content)
+    print(f"Wrote {registry_file}")

tests/unit_tests/test_loader.py

@@ -8,6 +8,7 @@ from unittest.mock import MagicMock, patch
 
 import pytest
 
+from esphome.component_aliases import COMPONENT_ALIASES
 from esphome.loader import (
     AliasMeta,
     ComponentManifest,
@@ -481,6 +482,33 @@ def test_real_alias_map_includes_rp2040() -> None:
     assert meta["rp2040"].removal_version == "2027.7.0"
 
 
+def test_alias_registry_matches_component_tree() -> None:
+    """The checked-in registry must match a live scan of the component tree."""
+    _, meta_map = _build_alias_map()
+    expected = {
+        alias: (meta.canonical, meta.removal_version)
+        for alias, meta in meta_map.items()
+    }
+    assert expected == COMPONENT_ALIASES, (
+        "esphome/component_aliases.py is out of date; "
+        "run script/build_alias_registry.py"
+    )
+
+
+def test_alias_map_built_from_registry() -> None:
+    """The runtime alias map comes from the generated registry, not a scan."""
+    with (
+        patch(
+            "esphome.component_aliases.COMPONENT_ALIASES",
+            {"legacy": ("modern", "2099.1.0")},
+        ),
+        patch("esphome.loader._ALIAS_META_CACHE", None),
+    ):
+        assert get_alias_metadata() == {
+            "legacy": AliasMeta(canonical="modern", removal_version="2099.1.0")
+        }
+
+
 def test_get_component_resolves_alias() -> None:
     """``get_component('rp2040')`` should return the rp2 manifest — every
     caller of the loader (dep checker, schema validator, codegen) hits