Merge branch 'host-pch' into esp32-pio-pch

script/clang_tidy_hash.py

@@ -1,13 +1,10 @@
-"""Files that affect clang-tidy results, and a content hash over them.
+"""Files that affect clang-tidy results and the idedata built from them.
 
-``CLANG_TIDY_GLOBAL_FILES`` (plus ``SDKCONFIG_DEFAULTS_PREFIX``) is the single
-source of truth for which files influence clang-tidy output. A change to any of
-them can surface warnings in source files a PR didn't touch, so:
-
-* ``script/determine-jobs.py`` runs a full clang-tidy scan when one changes, and
-* ``calculate_clang_tidy_hash()`` folds them into the idedata cache key used by
-  ``script/helpers.py`` (a content hash, unlike an mtime check, stays correct
-  across git checkouts).
+``CLANG_TIDY_GLOBAL_FILES`` (plus ``SDKCONFIG_DEFAULTS_PREFIX``) lists the files
+that influence clang-tidy output; ``script/determine-jobs.py`` runs a full scan
+when one changes. ``ESP_IDF_INFRA_TRIGGER_*`` lists the native ESP-IDF build
+code. ``idedata_cache_hash()`` folds the right set into the idedata cache key
+used by ``script/helpers.py`` and the CI cache action.
 """
 
 from __future__ import annotations
@@ -31,6 +28,18 @@ CLANG_TIDY_GLOBAL_FILES = (
 # this prefix at the repo root.
 SDKCONFIG_DEFAULTS_PREFIX = "sdkconfig.defaults"
 
+# Native ESP-IDF build infra: determine-jobs forces an esp32 compile when these
+# change, and they feed the clang-tidy idedata cache key.
+ESP_IDF_INFRA_TRIGGER_PATH_PREFIXES = ("esphome/espidf/", "esphome/build_helpers/")
+ESP_IDF_INFRA_TRIGGER_FILES = frozenset(
+    {
+        "esphome/build_gen/espidf.py",
+        "esphome/framework_helpers.py",
+        "esphome/platformio/library.py",
+        "esphome/platformio/extra_script.py",
+    }
+)
+
 
 def read_file_bytes(path: Path) -> bytes:
     """Read bytes from a file."""
@@ -66,3 +75,33 @@ def calculate_clang_tidy_hash(repo_root: Path | None = None) -> str:
         hasher.update(read_file_bytes(path))
 
     return hasher.hexdigest()
+
+
+def calculate_idedata_cache_hash(repo_root: Path | None = None) -> str:
+    """Clang-tidy hash plus the Python that generates the idedata."""
+    repo_root = _ensure_repo_root(repo_root)
+
+    hasher = hashlib.sha256()
+    hasher.update(calculate_clang_tidy_hash(repo_root).encode())
+
+    paths = {repo_root / name for name in ESP_IDF_INFRA_TRIGGER_FILES}
+    for prefix in ESP_IDF_INFRA_TRIGGER_PATH_PREFIXES:
+        # .pyc files appear between the CI key computation and load_idedata's.
+        paths.update(
+            path
+            for path in (repo_root / prefix).rglob("*")
+            if "__pycache__" not in path.parts
+        )
+    for path in sorted(paths):
+        if path.is_file():
+            hasher.update(str(path.relative_to(repo_root)).encode())
+            hasher.update(read_file_bytes(path))
+
+    return hasher.hexdigest()
+
+
+def idedata_cache_hash(environment: str, repo_root: Path | None = None) -> str:
+    """Hash gating the cached idedata of one clang-tidy environment."""
+    if "esp32" in environment:
+        return calculate_idedata_cache_hash(repo_root)
+    return calculate_clang_tidy_hash(repo_root)

script/determine-jobs.py

@@ -59,7 +59,12 @@ from pathlib import Path
 import sys
 from typing import Any
 
-from clang_tidy_hash import CLANG_TIDY_GLOBAL_FILES, SDKCONFIG_DEFAULTS_PREFIX
+from clang_tidy_hash import (
+    CLANG_TIDY_GLOBAL_FILES,
+    ESP_IDF_INFRA_TRIGGER_FILES,
+    ESP_IDF_INFRA_TRIGGER_PATH_PREFIXES,
+    SDKCONFIG_DEFAULTS_PREFIX,
+)
 from helpers import (
     CPP_FILE_EXTENSIONS,
     ESPHOME_TESTS_COMPONENTS_PATH,
@@ -551,28 +556,6 @@ def _esp32_platformio_path_or_file_trigger(files: list[str]) -> bool:
     )
 
 
-# Native-build infra: changes under esphome/espidf/, the shared
-# esphome/build_helpers/ package, or the modules the native ESP-IDF build
-# imports affect every esp32 IDF build (now the default toolchain) but aren't
-# components, so the component matrix wouldn't otherwise force any esp32
-# compile. When they change we fold the `esp32` component into the matrix so
-# the default native-IDF build path is still compiled on an infra-only PR.
-ESP_IDF_INFRA_TRIGGER_PATH_PREFIXES = ("esphome/espidf/", "esphome/build_helpers/")
-# Shared library-conversion modules every native build imports; a new shared
-# module added to one native trigger set must not silently skip the other's
-# smoke test, so both sets union this one.
-_NATIVE_SHARED_TRIGGER_FILES = frozenset(
-    {
-        "esphome/framework_helpers.py",
-        "esphome/platformio/library.py",
-        "esphome/platformio/extra_script.py",
-    }
-)
-ESP_IDF_INFRA_TRIGGER_FILES = _NATIVE_SHARED_TRIGGER_FILES | {
-    "esphome/build_gen/espidf.py",
-}
-
-
 def _esp_idf_infra_changed(files: list[str]) -> bool:
     """Whether any changed file is ESP-IDF build/runner infrastructure."""
     return _path_or_file_trigger(
@@ -676,6 +659,18 @@ ESP8266_NATIVE_TRIGGER_PATH_PREFIXES = (
     "esphome/arduino/",
     "esphome/build_helpers/",
 )
+# Shared library-conversion modules every native build imports; espidf-only
+# infra (build_gen/espidf.py) deliberately stays out of the esp8266 set.
+_NATIVE_SHARED_TRIGGER_FILES = frozenset(
+    {
+        "esphome/framework_helpers.py",
+        "esphome/platformio/library.py",
+        "esphome/platformio/extra_script.py",
+    }
+)
+# Tripwire: the shared modules must stay in the ESP-IDF trigger set too
+# (now defined in clang_tidy_hash), or its smoke test silently skips them
+assert _NATIVE_SHARED_TRIGGER_FILES <= ESP_IDF_INFRA_TRIGGER_FILES
 ESP8266_NATIVE_TRIGGER_FILES = (
     _NATIVE_SHARED_TRIGGER_FILES
     | _SMOKE_HARNESS_TRIGGER_FILES

script/helpers.py

@@ -809,17 +809,14 @@ def load_idedata(environment: str) -> dict[str, Any]:
     start_time = time.time()
     print(f"Loading IDE data for environment '{environment}'...")
 
-    # Reuse the clang-tidy input hash as the cache key: it already covers every
-    # file baked into the generated idedata (platformio.ini, sdkconfig.defaults,
-    # esphome/idf_component.yml), so this can't drift from that file list. A
-    # content hash -- unlike an mtime comparison -- stays correct across git
-    # checkouts, which don't preserve mtimes.
-    from clang_tidy_hash import calculate_clang_tidy_hash
+    # Content hash of the idedata inputs (data files and the generator code); a
+    # content hash, unlike mtimes, stays correct across git checkouts.
+    from clang_tidy_hash import idedata_cache_hash
 
     temp_idedata = Path(temp_folder) / f"idedata-{environment}.json"
     temp_hash = Path(temp_folder) / f"idedata-{environment}.hash"
 
-    cache_key = calculate_clang_tidy_hash()
+    cache_key = idedata_cache_hash(environment)
     changed = (
         not temp_idedata.is_file()
         or not temp_hash.is_file()