Merge remote-tracking branch 'origin/external-files-parallel' into integration

esphome/components/audio_file/__init__.py

@@ -1,4 +1,5 @@
 from dataclasses import dataclass, field
+from functools import partial
 import hashlib
 import logging
 from pathlib import Path
@@ -199,7 +200,7 @@ def _validate_supported_local_file(config: list[ConfigType]) -> list[ConfigType]
 CONFIG_SCHEMA = cv.All(
     cv.only_on_esp32,
     cv.ensure_list(MEDIA_FILE_TYPE_SCHEMA),
-    lambda c: download_web_files_in_config(c, _compute_local_file_path),
+    partial(download_web_files_in_config, path_for=_compute_local_file_path),
     _validate_supported_local_file,
 )
 

esphome/components/speaker/media_player/__init__.py

@@ -1,5 +1,6 @@
 """Speaker Media Player Setup."""
 
+from functools import partial
 import hashlib
 import logging
 from pathlib import Path
@@ -277,7 +278,9 @@ CONFIG_SCHEMA = cv.All(
             cv.Optional(CONF_CODEC_SUPPORT_ENABLED): cv.Any(cv.boolean, cv.string),
             cv.Optional(CONF_FILES): cv.All(
                 cv.ensure_list(MEDIA_FILE_TYPE_SCHEMA),
-                lambda c: download_web_files_in_config(c, _compute_local_file_path),
+                partial(
+                    download_web_files_in_config, path_for=_compute_local_file_path
+                ),
             ),
             cv.Optional(CONF_TASK_STACK_IN_PSRAM): cv.All(
                 cv.boolean, cv.requires_component(psram.DOMAIN)

esphome/external_files.py

@@ -216,13 +216,12 @@ def download_content_many(
 
     Items are de-duplicated by `path` -- two callers asking for the same
     cache file (e.g. the same URL referenced twice in a config) would
-    otherwise race on `download_content`'s non-atomic write.
+    otherwise race on `download_content`'s non-atomic write. When the
+    same `path` appears more than once, the last URL wins (standard dict
+    comprehension semantics); in practice duplicate paths only arise when
+    the URL is duplicated, so the choice doesn't matter.
     """
-    seen: dict[Path, str] = {}
-    for url, path in items:
-        if path in seen:
-            continue
-        seen[path] = url
+    seen: dict[Path, str] = {path: url for url, path in items}
     if not seen:
         return
     if len(seen) == 1:

tests/unit_tests/test_external_files.py

@@ -567,7 +567,9 @@ def test_download_content_many_dedupes_by_path(
     mock_download_content: MagicMock, setup_core: Path
 ) -> None:
     """Two items pointing at the same cache path must collapse to one
-    download -- otherwise concurrent writes race on the same file.
+    download -- otherwise concurrent writes race on the same file. Which
+    URL wins doesn't matter (in practice duplicate paths only arise when
+    the URL is duplicated), so we only assert the call count and path.
     """
     path = setup_core / "shared"
     items = [
@@ -577,9 +579,7 @@ def test_download_content_many_dedupes_by_path(
     ]
     external_files.download_content_many(items)
     assert mock_download_content.call_count == 1
-    # First-seen URL wins, matching dict-insertion order.
     args, _ = mock_download_content.call_args
-    assert args[0] == "https://example.com/a"
     assert args[1] == path