Merge branch 'esp8266-native-toolchain-plumbing' into esp8266-native-build-infra

esphome/platformio/library.py

@@ -15,6 +15,7 @@ regardless of which toolchain consumes the result.
 from collections import deque
 from collections.abc import Callable, Iterable
 from concurrent.futures import ThreadPoolExecutor
+import contextlib
 from dataclasses import dataclass, field
 import glob
 import hashlib
@@ -103,14 +104,7 @@ class URLSource(Source):
     def __init__(self, url: str):
         self.url = url
 
-    def download(
+    def _cache_dir(self, dir_suffix: str, salt: str, namespace: str) -> Path:
-        self,
-        dir_suffix: str,
-        force: bool = False,
-        salt: str = "",
-        namespace: str = "",
-        progress: Callable[[int], None] | None = None,
-    ) -> Path:
         # Namespace the cache per backend (e.g. pio_components/idf, .../zephyr) so
         # the build files each backend writes into the library dir can't collide.
         base_dir = Path(CORE.data_dir) / DOMAIN
@@ -120,7 +114,23 @@ class URLSource(Source):
         h.update(self.url.encode())
         if salt:
             h.update(salt.encode())
-        path = base_dir / h.hexdigest()[:8] / dir_suffix
+        return base_dir / h.hexdigest()[:8] / dir_suffix
+
+    def is_cached(self, dir_suffix: str, salt: str = "", namespace: str = "") -> bool:
+        """Whether a completed extraction already exists for this source."""
+        return (
+            self._cache_dir(dir_suffix, salt, namespace) / ".esphome_extracted"
+        ).is_file()
+
+    def download(
+        self,
+        dir_suffix: str,
+        force: bool = False,
+        salt: str = "",
+        namespace: str = "",
+        progress: Callable[[int], None] | None = None,
+    ) -> Path:
+        path = self._cache_dir(dir_suffix, salt, namespace)
         # Marker file written last to signal a complete extraction. Using a
         # marker (instead of just `path.is_dir()`) means an interrupted
         # extraction is correctly detected and re-run on the next invocation,
@@ -918,6 +928,13 @@ def _prefetch_wave(
         if component.source.url in seen:
             continue
         seen.add(component.source.url)
+        with contextlib.suppress(Exception):
+            if component.source.is_cached(
+                component.get_sanitized_name(), salt=salt, namespace=namespace
+            ):
+                # A completed extraction downloads nothing; a warm build
+                # must stay silent
+                continue
         components.append(component)
     if len(components) < 2:
         return

tests/unit_tests/test_platformio_library.py

@@ -644,6 +644,50 @@ def test_content_lengths_head_requests(monkeypatch: pytest.MonkeyPatch) -> None:
     assert lib._content_lengths(["https://x/a", "https://x/bad"]) == [123, 0]
 
 
+def test_prefetch_wave_cache_probe_failure_still_prefetches(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """The cache probe is best-effort; a failing probe prefetches anyway."""
+    calls: list[str] = []
+    monkeypatch.setattr(
+        ConvertedLibrary,
+        "download",
+        lambda self, **kw: calls.append(self.source.url),
+    )
+    monkeypatch.setattr(
+        URLSource,
+        "is_cached",
+        lambda self, *a, **kw: (_ for _ in ()).throw(RuntimeError("no core")),
+    )
+    wave = [
+        ("a", ConvertedLibrary("a", "1.0", URLSource("https://x/a.tar.gz"))),
+        ("b", ConvertedLibrary("b", "1.0", URLSource("https://x/b.tar.gz"))),
+    ]
+    lib._prefetch_wave(wave, "", "idf")
+    assert sorted(calls) == ["https://x/a.tar.gz", "https://x/b.tar.gz"]
+
+
+def test_prefetch_wave_warm_cache_is_silent(
+    setup_core, monkeypatch: pytest.MonkeyPatch, caplog: pytest.LogCaptureFixture
+) -> None:
+    """Already-extracted archives download nothing; a warm build must not
+    print a Downloading line or draw a bar."""
+    monkeypatch.setattr(
+        ConvertedLibrary,
+        "download",
+        lambda self, **kw: (_ for _ in ()).throw(AssertionError("downloaded")),
+    )
+    wave = []
+    for name in ("a", "b", "c"):
+        comp = ConvertedLibrary(name, "1.0", URLSource(f"https://x/{name}.tar.gz"))
+        marker_dir = comp.source._cache_dir(comp.get_sanitized_name(), "", "idf")
+        marker_dir.mkdir(parents=True)
+        (marker_dir / ".esphome_extracted").touch()
+        wave.append((name, comp))
+    lib._prefetch_wave(wave, "", "idf")
+    assert "Downloading" not in caplog.text
+
+
 def test_prefetch_wave_single_archive_skips_the_pool(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None: