v1.46.1: re-check of v1.46.0 — HP-1460-01, -02, -03

HP-1460-01: v1.46.0 stopped overwriting attachments, but picking a free name
and taking it were two steps. Two uploads racing between them agreed on the
same name, both answered 200, and one set of bytes replaced the other;
files/migrate had the same check-then-copy gap. reserve_filename now claims the
name with O_CREAT|O_EXCL as it picks it, and both paths use it. It also splits
the extension off the RAW name and budgets the stem against MAX_FILENAME
including the collision tag — a maximal name lost its '.pdf' and then grew past
the limit, so the view sanitised the request back to a different name and the
attachment 404'd for good.

HP-1460-02: cleanup lived in an 'except Exception', which CancelledError walks
past, only one tmp_path was tracked, promotion had no finally, and the
collector only walks marker folders — an aborted transfer stranded a .upload-*
that nothing would ever remove. An outer finally owns every temporary, a second
'file' part is refused, promotion failure cleans up, and sweep_upload_temps
runs at setup, daily, and inside the commit-scoped collector. Chunks are
batched to 1 MB per disk task instead of one per 64 KB.

HP-1460-03: the layout event reached the static card and not the full one, so
two full cards diverged until a reload. The full card subscribes now and
re-reads ONLY the layout, keyed on its revision. Two hazards handled: it
records revisions it produced itself, and the reaction is deferred ~200 ms
because the event can beat the reply to our own write over the same socket;
positions dragged but not yet sent are flushed and merged on top, so a fix for
a stale UI cannot become a lost drag.

Tests: smoke_layout_sync (fails on a v1.46.0 build), four pure tests for atomic
reservation incl. 20-thread concurrency and the length boundary, a backend test
walking every failing exit path of an upload, and — as the report asked — an
HA-harness test that a repair issue disappears with its space.
Docs: CHANGELOG.md + CHANGELOG.ru.md + ARCHITECTURE.md + TESTING.md + STATUS.md.

tests_backend/test_ha_websocket.py

@@ -696,3 +696,90 @@ async def test_upload_never_overwrites_an_existing_attachment(
     assert await got.read() == b"ONE", "the first file is untouched"
     got2 = await http.get(second)
     assert await got2.read() == b"TWO"
+
+
+async def test_upload_leaves_no_temporary_behind(
+    hass: HomeAssistant, hass_ws_client: WebSocketGenerator, hass_client, monkeypatch
+) -> None:
+    """HP-1460-02: every exit path must take its temporary file with it.
+
+    The streaming rewrite kept one `tmp_path` and cleaned it in an
+    `except Exception`, which cancellation (a BaseException) walks straight
+    past — and the attachment collector only ever looks inside marker folders,
+    so a stranded `.upload-*` was never seen again.
+    """
+    import os
+
+    from custom_components.houseplan import http_api
+    from custom_components.houseplan.const import FILES_DIR
+    from custom_components.houseplan.plans import TMP_PREFIX
+
+    await _setup(hass)
+    http = await hass_client()
+    root = hass.config.path(FILES_DIR)
+
+    def temps() -> list[str]:
+        return [n for n in os.listdir(root) if n.startswith(TMP_PREFIX)]
+
+    import aiohttp
+
+    def form(*files, marker="m8"):
+        w = aiohttp.FormData()
+        w.add_field("marker_id", marker)
+        for name, data in files:
+            w.add_field("file", data, filename=name)
+        return w
+
+    # two file parts: refused, and nothing left over
+    resp = await http.post("/api/houseplan/upload", data=form(("a.pdf", b"A"), ("b.pdf", b"B")))
+    assert resp.status == 400 and (await resp.json())["error"] == "one_file_only"
+    assert temps() == []
+
+    # a rejected extension after the temporary already exists
+    resp = await http.post("/api/houseplan/upload", data=form(("evil.exe", b"X")))
+    assert resp.status == 400
+    assert temps() == []
+
+    # promotion itself blows up
+    real = http_api.reserve_filename
+
+    def _boom(*_a, **_k):
+        raise OSError("disk on fire")
+
+    monkeypatch.setattr(http_api, "reserve_filename", _boom)
+    resp = await http.post("/api/houseplan/upload", data=form(("c.pdf", b"C")))
+    assert resp.status == 500
+    assert temps() == [], "a failed promotion must not strand the upload"
+    monkeypatch.setattr(http_api, "reserve_filename", real)
+
+    # and the happy path leaves nothing either
+    resp = await http.post("/api/houseplan/upload", data=form(("d.pdf", b"D")))
+    assert resp.status == 200
+    assert temps() == []
+
+
+async def test_repair_issue_goes_when_its_space_does(
+    hass: HomeAssistant, hass_ws_client: WebSocketGenerator
+) -> None:
+    """HP-1454-09: the cleanup used to walk only spaces that still exist.
+
+    So deleting or renaming a space with a missing plan left its warning in
+    Repairs with nothing able to clear it.
+    """
+    from homeassistant.helpers import issue_registry as ir
+
+    from custom_components.houseplan.const import DOMAIN as HP_DOMAIN
+
+    await _setup(hass)
+    client = await hass_ws_client(hass)
+    registry = ir.async_get(hass)
+
+    gone = "/api/houseplan/content/plans/_/nosuchfile.png"
+    rev = (await _save(client, await _cfg([{"id": "r7", "plan_url": gone}]), 0))["result"]["rev"]
+    await hass.async_block_till_done()
+    assert registry.async_get_issue(HP_DOMAIN, "broken_plan_r7") is not None
+
+    # the space is deleted entirely — the warning must not outlive it
+    await _save(client, await _cfg([{"id": "other", "plan_url": None}]), rev)
+    await hass.async_block_till_done()
+    assert registry.async_get_issue(HP_DOMAIN, "broken_plan_r7") is None

tests_backend/test_validation.py

@@ -386,25 +386,81 @@ def test_every_room_fill_mode_the_editor_offers_is_accepted():
 # ---------- attachments & inner limits (HP-1454-02, -05) ----------
 
 
-def test_unique_filename_never_returns_a_taken_name(tmp_path):
-    unique_filename = plans.unique_filename
-    d = tmp_path / "m1"
-    d.mkdir()
-    assert unique_filename(d, "manual.pdf") == "manual.pdf"
-    (d / "manual.pdf").write_bytes(b"x")
-    assert unique_filename(d, "manual.pdf") == "manual-2.pdf"
-    (d / "manual-2.pdf").write_bytes(b"x")
-    assert unique_filename(d, "manual.pdf") == "manual-3.pdf"
-    # no extension, and a name that needs sanitising
-    (d / "readme").write_bytes(b"x")
-    assert unique_filename(d, "readme") == "readme-2"
-    assert unique_filename(d, "../../etc/passwd") == "passwd"
+def test_reserve_filename_claims_the_name_atomically(tmp_path):
+    """HP-1460-01: choosing a name and taking it must be one operation.
 
-    # A generated name has to survive the sanitiser the CONTENT VIEW applies to
-    # the request, or the file is written and then never served: " (2)" became
-    # "_2_" and every collision-renamed attachment 404'd.
-    for candidate in ("manual-2.pdf", "manual-3.pdf", "readme-2"):
-        assert v.sanitize_filename(candidate) == candidate
+    The old helper asked `exists()` and returned a string; two uploads racing
+    between the check and the write agreed on the same name and one overwrote
+    the other, both reporting success.
+    """
+    reserve = plans.reserve_filename
+    d = tmp_path / "m1"
+
+    first = reserve(d, "manual.pdf")
+    assert first == "manual.pdf"
+    assert (d / first).is_file(), "the name is taken, not merely picked"
+    second = reserve(d, "manual.pdf")
+    assert second == "manual-2.pdf" and (d / second).is_file()
+    assert reserve(d, "manual.pdf") == "manual-3.pdf"
+
+    assert reserve(d, "readme") == "readme"
+    assert reserve(d, "readme") == "readme-2"
+    assert reserve(d, "../../etc/passwd") == "passwd"
+
+
+def test_reserve_filename_result_survives_the_content_sanitizer(tmp_path):
+    """A name the view would rewrite is a file written and then never served."""
+    reserve = plans.reserve_filename
+    d = tmp_path / "m2"
+    long_stem = "x" * 200                      # far past the limit
+    names = [reserve(d, f"{long_stem}.pdf") for _ in range(12)]
+    assert len(set(names)) == 12, "each call takes its own name"
+    for n in names:
+        assert len(n) <= v.MAX_FILENAME
+        assert v.sanitize_filename(n) == n, n
+        assert n.endswith(".pdf")
+
+    # a name already exactly at the limit still leaves room for the tag
+    exact = "y" * (v.MAX_FILENAME - 4) + ".pdf"
+    assert len(exact) == v.MAX_FILENAME
+    a = reserve(d, exact)
+    b = reserve(d, exact)
+    assert a != b and len(b) <= v.MAX_FILENAME and v.sanitize_filename(b) == b
+
+
+def test_reserve_filename_is_safe_under_concurrency(tmp_path):
+    """Twenty threads, one filename: twenty distinct files, nothing overwritten."""
+    from concurrent.futures import ThreadPoolExecutor
+
+    reserve = plans.reserve_filename
+    d = tmp_path / "m3"
+    d.mkdir(parents=True)
+    with ThreadPoolExecutor(max_workers=20) as pool:
+        names = list(pool.map(lambda _: reserve(d, "manual.pdf"), range(20)))
+    assert len(set(names)) == 20
+    assert sorted(p.name for p in d.iterdir()) == sorted(names)
+
+
+def test_sweep_upload_temps(tmp_path):
+    """HP-1460-02: a crashed transfer leaves a temp no other collector sees."""
+    import os
+    import time
+
+    files = tmp_path / "files"
+    files.mkdir()
+    fresh = files / f"{plans.TMP_PREFIX}fresh"
+    old = files / f"{plans.TMP_PREFIX}old"
+    keep = files / "notes.txt"
+    for f in (fresh, old, keep):
+        f.write_bytes(b"x")
+    t = time.time() - const.PLAN_ORPHAN_TTL_S - 60
+    os.utime(old, (t, t))
+
+    assert plans.sweep_upload_temps(files) == 1
+    assert not old.exists(), "an aged temporary goes"
+    assert fresh.is_file(), "a fresh one may belong to a request in flight"
+    assert keep.is_file(), "nothing else is touched"
+    assert plans.sweep_upload_temps(tmp_path / "nope") == 0
 
 
 def _acfg(*pairs):