soulsync/tests/test_soulseek_album_poll_stall.py
BoulderBadgeDad a60eae9315 Soulseek album poll: treat 'Aborted'/'Cancelled' transfers as failed
Live testing surfaced that slskd reports a peer-side abort as 'Completed,
Aborted' at 0 bytes (peer accepts then drops every transfer). That string
contains 'Completed', so the poll's completed-branch ran first and misread it as
'completed but file missing' — routing it into the #715 unresolved/download_path
grace (gives up after 45s with a misleading 'download_path mismatch' log)
instead of recognizing it as a failure.

Add 'Aborted' and 'Cancelled' to the failure-token check (which runs before the
completed branch), so these resolve immediately and correctly as failed. Test
added for the all-aborted folder.
2026-05-29 19:29:55 -07:00

162 lines
6.7 KiB
Python

"""Regression for the Soulseek album-bundle poll hanging when the peer stalls.
`_poll_album_bundle_downloads` waits for every transfer in the selected folder
to reach a terminal state (completed or failed). A transfer the peer stalls on —
stuck InProgress/Queued, or dropped by slskd — is never failed, never completed,
and never marked "completed-but-unresolved", so it used to block both the
all-terminal finish check AND the #715 grace exit, and the poll spun to the full
~6h timeout (the Slipknot hang).
The fix adds a bundle-level stall guard: if NOTHING progresses (no transfer
reaches terminal AND no pending transfer's byte count moves) for `_stall_grace`
seconds, the stuck transfers are marked failed so the bundle resolves with what
completed. These tests drive the real poll with a fake clock + scripted slskd
states.
"""
from __future__ import annotations
import os
from pathlib import Path
from types import SimpleNamespace
from unittest.mock import patch
from core.soulseek_client import SoulseekClient
class _Clock:
def __init__(self):
self.now = 0.0
def monotonic(self):
return self.now
def sleep(self, seconds):
self.now += seconds
def _dl(username, filename, state, *, size=100, transferred=100):
return SimpleNamespace(
username=username, filename=filename, state=state,
size=size, transferred=transferred,
)
class _StubClient:
"""Minimal stand-in for SoulseekClient with just what the poll touches."""
def __init__(self, states, resolvable):
self._states = states # list of download-lists, one per poll; last repeats
self._call = 0
self._resolvable = set(resolvable)
def get_all_downloads(self):
i = min(self._call, len(self._states) - 1)
self._call += 1
return self._states[i]
def _resolve_downloaded_album_file(self, filename):
base = os.path.basename((filename or "").replace("\\", "/"))
if base in self._resolvable or filename in self._resolvable:
return Path(f"/staged/{base}")
return None
def _run_poll(stub, transfer_keys, *, timeout=7200.0, interval=2.0):
clock = _Clock()
emits = []
with patch("core.soulseek_client.time", clock), \
patch("core.soulseek_client.run_async", lambda x: x), \
patch("core.soulseek_client.get_poll_timeout", lambda: timeout), \
patch("core.soulseek_client.get_poll_interval", lambda: interval):
result = SoulseekClient._poll_album_bundle_downloads(
stub, transfer_keys, lambda phase, **kw: emits.append((phase, kw))
)
return result, clock, emits
def _keys(*names, user="peer"):
"""Build {(user, name): TrackResult-ish} preserving order."""
return {(user, n): SimpleNamespace(filename=n) for n in names}
def test_stalled_peer_gives_up_and_returns_completed_subset():
tk = _keys("01.flac", "02.flac", "03.flac")
# 01 completes (resolvable); 02/03 stuck InProgress with FROZEN byte counts.
frozen = [
_dl("peer", "01.flac", "Completed", transferred=100, size=100),
_dl("peer", "02.flac", "InProgress", transferred=50, size=100),
_dl("peer", "03.flac", "InProgress", transferred=30, size=100),
]
stub = _StubClient([frozen], resolvable={"01.flac"})
result, clock, _ = _run_poll(stub, tk, timeout=7200.0, interval=2.0)
# Resolved with the one completed track instead of hanging to the deadline.
assert result == [Path("/staged/01.flac")]
# Gave up around the stall window (~180s), nowhere near the 7200s timeout.
assert clock.now < 600.0
def test_progressing_bundle_is_not_falsely_stalled():
tk = _keys("01.flac", "02.flac")
# 02 keeps downloading more bytes each poll, then completes — must NOT trip
# the stall guard even though it takes a while.
states = [
[_dl("peer", "01.flac", "Completed"), _dl("peer", "02.flac", "InProgress", transferred=10, size=100)],
[_dl("peer", "01.flac", "Completed"), _dl("peer", "02.flac", "InProgress", transferred=40, size=100)],
[_dl("peer", "01.flac", "Completed"), _dl("peer", "02.flac", "InProgress", transferred=80, size=100)],
[_dl("peer", "01.flac", "Completed"), _dl("peer", "02.flac", "Completed", transferred=100, size=100)],
]
stub = _StubClient(states, resolvable={"01.flac", "02.flac"})
result, _clock, _ = _run_poll(stub, tk, timeout=7200.0, interval=2.0)
assert set(result) == {Path("/staged/01.flac"), Path("/staged/02.flac")}
def test_all_transfers_stalled_returns_empty():
tk = _keys("01.flac", "02.flac")
frozen = [
_dl("peer", "01.flac", "InProgress", transferred=10, size=100),
_dl("peer", "02.flac", "Queued", transferred=0, size=100),
]
stub = _StubClient([frozen], resolvable=set())
result, clock, _ = _run_poll(stub, tk, timeout=7200.0, interval=2.0)
assert result == [] # nothing completed → empty (caller falls back)
assert clock.now < 600.0 # didn't spin to the deadline
def test_dropped_transfers_also_stall_out():
"""slskd dropping the transfers entirely (dl=None) must also trip the guard,
not hang — there's no byte progress and nothing terminal."""
tk = _keys("01.flac", "02.flac")
stub = _StubClient([[]], resolvable=set()) # get_all_downloads returns nothing
result, clock, _ = _run_poll(stub, tk, timeout=7200.0, interval=2.0)
assert result == []
assert clock.now < 600.0
def test_completed_aborted_classified_as_failed_not_unresolved():
"""slskd reports a peer-side abort as 'Completed, Aborted' at 0 bytes. Because
that string contains 'Completed', it was misread as 'completed but file
missing' (#715 path). It must be classified as FAILED — so an all-aborted
folder resolves immediately, not after the unresolved/stall grace."""
tk = _keys("01.flac", "02.flac")
aborted = [
_dl("peer", "01.flac", "Completed, Aborted", transferred=0, size=3_600_000),
_dl("peer", "02.flac", "Completed, Aborted", transferred=0, size=24_700_000),
]
stub = _StubClient([aborted], resolvable=set())
result, clock, _ = _run_poll(stub, tk, timeout=7200.0, interval=2.0)
assert result == [] # all failed → empty (caller falls back)
assert clock.now < 30.0 # resolved on the first poll, no 45s/180s wait
def test_clean_finish_unaffected():
tk = _keys("01.flac", "02.flac")
done = [_dl("peer", "01.flac", "Completed"), _dl("peer", "02.flac", "Succeeded")]
stub = _StubClient([done], resolvable={"01.flac", "02.flac"})
result, clock, _ = _run_poll(stub, tk)
assert set(result) == {Path("/staged/01.flac"), Path("/staged/02.flac")}
assert clock.now < 10.0 # resolves on the first couple polls