mirror of
https://github.com/kuhyx/testsAndMisc.git
synced 2026-11-10 18:07:06 +01:00
Krzysztof kuhy Rudnicki
cef72fbab0
TachiyomiSY 1.13.2 can report a restore as SUCCESS while writing almost nothing, then sync that near-empty library up and replace the only good copy. On 2026-08-09 a 14 MB payload holding 2182 manga became a 267 KB stub, and the real library survived only as stale pages in the SQLite write-ahead log. The upstream bug is Kotlin in an APK and cannot be fixed here, so this targets the fixable half: decode the payload, count what is actually in it, and refuse to treat a collapse as normal drift. A pure is-it-zero check would have caught this incident but not the harder case upstream describes, where a restore keeps some entries and drops the rest, so the threshold is proportional. Categories get their own rule because all 2182 manga survived the round trip and all 18 categories did not. Adds the setup script and a runbook covering the WAL-page salvage used to recover the library on the day. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YQLwdquDCNmWtvgi3Lk72J
153 lines
4.1 KiB
Python
153 lines
4.1 KiB
Python
"""Tests for the collapse-detection rules.
|
|
|
|
Anchored on the real 2026-08-09 numbers so the thresholds are judged against
|
|
what actually happened, not against invented ones.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pytest
|
|
|
|
from python_pkg.syncyomi_guard.payload import PayloadStats
|
|
from python_pkg.syncyomi_guard.verdict import Status, Thresholds, compare
|
|
|
|
# The library as it stood before the failed restore.
|
|
_GOOD = PayloadStats(
|
|
size_bytes=14_083_674,
|
|
manga=2182,
|
|
categories=18,
|
|
chapters=75_376,
|
|
sources=24,
|
|
)
|
|
|
|
# What the "successful" restore pushed to the server at 22:53.
|
|
_STUB = PayloadStats(
|
|
size_bytes=267_737,
|
|
manga=0,
|
|
categories=0,
|
|
chapters=0,
|
|
sources=24,
|
|
)
|
|
|
|
# After the good restore: manga intact, duplicate chapters collapsed,
|
|
# categories silently gone.
|
|
_DEDUPED = PayloadStats(
|
|
size_bytes=12_277_887,
|
|
manga=2182,
|
|
categories=0,
|
|
chapters=68_418,
|
|
sources=24,
|
|
)
|
|
|
|
|
|
def test_the_real_incident_is_reported_as_collapsed() -> None:
|
|
verdict = compare(_STUB, _GOOD)
|
|
assert verdict.status is Status.COLLAPSED
|
|
assert verdict.is_failure
|
|
assert "manga went from 2182 to 0" in verdict.reason
|
|
|
|
|
|
def test_category_only_loss_is_still_a_collapse() -> None:
|
|
"""The subtle half of the incident.
|
|
|
|
Every manga and every distinct chapter survived; only the categories went.
|
|
A manga-only check would have called this a clean sync.
|
|
"""
|
|
verdict = compare(_DEDUPED, _GOOD)
|
|
assert verdict.is_failure
|
|
assert "categories went from 18 to 0" in verdict.reason
|
|
|
|
|
|
def test_chapter_dedup_alone_does_not_trip_the_guard() -> None:
|
|
"""68418 vs 75376 is a 9.2 % drop caused by removing duplicates.
|
|
|
|
Real event: the restore collapsed 6958 duplicate chapter entries while the
|
|
distinct chapter count was unchanged. Failing here would make the guard cry
|
|
wolf on a legitimate cleanup.
|
|
"""
|
|
previous = PayloadStats(
|
|
size_bytes=14_083_674,
|
|
manga=2182,
|
|
categories=18,
|
|
chapters=75_376,
|
|
sources=24,
|
|
)
|
|
current = PayloadStats(
|
|
size_bytes=12_277_887,
|
|
manga=2182,
|
|
categories=18,
|
|
chapters=68_418,
|
|
sources=24,
|
|
)
|
|
assert compare(current, previous).status is Status.OK
|
|
|
|
|
|
def test_growth_is_healthy() -> None:
|
|
bigger = PayloadStats(
|
|
size_bytes=15_000_000,
|
|
manga=2200,
|
|
categories=19,
|
|
chapters=76_000,
|
|
sources=24,
|
|
)
|
|
assert compare(bigger, _GOOD).status is Status.OK
|
|
|
|
|
|
def test_small_drop_is_tolerated() -> None:
|
|
"""Finishing and removing a few series is normal."""
|
|
smaller = PayloadStats(
|
|
size_bytes=14_000_000,
|
|
manga=2100,
|
|
categories=18,
|
|
chapters=75_000,
|
|
sources=24,
|
|
)
|
|
assert compare(smaller, _GOOD).status is Status.OK
|
|
|
|
|
|
def test_drop_past_the_threshold_fails() -> None:
|
|
halved = PayloadStats(
|
|
size_bytes=7_000_000,
|
|
manga=1000,
|
|
categories=18,
|
|
chapters=40_000,
|
|
sources=24,
|
|
)
|
|
verdict = compare(halved, _GOOD)
|
|
assert verdict.is_failure
|
|
assert "manga dropped" in verdict.reason
|
|
|
|
|
|
def test_missing_baseline_is_first_run_not_success() -> None:
|
|
"""A lost state file must not silently turn the guard into a no-op."""
|
|
verdict = compare(_GOOD, None)
|
|
assert verdict.status is Status.FIRST_RUN
|
|
assert not verdict.is_failure
|
|
assert verdict.previous is None
|
|
|
|
|
|
def test_growth_from_a_zero_baseline_is_not_a_drop() -> None:
|
|
assert compare(_GOOD, _STUB).status is Status.OK
|
|
|
|
|
|
def test_threshold_is_configurable() -> None:
|
|
slightly_smaller = PayloadStats(
|
|
size_bytes=13_000_000,
|
|
manga=2100,
|
|
categories=18,
|
|
chapters=75_000,
|
|
sources=24,
|
|
)
|
|
strict = Thresholds(max_drop_ratio=0.01)
|
|
assert compare(slightly_smaller, _GOOD, strict).is_failure
|
|
|
|
|
|
@pytest.mark.parametrize("ratio", [0.0, 1.0, -0.5, 2.0])
|
|
def test_nonsense_thresholds_are_rejected(ratio: float) -> None:
|
|
with pytest.raises(ValueError, match="max_drop_ratio"):
|
|
Thresholds(max_drop_ratio=ratio)
|
|
|
|
|
|
def test_reason_describes_a_healthy_payload() -> None:
|
|
assert "2182 manga" in compare(_GOOD, _GOOD).reason
|