feat(career): extract the whole setlist before the gig starts (no more waiting between songs) (#971)
ship-ci / ci (push) Waiting to run

* feat(career): extract the whole setlist before the gig starts

A feedpak is a zip, and the first play of one pays for its extraction into
sloppak_cache. Inside a set that cost landed BETWEEN songs: the player finished
a number and then sat there waiting for the next one to unpack, mid-gig.

A setlist is a known list up front, so unpack it all while the poster is still on
screen. New POST /gigs/prepare walks the set through resolve_source_dir; the
poster's Play button shows "Preparing set…" while it runs.

Best-effort by design, at every level:
  - a corrupt pak in the set does not sink the prepare (it is reported in
    `failed`; the play itself surfaces the error exactly as it does outside a
    gig — slow beats blocked)
  - a host without the library resolvers degrades to a no-op rather than 500
  - a failed request just falls through to the old lazy extraction

Ordering matters and is pinned: the set is unpacked BEFORE the stage is borrowed
(venue/viz overwritten) and before the queue starts, so a proposal cancelled
while unpacking leaves nothing half-applied to unwind.

Tests unpack REAL zips rather than mocking the extractor: every song of the set
lands on disk before the first note, a re-prepare does not duplicate the unpack,
one bad pak still leaves the good one prepared, and no-library / empty-setlist
degrade cleanly. 18/18.

NB the other half of the gig report — the per-song results popup interrupting
the set (and worse, claimAutoExit'ing so the queue would not advance until it was
dismissed) — is fixed in the note_detect plugin repo, which is not part of this
checkout.

* fix(career): bound the prepare request; validate the setlist (PR #971 review)

Both CodeRabbit findings were right.

1. A HUNG PREPARE COULD BLOCK THE GIG FOREVER.

   `await fetch(...)` only rejects on a network ERROR. A server that accepts the
   connection and then never answers hangs indefinitely — and the gig would never
   start. That makes this optimisation the exact thing the PR promises it can
   never be: the reason you cannot play.

   The request is now bounded by an AbortController (PREPARE_TIMEOUT_MS, generous
   because unpacking a setlist is real work — but a CEILING, not a wait). Past it
   we start the gig and let the first play extract lazily, as it always did. The
   Play button is restored in a `finally`, so a timeout cannot strand the poster
   on "Preparing set…" with Play disabled — which would have been the same bug
   wearing a different hat.

2. THE `songs` BODY WAS UNVALIDATED.

   A str is iterable: "abc" would have prepared three one-character "songs". And
   the endpoint unpacks zips, so an arbitrary caller could ask for unbounded work.
   Now list-only, string entries, blanks dropped, capped at MAX_GIG_SONGS.

Tests: the fetch is abortable and the button is re-enabled on EVERY path
including the abort; non-list bodies, non-string/blank entries, and an
oversized setlist. 50 career tests, JS 5/5, eslint clean.

* fix(career): path-traversal guard on prepare; a cap test that actually tests the cap

CodeRabbit again, and the first one is a real hole I put there.

1. PATH TRAVERSAL. sloppak.resolve_source_dir() does a bare `dlc_root / filename`
   with NO containment guard — so `../../x` walks straight out of the library, and
   my new endpoint handed it attacker-supplied filenames. Every filename now goes
   through _resolve_dlc_path first, the same check every other filename-bound
   handler applies. Pinned: `..`, backslash traversal, an absolute POSIX path and
   a Windows drive path are all refused, and nothing outside the library is
   unpacked.

2. THE CAP TEST WAS VACUOUS. It asserted `prepared == 0` against a fixture with no
   library — where the endpoint exits before extraction — so it passed whether or
   not MAX_GIG_SONGS existed. It now runs against a real library and asserts the
   endpoint CONSIDERED at most MAX_GIG_SONGS of the 82 it was handed. Verified to
   fail when the cap is removed.

   Same class of mistake as the notedetect gigBlock: a test that passes for the
   wrong reason. Worth saying out loud since it is twice in one day.

3. E702 — semicolon-joined statements in the new tests, split.

51 career tests; full suite green.
This commit is contained in:
Byron Gamatos
2026-07-15 00:36:20 +02:00
committed by GitHub
parent 917d81c2d2
commit 1702afa379
4 changed files with 380 additions and 2 deletions
+98
View File
@@ -0,0 +1,98 @@
// A gig is a SET, not a run of unrelated songs.
//
// Reported from a live gig: the player finished the first song and had to sit
// through the per-song results popup before the next one would start, and then
// wait again while that song was extracted from its feedpak zip.
//
// This file covers the CORE half — career pre-extracts the whole setlist before
// the first note. The other half (note_detect must not show its per-song summary
// inside a gig) lives in the note_detect plugin repo, which is not part of this
// checkout: plugins/*/ is gitignored here and note_detect ships from
// feedBack-plugin-notedetect. A test reading it from core would pass on a dev
// box (where the plugin happens to be bundled) and fail in CI, which is worse
// than no test.
//
// The pre-extraction is tested for REAL behaviour — actually unpacking zips — in
// tests/plugins/career/test_routes.py. These are the wiring guards around it.
'use strict';
const { test } = require('node:test');
const assert = require('node:assert/strict');
const fs = require('node:fs');
const path = require('node:path');
const ROOT = path.join(__dirname, '..', '..');
const CAREER = fs.readFileSync(path.join(ROOT, 'plugins', 'career', 'screen.js'), 'utf8');
const CAREER_ROUTES = fs.readFileSync(path.join(ROOT, 'plugins', 'career', 'routes.py'), 'utf8');
function extractBlock(src, signature) {
const start = src.indexOf(signature);
assert.ok(start !== -1, `signature '${signature}' not found`);
const openBrace = src.indexOf('{', start);
let depth = 1;
let i = openBrace + 1;
while (i < src.length && depth > 0) {
const ch = src[i];
if (ch === '{') depth++;
else if (ch === '}') depth--;
i++;
}
assert.ok(depth === 0, `unbalanced braces after '${signature}'`);
return src.slice(start, i);
}
test('startGig extracts the whole setlist before starting the queue', () => {
const fn = extractBlock(CAREER, 'async function startGig(');
const prepIdx = fn.search(/await\s+prepareGigSongs\s*\(/);
const startIdx = fn.search(/q\.start\s*\(/);
assert.ok(prepIdx !== -1, 'startGig must pre-extract the set');
assert.ok(startIdx !== -1, 'q.start not found');
assert.ok(prepIdx < startIdx,
'the set must be unpacked BEFORE the queue starts — otherwise the player ' +
'waits between songs, which is the bug');
});
test('the stage is only borrowed once the set is ready', () => {
const fn = extractBlock(CAREER, 'async function startGig(');
const prepIdx = fn.search(/await\s+prepareGigSongs\s*\(/);
const stageIdx = fn.search(/VENUE_OVERRIDE_KEY/);
assert.ok(prepIdx < stageIdx,
'a gig cancelled while unpacking must not leave the venue/viz overwritten');
assert.match(fn, /_ppGigProposal\s*!==\s*prop/,
'a proposal dismissed while unpacking must not then start a gig');
});
test('pre-extraction never blocks the gig from starting', () => {
const fn = extractBlock(CAREER, 'async function prepareGigSongs(');
assert.match(fn, /catch\s*\(/,
'a failed prepare must fall through to the old lazy extraction, not abort the gig');
});
test('the prepare route degrades instead of failing', () => {
assert.match(CAREER_ROUTES, /def prepare_gig/, 'prepare route missing');
assert.match(CAREER_ROUTES, /context\.get\(\s*["']get_dlc_dir["']\s*\)/,
'a host without the library resolvers must degrade, not 500 — pre-extraction ' +
'is an optimisation and can never be why a gig will not start');
});
// ── the prepare must never be able to BLOCK the gig (CodeRabbit, #971) ──────
//
// A bare `await fetch(...)` only rejects on a network error. A server that
// accepts the connection and then never answers hangs forever — and the gig
// would never start. That would make this optimisation the exact thing it
// promises never to be: the reason you cannot play.
test('the prepare fetch is bounded — a hung server cannot block the gig', () => {
const fn = extractBlock(CAREER, 'async function prepareGigSongs(');
assert.match(fn, /AbortController/, 'the request must be abortable');
assert.match(fn, /setTimeout\([\s\S]{0,40}abort\s*\(\s*\)/,
'a hung request must be aborted, not awaited forever');
assert.match(fn, /signal:\s*ctrl\.signal/, 'the signal must actually be passed to fetch');
assert.match(fn, /clearTimeout/, 'the timer must be cleared on the happy path');
assert.match(CAREER, /const\s+PREPARE_TIMEOUT_MS\s*=\s*\d+/, 'the ceiling must be named');
// The button must be restored however we leave — otherwise a timeout strands
// the poster on "Preparing set…" with Play disabled: unplayable.
assert.match(fn, /finally\s*\{[\s\S]{0,220}btn\.disabled\s*=\s*false/,
'the Play button must be re-enabled on EVERY path, including the abort');
});
+173
View File
@@ -174,3 +174,176 @@ def test_double_download_409s(client, monkeypatch):
career_routes._state["downloads"]["bar"] = {"status": "running"}
assert client.post("/api/plugins/career/packs/bar/download").status_code == 409
assert client.delete("/api/plugins/career/packs/bar").status_code == 409
# ── gig pre-extraction (the wait between songs) ─────────────────────────────
#
# A feedpak is a zip: the first play of one pays for its extraction into
# sloppak_cache. Inside a set that cost landed BETWEEN songs — the player
# finished a number and then sat waiting for the next one to unpack, mid-gig.
# The setlist is known up front, so extract it all while the poster is up.
def _career_client_with_library(tmp_path, meta_db, dlc, cache):
from fastapi import FastAPI
from fastapi.testclient import TestClient
import routes as career_routes
app = FastAPI()
career_routes.setup(app, {
"config_dir": str(tmp_path),
"meta_db": meta_db,
"get_dlc_dir": lambda: dlc,
"get_sloppak_cache_dir": lambda: cache,
})
return TestClient(app)
def _write_feedpak(dlc, name, title="T"):
"""A minimal but REAL feedpak zip, so resolve_source_dir genuinely unpacks."""
import json as _json
import zipfile as _zip
p = dlc / name
with _zip.ZipFile(p, "w") as z:
z.writestr("manifest.json", _json.dumps({"title": title, "artist": "A", "arrangements": []}))
return p
def test_gig_prepare_extracts_every_song_up_front(tmp_path, meta_db):
dlc = tmp_path / "dlc"
dlc.mkdir()
cache = tmp_path / "cache"
cache.mkdir()
for n in ("one.feedpak", "two.feedpak", "three.feedpak"):
_write_feedpak(dlc, n)
client = _career_client_with_library(tmp_path, meta_db, dlc, cache)
before = list(cache.iterdir())
assert before == [], "nothing unpacked yet"
res = client.post("/api/plugins/career/gigs/prepare",
json={"songs": ["one.feedpak", "two.feedpak", "three.feedpak"]})
assert res.status_code == 200
body = res.json()
assert body["ok"] is True
assert body["prepared"] == 3, body
assert body["failed"] == []
# The point of the whole exercise: the set is on disk BEFORE the first note.
assert len(list(cache.iterdir())) == 3, "every song of the set must be unpacked"
def test_gig_prepare_is_idempotent_on_a_warm_cache(tmp_path, meta_db):
dlc = tmp_path / "dlc"
dlc.mkdir()
cache = tmp_path / "cache"
cache.mkdir()
_write_feedpak(dlc, "one.feedpak")
client = _career_client_with_library(tmp_path, meta_db, dlc, cache)
first = client.post("/api/plugins/career/gigs/prepare", json={"songs": ["one.feedpak"]}).json()
second = client.post("/api/plugins/career/gigs/prepare", json={"songs": ["one.feedpak"]}).json()
assert first["prepared"] == second["prepared"] == 1
assert len(list(cache.iterdir())) == 1, "a re-prepare must not duplicate the unpack"
def test_one_bad_feedpak_does_not_stop_the_set(tmp_path, meta_db):
# A corrupt pak in the setlist must not block the gig: the play itself will
# surface the error exactly as it does outside a gig. Slow beats blocked.
dlc = tmp_path / "dlc"
dlc.mkdir()
cache = tmp_path / "cache"
cache.mkdir()
_write_feedpak(dlc, "good.feedpak")
(dlc / "bad.feedpak").write_bytes(b"not a zip at all")
client = _career_client_with_library(tmp_path, meta_db, dlc, cache)
body = client.post("/api/plugins/career/gigs/prepare",
json={"songs": ["good.feedpak", "bad.feedpak"]}).json()
assert body["ok"] is True, "a bad pak must not fail the whole prepare"
assert body["prepared"] == 1
assert body["failed"] == ["bad.feedpak"]
def test_gig_prepare_degrades_without_a_library(tmp_path, meta_db, client):
# The stock fixture's context has no dlc/cache resolvers. That must be a
# graceful no-op, not a 500 — pre-extraction is an optimisation and can
# never be the reason a gig won't start.
res = client.post("/api/plugins/career/gigs/prepare", json={"songs": ["x.feedpak"]})
assert res.status_code == 200
assert res.json()["prepared"] == 0
def test_gig_prepare_empty_setlist(tmp_path, meta_db, client):
res = client.post("/api/plugins/career/gigs/prepare", json={"songs": []})
assert res.status_code == 200
assert res.json() == {"ok": True, "prepared": 0, "failed": []}
def test_prepare_rejects_a_non_list_songs_value(tmp_path, meta_db, client):
# A str is iterable: without the list check, "abc" would prepare three
# one-character "songs".
for bad in ("abc", 42, {"a": 1}, None):
res = client.post("/api/plugins/career/gigs/prepare", json={"songs": bad})
assert res.status_code == 200
assert res.json()["prepared"] == 0
def test_prepare_ignores_non_string_and_blank_entries(tmp_path, meta_db):
dlc = tmp_path / "dlc"
dlc.mkdir()
cache = tmp_path / "cache"
cache.mkdir()
_write_feedpak(dlc, "good.feedpak")
client = _career_client_with_library(tmp_path, meta_db, dlc, cache)
body = client.post("/api/plugins/career/gigs/prepare",
json={"songs": ["good.feedpak", "", " ", 7, None, {"x": 1}]}).json()
assert body["prepared"] == 1
assert body["failed"] == []
def test_prepare_caps_the_setlist(tmp_path, meta_db):
# This endpoint unpacks zips — an arbitrary caller must not be able to ask for
# unbounded work.
#
# The first version of this test asserted `prepared == 0` against a fixture
# with NO library: the endpoint exits before extraction there, so it passed
# whether or not the cap existed. Give it a real library, ask for far more than
# the cap, and assert the endpoint only ever considered MAX_GIG_SONGS of them.
import routes as career_routes
assert career_routes.MAX_GIG_SONGS <= 64
dlc = tmp_path / "dlc"
dlc.mkdir()
cache = tmp_path / "cache"
cache.mkdir()
client = _career_client_with_library(tmp_path, meta_db, dlc, cache)
n = career_routes.MAX_GIG_SONGS + 50
# None of these exist, so every song the endpoint LOOKS AT lands in `failed`.
# That makes `failed` an exact count of how many it considered.
body = client.post("/api/plugins/career/gigs/prepare",
json={"songs": [f"missing{i}.feedpak" for i in range(n)]}).json()
assert body["prepared"] == 0
assert len(body["failed"]) == career_routes.MAX_GIG_SONGS, (
f"the endpoint must consider at most MAX_GIG_SONGS "
f"({career_routes.MAX_GIG_SONGS}), not all {n}"
)
def test_prepare_refuses_to_escape_the_library(tmp_path, meta_db):
# resolve_source_dir() does a bare `dlc_root / filename` with no containment
# guard, so a crafted path would walk straight out of the library. Every
# filename must go through _resolve_dlc_path first.
dlc = tmp_path / "dlc"
dlc.mkdir()
cache = tmp_path / "cache"
cache.mkdir()
(tmp_path / "outside.feedpak").write_bytes(b"secret")
client = _career_client_with_library(tmp_path, meta_db, dlc, cache)
for evil in ("../outside.feedpak", "..\\outside.feedpak",
"a/../../outside.feedpak", "/etc/passwd", "C:/Windows/x.feedpak"):
body = client.post("/api/plugins/career/gigs/prepare",
json={"songs": [evil]}).json()
assert body["prepared"] == 0, f"{evil!r} must never be prepared"
assert body["failed"] == [evil]
# Nothing outside the library may have been unpacked.
assert list(cache.iterdir()) == []