Window is [begin, end) — exclusive upper bound. Ordered priority-DESC
then Id-ASC so the controller can break on the first match.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Phase 3 shipped the AI rank battle bot pool as a hardcoded 8-entry list
inlined in SVSim.EmulatedEntrypoint/Matching/BotRoster.cs — editing meant
recompiling. Per PLAN.md 2026-06-02 item (d), move it to a Bootstrap
importer so the roster lives in seeds/bot-roster.json and the DB.
Shape mirrors PracticeOpponent end-to-end:
- BotRosterEntry (SVSim.Database/Models) — PK = AiId via the Id passthrough
pattern. DbSet<BotRosterEntry> BotRoster on SVSimDbContext.
- AddBotRoster migration (DDL only, per migrations-are-DDL-only rule).
- seeds/bot-roster.json — 8 rows preserving the current prod-verified
cosmetic ids (sleeve 704141010 / emblem 400001100 / degree 120027 /
field 5) and series-1 ai_ids from rm_ai_setting.csv (1111..1181).
- BotRosterSeed POCO + BotRosterImporter (idempotent upsert keyed by AiId,
leaves seed-missing rows intact). Wired into SVSim.Bootstrap/Program.cs
next to PracticeOpponentImporter.
- IGlobalsRepository.GetBotRoster() + impl.
IBotRoster.Pick → PickAsync because BotRoster now depends on the transient
IGlobalsRepository. RankBattleController awaits the new signature. The
deterministic hash-on-ctx invariant (same ctx → same bot, so /ai_<fmt>/start
retries pick the same opponent) is preserved.
DI: AddSingleton<IBotRoster> → AddTransient (matches IGlobalsRepository's
lifetime). Test fixture's SeedGlobalsAsync also runs the importer so
RankBattleControllerTests + the rewritten BotRosterTests both see seeded
rows.
Tests: 931 → 936 passing. Existing 3 BotRosterTests reshaped for the DB
backing + 1 new "throws on empty roster" guard; 4 new
BotRosterImporterTests mirror PracticeOpponentImporterTests
(round-trip / idempotent / seed-missing-row-intact / ai_id=0 skip).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Focused AsNoTracking load with Info.SelectedEmblem/SelectedDegree includes
for the new MatchContextBuilder. Single test locks the include graph.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
GameStart already detects the Steam-vs-UDID mismatch produced by
wipe-and-resignup; it now also reclaims the orphan. New
ViewerRepository.MergeAnonymousViewerInto transfers the fresh UDID
from V_new onto V_old in one save (freeing the unique-index slot),
then deletes V_new in a second save. Partial-failure mode is a
benign null-UDID viewer; two rows never contend for the same UDID.
Side benefit: future GetViewerByUdid lookups now short-circuit to
V_old without going through the Steam handler.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
BattlePassRepository._curveCache and MissionCatalogRepository._maxLevelCache
were private-static fields populated lazily on first read from whatever
DbContext happened to be in scope. In production "one DbContext lineage
per process" makes that fine. Under parallel test execution each
SVSimTestFactory owns its own SQLite :memory: DB, so the first reader's
DB (often empty, in tests that don't seed BP) poisoned the cache for
concurrent readers from a seeded DB — assertions like "BP level info
must be present after seeding" failed because the process-static cache
returned an empty list populated by the other test's empty DB.
The first patch attempted a `BypassCacheForTests` static flag, which is
exactly the kind of test-only seam that rots the production code: future
caches get the same flag, repos accumulate hidden knobs, and the
underlying invariant ("a cache populated from arbitrary scope serves
arbitrary scope") goes unaddressed.
Instead, move both caches into the DI-registered IMemoryCache.
AddMemoryCache() registers it as singleton-per-service-provider:
production has one provider → one IMemoryCache → identical caching
semantics to before. Each WebApplicationFactory builds its own
provider → its own IMemoryCache → cache is naturally scoped per fixture,
no cross-test bleed possible.
The ResetLevelCurveCache() method and its three call sites
(SVSimTestFactory.SeedGlobalsAsync, BattlePassServiceTests,
LoadControllerTests) are deleted — a fresh factory owns a fresh empty
cache, no manual invalidation needed.
With this and the previous StoryService fixture-instance fix in place,
ParallelScope.All works: 776/776 in 57s wall clock (down from 59s on
Fixtures, 2m13s pre-parallelism).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
RedEther debit now goes through tx.TrySpendAsync (freeplay-aware);
Card grants route through tx.GrantAsync (cosmetic cascade for first-time
owners). Validation phase unchanged. DestructCards left on direct-viewer
path (structural mismatch: validation on one viewer, mutation on same
instance — clean tx port deferred to follow-up).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds GetRewardsByWinCountAsync and GetMaxWinCountAsync (short-circuits
to 0 on empty table). Registers as AddTransient in Program.cs alongside
other global catalog repos. 3 NUnit tests pass.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Mirror of the outer-repo data_dumps/ reorganization (commit e1e595d in
the SVSim outer repo): updates all data_dumps/extract/ → data_dumps/scripts/,
data_dumps/client_master_csv → data_dumps/client-assets, data_dumps/traffic
→ data_dumps/captures/traffic in XML doc-comments and inline comments
across importers, controllers, middlewares, DTOs, and tests. Doc-only;
no logic changes; build green.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Aggregate (Config + SlotRates + CardWeights) and a single-pack getter
loaded as one unit per /pack/open. PackOpenService consumes the
aggregate; tests use the production seed (fixture overlay) to validate
shape.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
PackImporter now runs a second pass over pack-stubs.json, inserting
PackConfigEntry placeholders for any pack_id NOT already present from
the live-capture packs.json pass. Synthesized stubs default
IsEnabled=false; live-capture rows default IsEnabled=true.
PackRepository.GetActivePacks filters by IsEnabled in addition to the
date window, so synthesized stubs stay hidden until an operator opts
them in (UPDATE Packs SET IsEnabled=true WHERE Id=...).
Bundles Task 6 + Task 11 because adding pack-stubs.json to the
test-fixture set surfaces an extra row in PackControllerFullCatalogTests'
35-pack count assertion; the filter is what makes the test resilient.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds StoryDeckView projection, IBuildDeckRepository.GetStoryDecksByClass interface method,
and BuildDeckRepository implementation that loads StoryDeckEntry rows for a class, fetches
matching BuildDeckProductEntry card lists, and expands each card by Number into a flat
CardIdArray. TDD: 2 tests in StoryDeckRepositoryTests (expand + empty-class).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Verified against Wizard.Title/UserNameInput.cs:30 in the 2026-05-23
decompile:
IsFinished = !string.IsNullOrEmpty(PlayerStaticData.UserName);
Any non-empty seeded value — including the prior " - " placeholder
this method was passing — sets IsFinished=true on the first frame and
silently skips both the input dialog and the /tutorial/update_action #1
+ /account/update_name calls that travel with it. The in-source comment
described the opposite behavior; empty string is what actually triggers
the dialog.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
BuildDefaultViewer hardcoded TutorialState=1 — correct for fresh anonymous
signups (RegisterAnonymousViewer) but wrong for AdminController.ImportViewer
and Steam-social signups, which both go through RegisterViewer and expect a
prod-replica viewer that boots to the home screen. Add an initialTutorialState
parameter (default 1 preserves RegisterAnonymousViewer behavior); RegisterViewer
passes 100.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Four targeted fixes that together let /tutorial/pack_info display
the legendary starter at index 0, let /tutorial/pack_open succeed
on it, and let the pack drop out of the shop after.
1. /pack/info now loads viewer.Items into a Dictionary<long,int>
and threads it through ToDto so child_gacha_info.item_number
reflects the viewer's actual owned count of item_id. Previously
defaulted to 0 for every pack, so the legendary pack 99047
reported item_number=0 immediately after the gift granted 1×
ticket id=90001. Verified against the prod tutorial capture.
2. PackRepository.GetActivePacks now orders parent_gacha_id DESC
to match prod's /pack/info wire order (99047, 92001, 80047,
16015...10001). The tutorial pack UI runs with controls locked
and auto-selects index 0 via GachaUI.GetCurrentLegendPackId
(FirstOrDefault on IsLegendPackId), so the legendary starter
needs to be the first legend pack in the list.
3. DbCardPoolProvider.GetPool falls back to all in-rotation cards
when a LegendCardPack's base set has no rows. Pack 99047's
base_pack_id is 90001, a synthetic "Throwback Rotation" category
that doesn't correspond to a real card_set in the prod card
master — its real pool is curated across older rotation sets
(Altersphere through Colosseum). We don't have that membership
map captured yet; the rotation fallback is broader than prod
but produces a valid 8-card draw, which is what the tutorial
needs to advance to step 100. TODO in code points at the
real fix.
4. PackController.Open's tutorial path now consumes the granted
ticket (decrement viewer.Items by packNumber for child.ItemId)
and emits the post-state count in reward_list as
{reward_type:4, reward_id:item_id, reward_num:post_count}.
Without this, the pack stayed at item_number=1 forever, the
shop kept showing it post-tutorial, and the next click hit
/pack/open (not /tutorial/pack_open) which 501s on type_detail=5.
Also: docstring on PackConfigDto.SalesPeriodInfo flags the deferred
wire-fidelity fix (prod emits {"sales_period_time": "<complete_date>"}
for limited windows, [] for evergreens; we always emit {}) and the
retype from Dictionary<string,string?> to a typed
PackSalesPeriodInfoDto. Doesn't affect tutorial flow, deferred for
the pack-system rework.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three corrections to BuildDefaultViewer + RegisterAnonymousViewer
verified against data_dumps/traffic_prod_tutorial.ndjson:
- TutorialState now 1 (TUTORIAL_STEP0), not 0 (PRE_TUTORIAL_STEP).
Wizard.Title.NextSceneSwitcher routes step==1 to the Prologue scene;
any other non-{31,41,100} step routes to AreaSelect at section 0,
which has no chapter data and crashes the client with a LINQ
Single() "Sequence contains no matching element" from
AreaSelectUI.SelectChapter. Prod's first /check/game_start returns
now_tutorial_step="1"; step 0 is a pre-existence state we never
want to expose on the wire.
- DisplayName " - " (literal space-dash-space), not "Player".
Wizard.Title.UserNameInput.Start short-circuits with
IsFinished=true on !string.IsNullOrEmpty(PlayerStaticData.UserName),
silently skipping the name dialog AND the tutorial sub-step it
drives (/tutorial/update_action #1 + /account/update_name). Prod
uses " - " as the unset placeholder.
- viewer.Info.SelectedEmblem/SelectedDegree assigned from the default
emblem/degree the loadout grants. Without this, /load/index emits
selected_emblem_id=0 and selected_degree_id=0 for a fresh viewer
that owns those cosmetics — prod sends the real granted IDs.
Also surfaces the actual Postgres constraint name in the unique-
violation re-raise (ExtractConstraintName), instead of always
saying "UDID" — the original message was misleading whenever the
real constraint was on owned-collection rows.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Previously RegisterAnonymousViewer auto-completed the tutorial, which
prevented the client from ever entering the tutorial flow. SeedViewerAsync
gains a tutorialState parameter (default 100) so existing tests keep
their pre-completed-tutorial assumption.
Implements ICardInventoryRepository.SetProtected — loads only the
owned-cards collection (no decks/currency), guards against the EF
owned-nav Card.Id==0 default-init quirk, and accepts Count=0 rows
(destruct→re-protect round-trip). Covered by 4 new NUnit tests
(flip, unset, zero-count-row, unknown-card error). Full suite: 533/533.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
(1) RegisterAnonymousViewer now catches the unique-violation
race (SQLSTATE 23505 on Postgres / code 19 on SQLite) and
re-reads by UDID, returning the existing row instead of
surfacing 500 to the second concurrent /tool/signup caller.
New repo test exercises the back-to-back register path.
(2) Add unique index on SocialAccountConnection (AccountType,
AccountId). The auth handler's find-or-link path claimed
this index existed as the dedup backstop; the claim was
accurate as design intent but the schema was missing. Now
matched. Comment in handler updated.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Review polish on the prior commit (30874c6):
- BuildDefaultViewer extract dropped the "filter out Id=0 placeholders
and dedupe" comment from the leader-skin grant block — restored.
- LinkSteamToViewer test now calls link twice and asserts count stays
at 1, exercising the alreadyLinked short-circuit.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Extracts the default-loadout body into a private BuildDefaultViewer
helper shared by the existing Steam-import path and a new
RegisterAnonymousViewer for /tool/signup. LinkSteamToViewer is the
seam SteamSessionAuthenticationHandler will call on first-Steam-touch
of a UDID-keyed viewer.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three endpoints + 9 integration tests. Captured-data-is-catalog: viewer's
achievement Level starts at MIN(Level) per type from the catalog (not 1),
so the assembler always has a row to render against.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GetOrCreateProgressAsync now persists the new row itself and catches
DbUpdateException on unique-constraint violations — concurrent /info
calls no longer throw 500s. BattlePassService no longer calls
SaveChangesAsync after the get-or-create. FormatWireDate uses a named
JstOffset constant instead of an inline TimeSpan.FromHours(9).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Also fixes BattlePassRepository.GetActiveSeasonAsync to use client-side
DateTimeOffset filtering (SQLite provider cannot translate DateTimeOffset
comparisons in LINQ WHERE/ORDER BY clauses).
- IndexResponse.BattlePassLevelInfo widened to IReadOnlyDictionary<string,BattlePassLevel>?
so any IReadOnlyDictionary impl (FrozenDictionary, wrapper, etc.) serializes correctly
instead of silently null-ing via a failed as-cast
- LoadController.Index now takes CancellationToken ct and threads it to GetLevelCurveAsync
instead of CancellationToken.None
- BattlePassRepository.ResetLevelCurveCache changed from public to internal; added
InternalsVisibleTo("SVSim.UnitTests") to SVSim.Database.csproj (was absent)
Wire IBattlePassService.GetLevelCurveAsync into LoadController so /load/index
emits the 100-entry battle_pass_level_info dict when levels are seeded.
Also adds BattlePassRepository.ResetLevelCurveCache() to bust the process-level
static cache in tests that seed levels after earlier HTTP calls have primed it
with an empty list, and updates SVSimTestFactory.SeedGlobalsAsync + the stale
Index_surfaces_seeded_globals_after_bootstrap assertion accordingly.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add ValueGeneratedOnAdd to ViewerBattlePassProgress.Id and
ViewerBattlePassClaims.Id so Postgres generates IDENTITY values at
runtime. Regenerate AddBattlePass migration in-place to include the
IdentityByDefaultColumn annotations. Add IBattlePassRepository /
BattlePassRepository (season lookup + level-curve cache) and
IViewerBattlePassRepository / ViewerBattlePassRepository
(get-or-create progress, claim reads/writes).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
