Every write path normalises a bundle's texture to WebP Lossless, so the
catalog icon was being read back with a PNG-only decoder. decodePng throws
on anything that is not a PNG, the callers caught that and returned null,
and the user-visible result was "no icon (not in source or bundle)" for
every furniture whose source does not serve a standalone icon.
Measured against the production asset tree, all 18,505 bundles were WebP;
icon extraction succeeded on 0 of them. src/lib/services/imager/
decode-texture.ts keeps PNG on the dependency-free decoder and routes WebP
through sharp, which is already a dependency and already encodes these
textures. extractFurniIconPng and getPetIconPng become async; the five
call sites (upload, clone import, furni import, icon repair and both icon
routes) already awaited their surrounding work.
Same root cause, second bug: the spritesheet frame key. Converters disagree
on packing — some keep a trailing ".png", and some lowercase the whole key
while leaving the bundle name mixed-case, so "LTD_fashionistaf" looks up
frame "LTD_fashionistaf_LTD_fashionistaf_icon_a" and never finds
"ltd_fashionistaf_ltd_fashionistaf_icon_a". Any mixed-case classname could
therefore never match, which is most of the catalogue. findFrame tries the
two exact spellings, then falls back to one case-insensitive pass.
Extraction now succeeds on 18,483 of 18,505 bundles (99.88%); the 22
remainder are data, not code — 9 bundles ship no icon asset, 11 do not
parse.
Third: three catalogue icons exist only as .gif while catalogueIconUrl
hardcoded .png, so the picker offered icons that could only ever 404, and
291 icons that ship as both formats were listed twice. The API now dedupes
per id and the two renderers retry with .gif before falling back to the
placeholder, matching what catalog-image-picker already did. Verified live:
/gamedata/.../icon_1542.png returns 404 while icon_1542.gif returns 200.
Separately, close the last hole in the memory cap. Every script in
package.json routes through scripts/with-memory-cap.sh, but invoking the
builder directly — from a terminal, an IDE or an agent — skipped the
wrapper and ran unbounded, on a host with no swap where the OOM killer
picks its victim across the whole machine. next.config.ts now refuses a
production build that the wrapper has not marked, before anything
allocates. next dev and next start are deliberately unaffected.
README gains a Memory-capped commands section covering the per-script
ceilings, the backends and the ulimit -v trap, and its stale version and
script tables are corrected.
Every bundle the CMS writes — upload, clone, sync, repair and the pet /
effect / figure importers — now lands as `<classname>.hab`, the extension
this deployment's renderer asks for. `.hab` and `.nitro` are the same
container, so an upload of either extension is accepted.
Resolution goes through one module, src/lib/furni/bundle-file.ts, so
nothing has to know the extension twice. Every existence check probes
`.hab` first and falls back to `.nitro`: the on-disk asset set is still
predominantly `.nitro`, and without the fallback Studio would report every
imported item as missing and the cleanup scan would classify 18k live
bundles as fake leftovers. Downloads are unchanged — Habbo's CDN and every
configured clone source still serve `.nitro`, so the conversion happens on
write, not on request.
Deliberately unchanged: the staged-attachment store in furni-attachment.ts
keys on a UUID and never reaches the client, so renaming it would break
in-flight recovery jobs.
Adds scripts/migrate-nitro-to-hab.ts to rename the existing asset set. It
refuses to run without --dry-run or --yes, never overwrites an existing
.hab, never deletes, and is idempotent.
Note: renderer-config.json lives outside this repo and was patched to
.hab separately; that file is served with a 30-day max-age, so returning
clients need a cms-client cache purge to pick the change up.
Importing from a hotel wrote the downloaded .nitro to disk untouched, so
official PNG textures stayed PNG and only SWF imports ended up as WebP.
Every Studio import should produce the same format regardless of where the
bytes came from, so both clone and upload paths now run the bundle through
toWebpLosslessBundle.
The helper decodes the texture and re-encodes it with the same VP8L options
the SWF importer uses, so the artwork round-trips bit-for-bit, and lets
createNitroBundle relabel the member and repair the meta.image pointer. A
bundle that is already lossless WebP is returned untouched, making the
operation idempotent and safe to run on re-import. A colour variant that
shares a library keeps the member base name it arrived with.
A texture that cannot be decoded keeps its original format with a warning
instead of failing the import: the bundle is valid, and losing a furniture
item over a codec edge case is worse than a slightly larger texture.
Replace raw db.execute tuple casts with queryRows/rowsFrom/execResult/
affectedRows helpers from lib/db, drop redundant mysql2 casts on typed
query builders, and centralize per-test fakeForm into test/fake-form.
Update db mocks in tests so helpers resolve against mocked execute.
Every write path now sets offer_id = catalog_items.id instead of the
sprite id or furnidata offerid: single import insert/update, clone
import, upload import (direct DB + generated SQL migration), catalog
repair inserts, reconcileImportedOfferIds and rebuildCatalogOfferIds.
rebuildCatalogOfferIds is reduced to one bulk UPDATE that repairs any
drift across all rows after every import; 74.838 legacy rows were
repaired on the live database with this change.
Interaction detection:
- Trust SWF-derived sit/lay/stand flags only when the logic XML actually
contains action data (new hasActions metadata); otherwise fall back to
keyword detection so custom furni without <action> nodes are still
classified correctly
- Add French/Dutch/German/Spanish/Italian keywords (chaise, banquette,
stoel, silla, sedia, stuhl, tafel, mesa, ...) to sit/lay/stand detection
with token-boundary matching to avoid false positives like bedside_table
- Unify interaction_modes_count priority: mechanic fixed modes, then raw
animation state count, then sit/lay fallback, then keyword default
- Detect mechanic type even when real flags are not used
Catalog integrity:
- Skip duplicate catalog_items inserts in clone and upload imports
- Re-check live catalog rows before applying generated repair SQL
- Add rebuildCatalogOfferIds() which rebuilds every catalog_items.offer_id
from the local FurnitureData.json (matched by entry id, then classname)
and run it after every import path (single, batch, batch-regen, clone)
- importSingleFurni: actually update an existing item instead of failing on a primary key INSERT collision; refuse to reassign a spriteId owned by a different classname
- reconcileFurniDataWithItemsBase: also sync each entry's offerid to the DB sprite id, not just the id
- Move the full-table interaction_modes_count verification out of the per-item import hot path and run it once per batch/single import
- clone-import: apply the auto-detected interaction_type (was hardcoded 'default'), set FurnitureData offerid and catalog_items.offer_id to the new local sprite id (was the source hotel's id / -1)
- clone batch route: run the same post-import reconcile/verify/ownership/RCON consolidation as the furni batch route
- upload-import: set offer_id to the allocated id in both the direct insert and the generated SQL migration (was -1)
- generateCatalogSql: use the classname as catalog_name and the item id as offer_id so rows match the app-managed import convention
- stats + missing-nitro endpoints: match catalog membership via item_ids instead of the catalog_name join, so translated display names no longer break counts
- deleteImportedItem/rollback: delete catalog rows by the canonical item_ids link only
- classifyFurni: wall items always classify as 'walls' before prefix rules (rare_/val_/xmas_) can misfile them
Replace classname/name keyword guessing with real data so sit/lay/stand
has zero false positives:
- autoDetectInteraction() now accepts real flags (cansiton/canlayon/
canstandon), logicType, and animation-state count; real data wins and
keywords are only a fallback when no real data exists.
- New furni-real-interaction helper reads the local FurnitureData.json
(highest-revision entries win; a true flag in any duplicate is kept)
and falls back to the official Habbo furnidata.
- interaction_modes_count now uses emulator conventions: dice=6,
gate/teleport=2, roller=1, generic multistate = animation-state count,
chairs/beds=1, plain items=0.
- verifyAndFixInteractionModesCount only touches items with real data
and skips everything else (no keyword overwrites).
- parseNitroBundle falls back to gunzip for gzip-compressed .nitro files.
- Wired through furni-import, upload-import, clone-import, and the nitro
editor auto-detect button.
- fetchSourceFurnidata now parses JSON regardless of content-type,
so GitHub raw mirrors (Kyzegs, sphynxkitten) that serve JSON as
text/plain work again instead of forcing a FlareSolverr round-trip.
- Add curl subprocess fallback (curl-fetch.ts) for CDNs that block
Node's fetch by TLS fingerprint (Leet.city) — used for furnidata,
nitro bundle and icon downloads before giving up.
- Merge verified default clone source presets with stored user sources
so the admin always offers many working sources.
- New auto-interaction.ts: detects canSit/canLay/canStand from classname/name keywords
- Maps keywords to interaction types (bench, chair, bed, table, etc.)
- Returns clickLimit and interactionModesCount for proper click/interaction limits
- Nitro editor: added auto-detect button with preview of detected settings
- furni-import.ts: auto-sets interaction_modes_count during import
- upload-import.ts: auto-detects for direct insert and SQL migrations
- clone-import.ts: auto-detects when cloning from other hotels
- CloneSource interface now has a hotel field (maps to habbo_gamedata_hotel)
- DEFAULT_SOURCES presets include their associated hotel (it/com)
- When cloning from a source with a hotel configured, habbo_gamedata_hotel
is set automatically so official furnidata enrichment uses the correct locale
- Clone source form UI now has a hotel select dropdown
- Source list shows the hotel label when configured
- Clone API route passes hotel through to upsertSource
Leet serves its furnidata as HTML with the JSON embedded in a <pre>
block, and Cloudflare blocks Node's direct fetch. Extract the JSON
payload from the HTML (direct or via the FlareSolverr fallback) before
giving up on a Cloudflare challenge.
Also skip the standalone icon download when a clone source has no
iconBaseUrl configured (Habbo, Hubba, Fresh, Kyzegs, RidgeRP), which
previously produced invalid relative URLs like /xxx_icon.png and a
flood of download errors before falling back to extracting the icon
from the .nitro bundle.
Replace the serialized SELECT MAX + INSERT id-allocation chains for
items_base and catalog_items with a lazy-seeded in-process counter so
concurrent clone workers no longer queue on a global lock per item.
Re-seeds after 60s idle to avoid colliding with externally added rows.
Raise the clone import concurrency default from 6 to the batch cap of 10.
- Clone import: defer FurnitureData.json writes and append all entries in a
single batched write instead of one read-modify-write per item, removing
the main serialization bottleneck for large batches.
- Clone import: raise SSE batch concurrency cap from 5 to 10 and bump the
clone client/route default from 2 to 6.
- Add a flush hook to runSseBatch so callers can batch deferred work before
batch_complete is emitted, and surface flush errors as an error event.
- Enable Next.js Cache Components (instant: false opt-out) and silence the
related build warnings in next.config.ts.
- Switch isomorphic-dompurify to dompurify and refresh dependencies.
Previously, if FlareSolverr itself failed (timeout, connection error),
the error would propagate as generic 'network error'. Now it throws
a descriptive error message.
- Detect HTML responses from FlareSolverr and throw descriptive error
instead of letting JSON.parse fail with cryptic 'Unexpected token' error
- Add try/catch to clone/route.ts GET handler to return 502 with
clear message instead of Internal Server Error 500
- Add FLARESOLVERR_URL to .env
- Add FLARESOLVERR_URL env var to .env.example
- Update fetchSourceFurnidata to fall back to FlareSolverr on CF challenges (403/HTML)
- Add docker-compose.yml with FlareSolverr service
- Add scripts/health-check.sh for FlareSolverr readiness check
- Add health:check script to package.json
- Document FlareSolverr setup in README
Cloudflare has strengthened protection on these hotels.
Puppeteer-based bypass returns HTML instead of JSON.
cloudscraper has dependency issues with Node.js v26.
Reverted to simple HTTP fetch with clear error messages.
- Add cf-fetch.ts with puppeteer-extra + stealth plugin for CF bypass
- Modify fetchSourceFurnidata to detect CF challenge and retry with puppeteer
- Restore Leet, Hubbly, and Habblet City to clone sources (now CF-protected)
openhands
Simo
Remco