Fetch QQ's versioned robot UIN ranges, classify and persist member status, and surface is_robot from both group member actions.
Preserve unknown legacy cache entries and fail visibly when classification is unavailable.
Closes#265
Encode poke message segments as private window shake messages while keeping the dedicated send_poke action unchanged.
Restrict window shakes to a single ordinary private-chat segment. Reject group, temporary-session, mixed-message, and forwarded usage before identity lookup, media upload, or packet transmission.
Add full outbound request coverage, OneBot action regressions, and live private-chat verification for issue #259.
Closes#259
Decode the verified album cover response field and expose normalized cover details and last-upload timestamps through both album list actions. Preserve the legacy numeric timestamp shape, retain string timestamps for the native response, return null when no cover exists, and avoid additional network requests.
Document the new output contracts in the OneBot and MCP catalogs. Add independent wire fixtures and action-level regression tests for the protocol mapping and public response shapes.
Refs #257
Use the C2C roam cursor and canonical NT sequence for bidirectional history. Persist direction-aware recall tombstones so recalled messages cannot reappear.
Recognize both raw UTF-8 UIDs and nested OperatorInfo envelopes in GroupChange pushes while keeping self-kick and voluntary-leave semantics explicit.\n\nFixes #253.
Wait for the canonical self group-message event after AI voice synthesis instead of interpreting undocumented response fields as a message sequence.
Correlate receipts by exact media UUID, buffer response races, and claim each semantic message identity only once across concurrent calls and duplicate multi-process pushes. Reject missing identifiers, invalid sequences, bounded-buffer overflow, and receipt timeouts with explicit diagnostics.
Keep fetchAiVoice focused on returning the media node and add regression coverage for race ordering, unrelated events, identical concurrent audio, duplicate pushes, malformed receipts, and timeout cleanup.
Fixes#250
Replace the lossy Qzone album-list endpoint with the QQ NT AlbumService protobuf request. Preserve creator nicknames and expose native pagination for get_qun_album_list.
Replace the get_collection_list placeholder with authenticated, read-only collection retrieval through the existing bridge API.
Add strict binary envelope validation, bounded pagination, accurate category filtering, preserved 64-bit identifiers, and explicit ticket, transport, service, and malformed-response failures. Cap item counts, page traversal, response bytes, per-request waits, and the total operation time.
Expose the compatible response model through Core, OneBot, the SDK, and the generated MCP catalog. Cover protocol framing, cursor behavior, filtered hasMore semantics, cleanup failures, bridge ticket acquisition, and Action validation with regression tests.
Verified with the live test bot without changing native components or adding any collection write behavior.
Store the latest online-device snapshot for each Bridge lifecycle and expose it through the bridge contract as an immutable, owned value.
Distinguish a snapshot that has not yet been observed from a valid empty snapshot, replace state atomically on each QQ update, clear it during disposal, and cover lifecycle, immutability, and cleanup behavior with focused tests.
Decode Event 0x210 subtype 349 into typed online-device snapshots, including QQ-compatible low-byte client type normalization and one-entry-per-device-category projection.
Keep replacement snapshots outside system-message deduplication, reject malformed device identities observably, and add wire-level regressions for complete, packed, repeated, and invalid payloads. Live verification covered mobile account switching without requiring native changes.
Decode the folder modifier metadata returned by group file listings and project it as last upload time, uploader ID, and uploader name for both folder listing actions.
Regenerate the MCP catalog and add protocol, bridge, OneBot projection, and action schema coverage for the new fields.
Closes#248
Encode the opaque media-list cursor in the protobuf page-info field while leaving the reserved field empty. This restores subsequent-page requests without changing the OneBot action or response shape.
Add an independent wire oracle so future protocol definition drift cannot make the encoder regression test pass against itself.
Fixes#245
Introduce a CSS-driven adaptive cursor that morphs for controls, text editing, dragging, resizing, busy states, and other native pointer semantics on desktop devices.
Add a themed context menu for editing, selections, links, images, and page navigation while preserving explicit native-menu escape paths. Store a default-on interaction preference through the normalized appearance model and cover its migration and opt-out behavior.
Walk bounded QQ history windows in either direction while keeping all returned messages in chronological order.
Resolve short-capped forward ranges without skipping messages nearest the anchor, share the existing request throttle across group and private history, and surface incomplete fetches instead of returning silent partial results.
Refs #229
Clear only the membership flag bit when normalizing group increase operation codes so unrelated high bits cannot collapse into the invitation value.
Add regression coverage for extended operation codes.
Derive group increase approval and invitation subtypes from the member-change operation code instead of actor identity.
Treat voluntary leave field 5 as structured metadata and attribute the operation to the departing member so valid notices are not discarded.
Add protocol and OneBot regression coverage for issue #237.
Parse RFC 2397 base64 data URLs through the shared binary source classifier so inline media cannot fall through to local file handling.
Reuse the same parsing path for buffered and disk-staged uploads, reject malformed or oversized inputs explicitly, and add regression coverage for both paths.
Refs #236
Treat an absent media-list return code as the protobuf zero default while preserving failures for explicit non-zero codes.
Add a regression test that exercises the encoded response path with media data and no status field.
Refs #234
Track receive health for every process attached to an account while treating unarmed monitoring as compatible rather than degraded.
Report the account as healthy while any attached process remains healthy, and mark it degraded only after every attached process has been confirmed stale. Keep health ownership synchronized with process attach, migration, and detach transitions.
Fail fast when process ownership and health state diverge instead of silently presenting a healthy status. Add multi-process regression coverage for partial degradation, full degradation, and recovery.
Refs #233.
Replace route-local restore writes with the core transaction and map preflight, rollback, and post-commit cleanup failures to precise HTTP responses. Treat only ENOENT as a missing export file and return transaction, restored, skipped, migrated, and restart facts.
Update the settings panel to describe the full sensitive configuration group, avoid restart guidance for empty plans, and reload settings after both successful and failed imports so the UI reflects committed disk state. Add HTTP regression coverage for export and restore error semantics.
Move restore execution into a dedicated module with explicit preflight, snapshot, stage, commit, rollback, and cleanup phases. Keep live files untouched until staging completes, roll back partial commits, and retain the transaction directory when rollback or cleanup cannot finish.
Report the transaction ID, active phase, failed files, commit state, rollback result, and recovery path without logging configuration contents. Cover every filesystem boundary with a focused fault matrix, including unreadable current files and cleanup failures.
Route selected backup entries through owner-owned config parsers before any filesystem write. Reject unsupported paths, mismatched encodings, malformed base64, lossy normalization, invalid effective overlays, and decoded payloads larger than 32 MiB.
Treat WebUI state, private keys, and every OneBot config as a single sensitive group. Skip unselected sensitive entries without inspecting their contents, preserve recognized migrations, and assign restrictive modes to credential files.
Handle member joins and ordinary removals using their direct operator identity while preserving the special nested form used when the bot is removed.
Reject malformed values with bounded diagnostic context and add regression coverage for #228 and related member-decrease events.
Route regular and new-member announcements through their correct web endpoints. Validate option conflicts, preserve existing defaults, merge returned notice collections, and surface transport and upload failures.
Present Windows x64, Linux x64, and Linux arm64 as fully supported release targets in a concise download matrix.
Add quick-start and unattended-consent guidance, link commits and version comparisons, and filter merge and release housekeeping from the change list.
Refresh the adjacent packaging documentation so it matches current platform support.
Allow unattended deployments to accept the EULA and Privacy Notice through SNOWLUMA_ACCEPT_EULA and SNOWLUMA_ACCEPT_PRIVACY.
Require both values, reject invalid configuration, keep environment consent non-persistent, and log whether acceptance is complete or partial.
Document the deployment contract and add regression coverage for parsing, persistence isolation, and stored-consent compatibility.
Replace message-header sequence forwarding with a server-guided read flow for group and private conversations.
Batch read reports conservatively, validate per-conversation results and cursor advancement, and fail with actionable errors when the server does not apply an update.
Add regression coverage for mixed batches, pagination, already-current sessions, malformed responses, and private sequence-domain mismatches.
Serialize desired network plans, isolate per-adapter apply failures, restore prior bindings when possible, and report saved versus applied state. Quiesce Action and event ingress, await in-flight dispatches and transport closes, retain failed generations for retry, and propagate shutdown failures.
Replace the open message bag with a validated discriminated union and preflight complete messages before uploads or sends. Return BAD_REQUEST for invalid or unsendable segments instead of dropping fields, sending partial content, or silently changing private messages.
Cap long-message gunzip output at 32 MiB and fail before an oversized QQ response can exhaust process memory. Include the resource id, compressed size, and configured limit in observable errors.
Reject malformed varints, out-of-bounds fields, and invalid groups before they escape their parent frame. Aggregate unknown-field metadata and decode signed int32 values while preserving SnowLuma's existing outbound encoding bytes.
send_private_msg / send_msg can now reply to a group temp-session message
by passing the source group_id. Inbound temp messages surface their
source group on sender.group_id; a reply is only made into a temp session
the bot has an inbound record of. Group temp replies route through
RoutingHead.grpTmp (field 3).
core's typecheck covers its own tests (protocol/onebot typecheck src
only), so the newly-required GroupMessage.groupName field failed the
Dev Build typecheck on test fixtures that construct group_message
events. Add groupName to every such fixture across core/onebot/protocol
tests.
The #211 concurrency attempt (c17fff6) was disproven on a live endpoint.
QQ's highway server requires blocks to arrive in offset order and rejects
an out-of-order PUT with error_code=102902 — ~20% failure under a real
account even after deferring the finalize block. Parallel connections
can't guarantee arrival order, so that approach is unshippable.
Replace it with a single persistent connection that PUTs blocks in strict
offset order, reusing the socket across the whole file via HTTP keep-alive:
- In order: the server never sees an out-of-order block, so no 102902.
- One warm connection: the TCP congestion window stays open across the
file instead of paying slow-start on a fresh connection per 1 MiB block.
That per-block cold start was the real throughput cap behind #211.
Adaptive to QQ's edge nodes FIN'ing between chunks (the reason #118 went
one-connection-per-chunk): on a transport failure, or a peer half-close
after a response, drop the socket, reconnect, and re-send the SAME block
(re-PUTting an offset is idempotent). Worst case degrades to one
connection per block — no worse than #118, and still strictly in order.
Also fixes a latent bug on the reconnect path, surfaced by review: a
response truncated mid-stream (header arrived, Content-Length unmet, then
FIN) was resolved as a partial frame and then fatally rejected by
unpackHighwayFrame OUTSIDE the retry loop, turning a recoverable mid-body
FIN into a failed upload. A truncated response now rejects as a retryable
transport failure so the block is re-sent. Added an idle read timeout so
a silent peer can't hang the upload forever (the persistent socket widens
that exposure).
Verified live (tempserver, repeated 48 MiB group-file uploads): 1
connection per file, 0 failures across dense repeats, ~1.4x throughput
over the old per-chunk-connection serial path in a low-RTT datacenter —
the win is larger on high-RTT home links where per-block slow-start hurt
most (the original #211 report). Unit tests cover keep-alive reuse,
FIN-after-each-response degradation, mid-stream reconnect+resume, body-
truncation retry, in-order exactly-once coverage, and close-once.
Group/private file uploads (and all highway media) topped out around
500 KB/s regardless of link speed. Two costs stacked up in
uploadHighwayHttp: chunks were sent strictly one at a time (throughput
capped at one 1 MiB block per round-trip), and each block opened a fresh
connection that paid TCP slow-start from a cold congestion window — so a
1 MiB block finished before the window ramped and never reached line
rate.
Drive the (already offset-addressed, self-contained) blocks through a
small worker pool instead of a sequential loop. Highway frames carry
their own dataOffset/chunkMd5/fileMd5, so blocks are independent and the
server reassembles by offset; a handful of connections overlap the
round-trip stalls and ramp their windows together. Concurrency is kept
modest (4) to stay friendly to the edge nodes' rate limiting, and the
one-connection-per-block invariant from #118 is preserved — parallelism
just puts several of those short-lived connections in flight at once.
Correctness preserved and covered by new tests:
- offset claim is synchronous (no await between read and advance of
nextOffset), so every byte range is uploaded exactly once — no skip,
no duplicate, ragged final block correct;
- concurrent ChunkSource reads are race-free (positional FileHandle
reads into per-call buffers; read-only Buffer subarrays);
- first permanent block failure aborts the whole upload and rejects;
peers stop claiming new blocks and drain the one in flight;
- workers never reject, so Promise.all settles all in-flight
reads/sockets before the source is closed exactly once — no
close-during-read, no masked upload error;
- 0-byte source still opens zero connections; peak memory bounded to
concurrency x ~1 MiB, never the whole file.