Files
multica/server/internal/handler/handler.go
beast 60048172a7 fix(lark): ingest inbound images and videos as chat attachments (MUL-4934) (#5580)
* fix: ingest feishu media as chat attachments

* fix: ingest feishu post embedded media

* fix(lark): make inbound media retries safe

* fix lark media resource limit

* fix(lark): move inbound media off ack path

* fix(channel): make inbound media runs durable

* fix(channel): close enqueue-vs-append race on media deferral

EnqueueChatTask read the session-wide media deadline in one statement and
sealed the input batch in a later one. Under READ COMMITTED a media message
committing between the two got sealed into a task the deadline read had
already decided was 'queued', so the daemon could claim it before its
attachment bound — the agent received the bare placeholder, and the later
media-ready promotion was a no-op against a non-deferred task.

After the seal, re-derive the deferral from the sealed batch itself in the
same transaction (DeferChatTaskForSealedPendingMedia): if any sealed message
still carries an unexpired media marker, flip the task to deferred with
fire_at aligned to the latest marker. The existing post-commit promote fence
already covers the opposite direction (marker cleared mid-transaction).

Adds a deterministic regression test that injects the media append between
the deadline read and the seal via a wrapped pgx.Tx.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): keep committed chat task out of enqueue error path

The post-commit media-ready fence returned its error from EnqueueChatTask
even though the deferred task was already durably committed. The router
flush treats any enqueue error as "no task exists": it clears the typing
indicator and logs an enqueue failure while the run still happens at its
fire_at deadline. Log the fence failure instead — the claim-path deferred
promoter re-queues the task regardless.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): cap global media resolution concurrency

Media jobs were serialized per session but unbounded across sessions: a
burst could open arbitrarily many concurrent 45s Lark downloads, and each
unknown-length upload may buffer up to the 100 MiB resource cap in memory.
Gate resolveAndBindMedia behind a global slot semaphore (default 8,
RouterConfig.MediaConcurrency). Per-session ordering is unchanged; on
shutdown a job cancelled while waiting for a slot proceeds straight to the
bounded DB finalize so marker clearing stays prompt. Also document that the
per-message media budget spans queue/slot waits (it must match the
persisted fire_at) and why timed-out uploads cannot leak unbounded orphans.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(chat): keep channel-sealed user messages on task cancel

Sealing the channel input batch stamps task_id onto channel user
messages, which exposed them to the cancel draft-restore path: an
empty-transcript cancel would DeleteUserChatMessageByTask the sealed
Feishu/Slack messages and detach their attachments. Those messages are
the durable record of what the platform sender wrote — the sender has
no Multica composer to restore a draft into.

Gate the restore-delete on ChatSessionHasChannelBinding in both the
synchronous finalize and the deferred finalize (the latter covers
markers left by an older replica during a rolling deploy); a bound
session now settles as "Stopped." instead.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): skip the media pipeline for messages without media

Every inbound message on a Media-enabled platform persisted a 45s
media deadline and queued a resolution job, so a plain text message
could wait behind the global media semaphore (its task deferred while
other sessions download 100 MiB videos) and a crash between append and
clear delayed a pure-text run to the full 45s fallback.

Add MediaResolver.HasMedia — a pure in-memory probe the Router calls
on the ACK path — and only persist the deadline / enqueue the job when
the message actually references platform media. The Feishu resolver
decodes the already-received payload and reports standalone image or
video keys and post-embedded img/media spans.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(chat): gate cancel restore on immutable channel provenance

The previous guard keyed the cancel restore-delete off
ChatSessionHasChannelBinding, but a binding only proves routing exists
right now: archiving a session and rebinding an installation both
delete the binding while preserving chat history, so a still-cancellable
sealed task could again restore-delete the original inbound messages.

Persist provenance on the message instead: migration 203 adds
chat_message.channel_ingested, stamped inside the channel append
transaction and never mutated, and both cancel finalize paths now gate
on TaskHasChannelIngestedMessages over the task's sealed batch. The
binding-existence query is removed. Regression tests cover ingest ->
archive/unbind -> cancel for a queued and a started task.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): reclaim media uploads that never gain an attachment row

Deadline expiry dropped already-resolved refs and a BindMedia failure
was log-only, leaving uploaded objects with no attachment row and no
reclaim path — the dedup mark commits with the message before media
runs, so a redelivery is dropped as a duplicate and never re-resolves
(and thus never overwrites) those keys, and workspace/session deletion
only enumerates the attachment table.

Add MediaResolver.DiscardMedia — a best-effort delete by StorageKey —
and call it from both failure paths in resolveAndBindMedia. The Feishu
resolver forwards to the storage backend's Delete. Tests cover a
partial upload discarded at the deadline, discard on bind failure, and
key-level deletion in the resolver.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* docs(server): refresh comments stale after detached media ingestion

Channel tasks now seal a self-owned input batch, media ingestion is no
longer out of scope for the flattener, and MediaRefs are filled by the
detached resolver after append rather than by feishuChannel pre-engine.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(chat): stop keying channel empty-completion silence off chat_input_task_id

Sealing gave channel tasks a self input-owner, which broke
writeChatCompletionOutcome's discriminator: it treated any owned task
as direct, so an empty channel completion wrote the no_response
fallback row and the outbound patcher — which forwards any non-empty
chat:done content verbatim — pushed the English fallback body to
Feishu/Slack, violating the MUL-4351 contract.

Silence is now decided by the immutable channel_ingested provenance of
the task's input batch, looked up by the batch OWNER id
(chat_input_task_id): auto-retry clones inherit the owner while their
sealed messages stay tagged with the parent's id, so keying off the
task's own id would misread a channel retry as direct. The cancel-path
provenance gates switch to the same owner key via chatInputOwnerID.
chat_input_task_id is back to meaning only "input batch owner".

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore(migrations): renumber to 203/204 after upstream took 202

Upstream main merged 202_runtime_profile_add_qwen while this branch
held 202/203, tripping TestMigrationNumericPrefixesStayUniqueAfterLegacySet
on the CI merge tree. channel_media_pending becomes 203 and
channel_ingested becomes 204; no content changes.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channels): gate outbound delivery on channel provenance, not owner

Merging main brought #5645 (keep direct chat replies in Multica),
whose outbound gate assumed channel tasks leave chat_input_task_id
NULL. Sealed channel tasks own an input batch too, so on the merge
tree every channel reply and failure notice was classified as direct
and silently dropped — agents stopped replying in Feishu/Slack.

Both outbound gates now call engine.TaskInputIsChannelIngested: a NULL
owner keeps #5645's deliver-by-default for pre-sealing tasks, an owned
batch delivers only when it carries the immutable channel_ingested
stamp (keyed by the owner id, so auto-retry clones inherit the
verdict). Direct replies stay in Multica; sealed channel replies reach
the platform. Tests cover both directions on both platforms.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): discard media orphans on a fresh context, after finalize

DiscardMedia shared finalizeCtx with BindMedia, so a bind that failed
because the finalize deadline expired handed the storage deletes an
already-dead context — the compensation silently no-opped and the
orphans leaked anyway. The deadline path also ran S3 deletes before
the marker clear, eating the same 5s budget the user-facing
bind/promotion needed.

Collect the refs from both failure paths, run bind + promotion on the
finalize budget first, then delete on a fresh discard context. The
bind-failure test now pins that discard receives a live context.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): compensate result-uncertain media uploads and commits

The compensation protocol treated "the call returned an error" as "the
side effect did not happen", which is wrong in both directions across
the result-uncertain windows:

- An upload error can follow a server-side write (lost response,
  deadline mid-write). The attempted key never reached the router, so
  nothing could reclaim it and dedup guarantees no re-resolve. The
  resolver now idempotently deletes the deterministic key on a fresh
  budget right at the failure site.

- A commit error is not a rollback guarantee: a lost ack can report
  failure after Postgres durably committed the attachment rows, and
  the router's discard would then delete objects those rows reference.
  BindMediaRefs now converges the ambiguity on a fresh budget — any of
  the batch's URLs present proves the atomic commit landed (bind
  reports success); none proves the rollback (discard stays safe); a
  failed verification returns ErrMediaBindResultUnknown and the router
  keeps the uploads, preferring a rare orphan over a broken attachment.

Fault-injection coverage: an upload error deletes the attempted key; a
lost-ack commit keeps the bound attachment and reports success; a
verified rollback stays a discardable error; the router keeps uploads
on the unknown-outcome sentinel.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore(migrations): renumber to 207/208 after upstream took 203-206

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* docs(channel): note DiscardMedia self-invocation and the unknown-outcome skip

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore(migrations): renumber to 212/213 after upstream took 207-211

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* feat(channel): replace inline media compensation with an intent ledger and reconciler

Inline best-effort compensation cannot answer "did my side effect
happen?" at the moment it needs the answer — the DELETE/PUT reordering
and the empty-read-vs-in-flight-COMMIT gaps were both instances of the
same two-system atomicity problem. Persist the intent instead and let
an asynchronous reconciler settle it:

- channel_media_pending_object (migration 214; claim index 215 as its
  own single-statement CONCURRENTLY migration): a state machine row
  ('pending' -> 'deleting') with lease, attempt, and backoff columns.
- The resolver upserts the row BEFORE each PUT, state-guarded so a key
  the reconciler owns is never resurrected (the resource is skipped).
  ObjectURL is a pure function of configuration, so the row carries the
  attachment URL pre-upload.
- BindMediaRefs deletes the batch's rows INSIDE the attachment-insert
  transaction: commit landed <=> intents gone, atomically, so an
  ambiguous COMMIT never needs adjudication. A key already claimed to
  'deleting' is skipped (placeholder stays).
- Nothing is ever deleted inline. The reconciler — an independent
  worker so storage latency cannot starve other sweepers — claims due
  rows ('pending' past the settle delay, or expired leases) under a
  fresh lease, checks for a durable attachment reference only AFTER the
  claim (race-free: bind can no longer succeed on the key), deletes
  unreferenced objects outside any transaction, and backs off failed
  deletes with attempt-based retry. Crash windows converge for free.
- The settle delay is a fixed constant carrying NO correctness weight;
  invariant tests pin it at >=10x every pipeline budget. Metrics cover
  deletes, referenced clears, delete failures, and ledger backlog.

Removed: MediaResolver.DiscardMedia, ErrMediaBindResultUnknown, the
post-commit verification, and both router discard branches.

Tests: intent-before-upload ordering; upload error leaves the row and
deletes nothing; bind-wins vs reconciler-wins on the same key; lost-ack
and rolled-back commit injections (intent cleared iff the attachment
landed); reconciler three-state settle; expired-lease reclaim; delete
failure backoff and retry; settle invariants.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): never build or sweep the media reconciler without storage

store is nil when S3 is unconfigured AND the local upload dir fails to
initialize, but the reconciler was constructed unconditionally and
main only gates the goroutine on the reconciler pointer — the first
unreferenced ledger row (rows can pre-exist from a boot where storage
worked) would nil-pointer panic a bare goroutine and take down the
process.

Construct the reconciler only when a storage backend exists, and guard
RunOnce defensively: with no deleter it skips the sweep without
claiming, so rows are not stranded in 'deleting' until lease expiry.
Test covers the pre-existing-row + missing-storage boot.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore(migrations): renumber to 213-216 after upstream took 212

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* refactor(channel): remove the dead pre-resolved MediaRefs ingress path

lark.InboundMessage.MediaRefs and the resolver's early-returns for
pre-populated refs were vestiges of the pre-detached synchronous design
— no producer fills them before the router anymore. Worse, the intent
ledger made the path actively misleading: refs arriving without ledger
rows would be silently skipped at bind (with a log blaming the
reconciler), contradicting the field's "already persisted" contract.

Delete the field, its channelMessageFromLark mapping, and both
early-returns; channel.InboundMessage.MediaRefs is now documented as
what it actually is — ResolveMedia's output channel, always empty on
ingress, attachable only through a claimed ledger intent.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): enforce workspace tenancy on every ledger query

The intent-ledger upsert's conflict branch guarded only on state, so a
cross-workspace storage_key collision could rewrite the row's
workspace/message/url ownership; release and delete keyed on
(storage_key, lease_token) alone. The derived key embeds the workspace
UUID so none of this is reachable today — but tenancy must be enforced
by the workspace column in every query, never derived from the key
string (MUL-3515 rule, restated in this PR's review).

The upsert now updates only within the same workspace (a cross-tenant
conflict updates nothing, returns no row, and the resolver skips the
upload — the fail-safe direction), and release/delete take
(workspace_id, storage_key, lease_token). Tests pin that a foreign
workspace can neither steal, release, nor delete a row.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(migrations): build the ledger primary key via a concurrent index

storage_key TEXT PRIMARY KEY created its unique index implicitly at
CREATE TABLE, against the repo convention that every migration index —
including a new table's unique index — is built CONCURRENTLY in its
own single-statement migration (the exact three-step pattern
client_usage_daily shipped in 207-209). The table now declares
storage_key NOT NULL, 216 builds the unique index concurrently, and
217 attaches the primary key USING INDEX; the claim index moves to
218. ON CONFLICT (storage_key) still resolves against the constraint,
and the full down/up round-trip is verified.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): bound each reconciler object delete with its own timeout

DeleteObject ran on the worker-lifetime context and the SDK's default
HTTP client has no overall request timeout, so one black-holed
connection would wedge the sequential sweep loop — and with it every
later batch and the backlog gauge — forever; a single-replica
deployment has no other worker to reclaim the lease. Each delete now
gets a 30s timeout (well under the 2min lease), and a timed-out delete
takes the existing release/backoff path. Covered by a blocking-deleter
test with an injectable timeout.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): anchor the media deadline to the DB clock and bound queue waits by it

Two deadline gaps from review:

- The persisted marker was an application-clock timestamp compared
  against SQL now() everywhere it is read, so a skewed app node could
  shrink the fallback window and hand the agent a placeholder before
  the resolver's local budget ended. The append transaction now anchors
  a relative budget (MediaPendingSeconds) with now() + make_interval,
  writer and readers sharing one clock; the local resolve budget stays
  monotonic app-side. A DB test pins that the remaining budget measured
  by the DB clock equals the requested one.

- enqueueMedia's waits (per-session order, global slot) only watched
  shutdown, so in a burst an already-expired job kept its goroutine and
  payload until it reached the front. Both waits now also watch the
  message's deadline; on expiry the job skips the resolver entirely and
  runs only the empty finalize (marker clear + promotion), which also
  unblocks the session's later messages. Covered by a queued-expiry
  test that finalizes while the only slot is deterministically held.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore(migrations): renumber to 216-221 after upstream took 213-215

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): start the local media budget before the append transaction

The DB anchors the durable fallback at insert-time now(), but the
local monotonic budget started only after AppendMessage returned — so
the resolver outlived the fallback by the append/commit latency, a
window where the deferred task is already claimable while the resolver
still runs and the agent reads a placeholder that binds moments later.
Capture the local deadline before calling AppendMessage, restoring the
ordering local-gives-up <= durable-fallback-fires. A slow-append test
pins that the resolver's context deadline is measured from the
pre-append instant.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore(migrations): renumber to 224-229 after upstream took 216-223

Verified against the merged tree: the numeric-prefix uniqueness test
passes and the full migration set applies cleanly from scratch.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): heartbeat the reconciler lease per row

One claim covers up to 50 rows under a single 2-minute lease, but the
batch is processed sequentially and each delete may run its full 30s
timeout — a few stalled deletes could outlive the lease mid-batch,
letting another replica reclaim the tail: duplicate concurrent
deletes, inflated attempt/backoff on rows whose owner was alive, and
skewed metrics.

The lease is now renewed before EACH row's settle work, so it only
ever needs to cover one row's worst case (invariant-tested: lease >=
2x the per-delete timeout). A renewal that matches no row means
another worker reclaimed it after a genuine expiry — the row is
skipped, leaving the new owner's state untouched. Test simulates a
mid-batch reclaim and pins the skip.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): dedup post media resources and make local writes atomic

A rich post may reference the same image_key/file_key in several spans.
The object key derives from (message, type, key), so duplicates uploaded
to the SAME key twice: LocalStorage.UploadStream truncated the
destination up front and removed it outright on a copy error, so a
second failing attempt destroyed the object the first success had
produced — leaving an attachment row pointing at nothing. A second
succeeding attempt instead produced two attachment rows for one object.

Collapse duplicate spans by (fetch type, platform key) before the
upload loop, and write local uploads through a temp file renamed into
place so a failed write can only discard its own temp file. Tests cover
a duplicated span uploading once and a failed re-upload leaving the
previous object intact.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): fence late-materializing PUTs with a tombstone schedule

A DELETE cannot be ordered against a PUT the client already abandoned:
the store may materialize the object after the delete completes. The
reconciler cleared the ledger row right after deleting, so such an
object had no row and nothing to reclaim it — which made the settle
delay the de-facto correctness barrier for the PUT/DELETE race, exactly
what the design says it must not be.

The row is now kept as a tombstone ('tombstoned' state, migration 226's
CHECK) and re-deleted on a widening schedule (15m, 1h, 6h, 24h, the
pass index carried in last_error), so a late materialization is
reclaimed by a later pass; only after the schedule is exhausted is the
row dropped. Claim, heartbeat, lease, and tenancy predicates are
unchanged — a tombstone is claimed exactly like any other due row. A
separate gauge reports tombstones so they cannot be mistaken for a
backlog of objects awaiting reclaim, and the header comment now states
precisely what state fences (bind/commit) versus what the schedule
fences (late PUTs).

Tests: the reviewer's interleaving — DELETE completes, the abandoned PUT
materializes right after, and the object is gone by the end of the
schedule — plus a full schedule walk asserting the object is counted
once and the row clears at the end.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): tombstones must re-delete, not re-ask the reference question

A tombstone revisit ran the same reference check as a first settle, so
an attachment carrying the same URL — a re-ingested copy of the object —
sent the row down the "referenced, keep it" branch: the object was kept
and the row cleared, abandoning the re-delete schedule that fences the
ORIGINAL object against an abandoned PUT. A tombstone has already been
judged unreferenced and deleted; it exists only to re-delete whatever
materializes later, so it now goes straight to the delete + schedule
tail (extracted as settleDeletedObject, shared with the first settle).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(channel): keep the tombstone schedule position in its own column

The re-delete pass index was encoded into last_error, which the failure
path also writes: one failed re-delete erased the position and restarted
the walk. A store failing intermittently could therefore keep a tombstone
alive indefinitely — every recovery would resume at pass 1 and the row
would never reach the end of the schedule to be dropped.

tombstone_pass is now its own column (the table is introduced in this PR,
so migration 226 carries it), advanced only by a successful delete, and
the tombstone write clears the now-stale last_error. Test walks the
schedule across a failed re-delete and asserts it resumes rather than
restarts, and that the row still terminates.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(lark): derive media object keys per chat message

The object key was derived from the platform message alone, so a second
ingest of the same Feishu message reused the first ingest's ledger row.
That row can be a tombstone (up to ~31h while the re-delete schedule
runs), and the intent upsert refuses anything that has left 'pending', so
the second ingest skipped the upload and silently produced a placeholder
with no attachment. A re-ingest is reachable: the inbound dedup claim is
reclaimable once 60s stale and the dedup row is only vacuumed after 24h.

Keying on the chat message the object will attach to keeps the two
ingests independent, and nothing leaks: each one's objects are covered by
its own ledger row.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* refactor(storage): route both local upload paths through one atomic write

UploadStream wrote through a temp file and renamed into place, but the
buffered Upload path still truncated the destination up front — the
destructive shape the stream path exists to avoid, one caller away from
coming back. Both now share writeAtomic, which also restores the 0644 the
direct write used (CreateTemp makes files 0600).

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* chore(channel): gofmt the media-pending append fields

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(storage): keep the local upload chmod best-effort

The rename-into-place rewrite made a failed chmod fail the whole upload.
CreateTemp's 0600 has to be widened to the 0644 the direct write used, but
an upload dir on a mount that ignores chmod (SMB/NFS/FUSE) accepted the
old direct write fine — turning those deployments' uploads into hard
errors would be a regression for a cosmetic property. Log and continue.

Tests pin 0644 on both upload paths, and that a failed buffered upload
leaves no temp litter and no damage to a previous object.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(channel): never re-delete an object an attachment references

The tombstone pass skipped the reference check and deleted unconditionally,
so a durable attachment carrying that URL lost the only object it can read
— the dangling attachment the intent ledger exists to prevent, and the
opposite of the posture every other path here takes ("a reclaimable orphan
beats a broken attachment").

The check now runs on every pass. A positive result on a tombstone is
unreachable by design — keys are per (chat message, resource) and a bind
cannot attach a key that has left 'pending' — so reaching it means an
invariant broke: keep the object, clear the row, log it, and count it on
a dedicated reconciler_tombstone_referenced_total counter. The test's
contract is flipped to assert the referenced object survives.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(storage): make the local staging file reclaimable after a crash

os.CreateTemp's random suffix meant a crash between the staging write and
the rename left a file nothing could name: the ledger records only the
final storage key, and DeleteObject removed only the object and its
sidecar. Each leftover can approach the 100 MiB resource cap and they
accumulate without bound.

The staging path is now derived from the object key, so DeleteObject
removes it alongside the object — which makes the media reconciler reclaim
it too, since the intent row is written before the upload. Opening it 0644
directly also drops the chmod the previous commit had to make best-effort.
Both read paths refuse the staging name (keys come from the request URL,
and a half-written body should not be readable); a user-supplied ".tmp"
extension is unaffected, since object keys are generated and never
dot-prefixed.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* chore(channel): renumber the media migrations after merging main

main took 224 (agent_task_session_rollout_missing), so the ledger group
moves to 225-230 and the cross-references inside the table migration follow.
main's CompleteTask also grew a sessionRolloutMissing parameter; the three
call sites this PR added to chat_input_ownership_test.go pass false.

Verified the way the numbering is meant to be verified: full migration set
applied from scratch on the merged tree, and the whole server suite run
against that database.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(channel): let Postgres compute every reconciler deadline

The reconciler built settle cutoffs, lease expiry, backoff and re-delete
times from the process clock and compared them against the database's
now(). A replica whose clock had drifted would therefore settle rows whose
upload was still in flight (the object is deleted and the bind then refuses
to attach — media silently lost), hand out leases that are born expired
(rows churn between workers, attempt/backoff inflate), or compress the
tombstone schedule that fences a late-materializing PUT.

The four settle queries now take durations and derive their timestamps from
now(), so every replica reads one clock. The parameter types are the guard:
an app-side timestamp can no longer be passed. Test asserts the persisted
lease, backoff and re-delete deadlines all track the database's now().

The generated code also picks up main's new agent_task_queue column in the
two RETURNING task.* queries this PR adds.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* chore(lark): drop unrelated gofmt-only churn from this PR

Six files carried whitespace/comment-reformatting with no functional
change, unrelated to the inbound media pipeline. Reverting them to the
base revision keeps the diff focused on the feature (75 -> 69 files):

  server/internal/service/empty_claim_cache.go
  server/internal/integrations/lark/markdown_detect.go
  server/internal/integrations/lark/ws_chunk_assembler.go
  server/internal/integrations/lark/ws_chunk_assembler_test.go
  server/internal/integrations/lark/ws_frame_test.go
  server/internal/integrations/lark/registration_test.go

Verified: `git diff -w` against these files was already empty, so no
behavior is affected. go vet clean; tests covering these files pass.

Co-authored-by: multica-agent <github@multica.ai>

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: Bohan-J <bohan@devv.ai>
Co-authored-by: multica-agent <github@multica.ai>
2026-07-27 15:44:39 +08:00

933 lines
36 KiB
Go

package handler
import (
"context"
"crypto/rand"
"encoding/hex"
"encoding/json"
"errors"
"log/slog"
"net/http"
"net/netip"
"strconv"
"time"
"github.com/go-chi/chi/v5"
"github.com/jackc/pgx/v5"
"github.com/jackc/pgx/v5/pgconn"
"github.com/jackc/pgx/v5/pgtype"
"github.com/multica-ai/multica/server/internal/analytics"
"github.com/multica-ai/multica/server/internal/auth"
"github.com/multica-ai/multica/server/internal/cloudruntime"
"github.com/multica-ai/multica/server/internal/daemonws"
"github.com/multica-ai/multica/server/internal/events"
"github.com/multica-ai/multica/server/internal/integrations/channel/engine"
composio "github.com/multica-ai/multica/server/internal/integrations/composio"
"github.com/multica-ai/multica/server/internal/integrations/ghsnapshot"
"github.com/multica-ai/multica/server/internal/integrations/lark"
"github.com/multica-ai/multica/server/internal/integrations/slack"
obsmetrics "github.com/multica-ai/multica/server/internal/metrics"
"github.com/multica-ai/multica/server/internal/middleware"
"github.com/multica-ai/multica/server/internal/realtime"
"github.com/multica-ai/multica/server/internal/service"
"github.com/multica-ai/multica/server/internal/storage"
"github.com/multica-ai/multica/server/internal/util"
"github.com/multica-ai/multica/server/internal/util/secretbox"
db "github.com/multica-ai/multica/server/pkg/db/generated"
"github.com/multica-ai/multica/server/pkg/featureflag"
"github.com/multica-ai/multica/server/pkg/llm"
)
// randomID returns a random 16-byte hex string used as a request ID for
// in-memory stores (model list, local skills, CLI update, etc.).
func randomID() string {
b := make([]byte, 16)
rand.Read(b)
return hex.EncodeToString(b)
}
type txStarter interface {
Begin(ctx context.Context) (pgx.Tx, error)
}
type dbExecutor interface {
Exec(ctx context.Context, sql string, arguments ...any) (pgconn.CommandTag, error)
Query(ctx context.Context, sql string, args ...any) (pgx.Rows, error)
QueryRow(ctx context.Context, sql string, args ...any) pgx.Row
}
type Config struct {
AllowSignup bool
AllowedEmails []string
AllowedEmailDomains []string
// DisableWorkspaceCreation, when true, makes POST /api/workspaces return
// 403 for every caller. There is no role/owner exception because the repo
// has no platform-admin concept; operators bootstrap the workspace with
// the flag off, then flip it on and restart so subsequent users join via
// invitation only. The public /api/config endpoint mirrors this flag so
// the UI can hide every "Create workspace" affordance — see #3433.
DisableWorkspaceCreation bool
// VCSIntegrationEnabled gates the self-hosted Git provider integration
// (Forgejo / Gitea / GitLab) at the deployment level, independent of whether
// MULTICA_VCS_SECRET_KEY is set. It is the product boundary: the feature is
// intended for self-hosted Multica only (where Multica and the Git instance
// can share a network), and is left off on the managed cloud — connect,
// rotate, and webhook handlers reject when it is false, and /api/config
// omits it so the UI hides the whole section rather than showing a
// "missing key" message a cloud user cannot act on. Populated from
// MULTICA_VCS_INTEGRATION_ENABLED; the self-host compose defaults it on.
VCSIntegrationEnabled bool
// PublicURL is the absolute base URL the API is reachable at from the
// public internet, with no trailing slash (e.g. "https://multica.ai").
// Used only to build webhook_url responses for autopilot webhook triggers
// — never for auth, routing, or workspace resolution. Empty when unset,
// in which case clients fall back to webhook_path + their own origin.
// Reading the public host from request headers (Host / X-Forwarded-Host)
// is intentionally avoided so a misconfigured reverse proxy cannot trick
// the server into minting webhook URLs pointing at an attacker-controlled
// host.
PublicURL string
// TrustedProxies are CIDRs whose source IP we trust to set
// X-Forwarded-For / X-Real-IP. Empty means "trust nothing": the rate
// limiter uses r.RemoteAddr exclusively. Populated via the
// MULTICA_TRUSTED_PROXIES env var (comma-separated CIDRs, e.g.
// "10.0.0.0/8,127.0.0.1/32"). This is specifically to keep the per-IP
// webhook limiter from being bypassed by a spoofed XFF on deployments
// without a header-stripping reverse proxy in front.
TrustedProxies []netip.Prefix
// CloudRuntimeFleetURL enables the SaaS-only remote Fleet adapter when set.
// Empty keeps self-hosted deployments explicit: cloud runtime endpoints
// return 503 instead of attempting to dial a hard-coded private service.
CloudRuntimeFleetURL string
CloudRuntimeFleetTimeout time.Duration
AttachmentDownloadMode string
AttachmentDownloadURLTTL time.Duration
// AttachmentFrameAncestors are trusted browser origins allowed to embed
// attachment preview responses. In production this should mirror the
// frontend/CORS origin allowlist so split app/api self-hosted deployments
// can frame API-hosted PDFs without allowing arbitrary third-party frames.
AttachmentFrameAncestors []string
// LLM* configure the basic LLM API layer (MUL-4238). They back the
// server-internal LLM helpers in pkg/llm (e.g. chat title generation).
// The generic OpenAI-compatible passthrough endpoints were removed in
// MUL-4309; LLM access is internal-only now. When both LLMAPIKey and
// LLMBaseURL are empty the layer is disabled and callers fall back
// silently (see maybeGenerateChatTitleAsync).
// - LLMAPIKey -> MULTICA_LLM_API_KEY
// - LLMBaseURL -> MULTICA_LLM_BASE_URL (OpenAI or any compatible gateway)
// - LLMDefaultModel -> MULTICA_LLM_DEFAULT_MODEL (used when a request omits `model`)
LLMAPIKey string
LLMBaseURL string
LLMDefaultModel string
// ServerVersion is the build version of the running API binary (the same
// value main.go stamps via -X main.version and reports on /metrics).
// Surfaced through /api/config so self-hosted operators can confirm which
// server build is deployed. Empty in dev builds.
ServerVersion string
}
type cloudRuntimeProxy interface {
Enabled() bool
Do(ctx context.Context, req cloudruntime.Request) (*cloudruntime.Response, error)
}
type RuntimeProfileRefreshNotifier interface {
NotifyRuntimeProfilesChanged(workspaceID, profileID string)
}
type WorkspaceSetRefreshNotifier interface {
NotifyWorkspacesChanged(userID string)
}
type Handler struct {
Queries *db.Queries
DB dbExecutor
TxStarter txStarter
Hub *realtime.Hub
DaemonHub *daemonws.Hub
DaemonProfileRefresh RuntimeProfileRefreshNotifier
DaemonWorkspaceRefresh WorkspaceSetRefreshNotifier
Bus *events.Bus
TaskService *service.TaskService
IssueService *service.IssueService
AutopilotService *service.AutopilotService
EmailService *service.EmailService
UpdateStore UpdateStore
ModelListStore ModelListStore
LocalSkillListStore LocalSkillListStore
LocalSkillImportStore LocalSkillImportStore
FeatureFlags *featureflag.Service
LivenessStore LivenessStore
HeartbeatScheduler HeartbeatScheduler
Storage storage.Storage
CFSigner *auth.CloudFrontSigner
Analytics analytics.Client
// Metrics is the shared business-metrics collector built by main.go.
// May be nil in tests / self-hosted with the metrics listener disabled;
// every Record* method is nil-safe and obsmetrics.RecordEvent treats a
// nil Metrics as "PostHog only".
Metrics *obsmetrics.BusinessMetrics
PATCache *auth.PATCache
DaemonTokenCache *auth.DaemonTokenCache
MembershipCache *auth.MembershipCache
WebhookRateLimiter WebhookRateLimiter
WebhookIPRateLimiter WebhookRateLimiter
WebhookAbsoluteIPRateLimiter WebhookRateLimiter
WebhookDeliveryWorker *WebhookDeliveryWorker
CloudRuntime cloudRuntimeProxy
// Lark integration. All three are nil when the Lark master key
// (MULTICA_LARK_SECRET_KEY) is unset; the corresponding HTTP
// handlers return 503 in that case so a misconfigured self-host
// deployment surfaces a clear error instead of silently using a
// zero key. Wired in cmd/server/router.go after handler.New.
LarkInstallations *lark.InstallationService
LarkBindingTokens *lark.BindingTokenService
// LarkRegistration owns the device-flow install lifecycle: begin
// a registration session against accounts.feishu.cn, poll, and
// on success write lark_installation + the installer's
// lark_user_binding in one DB transaction. Nil when the at-rest
// key is unset or the RegistrationService failed to construct at
// boot.
LarkRegistration *lark.RegistrationService
// LarkAPIClient is the live transport that backs SendInteractiveCard,
// PatchInteractiveCard, SendBindingPromptCard, GetBotInfo. The
// router wires the real Lark HTTP client whenever
// MULTICA_LARK_SECRET_KEY is set; tests that need a no-op
// behaviour can swap in `lark.NewStubAPIClient(...)` directly. The
// UI consults IsConfigured() to decide whether to surface install
// entry points.
LarkAPIClient lark.APIClient
// Composio integration (MUL-3720). Nil when COMPOSIO_API_KEY is unset;
// the composio HTTP handlers return 503 in that case. Wired in
// cmd/server/router.go after handler.New.
Composio *composio.Service
// ChannelSupervisor owns the per-installation supervisor goroutines
// that hold the §4.4 WS lease and drive each channel.Channel
// (MUL-3620 generalized the Feishu-only Hub into this channel-agnostic
// engine). The router constructs it UNCONDITIONALLY — it drives any
// channel type, not just Feishu, so it does not depend on the Lark
// master key; each platform registers its Factory only when configured
// (Feishu when MULTICA_LARK_SECRET_KEY is set). The router does NOT
// call Run; the process owner (main.go) starts it under a long-running
// context and joins via WaitWithTimeout (bounded, fenced by
// ShutdownTimeout) during graceful shutdown so the lease renewer yields
// cleanly when the DB is healthy without blocking process exit if the
// pool is frozen — at worst the next replica waits the full TTL.
ChannelSupervisor *engine.Supervisor
// ChannelRouter is the channel-agnostic inbound pipeline (the shared
// handler the Supervisor injects into every Channel). main.go calls
// Drain on it during shutdown, after the Supervisor has stopped
// delivering events, to flush debounced run triggers and join in-flight
// reply goroutines. Built unconditionally (even without Lark).
ChannelRouter *engine.Router
// ChannelMediaReconciler settles the channel-media intent ledger
// (uploaded-but-unbound object reclaim). Built in cmd/server/router.go
// where the storage backend exists; main.go starts it as an independent
// worker goroutine. Nil when no storage backend is configured.
ChannelMediaReconciler *service.ChannelMediaReconciler
// SlackInstall owns the bring-your-own-app Slack install lifecycle (register
// pasted tokens / list / revoke) and the at-rest encryption of each app's bot
// + app tokens (MUL-3666). Nil unless MULTICA_SLACK_SECRET_KEY is set.
SlackInstall *slack.InstallService
// SlackBindingTokens mints/redeems the user-binding tokens behind the
// "link your Slack account" prompt (MUL-3666). Nil unless Slack is
// configured (MULTICA_SLACK_SECRET_KEY set).
SlackBindingTokens *slack.BindingTokenService
// SlackHistory backs the agent-facing `multica chat history` command: it
// reads a chat session's bound Slack conversation on demand (MUL-3871). Nil
// unless Slack is configured; GetChatChannelHistory then reports "no channel
// integration". A future platform satisfies the same reader interface.
SlackHistory ChatChannelHistoryReader
// LLM is the basic LLM API layer (MUL-4238): a thin wrapper over the
// OpenAI Go SDK backing server-internal one-shot LLM helpers such as chat
// title generation. The generic passthrough endpoints were removed in
// MUL-4309, so it is internal-only now. Always non-nil (New builds it from
// Config); when unconfigured its Enabled() reports false and callers fall
// back silently.
LLM *llm.Client
// VCSSecretBox encrypts/decrypts per-workspace Git provider access tokens and
// webhook secrets at rest (Forgejo / Gitea / GitLab). Nil when
// MULTICA_VCS_SECRET_KEY is unset; the connect/webhook handlers return 503
// in that case so a misconfigured self-host deployment surfaces a clear
// error rather than silently storing plaintext. Wired in
// cmd/server/router.go after New.
VCSSecretBox *secretbox.Box
// PRRefresh drives the GitHub API snapshot pipeline for PR cards (MUL-5265):
// webhook / page-visit / TTL triggers → authenticated GraphQL fetch →
// head-SHA-guarded atomic snapshot write. Always non-nil, but inert (every
// trigger is a no-op) when GITHUB_APP_ID / GITHUB_APP_PRIVATE_KEY are unset,
// so the feature degrades cleanly on deployments without a private key.
// Wired in cmd/server/router.go after New.
PRRefresh *ghsnapshot.Manager
cfg Config
}
func New(queries *db.Queries, txStarter txStarter, hub *realtime.Hub, bus *events.Bus, emailService *service.EmailService, store storage.Storage, cfSigner *auth.CloudFrontSigner, analyticsClient analytics.Client, cfg Config, daemonHubs ...*daemonws.Hub) *Handler {
var executor dbExecutor
if candidate, ok := txStarter.(dbExecutor); ok {
executor = candidate
}
if analyticsClient == nil {
analyticsClient = analytics.NoopClient{}
}
if mode, ok := normalizeAttachmentDownloadMode(cfg.AttachmentDownloadMode); ok {
cfg.AttachmentDownloadMode = string(mode)
} else {
slog.Warn("invalid ATTACHMENT_DOWNLOAD_MODE, using auto", "value", cfg.AttachmentDownloadMode)
cfg.AttachmentDownloadMode = string(attachmentDownloadModeAuto)
}
if cfg.AttachmentDownloadURLTTL <= 0 {
cfg.AttachmentDownloadURLTTL = defaultAttachmentDownloadURLTTL
}
var daemonHub *daemonws.Hub
if len(daemonHubs) > 0 {
daemonHub = daemonHubs[0]
}
var daemonProfileRefresh RuntimeProfileRefreshNotifier
var daemonWorkspaceRefresh WorkspaceSetRefreshNotifier
if daemonHub != nil {
daemonProfileRefresh = daemonHub
daemonWorkspaceRefresh = daemonHub
}
taskSvc := service.NewTaskService(queries, txStarter, hub, bus, daemonHub)
taskSvc.Analytics = analyticsClient
h := &Handler{
Queries: queries,
DB: executor,
TxStarter: txStarter,
Hub: hub,
DaemonHub: daemonHub,
DaemonProfileRefresh: daemonProfileRefresh,
DaemonWorkspaceRefresh: daemonWorkspaceRefresh,
Bus: bus,
TaskService: taskSvc,
IssueService: service.NewIssueService(queries, txStarter, bus, analyticsClient, taskSvc),
AutopilotService: service.NewAutopilotService(queries, txStarter, bus, taskSvc),
EmailService: emailService,
UpdateStore: NewInMemoryUpdateStore(),
ModelListStore: NewInMemoryModelListStore(),
LocalSkillListStore: NewInMemoryLocalSkillListStore(),
LocalSkillImportStore: NewInMemoryLocalSkillImportStore(),
LivenessStore: NewNoopLivenessStore(),
HeartbeatScheduler: NewPassthroughHeartbeatScheduler(queries),
Storage: store,
CFSigner: cfSigner,
Analytics: analyticsClient,
WebhookRateLimiter: NewMemoryWebhookRateLimiter(DefaultWebhookRateLimit()),
WebhookIPRateLimiter: NewMemoryWebhookIPRateLimiter(DefaultWebhookIPRateLimit()),
WebhookAbsoluteIPRateLimiter: NewMemoryWebhookAbsoluteIPRateLimiter(DefaultWebhookAbsoluteIPRateLimit()),
CloudRuntime: cloudruntime.NewClient(cloudruntime.Config{
BaseURL: cfg.CloudRuntimeFleetURL,
Timeout: cfg.CloudRuntimeFleetTimeout,
}),
LLM: llm.New(llm.Config{
APIKey: cfg.LLMAPIKey,
BaseURL: cfg.LLMBaseURL,
DefaultModel: cfg.LLMDefaultModel,
}),
cfg: cfg,
}
h.WebhookDeliveryWorker = NewWebhookDeliveryWorker(h)
// GitHub API snapshot pipeline for PR cards (MUL-5265). Built
// unconditionally but inert (every trigger no-ops) when the App private key
// is unconfigured, so the feature degrades cleanly. main.go calls
// h.PRRefresh.Start(ctx) to launch its worker pool + TTL sweeper.
ghClient, err := ghsnapshot.NewClientFromEnv()
if err != nil {
// Malformed key is operator-actionable; the pipeline stays disabled.
slog.Warn("github: PR snapshot pipeline disabled (invalid App private key)", "err", err)
}
h.PRRefresh = ghsnapshot.NewManager(ghClient, queries, txStarter, h.broadcastPRSnapshotApplied)
return h
}
func writeJSON(w http.ResponseWriter, status int, v any) {
// Marshal the body up front so we can advertise an accurate Content-Length
// header. Streaming straight into the ResponseWriter after WriteHeader forces
// net/http into chunked transfer encoding, which omits Content-Length; buffering
// first lets clients (and proxies) see the exact body size.
body, err := json.Marshal(v)
if err != nil {
// Fall back to a minimal, self-describing error payload rather than leaving
// the client with a half-written response.
body = []byte(`{"error":"failed to encode response"}`)
status = http.StatusInternalServerError
}
// Match the trailing newline that json.Encoder.Encode historically appended.
body = append(body, '\n')
w.Header().Set("Content-Type", "application/json")
w.Header().Set("Content-Length", strconv.Itoa(len(body)))
w.WriteHeader(status)
_, _ = w.Write(body)
}
// writeMeasuredJSON behaves like writeJSON but returns the encoded body size so
// callers can record payload bytes in slow-endpoint diagnostics. It measures the
// uncompressed JSON length and is unrelated to transport compression.
func writeMeasuredJSON(w http.ResponseWriter, status int, v any) (int, error) {
body, err := json.Marshal(v)
if err != nil {
writeError(w, http.StatusInternalServerError, "failed to encode response")
return 0, err
}
body = append(body, '\n')
w.Header().Set("Content-Type", "application/json")
w.Header().Set("Content-Length", strconv.Itoa(len(body)))
w.WriteHeader(status)
if _, err := w.Write(body); err != nil {
return len(body), err
}
return len(body), nil
}
func writeError(w http.ResponseWriter, status int, msg string) {
writeJSON(w, status, map[string]string{"error": msg})
}
// Thin wrappers around util functions.
//
// parseUUID is intentionally the panicking variant: any handler call site
// reachable here is expected to feed a UUID that is either (a) a sqlc round-trip
// of a DB-sourced value, or (b) a raw request input that has already been
// validated upstream. A panic here means an unguarded user-input string slipped
// in — that is a real bug we want surfaced loudly (chi's middleware.Recoverer
// converts it to a 500) instead of silently corrupting data via a zero UUID.
//
// For unvalidated user input at request boundaries, use parseUUIDOrBadRequest
// (writes 400) — never feed raw chi.URLParam / request-body strings into
// parseUUID directly when the call writes to the database.
func parseUUID(s string) pgtype.UUID { return util.MustParseUUID(s) }
func uuidToString(u pgtype.UUID) string { return util.UUIDToString(u) }
func textToPtr(t pgtype.Text) *string { return util.TextToPtr(t) }
func ptrToText(s *string) pgtype.Text { return util.PtrToText(s) }
func strToText(s string) pgtype.Text { return util.StrToText(s) }
func timestampToString(t pgtype.Timestamptz) string { return util.TimestampToString(t) }
func timestampToPtr(t pgtype.Timestamptz) *string { return util.TimestampToPtr(t) }
func dateToPtr(d pgtype.Date) *string { return util.DateToPtr(d) }
func uuidToPtr(u pgtype.UUID) *string { return util.UUIDToPtr(u) }
// uuidsToStrings maps a UUID array column to string ids, skipping NULL/invalid
// entries. Returns nil (not an empty slice) when there is nothing to emit so
// `omitempty` JSON fields drop out cleanly (MUL-4195).
func uuidsToStrings(us []pgtype.UUID) []string {
if len(us) == 0 {
return nil
}
out := make([]string, 0, len(us))
for _, u := range us {
if u.Valid {
out = append(out, uuidToString(u))
}
}
if len(out) == 0 {
return nil
}
return out
}
// uuidStringsOrEmpty preserves the distinction between a modern, authoritative
// empty UUID-array value (`[]`) and a field omitted by a legacy server. Delivery
// receipts use this so clients never mistake zero delivered comments for an
// unknown receipt and fall back to the enqueue-time plan.
func uuidStringsOrEmpty(us []pgtype.UUID) []string {
out := uuidsToStrings(us)
if out == nil {
return []string{}
}
return out
}
func int8ToPtr(v pgtype.Int8) *int64 { return util.Int8ToPtr(v) }
func int4ToPtr(v pgtype.Int4) *int32 { return util.Int4ToPtr(v) }
func ptrToInt4(v *int32) pgtype.Int4 { return util.PtrToInt4(v) }
// parseUUIDOrBadRequest validates a UUID string sourced from user input
// (URL params, request body, headers). On invalid input it writes a 400
// response and returns ok=false; callers must return immediately.
//
// Use this anywhere a malformed UUID would otherwise reach a write query
// (DELETE / UPDATE) — the silent zero-UUID behavior of the old ParseUUID
// caused real silent-data-loss bugs (#1661).
func parseUUIDOrBadRequest(w http.ResponseWriter, s, fieldName string) (pgtype.UUID, bool) {
u, err := util.ParseUUID(s)
if err != nil {
writeError(w, http.StatusBadRequest, "invalid "+fieldName)
return pgtype.UUID{}, false
}
return u, true
}
func parseUUIDSliceOrBadRequest(w http.ResponseWriter, ids []string, fieldName string) ([]pgtype.UUID, bool) {
uuids := make([]pgtype.UUID, len(ids))
for i, id := range ids {
u, err := util.ParseUUID(id)
if err != nil {
writeError(w, http.StatusBadRequest, "invalid "+fieldName)
return nil, false
}
uuids[i] = u
}
return uuids, true
}
// publish sends a domain event through the event bus.
func (h *Handler) publish(eventType, workspaceID, actorType, actorID string, payload any) {
h.Bus.Publish(events.Event{
Type: eventType,
WorkspaceID: workspaceID,
ActorType: actorType,
ActorID: actorID,
Payload: payload,
})
}
func (h *Handler) notifyDaemonWorkspacesChanged(userIDs ...string) {
if h.DaemonWorkspaceRefresh == nil {
return
}
seen := make(map[string]struct{}, len(userIDs))
for _, userID := range userIDs {
if userID == "" {
continue
}
if _, ok := seen[userID]; ok {
continue
}
seen[userID] = struct{}{}
h.DaemonWorkspaceRefresh.NotifyWorkspacesChanged(userID)
}
}
// publishTask is publish() plus a TaskID hint so the realtime layer can route
// the event to the per-task scope rather than the whole workspace.
func (h *Handler) publishTask(eventType, workspaceID, actorType, actorID, taskID string, payload any) {
h.Bus.Publish(events.Event{
Type: eventType,
WorkspaceID: workspaceID,
ActorType: actorType,
ActorID: actorID,
TaskID: taskID,
Payload: payload,
})
}
// publishChat is publish() plus a ChatSessionID hint so the realtime layer
// can route the event to the per-chat-session scope.
func (h *Handler) publishChat(eventType, workspaceID, actorType, actorID, chatSessionID string, payload any) {
h.Bus.Publish(events.Event{
Type: eventType,
WorkspaceID: workspaceID,
ActorType: actorType,
ActorID: actorID,
ChatSessionID: chatSessionID,
Payload: payload,
})
}
func isNotFound(err error) bool {
return errors.Is(err, pgx.ErrNoRows)
}
func isUniqueViolation(err error) bool {
var pgErr *pgconn.PgError
return errors.As(err, &pgErr) && pgErr.Code == "23505"
}
// isCheckViolation reports whether err is a PostgreSQL CHECK constraint
// violation (SQLSTATE 23514). Used to translate column-level CHECK failures
// into a 4xx instead of a generic 500.
func isCheckViolation(err error) bool {
var pgErr *pgconn.PgError
return errors.As(err, &pgErr) && pgErr.Code == "23514"
}
func requestUserID(r *http.Request) string {
return r.Header.Get("X-User-ID")
}
// resolveActor determines whether the request is from an agent or a human member.
//
// First-class signal: X-Actor-Source set to "task_token" means the request
// authenticated via an `mat_` task-scoped token. The auth middleware sets
// that header (and stripped any client-supplied value first), so it is
// authoritative — the bound (agent_id, task_id) cannot be forged or
// stripped by the agent process. This is the path MUL-2600 relies on to
// reject agent-process traffic on owner-only endpoints.
//
// Fallback signal (legacy CLI / member-token paths): the request MUST
// carry both X-Agent-ID and a valid X-Task-ID, and the task must belong
// to the claimed agent. Otherwise we fall back to "member".
//
// X-Agent-ID alone is not trusted: any workspace member can guess or observe
// an agent's UUID, and a member-supplied X-Agent-ID would otherwise let that
// member impersonate the agent and bypass the private-agent gate (#2359
// review). The daemon always pairs the two headers, so requiring both has
// no effect on legitimate agent callers but closes the impersonation path.
//
// Returns ("agent", agentID) on success, ("member", userID) otherwise.
func (h *Handler) resolveActor(r *http.Request, userID, workspaceID string) (actorType, actorID string) {
if r.Header.Get("X-Actor-Source") == "task_token" {
// Server-set header — auth middleware also forced X-Agent-ID
// from the token row. Trust it directly without re-querying.
return "agent", r.Header.Get("X-Agent-ID")
}
agentID := r.Header.Get("X-Agent-ID")
if agentID == "" {
return "member", userID
}
taskID := r.Header.Get("X-Task-ID")
if taskID == "" {
slog.Debug("resolveActor: X-Agent-ID present but X-Task-ID missing, refusing to trust agent identity", "agent_id", agentID)
return "member", userID
}
agentUUID, err := util.ParseUUID(agentID)
if err != nil {
slog.Debug("resolveActor: X-Agent-ID is not a valid UUID, falling back to member", "agent_id", agentID)
return "member", userID
}
// Validate the agent exists in the target workspace.
agent, err := h.Queries.GetAgent(r.Context(), agentUUID)
if err != nil || uuidToString(agent.WorkspaceID) != workspaceID {
slog.Debug("resolveActor: X-Agent-ID rejected, agent not found or workspace mismatch", "agent_id", agentID, "workspace_id", workspaceID)
return "member", userID
}
taskUUID, err := util.ParseUUID(taskID)
if err != nil {
slog.Debug("resolveActor: X-Task-ID is not a valid UUID, falling back to member", "task_id", taskID)
return "member", userID
}
task, err := h.Queries.GetAgentTask(r.Context(), taskUUID)
if err != nil || uuidToString(task.AgentID) != agentID {
slog.Debug("resolveActor: X-Task-ID rejected, task not found or agent mismatch", "agent_id", agentID, "task_id", taskID)
return "member", userID
}
return "agent", agentID
}
func requireUserID(w http.ResponseWriter, r *http.Request) (string, bool) {
userID := requestUserID(r)
if userID == "" {
writeError(w, http.StatusUnauthorized, "user not authenticated")
return "", false
}
return userID, true
}
// resolveWorkspaceID returns the workspace UUID for this request. Delegates
// to middleware.ResolveWorkspaceIDFromRequest so middleware-protected routes
// and middleware-less routes (e.g. /api/upload-file) share identical
// resolution behavior — including slug → UUID translation via the DB.
//
// Returns "" when no workspace identifier was provided or a slug was provided
// but doesn't match any workspace.
func (h *Handler) resolveWorkspaceID(r *http.Request) string {
return middleware.ResolveWorkspaceIDFromRequest(r, h.Queries)
}
// ctxMember returns the workspace member from context (set by workspace middleware).
func ctxMember(ctx context.Context) (db.Member, bool) {
return middleware.MemberFromContext(ctx)
}
// ctxWorkspaceID returns the workspace ID from context (set by workspace middleware).
func ctxWorkspaceID(ctx context.Context) string {
return middleware.WorkspaceIDFromContext(ctx)
}
// workspaceIDFromURL returns the workspace ID from context (preferred) or chi URL param (fallback).
func workspaceIDFromURL(r *http.Request, param string) string {
if id := middleware.WorkspaceIDFromContext(r.Context()); id != "" {
return id
}
return chi.URLParam(r, param)
}
// workspaceMember returns the member from middleware context, or falls back to a DB
// lookup when the handler is called directly (e.g. in tests).
func (h *Handler) workspaceMember(w http.ResponseWriter, r *http.Request, workspaceID string) (db.Member, bool) {
if m, ok := ctxMember(r.Context()); ok {
return m, true
}
return h.requireWorkspaceMember(w, r, workspaceID, "workspace not found")
}
func roleAllowed(role string, roles ...string) bool {
for _, candidate := range roles {
if role == candidate {
return true
}
}
return false
}
func countOwners(members []db.Member) int {
owners := 0
for _, member := range members {
if member.Role == "owner" {
owners++
}
}
return owners
}
func (h *Handler) getWorkspaceMember(ctx context.Context, userID, workspaceID string) (db.Member, error) {
userUUID, err := util.ParseUUID(userID)
if err != nil {
return db.Member{}, err
}
wsUUID, err := util.ParseUUID(workspaceID)
if err != nil {
return db.Member{}, err
}
return h.Queries.GetMemberByUserAndWorkspace(ctx, db.GetMemberByUserAndWorkspaceParams{
UserID: userUUID,
WorkspaceID: wsUUID,
})
}
func (h *Handler) requireWorkspaceMember(w http.ResponseWriter, r *http.Request, workspaceID, notFoundMsg string) (db.Member, bool) {
if workspaceID == "" {
writeError(w, http.StatusBadRequest, "workspace_id is required")
return db.Member{}, false
}
userID, ok := requireUserID(w, r)
if !ok {
return db.Member{}, false
}
member, err := h.getWorkspaceMember(r.Context(), userID, workspaceID)
if err != nil {
writeError(w, http.StatusNotFound, notFoundMsg)
return db.Member{}, false
}
return member, true
}
func (h *Handler) requireWorkspaceRole(w http.ResponseWriter, r *http.Request, workspaceID, notFoundMsg string, roles ...string) (db.Member, bool) {
member, ok := h.requireWorkspaceMember(w, r, workspaceID, notFoundMsg)
if !ok {
return db.Member{}, false
}
if !roleAllowed(member.Role, roles...) {
writeError(w, http.StatusForbidden, "insufficient permissions")
return db.Member{}, false
}
return member, true
}
// isWorkspaceEntity checks whether a user_id belongs to the given workspace,
// as either a member or an agent depending on userType.
func (h *Handler) isWorkspaceEntity(ctx context.Context, userType, userID, workspaceID string) bool {
switch userType {
case "member":
_, err := h.getWorkspaceMember(ctx, userID, workspaceID)
return err == nil
case "agent":
userUUID, err := util.ParseUUID(userID)
if err != nil {
return false
}
wsUUID, err := util.ParseUUID(workspaceID)
if err != nil {
return false
}
_, err = h.Queries.GetAgentInWorkspace(ctx, db.GetAgentInWorkspaceParams{
ID: userUUID,
WorkspaceID: wsUUID,
})
return err == nil
default:
return false
}
}
func (h *Handler) loadIssueForUser(w http.ResponseWriter, r *http.Request, issueID string) (db.Issue, bool) {
if _, ok := requireUserID(w, r); !ok {
return db.Issue{}, false
}
workspaceID := h.resolveWorkspaceID(r)
if workspaceID == "" {
writeError(w, http.StatusBadRequest, "workspace_id is required")
return db.Issue{}, false
}
// Try identifier format first (e.g., "JIA-42"). resolveIssueByIdentifier
// silently returns false for non-identifier strings, falling through to
// the UUID path below.
if issue, ok := h.resolveIssueByIdentifier(r.Context(), issueID, workspaceID); ok {
return issue, true
}
issueUUID, err := util.ParseUUID(issueID)
if err != nil {
// Not a valid UUID and didn't match identifier format → 404 (consistent
// with previous silent-zero behavior, which would also have produced 404).
writeError(w, http.StatusNotFound, "issue not found")
return db.Issue{}, false
}
wsUUID, err := util.ParseUUID(workspaceID)
if err != nil {
writeError(w, http.StatusBadRequest, "invalid workspace_id")
return db.Issue{}, false
}
issue, err := h.Queries.GetIssueInWorkspace(r.Context(), db.GetIssueInWorkspaceParams{
ID: issueUUID,
WorkspaceID: wsUUID,
})
if err != nil {
writeError(w, http.StatusNotFound, "issue not found")
return db.Issue{}, false
}
return issue, true
}
// resolveIssueByIdentifier tries to look up an issue by "PREFIX-NUMBER" format.
func (h *Handler) resolveIssueByIdentifier(ctx context.Context, id, workspaceID string) (db.Issue, bool) {
parts := splitIdentifier(id)
if parts == nil {
return db.Issue{}, false
}
if workspaceID == "" {
return db.Issue{}, false
}
wsUUID, err := util.ParseUUID(workspaceID)
if err != nil {
return db.Issue{}, false
}
issue, err := h.Queries.GetIssueByNumber(ctx, db.GetIssueByNumberParams{
WorkspaceID: wsUUID,
Number: parts.number,
})
if err != nil {
return db.Issue{}, false
}
return issue, true
}
type identifierParts struct {
prefix string
number int32
}
func splitIdentifier(id string) *identifierParts {
idx := -1
for i := len(id) - 1; i >= 0; i-- {
if id[i] == '-' {
idx = i
break
}
}
if idx <= 0 || idx >= len(id)-1 {
return nil
}
numStr := id[idx+1:]
num := 0
for _, c := range numStr {
if c < '0' || c > '9' {
return nil
}
num = num*10 + int(c-'0')
}
if num <= 0 {
return nil
}
return &identifierParts{prefix: id[:idx], number: int32(num)}
}
// getIssuePrefix fetches the issue_prefix for a workspace.
// Falls back to generating a prefix from the workspace name if the stored
// prefix is empty (e.g. workspaces created before the prefix was introduced).
func (h *Handler) getIssuePrefix(ctx context.Context, workspaceID pgtype.UUID) string {
ws, err := h.Queries.GetWorkspace(ctx, workspaceID)
if err != nil {
return ""
}
if ws.IssuePrefix != "" {
return ws.IssuePrefix
}
return generateIssuePrefix(ws.Name)
}
func (h *Handler) loadAgentForUser(w http.ResponseWriter, r *http.Request, agentID string) (db.Agent, bool) {
if _, ok := requireUserID(w, r); !ok {
return db.Agent{}, false
}
workspaceID := h.resolveWorkspaceID(r)
if workspaceID == "" {
writeError(w, http.StatusBadRequest, "workspace_id is required")
return db.Agent{}, false
}
agentUUID, ok := parseUUIDOrBadRequest(w, agentID, "agent id")
if !ok {
return db.Agent{}, false
}
wsUUID, ok := parseUUIDOrBadRequest(w, workspaceID, "workspace id")
if !ok {
return db.Agent{}, false
}
agent, err := h.Queries.GetAgentInWorkspace(r.Context(), db.GetAgentInWorkspaceParams{
ID: agentUUID,
WorkspaceID: wsUUID,
})
if err != nil {
writeError(w, http.StatusNotFound, "agent not found")
return db.Agent{}, false
}
if agent.Kind != "user" {
writeError(w, http.StatusNotFound, "agent not found")
return db.Agent{}, false
}
return agent, true
}
func (h *Handler) loadInboxItemForUser(w http.ResponseWriter, r *http.Request, itemID string) (db.InboxItem, bool) {
userID, ok := requireUserID(w, r)
if !ok {
return db.InboxItem{}, false
}
workspaceID := h.resolveWorkspaceID(r)
if workspaceID == "" {
writeError(w, http.StatusBadRequest, "workspace_id is required")
return db.InboxItem{}, false
}
itemUUID, ok := parseUUIDOrBadRequest(w, itemID, "inbox item id")
if !ok {
return db.InboxItem{}, false
}
wsUUID, ok := parseUUIDOrBadRequest(w, workspaceID, "workspace id")
if !ok {
return db.InboxItem{}, false
}
item, err := h.Queries.GetInboxItemInWorkspace(r.Context(), db.GetInboxItemInWorkspaceParams{
ID: itemUUID,
WorkspaceID: wsUUID,
})
if err != nil {
writeError(w, http.StatusNotFound, "inbox item not found")
return db.InboxItem{}, false
}
if item.RecipientType != "member" || uuidToString(item.RecipientID) != userID {
writeError(w, http.StatusNotFound, "inbox item not found")
return db.InboxItem{}, false
}
return item, true
}