Files
OwnCord/Server/api/emoji_handler.go
T
J3vbandClaude Fable 5 f5faf82a60 infra: observability, backups, guardrails, and deployment hardening (#1376)
* docs: add infrastructure roadmap plan

Records the verified recommendations from an infrastructure review in three
tracks: raising the single-instance ceiling, cheap seams for a possible
multi-instance future, and ops hygiene. Includes explicit anti-recommendations
and sequencing. Security-sensitive detail is intentionally excluded per
docs/security.md.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* feat(server): real health checks and saturation metrics

/api/v1/metrics now exposes signals that were already computed in memory but
never surfaced: reconnect replay tier hits, event-persister counters, SQLite
writer-pool wait stats, aggregate per-client backpressure counters (including
previously invisible low-priority drops), and permission-cache hit/miss.

/health now returns a real verdict: hub dispatch-loop liveness, a bounded
database ping, and a free-disk check, returning 503 with a subsystem reason
when degraded. Checks are cached so the unauthenticated endpoint cannot
amplify load. The hub's panic breaker now exits the process so a supervisor
can restart it, instead of leaving broadcast delivery silently dead while
clients still appear online.

OTel instruments that were declared but never recorded are now wired
(ws_active_connections, ws_broadcast_latency_seconds, ws_messages_total,
ws_events_dropped_total, voice gauges) or removed (db_query_duration_seconds).
Also corrects the docs/api.md description of broadcast_drops, which counts
hub-queue overflow, not client send-queue overflow.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* feat(server): implement scheduled backups, retention, and backup verification

The backup_schedule and backup_retention settings have existed in the admin
panel and API since the initial schema but were never read by any code. The
15-minute maintenance loop now enforces them: a scheduled backup is taken
when the newest backup on disk is older than the schedule interval (manual
backups reset the clock), and retention prunes backups older than the
configured days while always keeping the newest one.

Backups are now verified with PRAGMA integrity_check immediately after
VACUUM INTO (a failed backup is removed rather than listed as restorable)
and again before a restore may overwrite the live database. A failed VACUUM
INTO also cleans up its partial output file — but never a pre-existing one.

The backup directory is configurable via a new backup.dir key (default
data/backups) so operators can point backups at another disk or an off-host
mount, mirroring the SetDatabasePath plumb.

Restore-handler tests now use real SQLite fixtures (the integrity gate
correctly refuses text files) with the mid-copy failure injected through a
test-only copy hook. Also adds audited gosec suppressions to the Windows
disk-free syscall added in the previous commit, which the Windows lint leg
flagged.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* feat(server): capacity and failure-mode guardrails

- server.max_ws_connections: optional cap on concurrent WebSocket clients,
  checked before the upgrade with a 503 + Retry-After; rejections are counted
  and exposed as ws_conn_rejects in /api/v1/metrics.
- Single-process database lock: an OS-level advisory lock (flock / exclusive
  handle) beside the SQLite file makes a second server process fail fast with
  a clear message instead of silently fighting the first over process-local
  state. A bounded retry covers the self-update/restore restart handoff, and
  the lock mechanism failing (e.g. network filesystems) only warns.
- Disk-space awareness: boot-time warnings for the data and backup volumes,
  plus a disk_free_mb metrics field, via a small cross-platform diskutil
  package (already used by /health).
- Upload storage failures: storage.Save now marks server-side filesystem
  failures with a sentinel (storage.ErrIO); handlers return 507 for those
  instead of blaming the client with a 400, and the emoji route stops echoing
  raw storage errors (which embed absolute paths) into responses.
- Unknown config keys now warn at startup — a typo like admin_alowed_cidrs
  previously kept the default silently while the operator believed the
  setting changed. Never fatal: newer servers tolerate older configs.
- Admin settings honesty: the three stored-but-inert settings (server_icon,
  max_upload_bytes, voice_quality) are shown read-only with a note pointing
  at the real config.yaml keys, instead of pretending to apply.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* perf(db): write-path efficiency and capacity knobs

- channel_focus/mark_read now skip the read-state UPSERT when the stored row
  already matches (same last_message_id, no mentions) — refocus events fire
  at up to 10/s/user and every no-op write still occupied the single SQLite
  writer connection. The extra existence check runs on the reader pool, which
  doesn't serialize. Same shape as the session-touch throttle.
- DeleteExpiredSessions is now sargable: migration 031 normalizes legacy
  expiry formats to the RFC3339-Z layout the server writes and indexes
  expires_at, replacing the strftime full-table scan that ran on the writer
  every 15 minutes.
- Boot-time ANALYZE runs only when a migration actually applied; unchanged
  schemas get the cheap PRAGMA optimize instead (which also covers
  crash-restarts that never reached the shutdown optimize).
- The read/write SQL router gets a table-driven test with explicit expected
  values (INSERT ... RETURNING must hit the writer despite being :one).
- New knobs, all defaulting to current behavior: database.max_readers,
  security.auth_rate_limit_multiplier (for shared-NAT communities),
  event_persistence.replay_ring_size and replay_cold_limit.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* fix(server): shutdown lifecycle ordering

- The event pruner and maintenance loop are now joined (bounded) before the
  database closes: bgCtx cancellation used to run AFTER database.Close via
  LIFO defers, contradicting its own comment, and neither goroutine was ever
  waited on — a mid-tick scheduled backup or prune could still hold the
  writer while the pool tore down. StartEventPruner returns a done channel
  with the same join contract EventPersister.Stop already had.
- srv.Shutdown now runs before hub.GracefulStop, so in-flight HTTP handlers'
  broadcasts still reach a live hub and the event persister instead of
  vanishing from the replay/event store across a restart. Shutdown does not
  wait on hijacked WebSocket connections, so the swap adds no delay.
- GracefulStopContext threads the 30s shutdown budget into the hub: the 5s
  client-notice window (matching the countdown clients are shown) ends early
  when the budget expires, and is skipped entirely when nobody is connected —
  early-return startup paths and idle servers no longer sleep 5s for an
  audience of zero.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* build(deploy): systemd unit, compose hardening, boot-smoked releases, CI polish

- deploy/owncord.service: hardened systemd unit template with the two
  verified caveats encoded (install dir stays writable for self-update under
  ProtectSystem=strict; CAP_NET_BIND_SERVICE for ACME's :80), plus a
  'Linux (systemd)' deployment docs section — the Linux service story was
  previously 'Docker or nothing'.
- New 'Reverse Proxy Topology' docs section with a working nginx snippet and
  the correct signaling-vs-media distinction: /livekit/* is already proxied
  by the server, only WebRTC media ports must be directly reachable.
- docker-compose: log rotation, commented resource limits, and a healthcheck
  backed by a new 'chatserver healthcheck' subcommand (the distroless image
  has no shell) that probes /health without config side effects.
- release.yml: a concurrency group (queue, never cancel), and boot-smoke
  gates — the freshly built server binaries and the Docker image are cold
  booted and probed healthy BEFORE anything is signed or pushed. The release
  feed drives signed self-updates, so a binary that compiles but dies on
  boot previously would have shipped itself to every auto-updating instance.
- ci.yml: client-check/client-tests move to ubuntu with the reasoning
  recorded (no win32 code paths, LF enforced repo-wide); admin-e2e gets a
  written graduation criterion instead of an open-ended non-blocking status.
- docs: Tailscale guide notes the CGNAT range vs the default admin CIDRs;
  architecture overview records presence/voice state as the fifth
  single-instance blocker and the macOS client scope decision.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* perf(server): measured load tooling, narrowed invalidation, presence coalescing, storage and CIDR seams

- Fix scripts/k6/ws-load.js against the real wire protocol: envelope-wrapped
  frames, correct message types (typing_start, presence_update), the correct
  /api/v1/ws path, and thresholds that fail a run where nobody authenticated
  or went ready — the script had drifted to pre-envelope framing and reported
  100% green while every auth failed on the first frame. A new
  workflow_dispatch-only load-baseline workflow boots a real server, seeds
  users through the setup/invite APIs, runs the script, and uploads the k6
  summary plus a metrics snapshot for before/after comparison.
- Role-scoped channel-override changes now evict only the affected role's
  members from the permission cache (fail-safe: unreadable member list still
  flushes everything). InvalidateAll here repopulated every connected user —
  two reads each — synchronously inside the admin request via
  RefreshChannelVisibility, a stampede that scaled with total population
  rather than the role's size. Same pattern the per-user override endpoints
  already used.
- Connect/disconnect presence broadcasts now pass through a 300ms latest-wins
  coalescer (QueuePresence): each un-coalesced presence change is a sequenced
  global broadcast (an O(clients) fan-out under seqMu), so a reconnect storm
  fired O(users) of them from the connect critical path. A flap inside the
  window collapses to its final state; the wire format, seq ordering, and
  replay behaviour are unchanged, and the delivery path (BroadcastPresence)
  is untouched.
- Storage seam: api handlers now consume a FileStore interface (consumer-side,
  same pattern as service.Store) with Open returning a seekable storage.File —
  writing down the contract (range-request seeks included) an alternative
  backend would have to meet, without building one.
- The metrics surfaces and the LiveKit webhook/health endpoints get their own
  allowlist keys (metrics_allowed_cidrs, livekit_webhook_allowed_cidrs, both
  defaulting to admin_allowed_cidrs), so a central Prometheus scraper or an
  externally-hosted LiveKit no longer requires widening the admin panel's
  perimeter. Startup now also warns when admin_allowed_cidrs is customized
  while trusted_proxies is empty — behind a proxy or container network the
  check would otherwise compare the proxy's private address, not the client's.
- The container healthcheck probe now PINS the server's own certificate from
  disk (VerifyConnection, exact-match) instead of skipping TLS verification,
  addressing the CodeQL finding on the previous commit; WebPKI verification
  is used when no local cert exists (ACME).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* fix(server): address self-review findings on the hardening branch

Seven fixes from a high-effort review of the full branch diff:

- healthcheck CLI now works under tls.mode acme: it overrides ServerName
  with the configured domain for WebPKI verification instead of pinning a
  cert that doesn't exist (or is stale) in that mode. Previously an ACME
  deployment's container healthcheck failed forever.
- /health pings the READER pool (new db.PingRead): the writer ping queued
  behind a scheduled backup's VACUUM INTO and reported the server degraded
  for the whole backup — which an autoheal watchdog would turn into a
  nightly mid-backup restart.
- /health runs its cached checks under context.WithoutCancel so a probe
  that disconnects mid-request cannot poison the shared cache with a false
  degraded verdict for the next 5 seconds.
- The token CLI uses a new db.OpenShared that skips the single-process
  lock: minting a token against a running server is safe under WAL and was
  a documented workflow the lock had broken.
- The per-user TOTP failure cap is no longer scaled by
  security.auth_rate_limit_multiplier — that knob exists for per-IP limits;
  scaling the only cross-IP brute-force defence multiplied an attacker's
  distributed guess budget. Mirrors the unscaled per-user login threshold.
- A direct presence_update now drops the user's queued entry in the
  connect/disconnect coalescer, so a stale connect-time presence can no
  longer flush 300ms later over the user's fresher chosen status.
- The scheduled-backup filename collision loop breaks on any stat error
  and bounds its suffix probing, instead of spinning the maintenance
  goroutine forever on a persistent EACCES.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

* test(admin): real SQLite fixture for the merged Close-failure restore test

TestHandleRestoreBackup_RestartsWhenCloseFails arrived from main (#1375)
with a plain-text backup fixture; this branch's restore handler verifies
backups with integrity_check before touching the live database, so the text
fixture was (correctly) refused with 400 before the Close-failure branch
under test was reached. Use a real backup via BackupToSafe, matching the
other restore tests.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RtDNHSYWwPKArL8MsRdbj

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-08-15 20:50:47 +02:00

411 lines
15 KiB
Go

package api
import (
"bytes"
"context"
"encoding/binary"
"errors"
"fmt"
"image"
_ "image/gif" // register the GIF decoder for image.DecodeConfig
_ "image/jpeg" // register the JPEG decoder for image.DecodeConfig
_ "image/png" // register the PNG decoder for image.DecodeConfig
"io"
"log/slog"
"net/http"
"strconv"
"time"
"github.com/go-chi/chi/v5"
"github.com/google/uuid"
"github.com/owncord/server/auth"
"github.com/owncord/server/db"
"github.com/owncord/server/service"
)
// EmojiBroadcaster is the slice of the hub the emoji routes need: after every
// mutation the full set is pushed to every connected client so pickers,
// message rendering and reaction pills converge without a reconnect.
type EmojiBroadcaster interface {
BroadcastEmojiUpdate(list []*db.Emoji)
}
// emojiResponse is the JSON shape of one emoji in GET/POST /api/v1/emoji.
// Deliberately not db.Emoji: the storage id and sniffed mime type are
// server-side details, and `url` is what a client actually needs.
type emojiResponse struct {
ID int64 `json:"id"`
Shortcode string `json:"shortcode"`
URL string `json:"url"`
}
func toEmojiResponse(e *db.Emoji) emojiResponse {
return emojiResponse{ID: e.ID, Shortcode: e.Shortcode, URL: service.EmojiImageURL(e.ID)}
}
func toEmojiResponses(list []*db.Emoji) []emojiResponse {
out := make([]emojiResponse, 0, len(list))
for _, e := range list {
if e == nil {
continue
}
out = append(out, toEmojiResponse(e))
}
return out
}
// allowedEmojiMIME is the set of image types an emoji may be, matched against
// the type sniffed from the file's own bytes. SVG is absent on purpose: it is
// markup with script and external-fetch capability, which is exactly what
// isUnsafeInlineMIME forces to a download on the attachment route -- an emoji
// is by definition rendered inline, so the format simply cannot be allowed.
var allowedEmojiMIME = map[string]bool{
"image/png": true,
"image/jpeg": true,
"image/gif": true,
"image/webp": true,
}
// MountEmojiRoutes registers the custom-emoji endpoints.
//
// Every route requires authentication: reading the set is ungated beyond that
// (emoji are server-wide and every member renders them), while POST and DELETE
// are gated on MANAGE_SERVER inside EmojiService. The image route is
// authenticated rather than public so an emoji cannot be used as an
// unauthenticated tracking pixel hosted on someone else's server.
func MountEmojiRoutes(r chi.Router, database *db.DB, svc *service.Services, store FileStore, limiter *auth.RateLimiter, broadcaster EmojiBroadcaster) {
r.Route("/api/v1/emoji", func(r chi.Router) {
r.Use(AuthMiddleware(database))
r.Get("/", handleListEmoji(svc))
r.Get("/{id}/image", handleServeEmojiImage(svc, store))
r.With(MaxBodySize(emojiMaxBodySize)).
Post("/", handleCreateEmoji(svc, store, limiter, broadcaster))
r.Delete("/{id}", handleDeleteEmoji(svc, store, broadcaster))
})
}
func handleListEmoji(svc *service.Services) http.HandlerFunc {
return func(w http.ResponseWriter, r *http.Request) {
list, err := svc.Emoji.List(r.Context())
if err != nil {
writeServiceError(r.Context(), w, err)
return
}
writeJSON(w, http.StatusOK, toEmojiResponses(list))
}
}
// handleCreateEmoji processes POST /api/v1/emoji (multipart: `file` + `shortcode`).
//
// Order matters here. The permission gate runs BEFORE the multipart parse, so a
// member without MANAGE_SERVER cannot make the server spool a body to disk; the
// shortcode is validated next, so a malformed name costs nothing either; only
// then are the bytes read, sniffed, measured and stored.
func handleCreateEmoji(svc *service.Services, store FileStore, limiter *auth.RateLimiter, broadcaster EmojiBroadcaster) http.HandlerFunc {
return func(w http.ResponseWriter, r *http.Request) {
user, ok := r.Context().Value(UserKey).(*db.User)
if !ok || user == nil {
writeJSON(w, http.StatusUnauthorized, errorResponse{
Error: "UNAUTHORIZED", Message: "not authenticated",
})
return
}
if err := svc.Emoji.RequireManage(r.Context(), user.ID); err != nil {
writeServiceError(r.Context(), w, err)
return
}
if limiter != nil && !limiter.Allow(auth.Key("emoji_upload", user.ID), emojiUploadRateLimitPerMinute, time.Minute) {
writeJSON(w, http.StatusTooManyRequests, errorResponse{
Error: "RATE_LIMITED", Message: "emoji upload rate limit exceeded, try again later",
})
return
}
// Bound the body before the multipart parser touches it. The route also
// carries MaxBodySize, but a handler that parses a form has to state
// its own limit — the parser is what turns an unbounded body into heap.
r.Body = http.MaxBytesReader(w, r.Body, emojiMaxBodySize)
if err := r.ParseMultipartForm(emojiMultipartMemoryLimit); err != nil {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST", Message: "invalid multipart form",
})
return
}
shortcode, err := service.ValidateShortcode(r.FormValue("shortcode"))
if err != nil {
writeServiceError(r.Context(), w, err)
return
}
file, _, err := r.FormFile("file")
if err != nil {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST", Message: "missing file field",
})
return
}
defer file.Close() //nolint:errcheck
// Read at most one byte past the cap so "exactly at the limit" passes
// and "one byte over" is caught, without buffering an unbounded body.
raw, err := io.ReadAll(io.LimitReader(file, maxEmojiFileBytes+1))
if err != nil {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST", Message: "failed to read uploaded file",
})
return
}
if int64(len(raw)) > maxEmojiFileBytes {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST",
Message: fmt.Sprintf("emoji must be at most %d KB", maxEmojiFileBytes>>10),
})
return
}
mimeType := http.DetectContentType(raw)
if !allowedEmojiMIME[mimeType] {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST",
Message: "emoji must be a PNG, JPEG, GIF or WebP image",
})
return
}
width, height, err := imageDimensions(raw, mimeType)
if err != nil {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST", Message: "could not read image dimensions",
})
return
}
// Re-check the sniffed dimensions rather than trusting anything the
// client said about the image: the cap is what keeps an "emoji" from
// being a full-size picture inlined into every message that names it.
if width <= 0 || height <= 0 || width > maxEmojiDimension || height > maxEmojiDimension {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST",
Message: fmt.Sprintf("emoji must be at most %dx%d pixels (got %dx%d)", maxEmojiDimension, maxEmojiDimension, width, height),
})
return
}
storedAs := uuid.New().String()
if _, saveErr := store.Save(storedAs, bytes.NewReader(raw)); saveErr != nil {
// writeStorageSaveError also stops echoing raw storage errors
// (which embed absolute paths) into the response body.
writeStorageSaveError(w, saveErr, "emoji upload")
return
}
created, err := svc.Emoji.Create(r.Context(), user.ID, shortcode, storedAs, mimeType)
if err != nil {
// The row never landed, so the file is an orphan — unlink it.
if delErr := store.Delete(storedAs); delErr != nil {
slog.Error("failed to clean up orphaned emoji file", "stored_as", storedAs, "error", delErr)
}
writeServiceError(r.Context(), w, err)
return
}
broadcastEmojiSet(r.Context(), svc, broadcaster)
writeJSON(w, http.StatusCreated, toEmojiResponse(created))
}
}
func handleDeleteEmoji(svc *service.Services, store FileStore, broadcaster EmojiBroadcaster) http.HandlerFunc {
return func(w http.ResponseWriter, r *http.Request) {
user, ok := r.Context().Value(UserKey).(*db.User)
if !ok || user == nil {
writeJSON(w, http.StatusUnauthorized, errorResponse{
Error: "UNAUTHORIZED", Message: "not authenticated",
})
return
}
id, err := strconv.ParseInt(chi.URLParam(r, "id"), 10, 64)
if err != nil || id <= 0 {
writeJSON(w, http.StatusBadRequest, errorResponse{
Error: "BAD_REQUEST", Message: "invalid emoji id",
})
return
}
removed, err := svc.Emoji.Delete(r.Context(), user.ID, id)
if err != nil {
writeServiceError(r.Context(), w, err)
return
}
// The row is already gone, so a failed unlink leaves an orphaned blob,
// not a broken emoji — log it rather than failing a successful delete.
if delErr := store.Delete(removed.StoredAs); delErr != nil {
slog.Warn("failed to remove emoji file", "stored_as", removed.StoredAs, "error", delErr)
}
broadcastEmojiSet(r.Context(), svc, broadcaster)
w.WriteHeader(http.StatusNoContent)
}
}
// broadcastEmojiSet re-reads the set and pushes it to every client. A failure
// here is logged and swallowed: the mutation itself already succeeded, and the
// caller's own response carries the change.
//
// Called after the mutation has already committed, so the caller's request
// context may be canceled by the time this runs (client aborted, deadline
// fired) -- context.WithoutCancel detaches the re-read from that, matching
// the pattern service/emoji.go already uses for its post-commit audit write.
func broadcastEmojiSet(ctx context.Context, svc *service.Services, broadcaster EmojiBroadcaster) {
if broadcaster == nil {
return
}
list, err := svc.Emoji.List(context.WithoutCancel(ctx))
if err != nil {
slog.Error("failed to load emoji for broadcast", "error", err)
return
}
broadcaster.BroadcastEmojiUpdate(list)
}
// handleServeEmojiImage serves the stored bytes of one emoji.
//
// Unlike /api/v1/files/{id} there is no per-channel ACL to apply: an emoji is
// server-wide by construction, so authentication is the whole check. The
// response is immutable for the id's lifetime (an emoji's bytes never change
// — a replacement is a new row), which is what lets it be cached hard.
func handleServeEmojiImage(svc *service.Services, store FileStore) http.HandlerFunc {
return func(w http.ResponseWriter, r *http.Request) {
id, err := strconv.ParseInt(chi.URLParam(r, "id"), 10, 64)
if err != nil || id <= 0 {
http.NotFound(w, r)
return
}
e, err := svc.Emoji.Get(r.Context(), id)
if err != nil {
if errors.Is(err, service.ErrNotFound) {
http.NotFound(w, r)
return
}
writeServiceError(r.Context(), w, err)
return
}
f, err := store.Open(e.StoredAs)
if err != nil {
http.NotFound(w, r)
return
}
defer f.Close() //nolint:errcheck
w.Header().Set("Content-Type", e.MimeType)
w.Header().Set("X-Content-Type-Options", "nosniff")
w.Header().Set("Content-Disposition", "inline")
// Private (it needed a session to fetch) but immutable, so a client may
// keep it for a day rather than re-fetching it for every message.
w.Header().Set("Cache-Control", "private, max-age=86400, immutable")
var modTime time.Time
if info, statErr := f.Stat(); statErr == nil {
modTime = info.ModTime()
}
http.ServeContent(w, r, e.Shortcode, modTime, f)
}
}
// ─── Dimension extraction ────────────────────────────────────────────────────
// imageDimensions returns the pixel size of an image already known to be one of
// the allowed types. PNG/JPEG/GIF go through image.DecodeConfig; WebP has no
// decoder in the standard library and none is vendored, so its header is read
// directly — which is all that is wanted here anyway, since decoding a whole
// frame just to learn its size is work the cap exists to avoid.
//
// Shared by the emoji and avatar upload routes: both refuse an image too big
// for the surface it renders on, and both have to answer the same question
// about the same four formats.
func imageDimensions(raw []byte, mimeType string) (width, height int, err error) {
if mimeType == "image/webp" {
width, height, err = webpDimensions(raw)
} else {
var cfg image.Config
cfg, _, err = image.DecodeConfig(bytes.NewReader(raw))
if err != nil {
return 0, 0, fmt.Errorf("decoding image config: %w", err)
}
width, height = cfg.Width, cfg.Height
}
if err != nil {
return 0, 0, err
}
// Both callers treat a non-error return as trustworthy enough to compare
// straight against their pixel cap. A width or height of zero is not a
// "small" image, it is a decoder -- Go's own GIF DecodeConfig happily
// reports height=0 for a malformed logical screen descriptor -- accepting
// a degenerate header as valid. Rejecting it here means the invariant
// holds even for a caller that forgets to re-check, instead of relying on
// every call site getting its own bounds check right.
if width <= 0 || height <= 0 {
return 0, 0, fmt.Errorf("decoding image config: non-positive dimensions %dx%d", width, height)
}
return width, height, nil
}
// errBadWebP is returned for any WebP whose header does not parse; the caller
// turns it into the same 400 a corrupt PNG gets.
var errBadWebP = errors.New("malformed WebP header")
// webpDimensions reads the canvas size out of a RIFF/WEBP container. All three
// chunk flavours are handled: VP8 (lossy), VP8L (lossless) and VP8X (extended,
// which is what an animated or alpha WebP uses).
func webpDimensions(raw []byte) (width, height int, err error) {
// 12-byte RIFF header + at least a 4-byte chunk fourcc.
if len(raw) < 16 || string(raw[0:4]) != "RIFF" || string(raw[8:12]) != "WEBP" {
return 0, 0, errBadWebP
}
switch string(raw[12:16]) {
case "VP8 ":
// Chunk payload starts at 20; the keyframe start code sits 3 bytes in,
// followed by two 14-bit dimensions (the top 2 bits are a scale field).
if len(raw) < 30 {
return 0, 0, errBadWebP
}
if raw[23] != 0x9d || raw[24] != 0x01 || raw[25] != 0x2a {
return 0, 0, errBadWebP
}
w := int(binary.LittleEndian.Uint16(raw[26:28]) & 0x3FFF)
h := int(binary.LittleEndian.Uint16(raw[28:30]) & 0x3FFF)
// Unlike VP8L/VP8X (which store size-1, so they can never encode
// zero), the VP8 keyframe stores the size directly: an all-zero
// dimension field is a validly-shaped but degenerate header, not a
// real 0x0 canvas. Reject it here rather than reporting "success"
// with a size no image actually has.
if w == 0 || h == 0 {
return 0, 0, errBadWebP
}
return w, h, nil
case "VP8L":
// Payload starts at 20 with a 0x2F signature byte, then a packed
// 14+14-bit (width-1, height-1) pair.
if len(raw) < 25 || raw[20] != 0x2F {
return 0, 0, errBadWebP
}
bits := binary.LittleEndian.Uint32(raw[21:25])
w := int(bits&0x3FFF) + 1
h := int((bits>>14)&0x3FFF) + 1
return w, h, nil
case "VP8X":
// Payload starts at 20: flags byte, 3 reserved bytes, then canvas
// width-1 and height-1 as 24-bit little-endian values.
if len(raw) < 30 {
return 0, 0, errBadWebP
}
w := int(uint32(raw[24]) | uint32(raw[25])<<8 | uint32(raw[26])<<16)
h := int(uint32(raw[27]) | uint32(raw[28])<<8 | uint32(raw[29])<<16)
return w + 1, h + 1, nil
default:
return 0, 0, errBadWebP
}
}