Bring the feature-rich main branch into the frontend Svelte rewrite
(PR #478, base bcn/frontend-svelte-rewrite). main moved well ahead of the
PR's branch point (b8a0018): it added the Places photo-map and People
(faces) backends, photos enhancements, the ReBAC→role-grants migration,
load tests, and more.
Conflicts resolved (4 files):
- Dockerfile: combine the explicit --bin allowlist (defence-in-depth from
main) with the SPA copy from the frontend build stage (PR).
- .github/workflows/ci.yml: keep the PR's Svelte frontend job
(svelte-check + eslint + stylelint + prettier + vitest); the legacy
static/-targeted tsc/locale/icon advisory steps don't fit the new
working-directory: frontend job and svelte-check supersedes them.
- justfile: keep both the new fe-* / dev recipes (PR) and the load-* k6
recipes (main).
- static/locales: keep the PR's symlink (-> ../frontend/static/locales);
main's new photos/people locale keys are folded into the Svelte locale
files alongside the ported views.
Backend (people/places/faces handlers, routes, DI, migrations) merged
cleanly. `cargo check --bins` passes. The new Places/People UI is not yet
in the Svelte app; that is ported in follow-up commits.
Implements the last Phase 2 piece: a working face detector/embedder behind
the new `faces-onnx` cargo feature (mirrors how `plugins` gates wasmtime).
Inert by default — the default build is unchanged and ships the no-op
analyzer.
Pipeline (InsightFace/immich pattern): SCRFD detection with 5-point
landmarks → least-squares similarity alignment to the canonical 112×112
template → ArcFace embedding → L2-normalized 512-d vector.
- face_geometry.rs (always compiled, unit-tested): SCRFD anchor/distance
decode, NMS, the closed-form (complex-number) similarity transform,
bilinear affine warp, NCHW normalization, L2-norm, Laplacian sharpness.
11 unit tests cover the error-prone math with no model needed.
- onnx_face_analyzer.rs (feature `faces-onnx`): wires the geometry to ONNX
Runtime via `ort` (load-dynamic, so libonnxruntime is dlopen'd at runtime
and the crate builds without it). Inference runs on spawn_blocking; each
session is serialized behind a Mutex. Loads via `ort::init_from` (fallible)
not ORT's lazy loader, which would panic under `panic = "abort"`.
- config: FacesConfig + OXICLOUD_FACES_{ORT_DYLIB,DETECTOR_MODEL,
EMBEDDER_MODEL,DET_SIZE,DET_THRESHOLD,NMS_THRESHOLD,INTRA_THREADS}.
- di: build_face_analyzer() loads the real analyzer when the feature is
compiled in and runtime+models are configured; any missing piece or load
failure degrades to the no-op analyzer (logged) so startup never fails.
- ort/ndarray added as optional deps; example.env documents the setup.
Models and the ONNX Runtime dylib are operator-provided at runtime and are
never committed. Cannot be exercised in CI (no models/dylib); the geometry
is unit-tested and the ONNX seam is isolated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
Phase 2 increment 6 — the People HTTP API (mounted only when
OXICLOUD_ENABLE_FACES is on; every handler is caller-scoped):
- GET /api/people list identity clusters
- GET /api/people/{id}/photos a person's photo file ids
- PATCH /api/people/{id} name / rename a person
- POST /api/people/{id}/hide hide / unhide
- POST /api/people/merge merge two clusters
- POST /api/people/recluster re-run clustering
- DELETE /api/people/data erase all face data (opt-out)
- GET /api/people/faces/{file_id} face boxes for lightbox tagging
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
Phase 2 increment 5:
- FaceIndexingService: a FileLifecycleHook that, on image upload, detects +
embeds faces in a background task and stores them. Dedup-aware (clones an
identical blob's faces instead of re-running inference), reindexes on
overwrite, and relies on the DB cascade for deletes. Completely inert when
no model is ready.
- DI: registers the hook in the FileLifecycleService chain and exposes
PeopleService in AppState — both gated on OXICLOUD_ENABLE_FACES, both
using the default no-op analyzer until the operator wires a real ONNX
model.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
Phase 2 increment 4:
- recluster(user): union-find connected-components over the user's face
embeddings (cosine threshold); groups of >= min_faces become persons, and
an existing person's name is preserved across reclusters. O(n^2) — fine
for moderate libraries; an ANN index is the documented scale-up.
- caller_id-scoped use cases for the HTTP layer: list_people (non-empty
clusters, cover thumbnail, most-photographed first), person_photos,
faces_for_file, rename, hide, merge, and delete_all (right to erasure).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
Phase 2 increment 3: FacePgRepository implements FaceRepository.
Embeddings stored/read as BYTEA (512 little-endian f32), bbox as REAL[].
Every query is user-scoped. Covers face CRUD, person CRUD (create / rename
/ cover / hide), files-for-person, and a transactional delete-all-for-user
(right to erasure).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
Phase 2 increment 2:
- domain: Face, Person, BoundingBox, DetectedFace (512-d embeddings).
- ports: FaceAnalyzerPort (detect + embed from raw bytes; decodes internally
so the application layer stays image/ML-crate agnostic) and FaceRepository
(user-scoped face/person persistence).
- DTOs: PersonDto, FaceBoxDto.
- NoopFaceAnalyzer — reports is_ready()==false and returns no faces, so the
whole People pipeline compiles and runs inert until the operator wires a
real ONNX-backed analyzer.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
First slice of Phase 2 (People / faces):
- migration: `faces` schema with `faces.persons` and `faces.faces`.
Embeddings are stored as BYTEA (512 x f32) — no pgvector extension
dependency; similarity is computed in-app (pgvector/VectorChord is the
documented scale-up). Cascade deletes (by user and by source file)
satisfy the right to erasure.
- OXICLOUD_ENABLE_FACES feature flag, OFF by default (biometric data,
opt-in per deployment).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
Three list endpoints resolved each resource with one query per id:
- GET /api/grants/incoming and /api/grants/outgoing used
join_all(ids.map(get_file)) + join_all(ids.map(get_folder)), so a single
page (limit ≤ 200) could demand ~200 concurrent connections from the
20-connection primary pool, causing acquire-timeouts and head-of-line
blocking under load.
- The NextCloud favorites REPORT (oc:filter-files) fetched get_file/
get_folder once per favorite — up to N serial round-trips per sync.
Add by-ids batch reads that mirror the existing get_file/get_folder column
mapping and NOT is_trashed filter:
- FileBlobReadRepository::get_files_by_ids / FolderDbRepository::get_folders_by_ids
(one SELECT ... WHERE id = ANY($1)), exposed as FileRetrievalService::
get_files_by_ids / FolderService::get_folders_by_ids returning DTOs.
- Both grant handlers and the favorites REPORT now issue two batch queries
total and look results up by id, preserving original order. Missing ids
(stale grants whose resource was deleted, or trashed/removed favorites)
drop out exactly as before. No auth-semantics change: these paths already
resolved ids vetted by the authorization engine / favorites table.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAzLEQDaLak3dnrEN3YT35
Adds a Places tab to the Photos section: a self-hosted MapLibre GL map of
the user's geotagged photos.
- places.js: lazy-loads vendored MapLibre + pmtiles.js on first open; draws
the server-aggregated clusters (GET /api/photos/geo) as HTML thumbnail
markers (no glyphs/sprites, no client-side clustering); pan/zoom refetches
for the new viewport; click a cluster to zoom in, or a single photo to
open it in the lightbox.
- Optional vector basemap read directly from /basemaps/basemap.pmtiles over
HTTP Range (label-light Protomaps style, light/dark aware); falls back to
a themed background when no basemap is present. ODbL attribution shown.
- "Moments | Places" sub-nav wired into the Photos section.
- enable_places now defaults on, since the route + UI are ready.
No new backend serving code: tower-http ServeDir already serves static/
with Range, so the .pmtiles basemap is just a static file.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
JWT access tokens freeze role/identity at login (access 1h, refresh 7d)
and validated tokens are cached for 30s. The Bearer and cookie auth paths
trusted claims.role and never re-checked the account, so demoting an admin,
or disabling/deleting an account, did not revoke access until the token
expired. The app-password path already re-read role/active from the DB;
only the JWT/cookie path had the gap.
Re-validate the caller against the live user record on the token path via
the already-cached get_user_flags (role / is_external / active), bounded by
USER_FLAGS_CACHE_TTL and invalidated eagerly on set_user_active /
change_user_role / delete_user_admin:
- middleware/user.rs: new resolve_live_role helper (+ pure decide_live_role
core) — returns the *current* role, rejects deleted (NotFound) and
deactivated accounts, and fails open on transient lookup errors (mirrors
require_internal_user). Login/refresh remain the canonical active gate.
- middleware/auth.rs: auth_middleware (Bearer + cookie) now populates
CurrentUser with the live role and rejects revoked accounts (Bearer ->
401 AccountInactive; cookie -> fall through to 401/login redirect).
require_admin emits an audit line on denial.
- middleware/admin.rs: require_admin / require_authenticated re-check the
live record, return the live role, and audit admin denials.
Downstream admin gates (dedup_handler, subject_group_handler, OCS) inherit
the live role automatically via CurrentUser / require_authenticated.
Tests: decide_live_role policy (active / demoted / deactivated / deleted /
transient fail-open) and AccountInactive -> 401.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAzLEQDaLak3dnrEN3YT35
Phase 1 server side, gated on OXICLOUD_ENABLE_PLACES (off by default):
- migration: partial index on storage.file_metadata(longitude, latitude).
- FileBlobReadRepository::list_geo_clusters — plain-SQL grid aggregation
(no PostGIS) scoped to the caller's own non-trashed photos, returning a
centroid, count and a representative file id per non-empty cell.
- PlacesService (caller_id-scoped; user-scoped data needs no authz check,
mirroring RecentService) with a zoom→cell-size mapping.
- GET /api/photos/geo?bbox=w,s,e,n&zoom=N returning GeoCluster[]. The route
is mounted only when the Places service is present, and is registered in
the OpenAPI path list.
The map frontend (PMTiles serving + MapLibre module) is deferred pending
the basemap-sourcing and vendoring decision.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
list_media_files now LEFT JOINs storage.file_metadata and returns each
photo's pixel dimensions next to the sort date. The endpoint wraps FileDto
in a flattened PhotoDto carrying width/height, so the gallery can lay tiles
out at their true aspect ratio (justified layout) without a second per-file
metadata round-trip and without layout shift. FileItem gains optional
width/height. No change to FileDto or its other construction sites.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JW6ghFMDtnRYuYNzZhb47M
remove_manifest_reference unlinked a chunk's backing file right after the
row-delete committed — the same TOCTOU the GC grace window was added to
close: a concurrent upload of identical content can re-reference (pin) the
chunk in the gap between commit and unlink, after which the deferred
unlink strands a referenced chunk with no bytes.
Route physical chunk reclamation through the single grace-protected path:
on last reference, delete the manifest and decrement its chunks (stamping
orphaned_at on the ones that reach 0), but leave the chunk rows and files
for garbage_collect() to reclaim once orphaned past the grace window. The
manifest deletion and its blob-keyed thumbnail hook stay eager.
remove_legacy_reference and cleanup_if_orphaned's legacy path are left as
eager deletes on purpose: a legacy whole-file hash can never be re-created
by an ingest (uploads are always CDC now), so there is no writer to race —
the existing "row gone ⇒ no resurrection" reasoning holds for them.
Adds an integration test asserting a CDC manifest dereference leaves
chunks orphaned-but-present, then reclaimed by a post-grace GC.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0172rsVwzTwD216R9HXT2aU4
Deduplication GC (garbage_collect, Phase 2):
- Add an orphan grace period before a ref_count=0 blob's backing file is
physically deleted, mirroring git's gc.pruneExpire. New
storage.blobs.orphaned_at records when a blob last reached ref_count 0;
the delete trigger and every decrement / 0-ref insert path stamp it,
every re-reference clears it.
- Cross-check that no manifest lists the chunk and no file points at the
blob before deleting it (mirrors Phase 1's file check), so a stale
ref_count can only delay collection, never delete live content.
- Unlink the backing files with bounded parallel fan-out.
Together these close a TOCTOU where a concurrent upload of identical
content could re-reference a chunk in the window between the GC row
delete committing and the backing file being unlinked. Individual file
deletes still reclaim eagerly; only bulk empty-trash and the periodic
sweep observe the grace window.
Trash: match ErrorKind::NotFound instead of substring-matching the error
message when treating an already-deleted item as success.
Content-index worker: supervise the drain loop and restart it with
backoff after a panic, instead of letting a panic silently freeze the
search index while the dirty queue grows unbounded.
Adds migration 20260802000000_blob_gc_grace.sql and an integration test
covering the grace window and reference cross-checks.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0172rsVwzTwD216R9HXT2aU4
Standard CalDAV/CardDAV clients (Thunderbird, DAVx5, Apple
Calendar/Contacts) failed to connect, mounted collections read-only, or
could not discover address books, even though curl worked. Three
protocol-compliance gaps caused this:
1. Missing Basic-auth challenge on /caldav and /carddav.
The 401 returned for these surfaces carried no `WWW-Authenticate`
header (only /webdav did). Spec-compliant clients never send
credentials preemptively the way `curl -u` does — they wait for the
challenge — so Thunderbird never authenticated and failed with
"discovery failed" / 401. Extend the challenge to all DAV surfaces via
shared `is_dav_path` / `dav_basic_auth_challenge` helpers.
2. Calendars always advertised read-only.
The `current-user-privilege-set` write gate compared `owner_id`
against the literal string "current_user_id", which never matched a
real UUID, so `<D:write/>` was never emitted and clients mounted every
calendar read-only. Thread the caller's id through the CalDAV adapter
and grant write when the caller owns the calendar.
3. CardDAV discovery was incomplete.
There was no `/.well-known/carddav` route and the root PROPFIND
exposed neither `current-user-principal` nor `addressbook-home-set`,
so clients could not locate address books. Add the well-known redirect
and root/principal discovery responses mirroring the CalDAV adapter.
Adds unit tests for the auth challenge predicate, the calendar
owner/non-owner privilege split, and the CardDAV root/principal discovery
responses.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016cVV9nRQjP6G6a8zbNUWMw
D-Prep (migration 20260801000002_drop_access_grants) drops
storage.access_grants and replaces it with storage.role_grants — one
row per role assignment instead of N rows per permission bundle. The
load seeder still spoke the old per-permission shape and failed every
nightly with `relation "storage.access_grants" does not exist`.
Both seeder call sites already grant the read bundle (single
permission), which maps cleanly to the `viewer` role; switching them
to the new schema is a one-row INSERT with the role name. The
conflict key drops `permission` since uniqueness is now per
(subject, resource).
prepare migration of permission to roles
this simplify drastically database (permission are now simply defined in code)
and will permit reuse of the same ReBAC engine to define owners of drives
mapping:
```
Role::Viewer => &[Permission::Read],
Role::Commenter => &[Permission::Read, Permission::Comment],
Role::Contributor => &[Permission::Read, Permission::Create],
Role::Editor => &[
Permission::Read,
Permission::Comment,
Permission::Create,
Permission::Update,
],
Role::Owner => &[
Permission::Read,
Permission::Comment,
Permission::Create,
Permission::Update,
Permission::Share,
Permission::Delete,
Permission::Manage,
],
```
handle_head was declaring `Content-Length: file.size` while writing
`Body::empty()` — on a keep-alive connection the client waits forever
for N bytes that never come. Hyper now derives Content-Length: 0 from
the actual body, which is honest about what's on the wire.
RFC 7231 §4.3.2 suggests HEAD return the same headers as GET, but
lying about Content-Length is worse than omitting it: NC and Sabre
clients use PROPFIND for size anyway, and curl -I (and any client
applying HEAD semantics) gets the same ETag/MIME/Last-Modified it
needs. Caught by the F6b test which uses `curl -X HEAD` to read the
current ETag before a conditional PUT.
Also adds `nc_status_propfind_depth0` to lib/dav_helpers.sh so the
F11/F11b assertions ("did the intermediate parent get auto-created?")
can compile.
When the client sends `MOVE /trashbin/{id}` with a `Destination` header,
handle_restore now resolves the destination path and returns 412
Precondition Failed if a live file or folder already sits there —
matching Sabre/DAV and the NC desktop client's expectation. There is
no `Overwrite: T` workflow for trash restore in either reference
implementation (silently replacing a live file with an undeleted one
is a footgun), so the refusal is unconditional.
The destination header is extracted at the dispatch site as an owned
String so the future stays Send-compatible (`&Request<Body>` is not
Sync because the body trait object is Send-only).
`extract_nc_subpath_from_dest` is promoted to `pub` so trashbin_handler
can share the same URL parser as handle_move.
Threads `new_name: Option<&str>` through FileWritePort::copy_file and
FileManagementUseCase::copy_file_with_perms so a same-folder
COPY /a.txt → /b.txt picks up the destination name via a single
COALESCE($3::text, name) in the CTE. Without it the new row inherits
the source's filename and collides on the (folder_id, name, user_id)
unique index — the "Already Exists" 500 M8 was hitting.
handle_copy in the native WebDAV surface now passes
`(file.name != dest_name).then(|| dest_name.into())`, keeping the
"same name in a different folder" case at None so existing semantics
are preserved.
Extends the N2/PUT lock guard introduced earlier to the rest of the
native mutator surface. Same helper, same If: capture before body
consumption, same 423-on-reject shape:
- handle_delete : check source path
- handle_proppatch : check source path
- handle_move : check source AND destination paths
- handle_copy : check destination path only (source isn't
modified by a copy)
The class-2 DAV advertisement in OPTIONS is now honest across the
full surface, not just PUT.
New tests N2c-N2f run while n-locked.txt is still LOCKed (before the
existing N3 UNLOCK). Each asserts 423 without the token and verifies
the operation didn't half-apply: file present after DELETE-423,
source untouched + no destination after MOVE-423, locked destination's
content unchanged after COPY-423.
Positive (with-token) coverage is implicit via the M-series happy-
path tests that exercise each method on unlocked resources — a
regression that hard-rejected every call would fail there too.
Closes F11.
The NC MKCOL handler previously had `mkdir -p` semantics:
sending MKCOL on /a/b/c/ where neither a nor b exists silently
created both intermediates and returned 201. Sabre/DAV and the
actual NC server both 409 on that — our auto-create deviated.
NC desktop walks ancestors one MKCOL at a time during sync so
nothing real depended on the old behaviour.
Drop the segment-walking creation loop. New flow:
target exists → 405
parent path missing → 409
parent ok, target new → 201
The race-recovery branch for the loop's per-segment create is
also gone — single parent lookup, single create, no window.
Test F11 flipped from pinned-201 to strict 409 and asserts the
intermediate parent was not silently created. F11b and F11c added
as regression guards for the success path and the 'target already
exists' case.
Closes G9.
DELETE on a folder now flips is_trashed on every descendant folder
and file under it in a single CTE pipeline (lpath <@ root.lpath
covers the whole subtree via the GiST index). Previously only the
root row was flipped — descendants stayed live, directly addressable
via their full path, and confused desktop-sync tree walks that
expected the parent-collection 404 to imply the children were gone
too.
restore_from_trash mirrors the cascade: descendants where the
original_*_parent_id column is NULL are the ones we cascade-trashed,
so they get cascade-restored too. Descendants that were independently
trashed before the parent went to trash have original_*_parent_id
set, so they stay in trash and remain visible as top-level entries
in storage.trash_items.
No schema migration needed — both original_parent_id (folders) and
original_folder_id (files) were already nullable and already encoded
'where this came from when it was independently trashed'; using NULL
as the cascade-marker reuses that existing distinction cleanly.
Test G9 flipped from KNOWN BUG to assert every descendant 404s after
the parent DELETE; new G9b proves the inverse cascade by restoring
the trashed root and verifying every descendant comes back at its
original path.
Closes M5 / M7 / M8a / M8b.
The optimized PathResolver and the read-side find_*_by_path queries
disagree on what counts as 'a path that hits a row'. After the drive-
refactor migration rewrote the path column to drop the
"My Folder - <user>/" prefix, files PUT through the WebDAV surface
stayed reachable by GET (legacy lookup) but vanished from the
optimized resolver (strict path-match). MOVE/DELETE/COPY 404'd on
every root-level file as a result.
Introduces resolve_or_legacy: optimized resolver first, then the
GET-style legacy lookups as a strict superset. Ownership is enforced
in both branches. handle_delete / handle_move / handle_copy each
collapsed from two near-identical resolver-only + legacy-only branches
into a single match using the helper — fewer lines, identical
semantics, root-level paths now resolve.
handle_copy also fixes M8b: copy_file_with_perms takes no destination
name, so a copy to a different filename in the same folder collided
with the source. After the copy, rename the new file when dest_name
differs from source name. Mirrors what handle_move already does.
Closes G4 / G5 / K5.
handle_move now resolves the destination once before the file/folder
dispatch and applies RFC 4918 §9.9.4:
- Overwrite: F on a collision → 412 Precondition Failed, source
untouched, destination untouched.
- Overwrite: T (or absent) on a collision → delete the existing
destination, then proceed → 204 No Content.
- No collision → 201 Created (unchanged).
The same destination lookup powers the 201-vs-204 status decision, so
adding the precondition guard adds zero extra DB hits on the happy
path.
handle_restore now catches the unique-index collision out of
restore_item and returns 412 instead of letting it bubble as 500.
Mirrors the G4 semantics for the trashbin surface (restore has no
Overwrite header so the refusal is unconditional; client resolves by
renaming the live file first).
Sabre/DAV's CorePlugin and our test pins agreed independently — NC
clients expect this exact behavior, and the new G5b/G5c positive-case
tests guard against a regression that hard-rejected every MOVE.
Closes N2. The native WebDAV PUT handler now consults the lock
store before accepting a write: if the target path is exclusively
locked, the request must carry the lock token in its If: header
or the server returns 423 Locked. Without a matching token, the
body is never consumed — a rejected PUT no longer wastes the
upload bandwidth or hits the CDC ingester.
Two helpers are introduced so the same enforcement plugs into
the other mutator methods (delete/move/copy/proppatch) when their
fixes land:
extract_if_header_tokens — angle-bracket-scoop view of If:
(sufficient for one-target writes;
full §10.4 tagged-list grammar would
only matter for multi-resource Ifs)
enforce_native_lock — Some(423) when locked + no/wrong
token, None otherwise
Test N2 flipped from pinned 204 to assert 423. Added N2b: same
PUT with the captured Lock-Token in If:(<...>) returns 204, so a
regression that hard-rejected every PUT would still fail loudly.
Closes F5/F6. The NC PUT handler now evaluates conditional
preconditions before body ingestion and returns 412 Precondition
Failed when they fail:
- If-None-Match: * on an existing target → 412 (create-if-absent)
- If-None-Match with matching ETag → 412 (weak compare)
- If-Match: * with no current representation → 412
- If-Match with no listed ETag strong-matching the current → 412
The lookup that drives the precondition reuses the same query the
handler already needed for the 201-vs-204 distinction, so this adds
no extra DB round-trip. Rejected requests skip body ingestion
entirely so a 412 doesn't waste megabytes of bandwidth + disk I/O.
Test F5/F6 flipped from 'pinned current 204' to assert 412, plus
mirror cases F5b/F6b/F6c/F6d covering the legitimate-success paths
so logical-operator regressions can't slip past silently.
- permit reuse of blob where file is trashed (by-hash and chunked)
- add hurl test on /api/files/by-hash
- add anti enumeration of blob (404 is always blob_not_owned_by_caller)
After every upload, maybe_update_storage_usage spawned a full
`SUM(size) OVER all the user's non-trashed files` to refresh
auth.users.storage_used_bytes — O(N) in the user's file count per upload,
i.e. O(N²) for a bulk upload. (The covering index makes it index-only but
still scans N rows.)
Replace it with an O(1) incremental `storage_used_bytes += size`, keyed by the
file's owner_id (dropping the brittle "My Folder - <user>" path-parsing hack).
Deletes/trash never decremented this value — they already rely on the periodic
reconciliation sweep — so the model is unchanged: the sweep remains the
correctness backstop for every mutation, and the counter is clamped at 0.
Both stay fire-and-forget on a background task, off the upload's latency path.
Benchmarked (per-call, vs the user's existing file count N):
N=1k: full-SUM 202us vs incremental 123us
N=10k: full-SUM 1185us vs incremental 113us (10x)
N=50k: full-SUM 5397us vs incremental 114us (47x — incremental is flat O(1))
Bulk upload of 10k files (insert + usage update each):
full-SUM (O(N²)) 10.37s -> incremental (O(N)) 4.89s (>2x, diverges with scale)
https://claude.ai/code/session_01DCszkkU11LYxMEUWr4setK
save_file_with_blob_impl did three sequential DB round-trips per upload:
resolve_user_id (SELECT folders.user_id), the INSERT, then lookup_folder_path
(SELECT folders.path) — the first and third re-reading the same folders row.
Replace them with a single statement: a `parent` CTE reads the folder once and
the INSERT derives user_id from it and returns the path via the CTE, in one
round-trip. An empty CTE (folder vanished between ingest and insert) inserts
zero rows and now surfaces as a clean NotFound instead of a generic owner
error. Deadlock retry, blob-ref compensation, and the 23505 (duplicate name)
mapping are preserved; owner resolution + insert are now atomic (no TOCTOU).
What it does and doesn't buy (benchmarked, honest):
- Server CPU: UNCHANGED. A server-side 50k loop is identical (13.6s vs 13.6s)
— the two extra folder reads are cached point lookups, negligible against
the INSERT + per-statement triggers + 13 indexes.
- Client-observed latency: 2 fewer client<->DB round-trips per upload. At the
measured ~189us/round-trip on localhost that's ~0.38ms/upload; on a
networked DB (RTT 0.5-1ms) ~1-2ms/upload.
- Connection pool: the metadata phase holds a pooled connection for 1
round-trip instead of 3, freeing it ~3x sooner under upload concurrency.
So this is a latency + connection-utilization win (and an atomicity/cleanup),
not a server-CPU win. register_file_deferred keeps its own path.
https://claude.ai/code/session_01DCszkkU11LYxMEUWr4setK
FileContentCache (moka, 512 MiB) was keyed by the file UUID, so content that
the CDC store already deduplicates to ONE blob on disk was cached once PER
FILE in RAM: N files sharing a blob held N copies, all counting against the
512 MiB cap. With effective dedup the cache filled with duplicates and
thrashed.
Key it by the blob hash instead (already on FileDto::content_hash):
- The in-RAM cache now benefits from dedup like the disk does — each distinct
blob is cached once and shared across every file/user that references it,
so a download by user A warms the cache for user B's identical content.
- Content is immutable by hash, so entries never go stale; the existing
invalidate(file_id) calls become harmless no-ops (a UUID never matches a
hash key) and can be removed in a later cleanup.
- ETag is now the immutable content hash (strong validator).
- Guarded: a hash-less stub DTO disables caching for that request rather than
colliding every such file on the empty key.
Response Content-Type still comes from the DTO, not the cache, so keying does
not affect the served MIME (verified).
Benchmark (real moka, exact 512 MiB/weight config, 400 files x 4 MiB = 1600 MiB
working set, 4000 uniform-random accesses):
dedup 1x : file_id 35.8% hit / 2568 reads vs hash 35.6% / 2577 (no dedup -> no change; control)
dedup 5x : file_id 35.8% hit / 2570 reads vs hash 98.0% / 80 (32x fewer disk reads, RAM 512->320 MiB)
dedup 20x: file_id 35.1% hit / 2596 reads vs hash 99.5% / 20 (130x fewer disk reads, RAM 512->80 MiB)
The win scales with the dedup ratio; with no dedup it is a no-op.
https://claude.ai/code/session_01DCszkkU11LYxMEUWr4setK
read_blob_stream / read_blob_range_stream reassembled a CDC file by fetching
its chunks with `buffered(1)` — strictly sequential, so the next chunk's
backend fetch (a file `open` locally; a full request round-trip on S3/Azure)
only started after the current chunk was fully drained.
A benchmark of the exact pipeline (stream::iter(chunks).map(get).buffered(K)
.try_flatten()) showed a blind `buffered(4)` is the WRONG fix: on a local
disk it is neutral on a warm page cache and ~37% SLOWER cold, because
concurrent opens turn one sequential read into several competing random-I/O
streams over content-addressed (scattered) chunk files. The win is entirely
on remote backends, where per-chunk request latency dominates and overlapping
fetches hide it (≈ linear in K).
So the read-ahead depth is now a backend hint, not a constant:
- BlobStorageBackend::read_prefetch() default 1 (sequential; safe for local).
- S3 / Azure override to 8 (overlap GETs to hide TTFB).
- cached / encrypted / retry / migration delegate to the backend that serves
the bytes.
- Both CDC read paths use `self.backend.read_prefetch().max(1)`.
Net: local backend unchanged (no regression); remote reassembly ~4-8x faster.
Ordered `buffered` (not buffer_unordered) keeps chunks in sequence.
Bench (per-chunk fetch-latency model): buffered(1)->(4)/(8) = x3.9 / x7.8
@1ms, x4.0 / x8.1 @5ms, x4.0 / x8.0 @20ms. Local warm: 230ms@1 vs 227ms@4
(noise); local cold: 425ms@1 vs 585ms@4 (why local stays at 1).
https://claude.ai/code/session_01DCszkkU11LYxMEUWr4setK
The global tower-http CompressionLayer compressed every response whose
Content-Type was not in the already-compressed exclusion list — including
large text-ish file downloads (.csv/.log/.sql/.json/source). That (a) burned
CPU re-encoding multi-GB bodies on the request path with no cached result,
and (b) made tower-http strip Content-Length and Accept-Ranges, breaking
byte-range seek and download resume.
Add a NotForDownloads predicate that skips compression for any response
carrying Content-Disposition (every download surface: REST file, share,
folder/zip, batch-zip, inline previews). API JSON and static assets never
set Content-Disposition, so they stay compressed.
Verified with the real tower-http layer + this exact predicate (64 MiB
text/plain download):
- Full download regains Content-Length + Accept-Ranges (were stripped);
/api/data stays brotli-compressed (fix is surgical).
- CPU: 2.7-3.6x less per download sequential; 6-7x less under 8-way concurrency.
- TTFB: 44ms->1ms (gzip), 110ms->1ms (brotli).
- Delivered content throughput: 2.4-2.8x higher.
Tradeoff: genuinely-compressible downloads now send more bytes on the wire;
reclaim via compress-at-rest if it ever matters.
https://claude.ai/code/session_01DCszkkU11LYxMEUWr4setK
Pre-existing in the delta-upload service: Rust 1.93's stricter large_enum_variant lint flags the FileDto-carrying Done variant (the sibling variant is tiny). The value is short-lived (one per commit), so silence the lint rather than box FileDto and complicate every call site. Keeps the CI clippy -D warnings gate green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Backend — Photos timeline now groups by real capture date instead of upload time. New MediaMetadataService (FileLifecycleHook) extracts EXIF DateTimeOriginal from images and container creation_time from videos (mov/mp4/mkv) via nom-exif, timezone-correct (OffsetTimeOriginal), persisting captured_at so the existing media_sort_date trigger takes over. Adds POST /admin/photos/metadata/reextract to backfill existing media. Falls back to upload date when no embedded date exists.
Frontend — premium grid cards: combined metadata line (relative date · size, owner avatar when shared), custom selection checkbox with a clear checked state, uniform full-width 4:3 thumbnail tiles independent of filename length, centered file-type icons, and a hit-test fix so checkbox/star/kebab clicks reach the controls (the decorative thumbnail no longer captures pointer events). Notification messages internationalised across all 16 locales. Broader polish: design tokens, a11y/focus-visible states, brand + PWA assets.
Chore — bump semver-compatible dependencies (cargo upgrade); add nom-exif 3.6.1.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Edouard Vanbelle