Commit Graph

1471 Commits

Author SHA1 Message Date
MHSanaei
dd14e9b3b0 feat(inbounds): attach existing clients to an inbound in one click
Adds an 'Attach Existing Clients' row action on multi-user inbounds (shown even when the inbound is empty). It opens a modal listing the whole client pool with search and group filter, all attachable clients pre-selected, and attaches the selection to that inbound via the existing bulkAttach endpoint. Clients already on the inbound are shown disabled and skipped. Translations added for all 13 locales.
2026-06-01 07:26:30 +02:00
MHSanaei
971843f669 feat(nodes): bulk panel self-update with live online indicator
Adds the ability to update node panels to the latest release from the Nodes
page: select online, enabled nodes (checkboxes) and trigger their official
self-updater, or use the per-row Update action. A node whose reported panel
version trails the latest GitHub release is flagged with an 'update available'
tag (compared via lib/panel-version, mirroring the Go isNewerVersion).

Backend: Remote.UpdatePanel calls the node's existing
POST /panel/api/server/updatePanel; NodeService.UpdatePanels fans out over the
selected ids, skipping disabled/offline nodes with a per-node reason; exposed
as POST /panel/api/nodes/updatePanel (documented in endpoints.ts + openapi.json).

The bulk request sends a JSON body, so it sets Content-Type: application/json
explicitly — axios defaults POST to form-urlencoded, which made ShouldBindJSON
fail with 'invalid character i'.

Also reuses the clients-page online cue on the Nodes page: a pulsing green dot
plus green label for an online node. The .online-dot style moved to the shared
styles/utils.css so both pages load it.

Translations for all new node keys added across every language file.
2026-06-01 07:03:06 +02:00
MHSanaei
fb311afa6f fix(sub): keep listen/bind IP out of subscription page URLs
The subscription page leaked an inbound's server-side Listen IP into the
client-facing URLs when a bind address was set:

- Per-config links: resolveInboundAddress returned the bind Listen IP
  (loopback/private/public alike) instead of the host the subscriber
  reached the panel on. It now returns the node address for node-managed
  inbounds, otherwise the subscriber host; the bind Listen is ignored
  (External Proxy remains the way to advertise a specific endpoint).

- Subscription Copy URL (SUB/JSON/CLASH): BuildURLs composed the base
  differently from the panel's Client Information page and never
  normalized the request host, so a loopback/bind request leaked the raw
  IP. The composition is extracted into the shared
  SettingService.BuildSubURIBase, used by both the panel and the sub page
  so they render identically, and fed the already-normalized subscriber
  host.
2026-06-01 05:47:18 +02:00
MHSanaei
eb78b8666f fix(inbound): re-derive auto tags on edit and keep node tags consistent
Auto-generated inbound tags (in-<port>-<l4>, n<id>- prefixed for node inbounds) now re-derive when port/listen/transport change on update instead of keeping the stale round-tripped value. The resolved tag is mirrored onto the API response, and NodeID is pinned to the stored row so a node inbound never loses its n<id>- prefix on edit. The edit form recomputes the tag live via a Go-parity helper so the JSON preview matches what gets saved.

Make node/central tag matching prefix-agnostic in all three places (traffic attribution, remote-id resolution, and the orphan sweep) so an n<id>- prefix present on only one side can no longer spawn duplicate inbounds or drop traffic on sync.

Force LF on shell scripts via .gitattributes (CRLF broke the Docker build shebang when the repo is checked out on Windows) and add a .dockerignore to keep node_modules/.git out of the build context.

Adds Go and frontend tests covering tag re-derivation, prefix-agnostic matching, and node-snapshot prefix mismatch.
2026-06-01 05:08:29 +02:00
MHSanaei
4a11375f36 fix(tgbot): send login notification asynchronously
UserLoginNotify ran SendMsgToTgbotAdmins synchronously on the login request goroutine. When Telegram was unreachable, the send retried up to 3x with a 30s timeout each, blocking the login handler for ~90s+ and effectively locking users out (issue #4585).

Dispatch the send in a goroutine after the cheap bot-running/login-notify-enabled guards so login always returns promptly; the existing per-send 30s context timeout and bounded retries keep the background goroutine from leaking.
2026-06-01 02:38:06 +02:00
MHSanaei
3f5e37b038 fix(postgres): record client traffic when inbound_id is stale
When an inbound is deleted and recreated it gets a new id, but the shared-by-email client_traffics row keeps the old (now deleted) inbound_id because AddClientStat's OnConflict-DoNothing never refreshes it. The traffic updater matched rows with inbound_id IN (local inbounds), so those orphaned rows were dropped: client traffic and online status stopped updating and auto-renew skipped them, while inbound-level traffic (matched by tag) kept working and the client count still showed (matched by email).

Match by email and exclude only rows owned by a node inbound (inbound_id NOT IN (node inbounds)) in addClientTraffic and autoRenewClients. The local Xray only reports local-client emails, so a stale local pointer no longer hides the row, while genuine node-owned rows stay protected. Verified against a real affected dump: visible rows went from 4/668 to 668/668.
2026-06-01 01:39:21 +02:00
MHSanaei
61ba5754ca fix(postgres): commit client traffic backfill in migration
MigrationRequirements backfills missing client_traffics rows from each inbound's settings.clients, but the later MultiDomain->ExternalProxy detection query used SQLite-only json_extract and executed via .Scan. On PostgreSQL it errored, rolling back the whole transaction including the backfill, so clients had no traffic rows: client traffic was never recorded, clients showed offline, and the inbound list showed 0 clients until each inbound was edited and saved.

Make the detection query dialect-aware (NULLIF(stream_settings,'')::jsonb #>> / #>) so the function runs to completion and commits on both dialects.
2026-06-01 00:43:42 +02:00
MHSanaei
575355e4f1 fix(inbounds): only reset id sequence when all inbounds are deleted
Commit 80110f9 realigned sqlite_sequence to MAX(id) after every delete,
which recycled freed ids and let a newly added inbound take an old
inbound id. Now the sequence row is cleared only when the table is empty,
so the counter keeps climbing while any inbound remains and existing ids
are never reused. Still guarded behind !IsPostgres().
@
2026-05-31 23:04:15 +02:00
MHSanaei
76dbbfc1f8 feat(inbounds): clearer client validation errors on save
When an inbound save fails Zod validation, the toast previously showed a
raw path like `settings.clients.494.tgId: Invalid input`, which gave no
hint which of hundreds of clients was at fault. Resolve the client array
index back to the client email, name the field, and append a "(+N more)"
count when several fields fail. console.error now logs a readable list of
every issue instead of dumping the whole form.

Adds the invalidClientField/invalidField/moreIssues toast strings across
all 13 translations.
2026-05-31 22:41:58 +02:00
MHSanaei
998fa0dfe1 fix(postgres): stop FK constraint from blocking inbound delete
The schema was written for SQLite, which never enforces foreign keys, so
relationships are managed in application code and deleting an inbound keeps
its client_traffics by design. On Postgres GORM auto-created the
fk_inbounds_client_stats constraint, which rejected those deletes with
SQLSTATE 23503.

Set DisableForeignKeyConstraintWhenMigrating so neither backend creates the
constraint, and drop the already-created one on existing Postgres DBs via
dropLegacyForeignKeys. Also revert the client_traffics deletion that
c20ee00f added to DelInbound so traffic is preserved.
2026-05-31 21:45:41 +02:00
MHSanaei
c20ee00fa3 fix(postgres): clear client_traffics before deleting inbound
DelInbound removed the client_inbounds join rows but never deleted the
inbound's client_traffics, so Postgres rejected the inbound delete with
fk_inbounds_client_stats (SQLSTATE 23503). SQLite never enforced the FK
so this went unnoticed. Delete client_traffics first, matching the order
already used in the sync path.
2026-05-31 19:48:19 +02:00
MHSanaei
b1c141a515 fix(settings): sync generated schemas
- entity.go: tighten SessionMaxAge validate tag gte=0 -> gte=1 to match the panel UI (min 60) and the hand-written setting.ts schema

- GeneralTab.tsx: add max bounds to sessionMaxAge (525600) and pageSize (1000), raise pageSize min to 1

- regenerate zod.ts/types.ts, picking up pending drift: panelProxy field, client group field, InboundFallback.dest, and dropping the stale hysteria2 protocol enum value
2026-05-31 19:00:26 +02:00
MHSanaei
cc34dc381c feat(postgres): in-panel backup/restore and consistent CLI backend
Two PostgreSQL gaps on the panel:

1. x-ui setting and other CLI subcommands read XUI_DB_TYPE/XUI_DB_DSN from
   the process environment, which systemd injects via EnvironmentFile but a
   plain shell invocation does not. On a PostgreSQL install the CLI silently
   fell back to SQLite, so changes made from the management menu never
   reached the panel's database. Load the systemd EnvironmentFile
   (/etc/default/x-ui and distro equivalents) at startup; godotenv.Load does
   not override existing vars, so it stays a no-op for the managed service.

2. DB backup/restore (panel endpoints and the Telegram bot) only handled the
   SQLite file, so on PostgreSQL Back Up returned a stale/absent x-ui.db and
   Restore silently did nothing. Add pg_dump/pg_restore based backup/restore:
   - GetDb/ImportDB run pg_dump (custom format) / pg_restore, passing
     credentials via the PG* environment instead of argv.
   - getDb downloads x-ui.dump on Postgres, x-ui.db on SQLite.
   - Telegram backup sends the matching file via GetDb.
   - BackupModal shows a Postgres note and accepts .dump; the dist page
     injects window.X_UI_DB_TYPE; new strings translated for all locales.
   - install.sh installs postgresql-client for the external-DSN path and
     points the user to in-panel Backup & Restore.

Closes #4658
2026-05-31 17:53:34 +02:00
MHSanaei
7028c15e8c i18n(nodes): translate basePath and apiToken labels
Localize the node form 'basePath' (zh-CN, zh-TW, tr-TR) and 'apiToken'
(zh-CN, zh-TW, uk-UA) labels that were still showing the English defaults.
2026-05-31 16:17:06 +02:00
MHSanaei
24d0e4ec7c fix(clients): persist group for node-inbound clients
The client create/edit form left `group` out of the request payload, so choosing a group in the form was silently dropped (bulkAdd from the Groups page still worked because it writes the column directly). Add `group` to the payload next to `comment`.

SyncInbound also overwrote group_name unconditionally; a group set via bulkAdd is never pushed to the node, so the next node snapshot — which lacks it — wiped the column. Keep group sticky (only overwrite when the incoming value is non-empty); group is only ever set/cleared via the Groups page. Preserve comment for node clients during snapshot sync the same way. Add tests.
2026-05-31 15:25:21 +02:00
MHSanaei
b94e859e73 test: name temp sqlite db x-ui.db to match the real db filename 2026-05-31 15:25:05 +02:00
MHSanaei
80110f9404 fix(inbounds): reset id sequence on delete so old ids are reused
SQLite AUTOINCREMENT keeps a high-water mark in sqlite_sequence that
deleting rows never lowers, so after removing inbounds the next add kept
climbing instead of reusing freed ids. DelInbound now realigns the
counter to MAX(id) after each delete, clearing the sqlite_sequence row
entirely when the table is empty so the next inbound starts at id 1.
Guarded behind !IsPostgres(); Postgres sequences are left untouched.
2026-05-31 00:43:26 +02:00
MHSanaei
cf50952921 feat(inbounds): add multi-select and bulk delete
Mirror the clients page: checkbox selection on the desktop table and on
mobile cards, with a danger Delete button in the toolbar that removes all
selected inbounds in one call.

Backend adds POST /panel/api/inbounds/bulkDel, which loops the existing
DelInbound per id (xray restarts at most once) and returns {deleted,
skipped}. Frontend shows a confirm modal plus a result toast, clears the
selection on success, adds bulk-delete i18n keys across all 13 languages,
and documents the endpoint in the in-panel API docs.
2026-05-31 00:29:24 +02:00
MHSanaei
6bb5a3b56b fix(inbounds): preserve client data on delete and show traffic in detail
Deleting an inbound now only detaches its clients (removes the
client_inbounds rows). It no longer deletes client_traffics or client IP
logs: those are keyed centrally by email (one row per client) and must
survive, since a client may stay attached to other inbounds and is
managed from the Clients page.

Separately, /get/:id now uses a new GetInboundDetail that preloads and
enriches ClientStats, so hydrated records (info / QR / export) carry
per-client traffic instead of null. DBInbound.toJSON drops the internal
_clientStatsMap cache so it no longer leaks into the exported JSON.
2026-05-30 23:53:28 +02:00
MHSanaei
a08bb91f58 fix(settings): reject spaces, '\' and control chars in URI path settings
webBasePath, subPath, subJsonPath and subClashPath are URL paths, so '/'
stays allowed, but spaces, backslashes and control characters break
routing. Strip them as you type (shared sanitizePath helper, now also
applied to the panel base path) and reject them on save in
AllSetting.CheckValid so direct API callers are covered too.
2026-05-30 23:29:08 +02:00
MHSanaei
2fa7be86dc fix(clients): reject spaces, '/', '\' and control chars in subscription ID
Like the client email, the subId is embedded directly in subscription
URLs, so the same characters break it. Validate it on the backend
(Create + Update) and the frontend (Zod), with a localized message
across all 13 locales. An empty subId stays allowed (it is then
auto-generated).
2026-05-30 23:28:58 +02:00
MHSanaei
a0865a67fd fix(clients): reject spaces, '/', '\' and control chars in client email
Client emails containing a slash broke the path-param routes
(edit/delete/view returned 404 / "client not found"), leaving stale
records that could only be cleared with manual SQLite edits. Validate
the email on both the backend (Create + Update, which also covers the
bulk paths) and the frontend (Zod) so these characters are rejected at
save time with a clear, localized message across all 13 locales.

Closes #4695
2026-05-30 22:40:48 +02:00
MHSanaei
987a6dd1e5 feat(clients/inbounds): IP log popups, clearer titles, tag-based inbound labels
Add an IP Log popup (view list + refresh + clear) to the client edit form and the Client Information modal, with IPs stacked vertically.

Identify inbounds by their xray tag (not remark/protocol:port) across every picker and chip: attach/detach modals, the attached-inbounds column and field, the filter drawer, and bulk-add. Add the tag field to the InboundOption schema (the backend already returned it).

Clarify modal titles/labels: Client Information (was More Information) and Inbound Information (was Inbound's Data); Client Information / QR Code titles now include the client email.

i18n: rename keys moreInformation->clientInfo and inboundData->inboundInfo with proper translations in all languages; addTitle->addClient, editTitle->editClient, addToGroupPlaceholder->groupName.
2026-05-29 23:22:49 +02:00
MHSanaei
cb7af04cd3 fix(xray): test UDP outbounds via xray probe (#4657) + Vision testseed & Flow form fixes
Outbound connection tester (#4657): UDP-based outbounds (wireguard,
hysteria, kcp/quic transports) were probed with a raw UDP dial that
treated the inevitable read timeout as success, so every one reported a
fake ~5s 'alive'. Route them through the authoritative xray
burstObservatory probe and drop the broken raw-UDP path. Test All now
runs a parallel TCP lane and a serial HTTP lane so xray-probe outbounds
don't collide on the test semaphore.

Vision testseed: the [900, 500, 900, 256] default repeats 900, and a
tags Select keys each tag by value -> 'two children with the same key,
900'. Render it as four InputNumbers (inbound + outbound forms); the
field is a fixed 4-tuple where repeats are valid.

Inbound form: drop the null-valued 'Local Panel' Select option (AntD
rejects null option values; placeholder + allowClear already cover it).

Outbound form: add an explicit 'None' option to the Flow selector.
2026-05-29 21:07:01 +02:00
MHSanaei
62c293e034 fix(outbounds): support proxyProtocol on freedom outbound
Xray's freedom outbound accepts a numeric proxyProtocol (0 disabled,
1 or 2 for the PROXY protocol version), but the panel had no field for
it and the typed form adapter dropped the key on save — so a value set
via the JSON editor disappeared the moment the outbound was saved.

Model proxyProtocol through the freedom wire schema, the form schema,
and both adapter directions (clamped to 0/1/2, omitted from the wire
when 0), and add a Select (none / v1 / v2) to the freedom section of
the outbound form. Add round-trip test coverage and the proxyProtocol
label across all locales.

Closes #4486
2026-05-29 17:18:21 +02:00
MHSanaei
7ea88e3e37 fix(clients): store flow per-inbound for shared clients
A client shared across inbounds (e.g. VLESS+TCP+Reality and VLESS+WS+TLS)
had its `flow` applied globally, so enabling xtls-rprx-vision for Reality
broke the WS+TLS inbound for the same client (#4628).

Gate flow per inbound at every fan-out site via clientWithInboundFlow,
reusing inboundCanEnableTlsFlow (VLESS+TCP+TLS/Reality only), and make
ListForInbound treat flow_override as authoritative so an empty override
means "no flow on this inbound" instead of inheriting the record's global
flow. Also tighten buildTargetClientFromSource (copy-clients) to gate on
transport, not just protocol.
2026-05-29 02:35:53 +02:00
MHSanaei
8a28373a01 fix(nodes): use GREATEST for last_online merge on PostgreSQL
setRemoteTrafficLocked merged last_online with MAX(last_online, ?), which
is SQLite's two-argument scalar max. PostgreSQL's MAX() is aggregate-only,
so node traffic sync failed every cycle with "function max(bigint, unknown)
does not exist (SQLSTATE 42883)", flooding the logs.

Add a dialect-aware database.GreatestExpr helper (GREATEST on Postgres,
MAX on SQLite) and use it for the last_online merge. last_online is a
non-null int64, so the two functions are semantically identical here.

Closes #4633
2026-05-29 02:04:02 +02:00
MHSanaei
169068d8fb fix(nodes): clean up orphaned client_inbounds on node inbound removal
When a remote node disconnects or one of its inbounds vanishes from the
traffic snapshot, setRemoteTrafficLocked deleted the central inbound row
but left the client_inbounds join rows behind. Affected clients ended up
linked to hundreds of phantom inbounds, and editing one then failed with
"record not found" / "Load Old Data Error" because Update aborted on the
first GetInbound miss.

- Detach client_inbounds rows when deleting a vanished node inbound
- Prune stale links during client Update instead of aborting the save
- Drop orphaned client_inbounds rows on startup to heal existing DBs

Closes #4636
2026-05-29 01:41:52 +02:00
MHSanaei
798e18b6ee feat(fallbacks): add per-rule dest override
Operators can now type an explicit dest (e.g. "8443", "127.0.0.1:8443",
"/dev/shm/x.sock") on each fallback row to override the auto-resolved
child listen+port. Empty keeps the existing auto behavior.

Adds the column to inbound_fallbacks (GORM AutoMigrate), threads it
through the panel form, API docs, and translations.
2026-05-28 21:17:49 +02:00
MHSanaei
3f0b7fbe97 feat(tls): surface pinnedPeerCertSha256 in panel, share links, and subs
Adds a panel-only `pinnedPeerCertSha256` field on TLS settings with a tags input and a random-hash generator. The hashes ride share links as `pcs` (v2rayN-compatible), Clash sub as `pin-sha256`, and JSON sub as `pinnedPeerCertSha256`, while remaining stripped from the run-config sent to xray-core.
2026-05-28 19:32:10 +02:00
MHSanaei
c5b5606bf5 i18n(panel): translate Copy/Cancel buttons, Stream/Sniffing tabs, and All-Inbounds filenames
- TextModal: route the Copy button label and the post-copy toast
  through t('copy')/t('copied') instead of hardcoded English.
- PromptModal: route cancelText through t('cancel') and default okText
  through t('confirm') so the import-inbound prompt stops showing
  "Cancel" in non-English UI.
- InboundsPage: pass the All-Inbounds and All-Inbounds-Subs download
  filenames through t(...) so each locale can localize them.
- en-US.json: add pages.inbounds.exportAllLinksFileName and
  pages.inbounds.exportAllSubsFileName.
- All 12 non-English locales: translate streamTab and sniffingTab
  (previously left as literal English) and add the two new filename
  keys with appropriate translations.

All 13 locale files now have 1541 lines.
2026-05-28 18:45:59 +02:00
MHSanaei
99df5d70a8 fix(clients): backfill missing subId on startup and guard create/update
Legacy clients (and any API consumer that POSTs to AddInboundClient
without a subId) ended up with an empty SubID, which breaks the panel's
sub-link generation. Backfill them once at startup and stop the gap at
the write path so new clients can't reintroduce it.

- util/random: add NumLower(n) — 16-char [0-9a-z] generator that matches
  the frontend's RandomUtil.randomLowerAndNum convention.
- database/db.go: new InboundClientSubIdFix seeder, modeled on
  InboundClientTgIdFix. Loops every inbound, parses settings.clients,
  fills empty/missing subId with random.NumLower(16), persists via the
  same transaction-wrapped Update("settings", …) path, then records in
  HistoryOfSeeders so it runs at most once.
- web/service/client.go: defense-in-depth in AddInboundClient and
  UpdateInboundClient — fill subId on the persisted settings map when
  the payload omits it (Update prefers the previous value before
  generating a fresh one).
- database/db_seed_test: cover empty subId, missing-key subId, and
  preserved-existing subId; assert exactly one HistoryOfSeeders row.
2026-05-28 18:20:34 +02:00
MHSanaei
72b97efa8a i18n(panel): migrate hardcoded panel strings to en-US and translate all locales
Surface ~400 hardcoded English labels, tooltips, placeholders, dt/divider
text, modal okText/cancelText, and Spin loading from the panel pages
(clients/groups/inbounds/nodes/settings/xray/sub/index) into
web/translation/en-US.json under existing pages.<page>.* namespaces, with
JSX swapped to t(...). Brand and protocol identifiers (TLS, MTU, SNI,
NordVPN, Cloudflare WARP, etc.) stay literal.

Sync all 12 non-English locales (ar-EG, es-ES, fa-IR, id-ID, ja-JP,
pt-BR, ru-RU, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW) to match en-US's
structure and translate the 521 new key paths per locale. Every locale
file now has 1539 lines, mirroring en-US ordering.

Also remove a dead duplicate "info": "Info" key under pages.inbounds
that collided with the new pages.inbounds.info.* object.

Backend: bulk attach/detach errors in web/service/client.go now route
through logger.Warningf (so they appear under /panel/api/server/logs/)
instead of only living on the response payload.
2026-05-28 18:03:07 +02:00
Aleksey Surkov
0829f1ecd4 change tg message when send qrCode (#4623) 2026-05-28 17:26:45 +02:00
MHSanaei
c5dc84d314 refactor(inbound-tag): drop protocol segment from canonical shape
The canonical tag is now "[n<id>-]in-[<listen>:]<port>-<transport>"
instead of "[n<id>-]in-[<listen>:]<port>-<protocol>-<transport>".
Two TCP inbounds on the same port are already blocked by
checkPortConflict, so only the transport segment is needed to
disambiguate the legitimate tcp/udp coexistence case.

Existing DB rows keep their current tags via resolveInboundTag's
"reuse if free" branch — no migration needed (protocol-segment
form was never released).
2026-05-28 15:50:00 +02:00
MHSanaei
b42a4d93fc fix(inbounds): heal legacy client data and TLS cert form hydration
- Detach preserves client traffic stats. DelInboundClient,
  DelInboundClientByEmail, and bulkDelInboundClients now take a
  keepTraffic flag; Detach passes true, delete-paths keep prior
  behavior. Runtime user removal still runs so xray drops the session.
- Two startup seeders normalize legacy inbound settings JSON:
  clients:null -> [] and any non-numeric tgId -> 0 (string, bool,
  NaN, Inf, non-integer floats). Each records itself once in
  history_of_seeders.
- MigrationRequirements no longer rewrites empty clients arrays back
  to null: newClients is initialized as a non-nil slice and incoming
  clients:null is coerced before the type assertion.
- TLS cert form: rawInboundToFormValues synthesizes a useFile
  discriminator per cert from whichever side carries data, so the
  edit modal can show file-mode paths again. formValuesToWirePayload
  strips useFile so saved JSON stays in wire shape.
2026-05-28 15:11:53 +02:00
MHSanaei
530e338c66 refactor(clients): coherent group management — rename, split, extract
This bundles a set of group-related improvements that built up across
one session and only make sense together.

Terminology / API surface:
- Rename "assign group" → "add to group" everywhere: i18n keys,
  callback names (bulkAddToGroup), component + file names
  (BulkAddToGroupModal, AddClientsToGroupModal), Go controller/struct
  names (bulkAddToGroup, AddToGroup), OpenAPI summaries. Nothing keeps
  the word "assign" anymore.
- Move group routes under /panel/api/clients/groups/* (was
  /bulkAssignGroup at the clients root).
- Split add and remove into two endpoints: /groups/bulkAdd now rejects
  empty group; new /groups/bulkRemove clears the label for the given
  emails. The old "submit empty to clear" UX is gone — Ungroup is its
  own action.

UI affordances on Clients page:
- Promote Group + Ungroup to visible bar buttons next to Attach +
  Detach. Group reuses BulkAddToGroupModal; Ungroup pops a danger
  confirm and calls bulkRemoveFromGroup.
- Custom UngroupIcon (TagsOutlined with a diagonal strike) for the
  Ungroup button so the pairing reads at a glance.
- Hide the Group column when no clients have a group label yet —
  removes a column of em-dashes on fresh installs.

UI on Groups page:
- New per-row Add clients… / Remove clients… actions backed by
  GroupAddClientsModal and GroupRemoveClientsModal: rich client picker
  (email / comment / current group / enable) with search and
  preserveSelectedRowKeys, mirroring the inbounds Attach modal UX.

Controller split:
- Move all /groups/* routes, handlers, and request bodies out of
  web/controller/client.go into a dedicated web/controller/group.go
  (GroupController with leaner clientService + xrayService
  dependencies). URLs are byte-identical because the new controller
  registers on the same parent gin.RouterGroup; api_docs_test.go gets
  a group.go → /panel/api/clients basePath entry so its route
  extraction keeps working.

Invalidation dedup:
- Removing a client from a group on the Groups page used to refetch
  /clients/groups and /clients/onlines three times: once from the
  mutation's onSuccess, once from a redundant invalidate() in the
  page's onSubmit, once from the WebSocket invalidate broadcast that
  the backend fires after every mutation. The manual invalidate() is
  gone, and a small invalidationTracker module lets websocketBridge
  skip WS-driven invalidates that arrive within 1.5s of a local
  invalidate — bringing the refetch count down to one. The WS path
  still works for changes made by another tab or user.
2026-05-28 12:59:20 +02:00
MHSanaei
bf1b488a63 feat(clients): tidier bulk action toolbar
When at least one client is selected, the toolbar now collapses to a
small selection indicator plus the three most-used actions instead of
spreading six count-suffixed buttons across the row:

- Replaces every per-button "(N)" with a single closable "{N} selected"
  tag on the left — one click on its × clears the selection.
- Hides "+ Add Clients" while a selection is active (focus mode).
- Keeps Attach, Detach, and Delete as visible buttons; Delete is pushed
  to the right with auto margin so it doesn't sit flush against the
  non-destructive actions.
- Folds Adjust, Group, and Sub links into the existing "more"
  dropdown, which is now context-aware: selection-scoped overflow when
  rows are picked, global actions (Add Bulk / Reset all / Del depleted)
  otherwise.

On mobile the new buttons collapse to icon-only the same way as the
rest of the toolbar.
2026-05-28 11:24:21 +02:00
MHSanaei
8d6d845262 feat(settings): include email in default remarkModel pattern
Change the out-of-the-box remarkModel from "-io" (Inbound, Other) to
"-ieo" so newly provisioned panels include the client's email between
the inbound name and the other slot — much easier to identify which
client a generated remark belongs to. Existing installs that have
already written a remarkModel value are unaffected; only first-run /
unset defaults inherit the new pattern.
2026-05-28 11:08:59 +02:00
MHSanaei
72b68cce22 feat(clients): selective bulk attach + new bulk detach
Inbounds page:
- AttachClientsModal now shows a per-client selection table (email,
  comment, enabled tag) with search and a live "selected of total"
  counter; all clients are pre-selected so the old "attach all"
  workflow stays a single OK click.
- New DetachClientsModal on the inbound row menu lets you pick which
  clients to remove from that inbound (records are kept so they can be
  re-attached later; for full removal use Delete).

Clients page:
- New "Attach (N)" bulk-action button + BulkAttachInboundsModal that
  attaches selected clients to one or more multi-user inbounds.
- New "Detach (N)" bulk-action button + BulkDetachInboundsModal that
  removes selected clients from chosen inbounds; (email, inbound) pairs
  where the client isn't attached are silently skipped.

Backend adds POST /panel/api/clients/bulkDetach, wrapping the existing
Detach service for each email and reporting per-email
detached/skipped/errors. ClientRecord rows are kept on detach to match
the single-client endpoint; bulkDel remains the path for full removal.
2026-05-28 11:08:52 +02:00
MHSanaei
9e005ffcf9 feat(inbounds): restore "Set Cert from Panel" / Clear buttons in TLS certs
Bring back the per-certificate buttons in the inbound TLS section (File Path
mode): "Set Cert from Panel" fetches the panel's own webCertFile/webKeyFile via
/panel/setting/all and fills the cert's certificateFile/keyFile, warning when no
panel cert is configured; "Clear" empties both paths.

Reuses the existing pages.inbounds.setDefaultCert label and adds a
setDefaultCertEmpty warning string.
2026-05-28 02:41:39 +02:00
MHSanaei
1a096d72f1 feat(inbounds): bulk-attach & assign-group client actions + form defaults
- Bulk-attach an inbound's clients onto other inbounds (same identity, shared traffic): new ClientService.BulkAttach + POST /clients/bulkAttach, an inbound row action, and AttachClientsModal.
- Assign all of an inbound's clients to a group from the inbound page, reusing /clients/bulkAssignGroup and the existing BulkAssignGroupModal.
- Default a random user/pass account for new Mixed and HTTP inbounds instead of an empty accounts list.
- Capitalize the inbound Security toggle labels (None/TLS/Reality).
2026-05-28 01:54:32 +02:00
MHSanaei
9d9737f470 feat(settings): panel network proxy for the panel's own outbound requests
Add a panelProxy setting that routes the panel's self-initiated HTTP requests (geo updates, Xray version/core download, panel update check) through an admin-configured socks5/http(s) proxy, to bypass server-side filtering of GitHub/Telegram. The Telegram bot falls back to it when tgBotProxy is empty (socks5 only). New util/netproxy.NewHTTPClient builds the proxied client.

Also fix the Mixed-inbound SOCKS/HTTP share URLs that had host:port and user:pass in the wrong order, and consolidate the Telegram settings tab (move API server into the general tab, drop the empty Proxy & Server tab).
2026-05-28 00:45:32 +02:00
Sanaei
272854df91 Client/inbound resilience + Postgres pool tuning + schema fixes (#4607)
* fix(clients): fall back to inbound scan when ClientRecord is missing

DeleteByEmail looked up the email in client_records and returned the
raw "record not found" gorm error when nothing matched, even though
the client could still live inside an inbound's settings.clients JSON
(legacy entries that SyncInbound never picked up, or rows deleted out
from under a stale inbound). The user-visible delete then fails
mysteriously while xray happily keeps serving the client.

When GetRecordByEmail returns ErrRecordNotFound, walk inbounds whose
settings JSON references the email and run DelInboundClientByEmail on
each. The traffic / IP rows are cleaned up at the end unless keepTraffic
is set. If no inbound carries the email either, surface a clear
"client %q not found in any inbound or client record" error instead.

* chore(logging): include request + caller context in jsonMsgObj warnings

The generic "X-UI: Something went wrong. Error: record not found" log
gave no clue about which endpoint, client, or controller line emitted
it. Prepend a context block:

  [POST /panel/api/clients/del/ADMIN ip=109.124.234.127
   handler=controller.(*ClientController).delete client.go:146]

Handler frame is located by scanning the stack for the first caller
outside util.go, so it points at the right controller method whether
the path went through jsonMsg, jsonObj, or jsonMsgObj directly.

* fix(clients): tolerate orphan client_inbounds rows in Delete

DeleteByEmail's previous fix only covered the case where GetRecordByEmail
returned ErrRecordNotFound. When the ClientRecord exists but a client_inbounds
row points to an inbound that has been removed out-of-band (failed mid-delete,
manual SQL, pre-SyncInbound migration), Delete bubbled the raw gorm
"record not found" from inboundSvc.GetInbound and aborted before any cleanup
ran — leaving the client un-deletable through the UI/API.

Match the tolerance bulkDelInboundClients already has: when GetInbound
returns gorm.ErrRecordNotFound for a join row, log a warning and continue.
The unconditional Delete(&model.ClientInbound{}) later in the function then
removes the stale row, and the ClientRecord delete succeeds.

* fix(schemas): accept empty-string fingerprint on externalProxy

The External Proxy form offers a "Default" option with value '' for the
uTLS fingerprint dropdown, but UtlsFingerprintSchema.optional() rejects
empty strings (only undefined or a valid enum member). Saving an inbound
with externalProxy rows failed with `expected one of "360"|"chrome"|...`.

Preprocess '' to undefined before the optional enum, matching the existing
pattern used for VmessSecuritySchema.

* chore(logging): drop noisy orphan client_inbounds warning

Per-row WARNINGs spammed logs whenever a client referenced multiple
already-deleted inbounds. The continue keeps the orphan-tolerant
behavior; just no longer announces each skipped row.

* feat(clients): per-client VMess security in client form

Restores the VMess `security` selector on the client form (auto, aes-128-gcm,
chacha20-poly1305, none, zero) and surfaces it only when at least one attached
inbound is VMess. The value rides into the share link via the existing
`scy=` field in genVmessLink; the panel persists it on ClientRecord and in
the inbound's settings.clients so the link generator can read it back.

Adds the pages.clients.vmessSecurity i18n key in en-US and fa-IR.

* fix(xray-config): strip panel-only fields from inbound config

Two fields the panel stores but Xray doesn't accept on the inbound side:

- VMess clients[].security — panel persists it so the share-link generator
  can write `scy=...`, but xray's vmess inbound spec has no per-client
  security. The field was leaking into the inbound JSON pushed to xray-core.
- VLESS settings.encryption — per the xray spec the inbound only takes
  `decryption`; `encryption` is for the matching client outbound. The panel
  keeps it for operator reference, but it must not appear in the inbound
  payload.

Add two strip helpers next to HealShadowsocksClientMethods and wire them
into GenXrayInboundConfig via a per-protocol switch, so both local and
remote runtime paths get the cleaned config.

* chore(db): backend-aware pool sizes with env overrides

Per-backend defaults:
- Postgres: 25 max open / 25 max idle. Matching idle to open removes
  pool churn under bursts (Postgres handles concurrency at the server,
  idle connections are cheap).
- SQLite: 1 max open / 1 max idle. Single-writer model means a wider
  cap just queues behind busy_timeout; tight cap is honest.

Both back ends share ConnMaxLifetime=1h and ConnMaxIdleTime=30m so
stale connections (vault rotation, pgbouncer drops, load-balancer
idle eviction) rotate out without operator intervention.

Operators can override either default at boot via:
  XUI_DB_MAX_OPEN_CONNS=...
  XUI_DB_MAX_IDLE_CONNS=...

envInt parses these; missing/empty/non-positive values fall back to
the per-backend default.

* fix(schemas): accept boolean acceptProxyProtocol on TCP stream

TcpStreamSettingsSchema declared `acceptProxyProtocol: z.literal(true).optional()`,
so saving an inbound where the AntD Switch sat in the off state failed
validation with `Invalid input` because the Switch always emits a plain
boolean.

Switch to `z.boolean().default(false)` — same shape ws/sockopt/httpupgrade
already use, and matches the actual wire payload (golden fixtures and
other settings blocks all store `acceptProxyProtocol: false`).

Snapshots for stream.test and inbound-full.test pick up the new defaulted
field on TCP fixtures.
2026-05-27 22:51:37 +02:00
MHSanaei
3046d96145 refactor(inbound-tag): add short protocol segment, rename tcpudp suffix
Tag shape becomes "[n<id>-]inbound-[<listen>:]<port>-<proto>-<net>"
where <proto> is a 2-char alias (vmess→vm, vless→vl, trojan→tr,
shadowsocks→ss, mixed→mx, wireguard→wg, hysteria→hy, tunnel→tn;
http stays as "http"), and <net> uses "tcpudp" for the TCP+UDP combo
instead of the previous "mixed" (which clashed visually with the
mixed protocol name).

Examples:
  local VLESS TCP 443        → inbound-443-vl-tcp
  local Hysteria UDP 443     → inbound-443-hy-udp
  local Mixed protocol dual  → inbound-22912-mx-tcpudp
  local Tunnel allow=tcp,udp → inbound-51542-tn-tcpudp
  node 1 VLESS TCP 443       → n1-inbound-443-vl-tcp

protocolShortName returns the raw protocol identifier for anything not
in the table, so future protocols still get a tag without a code edit.
Existing inbound tags are left alone — only newly generated tags adopt
the shape.
2026-05-27 19:47:02 +02:00
MHSanaei
7ade9d9a1f refactor(inbound-tag): node-prefixed + transport-suffixed canonical shape
Tag scheme moves to "[n<nodeID>-]inbound-[<listen>:]<port>-<transport>"
so two long-standing collision classes go away on the create path:
  - tcp/443 and udp/443 on the same listener (independent sockets)
  - same listen+port living on the central panel and on a remote node

Examples:
  local TCP 443    → inbound-443-tcp
  local UDP 443    → inbound-443-udp
  node 1 TCP 443   → n1-inbound-443-tcp

Refactor:
  - composeInboundTag is the single source of truth, called from
    generateInboundTag. Transport segment is now always present
    (used to appear only on collision); n<id>- prefix is added when
    Inbound.NodeID != nil.
  - addInbound / importInbound drop their inline "inbound-<port>"
    fallback; an empty Tag now flows through resolveInboundTag, which
    keeps caller-supplied tags verbatim when free and otherwise
    delegates to generateInboundTag.
  - setRemoteTrafficLocked indexes tagToCentral under both the stored
    tag and the prefix-stripped form, so a node sending its bare tag
    still resolves to a row we may have rewritten at materialization.
    The create branch now picks between snap.Tag and the n<id>-
    prefixed form before falling back to the warn-once skip.
  - Tests updated for the always-on transport suffix, and two new
    cases cover the node-prefix behaviour.

Existing inbounds keep their tags — only newly generated tags adopt
the new shape, so user routing rules pointing at "inbound-443" still
match the row they always did until the row is recreated.
2026-05-27 19:14:22 +02:00
MHSanaei
d347605233 fix(remote-traffic): handle tag collisions + readable warning format
setRemoteTrafficLocked attempted to INSERT a new central inbound for
every snap whose tag was not in tagToCentral (which is scoped by
node_id). When a different owner — the local panel or another node —
already held the tag, the INSERT tripped the UNIQUE constraint on
inbounds.tag and re-fired on every periodic snap.

Pre-check for tag ownership before the INSERT. If a different owner
holds it, log once per (nodeID, tag) via a sync.Map dedupe and skip
silently from then on. Real DB errors still surface.

Also switch the six setRemoteTraffic warnings from logger.Warning(...)
to logger.Warningf("%q ... %v", ...) — fmt.Sprint only inserts spaces
between adjacent non-string operands, so the all-string call sites
produced runs like "taginbound-443failed:UNIQUE...". Format strings
also let us quote the tag with %q so it stands out from the prose.
2026-05-27 18:34:18 +02:00
MHSanaei
e23599cb18 feat(inbounds): row action to delete all clients of an inbound
Adds POST /panel/api/inbounds/:id/delAllClients that collects every
client email from settings.clients[] and runs ClientService.BulkDelete
in one pass. Row action lives in the More menu as a danger item, only
shown for multi-user inbounds that currently have at least one client;
confirmation modal displays the live client count.
2026-05-27 18:17:44 +02:00
MHSanaei
93eda06878 feat(clients,groups): client groups + sub-links export + dedicated groups page
Persistent client groups
- New ClientGroup model + client_groups table that holds empty
  (placeholder) groups so a user can define a label before any client
  references it. ListGroups merges these with the distinct group_name
  values already stored on clients and reports {name, clientCount}.
- ClientRecord gains group_name column; the model.Client wire shape
  gains a matching `group` JSON field that survives the
  inbound.settings → SyncInbound round-trip.
- Rename/Delete on a group mutates client_groups (rename row / delete
  row) AND propagates to all matching clients in ClientRecord and in
  every owning inbound's settings JSON, all in one transaction.

Bulk operations
- AssignGroup(emails, group) updates clients.group_name + patches each
  affected inbound's settings JSON in one read-modify-write per inbound.
  Empty group clears the label. Auto-creates the client_groups row when
  the user assigns to a brand-new name.
- BulkResetTraffic(emails) loops the existing single-reset path so the
  caller can zero traffic across a whole selection or a whole group.
- EmailsByGroup(name) returns just the email list (used by the groups
  page to fan a single bulk action over every member).

Endpoints (all under /panel/api/clients)
- GET  /groups                         — summaries with counts
- GET  /groups/:name/emails            — emails in a group
- POST /groups/create                  — empty placeholder group
- POST /groups/rename                  — rename (table + clients + JSON)
- POST /groups/delete                  — drop label everywhere (clients survive)
- POST /bulkAssignGroup                — assign N selected clients
- POST /bulkResetTraffic               — reset traffic on a list

Clients page UX
- New Group column (Actions → Client → Group → Inbounds → …) with a
  click-to-filter chip.
- FilterDrawer gains a multi-select Group filter whose options come
  from the new ClientPageResponse.groups field (sourced from ListGroups
  so empty/placeholder groups are pickable too).
- Single-client and bulk-add forms gain a Group AutoComplete pre-loaded
  with all known group names.
- New toolbar buttons when selection > 0: "Group ({n})" opens
  BulkAssignGroupModal, "Sub links ({n})" opens SubLinksModal.

Sub-links export modal (new SubLinksModal.tsx)
- Table of selected clients with their subscription URL (and JSON URL
  when subJsonEnable is on), per-row copy, Copy all, and Download as
  sub-links-<timestamp>.txt. Warns when subscription is disabled or
  none of the selected clients have a subId.

Dedicated Groups page (new pages/groups/GroupsPage.tsx)
- /groups route + sidebar entry (TagsOutlined icon) + page title key.
- Card-based layout matching Clients/Inbounds/Nodes — summary card with
  Total/Grouped/Empty stats, main card with Add Group button + table.
- Per-row More dropdown (icon-first column on the left): Sub links,
  Adjust (days+traffic), Reset traffic, Rename, Delete clients in
  group, Delete group (keep clients). Empty groups disable the
  client-targeted actions.
- Reuses SubLinksModal and ClientBulkAdjustModal — emails for the
  group are fetched on demand from GET /groups/:name/emails.

Other polish
- /groups + groups-page selectors added to page-shell.css and
  page-cards.css so the new page inherits the same background, padding,
  card borders, hover shadow, and summary-card padding.
- .card-toolbar gains a small vertical padding so the larger toolbar
  buttons (now default size, matching Inbounds) don't crowd the top of
  the card-head on Clients and Groups pages.
2026-05-27 17:30:55 +02:00
MHSanaei
7680e27d1d feat(clients): toolbar sort selector + preserve updated_at on unchanged rows
Frontend
- New Sort dropdown in the clients toolbar covering oldest/newest,
  recently updated, recently online, email A↔Z, most traffic, highest
  remaining, expiring soonest. Default is Oldest first.
- Strip per-column sorter arrows from the Table — all sorting now flows
  through the single dropdown, so the column headers stop competing
  with it.
- Empty state: TeamOutlined icon, t('noData'), text-secondary color
  (matching the inbound/node polish).

Backend
- sortClients: add createdAt, updatedAt and lastOnline cases (with id
  tie-break for stable ordering when timestamps collide).
- Fix Recently updated: SyncInbound was calling tx.Save on every client
  in the inbound, and GORM's autoUpdateTime tag stamped updated_at to
  time.Now() each time — so editing one client bumped ALL of them.
  After the Save, restore each row's preserved updated_at via
  UpdateColumn (skips hooks). The actually-edited client gets its
  fresh stamp from the explicit UpdateColumn at the end of Update().
- Fix periodic updated_at churn: adjustTraffics unconditionally set
  c["updated_at"] = now() for every client in any inbound that had a
  delayed-start expiry, every traffic-stats pass. Turn that into a
  backfill (only when the key is missing), matching the created_at
  treatment one line above.
2026-05-27 15:07:17 +02:00