rmyndharis/OpenWA v0.4.7

on GitHub

A webhooks, reliability, and dashboard release — no breaking changes; everything is additive or a
fix. Webhooks gain optional smart pre-dispatch filters: a trigger can carry AND-ed conditions
(sender/recipient/body/type/mentions/fromMe/hasMedia/isGroup) and fires only when they all match,
with engine-neutral WaId contact matching and a FilterBuilder UI — a webhook with no filters
behaves exactly as before. The whatsapp-web.js engine's first-boot init timeout is now configurable
(WWEBJS_AUTH_TIMEOUT_MS) for slow environments. Fixed: the dashboard no longer crashes on
PostgreSQL when a webhook exists (a JSON column type mismatch). Dashboard: a downed backend no
longer floods the screen with error toasts.

Added

• Smart webhook filters (optional, additive). A webhook trigger can now carry an optional set of
  pre-dispatch conditions, evaluated per event before delivery: it fires only when all conditions
  match (AND). Conditions match on sender / recipient / body / type / mentions / fromMe /
  hasMedia / isGroup with is / isNot / contains / equals operators;
  message-only conditions are skipped for non-message events, so a *-subscribed webhook still fires on
  session events. A webhook with no filters behaves exactly as before. Contact-id conditions
  (sender/recipient/mentions) match by the engine-neutral WaId key, so a filter written as a
  plain number or in any dialect (@c.us / @s.whatsapp.net / @lid) matches the same person - and a
  lid-addressed sender (e.g. an unresolved @lid group participant) matches a phone filter once the
  persistent lid -> phone table knows the mapping. Configurable via the API (filters on create/update)
  and a new FilterBuilder UI on the dashboard's Webhooks page. (#379)

• Configurable first-boot init timeout for the whatsapp-web.js engine (WWEBJS_AUTH_TIMEOUT_MS).
  On slow first boots (e.g. WSL2 or low-resource containers) the engine's fixed 30s wait for WhatsApp
  Web to finish loading could expire before the QR code was generated, aborting startup. Set
  WWEBJS_AUTH_TIMEOUT_MS to a larger value in milliseconds (e.g. 120000) to extend it; unset keeps
  the previous 30000ms default, so existing deployments are unchanged. (#353)

Changed

• Dashboard collapses connection-error spam into a single toast. When the backend is unreachable
  (failed to fetch, network errors, HTTP 502/503), the dashboard now shows one translated "Server
  Connection Lost" toast that auto-dismisses, instead of stacking an error toast per failed request —
  de-duplicated on a stable key so translation can't break it. Original work by @quinton-8. (#293)

Fixed

• Dashboard no longer crashes ("Something went wrong") when a webhook exists on PostgreSQL. JSON
  columns (webhooks.events/headers, sessions.config, messages.metadata, message_batches.*)
  were declared jsonb in their entities but created as text by the baseline migration, so on
  Postgres the driver returned them as raw JSON strings and the dashboard's events.map() threw an
  uncaught error. jsonColumnType() now resolves to simple-json on both dialects (parsed in JS on
  read) — no schema migration or data conversion, since the write format was already identical. This
  also corrects the same latent string-instead-of-object behavior for session reconnect config,
  message-reaction persistence, and bulk-send batches on Postgres. The dashboard additionally
  normalizes webhook events to an array at the query boundary as defense-in-depth. (#385)