The let(:channel) and the service's message.inbox.channel were different
object instances, so the chat_id / business_connection_id stubs on the
test instance never applied to the service. Other examples masked it
because they only counted requests; the business-chat example matches
on body and tripped. Set the values via conversation.additional_attributes
(where the channel methods actually read from) instead of stubbing.
- call_permission_reply specs: enable channel_voice and set
provider_config['source'] = 'embedded_signup' so voice_enabled? returns
true; without it the service short-circuits before any handling.
- whatsapp_cloud_service specs: stub /v22.0/<phone_id>/calls (and
/messages) to match calls_phone_id_path, which uses
WHATSAPP_API_VERSION (fallback v22.0). The OSS v13.0 path is locked
for legacy /messages compatibility only.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
# Pull Request Template
## Description
This PR fixes the non-functional resend confirmation feature on the V3
login page where clicking "Resend confirmation" did nothing. The issue
was caused by the V3 store not having the `resendConfirmation` action
that the login page was trying to dispatch.
**Key improvements:**
- Fixed V3 store integration by importing `resendConfirmation` directly
from auth API
- Added comprehensive UX improvements with loading states and 60-second
cooldown timer
- Implemented environment-aware debug logging for development
- Added proper error handling and user feedback
- Enhanced backend test coverage
**Context:** Users with unconfirmed accounts were unable to resend
confirmation emails from the login page, creating a poor user experience
and potential support burden.
Fixes#3157
## Type of change
- [x] Bug fix (non-breaking change which fixes an issue)
- [x] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality not to work as expected)
- [ ] This change requires a documentation update
## How Has This Been Tested?
**Backend Testing:**
- All existing resend_confirmation tests passing (7/7)
- Added comprehensive new test suite in
`spec/requests/api/v1/resend_confirmation_spec.rb`
- API endpoint returns 200 OK responses in ~0.39 seconds
- Email delivery confirmed via SMTP with test user `info@airbonar.com`
**Frontend Testing:**
- All frontend tests passing
- ESLint compliant code with automatic corrections applied
- Manual testing of login page functionality:
- 60-second cooldown timer with countdown display
- Error handling with user-friendly messages
- Development logging works (console output in dev mode only)
**Test Configuration:**
- Ruby/Rails backend with RSpec test suite
- Vue.js frontend with Jest/testing-library
- Development environment with Gmail SMTP configured
- Test user: unconfirmed account `info@airbonar.com`
**Reproduction Steps:**
1. Navigate to login page with unconfirmed account
2. Click "Resend confirmation link"
3. Observe loading state, API call, and success feedback
4. Verify 60-second cooldown prevents spam
5. Check email delivery.
## Checklist:
- [ ] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [x] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [ ] Any dependent changes have been merged and published in downstream
modules
---------
Co-authored-by: Sivin Varghese <64252451+iamsivin@users.noreply.github.com>
Co-authored-by: Sojan Jose <sojan@pepalo.com>
Co-authored-by: Sony Mathew <2040199+sony-mathew@users.noreply.github.com>
Co-authored-by: Sony Mathew <sony@chatwoot.com>
- Enable channel_voice and set provider_config['source'] = 'embedded_signup' in
the before block so Channel::Whatsapp#voice_enabled? actually returns true;
otherwise the service short-circuits before any handling runs.
- Outbound `connect` only stores the SDP answer and broadcasts
voice_call.outbound_connected; the in_progress / started_at transition has
moved to the separate status=ACCEPTED webhook. Update the existing-call spec
to expect status: 'ringing' and started_at: nil instead.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Wrap mark_outbound_accepted, accept_outbound_call, and handle_terminate in
call.with_lock so they serialize against Whatsapp::CallService (which holds
call.with_lock across the Meta API call). Without this, an ACCEPTED webhook
arriving mid-terminate could overwrite the freshly-finalized terminal status
back to in_progress and broadcast an erroneous outbound_accepted.
- Drop build_missed_inbound_call. An outbound terminate landing in the tiny
window between provider_service.initiate_call and Call.create! had no local
row, so the fallback materialised an inbound row with the same
provider_call_id and tripped the unique (provider, provider_call_id) index
on the controller's create. The "out-of-order webhook" case it protected is
rare in practice; trade that for eliminating the false-positive collision.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
# Pull Request Template
## Description
Bridges WhatsApp Cloud Calling onto Chatwoot's voice pipeline so agents
can answer and place calls on WhatsApp Cloud inboxes with the same UX as
Twilio voice. Browser ↔ Meta WebRTC is direct (no media-server hop);
Chatwoot owns signalling, recording upload, and `Call`/`Message`
lifecycle state.
Companion PRs:
- FE: #14346 (`feat/whatsapp-call-ui`) — UI consumer for these
endpoints.
- Specs: #14357 (`feat/whatsapp-call-meta-bridge-specs`) — backend test
coverage.
## Type of change
- [x] New feature (non-breaking change which adds functionality)
## What changed
- **API** — new `Api::V1::Accounts::WhatsappCallsController` with `show
/ accept / reject / terminate / initiate / upload_recording`.
Conversation-level Pundit auth. `initiate` is gated on the channel being
embedded-signup voice-enabled. Permission-blocked initiations (Meta
error 138006) return **422** with `{ status: 'permission_requested' |
'permission_pending' }` and trigger a throttled opt-in template send
under a conversation lock.
- **State machine** — `Whatsapp::CallService` handles accept / reject /
terminate with call-level locking. Meta failures are wrapped as
`Voice::CallErrors::CallFailed` so the controller renders 422 instead of
leaking 500s.
- **Webhooks** — `Whatsapp::IncomingCallService` processes Meta
connect/terminate events. Pins `setup:active` on outbound answers,
materialises a missed-call record when terminate arrives before connect
(out-of-order delivery), and broadcasts `voice_call.*` events to the
assignee's pubsub stream when assigned, otherwise account-wide.
- **Provider** — `Whatsapp::Providers::WhatsappCloudService` adds
`pre_accept_call / accept_call / reject_call / terminate_call /
initiate_call / send_call_permission_request`. Uses configurable
`WHATSAPP_API_VERSION` (default v22) since Calls API needs v17+ and the
OSS `phone_id_path` is locked at v13 for legacy `/messages`
compatibility.
- **Audio recordings** — `Attachment after_create_commit` enqueues
transcription and rebroadcasts the message so the FE bubble updates
immediately. Active Storage initializer allows audio MIME types to serve
inline.
## How to test
1. Set up a WhatsApp Cloud Embedded Signup inbox; flip
`provider_config.calling_enabled = true` (Calls tab in inbox settings —
ships with FE PR #14346).
2. Place a real call from a phone number that has previously messaged
the business → expect `voice_call.incoming` cable + `Call` row +
`voice_call` Message bubble.
3. Click Accept (FE PR) → `/whatsapp_calls/:id/accept` 200, audio flows
browser ↔ Meta, recording uploads on hangup, Whisper transcript appears
within ~10s.
4. From Chatwoot, click outbound → `/whatsapp_calls/initiate` 200 if the
contact is opted-in, otherwise **422** with `status:
'permission_requested'` and a template send (FE shows a banner instead
of an error toast).
5. Refresh during a ringing call → call survives on Meta, FE seeds it
back from the conversation cache, agent can still accept.
6. Refresh during an active call → `pagehide` beacon terminates the call
on Meta with `status: 'completed'` (no orphan).
## Checklist
- [x] Code follows project style (RuboCop / ESLint clean)
- [x] Self-review done
- [x] Hard-to-understand areas commented
- [ ] Documentation update (architecture doc lives in companion branch)
- [x] No new warnings
- [x] Tests live in companion PR #14357
- [x] Existing tests pass locally
- [x] Dependent changes merged downstream (FE #14346 consumes these
endpoints)
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
If Meta delivered a connect for an outbound call before the controller
committed the Call row (race between Meta's API response and our INSERT),
the handler treated the unknown call_id as inbound and either tripped the
unique (provider, provider_call_id) index or broadcast voice_call.incoming
for an outbound. Use session.sdp_type to gate inbound creation on 'offer'
only; 'answer' with no row is logged and skipped — the next status webhook
finds the now-committed row.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Description
This pull request addresses issue #11948, where inline images embedded
in emails (such as those sent from Outlook) are not rendered correctly
if the Content-Disposition header is missing.
The solution ensures that images referenced via cid: in the HTML body
are correctly identified and rewritten using url_for.
Fixes#11948
## Type of change
Bug fix (non-breaking change which fixes an issue)
## How Has This Been Tested?
Added test: detects image inline attachment by cid reference when
Content-Disposition is missing
## Checklist:
- [X] My code follows the style guidelines of this project
- [X] I have performed a self-review of my code
- [X] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [ ] My changes generate no new warnings
- [X] I have added tests that prove my fix is effective or that my
feature works
- [ ] New and existing unit tests pass locally with my changes
- [ ] Any dependent changes have been merged and published in downstream
modules
---------
Co-authored-by: Pranav <pranav@chatwoot.com>
Co-authored-by: Sojan Jose <sojan@pepalo.com>
Co-authored-by: Sony Mathew <sony@chatwoot.com>
Co-authored-by: Sony Mathew <2040199+sony-mathew@users.noreply.github.com>
Adds label support to contact import and export so teams can carry
approved contact labels through CSV workflows. Imports accept a `labels`
column with labels that already exist in the account; multiple labels
should be entered as a quoted comma-separated CSV value, for example
`"customer,vip"`.
Imports are additive: they add labels to contacts and do not remove
labels already on a contact. Removing a label from the CSV row or
leaving the `labels` cell blank will not clear existing contact labels.
To remove a label, edit the contact directly.
## Closes
- Closes#8535
## How to test
1. Create a few contact labels in the account, such as `customer`,
`vip`, and `lead`.
2. Go to Contacts -> Import contacts and download the sample CSV.
3. Import contacts with a `labels` column. Use a single label like
`lead`, or quote multiple labels like `"customer,vip"`.
4. Confirm imported contacts are created with the expected labels.
5. Re-import an existing contact with a new label and confirm the new
label is added without removing existing labels.
6. Try a row with an unknown label, such as `"vip,unknown_label"`, and
confirm only that row is rejected in the failed records CSV while the
other valid rows are imported.
7. Export contacts and confirm the CSV includes a `labels` column with
comma-separated approved labels.
## What changed
- Contact exports include approved `labels` in the default CSV columns.
This adds a new default export column for CSV consumers.
- Contact imports parse `labels` as comma-separated values inside the
CSV cell.
- Imported labels are validated against labels that already exist in the
account.
- Rows with unknown labels are rejected with an `Unknown labels: ...`
error; valid rows in the same import continue to process.
- Imported labels are additive and do not remove existing contact
labels.
- Label application during import does not dispatch an additional
per-contact update event.
- The sample CSV includes an import-safe `labels` column. The modal
keeps the existing generic CSV import copy.
---------
Co-authored-by: Sojan Jose <sojan@pepalo.com>
# Pull Request Template
## Description
skip documents that fail with ActiveRecord errors possibly due to
stale/corrupt data and not crash scheduler
How did we find out about this error?
before October 28th, 2025, we did not have url normalisation.
so we had document rows as:
id: 123 `https://example.com` status: `in_progress` --> likely stuck
crawl
id 234: `https://example.com/` status: `available`
When the schedule sync job ran, it ran an `document.update!(sync_status:
:syncing, last_sync_attempted_at: Time.current)` on the 234 one since it
was `available`
now `update!` runs `before_validation :normalize_external_link`
so `https://example.com/` became `https://example.com`
which invalidated:
`validates :external_link, uniqueness: { scope: :assistant_id }`
so the scheduler crashed.
This PR logs the skipped ones with their errors and continues to pick
other documents to scheduler doesn't crash
## Type of change
- [x] Bug fix (non-breaking change which fixes an issue)
## How Has This Been Tested?
Please describe the tests that you ran to verify your changes. Provide
instructions so we can reproduce. Please also list any relevant details
for your test configuration.
spec
## Checklist:
- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [x] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [x] Any dependent changes have been merged and published in downstream
modules
A retried terminate webhook for an answered short call (duration=0)
would re-enter handle_terminate, find call.in_progress? false, fall
through to no_answer, and corrupt a completed call's status. Bail when
the call is already terminal so retries don't recompute the status.
Inbound WhatsApp calls were matching contacts by stripping `+` from the
raw from_number, bypassing the BR/AR wa_id normalization that
Whatsapp::IncomingMessageServiceHelpers#processed_waid runs on the
messaging path. A wa_id whose webhook variant differs from the stored
normalized source_id (notably Brazil's added "9") would miss the
existing ContactInbox and create a fresh contact/conversation,
fragmenting history. Route the voice path through the same
Whatsapp::PhoneNumberNormalizationService so the lookup is consistent.
Embed templates interpolate regex captures from user-authored article
URLs into HTML attribute values. CommonMark's angle-bracket link
destination syntax allows characters that the capture regexes don't
filter, so the unescaped substitution could produce malformed attribute
output. Escaping at substitution time keeps the render deterministic
regardless of the URL.
### How was this tested?
Added specs.
Fixes [CW-6934](https://linear.app/chatwoot/issue/CW-6934/)
Co-authored-by: Sony Mathew <sony@chatwoot.com>
Co-authored-by: Sivin Varghese <64252451+iamsivin@users.noreply.github.com>
## Description
* Added Meta webhook HMAC validation in meta_token_verify_concern.rb.
* Wired it into instagram_controller.rb and whatsapp_controller.rb.
* WhatsApp now verifies X-Hub-Signature-256 with WHATSAPP_APP_SECRET.
* Instagram now verifies with either FB_APP_SECRET or
INSTAGRAM_APP_SECRET.
* Updated request specs so missing/invalid signatures return 401 and
valid signatures still enqueue jobs.
Fixes # (issue):
[CW-6786](https://linear.app/chatwoot/issue/CW-6786/ghsa-7rw7-pc8v-mrr3-unauthenticated-message-injection-via-missing)
## Type of change
Please delete options that are not relevant.
- [x] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality not to work as expected)
- [ ] This change requires a documentation update
## How Has This Been Tested?
* Updated the controller specs and ran them successfully.
* The original issue is no longer reproducible.
## Checklist:
- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [ ] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [ ] Any dependent changes have been merged and published in downstream
modules
---------
Co-authored-by: Muhsin Keloth <muhsinkeramam@gmail.com>
# Pull Request Template
## Description
Captain (v1) makes false promises by saying it will handoff but doesn't.
This happens due to an exact string match comparison and the prompt
gives the model a lot of responsibilities:
- identity
- what to respond
- obey custom instructions
- decide on tool calls
This PR decouples responsibility, the core prompt responds, and an
additional llm call evaluates if handoff was needed or not after that
message.
## Type of change
- [x] Bug fix (non-breaking change which fixes an issue)
## How Has This Been Tested?
Please describe the tests that you ran to verify your changes. Provide
instructions so we can reproduce. Please also list any relevant details
for your test configuration.
Locally
## Checklist:
- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [x] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [x] Any dependent changes have been merged and published in downstream
modules
Two production-grade fixes to the existing audio transcription service.
**Independent of the WhatsApp Calling work** — these affect every audio
attachment that goes through Whisper (voice notes, call recordings,
voicemails, etc.).
## Closes
- [PLA-151 — PR-5: Recording Upload + Transcription
Pipeline](https://linear.app/chatwoot/issue/PLA-151/pr-5-recording-upload-transcription-pipeline)
## Why this is needed
### 1. Whisper rejects payloads larger than 25 MB
OpenAI's [Whisper
API](https://platform.openai.com/docs/guides/speech-to-text) hard-caps
file uploads at 25 MB. Long audio recordings — voice notes from chatty
contacts, ~70+ min Opus call recordings — currently hit OpenAI with the
full payload and 413 (\`Payload Too Large\`). The job retries via the
existing \`Faraday::BadRequestError\` discard path, but the agent still
sees a transcription failure for an attachment we knew was too big up
front.
This PR adds a pre-flight \`audio_too_large?\` check via the blob's
\`byte_size\` and returns a controlled error without hitting OpenAI. The
audio attachment is preserved (agents can still listen), only the
transcription is skipped.
### 2. Whisper hallucinates on silence at non-zero temperature
At \`temperature: 0.4\` (the previous value), Whisper produces
well-documented hallucinated repeats on silence and near-silent segments
— e.g. \`Oh, dear. Oh, dear. Oh, dear.\` filling the transcript. This
shows up in real recordings whenever there's a hold or quiet moment.
\`temperature: 0.0\` matches OpenAI's recommended default for
transcription and eliminates the spirals.
Reference:
[openai/whisper#928](https://github.com/openai/whisper/discussions/928),
[openai-python#1010](https://github.com/openai/openai-python/issues/1010).
## Are WhatsApp call recordings already handled?
Yes — by the existing pipeline, **before this PR**:
\`\`\`
Browser MediaRecorder → upload_recording (PR-4)
→ @call.message.attachments.create!(file_type: :audio, ...)
→ Enterprise::Concerns::Attachment#enqueue_audio_transcription
(after_create_commit hook)
→ Messages::AudioTranscriptionJob.perform_later(attachment.id)
→ Messages::AudioTranscriptionService → Whisper
\`\`\`
The \`after_create_commit\` hook already fires for every audio
attachment regardless of source. PR-4's \`upload_recording\` endpoint
creates the attachment; the existing job/service take it from there. No
new wiring needed.
This PR just makes the existing service more robust:
- Calls longer than ~70 min (Opus 48 kbps) no longer 413 against OpenAI
- Quiet recordings no longer produce hallucinated transcripts
## How to test
\`\`\`ruby
# In rails console with a real audio attachment:
service = Messages::AudioTranscriptionService.new(Attachment.audio.last)
# Normal-sized audio: unchanged behaviour
service.perform # => { success: true, transcriptions: ... }
# Large audio: new guard returns error instead of 413-ing OpenAI
allow(attachment.file.blob).to
receive(:byte_size).and_return(30.megabytes)
service.perform # => { error: 'Audio too large for Whisper' }
\`\`\`
Existing transcription specs cover the happy path; one new spec
exercises the byte-limit guard.
## Risk
Low. Both changes are pre-flight guards or parameter values — they
reduce the surface of OpenAI calls that can fail. Failure to transcribe
is already non-fatal (the audio attachment is preserved either way).
When an inbound voice call ends, the conversation bubble now (1) renders
an inline audio player as soon as Twilio finishes the recording and (2)
shows the call duration alongside "Call ended" so the agent gets the
at-a-glance summary without opening the recording.
Fixes
https://linear.app/chatwoot/issue/PLA-118/feat-recordings-on-calls-should-be-attached-on-the-conversation
and
https://linear.app/chatwoot/issue/PLA-119/duration-of-the-call-is-not-visible-on-the-chat-bubble
## How to test
1. Set up a Twilio voice inbox and trigger an inbound call.
2. Answer the call from an agent, talk for a few seconds, then hang up.
3. As soon as the call ends, the bubble should read **"Call ended —
0:NN"** (where NN is the call duration in seconds).
4. Wait a few seconds for Twilio to finish processing the recording
(usually <30s after hangup).
5. The same bubble should now show an inline audio player below the
duration. Press play; the recording should be audible.
6. Refresh the page — both the duration and the player should still be
there.
7. End a second call on the same conversation — its bubble should get
its own duration + player, independent of the first.
---------
Co-authored-by: Muhsin <12408980+muhsin-k@users.noreply.github.com>
This PR limits IMAP email fetching to 500 messages per sync run to avoid
expensive/long-running mailbox scans. It also filters out
already-imported emails and Chatwoot-generated notification emails
during the header fetch phase, before fetching full email bodies,
reducing unnecessary IMAP work.
Fixes #CW-7001 (issue) :
https://linear.app/chatwoot/issue/CW-7001/emails-not-syncing
Adds the Meta WhatsApp Cloud API surface needed for browser-based
calling. This is the second slice of the WhatsApp calling feature,
sitting on top of `feat/voice-call-model-wiring` and consumed by later
PRs (incoming-webhook pipeline, call service, frontend).
This PR ships only the provider-level HTTP wrapper and one error class.
It is feature-flag-free and does not change any user-visible behaviour
on its own — without later PRs, no caller invokes these methods.
## Linear
-
https://linear.app/chatwoot/issue/PLA-148/pr-2-meta-cloud-api-provider-methods
## What changed
- Add `Whatsapp::Providers::WhatsappCloudCallMethods`
(`enterprise/app/services/whatsapp/providers/whatsapp_cloud_call_methods.rb`)
wrapping six Meta endpoints:
- `pre_accept_call`, `accept_call`, `reject_call`, `terminate_call` —
`POST /{phone_id}/calls` with the relevant action payload.
- `send_call_permission_request` — `POST /{phone_id}/messages`
interactive `call_permission_request`.
- `initiate_call` — `POST /{phone_id}/calls` with `audio`/`offer`
session.
- Prepend the module into `Whatsapp::Providers::WhatsappCloudService`
only if defined, so OSS continues to work without the enterprise
overlay.
- Add `Voice::CallErrors::NoCallPermission`
(`enterprise/lib/voice/call_errors.rb`) — raised when Meta returns error
code `138006` from `initiate_call`. The remaining call-service errors
(`NotRinging`, `AlreadyAccepted`, `CallFailed`) will land with PR-4.
## How to test
There is no UI in this PR. Smoke-test from a Rails console with a
WhatsApp inbox configured for calling:
```ruby
inbox = Inbox.find(<id>)
svc = inbox.channel.provider_service
svc.respond_to?(:initiate_call) # => true
svc.respond_to?(:send_call_permission_request) # => true
# Optional live calls (require a real phone + Meta call-permission opt-in):
svc.send_call_permission_request('15551234567')
svc.initiate_call('15551234567', '<sdp_offer>')
```
Failure path: `initiate_call` against a contact who has not granted call
permission should raise `Voice::CallErrors::NoCallPermission` with
Meta's user-facing message.
Twilio voice now uses first-class `Call` records as the source of truth
for call state, instead of storing it on
`conversation.additional_attributes` and `conversation.identifier`. Each
call gets its own record, its own `voice_call` bubble matched by
`call_sid`, and its own conference name keyed off `Call.id`. Multiple
calls on the same conversation (for `lock_to_single_conversation`
inboxes) now work correctly, and the conversation card stays in sync
with the real latest message.
Fixes https://linear.app/chatwoot/issue/PLA-121/lock-to-single-thread
---------
Co-authored-by: Muhsin <12408980+muhsin-k@users.noreply.github.com>
When an agent mentions themselves in a private note, they no longer
receive a redundant notification for their own mention.
Closes: #4096
# Pull Request Template
## Description
Agents who mention themselves in a private note no longer receive a
conversation_mention notification. Previously, the mention service would
generate a notification for every mentioned user without checking
whether the sender and the
mentioned user were the same person.
This PR fixes an issue where Slack emojis are rendered as text
shortcodes (e.g. 🚀) instead of the actual emoji characters in
Chatwoot messages.
It introduces a new EmojiFormatter class that uses the emoji-data
mapping to convert shortcodes to unicode characters.
---------
Co-authored-by: Sony Mathew <sony@chatwoot.com>
Co-authored-by: Sony Mathew <2040199+sony-mathew@users.noreply.github.com>
# Pull Request Template
## Description
- Wires up Controllers to auto-sync job
- adds plan based sync schedule
- a scheduler that runs every hour to check syncable documents
- guards the whole feature behind feature flag by reclaiming
`twilio_content_templates`
- Adds a global and account level cap on how many documents to enqueue
to prevent sudden burst at first run
- some refactor to simplify code
- specs
Fixes # (issue)
## Type of change
- [x] New feature (non-breaking change which adds functionality)
## How Has This Been Tested?
Please describe the tests that you ran to verify your changes. Provide
instructions so we can reproduce. Please also list any relevant details
for your test configuration.
specs and locally
## Checklist:
- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [x] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [x] Any dependent changes have been merged and published in downstream
modules
Voice calling is now a capability on the existing TwilioSms rather than
a separate Voice model. A single Twilio phone number handles both SMS
and voice calls through one inbox.
Fixes
https://linear.app/chatwoot/issue/CW-6683/add-voice-calling-as-a-capability-on-twilio-sms-channel
and https://linear.app/chatwoot/issue/PLA-120/add-the-support-for-sms
**What changed**
- Replaced Channel::Voice with voice_enabled flag on Channel::TwilioSms
- Added voice_enabled, twiml_app_sid, api_key_secret columns to
channel_twilio_sms table
- Dropped channel_voice table (no production data)
- All voice logic lives in Enterprise layer via
prepend_mod_with('Channel::TwilioSms')
- Added Voice settings tab on Twilio SMS inbox settings to
enable/disable voice
- Validates Twilio number voice capability before provisioning
- Teardown service cleans up TwiML app and credentials when voice is
disabled
- Frontend voice detection uses isVoiceCallEnabled() /
getVoiceCallProvider() helpers — extensible to future providers
- Gated by channel_voice feature flag
**How to test**
1. Enable feature flag:
Account.find(<id>).enable_features('channel_voice')
2. Create voice inbox: Inboxes → Voice tile → enter Twilio credentials →
verify incoming/outgoing calls and SMS work
3. Enable voice on existing SMS inbox: Inboxes → select Twilio SMS inbox
→ Voice tab → toggle on → provide API key credentials → verify calls
work
4. Disable voice: Voice tab → toggle off → verify TwiML app is deleted,
credentials cleared, SMS still works
5. Re-enable voice: Toggle on again → must provide api_key_secret again
→ new TwiML app provisioned
---------
Co-authored-by: Muhsin <12408980+muhsin-k@users.noreply.github.com>
## Description
Fix a Postgres planner trap on the "Pending Response: Longest first"
sort that causes the conversation list to hang on busy accounts.
The current `sort_on_waiting_since` generates query with `ORDER BY
waiting_since ASC NULLS LAST, created_at ASC`. That order-by is exactly
the shape of the single-column `index_conversations_on_waiting_since`
btree, so the planner picks a forward index walk thinking `LIMIT 25`
will stop early. In practice the per-account matches are spread along
the global waiting_since timeline, so the scan reads tens of millions of
rows from other accounts and discards them via the filter before
producing any results which in turn causes the requests to time out and
the conversation list spinner never resolves.
DESC direction and every other sort (`priority`, `created_at`,
`last_activity_at`) are unaffected. They fall through to
`conv_acid_inbid_stat_asgnid_idx` (account-scoped composite), which is
the right index for this access pattern.
This change leads the ORDER BY with the expression `(waiting_since IS
NULL)`, which no column-only btree can satisfy. The planner falls back
to the same account-scoped index used by every other sort, and sorts in
memory. Same logical NULLS LAST output for both directions; no behavior
change for users.
Fixes CW-6965
## Type of change
- [x] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality not to work as expected)
- [ ] This change requires a documentation update
## How Has This Been Tested?
- [x] Existing specs pass
- [x] Added new specs to cover NULL case
- [x] Verify results and order for old and new query in prod
- [x] Tested in prod since staging data was not sufficient
`EXPLAIN (ANALYZE, BUFFERS)` for the same query (status=0, ASC, LIMIT
25) on a representative production account, before vs after:
| Metric | Before | After |
| --- | --- | --- |
| Execution time | 34,679 ms | 0.71 ms |
| Rows discarded by filter | 36,906,962 | 0 |
| Shared buffer hits | 12,699,924 | 11 |
| Blocks read from disk | 851,226 | 112 |
| I/O read time | 17,785 ms | 0.3 ms |
| Pages dirtied / written | 98 / 31,043 | 0 / 0 |
Verified on two production accounts: identical row IDs in identical
order between the old and new ORDER BY for both ASC and DESC. A
NULL-bucket regression spec was added covering ASC/DESC tail ordering
when some conversations have a null `waiting_since`.
Roughly `49,000×` faster on this query (34,679 ms → 0.71 ms), and
trivially less I/O and buffer pressure on the cluster while it runs.
## Checklist:
- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [ ] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [ ] Any dependent changes have been merged and published in downstream
modules
Co-authored-by: Tanmay Deep Sharma <32020192+tds-1@users.noreply.github.com>
When a customer responds to a bot's interactive prompt (input_select,
input_csat, form, input_email) from the widget, the response shows up in
the Chatwoot agent UI but is not reflected in the linked Slack channel —
Slack only ever shows the original question. This happens because the
widget submits the answer as an UPDATE to the original message (writing
`content_attributes.submitted_values` or `submitted_email`), but the
Slack hook only listened to `message.created`, so updates were ignored.
Closes https://linear.app/chatwoot/issue/PLA-147
### Preview
<img width="1290" height="1106" alt="CleanShot 2026-04-21 at 13 19
19@2x"
src="https://github.com/user-attachments/assets/cd2a9d3f-89d3-4e81-9230-5b078e1b7b44"
/>
### How to test
1. Connect a web widget inbox to a Slack channel.
2. Trigger each bot message type (input_select, form, input_csat,
input_email) in a conversation.
3. Submit responses from the widget.
4. Verify each response now appears in the Slack thread, appended to the
original bot question.
---------
Co-authored-by: Muhsin <12408980+muhsin-k@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
## Linear Ticket
-
https://linear.app/chatwoot/issue/CW-6883/allow-disabling-2fa-using-a-backup-code
## Description
When a user loses access to their authenticator app, they can now
disable 2FA using one of their saved backup codes (in addition to their
password), so they can re-enroll a new authenticator. The disable dialog
includes a toggle to switch between entering a verification code and a
backup code.
## Type of change
- [x] Bug fix (non-breaking change which fixes an issue)
## How Has This Been Tested?
- Via UI flows
<img width="495" height="423" alt="Screenshot 2026-04-20 at 2 17 21 PM"
src="https://github.com/user-attachments/assets/cc6b3dc5-39e6-4104-b5b9-cdabdc46947e"
/>
<img width="475" height="409" alt="Screenshot 2026-04-20 at 2 17 36 PM"
src="https://github.com/user-attachments/assets/97c7304d-4adb-42ed-b7b4-50f5b38585a3"
/>
## Checklist:
- [ ] My code follows the style guidelines of this project
- [ ] I have performed a self-review of my code
- [ ] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [ ] My changes generate no new warnings
- [ ] I have added tests that prove my fix is effective or that my
feature works
- [ ] New and existing unit tests pass locally with my changes
- [ ] Any dependent changes have been merged and published in downstream
modules
Updating portal settings (name, header text, page title, homepage link)
on a portal that already has a logo attached returns 500. The error is
\`NoMethodError: undefined method 'valid_encoding?' for an instance of
Integer\`. The fix is a two-character change in
\`process_attached_logo\`.
Closes#13300
## Root cause
\`ActiveStorage::Blob.find_signed\` expects a signed ID string (e.g.
\`"eyJfcmFpbH..."\`). Internally it calls \`valid_encoding?\` on the
argument to validate the signature payload — a method that exists on
\`String\` but not \`Integer\`.
When a portal already has a logo, the frontend includes the blob's raw
database integer ID (e.g. \`blob_id: 170\`) in the update request
payload. The controller passes this integer directly to \`find_signed\`,
which immediately raises \`NoMethodError\` before any database query is
made.
\`\`\`ruby
# before, crashes when blob_id is an Integer
blob_id = params[:blob_id]
blob = ActiveStorage::Blob.find_signed(blob_id) # NoMethodError here
@portal.logo.attach(blob)
\`\`\`
## What changed
\`\`\`ruby
# after, safe for any input type
blob = ActiveStorage::Blob.find_signed(params[:blob_id].to_s)
@portal.logo.attach(blob) if blob
\`\`\`
\`.to_s\` on an Integer produces a plain decimal string (\`"170"\`),
which is not a valid signed ID. \`find_signed\` returns \`nil\` for any
invalid signature rather than raising, so the nil guard prevents a
broken \`attach\` call. The existing logo remains attached and the
settings update succeeds.
## Trade-offs considered
| Option | Decision |
|---|---|
| \`find(blob_id)\` when input is an Integer | Bypasses signature
verification — any authenticated user knowing a blob ID could attach
arbitrary files to a portal. Security risk. Rejected. |
| Raise a 422 for non-string blob_id | Overly strict — the frontend
sending an integer is pre-existing behaviour this PR shouldn't break. |
| Silently no-op for invalid blob_id (chosen) | Correct product
behaviour: if no valid signed upload is provided, leave the logo
unchanged. The settings update still succeeds. |
## Known limitation
The correct long-term fix is also on the frontend: it should only send
\`blob_id\` when attaching a **new** upload (using the signed ID from
the direct-upload flow), not when re-submitting the existing logo's raw
database integer ID. This PR makes the server robust against the current
frontend behaviour without requiring a coordinated frontend change.
## How to reproduce
1. Create a Help Center portal and upload a logo
2. Update any text field via \`PUT /api/v1/accounts/:id/portals/:slug\`
while including \`blob_id: <integer>\` in the payload
3. Observe 500 with \`NoMethodError: undefined method 'valid_encoding?'
for an instance of Integer\`
After this fix, the request returns 200, settings are updated, and the
existing logo is preserved.
Co-authored-by: Ramalau Debeila <rdebeila@datacentrix.co.za>
This routes external downloads used by webhook fetch used by macros and
acutomations through SafeFetch. It closes the SSRF exposure from raw
Down.download paths, preserves provider-specific auth and header flows,
and adds regression coverage for blocked internal URLs plus
authenticated downloads.
Fixes # (issue):
[CW-6940](https://linear.app/chatwoot/issue/CW-6940/ssrf-via-webhooksautomationmacros-non-upload-non-avatar)
\`GET /platform/api/v1/agent_bots\` returns 500 when any \`AgentBot\`
that was previously registered with a Platform App has since been
deleted. The bug was introduced by a missing \`dependent: :destroy\` on
the \`AgentBot\` model — deleting a bot left orphaned rows in
\`platform_app_permissibles\`, which the index action later iterated
over and crashed rendering with a \`NoMethodError\` on \`nil\`.
Closes#13407
## Root cause
The index action loads all \`platform_app_permissibles\` for the
platform app and passes each \`resource.permissible\` (the associated
\`AgentBot\`) to a Jbuilder partial. When the \`AgentBot\` no longer
exists, \`resource.permissible\` returns \`nil\` and the partial crashes
calling \`.id\`, \`.name\`, etc. on it.
Every other \`AgentBot\` association (\`agent_bot_inboxes\`,
\`messages\`, \`assigned_conversations\`) had a \`dependent:\` option —
\`platform_app_permissibles\` was the only one missing it. There was
also an N+1 query: the index fired a separate SQL query per permissible
to load each bot.
## What changed
**1. Model — prevent orphans at deletion time**
\`\`\`ruby
has_many :platform_app_permissibles, as: :permissible, dependent:
:destroy
\`\`\`
**2. Controller — eager-load to eliminate N+1**
\`\`\`ruby
@resources = @platform_app.platform_app_permissibles
.where(permissible_type: 'AgentBot')
.includes(:permissible)
\`\`\`
**3. Jbuilder — defensive nil guard for pre-existing orphans**
\`\`\`ruby
bot = resource.permissible
next if bot.nil?
json.partial! '...', resource: bot
\`\`\`
## Trade-offs considered
| Option | Decision |
|---|---|
| Rescue \`NoMethodError\` in jbuilder | Hides the failure rather than
fixing it. Rejected. |
| Only add the nil guard, skip the model fix | Leaves the data integrity
gap open — future deletions continue creating orphans. Rejected. |
| Both layers (chosen) | Model fix prevents new orphans; nil guard is
defence-in-depth for any orphans that survived before deployment. |
| \`dependent: :nullify\` | Doesn't apply — a nullified permissible
would still cause the same nil dereference. Rejected. |
## How to reproduce
1. Create an AgentBot via the Platform API
2. Delete the AgentBot via any path (admin UI, API, or direct model
call)
3. Call \`GET /platform/api/v1/agent_bots\` with a Platform App token
4. Observe 500
After this fix, the endpoint returns 200 with an empty array.
Co-authored-by: Ramalau Debeila <rdebeila@datacentrix.co.za>
Standardizes the contact company import/filter/automation contract on
`company_name`.
Closes#14096
Revives #9907
## Why
Contact company is read across the current CRM/contact UI from
`additional_attributes['company_name']`, but CSV import and a few
backend filter/automation paths still used the older `company` key. That
meant imported company values could be saved in a place the dashboard,
sorting, filters, and automation conditions did not consistently read
from.
Based on the production data check, the legacy `company` automation
configuration is effectively dead: the affected account did not have
contacts populated with `additional_attributes['company']`. So this PR
intentionally avoids adding long-term fallback behavior and uses
`company_name` as the single key going forward.
## What changed
- Contact CSV import now writes only `company_name` into
`additional_attributes['company_name']`.
- The example contact import CSV now uses the `company_name` header.
- Contact company sorting/filter config now uses `company_name`.
- Automation condition config now uses `company_name`.
- Existing standard automation conditions with `attribute_key:
'company'` are migrated to `company_name`.
- Existing saved contact filters with standard `attribute_key:
'company'` are migrated to `company_name`.
- Custom attributes named `company` are preserved and are not rewritten
by the migration.
## How to test
- Import a contact CSV with a `company_name` column and confirm the
Contact Company field is populated.
- Sort contacts by Company and confirm imported contacts are ordered
correctly.
- Create/edit an automation with Company as a condition and confirm it
saves with `company_name`.
- Verify existing saved contact filters and automation rules using the
old standard `company` key are migrated to `company_name`.
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Sojan Jose <sojan@pepalo.com>
Tanmay Deep Sharma