This PR improves error handling in `handleDriverException` by adding a
duck-typing check for `MessageImportDriverException` objects. It ensures
that errors are correctly identified and processed even if they are
received as plain objects rather than class instances. This change
prevents missed exception handling and increases the robustness of the
message import process.
This PR fixes this issue from the connected account refresh token
service that is
This PR fixes error handling in `handleDriverException` by ensuring that
errors resembling `MessageImportDriverException` are correctly detected,
even if they are plain objects and not true class instances. This
prevents missed exception handling due to failed `instanceof` checks.
Was introduced by [this
PR](https://github.com/twentyhq/twenty/pull/12233) that did not know all
provider cases that can occur.
Fixes https://github.com/twentyhq/twenty/issues/12589
In order to put back relations on track, we need to allow the
preexisting indexes to be created in indexmetadata, but also to avoid
failure when the migration runner will try to re-apply a pre-existing
index
Why : we had an issue impoting events du to CalendarEvents not yet
existing while inserting the CalendarChannelAssociation due to inverted
method in the service
This PR refactors the calendar event import logic by
- renaming
- splitting utility functions for better clarity and maintainability.
- adding TSDoc comments to explain the purpose and uniqueness of the
`eventExternalId` field in calendar event associations
Fixes#12690
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
## Why
After the changes on relations, index on relations were skipped by the
syncmetadata service, so no more migrations were generated for relation
fields.
We wanted to fix this.
## Test
This PR adds unit tests for the `createIndexMigration` utility in the
workspace migration builder. The tests cover:
- Creating index migrations for simple fields (e.g., text fields)
- Creating index migrations for relation fields (ensuring correct column
naming, e.g., `authorId` for the `author` objectmetadataname)
## Excluded
The delete index on relation does not need the column names so i don't
think i needed to work on this method. I might be wrong.
## Checklist
- [x] Added/updated unit tests for index migration creation
- [x] Verified correct handling of simple and relation fields
- [x] Ensured all tests pass
---------
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Co-authored-by: Charles Bochet <charles@twenty.com>
Fix inconsistent domain URL formats : removing the last / that was
caused by URL method
Standardize URL formatting to ensure consistent links storage and
retrieval of domain URLs across the application. Will improve the
dedpulicates in the links
Note: there is another temporary issue from google that was solved on
the 13th of june https://groups.google.com/g/adwords-api/c/tRSQMRZrJYM
but we consider this out of this scope
Fixes#12621
it is an unusual 2.75 here.
I tried to set it to 3 however it requires a big re-work on all cells of
the table. The main concern being the new border left of 1px for the
"focus mode"
Note: to be fair, we are not 100% aligned with the Figma on the paddings
and alignment contruction. It is not related to this issue, but makes
this fix not ideal since not aligned with our CSS standards.
Fixes https://github.com/twentyhq/twenty/issues/12607
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
### Issue:
overflow of input fields when opening the side panel.
Could be imporved with a better command menu state handling
### Solution:
In command menu hooks, we now close all dropdowns when opening the side
panel. This ensures all UI elements are close properly before the
sidepanel shows up
Fixes : https://github.com/twentyhq/twenty/issues/12485
Co-authored-by: Charles Bochet <charles@twenty.com>
Update the default set of system fields for custom objects, to ensure
position is not nullabel and has a default value to 0
Steps to reproduce :
create a custom object,
send a POST request with body ```{position:null}```
the record should be created
After the change,
an error will be thrown
<img width="754" alt="Screenshot 2025-06-13 at 17 16 56"
src="https://github.com/user-attachments/assets/d40931f7-16cc-4b68-8dbb-deb0fa292be5"
/>
After release 55, we found out that CRON job monitor was red for
CronTriggerCronJob
While only 1 workspace was not in the appropriate state, meaning the
whole command was probably failing for only 1 workspace failing.
We suggest here to catch errors per worksspace and simply push to sentry
the error of the errored workspace relative to workflow trigger.
# Integration test failing
- fix the local run by renaming folder in the jest configuration.
Otherwise "clickhouse" tests were failing
- falsy test introduced 2 days ago in
https://github.com/twentyhq/twenty/pull/12271/files
# Summary
Enhanced the Google OAuth flow to better handle missing permissions and
improved user experience by redirecting to settings/account page.
## Changes
- Added new google-apis-scopes.ts service for better scope management
- Updated Google APIs auth controller for better flow control
- New tests for this logic
## User request
From @bonapara email test and need to better handle user flow during the
connect email flow
Before :
<img width="574" alt="Screenshot 2025-05-28 at 17 58 59"
src="https://github.com/user-attachments/assets/fd54625b-e211-4b2f-b76a-48bcb08b5222"
/>
After :
<img width="1143" alt="Screenshot 2025-05-28 at 16 29 05"
src="https://github.com/user-attachments/assets/8f3d1f2c-9e02-4d25-b949-fe2b20f048f4"
/>
## Reference :
For google specialities, I added this link in the `export const
getGoogleApisOauthScopes` in order to keep that in mind
https://developers.google.com/identity/protocols/oauth2/scopes
# extracting domain emails
Added new test cases covering weird but valid email formats (plus
addressing, subdomains, international domains, etc.) to identify
potential failures in the current implementation.
Two tests with quoted local parts containing @ symbols or quotes are
marked as skipped since they're expected to fail with the current simple
string splitting approach. They are too exotic IMO, we should throw
errors.
## Next
We will monitor errors related to this and update accordingly later on.
### Note
technically, quotes are possible in RFC see
[here](https://stackoverflow.com/questions/4816424/are-single-quotes-legal-in-the-name-part-of-an-email-address)
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
# Gmail OAuth authentication flow issues
### TLDR
This error is not an error and therefore should be treated as a simple
redirect with a snackbar.
### More details
Fixing incomplete OAuth token exchange processes and improving error
handling for empty Gmail inboxes.
The changes include modifications to OAuth guards, to ensure that if a
user clicks "cancel" instead of completing the authentication workflow
if fails
## Before:
Redirection from `/settings/accounts` to `app.twenty.com` with an
`UNAUTHORIZED` error
## After :
<img width="948" alt="Screenshot 2025-05-26 at 18 04 37"
src="https://github.com/user-attachments/assets/62c8721e-c2b3-4e3d-ad0b-e4059dfb7a98"
/>
Fixes https://github.com/twentyhq/twenty/issues/11895
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
Catching "no licence - removed" microsoft message channels.
Current behabiour
> ` MessageImportException [Error]: The mailbox is either inactive,
soft-deleted, or is hosted on-premise.`
Goal:
better track errors VS user mistakes
Context:
A similar logic was already implemented for the calendar channels. I
just replicated it to message channels
# Handling Google error
This bug is hard to reproduce so the resolution is based only on the
Sentry logs. There is not definite error code
for this error message. in the documentation.
https://developers.google.com/workspace/gmail/api/guides
This is why i added more 5xx code types in order to be sure we catch
this. In order to refine this later, i added the error code to the
message.
Fixes https://github.com/twentyhq/twenty/issues/12025
## Description
When a calendar channel fails, its status is not reset during the
reconnect step.
In some cases, resetting is necessary—especially when we’ve deployed a
fix and users try to reconnect their accounts after the patch.
## Why reseting to initial state
Our processes are idempotent, so we can safely set the status to null to
restart the flow. This approach covers all cases and avoids edge
conditions caused by inconsistent statuses.
Fixes : https://github.com/twentyhq/twenty/issues/12026
# Display "Soft-Deleted Workspace Members" in Actor Field Display
Reminder of the issue :
<img width="154" alt="Screenshot 2025-05-07 at 12 11 59"
src="https://github.com/user-attachments/assets/168f8743-2684-4d9a-b1a4-e86bb335f7a4"
/>
- `ActorFieldDisplay` component : display soft-deleted members
- `UserService` includes soft-deleted records when fetching workspace
members. This is the tricky part : do we want that for all workspace
members or maybe i could create another property dedicated to workspace
members and softdeleted ones. To be discussed
Result looks like this (we loose the source and the context in this
impleentation)
<img width="114" alt="Screenshot 2025-05-07 at 12 05 28"
src="https://github.com/user-attachments/assets/3cdddd91-454f-4e96-8d6d-6fe671658945"
/>
Fixes https://github.com/twentyhq/twenty/issues/11870
Another way we could also get into :
We could also, when a workspace user is softDeleted, change the current
implementation : we could avoid to delete the ActorMetadata like
CreatedByName (and context and source) in the "Person" table.
It would look more like this
<img width="111" alt="Screenshot 2025-05-07 at 12 06 16"
src="https://github.com/user-attachments/assets/daa4ece2-200a-41f0-ba24-177375c72983"
/>
However, this implementation is requires more work, and IMO harder to
maintain since is decouples completely the record from the workspace
member. This could be an issue in case we want tohard delete a user, or
decide another logic to display the Actor name.
Since the usecase should be pretty rare, I chose the first one but
willing to discuss it
---------
Co-authored-by: prastoin <paul@twenty.com>
# TLDR
fix bug due to some event properties coming from the google calendar API
containing weird characters like this
"\u0000�4\u000b\u00042��K\u0001�z,\u001cm�",
it made the Postgres select and insert operator fail
## Details
We can have event properties (like cal UID below) encoded in a strange
way. From my research, the character \u0000 comes from `C` language to
signal end of line. It is wrongly interpreted by Postgres so must be
escaped. I decided to remove all possibility of failure with this regex
`[^\x20-\x7E]` basically "any character that is not a printable ASCII
character"
```
[
"5foijj28qb8smqiafjablo17vd@google.com",
"\u0011�\"�f�\\\u0019G_=��\u0005]x",
"?}|��\f}l��+�弴�",
"%���?t\u0007��n\u001e\u000eY�T<",
".\u0011�\u0016�!�\u000eIǹ� ��\u001f",
"!h\u0004��DŽ�6���h�]E",
"(�CX]�Q�7�^��n\u0006�",
"_040000008200E00074C5B7101A82E0080000000070105B958DEFD801000000000000000010000000EA30DB22E888B943A8EE0AD483F8DB35",
"\t�N�#�D��Ic�h",
"+�)�H���jJ|Ժ�'�",
"_040000008200E00074C5B7101A82E0080000000070A54736C6EED8010000000000000000100000006502334AFE61904595C2831FA4391034",
"_040000008200E00074C5B7101A82E0080000000072C80C3590EFD8010000000000000000100000001BA9FD5B330C1A4D85462AC9D70B9D9D",
"sg�fvUa:St>-<�d\u0006",
"\u0017ڦ��_\u001e\u001fGm-1����",
"_040000008200E00074C5B7101A82E00800000000F0F4F01F4EF0D8010000000000000000100000004C11CE0950C85549B79C456C13987AB8",
"$�����\u0007V\u0007��\u001e�OLN",
"_040000008200E00074C5B7101A82E00800000000341DA81151EDD8010000000000000000100000007453CEFB19AA2D4899B17F0BDB000493",
"_01C756CA-98DC-4799-9F06-883A540A065C",
"_040000008200E00074C5B7101A82E00800000000919548BBF1EDD80100000000000000001000000050AE1E41F3CD314CA9215F193EBE1D39",
"_040000008200E00074C5B7101A82E0080000000039CF64D718EDD801000000000000000010000000B3824EF0711436488CB5459BE83733B1",
"_040000008200E00074C5B7101A82E00800000000C50BDADCB5E4D8010000000000000000100000005B95FE762B2EF84B9C5AC53907B7E5E8",
"�Spx�\u0003ve��ss�X��",
"\b���>\u0013̈�ыh��0�",
"_040000008200E00074C5B7101A82E008000000005BCD492230E8D801000000000000000010000000B243A9CE99E94C4DAD201129A8F2A2F7",
"_040000008200E00074C5B7101A82E008000000009B540B82D1DCD801000000000000000010000000B4AFC9825D94994AA6C528C953BD3D96",
"_040000008200E00074C5B7101A82E008000000000D39EABDB7E4D801000000000000000010000000CD07BA2D05E61B47A597EE538ECE8CA5",
"\u0016���\u0003inv�=O����",
"4?b�-���\u001c\u0013ת�E�p",
">\u0000\u0015-;_�^�W&�p\u001f�",
"_040000008200E00074C5B7101A82E008000000006258818C85D9D80100000000000000001000000003346C625768FE43AF3F8D09917CA3C9",
"+K��ٔ�\u0006�\u0018G\u000b\u0000�s\u000e",
"/\f�\rj�IOD�g脅��",
"_68d820f9-e2c1-4d9a-bf61-83e4957c8261",
"xĠ�W4>���t�\u001d���",
"_040000008200E00074C5B7101A82E008000000003F1F314328E3D8010000000000000000100000002C8DC60F5369F44DA2B066441F882B35",
"_040000008200E00074C5B7101A82E008000000006DFCC204FDDED80100000000000000001000000046ECD444737D2E4992C349B2EA05637F",
"_1FF7C1BB-4E1A-485F-ADD0-5C4768601179",
"_040000008200E00074C5B7101A82E00800000000A0AA1223EFD3D801000000000000000010000000C83B72C2B424944199F3353D1442B27E",
"_040000008200E00074C5B7101A82E0080000000063D101EB06E4D801000000000000000010000000E489F61C89B8914BA6F8C8A2606405FE",
"\u0011��f�\"�b���rB�[�",
";t��\u001d\u001euDY+T\u001d��v",
]
```
Fixes https://github.com/twentyhq/core-team-issues/issues/946
Fixes
https://twenty-v7.sentry.io/issues/6568530279/?environment=prod&project=4507072499810304&query=is%3Aunresolved%20issue.priority%3A%5Bhigh%2C%20medium%5D%20b09d7a84-e6bc-45cf-b3ca-1e6047dddeed&referrer=issue-stream&stream_index=0
### Edit :
Changed the regex to match the chars to remove from
`[^\x20-\x7E]` to `replace('\u0000', '');`
After reading the blocknote documentation :
- we decided to increase to 100% the lines width
- we decided to reduce as much as possible inner padding
- we decided it's on the parent to decide the padding of the richtext
The two last points are recommended in a discussion on the project
blocknote. This way clicking on padding won't trigger weird behaviour on
Chrome.
Fixes
https://github.com/twentyhq/core-team-issues/issues/827#issuecomment-2842350359
parseBatchResponse:
we need more logs to understand why we have some empty repsonse in the
body.
For anothe PR:
~~UNKNOWN_NETWORK_ERROR : moving from handleUnknownException to
handleTemporaryException. This is important since in the logs I saw "The
service is currently unavailable" which makes me think we should handle
this type of error with as temporary~~
Issue : When I create a task in the "Assigned to me" task view, it will
disappear from the view because the Assignee field isn't automatically
populated.
Solution :
We created a "buildRecordInputFromFilters" funciron that will convert
filtered into their corresponding values for the input.
Fixes https://github.com/twentyhq/core-team-issues/issues/708
---------
Co-authored-by: etiennejouan <jouan.etienne@gmail.com>
Better error logging for messaging import exception handler.
Goal is to have better info on why Unknown errors are thrown and avoid
such messages `Unknown error occurred while importing messages for
message channel XXXXXXXX in workspace YYYYYYYYYY: Unknown error occurred
while importing messages for message channel XXXXXXXX...`
leave intact the `input` argument to avoid side effects on the parent
caller
---------
Co-authored-by: Charles Bochet <charlesBochet@users.noreply.github.com>
We found out that `RecordTableEmptyStateByGroupNoRecordAtAll` was used
only when `hasRecordGroups` was true in "RecordTableEmptyState"
However the only parent component is RecordTableEmpty and called
RecordTableEmptyState when `hasRecordGroups` was false
Fixes https://github.com/twentyhq/core-team-issues/issues/833
Meaning never called at all
Possiblity to reconnect you account on Failed Unkown errors.
This PR triggers a similar flow than the Failed errors for unsufficient
permission already existing in twenty.
Allows our users to force the synchro again even though they have a
strange error. It's been asked by some customers since we have had a
couple of issues in messaging lately
Fixes https://github.com/twentyhq/twenty/issues/11411
Seeing a couple of issues related to company creations in logs, I
suspect this to be the root cause
This should help a lot in all the support we have to do on email
synchronisation
# Unit test on the Messaging Module
Initially the issue was to create integration test for the messaging
module but after speaking with the core team, we decided to go for an
easier implementation: unit test only,
We decided to focus our test on three main components of the module :
- message list
- message import
- message save & create contact
Fixes https://github.com/twentyhq/core-team-issues/issues/56
Comparing the last date a datasource was used instead of a fixed TTL :
should fix workers issues "error: Error: Connection terminated"
FYI was done in pair prog with @Weiko
Swithcing plan overflows the modal soo Bonapara said to go back to
simple switch plan until we have a better version next week with the
stirpe pricing page
After investiagting the different options ([see related
issue](https://github.com/twentyhq/core-team-issues/issues/660#issuecomment-2766030972))
I decided to add a "Verify Component" and a to build a custom Layout for
this route.
Reason I cannot use the default one is to have all preloaded once the
user changes website and lands on the verify route.
Reason I did not modify the DefaultLayout to match our need is that is
would require many changes in order to avoid preloading states for our
specific usecase.
Fixes https://github.com/twentyhq/core-team-issues/issues/660
---------
Co-authored-by: Charles Bochet <charles@twenty.com>
Fixes : For phones I have to press "enter" to validate my changes but
for other fields it's saved automatically when I leave the cell
Bug related to onClickOutside on the MultiItemFieldMenuItem component
that shows phones (but also emails...)
Seen with @bonapara : we keep a consitent behaviour meaning
- saving input on click outside when primary item is being edited
- not saving input on click outside when other items are being edited
Fixes https://github.com/twentyhq/twenty/issues/11246
New options menu feature: table/kanban switching
This feature is the last one of the Optino Menu v2 update. It is
designed to enable the possibility to switch from table to kanban and
vice-versa.
Only the default tab is not editable for consitency in the UX of the
application as designed by our team
---------
Co-authored-by: Charles Bochet <charlesBochet@users.noreply.github.com>
Adding the possibility to change the view name and incon from the
Options menu dropdown
---------
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
workflow update to allow microsoft send email
- also handle the case were permissions are not enough
- update the redirection in case the user clicks on new account because
it's not anylonger as easy as simply google
fixes https://github.com/twentyhq/core-team-issues/issues/540
driver implementation for sending emails with microsoft
---------
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
This PR is supposed to solve an issue with the syncrhonisation of
messages, specifically with microsoft driver. Microsoft calls don't need
access_Token so refreshing toekns was not implemented.
However, microsoft rely on its client which calls its refresfh_token,
and I might have missed some underlying dependency from microsoft
impelemtation so I setup the access token process to refresh it
Needs a talk before to be merged
Fix : https://github.com/twentyhq/twenty/issues/10367
EDIT:
it was a problem with microsoft making refreshtoken expire (contrarily
to google) which needs to be handled.
Guillim