* adding events for success and failure for import and export from latex
* adding the operation property to capture the import/export keyword
GitOrigin-RevId: 2e5482b3c7517b402fc151966975ca8718729683
* [docstore] add useSecondary flag to projectHasRanges
The rev-check for unarchiving always consults with the primary.
Two extra changes:
- Add a projection argument to peekDoc in order to skip lines download
from projectHasRanges.
- Add one retry to peekDoc to reduce chances of surfacing a rev-check
violation.
* [web] resync_projects: use the secondaries for all reads
* [web] add default value for useSecondary
* [docstore] add default value for useSecondary
* [k8s] docstore: set MONGO_HAS_SECONDARIES=true
GitOrigin-RevId: f15ec4fdc1cabe74c1eab87bec85f28d6f7a587d
* Wait for parsing and syntax highlighting in autocomplete Cypress tests
* Wait for syntax highlighting in write-and-cite Cypress tests
GitOrigin-RevId: d48f10c864b0a170b4a02e95e3a989fdc4137dbb
* [web] Reject tracked changes notifications
feat: adding new tests
feat: adding rejected changes notifications
feat: adding tests for rejectchanges
feat: updating tests for rejecting notifications;
feat: adding in rejecting user, and improving subject and activity line
fix: moving to a params object instead of positionals for email building
feat: updating to use events triggered from applyUpdate in document-updater
feat: updating to send rejected author ids with rejected change notification instead of change ids
feat: moving rejected author notification determination to updateManager instead of RangesManager, which is used by other paths
feat: only map to author if changes were made
* fix: gate by user status not project status
* fix: unit tests post-rebase
---------
Co-authored-by: Kristina Hjertberg <kristina.hjertberg@overleaf.com>
GitOrigin-RevId: f992e1885c47d1a6cf776740769d6d4763f3cb7c
* [history-v1] add endpoint for downloading latest zip
* [web] address review feedback
* [web] tests: do not overwrite db.projects.overleaf, extend it
* [web] set includeReferer flag from downloading zip
GitOrigin-RevId: e63e549f004230086f82eccf03b43fd62bde6071
* [web] cleanup archived split-test assignments from user record on login
Co-authored-by: Anna Claire Fields <anna.fields@overleaf.com>
* [migrations] purge archived split tests from all users
Co-authored-by: Anna Claire Fields <anna.fields@overleaf.com>
* [web] add missing mock and update snapshot test
* [web] gracefully access db.users.splitTests
---------
Co-authored-by: Anna Claire Fields <anna.fields@overleaf.com>
GitOrigin-RevId: bd185074a402556d7b7c812208cf834dd52b27a5
* [web] Add Papers/ReadCube icon to plans page integrations
Closes#33493
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Delete old 200kB zotero logo
* Allow png use in logos
* Allow wrap
* [web] Share third-party integration icon list across plans and onboarding pages
Extract the icons array to services/web/app/src/util/third-party-icons.mjs
so the plans-2026 feature table and the try_premium onboarding page render
from a single source. The try_premium page now also includes Papers and
follows the plans page ordering.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* [web] Allow ciam try-premium logo row to wrap
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GitOrigin-RevId: f5a52418cbe01d9e343092b552183dffa3ae78bd
Plan cards inherited neutral-60 (=neutral-60) for description and
include-list text, which can fail WCAG contrast on the light card
background. Switch to content-secondary (=neutral-70) per Vee's
short-term recommendation; affects the free plan description, card
include items (e.g. "Basic AI allowance") and the in-card group
picker labels.
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GitOrigin-RevId: 42aff473a779b4b4f36b6c648d86097a79f820c8
* [web] add includeReferer flag to SplitTestHandler.getAssignment
* [web] tests: migrate User.getSplitTestAssignment to async/await
I don't want to fight with callbacks and optional arguments. Just move
it to async/await. New tests should use async/await, so there is no
point in making this work in callback-hell.
* [web] remove unused URL import
GitOrigin-RevId: 6251001e6ba7354f704fa663be8ef365ca0b9d23
* Set scale synchronously on pagesinit to prevent 1.333 DPI flash
PDF.js resets its internal scale to 1.0 when setDocument() is called,
causing pages to momentarily render at the default 96/72 DPI scale
(1.333) before the React restore effect can apply the correct value.
Setting currentScaleValue directly in the pagesinit handler eliminates
this one-frame wrong-scale flash.
* Override .page display to block to prevent horizontal jump on recompile
Overleaf's global .loading class sets display:inline-flex, which
collides with PDF.js's transient 'loading' class on .page elements.
When the loading class is applied, inline-flex breaks margin:auto
centering, causing the page to jump horizontally. Forcing display:block
at higher specificity prevents the global rule from taking effect.
* Fix scrollToPosition offset using marginTop instead of borderWidth
scrollPageIntoView aligns the page content edge with the container top,
leaving scrollTop equal to the page's top margin (12px) rather than 0.
The previous correction used borderWidth (effectively 0) so the margin
offset was never compensated. Using marginTop scrolls back the correct
amount so the margin above the first page is visible.
* Prevent PDF viewer collapsing during recompile by preserving height
When setDocument() is called with a new PDF, _resetView() synchronously
clears all page elements, briefly collapsing the .pdfViewer div to the
viewport height. This produces a visible flicker before pagesinit fires
and pages are re-added.
Fix: record the current height and pin it as min-height on the .pdfViewer
element before calling setDocument(). A one-shot pagesinit listener
removes the constraint once the new pages are initialised at the correct
scale, by which point the element is already at its correct final height.
* Suppress PDF.js page-level loading spinner in Overleaf viewer
The PDF.js loadingIcon/loading classes briefly add a ::after pseudo-element
with display:block and contain:strict to each page div. Overleaf has its
own loading state UI so the spinner is redundant, and its activation was
the root cause of the shifts 4-5 height oscillation (the display change
broke CSS margin collapse on .pdfViewer, adding 2x page margins to its
computed height).
The display:block rule already added to .page prevents the direct cause
(Overleaf's .loading{display:inline-flex} colliding with the PDF.js class).
This rule makes the intent explicit by zeroing the ::after entirely.
* Wrap PDF setDocument in startViewTransition
---------
Co-authored-by: Brian Gough <brian.gough@overleaf.com>
GitOrigin-RevId: 353ab865de3c7872363a61592d86390dfc34dacc
* Add info to errors in ProjectLocator
* Update ProjectLocator.test.mjs
* Add info to errors in SSOConfigManager
* Update SSOConfigManager.test.mjs
GitOrigin-RevId: 5a13350af1808f3a16a4bc8a9946cbe8f15e6b3a
* feat: adding audit log entries when users max out their AI usage
* feat: also log when user hits quota exactly, since support wants to know that
* feat: moving audit logging to the rate limiters themselves
* feat: moving to single quota breach event with tool in info
* feat: adding audit log for ai quota tests
GitOrigin-RevId: 64056632f142a9ea22a703b7621234f93e9f6ec7
