# Description of Changes
One of the Playwright tests is flaky, despite several attempts to fix it
before it made it into main. This disables the test for now so a
followup PR can try to fix it again.
Closes#6518
# Cause of the bug
This is a fix to the #6518 issue. The bug happened because the link
toolbar was rendered inside the PDF page layer. That layer can be
affected by the viewer/page rotation transform, so the toolbar was laid
out using local page coordinates and then visually transformed together
with the page.
As a result, the placement logic could calculate a position that was
correct in the page’s local coordinate space, such as above or below the
link, but the parent transform would rotate or shift that result after
layout. On rotated pages, this could make the toolbar appear on the
wrong side, inverted, or misaligned relative to the link.
More specifically, in the PDF that exposed the bug, the page content
appears to have been authored upside down and then corrected with a
180-degree page/viewer rotation so it looks normal to the user.
Because the toolbar was rendered inside the same transformed page layer,
it inherited that 180-degree rotation as well. The PDF content looked
upright because the rotation was part of how the page was displayed, but
the toolbar is viewer UI and should not be rotated with the page. As a
result, the tooltip appeared upside down even though the PDF itself
looked correct.
# Description of Changes
Fixes the inverted link tooltip/toolbar positioning in rotated PDF
viewer pages.
The link toolbar is now rendered through a body portal and positioned
from the link element’s real viewport bounds, so page rotation
transforms no longer flip or misalign it.
The update also keeps the toolbar within the viewport during scroll,
resize, zoom, and rotation changes, preserves the hover delay between
the link and toolbar, centralizes the z-index in a shared constant, and
improves label sizing to avoid clipped text.
Note: The link hover styling was also changed from an underline to a
subtle rectangular highlight based on the PDF link annotation bounds.
<!--
Please provide a summary of the changes, including:
- What was changed
- Why the change was made
- Any challenges encountered
Closes #(issue_number)
-->
---
## Checklist
### General
- [X] I have read the [Contribution
Guidelines](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/CONTRIBUTING.md)
- [X] I have read the [Stirling-PDF Developer
Guide](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/DeveloperGuide.md)
(if applicable)
- [ ] I have read the [How to add new languages to
Stirling-PDF](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/devGuide/HowToAddNewLanguage.md)
(if applicable)
- [X] I have performed a self-review of my own code
- [X] My changes generate no new warnings
### Documentation
- [ ] I have updated relevant docs on [Stirling-PDF's doc
repo](https://github.com/Stirling-Tools/Stirling-Tools.github.io/blob/main/docs/)
(if functionality has heavily changed)
- [ ] I have read the section [Add New Translation
Tags](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/devGuide/HowToAddNewLanguage.md#add-new-translation-tags)
(for new translation tags only)
### Translations (if applicable)
- [ ] I ran
[`scripts/counter_translation.py`](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/docs/counter_translation.md)
### UI Changes (if applicable)
- [X] Screenshots or videos demonstrating the UI changes are attached
(e.g., as comments or direct attachments in the PR)
UI behaviour before the changes :
<img width="1256" height="868" alt="Captura de tela de 2026-06-16
00-12-10"
src="https://github.com/user-attachments/assets/321edbb3-42a2-4bc3-96ad-3ccc70a355b8"
/>
<img width="762" height="496" alt="Captura de tela de 2026-06-16
00-11-46"
src="https://github.com/user-attachments/assets/be4c1af4-5488-4a54-9b6f-675e3bea73b8"
/>
<img width="1256" height="868" alt="Captura de tela de 2026-06-16
00-12-57"
src="https://github.com/user-attachments/assets/60f44cd5-c772-44a8-97c8-bde135764e53"
/>
UI behaviour after the changes :
<img width="1256" height="868" alt="Captura de tela de 2026-06-16
00-22-02"
src="https://github.com/user-attachments/assets/dda77bda-0780-4807-a70d-3bbc60683e5a"
/>
<img width="1256" height="868" alt="Captura de tela de 2026-06-16
00-23-07"
src="https://github.com/user-attachments/assets/5745c37e-438a-4bbe-ba1e-c6f2098421de"
/>
<img width="1256" height="868" alt="Captura de tela de 2026-06-16
00-23-24"
src="https://github.com/user-attachments/assets/85932541-4a6f-48e4-879f-41f34a6d79e6"
/>
### Testing (if applicable)
- [X] I have run `task check` to verify linters, typechecks, and tests
pass
- [X] I have tested my changes locally. Refer to the [Testing
Guide](https://github.com/Stirling-Tools/Stirling-PDF/blob/main/DeveloperGuide.md#7-testing)
for more details.
---------
Co-authored-by: James Brunton <[email protected]>
Co-authored-by: Anthony Stirling <[email protected]>
## What & why
Production reports of policy enforcement "hanging" traced to
large/many-page documents: the watermark step's flatten-to-image
(`convertPDFToImage`) on a 500+ page PDF takes minutes, exceeding both
the client poll cap and the backend per-step timeout. This makes the
slow case graceful instead of looking broken, and makes load-shedding
non-fatal.
### Poll runs to completion (no false "hang")
The client poll loop used a flat ~150s cap that was **shorter than the
backend's 300s per-step timeout**, so it abandoned long-but-healthy runs
mid-flight. The budget is now sized to the backend's real worst case —
`stepCount × per-step timeout + grace`, learned from the first status
report — so the client always polls long enough to surface the run's
**actual** terminal state (success or the backend's real error) rather
than a misleading client-side timeout.
### Per-step progress
The activity feed now shows `Enforcing… · step n/m` (from
`currentStep`/`stepCount`), so a slow step shows movement instead of a
dead spinner.
### Soft-retry on queue rejection
Under load the shared `JobQueue` rejects runs ("queue full"), which
previously surfaced as a hard failure needing a manual Retry. The
backend now tags that rejection with a stable `POLICY_QUEUE_FULL`
errorCode; the client treats it as transient backpressure and
**auto-retries the file in place** with exponential backoff (≈4s→64s, ~2
min), shown as a soft "Busy — retrying…" row, falling back to the manual
Retry only once the retry budget is spent.
## Testing
- **Frontend unit tests** (30 pass across the policies suite), including
a new `usePolicyAutoRun.retry.test.tsx` that drives the real controller
orchestration (poll → `POLICY_QUEUE_FULL` → relabel → backoff → in-place
re-dispatch), plus poll-budget, step-progress, and activity-feed relabel
cases.
- **Backend** `PolicyEngineTest` case asserting a queue-rejected run
carries the `POLICY_QUEUE_FULL` code.
- Typecheck clean on all three flavors (proprietary/saas/core); prettier
+ spotless clean.
- Poll-budget + progress + real-error surfacing were also verified live
end-to-end against a 599-page run (survived past the old cap, showed
step progress, reported the backend's real 300s-timeout failure,
recovered after a simulated network drop).
## Not included (follow-ups)
- The underlying flatten-to-image cost itself (bounded-memory/streaming
flatten, revisiting `convertPDFToImage` default and the 300s timeout) —
the real perf fix, deliberately out of scope here.
# Description of Changes
Fix Playwright failing test in SaaS (I think this is my third attempt
now so who knows if this will actually fix it for real this time, but
hopefully it does)
# Description of Changes
> [!warning]
> **Do not** squash this on merge. It should be merged via a merge
commit
Fixes conflicts in `pgvector_store.py`.
Also since codespell is failing, add comments to ignore the errors in
`sync_en_us_spelling.py`
---------
Co-authored-by: Ludy <[email protected]>
Co-authored-by: Anthony Stirling <[email protected]>
# Description of Changes
* Remove complex port selection logic from `engine.yml`. It's
inconsistent with the frontend & backend task files, and caused issues
with Docker, which have been worked around but would be simpler to just
get rid of the problem altogether
* Fix Ruff formatting of Python script
* Remove payg tests which are failing and have drifted too far from the
implementation to save directly
`PolicyController` was annotated `@PremiumEndpoint` (requires a
Pro-or-higher server license). Policies don't need a server-license
gate:
- On SaaS the server runs in Pro mode, so the check is always satisfied
anyway — it gates nothing in practice.
- Access is already governed by team scoping (#6632) plus the per-user /
guest gates.
So the annotation is dead weight and misleading. This removes it (and
its import) — a 2-line change.
## Verification
- `:proprietary:compileJava` succeeds; spotless clean. No other premium
gate on policy classes.
Guests (anonymous users on a login-enabled deployment) could open a
policy's setup/detail. Policies are an account feature, so a guest
clicking a policy should be nudged to sign up rather than opening it.
## Behaviour
A guest clicking a policy row — or a collapsed-rail icon — now
**re-summons the existing guest sign-up banner** ("You're using Stirling
PDF as a guest!…") and does **not** open the policy.
- `GuestUserBanner` listens for a `stirling:show-guest-banner` window
event and re-shows (even if previously dismissed; the render guard still
hides it for non-anonymous users).
- The policy sidebar dispatches that event on a guest click (cross-layer
via `CustomEvent`, same pattern as `payg:signupRequired`; a no-op on
builds without the banner).
- `usePolicyGuestBlocked()` gates it: `config.enableLogin === true &&
user.is_anonymous === true`.
- **Login-disabled single-user** deployments have an anonymous local
operator with full access → not gated.
## Verification
- Typecheck clean (proprietary + saas); eslint clean; sidebar tests
pass.
## Note
No dedicated guest unit test — the suite mocks `useAppConfig` at module
scope, and making it per-test controllable needs `vi.hoisted` plumbing
that risked the existing tests. Easy follow-up.
## Problem
We're getting 402s when an AI **agent** (chat) run hits the free
allowance / spending cap, but the frontend handles them poorly and never
pops the usage-limit modal.
The agent runs its tool calls **server-side** (loopback HTTP via
`PolicyExecutor`), so the 402 never reaches the `apiClient` interceptor
that pops the modal for direct calls. It was caught by the generic
tool-failure handler and flattened into a `CANNOT_CONTINUE` reason
string (`"The /api/v1/… tool failed: 402…"`), streamed as a `result`
event, and rendered as a scary chat bubble. This is the same gap the
policy auto-run path bridges (#6626) — one layer up.
## Fix
**Backend** (`proprietary`)
- `AiWorkflowResponse` gains `errorCode` + `errorSubscribed`.
- `AiWorkflowService` detects a downstream 401/402 entitlement sentinel
in its three tool-exec catch sites (`onToolCall`, `runPlan`,
`onConvertMarkdown`) and surfaces the structured code (+ `subscribed`)
on the terminal response instead of the raw failure text.
- Factored the 401/402 body extraction `PolicyEngine` already had into a
shared `DownstreamEntitlementError` util so the two server-side paths
can't drift.
**Frontend**
- New `usageLimitBridge` (`PAYG_LIMIT_REACHED_EVENT` +
`dispatchPaygLimitReached`) generalises the previously policy-only
bridge. Proprietary can't import the saas modal API (layering), so
server-side limit hits broadcast a window event the saas
`UsageLimitModalHost` opens the modal from. Migrated the policy path
onto it.
- `ChatContext` fires the matching modal (free → subscribe, subscribed →
raise cap) on the limit result **and** on a direct 402, replacing the
raw reason with a brief friendly line
(`chat.responses.usage_limit_reached`).
No Python engine changes — the charge/402 happens on the Java tool
endpoint that Java itself calls.
## Test plan
- [x] `:proprietary:compileJava` + `spotlessCheck` clean
- [x] `AiWorkflowServiceTest` + `PolicyEngineTest` green
- [x] eslint, proprietary + saas typechecks clean
- [ ] Manual: drive an agent run over the limit → brief line in chat +
the right modal (free vs cap)
> Note: proprietary test compilation is currently blocked on the
pre-existing `InitialSecuritySetupTest` 6-arg ctor break (unrelated,
tracked separately); verified locally by temporarily patching it.
add en-US changes to SaaS, previously merged into main. So this is
effectively a main -> SaaS PR also. It seems to be all additive.
Also take the 230 ish missing translations from en-GB over to en-US
using a script, and also make and english spellings American when adding
them to the en-US file, and fix any existing American spellings in the
en-GB file.
# Description of Changes
Search has got significantly worse since #6581, where I added all the
missing tags for tools that should have been there for months. Turns out
that the fuzzy matching search logic has always been way too permissive
to match words with Levenshtein distances way too far away from the
target word, so long searches include way too much stuff. The new tags
just exposed that underlying logic issue. This PR makes the Levenshtein
logic much stricter, so it is still tolerant to minor typos in tool
names, but doesn't match completely inappropriate strings.
Frontend follow-up to #6632 (team-scoped policies, editing gated to team
leaders on the backend). Brings the UI's edit gate in line.
## Problem
The policy config UI gated editing to `config.isAdmin`. On SaaS, org
users are never the single global admin, so **no one could open the
policy editor** — the same lockout #6632 fixed on the backend.
## Fix
`usePolicies` now allows a **team leader** to configure, falling back to
a global admin self-hosted:
```ts
canConfigure =
config != null && (!config.enableLogin || isTeamLeader || config.isAdmin === true);
```
- SaaS → `isTeamLeader` (from `useSaaSTeam()`) — team leaders can
configure; members get the read-only surface.
- Self-hosted → `config.isAdmin` (the core `useSaaSTeam` stub returns
`false`, so admins aren't locked out).
- Login disabled (single-user) → always allowed.
- The `config != null` guard keeps the gate closed until app-config
resolves, so edit controls never flash for users who can't use them.
The two locked-policy banners now read "Contact a team leader to change
this policy" (updated in the `t()` defaults and the `en-GB`
translations).
## Verification
- Typecheck clean (proprietary + saas); eslint clean.
- Tests pass: `usePolicies`, `PoliciesSidebar`.
Anthony StirlingandGitHub
Reece BrowneandGitHub
Anthony Stirling