# Description of Changes
Adds an eslint rule to disallow importing any Tauri APIs outside the
desktop folder to help hint to developers that they should be following
the frontend architecture.
While doing this, I also discovered that you can provide a custom
message in the `no-restricted-imports` rule, which is nicer than the
comments that I'd previously added to the eslint config file to explain
why they weren't allowed:
```text
/Users/jamesbrunton/Dev/spdf1/frontend/src/core/components/shared/config/configSections/GeneralSection.tsx
19:1 error 'src/core/contexts/PreferencesContext' import is restricted from being used by a pattern. Use @app/* imports instead of absolute src/ imports no-restricted-imports
20:1 error '../../../../../core/contexts/AppConfigContext' import is restricted from being used by a pattern. Use @app/* imports instead of relative imports no-restricted-imports
21:1 error '@tauri-apps/core' import is restricted from being used by a pattern. Tauri APIs are desktop-only. Review frontend/DeveloperGuide.md for structure advice no-restricted-imports
```
# Description of Changes
Redesign the Python AI engine to be properly agentic and make use of
`pydantic-ai` instead of `langchain` for correctness and ergonomics.
This should be a good foundation for us to build our AI engine on going
forwards.
# Description of Changes
Currently, cmd-r is set to rotate the PDF in the viewer instead of
perform refresh in the browser. This is unintuitive and confusing for
Mac users, and for Windows users (who are less used to doing ctrl-r for
refresh) it only works some of the time, if the Viewer is active, so
removing the override is no great loss.
# Description of Changes
Add frontend developer guide describing the path alias architecture.
There's probably more needed in here which we should flesh out over
time, but this is a start.
# Description of Changes
Ages ago I made #4835 to try and fix all the `any` type usage in the
system but never got it finished, and there were just too many to review
and ensure it still worked. There's even more now.
My new tactic is to fix folder by folder. This fixes the `any` typing in
the `saas/` folder, and also enables `no-unnecessary-type-assertion`,
which really helps reduce pointless `as` casts that AI generates when
the type is already known. I hope to expand both of these to the rest of
the folders soon, but one folder is better than none.
# Description of Changes
Fix#5164
As I mentioned on the bug
https://github.com/Stirling-Tools/Stirling-PDF/issues/5164#issuecomment-4045170827,
it's impossible to print on Mac currently because
`iframe.contentWindow?.print()` silently does nothing in Tauri on Mac,
but [it seems unlikely that this will be
fixed](https://github.com/tauri-apps/tauri/issues/13451#issuecomment-4048075861).
Instead, I've linked directly to the Mac `PDFKit` framework in Rust to
use its printing functionality instead of Safari's. I believe that
`PDFKit` is what `Preview.app` is using and the print UI that it
generates seems to perform identically, so this should solve the issue
on Mac. Hopefully one day the TS iframe print API will be fixed and
we'll be able to get rid of this code, or [there'll be an official Tauri
plugin for printing which we can use
instead](https://github.com/tauri-apps/plugins-workspace/issues/293).
This implementation should be entirely Mac-specific. Windows & Linux
will continue to use their TS printing (which comes from EmbedPDF)
unless we have a good reason to change them to use a native solution as
well.
# Description of Changes
Previously, `VITE_*` environment variables were scattered across the
codebase with hardcoded fallback values inline (e.g.
`import.meta.env.VITE_STRIPE_KEY || 'pk_live_...'`). This made it
unclear which variables
were required, what they were for, and caused real keys to be silently
used in builds where they hadn't been explicitly configured.
## What's changed
I've added `frontend/.env.example` and `frontend/.env.desktop.example`,
which declare every `VITE_*` variable the app uses, with comments
explaining each one and sensible defaults where applicable. These
are the source of truth for what's required.
I've added a setup script which runs before `npm run dev`, `build`,
`tauri-dev`, and all `tauri-build*` commands. It:
- Creates your local `.env` / `.env.desktop` from the example files on
first run, so you don't need to do anything manually
- Errors if you're missing keys that the example defines (e.g. after
pulling changes that added a new variable). These can either be
manually-set env vars, or in your `.env` file (env vars take precedence
over `.env` file vars when running)
- Warns if you have `VITE_*` variables set in your environment that
aren't listed in any example file
I've removed all `|| 'hardcoded-value'` defaults from source files
because they are not necessary in this system, as all variables must be
explicitly set (they can be set to `VITE_ENV_VAR=`, just as long as the
variable actually exists). I think this system will make it really
obvious exactly what you need to set and what's actually running in the
code.
I've added a test that checks that every `import.meta.env.VITE_*`
reference found in source is present in at least one example file, so
new variables can't be added without being documented.
## For contributors
New contributors shouldn't need to do anything - `npm run dev` will
create your `.env` automatically.
If you already have a `.env` file in the `frontend/` folder, you may
well need to update it to make the system happy. Here's an example
output from running `npm run dev` with an old `.env` file:
```
$ npm run dev
> [email protected] dev
> npm run prep && vite
> [email protected] prep
> tsx scripts/setup-env.ts && npm run generate-icons
setup-env: see frontend/README.md#environment-variables for documentation
setup-env: .env is missing keys from config/.env.example:
VITE_GOOGLE_DRIVE_CLIENT_ID
VITE_GOOGLE_DRIVE_API_KEY
VITE_GOOGLE_DRIVE_APP_ID
VITE_PUBLIC_POSTHOG_KEY
VITE_PUBLIC_POSTHOG_HOST
Add them manually or delete your local file to re-copy from the example.
setup-env: the following VITE_ vars are set but not listed in any example file:
VITE_DEV_BYPASS_AUTH
Add them to config/.env.example or config/.env.desktop.example if they are required.
```
If you add a new `VITE_*` variable to the codebase, add it to the
appropriate `frontend/config/.env.example` file or the test will fail.
# Description of Changes
Inspired by https://github.com/pydantic/pydantic-ai/pull/4169, this PR
moves our `CLAUDE.md` advice to the more generic `AGENTS.md` file (which
works on Codex, Gemini, etc). It also adds a symlink from `CLAUDE.md` to
`AGENTS.md`, which Claude follows properly, so all AIs should get the
same advice and we only need to keep one file up-to-date.
# Description of Changes
Adds the code for the SaaS frontend as proprietary code to the OSS repo.
This version of the code is adapted from 22/1/2026, which was the last
SaaS version based on the 'V2' design. This will move us closer to being
able to have the OSS products understand whether the user has a SaaS
account, and provide the correct UI in those cases.
# Description of Changes
Fix#5779. Cookie consent persistance doesn't work on desktop (on Mac at
least, not sure about Windows) because of permission differences with
Tauri. We are allowed to store things in local storage fine, so this
switches the cookie consent module to store in local storage for
desktop, and leaves it alone for web, where it already worked correctly.
# Description of Changes
Change the SAML support for SSO to understand when a request is coming
from the desktop app, and use the alternate auth flow that the desktop
app requires.
# Description of Changes
This PR fixes false-positive TypeScript errors in our layered build
setup (core → proprietary → desktop) by ensuring each build’s typecheck
only evaluates files that are actually part of that build’s reachable
module graph. This prevents overridden core implementations from being
typechecked in higher-layer builds where they are effectively
unreachable due to alias-based overrides.
## Background
We maintain multiple build targets from a layered source tree:
- core: open source baseline
- proprietary: core + proprietary additions/overrides
- desktop: proprietary + desktop-specific additions/overrides
We implement overrides via paths/aliases such that placing a file in a
higher layer at the same relative path supersedes the lower-layer file
at runtime.
For safety, we run TypeScript typechecking independently per build
target to ensure all builds remain valid.
## Problem
Our existing tsconfig setup often typechecked files that are not
actually reachable in a given build. Specifically:
- When a file in core is overridden by a file in proprietary or desktop,
the overridden core file can still be included in the TypeScript Program
for the higher-layer build (typically due to broad include globs).
- This produces false-positive type errors in higher-layer typecheck
runs, even though those core files are effectively unreachable in the
build.
This created friction and noise, and meant we had to make unnecessary
changes to `core` to make the other builds happy, reducing type safety
in the process.
## Solution
This PR adjusts the tsconfig strategy so each build target's typecheck
is driven by reachable entrypoints rather than blanket inclusion of all
layer source trees. Concretely:
- Each build’s tsconfig now includes only:
- that build’s entrypoints and layer sources that are intended to be
compiled for the target
- any shared/top-level sources required by the target
- Lower layers (e.g., core) are not globally included in higher-layer
builds; they are instead pulled in through module resolution only when
actually referenced (with paths ordering ensuring the correct override
wins).
- This means that we still check all the files that will actually be run
with whatever the overridden logic is, but avoid wasting time and
introducing false-positives by not checking files which have been
overridden.
## Notes
Unfortunately, the config we use for the type checking can't be the same
as the one we use for Vite in this strategy. Vite needs to know about
the entire source tree, so it can't only include the subfolders because
it causes build errors. Because of this, I've duplicated the existing
(valid) tsconfig files and use them for Vite. This is a little clunky
but it does the job. Some day hopefully I'll come back to it and be able
to figure out a nicer way to do it, but for now at least, this solves
the type checking issues without impacting the runtime builds.
Also, I noticed that `@desktop` is defined as an alias, which was
presumably missed when I was removing the self-aliases from the files. I
don't see why you'd ever need to have a desktop file reference
`@desktop` to say "import this but make it impossible for something else
to override the import". I've removed the `@desktop` alias in this PR
while I was in there.
# Description of Changes
- Just build proper installers in CI for each platform
- Provide commands to build just the bundled apps without the need for
installers locally
- `tauri-dev` - Builds quickly for an unoptimised version of the app
- `tauri-build` - Builds the full optimised app installer for release
- `tauri-build-dev` - Builds an optimised app with no bundling (builds
to a folder on Mac; raw `.exe` on Windows; etc.)
- `tauri-build-dev-mac` - Builds an optimised bundled Mac app with no
installer (as an `.app` file)
- `tauri-build-dev-windows` - Builds an optimised bundled Windows app as
an `.nsis` installer
- `tauri-build-dev-linux` - Builds an optimised bundled Linux app as an
`.appimage`
# Description of Changes
PDF editor has been impossible to access since #5169. This fixes the bug
introduced in that which made it impossible to access. The main issue
was the refactor removed the `setLeftPanelView` call.
# Description of Changes
Currently when applying stamps, the preview of the PDF is not being
treated as private content. This correctly marks the content as private
# Description of Changes
Fix#5206
Required splitting out the logic for the text with font entry so that
the labels are configurable from the call-site instead of all using the
same ones for Sign.
# Description of Changes
Fix#5189.
Fix Mac app not being able to open files with spaces in their name,
which was happening because the URL was not being decoded on input.
# Description of Changes
Currently, the Quick Access Bar only renders well in Chrome. It's got
all sorts of layout issues in Firefox and Safari. This PR attempts to
retain the recent changes to make the bar thinner etc. but make it work
better in all browsers.
# Description of Changes
Add lite mode for CI which just runs the most important jobs for
deployment. This won't be used in this repo, but allows other repos
containing Stirling to easily disable jobs like desktop builds etc. if
they're unnecessary, without needing to deal with conflicts in the
files. They'll just need to set the repo variable `CI_PROFILE` to
`lite`. We have an upstream repo that we'd like these changes for.
# Description of Changes
- Add new skeleton loader style type (block - nothing's currently using
it but it might as well be available)
- Make Dev API overridable (and set to the new docs that actually work
while Swagger docs don't work properly)
# Description of Changes
Start bundled backend instantly on startup of app and don't wait on it
being fully up to spawn app. This is techincally wasteful curently on
self-hosted mode where everything runs remotely, but in the future we'll
probably route simple operations to the local machine regardless of
connection, and it stops unnecessary waiting in the offline mode.
# Description of Changes
Fix#5017
Changes the handling of jwt tokens to be stored in local storage as well
as OS keyring so the rest of the app knows that you're logged in.
# Description of Changes
In the 2.0.0 release version, the frontend can't connect to the backend
on my machine because all the network requests 403. I think this is
because of CORS issues, and supposedly these will be fixed by using a
different Spring function, which is more lenient on URL schemes (needs
to allow `tauri://localhost` here, which isn't a standard URL)
# Description of Changes
Fixes two distinct but related issues in the backend of the desktop app:
- Correctly shows tools as unavaialable when the backend doesn't have
the dependencies or has disabled them etc. (same as web version - this
primarily didn't work on desktop because the app spawns before the
backend is running)
- Fixes infinite re-rendering issues caused by the app polling whether
the backend is healthy or not
# Description of Changes
- ~Force classic logo~
- Refer to email instead of username in SaaS sign in flow
- Allow drag-and-drop files into desktop app
- Convert terminology & icons from upload/download to open/save in
desktop version
# Description of Changes
Makes the desktop options to sign in with your Stirling account, or sign
into self-hosted:
<img width="608" height="456" alt="image"
src="https://github.com/user-attachments/assets/a49988ab-db3f-4333-b242-790aee5c07c6"
/>
The first option still runs everything locally, just enforces that
you've signed in for now. Future work will enable sending operations
that can't be run locally to the server.
# Description of Changes
Changes the desktop app to allow connections to self-hosted servers on
first startup. This was quite involved and hit loads of CORS issues all
through the stack, but I think it's working now. This also changes the
bundled backend to spawn on an OS-decided port rather than always
spawning on `8080`, which means that the user can have other things
running on port `8080` now and the app will still work fine. There were
quite a few places that needed to be updated to decouple the app from
explicitly using `8080` and I was originally going to split those
changes out into another PR (#4939), but I couldn't get it working
independently in the time I had, so the diff here is just going to be
complex and contian two distinct changes - sorry 🙁
# Description of Changes
The onboarding tour wasn't presenting itself as having any dependencies
to React so would never be re-rendered after language changes.
# Description of Changes
- Adds a reusable banner component/system to the core app
- Adds banner at the top of the desktop app if Stirling isn't your
default PDF editor, with a button to make it your default
- Adds a permanent button in the settings to do it manually (in case
you've dismissed the banner)
- Simplifies the file loading logic to fix a bug where the input file
could be duplicated occasionally. Now, the TS just receives files from
one buffer, regardless of how they've been passed to the app in Rust.
## Caveats
I've only been able to get the setting of default apps working properly
on Mac. The Windows build isn't signed (yet) so we can't use the proper
API for it, so currently it just sends you to the Settings UI. I've also
not been able to test it on Linux at all.
# Description of Changes
Re-enable macOS signing and tweak so it runs successfully through CI.
Also changes the runner to use macOS 15 instead of 13, which was
throwing a deprecation warning in GitHub. Note that the runner doesn't
affect the minimum target, which I've still got set to 10.15 (no idea if
it actually works on 10.15 but let's assume that if it builds it works
until someone can test it)
# Description of Changes
Locking to just having one instance of the app running unifies the
experience across all OSs. Opening new files in Stirling will cause the
files to be opened in the existing window rather than spawning a new
instance of the app with just that file in the new instance.
There's much more to explore here to allow multiple windows open at
once, but that can be done all from one instance of the app, and will
likely make it easier to allow movement of files etc. across different
windows.
Also fixes extra newlines in the logs and directly builds to `.app` on
Mac because it's frustrating during development to have to repeatedly
mount & unmount the `.dmg`.
# Description of Changes
Fix warnings in Rust code on Mac. They were all being caused by file
handling logic which is now built into Tauri, so I've just been able to
remove all of the Mac specific file handling code.
I've also set warnings to be treated as errors because it'll be really
easy to accidentally introudce warnings on individual platforms which
I'm not developing, and I'd like to know about them so we can fix it
before getting dodgy code.
# Description of Changes
`i18next` allows this pattern for translations, which we use quite a few
times in our current translation files:
```json
{
"a": {
"b": "hello"
},
"a.b": "world"
}
```
This makes it ambiguous when selecting `a.b` which string will be
retrieved. We have seen issues in other languages in the current release
like this:
<img width="325" height="249" alt="image"
src="https://github.com/user-attachments/assets/f24a29f0-550f-49b8-b355-c5e5eb436558"
/>
because we are expecting this:
<img width="1022" height="210" alt="image"
src="https://github.com/user-attachments/assets/b6d5cdd4-96cd-4b2b-8f1a-465da8bf70c8"
/>
but the Spanish file has:
<img width="312" height="136" alt="image"
src="https://github.com/user-attachments/assets/1e13392c-8484-47d1-b0c4-19d52b3ea5eb"
/>
and no `removeDigitalSignature` key on its own.
This PR resolves all of these ambiguities in the source by restructuring
all of the keys to uniquely target either an object or a string, not
both. It also adds a test which will fail on any keys with a `.` in
their name, therefore making it impossible to add anything ambiguous.
# Description of Changes
Remove path aliases from self folder (e.g. remove `@core` from
`tsconfig.core.json`). It's not necessary and using it means that it's
impossible for the other folders to override the behaviour. The only
reason we should currently be using `@core` is in `proprietary` where we
need to explicitly import the `core` version of the thing we're
overriding so that we can re-expose or use the objects.
# Description of Changes
Add linting to ensure correct imports style is used.
I've disabled the linting for two imports which use relative paths
because the files they're importing are siblings to core and
proprietary. They could probs be imported by `@app/../assets/xxx` but it
seems silly. The other thing we could do is add an explicit `@assets`
path alias or something, but it seemed more complex than just disabling
the lint for those two imports at this stage. We could always do it in
the future if we want to import stuff up there a lot in the future.
# Description of Changes
[See my comment here on why I think we should never allow lint warnings
to be merged into our
source](https://github.com/Stirling-Tools/Stirling-PDF/pull/4738#issuecomment-3451053692).
This doesn't change how ESLint behaves at all other than if only
warnings are reported, it'll report failure instead of success.
# Description of Changes
Refactors code to avoid circular imports everywhere and adds linting for
circular imports to ensure it doesn't happen again. Most changes are
around the tool registry, making it a provider, and splitting into tool
types to make it easier for things like Automate to only have access to
tools excluding itself.
# Description of Changes
Adds a test to scan the code for any static translation keys which are
not present in the GB translations file. The test won't catch every
missing translation present in our code, but it should greatly help us
keep the translations file up to date.
Replace kebab menu in file editor with on hover menu by refactoring page
editor's menu into a new component. In mobile sizes, the hover menus are
always visible.
# Description of Changes
Refactor user preferences to all be in one service and all stored in
localStorage instead of indexeddb. This allows simpler & quicker
accessing of them, and ensures that they're all neatly stored in one
consistent place instead of spread out over local storage.
# Description of Changes
Change shortcuts to just be a limited set for Quick Access tools rather
than for everything to avoid breaking browser key commands by default.
Also fixes a bunch of types of variables that were representing
`ToolId`s (I stopped at `automate` because there's loads in there so
I've just introduced some `any` casts for now 😭)
# Description of Changes
Make Viewer always accessible. Also deletes some dead code I noticed
while I was working.
---------
Co-authored-by: EthanHealy01 <[email protected]>
Co-authored-by: Reece Browne <[email protected]>
# Description of Changes
For the first release of V2, we'll not have any reasonable translations
for anything other than English GB, so with that in mind, this PR
disables language selection for anything other than English GB, with a
tooltip saying the other languages are coming soon. I also split the JSX
up a little bit while I was at it to make it easier to manage.
---------
Co-authored-by: EthanHealy01 <[email protected]>
# Description of Changes
Remove custom response handler from Merge. Also make `filePrefix`
mandatory for `multiFile` tools to make the output more visually
different since you get a new 'V1' file rather than 'V2' of the current
file.
# Description of Changes
Delete Claude local settings, which shouldn't really be in the repo.
Note that this is already in the `.gitignore` file, so there's no need
to change that as well.
# Description of Changes
Changes it so that callers of `useBaseTool` know what actual type the
parameters hook that they passed in returned, so they can actually make
use of any extra methods that that params hook has.
# Description of Changes
Adds auto-redact tool to V2, with manual-redact in the UI but explicitly
disabled.
Also creates a shared component for the large buttons we're using in a
couple different tools and uses consistently.
# Description of Changes
Change NPM scripts so they call each other (single source of truth) and
add a command to run type checking, linting and tests (to give
confidence CI will pass).
# Description of Changes
There's no current linter running over our TypeScript code, which means
we've got a bunch of dead code and other code smells around with nothing
notifying us. This PR adds ESLint with the typescript-eslint plugin and
enables the recommended settings as a starting point for us.
I've disabled all of the failing rules for the scope of this PR, just to
get linting running without causing a massive diff. I'll follow up with
future PRs that enable the failing rules one by one.
Also updates our version of TypeScript, which introduces a new type
error in the code (which I've had to fix)
# Description of Changes
Split was previously incorrectly marked as a multi-file interface, which
meant that if you fed 2 files into it, it'd just process the first and
discard the second.
This PR changes it to a single-file interface, and implements a custom
response handler because Split returns Zip files instead of PDFs, so the
response you get when running Split now is the union of all of the split
input files in the workbench (or them all zipped if you download it).
# Description of Changes
When adding files, the user probably wants to use them straight away in
the next tool that they use, so automatically select any added files (in
addition to whatever they previously had selected).
# Description of Changes
Reduce boilerplate in tool frontends by creating a base frontend hook
for the simple tools to use.
I've done all the simple tools here. It'd be nice to add in some of the
more complex tools as well in the future if we can figure out how.
# Description of Changes
The `FileId` type in V2 currently is just defined to be a string. This
makes it really easy to accidentally pass strings into things accepting
file IDs (such as file names). This PR makes the `FileId` type [an
opaque
type](https://www.geeksforgeeks.org/typescript/opaque-types-in-typescript/),
so it is compatible with things accepting strings (arguably not ideal
for this...) but strings are not compatible with it without explicit
conversion.
The PR also includes changes to use `FileId` consistently throughout the
project (everywhere I could find uses of `fileId: string`), so that we
have the maximum benefit from the type safety.
> [!note]
> I've marked quite a few things as `FIX ME` where we're passing names
in as IDs. If that is intended behaviour, I'm happy to remove the fix me
and insert a cast instead, but they probably need comments explaining
why we're using a file name as an ID.
James BruntonandGitHub
