# Description of Changes
Playwright tests currently fail in Firefox and Safari because of
inconsistent behaviour across the browsers. This is causing the
nightlies to fail every night. This PR fixes the test behaviour to work
consistently across browsers (most of the issues were to do with the
tests opening the file picker, which was being automatically suppressed
in Chromium, but not the other browsers).
# Description of Changes
We don't have any strong reasons to keep the Portal as a separate Vite
app, and it needs access to so many things from the Editor that it no
longer makes sense to keep them separate. This PR moves the Portal code
to have direct access to the Editor code and gets rid of the shared
folder.
# Description of Changes
Fix#6801, along with fixing policies on desktop, which would attempt to
download policy outputs from the local backend instead of the server,
where they actually live. I've changed the policies logic to maintain
the same backend for the file retrieval as it used for the policy
running, so when we support running policies locally, it should still
work correctly.
# Description of Changes
Continued effort to expand linting scope to ban the `any` type in our
codebase. This PR pulls in a lot of subfolders into the linting scope,
because the excluded list was getting short enough that it was feasible
to move a layer down. I then fixed all the trivially fixable `any` type
violations in the subfolders, which just required local changes to the
one file. The aim of this PR is more to expand the scope to all the
folders we can that already avoid `any` types, rather than actually fix
violations.
# Description of Changes
We can't convert to TS7 completely yet because it lacks the TS API, so
ESLint and some of our scripts don't work, but we can do [what the TS
team suggest and run TS6 and TS7
side-by-side](https://devblogs.microsoft.com/typescript/progress-on-typescript-7-december-2025/#compiler).
When we do that, we take the `task frontend:typecheck:all` job from ~76s
to ~13s, and everything else continues to work as it did before.
I've set it so that CI will still use TS6 for the time being and locally
we use TS7 out of an abundance of caution because CI time doesn't really
matter but local time does. I do think it was a bit pointless doing that
since the TS team claim the type checking performs identically, but we
might as well have it like that for now. If it happens to go badly
locally for any devs, they can use `CI=true task frontend:typecheck` to
revert to use TS6 trivially.
# Description of Changes
[Our nightlies have literally never passed
before](https://github.com/Stirling-Tools/Stirling-PDF/actions/workflows/nightly.yml).
As far as I can tell, that's because the frontend was never being built,
so the Playwright tests would just never start up.
I've forced a nightly run from this branch, and the Playwright tests
still fail, but for legitimate failures now. It's a separate job to
track down why they're actually failing, so I'm leaving that for
followup work.
# Description of Changes
Redesign policies backend to treat sources a lot closer to how the
frontend imagined them working (they're persistent now and have an API).
Then connect the portal to the sources when mocks are off to allow for
source creation in the UI. It's not particularly useful to do that right
now because there's no policies UI, but I've tested manually that
sources set up in the UI are usable by policies created via the API.
I had to change the portal so that when mocks are off, it doesn't just
hard crash when attempting to connect to all the backend APIs that don't
exist yet. It'll still log the errors, but just continues on rendering
the UI now.
I also changed all the policies backend APIs to be gated behind a flag
instead of behind the SaaS profile. This is because we haven't yet got
the payment model sorted, but we're going to need this stuff running
self-hosted to be able to test it locally.
# Description of Changes
- Change the nightly build to not sign any of the desktop builds, since
we just care about the compiled code. The restored code will still be
signed dependent on the OS in the PR builds.
- Change RPM Linux to use zstd for compression because the one it was
using runs really slowly, and the Jar is already compressed so it makes
basically no difference (arguably we shouldn't compress at all)
- ~Switch to consistently use Depot for Docker caching to stop filling
up the GHA cache and evicting the Rust cache~ Decided against switching
to Depot because we're probably doing another PR to remove Depot
altogether in the near future
# Description of Changes
Fixes one of the main causes of `any` typing left in tools, the way that
we register tool parameters in the registry. Currently, it just accepts
tool params via `any`, but instead we can explicitly change them to
`Record<string, unknown)`, so on the way back out they can more safely
be cast back to their correct type when known.
One consequence of this is that I had to redesign the way we
special-case the Convert tool, which previously was a different shape
than all the other param types. Now it's just got optional parameters on
it, which isn't quite as type-safe as before, but it does mean all tools
are a consistent shape now, which I think is worth the tradeoff.
# Description of Changes
TS6 introduced backwards-incompatible changes which affected us a little
bit. Other than that, I don't think it significantly changes things for
us, but we will need to deal with these breaking changes to be able to
upgrade to TS7 (the version written in Go, so dramatically faster), so
I'd rather do the work now before TS7 actually releases.
Main things I've done:
- Removed the use of `baseUrl` in the `tsconfig.json` files
- Explicitly provide the `node` types where needed
- Explicitly reference the un-referenced but required Google API types
- Dropped the installation of `madge` which we weren't using and wasn't
directly compatible with TS6
- Updated the `i18next` packages for explicit TS6 compatibility
- Explicitly override `tsconfck` to force TS6 compatibility since we
can't upgrade it. We're only using that for `vite-tsconfig-paths` and it
all still seems to work fine, so I don't think this is an issue. I think
we can theoretically drop `vite-tsconfig-paths` when we upgrade to Vite
8 ([because it supports
`paths`](https://v8.vite.dev/guide/features#paths)), but that's a bigger
job than I want to do in this PR
# Description of Changes
Refactor frontend auth to the shared folder and hook it up to both the
portal and editor so they share the same system. Also adds various tasks
to help run the portal, including `task dev:portal` to spawn the portal
with the backend, and `task dev:portal:proxy` to spawn the editor,
portal and backend, and a reverse proxy (at localhost:3000) to allow you
to use both at once to simulate how this will actually be deployed,
allowing you to check whether the seamless transition between the two
actually works.
# Description of Changes
> [!note]
> GitHub absolutely mangles the diff unless you change to ignore
whitespace changes
This page has never had the Enter key bound to the Change Password
button:
<img width="673" height="745" alt="image"
src="https://github.com/user-attachments/assets/7a1b06f0-2945-4270-a795-f799ec556c12"
/>
This PR changes the modal to be properly wrapped in a form so key
commands work correctly on it.
# Description of Changes
Rust cache added in #6732 never fired because `main` builds don't
include building the desktop apps. We could build them on `main` builds,
but that's fairly expensive, so just build them on nightlies instead to
warm the cache for any desktop PRs the next day
# Description of Changes
The Tauri jobs are very slow, especially the Linux ones, which can take
>1hr to build all the necessary code. A lot of that is because of the
actual Rust compilation, which isn't cached at all as far as I can tell.
This introduces a cache step for the Rust dependencies, so PRs will just
reuse the compiled Rust from the last build of main (if it's safe to do
so).
# Description of Changes
Continued effort to remove the remaining uses of the `any` type from our
TS code. The vast majority of these uses that it cleans up was just
catching errors as `any`, which are pretty simple to fix. I couldn't
completely remove the `any` type usage from `core/tools` because there
were cascading issues from a couple of the files in there (most notably
Automate) but still, moving in the right direction.
# Description of Changes
#6727 introduced frontend code which goes against the architecture, so
this PR re-implements it in the architecture properly, along with
another bad Tauri check that I found in the source. I also updated the
`AGENTS.md` file to use Claude's "read this file" syntax to try and
force AI to actually read the file instead of just suggesting that it
does it.
# Description of Changes
Add message describing the most common tasks when running `task` with no
arguments. I think this should help newcomers because `task --list` is
massive at this point and nobody's going to read through it all. Let me
know if you think any other commands should be in the default message.
# Description of Changes
The `pre-commit` commands in this repo are inconsistent with the rest of
the dev workflow, as they are impossible to run through Task and they
can cause CI to fail with no way for a developer to run the `pre-commit`
scripts after they've failed. This PR adds `task pre-commit` (and `task
pre-commit:fix`) and then hooks up the existing `pre-commit` hooks and
CI to call the Task rule, so if developers are using pre-commit hooks
then they should still work, but they're also runnable without using
pre-commit at all.
I think it'd be worth reviewing what we're actually running at
pre-commit in the future because I'm not entirely convinced by all of
the scripts that we are running, but this should at least make what we
have properly enforced and usable by all devs.
# Description of Changes
Currently, `task dev` explicitly calls the backend with
`AIENGINE_ENABLED=true` even though it isn't being spawned, so you just
get a dead FAB in the UI. This PR fixes it so that the engine will only
be enabled for tasks that will actually spawn the engine.
It also fixes a bug with the chat which makes it unusable locally. The
API path was not going through `apiClient` so for local dev you end up
with `//api/v1/...` which is not a valid path, so you get CORS errors
when trying to connect to the AI engine.
# Description of Changes
Redesign policy running so the server is in charge of policy IDs and
running, to make it impossible to have the frontend miss the results.
This solves a minor bug that we currently have in policies, where if you
load a file and then refresh while the policy is running, you'll never
receive the outputted file.
# 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.
# 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 <Ludy87@users.noreply.github.com>
Co-authored-by: Anthony Stirling <77850077+Frooodle@users.noreply.github.com>
# 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
# 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.
# Description of Changes
Kill off agents pane now that we have the FAB. Also fixes a bug with the
FAB where it would sometimes fail to render the chat, and fixes a
duplicated entry in the Vite config which was throwing a warning
# Description of Changes
- Use pool for postgres connections
- Add ability to require user ID to be set on API calls to the engine
- Add process-wide concurrency cap on AI access (in addition to existing
user caps)
- Allow number of workers (threads) to be specified for stirling engine
- Update env var names to reflect that the DB is not just for RAG
# Description of Changes
Remove the Pro guards from the Team settings page and also fix the
styling of the MCP settings screen (the code sections were black text on
black background in light mode)
# Description of Changes
* Improve typing of API (breaking change but unreleased, frontend also
updated in this PR)
* Add ownership concept to policies
* De-AI the comments
* Update the `task dev:saas` rule to spawn the engine as well
# Description of Changes
Add team settings UI to SaaS, which is currently only available in
desktop. It'd be nice to refactor this so they're more shared, but
they're slightly different so needs to be done with some care. Leaving
for followup work.
# Description of Changes
Add new triggers:
- Schedule (fires every X amount of time)
- Folder watch (fires whenever the OS tells us a folder has a new file
in it; on Mac this is technically a 2s schedule but that's just how Java
implements it)
Add new sources:
- Folder (reads from this directory)
Add new sinks:
- Inline (stores in FileStorage)
- Folder (stores in specified directory)
Still want to do S3 buckets and web hooks and stuff, but they can come
in a future PR. I'm hoping this should make it sufficient to be able to
integrate with processing folders frontend etc. I've also changed it so
that policies can have multiple sources and triggers at once, which
seems like it might be useful.
# Description of Changes
Adds all the missing translations that I could find (they're all dynamic
ones that the existing test can't detect are required) and adds a new
test to find as many unused translations as possible. The test has an
ignore list for translations that are used, but dynamically so the test
can't find them (most of the settings UI translations are built up
dynamically like that).
This PR is scoped to just include en-GB translation changes, since
that's the main supported language. We'll need to do a translation PR to
trim all the dead keys from the other languages, and add the missing
ones.
# Description of Changes
Disallow warnings and errors from being thrown in the browser console
during tests unless explicitly expected in the test. Also adds a
Playwright test to prod around some main UI areas and checks that no
warnings/errors have been thrown.
# Description of Changes
The changes in
[#6279](https://github.com/Stirling-Tools/Stirling-PDF/pull/6279) broke
the desktop app because the wasm URL handling didn't deal with
`tauri://` paths. Also I noticed that `task desktop:build:dev:mac`
failed locally because it was attempting to sign the app with
credentials that developers won't have (and shouldn't need), so I fixed
that too.
# Description of Changes
Add a backend for running any multi-step PDF operations. This is
designed to be used for the upcoming Policies feature, along with
anything else that will require automated running of PDF operations,
like the Automate tool or Processing Folders.
The implementation is not complete. I've tried to get all the
infrastructure in there so that we can add in whichever triggers we need
in the future (like cron triggers or watching folders on disk) but
currently it just supports manual triggering of the policy.
The basis of this work was the operation running from the Stirling
Engine, which this PR removes in favour of this new system. The only
currently accessible frontend way to test this work is to ask the AI
chat to execute multiple operations on a PDF, but I've also extensively
tested with direct API calls to make sure that the policies work and
persist properly.
# Description of Changes
Reconfigure linting of `any` type to an opt-out instead of an opt-in
strategy now that we're close enough to everything supporting it. Also
slightly expands the scope of things included in the linting.
# Description of Changes
[#6474](https://github.com/Stirling-Tools/Stirling-PDF/pull/6474)
updated the IndexedDB schema number to v9, but a couple of Playwright
tests were explicitly creating a DB in v4 schema, which then caused
inconsistently failing tests because the DB upgrade process is
asynchronous and sometimes was too slow to upgrade, causing the test to
get into an invalid state.
Also fixes the screenshots directory exclusion since the frontend folder
was restructured.
# Description of Changes
`mockServiceWorker.js` is a third-party managed file, which is included
in our `.prettierignore` file, and is rewritten to be in the module's
standard format whenever `msw` runs. At some point, it was reformatted
in our style, but shouldn't have been. This puts it back to `msw`'s
style, which should make it stop appearing in diffs.
# Description of Changes
Currently, it's not possible to develop the backend on Mac without
manually signing the JPDFium binaries yourself since macOS will reject
running the unsigned binaries. [We've now updated JPDFium to sign the
Mac binaries in
v1.0.2](https://github.com/Stirling-Tools/JPDFium/releases/tag/v1.0.2),
so update to use that version.
# Description of Changes
Change Stirling Engine to support deleting documents automatically. This
happens both on user logout and after an amount of time specified by the
Java when ingesting a document (allowing for personal documents to have
short lifetimes but org documents to be left in the db with no expiry
date). Also sets up an [ACL
policy](https://en.wikipedia.org/wiki/Access-control_list) for the
documents so the database knows which users have access to which
documents. This is not fully implemented in the Java, so currently all
docs are treated as having a single owner, the uploader, but
theoretically when we need to support org storage, we shouldn't need to
change the db schema.
# Description of Changes
Make zoom key command behave the same regardless of mouse position.
Previously only zoomed the editor if the mouse was over the editor.
# Description of Changes
The production SaaS is currently on v8 of IndexedDB due to various
schema changes for Smart Folders, which haven't made their way into OSS.
OSS is currently on v4 of IndexedDB, so if we release an OSS build to
the SaaS deployment, existing users will not be able to use it because
the DB version is 'too old'.
This PR updates the IDB version number to v9 so both OSS and SaaS users
will be able to upgrade to it. Theoretically both types of user should
be able to keep their IDB files without issue. SaaS previously actively
wiped the user's files in an old version (v6/v7) and users who haven't
used it since then will have their DBs wiped, but that'd happen anyway
if they use current SaaS so I don't think that matters.
# Description of Changes
Main fixes:
- Fix the display of the username in the bottom left
- Now displays as "User" when not logged in on self-hosted (desktop) and
"Guest" on SaaS when logged in anonymously
- Now updates properly when the user logs in/out in SaaS, desktop and
self-hosted
- Fix incremental build issues in the desktop app that have been here
since the start (I hope at least - I think the issue is that the JLink
is built read-only and then on subsequent builds you get OS errors when
trying to override the JLink with the new version. There's no real need
for it to be read-only that I know of, so we might as well just make it
R/W and ship like that)
# Description of Changes
Set CI backend & engine comments to auto-delete once the CI has passed.
Also redesign the engine CI to call `task engine:check` like it should
have been, and make it post a comment when the tool models need to be
updated.
Also makes the comment wording more consistent between the three
languages.
# Description of Changes
#6402 introduced a Rust test `refresh_token_fallback.rs`, but it wasn't
moved properly after the restructure of the `frontend/` folder in #6404.
This PR moves the file to the right place, and also hooks up Task and CI
rules for `cargo test` since nothing was actually running the test in
the first place.
# Description of Changes
Various fixes and improvements I made while testing the SaaS code:
- Changes the new `.env.saas` file to live in `app/` and match the
semantics of the other `.env` files
- Adds top-level `task dev:saas` command to spawn SaaS frontend &
backend
- Deletes dead SaaS code and improves some overriding logic
- Fixes refreshing issue when coming back to the tab
- Fix the Compare tool's selection logic
- Make Compare handle error cases properly
- Fixes the location of the "Dismiss All Errors" button (was rendering
on top of the top-bar with a transparent background previously so it
looked rubbish)
- Fixes file selection in PDF Editor
# Description of Changes
Adds a cancel button to the AI chat to allow the user to abort
long-running AI tasks. Just disconnects the SSE stream (all the backend
code already interrupts when it notices the stream is dead).
# Description of Changes
Give Edit Agent access to descriptions of the request from the Java API.
This opens the door to us better documenting our Java APIs to give the
stirling engine better knowledge of what the various tools are and how
to use them.
Also improves the tool selection sub-agent to get the tool parameters
and descriptions so it can more intelligently decide which operations
should be used to fulfil the user's request. Also provides it more
encouragement to string together multiple operations if necessary.
# Description of Changes
Adds storage in the database for full document content alongside the RAG
content (and changes the service to `DocumentService` instead of
`RagService`). Then adds a generic capability that should be usable by
any agent (currently just used by the Question Agent) which allows the
agent to pull out the full contents of the doc, chunks it into various
sections that will fit in the context window, and then processes them in
parallel to create an intermediate result, and then processes the
intermediate result into a final answer. It will re-chunk as many times
as necessary to get the content small enough for the actual answer to be
analysed (I've tested on PDFs ~3500 pages long, which is well above the
context limit and requires maybe 3 rounds of compression to get an
answer).
The new full doc analysis stuff is heavier than the RAG lookup so both
remain. The agents should use RAG for targeted info and the chunked
reasoner for info that requires reading the full doc.
# Description of Changes
#6312 reformatted `tauri.conf.json` via the Gradle script, which
reformats the entire file to not match the Prettier style. This PR
reformats the file back to Prettier format and changes the script to
update the version number without reformatting the entire file.
To be honest I'm not a huge fan of updating the version number with
regexes but it'd be a fool's errand to try and get Gradle to output JSON
in Prettier format, and this seems simpler than shelling out to run
Prettier over the file after the version string has been updated. Any
better ideas, let me know.
# Description of Changes
Hooks up the (alpha) PDF Editor backend to the AI engine Edit Agent via
an intermediary API which is easier for the agent to call. It suffers
from all the same issues that the PDF Editor does in actually editing
the text, but should also benefit from any fixes to that.
It also adds protection against the underlying tools misbehaving by
hanging, and fixes a hanging bug in the PDF Editor.
---------
Co-authored-by: EthanHealy01 <80844253+EthanHealy01@users.noreply.github.com>
# Description of Changes
Save indicator stopped showing up after #6050, which fixed the missing
truncation on filenames, but accidentally bypassed the save indicator
component at the same time. This PR puts the component back in and makes
it support truncation so we can have both.
<img width="586" height="166" alt="image"
src="https://github.com/user-attachments/assets/529c3dcb-ee00-4a6d-ae53-ef8657204369"
/>
# Description of Changes
Vite currently warns that when it's bundling our code that the chunk
size is way too high because most of the imports are static so it can't
split them into smaller chunks. This PR changes a few key areas to use
lazy imports to try and make the chunks as small as possible with
minimal code changes.
Vite's warnings kick in at minified chunks being >500kB, and we've got a
little way to go still to reach that, but we can keep chipping away at
this and I'd rather get the biggest wins done now. I've also included
Lighthouse scores because there's been discussion about improving ours
recently. It's not the aim of this PR to improve it, but it's nice that
it makes it a little better.
## Current main chunks
Build split into 12 chunks. Largest chunk in build is:
```
[frontend:build] dist/assets/index-B6JiWDxZ.js 5,175.51 kB │ gzip: 1,495.85 kB
```
<img width="1442" height="775" alt="image"
src="https://github.com/user-attachments/assets/b0e8a3fa-4ef3-4ccd-8c1d-bfed2d99bd27"
/>
Lighthouse score:
<img width="423" height="146" alt="before"
src="https://github.com/user-attachments/assets/c62056e8-2e77-49a6-a1ae-f08ec8021fb3"
/>
## This PR's chunks
Build split into 176 chunks. Largest chunk in build is:
```
[frontend:build] dist/assets/index-qCgeCY4B.js 2,878.54 kB │ gzip: 861.03 kB
```
<img width="1447" height="776" alt="image"
src="https://github.com/user-attachments/assets/8d0c3cf0-cc25-41c3-b114-4940d3e99349"
/>
Lighthouse score:
<img width="402" height="145" alt="after"
src="https://github.com/user-attachments/assets/99a26eb3-bd15-4b92-bf22-82b58b458f52"
/>
---------
Co-authored-by: EthanHealy01 <80844253+EthanHealy01@users.noreply.github.com>
# Description of Changes
Have the Java send a list of enabled endpoints to the AI engine so it
can intelligently respond to the user that the tool does exist but is
disabled on the server so it can't acutally run the operation, instead
of the current behaviour where it sends the API call back and then 503
errors because the execution fails when the URL is disabled.
<img width="380" height="208" alt="image"
src="https://github.com/user-attachments/assets/5842fb2e-2e55-45a5-8205-25515636daae"
/>
---------
Co-authored-by: EthanHealy01 <80844253+EthanHealy01@users.noreply.github.com>
# Description of Changes
Flesh out the RAG system and connect it to the PDF Question Agent so it
can respond to questions about PDFs of an extremely large size.
I'd expect lots more work will need to be done to finish off the RAG
system to really be what we need, but this should be a reasonable start
which will let us connect it to tools and have the ingestion mostly
handled automatically. I'm leaving file deletion and proper file ID
management to be done in a future PR. We also need to consider whether
all tools should retrieve content exclusively via RAG, or whether it's
beneficial to have tools sometimes fetch the direct content and other
times fetch it from RAG.
A diagram of the expected interaction is as follows:
```mermaid
sequenceDiagram
autonumber
actor U as User
participant FE as Frontend<br/>(ChatPanel)
participant J as Java<br/>(AiWorkflowService)
participant O as Engine:<br/>OrchestratorAgent
participant QA as Engine:<br/>PdfQuestionAgent
participant RAG as Engine:<br/>RagService + SqliteVecStore
participant V as VoyageAI<br/>(embeddings)
participant L as LLM<br/>(Claude / etc.)
U->>FE: types "Summarise this PDF"<br/>(PDF already uploaded)
FE->>J: POST /api/v1/ai/orchestrate/stream<br/>multipart: fileInputs[], userMessage
Note over J: ByteHashFileIdStrategy<br/>id = sha256(bytes)[:16]
J->>O: POST /api/v1/orchestrator<br/>{ files:[{id,name}], userMessage }
O->>L: route via fast model
L-->>O: delegate_pdf_question
O->>QA: PdfQuestionRequest
loop for each file
QA->>RAG: has_collection(file.id)
RAG-->>QA: false
end
QA-->>O: NeedIngestResponse(files_to_ingest)
O-->>J: { outcome:"need_ingest", filesToIngest:[...] }
Note over J: onNeedIngest
loop per file
J->>J: PDFBox: extract page text
J->>O: POST /api/v1/rag/documents<br/>(long-running timeout)
O->>RAG: chunk + stage documents
O->>V: embed_documents (batches of 256)
V-->>O: embeddings
O->>RAG: add_documents
O-->>J: { chunks_indexed: N }
end
Note over J: retry with resumeWith=pdf_question
J->>O: POST /api/v1/orchestrator
Note over O: fast-path to PdfQuestionAgent
O->>QA: PdfQuestionRequest
Note over QA: build RagCapability<br/>pinned to file IDs
QA->>L: run(prompt) with search_knowledge tool
loop up to max_searches
L->>QA: search_knowledge(query)
QA->>V: embed_query
V-->>QA: query vector
QA->>RAG: search(vector, collections=[file.id])
RAG-->>QA: top-k chunks
QA-->>L: formatted chunks
end
Note over QA: once budget spent,<br/>prepare() hides the tool
L-->>QA: PdfQuestionAnswerResponse
QA-->>O: answer
O-->>J: { outcome:"answer", answer, evidence }
J-->>FE: SSE "result"
FE->>U: assistant bubble
```
# Description of Changes
Adds the ability for the Edit agent to request the content of the
document before it decides which parameters it needs. This makes it able
to process requests like `Split the document after the page containing
the "My Section" section`, allowing for document context-based requests
for all[^1] tools.
I had to make a few changes elsewhere to make this work, including:
- Moving the requesting of content out of the Question Agent and into a
common location
- Added specific API docs for the Split param because the generic ones
were not specific enough for the AI to be able to reliably perform the
correct operation
- Fixed an issue in the tool models generator which caused the Redact
params to only be half-generated (causing Pydantic to crash when the AI
tried to run Redact)
- Added missing logging to a bunch of tools and hooked it up properly so
it'll print to stderr
- Made the limits for the max pages/chars to extract from PDFs
configurable via env var
[^1]: Many of the tools can't actually do anything useful with the
context at this stage, but will just need the tool API to be extended
with new features like page-specific operations to be automatically able
to do smart operations without needing to change the Edit agent itself.
# Description of Changes
We keep adding stuff to `engine/config/.env.example` and have to
manually update `.env` because of it, which is really clunky, especially
when working on multiple worktrees at once. This PR changes it so that
we just have a committed `.env` file and have an `.env.local` override
to put the actual private keys into, which should make it a bit easier
to manage.
> [!warning]
>
> After this goes in, be very careful for a little while not to
accidentally commit any keys that you've got inside your `.env` file!
# Description of Changes
Add an extra parameter to every agent to receive the conversation
history in addition to the current message. This will make it possible
to answer followup questions from the AI without needing to give full
context in your message.
# Description of Changes
Redesign AI engine so that it autogenerates the `tool_models.py` file
from the OpenAPI spec so the Python has access to the Java API
parameters and the full list of Java tools that it can run. CI ensures
that whenever someone modifies a tool endpoint that the AI enigne tool
models get updated as well (the dev gets told to run `task
engine:tool-models`).
There's loads of advantages to having the Java be the one that actually
executes the tools, rather than the frontend as it was previously set up
to theoretically use:
- The AI gets much better descriptions of the params from the API docs
- It'll be usable headless in the future so a Java daemon could run to
execute ops on files in a folder without the need for the UI to run
- The Java already has all the logic it needs to execute the tools
- We don't need to parse the TypeScript to find the API (which is hard
because the TS wasn't designed to be computer-read to extract the API)
I've also hooked up the prototype frontend to ensure it's working
properly, and have built it in a way that all the tool names can be
translated properly, which was always an issue with previous prototypes
of this.
---------
Co-authored-by: Anthony Stirling <77850077+Frooodle@users.noreply.github.com>
Co-authored-by: EthanHealy01 <80844253+EthanHealy01@users.noreply.github.com>
# Description of Changes
Follow on from #5949, expanding any type usage ban to the `desktop/`
folder
Also gets rid of a bunch of really verbose desktop logging that I don't
think we really need anymore (or ever needed tbh, most of it doesn't
make sense) because it was using a bunch of `any` typing and wasn't
worth fixing.
# Description of Changes
When I added Prettier formatting in #6052, my aim was to use just the
default settings in Prettier. Turns out, Prettier looks _really hard_
for any config files if it's not explicitly given one, which means that
if a developer has some sort of Prettier config file lying around on
their system, Prettier might find it and use it. Also, Prettier changes
its defaults based on stuff in `.editorconfig` without any good way of
disabling that behaviour explicitly in its config file.
To solve both of these issues, I've introduced a `.prettierrc` file
which sets Prettier's defaults explicitly, and then reformatted all our
code _again_ in Prettier's actual default settings. This should achieve
the aim of #6052 and remove the possibility for it breaking on different
dev computers.
# Description of Changes
Adds a streaming endpoint to the Java AI orchestrator
(`/api/v1/ai/orchestrate/stream` in addition to the existing
`/api/v1/ai/orchestrate`). This allows the caller to get updates of what
stage of orchestration is being run at the time so UIs can give the user
feedback.
Also contains some dubious Gradle changes to suppress errors coming from
Spotless, when it crashes in Google stuff. I'm not sure if that's
appropriate to add, feel free to ask for changes in review.
# Description of Changes
Changes the strategy for autoformatting to reject PRs if they are not
formatted correctly instead of allowing them to merge and then spawning
a new PR to fix the formatting. The old strategy just caused more work
for us because we'd have to manually approve the followup PR and get it
merged, which required 2 reviewers so in practice it rarely got done and
just meant everyone's PRs ended up containing reformatting for unrelated
files, which makes code review unnecessarily difficult. If the PR's code
is not formatted correctly after this PR, a comment will be added
automatically to tell the author how to run the formatter script to fix
their code so it can go in.
This also enables autoformatting for the frontend code, using Prettier.
I've enabled it for pretty much everything in the frontend folder, other
than 3rd party files and files it doesn't make sense for. I also
excluded Markdown because it sounds likely to be more annoying to have
to autoformat the Markdown in the frontend folder but nowhere else. Open
to changing this though if people disagree.
> [!note]
>
> Advice to reviewers: The first commit contains all of the actual logic
I've introduced (CI changes, Prettier config, etc.)
> The second commit is just the reformatting of the entire frontend
folder.
> The first commit needs proper review, the second one just give it a
spot-check that it's doing what you'd expect.
# Description of Changes
Add prototypes folder to test new functionality in. This build of the
app is spawnable with `npm run dev:prototypes`.
Currently just contains a very developer-y chat interface to help us
develop & explore the AI backend before we make the frontend for it for
real.
# Description of Changes
Add Java orchestration layer which can connect and go back and forth
with the AI engine to get results for the user. It's expected that the
AI engine will not be publicly available and this Java layer will always
be in front of it, to manage sessions and auth etc.
# 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
> frontend@0.1.0 dev
> npm run prep && vite
> frontend@0.1.0 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.
James Brunton