Files
windmill/frontend/src/lib/components/Editor.svelte
T
3dae553576 feat: pipeline editor: assets-only view, Fix actions and draft triggers (#11440)
* wip: one sidebar for operators and developers, with a detach mode

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: snap-detach the operator sidebar, and join the handle to the picker

Dragging the sidebar into the left snap zone detaches it for an operator, who
has no icon-only rail to collapse into; the width the drag started from is what
gets persisted, so re-attaching restores the sidebar's size.

The handle's colour pastille slides onto the card's workspace pastille as the
card opens, alongside the workspace name, and both buttons that attach or
detach the sidebar carry a tooltip.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: attach and detach from the sidebar header, open the card under its handle

The button that swaps the sidebar between docked and detached moves to the
header, next to the workspace name, so it stays in one place across the swap;
the icon-only rail has no room for it and keeps the footer copy.

The handle itself becomes a component of its own, carries the workspace icon
its picker uses, and opens the card rather than docking — docking is the header
button's job. The card is placed under the handle measured at open time instead
of at a fixed offset.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: one page header across routes, with the workspace picker in it

Every route now wears the same bar above its content: the sidebar toggle, the
workspace picker, and whatever the route registers — an item's breadcrumb and
summary, or a section name — with the route's own buttons at the far end. The
bar's row is level with the sidebar's own top row, which the workspace ⇄
sessions switch now owns, so the two read as one band.

A route fills the bar by mounting PageHeaderContent, which registers a getter
rather than a snapshot: the layout calls it while rendering, so a path or
summary the page edits reaches the bar without an effect keeping two copies in
step. Nothing registers yet; the routes follow.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: detail pages fill the page header, which now spans above the sidebar

A script or flow detail page no longer draws a bar of its own: its path becomes
the header's breadcrumb, its summary keeps the rename-and-labels popover next to
it, and its badges, menu and buttons ride along as the header's actions. The
band runs the full width with the sidebar hanging below it.

A snippet rendered in the header initialises its components under the header,
not under the page that wrote it, so anything the page's tree provides is lost —
the detail page's buttons read TriggerContext and threw. ContextBridge carries
the entries a registration names back around the render.

The floating handle is gone with it: the header's own toggle is the one control,
hovering it peeks the sidebar in and clicking pins it. The workspace picker in
the header is a denser variant of the sidebar's, and the card's outside-click
catcher starts at the card rather than the viewport top, so it no longer
swallows clicks meant for the toggle.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: one breadcrumb in the page header, and the banners that used to sit under it

The header band now carries a single breadcrumb — workspace, fork, then the item's
path — built from one component, so every segment has the same size, weight and icon
size, and slashes separate them. The item's kind icon rides with its summary, after a
dot, rather than with the path.

The fork and drafts banners move into the band as badges plus a subtle Review button:
ahead/behind counts instead of a bare change total, and the per-kind breakdown stays on
the compare page. The home hero's CLI, MCP and Hub links move into an ellipsis menu,
alongside the toggle that pins the session chat to the page.

The flow editor and the app viewer hand their bar to the band too, and a route can drop
the band's bottom border when its own content draws one — the border stays in the box
with a transparent colour, so contents never shift a pixel across a navigation.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the runs page's own header row moves into the page header

The title, the queue counters, the kind and status filters, the schedule and
future-job switches, the search field and the timeframe picker all ride the band
now, and the page starts at the chart — one row of vertical space back.

The two toggle groups appear from 1500px: below that the bar cannot hold them next
to the search field, and the same filtering is reachable from the search field's own
kind and status chips. The search field is what gives width back, and the breadcrumb
yields last, so a page's own name stays whole while the workspace and fork names
truncate.

The queue counters keep their icon form at every width, since the bar has room for
two counters and never for the two labelled blocks. Their counts hang outside their
icons, so the group carries the margins that keep a badge off its neighbour, and the
suspended counter answers a hover at zero rather than sitting there as a dead icon.

The band is 44px rather than 40, and a page can hand the breadcrumb a hint — the
documentation tooltip that belongs with the page's name rather than with its buttons.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: no border under the page header, and the peek card closes on the way out

The band draws no bottom edge and no page asks it to, so `barBorder` goes with it —
a page that needs the separation draws its own line, as the detail pages now do above
the run form and the tab strip. With nothing toggling a border, the band is a flat
44px on every route.

The detached sidebar's card is a hover affordance, so it leaves when the pointer does:
off the handle and off the card both, on the same debounce a hover menu uses, because
the two do not touch and the leave of one fires before the enter of the other. The
mobile drawer is opened deliberately and still stays until it is dismissed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: one band, one workspace part, and the pickers that hang off it

The band is a single element whose placement is styled rather than branched: a page
registering a side-panel inset used to swap it between two `{#if}` arms, destroying the
header and rebuilding it a frame later — a blink of no header with the content sliding
up into the gap.

The workspace and its fork become one breadcrumb part: the name leads home (inert in a
session, where the workspace is the session's scope rather than a place to go) and one
chevron opens the picker, which already expands a family to reach its forks. The
picker's panel hangs from the part's left edge, opens with the active family expanded,
and closes before the switch runs — the switch renames the workspace, which moves the
trigger the panel hangs from, and floating-ui was chasing it across the viewport.

The item's path is a plain label rather than a row of pickers: same separators and
weight, and a click copies it with its own confirmation. The kind icon goes, the page
below says what kind it is.

Around them: a "prod" badge for a root workspace, the session's acting fork beside the
session's name rather than in the breadcrumb, the edit-in-dev affordance as a button in
the header, the drafts count as a sentence, and a fade when the workspace changes under
a page that stays mounted.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: keep a page's own controls when the workspace navigation is hidden

`menuHidden` means "this is an embed, don't show the workspace's navigation" — but
Runs, the detail pages and the flow editor now keep their controls in the band and
nowhere else, so hiding it took the filters, the timeframe and the page's buttons with
it. A session's preview iframe opens exactly that way. The band now renders for an
embed too, without its breadcrumb, so the page keeps its controls and the embed still
gains no way to navigate the host's workspace.

The flow editor's autosave indicator moves into the header's actions with the rest of
its bar: saving state, the autosave toggle and other people's drafts were left behind
in the branch the full-page editor no longer takes.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the header carries what it promised, and nothing it does not

The registry's item described pickers the breadcrumb no longer has: nine fields — the
editability flags, the change callbacks, the navigation hook — that nothing read, while
the flow editor passed eight of them and quietly lost its inline rename. The type now
says what the bar does: a path, and either a summary or the page's own control for one.
The flow editor renames through that control, as the detail page already did.

`barPlacement` went the same way: no page registered 'inline', so the layout's guard
never fired.

The sessions page registered an inset of 0 whenever its preview pane never mounted —
the "no session selected" and opted-out arms — and 0 reads as "a page owns the right
edge", floating the band over the page's first line. It registers an inset only once
the pane has a width.

The band's sidebar handle docks through `setDetached` like the card's own toggle, so
docking still suppresses the rail's entry animation and closes the card; a summary
without its breadcrumb no longer leads with a stray separator in an embed; a page's
hint sits with its name whichever way the page named itself; and `Menu.closeInstantly`
and `EditorHeader`'s `inline`/`hideSummary` go, their callers having been replaced.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: a deployed app gets the viewport, and the header waits in the corner

An app is the page, not a page inside Windmill's chrome, so the band leaves the flow
and floats above the viewport until the handle in the app's corner calls it down. The
handle is faded until the pointer nears it, hovering peeks the band over the app, and a
click pins it — one flag for every deployed app, in this browser, because someone who
wants the breadcrumb over a dashboard wants it over the next one too.

Peeking overlays rather than pushes: the app is an iframe, and taking 44px off its box
on every pass of the pointer would re-lay-out whatever it draws. Pinned, the band is in
flow like any other route's and the app starts under it.

The sidebar is untouched — docked stays docked, detached stays a peek — since the pin
is about the header alone. A page says it wants the viewport with one registry field,
so the layout keeps asking the registry rather than knowing about apps.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: one corner control for the chrome an app page hides

Hovering the button overlaid on a deployed app brings back whatever that page is
hiding: the band always, and the sidebar card too when the sidebar is detached. The
card hangs from this button rather than from the header's own handle, which is
off-screen with the band.

The handle in the band no longer attaches the sidebar — it reveals it and nothing more.
A control that hides the thing it sits on cannot also be the way to bring it back, so
attaching stays with the toggle in the sidebar's own footer, where detaching happens.

The band's two buttons go in the order the things nest: the sidebar's handle, then the
header's pin. The overlaid button stands where the pin will be once pinned — same 28px
box, and a slot further along when the sidebar's handle takes the first — so pinning
moves nothing. It wears the panel pair the sidebar's toggle uses: solid while the band
is away, dashed while it is shown.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the access point sits in the corner, and the header covers it

The button is in the page's top-left corner — 8px in, or just past the rail when the
sidebar is docked — rather than shifted along to where the header's pin would land.
Hovering brings the header down over it: an access point that the regular view covers
once it arrives, so there is nothing to line up and nothing to move.

Because the header lands on the button, the pointer leaves it the instant it arrives.
The band holds both itself and the sidebar card open while the pointer is anywhere on
it, or the card would retract under a header that stayed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: on an app page the floating button owns both, and they move together

The band drops its sidebar handle there: the floating control in the corner reveals the
sidebar along with the band, and two handles for one gesture is one too many. The two
are then linked — the pointer resting on either holds both, and leaving either collapses
both — so they behave as the single thing the one button revealed.

The card meets the header in a line: it hangs from the header's bottom edge rather than
from whichever control opened it, which had it starting a couple of pixels high. With no
header to sit under — an embed — it still hangs from the control that reached for it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the revealed sidebar is a panel under the header, not a card over the page

No rounded corners, no inset from the window: it starts at the header's bottom edge and
runs to the left and bottom edges, with one border down its right side and a shadow to
lift it off the page. A card floating in from nowhere read as a separate thing; this
reads as the sidebar the header sits on top of.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: only an app page gets the flush panel; elsewhere the card still floats

Over an app the sidebar reads as the panel the header sits on — flush left and bottom,
square, one border down its right side. On every other page it is what it was: a card
inset from the window with rounded corners and a border all round.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: an app opens with the header already away

Opening an app showed the header and then slid it out: the page claimed the viewport on
mount, and the band animated from where it had been rendering. The slide belongs to the
hover, not to arriving, so it is armed a frame after the claim — the first hide is a cut.

The claim moves to the route, which mounts before the viewer waiting on the workspace to
resolve. A cold load now paints the band away from the first frame instead of showing a
header that leaves again.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: an app's Edit sits with its name

At the far end of the bar it was a journey across the window from the thing it edits —
on a page whose header is only there while the pointer holds it, a journey with a time
limit. It goes beside the app's path instead.

The slot it uses is the one Runs' documentation tooltip already had, renamed from `hint`
to `afterName` and documented for both: what belongs to the page's name, whether that is
a mark or a control that acts on what the name points at.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: developers collapse, operators detach

One control each, rather than three for everyone. A developer switches the sidebar
between icon-only and full width and never detaches it; an operator, who has no
icon-only rail to collapse to, hides it instead — by the button in its footer or by the
same drag that collapses it for a developer.

The way back survives either way: a detached sidebar keeps its "Show sidebar" button for
anyone, so a developer who detached before this was so is not left with a sidebar they
cannot return.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: configure pipeline triggers and outputs as drafts in the editor

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: one card for pipeline nodes, and the pipeline pages in the page header

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: drop a removed node's trigger drafts, and confirm triggers when leaving

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: show pipeline assets red when unused or awaited but never produced

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: route long pipeline edges around nodes in the target column, and flag scripts waiting on unproduced assets

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: a page is as tall as it is given, not as tall as the screen

The band takes 44px off the top of every route, so a page sized to the
viewport overflows by exactly that much: h-screen inside a column that
already lost the band's height. These pages measure the box they are in
instead.

CenteredModal gains `inPage` for the five pages that render it inside the
layout rather than over it — the same modal, centred in what is left of
the page rather than in the window.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the band narrows, wraps, and draws its own edge

The band publishes its width, so the pages filling it can answer it. Two
tiers: under 1000px the breadcrumb keeps the workspace disc and its env
badge and drops the name, and under PHONE_BAR (800) the trail stops
yielding at 10rem — that floor is what makes the bar wrap, so a page's
buttons take the line below rather than squeezing the name to an ellipsis.

The band also draws the line between itself and the page, and only while
something is scrolled under it. One capture listener on the window sees
every scroller, since most pages scroll inside their own box and never
move the layout's. An inset shadow, not a border: a border would move
every page down a pixel the moment the line appeared.

DetailPageLayout and InWorkspaceAppViewer drew that first line themselves;
two lines would stack, so theirs go.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the script, app, runs and audit pages fill the band

Each of these kept a header of its own above the page. They register with
the band instead: the breadcrumb names what the page is on, and the page's
buttons sit at the far end of the one row.

They answer the band's width rather than the window's, since the buttons
share that row with the breadcrumb. The script editor folds Diff and
Settings into the menu it already shows when compact, the app editor drops
the panel buttons, Awareness and Export and goes icon-only on Deploy and
the mode switch, the flow editor menus its diff, and runs hides its
counters. Runs also declares `actionsFlexible`: its search is the one
control here that can give width back, and a box that refused to shrink
would make it take its maximum instead.

The run detail page loses its ActionRow, and the audit log its big title —
the breadcrumb says where you are, and the page keeps its tooltip.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: one sidebar column, shown as a rail or as a panel

The rail and the drawer held two copies of the same column — the same
sections, in the same order, in two pieces of markup, so a change to one
silently skipped the other. Both now render `sidebarColumn`, and the
shells keep only what differs: the rail its width, resize handle and
collapse button, the panel its outside-click catcher and its transform.

A phone gets that panel rather than a modal. There was a width test
deciding between two presentations; the card under the band's handle works
at any width, so it goes. What is left of the width test is the choice
between a rail and a panel, plus a `(hover: hover)` test — the old
`innerWidth >= 768` was standing in for "has a pointer", which is what
decides whether the panel closes when the pointer leaves it.

With the card everywhere, the backdrop, the `aria-modal` (nothing dims and
nothing traps focus) and the drawer's close cross all go: the band's
handle opens the panel and a tap outside closes it. The burger row goes
with them. Entering panel width now collapses the rail transiently, so
coming back to a wide window does not land on a rail the page was resized
away from.

The band also stops easing its right edge. That transition belongs to the
rail, which moves the band's left; its right follows a splitter the page
owns, and easing it left the band lying over a session's side panel — a
blank strip across that panel's own header for 200ms after every drag.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the home composer flies to the menu that restores it

Removing the composer puts it away rather than throws it out, and the way
back is one item in the band's menu — easy to miss on a page the composer
has just vacated. A mark flies from the control that removed it to that
menu, and the menu answers with one ring as the mark dissolves, so where
it went is witnessed rather than explained.

The cross and the menu item now read "Remove session chat", one name for
one thing whichever a reader reaches for.

Reduced motion skips both.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: give pipeline asset nodes the trigger nodes' background again

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: taller pipeline nodes, with the asset + back on the node's edge

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: judge dbt relations by provenance, keep dynamic_skip on reopen, refresh saved scripts' titles

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: label the pipeline + buttons: add data source, add downstream step on hover

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: name a schedule node's cadence, like "Every day at 12:00"

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: keep the downstream-step pill open while its menu is, without flicker

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: room between pipeline node chips and the node's edge

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: refuse renaming a pipeline draft out of the pipeline's folder

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: title pipeline schedule nodes with the schedule's summary

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* revert: keep operators out of AI sessions

That change is its own PR; this one is about the navigation. `prefersSessionHandoff`
takes the role again and every entry point passes it, the sessions page refuses an
operator arriving by URL, and the layout leaves them on the docked chat.

Two clauses are new rather than restored, because the nav rework is what gives an
operator a sidebar at all: the Workspace/AI Sessions switch and session mode are both
hidden from them, since the page behind them declines. `askAiShown` gains the operator
case for the same reason — they keep the docked pane whatever the beta gate says, so
the sidebar has to offer a way into it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: assets-only view of a pipeline, with producer and trigger chips

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: default schedule handlers, unique trigger draft paths, zoned cadence, valid script names

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: setup banner without icon, button actions, plainer activity rows

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: assets-only trigger chip under the title, editable and selectable

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: await schedule defaults before confirming, refuse colliding trigger renames

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: size assets-only cards to their content

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: pipeline draft sync status beside the folder name in edit mode

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: autosave edits to a deployed pipeline script from the first keystroke

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: show schedule handlers once their defaults load, draft suffix on static chips

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: keep the pipeline folder while the page tears down on leaving

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: discard the open pane's autosaved edits with Discard all

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: pipeline node context menus, add-at-cursor, and assets-only delete

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* style: borderless pipeline cards on the main surface, like flow steps

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: remember which pipelines are in the assets-only view

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: edit an asset's producing script from its pane and multi-script chip

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* style: more room between pipeline layers for the bottom + button

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* style: background on the trigger chip only when it opens an editor

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: assets-only delete in both modes, why it is blocked, and a Show scripts toggle

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: remove the script before its trigger, archive it for non-admins

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: scope assets-only memory by workspace, no delete for operators, flow reason

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* style: fly in the pipeline node menus

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* style: size the Show drafts toggle like the other header buttons

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: a No asset node for scripts that build nothing, in the assets-only view

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: describe a pipeline edge's relationship on hover

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: name assets-only edges by what is built from what

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: open an asset's upstream trigger from its pane header

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: reopen the edit-mode pane when coming back from View

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* style: no red border on errored pipeline nodes

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: Fix actions on misconfigured pipeline nodes, and trigger toggles on edges

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: folded errors and edge toggles in assets-only view, preset writer wizard

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: flag pipeline scripts nothing starts, with schedule/webhook/remove fixes

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: an unwritten subscription is fixed on its script only, by removing it

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: hide subscribed-only unwritten assets, review fixes for the edit pane and edge actions

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: an unwritten subscription is the subscriber's error, not the asset's

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: a script only started by reads of unwritten assets is flagged too

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: no add-downstream + on assets nothing in the workspace writes

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: Save all treats a draft identical to the deployed script as saved

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: remove-from-View opens its dialog, error chip fits, no link to an undrawn asset

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: drag a script's dot onto an asset to subscribe it; unbuilt asset opens its script

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: toggle to hide reads that don't trigger runs in the pipeline graph

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: drag an asset's + onto a script or the canvas; tell reactive assets-only edges apart

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: a programmatic on/mute line joins the existing ones in the header

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: address review — live-version save check, bounded writer lookups, script-only trigger writes, dbt sources, hidden pane

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: review — compare every deployed field, passive-read filter only with its toggle, drag cancel, asset escape hatch

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: keep every control the band took over reachable

Four controls the page-header conversion could put out of reach, each found by a
review pass and reproduced in the browser:

- The flow editor's preview buttons. `topbarWidth` is bound only in the branch a
  full-page editor does not render, so `compactTopbar` was permanently false
  there; below the 800px phone bar `phoneTopbar` dropped the buttons inline and
  the graph overlay never took them, leaving FlowPreviewButtons mounted nowhere
  and step testing, whole-flow testing and "Test flow & record" no-ops through
  optional calls. It now measures the header the way ScriptBuilder and
  AppEditorHeader do, and the inline and overlay conditions are exact
  complements.
- The pin on a deployed app. The band lands over the corner handle that called
  it down, so the pointer that arrives there hits the band. The band carries the
  pin itself now, on the same pixels, in place of the sidebar handle it hides.
- The detached sidebar's own menus. They portal to `body`, so moving into one
  left the card and closed it 200ms later, sliding it out from under the menu
  being read. The card holds while the pointer is in a menu and takes the close
  from the leave of that.
- A deployed app's name for a read-only viewer. The breadcrumb was registered
  behind write access, so without it the trail fell back to the section name
  "Apps"; and the Edit button rode `afterName`, which the band drops for an
  embed along with the trail. The name is registered for whoever opened the app,
  and what sits after it survives `navHidden` as the page's own control.

navPageFor gets a table test: it answers both the sidebar links and the
breadcrumb label, and its longest-prefix and first-segment rules are what made
the "Apps" fallback look like a correct answer.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: drop what the operator revert and this refactor left behind

The operator tour's last step named AI sessions among the pages the sidebar
opens, which operators have no way to reach. The nav handle's comment named the
AI chat layout's menu button as a second way into the card; that button and its
prop are gone, and the corner control on a page that hides the band is what
stands in for the handle now.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* style: prettier the three files this branch left unformatted

`prettier --check` fails on all three at the previous head. Whitespace only,
apart from one Button tag that fits on a line.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: assets-only nodes name their script in a header tab instead of a side chip

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: no loading spinner in the pane while a clicked script loads

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: no subscribe drag dot in view mode

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: opening a draft no longer autosaves; header tab styling, selected edge, taller cards

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* revert: clicking an unbuilt asset opens the asset again, not its script

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: header tab sets card width, edit icons on script header and trigger chip, no pane header shift on open

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: no trash in the pane header; card width ignores the run chip so a click never re-lays the graph

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: taller trigger chip; node menu and guard badge on the card corner, below the header

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: give a page without the band's chrome what the band was carrying

Second review round, three reachable defects plus one decision:

- A public raw app was 150px tall. The bundle's iframe is `h-full`, and dropping
  its `min-h-screen` left nothing to resolve against on the share, embed and
  custom-path routes — none of which has a layout above them. Measured: 150px in
  a 900px viewport. The floor goes back for those routes and stays off inside the
  workspace, where the layout hands a height down and `min-h-screen` would
  overflow it by the band.
- The band retracted out from under its own open menu on a deployed app. Last
  round taught the sidebar card to hold while the pointer is in a portaled menu
  and left its sibling alone, but the band opens menus too — the workspace disc,
  the fork family, every path level. One helper now, used by both closers.
- `bandPeeked` outlived the page it belonged to. Clicking a control in the band
  navigates away, and the next page may not be one the band floats over, so the
  hover handlers that would clear it are gone — leaving the band already down
  when a full-bleed page came back. `beforeNavigate` clears it.
- An embed has no band to carry a control, so the app's Edit floats over the
  canvas there, where it sat before the band existed. Registering it as the
  band's `afterName` under `navHidden` instead would have cost the app 44px of
  the iframe for a row holding one button.

Reading "is the workspace navigation hidden" now has one definition
(`isMenuHidden`), since a page has to answer it too: the flag is sticky inside an
iframe, so the raw query param is the wrong thing to read.

Also inset the band by the rail only when there is a rail — an embed reserved
its width and crowded its own controls into a compact layout early.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: a path that names nothing yet has no breadcrumb leaf

`splitItemPath` documents that it gives up on a path still being typed, but
`f/demo/` is three parts and returned a leaf with an empty name — a blank
segment in the trail. The editor binds the path live, so the breadcrumb sees
every keystroke.

Tested here rather than `navPageFor`'s sort: no two `NAV_PAGES` paths nest, so
the longest-prefix case that test claimed to pin has only one candidate. Renamed
it to what it actually checks.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* style: prettier the four remaining files this branch left unformatted

All four fail `prettier --check` on the branch and pass at the merge base.
`raw_apps/utils.ts` also fails but already did at the base, so it is left alone.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: pipelines open in the assets-only view by default

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: a No asset node per script, banner-only when writes start it; no On asset change chip

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: delete a node's script, trigger, or both from its menu; red header for a script with errors

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: an errored script header stays red when selected

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: coloured run badge border; per-script delete when no combined delete; run button centred on the card

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: Add data source pill sits one small gap above the graph

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: Multiple triggers chip picks the trigger to edit; each folded trigger deletable from the node menu

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: add-step menu labels its path field Path

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: iterate NodeLists with Array.from so svelte-check passes

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: no trigger delete for flow-built assets; Multiple triggers chip shows the one being edited

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: Archive button for archive-only deletes, single delete doc, inert Multiple triggers label in View

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: the raw app editor fills the band like the other three

It was the one editor route still rendering its own bar under the band, so it
showed two header rows and the band named the app "Apps raw" — `navPageFor`'s
first-segment fallback, because nothing on the route registered an item. Its top
bar now becomes the page header the way the script, flow and low-code app
editors' do: the path and summary are the breadcrumb's, the file-sidebar toggle,
autosave cloud, presence mark, menu, Diff, Jobs, export, AI and Deploy are the
header's actions, and the path is a plain label since this editor renames from
its deploy drawer.

The actions render under the band, outside the subtree `RawAppEditor` declares
the operating workspace in, and `OpenInSessionButton` asks that context who the
acting user is — so the workspace travels with them.
`operatingWorkspaceContextEntry()` hands it over without exporting the key.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: keep the folders a half-typed path already names

The trailing-separator guard gave up on `f/demo/` entirely, which dropped the
whole trail to the kind chip and brought `f/demo` back on the first character of
the name. The levels the path does name come back now; only the leaf waits.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: pin holds the first slot, and the nav table says what it shares

Pinning brings the sidebar's handle back, and from behind it the unpin button sat
one slot over from where the pin had just been clicked — so a second click in the
same place opened the sidebar instead. An operator is detached by default, which
makes that the default state of the feature rather than an edge case.

`NAV_PAGES` claimed one table stops a page being called one thing in the nav and
another in the header, which holds for the seven entries the sidebar's main group
is built from and not for the four grouped separately — `/routes` was "Triggers"
in the header and "HTTP" in the nav. The comment now says which is which, and
that entry is named for the page rather than its group.

`sidebarPageAllowed` gets a table test: it decides what an operator sees, and its
`=== true` is what a future `Option`-style default would silently change.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the band names the path Deploy will create, and keeps the summary readable

A brand-new app is parked at a `draft_<uuid>` placeholder, and the band named
that rather than the path its Deploy will create — `newEditedPath` refuses the
placeholder for exactly this reason, and the editor's own bar read it. Both app
editors register it now. Verified: the trail reads `u/admin/gentle_app` while the
URL is still the draft.

The band's summary gets the `title` the editor's own bar had, so one the bar
truncates can still be read on hover.

The wrapper that hid AppExportButton on a phone never hid anything:
the component draws no trigger — it is the export drawer, which portals to
`body` — so the class toggle was inert in both states and the comment read as if
something moved. Dropped from both app editors.

Two comments corrected: `NAV_PAGES`' `icon` does reach the breadcrumb for all
eleven entries (`NavBreadcrumb` renders `routePage.icon`), so only `operatorKey`
goes unread for the four the sidebar groups separately; and the raw editor's
phone note described folding that does not happen — the file-sidebar toggle stays
in the band, Jobs and the export are already in the menu, and AI is what stands
down.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: an item the picker files under no kind keeps its name in the band

The merge brought main's agents detail page, which reaches DetailPageHeader with
no `errorHandlerKind` — and the band's item required one. `WorkspaceItemKindIcon`
falls through to the script glyph for anything it does not know, so defaulting
would have labelled an agent a script; the kind is optional now and the trail
names the item without claiming it is one of the three.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the file sidebar's toggle leaves the band

The band is the workspace's header, not the raw editor's layout controls. ⌘B
still folds the file sidebar, and the editor's own bar keeps the button where it
draws one — a session pane.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the scroll edge stops measuring, and the comments describe the code

The band's edge listener captures every scroller in the app — Monaco, the audit
table, a run's logs — about once a frame while any of them moves, and it took two
`getBoundingClientRect()` reads before looking at `scrollTop`. It now answers from
`scrollTop` first and only measures when that answer differs from what the band
draws, which is the transition rather than the steady state. The rect check still
authorizes every change, so a scroller that does not start at the content's top
still cannot claim the edge.

The two agent routes main just added size themselves to the viewport, which is
44px more than the box the band leaves them. They measure 856 either way — the
layout's `min-h-0` column clamps them — so this is the sweep the rest of the
branch already had, making the height follow the box rather than depend on a
parent's clamp.

Six comments described the conversion rather than the code, which AGENTS.md calls
out by name. Each now states the constraint a reader could break: why the summary
keeps its popover here, why the audit hint is `afterName` and not an action, why a
run registers a `section` rather than an item, why the Runs filters are
`actionsFlexible`.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* Revert "fix: the file sidebar's toggle leaves the band"

This reverts commit 1da8d16740.

A folded file sidebar renders nothing of its own, so the toggle on the bar is the
only way back short of knowing ⌘B.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the editors name their item in the band, and a pen renames it

The band showed each editor's summary as an inline field, which put editing
where the page names what it is working on. The summary is text there now, with
its `title` for one the band truncates, and the pen that opens the path for
editing sits after it — the same popover the editors' own bar carries, at the end
of the name it renames. Every editor already has a field for the summary itself:
the script's settings, the flow's Settings tab, and both app editors' deploy
drawer.

The popover moves into `PathEditPopover` rather than being written out four more
times. It owns the open state and hands back the path as it was when it opened:
the band's trail reads that while the popover is open, so a path growing under
the pointer does not drag the pen — and floating-ui the popover — along with it.
`EditorHeader` renders the same component, so the session pane and the band
cannot drift apart.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the pen renames the item and its summary

The pen opened the path alone, which left the summary with nowhere to be edited
from the band. It carries a Summary field above the path now — the item's name
first, the address it lives at below — and the band's text follows it as the user
types.

`TextInput` with a label rather than the bar's inline field: `EditableInput`'s own
docs rule itself out inside a popover, and it wraps this same input in its edit
mode, so the field, the "Add a summary..." placeholder and the trim on commit are
the behaviour that was there, in the shape the deploy drawer's Summary field
already uses. The keydown stops at the field because every editor binds keys of
its own and a summary is prose.

`summaryEditable` carries the hosts' `customUi.topBar.editableSummary` through,
and the editors' own bar passes it false: that bar keeps its inline field, and
`hidePath` would take the pen with it, leaving a condensed session pane no way to
name the item at all.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the agent editor fills the band too

It was the last editor route drawing its own header row under the band, and the
band named it "Agents" — `navPageFor`'s first-segment fallback, because nothing
on the route registered an item. The agent's path is the breadcrumb now, its
version badge reads after that name, the sentence that was the row's second line
became the hint beside it, and the Chat/Form switch, Settings, Evals, version
history and Deploy are the band's actions.

No trail levels in the registration: this layout opens settings in a drawer and
evals over the page, so unlike the dialog it never stands on a level — `trail` is
its root alone here.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the pipeline page fills the band

Found by sweeping every route under the band: the pipeline editor drew its own
row beneath it, 48px of a second header, while the band named only the section.
Its name is the band's now, the folder it is open on sits with that name — the
switcher is how you change pipelines, so it belongs to the name rather than to
the actions — and the mode switch, deploy-all, activity, refresh and overflow
menu are the bar's actions. The arrow back to the list stays: nothing else on the
page points there, and the sidebar does not list pipelines.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the app canvas controls stay reachable, and an unnamed item says so

Codex found the app editor's canvas controls unreachable on an ordinary laptop:
the owning route collapsed at 1350px, and a 1440 window with the rail out leaves
1232, so the width, theme and breakpoint toggles were hidden with nothing in the
menu to replace them — and the breakpoint is the only way onto the mobile layout.
Measured on the band, the toggles are 240px beside 334 of other actions and a
~430 trail, so they fit to about 1050: the threshold is 1100 now, and below it
each of the three keeps a menu entry rather than disappearing.

An item with no summary read as a bare `·` with an invisible pen after it, since
the plain text replaced a field that had a placeholder. It now reads "Add a
summary..." in the italic SummaryPathDisplay uses for the same state, and the pen
stops hiding when there is nothing to hover.

Three more from the review:
- `actionsFlexible` says "these actions can give width back", which is true of the
  Runs filter row and not of the pipeline's row of fixed buttons.
- `pathSnapshot ?? (newEditedPath || …)` keeps the `||` chain the five rename
  sites read and the declaration documents; `??` had made the rest unreachable,
  since `newEditedPath` is a string that is never undefined.
- The band bridged `contexts` onto the actions alone, while the pen that decides
  whether a path is the caller's to change renders in `summaryContent`. The
  breadcrumb's `afterName` and the summary are bridged the same way now, so that
  permission check reads the tree that registered it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the widest action rows scroll, and the canvas toggles show only where they fit

Measuring the band instead of estimating it: with the canvas toggles shown the
app editor's actions are 826-911px and the trail with the summary another ~500,
so the row overflowed the bar at every width up to about 1470 — the 1100 I set
from an estimate was wrong, and the 1350 before it was too. The toggles come out
at 1500 now, and the menu entries are what keeps all three reachable below that
rather than merely out of the way.

The two widest rows — this editor's and the pipeline's — scroll inside their own
box, which is the escape the rows they replaced had (`overflow-x-auto
whitespace-nowrap`). That answers pi's finding and clears the last 23-40px of
band overflow at middling widths, which predated the conversion. Measured after:
zero horizontal overflow from 852px of bar to 1392.

From the CI round's other findings:
- The pen is offered when either the summary or the path is editable, not the
  path alone: a host that allowed one and refused the other could no longer edit
  the summary from the band. `PathEditPopover` takes `pathEditable` to match, and
  its label names whichever it is given.
- The four editors' comments called the summary read-only, which stopped being
  true when the pen beside it started editing it.
- The pipeline's "centre group" comment described a layout it no longer has.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: actionsFlexible covers both ways a row gives width back

It read as if a search field were the only way to earn it, which stopped being
true when the app and pipeline editors' rows started scrolling inside their own
box. Both are the same bargain — the box may shrink because its contents can —
and the flag is what the bar reads to allow it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: data pipelines in AI sessions (alpha) (#11453)

* feat: pipelines in AI sessions with several folders open at once

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: deploy and discard a session's pipeline edits as one changes-list row

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: pipelines in the chat's listing and the preview picker; session tab shares the page's asset prefetch

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: a masked pipeline preselects only the scripts and triggers of its folder

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: review round — exact pipeline scripts in the fork mask, refuse a failed hand-off flush, deploy guard before awaits

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: review round — failed bundle save fails the deploy, removing a draft records no change

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: a session's pipeline tab opens node details in a modal, as the flow editor does

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: route pipeline graph edges orthogonally with spread ports, separate tracks and lanes

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: pipeline graph edges merge by shared end, enter nodes by the side, keep fewest crossings

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: review round — incremental crossing search and cached routes, retry the leftover draft save, drop the pipeline row when its last draft goes

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: retry the leftover draft save with what failed, never an empty draft

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* revert: pipeline graph edges back to bezier curves, detours curved too

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: hovering a pipeline graph node fades everything outside its lineage

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: hover fade eases edges too, through a stylesheet instead of new node objects

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: gentler hover fade without flicker on click; implied trigger edges recede in the assets-only view

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* docs: pipeline nodes ingest straight into DuckLake or data tables from Python/TS

* fix: pipeline tools callable right after open_preview; write_script refuses pipeline nodes in sessions

* feat: delete a data pipeline from the home list; pipeline rows match item rows

* fix: upward detour edges keep their gutter lane

* fix: pipeline delete takes only its own triggers and discards draft-only nodes by storage key

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: AI attaches trigger drafts to the pipeline nodes that declare them; materialize nodes keep no stale output

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: AI gives every pipeline node a succinct summary

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: session pipeline tools survive another tab taking the panel, reopening the editor on call

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: pipeline trigger lookups skip same-path flows; a failed trigger never hides its built node

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: deleting a pipeline clears it from open sessions so autosave cannot restore it

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: data upload opens on its inputs alone, works in sessions, with a prominent Upload data button

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: closing the pipeline pane closes its modal instead of leaving it blank

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* feat: a DuckLake asset opens on a Data tab with the database manager

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: inputs-only view infers draft inputs, keeps stored lineage, runs the live body

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: lineage follows the editor that inferred it; upload runs preview unsaved edits of deployed nodes

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: readyDataUploadPaths is a ReadonlySet in the pipeline editor

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: selected pipeline nodes get their blue edge in dark mode too

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: no red flash on load while writer lookups and script bodies are in flight

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: deploy an open deployed script's edits, and don't resurrect them after a discard

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* refactor: deploy and discard close the open pane first instead of patching its edits

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: keep the pane closed while deploying, and clear its buffer once its edits are a draft

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: a draft edited while it deploys keeps the newer edits, chained on the new version

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: a draft whose summary or labels changed mid-deploy is kept too

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: the app viewer names the page it owns, and stops claiming the viewport

A deployed app opened in an AI session's preview panel renamed the band from
the session to the app, and put the app's Edit beside the session's name. The
viewer registered wherever it was mounted; it takes `ownsPageHeader` now, the
way the detail pages already did, and keeps Edit over its own canvas otherwise.

Both viewers also claimed the whole viewport under a band that takes 44px of
it, so a deployed app scrolled by exactly the height of its own header (the
low-code one measured 900px in an 856px box). The height comes from the host
now: the three standalone routes floor themselves at the viewport and
everything inside them is `h-full`. A raw app reached through `/app_embed` had
no height above it at all and collapsed to nothing; it fills the frame now,
though that route still fetches no bundle secret, so what it shows is blank —
as it is on main.

The agent editor's pen was a second editor of the same path that reported
nothing: an invalid path never reached the deploy gate the settings drawer's
field reports to. It binds to the same slot now. Two things that surfaced
with it: a rename never reached the trail, which read the deployed path; and
typing the saved path back raised "path already used", since the popover's
`initialPath` moves with the rename — `allowedExistingPath` is what says the
item already occupies it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* refactor: the band's controls come from the design system, and the row has one copy

`NavBreadcrumbTrigger` was `MeltButton` written again — same `use:melt` on a
plain button, same spread — so the workspace chevron uses the component that
already exists, and the sidebar handle is a `Button`.

`DetailPageHeader` carried its action row twice, once for the band and once
for the row it draws inside a session's preview panel, and the two had already
drifted: different dropdown size, no `unifiedSize` on the buttons, and the
deprecated `variant="contained" color="light" size="xs"` on the schedule
button. Both placements render the same two snippets now, at the size each
needs.

`OperatorPageKey` admitted every `operator_settings` field, including the three
write rights, which `sidebarPageAllowed`'s `=== true` rule answers wrongly: a
write right is granted unless withdrawn, so its absence means "never
configured". It is the ten visibility flags now.

A test over the registry's merge: later registration wins field by field,
actions sort by `actionsOrder` then registration, release gives its fields
back. That rule is why a component registering where it is not the page
renames someone else's band.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the band owns its bottom edge, and one pen renames everywhere

The page header decides whether a separator is drawn, so a page cannot draw a
second line under it. A registration asks for `onScroll` (the default, drawn
only while something passes under the bar), `always` for a page whose content
starts right beneath it, or `none` where the page genuinely draws its own first
line. Detail pages, the agent editor, workspace settings and the pipeline list
take `always`; the raw app editor takes `none`, since its splitpanes already
draw that line.

The pen that opens the rename popover becomes PathEditPen, shared by the
editors and the deployed pages so the two cannot drift, and the deployed pages
gain it. SummaryPathDisplay makes the pen the only trigger rather than the whole
name block, so the name is selectable text again. Renaming an agent goes through
ResourceService with only the fields the popover edits: the backend sets a
column per field it is given, so writing the config back would clobber a
concurrent edit. The popover also stops clearing a real collision verdict by
telling Path which path is its own.

Two Button fixes, both reaching call sites beyond this work:

- `dropdownItems` given as a function was asked for `.length`, which is its
  arity and therefore 0, so a split button built that way rendered no caret.
  Both app editors' Deploy dropdowns were unreachable.
- the anchor branch dropped `title`, so a link button had no tooltip.

The audit log filters move into the header actions at the design system's sizes
and trade absolutely positioned labels for the form layout in
brand-guidelines.md: a label above its input, 4px apart, fields 8px apart.

The deployed app viewer floors itself at `min-h-full` instead of claiming the
viewport, which would overhang the page box by the height of the band, and
carries Edit and a public-url link that fold into a menu on a narrow bar. The
root layout drops the legacy menu block, which rendered a second hidden sidebar
and 54 invisible tab stops, and marks the closed detached sidebar `inert`.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: service logs and the audit filters move into the band

Service logs: the page's name is the breadcrumb's, so the `<h1>` is gone and the
search field, the timeframe and the two switches sit in the band beside it, with
the documentation tooltip after the name. ServiceLogsInner registers the band
itself, the way RunsPage does, since that route is its only caller, and the
`title` snippet the page used to hand it goes with the heading.

The search is the widest control there, because searching is what the page is
for, and it is the one that gives width back under pressure. The two switches
fold into a popover below a 1200px bar rather than at a phone's width, so the
field keeps its full width down to a much narrower window — folded rather than
dropped, because the page no longer carries a toolbar they could fall back to.

The timeframe control took no size of its own, so it defaulted to `md` beside a
`sm` field, and sat in a `w-48` box 20px narrower than its content, whose
overflow is visible and so spilled into its neighbours.

No `separator`: the two panes below start under the bar with borders of their
own, so at rest there is nothing for the band to divide.

Audit logs: the filters lose their labels. Each select says what it filters
while unset — "all usernames", "all resources" — and once set the value names
the field by itself. From and To become the TimeframeSelect the runs page uses,
which absorbs the standalone Refresh button, since that control's left half is
the refresh. `auditTimeframes` joins the runs and service-logs presets, with
"All time" first. The bounds stay the filters' own `after`/`before` rather than a
remembered preset label: the timeline's zoom writes them too, and a label could
then contradict the chart.

That takes the row from 1326px to 898px, so it stays inline down to a ~1500px
viewport instead of ~2500. The fold measures the row against the bar less what
the breadcrumb takes, plus a gap the two must never close — the breadcrumb
yields width grudgingly and the row not at all, so sized to just meet they
overlap rather than compress, and mid-resize the tooltip after the page's name
lands on the timeframe button.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: an embedded page keeps its name in the band

A page in a session's preview panel loads in a frame that sets `nomenubar`, so the
layout renders the band with `navHidden`. That flag hid the whole breadcrumb —
which is what carries the page's name — while the actions beside it had no such
guard, leaving a row of controls belonging to nothing.

Hiding the trail is still right: it leads to the host's workspace and offers a
picker that would take the reader out of the session. So the breadcrumb gains a
`nameOnly` mode that drops the workspace part and its picker and keeps the page's
own name, and the band asks for it instead of rendering nothing. The summary line
loses the same guard, so an item in there shows its path and summary.

Only the pages whose own title the band took over are affected. Variables,
Resources, Schedules and Triggers still draw their own `PageHeader` and render no
band when embedded, so nothing is doubled.

The bar's left padding is there to sit a button against — the sidebar handle, or
the workspace disc, both of which carry their own margin. Embedded there is
neither, and the name started as bare text 8px from the edge, so that case takes
the same 16px the bar already keeps on its right.

Workspace settings drops the copy-id button from the band: the General tab names
the workspace ID with a field of its own, which is where someone looks for it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: a session in full screen keeps its tabs in the band

Full screen used to drop the session's name from the breadcrumb and leave the
tab strip inside the panel. The band now carries both: the session's name where
"Sessions" stood, and from the breadcrumb to the right edge one filled surface
holding the tabs and the preview's controls, the bar's full height.

The switch between the two layouts is the panel expanding to the left. The chat
column is resized to nothing rather than unmounted, so it can be seen taking its
place -- and so a toggle no longer tears down every mounted chat and rebuilds it
on the way back. Panes transition their width only while that switch is in
flight, since a splitter drag sets the same inline width and would trail the
pointer.

The band floats over the content in full screen as it already did beside a
panel, so only its right edge moves and the page below never shifts by the
bar's height. Leaving full screen, the panel travels back across ground the
band has already claimed, so it passes one layer above it for the length of the
move: pinned to the width the panel will end at, the bar would otherwise show
through as a white patch over the panel's left edge.

The breadcrumb also gains the acting fork, the way every other page shows it,
which replaces the "Acting on" control the session header carried.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the band's inset survives a resize made in full screen

The panel was remembered as a width, sampled while it stood still. A window
resized during full screen left that width describing a layout that no longer
existed, and the way out pinned the band to it for the length of the move. It
is now remembered as the share of the row the panel takes, which the new row
width turns back into the right inset: leaving full screen at 1200px after
entering it at 1600px, the band starts at 628 rather than at 909.

The pane transition's duration is read from the property the timer is set from,
so the class cannot outlive the movement it describes, and two comments that
described earlier drafts are rewritten: the band's name in full screen comes
from the session's own header once it has mounted, and the tab strip's wrapper
carries a fade, not a measured move.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: an agent's pen keeps the labels the agent already carries

The pen on an agent's page was handed the summary and the path but not the
labels, so the popover offered an empty list over a resource that had some. It
saves every field it shows in one write, and the resource endpoint sets the
column it is given, so adding a label there replaced the agent's own: an agent
labelled `production` and `scheduled` came back from one edit labelled only with
the label just added. A flow's page and a script's page both pass theirs; this
one now does too.

Verified against the running instance: an agent created with
`["production","scheduled"]`, one label added through the pen, reads back
`["production","scheduled","new"]` — and `["new"]` with the prop removed again.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the sidebar remembers detaching per role, not per account

Whether the sidebar starts detached depends on the role: an operator gets it out
of the way, a developer gets it docked. That default held only until the control
was pressed once, because the choice went to a single key read before the role.
From then on both roles wore whichever was pressed last, and the one that had
not asked for it could never get its own default back — switch a workspace where
you are an operator for one where you are a developer, and the sidebar stayed
detached.

Each role now keeps its own choice, and falls back to its own default until it
has one.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: the handle's comment describes what the handle does

It claimed a click holds the sidebar card open. Nothing distinguishes a click
from a hover there — both call the same opener, and the pointer leaving
schedules the close either way — so on a desktop the card follows the pointer
and goes when it goes. Measured: hovering slides the rail from -213 to 5, the
click moves nothing, and the pointer leaving puts it back at -213.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: a dev workspace's badge wears its fork's own colour

Every surface that names a dev workspace drew the same blue badge, which said
nothing about which workspace it belonged to. One component, DevWorkspaceBadge,
now draws it: a fork with a colour fills the badge with that hue and writes the
label in white, and a fork without one keeps whatever its row had before.

The hue cannot carry white at a fixed lightness — HSL lightness is not perceived
brightness, so yellow and purple at the same L differ by several stops — so
forkAccentStyle solves the badge's lightness by bisection instead, taking the
lightest shade of the hue that still holds white at WCAG AA. Measured across the
wheel the contrast lands between 4.52 and 4.64.

The session picker opens with every family collapsed, since the fork a session
runs in is the session's own and not something that picker should offer to
change.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the name is what opens its rename editor, and the trail truncates from its start

The pen beside a name is gone. The name itself is the trigger, with a hover fill
behind it where the click opens an editor rather than going anywhere — the path
segment that only copies keeps no fill, which is what tells the two apart. A bar
with a summary field of its own (EditorHeader's condensed pane) still gets the
pen, since it has no name to hang the trigger on.

Clicking either half of a name opens the editor under that half with the cursor
in the matching field. The band draws the path and the page draws the summary, so
one popover cannot anchor under both: a page hands the band a `pathTrigger`
snippet and the band renders its path segment inside it, passing the classes its
own segment wears. Each editor supplies a second PathEditPopover for it, and
`focusField` names the field to focus — `#path` explicitly, since melt otherwise
lands on the row's first button.

Those snippets render from the band's tree, which outlives the values they read
by a flush, so each is gated on the value it reads. Without that an agent page
whose draft had not loaded threw inside the band and left the breadcrumb empty.

The trail now carries two weights: the segment that ends it is what the page is
about and takes `font-medium text-emphasis`, everything before it is a step
quieter. Which segment that is follows the route — the workspace on home (the
fork alone inside a fork), a list page's own name, an item's summary.

And it truncates instead of switching to icons at a breakpoint. The workspace
part yields first, the family name before the fork, then the path's folders, and
the item's own name last. The bar no longer wraps on a phone: a wrapping row
never squeezes anything, it hands the trail a line of its own and leaves it
whole.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: a page keeps its path segment when it cannot offer a rename

Four things the local reviews caught in the rename work:

A detail page handed the band a `pathTrigger` whether or not it could save one.
Without `onSaved` — every script, flow or agent page the reader cannot write —
`SummaryPathDisplay` fell through to its read-only branch, which ignores the
`label` it was given and draws its own block: the trail lost its path and the
summary appeared twice, two lines deep, in a 44px bar. The page now offers the
trigger only where a rename can be saved, and the component renders what it was
handed rather than its own row.

The band gives up its own segment the moment a page offers a `pathTrigger`, so a
trigger whose own gate is closed left the trail with no path at all — for the
whole of an agent draft's load, not just a flush. Each gated snippet now draws
the path itself in that case.

`openFocus` named the path field as `#path`, which melt resolves with a
document-wide query; `Path` hardcodes that id at every call site, and the
popover is portalled to the body, so it lost document order to any other `Path`
on the page. It is scoped to the popover's own field now.

And `open` was an exported `$bindable(false)` on an optional prop, which the
frontend's own rules ban. No caller bound it, so it is local state.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the badge solves its lightness against the colour the browser paints

Three nits from the review round.

The contrast bisection measured the hue at full float precision, but a browser
paints 8-bit channels: rounding them moved the result up to 0.025 off the target,
which took #993388 to 4.478 against white — under the AA floor the comment
claims. Quantizing the channels inside `paintedLuminance` puts the worst case
across 2880 hue and saturation combinations at 4.500, and your own fork's badge
measures 4.53 as painted.

An agent nobody can write still offered the rename behind its path segment: an
editor with two read-only fields, in place of the click that copies the path.
It is gated on `readOnly` now, the way a detail page is gated on `onSaved`. The
script and flow editors take the same gate from the `customUi` flags that
already decide whether their summary offers one — unreachable today, since the
editors own the band only on routes that pass no `customUi`, but the four
editors now answer the question the same way.

And the registry's doc named an `onEditName` field that has never existed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: an operator is not offered a fork of a workspace

A fork does not carry the role. The server writes the forker's membership row in
the new workspace from the parent's `is_admin` and lets `operator` fall to its
`false` default, so an operator who forks lands in a full-rights copy of the
workspace they could only run things in.

`canCreateFork` refuses an operator, which takes the entry out of the workspace
picker, the session family picker and the session changes bar;
`editInForkAllowed` refuses one too, so no editor offers "Edit in fork". The
fork page refuses on its own, since a URL reaches it whatever the pickers show.
And a workspace the user is an operator in is no longer a base the fork form
will accept, so the base picker cannot route around the gate.

This hides the affordance; it does not enforce the rule. `create_fork` asks only
for membership — `require_admin` covers the dev-workspace case alone — so the
API still answers an operator who calls it directly. The server-side fix is to
carry `operator` into that membership row, the way the member-copy path already
does, and it belongs in a backend change rather than here.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* style: the fork refusal uses the design system's alert

The refusal panel was hand-built from a div and two paragraphs with literal
colors, copied from the Runs page's own operator refusal. The frontend's rules
say to use the design-system component, and `Alert` already draws exactly this.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: frontend type errors failing npm_check

The dev workspace badge accepts the nullable label the API returns, the
session strip transitions take the node Svelte passes them, and an unused
import is dropped.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

* fix: the frontend type-checks again

`check frontend build` has been red on every head for a day, through four review
rounds that each came back clean — none of the reviewers type-check, and they
say so in their own coverage sections. Ten errors:

The dev badge's `label` and `color` were typed `string | undefined`, but the
workspace rows carry `string | null`, so all seven of its call sites failed.
`forkAccentStyle` took the same widening.

A Svelte transition is called with the node it animates, and `stripArrive` and
`stripLeave` were declared with no parameters at all.

And `DevWorkspaceSetting` still imported `Badge` after its badge moved into the
shared component.

`svelte-check` now reports 0 errors. Note that it only agrees with CI once
`src/lib/gen` has been regenerated — the gitignored client goes stale against
main and invents errors CI never sees.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the home bar carries the CLI and hub buttons while it has the room

A menu is where a thing goes when the bar cannot hold it, not where it belongs.
Above 1024px of bar, "CLI / MCP" and "Hub" stand as their own buttons; below it
they fold back into the ellipsis, which is still where they live on a phone.

The threshold reads `pageHeader.barWidth`, not the viewport, so it tracks what
the sidebar actually leaves behind. Home's trail is ~250px and its other
registered actions ~400, so the two labelled buttons fit from about 850 up;
1024 leaves margin before the trail would start truncating, and it does not
truncate at any width between 760 and 1600.

The ellipsis hides itself when it would open an empty menu, which happens only
once both side trips are in the bar and the composer is already on the page.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: a fork's band counts its drafts beside its drift, and every count opens what it names

A fork that had drifted stopped saying anything about its drafts: the banner
printed the draft count only while the fork was in sync, and the drift badges
replaced it the moment there was anything to deploy. A fork can be 86 ahead, 39
behind and holding 128 drafts; it now says so.

All four counts open the page that holds them — ahead the deploy direction,
behind `?dir=update`, conflicts the side they are resolved from before
deploying, drafts the drafts half. The button beside them still chooses a
direction on the reader's behalf; a badge means the one thing it names.

The colours and the icons are the compare page's, so the counts in the bar and
the rows behind them say the same thing the same way: ahead green, behind blue,
conflicts orange. Drafts share blue with behind and are told apart by a pencil
and by naming what they count, rather than by an arrow that would read as a
third direction. Both arrows are straight, here and on the compare page.

"Review changes" replaces "Review & Deploy Changes", and it renders in every
state. What the comparison has to say — still counting, or failed — is a mark
beside the button rather than a branch that replaces it: a fork whose count is
unfinished is exactly when a reader wants the page that counts properly.

The drafts banner outside a fork drops "This workspace has", so the count reads
the same wherever it appears.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the home band keeps its hero's own cross, and the side trips stay quiet

The CLI and hub buttons take the shape the hero gave them before the bar did —
`xs`, hint-coloured, the hub keeping its outbound mark — so a side trip does not
read as loudly as the thing the page is for.

The ellipsis is gone while the hero is on the page: the hero's cross already
puts it away, so a menu whose only item repeats that is a button that does
nothing new. It comes back when it has something of its own to carry, which is
the side trips on a narrow bar and the way back to a hero that has been put
away. That entry is now "Pin session hero to homepage", and "Remove session
chat" survives only as the cross's own words, where one action keeps one name.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: a fork's counts and the way to review them are one control

Four badges beside a button read as four things that happen to sit together.
They are one thing — every segment opens the same page on the part it names — so
they now share a box: one outline, square corners inside it, and the rounding on
the box alone, which means no segment has to know where in the row it sits.

The only rule in the strip is the one before the end cap: everything left of it
is a reading, that is an action.

With room every count says what it counts; below 1280px of bar they keep their
icon and their number, which is the part a reader scans for, and the tooltip
names them either way. The end cap keeps its words at every width — the counts
have icons to fall back on and it has none.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the compare page asks once, says it is working, and knows its own name

The page re-requested the same summaries and on-behalf-of emails forever. Two
faults compounding:

`fetchOnBehalfOfInfo` skipped a key it had already answered with
`onBehalfOfInfo[k] !== undefined`, but stored `undefined` on failure — so
"asked, and there was nothing" could not be told from "never asked". An item
that is only ahead does not exist in the parent, so that side always fails:
a permanently un-cacheable set, re-requested on every pass.

And the effect that drives both fetchers depended on what they write. Each reads
its cache synchronously before its first `await`, which registers that cache as
a dependency; the write that follows re-runs the effect. Write, re-run, write.

Both now record what they asked for in a plain Set, marked before the request
rather than after the answer, and the effect is untracked so it depends on the
comparison alone. The caches are cleared when the compared pair changes, since a
summary's key carries no workspace and those Sets would otherwise never ask
again.

While the comparison was in flight the page printed "No changes between this
workspace and X" — an empty list and an unanswered one look alike from the
layout, and that is a claim with no answer behind it. The page tracks the
request now, and the layout spins instead.

The breadcrumb called the page "Forks", which is what `navPageFor` falls back to
when a route names itself nowhere: the first segment, capitalised. It registers
"Compare" instead — in the page rather than in NAV_PAGES, which is what the
sidebar renders from.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* Revert "feat: a fork's counts and the way to review them are one control"

This reverts commit e4504d2843. The counts go back to separate badges and
"Review changes" to a button of its own.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the compare page follows the fork picked in the trail

Switching a fork from the breadcrumb left the page comparing the pair it opened
with. Two reasons, neither reactive: the page seeds `currentWorkspaceId` from
`?workspace_id` once at mount and never reads it again, and the switch rewrites
`?workspace`, a different param that this page does not read.

It follows `$workspaceStore` now — picking a fork in the trail means comparing
that fork. The param is rewritten so a reload keeps the pair on screen, and an
arbitrary `?target=` is dropped, since it named a destination for the workspace
just left.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: the band is a step taller, and the runs row sizes with it

The bar goes from h-11 to h-12, taken from one constant beside the top
padding a floating band owes the page under it, so the two cannot drift.

The runs page's controls now share the sm height the rest of a header's
actions use: FilterSearchbar learns the size, dropping the top margin
that only centres text at md and taking the shared leading instead, and
sm buttons get px-3 so a label reads as a button. The actions sit in
their own row with a wider gap, and the chart below meets the header
with no gap of its own.

The search field holds one width through typing. As flex-1 it was
whatever the row had left over, so committing a chip grew it from 226 to
288 and took those pixels off the breadcrumb.

A fork's count badges keep their labels while the bar is at least 1280
wide and fall back to icon and number below that.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the trail holds a path's place instead of naming its kind

A page registers its kind when it mounts and its path once it has loaded
one. In between the trail printed "script" or "flow", which reads as the
answer and then jumps when the real path lands. A skeleton of a
plausible path's width holds the place, with aria-busy and a label
naming what is loading.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the compare page reads the workspace it was linked with

The effect that follows the trail's fork also ran on mount, so a
`?workspace_id` that differs from the navigation workspace was thrown
away. That difference is the contract: a session's Review button passes
the fork it committed to and leaves the navigation workspace alone, as
does the prod -> dev link in UpdateDevWorkspaceModal. Both opened on the
sidebar's workspace instead, comparing the wrong pair or none at all.

Only a change in the store now retargets the page, measured against what
it held when the page opened.

The URL rewriting goes with it. The picker's items are links that carry
every other param forward, so their navigation lands after the effect
and put back the `workspace_id` it had just rewritten: switching fork
and reloading left the trail naming one workspace and the page comparing
another. The page reads `?workspace=` ahead of `?workspace_id=` instead
— the param the picker writes and the root layout already applies to the
store — and `?target=` lapses while the page is on another workspace
rather than being deleted from a URL someone else is about to replace.

Two 44px literals left over from the h-11 band: the detached sidebar
card's measured-top fallback and the rail morph's end state, both now
from BAR_HEIGHT_PX beside the class the band wears.

The runs search field keeps its 18rem without growing, but as `shrink-0`
nothing in that row could give width back, so a short bar had no answer
but the trail's truncation and an overlong row would have run past the
edge. `min-w-0` restores the yield without restoring the growth: a flex
item's floor is its min-content width, which the chips inside the field
raise.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the compare page's target picker works, and a switch survives reload

`?target=` was gated on the page still being on the workspace the URL
named. Opened from "Merge into another workspace" in the workspace
settings the URL names no workspace at all, so that comparison was
`<id> === undefined` and the param was ignored for good: picking a
target left the page on the "no parent workspace" prompt, which is the
only place the migration can be started from. Read the param as it was
read before.

The follow effect writes `?workspace=` again. A plain click in the
picker preventDefaults its own link and switches the store directly, so
nothing navigates afterwards and nothing overwrites the write — while
`fixupUrlAfterWorkspaceSwitch` rewrites the param only once it is
already there, which on a page opened with `workspace_id` alone it never
is. Without it, switching fork and reloading left the trail naming one
workspace and the page comparing another.

Exercised on a fork pair: the settings entry's picker loads the
comparison it names, a plain-click switch survives a reload, and a
`workspace_id` deep link still outranks the navigation workspace.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* refactor: the compare page reads its pair instead of following it

Net effect count: three removed, none added.

The page derives the workspace it compares from the URL and the store
rather than keeping it in state and running an effect to follow a
switch. The workspace-switch link is what moves it: the picker's items
carry every param forward, so a page-side rewrite raced that navigation
and lost — the trace showed the fixup's replaceState landing first and
the link's pushState putting the old params straight back. The href now
swaps `workspace_id` alongside `workspace` and drops `target`, and
`fixupUrlAfterWorkspaceSwitch` does the same for the paths that do not
navigate, so both writers agree and the order stops mattering.

Dropping `target` on a switch also closes the case where switching onto
the workspace that was the target left the page comparing it with
itself.

CompareWorkspaces remounts on either side of the pair changing, not just
the target: two forks of one parent shared a mount, since the target is
the parent they have in common, and with it the selection, the deploy
direction, the summaries, the pinned rows, the deploy permissions and
the CI results. The cache-clearing effect that stood in for that is
gone. Its summary loops outlive their mount, so they read the pair once
at the start and stop writing when the instance is destroyed.

The app viewer asked for a public secret on every app view to decide
whether to draw a button; it reads the mode off the app it has already
loaded and fetches the secret when the button is used.

The summary popover seeds its fields on the open transition rather than
from an effect that also read the item: a save landing while it was open
reseeded the fields over what was being typed.

Exercised against a running instance, 18 cases: idle issues nothing, a
switch moves page, trail and URL together and survives a reload, a deep
`workspace_id` link still outranks the navigation workspace, four fast
switches settle with no foreign row, and the migration target picker
loads the pair it names.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the public url opens the tab the click itself opened

The secret is fetched when the button is used, and the tab was opened
once it arrived — after an await, where Safari no longer counts the
click as the gesture that opened it and blocks the popup without saying
so. Open the tab synchronously and point it at the url once the secret
is known, closing it and saying why when there is none.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the public url button survives a sandboxed app

A sandboxed low-code app loads inside the opaque /app_embed iframe, so
`loadApp` never runs on this page and `app` stays undefined — taking the
button and its compact-menu entry with it, for the owner too. The mode
now comes from the lite read that already runs for the Edit button,
which is there for exactly this reason: it is the one description of the
app this page has whether or not the app itself loads here.

Exercised on four apps: anonymous sandboxed low-code, anonymous
unsandboxed low-code and an anonymous raw app all show the button and
fetch the secret only on click; a publisher app shows neither and asks
nothing. At 640px the button folds into the menu, which carries it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* refactor: the removed hero leaves a blue flash on the ellipsis

The mark that flew from the cross to the menu is gone, with its two
element refs, its flight state and timer, and the `data-home-ai-menu`
hook that existed only as its destination. What is left is the thing it
was there to say: the ellipsis holds the way back, so it flashes blue as
it appears.

Written as a keyframe animation, not a transition. A transition needs
the element painted at its start value before the end value is set, and
this mounts in the same flush that collapses the hero, so the frame that
would have shown it at rest is the frame it is already told to leave —
one `requestAnimationFrame` is not reliably past that paint, and the
flash was never drawn at all. `animate-fade-out` already exists for this
(AssetsDropdownButton flashes a background the same way); mounting it
only while pulsing gets a fresh run per removal without that component's
reflow juggling.

The flash rides the ellipsis rather than the row: with the CLI and hub
buttons beside it the row is three controls wide, and marking all three
names none of them.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the flash rests invisible, and its comment describes what is there

`animate-fade-out` carries no fill mode, so at the end of the run the
element returns to its own opacity. At the default 1 that is the flash
coming back at full strength for as long as the unmount timer lags the
animation — which the hero's collapse, happening right there, is well
placed to cause. `opacity-0` at rest makes the revert land on invisible.

The comment above it still described the mark that used to fly to the
menu, and made its point three times over.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the compare page asks only about the pair it is showing

Switching workspace on this page fetched the previous pair's item paths
from the new workspace, where none of them exist: 97 of 174 requests
404ing on one switch. The comparison was plain state, and the card is
remounted when the pair changes — in the same render that changes it,
before the effect that cleared the comparison could run, so the new
mount started from the old pair's diffs. It now carries the pair it
describes and is read only while that matches, which leaves nothing to
clear.

`computeFullScan` had the same shape: it read the target again after its
await, so a switch mid-scan invalidated another pair's cache, named the
wrong workspace in its toast, and left the scan button spinning on a
pair that was never scanned. It captures the pair up front, re-compares
only while the page is still on it, and reports it as its own.

Each diff says which side holds the item, so the summary and
on-behalf-of loops stop asking the side that does not: an item that is
only ahead cannot answer in the parent, and asking was a guaranteed 404
and a console error per one-sided row.

Those two loops also read the same object — a summary and an identity
are fields of one payload — so every row was fetched twice per
workspace. One read per item per workspace now serves both. A failed
read is not kept: held, it would read as "this item has no identity",
which is what decides whether a deploy offers to preserve one, and an
item deployed without it silently becomes the deployer's.

The header names the workspace being compared when that is not the one
the app is pointed at, which is how a session's Review button and the
prod -> dev link open this page. The trail said the parent while the
page compared the fork.

Measured on a 90-item pair, before -> after: a cold load 247 -> 170
requests and 167 -> 90 item reads; a workspace switch 174 requests with
97 404s -> 71 with none. Sixteen cases over cold load, idle, switching
in both directions, four switches with no time to settle, a cross-family
switch, reload, deep link, the migration picker and leaving an arbitrary
target: no request in any of them names a pair the page has left.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: the scan toast reports the pair it scanned

Off the scanned pair the re-compare is skipped, so there is nothing to
report as compared — but the success branch said "Compared N items"
anyway, and the error it checked first belonged to whichever pair the
page had moved to. Each outcome now says what happened: the seed alone
when the page has moved on, and the comparison's result only where one
was asked for.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Guilhem Lemouel <guilhemlemouel@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
2026-10-08 14:23:29 +02:00

2352 lines
70 KiB
Svelte

<script module>
import '@codingame/monaco-vscode-standalone-languages'
import '@codingame/monaco-vscode-standalone-typescript-language-features'
import { typescriptDefaults } from '@codingame/monaco-vscode-standalone-typescript-language-features'
</script>
<script lang="ts">
import { BROWSER } from 'esm-env'
import { buildWsUrl } from '$lib/wsUrl'
import { sendUserToast } from '$lib/toast'
import { createEventDispatcher, getContext, onDestroy, onMount, tick, untrack } from 'svelte'
// import libStdContent from '$lib/es6.d.ts.txt?raw'
// import domContent from '$lib/dom.d.ts.txt?raw'
// import denoFetchContent from '$lib/deno_fetch.d.ts.txt?raw'
import * as vscode from 'vscode'
// import '@codingame/monaco-vscode-typescript-basics-default-extension'
// import '@codingame/monaco-vscode-typescript-language-features-default-extension'
// import 'vscode/localExtensionHost'
import { MonacoLanguageClient } from 'monaco-languageclient'
import { toSocket, WebSocketMessageReader, WebSocketMessageWriter } from 'vscode-ws-jsonrpc'
import { CloseAction, ErrorAction, RequestType } from 'vscode-languageclient'
import type { DocumentUri, MessageTransports } from 'vscode-languageclient'
import { MonacoBinding } from 'y-monaco'
import {
dbSchemas,
type DBSchema,
codeCompletionSessionEnabled,
lspTokenStore,
formatOnSave,
vimMode,
relativeLineNumbers
} from '$lib/stores'
import { editorConfig, registerWebviewPaste, updateOptions } from '$lib/editorUtils'
import { editorFontSize } from '$lib/editorFontSize.svelte'
import { createHash as randomHash } from '$lib/editorLangUtils'
import DdlMigrationGuard from './DdlMigrationGuard.svelte'
import {
type Preview,
ResourceService,
type ScriptLang,
UserService,
WorkspaceService
} from '$lib/gen'
import type { Text } from 'yjs'
import {
initializeVscode,
keepModelAroundToAvoidDisposalOfWorkers,
MONACO_Y_PADDING
} from '$lib/components/vscode'
// import { initializeMode } from 'monaco-graphql/esm/initializeMode.js'
// import type { MonacoGraphQLAPI } from 'monaco-graphql/esm/api.js'
import {
editor as meditor,
languages,
KeyCode,
KeyMod,
Uri as mUri,
type IRange,
type IDisposable
} from 'monaco-editor'
import EditorTheme from './EditorTheme.svelte'
import {
BIGQUERY_TYPES,
DUCKDB_TYPES,
MSSQL_TYPES,
MYSQL_TYPES,
ORACLEDB_TYPES,
POSTGRES_TYPES,
SNOWFLAKE_TYPES
} from '$lib/consts'
import { setupTypeAcquisition, type DepsToGet } from '$lib/ata/index'
import { initWasmTs, type InferAssetsSqlQueryDetails } from '$lib/infer'
import { initVim } from './monaco_keybindings'
import { updateSqlQueriesInWorker, waitForWorkerInitialization } from './sqlTypeService'
import { parseTypescriptDeps } from '$lib/relative_imports'
import { scriptLangToEditorLang } from '$lib/scripts'
import {
listWorkspaceMacrosCached,
macroDefinitionSql
} from '$lib/components/assets/workspaceMacros'
import {
fetchLatestSchema,
normalizeAssetPath,
type ContractMarker
} from '$lib/components/assets/AssetGraph/schemaContracts'
import * as htmllang from '$lib/svelteMonarch'
import { conf, language } from '$lib/vueMonarch'
import { Autocompletor } from './copilot/autocomplete/Autocompletor'
import { AIChatEditorHandler, type ReviewChangesOpts } from './copilot/chat/monaco-adapter'
import GlobalReviewButtons from './copilot/chat/GlobalReviewButtons.svelte'
import AIChatInlineWidget from './copilot/chat/AIChatInlineWidget.svelte'
import { writable } from 'svelte/store'
import { formatResourceTypes } from './copilot/chat/script/core'
import type { ScriptLintResult } from './copilot/chat/shared'
import FakeMonacoPlaceHolder from './FakeMonacoPlaceHolder.svelte'
import { editorPositionMap } from '$lib/utils'
import { extToLang, langToExt } from '$lib/editorLangUtils'
import { aiChatManager, type AIChatManager } from './copilot/chat/AIChatManager.svelte'
import { chatState } from './copilot/chat/sharedChatState.svelte'
import type { Selection } from 'monaco-editor'
import { canHavePreprocessor, getPreprocessorModuleCode } from '$lib/script_helpers'
import { setMonacoTypescriptOptions } from './monacoLanguagesOptions'
import { copilotInfo } from '$lib/aiStore'
import { getDbSchemas } from './apps/components/display/dbtable/metadata'
import { rawAppLintStore, type MonacoLintError } from './raw_apps/lintStore'
import { MarkerSeverity } from 'monaco-editor'
import { resource, useDebounce, watch } from 'runed'
import { useOperatingWorkspace } from '$lib/components/operatingWorkspace.svelte'
const operatingWorkspace = useOperatingWorkspace()
// import EditorTheme from './EditorTheme.svelte'
let divEl: HTMLDivElement | null = $state(null)
let editor: meditor.IStandaloneCodeEditor | null = $state(null)
let pasteCleanup: (() => void) | undefined = undefined
interface Props {
code?: string
cmdEnterAction?: (() => void) | undefined
formatAction?: (() => void) | undefined
automaticLayout?: boolean
websocketAlive?: any
shouldBindKey?: boolean
fixedOverflowWidgets?: boolean
path?: string | undefined
yContent?: Text | undefined
awareness?: any | undefined
folding?: boolean
args?: Record<string, any> | undefined
/** Schema to complete SQL against when the caller already holds it; otherwise it is
* looked up from `args.database`. */
sqlSchema?: DBSchema | undefined
useWebsockets?: boolean
small?: boolean
scriptLang: Preview['language'] | 'bunnative' | 'tsx' | 'jsx' | 'json' | undefined
workflowAsCode?: boolean
disabled?: boolean
lineNumbersMinChars?: number
files?: Record<string, { code: string; readonly?: boolean }> | undefined
extraLib?: string | undefined
/** Trailing debounce window (ms) on Monaco's onDidChangeModelContent.
* Each keystroke schedules (or reschedules) an `updateCode` call this
* far in the future. */
changeTimeout?: number
/** Hard ceiling (ms) on how long `updateCode` can be deferred while
* the user is typing continuously — measured from the FIRST
* keystroke of the burst (the leading fire). Without this cap,
* uninterrupted typing would hold the bindable `code` prop stale
* indefinitely and downstream consumers (autosave, lint, live
* preview) would never see the latest text. */
maxChangeTimeout?: number
loadAsync?: boolean
key?: string | undefined
class?: string | undefined
moduleId?: string
enablePreprocessorSnippet?: boolean
/** When set, enables raw app lint collection mode and reports Monaco markers to the lint store under this key */
rawAppRunnableKey?: string | undefined
// Used to provide typed queries in TypeScript when detecting assets
preparedAssetsSqlQueries?: InferAssetsSqlQueryDetails[] | undefined
// To execute preview scripts with the right worker group
customTag?: string
// Live schema-contract diagnostics (pipelines gap #2b): owner-scoped
// warning markers computed by the caller (ScriptEditor's contract
// mirror) from the buffer's asset refs vs captured producer schemas.
// Warning severity only — contracts never block; empty clears.
schemaContractMarkers?: ContractMarker[]
}
let {
code = $bindable(),
cmdEnterAction = undefined,
formatAction = undefined,
automaticLayout = true,
websocketAlive = $bindable(),
shouldBindKey = true,
fixedOverflowWidgets = true,
path = undefined,
yContent = undefined,
awareness = undefined,
folding = false,
args = undefined,
sqlSchema = undefined,
useWebsockets = true,
small = false,
scriptLang,
workflowAsCode = false,
disabled = false,
lineNumbersMinChars = 3,
files = {},
extraLib = undefined,
changeTimeout = 500,
maxChangeTimeout = 1000,
loadAsync = false,
key = undefined,
class: clazz = undefined,
moduleId = undefined,
enablePreprocessorSnippet = false,
rawAppRunnableKey = undefined,
preparedAssetsSqlQueries,
customTag,
schemaContractMarkers = []
}: Props = $props()
$effect.pre(() => {
if (websocketAlive == undefined) {
websocketAlive = {
pyright: false,
ruff: false,
deno: false,
go: false,
shellcheck: false
}
}
})
let lang = $state(scriptLangToEditorLang(untrack(() => scriptLang)))
// On a postgres script targeting a datatable, DDL statements are intercepted
// on run (cmd+enter) and offered as migrations instead.
let datatableForMigrations = $derived(
scriptLang === 'postgresql' &&
typeof args?.database === 'string' &&
args.database.startsWith('datatable://')
? args.database.slice('datatable://'.length).split('/')[0]
: undefined
)
let ddlGuard = $state<DdlMigrationGuard | undefined>(undefined)
// Run the DDL migration guard against the current code. Returns false when the
// user cancels (the run must be aborted); may rewrite the code (migrated
// statements stripped). Exported so run paths that bypass the Monaco
// Cmd+Enter binding (e.g. the Test button) can guard too.
export async function guardDdlBeforeRun(): Promise<boolean> {
if (datatableForMigrations && ddlGuard) {
const res = await ddlGuard.guard(getCode())
if (!res.proceed) return false
if (res.code !== getCode()) setCode(res.code)
}
return true
}
async function runCmdEnterWithDdlGuard() {
if (!(await guardDdlBeforeRun())) return
cmdEnterAction?.()
}
let filePath = $state(computePath(untrack(() => path)))
let initialPath: string | undefined = $state(untrack(() => path))
let websockets: WebSocket[] = []
let languageClients: MonacoLanguageClient[] = []
let websocketInterval: number | undefined
let lastWsAttempt: Date = new Date()
let nbWsAttempt = 0
let disposeMethod: (() => void) | undefined
const absolutePathExtraLibs = new Map<string, { dispose: () => void }>()
const dispatch = createEventDispatcher()
// Set by the sessions pane; undefined everywhere else. ⌘L targets it so the
// shortcut reaches the chat the user is actually looking at. Read directly
// rather than via getAiChatManager(), which collapses "no session" into the
// singleton — the distinction is what tells ⌘L whether a pane exists to open.
const sessionScopedChatManager = getContext<AIChatManager | undefined>('aiChatManager')
// let graphqlService: MonacoGraphQLAPI | undefined = undefined
let fetchedDbSchema: DBSchema | undefined = $state(undefined)
let dbSchema: DBSchema | undefined = $derived(sqlSchema ?? fetchedDbSchema)
let destroyed = false
const uri = computeUri(
untrack(() => filePath),
untrack(() => scriptLang)
)
console.log('uri', uri)
function computeUri(filePath: string, scriptLang: string | undefined) {
let file
if (filePath.includes('.')) {
file = filePath
} else {
file = `${filePath}.${scriptLang == 'tsx' ? 'tsx' : langToExt(lang)}`
}
if (file.startsWith('/')) {
file = file.slice(1)
}
return !['deno', 'go', 'python3'].includes(scriptLang ?? '')
? `file:///${file}`
: `file:///tmp/monaco/${file}`
}
function computePath(path: string | undefined): string {
if (
['deno', 'go', 'python3'].includes(scriptLang ?? '') ||
path == '' ||
path == undefined //||path.startsWith('/')
) {
return randomHash()
} else {
// console.log('path', path)
return path as string
}
}
export function switchToFile(path: string, value: string, lang: string) {
if (editor) {
const uri = mUri.parse(path)
console.log('switching to file', path, lang)
// vscode.workspace.fs.writeFile(uri, new TextEncoder().encode(value))
let nmodel = meditor.getModel(uri)
if (nmodel) {
console.log('using existing model', path)
editor.setModel(nmodel)
} else {
console.log('creating model', path)
nmodel = meditor.createModel(value, lang, uri)
editor.setModel(nmodel)
}
model = nmodel
setTypescriptExtraLibs()
}
}
let valueAfterDispose: string | undefined = undefined
export function getCode(): string {
if (valueAfterDispose != undefined) {
return valueAfterDispose
}
return editor?.getValue() ?? ''
}
export function getModel(): meditor.IEditorModel | undefined {
return editor?.getModel() ?? undefined
}
export function insertAtCursor(code: string): void {
if (editor) {
editor.trigger('keyboard', 'type', { text: code })
}
}
export function insertAtCurrentLine(code: string): void {
if (editor) {
insertAtLine(code, editor.getPosition()?.lineNumber ?? 0)
}
}
export function arrowDown(): void {
if (editor) {
let pos = editor.getPosition()
if (pos) {
editor.setPosition({ lineNumber: pos.lineNumber + 1, column: pos.column })
}
}
}
export function backspace(): void {
if (editor) {
editor.trigger('keyboard', 'deleteLeft', {})
}
}
export function insertAtBeginning(code: string): void {
if (editor) {
const range = { startLineNumber: 1, startColumn: 1, endLineNumber: 1, endColumn: 1 }
const op = { range: range, text: code, forceMoveMarkers: true }
editor.executeEdits('external', [op])
}
}
export function insertAtLine(code: string, line: number): void {
if (editor) {
const range = { startLineNumber: line, startColumn: 1, endLineNumber: line, endColumn: 1 }
const op = { range: range, text: code, forceMoveMarkers: true }
editor.executeEdits('external', [op])
}
}
export function getSelectedLines(): string | undefined {
if (editor) {
const selection = editor.getSelection()
if (selection) {
const range: IRange = {
startLineNumber: selection.startLineNumber,
startColumn: 1,
endLineNumber: selection.endLineNumber + 1,
endColumn: 1
}
return editor.getModel()?.getValueInRange(range)
}
}
}
export function onDidChangeCursorSelection(f: (e: meditor.ICursorSelectionChangedEvent) => void) {
if (editor) {
return editor.onDidChangeCursorSelection(f)
}
}
export function show(): void {
divEl?.classList.remove('hidden')
}
export function hide(): void {
divEl?.classList.add('hidden')
}
export function setCode(ncode: string, noHistory: boolean = false): void {
// Track whether the code actually changed before updating.
const changed = code != ncode
if (changed) {
code = ncode
}
// setCode is an authoritative overwrite (reset, AI apply, module switch).
// Cancel any in-flight keystroke debounce first: otherwise alignCodeWithEditor
// skips on the `timeoutModel` guard (leaving Monaco stale), and the pending
// updateCode later reads the old buffer and writes it back over `ncode`.
cancelPendingChanges()
alignCodeWithEditor(!noHistory)
// Dispatch change immediately when code actually changed. This ensures
// callers like the Reset button and copilot trigger on:change handlers.
// The debounced onDidChangeModelContent handler will no-op since code
// will already match by the time it fires.
if (changed) {
dispatch('change', ncode)
}
updateRawAppLintDiagnostics()
}
/** Collect Monaco markers and update the raw app lint store */
function updateRawAppLintDiagnostics(): void {
if (!rawAppRunnableKey || !model) return
const markers = meditor.getModelMarkers({ resource: model.uri })
const lintErrors: MonacoLintError[] = markers
.filter((m) => m.severity === MarkerSeverity.Error || m.severity === MarkerSeverity.Warning)
.map((m) => ({
message: m.message,
severity: m.severity === MarkerSeverity.Error ? 'error' : 'warning',
startLineNumber: m.startLineNumber,
startColumn: m.startColumn,
endLineNumber: m.endLineNumber,
endColumn: m.endColumn
}))
rawAppLintStore.setDiagnostics(rawAppRunnableKey, lintErrors)
}
function updateCode() {
const ncode = getCode()
if (code == ncode) {
return
}
code = ncode
lastEditorCode = ncode
dispatch('change', ncode)
}
/** Force-materialize the latest Monaco content into the bindable
* `code` prop right now, bypassing the trailing debounce. Use for
* explicit "save now" shortcuts (Ctrl/Cmd+S) — without this, anything
* the user typed within the last `changeTimeout` ms is still sitting
* in Monaco's buffer and downstream consumers (autosave, lint) won't
* see it. Clears the chain state so the next keystroke after this
* flush is a fresh leading fire. */
export function flushPendingChanges(): void {
cancelPendingChanges()
updateCode()
}
/** Discard any in-flight keystroke debounce without materializing it, so a
* deferred updateCode can't fire later. Resets chain state to a fresh leading
* fire on the next keystroke. */
function cancelPendingChanges(): void {
if (timeoutModel !== undefined) {
clearTimeout(timeoutModel)
timeoutModel = undefined
}
changeChainStart = undefined
}
export function append(code: string): void {
if (editor) {
const lineCount = editor.getModel()?.getLineCount() || 0
const lastLineLength = editor.getModel()?.getLineLength(lineCount) || 0
const range: IRange = {
startLineNumber: lineCount,
startColumn: lastLineLength + 1,
endLineNumber: lineCount,
endColumn: lastLineLength + 1
}
editor.executeEdits('append', [
{
range,
text: code,
forceMoveMarkers: true
}
])
editor.revealLine(lineCount)
}
}
export async function format() {
if (editor) {
updateCode()
if (lang != 'shell' && lang != 'nu') {
if ($formatOnSave != false) {
if (scriptLang == 'deno' && languageClients.length > 0) {
languageClients.forEach(async (x) => {
let edits = await x.sendRequest(new RequestType('textDocument/formatting'), {
textDocument: { uri },
options: {
tabSize: 2,
insertSpaces: true
}
})
console.debug(edits)
if (Array.isArray(edits)) {
edits = edits.map((edit) =>
edit.range.start != undefined &&
edit.range.end != undefined &&
edit.newText != undefined
? {
range: {
startLineNumber: edit.range.start.line + 1,
startColumn: edit.range.start.character + 1,
endLineNumber: edit.range.end.line + 1,
endColumn: edit.range.end.character + 1
},
text: edit.newText
}
: {}
)
//@ts-ignore
editor?.executeEdits('fmt', edits)
}
})
} else {
await editor?.getAction('editor.action.formatDocument')?.run()
}
}
updateCode()
}
if (formatAction) {
formatAction()
}
}
}
export function getScriptLang(): string | undefined {
return scriptLang
}
export function getEditor(): meditor.IStandaloneCodeEditor | null {
return editor
}
/** Get lint errors and warnings from the Monaco editor */
export function getLintErrors(): ScriptLintResult {
if (!model) {
return { errorCount: 0, warningCount: 0, errors: [], warnings: [] }
}
const markers = meditor.getModelMarkers({ resource: model.uri })
const errors = markers.filter((m) => m.severity === MarkerSeverity.Error)
const warnings = markers.filter((m) => m.severity === MarkerSeverity.Warning)
return {
errorCount: errors.length,
warningCount: warnings.length,
errors,
warnings
}
}
let command: IDisposable | undefined = undefined
let sqlTypeCompletor: IDisposable | undefined = $state(undefined)
let resultCollectionCompletor: IDisposable | undefined = $state(undefined)
function addSqlTypeCompletions() {
sqlTypeCompletor?.dispose()
resultCollectionCompletor?.dispose()
resultCollectionCompletor = languages.registerCompletionItemProvider('sql', {
triggerCharacters: ['='],
provideCompletionItems: function (model, position) {
const lineContent = model.getLineContent(position.lineNumber)
const match = lineContent.match(/^--\s*result_collection=/)
if (!match) {
return { suggestions: [] }
}
const word = model.getWordUntilPosition(position)
const range = {
startLineNumber: position.lineNumber,
endLineNumber: position.lineNumber,
startColumn: word.startColumn,
endColumn: word.endColumn
}
const suggestions = [
'last_statement_all_rows',
'last_statement_first_row',
'last_statement_all_rows_scalar',
'last_statement_first_row_scalar',
'all_statements_all_rows',
'all_statements_first_row',
'all_statements_all_rows_scalar',
'all_statements_first_row_scalar'
].map((label) => ({
label: label,
kind: languages.CompletionItemKind.Function,
insertText: label,
range,
sortText: 'a'
}))
return { suggestions }
}
})
sqlTypeCompletor = languages.registerCompletionItemProvider('sql', {
triggerCharacters: scriptLang === 'postgresql' ? [':'] : ['('],
provideCompletionItems: function (model, position) {
const lineUntilPosition = model.getValueInRange({
startLineNumber: position.lineNumber,
startColumn: 1,
endLineNumber: position.lineNumber,
endColumn: position.column
})
let suggestions: languages.CompletionItem[] = []
if (
scriptLang === 'postgresql'
? lineUntilPosition.endsWith('::')
: lineUntilPosition.match(/^-- .* \(/)
) {
const word = model.getWordUntilPosition(position)
const range = {
startLineNumber: position.lineNumber,
endLineNumber: position.lineNumber,
startColumn: word.startColumn,
endColumn: word.endColumn
}
suggestions = (
scriptLang === 'postgresql'
? POSTGRES_TYPES
: scriptLang === 'mysql'
? MYSQL_TYPES
: scriptLang === 'snowflake'
? SNOWFLAKE_TYPES
: scriptLang === 'bigquery'
? BIGQUERY_TYPES
: scriptLang === 'mssql'
? MSSQL_TYPES
: scriptLang === 'oracledb'
? ORACLEDB_TYPES
: scriptLang === 'duckdb'
? DUCKDB_TYPES
: []
).map((t) => ({
label: t,
kind: languages.CompletionItemKind.Function,
insertText: t,
range: range,
sortText: 'a'
}))
}
return {
suggestions
}
}
})
}
let workspaceMacroCompletor: IDisposable | undefined = undefined
// Workspace DuckDB macros (deployed `// macros` libraries): suggest each
// macro while typing an identifier, with its signature + body as docs and
// a snippet insert that parks the cursor inside the call parens. Fetched
// via a short-TTL cache — macros are late-bound, so mild staleness is fine.
async function addWorkspaceMacroCompletions() {
workspaceMacroCompletor?.dispose()
const workspace = $operatingWorkspace
if (!workspace) return
let macros: Awaited<ReturnType<typeof listWorkspaceMacrosCached>> = []
try {
macros = await listWorkspaceMacrosCached(workspace)
} catch (e) {
console.error('error listing workspace macros', e)
return
}
if (macros.length === 0) return
workspaceMacroCompletor = languages.registerCompletionItemProvider('sql', {
provideCompletionItems: function (model, position) {
const word = model.getWordUntilPosition(position)
const range = {
startLineNumber: position.lineNumber,
endLineNumber: position.lineNumber,
startColumn: word.startColumn,
endColumn: word.endColumn
}
const suggestions = macros.map((m) => ({
label: `${m.name}(${m.params})`,
kind: m.is_table
? languages.CompletionItemKind.Interface
: languages.CompletionItemKind.Function,
detail: `${m.is_table ? 'table macro' : 'macro'} · ${m.provider_path}`,
documentation: {
value: '```sql\n' + macroDefinitionSql(m) + '\n```'
},
insertText: `${m.name}($0)`,
insertTextRules: languages.CompletionItemInsertTextRule.InsertAsSnippet,
filterText: m.name,
range,
sortText: 'b' + m.name
}))
return { suggestions }
}
})
}
let schemaContractCompletor: IDisposable | undefined = undefined
// Column-name completion for pipeline annotation refs (`// column out <-
// ducklake://lake/orders.|`, `// data_test relationships col ->
// ducklake://lake/customers.|`): suggests the referenced asset's *captured*
// columns (with types) so a broken ref never gets typed — the prevention
// side of the schema-contract check. Only fires on annotation comment lines
// with a ducklake URI right before the cursor's `.`; schemas come from the
// short-TTL contract cache, so per-keystroke cost is a map lookup.
function addSchemaContractCompletions() {
schemaContractCompletor?.dispose()
schemaContractCompletor = languages.registerCompletionItemProvider(lang, {
triggerCharacters: ['.'],
provideCompletionItems: async function (model, position) {
// Read the store per request, not at registration — the provider
// outlives a workspace switch.
const workspace = $operatingWorkspace
if (!workspace) return { suggestions: [] }
const before = model.getLineContent(position.lineNumber).slice(0, position.column - 1)
if (!/^\s*(\/\/|--|#)\s*(column|data_test|on|materialize)\b/.test(before)) {
return { suggestions: [] }
}
const uri = before.match(/ducklake:\/\/([\w/.{}-]+?)\.$/)
if (!uri) return { suggestions: [] }
const schema = await fetchLatestSchema(workspace, normalizeAssetPath(uri[1]))
if (!schema) return { suggestions: [] }
const word = model.getWordUntilPosition(position)
const range = {
startLineNumber: position.lineNumber,
endLineNumber: position.lineNumber,
startColumn: word.startColumn,
endColumn: word.endColumn
}
return {
suggestions: schema.columns.map((c) => ({
label: c.name,
kind: languages.CompletionItemKind.Field,
detail: `${c.type} · captured schema v${schema.version}`,
insertText: c.name,
range,
sortText: 'a' + c.name
}))
}
}
})
}
let sqlSchemaCompletor: IDisposable | undefined = undefined
async function updateSchema(newSchemaRes: string | undefined) {
if (typeof newSchemaRes === 'string') {
const resourcePath = newSchemaRes.replace('$res:', '')
fetchedDbSchema = $dbSchemas[resourcePath]
if (fetchedDbSchema === undefined) {
$dbSchemas[resourcePath] = await getDbSchemas(
lang === 'graphql' ? 'graphql' : (scriptLang ?? ''),
resourcePath,
$operatingWorkspace,
(e) => console.error(`error getting ${lang} (${scriptLang}) db schema`, e),
{ customTag }
)
}
fetchedDbSchema = $dbSchemas[resourcePath]
} else {
fetchedDbSchema = undefined
}
}
function disposeSqlSchemaCompletor() {
sqlSchemaCompletor?.dispose()
}
function disposeGaphqlService() {
// graphqlService = undefined
}
function addDBSchemaCompletions() {
const { lang: schemaLang, schema } = dbSchema || {}
if (!schemaLang || !schema) {
return
}
console.log('adding db schema completions', schemaLang)
if (schemaLang === 'graphql') {
//graphql depreciated until https://github.com/graphql/graphiql/issues/4104 is fixed with monaco > 0.52.2
// languages.register({ id: 'graphql' })
// graphqlService ||= initializeMode()
// console.log('setting schema config', schema)
// graphqlService?.setSchemaConfig([
// {
// uri: 'my-schema.graphql',
// introspectionJSON: schema
// }
// ])
} else {
if (sqlSchemaCompletor) {
sqlSchemaCompletor.dispose()
}
sqlSchemaCompletor = languages.registerCompletionItemProvider('sql', {
triggerCharacters: ['.', ' ', '('],
provideCompletionItems: function (model, position) {
// Registered for the whole `sql` language: every other mounted SQL editor
// would otherwise add its own schema's suggestions to this one's.
if (model !== editor?.getModel()) return { suggestions: [] }
const textUntilPosition = model.getValueInRange({
startLineNumber: 1,
startColumn: 1,
endLineNumber: position.lineNumber,
endColumn: position.column
})
const word = model.getWordUntilPosition(position)
const range = {
startLineNumber: position.lineNumber,
endLineNumber: position.lineNumber,
startColumn: word.startColumn,
endColumn: word.endColumn
}
let suggestions: languages.CompletionItem[] = []
const noneMatch = textUntilPosition.match(/(?:add|create table)\s/i)
if (noneMatch) {
return {
suggestions
}
}
for (const schemaKey in schema) {
suggestions.push({
label: schemaKey,
detail: 'schema',
kind: languages.CompletionItemKind.Function,
insertText: schemaKey,
range: range,
sortText: 'z'
})
for (const tableKey in schema[schemaKey]) {
suggestions.push({
label: tableKey,
detail: `table (${schemaKey})`,
kind: languages.CompletionItemKind.Function,
insertText: tableKey,
range: range,
sortText: 'y'
})
const noColsMatch = textUntilPosition.match(
/(?:from|insert into|update|table)\s(?![\s\S]*(\b(where|order by|group by|values|set|column)\b|\())/i
)
if (!noColsMatch) {
for (const columnKey in schema[schemaKey][tableKey]) {
suggestions.push({
label: columnKey,
detail: `${schema[schemaKey][tableKey][columnKey]['type']} (${schemaKey}.${tableKey})`,
kind: languages.CompletionItemKind.Function,
insertText: columnKey,
range: range,
sortText: 'x'
})
}
}
if (textUntilPosition.match(new RegExp(`${tableKey}.$`, 'i'))) {
suggestions = suggestions.filter((x) =>
x.detail?.includes(`(${schemaKey}.${tableKey})`)
)
return {
suggestions
}
}
}
if (textUntilPosition.match(new RegExp(`${schemaKey}.$`, 'i'))) {
suggestions = suggestions.filter((x) => x.detail === `table (${schemaKey})`)
return {
suggestions
}
}
}
return {
suggestions
}
}
})
}
}
let preprocessorCompletor: IDisposable | undefined = undefined
function addPreprocessorCompletions(lang: string) {
if (preprocessorCompletor) {
preprocessorCompletor.dispose()
}
const windmillLang = lang === 'typescript' ? 'deno' : lang === 'python' ? 'python3' : lang
const preprocessorCode = getPreprocessorModuleCode(windmillLang as ScriptLang)
if (!preprocessorCode) {
return
}
preprocessorCompletor = languages.registerCompletionItemProvider(lang, {
provideCompletionItems: function (model, position) {
const word = model.getWordUntilPosition(position)
if (word.word.length >= 3 && 'preprocessor'.startsWith(word.word)) {
const range = {
startLineNumber: position.lineNumber,
endLineNumber: position.lineNumber,
startColumn: word.startColumn,
endColumn: word.endColumn
}
return {
suggestions: [
{
label: 'preprocessor (windmill)',
kind: languages.CompletionItemKind.Function,
insertTextRules: languages.CompletionItemInsertTextRule.InsertAsSnippet,
insertText: preprocessorCode,
range,
additionalTextEdits: [
{
range: {
startLineNumber: position.lineNumber,
endLineNumber: position.lineNumber,
startColumn: 0,
endColumn: word.startColumn
},
text: ''
}
]
}
]
}
}
return {
suggestions: []
}
}
})
}
let reviewingChanges = $state(writable(false))
let aiChatEditorHandler: AIChatEditorHandler | undefined = $state(undefined)
// Inline ai chat widget
let showInlineAIChat = $state(false)
let inlineAIChatSelection: Selection | null = $state(null)
let selectedCode = $state('')
export async function reviewAndApplyCode(code: string, opts?: ReviewChangesOpts) {
await aiChatEditorHandler?.reviewChanges(code, opts)
}
export async function reviewAppliedCode(
originalCode: string,
opts?: { onFinishedReview?: () => void }
) {
await aiChatEditorHandler?.reviewChanges(originalCode, {
mode: 'revert',
onFinishedReview: opts?.onFinishedReview
})
}
export function getAiChatEditorHandler() {
return aiChatEditorHandler
}
function addChatHandler(editor: meditor.IStandaloneCodeEditor) {
try {
aiChatEditorHandler = new AIChatEditorHandler(editor)
reviewingChanges = aiChatEditorHandler.reviewingChanges
} catch (err) {
console.error('Could not add chat handler', err)
}
}
let autocompletor: Autocompletor | undefined = $state(undefined)
function addAutoCompletor(
editor: meditor.IStandaloneCodeEditor,
scriptLang: ScriptLang | 'bunnative' | 'jsx' | 'tsx' | 'json',
workflowAsCode: boolean
) {
if (autocompletor) {
autocompletor.dispose()
}
autocompletor = new Autocompletor(editor, scriptLang, { workflowAsCode })
}
const outputChannel = {
name: 'Language Server Client',
appendLine: (msg: string) => {
console.log(msg)
},
append: (msg: string) => {
console.log(msg)
},
clear: () => {},
replace: () => {},
show: () => {},
hide: () => {},
dispose: () => {}
}
export async function reloadWebsocket() {
await closeWebsockets()
if (
!useWebsockets ||
!(
(lang == 'typescript' && scriptLang === 'deno') ||
lang == 'python' ||
lang == 'go' ||
lang == 'shell'
)
) {
return
}
console.log('reloadWebsocket')
function createLanguageClient(
transports: MessageTransports,
name: string,
initializationOptions: any,
middlewareOptions: ((params, token, next) => any) | undefined
) {
const client = new MonacoLanguageClient({
name: name,
messageTransports: transports,
clientOptions: {
outputChannel,
documentSelector: [lang],
errorHandler: {
error: () => ({ action: ErrorAction.Continue }),
closed: () => ({
action: CloseAction.Restart
})
},
markdown: {
isTrusted: true
},
workspaceFolder:
name != 'deno'
? {
uri: vscode.Uri.parse(uri),
name: 'windmill',
index: 0
}
: undefined,
initializationOptions,
middleware: {
workspace: {
configuration:
middlewareOptions ??
((params, token, next) => {
return [{ enabled: true }]
})
}
}
}
})
return client
}
async function connectToLanguageServer(
url: string,
name: string,
initOptions: any,
middlewareOptions: any
) {
try {
const webSocket = new WebSocket(url)
websockets.push(webSocket)
webSocket.onopen = async () => {
const socket = toSocket(webSocket)
const reader = new WebSocketMessageReader(socket)
const writer = new WebSocketMessageWriter(socket)
const languageClient = createLanguageClient(
{ reader, writer },
name,
initOptions,
middlewareOptions
)
// if (middlewareOptions != undefined) {
// languageClient.registerNotUsedFeatures()
// }
const om = webSocket.onmessage
webSocket.onmessage = (e) => {
om && om.apply(webSocket, [e])
if (destroyed) {
webSocket.close()
console.log('Stopping client early because of mismatch')
}
}
languageClients.push(languageClient)
// HACK ALERT: for some reasons, the client need to be restarted to take into account the 'go get <dep>' command
// the only way I could figure out to listen for this event is this. I'm sure there is a better way to do this
if (name == 'go') {
const om = webSocket.onmessage
webSocket.onmessage = (e) => {
om && om.apply(webSocket, [e])
const js = JSON.parse(e.data)
if (js.method == 'window/showMessage' && js.params.message == 'completed') {
console.log('reloading websocket after go get')
reloadWebsocket()
}
}
}
reader.onClose(async () => {
try {
console.log('CLOSE')
websocketAlive[name] = false
await languageClient.stop()
} catch (err) {
console.error(err)
}
})
socket.onClose((_code, _reason) => {
websocketAlive[name] = false
})
try {
console.log('starting client')
await languageClient.start()
// for python we want to use the pyright client for signature help, not ruff
if (lang !== 'python' || (lang === 'python' && name == 'pyright')) {
autocompletor?.setLanguageClient(languageClient)
}
console.log('started client')
} catch (err) {
console.log('err at client')
console.error(err)
return
}
lastWsAttempt = new Date()
nbWsAttempt = 0
if (name == 'deno') {
command && command.dispose()
command = undefined
try {
command = vscode.commands.registerCommand(
'deno.cache',
(uris: DocumentUri[] = []) => {
languageClient.sendRequest(new RequestType('deno/cache'), {
referrer: { uri },
uris: uris.map((uri) => ({ uri }))
})
}
)
} catch (err) {
console.warn(err)
}
}
websocketAlive[name] = true
}
} catch (err) {
console.error(`connection to ${name} language server failed`)
}
}
const hostname = getHostname()
let encodedImportMap = ''
if (useWebsockets) {
if (lang == 'typescript' && scriptLang === 'deno') {
ata = undefined
let root = await genRoot(hostname)
const importMap = {
imports: {
'file:///': root + '/'
}
}
if (filePath && filePath.split('/').length > 2) {
let path_splitted = filePath.split('/')
for (let c = 0; c < path_splitted.length; c++) {
let key = 'file://./'
for (let i = 0; i < c; i++) {
key += '../'
}
let url = path_splitted.slice(0, -c - 1).join('/')
let ending = c == path_splitted.length - 1 ? '' : '/'
importMap['imports'][key] = `${root}/${url}${ending}`
}
}
encodedImportMap = 'data:text/plain;base64,' + btoa(JSON.stringify(importMap))
await connectToLanguageServer(
buildWsUrl('/ws/deno'),
'deno',
{
certificateStores: null,
enablePaths: [],
config: null,
importMap: encodedImportMap,
internalDebug: false,
lint: false,
path: null,
tlsCertificate: null,
unsafelyIgnoreCertificateErrors: null,
unstable: true,
enable: true,
codeLens: {
implementations: true,
references: true,
referencesAllFunction: false
},
suggest: {
autoImports: true,
completeFunctionCalls: false,
names: true,
paths: true,
imports: {
autoDiscover: true,
hosts: {
'https://deno.land': true
}
}
}
},
() => {
return [
{
enable: true
}
]
}
)
} else if (lang === 'python') {
await connectToLanguageServer(
buildWsUrl('/ws/pyright'),
'pyright',
{},
(params, token, next) => {
if (params.items.find((x) => x.section === 'python')) {
return [
{
analysis: {
useLibraryCodeForTypes: true,
autoImportCompletions: true,
diagnosticSeverityOverrides: { reportMissingImports: 'none' },
typeCheckingMode: 'basic'
}
}
]
}
if (params.items.find((x) => x.section === 'python.analysis')) {
return [
{
useLibraryCodeForTypes: true,
autoImportCompletions: true,
diagnosticSeverityOverrides: { reportMissingImports: 'none' },
typeCheckingMode: 'basic'
}
]
}
return next(params, token)
}
)
// `ruff server` resolves settings per workspace folder, and the folder registered
// below is the document URI rather than a directory containing it, so a ruff.toml
// on disk never applies unless named here (written by lsp/pyls_launcher.py).
connectToLanguageServer(
buildWsUrl('/ws/ruff'),
'ruff',
{ settings: { configuration: '/tmp/monaco/ruff.toml' } },
undefined
)
} else if (lang === 'go') {
connectToLanguageServer(
buildWsUrl('/ws/go'),
'go',
{
'build.allowImplicitNetworkAccess': true
},
undefined
)
} else if (lang === 'shell') {
connectToLanguageServer(
buildWsUrl('/ws/diagnostic'),
'shellcheck',
{
linters: {
shellcheck: {
command: 'shellcheck',
debounce: 100,
args: ['--format=gcc', '-'],
offsetLine: 0,
offsetColumn: 0,
sourceName: 'shellcheck',
formatLines: 1,
formatPattern: [
'^[^:]+:(\\d+):(\\d+):\\s+([^:]+):\\s+(.*)$',
{
line: 1,
column: 2,
message: 4,
security: 3
}
],
securities: {
error: 'error',
warning: 'warning',
note: 'info'
}
}
},
filetypes: {
shell: 'shellcheck'
}
},
undefined
)
} else {
closeWebsockets()
}
websocketInterval && clearInterval(websocketInterval)
websocketInterval = setInterval(() => {
if (document.visibilityState == 'visible') {
if (
!lastWsAttempt ||
(new Date().getTime() - lastWsAttempt.getTime() > 60000 && nbWsAttempt < 2)
) {
if (
!websocketAlive.deno &&
!websocketAlive.pyright &&
!websocketAlive.go &&
!websocketAlive.shellcheck &&
!websocketAlive.ruff &&
scriptLang != 'bun' &&
scriptLang != 'tsx'
) {
console.log('reconnecting to language servers')
lastWsAttempt = new Date()
nbWsAttempt++
reloadWebsocket()
} else {
if (nbWsAttempt >= 2) {
sendUserToast('Giving up on establishing smart assistant connection', true)
clearInterval(websocketInterval)
}
}
}
}
}, 5000)
}
}
let pathTimeout: number | undefined = undefined
let yPadding = MONACO_Y_PADDING
function getHostname() {
return BROWSER ? window.location.protocol + '//' + window.location.host : 'SSR'
}
function handlePathChange() {
console.log('path changed, reloading language server', initialPath, path)
initialPath = path
pathTimeout && clearTimeout(pathTimeout)
ata = undefined
pathTimeout = setTimeout(reloadWebsocket, 1000)
}
async function closeWebsockets() {
command && command.dispose()
command = undefined
console.debug(`disposing ${websockets.length} language clients and closing websockets`)
for (const x of languageClients) {
try {
await x.dispose()
} catch (err) {
console.debug('error disposing language client', err)
}
}
languageClients = []
for (const x of websockets) {
try {
await x.close()
} catch (err) {
console.debug('error closing websocket', err)
}
}
console.debug('done closing websockets')
websockets = []
websocketInterval && clearInterval(websocketInterval)
}
// let widgets: HTMLElement | undefined = document.getElementById('monaco-widgets-root') ?? undefined
let model: meditor.ITextModel | undefined = $state(undefined)
let monacoBinding: MonacoBinding | undefined = $state(undefined)
let initialized = $state(false)
let ata: ((s: string | DepsToGet) => void) | undefined = undefined
let statusDiv: Element | null = $state(null)
function saveDraft() {
dispatch('saveDraft', code)
}
let vimDisposable: IDisposable | undefined = $state(undefined)
function onVimDisable() {
vimDisposable?.dispose()
}
function onVimMode() {
if (editor && statusDiv) {
vimDisposable = initVim(editor, statusDiv, saveDraft)
}
}
let svelteRegistered = false
let vueRegistered = false
function onFileChanges() {
if (files && Object.keys(files).find((x) => x.endsWith('.svelte')) != undefined) {
if (!svelteRegistered) {
svelteRegistered = true
languages.register({
id: 'svelte',
extensions: ['.svelte'],
aliases: ['Svelte', 'svelte'],
mimetypes: ['application/svelte']
})
languages.setLanguageConfiguration('svelte', htmllang.conf as any)
languages.setMonarchTokensProvider('svelte', htmllang.language as any)
}
}
if (files && Object.keys(files).find((x) => x.endsWith('.vue')) != undefined) {
if (!vueRegistered) {
vueRegistered = true
languages.register({
id: 'vue',
extensions: ['.vue'],
aliases: ['Vue', 'Vue'],
mimetypes: ['application/svelte']
})
languages.setLanguageConfiguration('vue', conf as any)
languages.setMonarchTokensProvider('vue', language as any)
}
}
if (files && model) {
for (const [path, { code, readonly }] of Object.entries(files)) {
const luri = mUri.file(path)
if (luri.toString() != model.uri.toString()) {
let nmodel = meditor.getModel(luri)
if (nmodel == undefined) {
const lmodel = meditor.createModel(code, extToLang(path?.split('.')?.pop()!), luri)
if (readonly) {
lmodel.onDidChangeContent((evt) => {
// This will effectively undo any new edits
if (lmodel.getValue() != code && code) {
lmodel.setValue(code)
}
})
}
} else {
const lmodel = meditor.getModel(luri)
if (lmodel && code) {
lmodel.setValue(code)
}
}
}
}
}
}
let timeoutModel: number | undefined = undefined
/** Wall-clock start (ms) of the current debounce chain. Reset whenever
* the trailing fire lands — so a typing burst → pause → typing burst
* gets a fresh leading fire instead of inheriting the previous cap. */
let changeChainStart: number | undefined = undefined
async function loadMonaco() {
setMonacoTypescriptOptions()
console.log('path', uri)
try {
console.log("Loading Monaco's language client")
await initializeVscode('editor', divEl!)
console.log('done loading Monaco and vscode')
} catch (e) {
console.log('error initializing services', e)
}
// Unmounted while the services loaded: the container is detached, and
// creating the editor on it throws.
if (destroyed || !divEl?.isConnected) return
// vscode.languages.registerDefinitionProvider('*', {
// provideDefinition(document, position, token) {
// // Get the word under the cursor (this will be the import or function being clicked)
// const wordRange = document.getWordRangeAtPosition(position)
// const word = document.getText(wordRange)
// // Do something with the word (for example, log it or handle it)
// console.log('Clicked on import or symbol:', word)
// // Optionally, you can also return a definition location
// return null // If you don't want to override the default behavior
// }
// })
// console.log('bef ready')
// console.log('af ready')
initialized = true
try {
model = meditor.createModel(code ?? '', lang == 'nu' ? 'python' : lang, mUri.parse(uri))
} catch (err) {
console.log('model already existed', err)
const nmodel = meditor.getModel(mUri.parse(uri))
if (!nmodel) {
throw err
}
model = nmodel
}
model.updateOptions(lang == 'python' ? { tabSize: 4, insertSpaces: true } : updateOptions)
onFileChanges()
try {
editor = meditor.create(divEl as HTMLDivElement, {
...editorConfig(
code ?? '',
lang,
automaticLayout,
fixedOverflowWidgets,
$relativeLineNumbers
),
model,
fontSize: small ? editorFontSize.small : editorFontSize.regular,
lineNumbersMinChars,
// overflowWidgetsDomNode: widgets,
tabSize: lang == 'python' ? 4 : 2,
folding,
padding: { bottom: yPadding, top: yPadding }
})
if (key && editorPositionMap?.[key]) {
editor.setPosition(editorPositionMap[key])
editor.revealPositionInCenterIfOutsideViewport(editorPositionMap[key])
}
} catch (e) {
console.error('Error loading monaco:', e)
return
}
keepModelAroundToAvoidDisposalOfWorkers()
// In VSCode webview (iframe), clipboard operations need special handling
// because the webview has restricted clipboard API access
if (window.parent !== window) {
editor.addCommand(KeyMod.CtrlCmd | KeyCode.KeyC, function () {
document.execCommand('copy')
})
editor.addCommand(KeyMod.CtrlCmd | KeyCode.KeyX, function () {
document.execCommand('cut')
})
// Paste is scoped to this editor's container instead of a global
// Ctrl+V keybinding, which would leak across editor instances.
pasteCleanup?.()
pasteCleanup = registerWebviewPaste(divEl, () => editor)
}
// updateEditorKeybindingsMode(editor, 'vim', undefined)
// Raw app lint collection: listen for marker changes and report to store
let markerChangeDisposable: IDisposable | undefined = undefined
if (rawAppRunnableKey && model) {
markerChangeDisposable = meditor.onDidChangeMarkers((uris) => {
if (!model || !rawAppRunnableKey) return
const modelUri = model.uri.toString()
if (uris.some((u) => u.toString() === modelUri)) {
updateRawAppLintDiagnostics()
}
})
// Initial lint diagnostics collection
updateRawAppLintDiagnostics()
}
let ataModel: number | undefined = undefined
editor?.onDidChangeModelContent((event) => {
// Leading fire on the first keystroke of a burst: every
// downstream consumer (autosave's 1.5s debouncer, the
// `bind:code` chain, change listeners) sees text within the
// same tick instead of after `changeTimeout` ms of silence.
// Subsequent keystrokes within the burst are trailing-only
// (debounced by `changeTimeout`), with a hard ceiling at
// `chainStart + maxChangeTimeout` so continuous typing still
// materializes at least once per `maxChangeTimeout` window.
const now = Date.now()
if (changeChainStart === undefined) {
updateCode()
changeChainStart = now
}
timeoutModel && clearTimeout(timeoutModel)
const fireAt = Math.min(now + changeTimeout, changeChainStart + maxChangeTimeout)
timeoutModel = setTimeout(
() => {
updateCode()
timeoutModel = undefined
changeChainStart = undefined
},
Math.max(0, fireAt - now)
)
ataModel && clearTimeout(ataModel)
ataModel = setTimeout(() => {
if (scriptLang == 'bun' || scriptLang == 'bunnative') {
ata?.(getCode())
}
}, 1000)
})
editor?.onDidBlurEditorText(() => {
dispatch('blur')
})
editor?.onDidChangeCursorPosition((event) => {
if (key) editorPositionMap[key] = event.position
})
editor?.onDidFocusEditorText(() => {
dispatch('focus')
// for escape we use onkeydown instead of addCommand because addCommand on escape specifically prevents default behavior (like autocomplete cancellation)
editor?.onKeyDown((e) => {
if (e.keyCode === KeyCode.Escape) {
if (showInlineAIChat) {
closeAIInlineWidget()
}
aiChatEditorHandler?.rejectAll()
}
})
editor?.addCommand(KeyMod.CtrlCmd | KeyCode.DownArrow, function () {
if (aiChatManager.pendingNewCode) {
aiChatManager.scriptEditorApplyCode?.(aiChatManager.pendingNewCode)
if (showInlineAIChat) {
closeAIInlineWidget()
}
}
})
editor?.addCommand(KeyMod.CtrlCmd | KeyCode.KeyS, function () {
updateCode()
shouldBindKey && format && format()
// Monaco swallows the keydown (addCommand prevents default and
// stops propagation), so page-level Ctrl/Cmd+S handlers never
// see it. Re-broadcast as a window event so editors that flush
// a draft on the shortcut can react regardless of which Monaco
// has focus. Only after `tick()`: the autosave payload is parked
// by a `$effect`, so dispatching synchronously would make every
// listener flush the state from before `updateCode()`.
void tick().then(() => window.dispatchEvent(new CustomEvent('wm-monaco-save-shortcut')))
})
editor?.addCommand(KeyMod.CtrlCmd | KeyCode.Enter, function () {
updateCode()
if (!shouldBindKey || !cmdEnterAction) return
void runCmdEnterWithDdlGuard()
})
editor?.addCommand(KeyMod.CtrlCmd | KeyMod.Shift | KeyCode.Digit7, function () {
// CMD + slash (toggle comment) on some EU keyboards
editor?.trigger('keyboard', 'editor.action.commentLine', {})
})
editor?.addCommand(KeyMod.CtrlCmd | KeyCode.KeyL, function () {
const selectedLines = getSelectedLines()
const selection = editor?.getSelection()
const hasSelection =
selection &&
(selection.startLineNumber !== selection.endLineNumber ||
selection.startColumn !== selection.endColumn)
// Target whichever chat is actually on screen: the session's own in a
// session pane, else the docked one. With sessions on outside a pane
// there is neither, and both branches below would be silent no-ops.
const chat = sessionScopedChatManager ?? aiChatManager
if (!sessionScopedChatManager && !chatState.dockedChatAvailable) return
if (hasSelection && selectedLines) {
chat.addSelectedLinesToContext(
selectedLines,
selection.startLineNumber,
selection.endLineNumber,
moduleId
)
} else {
// A session chat is always visible — only the docked pane toggles.
if (!sessionScopedChatManager) chat.toggleOpen()
chat.focusInput()
}
})
editor?.addCommand(KeyMod.CtrlCmd | KeyCode.KeyK, function () {
if ($copilotInfo.enabled) {
aiChatEditorHandler?.rejectAll()
if (showInlineAIChat) {
closeAIInlineWidget()
} else {
showAIInlineWidget()
}
}
})
editor?.addCommand(KeyMod.CtrlCmd | KeyCode.KeyU, function () {
dispatch('toggleTestPanel')
})
if (
!websocketAlive.deno &&
!websocketAlive.pyright &&
!websocketAlive.ruff &&
!websocketAlive.shellcheck &&
!websocketAlive.go &&
!websocketInterval
) {
reloadWebsocket()
}
})
reloadWebsocket()
setTypescriptExtraLibs()
setTypescriptRTNamespace()
return () => {
console.log('disposing editor')
ata = undefined
try {
closeWebsockets()
vimDisposable?.dispose()
closeAIInlineWidget()
markerChangeDisposable?.dispose()
// Note: We don't clear lint diagnostics on dispose - they persist across runnable switches
// Diagnostics are only updated when Monaco reports new markers for this runnable
console.log('disposing editor')
model?.dispose()
editor && editor.dispose()
console.log('disposed editor')
} catch (err) {
console.log('error disposing editor', err)
}
}
}
export async function fetchPackageDeps(deps: DepsToGet) {
ata?.(deps)
}
let customTsTypesData = resource([() => lang], async () => {
if (lang !== 'typescript') return undefined
let datatables = (
await WorkspaceService.listDataTables({ workspace: $operatingWorkspace ?? '' })
).map((d) => d.name)
let ducklakes = await WorkspaceService.listDucklakes({ workspace: $operatingWorkspace ?? '' })
return { datatables, ducklakes }
})
function setTypescriptCustomTypes() {
if (!customTsTypesData.current) return
if (lang !== 'typescript') return
const ducklakeNames = customTsTypesData.current.ducklakes
const datatableNames = customTsTypesData.current.datatables
const ducklakeNameType = ducklakeNames.length
? ducklakeNames.map((name) => JSON.stringify(name)).join(' | ')
: 'string'
const datatableNameType = datatableNames.length
? datatableNames.map((name) => JSON.stringify(name)).join(' | ')
: 'string'
const isDucklakeOptional = ducklakeNames.includes('main')
const isDataTableOptional = datatableNames.includes('main')
let disposeTs = typescriptDefaults.addExtraLib(
`export {};
declare module 'windmill-client' {
import { type DatatableSqlTemplateFunction, type SqlTemplateFunction } from 'windmill-client';
export function ducklake(name${isDucklakeOptional ? '?' : ''}: ${ducklakeNameType}): SqlTemplateFunction;
export function datatable(name${isDataTableOptional ? '?' : ''}: ${datatableNameType}): DatatableSqlTemplateFunction;
}`,
'file:///custom_wmill_types.d.ts'
)
return () => {
disposeTs.dispose()
}
}
async function setTypescriptRTNamespace() {
if (
scriptLang &&
(scriptLang === 'bun' ||
scriptLang === 'deno' ||
scriptLang === 'bunnative' ||
scriptLang === 'nativets')
) {
// The generated client resolves `undefined` for any response whose body it
// cannot read — `getResponseBody` swallows a parse failure and returns
// nothing, which `catchErrorCodes` passes through on a 2xx. Typing says
// otherwise, so without this the editor dies on mount rather than losing
// its resource-type autocomplete.
const resourceTypes =
(await ResourceService.listResourceType({
workspace: $operatingWorkspace ?? ''
})) ?? []
const namespace = formatResourceTypes(
resourceTypes,
scriptLang === 'bunnative' ? 'bun' : scriptLang
)
typescriptDefaults.addExtraLib(namespace, 'rt.d.ts')
}
}
async function setTypescriptExtraLibs() {
if (extraLib) {
const uri = mUri.parse('file:///extraLib.d.ts')
typescriptDefaults.addExtraLib(extraLib, uri.toString())
}
if (
lang === 'typescript' &&
(scriptLang == 'bun' || scriptLang == 'tsx' || scriptLang == 'bunnative') &&
ata == undefined
) {
absolutePathExtraLibs.forEach((d) => d.dispose())
absolutePathExtraLibs.clear()
const hostname = getHostname()
const addLibraryToRuntime = async (code: string, _path: string) => {
const path = 'file://' + _path
let uri = mUri.parse(path)
console.log('adding library to runtime', path)
typescriptDefaults.addExtraLib(code, path)
try {
await vscode.workspace.fs.writeFile(uri, new TextEncoder().encode(code))
} catch (e) {
console.log('error writing file', e)
}
}
const addLocalFile = async (code: string, _path: string) => {
if (destroyed) return
let p = new URL(_path, uri).href
let nuri = mUri.parse(p)
console.log('adding local file', _path, nuri.toString())
// Monaco's TS service resolves relative imports against the importer's URI (finding the
// model), but absolute paths like "/u/admin/foo" are looked up as raw paths and miss the
// `file://` model. Register them as extra libs so TS can resolve them.
if (_path.startsWith('/')) {
absolutePathExtraLibs.get(_path)?.dispose()
absolutePathExtraLibs.set(_path, typescriptDefaults.addExtraLib(code, _path))
}
if (editor) {
let localModel = meditor.getModel(nuri)
if (localModel) {
localModel.setValue(code)
} else {
meditor.createModel(code, 'typescript', nuri)
}
try {
if (model) {
model?.setValue(model.getValue())
}
} catch (e) {
console.log('error resetting model', e)
}
}
}
await initWasmTs()
const root = await genRoot(hostname)
console.log('SETUP TYPE ACQUISITION', { root, path })
ata = setupTypeAcquisition({
projectName: 'Windmill',
depsParser: (c) => {
return parseTypescriptDeps(c)
},
root,
scriptPath: path,
logger: console,
delegate: {
receivedFile: addLibraryToRuntime,
localFile: addLocalFile,
progress: (downloaded: number, total: number) => {
// console.log({ dl, ttl })
},
started: () => {
console.log('ATA start')
},
finished: (f) => {
console.log('ATA done')
}
}
})
if (scriptLang == 'bun') {
ata?.('import "bun-types"')
}
if (scriptLang == 'bunnative' || scriptLang == 'bun') {
ata?.(code ?? '')
}
dispatch('ataReady')
}
}
export function addAction(
id: string,
label: string,
callback: (editor: meditor.IStandaloneCodeEditor) => void,
keybindings: number[] = []
) {
editor?.addAction({
id,
label,
keybindings,
contextMenuGroupId: 'navigation',
run: function (editor: meditor.IStandaloneCodeEditor) {
callback(editor)
}
})
}
function showAIInlineWidget() {
if (!editor) return
inlineAIChatSelection = editor.getSelection()
if (!inlineAIChatSelection) {
return
}
const model = editor.getModel()
selectedCode = ''
if (model) {
selectedCode = model.getValueInRange(inlineAIChatSelection)
}
showInlineAIChat = true
aiChatInlineWidget?.focusInput()
}
function closeAIInlineWidget() {
showInlineAIChat = false
inlineAIChatSelection = null
selectedCode = ''
}
let aiChatInlineWidget: AIChatInlineWidget | null = $state(null)
$effect(() => {
const fontSize = small ? editorFontSize.small : editorFontSize.regular
if (editor) {
editor.updateOptions({ fontSize })
}
})
let loadTimeout: number | undefined = undefined
onMount(async () => {
if (BROWSER) {
if (loadAsync) {
loadTimeout = setTimeout(() => loadMonaco().then((x) => (disposeMethod = x)), 0)
} else {
let m = await loadMonaco()
disposeMethod = m
}
}
})
onDestroy(() => {
// Flush a pending keystroke debounce, or unmounting discards the last
// `changeTimeout` ms of typing. Both guards are load-bearing: without a pending
// timer Monaco isn't the newer side (an external write may be), and before init
// `getCode()` is '' — flushing either case overwrites real content.
if (editor && timeoutModel !== undefined) {
updateCode()
}
valueAfterDispose = getCode()
pasteCleanup?.()
destroyed = true
disposeMethod && disposeMethod()
websocketInterval && clearInterval(websocketInterval)
sqlSchemaCompletor && sqlSchemaCompletor.dispose()
autocompletor && autocompletor.dispose()
sqlTypeCompletor && sqlTypeCompletor.dispose()
resultCollectionCompletor && resultCollectionCompletor.dispose()
workspaceMacroCompletor && workspaceMacroCompletor.dispose()
schemaContractCompletor && schemaContractCompletor.dispose()
preprocessorCompletor && preprocessorCompletor.dispose()
timeoutModel && clearTimeout(timeoutModel)
changeChainStart = undefined
loadTimeout && clearTimeout(loadTimeout)
aiChatEditorHandler?.clear()
absolutePathExtraLibs.forEach((d) => d.dispose())
absolutePathExtraLibs.clear()
})
async function genRoot(hostname: string) {
let token = $lspTokenStore
if (!token) {
let expiration = new Date()
expiration.setHours(expiration.getHours() + 72)
const newToken = await UserService.createToken({
requestBody: { label: 'Ephemeral lsp token', expiration: expiration.toISOString() }
})
$lspTokenStore = newToken
token = newToken
}
let root = hostname + '/api/scripts_u/tokened_raw/' + $operatingWorkspace + '/' + token
return root
}
function acceptCodeChanges() {
const mode = aiChatEditorHandler?.getReviewMode?.()
if (mode === 'revert') {
aiChatEditorHandler?.keepAll()
} else {
aiChatEditorHandler?.acceptAll()
}
}
function rejectCodeChanges() {
const mode = aiChatEditorHandler?.getReviewMode?.()
if (mode === 'revert') {
aiChatEditorHandler?.revertAll()
} else {
aiChatEditorHandler?.rejectAll()
}
}
function onKeyDown(e: KeyboardEvent) {
if (e.key === 'Escape') {
if (showInlineAIChat) {
closeAIInlineWidget()
}
rejectCodeChanges()
} else if ((e.ctrlKey || e.metaKey) && e.key === 'ArrowDown' && aiChatManager.pendingNewCode) {
acceptCodeChanges()
if (showInlineAIChat) {
closeAIInlineWidget()
}
}
}
$effect(() => {
lang = scriptLangToEditorLang(scriptLang)
})
$effect(() => {
filePath = computePath(path)
})
$effect(() => {
path != initialPath &&
(scriptLang == 'deno' || scriptLang == 'bun' || scriptLang == 'bunnative') &&
untrack(() => {
handlePathChange()
})
})
$effect(() => {
initialized && lang === 'sql' && scriptLang
? untrack(() => addSqlTypeCompletions())
: (sqlTypeCompletor?.dispose(), resultCollectionCompletor?.dispose())
})
$effect(() => {
initialized && lang === 'sql' && scriptLang === 'duckdb'
? untrack(() => {
addWorkspaceMacroCompletions()
})
: workspaceMacroCompletor?.dispose()
})
// Pipeline annotation grammar is language-agnostic (`//` / `--` / `#`
// comment headers), so contract-ref completions register for every script
// language that can be a pipeline member. The provider line-gates itself,
// so it is inert outside annotation lines.
$effect(() => {
initialized && ['duckdb', 'python3', 'bun', 'deno', 'nativets'].includes(scriptLang ?? '')
? untrack(() => addSchemaContractCompletions())
: schemaContractCompletor?.dispose()
})
// Schema-contract markers arrive as a prop because the mirror can finish
// computing before Monaco initializes on mount — reacting to `initialized`
// re-applies the pending set once the model exists. The ever-set latch
// keeps unrelated editors from calling setModelMarkers with [] forever.
let contractMarkersEverSet = false
$effect(() => {
const ms = schemaContractMarkers
if (!initialized || (ms.length === 0 && !contractMarkersEverSet)) return
contractMarkersEverSet = true
untrack(() => {
const model = editor?.getModel()
if (!model) return
meditor.setModelMarkers(
model,
'schema-contracts',
ms.map((m) => ({ ...m, severity: MarkerSeverity.Warning }))
)
})
})
$effect(() => {
initialized && canHavePreprocessor(lang) && enablePreprocessorSnippet
? untrack(() => addPreprocessorCompletions(lang))
: preprocessorCompletor?.dispose()
})
let lastArg = undefined
$effect(() => {
let newArg = lang === 'graphql' ? args?.api : args?.database
if (newArg !== lastArg) {
lastArg = newArg
$dbSchemas && untrack(() => updateSchema(newArg))
}
})
$effect(() => {
console.log('updating db schema completions', dbSchema, lang)
initialized &&
dbSchema &&
['sql', 'graphql'].includes(lang) &&
untrack(() => addDBSchemaCompletions())
})
$effect(() => {
;(!dbSchema || lang !== 'sql') && untrack(() => disposeSqlSchemaCompletor())
})
$effect(() => {
;(!dbSchema || lang !== 'graphql') && untrack(() => disposeGaphqlService())
})
$effect(() => {
const currentWorkflowAsCode = workflowAsCode
$copilotInfo.enabled &&
$codeCompletionSessionEnabled &&
Autocompletor.isProviderModelSupported($copilotInfo.codeCompletionModel) &&
initialized &&
editor &&
scriptLang &&
untrack(() => editor && addAutoCompletor(editor, scriptLang, currentWorkflowAsCode))
})
$effect(() => {
$copilotInfo.enabled && initialized && editor && untrack(() => editor && addChatHandler(editor))
})
$effect(() => {
;(!$codeCompletionSessionEnabled || !$copilotInfo.enabled) && autocompletor?.dispose()
})
$effect(() => {
if (yContent && awareness && model && editor) {
untrack(() => {
monacoBinding && monacoBinding.destroy()
monacoBinding = new MonacoBinding(
yContent,
model!,
new Set([editor as meditor.IStandaloneCodeEditor]),
awareness
)
})
}
})
$effect(() => {
editor && $vimMode && statusDiv && untrack(() => onVimMode())
})
$effect(() => {
!$vimMode && vimDisposable && untrack(() => onVimDisable())
})
$effect(() => {
files && model && untrack(() => onFileChanges())
})
$effect(() => {
editor?.updateOptions({
lineNumbers: $relativeLineNumbers ? 'relative' : 'on'
})
})
let applyExternalCode = useDebounce(() => alignCodeWithEditor(true), 800)
// Last `code` value the editor itself produced or aligned to. Used to tell an
// echo (the bindable changed because the user typed — Monaco is already
// ahead) from a genuine external write. Without this, a typing burst longer
// than the debounce window would sync the lagging `code` back over newer
// keystrokes. Must be kept in step with every editor↔`code` sync point.
let lastEditorCode = code
function alignCodeWithEditor(history: boolean) {
const ed = editor
if (!ed) return
const next = code ?? ''
const value = ed.getValue()
const model = ed.getModel()
// Some keystrokes are still being debounced, don't overwrite them.
// When the debounce is done, updateCode will be called and the code will be aligned with the editor.
if (timeoutModel !== undefined) return
if (!model) return
lastEditorCode = next
if (value === next) return
if (history) {
ed.pushUndoStop()
ed.executeEdits('external', [{ range: model.getFullModelRange(), text: next }])
ed.pushUndoStop()
} else {
ed.setValue(next)
}
// The write above went through `onDidChangeModelContent`, which arms the
// keystroke debounce as if the user had typed. Monaco now holds exactly
// `code`, so there is nothing to flush — and leaving the timer armed makes
// every "is Monaco the newer side?" check (notably the unmount flush) answer
// yes on the strength of our own write.
cancelPendingChanges()
}
// External `code` prop changes should flow into the Monaco editor. Skip
// echoes: when `code` matches what the editor last produced (`updateCode`)
// or aligned to, the change came from the editor itself, so syncing back
// would clobber input typed since. Only genuine external writes — where
// `code` diverges from `lastEditorCode` — schedule a sync. The `untrack`
// block reads/writes Monaco without subscribing, so we don't loop.
$effect(() => {
;[code, editor]
if (!editor) return
untrack(() => {
if (code === lastEditorCode) return
applyExternalCode()
})
})
let isTsWorkerInitialized = resource([() => lang, () => initialized], async () => {
if (lang !== 'typescript' || !initialized) return false
// Use the stable model URI (computed once at mount), not filePath which changes on rename
await waitForWorkerInitialization(uri)
return true
})
// Update SQL query type information in the TypeScript worker
// This enables TypeScript to show proper types for SQL template literals
let handleSqlTypingInTs = useDebounce(function handleSqlTypingInTs() {
if (lang !== 'typescript' || !isTsWorkerInitialized.current) return
if (!preparedAssetsSqlQueries || preparedAssetsSqlQueries.length === 0) {
// Clear SQL queries if none exist
updateSqlQueriesInWorker(uri, [])
return
}
// Send SQL query information to the custom TypeScript worker
// The worker will inject type parameters into the code that TypeScript analyzes
// Worker async function call freezes if we pass a Proxy, $state.snapshot() is very important here
// Filter out queries with raw interpolations — they can't be type-checked
let queriesToSend = $state
.snapshot(preparedAssetsSqlQueries)
.filter((q) => !q.has_raw_interpolation)
updateSqlQueriesInWorker(uri, queriesToSend)
}, 250)
watch([() => preparedAssetsSqlQueries, () => lang, () => isTsWorkerInitialized.current], () => {
handleSqlTypingInTs()
})
watch([() => customTsTypesData.current], setTypescriptCustomTypes)
</script>
<svelte:window onkeydown={onKeyDown} />
<EditorTheme />
{#if datatableForMigrations && $operatingWorkspace}
<DdlMigrationGuard
bind:this={ddlGuard}
workspace={$operatingWorkspace}
datatable={datatableForMigrations}
/>
{/if}
{#if !editor}
<div class="inset-0 absolute overflow-clip">
<FakeMonacoPlaceHolder {code} lineNumbersWidth={51} />
</div>
{/if}
<div bind:this={divEl} class="{clazz} editor {disabled ? 'disabled' : ''}"></div>
{#if $vimMode}
<div class="fixed bottom-0 z-30" bind:this={statusDiv}></div>
{/if}
{#if $reviewingChanges}
<GlobalReviewButtons onAcceptAll={acceptCodeChanges} onRejectAll={rejectCodeChanges} />
{/if}
{#if editor && $copilotInfo.enabled && aiChatEditorHandler}
<AIChatInlineWidget
bind:this={aiChatInlineWidget}
bind:show={showInlineAIChat}
{editor}
editorHandler={aiChatEditorHandler}
selection={inlineAIChatSelection}
{selectedCode}
/>
{/if}
<style global lang="postcss">
.editor {
@apply p-0;
}
.yRemoteSelection {
background-color: rgb(250, 129, 0, 0.5);
}
.yRemoteSelectionHead {
position: absolute;
border-left: orange solid 2px;
border-top: orange solid 2px;
border-bottom: orange solid 2px;
height: 100%;
box-sizing: border-box;
}
.yRemoteSelectionHead::after {
position: absolute;
content: ' ';
border: 3px solid orange;
border-radius: 4px;
left: -4px;
top: -5px;
}
</style>