Schema validation reported a generic "Invalid data" message. Report the expected schema and the received value, with a bounded cycle-safe renderer so the error path cannot itself crash.
Fixes#10072
Signed-off-by: Akshit Nassa <akshitnassa412@gmail.com>
Co-authored-by: Akshit Nassa <akshitnassa412@gmail.com>
* 🐛 Fix geometry sync between mains and rotated component copies
Rotating a copy instance as a whole marked every shape inside it as
touched for geometry, so later geometric changes in the main (e.g. a
resize) were no longer propagated to that copy, while non-geometric
ones (e.g. fills) still were. And on paths where geometry did get
written to a rotated copy (e.g. resetting overrides), the sync engine
compensated only the roots' position delta, so the written values wiped
the copy's rotation back to 0.
Model the instance root's transformation as inherited, overridable
content, asymmetric to position (which remains free per-instance
placement):
- An untouched copy follows the main's transformation verbatim,
including rotation and flips (preserving the BUG #13267 semantics
that rotating a main propagates to its copies).
- Transforming a copy as a whole overrides only its ROOT: check-delta
compares the root's rotation/flips absolutely, but the descendants
relative to their root, so they merely follow and stay untouched.
- When a copy root's geometry is overridden, update-attrs expresses the
main's geometry in the copy's own frame: reposition-shape applies the
roots' relative transformation (rotation/flips) around the dest root
center in addition to the position delta. Geometric changes from the
main then keep propagating to the rotated copy, landing correctly in
its rotated frame instead of destroying its placement.
Covered by the new composable test case
case-n-geometry-sync-with-rotated-instances: an 8-variant sweep over
optional copy rotation, optional main rotation, and one of a fills or
height edit on the main child, asserting the whole model through the
real workspace events (the new rotate operation dispatches
dwt/increase-rotation, whose apply-modifiers step runs the check-delta
classification under test; change-height dispatches
dwt/update-dimensions and implements IPropertyCheck so one-of sweeps
can mix property and geometry edits). Verified by temporarily reverting
the fix: the case then fails with 6 assertion failures and passes again
with the fix restored.
Fixes#10109
AI-assisted-by: claude-fable-5
* 🐛 Fix synchronization problems
---------
Co-authored-by: alonso.torres <alonso.torres@kaleidos.net>
Relative operators were accepted in operand position, so "10+*3" silently evaluated to 310 instead of being rejected. Make negation a first-class operand so legitimate negative operands ("10 + -3") keep working.
Fixes#9581
Signed-off-by: Akshit Nassa <akshitnassa412@gmail.com>
Co-authored-by: Akshit Nassa <akshitnassa412@gmail.com>
Co-authored-by: Andrey Antukh <niwi@niwi.nz>
* ✨ Add systematic component tests via a composable test model
Introduce a framework for systematically testing Penpot component behaviour
(synchronisation/propagation, swaps, variant switches, nesting), plus a first
suite of cases built on it.
A test is expressed as a COMPOSITION OF OPERATIONS over a "situation" (an
in-memory file value plus named role bindings). Operations are reified as data
and composed by two combinators — `in-sequence` (threads the situation) and
`one-of`/`optional` (alternatives, enumerated into concrete variants). So one
written case stands for a whole matrix of variants, and coverage grows by
composition rather than by copying tests. Operations drive the REAL production
change pipeline, and event-operations dispatch the REAL workspace events and
await settlement, so the production watcher's automatic propagation is what is
exercised — the tests reflect genuine app behaviour, not a reimplementation.
Structure (frontend/test/frontend_tests/composable_tests/):
- core — the domain-agnostic engine: situation, the operation and
enumeration protocols, the combinators, and the runners.
- comp/nodes — the component operations (create/instantiate/reset, nesting,
swap, the variant ops, child add/remove/move, change, undo,
library sync).
- comp/setups — component-shaped starting configurations.
- interpreter — runs a case against the real frontend store: sync-ops apply
directly, event-ops dispatch real events and await
settlement (absorbing sync-file's delayed status RPC, which
would otherwise leak an error into subsequent tests).
- comp/sync-test — the cases (B-F, H, I, K, L, M).
This is test-only code with a single consumer — the frontend test suite (the
layer that runs the real app) — so it lives entirely under the frontend test
tree as .cljs, not under app/common.
The framework and its cases are documented in the project memory
frontend/composable-component-tests, added alongside.
Co-authored-by: Claude <noreply@anthropic.com>
* 🐛 Guard WASM mock teardown against an empty snapshot
`teardown-wasm-mocks!` unconditionally restored from the `originals` atom.
When run without a matching setup (double teardown, or `with-wasm-mocks*`
misused around an async test body), the snapshot is empty and every WASM API
function was `set!` to nil — permanently, for the remainder of the test run.
Any later code calling one of them (e.g. a leaked debounced resize-wasm-text
event firing during a subsequent test namespace) then crashed with
"initialized? is not a function".
Make the restore a no-op when there is nothing to restore.
Co-authored-by: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
* 🐛 Fix too much recursion when clicking shape in comments mode
Remove `deselect-all` from `handle-interrupt` in comments mode. The
`select-shape` event emits `:interrupt` which the comments stream
watcher routes to `handle-interrupt`. In comments mode, calling
`deselect-all` cleared the selection and emitted a competing
`rt/nav`, creating a synchronous recursion cycle in the potok store
that overflowed the JS call stack.
Fixes#10620
AI-assisted-by: opencode
* 🐛 Add unit tests for comments handle-interrupt
Make `handle-interrupt` public (defn- → defn) and add 4 tests covering
each branch: draft thread, open thread, comments mode, and noop.
AI-assisted-by: opencode
Guard remaining text pipeline locations that accessed :selrect directly
against nil/zero-dimension selrects by using safe-size-rect, which
provides a 4-level fallback chain (selrect -> points -> shape fields ->
empty 0.01x0.01 rect).
Fixes:
- fix-position in viewport_texts_html.cljs: replaced dm/get-prop
:selrect with ctm/safe-size-rect for both old and new shape
- assoc-position-data in modifiers.cljs: replaced (:selrect ...)
with ctm/safe-size-rect for delta computation
- change-orientation-modifiers in modifiers.cljc: replaced raw
:selrect access with safe-size-rect for scale and origin computation
Closes#10617
AI-assisted-by: mimo-v2.5-pro
The InteractionProxy `delay` setter rejected `0` via `(not (pos? value))`,
so an immediate after-delay interaction (`interaction.delay = 0`) was
refused. The model allows it: `set-delay`
(common/.../shape/interactions.cljc) only asserts `check-safe-int`, with
no positivity constraint, and `safe-int` permits 0.
Replace the guard with `(or (not (sm/valid-safe-int? value)) (neg? value))`:
it allows 0 and positive integers, rejects negatives, and rejects
non-integers cleanly (the previous `number?` check let fractional values
through, which then failed the model's `check-safe-int` assertion
downstream).
Adds a regression test (delay round-trips a positive value, then 0).
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Filip Sajdak <filip.sajdak@siili.com>
The ShapeProxy `borderRadius` setters (`borderRadius` and the four
per-corner variants) validated with `sm/valid-safe-int?`, so a fractional
radius (e.g. `shape.borderRadius = 7.5`) was rejected as invalid. The data
model types `:r1`-`:r4` as `::sm/safe-number` (shape.cljc), the radius
sidebar input only constrains `min 0` (not integer), and `set-radius-*`
store the value verbatim — so the plugin API was stricter than the model.
Same class of defect as the merged #9780.
Switch those 5 guards to `sm/valid-safe-number?`, keeping the existing
non-negative guard on the all-corners setter. Integer-only setters in the
file (`getRange` bounds, `setParentIndex`) keep `valid-safe-int?`.
Adds a regression test asserting fractional radius values are accepted
(with throwValidationErrors enabled).
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Filip Sajdak <filip.sajdak@siili.com>
The flex and grid layout proxies validated `rowGap`, `columnGap` and the
four padding setters with `sm/valid-safe-int?`, so a fractional value
(e.g. `flex.rowGap = 10.5`) was rejected as invalid. The data model types
`:row-gap`/`:column-gap` and `:p1`-`:p4` as `::sm/safe-number`
(layout.cljc), and the sidebar accepts decimals, so the plugin API was
stricter than the model — the same class of defect as the merged #9780.
Switch those 16 gap/padding guards (8 in flex.cljs, 8 in grid.cljs) to
`sm/valid-safe-number?`, matching the model and the predicate already used
by the flex-element setters in the same file. Integer-only setters
(`zIndex`, grid track indices/counts and cell positions/spans) keep
`valid-safe-int?`. Also fixes a `:righPadding` typo in two grid
rightPadding error branches.
Adds a regression test asserting fractional gap/padding values are
accepted (with throwValidationErrors enabled) for both flex and grid.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Filip Sajdak <filip.sajdak@siili.com>
Setting or clearing `board.guides` from a plugin threw a malli
`:malli.core/invalid-schema` for every value, including `[]`. Causes:
- `shape.cljs` validated against `[:vector ::ctg/grid]`, but the
`:app.common.types.grid/grid` reference is no longer registered
(direct schemas replaced the namespaced-keyword registrations). Use
the direct var `ctg/schema:grid`, matching every other setter.
- `parser.cljs` `parse-frame-guide` returned the `parse-frame-guide-column`
/ `parse-frame-guide-row` fns instead of calling them with the guide,
so column/row guides parsed to a vector containing a function.
- `parse-frame-guide-square` used `parse-frame-guide-column-params`
instead of the dedicated `parse-frame-guide-square-params`.
Adds a regression test parsing column/square guides and validating the
result (and an empty vector) against `ctg/schema:grid`.
Fixes#9773
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Filip Sajdak <filip.sajdak@siili.com>
The plugin proxy `:on-error` handler funneled every throwable through
`handle-error`, which only produced a message for malli ExceptionInfo
carrying `::sm/explain`. Two paths lost the real cause and rendered the
unhelpful "[PENPOT PLUGIN] Value not valid. Code: :error":
- A plain JS error (TypeError, etc.) is not an ExceptionInfo, so
`(ex-data cause)` is nil and the message bound to nil; the real
`.-message` only reached the host-page console, invisible to the
plugin sandbox.
- A malli explain whose errors flatten to nothing made `error-messages`
return "", which rendered as "Value not valid: . Code: :error".
`error-messages` now returns nil (not "") when there is nothing to
render, and `handle-error` falls back to the raw explain, then to the
throwable's `ex-message`/`str`, so a useful message always surfaces.
Working cases (well-formed explain) are unchanged.
Adds regression tests for the plain-JS-error path, the empty-explain
path, and the `error-messages` nil contract.
Fixes#9692
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Filip Sajdak <filip.sajdak@siili.com>
* ✨ Adds static dispatch safe stubs in tests
* 🐛 Fix shapesColors metadata key to match ColorShapeInfo
* 🐛 Fix CommentThread.remove rejecting the owner's own threads
* 🐛 Fix page.removeCommentThread throwing on a spurious Promise
* ✨ Implement ShapeBase.swapComponent in the plugin API
* ✨ Expose File.revn in the plugin API
* 🐛 Fix FileVersion.createdAt calling Luxon method on a js/Date
* 🐛 Fix plugin font/typography application to text and ranges
* 🐛 Default plugin overlay interaction position for non-manual types
* 🐛 Fix plugin interaction setters passing an id-only shape
* 🐛 Fix grid addColumnAtIndex rejecting valid track types
* 🐛 Expose libraryId on library color/typography/component proxies
* ✨ Implement LibraryTypography.setFont in the plugin API
* 🐛 Fix typography.applyToTextRange reading unexposed range bounds
* 🐛 Fix utils.geometry.center argument mismatch
* 🐛 Fix localStorage.removeItem calling getItem
* 🐛 Fix shape backgroundBlur proxy key casing
* 🐛 Report boolean shape type as 'boolean' in the plugin API
* 🐛 Return the resulting paths from plugin flatten
* 🐛 Make plugin z-order methods act on the target shape
* 🐛 Make is-variant-container? return a boolean
* ✨ Implement Group.isMask in the plugin API
* 🐛 Return a shape proxy from TextRange.shape
* 🐛 Return the duplicated set from TokenSet.duplicate
* 🐛 Fix theme addSet/removeSet reading set name with a keyword
* 🐛 Accept string fontFamilies token value in the plugin API
* 🐛 Fix combineAsVariants ignoring the passed component ids
* 🐛 Fix board removeRulerGuide ignoring its argument
* 🐛 Fix board guides setter schema and parser
* 🐛 Avoid 0-byte allocation when syncing empty grid tracks
* 🐛 Validate grid track indices in the plugin API
* 🐛 Return null for empty input in group() and centerShapes()
* 🐛 Return TokenTypographyValue[] from a typography token's resolvedValue
* 🐛 Return TokenShadowValue[] from a shadow token's resolvedValue
* 🐛 Return string[] from a fontFamilies token's resolvedValue
* 🐛 Clear mutually-exclusive reps when setting LibraryColor gradient/image
* 🐛 Add readonly tags to types, deprecate Image type
* 📚 Update plugins changelog
The plugin text API rejected negative letter-spacing even though the
product UI allows -200..200 (typography.cljs). Two defects in
frontend/src/app/plugins/text.cljs:
- `letter-spacing-re` (`#"^\d*\.?\d*$"`) had no provision for a leading
minus, so any negative value failed validation.
- The text-range `:letterSpacing` setter inverted its guard: it used
`(or (empty? value) (re-matches ...))` to mean "invalid", which
rejected matching values and let non-numeric input through. The
text-shape setter and the sibling `lineHeight` range setter both
correctly use `(not (re-matches ...))`.
Fix the regex to allow an optional leading minus and add the missing
`not` so the range setter matches the shape setter. Adds regression
coverage for the regex accept/reject contract.
Fixes#9780
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Filip Sajdak <filip.sajdak@siili.com>
* 🎉 Add background blur
* 🎉 Add test
* 🎉 Add background blur info to plugins API
* 🎉 Suport in wasm for both layer and background blur
* 🐛 Fix failing test
* ♻️ Fix comments
---------
Co-authored-by: alonso.torres <alonso.torres@kaleidos.net>
* ✨ Batch multiple thumbnail deletions into a single RPC call
Replace the old per-object immediate thumbnail deletion with a
debounced batched approach. The frontend queues object-ids in state
and waits 200ms before sending a single RPC request with up to 200
object-ids. The backend deletes all matching thumbnails in one SQL
statement with a single RETURNING clause, then touches the affected
media objects.
This reduces RPC overhead when rapidly clearing thumbnails (e.g.
navigating pages) and makes deletions more efficient.
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
* 📎 Fix missing issues
---------
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
* ✨ Add MCP connection badge to the workspace toolbar
* ✨ Add MCP status button with single-tab connection control
* ♻️ Extract component for MCP indicator in the toolbar
* ♻️ Some improvements
---------
Co-authored-by: Luis de Dios <luis.dedios@kaleidos.net>
