Ruler guides were a list kept beside the layers. Each guide is now an
object of its layer or sublayer, always directly in it and above the
layer's art. These behave as on main:
- Guides draw on screen only. Exports (SVG, PDF, PNG, PSD), Fit All,
Select All, the art bounds, clicks and marquees, and a layer's
Pathfinder and Transform effects leave them out. The Layers panel
lists them as guide rows of their layer, the highest guide.list
index first.
- Deleting a layer deletes its guides, and merging layers keeps their
guides in the merged layer.
- The guide commands number guides as guide.list does, in the order
they were made.
New: each guide takes an id from the document's object ids. An object
command given a guide's id fails with an error, and node.move cannot
put a guide into a group or out of its layers. A guide made while an
opacity mask, a symbol or a pattern is edited moves to the top
visible, unlocked layer when editing ends.
File format: native files still save the guides in the document's
guides list, in every file version, so older builds read them. Each
entry may now also carry the guide's id, name, locked and hidden
flags; older builds ignore those fields. Opening a file puts each
saved guide on its layer (on the top visible, unlocked layer when the
file names none or a layer it does not have), and each saved guide
counts against the object limit for opening a file. A ruler guide node
that a file holds below a layer's art moves up to the layer's guides,
and one anywhere else (in a group or another object, beside the
layers, or in a symbol) is removed. Ruler guides read from an .ai file
become guide objects on the layer they were on, numbered in the order
the file lists them.
A centred or right-aligned line of tracked type was measured with the
tracking after its last character, so it sat half that tracking
(centred) or the whole tracking (right-aligned) left of where the page
of users' .ai files draws it. That page aligns a line as wide as its
characters without that tracking.
Shaping now records the tracking each glyph's advance holds
(SGlyph::tracking), and a line that isn't justified leaves its last
character's out of the width it is aligned by. The last character is
the last one the line draws before a paragraph's end, a forced line
break or the spaces a wrap leaves; spaces or a tab typed before a
paragraph's end or a forced line break keep the letter's tracking, as
the page does. Forced line breaks and wraps aren't in the files: they
follow the rule so that turning a wrap into a paragraph or a forced line
break keeps the line in place. Hyphenated lines, lines whose last
character hangs, vertical and right-to-left lines and type on a path
keep it. The tracking stays in the glyph's advance, so the caret,
selections, line breaking and Fit Headline are unchanged.
LineInfo::trailing gives the amount. Right-aligned area type sets the
text after its last tab the same way.
What aligns by line widths follows. Convert To Area Type makes its
frame end where the lines are aligned, with room for the tracking, so
tracked type converted either way stays in place. Affinity import lines
up differently aligned paragraphs of artistic text by the aligned
widths. SVG import moves a centred or right-aligned letter-spaced line
back by its share of the last letter spacing, so it stays where a reader
draws it and SVG export, which anchors lines at the end of their
advances, reads back unchanged.
Centred and right-aligned tracked type lays out differently than before
in every document.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A <text> in a <clipPath> was dropped with a warning, so a clip path made
only of text had no content and the art it clipped was hidden. The
clipPath content model in CSS Masking Module Level 1 lists text and use
(https://www.w3.org/TR/css-masking-1/#ClipPathElement).
The importer now clips by the text's glyph outlines, together with any
shapes in the same clip path, and keeps clipPathUnits, the transforms,
clip-rule, visibility, the conditional processing attributes and a clip
path on the text. A <use> in a clip path that shows a text clips the
same way, placed by the <use>; a clip path on the <use> applies in the
<use>'s user space, and a reference goes to the first element with the
id. When the clip path holds only one visible text, with no clip path
of its own and no even-odd clip rule, the clipping path of the clip
group is the text object with no fill or stroke, the same structure
Make Clipping Mask produces for type. A text or a <use> with an id in a
clip path can still be shown by a <use> outside it. Each text's
outlines are laid out once, however many elements use the clip path.
A transform set by a style rule or a style attribute on a text now
places it, as it does a shape. Before, only the transform attribute was
read, inside and outside clip paths.
In horizontal type, tracking and manual kerning are now thousandths of
the em as the Horizontal Scale draws it: at 50%, a tracking of 100 adds
0.05 em after each character. They were added at the full em while the
glyph advances took the scale, so narrow tracked type drifted from the
page of users' .ai files letter by letter. That page writes such type's
character spacing as the tracking itself in a text space as wide as the
scale, which PDF scales by it (ISO 32000-1 9.4.4). Vertical type keeps
the full em: its advances down the column don't all take the scale.
vectorcraft_text::spacing_scale gives the factor. Shaping, inline
graphics and the Optical pairs use it, and so do the conversions from
points back to thousandths of an em: the automatic kerning the Character
panel reads and Touch Type's limits (glyph_em), Combine Point Text, and
PDF import, whose tracking is now the file's Tc. Scale Text Attributes
now keeps tracked type where a horizontal stretch drew it.
SVG export writes letter-spacing in thousandths of the font size it
writes (the vertical scale), so uniformly scaled type keeps its spacing
in a viewer.
Type with both a Horizontal Scale other than 100% and tracking or manual
kerning lays out differently than before, in every document.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The existing entity tests are unit tests that place the references in
attributes. This adds an import-level test with the references inside
<desc>, which the import refuses, and checks that the namespace entities
drawing apps write (&ns_svg;) still import.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A mask, inverted mask or knockout group over type now keeps the region
that covers any artwork. Type is measured by its layout or, on a path,
by the path, and a reader that sets live type in its own font can draw
it wider, so a region sized from that box could cut glyphs off.
A live object (blend, envelope, repeat, mesh, placed document,
intertwine) under a mask is measured as the art the writer writes for
it, expanded on its own with the same limits.
The writer keeps its symbol boxes and symbols' art reach in plain
fields instead of cells.
Opacity masks, inverted and unclipped masks, and knockout groups were
written with mask and filter regions of 200,000 x 200,000 units around
the origin. CairoSVG draws each mask into an image of its region's size
and stops with "out of memory", so a file with any of them did not
open; Firefox drew inverted masks wrong and WebKit drew knockout groups
and inverted masks wrong.
Each region now covers what the masked or knocked-out art paints: its
visual box grown by its raster effects, brush art and members' reach,
the mask art's box, and in a scaled symbol instance its strokes and
effects at their own size. A mask whose art reaches past its object
still comes back whole on import.
A group with a visible member that carries a geometry, Pathfinder or 3D effect, also in an opacity mask, keeps its previous copies. Copies stop at the budget with what each copy's own effects make, and a group's copies count the art of opacity masks.
The PDF and SVG tests for a group's image copies now expect the original on top.
The canvas, the gradient chips and the SVG, PDF, EPS and EMF writers mixed
freeform points in display RGB, while Object > Expand and Create Gradient
Mesh mixed them in the model the points share. Points that are all CMYK,
Gray, Lab or RGB now mix in that model in those places too. Points in
different models mix as before: in display RGB, and on ink planes as the
inks of each point.
This changes how existing art looks:
- CMYK and Lab points. With the default Generic CMYK working space and no
ink planes, halfway between C100 and K100 the canvas drew (17, 106, 139)
and now draws (54, 118, 143).
- RGB points in a working RGB space other than sRGB mix in that space.
- On ink planes (a CMYK document that shows transparency), RGB and Lab
points mix first and then separate.
Gray mixes keep their values, and CMYK and Gray points on ink planes draw
as before. RGB points in an sRGB working space draw as before except on
ink planes.
- PDF and EPS write a gradient whose points are all CMYK, or all Gray in a
CMYK file, as a CMYK image. PDF's Output settings convert it as they
convert any CMYK image.
- Darkening applies to the points before they mix.
- With Appearance of Black set to Rich, the canvas mixes a gradient with a
pure CMYK black point in display RGB, and the point fades from RGB black.
SVG export wrote an SVG filter as a live <filter> when its SVG 1.1 form
was at most 64 KB, without counting the region and id it writes into
it. A filter close to the limit came back over it, SVG import dropped
it, and the reopened object had no filter.
Export now leaves room for the id and the measured region, checks the
SVG 1.1 form the way import does, and counts each character past U+00FF
as the character reference Latin-1 output writes. A filter that does not
fit is exported as the other pixel effects are. The SVG 1.1 form and its
check are kept for the filters seen last, as checked filters already
are, so export does not check one filter again for each object.
Type that SVG text can't draw is written as outlines: vertical type, ruby,
right-to-left text, type on a path with an effect other than Rainbow, an end
bracket, Align to Path, Spacing, hanging punctuation, bullets or inline
graphics. Only glyphs bent along a path gave the outlined group a
`role="img"` and `aria-label`; every other case left screen readers and
search with no text. Each of them now names the group with its text.
Convert to Outlines still writes the paths alone, as before.
Inline graphics' placeholder characters stay out of the name, and empty
type gets none. An object whose id is its name also gets `data-name`, so
opening the file again keeps the name instead of taking the label.
SVG viewers center each <textPath> glyph on its midpoint on the path and
turn it to follow the path, which is how the Rainbow effect draws. Type on
a path with any other effect and no spacing was written as a plain
<textPath>, so a viewer drew it as Rainbow. It is now written as outlines
with a warning, as type with an end bracket, Align to Path or Spacing
already is.
Docs: docs/mcp.md, docs/target-app-parity.md.
Images were keyed by their bytes alone, so two <image> elements that
link different files with the same bytes shared one image key. Every
missing file shows the same placeholder, and files can also be copies
of each other. Export reads a linked file's pixels into the image key
when the document holds only a preview, so once one of the files was
found or changed, the images of the other file exported and printed
that file's pixels.
A linked image's key now also holds a hash of its link's path, as the
.ai reader's linked images do. Images of one file still share a key.
The missing-file placeholder comes from
vectorcraft_doc::links::missing_placeholder instead of a copy in the
SVG importer.
SVG Tiny 1.2 lists class among its core attributes, so the Tiny export no longer drops an object's own classes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SVG text with a textLength came in at its natural width. Following
SVG 1.1 §10.9.1, the characters of each text or tspan with a textLength
are now fitted to it: by spacing them out (lengthAdjust="spacing", the
default), which becomes kerning, or by stretching them
(spacingAndGlyphs), which becomes horizontal scale (kept within
1–10000%). Inner elements are fitted first; vertical text is spaced out
only, with a warning for spacingAndGlyphs. Negative lengths are ignored.
A clipPath with several children was imported as one compound path under
the first child's fill rule, so children wound in opposite directions
cancelled where they overlapped, and an even-odd child after a non-zero
one lost its hole. SVG 1.1 §14.3.5 defines the clip as the union of the
children's silhouettes, each computed on its own.
A clipPath with several children now comes in as a group of them as the
clip group's clipping path, each child keeping its own rule; a group
clips to the union of its members. One child imports as before.
Found while testing crates/svg/src/export/nodes.rs:492 (Writer::node, `xlink:href` built with xml_escape)
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
An id the importer writes into an element no longer matches a style sheet's
#id selector or a <use> link that no element matched before. The ids are
read from the decoded attribute values and text of the file, once.
An ellipse whose radius is smaller than the point tolerance at its coordinates, but at least two
f32 steps, now imports as the four-anchor ellipse instead of usvg's curves with rounded handles.
The import gives an element an id when it needs to find it again in
usvg's tree (links, symbol uses, events, data attributes, and now
circles and ellipses). A made-up id could repeat one the file already
had, such as vectorcraft-id-0, so a <use> or paint reference to that
element showed the other one. Made-up ids now skip every id in the file.
Circles and ellipses in a file whose style sheet has an [id selector get
no made-up id and keep usvg's curves, so the selector matches the same
elements as before.
The tolerance that compares the points usvg makes of a <circle> or
<ellipse> grew with the center's distance from the origin, so a circle
of radius 400 at x 40,000,000 fell back to eight anchors. The tolerance
is now a few f32 steps at the size of the coordinates, and the on-ellipse
check allows the same steps relative to the smaller radius.
usvg turns each <circle> and <ellipse> into arcs and then into a number
of cubics that depends on f32 rounding: an ellipse that export wrote as
<ellipse> could come back with eight anchors instead of the four it was
written from. The renderer flattens the two outlines differently, so a
round trip of a flat filled ellipse redrew 158 edge pixels by up to 43
levels, and prop_svg::plain_art_survives_visually failed on main.
Import now records the ids of <circle> and <ellipse> elements in the
prepass and rebuilds each as the ellipse VectorCraft draws: four smooth
anchors on the axes, starting at the right as SVG defines, with handles
KAPPA of the radii long, then transformed as before.
Tests: prop_svg::filled_ellipse_on_the_artboard_edge_survives_visually
(the failing case) and basic_shapes::circles_and_ellipses_import_as_the_ellipse_drawn
fail without the change.
Found while testing crates/svg/src/export/output.rs:360 (dublin_core, around line 81)
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
SVG export ignored an image's colorize paint and wrote the image's own
black, white and gray pixels. EPS export flattened every colorized image
into a process-color raster, so a spot color lost its plate. Both now
write the image in the fill color, with the coverage the canvas and PDF
export use (darkness times alpha), so white and transparent pixels take
no color.
- SVG writes a coverage PNG as a mask over a rectangle in the color,
shared between repeated images while each keeps its own color. Linked
mode writes the coverage PNG as a file, and SVG Tiny 1.2 writes a PNG
in the color. A colorized image whose pixels are missing, or whose
grayscale pixels can't be decoded, is left out with a warning.
- EPS writes fully covered or uncovered pixels as an imagemask in the
color's process or Separation ink (Registration on every plate), and a
grayscale image as ink samples scaled by coverage. Level 3 masks its
uncovered pixels out and Level 2 writes them white; gray midtones are
written over white, with the existing warnings.
- EPS flattening no longer counts a solid colorize paint on a grayscale
or 1-bit image as transparency by itself. Opacity, blending modes,
opacity masks, effects and overlapping transparent art still flatten
it, and so does a colorize paint the writer can't draw.
The svg crate's tests turn on resvg's raster-images feature to render
the embedded PNGs.
Docs: docs/mcp.md, docs/target-app-parity.md.
usvg 0.48 casts the filter's order attribute through i32, so a value like
1066570220609 saturates and the u32 order_x * order_y multiply overflows.
Only debug and overflow-checks builds panic on that; release builds already
import the file — the multiply wraps, and VectorCraft drops the filter as
an unsupported primitive with a warning, so nothing reaches the document,
render or export (observed on main in a --release build: both repro files
import without error; with a one-entry kernelMatrix the wrapped product
2147483647² mod 2³² = 1 even passes usvg's length check and leaves a filter
claiming a 2147483647×2147483647 kernel, still dropped downstream).
The import now parses order the way usvg does (svgtypes numbers, the same
as-i32 cast with its saturation, the first side defaulting the second, both
positive or both back to the 3×3 default) and refuses exactly the one case:
when the product of the two sides overflows u32 (checked_mul). Any order
usvg can multiply keeps importing as before, however large on its own, and
a missing, unparseable or non-positive order still falls back to 3.
Regression tests cover the overflow repros, the release wrap case, large
but fitting orders (100, 80 80, 65535 65535, 2147483647 1), the
pass-through shapes, and sane kernels, in debug and in release.
Co-Authored-By: GLM-5.3-Flash <noreply@z.ai>
Each patch of a gradient mesh was cut into 8 x 8 flat quads for SVG export,
Object > Expand and the other outputs that cannot shade a mesh, whatever its
colours, so a patch of one colour became 64 identical pieces. Neighbouring
quads are now joined into square blocks wherever the colour changes by no
more than about one 8-bit step across the block; the block's outline runs
through its quads' corners, and a patch that folds over itself is not joined.
Preserve SVG class metadata together with upstream basic-shape export helpers and current file-format evidence.
Co-Authored-By: Codex <noreply@openai.com>
BFlat