push_clip_glyph kept a hash + linear-scan cache of emitted
clip paths to dedup repeated clip-glyph uses within a single
COLRv1 paint tree. With paint contexts being one-glyph per
use, the cache mostly churned; it also made hb_memcmp trip
on sanitizer-null when an empty path matched an equally-empty
existing entry (fuzzer-reached null-deref).
Drop the cache: always emit a fresh <path> + <clipPath> pair
per push_clip_glyph. Removes defined_paths / content_entry_t
from the paint struct.
Regenerate the two handwriting COLR expected SVGs whose
clip-reuse now produces distinct path defs.
Assisted-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
include_point used to flip Y so the computed line-box extents
matched the old Y-down internal model. With the library now
tracking extents in Y-up (font) space and wrapping SVG body
in <g transform="scale(1,-1)">, that flip sent the auto-fit
viewBox above the glyph ink instead of around it.
Regenerate the vector expected SVGs that had captured the
shifted viewBox.
Assisted-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Accept caller-supplied extents in any orientation and
normalize to min-corner origin with positive width/height.
Lets callers use the glyph-extents-style {top, -height} form
matching hb_glyph_extents_t convention, and brings set_extents
in line with the Y-up internal model the paint/draw contexts
now use.
Assisted-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Exposes the path separation as a public function so callers
using the draw funcs directly can separate glyph outlines.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Remove draw svg_prefix from New API list and sections.txt
- Document pen_x/pen_y removal, dedup removal, _glyph
simplification in the Vector library section
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Remove the initial base_transform push from ensure_initialized.
Instead, _glyph_impl pushes/pops the positioning transform via
the paint funcs, matching the vector paint model.
current_transform() returns identity when the stack is empty.
pop_transform allows popping to empty (for the final glyph-level
pop).
This makes changing set_transform between glyph calls work
correctly — the transform is applied fresh for each glyph.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Each _glyph_or_fail function now documents what it is
equivalent to in terms of the underlying font draw/paint
API + the context's own funcs.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Remove hb_pdf_paint_ensure_initialized — it was identical to
paint->ensure_initialized(). All PDF callbacks now call the
common method. Remove redundant call from _glyph_impl since
paint callbacks handle it.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SVG render now wraps all body content in a single
<g transform="scale(1,-1)">. All coordinates inside are
in Y-up (font) space. The viewBox maps the flipped range.
This eliminates:
- Per-path scale(1,-1) wrappers in draw flush_path_svg
- Per-glyph Y-flip in paint _glyph_impl's push_transform
- The {-xy, -yy, -ty} negation in extents transforms
- Format switch in paint _glyph_impl
paint _glyph_impl now passes the raw transform to
push_transform — no format knowledge needed.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Remove content-based color glyph dedup from paint — no more
<defs>/<use> for color glyphs. Each glyph's paint output is
emitted inline, wrapped by push_transform/pop_transform from
the paint funcs (which emit <g transform="..."> for SVG and
q cm ... Q for PDF).
_glyph_impl is now format-agnostic: extend extents,
push_transform, paint, pop_transform.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
_glyph_or_fail is now: extend extents, new_path, draw.
No format switch, no flush — just accumulate.
new_path() flushes any pending path and starts a fresh one,
separating glyph outlines so they don't interact under fill
rules.
Flush also happens on foreground change and at render time,
so the last pending path is always emitted.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Remove format switch from _glyph_or_fail — it now just extends
extents, calls hb_font_draw_glyph with the context's own funcs,
and flushes. All format-specific work is in flush_path() and
the draw callbacks.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Draw SVG output is now flat inline <path> elements — no
<defs>, no <use>, no id references. Each glyph is an
independent <path d="..." transform="..." fill="..."/>.
Remove hb_vector_draw_set_svg_prefix / get_svg_prefix since
draw output no longer emits any ids. Paint keeps svg_prefix
for gradient and clip ids which SVG requires.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The transform is transient per _glyph call, not a global
state for the entire document. Each glyph's <use> already
carries its own transform from the caller's set_transform.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
These are convenience APIs, not typesetting APIs — positioning
is the caller's responsibility via set_transform(). This
matches the GPU API which never had pen parameters.
All callers updated to use set_transform with pen position
before each _glyph call.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace glyph-ID-based dedup with content-based dedup for SVG
path definitions and color glyph groups. Path/paint output is
hashed and compared byte-for-byte against existing defs;
identical content reuses the same def regardless of glyph ID.
This eliminates the font reference held by draw/paint contexts
(cached_font + hb_font_reference), which was keeping the font
and its face alive longer than the caller expected. No font
pointer is stored at all now.
Also removes changed() — with content-based dedup, foreground
and palette changes produce different content that naturally
won't match cached entries.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Nastaliq and other deep-descender scripts were clipped because
the fit used only logical extents (one em height). Now uses
the union of ink and logical extents so all ink is visible.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- demo_buffer_set_features() sets global features for hb_shape
- web_set_features() wasm export parses comma-separated features
- Shell message handler accepts { kind: "features", value }
- Features applied to all shape calls in demo_buffer_add_text
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Allows the host page to simulate keypresses (e.g. 'b' for
dark mode toggle) via postMessage { kind: 'key', value }.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Malicious fonts with extreme scale values cause huge pixel
buffer allocations across multiple push_group surfaces. The
library correctly bounds individual surfaces at 1 GB, but
stacked group surfaces can exceed the fuzzer RSS limit.
Cap the font scale to 1000 in the harness so the exercised
surface stays within fuzzer memory constraints.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When clip nesting exceeds HB_GPU_PAINT_MAX_CLIP_DEPTH (3),
push_clip_glyph sets unsupported=true but still increments
clip_depth. emit_all_clip_sub_blobs then iterated clip_depth
without capping, reading past the fixed-size clip_stack array.
Cap the iteration at HB_GPU_PAINT_MAX_CLIP_DEPTH.
Fixes: https://oss-fuzz.com/testcase-detail/6573156621156352
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Coverage out parameter, premultiplied gradient interpolation,
foreground alpha fix, and the shader signature change.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Update the signature, parameter documentation, and sample code
to reflect the new edge coverage out parameter. Stem darkening
example now shows the coverage-based correction pattern.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
hb_gpu_paint() now takes an out float coverage parameter that
receives the maximum slug coverage across all paint layers.
This lets the caller apply stem darkening and gamma correction
to edge antialiasing only, without distorting interior paint
colors (gradients, semi-transparent layers).
The demo shaders (GLSL, MSL, HLSL, WGSL) use the new parameter
to share a single coverage-correction block between draw and
paint paths. Also fixes the blend func to GL_ONE for correct
premultiplied compositing.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Reference the new font before destroying the old one, in case
both pointers are the same object. In practice the caller always
holds its own reference so this cannot trigger, but the pattern
is safer and silences Coverity CID 503195/503196.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add sample code for both draw and paint contexts, note that
render clears accumulated content (get extents first), and
document the semi-transparent overlap compositing behavior.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The OpenType COLR spec requires gradient interpolation in
premultiplied space. The shader was interpolating unpremultiplied
stop colors directly, which produced washed-out transitions when
a stop fades toward transparency (e.g. opaque red to transparent
gives desaturated pink instead of red at decreasing opacity).
Premultiply stop colors before mix(), unpremultiply the result.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
hb_gpu_paint() returns premultiplied RGBA, as documented in
hb-gpu.cc. The GL demo was using GL_SRC_ALPHA which
double-applied alpha, darkening semi-transparent colors.
Use GL_ONE as the API docs prescribe.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
These backends use immediate-mode foreground: they pass their
foreground color to hb_font_paint_glyph() and use the resolved
color from the callback directly. The is_foreground substitution
was redundant (a no-op when foreground alpha is 255) and prevented
semi-transparent foreground alpha from working correctly.
Now the foreground is passed as-is (including its alpha) to
hb_font_paint_glyph(), and the color callback ignores
is_foreground — the color already carries the correct value.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The previous commit scaled all four components of the foreground
uniform by the baked paint-tree alpha, which darkened the RGB
towards black. The foreground is unpremultiplied, so only the
alpha channel should be modulated; the shader's later premultiply
step handles the rest.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Explain the two valid usage modes: immediate (use @color directly)
and deferred (substitute own foreground RGB, combine alpha). Note
that deferred mode requires passing an opaque foreground to
hb_font_paint_glyph().
hb_color_stop_t defers to hb_paint_color_func_t for the
is_foreground/color contract.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When is_foreground was set, the shader replaced the entire baked
color with the foreground uniform, discarding the paint-tree alpha
encoded in the color's alpha channel. This meant COLRv1 glyphs
with a separate alpha on their foreground color sentinel would
lose that alpha at render time.
Now the shader multiplies the foreground uniform by the baked alpha
(ct.a / 32767 for LAYER_SOLID, b.a / 32767 for gradient stops),
preserving the paint-tree modulation.
Updated all four shader languages (GLSL, HLSL, MSL, WGSL) and
regenerated the .hh artifacts.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The synthesized fallback should signal is_foreground=true so
backends like cairo that handle foreground specially can detect
it. Pass alpha=255 since the caller passes opaque foreground
to paint_glyph; backends that care about foreground will apply
their own alpha.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The COLR walker's get_color bakes foreground_alpha into
the returned color. Since our backends don't substitute
paint->foreground in their callbacks (they trust the walker's
resolved color), passing foreground with its own alpha would
double-count it.
Force alpha=255 when passing foreground to hb_font_paint_glyph
and _or_fail, matching what cairo does. The backend applies
foreground alpha through its own mechanisms (draw fill-opacity,
raster compositing).
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The PDF draw context was dropping foreground alpha because
it had no resource dict for ExtGState. Semi-transparent
fills (e.g. --rainbow) rendered fully opaque.
Add a per-flush ExtGState with /ca (fill opacity) when
alpha < 255. The ExtGState entries are inlined in the
page's resource dict — no separate PDF objects needed.
Assisted-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>