Previously, any builtin functions would be optimized as a side-effect of
optimizing programs that used them. Now that shared elements aren't
being optimized in that way, we explicitly optimize any shared modules
when they are first created. We don't remove dead elements, but we
we do substitute settings, simplify, and inline.
Bug: skia:10905
Change-Id: I701b5e9f52fb880ef3e6f4c67694d08602f47e95
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/336440
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This generic parsing will allow us to later remove kInherit and
"inherit" parsing from all other specific SVG types.
Also convert "filter" presentation attribute to new-style parsing
for compatibility with the setter.
Change-Id: Id917e12d77cdefb1a8f1404cac1e0c4e486d8b53
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/335660
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Tyler Denniston <tdenniston@google.com>
This CL adds a new SkSVGProperty<T,B> class and uses it instead of
SkTLazy for the presentation attributes. Ideally this will form the
foundation for improvements to our presentation attribute parsing
as well as correctness for inherited/non-inherited properties.
Change-Id: Ie1cdb3db9674c55376e127cc1a8b8cb303a1bd13
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334837
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
This reverts commit b776a105e5.
Reason for revert: This will be the first step of a staged change.
Original change's description:
> Some testing images in Google3 look slightly different with this change. Need to change them, too.
>
> Revert "Roll back #10487 fix (so now spaces always resolved in the main font)"
>
> This reverts commit ff5d8700b3.
>
> Reason for revert: <INSERT REASONING HERE>
>
> Original change's description:
> > Roll back #10487 fix (so now spaces always resolved in the main font)
> >
> > Bug: skia:10715
> > Change-Id: I06bb881998d00228c79a218967eda6468cbc5bed
> > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334053
> > Reviewed-by: Ben Wagner <bungeman@google.com>
> > Commit-Queue: Julia Lavrova <jlavrova@google.com>
>
> TBR=bungeman@google.com,jlavrova@google.com
>
> Change-Id: Icc1f262f60fa2ec9b4425ce3088b5eaac9313374
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: skia:10715
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334158
> Reviewed-by: Julia Lavrova <jlavrova@google.com>
> Commit-Queue: Julia Lavrova <jlavrova@google.com>
TBR=bungeman@google.com,jlavrova@google.com
Change-Id: I0f2f46fc21eec795026ea98696f68112915f7979
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:10715
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334159
Commit-Queue: Julia Lavrova <jlavrova@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
Drawing to a Premul Raster Surface and then copying out Unpremul pixels
was 15x slower. Drawing to an Unpremul Raster surface was a little
slower than the previous RasterDirect surface.
Out of an abundance of caution, I'm going back to the way things were.
We can revisit drawing to a Raster surface in the future.
This also adds better error handling to the SkottieFrames test which
thankfully caught this regression.
Bug: skia:10717
Change-Id: Ib8b9d490c92e013c3311adb1d08de5fa0fc0351c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334356
Reviewed-by: Kevin Lubick <kjlubick@google.com>
This also refactors some internal names to be more consistent
with the normal naming convention.
Bug: skia:10717
Change-Id: I69a0d95aaca359e121b3ff84002d0b5f7cadb260
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334041
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Nathaniel Nifong <nifong@google.com>
The only caller is CanvasKit, and SkAnimatedImage already has getBounds
(from SkDrawable) which does the same thing, except it returns
SkScalars. It also takes scaling and cropping into account, which
CanvasKit does not use, and in a future commit, it will consider
orientation. Switch CanvasKit to use getBounds.
Change-Id: Ia956f91d241641aec450f3aba99583e95a3ff386
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333222
Commit-Queue: Leon Scroggins <scroggo@google.com>
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Revert "Roll back #10487 fix (so now spaces always resolved in the main font)"
This reverts commit ff5d8700b3.
Reason for revert: <INSERT REASONING HERE>
Original change's description:
> Roll back #10487 fix (so now spaces always resolved in the main font)
>
> Bug: skia:10715
> Change-Id: I06bb881998d00228c79a218967eda6468cbc5bed
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334053
> Reviewed-by: Ben Wagner <bungeman@google.com>
> Commit-Queue: Julia Lavrova <jlavrova@google.com>
TBR=bungeman@google.com,jlavrova@google.com
Change-Id: Icc1f262f60fa2ec9b4425ce3088b5eaac9313374
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:10715
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/334158
Reviewed-by: Julia Lavrova <jlavrova@google.com>
Commit-Queue: Julia Lavrova <jlavrova@google.com>
Introduce a Builder helper and plumb the client-provided SkFontMgr for
font resolution.
Also clean up some of the legacy SkSVGDom factories.
Bug: skia:10840
Change-Id: I6e1eabe7c257cb75dfdb5bf67054f93f25769027
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333577
Commit-Queue: Florin Malita <fmalita@google.com>
Reviewed-by: Tyler Denniston <tdenniston@google.com>
- Added new SVG_OPTIONAL_ATTR macro for lazy properties
- Added SkSVGIRI type -- we need explicit types now to be able to
dispatch via the templated parse function
- Converted several attribute parse functions to templated version
Change-Id: I270d35983083f368e36afd96a62c768161d49942
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333518
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Requested by Flutter. This adds about 80k of code size
due in large part to the brotli decoding logic.
This also updates the check_deps error to point at
a doc on how to create a new GOB mirror.
Brotli version is 1.0.9; the latest as of today.
Change-Id: I1580cb2189ff1205a9bffca3d887ff2b98a4042e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/333218
Reviewed-by: Ben Wagner <bungeman@google.com>
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Kevin Lubick <kjlubick@google.com>
- Plumbing to store filter results by id and resolve them as inputs
when referenced
- Added implementation of feComposite filter
- Added call to resolve input in feColorMatrix
- Bugfix to SkSVGFilterType operator==
The tests filters-color-01-b and filters-composite-03-b should now be
passing.
Bug: skia:10841
Change-Id: I2cd099c60ac21710f25184806c5cc537656b42af
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332723
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Supporting this filter will also allow us to start implementing and
testing the filter result DAG via the filters-color-01-b test. That test
is one of the simplest that involves filter result storage, so it's a
good candidate to start with.
Bug: skia:10841
Change-Id: Id8317ccfb78031cc6af83c9f3e3f382dcb6dee98
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332599
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
This makes both APIs have the same arguments with the two
source coordinates first and all the destination params
(image info, optional buffer, optional rowBytes) after.
Bug: skia:10717
Change-Id: I483e4f33f24e226793db6113d5ba5b1955cd892e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332622
Reviewed-by: Mike Reed <reed@google.com>
- remove the custom-func SkSVGAttributeParser::parse version and always
dispatch via SkSVGAttributeParser::parse<T>(T*) - this should
avoid adding any other parse helper declarations in the future
- relocate the turbulence parse helpers to SkSVGAttributeParser (while
keeping the definition in SkSVGFETurbulence.cpp)
- update ParseResult initialization to use move semantics
Change-Id: I8ed9811671a6fbc5971f9d1f14e7b9c07da7dec0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332540
Commit-Queue: Florin Malita <fmalita@chromium.org>
Commit-Queue: Florin Malita <fmalita@google.com>
Reviewed-by: Tyler Denniston <tdenniston@google.com>
The previous behavior leaked Skia-internal concepts into public SkSL.
Users coming from GLSL will expect that bindable/sampleable objects are
uniform (just like texture2D). This keeps the old support around (and
tested), but updates all of our examples to use 'uniform'.
Bug: skia:10679
Change-Id: I0c98162f5e21dad7014d9778ceb26143d2f6030e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/332376
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
One significant source of boilerplate in the SVG frontend is the
plumbing to ensure type safety all the way down from the XML layer to
the SkSVGNode layer. This is mostly an artifact of the top-down parsing
approach currently used by the SkDOM -> SkSVGDom building process.
One way to help remove some boilerplate is to perform attribute parsing
bottom-up, where each SVG node knows how to parse and populate its own
attribute values from a string-valued KV pair.
Additionally, bottom-up parsing allows us to support the case of the
same SVG attribute name having different meanings on different nodes
(e.g. the "type" attribute has different meaning on <feTurbulence>
versus <feColorMatrix>).
This CL adds some initial work to start us down that road, and ports the
attributes previously added for <feTurbulence> to use the new code path.
Change-Id: I2973cfab96891475d05ebf1228117626ca48ef4d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331477
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
This ties the caps to the compiler instance, paving the way for
pre-optimizing the shared code. Most of the time, the compiler is
created and owned the GPU instance, so this is fine. For runtime
effects, we now use the shared (device-agnostic) compiler instance
for the first compile, even on GPU. It's configured with caps that
apply no workarounds. We pass the user's SkSL to the backend as
cleanly as possible, and then apply any workarounds once it's part
of the full program.
Bug: skia:10905
Bug: skia:10868
Change-Id: Ifcf8d7ebda5d43ad8e180f06700a261811da83de
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331493
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This had previously been incompletely exposed. This does so
properly, by returning a JSArray of JSObjects with the metrics.
I am purposely not exposing fLineMetrics until that is explicitly
requested, as it looks complicated.
Bug: skia:10614
Change-Id: If3a2154e8ee39d8adcf416b4513b33a955c2482f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331359
Reviewed-by: Nathaniel Nifong <nifong@google.com>
Reviewed-by: Julia Lavrova <jlavrova@google.com>
This will let clients have faster access to the drawn data
and give them control about calling Canvas2D.putImageData.
Bug: skia:10565
Change-Id: Ia3da7867be3f671ab20fdb182ec758b77c0f63b8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331316
Reviewed-by: Mike Reed <reed@google.com>
Reviewed-by: Nathaniel Nifong <nifong@google.com>
This was the simplest one to start with. I need to add support for the
"filter effect subregion" in order to handle tile stitching property,
so this isn't quite complete. But I believe this is enough for the
basic filters-turb-01-f test to pass, which gives us a baseline for
further filter work.
Summary of changes:
- Added attribute type and parsing for SVG integer datatype
- Added new node class for feTurbulence
- Added several new properties and parsing for feTurbulence
Bug: skia:10841
Change-Id: I8c877a5e1a837bfd527782253062eeb58febdde6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/330621
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Prior to rendering a node, check the filter presentation attribute and
create a new canvas layer with a corresponding SkImageFilter set.
Also added the computation of the filter effect region and added a naive
(and incomplete) construction of the image filter DAG for a
<filter> element.
Bug: skia:10841
Change-Id: Ie94299757e059c39540ad316cddf438df5726d97
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/330619
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
There are currently many tests skipped, but many more pass.
This changes the built binary to have a lot of debugging logic
in it so we should be able to get backtraces on those crashes
more easily when debugging.
gmtests.html was removed as it was superceded by run-wasm-gm-tests
and make run_local.
Bug: skia:10812, skia:10869
Change-Id: I72ab34d3db83a654dc8829831b3ecb795fe23d43
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/329170
Reviewed-by: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Nathaniel Nifong <nifong@google.com>
Commit-Queue: Kevin Lubick <kjlubick@google.com>
Added:
- SkSVGFilter: This corresponds to a <filter> element. Also added some
of the attributes of this element.
- SkSVGFe: This will be the base class of all <fe*> elements.
- SkSVGFilterContext: This will hold the contextual mapping of string id
-> image filter result, for constructing pipelines.
Bug: skia:10841
Change-Id: I15a29d39411a6255ab4e11f022baa10554b2bce6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/330618
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
It turns out that "userSpaceOnUse | objectBoundingBox" can be specified
as the unit type for multiple elements, such as clips, masks, filters
etc. and is not specific to gradients.
The full list:
https://www.w3.org/TR/SVG11/coords.html#ObjectBoundingBoxUnits
Bug: skia:10842
Change-Id: I995d588862fb43b9f130f0455d7af8de47046af4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/330616
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Currently only works properly when filling rects, since that is the only
node that implements bounds computation with this CL.
Summary of changes:
- Scale gradient coords by object bounds when units are
kObjectBoundingBox.
- Make fGradientUnits protected instead of private.
- Change default value of fGradientUnits to kObjectBoundingBox, which
is the default according to the spec.
- Change SkSVGNode::onObjectBoundingBox to take a full render context
instead of length context.
- Implement bounds computation for SkSVGRect, SkSVGContainer and
SkSVGUse.
Bug: skia:10842
Change-Id: I2e999985e67644e50da7f681fde190bcf4823eec
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/329223
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Tyler Denniston <tdenniston@google.com>
This adds an Init() which should be called before any resources
are loaded or GMs/Tests are called.
This also adds a little helper in compile.sh for building only
a single GM, which can make local development faster given the
fact that emsdk doesn't do its final compilation/linking in
parallel.
Bug: skia:10812
Change-Id: I1a8932549b9ae951c153c0d49927bdc933c29657
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/329358
Reviewed-by: Ben Wagner <bungeman@google.com>
