Motivation: as we implement more features in PDF, it would be nice to more easily see what is happening in the output. This change serializes page content as plain text rather than compressed text, but it has to be explicitly enabled with a GYP_DEFINE change:
export GYP_DEFINES='skia_pdf_less_compression=1'
bin/sync-and-gyp
ninja -C out/Debug dm
out/Debug/dm --config pdf --src gm -w /tmp
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1840103002
Review URL: https://codereview.chromium.org/1840103002
The following are currently unused in Android, Google3, Chromium, and Mozilla:
- SkEvent
- SkTime::GetMSecs
- SK_TIME_FACTOR (also unused in Skia)
- SkAutoTime
I left uses of SkMSec more-or-less intact for SkEvent, SkAnimator, and SkInterpolator. SkInterpolator is used in Chromium, so I did not want to change the API. The views/ and animator/ code is crufty, so it didn't seem worthwhile to refactor it. Instead, I added SkEvent::GetMSecsSinceStartup, which is likely to be adequate for use in SampleApp.
I also left SkMSec where it is used to measure a duration rather than a timestamp. With the exception of SkMovie, which is used in Android, all of the uses appear to measure the execution time of a piece of code, which I would hope does not exceed 2^31 milliseconds.
Added skiatest::Timer to support a common idiom in tests where we want to measure the wallclock time in integer milliseconds. (Not used in tests/PathOpsSkpClipTest.cpp because it redefines things in Test.h.)
Removed tabs in tests/StrokerTest.cpp.
BUG=skia:4632
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1811613004
Review URL: https://codereview.chromium.org/1811613004
Before this change, the PDFCanon held a map from BitmapKeys
to SkImages for de-duping bitmaps. Even if the PDFDocument
serialized images early, the Canon still held a ref to that
image inside the map. With this change, the Canon holds a
single map from BitmapKeys to PDFObjects. Now, Images are
only held by the PDFObject, which the document serializes
and drops early.
This change also:
- Moves SkBitmapKey into its own header (for possible
reuse); it now can operate with images as well as
bitmaps.
- Creates SkImageBitmap, which wraps a pointer to a bitmap
or an image and abstracts out some common tasks so that
drawBitmap and drawImage behave the same.
- Modifies SkPDFCreateBitmapObject to take and return a
sk_sp<T>, not a T*.
- Refactors SkPDFDevice::internalDrawImage to use bitmaps
or images (via a SkImageBitmap).
- Turns on pre-serialization of all images.
BUG=skia:5087
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1829693002
Review URL: https://codereview.chromium.org/1829693002
Page Contents:
serialize early, at endPage()
SkPDFDocument:
Rather than holding all SkPDFDevices until onClose(), store
fGlyphUseage and array of pages objects.
perform_font_subsetting function removed:
First half moved to onEndPage.
Second half moved to onClose.
create_pdf_page function removed:
Merged into onEndPage.
generate_page_tree:
Refactored to use SkTArray<sk_sp<T>> over SkTDArray<T*>.
(the former is explicit about ownership, the latter is unclear.)
No longer populates a structure of objects to be dumped,
unnecessary since dump is always called after serialization.
Takes ownership of fPages and returns root of tree-ified version.
Less reference churn.
SkPDFGlyphSetMap:
use new-style iterator.
BUG=skia:5087
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1823683005
Review URL: https://codereview.chromium.org/1823683005
Call drop() after calling emitObject() on top-level objects. In Debug
mode, assert that each object is emited exactly once by asserting that
emitObject is never called after drop(). Same for addResources().
To make sure that top level objects don't get deleted prematurely,
SkPDFObjNumMap takes a reference.
Motivation: save RAM. Allow even earlier serialization with later
changes.
Also: Switch some SkTDArrays to SkTArrays.
BUG=skia:5087
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1790023003
Review URL: https://codereview.chromium.org/1790023003
The C++ standard library uses the name "release" for the operation we call "detach".
Rewriting each "detach(" to "release(" brings us a step closer to using standard library types directly (e.g. std::unique_ptr instead of SkAutoTDelete).
This was a fairly blind transformation. There may have been unintentional conversions in here, but it's probably for the best to have everything uniformly say "release".
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1809733002
Review URL: https://codereview.chromium.org/1809733002
Move these singletons into SkPDFCanon (there is still a single object
per document; output PDF size will not change).
Motivation: After this change, all indirectly-referenced SkPDFObjects
are serialized exactly once. The next CL will add a memory saving
feature: a method to purge an object's resources immediately after
serialization. After that, further changes wil allow some objects to be
serialized *before* SkDocument::close(), leading to potentially very
large memory savings.
BUG=skia:5087
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1788263002
Review URL: https://codereview.chromium.org/1788263002
Note to reviewers: Start with changes to SkPDFTypes.h
Many places that had a bare pointer owning a reference are refactored to
use a sk_sp.
There remain several places where a non-owning pointer `T*` should be
replaced with `const sk_sp<T>&` to eliminate the common pattern
`sk_sp<T>(SkRef(x))`.
Review URL: https://codereview.chromium.org/1775043002
SkUtility.h and SkTLogic.h implement a number of type traits now
available through <type_traits> and <utility>. This removes SkUtility.h,
replacing it with <utility>, and moves a number of traits in
SkTLogic.h to use the std:: equivelents. This change only uses C++11
parts of the standard library; SkTLogic.h will continue to provide
C++14 and beyond for now in the skstd namespace.
The changes to SkTLogic.h are being done gradually so that safe changes
may be landed confidently, with more risky changes in the future.
Review URL: https://codereview.chromium.org/1561683002
Passing &SkGoodHash to SkTHashMap and SkTHashSet doesn't guarantee that it's actually instantiated. Using a functor does.
BUG=skia:
Review URL: https://codereview.chromium.org/1405053002
This is required to build Chromium's ui_base target with -fmodules, but it
seems like a good change regardless. Including a header in a namespace is
weird.
BUG=chromium:543704
Review URL: https://codereview.chromium.org/1398843007
Note: this format does not yet pass validation tests.
Add skia_pdf_generate_pdfa GYP flag. Default to off for now.
PDF/A files are not reproducable, so they make correctness
testing harder.
Turn the Metadata struct into te SkPDFMetadata struct. This
splits out a lot of functionality around both kinds of metadata.
When PDF/A is used, add an ID entry to the trailer.
Add SkPDFObjNumMap::addObjectRecursively.
Test with
GYP_DEFINES=skia_pdf_generate_pdfa=1 bin/sync-and-gyp
ninja -C out/Release dm
out/Release/dm --config pdf --src skp gm -w /tmp/dm
With skia_pdf_generate_pdfa=0, all PDFs generated from GMs and
SKPs are identical. With skia_pdf_generate_pdfa=1, all PDFs
generated from GMs and SKPs render identically in Pdfium.
BUG=skia:3110
Review URL: https://codereview.chromium.org/1394263003
Reason for revert:
SkMD5 is not really part of the Skia library. This is breaking the roll by using it, since Chromium doesn't build it.
Original issue's description:
> SkPDF: Optionally output PDF/A-2b archive format.
>
> Note: this format does not yet pass validation tests.
>
> Add skia_pdf_generate_pdfa GYP flag. Default to off for now.
> PDF/A files are not reproducable, so they make correctness
> testing harder.
>
> Turn the Metadata struct into te SkPDFMetadata struct. This
> splits out a lot of functionality around both kinds of metadata.
>
> When PDF/A is used, add an ID entry to the trailer.
>
> Add SkPDFObjNumMap::addObjectRecursively.
>
> Test with
>
> GYP_DEFINES=skia_pdf_generate_pdfa=1 bin/sync-and-gyp
> ninja -C out/Release dm
> out/Release/dm --config pdf --src skp gm -w /tmp/dm
>
> With skia_pdf_generate_pdfa=0, all PDFs generated from GMs and
> SKPs are identical. With skia_pdf_generate_pdfa=1, all PDFs
> generated from GMs and SKPs render identically in Pdfium.
>
> BUG=skia:3110
>
> Committed: https://skia.googlesource.com/skia/+/939c0fe51f157104758bcb268643c8b6d317a530TBR=tomhudson@google.com,halcanary@google.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:3110
Review URL: https://codereview.chromium.org/1398193002
Note: this format does not yet pass validation tests.
Add skia_pdf_generate_pdfa GYP flag. Default to off for now.
PDF/A files are not reproducable, so they make correctness
testing harder.
Turn the Metadata struct into te SkPDFMetadata struct. This
splits out a lot of functionality around both kinds of metadata.
When PDF/A is used, add an ID entry to the trailer.
Add SkPDFObjNumMap::addObjectRecursively.
Test with
GYP_DEFINES=skia_pdf_generate_pdfa=1 bin/sync-and-gyp
ninja -C out/Release dm
out/Release/dm --config pdf --src skp gm -w /tmp/dm
With skia_pdf_generate_pdfa=0, all PDFs generated from GMs and
SKPs are identical. With skia_pdf_generate_pdfa=1, all PDFs
generated from GMs and SKPs render identically in Pdfium.
BUG=skia:3110
Review URL: https://codereview.chromium.org/1394263003
The "zeroPath" and emptystroke GMs capture this issue.
This CL changes the following PDF GMs: emptystroke dashing4
lineclosepath dashing3 zeroPath linepath
complexclip3_complex complexclip3_simple roundrects
degeneratesegments filltypes strokerect pathfill
inverse_paths desk_chalkboard.skp
After this change, all PDF GMs look better (closer to 8888).
The dashing4, emptystroke, and zeroPath GMs still need a lot
of work to make them look right.
BUG=538726
Review URL: https://codereview.chromium.org/1374383004
drawImage calls now properly embeds the original jpeg.
NOTE: drawBitmap*() calls no longer embed JPEG files when
possible (this is in advance of eliminating bitmaps backed
by encoded data). Chromium has already moved from
drawBitmap to drawImage.
Comparisons:
control:
total PDF drawImage/drawBitmap calls: 8010
total PDF jpeg images: 0
total PDF regular images: 3581
experiament:
total PDF drawImage/drawBitmap calls: 8014
total PDF jpeg images: 271
total PDF regular images: 3311
total PDF regular images: 3582 (271 + 3311)
When comparing rendered output there were perceptual
differences in the following four GMs: colorcube, emboss,
colormatrix, and tablecolorfilter. All of these differences
were improvements (that is, closer to the 8888 rendering)
due fixing a bug with colorfilters and forgetting to call
notifyPixelsChanged.
No SKPs had perceptual differences.
Total PDF size dropped from 133964 kB to 126276 kB, a 5.7%
improvement (mostly due to restoring use of JPG images in
SKPs).
BUG=skia:4370
Review URL: https://codereview.chromium.org/1372783003
Previously annotations added between saveLayer/restore were lost.
Merge annotations in SkPDFDevice::drawDevice(). Also modified code to
apply correct transformation and clipping on annotations added between
saveLayer/restore:
- Apply the initial transform only when adding the annotations into the
doc, otherwise we need to unapply sub-device's initial transform
before merging the annotations into parent-device.
- Apply only device-local clipping. fClipStack is in global coordinates,
which is not suitable to clip rects in sub-devices.
BUG=skia:4080
BUG=503515
Review URL: https://codereview.chromium.org/1257533004
Reason for revert:
Breaks Chrome roll.
obj/skia/ext/skia_chrome.skia_memory_dump_provider.o
does not have -I include/private on its include path, but transitively includes SkMessageBus.h.
Original issue's description:
> Port uses of SkLazyPtr to SkOncePtr.
>
> This gives SkOncePtr a non-trivial destructor that uses std::default_delete
> by default. This is overrideable, as seen in SkColorTable.
>
> SK_DECLARE_STATIC_ONCE_PTR still just leaves its pointers hanging at EOP.
>
> BUG=skia:
>
> No public API changes.
> TBR=reed@google.com
>
> Committed: https://skia.googlesource.com/skia/+/a1254acdb344174e761f5061c820559dab64a74cTBR=herb@google.com,mtklein@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:
Review URL: https://codereview.chromium.org/1334523002
This gives SkOncePtr a non-trivial destructor that uses std::default_delete
by default. This is overrideable, as seen in SkColorTable.
SK_DECLARE_STATIC_ONCE_PTR still just leaves its pointers hanging at EOP.
BUG=skia:
No public API changes.
TBR=reed@google.com
Review URL: https://codereview.chromium.org/1322933005
