summaryrefslogtreecommitdiff
path: root/core/fxge/skia
AgeCommit message (Collapse)Author
2017-03-28Revert "Refcount all CFX_DIBSources (and subclasses) all the time."dsinclair
This reverts commit 0004f29bf6ee3c6060a272c79f14993e92e053c7. Reason for revert: Breaks build with skia_paths enabled (which will break the chrome roll). ../../third_party/pdfium/core/fxge/skia/fx_skia_device.cpp:1858:38: error: no member named 'get' in 'CFX_RetainPtr<CFX_DIBitmap>' ../../third_party/pdfium/core/fxge/skia/fx_skia_device.cpp:1861:42: error: no member named 'get' in 'CFX_RetainPtr<CFX_DIBitmap>' ../../third_party/pdfium/core/fxge/skia/fx_skia_device.cpp:2987:15: error: no viable overloaded '=' ../../third_party/pdfium/core/fxge/skia/fx_skia_device.cpp:2991:18: error: no viable overloaded '=' ../../third_party/pdfium/core/fxge/skia/fx_skia_device.cpp:2999:17: error: no viable overloaded '=' ../../third_party/pdfium/core/fxge/skia/fx_skia_device.cpp:3001:43: error: no member named 'GetObject' in 'CFX_RetainPtr<CFX_DIBitmap>' Original change's description: > Refcount all CFX_DIBSources (and subclasses) all the time. > > There are currently several ownership models for these objects, > including ad-hoc logic for sharing and deletion, and the > now-redundant CFX_DIBitmapRef externally-counted handle to the DIBs. > > Replace them all with the internal refcount scheme. > > Change-Id: I2db399dfc19219eda384f94cc989353b78ce2872 > Reviewed-on: https://pdfium-review.googlesource.com/3166 > Reviewed-by: dsinclair <dsinclair@chromium.org> > Commit-Queue: dsinclair <dsinclair@chromium.org> > TBR=thestig@chromium.org,tsepez@chromium.org,dsinclair@chromium.org,pdfium-reviews@googlegroups.com NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true Change-Id: I678b1fbc5e666cf7a19372ebaff3270fb115ba5e Reviewed-on: https://pdfium-review.googlesource.com/3243 Reviewed-by: dsinclair <dsinclair@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-03-28Refcount all CFX_DIBSources (and subclasses) all the time.Tom Sepez
There are currently several ownership models for these objects, including ad-hoc logic for sharing and deletion, and the now-redundant CFX_DIBitmapRef externally-counted handle to the DIBs. Replace them all with the internal refcount scheme. Change-Id: I2db399dfc19219eda384f94cc989353b78ce2872 Reviewed-on: https://pdfium-review.googlesource.com/3166 Reviewed-by: dsinclair <dsinclair@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-03-27Fix some ASAN issues in fx_skia_deviceNicolas Pena
- Flush is needed in ~CFX_FxgeDevice, otherwise it may be called after deleting the bitmap, when calling the destructor of the skia device driver. - SkDashPathEffect::Make copies the given intervals instead of owning them, so free the input to that method. - If StartDIBits creates a new CFX_ImageRenderer, then the corresponding CancelDIBits needs to delete the handle. Bug: chromium:705131 Change-Id: I22c7c51a4070e73538eb8af51a60afeaa67f8bb7 Reviewed-on: https://pdfium-review.googlesource.com/3230 Commit-Queue: Nicolás Peña <npm@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-03-27fix new tab crash in skia clip stackCary Clark
The crash on the new tab page is triggered by processing a transparency. This creates a new Skia device in CPDF_RenderStatus::LoadSMask(): // cpdf_renderstatus.cpp # 2557 if (!bitmap_device.Create(width, height, format, nullptr)) which sets the Skia clip stack to empty. It then calls RenderObjectList() RenderSingleObject() ProcessClipPath() which resets the clip stack; // cpdf_renderstatus.cpp # 1882 m_LastClipPath = ClipPath; m_pDevice->RestoreState(true); At this point m_LastClipPath contains {m_Ref={m_pObject={m_pObj=empty } } } The impelemntation in CFX_AggDeviceDriver::RestoreState() is // fx_agg_driver.cpp # 1283 if (m_StateStack.empty()) return; This hides unbalanced save/restores, but reworking PDFium to balance is nontrivial. R=dsinclair@chromium.org BUG=chromium:704442 Bug: Change-Id: Ia70d4dd7bd118e40adc5c029acbaa0b66372d3aa Reviewed-on: https://pdfium-review.googlesource.com/3191 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-03-24Fix CFX_FxgeDevice constructor and destructor in SkiaNicolas Pena
Attach and Create methods may be called without _SKIA_SUPPORT_, so the bitmap ownership code can't be if def'd to _SKIA_SUPPORT_. Bug:chromium:704835 Change-Id: If6a7da508156ca3c45898bf101d84dd8295e479f Reviewed-on: https://pdfium-review.googlesource.com/3210 Reviewed-by: Cary Clark <caryclark@google.com> Commit-Queue: Nicolás Peña <npm@chromium.org>
2017-03-23fix skia path debugCary Clark
SHOW_SKIA_PATH code had bit-rotted. R=dsinclair@chromium.org BUG= Change-Id: I8961853aa0373fb0a04f829b8e78722197e8f5af Reviewed-on: https://pdfium-review.googlesource.com/3122 Reviewed-by: dsinclair <dsinclair@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-03-23Handle the Clip command list being emptyDan Sinclair
If the command list is already empty when we attempt to restore the clip we will access outside the command array. BUG=chromium:704442 Change-Id: Ib7c44e14f0de175f1c10c1a538d16987aa49bf8a Reviewed-on: https://pdfium-review.googlesource.com/3159 Reviewed-by: Cary Clark <caryclark@google.com> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-03-14Replace FX_FLOAT with underlying float type.Dan Sinclair
Change-Id: I158b7d80b0ec28b742a9f2d5a96f3dde7fb3ab56 Reviewed-on: https://pdfium-review.googlesource.com/3031 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org> Reviewed-by: Nicolás Peña <npm@chromium.org>
2017-02-23Convert point x,y into CFX_PointFdan sinclair
This Cl converts the PointX,PointY pairs into a CFX_PointF. Change-Id: I46897832077c317a5bffb4e568550705decbc40c Reviewed-on: https://pdfium-review.googlesource.com/2821 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org> Reviewed-by: Nicolás Peña <npm@chromium.org>
2017-02-16Change FXTEXT_CHARPOS to use CFX_PointFDan Sinclair
This CL updates the Origin x,y coordinates in FXTEXT_CHARPOS to be an CFX_PointF. Change-Id: I67281db2cb82687e12490145f7c99aee908e5fa8 Reviewed-on: https://pdfium-review.googlesource.com/2718 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Nicolás Peña <npm@chromium.org>
2017-02-15Cleanup CFX_PathData.Dan Sinclair
This CL replaces the array of path points with a vector. Cleaning up the usage as required. Change-Id: Ifa386a2c847005fef68af748ebe99c4e08961238 Reviewed-on: https://pdfium-review.googlesource.com/2710 Reviewed-by: Nicolás Peña <npm@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-02-15remove skia clipstack debuggingCary Clark
SkClipStack is going away. It was only used for debugging in PDFium. R=dsinclair@chromium.org Change-Id: I476bfe84874916fc7c72ae8942afb021662935d1 Reviewed-on: https://pdfium-review.googlesource.com/2730 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: dsinclair <dsinclair@chromium.org>
2017-02-14Reland Cleanup CPDF_MeshStreamDan Sinclair
This CL fixes up the bits of the CPDF_MeshStream CL that depended on the Origin CL and integrates the fixes for the Skia bots. Change-Id: I470e49c35b809f00fecbaeb303de15c0db3fe590 Reviewed-on: https://pdfium-review.googlesource.com/2698 Reviewed-by: Nicolás Peña <npm@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-02-14Revert "Convert Origins to points"Dan Sinclair
This reverts commit da83d3a5cc09c4056310b3cf299dbbccd5c70d11. Reason for revert: Reverting chain to see if fixes Chrome roll. Original change's description: > Convert Origins to points > > This CL converts various OriginX, OriginY pairs into CFX_PointF objects. > > Change-Id: I9141f7fc713c710b2014d4fdcdec7dc93501f844 > Reviewed-on: https://pdfium-review.googlesource.com/2575 > Commit-Queue: dsinclair <dsinclair@chromium.org> > Reviewed-by: Nicolás Peña <npm@chromium.org> > TBR=tsepez@chromium.org,dsinclair@chromium.org,npm@chromium.org,pdfium-reviews@googlegroups.com NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Change-Id: I949fb4ec712e2587e7d0ef0191c34db198b61dcc Reviewed-on: https://pdfium-review.googlesource.com/2696 Reviewed-by: dsinclair <dsinclair@chromium.org> Commit-Queue: dsinclair <dsinclair@chromium.org>
2017-02-13Convert Origins to pointsDan Sinclair
This CL converts various OriginX, OriginY pairs into CFX_PointF objects. Change-Id: I9141f7fc713c710b2014d4fdcdec7dc93501f844 Reviewed-on: https://pdfium-review.googlesource.com/2575 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Nicolás Peña <npm@chromium.org>
2017-02-08Update to use CFX_Rect{F} and CFX_Matrix constructors.Dan Sinclair
This Cl updates the code to use the constructors instead of creating an empty object and calling Set(). It also removes the various memsets of the CFX_Rect{F} classes. Change-Id: I6e20cec00866a38372858dcba5a30d31103172e4 Reviewed-on: https://pdfium-review.googlesource.com/2550 Commit-Queue: dsinclair <dsinclair@chromium.org> Reviewed-by: Nicolás Peña <npm@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
2017-02-07Use enum class for PathPoint types.Nicolas Pena
This hopefully makes it less confusing what the description of a point is. Currently we have defines for the types, which is confusing because a point can only be one of the three. And it is mixed up with whether the point is closing a figure or not. Change-Id: Icd71355d69c77b3d52ca78e03bc379081ff87753 Reviewed-on: https://pdfium-review.googlesource.com/2552 Commit-Queue: Nicolás Peña <npm@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
2016-12-12fix linux buildcaryclark
Found when adding pdf_use_skia_paths canary to skia status R=dsinclair@chromium.org, borenet@google.com BUG=skia:5849 Review-Url: https://codereview.chromium.org/2569003002
2016-12-12sync with latest skia apicaryclark
The canvas clip op changed to a enum class to safeguard against a regular enum used as a bool in an overloaded interface. R=reed@google.com, dsinclair@chromium.org Review-Url: https://codereview.chromium.org/2568193002
2016-12-09update to latest skia apichromium/2949chromium/2948chromium/2947caryclark
Skia has changed its clipping API from one enum to another. Fix PDFium to match. R=dsinclair@chromium.org,reed@google.com,herb@google.com Review-Url: https://codereview.chromium.org/2559123004
2016-12-06re-enable skia cachingchromium/2944caryclark
This permits consecutive path drawing or text drawing to combine if their graphics state matches. This speeds up rendering considerably when PDFium draws in many small parts. It also allows discarding changes to the clip state that have no effect from draw to draw. All corpus tests draw equivalently with caching enabled or disabled. Change the member order in CPDF_PageRenderContext so the device is flushed before the referencing annotation is deleted. Add more printf style debugging for test draws. R=dsinclair@chromium.org Review-Url: https://codereview.chromium.org/2546803003
2016-12-05enable skia gradients, fix bugscaryclark
Allow shading to use Skia for some gradients; linear, radial, and coons patch. The corpus uncovered a few gradient-related bugs in the Skia driver which are fixed as well. Fix corpus 11.pdf by ignoring zero area filled paths. PDFium uses CFX_PathData::GetZeroAreaPath() to determine if a path that doesn't enclosing anything should be drawn anyway, and CFX_AggDeviceDriver::GetDriverType() returns 1 so that the empty path is tweaked so that it doesn't draw too often. There are no tests in the corpus that show the utility of this approach; 11.pdf draws correctly if the zero area paths are ignored. Standardize debugging output so that a separate tool can visualize it. R=dsinclair@chromium.org, thestig@chromium.org Review-Url: https://codereview.chromium.org/2530343004
2016-11-23Rename IFWL classes which do not have CFWL equivalentsdsinclair
This CL moves the IFWL classes that do not have CFWL class buddies to have the CFWL name. This CL leaves the tree in a weird state of having CFWL be two hierarchies, one of which is intertwined with the IFWL hierarchy. This should be commited just before the CL to move the rest of IFWL to CFWL. Review-Url: https://codereview.chromium.org/2525083002
2016-11-23fix negative font sizecaryclark
PDF can specify a negative font size to change the y-axis for text. Add support in the Skia driver for that. The cached form for text and paths had fallen out of sync with the uncached code used by skiapaths, so sync that up so that the mainline skia setting has those bug fixes as well. This does not fix the underlying bug in 528103, but gets Skia on parity with native PDFium text rendering. The bold text is still drawn in the wrong place. R=dsinclair@chromium.org BUG=528103 Review-Url: https://codereview.chromium.org/2528663002
2016-11-22handle antialiased rendering as premultipliedcaryclark
Transparencies and bitmap patterns need to be unpremultiplied after Skia renders them so that PDFium can use its own compositing. Also added some linear (i.e. axial) gradient support, although its unclear if any of the test corpus uses this feature. R=dsinclair@chromium.org Review-Url: https://codereview.chromium.org/2520073003
2016-11-21Fixup lint flags.Dan Sinclair
The -build/include setting was masking out build/include_what_you_use. This CL restores them, fixes any build errors, and adds NOLINT as needed. As well, the runtime/explicit and runtime/printf flags are aslo enabled and NOLINT'd. lint cleanups Change-Id: Ib013b3eb29c8d0e48cad74c5df9028684130719f Reviewed-on: https://pdfium-review.googlesource.com/2030 Reviewed-by: Tom Sepez <tsepez@chromium.org>
2016-11-21use skia textcaryclark
This adds text to the _SUPPORT_SKIA_PATHS_ variant. The output of the test corpus has more differences, but no feature changes or dropouts that I can see. The text is a bit lighter. Bungeman thinks this is because of our support of SRGB color, although the fonts also look different so maybe there's more to the story. The output looks compatible with the variations I see in Adobe Reader and Firefox on the same platform. There's probably more text tuning to do. Additionally, this turns off the caching feature for this variation. Caching can't work because the drawing alternates between Skia and PDFium's native blits, so any state caching won't know about changes and drawing performed by the other. R=dsinclair@chromium.org, bungeman@chromium.org Review-Url: https://codereview.chromium.org/2520483002
2016-11-17fix corpus testscaryclark
All corpus tests draw reasonably well. There are pixel differences, but no missing features or large scale errors. This was verified by running run_corpus_tests.py on skiapaths and agg build dirs, then running the skdiff tool to generate a report of the pixel difference. For example, after running $ ninja -C out/skia skdiff $ ./out/skia/skdiff /repo/pdfium/out/agg/gen/pdfium/testing/corpus/ /repo/pdfium/out/skiapaths/gen/pdfium/testing/corpus/ ~/aggskiadiffs view the report in a browser at ~/aggskiadiffs/index.html R=dsinclair@chromium.org BUG=skia:5973 Review-Url: https://codereview.chromium.org/2504863003
2016-11-16Create a subset of skia support for paths onlycaryclark
This is a continuation of https://codereview.chromium.org/2346483006/ This removes the need for agg, without providing full Skia support. It doesn't work yet, but it does compile and run for simple PDFs. R=dsinclair@google.com Committed: https://pdfium.googlesource.com/pdfium/+/5d223298b26c9b2b6284cba9a51521d3873b6e58 Review-Url: https://codereview.chromium.org/2491693002
2016-11-15Revert of Create a subset of skia support for paths only (patchset #10 ↵dsinclair
id:180001 of https://codereview.chromium.org/2491693002/ ) Reason for revert: Breaking the chrome roll. https://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_compile_dbg_ng/builds/306015/steps/generate_build_files%20%28with%20patch%29/logs/stdio Original issue's description: > Create a subset of skia support for paths only > > This is a continuation of https://codereview.chromium.org/2346483006/ > > This removes the need for agg, without providing > full Skia support. > > It doesn't work yet, but it does compile and run > for simple PDFs. > > R=dsinclair@google.com > > Committed: https://pdfium.googlesource.com/pdfium/+/5d223298b26c9b2b6284cba9a51521d3873b6e58 TBR=dsinclair@google.com,caryclark@google.com # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true Review-Url: https://codereview.chromium.org/2503883003
2016-11-15Create a subset of skia support for paths onlycaryclark
This is a continuation of https://codereview.chromium.org/2346483006/ This removes the need for agg, without providing full Skia support. It doesn't work yet, but it does compile and run for simple PDFs. R=dsinclair@google.com Review-Url: https://codereview.chromium.org/2491693002
2016-11-02Remove FX_BOOL from coretsepez
Review-Url: https://codereview.chromium.org/2477443002
2016-10-31remove deprecated skia transfer mode usagecaryclark
Skia has removed transfer modes as a class, and made it an enum. Update PDFium to match. R=reed@google.com,dsinclair@chromium.org Review-Url: https://codereview.chromium.org/2468443002
2016-10-28Fix FX_BOOL / int issue in fx_skia_device.cpptsepez
Review-Url: https://codereview.chromium.org/2449323006
2016-10-17Move fxge/android to if(is_android) on buildchromium/2894npm
core/fxge/android files all have if _FX_OS_ == _FX_ANDROID_. Instead, put the files inside if (is_android) on Build.gn. Also, the skia file is already under if (pdf_use_skia), so the if defined(_SKIA_SUPPORT_) can be removed. Review-Url: https://codereview.chromium.org/2413983004
2016-10-06switch to new skia blend mode enum classcaryclark
Skia's interface to transfer modes is not described by an enum instead of a class. R=reed@google.com, dsinclair@chromium.org Review-Url: https://codereview.chromium.org/2394683004
2016-10-04Move core/fpdfapi/fpdf_parser to core/fpdfapi/parserdsinclair
BUG=pdfium:603 Review-Url: https://codereview.chromium.org/2392603004
2016-10-04Move core/fpdfapi/fpdf_page to core/fpdfapi/pagedsinclair
BUG=pdfium:603 Review-Url: https://codereview.chromium.org/2386423004
2016-10-04update skia to latest sourcescaryclark
Update clip to use intersect verb from canvas. R=dsinclair@chromium.org Review-Url: https://codereview.chromium.org/2384283002
2016-10-03Add ptr_util.h from base until std::make_unique<> availabletsepez
Review-Url: https://codereview.chromium.org/2386273004
2016-09-29Move fpdfsdk/include to fpdfsdkdsinclair
BUG=pdfium:611 Review-Url: https://codereview.chromium.org/2384503003
2016-09-29Move core/fxge/include to core/fxgedsinclair
BUG=pdfium:611 Review-Url: https://codereview.chromium.org/2377393002
2016-09-29Move core/fxcrt/include to core/fxcrtdsinclair
BUG=pdfium:611 Review-Url: https://codereview.chromium.org/2382723003
2016-09-29Move core/fxcodec/codec/include and core/fxcodec/include files updsinclair
BUG=pdfium:611 Review-Url: https://codereview.chromium.org/2381063002
2016-09-29Move core/fpdfapi/fpdf_parser/include to core/fpdfapi/fpdf_parserdsinclair
BUG=pdfium:611 Review-Url: https://codereview.chromium.org/2383543002
2016-09-20Refactor CFX_FontCache to have only one in GE Module.art-snake
After this CL: only one global CFX_FontCache used. Any cached items from it, are released, when they are not being used. This is restore part of reverted CL: Original CL: https://codereview.chromium.org/2158023002 Revert reason: BUG=647612 Fix bug CL: https://codereview.chromium.org/2350193003 Review-Url: https://codereview.chromium.org/2350243002
2016-09-19Revert of Pdfium: Fix fonts leaking on ClosePage. (patchset #10 id:180001 of ↵dsinclair
https://codereview.chromium.org/2158023002/ ) Reason for revert: Causes heap-use-after-free. See crbug.com/647612. Original issue's description: > Fix memory leaking on ClosePage. > CFX_FontCache refactoring: > after this CL: Only one global CFX_FontCache used. Any cached items from it, are released, when its are not used. > > BUG=79367,48791 > > The fonts was not cleared after unloading pages. > > Test pdf: > > http://www.nasa.gov/pdf/750614main_NASA_FY_2014_Budget_Estimates-508.pdf > > For this file, we have ~5 fonts per page, which equal ~1 Mb per page. > In this PDF we have 670 pages, as result after slow scrolling(reading) full document we have ~600 Mb fonts data in memory. > > memory usage of PDF Plugin: > before this CL: ~660 Mb > after this CL: ~100 Mb > > Committed: https://pdfium.googlesource.com/pdfium/+/cde5101eb15b24519e89fa500fe37038bc8e2201 TBR=tsepez@chromium.org,brucedawson@chromium.org,npm@chromium.org,art-snake@yandex-team.ru # Not skipping CQ checks because original CL landed more than 1 days ago. BUG=79367,48791 Review-Url: https://codereview.chromium.org/2350763002
2016-09-15Fix memory leaking on ClosePage.chromium/2862art-snake
CFX_FontCache refactoring: after this CL: Only one global CFX_FontCache used. Any cached items from it, are released, when its are not used. BUG=79367,48791 The fonts was not cleared after unloading pages. Test pdf: http://www.nasa.gov/pdf/750614main_NASA_FY_2014_Budget_Estimates-508.pdf For this file, we have ~5 fonts per page, which equal ~1 Mb per page. In this PDF we have 670 pages, as result after slow scrolling(reading) full document we have ~600 Mb fonts data in memory. memory usage of PDF Plugin: before this CL: ~660 Mb after this CL: ~100 Mb Review-Url: https://codereview.chromium.org/2158023002
2016-09-15Rename dictionary set and get methodsdsinclair
This Cl makes the Get and Set methods consistenly use {G|S}et<Type>For. BUG=pdfium:596 Review-Url: https://codereview.chromium.org/2334323005
2016-08-19Apply structure packing to FXTEXT_CHARPOS to save space.npm
BUG=pdfium:562 Review-Url: https://codereview.chromium.org/2257313002