Repository navigation
Sync S9: merge upstream PDFium up to 06fbf69b3 (Oct 10 2026) - #72
Merged
Merged
Conversation
Track CFFL_FormField lifetime using ObservedPtr during focus transitions and PWL window updates. Return early if the form field is destroyed while handling these events. Fixed: 567936263 Change-Id: I992b06f6bfde5cc839e9d11b90a098b8affcf5f6 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159451 Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/external/github.com/llvm/llvm-project/libcxxabi.git/+log/fc1897a2c12a..09351f6ec00c Created with: roll-dep third_party/libc++abi/src Change-Id: I11104651dc89cade173200076ece221b65b98b6a Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159392 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
Add named constants for OpenType and TrueType platform IDs and encoding IDs in core/fxge/fx_font.h. Use these symbols in CFX_Face and CPDF_TrueTypeFont. Introduce CharMapIdToFontEncoding() in core/fxge/cfx_face.cpp to map platform and encoding IDs to fxge::FontEncoding, replacing ToFontEncoding() and unifying font charmap classification. Fontations requires this functionality because Skrifa exposes raw platform and encoding IDs directly from OpenType 'cmap' records without FreeType's internal translation layer. Unifying charmap classification under CharMapIdToFontEncoding() allows both backends to share identical logic and paves the way for Fontations non-Unicode SFNT charmap support. -- Rename CharMapID to CharMapIdPair - since it is a pair. -- use uint16_t members to match FT value size. -- use designated initializers to avoid accidental swapping of IDs. -- Deprecate use of FT_ENCODING_OLD_LATIN_2 per freetype header. Bug: 42271123 Change-Id: Ia6c0e32b4c55077ec7ca633312d13f578b3e18eb Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159350 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Change-Id: Ic040658c96e46e30e96521e488b2d806bf37ff13 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159370 Auto-Submit: Andy Phan <andyphan@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
When rendering or loading glyph paths for built-in generic substitute fonts (FoxitSansMM and FoxitSerifMM), adjust horizontal scaling by dest_width / EmAdjust(outline.advance_width) if dest_width is set. Also return dest_width in CFX_Face::GetGlyphWidth() for built-in generic fonts when dest_width > 0. This matches substituted glyph widths to PDF /Widths dictionaries without using FreeType's Multiple-Master API. Bug: 42271123 Change-Id: I62d90493e12675cae461f2e99f0df8230ed4993a Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159290 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
https://chromium.googlesource.com/external/github.com/llvm/llvm-project/libc.git/+log/320824188c37..14e51fd2cd7e Created with: roll-dep third_party/llvm-libc/src Change-Id: I6643e2fb3ba5d620bb524445771d90f3752a97e6 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159470 Commit-Queue: Andy Phan <andyphan@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/external/github.com/llvm/llvm-project/libunwind.git/+log/45120ee23311..eb1ca4993b53 Created with: roll-dep third_party/libunwind/src Change-Id: Id3f17835d1f9489437aa966c769d62556b1935ce Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159412 Commit-Queue: Lei Zhang <thestig@chromium.org> Commit-Queue: Andy Phan <andyphan@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/chromium/src/testing/scripts/rust.git/+log/ead1c91bc31e..9c16e2f4c17e Created with: roll-dep testing/scripts/rust Change-Id: I626bab33f65845a2331da841a6e98081e190232a Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159371 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Andy Phan <andyphan@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/chromium/src/tools/win.git/+log/13cb6e5d223d..ae34cdd9867c Created with: roll-dep tools/win Change-Id: Iab867e88fdaf1e4a27211ffcef8fa3a8ddfeb064 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159450 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/chromium/src/third_party/rust/+log/4dec6f65bcb3..b8c38a533c84 Created with: roll-dep third_party/rust Change-Id: I7d458102151436fd27fa7666538ce54467edf5d4 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159452 Commit-Queue: Andy Phan <andyphan@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/chromium/src/buildtools.git/+log/6f6a5dbf04b7..64e87f9fbdaf Created with: roll-dep buildtools Change-Id: Icbd902c1560d0d6752751c05ea2f81101d67d4c8 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159393 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
llvmorg-24-init-3796-g20e97c4b-4:llvmorg-24-init-7747-g62397f8b-57 https://chromium.googlesource.com/chromium/src/tools/clang/+log/450d823868ea..52b22f34fa14 Created with: roll-dep tools/clang Change-Id: I7a6c8a039ce0c0795b6c7f1faf1dbab9856225bd Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159490 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Andy Phan <andyphan@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/external/github.com/llvm/llvm-project/clang/tools/clang-format.git/+log/70510081984c..9f796802e5d6 Created with: roll-dep third_party/clang-format/script Change-Id: I8a2a719bd0f562a126987764c18c6f34ee55948c Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159411 Auto-Submit: Andy Phan <andyphan@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
Change-Id: Ie2e4f2995f3bb24839bf1411947445ade9579775 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159410 Auto-Submit: Andy Phan <andyphan@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
Change-Id: I7a65e74ee8572cf3cf8dd659daf492527068db58 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159390 Reviewed-by: Lei Zhang <thestig@chromium.org> Auto-Submit: Andy Phan <andyphan@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/external/github.com/fastfloat/fast_float.git/+log/34164f547b7d..b0ab987b3dfd Created with: roll-dep third_party/fast_float/src Change-Id: I7e835aacfadbc0008123637e4535e0a89297f07c Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159391 Auto-Submit: Andy Phan <andyphan@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/chromium/deps/icu.git/+log/8cc91d9b6ab9..5aa526207171 Created with: roll-dep third_party/icu Change-Id: Ia6292fea7453c157024fb93ad002488e6279e1bb Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159430 Reviewed-by: Lei Zhang <thestig@chromium.org> Auto-Submit: Andy Phan <andyphan@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
… rolls) https://skia.googlesource.com/skia.git/+log/5549c93c9a1c..5b1a7f7c0644 Created with: roll-dep third_party/skia Change-Id: I73e56b18a098af0fbe71349e268ff1f5f2084a55 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159550 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/external/github.com/harfbuzz/harfbuzz.git/+log/886fc1e64538..ff36703b9501 Created with: roll-dep third_party/harfbuzz/src Change-Id: I64cccba6cf41995941a56999125dc48efbdef68e Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159532 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
https://chromium.googlesource.com/v8/v8.git/+log/df65d5c8f293..e110f6a9dff3 Version 15.5.19 Created with: roll-dep v8 Change-Id: Ic479da3dff7dfe69e84fda3ff468f07534df9511 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159533 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
Roll clang+rust llvmorg-24-init-3796-g20e97c4b-4:llvmorg-24-init-7747-g62397f8b-57 Roll build/ 2cd94c0ab..ebdaeefa6 (166 commits) https://chromium.googlesource.com/chromium/src/build.git/+log/2cd94c0abead..ebdaeefa62db Created with: roll-dep build Roll tools/rust/ 54d114869..e61a63fae (22 commits) https://chromium.googlesource.com/chromium/src/tools/rust/+log/54d1148695db..e61a63fae238 Created with: roll-dep tools/rust Change-Id: I92c4d66c7081865df63f5f5e5fdb4b29088a587c Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159510 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Auto-Submit: Andy Phan <andyphan@chromium.org>
https://chromium.googlesource.com/chromium/src/third_party/abseil-cpp.git/+log/435e7d977fb3..a00d74bb55b3 Created with: roll-dep third_party/abseil-cpp Change-Id: I00ec81b79e7665fb5e95b495b79c7678a4253277 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159530 Auto-Submit: Andy Phan <andyphan@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
…mits) https://chromium.googlesource.com/chromium/src/base/allocator/partition_allocator.git/+log/89615a645f02..dc6144ac93b2 Created with: roll-dep base/allocator/partition_allocator Change-Id: Ibfdba6ca2e64397bfcaacaf279b3c559ddb90cba Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159531 Commit-Queue: Lei Zhang <thestig@chromium.org> Auto-Submit: Andy Phan <andyphan@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
TranslateScanline24bppDefaultDecode() walked both the source and the destination scanline with raw pointers, each of its three arms inside UNSAFE_TODO(). The 8 bpc arm is a per-pixel RGB-to-BGR byte swap, which is exactly what fxcodec::ReverseRGB() does on spans, and its disjoint-copy loop is written so the compiler vectorizes it; call that instead. The destination scanline buffer never overlaps the decoder's source scanline, and ReverseRGB() falls back to an in-place swap if the two are ever the same buffer. The other two arms now write through a destination pixel-struct span. The 16 bpc arm reads its source through a struct that names the big-endian sample layout, so keeping each component's high byte no longer looks like an unexplained stride-6 index. The sub-8 bpc arm already read its components out of the bit stream through a span; only its stores needed converting. Bug: 553155247 Change-Id: I55048b4a84179bd7f7a58802126788aa391d5ca1 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/155971 Reviewed-by: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
ContinueStretchHorz()'s kManyBpptoManyBpp arm unrolls its one- and two-tap cases, since at those trip counts the loop is all overhead. The kManyBpptoManyBppWithAlpha arm computes the same thing, differing only in the per-tap alpha premultiply and in writing a fourth channel, but still runs the generic loop for every destination column. Give it the same treatment. The terms are summed in tap order, so the result is identical, and columns with three or more taps keep the existing loop. Bug: 553937496 Change-Id: Ib9206eb8b4a56d4fe2bfd18843e4ebe725618f61 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/156110 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
Replace the separate value members in CJS_DelayData with a single std::variant member `data`, since each CJS_DelayData instance only stores one property value. Also remove the unused `widestring` and `color` fields. Update CJS_Field::DoDelay() and the CJS_Field::AddDelay_*() methods in fxjs/cjs_field.cpp to read and write `CJS_DelayData::data`. TAG=agy CONV=d7374919-0adf-44a3-8c57-4bc07cdb5cea Change-Id: I0e2cddfd1e0af2afe24d8c73207c14446a4606d7 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159690 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
Add FFI methods to SkrifaFont to query SFNT charmap encoding records (platform ID and encoding ID) and map character codes to glyph IDs using subtable-specific lookups. Update CFX_Face charmap enumeration, selection, and character-to-glyph mapping to use these methods under Fontations. Bug: 42271123 Change-Id: I75e0eef3a8c50debe7363e5c184e0dac9b64388e Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159211 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
…_DIB. In CPDF_DIB::TranslateScanline24bpp(), stores through dest_scan (uint8_t) and the virtual call color_space_->GetRGBOrZerosOnError() may alias any object in memory under C++ strict aliasing rules, preventing the compiler from hoisting GetWidth() out of the per-pixel loop and forcing a memory reload on every iteration. Cache GetWidth() in a local width variable before the loop. In CPDF_DIB::TranslateScanline24bppDefaultDecode(), load the 16-bpc src channels into local uint8_t variables before writing to dest (matching fxcodec::ReverseRGB()), avoiding interleaved loads and stores when the compiler cannot prove src and dest do not overlap. Co-authored-by: Richard Winterton <rrwinterton@intel.com> Change-Id: Ie605d688918b73054c7601a7373e480486859872 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159571 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
V8 added third_party/llvm-libclang in https://crrev.com/c/8219268 for tools/metagen (https://crrev.com/c/7864481), which was enabled by default in https://crrev.com/c/8365609. Change-Id: Ibddd0e7a0ff8b24046fa19fa4ba1e3b7a810be54 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159534 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Andy Phan <andyphan@chromium.org>
https://chromium.googlesource.com/v8/v8.git/+log/e110f6a9dff3..0d546234ecf0 Version 15.7.21 Created with: roll-dep v8 Change-Id: I3e3c046748be729561b68dc3405694e84ea021f8 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159551 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Andy Phan <andyphan@chromium.org>
Allow pdf_enable_freetype=false when pdf_enable_fontations=true and pdf_enable_skia=true. Guard FreeType specific paths across core/fxge, testing environments, and fpdfsdk behind defined(PDF_ENABLE_FREETYPE). Populate font_data_ in CFX_Font::LoadFaceFromSpanStream() so that Skia device drivers can render fonts loaded from span streams instead of falling back to FT for this one case. Update Skia + Fontations text rendering test checksums in cfde_textout_unittest.cpp. Bug: 42271123 Change-Id: I403fb18c50fc39962843b410ae745856af8dbd96 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159552 Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
V8 deprecated FunctionCallbackInfo::Data() and PropertyCallbackInfo::Data() in favor of DataV2(), which returns Local<Data>. This does not change callback behavior. Bug: 559925519 Change-Id: I0f16a5437d76a4acfe6c71d3dff36ef53cc6de49 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159990 Reviewed-by: Lei Zhang <thestig@chromium.org> Auto-Submit: Deepak Mohan (Robo) <hop2deep@gmail.com> Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
https://chromium.googlesource.com/chromium/src/third_party/freetype2.git/+log/b76738eff2f3..891f904e2e7b Created with: roll-dep third_party/freetype/src Then update test expectations to match the latest FreeType behavior. See https://pdfium-review.googlesource.com/159770 for corpus test expectation updates. Change-Id: Id4538089434ea8f7ee3b3981865b323b566992b2 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159790 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Andy Phan <andyphan@chromium.org>
https://chromium.googlesource.com/chromium/src/third_party/freetype2.git/+log/6f5bcbbad52e..891f904e2e7b Created with: roll-dep third_party/freetype/src Also skip drawing zero-sized text in CFX_RenderDevice::DrawNormalText(), as FreeType's FT_Set_Matrix() now rejects zero matrices and falls back to the identity matrix, which would otherwise cause zero-sized text to be rendered visibly. Change-Id: Ibe1857897484916871f06d6c2b931bcb36c6ff89 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159950 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Andy Phan <andyphan@chromium.org>
This is now being DEPS'd into the tree. Change-Id: I3941eef2036970beaf39ab5f5b30db33c48f1b13 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160130 Commit-Queue: Lei Zhang <thestig@chromium.org> Auto-Submit: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
Unify macro usage to follow the convention where USE_ is for backend choices (e.g., Skia) and ENABLE_ is for additive features (e.g., XFA). - Remove PDF_USE_XFA from public/fpdfview.h as it was redundant with PDF_ENABLE_XFA and is unused throughout the chromium codebase. - Replace PDF_ENABLE_SKIA with PDF_USE_SKIA in the testing directory and BUILD.gn, as this was one oddball usage. - Remove redundant #define block for PDF_USE_SKIA in pdfium_test.cc. - More accurately name pdfium_test_config. Change-Id: I1de149cd743bd24b1cb99d493e199524d9655144 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160170 Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
This CL implements CalculateSSIM in pixel_diff_util, modifies image_diff.cpp to output mean_ssim and min_window_ssim, and updates pngdiffer.py to parse them without taking any new action. Eventually, these can be used to provide another sanity check on Fontations/Skia results that would have large MSE issues otherwise and would have to their diffs ignored for the moment. A short-circuit is provided in image_diff.cpp to avoid SSIM calculation if the images are exactly identical. Change-Id: I40e1e4e9b2cbc3b9f8862fdae5820d846b076e05 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160150 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Simplify PaethPredictor() algebraically (pa = abs(b - c), pb = abs(a - c), pc = abs((a - c) + (b - c))) and use branchless conditional selection. Extend the per-pixel register-carry fast path (previously used only for 3- and 4-byte Sub rows) to Bpp = 1, 2, 3, and 4 across Sub, Average, and Paeth rows, and reuse PredictSubRow() on the first row of Paeth. Carrying the running left and upper_left pixel state in local registers eliminates per-byte store-to-load forwarding stalls on dest_span[i - bpp], redundant loads of last_span[i - bpp], and per-byte span bounds checks via fxcrt::Zip(). On a 2048x200 image, PngPredictLine() with Paeth is 3.72x faster for 4-Bpp (5911.4 us -> 1588.9 us) and 2.33x faster for 3-Bpp (5553.3 us -> 2386.0 us), while remaining bit-exact for all filter tags and inputs. Bug: 549230311 Change-Id: I45ed513938044b868cb7ef526706948501b567ca Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160071 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
Note: public_deps should now be used iff a public header includes the dependency. This is because public_deps no longer exposes headers from dependencies, so the advice to avoid public_deps is no longer relevant. This is a part of the LSC https://docs.google.com/document/d/1zYUC_zuFrZozo9ogFHcf0Rzkiw9W7RVnaY-1fzfQ0AM/edit?tab=t.0 Bug: 556959580 Change-Id: Id9954ce4c8685204030e22a7a23891eb6a6a6964 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160112 Auto-Submit: Matt Stark <msta@google.com> Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
- In core/fxge, mark code as NOTREACHED() for the case when AGG is disabled at build time, yet somehow Skia is not available. - Compile FPDFViewEmbedderTest.Bug2087 only for builds that have both AGG and Skia. Otherwise the test does not make sense to run. Change-Id: If0f93635ca1a2859db4065e0c3740e4d092ac4d6 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160190 Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
FPDF_CountNamedDests() counts every key in the catalog's old-style /Dests dictionary, and FPDF_GetNamedDest() falls back to that dictionary for indices past the /Dests name tree. The fallback used the raw dictionary value, so when the value was an indirect reference, which is valid PDF, it was neither an array nor a dictionary. FPDF_GetNamedDest() then returned nullptr without reporting the name, and a document could report named destinations that it could not return by index. All other named destination lookups already dereference the value: the /Dests name tree lookups by name and by index, and the old-style lookup by name. As a result, FPDF_GetNamedDestByName() found indirect old-style destinations that FPDF_GetNamedDest() dropped. Dereference only the value of the matching key, so that the loop does not parse the objects behind the keys it walks past. Direct values are unaffected, and a dangling reference resolves to nothing, so FPDF_GetNamedDest() returns nullptr for it. Add a named_dests_old_style_indirect.pdf fixture whose old-style /Dests values are all indirect: a destination array, a destination dictionary with a /D array, and a dangling reference. The new embedder test checks that FPDF_GetNamedDest() returns the same destinations as FPDF_GetNamedDestByName() along with their names, that those destinations are for the expected pages, and that the dangling entry gives neither a destination nor a name. An alternative would be to expose GetNamedDestFromObject() from cpdf_nametree.cpp through a new CPDF_NameTree static for the old-style lookup by index. This change keeps the fix local instead. Found while working on the pdfium-native Node.js binding. Change-Id: I380f9b07dbe1bbc680e3c5ae4a21f161a1d554fa Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159090 Reviewed-by: Lei Zhang <thestig@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Avoid hitting a check when initializing a library by providing the correct default based upon what is linked into the binary. Do the same for the embedder_test_environment.cc. -- No need to provide designated-initializer for 0s. Change-Id: I82745b19cc8c8e7b5b058cd2aee1338be44fb3af Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160131 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
As seen when running buildtools/checkdeps/checkdeps.py. Change-Id: Icc6fe0267e30706a9f899e181ec70378deddf876 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160132 Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
Revert a tiny portion of https://pdfium-review.googlesource.com/c/160112 so Chromium can successfully roll PDFium and build the fuzzers. Bug: 556959580 Change-Id: Ic002f56f0eb5a9fff4d7eebaa2a398fc7d19bde2 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160250 Commit-Queue: Andy Phan <andyphan@chromium.org> Auto-Submit: Lei Zhang <thestig@chromium.org> Reviewed-by: Andy Phan <andyphan@chromium.org>
Just use third_party/rust/chromium_crates_io/vendor/cxx-v1/include/cxx.h directly. Bug: 396397336 Change-Id: Icbc6afb2ff91d7c46fd54da932afa10ef6c47d0f Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159930 Reviewed-by: Tom Sepez <tsepez@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Tests should test one thing. In this case, that the Brotli filter correctly decodes an image stream using DecodeParms predictor parameters. Previously, brotli_decodeparms.in was derived from bug_1015233.in, a test originally created to demonstrate a tiling pattern bug. Consequently, it included fractional pattern matrices ([.202488 0 0 -.449947 0 791.999]), subpixel path strokes (0.24 pt), and pattern tiling. Those extraneous rendering operations made the test non-portable. AGG and Skia rasterized the patterns and subpixel strokes differently, causing 120,196 out of 120,200 image bytes to diverge. That divergence required separate brotli_decode_parms_agg.png and brotli_decode_parms_skia.png expectation files. Simplify brotli_decodeparms.in to render the 4x4 Brotli image directly to the page without patterns or strokes. The resulting output is bit-for-bit identical between AGG and Skia, allowing replacement of the backend-specific PNGs with a single brotli_decode_parms.png and switching the test to CompareBitmap(). Change-Id: I172f21b04b43494ca3e5d1906ae7a217c6c05a5b Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160090 Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
First step towards simplifying the number of different forms of device class constructors. Group knockout is fundamentally a piece of rendering state - it only matters when blending a new primitive onto the canvas. Pass as an argument when needed. Change-Id: I0e187f79d1f975ece739c0ba788c89dc07e04e1d Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160290 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Prevents CFX_RenderDevice from being a driver/bitmap factory, and allow extension for test-only drivers. -- Add CFX_RenderDevice::CreateWithDriver() static method. -- Make GetDeviceDriver() public. -- Rewrite CreateForWindowsDC() using CreateWithDriver(). -- Rewrite CreateForSkiaCanvas() using CreateWithDriver(). -- Remove CFX_RenderDevice::CreateForNewBitmap*() -- C++23: need complete types for defaulted arguments. -- Add FakeDeviceDriver for testing. Change-Id: I4bb2b31519546c92a6bee82ad068ae56717eb5ca Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159910 Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
The change at https://pdfium-review.googlesource.com/150952 eliminated two-stage initialization of CFX_RenderDevice and routed FPDF_FFLDrawSkia() to CFX_RenderDevice::CreateForSkiaCanvas(). Consequently, AttachSkiaImpl() is no longer invoked with a null bitmap pointer during canvas recording. Replace the dead null-check in AttachSkiaImpl() with a CHECK(pBitmap), matching AttachAggImpl(). Remove the now-obsolete comments regarding null bitmaps in both AttachSkiaImpl() and AttachAggImpl(). Change-Id: Ib213880b0d231b95b0e13758722f1b549581205b Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160361 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Stop attempting to scale down CPDF_ScaledRenderBuffer dimensions once both width and height reach 1. Subsequent scalings cannot reduce the integer bounds further and cause the loop to spin until floating-point underflow eventually collapses dimensions to 0.0 exactly. Then, termination via CFX_DIBitmap::CalculatePitchAndSize() failure is an incidental quirk of its implementation. Change-Id: I9676e5ffe57ed3fe1aef6309fd1ee32c8517c320 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160410 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Replace calls to ByteString::Format() that only perform string substitution with the ByteString constructor taking an initializer list of ByteStringViews. Change-Id: I3a21c18693df2cacff37d679061c55cb95d081da Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160310 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
Bug: 570068745 Change-Id: I2aed2f81118f78a47f3266d9275d3c80ebdb5a41 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160430 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
Introduce DisplayListDeviceDriver, an implementation of RenderDeviceDriverIface for testing that accumulates semantic drawing commands directly into a string display list without pixel rasterization. The intent is to avoid needing platform-specific .png files when some higher level function is all that is being tested. Add Render*ToDisplayList methods to EmbedderTest to render pages and forms to display lists for semantic verification. Add a test in fpdf_view_embeddertest.cpp verifying output on rectangles.pdf. -- Add --dump flag to pdfium_test. -- Retire the fake driver. Change-Id: Ibc77e853f7fe97bc3d8258f85f11d38144c1869b Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/159394 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org>
This CL updates library initialisation in fpdf_view.cpp to disable Brotli when the configuration version is less than 6 or when m_BrotliEnabled is not set to true by the embedder. Change-Id: I9ebe03a91d0f8b7d292cc7d44e98de394b9f57bb Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/158910 Reviewed-by: Tom Sepez <tsepez@chromium.org> Commit-Queue: Tom Sepez <tsepez@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org>
Change some signed integer math operations to use FX_SAFE_INT32 and avoid UB. Bug: 570876470 Change-Id: I9ecc7b1520420424d2d8f7bb5d62dfb967387c3f Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160390 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
GetNormalizedPoints() only drops points that repeat the previous point, so a rectangle with an extra vertex in the middle of an edge, such as: x1 y1 m x2 y1 l x2 y2 l xm y2 l x1 y2 l h is not a rectangle to CFX_Path::IsRect() and GetRect(). Some PDF generators (seen in Amyuni PDF Converter output) clip every image strip of a rasterized page with such a page-sized rectangle. CFX_AggDeviceDriver::SetClip_PathFill() then rasterizes a full clip mask for each strip instead of intersecting a rectangle. A page with 10,000 such clips takes 9.1 s to render at 200 DPI with pdfium_test, and 0.3 s with this change. Also skip a line point that lies strictly between the previous point and the next point on a horizontal or vertical line. Removing it does not change the shape of the path, for either fill rule. The "re" writers already handle such rectangles since https://pdfium-review.googlesource.com/160011. Filled rectangles with such points now take the FillRect() fast path in CFX_RenderDevice::DrawPathWithBlend(), just like the same rectangles written without the extra points. This changes the rendering of the tiny rectangles in the 1751_1 corpus test, so also roll DEPS to pick up the new corpus test expectations in https://pdfium-review.googlesource.com/c/pdfium_tests/+/160230 and https://pdfium-review.googlesource.com/c/pdfium_tests/+/160270. Bug: 570363704 Change-Id: I7d0c57c8137331bd1fc4068bfe215e557622012d Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160010 Commit-Queue: Lei Zhang <thestig@chromium.org> Reviewed-by: Lei Zhang <thestig@chromium.org> Reviewed-by: Tom Sepez <tsepez@chromium.org>
Ensure the rectangle coordinates of page objects are finite before processing them, avoiding potential issues when clamping and converting non-finite float values to integers. Also reuse the retrieved rectangle to avoid redundant calls to GetRect(). Fixes UBSan `float-cast-overflow` bug: ``` CRASH LOG: SUMMARY: UndefinedBehaviorSanitizer: float-cast-overflow third_party/pdfium/HEAD/core/fpdftext/cpdf_textpage.cpp:674:9 (REDACTED) ``` PiperOrigin-RevId: 995805511 Change-Id: I1bff4d6a9019bcd337d3707616b1efb6dc893292 Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/160330 Reviewed-by: Lei Zhang <thestig@chromium.org> Commit-Queue: Lei Zhang <thestig@chromium.org>
The linuxmusl-arm64 target was failing because libyuv arm64 sources require dot-product, SVE2, and SME instructions that the musl GCC toolchain rejects. This disables libyuv for that target by setting pdf_use_libyuv=false in build-target.sh and documents the reason in the sync notes alongside the related build findings.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
11 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.