* fix(text): read a TrueType font without an encoding as WinAnsi (fixes#1468)
* fix(text): describe the WinAnsi reading by pdftotext's output
---------
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Follow up #1237 by deriving Bradford scaling ratios from the source and
D65 white points in LMS. Preserve CalRGB and the merged Indexed/Range
handling. Add independent neutral/chromatic, identity, asymmetric Range,
and degenerate-cone render regressions.
Co-authored-by: Codex <noreply@openai.com>
A PDF whose fonts aren't embedded is drawn with PDFium's Foxit stand-ins
for the standard 14 fonts. Those carry only the 230-odd glyphs of the
Standard, WinAnsi and Mac Roman encodings, so a letter a font names beyond
them in its /Differences mapped to .notdef and drew nothing. iText 2
writes Turkish exactly this way (Times New Roman left unembedded,
/Gbreve /Idotaccent /Scedilla /gbreve /scedilla over Ð Ý Þ ð þ), and such
letters showed blanks for Ğ, İ, Ş, ğ and ş while ı, ç, ö and ü drew.
StandardKind now builds a Latin Extended-A letter the stand-in lacks from
the base letter and accent it does have, as Type 1 seac composites and
the standard-14 AFM composites are built: the accent is centred on the
letter, raised over capitals and letters with ascenders (and moved along
the slant in italics), cedillas hang below and ogoneks from the right
edge. Composite glyph ids sit above any font's glyph count, one per
character code; their width comes from the standard metrics of the name.
Test: raster::tests::accented_letters_the_standard_stand_ins_lack_are_built_from_their_parts
(fails without the patch: Gbreve drew no pixel).
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* M2: snap shrunk interpolated images to whole pixels so scans stay sharp
A scanned page (one 300 dpi image with /Interpolate true) was shrunk to a
ceil-sized raster and then sampled a second time at a fractional scale and
offset, which softened every edge next to Acrobat. An axis-aligned shrunk
interpolated image is now shrunk once to its rounded device rectangle and
drawn 1:1 at the rounded corner. Rotated, skewed, flipped, Type 3 and
non-interpolated images are unchanged.
On a synthetic scan page the mean error against an area-average reference
falls from 3.4 to 1.6 grey levels at 96 dpi (3.0 to 1.5 at 144 dpi).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* vendor: list the hayro snap patch as (14) in the Patches and Tests columns; correct gaps.md patch counts
The patch description had been appended to the Tests column of one of the
two hayro rows. gaps.md now counts the distinct patches vendor/README.md
lists (26 hayro-interpret, 15 hayro, 3 hayro-jbig2, 17 hayro-syntax).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Revisions 5-6 of the standard security handler key the file on the
password processed with SASLprep (ISO 32000-2 7.6.4.3.3, Algorithm 2.A
step a). pdfcraft-crypt writes and reads it so, but the vendored
hayro-syntax skipped the step ("TODO: Do the above.") and used the UTF-8
bytes as typed. A password that SASLprep changes (a decomposed "cafe" +
U+0301, a no-break space, fullwidth letters typed by an IME in full-width
mode) was wrong for the renderer: Protect with such an open password was
rolled back ("the password is incorrect"), and an AES-256 file with such a
password did not open in a session with it, although pdfcraft-cos
accepted it.
- hayro-syntax (vendored patch 16) tries an R5/R6 password in its SASLprep
form (as pdfcraft-crypt computes it: as given if SASLprep refuses it,
UTF-8, at most 127 bytes), then, when that differs, as given.
stringprep, already in the tree through pdfcraft-crypt, is an optional
dependency enabled by the std feature; without std nothing changes.
- Tests: hayro-syntax crypto::pdfcraft_password_tests (candidate list and
R5/R6 dictionaries built from prepared passwords) and two engine tests
(Protect and reopen, and pdfcraft-cos-encrypted files) for the three
cases, with precomposed "caf\u{E9}" as the control. The dictionary test
and both engine tests fail without the fix.
Fixes#1296.
* Text extraction: honour /ActualText, so Thai from Chrome PDFs is searchable
Chrome (Skia) writes /ActualText around each Thai, Indic or Arabic cluster
whose shaped glyphs (a tone mark moved clear of a vowel, a ligature) have no
Unicode of their own. Text extraction ignored it, so those marks were lost:
"น้ำหนัก" came out as "นาหนัก", "เป็น" as "เปน", and searching for any
such word found nothing.
- hayro-interpret patch (25): Device::actual_text hands the device a
marked content sequence's /ActualText, inline or named in /Properties.
- TextDevice turns the glyphs of a replacement into one glyph carrying the
replacement text, boxed around them. The outermost replacement wins, an
empty one stands for no text, and one spanning several lines or writing
directions keeps per-glyph text so selection and lines still work.
- Canva draws text twice (outline or shadow) and marks only one drawing:
the unmarked redraw at the same places takes the same replacement, so the
existing redraw detection still drops the copy.
Checked on 1,114 real-world PDFs against the previous build, with pdftotext
as the reference: 957 extract byte-identical text, 157 change and all of
them read closer to the reference (exact matches 3 -> 122), none got worse,
none newly fail, and total extraction time is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Text extraction: bound the work of the ActualText redraw dedupe
`TextDevice::finish()` compared every replacement with every unmatched glyph
in up to 4096 grid cells, so a hostile page stacking ~10^5 glyphs and ~10^5
replacements at one spot meant ~10^10 comparisons (a hang). It now charges
each visited cell and each candidate run against a per-page budget
(`MAX_REDRAW_WORK`, 2^22) and stops deduplicating once it is spent; redraws
found so far are still merged. A regression test feeds 100,000 replacements and
100,000 non-matching glyphs at one place.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: git-monchiant <4667243+git-monchiant@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* test: give faux-redaction audit its own workdir (parallel-test flake)
* hayro: ColorReplacement type and luminance mapping (reading colours, #897)
* fmt: engine dates (upstream Greek localization miss)
* hayro: apply ColorReplacement to solid text and line-art paints (#897)
* render: ReadingColors view setting flows into RenderSettings (#897)
* engine: per-document reading colours state (#897)
* automation: view_reading_colors tool; page_render honours it (#897)
* ui: Reading colours preference (Replace Document Colors, #897)
* ui: seed reading colours on revision and signed-version opens (#897)
* parity: a11y.replace-page-colors shipped (#897)
* render: map stencil-mask paints through reading colours (#897)
An ImageMask stencil's solid paint was baked into the raster bytes raw,
bypassing the reading-colours hook every other solid fill goes through:
under white-on-black, CCITT scans, JBIG2 text and stencil logos drew
black ink on the black mapped background and vanished. Route it through
replaced(paint, false) — stencils are line art in disguise — and cover
it with an ImageMask fixture test sampling mapped ink on a mapped
background.
* render: print previews render with reading colours off (#897)
Print previews went through the document's render pool, whose config now
carries the reading colours: a white-on-black user saw black sheets in
the Print dialog while the actual print stayed uncoloured. The request
tag is the per-request channel that already crosses the render boundary,
so PRINT_TAG now lives in pdfcraft_render and the renderer honours it by
rendering previews with reading colours off — a preview's job is
fidelity to the printout. The main view stays recoloured. Covered on
both render paths (inline PageRenderer and pool worker).
* ui: seed reading colours on created-document opens (#897)
open_created (comment summaries) and open_created_bytes (images, text,
blank pages, clipboard) pushed new DocViews without seeding the reading
colours, unlike every other open path. A created document now starts
with the app-wide preference too; the seed-on-open test drives both
paths.
* docs: reading-colours flag echo and display-fidelity parity notes (#897)
The view_reading_colors description now says a flag that differs from
the mode's preset default makes the echoed mode read custom, and the
a11y.replace-page-colors parity notes record that print previews and
view snapshots render with the document's true colours: reading colours
apply to the reading view only.
* Merge follow-ups: reading colours beside preview renders and the new catalogs
- Renderer::new carries both color_replacement and the new preview flag
- our raster tests construct RenderRequest with preview: false
- the preference dialog grew past the window: scroll the combo into view
and give the fixture a valid xref so no repair banner covers the dialog
- checkboxes use the crate's square-box widget (clippy disallows egui's)
- Kazakh and Turkish catalogs gain the ten reading-colours strings
* Reading colours: record the hayro patch in vendor/README.md; add missing translations
vendor/README.md gains hayro patch (13): `RenderSettings::color_replacement` /
`ColorReplacement`, the `render_pdf` signature change and the removed
`[dev-dependencies.log]`, with its tests. Adds the ten reading-colours labels
to el, es, pl, te and zh-hant.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix: recognize paragraph drafts as unsaved work (Issue #811)
* fix(color): honour the /Lab colour space /WhitePoint (#1203)
The old path fed every Lab value to the bundled ICC transform as if the
space's white point were D50, so a /Lab space with any other /WhitePoint
(e.g. D65 [0.9505 1 1.089]) rendered off by several levels per channel
and a D65 blue came out 45 levels too red.
Map L*a*b* to XYZ in the space's own white coordinates (ISO 32000-2
§8.6.5.4), Bradford-adapt it to D65 and convert to sRGB directly, as
Poppler does. Missing or degenerate white points fall back to a flat
white and hostile components are clamped without panicking.
Add a render regression that checks the five reported flat patches of
the D65 white-point fixture against their CIE sRGB values.
* Lab: map normalised Indexed palette components onto L*/Range; drop unused LAB.icc
An /Indexed palette over Lab passes its entries as byte/255 (manual_scale),
which the new Lab conversion read as raw L*a*b* and drew near black. Map
them onto L* 0..100 and the a*/b* /Range, with a regression test. Record the
patch as hayro-interpret (25) in vendor/README.md, and remove LAB.icc (no
longer referenced) with its ATTRIBUTION, NOTICE and assets README entries.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Main gates: rustfmt tests/viewer.rs; a type alias for the refresh-steps array (clippy type_complexity)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(x11): prefer server DPI over panel physical DPI in winit
Without Xft DPI, use the X server screen dimensions before the output physical density. Xorg can run a dense panel at 96 DPI; EDID sizes alone make only winit apps appear zoomed.
Supersedes #908: preserves Xft.dpi priority, configured HiDPI screens, and WINIT_X11_SCALE_FACTOR override (including randr). Falls back to RandR output density when server dimensions are missing.
Adds default_dpi_factor() with 4 unit tests plus vendored-copy guard test in apps/pdfcraft/tests/winit_x11_dpi.rs.
* vendor/README: record the winit X11 DPI patch as (11)
---------
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Heavy pages: cull off-tile images, cache decodes, preview backdrops, calmer queues
A large CAD sheet (page 24 of the benchmark file: ~5050 pt wide, 2104
images, 9 MB of content) cost a full page interpret per tile: 4.3 s for a
1024 px tile at 1000%, 10.5 s for the whole page, and a storm of renders on
every pan frame. Tiles now cost 0.6 s, the sheet is covered by a handful of
large tiles, and gestures no longer queue work faster than workers finish it.
- Vendored hayro patch (10): skip images whose stream-size box misses the
canvas before decoding them, so a tile no longer decodes, resamples and
rasterizes every image on its page. Pixel-identical; the skip regression
test now covers images too.
- Vendored hayro-interpret patch (23): decoded image pixels shared across
renders through the interpreter cache, keyed by stream and decode target
size within a 64 MiB budget.
- Preview renders (hayro patch 11, RenderRequest::preview through to the
device): backdrops, thumbnails, grid cells and signature-drag backgrounds
skip images under 16 device px. Tiles, whole pages, print and export always
draw everything. Previews may differ by those pixels; finals don't.
- Viewer: zoom/pan settle holds apply only to pages whose raster proved slow
(slow_pages, 500 ms), so fast pages render with no added latency; tile sides
grow with zoom and page size (giant sheets need ~8 tiles, not ~28), queued
center-first and keyed by exact coordinates; Edit-text outlines capped.
- vendor/README.md records the new patches with their tests.
Co-Authored-By: Muse Spark <noreply@meta.com>
* Decoded-image cache: key every decode input, survive a poisoned lock
The key was the dictionary's bytes and the target size, so two images with byte-identical
dictionaries (uncompressed, one size) shared pixels, and an image drawn under a transfer
function reused pixels decoded without it. DecodedImageKey now holds the dictionary's place
in the document (its data follows it), its bytes and object, mask/stencil/raster output,
interpolation, the target size and preview or final quality; images under a transfer
function and inline images (named colour spaces, dictionaries in content streams) are not
cached, and an image over the budget is never kept. A poisoned lock decodes uncached
instead of panicking. Soft masks: the per-renderer cache can't carry a preview mask into a
final render (documented). Regression tests for identical dictionaries, preview then final
through one cache, and transfer functions.
---------
Co-authored-by: Muse Spark <noreply@meta.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Fix#624: preserve hard edges for non-interpolated images
Select nearest-neighbour for Hayro's image pre-minification when /Interpolate is absent or false, retaining Catmull-Rom for true. Add a synthetic regression covering default, false, and true behaviour and record the vendored patch.
Co-authored-by: Codex <noreply@openai.com>
* M2: test non-interpolated minification with exact colours
The #624 regression used alternating black and white columns. Nearest 8→4 sampling reads only white columns from that pattern, so an image skipped entirely still passed, and the check only asked for near-black or near-white pixels.
Replace it with 2×2 colour blocks whose exact device output is asserted, with /Interpolate omitted equal to false, and /Interpolate true still blending. Add exact cases for gray, a non-integer 10→6 minification, and a matched soft mask. All pass on this head. On origin/main the same block image renders with blended mid-tones (for example 12eb12ff where pure green is expected), so these tests fail without the #624 change.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
* M2: plan nearest-neighbour pre-resize without filter limits
Non-interpolated images are pre-resized with nearest-neighbour, which copies source pixels and needs no filter tables, crossed intermediate or scratch. The preflight still applied the Catmull-Rom checks to it: a source wider than 2^20 pixels, or a filter plan over 128 MiB, was refused, and once such a source was wider than the backend's u16 limit the image was skipped.
resampling_byte_len and image_resampling_size take the image's interpolation flag. Nearest keeps the source and target buffer bounds and refuses any scratch it reports. Filtered images keep their limits unchanged, and the existing image_resampling_size tests pass true.
Measured: a 1.5M-pixel-wide RGB row drawn at 750×4 device pixels renders 0 non-white pixels on origin/main (the image is skipped) and all 3000 pixels with this commit. The 1.1M-pixel gray case is covered by wide_non_interpolated_sources_draw_after_shrinking.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
* M2: keep non-interpolated Type 3 glyph images nearest
Minified images inside Type 3 glyphs were drawn with ImageQuality::High, a bicubic sampler, whatever their /Interpolate. A non-interpolated image had already been pre-resized with nearest-neighbour, and bicubic sampling of that result blended it again. High is now used only for interpolated images, as elsewhere.
Before this commit the new test saw an interior glyph pixel of [28, 28, 27] (not a palette colour) for a 4.6 × 2.3 px glyph of a palette image; after it, every interior pixel is a palette colour. /Interpolate true still blends.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
* M2: nearest-sample masked minified images straight into premultiplied RGBA
A non-interpolated image with a matching soft mask used to build a full-resolution RGBA copy, nearest-resize it, then premultiply the target. For a minified image most of that copy was discarded.
nearest_premultiplied_rgba reads colour and mask from the same source pixel, selected with PicScale's nearest index (same fixed-point centre sampling as ResampleNearestPlan) and premultiplies only the target pixels. It falls back to the previous path if its target buffer cannot be allocated. Interpolated images, unmasked images and magnification are unchanged.
New test: a ten-column gray ramp with a soft mask drawn six pixels wide must pick the same columns for colour and mask.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
* M2: record the #624 image patch in vendor/README.md
The hayro row listed the #624 regression test but not the patch itself. Add patch item (10) covering the nearest pre-resize and its planning, the Type 3 bicubic restriction and the premultiplied soft-mask path, and list every regression test that covers them.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Claude Haiku 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
InDesign writes fades as luminosity soft masks whose group is one
extended axial `sh` over the whole page. hayro 0.7 sampled every such
shading pixel by pixel through the shading function, so a report page
with 40 of them took 10-15 s at 144 dpi.
Backport hayro 0.8.0's `gradient` module: an axial or radial shading
that a native gradient draws exactly becomes sRGB stops (within 0.01
per channel, a stop at every stitching bound), painted as a vello_cpu
gradient. Upstream's half-pixel shift is left out: it suits vello_cpu
0.3 but moves 0.0.8's gradients off pixel centres.
Page 4 of the reported file: 10.8 s -> 1.0 s.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* M2: stream page content past the 256 MiB cap (item 1 of storytold/pdfcraft#623)
Page content was decoded whole, and the decode stopped at
MAX_DECODED_STREAM (256 MiB), so the last ~10 MB of the #623 sample
was dropped without a warning. Top-level page content is now read a
window at a time, so the decoded content is never all in memory:
- the content tokenizer reads windows and holds back an instruction
that runs past one; Reader records reads that touch the end of the
data, including reads made through clones, so an instruction is
only decided once it has been seen whole;
- the Flate decoder is resumable and keeps its own 32 KiB history, so
it pauses at any output size and back-references do not depend on
what the caller has kept;
- page content may decode to max(256 MiB, 64 x its compressed size),
counting each distinct stream once; content that reaches that
budget is cut and the page reports ContentTruncated, as it did not
before;
- forms, tiling patterns, Type 3 glyph procedures and generic streams
keep their existing budgets and caps.
Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
* M2: Avoid duplicate page content raw reads
* M2: Document page content byte reuse
---------
Co-authored-by: Claude Haiku 5.5 <noreply@anthropic.com>
* M13: report unavailable symbolic-link fixtures in hostile-argument tests
Only include the planted link escape when symlink creation succeeds. On
non-elevated Windows the fixture can fail with OS error 1314; exercising
its nonexistent path otherwise produces false confinement leaks. Keep
all remaining traversal and absolute-path cases and report the skip.
Co-authored-by: Codex (GPT-6) <noreply@openai.com>
* M3: return renderer initialization errors to the OpenGL startup fallback
Guard renderer creation with the existing scoped GPU-error handler and
return CreateRenderer instead of invoking wgpu's default panic handler.
This lets Auto startup reach its existing OpenGL retry before app creation.
Add a real-device invalid/valid pipeline regression and Windows startup
reproduction notes. The host-specific DX12 shader compiler error itself
is not fixed; explicit renderer choices and the non-winit path are unchanged.
Co-authored-by: Codex (GPT-6) <noreply@openai.com>
* Leave the symlink-test fix to #949, which covers the same ground
---------
Co-authored-by: Codex (GPT-6) <noreply@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Revisions 2–4 of the standard security handler take the password in
PDFDocEncoding (ISO 32000-2 §7.6.4.3.2, Algorithm 2). PdfCraft authenticates
an encrypted file twice, and neither did that:
- pdfcraft-crypt took Latin-1 and dropped every other character, so "ı"
(0x9A) and "€" (0xA0) vanished and "ılık€" became "lk";
- the renderer (vendored hayro-syntax, "TODO: Convert to PDFDocEncoding")
used the UTF-8 bytes, so any non-ASCII password failed, even "Çok güzel",
which Latin-1 and PDFDocEncoding write alike;
- the engine handed the renderer an owner-recovered user password as Latin-1
characters, which the renderer then read as UTF-8 again.
Files that qpdf or Acrobat protected with such passwords could not be
opened with the right password.
- pdfcraft-crypt encodes R2–R4 passwords in PDFDocEncoding and, for a
password with characters PDFDocEncoding lacks (ğ, ş, İ), also tries its
UTF-8 bytes, which qpdf writes for those. The PDFDocEncoding table moves
here from pdfcraft-cos (which depends on this crate) and cos uses it.
- hayro-syntax (vendored patch 14) tries an R2–R4 password as given, then
in PDFDocEncoding.
- The engine passes a recovered user password to the renderer as UTF-8 when
its bytes are UTF-8 and as PDFDocEncoding otherwise, so the renderer gets
exactly those bytes back.
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Skipping a direct array or dictionary, and the lenient repair of a
dictionary that has an object where a key belongs, recursed once per
level with no bound, so deep nesting in a trailer entry or a content
stream overflowed the stack. Count the depth on the Reader and fail
beyond 128 levels like any other malformed object.
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
A compressed object's container was resolved with a fresh cycle set and
without checking that it is an ordinary object, and indirect /Length
chains had no depth limit, so self-containing or chained lookups
overflowed the stack. Pass the caller's set on, require an in-file
container and cap the chain at 16 lookups. Object-stream members were
each parsed to the end of the decoded stream, so repeated or nested
offsets kept many copies of one large object: parse each member within
its own extent and skip repeated offsets.
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
* M2: pages with /UserUnit are shown at their size in points, as in Acrobat
/UserUnit gives the size of a page's user-space unit in points; PdfCraft
ignored it everywhere except the measuring tools, so a /UserUnit 2 page
was shown, extracted, printed and edited at half its size, and the
measuring tools disagreed with the rest of the app.
The value is now read as Acrobat reads it (the page's own entry, a number
from 1 up, at most 75,000; anything else counts as 1) in two places:
- vendored hayro-syntax Page (patch 13): render_dimensions and
initial_transform give the page in points, so rendering, tiles,
PageInfo and the text layer follow; a page without it keeps exactly
the transform it had.
- pdfcraft-model Page: user_unit(), and display_size/view_matrix in
points, so print, header/footer, watermark, background, added content
and measuring agree with the screen. The measure crate drops its own
arithmetic. Added items record the /UserUnit they were written for;
items without the record (written before) read back converted.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* M2: an added item's recorded /UserUnit is clamped like the page's
An infinite or huge recorded /UserUnit made the scale factor 0 and collapsed the item.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* macOS: IME text committed after the marked text is cleared isn't lost (winit#4650)
winit 0.30.13 reports insertText as Ime::Commit only while marked text is
left, and after any commit it swallows the next key. Real paths where that
loses text:
- input methods that end a composition with unmarkText, then insertText
(rust-windowing/winit#3925): the committed text was dropped;
- Apple Korean 2-Set, for the key typed after a syllable: on 한 then 5 it
calls setMarkedText("한"), insertText("한"), setMarkedText(""),
insertText("5"), so the 5 (or ;, or any key that ends the syllable) was
dropped;
- a commit made outside a key press (a candidate clicked with the mouse,
dictation): no key follows to reset the state, so the next key (for
example Backspace) was swallowed.
vendor/winit now commits when a composition was started (pending_commit,
set in setMarkedText, cleared by the commit and when IME is disallowed);
other insertText calls still reach the app as KeyboardInput. A commit
outside keyDown clears the marked text and returns to the ground state.
validAttributesForMarkedText lists the underline and marked-clause
attributes, which some input methods need before they compose.
This is rust-windowing/winit#4650, the 0.30 backport of #4651 (merged on
winit master), plus the part of #4748 that fixes#4651's out-of-key-press
case (#4748's other parts are not included). One correction: upstream
spells the underline attribute "NSUnderlineStyle", but AppKit's
NSUnderlineStyleAttributeName is "NSUnderline" (printed on macOS 14), so
the AppKit value is used.
Checked with the real input methods on macOS 14.7.1 (a scratch winit app fed
key events through AppKit): with the patch 한+5 and 한+; commit the 5 and ;,
unmarkText-then-insertText commits, and Backspace after a commit outside a
key press is delivered; ABC typing, held keys, Japanese (日本, あ), Cangjie
(日) and Korean composition give the same events as before. A test fails if
a re-vendored winit loses the patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* macOS: the first Apple Korean key in a fresh PdfCraft composes (winit#4744)
On the first key after PdfCraft starts, Apple Korean 2-Set can answer
interpretKeyEvents with insertText and the raw compatibility jamo instead of
setMarkedText. winit sends that key to the app as an ordinary key, and the
next keys compose without it: g k s Space typed ㅎㅏㄴ instead of 한 (Search,
form fields, comments; rust-windowing/winit#3095).
vendor/winit now ports rust-windowing/winit#4744: keyDown interprets that
one key a second time, synchronously, before any KeyboardInput is sent, and
only when every condition holds: Apple Korean, IME allowed and not yet
composing, no pending commit or marked text, no Command/Control/Option, not
a repeat, the key window's first responder, the only callback is one
insertText of that same single jamo, and focus and input source unchanged
afterwards. The decision is ImeStartup (ime_startup.rs, copied with its
seven tests); no timer, no key queue, no text rewriting. It sits on top of
the previous macOS IME change (pending_commit, in_key_down); the retry runs
inside in_key_down. Command maps to ModifiersState::SUPER in winit 0.30
(META on winit master).
Checked with the real input method on macOS 14.7.1 (scratch winit app, fresh
process each time): without the patch g k s Space gave ㅎ as a key, then
ㅏ and ㄴ committed separately, 3 runs out of 3; with it, ㅎ → 하 → 한 → 한
committed, 3 out of 3. All other traces (Japanese, Cangjie, ABC, held
keys, later Korean keys) are unchanged. ImeStartup's tests pass with
rustc --test. A test fails if a re-vendored winit loses the patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* macOS: IME text committed after the marked text is cleared isn't lost (winit#4650)
winit 0.30.13 reports insertText as Ime::Commit only while marked text is
left, and after any commit it swallows the next key. Real paths where that
loses text:
- input methods that end a composition with unmarkText, then insertText
(rust-windowing/winit#3925): the committed text was dropped;
- Apple Korean 2-Set, for the key typed after a syllable: on 한 then 5 it
calls setMarkedText("한"), insertText("한"), setMarkedText(""),
insertText("5"), so the 5 (or ;, or any key that ends the syllable) was
dropped;
- a commit made outside a key press (a candidate clicked with the mouse,
dictation): no key follows to reset the state, so the next key (for
example Backspace) was swallowed.
vendor/winit now commits when a composition was started (pending_commit,
set in setMarkedText, cleared by the commit and when IME is disallowed);
other insertText calls still reach the app as KeyboardInput. A commit
outside keyDown clears the marked text and returns to the ground state.
validAttributesForMarkedText lists the underline and marked-clause
attributes, which some input methods need before they compose.
This is rust-windowing/winit#4650, the 0.30 backport of #4651 (merged on
winit master), plus the part of #4748 that fixes#4651's out-of-key-press
case (#4748's other parts are not included). One correction: upstream
spells the underline attribute "NSUnderlineStyle", but AppKit's
NSUnderlineStyleAttributeName is "NSUnderline" (printed on macOS 14), so
the AppKit value is used.
Checked with the real input methods on macOS 14.7.1 (a scratch winit app fed
key events through AppKit): with the patch 한+5 and 한+; commit the 5 and ;,
unmarkText-then-insertText commits, and Backspace after a commit outside a
key press is delivered; ABC typing, held keys, Japanese (日本, あ), Cangjie
(日) and Korean composition give the same events as before. A test fails if
a re-vendored winit loses the patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* macOS: Space in Apple Korean with no syllable in progress types one space (winit#4478)
With Apple Korean 2-Set on macOS 14, Space with no syllable in progress
(a second space, or a space after a digit or punctuation) makes AppKit
call setMarkedText(" "), insertText(" "), then setMarkedText(""). winit
reports the insertText as Ime::Commit(" "), but the trailing empty
setMarkedText returns its state to the ground state, so keyDown also sends
the key as KeyboardInput with text " ". egui inserts both, and PdfCraft's
text fields got two spaces (the double space of alacritty/alacritty#8079).
vendor/winit now records that the key being handled committed IME text and
then doesn't also send it as KeyboardInput, unless AppKit forwards it as a
command through doCommandBySelector (Enter after a syllable still commits
the syllable and sends Enter). keyDown consumes the record on every key,
so keys typed after IME is disallowed are unaffected.
rust-windowing/winit#4478 aims at the same double space, but its changes
(clearing the marked text on commit, forwarding a later insertText, and
dropping doCommandBySelector's guard) don't reach this call order, and the
key loss it also fixes is already fixed by the previous change
(pending_commit). This is the narrow form for the call order observed
here; it should go upstream with #4478.
Checked with the real input methods on macOS 14.7.1 (scratch winit app, on top
of the previous change): the duplicate KeyboardInput Space is gone and
nothing else changes (Korean syllables, 한+Enter, 한+5, Japanese, Cangjie,
ABC, held keys, keys after IME is disallowed). A test fails if a
re-vendored winit loses the patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Keep #678's emoji-picker branch: view.rs is main's plus this PR's Space fix (54b3d3c)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* macOS: IME text committed after the marked text is cleared isn't lost (winit#4650)
winit 0.30.13 reports insertText as Ime::Commit only while marked text is
left, and after any commit it swallows the next key. Real paths where that
loses text:
- input methods that end a composition with unmarkText, then insertText
(rust-windowing/winit#3925): the committed text was dropped;
- Apple Korean 2-Set, for the key typed after a syllable: on 한 then 5 it
calls setMarkedText("한"), insertText("한"), setMarkedText(""),
insertText("5"), so the 5 (or ;, or any key that ends the syllable) was
dropped;
- a commit made outside a key press (a candidate clicked with the mouse,
dictation): no key follows to reset the state, so the next key (for
example Backspace) was swallowed.
vendor/winit now commits when a composition was started (pending_commit,
set in setMarkedText, cleared by the commit and when IME is disallowed);
other insertText calls still reach the app as KeyboardInput. A commit
outside keyDown clears the marked text and returns to the ground state.
validAttributesForMarkedText lists the underline and marked-clause
attributes, which some input methods need before they compose.
This is rust-windowing/winit#4650, the 0.30 backport of #4651 (merged on
winit master), plus the part of #4748 that fixes#4651's out-of-key-press
case (#4748's other parts are not included). One correction: upstream
spells the underline attribute "NSUnderlineStyle", but AppKit's
NSUnderlineStyleAttributeName is "NSUnderline" (printed on macOS 14), so
the AppKit value is used.
Checked with the real input methods on macOS 14.7.1 (a scratch winit app fed
key events through AppKit): with the patch 한+5 and 한+; commit the 5 and ;,
unmarkText-then-insertText commits, and Backspace after a commit outside a
key press is delivered; ABC typing, held keys, Japanese (日本, あ), Cangjie
(日) and Korean composition give the same events as before. A test fails if
a re-vendored winit loses the patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* macOS: emoji from the emoji picker reach text fields (winit#4749)
The macOS emoji picker (Fn/Globe-E, Control-Command-Space) inserts its text
with insertText:replacementRange: outside any key press and without a
composition. winit 0.30.13 reports insertText only as part of a key press or
a composition, so an emoji picked into Search, a form field or a comment
was dropped.
vendor/winit now reports text inserted outside a key press, with IME
allowed and no composition, as Ime::Commit, as rust-windowing/winit#4749
(the 0.30 backport of #4748) does. Builds on the pending_commit and
in_key_down state of the previous macOS IME change.
#4748's selectedRange change, which lets dictation start, is left out:
checked on macOS 14.7.1, a valid selection switches on press-and-hold for
every held key in a winit window, so holding e kept sending e key events
under an accent menu, and the accent picked with 1 arrived as a separate
commit after "eeee…1". Without it, held keys repeat as before.
Checked with a scratch winit app on macOS 14.7.1: an insertText outside a key
press now arrives as Ime::Commit; held keys, ABC, Japanese, Cangjie and
Korean typing give the same events as before. A test fails if a re-vendored
winit loses the patch or gains the selectedRange change.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
egui disallows IME when a text field loses focus, for example when the user
clicks the page while composing. winit 0.30.13 then clears its own marked
text but never tells the input method, which keeps the composition and
continues it once IME is allowed again. Checked with the real input methods
on macOS 14.7.1: Korean 하, IME off and on, then r k gave 학, then 하 and 가
(하 committed a second time if the app already kept it); Japanese か, IME
off and on, then s a gave 傘.
vendor/winit now calls NSTextInputContext::discardMarkedText when IME is
disallowed with marked text present, before its existing reset, as
rust-windowing/winit#4745 does on winit master. With the patch the same
sequences give 가 and さ. AppKit may call unmarkText from
discardMarkedText while IME is still enabled, so its Preedit("") comes
before Disabled; the RefCell borrow in the check ends before the call.
Everything else in the scratch app's traces (ABC, Japanese, Cangjie,
Korean) is unchanged. A test fails if a re-vendored winit loses the patch.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
winit 0.30.13 reports insertText as Ime::Commit only while marked text is
left, and after any commit it swallows the next key. Real paths where that
loses text:
- input methods that end a composition with unmarkText, then insertText
(rust-windowing/winit#3925): the committed text was dropped;
- Apple Korean 2-Set, for the key typed after a syllable: on 한 then 5 it
calls setMarkedText("한"), insertText("한"), setMarkedText(""),
insertText("5"), so the 5 (or ;, or any key that ends the syllable) was
dropped;
- a commit made outside a key press (a candidate clicked with the mouse,
dictation): no key follows to reset the state, so the next key (for
example Backspace) was swallowed.
vendor/winit now commits when a composition was started (pending_commit,
set in setMarkedText, cleared by the commit and when IME is disallowed);
other insertText calls still reach the app as KeyboardInput. A commit
outside keyDown clears the marked text and returns to the ground state.
validAttributesForMarkedText lists the underline and marked-clause
attributes, which some input methods need before they compose.
This is rust-windowing/winit#4650, the 0.30 backport of #4651 (merged on
winit master), plus the part of #4748 that fixes#4651's out-of-key-press
case (#4748's other parts are not included). One correction: upstream
spells the underline attribute "NSUnderlineStyle", but AppKit's
NSUnderlineStyleAttributeName is "NSUnderline" (printed on macOS 14), so
the AppKit value is used.
Checked with the real input methods on macOS 14.7.1 (a scratch winit app fed
key events through AppKit): with the patch 한+5 and 한+; commit the 5 and ;,
unmarkText-then-insertText commits, and Backspace after a commit outside a
key press is delivered; ABC typing, held keys, Japanese (日本, あ), Cangjie
(日) and Korean composition give the same events as before. A test fails if
a re-vendored winit loses the patch.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
egui calls Window::set_ime_allowed(true) when a text field gains focus and
set_ime_allowed(false) when it loses it. On X11, winit 0.30.13 handles each
change by destroying the window's X input context and creating a new one, and
leaves focusing the new context to the next FocusIn. PdfCraft's window keeps
the focus while the user moves between Search, form fields and the canvas, so
no FocusIn comes and the new context stays unfocused: the input method
(IBus, fcitx5) doesn't follow the field until the window loses and regains
focus.
vendor/winit's Ime::set_ime_allowed now returns create_context's result, and
the event processor, after draining the queued IME requests, focuses the last
context replaced for the active window, once, as
rust-windowing/winit#4727 does on winit master. No public API changes. A
test fails if a re-vendored winit loses the patch.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
eframe creates PdfCraft's window hidden, restores the saved state
(app.ron, persist_window) and shows the window after the first frame is
drawn. For a window closed maximized, winit 0.30.13 applies the maximized
state with ShowWindow(SW_MAXIMIZE), which shows a hidden window: PdfCraft
appeared unpainted (white), was hidden again, then shown with SW_SHOW and
SW_MAXIMIZE once more. rust-windowing/winit#4586 also shows the window's
focus state going wrong on this path.
vendor/winit now skips the maximize while the window is hidden and shows a
hidden window that is maximized with one SW_MAXIMIZE, as
rust-windowing/winit#4587 (approved upstream) does on winit master. That is
the later, focus-safe form of #4575; #4575's other change (re-sending
SW_MINIMIZE on every flag change while minimized) is not taken. Transitions
between two visible states make the same ShowWindow calls as before. A test
fails if a re-vendored winit loses the patch.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
When a composition has no target clause, winit 0.30.13 reads the cursor with
ImmGetCompositionStringW(GCS_CURSORPOS) without checking that the
WM_IME_COMPOSITION lParam has GCS_CURSORPOS set. The Microsoft Korean IME
never sets it (lParam 0x6018), so the read gives 0 and winit reports
Ime::Preedit("한", Some((0, 0))). egui 0.36.2 recognises that exact report on
Windows and drops the range, so PdfCraft draws no caret while Korean is being
composed.
vendor/winit now reads GCS_CURSORPOS only when the flag says the IME set it
and otherwise puts the cursor at the end of the composition, as
rust-windowing/winit#4746 does on winit master (Firefox and SDL do the same).
IMEs that report a cursor or a target clause (Japanese, Chinese) are
unchanged. A test fails if a re-vendored winit loses the patch.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
winit 0.30.13 destroys a seat's zwp_text_input_v3 in
SeatHandler::remove_capability, whichever capability went away (a touch
screen or tablet unplugged, a remote-desktop session ending), and only
new_capability recreates it. A seat that loses a capability without gaining
one leaves PdfCraft without a text input: IME (fcitx5, IBus) stops working
until PdfCraft is restarted.
The text input belongs to the seat, so vendor/winit now destroys it in
remove_seat only, as rust-windowing/winit#4747 does on winit master. A test
fails if a re-vendored winit loses the patch.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* M3: harden render and CLI failure surfaces
Make content truncation observable, reject unknown CLI options, expose deterministic render work counters, and scope render failures to exact requests. Refuse silently reduced whole-page export rasters while preserving tiled rendering, and add heavy-vector and CLI regression coverage.
Co-authored-by: Codex <noreply@openai.com>
* Render: remember a failed render by page and kind again
A key with scale and tag made every zoom level and tile a new key: a
pathological page tied up a worker for the watchdog limit per request
(exhausting worker replacements) and was re-parsed each time. The
repeated-parsing check asks for a new generation and scale again.
* Export: pages too large at the chosen dpi export at the most the renderer allows, and say so
#669 made every Exporter refuse oversized whole-page renders, which
reached File > Export to image, doc_export_* and OCR: at 600 dpi any page
side over 13.65 in (Legal, A3, Tabloid) failed instead of exporting. The
exporter clamps again; Exporter::dpi_used reports the resolution a page
is drawn at, which the Export dialog's message and doc_export_png's
lower_dpi list mention. A source with reject_oversize still refuses
(CLI render keeps its strict config).
* doc_export_images: describe lower_dpi
* Render: ContentTruncated is reported once per page; the hayro-interpret patch is recorded
Nested forms, Type 3 glyphs and patterns are interpreted again and each
reported the exhausted budget; pdfcraft-render now keeps one. The warning
is part of hayro-interpret patch (18) (the page content budget): marked
in the code and described in vendor/README.md.
* CLI render: print render warnings (skipped content) to stderr
* CLI: the option check refuses unknown options only; commands report missing values themselves
text's own --page parsing (and its test on main) says what a page value
must be; the generic "needs a value" pre-empted it.
---------
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
A tile of a large page renders the whole page with an offset, and hayro
stroked and filled every path on it before vello discarded what fell
outside the tile. Vendored hayro patch (9), `may_paint_canvas`, skips a
fill or stroke whose box misses the canvas: the box of the path's points
and control points (a curve stays inside their hull), grown by the
stroke's reach (half the width times the larger of the miter limit and
sqrt 2), by 0.5 user units for vello's and kurbo's approximation
tolerances, and by 2 device pixels for antialiasing. Empty paths and
non-finite bounds or miter limits are always drawn.
Renders are byte-identical with and without skipping
(`skipping_paths_off_the_canvas_changes_no_pixel`; the hidden per-thread
`set_skip_offscreen_paths` exists for that test). On #623's sample at
scale 2, a 1024 px tile takes about 38% less CPU time; whole pages are
unchanged.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Adobe-Japan1-UCS2 maps the JIS X 0208 forms of the kanji that
JIS X 0213:2004 redrew to a variation sequence (CID 1247 is 噂 U+E0100).
hayro only looked up single characters in a substitute font, so a
non-embedded Japanese font drew 噂, 逢, 溢, 祇, 箸 and over 200 other
common kanji as unrelated glyphs. A character with a variation selector,
from Adobe-Japan1-UCS2 or a /ToUnicode, now draws the substitute's glyph
for that form when its cmap has one, else its glyph for the character.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
A Type 0 font that isn't embedded is drawn with a substitute (a craft-fonts
Japanese face for Adobe-Japan1). When hayro couldn't find a CID's character
in the substitute by Unicode, it took the CID for a glyph index of the
substitute, which numbers its glyphs its own way: 𠮷 (CID 13706) drew as
"Ｑ" in BIZ UDPGothic, ㈠ and the squared katakana words as unrelated
kanji. A CID of an Adobe character collection that the substitute can't
find now draws .notdef, unless the substitute is a CID-keyed CFF font,
which is looked up by CID as before.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Windows: vendor egui-wgpu 0.36.2 as published (#577)
The crate as released on crates.io, plus egui's LICENSE-APACHE and
LICENSE-MIT from the 0.36.2 tag (the published crate leaves them out).
Not used yet: the next commit patches it and adds it to [patch.crates-io].
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Windows: a window larger than the GPU's texture limit can't close PdfCraft (#577)
egui-wgpu 0.36.2 configures a window's surface at the window's own size.
Beyond the device's max_texture_dimension_2d, Surface::configure fails
validation and wgpu's default error handler panics. eframe configures the
surface before the app is created, so no app code can intercept it
(emilk/egui#8361). #578 raised the device's limit to the adapter's; this
removes the crash for windows larger than even that.
vendor/egui-wgpu: `winit::surface_fit` shrinks both sides by one factor so
neither exceeds the limit (the longer side lands exactly on it), and egui
draws at that factor into the surface. On DX12, PdfCraft's Windows default,
the swapchain is stretched over the window (DXGI_SCALING_STRETCH), so the
window stays complete and lined up with the pointer, only softer; wgpu's GL
backend copies it unscaled into a corner instead. It applies wherever a
surface is sized: install, resize (start-up restore, maximise, DPI change,
fullscreen) and a recreated surface, which starts again from the window's
own size. Depth and MSAA textures follow. A surface logs a warning when
it becomes fitted.
Why only an Outlook launch hit it (read from the code, not observed):
winit creates the window at CW_USEDEFAULT, so Windows places it, and a
launcher that hands ShellExecute its own window gets new windows on its
monitor. winit then converts the restored logical size at that monitor's
scale: 3440 points became 8600 pixels on a 250% panel. eframe's clamp
compares points against the largest monitor, so it let 3440 through.
Tests: surface_fit with the issue's 8600 x 3423 and edge sizes (zero,
u32::MAX, a zero limit, the longer side exactly on the limit); egui-wgpu
resolves to vendor/egui-wgpu and the fit stays wired into the paths that
size a surface.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Windows: a GPU that can't set up the window falls back to OpenGL (#519)
On a 2015 Intel GPU with no D3D12 driver, wgpu picked its GL adapter
over WARP and Surface::configure failed (GpuWaitTimeout). wgpu's default
error handler panics, and eframe configures the root surface before the
app is created, so the OpenGL retry from #530 never ran.
Vendored egui-wgpu (PdfCraft patch #519): winit::catch_errors runs a
closure inside wgpu error scopes (validation, out of memory, internal)
and returns the error. Painter::add_surface configures each new window's
surface inside it and returns a failure as WgpuError::ConfigureSurface,
so eframe returns Error::Wgpu and main retries with glow. Later
reconfigures are unchanged.
Tests: a real device returns a wgpu validation error from catch_errors
instead of panicking (red with the scopes removed), and add_surface
keeps the wiring (red with it removed). Docs: vendor/README.md patch 2,
docs/development.md "Renderer fallback" and PDFCRAFT_RENDERER, ROADMAP.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* vendor/egui-wgpu: drop the stray .cargo-ok marker
* Tests: the GPU error-scope test skips on a machine without a GPU adapter
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Windows: vendor egui-wgpu 0.36.2 as published (#577)
The crate as released on crates.io, plus egui's LICENSE-APACHE and
LICENSE-MIT from the 0.36.2 tag (the published crate leaves them out).
Not used yet: the next commit patches it and adds it to [patch.crates-io].
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Windows: a window larger than the GPU's texture limit can't close PdfCraft (#577)
egui-wgpu 0.36.2 configures a window's surface at the window's own size.
Beyond the device's max_texture_dimension_2d, Surface::configure fails
validation and wgpu's default error handler panics. eframe configures the
surface before the app is created, so no app code can intercept it
(emilk/egui#8361). #578 raised the device's limit to the adapter's; this
removes the crash for windows larger than even that.
vendor/egui-wgpu: `winit::surface_fit` shrinks both sides by one factor so
neither exceeds the limit (the longer side lands exactly on it), and egui
draws at that factor into the surface. On DX12, PdfCraft's Windows default,
the swapchain is stretched over the window (DXGI_SCALING_STRETCH), so the
window stays complete and lined up with the pointer, only softer; wgpu's GL
backend copies it unscaled into a corner instead. It applies wherever a
surface is sized: install, resize (start-up restore, maximise, DPI change,
fullscreen) and a recreated surface, which starts again from the window's
own size. Depth and MSAA textures follow. A surface logs a warning when
it becomes fitted.
Why only an Outlook launch hit it (read from the code, not observed):
winit creates the window at CW_USEDEFAULT, so Windows places it, and a
launcher that hands ShellExecute its own window gets new windows on its
monitor. winit then converts the restored logical size at that monitor's
scale: 3440 points became 8600 pixels on a 250% panel. eframe's clamp
compares points against the largest monitor, so it let 3440 through.
Tests: surface_fit with the issue's 8600 x 3423 and edge sizes (zero,
u32::MAX, a zero limit, the longer side exactly on the limit); egui-wgpu
resolves to vendor/egui-wgpu and the fit stays wired into the paths that
size a surface.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* vendor/egui-wgpu: drop the stray .cargo-ok marker
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Check source, destination, convolution and filter-plan sizes before allocation. Preserve original image geometry when a resize exceeds the policy, while allowing valid wide sources to shrink.
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Limit function construction and calculator work, detect path cycles, and handle input-derived integer and stack errors through the existing optional failure path.
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* M14: bound large-PDF memory use (#307)
Inspect large documents lazily, share parser/resource metadata, bound full-save
and raster retention, and cache signature candidate discovery safely.
Preserve upstream rendering deduplication, pixel alignment, signature previews,
and organizer insertion behavior, with regression coverage for their interaction.
Validated with 1,336 workspace tests, 193 parser tests, strict workspace Clippy,
30 WASM crate checks, 1x/2x GUI protocols, and same-PDF workflow comparisons.
* cos: drop the writer cache-policy probe and its test-only hooks in production paths
* cos: the temporary reader keeps parsed objects until a count or owned-bytes budget
Since #552 a stream read from the file is a view into the input, so clearing the whole parse
cache after every read saved little memory and cost most of the full-save slowdown (every
object parsed twice, indirect /Length objects re-parsed per stream). Clear only past
OBJECTS entries or STREAM_BYTES of data of the cache's own (decrypted streams, strings).
* cos: the writers refuse a document opened with a stream limit (read-only inspection)
* cos: a full save fails rather than writing null when an object reads differently the second time
* render: parsing is not timed by the watchdog; timing tests with wide margins
A slow first parse (a huge or damaged file) made the whole pool fail every page for its
lifetime. Parsing, shared or private, is now its own untimed stage, as on main before the
shared parser; rendering is timed from when parsing ends. The init-timeout tests become one
deterministic test (a 2 s parse against a 10 ms limit, shared and private). Watchdog tests
use limits and injected delays ~10x apart instead of 25-50 ms against real renders. Also:
saturating retired-worker count, main's stop-flag test runs on private parsers (the first
shared timeout is a private retry), set_queue documents that unlisted results are dropped,
and retire() documents the permanent per-worker-parser fallback.
* render: page numbers in the lazy page tree use checked conversion
* render: the lazy inspector is compared with lopdf on named destinations and object streams too
* ui: on-screen pages always render; pages turned back to stay cached; stable thumbnail tags
- Visible pages and tiles are always admitted and uploaded (the viewport bounds them); the
byte and count limits decide only how much prefetch fits. A HiDPI two-up view or a 5K
viewport left pages on 'Rendering...' for good.
- Pages no longer demanded stay cached, nearest the current page first, while the 128 MiB
allowance and the 24-page count have room: paging back no longer re-renders.
- Thumbnail density follows the demand count rounded up to a power of two, so rows scrolling
in and out of an organize grid don't change every thumbnail's tag and re-render it.
- A suspended (background) tab drains its pool's finished results instead of holding them.
- Organize drag-and-drop ends on the primary button only.
- Tests: the stale-result test uses valid pixels and checks nothing was uploaded; the
organizer wheel test runs 40 steps; regressions for the first three fixes.
* vendor: mark the hayro-syntax lock and SegmentList patches and list them as (8) and (9); clippy 1.99 as_chunks in a raster test
* render: clippy 1.99 — contains over iter().any() in a raster test
---------
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* hayro-syntax: fix float mantissa overflow on high-precision coordinates (#199)
jsPDF outputs double-precision floating-point coordinates with 16+ decimal
places in /MediaBox. In hayro-syntax, read_inner accumulated float mantissa
digits into u64 using wrapping arithmetic. Integer mantissas exceeding
u64::MAX wrapped modulo 2^64, severely corrupting page dimensions
(e.g., 3874.96 pt became 185.61 pt, rendering as a warped ribbon).
Accumulation now uses checked arithmetic: trailing fractional digits
exceeding u64 capacity are safely ignored without incrementing decimal_shift.
This preserves full IEEE-754 f64 precision while preventing overflow for
numbers with arbitrary numbers of decimal places.
Includes:
- Unit test in hayro-syntax with reported values and a 50-digit fraction.
- Synthetic PDF regression test in pdfcraft-render verifying inspect().
- Vendored patch (6) documentation in vendor/README.md.
Fixes#199
* Never crash: mesh shadings can't exhaust memory or run for minutes
`cargo xtask fuzz` (with the dash, JBIG2 and decompression fixes in
place) found a page whose Coons patch shading (type 6), a 147-byte
stream with 1-bit coordinates decoded onto +-40000, took the renderer
past 3 GB. hayro-interpret samples mesh shadings (types 4-7) pixel by
pixel over each triangle's device-space bounding box into a map, with
nothing but the u16 cast bounding it: up to 65536 x 65536 entries for
one triangle. A synthetic single patch passed 6 GB in six seconds; ten
thousand triangles that each cover half a 1000 x 1000 page ran for
minutes. Two rounds of review of earlier versions found more: every
patch was cut into 722 triangles (about 150 KB) before anything limited
them; a lattice mesh with `/VerticesPerRow 0` pushed empty rows forever,
and one with 4294967295 read the whole stream into one row; a mesh
painted a thousand times was triangulated a thousand times; and the
first caps cut content without a word.
Vendored hayro-interpret patch (18):
- `ShadingPattern::encode_within(area)`: the vendored hayro renderer
passes the area it will actually draw (the path's bounding box on the
canvas, which it already computes) plus a pixel for texels at a
fractional edge, and triangles are sampled only inside it, so the map
holds at most that area's pixels.
- `TriangleSampler`: at most 8 visits per pixel of that area plus 4 per
triangle in it. A triangle's bounding box is about twice its area and a
sub-pixel triangle's is at most 2 x 2 pixels, so tiling meshes and
small dense ones (a patch in a few pixels is 722 triangles) still
shade completely; overlapping ones stop.
- Coons and tensor-product patches are triangulated one at a time.
- Meshes hold at most `MAX_MESH_PATCHES` (2^16; a 200 x 200 gradient
mesh is 40 thousand) patches and `MAX_MESH_TRIANGLES` (2^20)
triangles, and a lattice needs two (ISO 32000-2 8.7.4.5.6) to 2^19
vertices per row.
- A page samples at most `MAX_PAGE_MESH_TRIANGLES` (2^25; the 200 x 200
mesh is 29 million) mesh triangles in all, reset per page like the
existing paint budget.
- Each of these cuts logs a warning.
`encode()` keeps its signature; with no area it samples within a fixed
2^24 visits.
Synthetic regression tests in render:
- `mesh_shadings_far_off_the_page_render`: a Coons patch and a triangle
mesh decoded onto +-40000 render and the rest of the page draws; a
patch on the page, the part of the huge one that covers it, and a
patch filling a 4 x 4 pt rectangle as a pattern are all shaded.
- `many_overlapping_mesh_triangles_finish`: 10,000 triangles each
covering half of a 1000 x 1000 page finish, and are drawn.
- `mesh_shadings_with_many_patches_or_no_rows_finish`: 40,000 patches
beside the page, and a lattice with `/VerticesPerRow 0`.
Red on main (out of memory, killed at 3.9 GB). Each part is pinned by a
mutation: no visit budget stalls past 60 s; no clip leaves the covering
patch unshaded; no per-triangle allowance leaves the small patch
unshaded; triangulating all patches first runs out of memory (killed at
5.8 GB); no rows check runs out of memory.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Never crash: decompression bombs can't exhaust the renderer's memory
The renderer's parser (vendored hayro-syntax) decoded Flate, LZW and
RunLength streams with no output limit. A 2 MB page whose content
stream inflates to 2 GB, or whose 8 x 8 image does, took `pdfcraft-cli
render` and `text` past 3 GB in under a second, and a few more MB of
file scale it further. On the desktop that is the first-page render of
an emailed PDF, and the allocation-failure abort takes every open
document with it. Our own layers already bound single streams (lopdf's
load-time decoding at 256 MiB, cos at 1 GiB), so `info` stays at a few
MB; the renderer path didn't. Random mutation almost never makes a
high-ratio stream, so `cargo xtask fuzz` doesn't find this; found by
hand while triaging the fuzz run.
Vendored hayro-syntax patch (6):
- `decode_limit` gives each filter an output limit, and Flate (both
flate2 and the fallback inflater), LZW and RunLength never write past
it (the buffer can still grow to under twice it):
`MAX_DECODED_STREAM`, 256 MiB, lopdf's figure; or, for the last filter
of an image whose size, bits per component and components are known,
twice its pixel bytes (a PNG predictor tag byte per byte at most),
capped at 1 GiB. Earlier filters in a chain produce compressed data
and get the general limit.
- `Stream::decoded_within(max)` for callers that budget.
- A page's `/Contents` array shares one `MAX_DECODED_STREAM` (naming one
bomb twenty times decoded 5 GB), reserving exactly what it appends.
Vendored hayro-interpret patch (18):
- The content a page holds at once (its contents, and the forms, tiling
patterns and Type 3 glyph procedures being painted) stays within
`MAX_PAGE_CONTENT`, 512 MiB, reset per page like the existing paint
budget. Forms nest fifty deep and each level kept its content while
the next ran: a form painting itself around a 256 MiB stream would
have held over 12 GB. Each level gives its bytes back when it is
dropped, so content painted again and again keeps drawing, and a
stream without filters (borrowed from the file) costs nothing.
- Type 0 sampled functions read every sample into a 4-byte value and
then a table entry: 256 MiB of 1-bit samples asked for over 8 GB. They
now take at most 2^21 table points and 2^23 values (the largest real
tables, DeviceN tint transforms such as 33^4 points x 3 outputs, are
1.2 million and 3.6 million), counted with checked arithmetic, and
read no further than `/Size` asks. `/Size 0` no longer underflows
building the default `/Encode`.
Three rounds of independent review found the aggregate cases, buffer
doubling past the limit, a chain/predictor shape the first image limit
would have truncated, the Type 0 amplifier, and a first, cumulative
budget that would have dropped legitimately repeated content, and a
first Type 0 cap low enough to refuse real tint transforms; all fixed
here.
Measured in a release build (render; OS peak commit):
- content stream inflating to 2 GB: 389 MB, 0.3 s
- 8 x 8 image inflating to 2 GB: 4 MB, 0.2 s
- /Contents naming it twenty times: 516 MB, 0.3 s
- a form painting itself around it: 644 MB, 0.4 s
Synthetic regression tests in render (flate2 is a new dev-dependency
there, already in the workspace):
- `decompression_bombs_stop_at_the_stream_limit`: 288 MiB bombs through
flate2, the fallback inflater (a stored block with a wrong length
check, which flate2 refuses and the fallback accepts), LZW and
RunLength each decode exactly 256 MiB, and at most 128 bytes as an
8 x 8 image.
- `page_contents_and_nested_forms_share_one_budget`: three /Contents
references decode one stream's worth; a self-painting form finishes.
- `content_painted_again_and_again_keeps_drawing`: a 2 MiB form painted
300 times draws the 300th time.
- `sampled_functions_with_billions_of_samples_are_refused`.
Ten mutations each turn them red: no limit at all; no limit in flate2,
the fallback, LZW, RunLength or for images; an unshared /Contents limit;
no page budget; no refund on drop; no Type 0 sample cap. (Taking a form
out of the budget no longer compiles: its content is the budget guard.)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* M2: normalize JPEG2000 samples without overflowing
Use bounded integer sample scaling for 1-32-bit JPX images and reject empty dimensions before chunking. Cover public decoding and continued page rendering with synthetic regressions.
Co-Authored-By: OpenAI Codex <codex@openai.com>
* Never crash: decompression cut at its limit is logged, not silent; vendor README renumbered after #575
---------
Co-authored-by: James Locke <336218001+jcl-bit@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Streetblock <github@streetblock.de>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Never crash: mesh shadings can't exhaust memory or run for minutes
`cargo xtask fuzz` (with the dash, JBIG2 and decompression fixes in
place) found a page whose Coons patch shading (type 6), a 147-byte
stream with 1-bit coordinates decoded onto +-40000, took the renderer
past 3 GB. hayro-interpret samples mesh shadings (types 4-7) pixel by
pixel over each triangle's device-space bounding box into a map, with
nothing but the u16 cast bounding it: up to 65536 x 65536 entries for
one triangle. A synthetic single patch passed 6 GB in six seconds; ten
thousand triangles that each cover half a 1000 x 1000 page ran for
minutes. Two rounds of review of earlier versions found more: every
patch was cut into 722 triangles (about 150 KB) before anything limited
them; a lattice mesh with `/VerticesPerRow 0` pushed empty rows forever,
and one with 4294967295 read the whole stream into one row; a mesh
painted a thousand times was triangulated a thousand times; and the
first caps cut content without a word.
Vendored hayro-interpret patch (18):
- `ShadingPattern::encode_within(area)`: the vendored hayro renderer
passes the area it will actually draw (the path's bounding box on the
canvas, which it already computes) plus a pixel for texels at a
fractional edge, and triangles are sampled only inside it, so the map
holds at most that area's pixels.
- `TriangleSampler`: at most 8 visits per pixel of that area plus 4 per
triangle in it. A triangle's bounding box is about twice its area and a
sub-pixel triangle's is at most 2 x 2 pixels, so tiling meshes and
small dense ones (a patch in a few pixels is 722 triangles) still
shade completely; overlapping ones stop.
- Coons and tensor-product patches are triangulated one at a time.
- Meshes hold at most `MAX_MESH_PATCHES` (2^16; a 200 x 200 gradient
mesh is 40 thousand) patches and `MAX_MESH_TRIANGLES` (2^20)
triangles, and a lattice needs two (ISO 32000-2 8.7.4.5.6) to 2^19
vertices per row.
- A page samples at most `MAX_PAGE_MESH_TRIANGLES` (2^25; the 200 x 200
mesh is 29 million) mesh triangles in all, reset per page like the
existing paint budget.
- Each of these cuts logs a warning.
`encode()` keeps its signature; with no area it samples within a fixed
2^24 visits.
Synthetic regression tests in render:
- `mesh_shadings_far_off_the_page_render`: a Coons patch and a triangle
mesh decoded onto +-40000 render and the rest of the page draws; a
patch on the page, the part of the huge one that covers it, and a
patch filling a 4 x 4 pt rectangle as a pattern are all shaded.
- `many_overlapping_mesh_triangles_finish`: 10,000 triangles each
covering half of a 1000 x 1000 page finish, and are drawn.
- `mesh_shadings_with_many_patches_or_no_rows_finish`: 40,000 patches
beside the page, and a lattice with `/VerticesPerRow 0`.
Red on main (out of memory, killed at 3.9 GB). Each part is pinned by a
mutation: no visit budget stalls past 60 s; no clip leaves the covering
patch unshaded; no per-triangle allowance leaves the small patch
unshaded; triangulating all patches first runs out of memory (killed at
5.8 GB); no rows check runs out of memory.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Never crash: decompression bombs can't exhaust the renderer's memory
The renderer's parser (vendored hayro-syntax) decoded Flate, LZW and
RunLength streams with no output limit. A 2 MB page whose content
stream inflates to 2 GB, or whose 8 x 8 image does, took `pdfcraft-cli
render` and `text` past 3 GB in under a second, and a few more MB of
file scale it further. On the desktop that is the first-page render of
an emailed PDF, and the allocation-failure abort takes every open
document with it. Our own layers already bound single streams (lopdf's
load-time decoding at 256 MiB, cos at 1 GiB), so `info` stays at a few
MB; the renderer path didn't. Random mutation almost never makes a
high-ratio stream, so `cargo xtask fuzz` doesn't find this; found by
hand while triaging the fuzz run.
Vendored hayro-syntax patch (6):
- `decode_limit` gives each filter an output limit, and Flate (both
flate2 and the fallback inflater), LZW and RunLength never write past
it (the buffer can still grow to under twice it):
`MAX_DECODED_STREAM`, 256 MiB, lopdf's figure; or, for the last filter
of an image whose size, bits per component and components are known,
twice its pixel bytes (a PNG predictor tag byte per byte at most),
capped at 1 GiB. Earlier filters in a chain produce compressed data
and get the general limit.
- `Stream::decoded_within(max)` for callers that budget.
- A page's `/Contents` array shares one `MAX_DECODED_STREAM` (naming one
bomb twenty times decoded 5 GB), reserving exactly what it appends.
Vendored hayro-interpret patch (18):
- The content a page holds at once (its contents, and the forms, tiling
patterns and Type 3 glyph procedures being painted) stays within
`MAX_PAGE_CONTENT`, 512 MiB, reset per page like the existing paint
budget. Forms nest fifty deep and each level kept its content while
the next ran: a form painting itself around a 256 MiB stream would
have held over 12 GB. Each level gives its bytes back when it is
dropped, so content painted again and again keeps drawing, and a
stream without filters (borrowed from the file) costs nothing.
- Type 0 sampled functions read every sample into a 4-byte value and
then a table entry: 256 MiB of 1-bit samples asked for over 8 GB. They
now take at most 2^21 table points and 2^23 values (the largest real
tables, DeviceN tint transforms such as 33^4 points x 3 outputs, are
1.2 million and 3.6 million), counted with checked arithmetic, and
read no further than `/Size` asks. `/Size 0` no longer underflows
building the default `/Encode`.
Three rounds of independent review found the aggregate cases, buffer
doubling past the limit, a chain/predictor shape the first image limit
would have truncated, the Type 0 amplifier, and a first, cumulative
budget that would have dropped legitimately repeated content, and a
first Type 0 cap low enough to refuse real tint transforms; all fixed
here.
Measured in a release build (render; OS peak commit):
- content stream inflating to 2 GB: 389 MB, 0.3 s
- 8 x 8 image inflating to 2 GB: 4 MB, 0.2 s
- /Contents naming it twenty times: 516 MB, 0.3 s
- a form painting itself around it: 644 MB, 0.4 s
Synthetic regression tests in render (flate2 is a new dev-dependency
there, already in the workspace):
- `decompression_bombs_stop_at_the_stream_limit`: 288 MiB bombs through
flate2, the fallback inflater (a stored block with a wrong length
check, which flate2 refuses and the fallback accepts), LZW and
RunLength each decode exactly 256 MiB, and at most 128 bytes as an
8 x 8 image.
- `page_contents_and_nested_forms_share_one_budget`: three /Contents
references decode one stream's worth; a self-painting form finishes.
- `content_painted_again_and_again_keeps_drawing`: a 2 MiB form painted
300 times draws the 300th time.
- `sampled_functions_with_billions_of_samples_are_refused`.
Ten mutations each turn them red: no limit at all; no limit in flate2,
the fallback, LZW, RunLength or for images; an unshared /Contents limit;
no page budget; no refund on drop; no Type 0 sample cap. (Taking a form
out of the budget no longer compiles: its content is the budget guard.)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* M2: normalize JPEG2000 samples without overflowing
Use bounded integer sample scaling for 1-32-bit JPX images and reject empty dimensions before chunking. Cover public decoding and continued page rendering with synthetic regressions.
Co-Authored-By: OpenAI Codex <codex@openai.com>
* Never crash: decompression cut at its limit is logged, not silent; vendor README renumbered after #575
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Streetblock <github@streetblock.de>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Never crash: decompression bombs can't exhaust the renderer's memory
The renderer's parser (vendored hayro-syntax) decoded Flate, LZW and
RunLength streams with no output limit. A 2 MB page whose content
stream inflates to 2 GB, or whose 8 x 8 image does, took `pdfcraft-cli
render` and `text` past 3 GB in under a second, and a few more MB of
file scale it further. On the desktop that is the first-page render of
an emailed PDF, and the allocation-failure abort takes every open
document with it. Our own layers already bound single streams (lopdf's
load-time decoding at 256 MiB, cos at 1 GiB), so `info` stays at a few
MB; the renderer path didn't. Random mutation almost never makes a
high-ratio stream, so `cargo xtask fuzz` doesn't find this; found by
hand while triaging the fuzz run.
Vendored hayro-syntax patch (6):
- `decode_limit` gives each filter an output limit, and Flate (both
flate2 and the fallback inflater), LZW and RunLength never write past
it (the buffer can still grow to under twice it):
`MAX_DECODED_STREAM`, 256 MiB, lopdf's figure; or, for the last filter
of an image whose size, bits per component and components are known,
twice its pixel bytes (a PNG predictor tag byte per byte at most),
capped at 1 GiB. Earlier filters in a chain produce compressed data
and get the general limit.
- `Stream::decoded_within(max)` for callers that budget.
- A page's `/Contents` array shares one `MAX_DECODED_STREAM` (naming one
bomb twenty times decoded 5 GB), reserving exactly what it appends.
Vendored hayro-interpret patch (18):
- The content a page holds at once (its contents, and the forms, tiling
patterns and Type 3 glyph procedures being painted) stays within
`MAX_PAGE_CONTENT`, 512 MiB, reset per page like the existing paint
budget. Forms nest fifty deep and each level kept its content while
the next ran: a form painting itself around a 256 MiB stream would
have held over 12 GB. Each level gives its bytes back when it is
dropped, so content painted again and again keeps drawing, and a
stream without filters (borrowed from the file) costs nothing.
- Type 0 sampled functions read every sample into a 4-byte value and
then a table entry: 256 MiB of 1-bit samples asked for over 8 GB. They
now take at most 2^21 table points and 2^23 values (the largest real
tables, DeviceN tint transforms such as 33^4 points x 3 outputs, are
1.2 million and 3.6 million), counted with checked arithmetic, and
read no further than `/Size` asks. `/Size 0` no longer underflows
building the default `/Encode`.
Three rounds of independent review found the aggregate cases, buffer
doubling past the limit, a chain/predictor shape the first image limit
would have truncated, the Type 0 amplifier, and a first, cumulative
budget that would have dropped legitimately repeated content, and a
first Type 0 cap low enough to refuse real tint transforms; all fixed
here.
Measured in a release build (render; OS peak commit):
- content stream inflating to 2 GB: 389 MB, 0.3 s
- 8 x 8 image inflating to 2 GB: 4 MB, 0.2 s
- /Contents naming it twenty times: 516 MB, 0.3 s
- a form painting itself around it: 644 MB, 0.4 s
Synthetic regression tests in render (flate2 is a new dev-dependency
there, already in the workspace):
- `decompression_bombs_stop_at_the_stream_limit`: 288 MiB bombs through
flate2, the fallback inflater (a stored block with a wrong length
check, which flate2 refuses and the fallback accepts), LZW and
RunLength each decode exactly 256 MiB, and at most 128 bytes as an
8 x 8 image.
- `page_contents_and_nested_forms_share_one_budget`: three /Contents
references decode one stream's worth; a self-painting form finishes.
- `content_painted_again_and_again_keeps_drawing`: a 2 MiB form painted
300 times draws the 300th time.
- `sampled_functions_with_billions_of_samples_are_refused`.
Ten mutations each turn them red: no limit at all; no limit in flate2,
the fallback, LZW, RunLength or for images; an unshared /Contents limit;
no page budget; no refund on drop; no Type 0 sample cap. (Taking a form
out of the budget no longer compiles: its content is the budget guard.)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* M2: normalize JPEG2000 samples without overflowing
Use bounded integer sample scaling for 1-32-bit JPX images and reject empty dimensions before chunking. Cover public decoding and continued page rendering with synthetic regressions.
Co-Authored-By: OpenAI Codex <codex@openai.com>
* Never crash: decompression cut at its limit is logged, not silent; vendor README renumbered after #575
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Streetblock <github@streetblock.de>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* M2: normalize JPEG2000 samples without overflowing
Use bounded integer sample scaling for 1-32-bit JPX images and reject empty dimensions before chunking. Cover public decoding and continued page rendering with synthetic regressions.
Co-Authored-By: OpenAI Codex <codex@openai.com>
* ROADMAP: keep the JPX entry with this session's entries
---------
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* M3: rasters reach egui without a copy on the UI thread
Every page, tile and thumbnail raster was copied pixel by pixel on the UI
thread (`ColorImage::from_rgba_premultiplied`), because egui's `Color32` is
4-byte aligned and a `Vec<u8>` isn't. That is 0.5 ms per 1024² tile, 5.4 ms
for a 33 MB page and 36 ms for a 197 MB page, against a 16.7 ms frame.
hayro, vendored patch 7: `render_size` and `render_into` render into a buffer
the caller allocates. `render` is `render_into` on a new pixmap, so its output
is unchanged. pdfcraft-render renders straight into `Pixels`, the same
premultiplied RGBA bytes kept in one 4-byte word per pixel. `Pixels`
dereferences to `[u8]`, so every reader of `RenderedPage::rgba` compiles
unchanged. The canvas reinterprets the words as `Color32`s and hands them to
egui as they are.
Pixels are identical: all 33 renders of #260's test file hash the same. The
UI-thread copy is gone, and rendering a 6182×8000 page takes 67 ms instead of
71–76 ms.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* M3: render_into leaves a buffer of the wrong length untouched instead of reaching vello's assert
* M3: regression test for render_into with a wrong-length buffer
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Issue #324: vendor winit 0.30.13 as published
The crate as released on crates.io, without its asset files (docs/res/*.svg and
examples/data/*.png, AGENTS.md §1). Not used yet: the next commit patches it and
adds it to [patch.crates-io].
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Issue #324: on Windows 11, a window dragged to another monitor takes its scale
winit 0.30.13 handles WM_DPICHANGED by checking that the new window rect is on
the monitor `MonitorFromWindow` names. During a drag that is still the monitor
being left, so winit nudged the window back onto it, which sent another
WM_DPICHANGED: the window grew and stayed on the wrong monitor at the wrong
scale. From Windows 11 (build 22000) on, apply the rect Windows suggests, as
winit master does since 488c036a (rust-windowing/winit#4341); no 0.30 release
has it. Windows 10 keeps the old adjustment, as on winit master.
The vendored copy goes in [patch.crates-io]; a test fails if winit resolves
from crates.io again or the vendored copy loses the patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
When a render runs past `stuck_after`, the watchdog answers with an error,
starts a replacement worker and drops the late result. The abandoned worker
still rendered the page to the end: #409's stop flag belonged to the whole
pool and was only set when the pool was dropped.
Each worker now has its own stop flag. Dropping the pool sets every worker's
flag, as before. The watchdog sets the flag of the worker it gives up on, so
that render stops at its next content operator. Its answer was already thrown
away, so nothing the user sees changes. Live workers are never stopped.
After giving up on 4 heavy renders, the process used 11.2–11.6 s of CPU in
the next 3 s with nothing left to do; now it uses none.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
`cargo xtask fuzz` found a 1.1 KB page whose JBIG2 image never finished
decoding (over two minutes, flat memory). It differs from a valid file by
one byte inside a symbol dictionary's arithmetic-coded data. Past the end
of its data the arithmetic decoder keeps returning values, and two loops
in symbol.rs only move on when those values make progress:
- Export runs (6.5.10): a run of length 0 doesn't advance, and the
decoder returned 0 for ever. This was the fuzz finding.
- Height classes (6.5.5): a class whose first width is already the
end-of-class marker decodes no symbol. A read-only sweep of the crate
found this one. In a synthetic dictionary every class after the first
symbol adds 13 to the height and decodes nothing, so it only fails
when the height overflows u32, after 330 million classes (about five
seconds per image, logged); a delta of 0 would never end.
The sweep found every other loop in hayro-jbig2 bounded by a symbol,
instance or pixel cap or by the end of the data. Bounded is not cheap
everywhere: refinement bitmaps in text regions and aggregate symbols
(up to 10,000 instances of up to 2^28 pixels each) are not changed
here.
Vendored hayro-jbig2 patch (3): a valid dictionary needs at most one
height class per new symbol and `2 x symbols + 1` export runs (they
alternate, so at most one zero-length run before each symbol's). Both
loops get that budget plus `EXTRA_EMPTY_STEPS` (1024), then fail the
dictionary as invalid. Encoders don't write empty steps back to back,
so the slack only tolerates oddities.
The fuzz finding now renders in 0.2 s. Synthetic regression test in
render: two dictionaries of a few arbitrary bytes each, found by
searching for streams that decode the same way (no corpus bytes). The
height-class one is drawn as twenty images so that it stalls well past
the 20 s timeout without the patch on any machine. Each case is red
with its own budget removed and green with it.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* M2: bound lopdf predictor scratch to available samples
Return empty PNG predictor frames before allocating row buffers. Reject
overflowing packed TIFF row dimensions and bound component scratch to the
samples present in the input, aligning with the existing lopdf#589 fix.
Cover hostile dimensions and valid packed rows with synthetic regressions
and record the vendored patch and its upstream proposal.
Co-Authored-By: OpenAI Codex <codex@openai.com>
* Keep main's M1 roadmap row after merging main
---------
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Mex
Marco Arnulfo
FredApps
TheLastRoadRunner
philip-tam
Joel Tony
git-monchiant
Eduardo Soares
Krzysztof Cieślik
gyurix
Maksim Popov
cyfung1031
mlkonrad
aAffliu
snipeTR
Ikki Nishihara
jcl-bit
Streetblock
skylar taylor-barrick
Phil Kwek