* M12: preserve document properties when importing legacy DOC and DOT
M12: preserve document properties when importing legacy DOC and DOT
Integrate current upstream while retaining import behavior and embedded-font loading.
* docbin: as_chunks for fixed-size records (clippy::chunks_exact_to_as_chunks on Rust 1.99); rustfmt
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: LabDennis <341377186+LabDennis@users.noreply.github.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* View Gridlines outlines header and footer tables (#1628)
Table Layout › View Gridlines drew only the body's table cells; a table
in a header or footer showed no gridlines. Walk the page's header and
footer items too.
Fixes#1628
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Test: Placed::Cell names its clip flag (#1617)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Word's cover pages are one paragraph holding a page break and then the
page-anchored text box with the title and the approval table. Word
draws a floating object on the page its anchor paragraph starts on;
WordCraft drew it with the line holding the anchor, which after the
break is the first line of page two, so the cover was blank and the
title box sat over the body text of page two.
`place_objects` now places a paragraph's floating objects from its first
piece, whichever of its lines holds the anchor, and none from later
pieces; inline objects still go with their own line. The wrap areas were
already registered from the paragraph's first page, so nothing changes
for text flowing around the object.
Co-authored-by: Daniele Mori <Dinasty@hotmail.it>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
- Envelopes: the size list was drawn as English text (`#10 4.13 × 9.5 in`)
with a decimal point in every region. The label is now a catalog template
(`{name} {height} × {width} in` / `mm`) filled with the region's decimal
separator; ru, hu, tr, bg, pl and it carry it. Regression test included.
- Italian: the 18 strings the interface audit looks up that it.tsv lacked
(Reading Mode bar, Go To, command search, status and command messages);
`Privacy Settings` no longer says "current".
- The audit's catalog-completeness check covers Italian too.
- docs/localization-parity.md: status line, Italian note, revision history.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
w:asciiTheme="minorBidi" (and majorBidi, and w:cstheme) read as the
theme's Latin font, so a template's titles came out in Calibri where
Word draws Arial. Measured in Word: the slot names the theme's a:cs
font, and an empty a:cs names nothing, so the font inherited from the
style applies.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
WordCraft chose a substitute for a font that isn't installed from its
name alone, so an unknown brand font always came out sans. Word reads the
font table instead: measured with the same missing font, it draws it in
Cambria when w:family is roman and in Calibri when it is swiss or auto.
A report whose font table calls its brand font roman got a sans footer
where Word draws serif.
The font table's roman fonts are read into the settings, and layout looks
a missing one up as Cambria (Caladea where Cambria isn't installed). The
other classes keep the existing substitutes, which already start with
Calibri's metric clone.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
space after and the second's space before, as Word does by default. Word
adds them instead when the document's compatibility options include
w:doNotUseHTMLParagraphAutoSpacing. Measured in Word with five paragraphs
of 12 pt before and after: 24 pt between them with the option, 12 pt
without, the same in compatibility modes 14 and 15. A 60-page report with
the option lost 6 pt per paragraph, about 60 pt on its cover.
The option is read, laid out in the body and in boxes, and saved back.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
A header or footer a section inherits (Same as Previous) keeps the
percentage widths of its tables relative to the section that defines it.
Measured in Word: a 100%-wide footer table defined by a portrait section
stays 451 pt wide, centred, on landscape pages that inherit it, and spans
the landscape width only when that section names the footer itself.
Paragraph alignment in the same footer follows the landscape page.
Found on a report whose landscape calculation pages drew their footer
table full width where Word draws it at the portrait width.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Layout: an empty section mark stays on a full page
An empty paragraph that carries a section break, when the next section
starts on a new page, needs no room of its own in Word. Measured: 50 lines
filling a page, then that paragraph, then the next section makes two pages
in Word; WordCraft made three, the middle one blank. A plain empty
paragraph in the same place still moves on, in both.
Found after a landscape drawing page in a 60-page report.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Layout: drop an unneeded mut in the section mark test
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>
* Tables: a row with less room than its minimum height moves on whole
Measured in Word with about 49 pt left on a page: a table row of eight
lines splits there, but the same row with an at-least height of 63.6 pt
goes to the next page whole, and so does one holding a nested table.
Without the height both split, as before. WordCraft split them all, so a
report's title row (a logo beside a small nested table) started at the
foot of its cover page.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Tables: preserve splitting for oversized minimum-height rows
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Tables: text running past an exact-height row is cut off
A row with an exact height (w:trHeight w:hRule="exact") keeps that
height whatever its cells hold. Word cuts the text that doesn't fit at
the cell's edge; WordCraft drew it over the rows below, so a certificate's
identification block came out as overlapping lines.
The cell is marked as clipped in layout (Placed::Cell clip), and display
draws its lines inside that cell's clip, on screen and in PDF. Hit
testing is unchanged: the lines stay where they are in the layout.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Tables: keep row clipping compatible with Rust 1.90
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
A heading whose text opens with a page break had its list number drawn on
the line holding the break, so the number stood alone at the bottom of one
page and the heading text started the next. Word draws the number with the
text. The label now goes on the first line after a leading break.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Word writes the next heading's TOC bookmark start into the paragraph that
ends with a page break. The line after the break then held that marker,
so it wasn't the empty mark kept with the break (#932): it went onto the
next page with the paragraph's spacing and pushed the heading 17 pt down.
Lines holding only zero-width markers (InlineObject::is_marker) now count
as empty there.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
A paragraph holding only a page break keeps its mark on the page before
(#932), but its space after still went onto the new page, so a heading
with page break before no longer stood at the top and broke again: a
blank page. The space after now stays with the mark. A next-page section
break carried by such a paragraph also started a second page; the empty
page the break started is now used instead.
Found on a 60-page inspection report: two blank pages Word doesn't have.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Follow-up to #1900: the fixture test vml_preset_shapes_are_drawn also
expected the isosceles triangle for VML shape type 6 (#1599).
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
#1599 made VML spt 6 (rtTriangle) a RightTriangle; #1442's mapping test
still expected the isosceles Triangle, so it failed on main.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Left of the left margin Word's next default tab stop is the first
multiple of the interval *at or after* the current position, not after
it: its arithmetic truncates toward zero. So a tab that already sits on
a stop, a second tab in a row or a leading tab on a line that starts at
-72 pt, doesn't advance; a tab preceded by text or a space, or a line
starting between stops, moves to the next multiple as usual, and right
of the margin nothing changes.
Measured on Word with synthetic documents (40 positions): "a\t\tX" at
-60 pt puts X at -36 pt where the old rule gave 0; "\tX" starting at
-72 pt leaves X at -72, at -60 moves it to -36; "\t \tX" at -60 reaches
0. Letterheads that line a footer up with spaces and tabs under a
negative indent rely on it: one had its registration number drawn 72 pt
too far right, over a vertical rule. It now lands within 0.2 pt of Word.
Co-authored-by: Daniele Mori <Dinasty@hotmail.it>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
DrawingML's rtTriangle was read as an isosceles triangle, so a cover
page's white corner cut-outs pointed up into a peak instead of cutting a
V into the picture above them. It is now its own shape kind, drawn with
the right angle at the bottom left, wrapped and glued along that outline,
and saved back as rtTriangle. The .doc reader maps msosptRightTriangle (6)
to it too.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Engine: the Word 2003 format check reads through io, so commands still never touch files themselves
is_word_2003_file (#1081) opened the file with std::fs, which the file-access guard test
(commands_do_not_touch_the_file_system_themselves) forbids outside io.rs; main failed it.
A new io::read_file_head reads the first bytes inside the read root.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* UI: a chrome click counts the page as focused at once, so the next keystroke isn't dropped; the ribbon-focus test passes again
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix: DOCのDIB画像変換でバイト上限を守る
* Drop the duplicate read_file_head left by the merge with main
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Dmitrii Khlapov <19991457+dkhlapov@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: haranishi <223786623+haranishi@users.noreply.github.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Italian proofing: po', qual è, È and un/un' grammar rules; more elisions in spelling
- grammar (works without a dictionary): un pò / un po -> un po'; qual'è -> qual è;
E' -> È; un'altro -> un altro and un altra -> un'altra (also nessun, ciascun,
qualcun, alcun), gender from the ending with masculine -ista/-ma/-cida... left alone;
more emphatic doubles allowed (pian piano, poco poco, così così...)
- spelling: more elisions (dov'è, cos'è, anch'io, senz'altro, nessun'altra, n'ha...),
truncations po'/mo'/be' accepted, pò and qual'è left to grammar (no double marks)
- tests: rule tests, grammar proptests, Italian suggestion proptest, It in mixed-span proptest
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Contributors: credit lupant by their GitHub profile name
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>
review.deleteComment removed anchors from the body only, so a comment
anchored in a header, footer, note or text box left its CommentStart and
CommentEnd behind, pointing at a deleted definition. It also collected only
direct replies, so a reply to a reply survived with its parent gone.
Anchors are now removed from every story, and the deleted set is the whole
thread below the comment. The thread is collected from a parent-to-replies
map built once, with a seen set, so a parent cycle from a hostile file ends
and a long reply chain doesn't take quadratic time. Undo restores everything.
Regression tests: a header comment (anchors checked in the header story), a
three-level reply chain (with its text parts and Undo), a parent cycle, and a
50,000-reply chain.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Save got focus only when nothing else had it, but the page keeps focus while typing and claims Tab and the arrow keys, so the prompt opened without a focused button and the keys moved the caret behind it. Save now takes the focus whenever none of the prompt's buttons has it.
Co-authored-by: Brandon Thomas <echelon@gmail.com>
w:pgNumType w:start="0" was read but laid out as 1: the page counter
was set to one less than the restart and could not go below zero. A
report numbering its cover 0 read "Page 2" on the page Word calls 1.
The restart is now the next page's number itself.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Layout: a table's percentage width past 100% is kept
Word keeps a preferred width over 100% of the text column (w:tblW
w:type="pct" above 5000). A cover-page footer table at 131% with the
left margin as a negative indent spans the page edge to edge in Word;
WordCraft capped it at 100%, so it ended three-quarters of the way across.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Contributors: add Fin Warman
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
The model's paragraph walk (each_para / each_para_mut) stopped 16 tables
down, while a .docx opens with tables nested up to 24 deep and pasting can
nest them further. Text box pruning, which runs after every edit, took a
box shown only from deeper for one shown nowhere and dropped its story,
so an unrelated edit lost the box's text. The same walk decides which
boxes a copy takes along and which pasted shapes get stories of their
own, and serves reading order, the next drawing id and note conversion.
The walk now keeps its own stack instead of recursing, so it reaches
every paragraph however deep tables nest, in the same document order and
with no call stack to run out of; a paragraph on its own still needs no
allocation. The signature is unchanged (the depth argument is now
unused), so every caller builds as before.
Tests: a box 17 tables down in an opened .docx survives an unrelated
edit and a save, while a box shown nowhere still goes; boxes 17 and 40
tables down, in a box and in a header are kept, counted in reading order,
and new drawing ids go above the deep shape's; a copied deep table
pastes into another document showing its own box's text.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The Word XML Document (Flat OPC) reader counted only the parts it kept:
a repeated or nameless part, a child of the package that isn't a part
and anything nested inside one were read without being counted, nesting
had no cap, and a part's inline XML was copied before its size was
checked. Every child of the package now counts against the part limit
and every part's data against the part and package size limits, kept or
not; nesting and the elements of each child are capped as the XML tree
reader caps a part's; and data is checked before it is copied.
Base64 in pkg:binaryData was taken from plain text only, so CDATA and
character references were dropped and a picture written that way lost
bytes. Both now read as the same text, and an entity there is an
invalid base64 error.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
russianLower/russianUpper showed 1), 2), 3). They now follow Word's 29-letter sequence (no ё, й, ъ or ь; ы kept), then repeat the letter (аа, бб, ааа), as observed black-box in Word.
Co-authored-by: Brandon Thomas <echelon@gmail.com>
With room for two of its three lines at the bottom of a page, a paragraph was
split two and one, leaving a widow; taking a line along would leave an orphan,
so Word moves the paragraph whole. A paragraph continued from the page before
may still take one line along from two (no orphan there).
Co-authored-by: przemek-arch <263159043+przemek-arch@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
The story tables of footnotes and endnotes (PlcffndTxt, PlcfendTxt),
headers and footers (Plcfhdd) and text boxes (PlcftxbxTxt) were read
one range at a time, and each range only had to run forwards: a table
stepping back to an earlier CP after every story pointed many notes,
header slots or text boxes at the same text, and each was read into a
part of its own (2,000 note references on one story made 1,001 copies
of it). The CPs of a well-formed PLC ascend, so a story that starts
before the end of the last one kept is now dropped (a header or footer
slot is left empty, which links it to the previous section's): every
subdocument CP becomes at most one story, and the stories together
never hold more than the file's text.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* A file name that isn't valid Unicode no longer crashes wordcraft-cli or the app (#1440)
Both binaries read std::env::args(), which panics on a non-Unicode
argument such as a Latin-1 file name. The CLI now reads args_os(): the
file a command opens keeps its real name and opens (convert, info, text,
inspect, render, zotero, run --file), and any other such argument is an
error naming it, exit 1. The app leaves such a file out with a warning,
as it does a file that fails to open, since file.open takes its path as
text; a --control port that isn't valid Unicode exits 2.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* The app opens a file whose name isn't valid Unicode too (#1440)
file.open takes its path as text but also accepts the file's bytes
under a name, as dropped files use. The app reads such a file, opens
its bytes under the lossy name and keeps the real path on the session,
so Save writes back to it. A unit test covers the open and the save.
The CLI test now checks that every command in OPENS_FILE reaches its
open with the file, and no longer pins `pos` counting a `--page` value.
ROADMAP says which saves go back to the file.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* #1440: skip non-UTF-8 file name tests on macOS
APFS refuses file names that aren't valid UTF-8, so the fixtures fail
with "Illegal byte sequence" there, and the app-start test keeps running
because macOS opens a window without a display variable. The unit test
for the CLI's argument checks and the control-port test still run.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Proofing: status bar flags a language whose dictionary isn't installed (#1167)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* vi: missing-dictionary status strings
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>
On macOS five of Word for Mac's default shortcuts were missing or unreachable:
- Cmd+Shift+S is Save As (it opened Apply Styles, which keeps Ctrl+Shift+S
on Windows and Linux).
- Cmd+Option+E inserts an endnote. macOS takes Cmd+Option+D to show and
hide the Dock, so the old key never reached the app.
- Cmd+Option+A adds a comment (Cmd+Option+M still works).
- Control+H opens Find and Replace (Cmd+H still works).
- Cmd+Option+R collapses or expands the ribbon (ui.collapseRibbon).
Keys on Windows, Linux and the web are unchanged. Tests pin the bindings
on each platform.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* contributors: add Nael-Nathanael
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* insert.picture and insert.icon leave the caret after the picture (#965)
A script that typed right after inserting a picture replaced it, so the
picture was missing from the saved file. Commands now leave the caret
after the new picture; Picture Format commands still act on it. The app
still selects a picture the user inserts, as in Word.
Fixes#965
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
* Select a pasted picture again, keep pictures inserted in a row (#965)
Desktop paste inserts through execute, so after the caret change it
left a pasted picture unselected. WordApp::select_inserted_picture
selects the picture just before the caret; run and the paste path call
it, paste once after the last picture so a multi-file paste keeps them
all. Tests for paste selection and for two inserts in a row.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
New parts, lists and comments took "largest id + 1": after an id of
u32::MAX (a .json model or a .docx can carry one) that overflowed, a
panic in debug builds and a wrap to 0 in release, and list.define's
saturating_add reused the taken id. free_id gives one past the largest,
or the lowest free id once that would overflow; try_add_part,
try_add_list and try_add_levels report ids running out and add nothing,
and the commands that add parts and lists use them.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
Paragraph::normalize now uses saturating sums, moves a run end that splits
a character to the next boundary, and makes the runs cover the text
exactly. Opening a .json document normalizes every paragraph in every story.
Co-authored-by: justinjohn0306 <babyjjbabyjj549@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* #1336: Markdown looks up the backtick run closing a code span instead of rescanning the text
Code spans and link brackets searched the rest of the text for the run
of backticks closing a code span. Every run nothing closes (one per
length) cost a scan to the end, and the bracket scan repeated those
searches for each `[` over them, outside its look-ahead budget: 2 MB of
such runs took 2-18 s. The runs are now indexed by length once per text
and found with a binary search; what reads as a code span or a link is
unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* #1452: Markdown brackets no longer turn later links in the paragraph into plain text
The link and image bracket scans share a look-ahead budget per text, so
hostile input stays linear, but it was charged more than they read: a
`[` nothing closes took the whole budget, and every bracket pair and
every link took the rest of the text, as if a destination had been read
to the end. An interval like `[0, 1)`, a dozen references in a long
paragraph or ten links were enough to make every later link plain text.
A scan now pays for the bytes it reads and records the pairs it passes,
and the brackets still open at the end of the text, so later brackets
are looked up. A destination is read, and paid for, only when `(`
follows the `]`; one that parses costs its own length.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Each time a page was redrawn (every keystroke on it, a zoom step, a page
scrolled into view) its raster was un-premultiplied (`Rendered::to_straight`)
and handed to `ColorImage::from_rgba_unmultiplied`, which premultiplied it
again. Both `Rendered` and egui's `Color32` hold premultiplied RGBA8, so the
round trip gave back the same pixels after two more passes over the whole
page.
The raster now goes across with `ColorImage::from_rgba_premultiplied`. A
test checks every premultiplied pixel value (each channel from 0 to its
alpha, for each alpha) and every page of the sample document against the
old path: identical.
Measured on an Apple Silicon MacBook Pro (120 Hz, 2x, 100% zoom, a
1632x2112 px page), typing into the sample document: the canvas redraw per
keystroke goes from 17.2 to 11.4 ms on page 1 and from 13.0 to 7.5 ms on
page 2.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
