The marker moves from a CSS ::before to a widget decoration: generated
content is no text node, so it reached neither the PDF nor a measurement.
The note also carries the paragraph style its file names, so font, size and
colour are the document's; only the hanging indent stays the editor's,
because the style model has no first-line half to pair with the margin.
Adds a unit test for the numbering and the settings, plus the architecture
doc and the two open parity findings.
Numbering format, start value, restart, position, prefix/suffix and the two
styles per class, plus the separator's length, thickness, spacing, alignment
and colour. The separator rides :root custom properties, so editor.css draws
the line and pageBreaks.ts reserves its band from the same values. A settings
change is no document change, so it carries a resync meta — without it the
numbering kept the format it was typed under.
The placement loop moves into placeLeaves(), which takes a per-page note
reservation off each contentEnd; a bounded fixed point re-runs it until the
reservation and the pages its own anchors land on agree. Each break now
carries the leaf offset it applies from, so an anchor below a line split
follows its half of the paragraph. Endnotes stay in flow and open their own
page, as LibreOffice collects them.
The anchor is an inline atom, the note text a block in the one noteSection
at the document end; a sync plugin keeps notes in anchor order, renumbers
them per class and drops the ones whose anchor is gone. Numbering, indent,
size and separator defaults are LibreOffice's, read out of a document it
saved. Endnotes still render in flow at the document end — lifting a
footnote to the foot of its page is the next step.
A pagination pass measures the DOM the last one changed, so a fresh
document searched for its layout in public — 12 of them over 550ms on a
two-column file, a phantom third page among them. Two real causes: the
flow's overflow test had no headroom, so it read its own last answer
(880px fits undecorated, 913px overflows once the decoration pins the
column height), and a split dropped the height decorations without
clearing their cache key. The rest is inherent to measuring rendered
DOM, so the page stays hidden until its layout repeats: one visible
layout instead of 12, and 21-139ms where there are no columns.
Two compounding bugs on a title/back-page background image:
- its host paragraph picks up position:relative from the line-height
rule, becoming the frame's containing block instead of the page and
offsetting it by a full page margin — now forced back to static.
- the frame always painted behind text (z-index -1); a graphic that
declares ODF's style:run-through="foreground" now paints in front,
as LibreOffice draws it, via a new inFront attr round-tripped through
ODF export/import.
Confirmed against LibreOffice's own render: the duplicated, ~2cm
displaced cover graphic now matches exactly.
Word paints a link through its Hyperlink character style, never for
being one, so the editor's own blue overwrote both the file's #0000FF
and the plain black of a link carrying no style at all. The DOCX
importer marks those `plain` and the CSS leaves them alone; LibreOffice
paints every ODF text:a, so those keep the visual.
A book's chapter heading is "Chapter 1" and its title on two lines, and both
LibreOffice and Word copy that break into the index. We joined it into one line,
so a fixture's contents came out 18 rows short. The leader and page
number ride the last of an entry's lines (align-items: last baseline).
.tiptap's padding draws one pair of side margins for the whole document, so a
title page declaring 3.18cm left sat 3.85mm off. The difference now rides a
--sec-inset-left/-right node decoration on the section's top-level blocks, and
everything that writes a block margin adds it in — an indent, a named style's
and a table's alignment all compose with it instead of losing to it.
The no-descent rule matched a block holding nothing but frames, so the
image-break-caption idiom kept a text descent under its picture — 0.9mm
per figure, and LibreOffice's own guides stack dozens of them in tables.
Test the run between two hard breaks instead and mark the frame itself.
Probed: an as-char frame alone in a paragraph makes a line exactly as tall as
itself whatever style:vertical-pos/-rel says — only from-top's offset still
lengthens it. The node view cannot see its siblings, so the block's class
overrides the alignment it wrote.
LibreOffice sizes such a line max(frame, line-height) — nothing of the text
sits under the frame (probed: an image-only paragraph is exactly as tall as
its image, down to a small one, where the text line wins). Our baseline
alignment left the strut's descent below it, ~1.1mm per formula. Whitespace
beside the frame counts as nothing, which CSS cannot ask, so image.ts marks
the block.
A run's own formatting wraps its hard break in a span, so the block's last
element was the span, not the BR — the trailing-break BR was hidden and the
empty line the break opens went missing. LibreOffice renders it (probed: a
Word paragraph closing on <w:br/> keeps a blank line before the next block).
The ¶ reserved after every block sat in the line's flow, so a trailing space
was no longer at the line end and could not hang into the margin: a last line
that exactly filled its width pushed the zero-width mark onto a line of its
own, and the block came out one line too tall. Out of flow at its static
position the mark is invisible to line breaking.
ODF style:vertical-pos/-rel on an inline frame, which every LibreOffice
formula image carries: baseline, character area, or a plain offset from the
baseline. The default moves from the line-box bottom to the baseline, so a
line with a picture in it keeps the descent below it, as LibreOffice and
Word both do.
A painted border is a whole pixel however thin the file declares it, and the
excess came off padding-bottom alone — a cell that declares none (a fixture's Mac
table) kept growing 0.25mm a row. What is left over now comes off padding-top.
The atomic leaf a text box becomes carried neither its space above nor its space
below, so the fit test ignored both and a page top never dropped the space above
the way it does for a paragraph. The box's space rides --space-top, which the
page-top decoration zeroes — the rotor centres itself in the padding box and has
to follow it.
An as-char figure frame sits in a paragraph of its own; the importer lifts the
box to top level and dropped that paragraph whole, so its space above and below
went with it — every figure in a fixture sat 2.1mm too tight. The box now carries
them (ParagraphSpacing covers textBox), space above as padding so it adds to the
block above, and the export mints an anchor style to write both back.
One fixture: 142 → 100 parity issues.
Two ways a table's height was off. Its space above was a margin, so it
collapsed against the paragraph above instead of adding to it — now it rides
--space-before like a paragraph's. And every cell border paints a whole pixel
however thin the file declares it, where LibreOffice reserves the declared
width; the excess now comes off the cell's padding-bottom, 0.25mm a row.
Two measured deviations, both worth ~1mm a row on a fixture: a cell's first block
got the space above that LibreOffice adds nowhere in a cell, and a repeating
header row rendered centred and bold because it is a <th> and the UA says so.
Pagination's page-top-drop detection now skips cell leaves, or the space that
is never there would read as a decoration to add back.
The TOC rendered every level with our 0.5cm-per-level indent, an editor-only
1.5 line height and no space above — so a fixture's contents drifted 7.6mm off
LibreOffice down one page and pushed entries onto the wrong one. The per-level
entry template names a paragraph style; the rows now carry it as data-style
and the document stylesheet does the rest. Our own indents stay the fallback
for an index whose file names none.
Every row carried 200 dots — clipped on screen, but the clipboard, the
PDF and any measurement got all of them. Cut each to what its gap holds,
and step the levels by 0.5cm, which is what LibreOffice's Contents styles
use. The parity runner now waits for the page numbers to settle too.
ODF text:anchor-type="page" is a cover graphic or a watermark: out of
flow, placed from its page's corner, behind the text. In flow, the two
full-page frames of a fixture's covers cost three pages.
LibreOffice adds the upper block's space-below to the lower one's
space-above, and drops the space-above where a page break put the block
at a page top — but for a Word document it takes only the larger of the
two (all probed against soffice). CSS margins can only collapse, so
space-before rides --space-before and editor.css renders it as padding
('add') or margin ('max'); the model is read from settings.xml, assumed
for DOCX, and written back on ODF export.
A positioned frame in a header is out of flow — the title-page background or
watermark. It keeps its wrap and its offset from the page corner, and the
layer paints it per page under the body instead of flowing it inline, where
it reserved a page-tall line of body space.
The page surface moves from .tiptap to .paper so the image has a layer
between the sheet and the text; ODF export writes it back page-anchored.
A formula wider than the text area painted over the app background
instead of being cut at the page edge, where LibreOffice cuts it
(thesis fixture, page 16). `overflow-x: clip` so .paper does not
become a scroll box.
Word's w:lvlJc / ODF's fo:text-align on the level: set right, the label grows
into the margin instead of over the text it labels. That is what the built-in
Roman numberings use, and a fixture's numerals collided without it.
A spacing above single puts all of its leading after the line (probed against
LibreOffice at 150%); CSS splits it. The block is shifted back up by that half,
which leaves the flow — and pagination — where it was.
A formula node stores only its LaTeX source; the MathML on screen, the ODF
formula object and Word's OMML are derived from it, and both importers parse
foreign markup back into that source, so an imported formula stays editable.
Word documents were losing their formulas outright: m:oMath sits beside the w:r
runs in its own namespace, past convertInline's `namespaceURI !== W` guard.
The left⇥centre⇥right layout every letterhead uses rides on them, and a
zone had none: Word keeps them on the Header style, ODF on the zone
paragraph's, and neither was read or written.
An inactive zone is static generateHTML output no plugin reaches, so it
is measured straight on the DOM. Two bugs surfaced doing it: a segment
its own advance had pushed onto the next line measured as wrapped, and
the alignment then had no way back — it takes the sum of the segment's
client rects now. And the scale came from offsetHeight, which rounds
half a percent away on a zone one line tall.
The default tab interval was fixed at 1.25cm, so a file's own
w:defaultTabStop / style:tab-stop-distance was ignored and the DOCX
export wrote 1.25 back over Word's 1.27. It is a document setting now.
A stop can also carry the character it repeats across its gap. odf-kit
has no leader option and mints one automatic style per option set, so
the leader rides style:type behind a sentinel — matching on position
and alignment alone gave every stop at 6cm the same leader, which
LibreOffice drew back plain as day.
CSS gives ::marker no position, so Chromium set it flush against the
text — about 4mm right of where Word and LibreOffice draw it. The marker
is now the item block's ::before, placed at the hanging indent both
exports already write (LIST_HANGING_CM 0.635cm): measured against LO's
own render, the three bullet levels land within 0.04mm.
Out of flow, because in flow it grew every list line by a pixel and cost
a fixture fixture a page. `list-style: none` drops only the marker box,
so the list-item counter and <ol start> keep numbering; a bullet's own
char rides an inherited property, since attr() in a pseudo would read the
block rather than the list.
A long URL or a German compound ran straight past the right margin, where
Word and LibreOffice break it — measured against LO in a 2cm column, the
break now falls on the same character.
`break-word` rather than `anywhere`, so a table column's minimum width is
still its longest word.
The reserved paragraph mark is inline content past the block's last BR,
so it opened a line of its own: LibreOffice set the next paragraph 4.7mm
higher than we did, and a fixture paid for every one of them.
The empty-block rule already pulled the mark out of flow for exactly this
reason — it now covers any block that ends in a break, seating the glyph
on the last line rather than the first.
The cell padding was a constant in three places that disagreed with each
other, so a narrow column lost 0.38cm it did not have: a bibliography
marker column of 0.71cm wrapped "[2] " onto three lines and every entry
grew with it, costing a fixture a whole page.
The table now carries its own margins (w:tblCellMar, direct or from the
table style; fo:padding on the ODF cell style) and both exports write
them back. A side within a twentieth of a millimetre of the default is
the default, or Word's 108 twips would tag every table.
Its leader is real dots now, not a border, so it scales with the font and
reaches the PDF as text. Title and entries follow the document's Heading 1
and default style, which is what the index styles of Word and LO are based on.
Word's Title style tightens by -0.5pt, so a title came out ~3mm wide of
LibreOffice's. letterSpacingPt joins TextProps and rides both formats
(w:spacing, fo:letter-spacing).
Line spacing needed the same care: it is proportional to the font's natural
line height, so the factor and that height are now separate CSS variables and
editor.css multiplies them in one place. Before, whichever of the two was
declared nearer simply overwrote the other — a block carrying its own font
silently dropped the style's spacing.
A paragraph holding an image rendered a line taller than it does anywhere
else: under white-space: pre-wrap the browser already shows the last line, so
ProseMirror's hack BR adds an empty one. The hack is kept where it earns its
keep — after a real break it carries the caret onto the empty line, and in an
empty block it is the only line box there is.
The marker took only the weight of the item's first text portion. It now takes
the family, slant, size and color too — as custom properties on every item,
since they inherit and a nested item must not take its parent's.
A named paragraph style reaches it a second way: the marker inherits the list
item's own font, and the document stylesheet only styled the item's paragraph,
so an imported document whose Standard style is Arial drew its numbers in the
editor's serif default. styleCss now emits the style's text half for the item
too (li:has(> …)).
Both file formats carry marker formatting per level, so a level definition is
written only for a list whose items agree: ODF mints a character style the level
points at, DOCX the level's w:rPr. A mixed list gets none on purpose, so
LibreOffice keeps applying its own first-portion rule.
A bold mark sits inside the item's paragraph and never reaches ::marker, so the
roman numerals of a numbered heading stayed light while their text was bold.
A plugin tags such items and editor.css weights the marker. The trigger is the
first text portion, which is the rule LibreOffice numbering itself uses.
Lists were padded 1.5rem per level on screen while the DOCX and ODT
exports place item text at LIST_LEFT_STEP_CM (1.27cm) per level, so the
editor disagreed with its own output and with Word and LibreOffice by
6.4mm on the first level, doubling with each nesting level.
Cells padded 0.1cm on all sides; Word and LibreOffice use 0.19cm left and
right and none top or bottom. Text sat 0.7mm too far left and every row
was 2mm too tall, so the offset accumulated down a table.
odf-kit only writes the fo:padding shorthand, which ODF defines as a
single length — LibreOffice drops a two-value form — so the export
expands it per side.
Steffen Becker