Form controls:
- Placement: a control hanging from the baseline (as LibreOffice places
those it makes) was saved with no placement. LibreOffice then puts it
at the line's top, so every control row shrank. Its raise is now kept
(`ContentControl::form_raise`, read as in-line frames are) and
written, standing on the baseline when it had none.
- Multi-line fields: these were saved without LibreOffice's `MultiLine`
property, so they showed one centred line. They now carry it.
- Text fields get LibreOffice's flat border.
- Also: `pt()` no longer writes "-0pt".
Custom shapes:
- A shape drawn from its own geometry was saved as WordCraft's path of
it, and a preset WordCraft draws as a rectangle (a cloud) was saved as
a rectangle. The geometry as read (type, path, equations, handles) is
now kept (`ShapeExtra::odf_geometry`) and written back, mirrored as
the shape now is.
- `shape.change` and the ink-to-shape conversion drop the kept geometry,
since the shape is then another kind. `ShapeExtra` holds it as an
`Arc<String>` to keep `Placed::Shape` small.
- The reader's `end` hands skipped elements to `end_skipped`.
Results:
- The forms document round-trips pixel-identical (2,949 pixels before),
with its text the same.
- The shape presets fall from 2,467 to 1,220. The rest is LibreOffice's
hairline outline, which WordCraft keeps as 0.5 pt.
- Across the corpus the round-trip difference falls from 45k to 41k
pixels. 42 of 125 documents round-trip pixel-identical, 92 with their
text the same, and none gets worse.
gaps.md row 19 and the ROADMAP log are updated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Charts: on save, WordCraft wrote each chart anew from its chart model
(type, data, title, legend, labels), losing the look it doesn't model:
fonts and sizes, axis and grid lines.
- The chart object as read is now kept with the chart (`Chart::source`
in the flow model, `Graphic::source` in the document).
- `chart::redraw` keeps it when the chart is drawn again on open or
resize.
- The ODT writer writes it back whole while it still reads as the same
chart. An edited chart is written from the model, as before; DOCX
still writes charts from their model.
- The chart documents now round-trip pixel-identical (chartcol: 5,917
pixels to 0).
LibreOffice's serif business letter (4,223 pixels to 116):
- A text frame outlined by a drawn stroke (`draw:stroke`, a 0 width a
hairline) and no `fo:border` lost its outline. It is now the frame's
line; an explicit "none" border still wins.
- A header or footer whose style states no height, or that has no style,
was 0 tall. LibreOffice makes it 0.5 cm (measured by varying the
height), so the text had moved up 1.5 pt.
Across the corpus the round-trip difference falls from 60k to 47k
pixels, and 41 of 125 documents round-trip pixel-identical. No document
gets worse.
gaps.md row 19 and the ROADMAP log are updated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Word's Font › Emphasis mark (a mark over or under each character, used
to stress East Asian text) wasn't modelled. DOCX `w:em` and ODT
`style:text-emphasize` were dropped.
- Model: `CharProps::emphasis` (dot, comma, circle, dot below, or none
to turn off a style's).
- Layout: a mark is drawn over (or under) each character, and the line
is made taller to fit it.
- `format.emphasisMark {"mark": …}` sets it, and Format state reports
it. The catalog lists it under Home › Font.
- DOCX: `w:em` is read, and written in its place in `w:rPr` (after
`w:cs`, before `w:lang`).
- ODT: `style:text-emphasize` maps to it: an accent is the comma, a
disc the dot, a dot below the under-dot.
Also fixes two problems from the previous commit (ODT variables):
- Field updates blanked a SET field's stored result. The layout
already draws nothing for SET, so its result stays as it was, as
`ask_and_set_fields_draw_nothing` pins.
- A clippy lint (manual char comparison) failed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
- ODT variables (`text:variable-set`, `text:variable-get`, §7.4) were
read as their text. A set is now Word's SET field, followed by a REF
when it is displayed; a get is a REF. They are written back declared
(`text:variable-decls`), the sets hidden with a get after them.
- The engine now evaluates SET (ECMA-376 §17.16.5.58). It shows
nothing and gives its name a value. REF to a name no bookmark has
shows that value; a REF before the SET shows the last value set, as
Word's bookmark does after an update. This works for Word documents
too.
- Custom character styles now keep what they turn off under a parent
that turns it on (`fo:font-weight="normal"`…), as paragraph styles
already do.
A LibreOffice document with variables round-trips with the same text,
and the saved file validates against the schema.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
LibreOffice ships 78 text templates (letters, faxes, agendas, reports,
CVs). Round-tripping them through WordCraft exposed these problems:
- Placeholders (`text:placeholder`, "<Recipient's name>") were
printed text. They now open as plain-text content controls showing
them, so a click selects one and typing replaces it. Their
description becomes the control's title. They are written back as
placeholders of their type while they still show; what was typed is
written as text.
- A fixed date with neither a value nor text was lost. It is now a date
field.
- `text:author-name` opens as USERNAME, which the engine left empty.
USERNAME and USERINITIALS now show the user's name and initials
from File › Options; a quoted text after the name is shown instead.
- A "First Page" master followed by another (`style:next-style-name`)
put the running header on the first page. It is now a section with a
different first page: its own (empty) header on page 1, the next
master's after that.
- A page layout without a size was Letter. It is now the default
page's, the user's region's, as LibreOffice uses its default paper.
ODT import takes the regional page like the other importers.
- Style-only templates (no `content.xml`) failed to open. They now
open empty with their styles.
- Table columns are now written with their share of the table
(`style:rel-column-width`) too. A table filling the margins of a
narrower page keeps its proportions in LibreOffice.
All 78 templates now open and round-trip.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Word brings a document's alternative format chunks (w:altChunk) in where they
stand when it opens the file; WordCraft left them out, so a document made of
one (as generators and mail-merge tools write them) opened empty.
- docx: read_with/ReadOptions with a ChunkImporter for formats other than
WordprocessingML; nested Word documents are read here (at most 4 deep,
256 chunks and 512 MB of chunk bytes a file, XML depth counted on).
- doc: Document::adopt (moved from paste) brings another document's media,
lists, styles, notes, comments and revisions along under new ids; pictures
in groups and charts are relinked. Insert > Text from File uses it and now
keeps the file's pictures and lists.
- engine: chunks in MHT, HTML (in their charset), RTF and plain text go
through the format importers.
- formats: MHT web archives open (MIME, quoted-printable/base64, charsets,
pictures by Content-Location or cid). HTML reads style element rules
(simple selectors, cascade, budgets), paragraph margins, indents, line
height, cell widths and borders; an opened page takes Word's web-page
defaults (Times New Roman 12 pt, single lines, black headings, automatic
spacing). html::decode knows the single-byte Windows code pages.
docxide-pdf suite: croatian_regulations_altchunk goes from one empty page to
Word's 8 pages, SSIM 2.7% -> 78.7%; no other document changes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Hidden form controls (`form:hidden`: a name and value the form sends,
not shown) were dropped. They are now kept verbatim, with their
properties, and written back into the form with ids of their own.
- OpenDocument text templates (`.ott`) weren't opened. They now open
as documents, and save as templates: the template's media type goes
in `mimetype` and in the manifest's root entry (ODF 1.3 Part 2 §3.3).
This works with a password too. `.ott` is listed wherever `.odt` is:
open and save formats, file dialogs, export, the web app and its
media type.
A LibreOffice template round-trips with the same text. 300 fuzzed
documents exercising custom shapes, cell formulas, kept parts, turned
text and forms caused no crashes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
- craft-proof now names ja, zh and th in missing_dictionary: packs::missing_dictionary hides it
for a pack language when no pack for it is listed (status bar and Options).
- de-CH and pt-PT are pack languages in craft-proof (de-ch, pt-pt): names, hyphenation and
grammar stay German and Portuguese (tests); pack names fall back to the region-aware name.
- About › Language data shows craft_proof::scowl::notice() when SCOWL is embedded (test).
- Flatpak manifest's craft-i18n commit follows craft-i18n.rev.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
- proof.packs reports index_override; the CLI prints the test source; the bar's Install tooltip
says it turns on downloading of fonts and Language packs, and names a test source.
- Docs: Install works once language-packs-1 is published (until then only a test index); with a
test index the sha256 only catches a broken download.
- About › Language data keeps its size: a fixed-height scrolling list, licence texts read again
only when the dictionaries change.
- The engine's Cancel is the font store's own flag (no snapshot), so a cancel stops a fetch or a
stalled download too.
- Index text shown without control, bidi or format characters; long names cut; the bar works out
what to offer again when the list changes (index generation, not its length).
- ui_shot's store is a temporary folder (or storeDir); tests reset the backend even on panic; a
fixture pack proves per-run proofing and hyphenation from a pack.
- The default-build check follows craft-proof's carry-over (de, pl, pt, it, ru embedded until packs
ship; CRAFT_PROOF_EMBED=en is English only); the Polish hyphenation check runs again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
- Install, Show available languages and Refresh turn on the font store's downloading only on the
person's own click (input_from_control, as in #959); a faked click says why instead.
- proof.installPack and proof.packs {refresh} run on the engine's thread in the app ("started"),
with progress, results and errors in proof.packs; the CLI and headless MCP stay synchronous.
- A session's layout is keyed on the dictionaries' generation, so parked documents proof again.
- proof.removePack refuses while that pack installs; pack commands that need a locale are left out
of Search; options text names Refresh; licence tooltip says Source; errors go through tl!;
the bar rescans at most once a second while typing; English names for every pack language and
sr-latn; About has one Close button; xtask doc comment back on cmd_i18n_report.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
- `review.issues` and `file.options` report a run's language with `ProofLang::tag()`, so a
Pack language reports its locale, not "".
- The test-source warning reads craft-font-store's `language_pack_index_overridden()`.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
- craft-proof (benandez/craft-libs 6ae4f03, storytold/craft-libs#485) and craft-font-store /
craft-font-catalog (630e42f, #486) by git rev instead of local paths.
- ProofLang::Pack: the bar offers every language with a pack (zxx/und never).
- Options › Language warns when CRAFT_LANGUAGE_PACK_INDEX supplies a test index.
- Pack `source` (upstream link) when the index gives it, else the archive.
- The bar's close button is the code-drawn close icon: "✕" had no glyph in the interface fonts
and egui panicked drawing it.
- `cargo xtask language` allows our own ru-extra, which stays embedded.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
- engine `packs`: the packs Backend (the font store on the desktop and CLI, a fixed list in tests
and ui_shot), the pinned index, installs and removals through craft-font-store's
`language-pack` kind, craft-proof reading the store's `language-packs/` folder at once.
Commands open nothing and never turn downloading on (only an Install click in the app does);
confined automation can't write the packs folder outside its write root.
- ui-egui `language_packs`: File › Options › Language › Language packs (name, size, licence,
source, Install/Remove, progress, errors), the missing-pack bar (dismissible per document,
never for a language without a pack), About › Language data (full licence texts).
- CP4.4: `cargo xtask language` (in `ci`): the default build embeds only en-* Language data.
- CP4.5: docs (help, MCP, control protocol, gaps #8, localization parity, roadmap).
- Path deps to Lane 4's worktrees while developing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
craft-proof moves to storytold/craft-libs#483's head (e6e948f, on #482's
9e5862a): each English checks against its own SCOWL rel-2026.02.25 dictionary
through the affix engine (ordinals via COMPOUNDRULE), with en-variants.txt as
the dialect policy. Lookup order: dictionaries folder, [English pack], SCOWL,
system Hunspell, the built-in Moby list; Moby stays for hyphenation.
craft-i18n.rev and the Flatpak source move to storytold/craft-i18n#109's merge
commit (efb1127), which adds the SCOWL files.
Typed-text tests pin British spellings in en-GB (#40), modern words (#520),
Canadian/Australian/NZ/Indian English (#1281, #1045) and ordinals; they skip
without the Language data, as the other data tests do. Docs: typing-parity,
localization-parity, ROADMAP progress log.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
CodeRabbit on #5202:
- AGENTS.md, ATTRIBUTION.md, README.md, docs/architecture.md: without
CRAFT_I18N_DIR English spelling falls back to an installed Hunspell English
or craft-proof's built-in Moby list, and only pattern hyphenation is off
(Bulgarian, Turkish and Russian hyphenate by rule); they said English
wasn't spell-checked. (CLAUDE.md carries the same sentence; left for the
maintainer, as agents don't edit it.)
- NOTICE: craft-i18n's patterns are language-data/<lang>/hyphenation.pat,
not en-us.pat, pl.pat, pt.pat, de.pat, it.pat.
- docs/spelling-dictionaries.md: the Flatpak's dictionaries folder is
~/.var/app/ai.storyteller.wordcraft/data/wordcraft/dictionaries; the host's
folders are outside its sandbox.
- printing.yml: the repository checkout doesn't persist credentials.
- tests.rs no_language_data: also probes the American/British variant table,
so a craft-i18n checkout missing some files skips the dialect tests
instead of failing them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
ODT cell formulas (`table:formula`) were lost, and only their value was
kept as text.
- Reading: LibreOffice's formulas, in its own syntax
(`ooow:sum <A1:A3>`, `<B1>*<B2>+<B3>`, `mean`, `min`, `max`,
`product`, `abs`, `sign`, `count`, `add`/`sub`/`mul`/`div`/`pow`, lists
joined by `|`), open as Word formula fields (`=SUM(A1:A3)`) showing
the cell's value. Update Field recalculates them.
- Writing: formula fields in cells are written back as cell formulas
with their value. ABOVE, LEFT, BELOW and RIGHT become the cells they
name.
- Cells holding a plain number are written as numbers, so LibreOffice
calculates with them. The `ooow` prefix is declared; without it,
LibreOffice reports "Expression is faulty".
A formula one side can't say keeps its value as text. A LibreOffice
table of sums round-trips with the same text, and the saved file
validates against the schema.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
ODT citations (`text:bibliography-mark`) were read as their text, and
the sources they carry were lost. So an ODT bibliography had nothing
to update from.
- Reading: each mark is now a `CITATION` field of its source, which
becomes one of the document's bibliography sources (the first mark
of an identifier gives it). The identifier is the tag, the type the
kind, plus author, title, year, publisher, address, journal, volume,
pages, URL, editor, edition, number and month.
- Writing: a `CITATION` of a known source is written back as a mark
carrying the source.
References › Update rebuilds an ODT bibliography from them. A
LibreOffice citation and bibliography round-trip pixel-identical and
validate against the schema.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
ODT custom document properties (`meta:user-defined` in `meta.xml`) were
dropped, and the fields showing them or a user field were read as
plain text.
- Properties: `meta:user-defined` values (string, number, true/false,
date) are now WordCraft's custom document properties, and so are
user field declarations (`text:user-field-decl`).
- Fields: `text:user-defined` and `text:user-field-get` are Word's
`DOCPROPERTY` fields, which WordCraft updates from those properties.
- Writing: custom properties go back into `meta.xml`, and a
`DOCPROPERTY` naming one is written as `text:user-defined`.
A LibreOffice document with a custom property and a user field
round-trips pixel-identical, valid against the schema. Its user field
comes back as a document property.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Polish/Portuguese/Czech review commands, the custom dictionary file and the two typing squiggle
tests pass without CRAFT_I18N_DIR at 48ed6f8. English dialects, regions' English and Hyphenation
Options still need the variant table and patterns, so they keep the skip.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
- wordcraft-proof is now `package = "craft-proof"` from craft-libs (local
path for now; TODO in Cargo.toml for the git+rev pin after the craft-libs
PR merges). `wordcraft_proof::` paths are unchanged.
- crates/proof and assets/{spelling,hyphenation} deleted with their
ATTRIBUTION.md rows; one row now points at craft-i18n's Language data;
NOTICE paths updated.
- craft-i18n.rev (and the flatpak source) moved to the craft-i18n commit that
adds language-data/ (CI sets CRAFT_I18N_REQUIRED=1, so the pin must have it).
- xtask layers: proof removed. Engine tests carry their own copy of the
synthetic Czech fixture (they reached into crates/proof/src).
- Docs: architecture.md, CLAUDE.md/AGENTS.md layer table, README build note
(CRAFT_I18N_DIR), localization-tr path, ROADMAP progress log.
Before/after wordcraft-cli on a 17-language proofing document: issues,
suggestions, manual hyphenation, PDF text and PDF bytes identical.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
ODT index marks (`text:alphabetical-index-mark`) were dropped. So
References › Update Index on an opened ODT index found no entries and
left the old list in place. Now:
- a mark at a point is Word's `XE` field, its `text:key1`/`text:key2`
the levels above its entry;
- a mark over text (`-mark-start` … `-mark-end`, in one paragraph) is
an `XE` field whose entry is the marked text, which stays text;
- `XE` fields are written back as point marks.
Engine: Update rebuilt only lists in WordCraft's own form, a paragraph
holding the field. A list held as a field around its entries, as ODT
and Word files hold an index, bibliography or table of figures, is now
first made the field's own paragraph, then rebuilt. That happens only
when the field starts its paragraph, so no document text is replaced.
A LibreOffice document with an index mark round-trips pixel-identical,
the mark kept.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Tests and notes for rules that #4129, #4113, #4109 and #4101 added and
left uncovered:
- A border line at the top of a story with formatting left out of the
Track Changes markup is not a formatting change, and Reject All keeps
it.
- The author's own earlier formatting change on that paragraph stays the
one change.
- Reject All keeps the table of a "+---+" line.
- A border line that a content control's lock stops Backspace from
deleting stays as it is, and Enter is refused, untracked too.
- With tab indents on, Tab in a table cell moves to the next cell; with
them off, Backspace before an indented paragraph joins it.
- AutoCorrect and the spelling check's web and e-mail addresses after a
narrow no-break space and a figure space.
docs/typing-parity.md describes each, and lists a no-break space typed
after a word, which does not finish the word for AutoCorrect, as a
known gap.
ODT form controls (ODF 1.3 §13: controls in `office:forms`, each
placed in the text by a `draw:control`) were dropped on open, values
and all. They now open as WordCraft content controls, each with its
value, entries and selection; its name becomes the tag and its label
the title:
- text fields and text areas: plain text controls;
- check boxes: check boxes;
- radio buttons: check boxes with round symbols;
- list boxes: drop-down lists;
- combo boxes: combo boxes;
- date fields: date controls.
Saved as DOCX they become Word content controls.
Content controls of those kinds in the body are written back as form
controls with a `draw:control` anchor; in headers and footers they stay
their text. LibreOffice shows the round-tripped controls with their
values, and the content validates against the ODF 1.3 schema.
Not read: buttons, images, hidden and other controls. Controls' sizes
and looks aren't kept.
Flow model: new `Inline::Control`, kept only by flows that ask for
boxes (ODT); other formats get its text.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
A password-protected ODT (ODF 1.3 Part 2 §3.4) failed to open. The
package is now decrypted with the password into a plain one, which the
reader opens as before.
- Supported: AES-256-CBC with a SHA-256 or SHA-1 start key, PBKDF2
with HMAC-SHA-1, and SHA-256 or SHA-1 1K checksums (LibreOffice's
encryption).
- Blowfish (the old default) is refused with a message naming it.
- A missing or wrong password gives the same errors as Word files, so
the app asks for the password.
- Hostile parameters are bounded: iteration count, IV size, ragged
data, inflated size.
Saving with a password (and saving a document opened with one) writes
the ODT encrypted as LibreOffice does. Every file but the `mimetype`
and manifest is deflated and encrypted, each with its own salt and IV,
and the manifest validates against the ODF manifest schema.
LibreOffice opens WordCraft's encrypted files with the password and
refuses a wrong one; WordCraft opens LibreOffice's.
No new dependencies: the AES, SHA, HMAC, miniz and getrandom crates
the workspace already uses for Word encryption.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
LibreOffice writes an embedded object's replacement picture in its own
metafile format, which WordCraft can't decode. So an embedded
spreadsheet (`draw:object`, ODF 1.3 §10.4.6) or OLE object
(`draw:object-ole`, §10.4.7) vanished on open, and saving lost it.
Such objects are now kept whole: their files, media type (from the
manifest or the sub-document's body) and replacement picture.
- Display: the replacement picture when WordCraft can show it, else a
placeholder of the object's size drawn in code (ruled for a
spreadsheet).
- ODT save: written back as the same object, with its manifest entries
and original replacement. LibreOffice renders a round-tripped
spreadsheet pixel-identically.
- Document model: the files live in `passthrough` under `odf-object/`,
linked from the picture's `ole` by a marker. The DOCX writer skips
the marker and writes the picture alone.
Flat ODT, a fix to the chart commit: inline sub-documents lacked the
required `office:mimetype`, so the schema rejected flat files holding
charts. Sub-documents now go inline as full `office:document`s (media
type, styles, settings) and are split back into their files on
opening. Replacement pictures and OLE files go inline as base64. A
picture's data is matched to the element that holds it, no longer to
the last `draw:image` before it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Review of the chart commit: the engine drew opened charts only at the
paragraph level, so a chart that was a member of a `draw:g` group
opened blank. The pass now goes into groups (bounded depth).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
A chart object (`draw:object` whose sub-document is an `office:chart`,
ODF 1.3 §10.4.3, Part 3 §11) used to open as its replacement picture.
It is now read into WordCraft's chart model: column, bar (stacked or
not), line, area, pie, doughnut and scatter charts, with categories,
series names and values (gaps included) taken from the chart's own
table through the series' cell range addresses, plus title, legend
position and data labels. The engine draws opened charts in the
document's theme colours, and Chart Design's commands edit them.
Charts WordCraft can't edit as they are (3D, percent-stacked, mixed
plot types, a second value axis, radar and the like) keep their
picture.
WordCraft charts are written as ODF chart objects shaped like
LibreOffice's: a data table, each series and slice in its accent
colour, pies clockwise from the top. They validate against the ODF 1.3
schema, and LibreOffice draws them like WordCraft does. In flat
documents they are written inline as an `office:document`.
Bug fixed: flat ODT reading ended the body at the first nested
`</office:body>`, so an inline sub-document dropped the text after it.
Element lookup now follows nesting, with a bounded depth.
Flow model: new `Inline::Chart`, kept only by flows that ask for
boxes (ODT); other formats still get the alt text.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
A flat ODT holds the whole document in one `office:document` element
(ODF 1.3 Part 3 §2.1.2), its pictures inline as base64
`office:binary-data` and its formulas inline in their `draw:object`.
It is opened by taking it apart into a package's files (content,
styles, meta, settings; pictures and formulas as entries, each part's
root keeping the document's namespaces) and reading that package, and
saved by putting a package's files together: one set of automatic
styles, every part's namespaces declared once on the root, pictures and
formulas inline without the link attributes.
`.fodt` opens and saves from Open, Save As, the CLI and the web app.
LibreOffice's own flat documents round-trip as their packages do, and
the written files validate against the ODF 1.3 schema.
The parity doc's ODT fidelity is re-estimated at ~80%, and its status
line, which hadn't been updated since the Asian fonts change (a text
replacement that matched nothing), now lists every change since.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Fields that were read as plain text now open as Word's, so Update Field
recomputes them, and are written back as ODF's:
- dates and times in their date or time style (`number:date-style`,
`number:time-style`) ↔ `DATE`/`TIME \@ "picture"`: day, month (by
number or name), year, weekday, hours (12-hour with AM/PM), minutes,
seconds and the text between, both ways; data styles are written as
automatic styles and the `number` namespace declared;
- sequences numbering figures and tables (`text:sequence`) ↔ `SEQ` with
its number format; on save the sequences are declared
(`text:sequence-decls`); on open a bookmark named by the sequence's
reference name covers the caption's label and number ("Figure 1"),
what references to it show;
- chapter fields (`text:chapter`) ↔ `STYLEREF "Heading n"`, by name or
number (`\n`);
- reference marks as bookmarks, and references to them and to sequences
(`text:reference-ref`, `text:sequence-ref`) as `REF`/`PAGEREF`.
Captions renumber, references to them and chapter fields refresh after
Update Field (engine test `odt_fields.rs`). A LibreOffice document with
all of them round-trips 13 px apart with the same text; the written
content and styles validate against the ODF 1.3 schema.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Word's date-time pictures (`DATE \@ "…"`) put literal text in single
quotes, two quotes in a row standing for one: `H' o''clock'` shows
"8 o'clock". The formatter read the quotes as text and the letters
inside as instructions (an `a` became AM or PM). Quoted text is now
shown as it is; an unclosed quote runs to the end.
Regression test: `date_pictures_keep_quoted_text`.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
A bookmark over text (`text:bookmark-start`…`text:bookmark-end`, in one
paragraph or across several) was read as a point at its start, and a
Word bookmark over text was written as a point: a reference to it then
updated to nothing. The flow model gets the end of a bookmark
(`Inline::AnchorEnd`, in flows asked for fields, as ODT's is), the
document gets the bookmark's start and end where they are, and the
writer writes a start and an end (a point, `text:bookmark`, when it
ends where it starts). A bookmark started twice, ended twice or ended
before it starts is kept once, as a point when it has no end after it.
References to bookmarks, LibreOffice's references to headings included,
now update to the bookmarked text (engine test `odt_fields.rs`, through
a save and reopen).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC
Tables of contents, tables of figures (`text:illustration-index`,
`text:table-index`), alphabetical indexes and bibliographies open as
WordCraft's: Word's `TOC` (levels from the source's
`text:outline-level`; `\c` and the caption name for figures), `INDEX`
and `BIBLIOGRAPHY` fields over their entries as last generated. With a
title the field goes in the title paragraph, as References › Table of
Contents inserts one; without, it spans the entries (new flow inlines
`FieldStart`/`FieldEnd`, mirroring the document's range fields). Update
Table then regenerates them from the headings (engine test
`odt_toc.rs`, titled and untitled, through a save and reopen).
LibreOffice's Contents 1–10, Contents Heading and Figure Index 1 styles
are WordCraft's TOC 1–9, TOC Heading and Table of Figures, both ways,
also when a file names them without defining them. WordCraft's tables
are written as ODF indexes with their source and entry templates, so
LibreOffice updates them, the entries' links in the Index Link style.
User and object indexes are read as their paragraphs; index marks are
not read yet.
Found on the way, both shown by the table of contents' round trip:
- a style based on LibreOffice's generic Heading style (as its headings
and index headings are) lost that parent on save, and with it its
font and spacing;
- a paragraph's tab stops were measured from its own indent only, not
from an indent its style gives (LibreOffice's `TabsRelativeToIndent`),
and a named style's stops not from its indent at all.
A LibreOffice document with a table of contents, a table of figures and
an alphabetical index round-trips pixel for pixel (it was 29,717 px
apart and a page short). The parity doc's ODT fidelity is re-estimated
at ~72%.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBVQLyDnrG8Tn1yDQcUabC