aleleba/wordcraft
1
0
Fork 0
mirror of https://github.com/storytold/wordcraft.git synced 2026-10-30 05:17:25 -06:00
Code Issues Packages Projects Releases Wiki Activity

#3735: docs — automation roots, type_text limit, Copy Styles under protection

Browse Source 
Progress log row for group J (findings 9, 18, 34, 41, 50, 54, 69, 78);
docs/mcp.md says what the roots now cover; the Restrict Editing row notes
Copy Styles.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Martin Ball
2026-10-23 20:19:49 +01:00
co-authored by Claude Opus 5.5
parent c7dfd92ff1
commit 4f5e596c71
2 changed files with 10 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/file-format-parity.md
+2 -1
View File
@@ -93,7 +93,7 @@ Measured by grepping `crates/docx/src/read` and `src/write` for the OOXML elemen
| Fields (`w:fldChar`, `w:fldSimple`), TOC, cross-references, `ADDIN` citations | ✅ | ✅ | Field codes Word supports but we don't evaluate keep their cached result. Textual `REF` fields with `\h` export internal PDF links without refreshing cached text; existing enclosing links take precedence. `wordcraft-cli convert --update-fields` explicitly runs field updates before saving; ordinary conversion preserves caches and locked fields stay unchanged. Instruction-only paragraphs carry hidden marks instead of blank lines; an empty result paragraph keeps its line. Imported TOCs preserve actual result boundaries, formatted entries and nested fields without relying on TOC styles. `EQ \o` (overstrike, enclosed characters) is drawn; enclosed characters are written as Word stores them, their style in the code's run sizes (#297); a cached picture (`INCLUDEPICTURE`) stays as the content of a range field (#465). A `HYPERLINK` field reads as a link, its result ordinary text (tabs, nested `PAGEREF`), and is written back as `w:hyperlink`; its tooltip (`\o`) and target frame (`\t`) are dropped (#2174) |
| Move tracking (`w:moveFrom`/`w:moveTo`, `w:moveFromRangeStart/End`, `w:moveToRangeStart/End`), moved paragraph marks | ✅ | ✅ | #400: both ends linked by the range name; ranges rewritten around each moved stretch (ids renumbered); a move missing an end or a range end opens as a plain deletion/insertion. ODT/PDF export the result (moved-away text hidden like deletions); RTF writes a move as a plain tracked deletion and insertion |
| Content controls (`w:sdt`) | ✅ | ✅ | Block, inline, row and cell level, nested; title, tag, id, lock, placeholder, temporary, rich/plain text, `w14:checkbox`, combo box, drop-down list, date, picture, building block gallery, `w15` repeating sections; other `w:sdtPr` children (data binding, appearance) kept verbatim (#381). A TOC's control is rebuilt by the TOC when the TOC has the heading-and-entries form; around any other TOC it is kept as read (#2427). Replace All respects content-edit and control-deletion locks and refuses a batch before making any edit |
| Restrict Editing (`w:documentProtection` edit/formatting/enforcement and password hash, `w:locked` styles, `w:permStart`/`w:permEnd` exceptions) | ✅ | ✅ | #412: SHA-512 hash with salt and spin count written; Word's legacy `w:cryptAlgorithmSid`/`w:hash` form (SHA-1/-256/-384/-512, ECMA-376 Part 4 §14.8.1) is kept and verified; MD2/MD4/MD5, `w:algIdExt` and other unsupported forms are refused, never unlocked; `w:colFirst`/`w:colLast` table-column ranges read as plain ranges; forms mode (`w:edit="forms"`) lets edits inside fillable content controls (`w:lock` content-editable, innermost wins) run, everything else rolls back; row-wrapped controls and control deletion remain refused; `review.restrict` reports `formFields`; Find Next Region covers form fields (the pane button shows for forms documents); under active protection, `developer.properties` refuses `lockEdit`/`lockDelete` changes, as Word blocks property edits until Stop Protection; under forms protection it refuses every property, and inserting or removing content controls is refused too (Word greys out the Developer tab's control buttons and Properties and offers no Remove Content Control), while under the other modes the title, tag and placeholder stay editable; a file command that changes the open document is checked like any other, so the privacy options, the properties and the inspector's removals are refused on a protected document, and restoring a stored version keeps the protection and its password rather than the snapshot's stale ones; the protection of a protected document only changes through the explicit stop (`review.restrict {"stop": true}` — the "none" mode is the shorthand for it and doesn't stop a protected document), the session author used for new tracked changes and comments can't change through `file.setAuthor` while the document is protected (the CLI's `--author` uses the same protection check), and the post-check compares the document's settings, metadata, custom properties, list definitions, bibliography sources and the comments' resolved flags as well as its text and formatting; named exceptions match a self-declared user name, not an authenticated identity |
| Restrict Editing (`w:documentProtection` edit/formatting/enforcement and password hash, `w:locked` styles, `w:permStart`/`w:permEnd` exceptions) | ✅ | ✅ | #412: SHA-512 hash with salt and spin count written; Word's legacy `w:cryptAlgorithmSid`/`w:hash` form (SHA-1/-256/-384/-512, ECMA-376 Part 4 §14.8.1) is kept and verified; MD2/MD4/MD5, `w:algIdExt` and other unsupported forms are refused, never unlocked; `w:colFirst`/`w:colLast` table-column ranges read as plain ranges; forms mode (`w:edit="forms"`) lets edits inside fillable content controls (`w:lock` content-editable, innermost wins) run, everything else rolls back; row-wrapped controls and control deletion remain refused; `review.restrict` reports `formFields`; Find Next Region covers form fields (the pane button shows for forms documents); under active protection, `developer.properties` refuses `lockEdit`/`lockDelete` changes, as Word blocks property edits until Stop Protection; under forms protection it refuses every property, and inserting or removing content controls is refused too (Word greys out the Developer tab's control buttons and Properties and offers no Remove Content Control), while under the other modes the title, tag and placeholder stay editable; a file command that changes the open document is checked like any other, so the privacy options, the properties and the inspector's removals are refused on a protected document, and restoring a stored version keeps the protection and its password rather than the snapshot's stale ones; the protection of a protected document only changes through the explicit stop (`review.restrict {"stop": true}` — the "none" mode is the shorthand for it and doesn't stop a protected document), the session author used for new tracked changes and comments can't change through `file.setAuthor` while the document is protected (the CLI's `--author` uses the same protection check), and the post-check compares the document's settings, metadata, custom properties, list definitions, bibliography sources and the comments' resolved flags as well as its text and formatting; named exceptions match a self-declared user name, not an authenticated identity; formatting-only protection also refuses Copy Styles (`styles.copyFrom`, #3692) |
| Legacy form fields (`w:ffData`), check boxes, drop-downs | 🟡 | ✅ | The field definition is kept raw and saved back. Check boxes (`FORMCHECKBOX`) are drawn as Word draws them: a square as tall as the text's line, inset 1.16 pt, 0.72 pt border, an X when checked (`w:checked`, else `w:default`), `w:size` honoured; drop-downs (`FORMDROPDOWN`) show their selected entry; text fields (`FORMTEXT`) show their result. Not yet: filling them in, Protect Document's forms mode for legacy fields |
| DrawingML pictures inline and anchored (`wp:anchor`), wrap square/tight/through/top-bottom | ✅ | ✅ | Tight/through read and written with their `wp:wrapPolygon` (and `edited`); without one, a shape's outline is written (a picture's rectangle). `wrapText` (both sides, left, right, largest) read and written for square, tight and through |
| Legacy form fields (`w:ffData`), check boxes, drop-downs | ❌ | ❌ | |
@@ -252,6 +252,7 @@ Imported text fields use the first nonempty cached result run's formatting on th
| Date | Change | Summary |
|---|---|---|
| 2026-10-22 | trivial | Formatting-only protection refuses Copy Styles, as it does Update Styles (#3692) |
| 2026-10-23 | trivial | A relationship the markup names without the prefix is read as it wrote: an unqualified `id` on an element that carries a relationship names that relationship, and the writer keeps the attribute with the id the new package holds the part under; an ordinary `id` — a VML shape's, a drawing's — names no part (#3411 follow-up) |
| 2026-10-21 | trivial | The reader takes the relationship ids the markup refers to from the relationship attributes of the main document it has already parsed, so a document with many relationships opens in the time its part takes, and an id that only appears as text is not a reference (#3411) |
| 2026-10-21 | trivial | HTML export writes ruby (furigana, phonetic guides) as `<ruby>base<rt>reading</rt></ruby>`, in the base text's run formatting, and HTML import reads `<ruby>`/`<rb>`/`<rt>` back into one ruby object: `<rp>` fallback parentheses and the tags are not text, an unclosed element ends at the next block, and malformed ruby keeps its text. HTML states no alignment, size or raise, so the Phonetic Guide's defaults apply (#1681) |
docs/mcp.md
+8 -4
View File
@@ -63,11 +63,15 @@ config works across the suite) keep the agent to the folders you give it:
`file.newFromTemplate`, `insert.textFromFile` and `insert.object`, `insert.picture` and
`picture.change` with a `path`, `mailings.recipients` with a `path` (csv, tsv, txt, xlsx, xlsm,
ods), `review.compare` (`path` and `original`), the templates and documents `tools.updateStyles`,
`styles.organizer` and `styles.copyFrom` read. The pictures an opened HTML file names are read
from beside that file, so they are inside the read root too (a link that leads out is not followed).
`styles.organizer` and `styles.copyFrom` read, and the folders `file.options` takes
(`dictionariesFolder`, `userTemplatesFolder`). The pictures an opened HTML file names are read
from beside that file and must be inside the read root too, even when the file was opened through
a link (#3736).
- Every file a command writes must lie inside the **write root**: `save_document`, `file.save`,
`file.saveAs`, `file.rename` (the file and its new name), `file.exportPdf`, `file.exportPng`, `file.printPdf`,
`file.saveTemplate`, and `mailings.finish` with a `path`.
`file.saveTemplate`, and `mailings.finish` with a `path`. A save replaces the name it is given,
so that name must be inside the write root as well as what it leads to: a link outside the root
pointing in is refused (#3735).
- Read and write access are separate. Give the same folder twice for both; a root left out grants
none of its access (with only `--automation-read-root`, nothing can be written to disk). Without
either flag nothing changes.
@@ -108,7 +112,7 @@ files it was given rather than guarding against other software on the computer.
| `batch` | run several commands in order |
| `new_document` | blank, sample, letter, resume, report |
| `open_document` / `save_document` | docx, doc (open only), odt, rtf, html, md, tex, txt, json; save also pdf, png |
| `type_text` | type at the caret (`paragraphs: true` splits lines into paragraphs) |
| `type_text` | type at the caret (`paragraphs: true` splits lines into paragraphs: one undo step, at most 10,000 lines) |
| `select_text` | select the n-th occurrence of some text |
| `get_text` / `inspect_document` | read the document back (verify without screenshots) |
| `render_page` | a page as PNG |