Merge branch 'main' into feat/bases

This commit is contained in:
Philipinho
2026-06-20 18:38:05 +01:00
74 changed files with 4280 additions and 2397 deletions
+1
View File
@@ -2,6 +2,7 @@
"name": "@docmost/editor-ext",
"homepage": "https://docmost.com",
"private": true,
"sideEffects": false,
"scripts": {
"build": "tsc --build",
"dev": "tsc --watch"
+4
View File
@@ -33,4 +33,8 @@ export * from "./lib/status";
export * from "./lib/pdf";
export * from "./lib/page-break";
export * from "./lib/resizable-nodeview";
export {
pageNodeToDocxBuffer,
type DocxImageResolver,
} from "./lib/prosemirror-docx";
export * from "./lib/base-embed";
@@ -0,0 +1,167 @@
# `prosemirror-docx`
[![prosemirror-docx on npm](https://img.shields.io/npm/v/prosemirror-docx.svg)](https://www.npmjs.com/package/prosemirror-docx)
[![prosemirror-docx on GitHub](https://img.shields.io/github/stars/curvenote/prosemirror-docx.svg?style=social)](https://github.com/curvenote/prosemirror-docx)
[
[![MIT License](https://img.shields.io/badge/license-MIT-blue.svg)](https://github.com/curvenote/prosemirror-docx/blob/master/LICENSE)
![CI](https://github.com/curvenote/prosemirror-docx/workflows/CI/badge.svg)
Export a [prosemirror](https://prosemirror.net/) document to a Microsoft Word file, using [docx](https://docx.js.org/).
![image](https://user-images.githubusercontent.com/913249/134953610-886047eb-2a21-4929-9a53-9a29d8f6184f.png)
## Overview
`prosemirror-docx` has a similar structure to [prosemirror-markdown](https://github.com/prosemirror/prosemirror-markdown), with a `DocxSerializerState` object that you write to as you walk the document. It is a light wrapper around <https://docx.js.org/>, which actually does the export. Currently `prosemirror-docx` is write only (i.e. can export to, but cant read from `*.docx`), and has most of the basic nodes covered (see below).
[Curvenote](https://curvenote.com) uses this to export from [@curvenote/editor](https://github.com/curvenote/editor) to word docs, but this library currently only has dependence on `docx`, `prosemirror-model` and `image-dimensions` - and similar to `prosemirror-markdown`, the serialization schema can be edited externally (see `Extended usage` below).
## Basic usage
```ts
import { defaultDocxSerializer, writeDocx } from 'prosemirror-docx';
import { EditorState } from 'prosemirror-state';
import { writeFileSync } from 'fs'; // Or some other way to write a file
// Set up your prosemirror state/document as you normally do
const state = EditorState.create({ schema: mySchema });
// If there are images, we will need to preload the buffers
const opts = {
getImageBuffer(src: string) {
return anImageBuffer;
},
};
// Create a doc in memory, and then write it to disk
const wordDocument = defaultDocxSerializer.serialize(state.doc, opts);
await writeDocx(wordDocument).then((buffer) => {
writeFileSync('HelloWorld.docx', buffer);
});
```
### Advanced usage
If you need to access the underlying state and modify the final docx `Document` you can use the last argument of `serialize` to pass in a callback function that receives the `DocxSerializerState`.
This function needs to return an `IPropertiesOptions` type, ie. the config that should be passed to a `Document`. Your options will be spread with the default options, so you can override any of the defaults.
```ts
const wordDocument = defaultDocxSerializer.serialize(state.doc, opts, (state) => {
return {
numbering: {
config: state.numbering,
},
fonts: [], // embed fonts,
styles: {
paragraphStyles,
default: {
heading1: paragraphStyles[1],
},
},
};
});
```
See the [docx documentation](https://docx.js.org/#/usage/document) for more details on the options you can pass in.
## Extended usage
Instead of using the `defaultDocxSerializer` you can override or provide custom serializers.
```ts
import { DocxSerializer, defaultNodes, defaultMarks } from 'prosemirror-docx';
const nodeSerializer = {
...defaultNodes,
my_paragraph(state, node) {
state.renderInline(node);
state.closeBlock(node);
},
};
export const myDocxSerializer = new DocxSerializer(nodeSerializer, defaultMarks);
```
The `state` is the `DocxSerializerState` and has helper methods to interact with `docx`.
If the exported content includes image links that require fetching the image data, you can use asynchronous APIs. Here's a demo example:
```ts
import { DocxSerializerAsync, defaultAsyncNodes, defaultMarks } from 'prosemirror-docx';
import { EditorState } from 'prosemirror-state';
import { writeFileSync } from 'fs';
const state = EditorState.create({ schema: mySchema });
export const docxSerializer = new DocxSerializerAsync(
{
...defaultAsyncNodes,
async image(state, node) {
const { src } = node.attrs;
await state.image(src, 70, 'center', undefined, 'png');
state.closeBlock(node);
},
},
defaultMarks,
);
// If there are images, we will need to preload the buffers
const opts = {
async getImageBuffer(src: string) {
const arrayBuffer = await fetch(src).then((res) => res.arrayBuffer());
return new Uint8Array(arrayBuffer);
},
};
// Create a doc in memory, and then write it to disk
const wordDocument = docxSerializer.serializeAsync(state.doc, opts);
await writeDocx(wordDocument).then((buffer) => {
writeFileSync('HelloWorld.docx', buffer);
});
```
## Supported Nodes
- text
- paragraph
- heading (levels)
- TODO: Support numbering of headings
- blockquote
- code_block
- TODO: No styles supported
- horizontal_rule
- hard_break
- ordered_list
- unordered_list
- list_item
- image
- math
- equations (numbered & unnumbered)
- tables
Planned:
- Internal References (e.g. see Table 1)
## Supported Marks
- em
- strong
- link
- Note: this is actually treated as a node in docx, so ignored as a prosemirror mark, but supported.
- code
- subscript
- superscript
- strikethrough
- underline
- smallcaps
- allcaps
## Resources
- [Prosemirror Docs](https://prosemirror.net/docs/)
- [docx](https://docx.js.org/)
- [prosemirror-markdown](https://github.com/ProseMirror/prosemirror-markdown) - similar implementation for markdown!
@@ -0,0 +1,24 @@
// MIT - https://github.com/curvenote/prosemirror-docx/
export type { SectionConfig, SerializationState } from './types';
export type {
MarkSerializer,
NodeSerializer,
NodeSerializerAsync,
Options,
OptionsAsync,
} from './serializer';
export {
DocxSerializerStateAsync,
DocxSerializerAsync,
DocxSerializerState,
DocxSerializer,
MAX_IMAGE_WIDTH,
} from './serializer';
export {
defaultAsyncNodes,
defaultMarks,
pageNodeToDocxBuffer,
type DocxImageResolver,
} from './schema';
export { writeDocx, createDocFromState, buildDoc } from './utils';
@@ -0,0 +1,47 @@
import { AlignmentType, convertInchesToTwip, ILevelsOptions, LevelFormat } from 'docx';
import { INumbering } from './types';
function basicIndentStyle(indent: number): Pick<ILevelsOptions, 'style' | 'alignment'> {
return {
alignment: AlignmentType.START,
style: {
paragraph: {
indent: { left: convertInchesToTwip(indent), hanging: convertInchesToTwip(0.18) },
},
},
};
}
const numbered = Array(3)
.fill([LevelFormat.DECIMAL, LevelFormat.LOWER_LETTER, LevelFormat.LOWER_ROMAN])
.flat()
.map((format, level) => ({
level,
format,
text: `%${level + 1}.`,
...basicIndentStyle((level + 1) / 2),
}));
const bullets = Array(3)
.fill(['●', '○', '■'])
.flat()
.map((text, level) => ({
level,
format: LevelFormat.BULLET,
text,
...basicIndentStyle((level + 1) / 2),
}));
const styles = {
numbered,
bullets,
};
export type NumberingStyles = keyof typeof styles;
export function createNumbering(reference: string, style: NumberingStyles): INumbering {
return {
reference,
levels: styles[style],
};
}
@@ -0,0 +1,250 @@
import { HeadingLevel, ShadingType } from 'docx';
import { Node } from 'prosemirror-model';
import {
DocxSerializerAsync,
MarkSerializer,
NodeSerializerAsync,
OptionsAsync,
} from './serializer';
import { writeDocx } from './utils';
export type DocxImageResolver = OptionsAsync['getImageBuffer'];
// docx requires a 6-digit hex color (no leading #). Convert #rgb, #rrggbb,
// and rgb()/rgba() inputs to 6-digit hex; return undefined for anything else
// (named colors, hsl, etc.) so the caller omits the color rather than letting
// docx throw "Invalid hex value".
function toDocxColor(input?: string): string | undefined {
if (!input) return undefined;
const value = input.trim().toLowerCase();
const hex = value.startsWith('#') ? value.slice(1) : value;
if (/^[0-9a-f]{6}$/.test(hex)) return hex;
if (/^[0-9a-f]{3}$/.test(hex)) {
return hex
.split('')
.map((ch) => ch + ch)
.join('');
}
const rgb = value.match(/^rgba?\(\s*(\d+)\s*,\s*(\d+)\s*,\s*(\d+)/);
if (rgb) {
const channel = (n: string) =>
Math.max(0, Math.min(255, parseInt(n, 10)))
.toString(16)
.padStart(2, '0');
return channel(rgb[1]) + channel(rgb[2]) + channel(rgb[3]);
}
return undefined;
}
// Images and diagrams embed via the image resolver; the URL (with its file
// extension) is passed through so docx can infer the image type.
const renderImage: NodeSerializerAsync[string] = async (state, node) => {
const src = node.attrs?.src || node.attrs?.attachmentId;
if (src) {
try {
await state.image(src, 100);
} catch {
// Unrenderable/missing image: skip rather than fail the whole export.
}
}
state.closeBlock(node);
};
// Non-embeddable media render as a labelled line.
const renderFileLine: NodeSerializerAsync[string] = (state, node) => {
const label =
node.attrs?.name || node.attrs?.src || node.attrs?.url || 'attachment';
state.text(label);
state.closeBlock(node);
};
const renderEmbedLine: NodeSerializerAsync[string] = (state, node) => {
const label = node.attrs?.src || node.attrs?.url || 'embed';
state.text(label);
state.closeBlock(node);
};
export const defaultAsyncNodes: NodeSerializerAsync = {
text(state, node) {
state.text(node.text ?? '');
},
async paragraph(state, node) {
await state.renderInline(node);
state.closeBlock(node);
},
async heading(state, node) {
await state.renderInline(node);
const heading = [
HeadingLevel.HEADING_1,
HeadingLevel.HEADING_2,
HeadingLevel.HEADING_3,
HeadingLevel.HEADING_4,
HeadingLevel.HEADING_5,
HeadingLevel.HEADING_6,
][(node.attrs.level ?? 1) - 1];
state.closeBlock(node, { heading });
},
async blockquote(state, node) {
await state.renderContent(node, { style: 'IntenseQuote' });
},
async codeBlock(state, node) {
await state.renderContent(node);
state.closeBlock(node);
},
horizontalRule(state, node) {
state.closeBlock(node, { thematicBreak: true });
state.closeBlock(node);
},
hardBreak(state) {
state.addRunOptions({ break: 1 });
},
async bulletList(state, node) {
await state.renderList(node, 'bullets');
},
async orderedList(state, node) {
await state.renderList(node, 'numbered');
},
async listItem(state, node) {
await state.renderListItem(node);
},
async taskList(state, node) {
await state.renderList(node, 'bullets');
},
async taskItem(state, node) {
if (state.currentNumbering) {
state.addParagraphOptions({ numbering: state.currentNumbering });
}
state.text(node.attrs?.checked ? '☑ ' : '☐ ');
await state.renderContent(node);
},
async table(state, node) {
await state.table(node);
},
// Docmost stores LaTeX in attrs.text.
mathInline(state, node) {
state.math(node.attrs?.text ?? '', { inline: true });
},
mathBlock(state, node) {
state.math(node.attrs?.text ?? '', { inline: false, numbered: false });
state.closeBlock(node);
},
image: renderImage,
drawio: renderImage,
excalidraw: renderImage,
video: renderFileLine,
audio: renderFileLine,
pdf: renderFileLine,
attachment: renderFileLine,
embed: renderEmbedLine,
youtube: renderEmbedLine,
async callout(state, node) {
await state.renderContent(node, { style: 'IntenseQuote' });
},
async details(state, node) {
await state.renderContent(node);
},
async detailsSummary(state, node) {
await state.renderInline(node);
state.closeBlock(node, { heading: HeadingLevel.HEADING_4 });
},
async detailsContent(state, node) {
await state.renderContent(node);
},
async columns(state, node) {
await state.renderContent(node);
},
async column(state, node) {
await state.renderContent(node);
},
async transclusionSource(state, node) {
await state.renderContent(node);
},
mention(state, node) {
state.text(`@${node.attrs?.label ?? ''}`);
},
status(state, node) {
state.text(`[${node.attrs?.text ?? ''}]`);
},
pageBreak(state, node) {
state.closeBlock(node, { pageBreakBefore: true });
},
// No usable static export representation: skip without failing.
subpages() {},
transclusionReference() {},
};
export const defaultMarks: MarkSerializer = {
bold() {
return { bold: true };
},
italic() {
return { italics: true };
},
strike() {
return { strike: true };
},
underline() {
return { underline: {} };
},
code() {
return {
font: { name: 'Monospace' },
color: '000000',
shading: { type: ShadingType.SOLID, color: 'D2D3D2', fill: 'D2D3D2' },
};
},
superscript() {
return { superScript: true };
},
subscript() {
return { subScript: true };
},
link() {
// Handled specifically in the serializer; Word treats links as nodes.
return {};
},
highlight(_state, _node, mark) {
const fill = toDocxColor(mark.attrs?.color);
return fill
? { shading: { type: ShadingType.CLEAR, fill } }
: { highlight: 'yellow' };
},
// @tiptap/extension-color stores the color on the textStyle mark.
textStyle(_state, _node, mark) {
const color = toDocxColor(mark.attrs?.color);
return color ? { color } : {};
},
// Comments are editor-only; drop the annotation in the export.
comment() {
return {};
},
};
export async function pageNodeToDocxBuffer(
doc: Node,
getImageBuffer: DocxImageResolver,
): Promise<Buffer> {
const serializer = new DocxSerializerAsync(defaultAsyncNodes, defaultMarks);
const wordDoc = await serializer.serializeAsync(
doc,
{ getImageBuffer },
// docx's built-in heading styles are blue (#2E74B5 / #1F4D78). The editor
// has no heading color, so override the default heading run colors to the
// normal text color. Sizes/italics mirror docx's own defaults so only the
// color changes.
() =>
({
styles: {
default: {
heading1: { run: { color: '000000', size: 32 } },
heading2: { run: { color: '000000', size: 26 } },
heading3: { run: { color: '000000', size: 24 } },
heading4: { run: { color: '000000', italics: true } },
heading5: { run: { color: '000000' } },
heading6: { run: { color: '000000' } },
},
},
}) as any,
);
return writeDocx(wordDoc);
}
@@ -0,0 +1,925 @@
import { Node, Mark } from 'prosemirror-model';
import {
IParagraphOptions,
IRunOptions,
Paragraph,
TextRun,
ExternalHyperlink,
ParagraphChild,
MathRun,
Math,
TabStopType,
TabStopPosition,
SequentialIdentifier,
Bookmark,
ImageRun,
AlignmentType,
Table,
TableRow,
TableCell,
ITableCellOptions,
InternalHyperlink,
SimpleField,
FootnoteReferenceRun,
IImageOptions,
Document,
ITableOptions,
ITableRowOptions,
IPropertiesOptions,
} from 'docx';
import { imageDimensionsFromData } from 'image-dimensions';
import { createNumbering, NumberingStyles } from './numbering';
import { buildDoc, createShortId } from './utils';
import { IFootnotes, INumbering, Mutable, SectionConfig, SerializationState } from './types';
// This is duplicated from @curvenote/schema
export type AlignOptions = 'left' | 'center' | 'right';
export type NodeSerializer = Record<
string,
(state: DocxSerializerState, node: Node, parent: Node, index: number) => void
>;
export type NodeSerializerAsync = Record<
string,
(state: DocxSerializerStateAsync, node: Node, parent: Node, index: number) => void | Promise<void>
>;
export type MarkSerializer = Record<
string,
(state: DocxSerializerState | DocxSerializerStateAsync, node: Node, mark: Mark) => IRunOptions
>;
export type Options = {
getImageBuffer: (src: string) => Uint8Array;
sections?: SectionConfig[];
};
export type OptionsAsync = {
getImageBuffer: (src: string) => Uint8Array | Promise<Uint8Array>;
sections?: SectionConfig[];
};
export type IMathOpts = {
inline?: boolean;
id?: string | null;
numbered?: boolean;
};
export type ImageType = 'jpg' | 'png' | 'gif' | 'bmp';
export const MAX_IMAGE_WIDTH = 600;
function createReferenceBookmark(
id: string,
kind: 'Equation' | 'Figure' | 'Table',
before?: string,
after?: string,
) {
const textBefore = before ? [new TextRun(before)] : [];
const textAfter = after ? [new TextRun(after)] : [];
return new Bookmark({
id,
children: [...textBefore, new SequentialIdentifier(kind), ...textAfter],
});
}
export class DocxSerializerState {
nodes: NodeSerializer;
options: Options;
marks: MarkSerializer;
children: (Paragraph | Table)[];
sections: Array<{
config: SectionConfig;
children: (Paragraph | Table)[];
}>;
currentSectionIndex = 0;
numbering: INumbering[];
footnotes: IFootnotes = {};
nextRunOpts?: IRunOptions;
current: ParagraphChild[] = [];
currentLink?: { link: string; children: IRunOptions[] };
// Optionally add options
nextParentParagraphOpts?: IParagraphOptions;
currentNumbering?: { reference: string; level: number };
constructor(nodes: NodeSerializer, marks: MarkSerializer, options: Options) {
this.nodes = nodes;
this.marks = marks;
this.options = options ?? ({} as Options);
this.children = [];
this.numbering = [];
// Initialize sections
if (options.sections && options.sections.length > 0) {
this.sections = options.sections.map((config) => ({
config,
children: [],
}));
this.children = this.sections[0].children;
} else {
this.sections = [];
}
}
renderContent(parent: Node, opts?: IParagraphOptions) {
parent.forEach((node, _, i) => {
if (opts) this.addParagraphOptions(opts);
this.render(node, parent, i);
});
}
render(node: Node, parent: Node, index: number) {
if (typeof parent === 'number') throw new Error('!');
if (!this.nodes[node.type.name])
throw new Error(`Token type \`${node.type.name}\` not supported by Word renderer`);
this.nodes[node.type.name](this, node, parent, index);
}
renderMarks(node: Node, marks: Mark[]): IRunOptions {
return marks
.map((mark) => {
return this.marks[mark.type.name]?.(this, node, mark);
})
.reduce((a, b) => ({ ...a, ...b }), {});
}
renderInline(parent: Node) {
// Pop the stack over to this object when we encounter a link, and closeLink restores it
let currentLink: { link: string; stack: ParagraphChild[] } | undefined;
const closeLink = () => {
if (!currentLink) return;
const hyperlink = new ExternalHyperlink({
link: currentLink.link,
// child: this.current[0],
children: this.current,
});
this.current = [...currentLink.stack, hyperlink];
currentLink = undefined;
};
const openLink = (href: string) => {
const sameLink = href === currentLink?.link;
this.addRunOptions({ style: 'Hyperlink' });
// TODO: https://github.com/dolanmiu/docx/issues/1119
// Remove the if statement here and oneLink!
const oneLink = true;
if (!oneLink) {
closeLink();
} else {
if (currentLink && sameLink) return;
if (currentLink && !sameLink) {
// Close previous, and open a new one
closeLink();
}
}
currentLink = {
link: href,
stack: this.current,
};
this.current = [];
};
const progress = (node: Node, offset: number, index: number) => {
const links = node.marks.filter((m) => m.type.name === 'link');
const hasLink = links.length > 0;
if (hasLink) {
openLink(links[0].attrs.href);
} else if (!hasLink && currentLink) {
closeLink();
}
if (node.isText) {
this.text(node.text, this.renderMarks(node, [...node.marks]));
} else {
this.render(node, parent, index);
}
};
parent.forEach(progress);
// Must call close at the end of everything, just in case
closeLink();
}
renderList(node: Node, style: NumberingStyles) {
if (!this.currentNumbering) {
const nextId = createShortId();
this.numbering.push(createNumbering(nextId, style));
this.currentNumbering = { reference: nextId, level: 0 };
} else {
const { reference, level } = this.currentNumbering;
this.currentNumbering = { reference, level: level + 1 };
}
this.renderContent(node);
if (this.currentNumbering.level === 0) {
delete this.currentNumbering;
} else {
const { reference, level } = this.currentNumbering;
this.currentNumbering = { reference, level: level - 1 };
}
}
// This is a pass through to the paragraphs, etc. underneath they will close the block
renderListItem(node: Node) {
if (!this.currentNumbering) throw new Error('Trying to create a list item without a list?');
this.addParagraphOptions({ numbering: this.currentNumbering });
this.renderContent(node);
}
addParagraphOptions(opts: IParagraphOptions) {
this.nextParentParagraphOpts = { ...this.nextParentParagraphOpts, ...opts };
}
addRunOptions(opts: IRunOptions) {
this.nextRunOpts = { ...this.nextRunOpts, ...opts };
}
text(text: string | null | undefined, opts?: IRunOptions) {
if (!text) return;
this.current.push(new TextRun({ text, ...this.nextRunOpts, ...opts }));
delete this.nextRunOpts;
}
math(latex: string, opts: IMathOpts = { inline: true }) {
if (opts.inline || !opts.numbered) {
this.current.push(new Math({ children: [new MathRun(latex)] }));
return;
}
const id = opts.id ?? createShortId();
this.current = [
new TextRun('\t'),
new Math({
children: [new MathRun(latex)],
}),
new TextRun('\t('),
createReferenceBookmark(id, 'Equation'),
new TextRun(')'),
];
this.addParagraphOptions({
tabStops: [
{
type: TabStopType.CENTER,
position: TabStopPosition.MAX / 2,
},
{
type: TabStopType.RIGHT,
position: TabStopPosition.MAX,
},
],
});
}
// not sure what this actually is, seems to be close for 8.5x11
maxImageWidth = MAX_IMAGE_WIDTH;
image(
src: string,
widthPercent = 70,
align: AlignOptions = 'center',
imageRunOpts?: IImageOptions,
imageType?: ImageType,
) {
const buffer = this.options.getImageBuffer(src);
const dimensions = imageDimensionsFromData(buffer);
/* If the image is not a valid image, don't add it */
if (!dimensions) return;
const aspect = dimensions.height / dimensions.width;
const width = this.maxImageWidth * (widthPercent / 100);
let it;
try {
it = imageType || (src.replace(/.*\./, '').toLowerCase() as any);
} catch (e) {
it = 'png';
}
this.current.push(
new ImageRun({
data: buffer,
...imageRunOpts,
type: it,
transformation: {
...(imageRunOpts?.transformation || {}),
width,
height: width * aspect,
},
}),
);
let alignment: string;
switch (align) {
case 'right':
alignment = AlignmentType.RIGHT;
break;
case 'left':
alignment = AlignmentType.LEFT;
break;
default:
alignment = AlignmentType.CENTER;
}
this.addParagraphOptions({
alignment: alignment as any,
});
}
table(
node: Node,
opts: {
getCellOptions?: (cell: Node) => ITableCellOptions;
getRowOptions?: (row: Node) => Omit<ITableRowOptions, 'children'>;
tableOptions?: Omit<ITableOptions, 'rows'>;
} = {},
) {
const { getCellOptions, getRowOptions, tableOptions } = opts;
const actualChildren = this.children;
const rows: TableRow[] = [];
node.content.forEach((row) => {
const cells: TableCell[] = [];
// Check if all cells are headers in this row
let tableHeader = true;
row.content.forEach((cell) => {
if (cell.type.name !== 'tableHeader') {
tableHeader = false;
}
});
// This scales images inside of tables
this.maxImageWidth = MAX_IMAGE_WIDTH / row.content.childCount;
row.content.forEach((cell) => {
this.children = [];
this.renderContent(cell);
const tableCellOpts: Mutable<ITableCellOptions> = { children: this.children };
const colspan = cell.attrs.colspan ?? 1;
const rowspan = cell.attrs.rowspan ?? 1;
if (colspan > 1) tableCellOpts.columnSpan = colspan;
if (rowspan > 1) tableCellOpts.rowSpan = rowspan;
cells.push(
new TableCell({
...tableCellOpts,
...(getCellOptions?.(cell) || {}),
}),
);
});
rows.push(new TableRow({ ...(getRowOptions?.(row) || {}), children: cells, tableHeader }));
});
this.maxImageWidth = MAX_IMAGE_WIDTH;
const table = new Table({ ...tableOptions, rows });
actualChildren.push(table);
// If there are multiple tables, this seperates them
actualChildren.push(new Paragraph(''));
this.children = actualChildren;
}
captionLabel(id: string, kind: 'Figure' | 'Table', { suffix } = { suffix: ': ' }) {
this.current.push(...[createReferenceBookmark(id, kind, `${kind} `), new TextRun(suffix)]);
}
$footnoteCounter = 0;
footnote(node: Node) {
const { current, nextRunOpts } = this;
// Delete everything and work with the footnote inline on the current
this.current = [];
delete this.nextRunOpts;
this.$footnoteCounter += 1;
this.renderInline(node);
this.footnotes[this.$footnoteCounter] = {
children: [new Paragraph({ children: this.current })],
};
this.current = current;
this.nextRunOpts = nextRunOpts;
this.current.push(new FootnoteReferenceRun(this.$footnoteCounter));
}
closeBlock(node: Node, props?: IParagraphOptions) {
const paragraph = new Paragraph({
children: this.current,
...this.nextParentParagraphOpts,
...props,
});
this.current = [];
delete this.nextParentParagraphOpts;
this.children.push(paragraph);
}
/**
* Move to the next section. If no more sections are available,
* this will be ignored (content continues in current section).
*/
nextSection() {
if (this.currentSectionIndex < this.sections.length - 1) {
this.currentSectionIndex += 1;
this.children = this.sections[this.currentSectionIndex].children;
}
}
/**
* Update the current section's configuration
*/
setSectionConfig(config: Partial<SectionConfig>) {
this.sections[this.currentSectionIndex].config = {
...this.sections[this.currentSectionIndex].config,
...config,
};
}
/**
* Add a new section with the given configuration and switch to it
*/
addSection(config: SectionConfig = {}) {
this.sections.push({
config,
children: [],
});
this.currentSectionIndex = this.sections.length - 1;
this.children = this.sections[this.currentSectionIndex].children;
}
/**
* Get the current section index
*/
getCurrentSectionIndex(): number {
return this.currentSectionIndex;
}
/**
* Get the current section configuration
*/
getCurrentSectionConfig(): SectionConfig {
return this.sections[this.currentSectionIndex].config;
}
/**
* Get the current serialization state for document creation
*/
getSerializationState(): SerializationState {
return {
numbering: this.numbering,
sections: this.sections,
footnotes: this.footnotes,
};
}
createReference(id: string, before?: string, after?: string) {
const children: ParagraphChild[] = [];
if (before) children.push(new TextRun(before));
children.push(new SimpleField(`REF ${id} \\h`));
if (after) children.push(new TextRun(after));
const ref = new InternalHyperlink({ anchor: id, children });
this.current.push(ref);
}
}
export class DocxSerializer {
nodes: NodeSerializer;
marks: MarkSerializer;
constructor(nodes: NodeSerializer, marks: MarkSerializer) {
this.nodes = nodes;
this.marks = marks;
}
serialize(
content: Node,
options: Options,
getDocumentOptions?: (state: SerializationState) => IPropertiesOptions,
): Document {
const state = new DocxSerializerState(this.nodes, this.marks, options);
state.renderContent(content);
return buildDoc(state, getDocumentOptions?.(state));
}
}
export class DocxSerializerStateAsync {
nodes: NodeSerializerAsync;
options: OptionsAsync;
marks: MarkSerializer;
children: (Paragraph | Table)[];
sections: Array<{
config: SectionConfig;
children: (Paragraph | Table)[];
}>;
currentSectionIndex = 0;
numbering: INumbering[];
footnotes: IFootnotes = {};
nextRunOpts?: IRunOptions;
current: ParagraphChild[] = [];
currentLink?: { link: string; children: IRunOptions[] };
// Optionally add options
nextParentParagraphOpts?: IParagraphOptions;
currentNumbering?: { reference: string; level: number };
constructor(nodes: NodeSerializerAsync, marks: MarkSerializer, options: OptionsAsync) {
this.nodes = nodes;
this.marks = marks;
this.options = options ?? ({} as OptionsAsync);
this.children = [];
this.numbering = [];
// Initialize sections
if (options.sections && options.sections.length > 0) {
this.sections = options.sections.map((config) => ({
config,
children: [],
}));
this.children = this.sections[0].children;
} else {
this.sections = [];
}
}
async renderContent(parent: Node, opts?: IParagraphOptions) {
for (let i = 0; i < parent.childCount; i += 1) {
const node = parent.child(i);
if (opts) this.addParagraphOptions(opts);
// eslint-disable-next-line no-await-in-loop
await this.render(node, parent, i);
}
}
async render(node: Node, parent: Node, index: number) {
if (typeof parent === 'number') throw new Error('!');
if (!this.nodes[node.type.name])
throw new Error(`Token type \`${node.type.name}\` not supported by Word renderer`);
await Promise.resolve(this.nodes[node.type.name](this, node, parent, index));
}
renderMarks(node: Node, marks: Mark[]): IRunOptions {
return marks
.map((mark) => {
return this.marks[mark.type.name]?.(this, node, mark);
})
.reduce((a, b) => ({ ...a, ...b }), {});
}
async renderInline(parent: Node) {
// Pop the stack over to this object when we encounter a link, and closeLink restores it
let currentLink: { link: string; stack: ParagraphChild[] } | undefined;
const closeLink = () => {
if (!currentLink) return;
const hyperlink = new ExternalHyperlink({
link: currentLink.link,
// child: this.current[0],
children: this.current,
});
this.current = [...currentLink.stack, hyperlink];
currentLink = undefined;
};
const openLink = (href: string) => {
const sameLink = href === currentLink?.link;
this.addRunOptions({ style: 'Hyperlink' });
// TODO: https://github.com/dolanmiu/docx/issues/1119
// Remove the if statement here and oneLink!
const oneLink = true;
if (!oneLink) {
closeLink();
} else {
if (currentLink && sameLink) return;
if (currentLink && !sameLink) {
// Close previous, and open a new one
closeLink();
}
}
currentLink = {
link: href,
stack: this.current,
};
this.current = [];
};
const progress = async (node: Node, offset: number, index: number) => {
const links = node.marks.filter((m) => m.type.name === 'link');
const hasLink = links.length > 0;
if (hasLink) {
openLink(links[0].attrs.href);
} else if (!hasLink && currentLink) {
closeLink();
}
if (node.isText) {
this.text(node.text, this.renderMarks(node, [...node.marks]));
} else {
await this.render(node, parent, index);
}
};
// Process nodes sequentially to maintain order
for (let i = 0; i < parent.childCount; i += 1) {
// eslint-disable-next-line no-await-in-loop
await progress(parent.child(i), 0, i);
}
// Must call close at the end of everything, just in case
closeLink();
}
async renderList(node: Node, style: NumberingStyles) {
if (!this.currentNumbering) {
const nextId = createShortId();
this.numbering.push(createNumbering(nextId, style));
this.currentNumbering = { reference: nextId, level: 0 };
} else {
const { reference, level } = this.currentNumbering;
this.currentNumbering = { reference, level: level + 1 };
}
await this.renderContent(node);
if (this.currentNumbering.level === 0) {
delete this.currentNumbering;
} else {
const { reference, level } = this.currentNumbering;
this.currentNumbering = { reference, level: level - 1 };
}
}
// This is a pass through to the paragraphs, etc. underneath they will close the block
async renderListItem(node: Node) {
if (!this.currentNumbering) throw new Error('Trying to create a list item without a list?');
this.addParagraphOptions({ numbering: this.currentNumbering });
await this.renderContent(node);
}
addParagraphOptions(opts: IParagraphOptions) {
this.nextParentParagraphOpts = { ...this.nextParentParagraphOpts, ...opts };
}
addRunOptions(opts: IRunOptions) {
this.nextRunOpts = { ...this.nextRunOpts, ...opts };
}
text(text: string | null | undefined, opts?: IRunOptions) {
if (!text) return;
this.current.push(new TextRun({ text, ...this.nextRunOpts, ...opts }));
delete this.nextRunOpts;
}
math(latex: string, opts: IMathOpts = { inline: true }) {
if (opts.inline || !opts.numbered) {
this.current.push(new Math({ children: [new MathRun(latex)] }));
return;
}
const id = opts.id ?? createShortId();
this.current = [
new TextRun('\t'),
new Math({
children: [new MathRun(latex)],
}),
new TextRun('\t('),
createReferenceBookmark(id, 'Equation'),
new TextRun(')'),
];
this.addParagraphOptions({
tabStops: [
{
type: TabStopType.CENTER,
position: TabStopPosition.MAX / 2,
},
{
type: TabStopType.RIGHT,
position: TabStopPosition.MAX,
},
],
});
}
// not sure what this actually is, seems to be close for 8.5x11
maxImageWidth = MAX_IMAGE_WIDTH;
async image(
src: string,
widthPercent = 70,
align: AlignOptions = 'center',
imageRunOpts?: IImageOptions,
imageType?: ImageType,
) {
const buffer = await Promise.resolve(this.options.getImageBuffer(src));
const dimensions = imageDimensionsFromData(buffer);
/* If the image is not a valid image, don't add it */
if (!dimensions) return;
const aspect = dimensions.height / dimensions.width;
const width = this.maxImageWidth * (widthPercent / 100);
let it;
try {
it = imageType || (src.replace(/.*\./, '').toLowerCase() as any);
} catch (e) {
it = 'png';
}
this.current.push(
new ImageRun({
data: buffer,
...imageRunOpts,
type: it,
transformation: {
...(imageRunOpts?.transformation || {}),
width,
height: width * aspect,
},
}),
);
let alignment: string;
switch (align) {
case 'right':
alignment = AlignmentType.RIGHT;
break;
case 'left':
alignment = AlignmentType.LEFT;
break;
default:
alignment = AlignmentType.CENTER;
}
this.addParagraphOptions({
alignment: alignment as any,
});
}
async table(
node: Node,
opts: {
getCellOptions?: (cell: Node) => ITableCellOptions;
getRowOptions?: (row: Node) => Omit<ITableRowOptions, 'children'>;
tableOptions?: Omit<ITableOptions, 'rows'>;
} = {},
) {
const { getCellOptions, getRowOptions, tableOptions } = opts;
const actualChildren = this.children;
const rows: TableRow[] = [];
for (let rowIndex = 0; rowIndex < node.content.childCount; rowIndex += 1) {
const row = node.content.child(rowIndex);
const cells: TableCell[] = [];
// Check if all cells are headers in this row
let tableHeader = true;
// Check if all cells in the row are headers
for (let cellIndex = 0; cellIndex < row.content.childCount; cellIndex += 1) {
const cell = row.content.child(cellIndex);
if (cell.type.name !== 'tableHeader') {
tableHeader = false;
}
}
// This scales images inside of tables
this.maxImageWidth = MAX_IMAGE_WIDTH / row.content.childCount;
// Iterate through cells and ensure order
for (let cellIndex = 0; cellIndex < row.content.childCount; cellIndex += 1) {
const cell = row.content.child(cellIndex);
this.children = [];
// eslint-disable-next-line no-await-in-loop
await this.renderContent(cell); // Ensure order
const tableCellOpts: Mutable<ITableCellOptions> = { children: this.children };
const colspan = cell.attrs.colspan ?? 1;
const rowspan = cell.attrs.rowspan ?? 1;
if (colspan > 1) tableCellOpts.columnSpan = colspan;
if (rowspan > 1) tableCellOpts.rowSpan = rowspan;
cells.push(
new TableCell({
...tableCellOpts,
...(getCellOptions?.(cell) || {}),
}),
);
}
rows.push(new TableRow({ ...(getRowOptions?.(row) || {}), children: cells, tableHeader }));
}
this.maxImageWidth = MAX_IMAGE_WIDTH;
const table = new Table({ ...tableOptions, rows });
actualChildren.push(table);
// If there are multiple tables, this separates them
actualChildren.push(new Paragraph(''));
this.children = actualChildren;
}
captionLabel(id: string, kind: 'Figure' | 'Table', { suffix } = { suffix: ': ' }) {
this.current.push(...[createReferenceBookmark(id, kind, `${kind} `), new TextRun(suffix)]);
}
$footnoteCounter = 0;
async footnote(node: Node) {
const { current, nextRunOpts } = this;
// Delete everything and work with the footnote inline on the current
this.current = [];
delete this.nextRunOpts;
this.$footnoteCounter += 1;
await this.renderInline(node);
this.footnotes[this.$footnoteCounter] = {
children: [new Paragraph({ children: this.current })],
};
this.current = current;
this.nextRunOpts = nextRunOpts;
this.current.push(new FootnoteReferenceRun(this.$footnoteCounter));
}
closeBlock(node: Node, props?: IParagraphOptions) {
const paragraph = new Paragraph({
children: this.current,
...this.nextParentParagraphOpts,
...props,
});
this.current = [];
delete this.nextParentParagraphOpts;
this.children.push(paragraph);
}
/**
* Move to the next section. If no more sections are available,
* this will be ignored (content continues in current section).
*/
nextSection() {
if (this.currentSectionIndex < this.sections.length - 1) {
this.currentSectionIndex += 1;
this.children = this.sections[this.currentSectionIndex].children;
}
}
/**
* Update the current section's configuration
*/
setSectionConfig(config: Partial<SectionConfig>) {
this.sections[this.currentSectionIndex].config = {
...this.sections[this.currentSectionIndex].config,
...config,
};
}
/**
* Add a new section with the given configuration and switch to it
*/
addSection(config: SectionConfig = {}) {
this.sections.push({
config,
children: [],
});
this.currentSectionIndex = this.sections.length - 1;
this.children = this.sections[this.currentSectionIndex].children;
}
/**
* Get the current section index
*/
getCurrentSectionIndex(): number {
return this.currentSectionIndex;
}
/**
* Get the current section configuration
*/
getCurrentSectionConfig(): SectionConfig {
return this.sections[this.currentSectionIndex].config;
}
/**
* Get the current serialization state for document creation
*/
getSerializationState(): SerializationState {
return {
numbering: this.numbering,
sections: this.sections,
footnotes: this.footnotes,
};
}
createReference(id: string, before?: string, after?: string) {
const children: ParagraphChild[] = [];
if (before) children.push(new TextRun(before));
children.push(new SimpleField(`REF ${id} \\h`));
if (after) children.push(new TextRun(after));
const ref = new InternalHyperlink({ anchor: id, children });
this.current.push(ref);
}
}
export class DocxSerializerAsync {
nodes: NodeSerializerAsync;
marks: MarkSerializer;
constructor(nodes: NodeSerializerAsync, marks: MarkSerializer) {
this.nodes = nodes;
this.marks = marks;
}
async serializeAsync(
content: Node,
options: OptionsAsync,
getDocumentOptions?: (state: SerializationState) => IPropertiesOptions,
) {
const state = new DocxSerializerStateAsync(this.nodes, this.marks, options);
await state.renderContent(content);
return buildDoc(state, getDocumentOptions?.(state));
}
}
@@ -0,0 +1,34 @@
import { INumberingOptions, Paragraph, ISectionOptions } from 'docx';
export type Mutable<T> = {
-readonly [k in keyof T]: T[k];
};
export type IFootnotes = Mutable<
Readonly<
Record<
string,
{
readonly children: readonly Paragraph[];
}
>
>
>;
export type INumbering = INumberingOptions['config'][0];
export interface SectionConfig {
properties?: ISectionOptions['properties'];
headers?: ISectionOptions['headers'];
footers?: ISectionOptions['footers'];
}
export interface SerializationState {
numbering: INumberingOptions['config'];
sections?: Array<{
config: SectionConfig;
children: ISectionOptions['children'];
}>;
children?: ISectionOptions['children'];
footnotes?: IFootnotes;
}
@@ -0,0 +1,91 @@
import {
Document,
INumberingOptions,
IPropertiesOptions,
ISectionOptions,
Packer,
SectionType,
} from 'docx';
import { Node as ProsemirrorNode } from 'prosemirror-model';
import { IFootnotes, SerializationState } from './types';
export function createShortId() {
return Math.random().toString(36).slice(2, 11);
}
export function buildDoc(state: SerializationState, opts?: IPropertiesOptions): Document {
let sections = state?.sections?.length
? state.sections.map((section) => ({
properties: section.config.properties || {
type: SectionType.CONTINUOUS,
},
headers: section.config.headers,
footers: section.config.footers,
children: section.children,
}))
: undefined;
if (!sections) {
sections = [
{
headers: undefined,
footers: undefined,
properties: {
type: SectionType.CONTINUOUS,
},
children: state?.children || [],
},
];
}
const doc = new Document({
footnotes: state.footnotes,
numbering: {
config: state.numbering,
},
sections,
...(opts || {}),
});
return doc;
}
/**
* @deprecated - use `buildDoc` instead
* Creates a docx document from the given state.
* */
export function createDocFromState(state: {
numbering: INumberingOptions['config'];
children: ISectionOptions['children'];
footnotes?: IFootnotes;
}) {
return buildDoc({
numbering: state.numbering,
sections: [
{
config: {},
children: state.children,
},
],
footnotes: state.footnotes,
});
}
export async function writeDocx(
doc: Document,
/**
* @deprecated use `.then()` or `await` instead
*/
write?: ((buffer: Buffer) => void) | ((buffer: Buffer) => Promise<void>),
) {
const buffer = await Packer.toBuffer(doc);
await write?.(buffer);
return buffer;
}
export function getLatexFromNode(node: ProsemirrorNode): string {
let math = '';
node.forEach((child) => {
if (child.isText) math += child.text;
// TODO: improve this as we may have other things in the future
});
return math;
}