reamkit 1.10.0 → 1.12.0

package/dist/esm/pptx/pptx-reader.js

@@ -3,8 +3,8 @@ import { ResourceStore } from "../core/ir/resources.js";
 import { FEATURES } from "../core/ir/features.js";
 import { EMPTY_STYLE_SHEET, resolveBodyStyles } from "../core/style-cascade/resolver.js";
 import "../core/style-cascade/index.js";
+import { poAttr, poChildren, poFindDescendant, poIntAttr, poIs } from "../core/po-helpers.js";
 import { bytesInclude } from "../core/bytes.js";
-import { poAttr, poChildren, poIntAttr, poIs } from "../core/po-helpers.js";
 import { DEFAULT_THEME_PALETTE, defaultColorResolver, makeColorResolver } from "../core/drawingml/colors.js";
 import { parseChart, withChartColorStyle } from "../core/drawingml/chart-parser.js";
 import { parseTheme } from "../core/drawingml/theme-parser.js";
@@ -81,7 +81,12 @@ function readPptx(bytes) {
 				colors: styles.colors,
 				resolveImage: makeSlideImageResolver(pkg, part.path, resources),
 				resolveChart: makeSlideChartResolver(pkg, part.path, charts, styles.colors),
-				resolveHyperlink: makeHyperlinkResolver(pkg, part.path)
+				resolveHyperlink: makeHyperlinkResolver(pkg, part.path),
+				resolveDiagram: makeSlideDiagramResolver(pkg, part.path),
+				onLoss: (loss) => losses.push(loss.where ? loss : {
+					...loss,
+					where: `slide ${i + 1}`
+				})
 			};
 			body.push(...parseSlide(part.data, ctx, styles.background, pageW, pageH));
 		}
@@ -167,6 +172,28 @@ function makeSlideChartResolver(pkg, slidePath, charts, colors) {
 		return key;
 	};
 }
+function makeSlideDiagramResolver(pkg, slidePath) {
+	const cache = /* @__PURE__ */ new Map();
+	return (relId) => {
+		if (cache.has(relId)) return cache.get(relId);
+		let spTree;
+		const dataRel = pkg.getPartRelationships(slidePath).find((r) => r.id === relId);
+		const data = dataRel ? pkg.resolveRelatedPart(slidePath, dataRel) : void 0;
+		if (data) {
+			const drawRel = pkg.getPartRelationships(data.path).find((r) => r.type.endsWith("/diagramDrawing"));
+			const draw = drawRel ? pkg.resolveRelatedPart(data.path, drawRel) : void 0;
+			if (draw) for (const root of parseXml(draw.data)) {
+				const found = poFindDescendant(root, "dsp:spTree");
+				if (found) {
+					spTree = found;
+					break;
+				}
+			}
+		}
+		cache.set(relId, spTree);
+		return spTree;
+	};
+}
 function slideStylesFor(pkg, slidePath, cache) {
 	const layoutRel = pkg.getPartRelationships(slidePath).find((r) => r.type.endsWith("/slideLayout"));
 	const layout = layoutRel ? pkg.resolveRelatedPart(slidePath, layoutRel) : void 0;
@@ -210,4 +237,4 @@ var pptxReader = {
 	read: (bytes) => readPptx(bytes)
 };
 //#endregion
-export { pptxReader };
+export { parseXml, pptxReader };

package/dist/esm/pptx/slide-parser.d.ts

@@ -1,6 +1,6 @@
-import { BodyElement, ShapeFill } from '../core/document-model/index.js';
+import { BodyElement, FloatAnchor, ShapeBlock, ShapeFill } from '../core/document-model/index.js';
 import { ColorResolver } from '../core/drawingml/colors.js';
-import { Pt, ResourceId } from '../core/ir/index.js';
+import { Loss, Pt, ResourceId } from '../core/ir/index.js';
 import { PoNode } from '../core/po-helpers.js';
 import { PlaceholderCascade } from './placeholder-cascade.js';
 import { ShapeBoxEmu } from './sp-helpers.js';
@@ -10,9 +10,15 @@ export interface SlideContext {
     readonly resolveChart?: (relId: string) => string | undefined;
     readonly colors?: ColorResolver;
     readonly resolveHyperlink?: (relId: string) => string | undefined;
+    readonly resolveDiagram?: (relId: string) => PoNode | undefined;
+    readonly onLoss?: (loss: Loss) => void;
 }
+type LinkResolver = ((relId: string) => string | undefined) | undefined;
 type GroupTransform = (box: ShapeBoxEmu) => ShapeBoxEmu;
 export declare function parseSlideShapes(container: PoNode, ctx?: SlideContext, transform?: GroupTransform): Array<BodyElement>;
+export declare function diagramTransform(spTree: PoNode, frame: ShapeBoxEmu): GroupTransform;
+export declare function noDiagramOverrideLoss(where?: string): Loss;
+export declare function parseDiagramDrawing(spTree: PoNode, transform: GroupTransform, makeFloat: (box: ShapeBoxEmu) => FloatAnchor, colors: ColorResolver, resolveLink: LinkResolver): Array<ShapeBlock>;
 export declare function parseBackgroundFill(bg: PoNode, colors: ColorResolver): ShapeFill | undefined;
 export declare function backdropElement(fill: ShapeFill, widthPt: Pt, heightPt: Pt): BodyElement;
 export {};

package/dist/esm/pptx/slide-parser.js

@@ -1,11 +1,13 @@
 import { emuToPt } from "../core/ir/units.js";
+import { FEATURES } from "../core/ir/features.js";
 import { poAttr, poChildren, poFindDescendant, poIntAttr, poIs, poText } from "../core/po-helpers.js";
 import { defaultColorResolver, resolveColorNode } from "../core/drawingml/colors.js";
-import { parseCustGeom, parseFill, parseLine, parsePrstGeom } from "../word/drawing-parser.js";
 import { boxFromXfrm, parsePh, parseXfrmBox, rPrToRunProps } from "./sp-helpers.js";
+import { parseCustGeom, parseFill, parseLine, parsePrstGeom } from "../word/drawing-parser.js";
 //#region src/pptx/slide-parser.ts
 var CHART_URI = "http://schemas.openxmlformats.org/drawingml/2006/chart";
 var TABLE_URI = "http://schemas.openxmlformats.org/drawingml/2006/table";
+var DIAGRAM_URI = "http://schemas.openxmlformats.org/drawingml/2006/diagram";
 var ALGN_TO_ALIGNMENT = {
 	l: "left",
 	ctr: "center",
@@ -33,10 +35,8 @@ function parseSlideShapes(container, ctx = {}, transform = IDENTITY_TRANSFORM) {
 			kind: "image",
 			image
 		});
-	} else if (poIs(child, "p:graphicFrame")) {
-		const el = parseGraphicFrame(child, ctx, transform);
-		if (el) out.push(el);
-	} else if (poIs(child, "p:grpSp")) out.push(...parseSlideShapes(child, ctx, composeGroupTransform(child, transform)));
+	} else if (poIs(child, "p:graphicFrame")) out.push(...parseGraphicFrame(child, ctx, transform));
+	else if (poIs(child, "p:grpSp")) out.push(...parseSlideShapes(child, ctx, composeGroupTransform(child, transform)));
 	return out;
 }
 function composeGroupTransform(grpSp, parent) {
@@ -106,7 +106,7 @@ function parseSp(sp, ctx, transform) {
 }
 function parseGraphicFrame(gf, ctx, transform) {
 	const own = boxFromXfrm(poChildren(gf).find((c) => poIs(c, "p:xfrm")));
-	if (!own) return void 0;
+	if (!own) return [];
 	const box = transform(own);
 	const graphicData = poFindDescendant(gf, "a:graphicData");
 	const uri = graphicData ? poAttr(graphicData, "uri") : void 0;
@@ -114,8 +114,8 @@ function parseGraphicFrame(gf, ctx, transform) {
 		const cChart = poFindDescendant(gf, "c:chart");
 		const relId = cChart ? poAttr(cChart, "id") : void 0;
 		const key = relId !== void 0 ? ctx.resolveChart?.(relId) : void 0;
-		if (key === void 0) return void 0;
-		return {
+		if (key === void 0) return [];
+		return [{
 			kind: "chart",
 			chart: {
 				float: floatAt(box),
@@ -124,15 +124,83 @@ function parseGraphicFrame(gf, ctx, transform) {
 				height: emuToPt(box.cy),
 				paragraphProperties: {}
 			}
-		};
+		}];
 	}
 	if (uri === TABLE_URI) {
 		const tbl = poFindDescendant(gf, "a:tbl");
-		return tbl ? {
+		return tbl ? [{
 			kind: "table",
 			table: parseTable(tbl, ctx.colors ?? defaultColorResolver)
-		} : void 0;
+		}] : [];
 	}
+	if (uri === DIAGRAM_URI) {
+		const relIds = poFindDescendant(gf, "dgm:relIds");
+		const dmRelId = relIds ? poAttr(relIds, "dm") : void 0;
+		const spTree = dmRelId !== void 0 ? ctx.resolveDiagram?.(dmRelId) : void 0;
+		if (!spTree) {
+			if (dmRelId !== void 0) ctx.onLoss?.(noDiagramOverrideLoss());
+			return [];
+		}
+		return parseDiagramDrawing(spTree, diagramTransform(spTree, box), floatAt, ctx.colors ?? defaultColorResolver, ctx.resolveHyperlink).map((shape) => ({
+			kind: "shape",
+			shape
+		}));
+	}
+	return [];
+}
+function diagramTransform(spTree, frame) {
+	const grpSpPr = poChildren(spTree).find((c) => poIs(c, "dsp:grpSpPr"));
+	const xfrm = grpSpPr ? poChildren(grpSpPr).find((c) => poIs(c, "a:xfrm")) : void 0;
+	const chOff = xfrm ? poChildren(xfrm).find((c) => poIs(c, "a:chOff")) : void 0;
+	const chExt = xfrm ? poChildren(xfrm).find((c) => poIs(c, "a:chExt")) : void 0;
+	const chExtCx = chExt ? poIntAttr(chExt, "cx") : void 0;
+	const chExtCy = chExt ? poIntAttr(chExt, "cy") : void 0;
+	const chOffX = (chOff ? poIntAttr(chOff, "x") : void 0) ?? 0;
+	const chOffY = (chOff ? poIntAttr(chOff, "y") : void 0) ?? 0;
+	const sx = chExtCx && chExtCx > 0 ? frame.cx / chExtCx : 1;
+	const sy = chExtCy && chExtCy > 0 ? frame.cy / chExtCy : 1;
+	return (b) => ({
+		x: frame.x + (b.x - chOffX) * sx,
+		y: frame.y + (b.y - chOffY) * sy,
+		cx: b.cx * sx,
+		cy: b.cy * sy
+	});
+}
+function noDiagramOverrideLoss(where) {
+	return {
+		severity: "dropped",
+		feature: FEATURES.smartArt,
+		detail: "SmartArt diagram has no pre-rendered drawing override; its layout is not reconstructed",
+		...where ? { where } : {}
+	};
+}
+function parseDiagramDrawing(spTree, transform, makeFloat, colors, resolveLink) {
+	const out = [];
+	for (const sp of poChildren(spTree)) {
+		if (!poIs(sp, "dsp:sp")) continue;
+		const spPr = poChildren(sp).find((c) => poIs(c, "dsp:spPr"));
+		const own = parseXfrmBox(spPr);
+		if (!own) continue;
+		const box = transform(own);
+		const txBody = poChildren(sp).find((c) => poIs(c, "dsp:txBody"));
+		const text = txBody ? parseTxBody(txBody, void 0, void 0, colors, resolveLink) : void 0;
+		const geometry = parseGeometry(spPr);
+		const fill = spPr ? parseFill(spPr, colors) : { kind: "none" };
+		const line = spPr ? parseLine(spPr, colors) : void 0;
+		const visibleLine = line !== void 0 && line.fill !== "none";
+		if (!text && fill.kind === "none" && !visibleLine) continue;
+		out.push({
+			float: makeFloat(box),
+			width: emuToPt(box.cx),
+			height: emuToPt(box.cy),
+			geometry,
+			fill,
+			...line ? { line } : {},
+			...text ? { text } : {},
+			paragraphProperties: {}
+		});
+	}
+	return out;
 }
 function parseBackgroundFill(bg, colors) {
 	const bgPr = poChildren(bg).find((c) => poIs(c, "p:bgPr"));
@@ -356,4 +424,4 @@ function cellFillHex(tcPr, colors) {
 	}
 }
 //#endregion
-export { backdropElement, parseBackgroundFill, parseSlideShapes };
+export { backdropElement, diagramTransform, noDiagramOverrideLoss, parseBackgroundFill, parseDiagramDrawing, parseSlideShapes };

package/dist/esm/word/document-parser.d.ts

@@ -1,6 +1,6 @@
-import { BodyElement, Section, SectionProperties } from '../core/document-model/index.js';
+import { BodyElement, Comment, Section, SectionProperties } from '../core/document-model/index.js';
 import { ColorResolver } from '../core/drawingml/colors.js';
-import { ResourceId } from '../core/ir/index.js';
+import { Loss, ResourceId } from '../core/ir/index.js';
 import { PoNode } from '../core/po-helpers.js';
 export type ImageResolver = (relId: string) => ResourceId | undefined;
 export type HyperlinkResolver = (relId: string) => string | undefined;
@@ -8,6 +8,9 @@ export interface ParseContext {
     readonly resolveColor: ColorResolver;
     readonly resolveImage?: ImageResolver;
     readonly resolveHyperlink?: HyperlinkResolver;
+    readonly resolveDiagram?: (relId: string) => PoNode | undefined;
+    readonly onLoss?: (loss: Loss) => void;
+    readonly openCommentRanges?: Set<string>;
 }
 export declare const DEFAULT_PARSE_CONTEXT: ParseContext;
 export declare function parseDocument(documentXml: Uint8Array, ctx?: ParseContext): Array<BodyElement>;
@@ -17,3 +20,15 @@ export declare function parseSections(documentXml: Uint8Array): Array<Section>;
 export declare function parseHeaderFooter(xml: Uint8Array, ctx?: ParseContext): Array<BodyElement>;
 export declare function parseBodyElements(children: ReadonlyArray<PoNode>, ctx?: ParseContext): Array<BodyElement>;
 export declare function parseNotes(notesXml: Uint8Array, rootTag: 'w:footnotes' | 'w:endnotes', noteTag: 'w:footnote' | 'w:endnote', ctx?: ParseContext): Map<string, Array<BodyElement>>;
+export declare function parseComments(commentsXml: Uint8Array, ctx?: ParseContext): Map<string, Comment>;
+/** A commentsExtended entry, keyed by the comment's last-paragraph w14:paraId. */
+export interface CommentExtension {
+    /** The parent comment's paraId — present when this comment is a reply. */
+    readonly paraIdParent?: string;
+    /** w15:done — the thread is resolved. */
+    readonly done: boolean;
+}
+export declare function parseCommentsExtended(xml: Uint8Array): Map<string, CommentExtension>;
+export declare function parseCommentThreads(commentsXml: Uint8Array, commentsExtendedXml: Uint8Array | undefined, ctx?: ParseContext): Map<string, Comment>;
+export declare function parsePeople(xml: Uint8Array): Map<string, string>;
+export declare function applyAuthorIds(comments: Map<string, Comment>, people: Map<string, string>): Map<string, Comment>;

package/dist/esm/word/document-parser.js

@@ -1,8 +1,9 @@
-import { twipsToPt } from "../core/ir/units.js";
-import { poAttr, poChildren, poFindByPath, poFindDescendant, poIntAttr, poIs, poText } from "../core/po-helpers.js";
+import { emuToPt, twipsToPt } from "../core/ir/units.js";
+import { poAttr, poAttrLocal, poChildren, poFindByPath, poFindDescendant, poIntAttr, poIs, poIsLocal, poText } from "../core/po-helpers.js";
 import { defaultColorResolver } from "../core/drawingml/colors.js";
-import { parseOMath } from "./omml-parser.js";
 import { expandMcChildren, parseDrawing, parseVmlPicture } from "./drawing-parser.js";
+import { diagramTransform, noDiagramOverrideLoss, parseDiagramDrawing } from "../pptx/slide-parser.js";
+import { parseOMath } from "./omml-parser.js";
 import { parseRunProperties } from "./run-properties.js";
 import { parseParagraphProperties } from "./paragraph-properties.js";
 import { poElementToFlat } from "./po-to-flat.js";
@@ -162,12 +163,15 @@ function parseBodyElements(children, ctx = DEFAULT_PARSE_CONTEXT) {
 		const bookmarkName = poAttr(child, "name");
 		if (bookmarkName !== void 0 && bookmarkName !== "" && bookmarkName !== "_GoBack") (pendingBookmarks ??= []).push(bookmarkName);
 	} else if (poIs(child, "w:p")) {
-		const drawing = tryExtractDrawingFromParagraph(child, ctx);
-		out.push(drawing ?? {
-			kind: "paragraph",
-			paragraph: parseParagraph(child, ctx, pendingBookmarks)
-		});
-		if (!drawing) pendingBookmarks = void 0;
+		const drawings = tryExtractDrawingFromParagraph(child, ctx);
+		if (drawings) out.push(...drawings);
+		else {
+			out.push({
+				kind: "paragraph",
+				paragraph: parseParagraph(child, ctx, pendingBookmarks)
+			});
+			pendingBookmarks = void 0;
+		}
 	} else if (poIs(child, "w:tbl")) out.push({
 		kind: "table",
 		table: parseTable(child, ctx)
@@ -220,7 +224,7 @@ function tryExtractDrawingFromParagraph(p, ctx) {
 	const paragraphProperties = pPrNode ? parseParagraphProperties(poElementToFlat(pPrNode)) : {};
 	if (content.kind === "image") {
 		const resource = ctx.resolveImage?.(content.imageId);
-		return {
+		return [{
 			kind: "image",
 			image: {
 				...resource ? { resource } : {},
@@ -230,9 +234,9 @@ function tryExtractDrawingFromParagraph(p, ctx) {
 				...content.altText ? { altText: content.altText } : {},
 				...content.float ? { float: content.float } : {}
 			}
-		};
+		}];
 	}
-	if (content.kind === "chart") return {
+	if (content.kind === "chart") return [{
 		kind: "chart",
 		chart: {
 			chartRelId: content.chartRelId,
@@ -242,8 +246,35 @@ function tryExtractDrawingFromParagraph(p, ctx) {
 			...content.altText ? { altText: content.altText } : {},
 			...content.float ? { float: content.float } : {}
 		}
-	};
-	return {
+	}];
+	if (content.kind === "diagram") {
+		const spTree = ctx.resolveDiagram?.(content.dmRelId);
+		if (!spTree) {
+			ctx.onLoss?.(noDiagramOverrideLoss());
+			return null;
+		}
+		const shapes = parseDiagramDrawing(spTree, diagramTransform(spTree, {
+			x: 0,
+			y: 0,
+			cx: content.widthEmu,
+			cy: content.heightEmu
+		}), (box) => ({
+			wrap: "none",
+			posH: {
+				relativeFrom: "column",
+				offsetPt: emuToPt(box.x)
+			},
+			posV: {
+				relativeFrom: "paragraph",
+				offsetPt: emuToPt(box.y)
+			}
+		}), ctx.resolveColor, void 0);
+		return shapes.length > 0 ? shapes.map((shape) => ({
+			kind: "shape",
+			shape
+		})) : null;
+	}
+	return [{
 		kind: "shape",
 		shape: {
 			...content.data,
@@ -251,7 +282,7 @@ function tryExtractDrawingFromParagraph(p, ctx) {
 			...content.altText ? { altText: content.altText } : {},
 			...content.float ? { float: content.float } : {}
 		}
-	};
+	}];
 }
 function parseParagraph(p, ctx, extraBookmarks) {
 	const bookmarks = [...extraBookmarks ?? []];
@@ -349,12 +380,24 @@ function applyFieldFsm(collected) {
 function collectRuns(container, out, ctx, href, anchor) {
 	for (const child of poChildren(container)) {
 		if (poIs(child, "w:pPr")) continue;
+		if (poIs(child, "w:commentRangeStart")) {
+			const id = poAttr(child, "id");
+			if (id !== void 0) ctx.openCommentRanges?.add(id);
+			continue;
+		}
+		if (poIs(child, "w:commentRangeEnd")) {
+			const id = poAttr(child, "id");
+			if (id !== void 0) ctx.openCommentRanges?.delete(id);
+			continue;
+		}
 		if (poIs(child, "w:r")) {
 			const parsed = parseRun(child, ctx);
-			const run = href !== void 0 || anchor !== void 0 ? {
+			const ranges = ctx.openCommentRanges && ctx.openCommentRanges.size > 0 ? [...ctx.openCommentRanges] : void 0;
+			const run = href !== void 0 || anchor !== void 0 || ranges !== void 0 ? {
 				...parsed.run,
 				...href !== void 0 ? { href } : {},
-				...anchor !== void 0 ? { anchor } : {}
+				...anchor !== void 0 ? { anchor } : {},
+				...ranges !== void 0 ? { commentRangeRefs: ranges } : {}
 			} : parsed.run;
 			out.push({
 				run,
@@ -417,6 +460,7 @@ function parseRun(r, ctx) {
 	let instrText;
 	let footnoteRef;
 	let endnoteRef;
+	let commentRef;
 	let noteNumber = false;
 	for (const child of expandMcChildren(poChildren(r))) {
 		if (poIs(child, "w:rPr")) continue;
@@ -439,6 +483,11 @@ function parseRun(r, ctx) {
 			if (id !== void 0) endnoteRef = id;
 			continue;
 		}
+		if (poIs(child, "w:commentReference")) {
+			const id = poAttr(child, "id");
+			if (id !== void 0) commentRef = id;
+			continue;
+		}
 		if (poIs(child, "w:footnoteRef") || poIs(child, "w:endnoteRef")) {
 			noteNumber = true;
 			continue;
@@ -479,6 +528,7 @@ function parseRun(r, ctx) {
 			...pageBreak ? { pageBreak: true } : {},
 			...footnoteRef !== void 0 ? { footnoteRef } : {},
 			...endnoteRef !== void 0 ? { endnoteRef } : {},
+			...commentRef !== void 0 ? { commentRef } : {},
 			...noteNumber ? { noteNumber: true } : {}
 		},
 		...fldChar ? { fldChar } : {},
@@ -503,5 +553,115 @@ function parseNotes(notesXml, rootTag, noteTag, ctx = DEFAULT_PARSE_CONTEXT) {
 	}
 	return out;
 }
+function parseCommentsRaw(commentsXml, ctx) {
+	const xml = decoder.decode(commentsXml);
+	const root = poFindByPath(parser.parse(xml), ["w:comments"]);
+	const comments = /* @__PURE__ */ new Map();
+	const paraIds = /* @__PURE__ */ new Map();
+	if (!root) return {
+		comments,
+		paraIds
+	};
+	for (const c of poChildren(root)) {
+		if (!poIs(c, "w:comment")) continue;
+		const id = poAttr(c, "id");
+		if (id === void 0) continue;
+		const author = poAttr(c, "author");
+		const initials = poAttr(c, "initials");
+		const date = poAttr(c, "date");
+		comments.set(id, {
+			content: parseBodyElements(poChildren(c), ctx),
+			...author !== void 0 ? { author } : {},
+			...initials !== void 0 ? { initials } : {},
+			...date !== void 0 ? { date } : {}
+		});
+		let paraId;
+		for (const child of poChildren(c)) {
+			if (!poIs(child, "w:p")) continue;
+			const pid = poAttrLocal(child, "paraId");
+			if (pid !== void 0) paraId = pid;
+		}
+		if (paraId !== void 0) paraIds.set(id, paraId);
+	}
+	return {
+		comments,
+		paraIds
+	};
+}
+function parseCommentsExtended(xml) {
+	const tree = parser.parse(decoder.decode(xml));
+	const out = /* @__PURE__ */ new Map();
+	const root = tree.find((n) => poIsLocal(n, "commentsEx"));
+	if (!root) return out;
+	for (const ex of poChildren(root)) {
+		if (!poIsLocal(ex, "commentEx")) continue;
+		const paraId = poAttrLocal(ex, "paraId");
+		if (paraId === void 0) continue;
+		const paraIdParent = poAttrLocal(ex, "paraIdParent");
+		const done = poAttrLocal(ex, "done");
+		out.set(paraId, {
+			...paraIdParent !== void 0 ? { paraIdParent } : {},
+			done: done === "1" || done === "true"
+		});
+	}
+	return out;
+}
+function linkCommentThreads(comments, paraIdByComment, ext) {
+	if (ext.size === 0) return comments;
+	const commentByParaId = /* @__PURE__ */ new Map();
+	for (const [id, pid] of paraIdByComment) commentByParaId.set(pid, id);
+	const out = /* @__PURE__ */ new Map();
+	for (const [id, c] of comments) {
+		const pid = paraIdByComment.get(id);
+		const e = pid !== void 0 ? ext.get(pid) : void 0;
+		if (!e) {
+			out.set(id, c);
+			continue;
+		}
+		const parentId = e.paraIdParent !== void 0 ? commentByParaId.get(e.paraIdParent) : void 0;
+		out.set(id, {
+			...c,
+			...parentId !== void 0 ? { parentId } : {},
+			...e.done ? { done: true } : {}
+		});
+	}
+	return out;
+}
+function parseCommentThreads(commentsXml, commentsExtendedXml, ctx = DEFAULT_PARSE_CONTEXT) {
+	const { comments, paraIds } = parseCommentsRaw(commentsXml, ctx);
+	if (!commentsExtendedXml) return comments;
+	return linkCommentThreads(comments, paraIds, parseCommentsExtended(commentsExtendedXml));
+}
+function parsePeople(xml) {
+	const tree = parser.parse(decoder.decode(xml));
+	const out = /* @__PURE__ */ new Map();
+	const root = tree.find((n) => poIsLocal(n, "people"));
+	if (!root) return out;
+	for (const person of poChildren(root)) {
+		if (!poIsLocal(person, "person")) continue;
+		const author = poAttrLocal(person, "author");
+		if (author === void 0) continue;
+		let userId;
+		for (const child of poChildren(person)) {
+			if (!poIsLocal(child, "presenceInfo")) continue;
+			const uid = poAttrLocal(child, "userId");
+			if (uid !== void 0) userId = uid;
+		}
+		if (userId !== void 0) out.set(author, userId);
+	}
+	return out;
+}
+function applyAuthorIds(comments, people) {
+	if (people.size === 0) return comments;
+	const out = /* @__PURE__ */ new Map();
+	for (const [id, c] of comments) {
+		const userId = c.author !== void 0 ? people.get(c.author) : void 0;
+		out.set(id, userId !== void 0 ? {
+			...c,
+			authorId: userId
+		} : c);
+	}
+	return out;
+}
 //#endregion
-export { DEFAULT_PARSE_CONTEXT, EMPTY_SECTION, parseBodyElements, parseDocument, parseHeaderFooter, parseNotes, parseSections };
+export { DEFAULT_PARSE_CONTEXT, EMPTY_SECTION, applyAuthorIds, parseBodyElements, parseCommentThreads, parseCommentsExtended, parseDocument, parseHeaderFooter, parseNotes, parsePeople, parseSections };

package/dist/esm/word/docx-reader.js

@@ -4,6 +4,7 @@ import { applyNumbering, applyNumberingToHeadersFooters } from "../core/numberin
 import { EMPTY_STYLE_SHEET, resolveBodyStyles, resolveHeadersFootersStyles } from "../core/style-cascade/resolver.js";
 import { resolveTableStyles } from "../core/style-cascade/table.js";
 import "../core/style-cascade/index.js";
+import { poFindDescendant } from "../core/po-helpers.js";
 import { bytesInclude } from "../core/bytes.js";
 import { DEFAULT_THEME_PALETTE, makeColorResolver } from "../core/drawingml/colors.js";
 import { parseChart, withChartColorStyle } from "../core/drawingml/chart-parser.js";
@@ -12,7 +13,8 @@ import { isOoxmlRel } from "../core/opc/relationship-types.js";
 import { OpcPackage } from "../core/opc/package.js";
 import { parseCoreProperties } from "../core/opc/core-properties.js";
 import "../core/opc/index.js";
-import { EMPTY_SECTION, parseDocument, parseHeaderFooter, parseNotes, parseSections } from "./document-parser.js";
+import { parseXml } from "../pptx/pptx-reader.js";
+import { EMPTY_SECTION, applyAuthorIds, parseCommentThreads, parseDocument, parseHeaderFooter, parseNotes, parsePeople, parseSections } from "./document-parser.js";
 import { parseStyles } from "./styles-parser.js";
 import { loadEmbeddedFonts } from "./font-table.js";
 import { EMPTY_NUMBERING, parseNumbering } from "./numbering-parser.js";
@@ -22,6 +24,9 @@ import "./index.js";
 var STYLES_PART = "word/styles.xml";
 var FOOTNOTES_PART = "word/footnotes.xml";
 var ENDNOTES_PART = "word/endnotes.xml";
+var COMMENTS_PART = "word/comments.xml";
+var COMMENTS_EXTENDED_PART = "word/commentsExtended.xml";
+var PEOPLE_PART = "word/people.xml";
 var NUMBERING_PART = "word/numbering.xml";
 var SETTINGS_PART = "word/settings.xml";
 var CORE_PROPS_PART = "docProps/core.xml";
@@ -33,10 +38,16 @@ function readDocx(docx) {
 	const main = pkg.getMainDocument();
 	const resolveColor = buildColorResolver(pkg);
 	const resources = new ResourceStore();
+	const resolveImage = makeImageResolver(pkg, resources);
+	const resolveHyperlink = makeHyperlinkResolver(pkg);
+	const losses = [];
 	const ctx = {
 		resolveColor,
-		resolveImage: makeImageResolver(pkg, resources),
-		resolveHyperlink: makeHyperlinkResolver(pkg)
+		resolveImage,
+		resolveHyperlink,
+		resolveDiagram: makeDiagramResolver(pkg, MAIN_DOCUMENT_PART),
+		onLoss: (loss) => losses.push(loss),
+		openCommentRanges: /* @__PURE__ */ new Set()
 	};
 	const body = parseDocument(main.data, ctx);
 	const rawSections = parseSections(main.data);
@@ -53,6 +64,11 @@ function readDocx(docx) {
 	const rawFootnotes = footnotesData ? parseNotes(footnotesData, "w:footnotes", "w:footnote", noteCtx(FOOTNOTES_PART)) : void 0;
 	const endnotesData = pkg.getPart(ENDNOTES_PART);
 	const rawEndnotes = endnotesData ? parseNotes(endnotesData, "w:endnotes", "w:endnote", noteCtx(ENDNOTES_PART)) : void 0;
+	const commentsData = pkg.getPart(COMMENTS_PART);
+	const commentsExtendedData = pkg.getPart(COMMENTS_EXTENDED_PART);
+	const peopleData = pkg.getPart(PEOPLE_PART);
+	let rawComments = commentsData ? parseCommentThreads(commentsData, commentsExtendedData, noteCtx(COMMENTS_PART)) : void 0;
+	if (rawComments && peopleData) rawComments = applyAuthorIds(rawComments, parsePeople(peopleData));
 	const settingsData = pkg.getPart(SETTINGS_PART);
 	const settings = settingsData ? parseSettings(settingsData) : EMPTY_SETTINGS;
 	const sections = rawSections.length > 0 ? rawSections.map((s) => ({
@@ -83,6 +99,7 @@ function readDocx(docx) {
 			numbering,
 			...rawFootnotes && rawFootnotes.size > 0 ? { footnotes: transformNotes(rawFootnotes, styles, numbering) } : {},
 			...rawEndnotes && rawEndnotes.size > 0 ? { endnotes: transformNotes(rawEndnotes, styles, numbering) } : {},
+			...rawComments && rawComments.size > 0 ? { comments: transformComments(rawComments, styles, numbering) } : {},
 			headersFooters: resolveHeadersFootersStyles(applyNumberingToHeadersFooters(headersFooters, numbering), styles),
 			charts,
 			resources,
@@ -90,7 +107,7 @@ function readDocx(docx) {
 			...info ? { info } : {},
 			...language ? { language } : {}
 		},
-		losses: []
+		losses
 	};
 }
 var docxReader = {
@@ -125,6 +142,28 @@ function infoFromCore(core) {
 		...core.modified ? { modificationDate: core.modified } : {}
 	};
 }
+function makeDiagramResolver(pkg, partName) {
+	const cache = /* @__PURE__ */ new Map();
+	return (relId) => {
+		if (cache.has(relId)) return cache.get(relId);
+		let spTree;
+		const dataRel = pkg.getPartRelationships(partName).find((r) => r.id === relId);
+		const data = dataRel ? pkg.resolveRelatedPart(partName, dataRel) : void 0;
+		if (data) {
+			const drawRel = pkg.getPartRelationships(data.path).find((r) => r.type.endsWith("/diagramDrawing"));
+			const draw = drawRel ? pkg.resolveRelatedPart(data.path, drawRel) : void 0;
+			if (draw) for (const root of parseXml(draw.data)) {
+				const found = poFindDescendant(root, "dsp:spTree");
+				if (found) {
+					spTree = found;
+					break;
+				}
+			}
+		}
+		cache.set(relId, spTree);
+		return spTree;
+	};
+}
 function makeImageResolver(pkg, store, partName = MAIN_DOCUMENT_PART) {
 	const byRelId = /* @__PURE__ */ new Map();
 	for (const rel of pkg.getPartRelationships(partName)) {
@@ -145,6 +184,17 @@ function transformNotes(notes, styles, numbering) {
 	for (const content of notes.values()) resolveTableStyles(content, styles);
 	return resolveHeadersFootersStyles(applyNumberingToHeadersFooters(notes, numbering), styles);
 }
+function transformComments(comments, styles, numbering) {
+	const contentById = /* @__PURE__ */ new Map();
+	for (const [id, c] of comments) contentById.set(id, [...c.content]);
+	const transformed = transformNotes(contentById, styles, numbering);
+	const out = /* @__PURE__ */ new Map();
+	for (const [id, c] of comments) out.set(id, {
+		...c,
+		content: transformed.get(id) ?? c.content
+	});
+	return out;
+}
 function makeHyperlinkResolver(pkg, partName = MAIN_DOCUMENT_PART) {
 	const byRelId = /* @__PURE__ */ new Map();
 	for (const rel of pkg.getPartRelationships(partName)) if (rel.type === REL_HYPERLINK && rel.targetMode === "External") byRelId.set(rel.id, rel.target);

package/dist/esm/word/docx-to-pdf.js

@@ -49,7 +49,7 @@ function prepareDocxStyledRender(docx, options) {
 	};
 }
 function detectDocxFamilyKeys(docx) {
-	const keys = new Set(["roboto"]);
+	const keys = new Set(["arimo"]);
 	let pkg;
 	try {
 		pkg = OpcPackage.open(docx);
@@ -111,7 +111,7 @@ async function resolveDocxAutoFonts(docx, options = {}) {
 			...fetchOpt
 		});
 		registriesByFamily.set(key, FontRegistry.fromBytes(bytes));
-		if (key === "roboto" || !baseBytes) baseBytes = bytes;
+		if (key === "arimo" || !baseBytes) baseBytes = bytes;
 	}
 	return {
 		fonts: baseBytes,