reamkit 1.16.0 → 1.17.0

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

@@ -1,16 +1,18 @@
-import { emuToPt, twipsToPt } from "../core/ir/units.js";
+import { emuToPt, pt, 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 { expandMcChildren, parseDrawing, parseVmlPicture } from "./drawing-parser.js";
-import { diagramTransform, noDiagramOverrideLoss, parseDiagramDrawing } from "../pptx/slide-parser.js";
+import { expandMcChildren, parseDrawing, parseVmlPicture, vmlFill } from "./drawing-parser.js";
+import { diagramTransform, noDiagramOverrideLoss, parseDiagramNodes, parseTxBody } from "../pptx/slide-parser.js";
+import { resolveInternalEntities } from "../core/opc/xml-entities.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";
-import { parseTable } from "./table-parser.js";
+import { parseBorders, parseTable } from "./table-parser.js";
 import { XMLParser } from "fast-xml-parser";
 //#region src/word/document-parser.ts
 var decoder = new TextDecoder("utf-8");
+var drawingTextParser = (txBody, resolveColor) => parseTxBody(txBody, void 0, void 0, resolveColor, void 0);
 var parser = new XMLParser({
 	htmlEntities: true,
 	ignoreAttributes: false,
@@ -18,7 +20,8 @@ var parser = new XMLParser({
 	preserveOrder: true,
 	parseAttributeValue: false,
 	parseTagValue: false,
-	trimValues: false
+	trimValues: false,
+	maxNestedTags: 1e3
 });
 var RUN_CONTAINER_TAGS = new Set([
 	"w:hyperlink",
@@ -40,19 +43,102 @@ var DEFAULT_PARSE_CONTEXT = { resolveColor: defaultColorResolver };
 * @param ctx         The document-wide parse context.
 * @returns The body elements; empty when the `w:body` is absent.
 */
-function parseDocument(documentXml, ctx = DEFAULT_PARSE_CONTEXT) {
-	const xml = decoder.decode(documentXml);
-	const body = poFindByPath(parser.parse(xml), ["w:document", "w:body"]);
-	if (!body) return [];
-	return parseBodyElements(poChildren(body), ctx);
+function parseDocument(documentXml, ctx = DEFAULT_PARSE_CONTEXT, blocks) {
+	const xml = resolveInternalEntities(decoder.decode(documentXml));
+	const children = documentBodyChildren(parser.parse(xml));
+	if (children.length === 0) return [];
+	return parseBodyElements(children, ctx, blocks);
+}
+/**
+* §17.2.2 — the `w:document`'s body content. The schema allows one `w:body`
+* and producers have written several: MultipleBodyBug.docx carries three, and
+* reading only the first printed one paragraph of its three. They are read as
+* the one body they stand for, in order.
+*
+* @param tree The parsed `document.xml`.
+* @returns Every body's children, concatenated.
+*/
+function documentBodyChildren(tree) {
+	const doc = poFindByPath(tree, ["w:document"]);
+	if (!doc) return [];
+	const out = [];
+	for (const child of poChildren(doc)) if (poIs(child, "w:body")) out.push(...poChildren(child));
+	return out;
+}
+/**
+* §17.2.1 `w:background` — the colour the whole page is painted, as a six-digit
+* hex. Read from `@w:color` (Word's own attribute), falling back to the VML
+* `v:background @fillcolor` beside it. `auto` and a gradient-only background
+* (which carries no flat colour of its own) both come back undefined.
+*
+* @param documentXml The raw `document.xml` bytes.
+* @returns The colour, or undefined when the document names none.
+*/
+function parseBackgroundColor(documentXml) {
+	const bg = backgroundNode(documentXml);
+	if (!bg) return void 0;
+	const vml = poChildren(bg).find((c) => poIs(c, "v:background"));
+	const hex = (poAttr(bg, "color") ?? (vml ? poAttr(vml, "fillcolor") : void 0))?.replace(/^#/u, "");
+	return hex !== void 0 && /^[0-9A-Fa-f]{6}$/u.test(hex) ? hex.toUpperCase() : void 0;
+}
+/**
+* §17.2.1 — the page background as the FILL it is. `@w:color` is only the flat
+* fallback: the `v:background` beside it carries the gradient or the picture
+* Word actually paints, and both references draw those. fill.docx runs a
+* five-stop radial sweep where we painted its fallback navy, and
+* tdf126533_pageBitmap.docx papers the page with an image.
+*
+* @param documentXml  The raw `document.xml` bytes.
+* @param resolveImage Resolver for a picture background's `r:id`.
+* @returns The fill, or undefined when the document names no background.
+*/
+function parseBackgroundFill(documentXml, resolveImage) {
+	const bg = backgroundNode(documentXml);
+	if (!bg || poAttr(bg, "color") === void 0) return void 0;
+	const vml = poChildren(bg).find((c) => poIs(c, "v:background"));
+	if (!vml) return void 0;
+	const fill = vmlFill(vml, void 0, resolveImage);
+	return fill.kind === "gradient" || fill.kind === "picture" ? fill : void 0;
+}
+function backgroundNode(documentXml) {
+	const doc = poFindByPath(parser.parse(resolveInternalEntities(decoder.decode(documentXml))), ["w:document"]);
+	return doc ? poChildren(doc).find((c) => poIs(c, "w:background")) : void 0;
+}
+/** A fresh {@link BlockCounter}. */
+function newBlockCounter() {
+	return {
+		index: [],
+		next: 0
+	};
+}
+/**
+* Translate a section boundary counted in source blocks into one counted in body
+* elements: the number of elements that came from a block before `blockEnd`.
+*
+* @param blocks   The counter filled while parsing the body.
+* @param blockEnd The section's `endIndex`, in source blocks.
+* @returns The matching body-element index.
+*/
+function bodyIndexForBlock(blocks, blockEnd) {
+	let n = 0;
+	while (n < blocks.index.length && blocks.index[n] < blockEnd) n++;
+	return n;
 }
 var HF_TYPES = new Set([
 	"default",
 	"first",
 	"even"
 ]);
-/** The empty {@link SectionProperties} fallback (no headers/footers). */
+/**
+* The empty {@link SectionProperties} fallback (no headers/footers). §17.6.13 —
+* on the page a document with no `w:sectPr` of its own is made on, which is the
+* same US Letter a section with no `w:pgSz` gets.
+*/
 var EMPTY_SECTION = {
+	pageSize: {
+		width: twipsToPt(12240),
+		height: twipsToPt(15840)
+	},
 	headers: [],
 	footers: []
 };
@@ -67,30 +153,38 @@ var EMPTY_SECTION = {
 * @returns The sections in document order; empty when the `w:body` is absent.
 */
 function parseSections(documentXml) {
-	const xml = decoder.decode(documentXml);
-	const body = poFindByPath(parser.parse(xml), ["w:document", "w:body"]);
-	if (!body) return [];
-	const children = poChildren(body);
+	const xml = resolveInternalEntities(decoder.decode(documentXml));
+	const children = documentBodyChildren(parser.parse(xml));
+	if (children.length === 0) return [];
 	const sections = [];
 	let bodyIdx = 0;
-	for (const child of children) {
-		if (poIs(child, "w:sectPr")) {
-			sections.push({
-				properties: parseSectPrNode(child),
-				endIndex: bodyIdx
-			});
-			continue;
+	let trailing;
+	const walk = (nodes) => {
+		for (const child of nodes) {
+			if (poIs(child, "w:sectPr")) {
+				trailing = parseSectPrNode(child);
+				continue;
+			}
+			if (poIs(child, "w:p")) {
+				const pPrNode = poChildren(child).find((c) => poIs(c, "w:pPr"));
+				const sectPrInPPr = pPrNode ? poChildren(pPrNode).find((c) => poIs(c, "w:sectPr")) : void 0;
+				bodyIdx++;
+				if (sectPrInPPr) sections.push({
+					properties: parseSectPrNode(sectPrInPPr),
+					endIndex: bodyIdx
+				});
+			} else if (poIs(child, "w:tbl")) bodyIdx++;
+			else if (poIs(child, "w:sdt")) {
+				const content = poChildren(child).find((c) => poIs(c, "w:sdtContent"));
+				walk(content ? poChildren(content) : []);
+			}
 		}
-		if (poIs(child, "w:p")) {
-			const pPrNode = poChildren(child).find((c) => poIs(c, "w:pPr"));
-			const sectPrInPPr = pPrNode ? poChildren(pPrNode).find((c) => poIs(c, "w:sectPr")) : void 0;
-			bodyIdx++;
-			if (sectPrInPPr) sections.push({
-				properties: parseSectPrNode(sectPrInPPr),
-				endIndex: bodyIdx
-			});
-		} else if (poIs(child, "w:tbl")) bodyIdx++;
-	}
+	};
+	walk(children);
+	if (trailing) sections.push({
+		properties: trailing,
+		endIndex: bodyIdx
+	});
 	if (sections.length === 0 || sections[sections.length - 1].endIndex < bodyIdx) sections.push({
 		properties: EMPTY_SECTION,
 		endIndex: bodyIdx
@@ -98,10 +192,18 @@ function parseSections(documentXml) {
 	return sections;
 }
 function parseSectPrNode(sectPr) {
-	let pageSize;
+	let pageSize = {
+		width: twipsToPt(12240),
+		height: twipsToPt(15840)
+	};
 	let margins;
 	let titlePg = false;
+	let pageNumberStart;
+	let lineNumbering;
 	let columns;
+	let sectionStart;
+	let pageBorders;
+	let gridLinePitchPt;
 	const headers = [];
 	const footers = [];
 	for (const child of poChildren(sectPr)) if (poIs(child, "w:pgSz")) {
@@ -120,30 +222,64 @@ function parseSectPrNode(sectPr) {
 		const left = poIntAttr(child, "left");
 		const header = poIntAttr(child, "header");
 		const footer = poIntAttr(child, "footer");
+		const gutter = poIntAttr(child, "gutter");
 		margins = {
 			top: twipsToPt(top ?? 1440),
 			right: twipsToPt(right ?? 1440),
 			bottom: twipsToPt(bottom ?? 1440),
 			left: twipsToPt(left ?? 1440),
 			...header !== void 0 ? { header: twipsToPt(header) } : {},
-			...footer !== void 0 ? { footer: twipsToPt(footer) } : {}
+			...footer !== void 0 ? { footer: twipsToPt(footer) } : {},
+			...gutter !== void 0 && gutter > 0 ? { gutter: twipsToPt(gutter) } : {}
 		};
 	} else if (poIs(child, "w:headerReference")) pushHeaderFooter(child, headers);
 	else if (poIs(child, "w:footerReference")) pushHeaderFooter(child, footers);
 	else if (poIs(child, "w:titlePg")) {
 		const val = poAttr(child, "val");
 		titlePg = val === void 0 || val === "" || val !== "0" && val !== "false";
+	} else if (poIs(child, "w:lnNumType")) {
+		const countBy = poIntAttr(child, "countBy") ?? 1;
+		const start = poIntAttr(child, "start") ?? 1;
+		const distance = poIntAttr(child, "distance");
+		const restartRaw = poAttr(child, "restart");
+		lineNumbering = {
+			countBy: countBy > 0 ? countBy : 1,
+			start,
+			...distance !== void 0 ? { distancePt: twipsToPt(distance) } : {},
+			restart: restartRaw === "newSection" || restartRaw === "continuous" ? restartRaw : "newPage"
+		};
+	} else if (poIs(child, "w:pgNumType")) {
+		const start = poIntAttr(child, "start");
+		if (start !== void 0) pageNumberStart = start;
 	} else if (poIs(child, "w:cols")) columns = parseColumns(child);
+	else if (poIs(child, "w:pgBorders")) {
+		const borders = parseBorders(child);
+		if (borders) pageBorders = {
+			borders,
+			offsetFrom: poAttr(child, "offsetFrom") === "page" ? "page" : "text"
+		};
+	} else if (poIs(child, "w:docGrid")) {
+		const type = poAttr(child, "type") ?? "default";
+		const pitch = poIntAttr(child, "linePitch");
+		if ((type === "lines" || type === "linesAndChars") && pitch !== void 0 && pitch > 0) gridLinePitchPt = twipsToPt(pitch);
+	} else if (poIs(child, "w:type")) sectionStart = poAttr(child, "val") === "continuous" ? "continuous" : "nextPage";
 	return {
-		...pageSize ? { pageSize } : {},
+		pageSize,
 		...margins ? { margins } : {},
 		headers,
 		footers,
 		...titlePg ? { titlePg: true } : {},
-		...columns ? { columns } : {}
+		...pageNumberStart !== void 0 ? { pageNumberStart } : {},
+		...lineNumbering ? { lineNumbering } : {},
+		...columns ? { columns } : {},
+		...sectionStart ? { sectionStart } : {},
+		...pageBorders ? { pageBorders } : {},
+		...gridLinePitchPt !== void 0 ? { gridLinePitchPt } : {}
 	};
 }
 function parseColumns(cols) {
+	const sepRaw = poAttr(cols, "sep");
+	const separator = sepRaw !== void 0 && sepRaw !== "0" && sepRaw !== "false";
 	const explicit = [];
 	for (const col of poChildren(cols)) {
 		if (!poIs(col, "w:col")) continue;
@@ -160,7 +296,8 @@ function parseColumns(cols) {
 	return {
 		count,
 		spacePt: twipsToPt(poIntAttr(cols, "space") ?? 720),
-		...explicit.length > 0 ? { explicit } : {}
+		...explicit.length > 0 ? { explicit } : {},
+		...separator ? { separator: true } : {}
 	};
 }
 function pushHeaderFooter(node, list) {
@@ -182,7 +319,7 @@ function pushHeaderFooter(node, list) {
 * @returns The body elements; empty when neither root is found.
 */
 function parseHeaderFooter(xml, ctx = DEFAULT_PARSE_CONTEXT) {
-	const root = parser.parse(decoder.decode(xml)).find((n) => poIs(n, "w:hdr") || poIs(n, "w:ftr"));
+	const root = parser.parse(resolveInternalEntities(decoder.decode(xml))).find((n) => poIs(n, "w:hdr") || poIs(n, "w:ftr"));
 	if (!root) return [];
 	return parseBodyElements(poChildren(root), ctx);
 }
@@ -196,29 +333,57 @@ function parseHeaderFooter(xml, ctx = DEFAULT_PARSE_CONTEXT) {
 * @param ctx      The document-wide parse context.
 * @returns The parsed body elements.
 */
-function parseBodyElements(children, ctx = DEFAULT_PARSE_CONTEXT) {
+function parseBodyElements(children, ctx = DEFAULT_PARSE_CONTEXT, blocks, sdtRunProps) {
 	const out = [];
+	const mark = (n) => {
+		if (!blocks) return;
+		for (let i = 0; i < n; i++) blocks.index.push(blocks.next);
+	};
 	let pendingBookmarks;
-	for (const child of children) if (poIs(child, "w:bookmarkStart")) {
-		const bookmarkName = poAttr(child, "name");
-		if (bookmarkName !== void 0 && bookmarkName !== "" && bookmarkName !== "_GoBack") (pendingBookmarks ??= []).push(bookmarkName);
-	} else if (poIs(child, "w:p")) {
-		const drawings = tryExtractDrawingFromParagraph(child, ctx);
-		if (drawings) out.push(...drawings);
-		else {
+	let pendingPageBreak = false;
+	for (const child of children) {
+		if (poIs(child, "w:br")) {
+			if (poAttr(child, "type") === "page") pendingPageBreak = true;
+			continue;
+		}
+		if (poIs(child, "w:bookmarkStart")) {
+			const bookmarkName = poAttr(child, "name");
+			if (bookmarkName !== void 0 && bookmarkName !== "" && bookmarkName !== "_GoBack") (pendingBookmarks ??= []).push(bookmarkName);
+		} else if (poIs(child, "w:p")) {
+			const drawings = tryExtractDrawingFromParagraph(child, ctx);
+			if (drawings) {
+				out.push(...drawings);
+				mark(drawings.length);
+			} else {
+				const anchored = [];
+				const parsed = parseParagraph(child, ctx, pendingBookmarks, anchored, sdtRunProps);
+				const paragraph = pendingPageBreak ? {
+					...parsed,
+					properties: {
+						...parsed.properties,
+						pageBreakBefore: true
+					}
+				} : parsed;
+				pendingPageBreak = false;
+				out.push(...anchored, {
+					kind: "paragraph",
+					paragraph
+				});
+				mark(anchored.length + 1);
+				pendingBookmarks = void 0;
+			}
+			if (blocks) blocks.next++;
+		} else if (poIs(child, "w:tbl")) {
 			out.push({
-				kind: "paragraph",
-				paragraph: parseParagraph(child, ctx, pendingBookmarks)
+				kind: "table",
+				table: parseTable(child, ctx)
 			});
-			pendingBookmarks = void 0;
+			mark(1);
+			if (blocks) blocks.next++;
+		} else if (poIs(child, "w:sdt")) {
+			const content = poChildren(child).find((c) => poIs(c, "w:sdtContent"));
+			if (content) out.push(...parseBodyElements(poChildren(content), ctx, blocks, sdtRunProps));
 		}
-	} else if (poIs(child, "w:tbl")) out.push({
-		kind: "table",
-		table: parseTable(child, ctx)
-	});
-	else if (poIs(child, "w:sdt")) {
-		const content = poChildren(child).find((c) => poIs(c, "w:sdtContent"));
-		if (content) out.push(...parseBodyElements(poChildren(content), ctx));
 	}
 	return out;
 }
@@ -233,12 +398,11 @@ function scanForLoneDrawing(container, acc) {
 	for (const child of expandMcChildren(poChildren(container))) {
 		if (poIs(child, "w:pPr")) continue;
 		if (poIs(child, "w:r")) {
-			for (const rc of expandMcChildren(poChildren(child))) if (poIs(rc, "w:drawing")) {
-				if (!acc.drawing) acc.drawing = rc;
-			} else if (poIs(rc, "w:pict") || poIs(rc, "w:object")) {
-				if (!acc.vml && poFindDescendant(rc, "v:imagedata") !== void 0) acc.vml = rc;
+			for (const rc of expandMcChildren(poChildren(child))) if (poIs(rc, "w:drawing")) acc.drawings.push(rc);
+			else if (poIs(rc, "w:pict") || poIs(rc, "w:object")) {
+				if (!acc.vml && (poFindDescendant(rc, "v:imagedata") ?? poFindDescendant(rc, "v:textpath")) !== void 0) acc.vml = rc;
 			} else if (poIs(rc, "w:t") && poText(rc).length > 0) acc.hasOther = true;
-			else if (poIs(rc, "w:tab") || poIs(rc, "w:br") || poIs(rc, "w:noBreakHyphen")) acc.hasOther = true;
+			else if (poIs(rc, "w:tab") || poIs(rc, "w:ptab") || poIs(rc, "w:br") || poIs(rc, "w:noBreakHyphen")) acc.hasOther = true;
 			continue;
 		}
 		if (poIs(child, "m:oMath") || poIs(child, "m:oMathPara")) {
@@ -251,25 +415,56 @@ function scanForLoneDrawing(container, acc) {
 	}
 }
 function tryExtractDrawingFromParagraph(p, ctx) {
-	const scan = { hasOther: false };
+	const scan = {
+		hasOther: false,
+		drawings: []
+	};
 	scanForLoneDrawing(p, scan);
-	const { drawing, vml } = scan;
-	if (scan.hasOther || !drawing && !vml) return null;
-	const fromVml = drawing === void 0;
+	const { drawings, vml } = scan;
+	if (scan.hasOther || drawings.length === 0 && !vml) return null;
+	if (!vml && drawings.every((d) => poChildren(d).some((c) => poIs(c, "wp:anchor")))) return null;
 	const parseBody = (children) => parseBodyElements(children, ctx);
-	const content = fromVml ? parseVmlPicture(vml) : parseDrawing(drawing, ctx.resolveColor, parseBody);
-	if (!content) return null;
-	if (fromVml && content.kind === "image" && ctx.resolveImage?.(content.imageId) === void 0) return null;
 	const pPrNode = poChildren(p).find((c) => poIs(c, "w:pPr"));
-	const paragraphProperties = pPrNode ? parseParagraphProperties(poElementToFlat(pPrNode)) : {};
+	const paragraphProperties = pPrNode ? parseParagraphProperties(poElementToFlat(pPrNode), ctx.autoSpacingPt) : {};
+	if (drawings.length === 0) {
+		const content = parseVmlPicture(vml, parseBody);
+		if (!content) return null;
+		if (content.kind === "image" && ctx.resolveImage?.(content.imageId) === void 0) return null;
+		return blocksForDrawing(content, paragraphProperties, ctx);
+	}
+	const contents = drawings.map((d) => parseDrawing(d, ctx.resolveColor, parseBody, ctx.resolveImage, ctx.resolveChartPart, ctx.themeLineWidths, ctx.themeStyles, drawingTextParser)).filter((c) => c !== null);
+	if (contents.length > 1 && contents.some((c) => !c.float)) return null;
+	const out = [];
+	for (const content of contents) out.push(...blocksForDrawing(content, paragraphProperties, ctx) ?? []);
+	return out.length > 0 ? out : null;
+}
+/**
+* One parsed drawing as the body elements it becomes: a picture, a chart, a
+* shape, or the several shapes a SmartArt diagram draws.
+*
+* @param content              The parsed drawing.
+* @param paragraphProperties  The owning paragraph's properties, which a block
+*                             carries for its spacing and alignment.
+* @param ctx                  The document-wide parse context.
+* @returns The blocks, or `null` when the drawing renders nothing.
+*/
+function blocksForDrawing(content, paragraphProperties, ctx) {
 	if (content.kind === "image") {
 		const resource = ctx.resolveImage?.(content.imageId);
 		return [{
 			kind: "image",
 			image: {
 				...resource ? { resource } : {},
+				...content.outline ? { outline: content.outline } : {},
+				...content.shadow ? { shadow: content.shadow } : {},
 				width: content.width,
 				height: content.height,
+				...content.crop ? { crop: content.crop } : {},
+				...content.rotation60k ? { rotation60k: content.rotation60k } : {},
+				...content.flipH ? { flipH: true } : {},
+				...content.flipV ? { flipV: true } : {},
+				...content.relativeSize ? { relativeSize: content.relativeSize } : {},
+				...content.effectExtent ? { effectExtent: content.effectExtent } : {},
 				paragraphProperties,
 				...content.altText ? { altText: content.altText } : {},
 				...content.float ? { float: content.float } : {}
@@ -288,31 +483,39 @@ function tryExtractDrawingFromParagraph(p, ctx) {
 		}
 	}];
 	if (content.kind === "diagram") {
-		const spTree = ctx.resolveDiagram?.(content.dmRelId);
-		if (!spTree) {
+		const diagram = ctx.resolveDiagram?.(content.dmRelId);
+		if (!diagram) {
 			ctx.onLoss?.(noDiagramOverrideLoss());
 			return null;
 		}
-		const shapes = parseDiagramDrawing(spTree, diagramTransform(spTree, {
+		const frame = {
 			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) => ({
+		};
+		const nodes = parseDiagramNodes(diagram.spTree, diagramTransform(diagram.spTree, frame), ctx.resolveColor, void 0, diagram.resolveImage);
+		return [{
 			kind: "shape",
-			shape
-		})) : null;
+			shape: {
+				width: emuToPt(content.widthEmu),
+				height: emuToPt(content.heightEmu),
+				children: nodes.map(({ box, shape }) => ({
+					shape,
+					xPt: emuToPt(box.x),
+					yPt: emuToPt(box.y)
+				})),
+				geometry: {
+					kind: "preset",
+					preset: "rect",
+					adjust: /* @__PURE__ */ new Map()
+				},
+				fill: { kind: "none" },
+				paragraphProperties,
+				...content.altText ? { altText: content.altText } : {},
+				...content.float ? { float: content.float } : {}
+			}
+		}];
 	}
 	return [{
 		kind: "shape",
@@ -324,7 +527,7 @@ function tryExtractDrawingFromParagraph(p, ctx) {
 		}
 	}];
 }
-function parseParagraph(p, ctx, extraBookmarks) {
+function parseParagraph(p, ctx, extraBookmarks, anchored, sdtRunProps) {
 	const bookmarks = [...extraBookmarks ?? []];
 	for (const child of poChildren(p)) {
 		if (!poIs(child, "w:bookmarkStart")) continue;
@@ -332,7 +535,11 @@ function parseParagraph(p, ctx, extraBookmarks) {
 		if (bookmarkName !== void 0 && bookmarkName !== "" && bookmarkName !== "_GoBack") bookmarks.push(bookmarkName);
 	}
 	const pPr = poChildren(p).find((c) => poIs(c, "w:pPr"));
-	let properties = parseParagraphProperties(pPr ? poElementToFlat(pPr) : void 0);
+	let properties = parseParagraphProperties(pPr ? poElementToFlat(pPr) : void 0, ctx.autoSpacingPt);
+	if (pPr && poChildren(pPr).some((c) => poIs(c, "w:sectPr"))) properties = {
+		...properties,
+		sectionBreak: true
+	};
 	const mathPara = poChildren(p).find((c) => poIs(c, "m:oMathPara"));
 	if (mathPara && properties.alignment === void 0) {
 		const paraPr = poChildren(mathPara).find((c) => poIs(c, "m:oMathParaPr"));
@@ -343,19 +550,79 @@ function parseParagraph(p, ctx, extraBookmarks) {
 			alignment
 		};
 	}
+	const pTabs = collectPositionTabs(p);
+	if (pTabs.length > 0) properties = {
+		...properties,
+		tabs: [...properties.tabs ?? [], ...pTabs]
+	};
 	const collected = [];
-	collectRuns(p, collected, ctx);
+	collectRuns(p, collected, ctx, void 0, void 0, anchored, sdtRunProps);
 	return {
 		properties,
 		runs: applyFieldFsm(collected),
 		...bookmarks.length > 0 ? { bookmarks } : {}
 	};
 }
+/**
+* §17.3.3.15 `w:ptab` — the absolute-position tabs a paragraph holds, as stops.
+*
+* `@w:alignment` says which edge of the text column the tab reaches for; the
+* `left` alignment reaches for no edge at all (it goes to where the text
+* already is) and states no stop. `@w:relativeTo` distinguishes the margin from
+* the paragraph's indent, which for a stop measured from the text margin is the
+* same place.
+*
+* @param p The `w:p` element.
+* @returns One stop per position tab, in document order.
+*/
+function collectPositionTabs(p) {
+	const out = [];
+	const visit = (node) => {
+		for (const child of poChildren(node)) {
+			if (poIs(child, "w:ptab")) {
+				const alignment = poAttr(child, "alignment");
+				if (alignment === "center" || alignment === "right") out.push({
+					positionPt: pt(0),
+					alignment,
+					relativeTo: alignment
+				});
+				continue;
+			}
+			if (poIs(child, "w:r") || poIs(child, "w:hyperlink") || poIs(child, "w:ins")) visit(child);
+		}
+	};
+	visit(p);
+	return out;
+}
+/**
+* §17.16.5.44 `MACROBUTTON macroName displayText` — the display text, which is
+* everything after the macro's name and is all a reader ever sees of the field.
+*
+* @param instr The field instruction.
+* @returns The display text, or `undefined` when this is not a MACROBUTTON.
+*/
+function macroButtonText(instr) {
+	const shown = /^\s*MACROBUTTON\s+\S+\s(.*)$/su.exec(instr)?.[1]?.replace(/\s+$/u, "");
+	return shown !== void 0 && shown !== "" ? shown : void 0;
+}
 function parseFieldInstr(instr) {
-	if (!instr) return void 0;
-	const kw = /^\s*([A-Za-z]+)/.exec(instr)?.[1]?.toUpperCase();
+	const kw = fieldKeyword(instr);
 	return kw === "PAGE" ? "PAGE" : kw === "NUMPAGES" ? "NUMPAGES" : void 0;
 }
+function fieldKeyword(instr) {
+	if (instr === void 0) return void 0;
+	return /^\s*([A-Za-z]+)/.exec(instr)?.[1]?.toUpperCase();
+}
+var CHECKBOX_CLEAR = "☐";
+var CHECKBOX_CHECKED = "☒";
+var SILENT_FIELDS = new Set([
+	"SET",
+	"ASK",
+	"XE",
+	"TC",
+	"RD",
+	"PRIVATE"
+]);
 function synthesizeFieldRun(result, field, href) {
 	const first = result[0];
 	const linked = href ?? result.find((r) => r.href !== void 0)?.href;
@@ -385,7 +652,9 @@ function applyFieldFsm(collected) {
 				phase: "instr",
 				instr: "",
 				result: [],
-				depth: 0
+				depth: 0,
+				...c.checkBox !== void 0 ? { checkBox: c.checkBox } : {},
+				properties: c.run.properties
 			};
 			continue;
 		}
@@ -403,8 +672,20 @@ function applyFieldFsm(collected) {
 				continue;
 			}
 			const field = parseFieldInstr(st.instr);
-			if (field) out.push(synthesizeFieldRun(st.result, field));
-			else out.push(...st.result);
+			if (SILENT_FIELDS.has(fieldKeyword(st.instr) ?? "")) {} else if (field) out.push(synthesizeFieldRun(st.result, field));
+			else if (st.result.length > 0) out.push(...st.result);
+			else {
+				const shown = macroButtonText(st.instr);
+				const box = fieldKeyword(st.instr) === "FORMCHECKBOX" && st.checkBox !== void 0 ? st.checkBox ? CHECKBOX_CHECKED : CHECKBOX_CLEAR : void 0;
+				if (shown !== void 0) out.push({
+					text: shown,
+					properties: {}
+				});
+				else if (box !== void 0) out.push({
+					text: box,
+					properties: st.properties ?? {}
+				});
+			}
 			st = null;
 			continue;
 		}
@@ -417,7 +698,7 @@ function applyFieldFsm(collected) {
 	if (st) out.push(...st.result);
 	return out;
 }
-function collectRuns(container, out, ctx, href, anchor) {
+function collectRuns(container, out, ctx, href, anchor, anchored, sdtRunProps) {
 	for (const child of poChildren(container)) {
 		if (poIs(child, "w:pPr")) continue;
 		if (poIs(child, "w:commentRangeStart")) {
@@ -431,10 +712,14 @@ function collectRuns(container, out, ctx, href, anchor) {
 			continue;
 		}
 		if (poIs(child, "w:r")) {
-			const parsed = parseRun(child, ctx);
+			const parsed = parseRun(child, ctx, anchored);
 			const ranges = ctx.openCommentRanges && ctx.openCommentRanges.size > 0 ? [...ctx.openCommentRanges] : void 0;
-			const run = href !== void 0 || anchor !== void 0 || ranges !== void 0 ? {
+			const run = href !== void 0 || anchor !== void 0 || ranges !== void 0 || sdtRunProps ? {
 				...parsed.run,
+				...sdtRunProps ? { properties: {
+					...sdtRunProps,
+					...parsed.run.properties
+				} } : {},
 				...href !== void 0 ? { href } : {},
 				...anchor !== void 0 ? { anchor } : {},
 				...ranges !== void 0 ? { commentRangeRefs: ranges } : {}
@@ -442,7 +727,8 @@ function collectRuns(container, out, ctx, href, anchor) {
 			out.push({
 				run,
 				...parsed.fldChar ? { fldChar: parsed.fldChar } : {},
-				...parsed.instrText !== void 0 ? { instrText: parsed.instrText } : {}
+				...parsed.instrText !== void 0 ? { instrText: parsed.instrText } : {},
+				...parsed.checkBox !== void 0 ? { checkBox: parsed.checkBox } : {}
 			});
 			continue;
 		}
@@ -467,11 +753,11 @@ function collectRuns(container, out, ctx, href, anchor) {
 			const field = parseFieldInstr(poAttr(child, "instr"));
 			if (field) {
 				const inner = [];
-				collectRuns(child, inner, ctx, href, anchor);
+				collectRuns(child, inner, ctx, href, anchor, anchored, sdtRunProps);
 				out.push({ run: synthesizeFieldRun(applyFieldFsm(inner), field, href) });
 				continue;
 			}
-			collectRuns(child, out, ctx, href, anchor);
+			collectRuns(child, out, ctx, href, anchor, anchored, sdtRunProps);
 			continue;
 		}
 		if (tag && RUN_CONTAINER_TAGS.has(tag)) {
@@ -486,18 +772,111 @@ function collectRuns(container, out, ctx, href, anchor) {
 					if (bookmark !== void 0 && bookmark !== "") childAnchor = bookmark;
 				}
 			}
-			collectRuns(child, out, ctx, childHref, childAnchor);
+			const childProps = tag === "w:sdt" ? sdtRunProperties(child) ?? sdtRunProps : sdtRunProps;
+			collectRuns(child, out, ctx, childHref, childAnchor, anchored, childProps);
 		}
 	}
 }
-function parseRun(r, ctx) {
+/**
+* §17.5.2.28 `w:sdt/w:sdtPr/w:rPr` — the run properties a content control
+* lends what it holds, when it states any.
+*
+* @param sdt The `w:sdt` element.
+* @returns The properties, or `undefined` when the control states none.
+*/
+function sdtRunProperties(sdt) {
+	const sdtPr = poChildren(sdt).find((c) => poIs(c, "w:sdtPr"));
+	const rPr = sdtPr ? poChildren(sdtPr).find((c) => poIs(c, "w:rPr")) : void 0;
+	if (!rPr) return void 0;
+	const props = parseRunProperties(poElementToFlat(rPr));
+	return Object.keys(props).length > 0 ? props : void 0;
+}
+var WINGDINGS_TO_UNICODE = new Map([
+	[108, "●"],
+	[109, "❍"],
+	[110, "■"],
+	[113, "❑"],
+	[115, "□"],
+	[116, "❒"],
+	[118, "❖"],
+	[167, "▪"],
+	[168, "▫"],
+	[252, "✔"],
+	[253, "✖"],
+	[254, "☒"],
+	[251, "☑"],
+	[224, "⇦"],
+	[225, "⇨"],
+	[226, "⇧"],
+	[227, "⇩"],
+	[231, "←"],
+	[232, "→"],
+	[233, "↑"],
+	[234, "↓"],
+	[235, "↖"],
+	[236, "↗"],
+	[237, "↙"],
+	[238, "↘"]
+]);
+var SYMBOL_TO_UNICODE = new Map([
+	[34, "∀"],
+	[36, "∃"],
+	[64, "≅"],
+	[92, "∴"],
+	[97, "α"],
+	[98, "β"],
+	[99, "χ"],
+	[100, "δ"],
+	[101, "ε"],
+	[102, "φ"],
+	[103, "γ"],
+	[104, "η"],
+	[105, "ι"],
+	[107, "κ"],
+	[108, "λ"],
+	[109, "μ"],
+	[110, "ν"],
+	[111, "ο"],
+	[112, "π"],
+	[113, "θ"],
+	[114, "ρ"],
+	[115, "σ"],
+	[116, "τ"],
+	[117, "υ"],
+	[119, "ω"],
+	[120, "ξ"],
+	[121, "ψ"],
+	[122, "ζ"],
+	[165, "∞"],
+	[177, "±"],
+	[179, "≥"],
+	[180, "×"],
+	[183, "•"],
+	[184, "÷"],
+	[185, "≠"],
+	[186, "≡"],
+	[187, "≈"],
+	[214, "√"],
+	[229, "∑"]
+]);
+function symbolChar(font, char) {
+	const cp = char !== void 0 ? Number.parseInt(char, 16) : NaN;
+	if (!Number.isFinite(cp) || cp <= 0) return "";
+	const low = cp & 255;
+	const mapped = (/wingdings|webdings|zapf/iu.test(font ?? "") ? WINGDINGS_TO_UNICODE : /symbol/iu.test(font ?? "") ? SYMBOL_TO_UNICODE : void 0)?.get(low);
+	if (mapped !== void 0) return mapped;
+	return String.fromCodePoint(cp);
+}
+function parseRun(r, ctx, anchored) {
 	const rPr = poChildren(r).find((c) => poIs(c, "w:rPr"));
 	const properties = parseRunProperties(rPr ? poElementToFlat(rPr) : void 0);
 	let text = "";
 	let pageBreak = false;
+	let columnBreak = false;
 	let inlineImage;
 	let fldChar;
 	let instrText;
+	let checkBox;
 	let footnoteRef;
 	let endnoteRef;
 	let commentRef;
@@ -507,6 +886,7 @@ function parseRun(r, ctx) {
 		if (poIs(child, "w:fldChar")) {
 			const t = poAttr(child, "fldCharType");
 			if (t === "begin" || t === "separate" || t === "end") fldChar = t;
+			if (t === "begin") checkBox = fieldCheckBoxState(child);
 			continue;
 		}
 		if (poIs(child, "w:instrText")) {
@@ -534,30 +914,49 @@ function parseRun(r, ctx) {
 		}
 		if (poIs(child, "w:t")) text += poText(child);
 		else if (poIs(child, "w:tab")) text += "	";
-		else if (poIs(child, "w:br")) if (poAttr(child, "type") === "page") pageBreak = true;
-		else text += "\n";
+		else if (poIs(child, "w:ptab")) text += "	";
+		else if (poIs(child, "w:br")) {
+			const breakType = poAttr(child, "type");
+			if (breakType === "page") pageBreak = true;
+			else {
+				if (breakType === "column") columnBreak = true;
+				text += "\n";
+			}
+		} else if (poIs(child, "w:sym")) text += symbolChar(poAttr(child, "font"), poAttr(child, "char"));
 		else if (poIs(child, "w:noBreakHyphen")) text += "‑";
 		else if (poIs(child, "w:softHyphen")) text += "­";
 		else if (poIs(child, "w:drawing")) {
-			const content = parseDrawing(child, defaultColorResolver);
-			if (content && content.kind === "image") {
+			const content = parseDrawing(child, ctx.resolveColor, (children) => parseBodyElements(children, ctx), ctx.resolveImage, ctx.resolveChartPart, ctx.themeLineWidths, ctx.themeStyles, drawingTextParser);
+			if (content?.float && anchored) anchored.push(...blocksForDrawing(content, {}, ctx) ?? []);
+			else if ((content?.kind === "chart" || content?.kind === "shape" || content?.kind === "diagram") && anchored) anchored.push(...blocksForDrawing(content, {}, ctx) ?? []);
+			else if (content && content.kind === "image") {
 				const resource = ctx.resolveImage?.(content.imageId);
 				inlineImage = {
 					...resource ? { resource } : {},
+					...content.outline ? { outline: content.outline } : {},
+					...content.shadow ? { shadow: content.shadow } : {},
 					width: content.width,
-					height: content.height
+					height: content.height,
+					...content.crop ? { crop: content.crop } : {},
+					...content.rotation60k ? { rotation60k: content.rotation60k } : {},
+					...content.flipH ? { flipH: true } : {},
+					...content.flipV ? { flipV: true } : {},
+					...content.effectExtent ? { effectExtent: content.effectExtent } : {}
 				};
 			}
 		} else if (poIs(child, "w:pict") || poIs(child, "w:object")) {
-			const content = parseVmlPicture(child);
-			if (content && content.kind === "image") {
+			const content = parseVmlPicture(child, (children) => parseBodyElements(children, ctx));
+			if (content?.float && anchored) anchored.push(...blocksForDrawing(content, {}, ctx) ?? []);
+			else if (content && content.kind === "image") {
 				const resource = ctx.resolveImage?.(content.imageId);
 				if (resource) inlineImage = {
 					resource,
+					...content.outline ? { outline: content.outline } : {},
+					...content.shadow ? { shadow: content.shadow } : {},
 					width: content.width,
 					height: content.height
 				};
-			}
+			} else if (content && anchored) anchored.push(...blocksForDrawing(content, {}, ctx) ?? []);
 		}
 	}
 	return {
@@ -566,14 +965,36 @@ function parseRun(r, ctx) {
 			properties,
 			...inlineImage ? { inlineImage } : {},
 			...pageBreak ? { pageBreak: true } : {},
+			...columnBreak ? { columnBreak: true } : {},
 			...footnoteRef !== void 0 ? { footnoteRef } : {},
 			...endnoteRef !== void 0 ? { endnoteRef } : {},
 			...commentRef !== void 0 ? { commentRef } : {},
 			...noteNumber ? { noteNumber: true } : {}
 		},
 		...fldChar ? { fldChar } : {},
-		...instrText !== void 0 ? { instrText } : {}
+		...instrText !== void 0 ? { instrText } : {},
+		...checkBox !== void 0 ? { checkBox } : {}
+	};
+}
+/**
+* §17.16.32 `w:fldChar/w:ffData/w:checkBox` — whether a form checkbox is
+* ticked: its current `w:checked` if it states one, else the `w:default` it was
+* created with, else clear.
+*
+* @param fldChar The field's `w:fldChar` (only a `begin` carries `w:ffData`).
+* @returns The state, or undefined when this is not a checkbox field.
+*/
+function fieldCheckBoxState(fldChar) {
+	const ffData = poChildren(fldChar).find((c) => poIs(c, "w:ffData"));
+	const box = ffData ? poChildren(ffData).find((c) => poIs(c, "w:checkBox")) : void 0;
+	if (!box) return void 0;
+	const flag = (name) => {
+		const el = poChildren(box).find((c) => poIs(c, name));
+		if (!el) return void 0;
+		const val = poAttr(el, "val");
+		return val === void 0 || val === "" || val !== "0" && val !== "false";
 	};
+	return flag("w:checked") ?? flag("w:default") ?? false;
 }
 function elementTag(node) {
 	for (const key of Object.keys(node)) if (key !== ":@" && key !== "#text") return key;
@@ -590,7 +1011,7 @@ function elementTag(node) {
 * @returns A map from note id to its body content.
 */
 function parseNotes(notesXml, rootTag, noteTag, ctx = DEFAULT_PARSE_CONTEXT) {
-	const xml = decoder.decode(notesXml);
+	const xml = resolveInternalEntities(decoder.decode(notesXml));
 	const root = poFindByPath(parser.parse(xml), [rootTag]);
 	const out = /* @__PURE__ */ new Map();
 	if (!root) return out;
@@ -605,7 +1026,7 @@ function parseNotes(notesXml, rootTag, noteTag, ctx = DEFAULT_PARSE_CONTEXT) {
 	return out;
 }
 function parseCommentsRaw(commentsXml, ctx) {
-	const xml = decoder.decode(commentsXml);
+	const xml = resolveInternalEntities(decoder.decode(commentsXml));
 	const root = poFindByPath(parser.parse(xml), ["w:comments"]);
 	const comments = /* @__PURE__ */ new Map();
 	const paraIds = /* @__PURE__ */ new Map();
@@ -649,7 +1070,7 @@ function parseCommentsRaw(commentsXml, ctx) {
 * @returns A map from `paraId` to its {@link CommentExtension}.
 */
 function parseCommentsExtended(xml) {
-	const tree = parser.parse(decoder.decode(xml));
+	const tree = parser.parse(resolveInternalEntities(decoder.decode(xml)));
 	const out = /* @__PURE__ */ new Map();
 	const root = tree.find((n) => poIsLocal(n, "commentsEx"));
 	if (!root) return out;
@@ -711,7 +1132,7 @@ function parseCommentThreads(commentsXml, commentsExtendedXml, ctx = DEFAULT_PAR
 * @returns A map from author name to userId.
 */
 function parsePeople(xml) {
-	const tree = parser.parse(decoder.decode(xml));
+	const tree = parser.parse(resolveInternalEntities(decoder.decode(xml)));
 	const out = /* @__PURE__ */ new Map();
 	const root = tree.find((n) => poIsLocal(n, "people"));
 	if (!root) return out;
@@ -751,4 +1172,4 @@ function applyAuthorIds(comments, people) {
 	return out;
 }
 //#endregion
-export { DEFAULT_PARSE_CONTEXT, EMPTY_SECTION, applyAuthorIds, parseBodyElements, parseCommentThreads, parseCommentsExtended, parseDocument, parseHeaderFooter, parseNotes, parsePeople, parseSections };
+export { DEFAULT_PARSE_CONTEXT, EMPTY_SECTION, applyAuthorIds, bodyIndexForBlock, newBlockCounter, parseBackgroundColor, parseBackgroundFill, parseBodyElements, parseCommentThreads, parseCommentsExtended, parseDocument, parseHeaderFooter, parseNotes, parsePeople, parseSections, sdtRunProperties };