@stll/folio-core 0.28.1 → 0.29.0

package/dist/prosemirror/conversion/toProseDoc.js

@@ -6,6 +6,7 @@ import { mergeTextFormatting } from "../../utils/textFormattingMerge.js";
 import { emuToPixels } from "../../utils/units.js";
 import { setAutospacingBaseValue } from "../autospacingBase.js";
 import { buildRunFormattingOverrideAttrs } from "../extensions/marks/RunFormattingOverrideExtension.js";
+import { stampNumberedRefFieldBaselines } from "../numberedRefFields.js";
 import { directionFromBidi } from "../paragraphDirection.js";
 import { schema } from "../schema/index.js";
 import { cascadeStyleTextFormatting } from "../styles/styleToggleCascade.js";
@@ -36,6 +37,56 @@ const createHyperlinkInstanceIndexAllocator = () => {
 	return () => index++;
 };
 /**
+* Find bookmark pairs across every editable inline wrapper in a document story.
+* Each endpoint is converted at its own structural position, so a range may
+* start outside a hyperlink and end inside it (or the inverse).
+*/
+const collectPairedBookmarkIds = (blocks) => {
+	const counts = /* @__PURE__ */ new Map();
+	let position = 0;
+	const countBoundary = (id, type) => {
+		const count = counts.get(id) ?? {
+			starts: 0,
+			ends: 0
+		};
+		if (type === "start") {
+			count.starts += 1;
+			count.firstStart ??= position;
+		} else {
+			count.ends += 1;
+			count.firstEnd ??= position;
+		}
+		position += 1;
+		counts.set(id, count);
+	};
+	const visitRun = (run) => {
+		for (const content of run.content) if (content.type === "shape" && content.shape.textBody) visitBlocks(content.shape.textBody.content);
+	};
+	const visitHyperlink = (hyperlink) => {
+		for (const child of hyperlink.children) if (child.type === "bookmarkStart") countBoundary(child.id, "start");
+		else if (child.type === "bookmarkEnd") countBoundary(child.id, "end");
+		else visitRun(child);
+	};
+	const visitParagraphContent = (content) => {
+		if (content.type === "bookmarkStart") countBoundary(content.id, "start");
+		else if (content.type === "bookmarkEnd") countBoundary(content.id, "end");
+		else if (content.type === "run") visitRun(content);
+		else if (content.type === "hyperlink") visitHyperlink(content);
+		else if (content.type === "simpleField") for (const child of content.content) if (child.type === "hyperlink") visitHyperlink(child);
+		else visitRun(child);
+		else if (content.type === "complexField") for (const run of [...content.fieldCode ?? [], ...content.fieldResult]) visitRun(run);
+		else if (content.type === "inlineSdt") for (const child of content.content) visitParagraphContent(child);
+		else if (content.type === "insertion" || content.type === "deletion" || content.type === "moveFrom" || content.type === "moveTo") for (const child of content.content) visitParagraphContent(child);
+	};
+	const visitBlocks = (nestedBlocks) => {
+		for (const block of nestedBlocks) if (block.type === "paragraph") for (const content of block.content) visitParagraphContent(content);
+		else if (block.type === "table") for (const row of block.rows) for (const cell of row.cells) visitBlocks(cell.content);
+		else visitBlocks(block.content);
+	};
+	visitBlocks(blocks);
+	return new Set([...counts].flatMap(([id, count]) => count.starts === 1 && count.ends === 1 && count.firstStart !== void 0 && count.firstEnd !== void 0 && count.firstStart < count.firstEnd ? [id] : []));
+};
+/**
 * Convert a Document to a ProseMirror document
 *
 * @param document - The Document to convert
@@ -50,7 +101,8 @@ function toProseDoc(document, options) {
 	const conversionContext = {
 		theme,
 		nextTextBoxGroupId,
-		nextHyperlinkInstanceIndex: createHyperlinkInstanceIndexAllocator()
+		nextHyperlinkInstanceIndex: createHyperlinkInstanceIndexAllocator(),
+		pairedBookmarkIds: collectPairedBookmarkIds(paragraphs)
 	};
 	const convertBodyBlocks = (blocks) => {
 		const out = [];
@@ -85,7 +137,7 @@ function toProseDoc(document, options) {
 	};
 	nodes.push(...convertBodyBlocks(paragraphs));
 	if (nodes.length === 0) nodes.push(schema.node("paragraph", {}, []));
-	const pmDoc = schema.node("doc", null, nodes);
+	const pmDoc = stampNumberedRefFieldBaselines(schema.node("doc", null, nodes));
 	assertValidProseMirrorDocument(pmDoc, "Document conversion produced an invalid ProseMirror document");
 	return pmDoc;
 }
@@ -126,7 +178,7 @@ function convertBlockSdt(blockSdt, convertBlocks) {
 * Resolves style-based text formatting and passes it to runs so that
 * paragraph styles (like Heading1) apply their font size, color, etc.
 */
-function convertParagraph(paragraph, styleResolver, nextHyperlinkInstanceIndex, activeCommentIds, extraRunFormatting, tableParagraphOverlay, textBoxAnchors) {
+function convertParagraph(paragraph, styleResolver, nextHyperlinkInstanceIndex, pairedBookmarkIds, activeCommentIds, extraRunFormatting, tableParagraphOverlay, textBoxAnchors) {
 	const attrs = paragraphFormattingToAttrs(paragraph, styleResolver, tableParagraphOverlay);
 	const isTocParagraph = attrs._tableOfContentsLevel !== void 0;
 	const inlineNodes = [];
@@ -209,42 +261,56 @@ function convertParagraph(paragraph, styleResolver, nextHyperlinkInstanceIndex,
 	const emitTrackedChange = (change, markType, moveKind) => {
 		emitInlineNodes(convertTrackedChange(change, markType, nextHyperlinkInstanceIndex, getInheritedRunFormatting, styleResolver, moveKind, textBoxAnchors));
 	};
-	for (const content of paragraph.content) {
-		if (content.type === "commentRangeStart") commentIds.add(content.id);
-		else if (content.type === "commentRangeEnd") commentIds.delete(content.id);
-		else if (content.type === "commentReference") anchorPointComment(inlineNodes, content.id);
-		else if (content.type === "run") emitInlineNodes(convertRun(content, getInheritedRunFormatting(content.formatting), styleResolver, textBoxAnchors));
-		else if (content.type === "hyperlink") {
-			const linkNodes = convertHyperlink(content, {
-				getInheritedRunFormatting,
-				styleResolver,
-				hyperlinkIndex: nextHyperlinkInstanceIndex(),
-				textBoxAnchors
-			});
-			if (linkNodes.length === 0) {
-				emptyHyperlinks ??= [];
-				emptyHyperlinks.push({
-					offset: inlineOffset,
-					...content.href !== void 0 ? { href: content.href } : {},
-					...content.anchor !== void 0 ? { anchor: content.anchor } : {},
-					...content.tooltip !== void 0 ? { tooltip: content.tooltip } : {},
-					...content.rId !== void 0 ? { rId: content.rId } : {}
-				});
-				continue;
-			}
-			emitInlineNodes(linkNodes);
-		} else if (content.type === "simpleField" || content.type === "complexField") emitInlineNode(convertField(content, getInheritedRunFormatting, styleResolver));
-		else if (content.type === "inlineSdt") emitInlineNode(convertInlineSdt(content, nextHyperlinkInstanceIndex, getInheritedRunFormatting, styleResolver, textBoxAnchors));
-		else if (content.type === "insertion" || content.type === "moveTo") emitTrackedChange(content, "insertion", content.type === "moveTo" ? "moveTo" : null);
-		else if (content.type === "deletion" || content.type === "moveFrom") emitTrackedChange(content, "deletion", content.type === "moveFrom" ? "moveFrom" : null);
-		else if (content.type === "mathEquation") emitInlineNode(convertMathEquation(content));
-		if (content.type === "bookmarkStart") {
-			if (!bookmarksArr) bookmarksArr = [];
-			bookmarksArr.push({
-				id: content.id,
-				name: content.name
+	for (const content of paragraph.content) if (content.type === "commentRangeStart") commentIds.add(content.id);
+	else if (content.type === "commentRangeEnd") commentIds.delete(content.id);
+	else if (content.type === "commentReference") anchorPointComment(inlineNodes, content.id);
+	else if (content.type === "run") emitInlineNodes(convertRun(content, getInheritedRunFormatting(content.formatting), styleResolver, textBoxAnchors));
+	else if (content.type === "hyperlink") {
+		const linkNodes = convertHyperlink(content, {
+			getInheritedRunFormatting,
+			styleResolver,
+			hyperlinkIndex: nextHyperlinkInstanceIndex(),
+			textBoxAnchors
+		});
+		if (linkNodes.length === 0) {
+			emptyHyperlinks ??= [];
+			emptyHyperlinks.push({
+				offset: inlineOffset,
+				...content.href !== void 0 ? { href: content.href } : {},
+				...content.anchor !== void 0 ? { anchor: content.anchor } : {},
+				...content.tooltip !== void 0 ? { tooltip: content.tooltip } : {},
+				...content.rId !== void 0 ? { rId: content.rId } : {}
 			});
+			continue;
 		}
+		emitInlineNodes(linkNodes);
+	} else if (content.type === "simpleField" || content.type === "complexField") emitInlineNode(convertField(content, {
+		getInheritedRunFormatting,
+		styleResolver,
+		nextHyperlinkInstanceIndex,
+		textBoxAnchors
+	}));
+	else if (content.type === "inlineSdt") emitInlineNode(convertInlineSdt(content, nextHyperlinkInstanceIndex, getInheritedRunFormatting, styleResolver, textBoxAnchors));
+	else if (content.type === "insertion" || content.type === "moveTo") emitTrackedChange(content, "insertion", content.type === "moveTo" ? "moveTo" : null);
+	else if (content.type === "deletion" || content.type === "moveFrom") emitTrackedChange(content, "deletion", content.type === "moveFrom" ? "moveFrom" : null);
+	else if (content.type === "mathEquation") emitInlineNode(convertMathEquation(content));
+	else if (content.type === "bookmarkStart" && pairedBookmarkIds.has(content.id)) emitInlineNode(schema.node("bookmarkBoundary", {
+		type: "start",
+		id: content.id,
+		name: content.name,
+		colFirst: content.colFirst,
+		colLast: content.colLast
+	}));
+	else if (content.type === "bookmarkEnd" && pairedBookmarkIds.has(content.id)) emitInlineNode(schema.node("bookmarkBoundary", {
+		type: "end",
+		id: content.id
+	}));
+	else if (content.type === "bookmarkStart") {
+		if (!bookmarksArr) bookmarksArr = [];
+		bookmarksArr.push({
+			id: content.id,
+			name: content.name
+		});
 	}
 	if (bookmarksArr) attrs.bookmarks = bookmarksArr;
 	if (emptyHyperlinks) attrs._emptyHyperlinks = emptyHyperlinks;
@@ -313,7 +379,7 @@ function convertTrackedChange(change, markType, nextHyperlinkInstanceIndex, getI
 	});
 }
 function canCarryTrackedRunMark(node, markType) {
-	return node.isText || node.isInline && node.type.allowsMarkType(markType) && (node.type.name === "image" || node.type.name === "shape" || node.type.name === "hardBreak" || node.type.name === "tab" || node.type.name === "symbol" || node.type.name === "textBoxAnchor");
+	return node.isText || node.isInline && node.type.allowsMarkType(markType) && (node.type.name === "image" || node.type.name === "shape" || node.type.name === "hardBreak" || node.type.name === "tab" || node.type.name === "symbol" || node.type.name === "bookmarkBoundary" || node.type.name === "textBoxAnchor");
 }
 /**
 * Convert ParagraphFormatting to ProseMirror paragraph attrs
@@ -1011,7 +1077,8 @@ function standaloneTableCellToProseMirror(cell, nodeType) {
 		context: {
 			theme: null,
 			nextTextBoxGroupId: createTextBoxGroupIdFactory(),
-			nextHyperlinkInstanceIndex: createHyperlinkInstanceIndexAllocator()
+			nextHyperlinkInstanceIndex: createHyperlinkInstanceIndexAllocator(),
+			pairedBookmarkIds: collectPairedBookmarkIds(cell.content)
 		},
 		isHeader: nodeType === "tableHeader",
 		gridWidthPercent: void 0,
@@ -1024,31 +1091,47 @@ function standaloneTableCellToProseMirror(cell, nodeType) {
 		defaultCellMargins: void 0
 	});
 }
-/**
-* Convert a SimpleField or ComplexField to a ProseMirror field node.
-* Preserves run formatting (bold, fontSize, color, etc.) as PM marks.
-* Accepts a run formatting resolver so fields inherit paragraph-level
-* formatting the same way regular text runs do.
-*/
-function convertField(field, getInheritedRunFormatting, styleResolver) {
+function convertField(field, { getInheritedRunFormatting, styleResolver, nextHyperlinkInstanceIndex, textBoxAnchors }) {
 	let displayText = "";
 	let fieldFormatting;
-	const runs = field.type === "simpleField" ? field.content : field.fieldResult;
-	for (const r of runs) if (r.type === "run") {
-		for (const c of r.content) if (c.type === "text") displayText += c.text;
-		if (!fieldFormatting && r.formatting) fieldFormatting = r.formatting;
+	const inlineNodes = [];
+	const hasStructuredSourceContent = field.type === "simpleField" && field.content.some((content) => content.type === "hyperlink");
+	const appendRun = (run) => {
+		for (const content of run.content) if (content.type === "text") displayText += content.text;
+		fieldFormatting ??= run.formatting;
+		if (!hasStructuredSourceContent) return;
+		inlineNodes.push(...convertRun(run, getInheritedRunFormatting(run.formatting, field.fieldType), styleResolver, textBoxAnchors));
+	};
+	if (field.type === "simpleField") for (const content of field.content) {
+		if (content.type === "run") {
+			appendRun(content);
+			continue;
+		}
+		for (const child of content.children) if (child.type === "run") {
+			for (const runContent of child.content) if (runContent.type === "text") displayText += runContent.text;
+			fieldFormatting ??= child.formatting;
+		}
+		inlineNodes.push(...convertHyperlink(content, {
+			getInheritedRunFormatting: (formatting) => getInheritedRunFormatting(formatting, field.fieldType),
+			styleResolver,
+			hyperlinkIndex: nextHyperlinkInstanceIndex(),
+			textBoxAnchors
+		}));
 	}
+	else for (const run of field.fieldResult) appendRun(run);
 	if (!fieldFormatting && field.type === "complexField" && field.fieldResult.length === 0) fieldFormatting = field.formatting;
 	const inheritedFormatting = getInheritedRunFormatting(fieldFormatting, field.fieldType);
 	const { marks } = buildRunMarks(fieldFormatting, inheritedFormatting, styleResolver);
-	return schema.node("field", {
+	const hasConvertedHyperlinkContent = inlineNodes.some((node) => node.marks.some((mark) => mark.type.name === "hyperlink"));
+	const createStructuredField = hasStructuredSourceContent && hasConvertedHyperlinkContent;
+	return schema.node(createStructuredField ? "structuredField" : "field", {
 		fieldType: field.fieldType,
 		instruction: field.instruction,
 		displayText,
 		fieldKind: field.type === "simpleField" ? "simple" : "complex",
 		fldLock: field.fldLock ?? false,
 		dirty: field.dirty ?? false
-	}, void 0, marks);
+	}, createStructuredField ? inlineNodes : void 0, marks);
 }
 /**
 * Convert a MathEquation to a ProseMirror math node.
@@ -1078,7 +1161,12 @@ function convertInlineSdt(sdt, nextHyperlinkInstanceIndex, getInheritedRunFormat
 		});
 		inlineNodes.push(...linkNodes);
 	} else if (content.type === "simpleField" || content.type === "complexField") {
-		const fieldNode = convertField(content, getInheritedRunFormatting, styleResolver);
+		const fieldNode = convertField(content, {
+			getInheritedRunFormatting,
+			styleResolver,
+			nextHyperlinkInstanceIndex,
+			textBoxAnchors
+		});
 		if (fieldNode) inlineNodes.push(fieldNode);
 	} else if (content.type === "inlineSdt") {
 		const nestedSdt = convertInlineSdt(content, nextHyperlinkInstanceIndex, getInheritedRunFormatting, styleResolver, textBoxAnchors);
@@ -1338,11 +1426,30 @@ function convertHyperlink(hyperlink, { getInheritedRunFormatting, styleResolver,
 		rId: hyperlink.rId,
 		_docxHyperlinkIndex: hyperlinkIndex
 	});
-	for (const child of hyperlink.children) if (child.type === "run") {
-		const inheritedFormatting = getInheritedRunFormatting(child.formatting);
-		const { marks: runMarks, mergedFormatting } = buildRunMarks(child.formatting, inheritedFormatting, styleResolver);
-		const allMarks = [...runMarks, linkMark];
-		for (const content of child.content) nodes.push(...convertRunContent(content, allMarks, mergedFormatting, textBoxAnchors));
+	for (const child of hyperlink.children) {
+		if (child.type === "bookmarkStart") {
+			nodes.push(schema.node("bookmarkBoundary", {
+				type: "start",
+				id: child.id,
+				name: child.name,
+				colFirst: child.colFirst,
+				colLast: child.colLast
+			}, void 0, [linkMark]));
+			continue;
+		}
+		if (child.type === "bookmarkEnd") {
+			nodes.push(schema.node("bookmarkBoundary", {
+				type: "end",
+				id: child.id
+			}, void 0, [linkMark]));
+			continue;
+		}
+		if (child.type === "run") {
+			const inheritedFormatting = getInheritedRunFormatting(child.formatting);
+			const { marks: runMarks, mergedFormatting } = buildRunMarks(child.formatting, inheritedFormatting, styleResolver);
+			const allMarks = [...runMarks, linkMark];
+			for (const content of child.content) nodes.push(...convertRunContent(content, allMarks, mergedFormatting, textBoxAnchors));
+		}
 	}
 	return nodes;
 }
@@ -1511,7 +1618,7 @@ function convertShape(shape) {
 }
 function convertParagraphWithTextBoxes(block, styleResolver, { textBoxGroupId, context, extraRunFormatting, tableParagraphOverlay }) {
 	const { textBoxes, textBoxAnchors } = extractTextBoxesFromParagraph(block, textBoxGroupId);
-	const pmParagraph = convertParagraph(block, styleResolver, context.nextHyperlinkInstanceIndex, void 0, extraRunFormatting, tableParagraphOverlay, textBoxAnchors);
+	const pmParagraph = convertParagraph(block, styleResolver, context.nextHyperlinkInstanceIndex, context.pairedBookmarkIds, void 0, extraRunFormatting, tableParagraphOverlay, textBoxAnchors);
 	const nodes = [];
 	const isEmptyAfterExtraction = textBoxes.length > 0 && !hasContentBesidesTextBoxAnchors(pmParagraph);
 	const keepWrapperParagraph = isEmptyAfterExtraction && hasParagraphBoundaryPayload(block, pmParagraph);
@@ -1773,7 +1880,8 @@ function headerFooterToProseDoc(content, options) {
 	const conversionContext = {
 		theme,
 		nextTextBoxGroupId,
-		nextHyperlinkInstanceIndex: createHyperlinkInstanceIndexAllocator()
+		nextHyperlinkInstanceIndex: createHyperlinkInstanceIndexAllocator(),
+		pairedBookmarkIds: collectPairedBookmarkIds(content)
 	};
 	const convertBlocks = (blocks) => {
 		const out = [];
@@ -1787,7 +1895,7 @@ function headerFooterToProseDoc(content, options) {
 	};
 	nodes.push(...convertBlocks(content));
 	if (nodes.length === 0) nodes.push(schema.node("paragraph", {}, []));
-	const pmDoc = schema.node("doc", null, nodes);
+	const pmDoc = stampNumberedRefFieldBaselines(schema.node("doc", null, nodes));
 	assertValidProseMirrorDocument(pmDoc, "Header/footer conversion produced an invalid ProseMirror document");
 	return pmDoc;
 }

package/dist/prosemirror/extensions/StarterKit.js

@@ -43,7 +43,8 @@ import { EmbossExtension, EmphasisMarkExtension, ImprintExtension, TextOutlineEx
 import { DeletionExtension, InsertionExtension, RunPropertyChangeExtension } from "./marks/TrackedChangeExtensions.js";
 import { UnderlineExtension } from "./marks/UnderlineExtension.js";
 import { BlockSdtExtension } from "./nodes/BlockSdtExtension.js";
-import { FieldExtension } from "./nodes/FieldExtension.js";
+import { BookmarkBoundaryExtension } from "./nodes/BookmarkBoundaryExtension.js";
+import { FieldExtension, StructuredFieldExtension } from "./nodes/FieldExtension.js";
 import { HardBreakExtension } from "./nodes/HardBreakExtension.js";
 import { HorizontalRuleExtension } from "./nodes/HorizontalRuleExtension.js";
 import { ImageExtension } from "./nodes/ImageExtension.js";
@@ -64,6 +65,11 @@ import { TextBoxExtension } from "./nodes/TextBoxExtension.js";
 function createStarterKit(options = {}) {
 	const disabled = options.disable ? new Set(options.disable) : /* @__PURE__ */ new Set();
 	const extensions = [];
+	let internalClipboardToken;
+	const getInternalClipboardToken = () => {
+		internalClipboardToken ??= globalThis.crypto.randomUUID();
+		return internalClipboardToken;
+	};
 	function add(name, ext) {
 		if (!disabled.has(name)) extensions.push(ext);
 	}
@@ -105,12 +111,13 @@ function createStarterKit(options = {}) {
 	add("insertion", InsertionExtension());
 	add("deletion", DeletionExtension());
 	add("runPropertyChange", RunPropertyChangeExtension());
+	add("bookmarkBoundary", BookmarkBoundaryExtension({ getInternalClipboardToken }));
 	add("hardBreak", HardBreakExtension());
 	add("tab", TabExtension());
 	add("symbol", SymbolExtension());
 	add("image", ImageExtension());
 	add("textBox", TextBoxExtension());
-	add("textBoxAnchor", TextBoxAnchorExtension());
+	add("textBoxAnchor", TextBoxAnchorExtension({ getInternalClipboardToken }));
 	add("shape", ShapeExtension());
 	add("imageDrag", ImageDragExtension());
 	add("imagePaste", ImagePasteExtension());
@@ -120,11 +127,12 @@ function createStarterKit(options = {}) {
 	add("pageBreak", PageBreakExtension());
 	add("renderedPageBreak", RenderedPageBreakExtension());
 	add("field", FieldExtension());
+	add("field", StructuredFieldExtension({ getInternalClipboardToken }));
 	add("sdt", SdtExtension());
 	add("blockSdt", BlockSdtExtension());
 	add("math", MathExtension());
 	if (!disabled.has("table")) extensions.push(...createTableExtensions());
-	add("pasteCleanup", PasteCleanupExtension());
+	add("pasteCleanup", PasteCleanupExtension({ getInternalClipboardToken }));
 	add("pasteStyleInliner", PasteStyleInlinerExtension());
 	add("list", ListExtension());
 	add("baseKeymap", BaseKeymapExtension());

package/dist/prosemirror/extensions/features/AutoBidiDetectionExtension.js

@@ -9,12 +9,16 @@ const autoBidiDetectionKey = new PluginKey("autoBidiDetection");
 const paragraphDirectionalText = (node) => {
 	let text = "";
 	node.descendants((child) => {
+		if (child.marks.some((mark) => mark.type.name === "deletion")) return false;
 		if (child.isText) {
-			if (!child.marks.some((mark) => mark.type.name === "deletion")) text += child.text ?? "";
-		} else if (child.type.name === "field") {
-			const display = child.attrs["displayText"];
-			if (typeof display === "string") text += display;
+			text += child.text ?? "";
+			return false;
 		}
+		if (child.isLeaf && child.textContent) {
+			text += child.textContent;
+			return false;
+		}
+		return true;
 	});
 	return text;
 };

package/dist/prosemirror/extensions/features/PasteCleanupExtension.d.ts

@@ -14,6 +14,9 @@ import { Extension } from "../types.js";
  * 2. `Mod-Alt-v` — "paste without formatting", inserting clipboard text with
  *    the source formatting stripped (see {@link pasteWithoutFormatting}).
  */
-declare const PasteCleanupExtension: (options?: Partial<Record<string, unknown>> | undefined) => Extension;
+type PasteCleanupOptions = {
+  getInternalClipboardToken?: () => string;
+};
+declare const PasteCleanupExtension: (options?: Partial<PasteCleanupOptions> | undefined) => Extension;
 //#endregion
 export { PasteCleanupExtension };

package/dist/prosemirror/extensions/features/PasteCleanupExtension.js

@@ -1,7 +1,7 @@
 import { pasteWithoutFormatting } from "../../commands/pastePlainText.js";
 import { createExtension } from "../create.js";
 import { Priority } from "../types.js";
-import { cleanPastedHtml } from "./pasteCleanup.js";
+import { cleanPastedHtml, removeUnpairedBookmarkBoundaries } from "./pasteCleanup.js";
 import { Plugin } from "prosemirror-state";
 //#region src/prosemirror/extensions/features/PasteCleanupExtension.ts
 /**
@@ -21,11 +21,15 @@ import { Plugin } from "prosemirror-state";
 const PasteCleanupExtension = createExtension({
 	name: "pasteCleanup",
 	priority: Priority.Highest,
-	onSchemaReady() {
+	onSchemaReady(_context, options) {
 		return {
-			plugins: [new Plugin({ props: { transformPastedHTML(html) {
-				return cleanPastedHtml(html);
-			} } })],
+			plugins: [new Plugin({ props: {
+				transformPastedHTML(html) {
+					const internalClipboardToken = options.getInternalClipboardToken?.();
+					return cleanPastedHtml(html, { ...internalClipboardToken ? { internalClipboardToken } : {} });
+				},
+				transformPasted: removeUnpairedBookmarkBoundaries
+			} })],
 			keyboardShortcuts: { "Mod-Alt-v": pasteWithoutFormatting }
 		};
 	}

package/dist/prosemirror/extensions/features/pasteCleanup.d.ts

@@ -1,31 +1,18 @@
+import { Slice } from "prosemirror-model";
 //#region src/prosemirror/extensions/features/pasteCleanup.d.ts
-/**
- * Office / web paste cleanup
- *
- * Content pasted from word processors and web pages arrives wrapped in a large
- * amount of producer-specific cruft: conditional comments, XML processing
- * instructions, namespaced markup (`<o:p>`, `<w:sdt>`, smart tags), `mso-*`
- * style declarations, and empty spans. None of it maps to the editor schema,
- * and some of it confuses the browser HTML parser that ProseMirror's clipboard
- * pipeline relies on. Producer class names (`MsoNormal`, ...) are deliberately
- * kept so the downstream style inliner can match `<style>` rules against them.
- *
- * `cleanPastedHtml` normalizes the raw clipboard HTML string into something the
- * schema's `parseDOM` rules can read cleanly. It is a pure string transform so
- * it runs in the editor and in tests without a DOM, and it is deliberately
- * conservative: it strips producer metadata but never rewrites visible text
- * (curly quotes, non-Latin scripts, and whitespace between words are left
- * untouched, since this editor targets an international, typography-sensitive
- * audience).
- */
 /**
  * Strip Office/web producer cruft from a raw clipboard HTML string.
  *
- * Best-effort and non-throwing: any unexpected failure returns the original
- * markup so a paste degrades to the browser default rather than losing content.
+ * Best-effort and non-throwing: any unexpected failure returns empty markup so
+ * reconstruction capabilities and other untrusted attributes fail closed.
  * `<style>` blocks are intentionally left in place for the downstream style
  * inliner, which resolves class-based CSS before the schema parser runs.
  */
-declare function cleanPastedHtml(html: string): string;
+type CleanPastedHtmlOptions = {
+  internalClipboardToken?: string;
+};
+declare function cleanPastedHtml(html: string, options?: CleanPastedHtmlOptions): string;
+/** Remove incomplete or duplicate bookmark pairs at copied slice edges. */
+declare function removeUnpairedBookmarkBoundaries(slice: Slice): Slice;
 //#endregion
-export { cleanPastedHtml };
+export { cleanPastedHtml, removeUnpairedBookmarkBoundaries };

package/dist/prosemirror/extensions/features/pasteCleanup.js

@@ -1,4 +1,6 @@
 import { stripXmlDeclarations } from "../../../utils/stripXmlDeclarations.js";
+import { readBookmarkBoundaryAttrs } from "../../bookmarkBoundaryAttrs.js";
+import { Fragment, Slice } from "prosemirror-model";
 //#region src/prosemirror/extensions/features/pasteCleanup.ts
 /**
 * Office / web paste cleanup
@@ -117,56 +119,109 @@ const STRAY_XML_TAG = new RegExp(`<\\/?xml${TAG_TAIL}>`, "gi");
 const NOISE_TAG = new RegExp(`<\\/?(?:font|meta|link)${TAG_TAIL}>`, "gi");
 const EMPTY_SPAN = new RegExp(`<span(?:\\s${TAG_TAIL})?><\\/span>`, "gi");
 const MAX_EMPTY_SPAN_PASSES = 5;
-const PM_SLICE_MARKER = /\bdata-pm-slice\s*=/i;
+const INTERNAL_CLIPBOARD_ATTR_PATTERN = /\s+data-docx-internal-clipboard(?:="([^"]*)"|='([^']*)'|=([^\s>]+))?/gi;
 const TEXTBOX_ANCHOR_ATTR = /\s+data-docx-textbox-anchor(?:="[^"]*"|='[^']*')?/gi;
+const BOOKMARK_BOUNDARY_ATTR = /\s+data-docx-bookmark-(?:boundary|id|name|col-first|col-last)(?:="[^"]*"|='[^']*'|=[^\s>]+)?/gi;
+const STRUCTURED_FIELD_ATTR = /\s+data-field-structured(?:="[^"]*"|='[^']*'|=[^\s>]+)?/gi;
 /**
-* Remove the `data-docx-textbox-anchor` marker from HTML that did not come
-* from a ProseMirror clipboard slice (see {@link PM_SLICE_MARKER}). Internal
-* copy/paste of a real text box carries the marker HTML unmodified so it
-* keeps working; anything else has the marker stripped defensively.
+* Treat reconstruction attributes as internal only when the HTML carries the
+* unguessable capability created with this editor schema. `data-pm-slice`
+* cannot establish trust because arbitrary external HTML can forge it.
 */
-function stripForeignTextBoxAnchors(html, originalHtml) {
-	if (PM_SLICE_MARKER.test(originalHtml)) return html;
+function hasInternalClipboardCapability(html, token) {
+	if (!token) return false;
+	INTERNAL_CLIPBOARD_ATTR_PATTERN.lastIndex = 0;
+	for (const match of html.matchAll(INTERNAL_CLIPBOARD_ATTR_PATTERN)) if ((match[1] ?? match[2] ?? match[3]) === token) return true;
+	return false;
+}
+function stripForeignTextBoxAnchors(html, internal) {
+	if (internal) return html;
 	return html.replace(TEXTBOX_ANCHOR_ATTR, "");
 }
+function stripForeignBookmarkBoundaries(html, internal) {
+	if (internal) return html;
+	return html.replace(BOOKMARK_BOUNDARY_ATTR, "");
+}
+function stripForeignStructuredFields(html, internal) {
+	if (internal) return html;
+	return html.replace(STRUCTURED_FIELD_ATTR, "");
+}
 /**
 * Remove empty spans left behind after stripping `mso-*` styles. Only truly
 * empty spans are removed (whitespace-only spans are kept so word-separating
 * spacer runs never collapse two words together). Repeated a bounded number of
 * times to unwrap nested empties (`<span><span></span></span>`).
 */
-function stripEmptySpans(html) {
+function stripEmptySpans(html, preserveInternalAtoms) {
 	let current = html;
 	for (let pass = 0; pass < MAX_EMPTY_SPAN_PASSES; pass++) {
-		const next = current.replace(EMPTY_SPAN, "");
+		const next = current.replace(EMPTY_SPAN, (span) => {
+			if (preserveInternalAtoms && /\bdata-docx-(?:bookmark-boundary|textbox-anchor)\s*=/i.test(span)) return span;
+			return "";
+		});
 		if (next === current) break;
 		current = next;
 	}
 	return current;
 }
-/**
-* Strip Office/web producer cruft from a raw clipboard HTML string.
-*
-* Best-effort and non-throwing: any unexpected failure returns the original
-* markup so a paste degrades to the browser default rather than losing content.
-* `<style>` blocks are intentionally left in place for the downstream style
-* inliner, which resolves class-based CSS before the schema parser runs.
-*/
-function cleanPastedHtml(html) {
+function cleanPastedHtml(html, options = {}) {
 	if (!html) return html;
 	try {
+		const internal = hasInternalClipboardCapability(html, options.internalClipboardToken);
 		let cleaned = stripHtmlComments(html);
 		cleaned = stripXmlDeclarations(cleaned);
 		cleaned = cleaned.replace(NAMESPACED_TAG, "");
 		cleaned = cleaned.replace(STRAY_XML_TAG, "");
 		cleaned = cleaned.replace(NOISE_TAG, "");
 		cleaned = stripMsoStyles(cleaned);
-		cleaned = stripEmptySpans(cleaned);
-		cleaned = stripForeignTextBoxAnchors(cleaned, html);
+		cleaned = stripEmptySpans(cleaned, internal);
+		cleaned = stripForeignTextBoxAnchors(cleaned, internal);
+		cleaned = stripForeignBookmarkBoundaries(cleaned, internal);
+		cleaned = stripForeignStructuredFields(cleaned, internal);
+		cleaned = cleaned.replace(INTERNAL_CLIPBOARD_ATTR_PATTERN, "");
 		return cleaned.trim();
 	} catch {
-		return html;
+		return "";
 	}
 }
+/** Remove incomplete or duplicate bookmark pairs at copied slice edges. */
+function removeUnpairedBookmarkBoundaries(slice) {
+	const counts = /* @__PURE__ */ new Map();
+	let boundaryIndex = 0;
+	slice.content.descendants((node) => {
+		if (node.type.name !== "bookmarkBoundary") return true;
+		const result = readBookmarkBoundaryAttrs(node);
+		if (!result.ok) return false;
+		const attrs = result.value;
+		const count = counts.get(attrs.id) ?? {
+			starts: 0,
+			ends: 0
+		};
+		if (attrs.type === "start") {
+			count.starts += 1;
+			count.firstStart ??= boundaryIndex;
+		} else {
+			count.ends += 1;
+			count.firstEnd ??= boundaryIndex;
+		}
+		boundaryIndex += 1;
+		counts.set(attrs.id, count);
+		return false;
+	});
+	const pairedIds = new Set([...counts].flatMap(([id, count]) => count.starts === 1 && count.ends === 1 && count.firstStart !== void 0 && count.firstEnd !== void 0 && count.firstStart < count.firstEnd ? [id] : []));
+	const filterFragment = (fragment) => {
+		const children = [];
+		fragment.forEach((node) => {
+			if (node.type.name === "bookmarkBoundary") {
+				const result = readBookmarkBoundaryAttrs(node);
+				if (result.ok && pairedIds.has(result.value.id)) children.push(node);
+				return;
+			}
+			children.push(node.childCount === 0 ? node : node.copy(filterFragment(node.content)));
+		});
+		return Fragment.fromArray(children);
+	};
+	return new Slice(filterFragment(slice.content), slice.openStart, slice.openEnd);
+}
 //#endregion
-export { cleanPastedHtml };
+export { cleanPastedHtml, removeUnpairedBookmarkBoundaries };

package/dist/prosemirror/extensions/nodes/BookmarkBoundaryExtension.d.ts

@@ -0,0 +1,9 @@
+import { NodeExtension } from "../types.js";
+//#region src/prosemirror/extensions/nodes/BookmarkBoundaryExtension.d.ts
+/** Zero-width bookmark boundary that preserves its position through ProseMirror edits. */
+type BookmarkBoundaryOptions = {
+  getInternalClipboardToken?: () => string;
+};
+declare const BookmarkBoundaryExtension: (options?: Partial<BookmarkBoundaryOptions> | undefined) => NodeExtension;
+//#endregion
+export { BookmarkBoundaryExtension };