@stll/folio-core 0.27.1 → 0.28.1

package/dist/docx/server/extractDocxText.js

@@ -1,9 +1,10 @@
 import { escapeTableCell } from "../../markdown/escape.js";
 import { RELATIONSHIP_TYPES, parseRelationships } from "../relsParser.js";
 import { findAllDeep, findDeep, getAttribute, getAttributeByNamespaceUri, getLocalName, getNamespaceUri, getTextContent, parseXml } from "../xmlParser.js";
-import { loadDocxArchive } from "./boundedArchive.js";
+import { DocxArchiveError, loadDocxArchive } from "./boundedArchive.js";
 //#region src/docx/server/extractDocxText.ts
 const DOCUMENT_RELS_PATH = "word/_rels/document.xml.rels";
+const ARCHIVE_LIMIT_REASON = "total-too-large";
 const WORDPROCESSINGML_NAMESPACES = /* @__PURE__ */ new Set(["http://schemas.openxmlformats.org/wordprocessingml/2006/main", "http://purl.oclc.org/ooxml/wordprocessingml/main"]);
 const childElements = (element) => element.elements?.filter((child) => child.type === "element") ?? [];
 const wordElementName = (element) => WORDPROCESSINGML_NAMESPACES.has(getNamespaceUri(element) ?? "") ? getLocalName(element.name) : null;
@@ -105,6 +106,23 @@ const MAX_TABLE_COLUMNS = 256;
 const MAX_NESTED_TABLE_DEPTH = 8;
 /** Rows collected from one `w:tbl`. The column cap alone leaves row count unbounded. */
 const MAX_TABLE_ROWS = 8192;
+/** Source and rendered characters retained across every table in one extraction. */
+const MAX_TABLE_CHARACTERS = 8 * 1024 * 1024;
+const createTableExtractionBudget = () => ({
+	remainingSourceCharacters: MAX_TABLE_CHARACTERS,
+	remainingRenderedCharacters: MAX_TABLE_CHARACTERS
+});
+const chargeTableCharacters = ({ budget, characters, kind }) => {
+	if (characters > (kind === "source" ? budget.remainingSourceCharacters : budget.remainingRenderedCharacters)) throw new DocxArchiveError({
+		message: `Extracted DOCX table ${kind} text exceeded the ${MAX_TABLE_CHARACTERS}-character limit`,
+		reason: ARCHIVE_LIMIT_REASON
+	});
+	if (kind === "source") {
+		budget.remainingSourceCharacters -= characters;
+		return;
+	}
+	budget.remainingRenderedCharacters -= characters;
+};
 /**
 * Characters one extraction emits, shared by the body and every header/footer
 * part. Element count is bounded at unzip, but a bounded element count still
@@ -112,19 +130,16 @@ const MAX_TABLE_ROWS = 8192;
 * GFM scaffolding are counted, so the emitted side carries its own ceiling.
 */
 const MAX_EXTRACTED_CHARS = 8e6;
-/**
-* Collect a table's `w:tr`, or a row's `w:tc`, seeing through the wrappers Word
-* puts around them (`w:sdt` / `w:sdtContent` content controls, `w:customXml`).
-* The walk stops at `w:tbl` and `w:p` so a nested table's rows and cells never
-* leak into the grid of the table that contains them.
-*/
-const collectTableParts = (parent, localName, limit) => {
+const collectTableParts = ({ parent, localName, limit }) => {
 	const parts = [];
 	const walk = (node) => {
 		for (const child of childElements(node)) {
-			if (parts.length >= limit) return;
 			const childName = wordElementName(child);
 			if (childName === localName) {
+				if (parts.length >= limit) throw new DocxArchiveError({
+					message: `Extracted DOCX table exceeded the ${limit}-${localName} limit`,
+					reason: ARCHIVE_LIMIT_REASON
+				});
 				parts.push(child);
 				continue;
 			}
@@ -152,7 +167,15 @@ const readCellSourceParagraphs = (cell, depth) => {
 			}
 			if (childName === "tbl") {
 				if (depth >= MAX_NESTED_TABLE_DEPTH) continue;
-				for (const row of collectTableParts(child, "tr", MAX_TABLE_ROWS)) for (const nestedCell of collectTableParts(row, "tc", MAX_TABLE_COLUMNS)) for (const paragraph of readCellSourceParagraphs(nestedCell, depth + 1)) paragraphs.push(paragraph);
+				for (const row of collectTableParts({
+					parent: child,
+					localName: "tr",
+					limit: MAX_TABLE_ROWS
+				})) for (const nestedCell of collectTableParts({
+					parent: row,
+					localName: "tc",
+					limit: MAX_TABLE_COLUMNS
+				})) for (const paragraph of readCellSourceParagraphs(nestedCell, depth + 1)) paragraphs.push(paragraph);
 				continue;
 			}
 			walk(child);
@@ -161,18 +184,26 @@ const readCellSourceParagraphs = (cell, depth) => {
 	walk(cell);
 	return paragraphs;
 };
-const readCellRenderedLines = (cell, depth) => {
+const readCellRenderedLines = (cell, { depth, budget }) => {
 	const lines = [];
 	const walk = (node) => {
 		for (const child of childElements(node)) {
 			const childName = wordElementName(child);
 			if (childName === "p") {
 				const text = collectText(child);
+				chargeTableCharacters({
+					budget,
+					characters: text.length,
+					kind: "source"
+				});
 				if (text.length > 0) lines.push(text);
 				continue;
 			}
 			if (childName === "tbl") {
-				if (depth < MAX_NESTED_TABLE_DEPTH) for (const line of flattenNestedTable(child, depth + 1)) lines.push(line);
+				if (depth < MAX_NESTED_TABLE_DEPTH) for (const line of flattenNestedTable(child, {
+					depth: depth + 1,
+					budget
+				})) lines.push(line);
 				continue;
 			}
 			walk(child);
@@ -181,10 +212,21 @@ const readCellRenderedLines = (cell, depth) => {
 	walk(cell);
 	return lines;
 };
-const flattenNestedTable = (table, depth) => {
+const flattenNestedTable = (table, { depth, budget }) => {
 	const lines = [];
-	for (const row of collectTableParts(table, "tr", MAX_TABLE_ROWS)) {
-		const cells = collectTableParts(row, "tc", MAX_TABLE_COLUMNS).map((cell) => readCellRenderedLines(cell, depth).join("\n"));
+	for (const row of collectTableParts({
+		parent: table,
+		localName: "tr",
+		limit: MAX_TABLE_ROWS
+	})) {
+		const cells = collectTableParts({
+			parent: row,
+			localName: "tc",
+			limit: MAX_TABLE_COLUMNS
+		}).map((cell) => readCellRenderedLines(cell, {
+			depth,
+			budget
+		}).join("\n"));
 		if (cells.some((text) => text.length > 0)) lines.push(cells.join(NESTED_TABLE_CELL_SEPARATOR));
 	}
 	return lines;
@@ -194,7 +236,7 @@ const emptyTableCell = () => ({
 	paragraphs: [],
 	gridSpan: 1
 });
-const readTableCell = (cell, depth) => {
+const readTableCell = (cell, { depth, budget }) => {
 	const properties = findWordChild(cell, "tcPr");
 	const gridSpanValue = getWordAttribute(findWordChild(properties, "gridSpan"), "val");
 	const parsedGridSpan = gridSpanValue === null ? 1 : Number.parseInt(gridSpanValue, 10);
@@ -205,10 +247,12 @@ const readTableCell = (cell, depth) => {
 		paragraphs: [],
 		gridSpan
 	};
-	const paragraphs = readCellSourceParagraphs(cell, depth);
 	return {
-		text: readCellRenderedLines(cell, depth).join("\n"),
-		paragraphs,
+		text: readCellRenderedLines(cell, {
+			depth,
+			budget
+		}).join("\n"),
+		paragraphs: readCellSourceParagraphs(cell, depth),
 		gridSpan
 	};
 };
@@ -235,18 +279,29 @@ const readRowGridOffset = (row, localName) => {
 	const parsed = value === null ? 0 : Number.parseInt(value, 10);
 	return Number.isFinite(parsed) && parsed > 0 ? Math.min(parsed, MAX_TABLE_COLUMNS) : 0;
 };
-const readTableGrid = (table) => {
+const readTableGrid = (table, budget) => {
 	const rows = [];
 	let columnCount = 0;
 	let firstRowIsHeader = false;
-	for (const [rowIndex, row] of collectTableParts(table, "tr", MAX_TABLE_ROWS).entries()) {
+	for (const [rowIndex, row] of collectTableParts({
+		parent: table,
+		localName: "tr",
+		limit: MAX_TABLE_ROWS
+	}).entries()) {
 		if (rowIndex === 0) firstRowIsHeader = declaresHeaderRow(row);
 		const columns = [];
 		const gridBefore = readRowGridOffset(row, "gridBefore");
 		for (let index = 0; index < gridBefore; index += 1) columns.push(emptyTableCell());
-		for (const cell of collectTableParts(row, "tc", MAX_TABLE_COLUMNS)) {
+		for (const cell of collectTableParts({
+			parent: row,
+			localName: "tc",
+			limit: MAX_TABLE_COLUMNS
+		})) {
 			if (columns.length >= MAX_TABLE_COLUMNS) break;
-			const extractedCell = readTableCell(cell, 0);
+			const extractedCell = readTableCell(cell, {
+				depth: 0,
+				budget
+			});
 			columns.push(extractedCell);
 			const padding = Math.min(extractedCell.gridSpan - 1, MAX_TABLE_COLUMNS - columns.length);
 			for (let index = 0; index < padding; index++) columns.push(emptyTableCell());
@@ -269,12 +324,17 @@ const toRowLine = (columns, columnCount) => {
 	return `| ${cells.join(" | ")} |`;
 };
 /** Render a `w:tbl` as GFM rows. A table with no cell at all renders nothing. */
-const renderTableRows = (table, tableIndex) => {
-	const { rows, columnCount, firstRowIsHeader } = readTableGrid(table);
+const renderTableRows = ({ table, tableIndex, budget }) => {
+	const { rows, columnCount, firstRowIsHeader } = readTableGrid(table, budget);
 	const [firstRow, ...remainingRows] = rows;
 	if (columnCount === 0 || firstRow === void 0) return [];
 	const rendered = [];
 	const pushScaffolding = (text, kind) => {
+		chargeTableCharacters({
+			budget,
+			characters: text.length,
+			kind: "rendered"
+		});
 		rendered.push({
 			text,
 			position: {
@@ -284,8 +344,14 @@ const renderTableRows = (table, tableIndex) => {
 		});
 	};
 	const pushCells = (cells) => {
+		const text = toRowLine(cells, columnCount);
+		chargeTableCharacters({
+			budget,
+			characters: text.length,
+			kind: "rendered"
+		});
 		rendered.push({
-			text: toRowLine(cells, columnCount),
+			text,
 			position: {
 				table: tableIndex,
 				kind: "cells",
@@ -304,7 +370,7 @@ const renderTableRows = (table, tableIndex) => {
 	return rendered;
 };
 const createCharBudget = () => ({ remaining: MAX_EXTRACTED_CHARS });
-const extractContainer = ({ container, source, startIndex, startTableIndex, budget }) => {
+const extractContainer = ({ container, source, startIndex, startTableIndex, budget, tableBudget }) => {
 	const paragraphs = [];
 	let charCount = 0;
 	let tableCount = 0;
@@ -341,7 +407,11 @@ const extractContainer = ({ container, source, startIndex, startTableIndex, budg
 			if (budget.remaining <= 0) return;
 			const childName = wordElementName(child);
 			if (childName === "tbl") {
-				for (const row of renderTableRows(child, startTableIndex + tableCount)) pushTableRow(row);
+				for (const row of renderTableRows({
+					table: child,
+					tableIndex: startTableIndex + tableCount,
+					budget: tableBudget
+				})) pushTableRow(row);
 				tableCount += 1;
 				continue;
 			}
@@ -356,7 +426,7 @@ const extractContainer = ({ container, source, startIndex, startTableIndex, budg
 		tableCount
 	};
 };
-const extractParts = async ({ archive, source, rootName, startIndex, startTableIndex, paths, budget }) => {
+const extractParts = async ({ archive, source, rootName, startIndex, startTableIndex, paths, budget, tableBudget }) => {
 	const paragraphs = [];
 	let charCount = 0;
 	let tableCount = 0;
@@ -371,7 +441,8 @@ const extractParts = async ({ archive, source, rootName, startIndex, startTableI
 			source,
 			startIndex: nextIndex,
 			startTableIndex: startTableIndex + tableCount,
-			budget
+			budget,
+			tableBudget
 		});
 		for (const paragraph of result.paragraphs) paragraphs.push(paragraph);
 		charCount += result.charCount;
@@ -443,6 +514,7 @@ const extractDocxText = async (bytes) => {
 	if (!body) return createEmptyResult();
 	const referencedParts = await resolveReferencedHeaderFooterParts(archive, root);
 	const budget = createCharBudget();
+	const tableBudget = createTableExtractionBudget();
 	const headers = await extractParts({
 		archive,
 		source: "header",
@@ -450,14 +522,16 @@ const extractDocxText = async (bytes) => {
 		startIndex: 0,
 		startTableIndex: 0,
 		paths: referencedParts.headers,
-		budget
+		budget,
+		tableBudget
 	});
 	const bodyResult = extractContainer({
 		container: body,
 		source: "body",
 		startIndex: headers.paragraphs.length,
 		startTableIndex: headers.tableCount,
-		budget
+		budget,
+		tableBudget
 	});
 	const footers = await extractParts({
 		archive,
@@ -466,7 +540,8 @@ const extractDocxText = async (bytes) => {
 		startIndex: headers.paragraphs.length + bodyResult.paragraphs.length,
 		startTableIndex: headers.tableCount + bodyResult.tableCount,
 		paths: referencedParts.footers,
-		budget
+		budget,
+		tableBudget
 	});
 	return {
 		paragraphs: [

package/dist/docx/server/materializeYjsDocx.d.ts

@@ -0,0 +1,31 @@
+//#region src/docx/server/materializeYjsDocx.d.ts
+/** Yjs fragment that stores Folio's canonical ProseMirror document. */
+declare const FOLIO_YJS_PROSEMIRROR_FRAGMENT_NAME = "prosemirror";
+/** Maximum accepted size of a complete Yjs collaboration state update. */
+declare const FOLIO_YJS_UPDATE_MAX_BYTES: number;
+/** Stable failure codes returned by server-side Yjs-to-DOCX materialization. */
+declare const FOLIO_YJS_DOCX_MATERIALIZATION_ERROR_CODES: readonly ["empty_update", "invalid_update", "missing_document", "update_too_large"];
+/** Failure code for a rejected Yjs-to-DOCX materialization request. */
+type FolioYjsDocxMaterializationErrorCode = (typeof FOLIO_YJS_DOCX_MATERIALIZATION_ERROR_CODES)[number];
+declare const FolioYjsDocxMaterializationError_base: import("better-result").TaggedErrorClass<"FolioYjsDocxMaterializationError">;
+/** Typed failure raised when a collaboration snapshot cannot be materialized. */
+declare class FolioYjsDocxMaterializationError extends FolioYjsDocxMaterializationError_base<{
+  code: FolioYjsDocxMaterializationErrorCode;
+  message: string;
+  cause?: unknown;
+}> {}
+/** Inputs for materializing a complete Yjs state update into a DOCX package. */
+type MaterializeYjsDocxOptions = {
+  /** Original DOCX whose package parts and non-body stories must be preserved. */
+  sourceDocx: ArrayBuffer | Uint8Array;
+  /** Complete Yjs state update containing Folio's ProseMirror fragment. */
+  yjsUpdate: Uint8Array;
+};
+/**
+ * Materialize a complete Folio Yjs state update into a DOCX while preserving
+ * package parts from the source document. This is the server-side equivalent
+ * of the browser editor's full save path for the main document story.
+ */
+declare const materializeYjsDocx: ({ sourceDocx, yjsUpdate }: MaterializeYjsDocxOptions) => Promise<ArrayBuffer>;
+//#endregion
+export { FOLIO_YJS_DOCX_MATERIALIZATION_ERROR_CODES, FOLIO_YJS_PROSEMIRROR_FRAGMENT_NAME, FOLIO_YJS_UPDATE_MAX_BYTES, FolioYjsDocxMaterializationError, FolioYjsDocxMaterializationErrorCode, MaterializeYjsDocxOptions, materializeYjsDocx };

package/dist/docx/server/materializeYjsDocx.js

@@ -0,0 +1,61 @@
+import { fromProseDoc } from "../../prosemirror/conversion/fromProseDoc.js";
+import { schema } from "../../prosemirror/schema/index.js";
+import { parseDocx } from "../parser.js";
+import { repackDocx } from "../rezip.js";
+import { Result, TaggedError } from "better-result";
+import { initProseMirrorDoc } from "y-prosemirror";
+import * as Y from "yjs";
+//#region src/docx/server/materializeYjsDocx.ts
+/** Yjs fragment that stores Folio's canonical ProseMirror document. */
+const FOLIO_YJS_PROSEMIRROR_FRAGMENT_NAME = "prosemirror";
+/** Maximum accepted size of a complete Yjs collaboration state update. */
+const FOLIO_YJS_UPDATE_MAX_BYTES = 10 * 1024 * 1024;
+/** Stable failure codes returned by server-side Yjs-to-DOCX materialization. */
+const FOLIO_YJS_DOCX_MATERIALIZATION_ERROR_CODES = [
+	"empty_update",
+	"invalid_update",
+	"missing_document",
+	"update_too_large"
+];
+/** Typed failure raised when a collaboration snapshot cannot be materialized. */
+var FolioYjsDocxMaterializationError = class extends TaggedError("FolioYjsDocxMaterializationError") {};
+const readProseMirrorDocument = (yjsUpdate) => {
+	if (yjsUpdate.byteLength === 0) throw new FolioYjsDocxMaterializationError({
+		code: "empty_update",
+		message: "Cannot materialize DOCX from an empty Yjs update."
+	});
+	if (yjsUpdate.byteLength > 10485760) throw new FolioYjsDocxMaterializationError({
+		code: "update_too_large",
+		message: "Yjs update exceeds the DOCX materialization limit."
+	});
+	const ydoc = new Y.Doc();
+	const parsed = Result.try({
+		try: () => {
+			Y.applyUpdate(ydoc, yjsUpdate);
+			const fragment = ydoc.getXmlFragment(FOLIO_YJS_PROSEMIRROR_FRAGMENT_NAME);
+			if (fragment.length === 0) throw new FolioYjsDocxMaterializationError({
+				code: "missing_document",
+				message: "Yjs update does not contain a Folio document."
+			});
+			return initProseMirrorDoc(fragment, schema).doc;
+		},
+		catch: (cause) => cause instanceof FolioYjsDocxMaterializationError ? cause : new FolioYjsDocxMaterializationError({
+			code: "invalid_update",
+			message: "Yjs update is not a valid Folio collaboration snapshot.",
+			cause
+		})
+	});
+	ydoc.destroy();
+	if (parsed.isOk()) return parsed.value;
+	throw parsed.error;
+};
+/**
+* Materialize a complete Folio Yjs state update into a DOCX while preserving
+* package parts from the source document. This is the server-side equivalent
+* of the browser editor's full save path for the main document story.
+*/
+const materializeYjsDocx = async ({ sourceDocx, yjsUpdate }) => {
+	return await repackDocx(fromProseDoc(readProseMirrorDocument(yjsUpdate), await parseDocx(sourceDocx, { preloadFonts: false })));
+};
+//#endregion
+export { FOLIO_YJS_DOCX_MATERIALIZATION_ERROR_CODES, FOLIO_YJS_PROSEMIRROR_FRAGMENT_NAME, FOLIO_YJS_UPDATE_MAX_BYTES, FolioYjsDocxMaterializationError, materializeYjsDocx };

package/dist/docx/styleParser.js

@@ -1,23 +1,24 @@
 import { isValidHexColor } from "../utils/colorResolver.js";
 import { mergeParagraphFormatting } from "../utils/paragraphFormattingMerge.js";
-import { mergeTextFormatting } from "../utils/textFormattingMerge.js";
+import { mergeStyleTextFormatting } from "../utils/textFormattingMerge.js";
 import { BorderStyleSchema, ConditionalStyleTypeSchema, EmphasisMarkSchema, FontHintSchema, FontThemeSchema, HighlightColorSchema, LineSpacingRuleSchema, ParagraphAlignmentSchema, ShadingPatternSchema, StyleTypeSchema, TabLeaderSchema, TabStopAlignmentSchema, TableCellTextDirectionSchema, TableRowHeightRuleSchema, TableWidthTypeSchema, TextEffectSchema, ThemeColorSlotSchema, UnderlineStyleSchema, narrowEnum } from "./parserEnums.js";
 import { resolveThemeFontRef } from "./themeParser.js";
 import { findChild, findChildren, getAttribute, getLocalName, parseBooleanElement, parseNumberingLevelAttribute, parseNumericAttribute, parseOnOffValue, parseTableMeasurementValue, parseXmlDocument } from "./xmlParser.js";
 //#region src/docx/styleParser.ts
+const findLastRunToggle = (rPr, localName) => findChildren(rPr, "w", localName).at(-1) ?? null;
 /**
 * Parse text formatting properties (w:rPr)
 */
 function parseRunProperties(rPr, theme) {
 	if (!rPr) return;
 	const formatting = {};
-	const b = findChild(rPr, "w", "b");
+	const b = findLastRunToggle(rPr, "b");
 	if (b) formatting.bold = parseBooleanElement(b);
-	const bCs = findChild(rPr, "w", "bCs");
+	const bCs = findLastRunToggle(rPr, "bCs");
 	if (bCs) formatting.boldCs = parseBooleanElement(bCs);
-	const i = findChild(rPr, "w", "i");
+	const i = findLastRunToggle(rPr, "i");
 	if (i) formatting.italic = parseBooleanElement(i);
-	const iCs = findChild(rPr, "w", "iCs");
+	const iCs = findLastRunToggle(rPr, "iCs");
 	if (iCs) formatting.italicCs = parseBooleanElement(iCs);
 	const u = findChild(rPr, "w", "u");
 	if (u) {
@@ -29,7 +30,7 @@ function parseRunProperties(rPr, theme) {
 			if (colorVal || themeColor) formatting.underline.color = parseColorValue(colorVal, themeColor, getAttribute(u, "w", "themeTint"), getAttribute(u, "w", "themeShade"));
 		}
 	}
-	const strike = findChild(rPr, "w", "strike");
+	const strike = findLastRunToggle(rPr, "strike");
 	if (strike) formatting.strike = parseBooleanElement(strike);
 	const dstrike = findChild(rPr, "w", "dstrike");
 	if (dstrike) formatting.doubleStrike = parseBooleanElement(dstrike);
@@ -38,11 +39,11 @@ function parseRunProperties(rPr, theme) {
 		const val = getAttribute(vertAlign, "w", "val");
 		if (val === "superscript" || val === "subscript" || val === "baseline") formatting.vertAlign = val;
 	}
-	const smallCaps = findChild(rPr, "w", "smallCaps");
+	const smallCaps = findLastRunToggle(rPr, "smallCaps");
 	if (smallCaps) formatting.smallCaps = parseBooleanElement(smallCaps);
-	const caps = findChild(rPr, "w", "caps");
+	const caps = findLastRunToggle(rPr, "caps");
 	if (caps) formatting.allCaps = parseBooleanElement(caps);
-	const vanish = findChild(rPr, "w", "vanish");
+	const vanish = findLastRunToggle(rPr, "vanish");
 	if (vanish) formatting.hidden = parseBooleanElement(vanish);
 	const color = findChild(rPr, "w", "color");
 	if (color) formatting.color = parseColorValue(getAttribute(color, "w", "val"), getAttribute(color, "w", "themeColor"), getAttribute(color, "w", "themeTint"), getAttribute(color, "w", "themeShade"));
@@ -154,13 +155,13 @@ function parseRunProperties(rPr, theme) {
 		const val = narrowEnum(getAttribute(em, "w", "val"), EmphasisMarkSchema);
 		if (val) formatting.emphasisMark = val;
 	}
-	const emboss = findChild(rPr, "w", "emboss");
+	const emboss = findLastRunToggle(rPr, "emboss");
 	if (emboss) formatting.emboss = parseBooleanElement(emboss);
-	const imprint = findChild(rPr, "w", "imprint");
+	const imprint = findLastRunToggle(rPr, "imprint");
 	if (imprint) formatting.imprint = parseBooleanElement(imprint);
-	const outline = findChild(rPr, "w", "outline");
+	const outline = findLastRunToggle(rPr, "outline");
 	if (outline) formatting.outline = parseBooleanElement(outline);
-	const shadow = findChild(rPr, "w", "shadow");
+	const shadow = findLastRunToggle(rPr, "shadow");
 	if (shadow) formatting.shadow = parseBooleanElement(shadow);
 	const rtl = findChild(rPr, "w", "rtl");
 	if (rtl) formatting.rtl = parseBooleanElement(rtl);
@@ -807,7 +808,7 @@ function resolveStyleInheritance(style, styleMap, visited = /* @__PURE__ */ new
 	const resolved = { ...style };
 	const mergedPPr = mergeParagraphFormatting(resolvedParent.pPr, style.pPr);
 	if (mergedPPr) resolved.pPr = mergedPPr;
-	const mergedRPr = mergeTextFormatting(resolvedParent.rPr, style.rPr);
+	const mergedRPr = mergeStyleTextFormatting(resolvedParent.rPr, style.rPr);
 	if (mergedRPr) resolved.rPr = mergedRPr;
 	if (style.type === "table") {
 		if (resolvedParent.tblPr || style.tblPr) resolved.tblPr = {

package/dist/layout-bridge/convert/toFlowBlocks.js

@@ -6,10 +6,11 @@ import { setParagraphFrame } from "../../layout-engine/paragraphFrame.js";
 import { setTextBoxGroupId } from "../../layout-engine/textBoxGroup.js";
 import { DEFAULT_TEXTBOX_MARGINS } from "../../layout-engine/types.js";
 import { getPageNumbering } from "../../paged-layout/sectionGeometry.js";
-import { expectBlockSdtAttrs, expectCharacterSpacingMarkAttrs, expectCommentMarkAttrs, expectEmphasisMarkAttrs, expectFieldAttrs, expectFontFamilyMarkAttrs, expectFontSizeMarkAttrs, expectFootnoteRefMarkAttrs, expectHardBreakAttrs, expectHighlightMarkAttrs, expectHyperlinkMarkAttrs, expectImageAttrs, expectLanguageMarkAttrs, expectMathAttrs, expectParagraphAttrs, expectRunFormattingOverrideMarkAttrs, expectRunShadingMarkAttrs, expectSymbolAttrs, expectTableAttrs, expectTableCellAttrs, expectTableRowAttrs, expectTextBoxAttrs, expectTextColorMarkAttrs, expectTextEffectMarkAttrs, expectTrackedChangeMarkAttrs, expectUnderlineMarkAttrs } from "../../prosemirror/attrs/index.js";
+import { expectBlockSdtAttrs, expectCharacterSpacingMarkAttrs, expectCharacterStyleMarkAttrs, expectCommentMarkAttrs, expectEmphasisMarkAttrs, expectFieldAttrs, expectFontFamilyMarkAttrs, expectFontSizeMarkAttrs, expectFootnoteRefMarkAttrs, expectHardBreakAttrs, expectHighlightMarkAttrs, expectHyperlinkMarkAttrs, expectImageAttrs, expectLanguageMarkAttrs, expectMathAttrs, expectParagraphAttrs, expectRunFormattingOverrideMarkAttrs, expectRunShadingMarkAttrs, expectSymbolAttrs, expectTableAttrs, expectTableCellAttrs, expectTableRowAttrs, expectTextBoxAttrs, expectTextColorMarkAttrs, expectTextEffectMarkAttrs, expectTrackedChangeMarkAttrs, expectUnderlineMarkAttrs } from "../../prosemirror/attrs/index.js";
 import { autospacingMatchesBase } from "../../prosemirror/autospacingBase.js";
 import { runShadingAttrsToShading } from "../../prosemirror/conversion/runShadingMark.js";
 import { directionToBidi } from "../../prosemirror/paragraphDirection.js";
+import { cascadeStyleTextFormatting } from "../../prosemirror/styles/styleToggleCascade.js";
 import { assertValidProseMirrorDocument } from "../../prosemirror/validation.js";
 import { resolveColor, resolveHighlightToCss } from "../../utils/colorResolver.js";
 import { resolveThemeFont } from "../../utils/fontResolver.js";
@@ -348,17 +349,49 @@ function mergeRunFormatting(paraDefaults, formatting) {
 	if (merged.letterSpacing === 0) delete merged.letterSpacing;
 	return merged;
 }
+/** Restore character-style toggle values that plain visual marks cannot represent. */
+function applyCharacterStyleToggleFormatting({ formatting, marks, paraDefaults }) {
+	const characterStyleMark = marks.find((mark) => mark.type.name === "characterStyle");
+	if (!characterStyleMark) return;
+	const styleRPr = expectCharacterStyleMarkAttrs(characterStyleMark)._styleRPr;
+	if (!styleRPr) return;
+	const effectiveStyleFormatting = cascadeStyleTextFormatting([{
+		formatting: {
+			bold: paraDefaults.bold ?? false,
+			boldCs: paraDefaults.complexScriptBold ?? false,
+			italic: paraDefaults.italic ?? false,
+			italicCs: paraDefaults.complexScriptItalic ?? false
+		},
+		type: "direct"
+	}, {
+		formatting: styleRPr,
+		type: "style"
+	}]).formatting;
+	if (styleRPr.bold !== void 0 && formatting.bold === void 0) formatting.bold = effectiveStyleFormatting?.bold ?? false;
+	if (styleRPr.boldCs !== void 0 && formatting.complexScriptBold === void 0) formatting.complexScriptBold = effectiveStyleFormatting?.boldCs ?? false;
+	if (styleRPr.italic !== void 0 && formatting.italic === void 0) formatting.italic = effectiveStyleFormatting?.italic ?? false;
+	if (styleRPr.italicCs !== void 0 && formatting.complexScriptItalic === void 0) formatting.complexScriptItalic = effectiveStyleFormatting?.italicCs ?? false;
+	if (styleRPr.allCaps === true && formatting.allCaps === void 0) formatting.allCaps = false;
+	if (styleRPr.emboss === true && formatting.emboss === void 0) formatting.emboss = false;
+	if (styleRPr.imprint === true && formatting.imprint === void 0) formatting.imprint = false;
+	if (styleRPr.outline === true && formatting.textOutline === void 0) formatting.textOutline = false;
+	if (styleRPr.shadow === true && formatting.textShadow === void 0) formatting.textShadow = false;
+	if (styleRPr.smallCaps === true && formatting.smallCaps === void 0) formatting.smallCaps = false;
+	if (styleRPr.strike === true && formatting.strike === void 0) formatting.strike = false;
+	if (styleRPr.hidden === true && formatting.hidden === void 0) formatting.hidden = false;
+}
 function applyRunFormattingOverrides(formatting, attrs) {
-	if (attrs.bold === false) formatting.bold = false;
-	if (attrs.italic === false) formatting.italic = false;
+	if (attrs.bold !== void 0) formatting.bold = attrs.bold;
+	if (attrs.italic !== void 0) formatting.italic = attrs.italic;
 	if (attrs.underline === "none") formatting.underline = false;
-	if (attrs.strike === false) formatting.strike = false;
-	if (attrs.allCaps === false) formatting.allCaps = false;
-	if (attrs.smallCaps === false) formatting.smallCaps = false;
-	if (attrs.emboss === false) formatting.emboss = false;
-	if (attrs.imprint === false) formatting.imprint = false;
-	if (attrs.shadow === false) formatting.textShadow = false;
-	if (attrs.outline === false) formatting.textOutline = false;
+	if (attrs.strike !== void 0) formatting.strike = attrs.strike;
+	if (attrs.allCaps !== void 0) formatting.allCaps = attrs.allCaps;
+	if (attrs.smallCaps !== void 0) formatting.smallCaps = attrs.smallCaps;
+	if (attrs.hidden !== void 0) formatting.hidden = attrs.hidden;
+	if (attrs.emboss !== void 0) formatting.emboss = attrs.emboss;
+	if (attrs.imprint !== void 0) formatting.imprint = attrs.imprint;
+	if (attrs.shadow !== void 0) formatting.textShadow = attrs.shadow;
+	if (attrs.outline !== void 0) formatting.textOutline = attrs.outline;
 	if (attrs.rtl === false) formatting.rtl = false;
 	if (attrs.boldCs !== void 0) formatting.complexScriptBold = attrs.boldCs;
 	if (attrs.italicCs !== void 0) formatting.complexScriptItalic = attrs.italicCs;
@@ -497,6 +530,11 @@ function paragraphToRuns(node, startPos, _options) {
 		if (child.type.name !== "sdt") leadingRenderedPageBreakPending = false;
 		if (child.isText && child.text) {
 			const formatting = extractRunFormatting(child.marks, theme);
+			applyCharacterStyleToggleFormatting({
+				formatting,
+				marks: child.marks,
+				paraDefaults
+			});
 			if (inTocParagraph) stripTocHyperlinkStyle(formatting);
 			const run = {
 				kind: "text",
@@ -513,6 +551,11 @@ function paragraphToRuns(node, startPos, _options) {
 			const text = decodeOoxmlSymbolCharacter(attrs.char);
 			if (text === null) return;
 			const formatting = extractRunFormatting(child.marks, theme);
+			applyCharacterStyleToggleFormatting({
+				formatting,
+				marks: child.marks,
+				paraDefaults
+			});
 			if (inTocParagraph) stripTocHyperlinkStyle(formatting);
 			runs.push({
 				kind: "text",
@@ -534,6 +577,11 @@ function paragraphToRuns(node, startPos, _options) {
 		}
 		if (child.type.name === "tab") {
 			const formatting = extractRunFormatting(child.marks, theme);
+			applyCharacterStyleToggleFormatting({
+				formatting,
+				marks: child.marks,
+				paraDefaults
+			});
 			const run = {
 				kind: "tab",
 				...mergeRunFormatting(paraDefaults, formatting),
@@ -561,6 +609,11 @@ function paragraphToRuns(node, startPos, _options) {
 			else if (ft === "DATE") mappedType = "DATE";
 			else if (ft === "TIME") mappedType = "TIME";
 			const extractedFieldFormatting = extractRunFormatting(child.marks, theme);
+			applyCharacterStyleToggleFormatting({
+				formatting: extractedFieldFormatting,
+				marks: child.marks,
+				paraDefaults
+			});
 			if (inTocParagraph) stripTocHyperlinkStyle(extractedFieldFormatting);
 			const fieldFormatting = markDefaultBlackTextColorSource(extractedFieldFormatting, paraDefaults);
 			const run = {
@@ -902,27 +955,25 @@ function convertParagraph(node, startPos, options) {
 	return block;
 }
 /**
-* Word keeps terminal empty body paragraphs after a final table as editable
-* anchors, but they do not create a page of their own. Preserve every block
-* and PM range while collapsing only the contiguous, run-free suffix; empty
-* paragraphs elsewhere still retain their normal line height.
+* Word keeps a final empty body paragraph after a table as an editable anchor,
+* but that final anchor does not create a page of its own. Earlier authored
+* empty paragraphs retain their height and may carry the document onto a blank
+* page. Preserve every block and PM range while collapsing only the final one.
 */
 function isPaintlessTerminalParagraph(block) {
 	if (block?.kind !== "paragraph" || block.runs.length !== 0) return false;
 	const attrs = block.attrs;
 	return !(attrs?.listMarker !== void 0 && !attrs.listMarkerHidden || attrs?.borders?.top || attrs?.borders?.bottom || attrs?.borders?.left || attrs?.borders?.right || attrs?.borders?.between || attrs?.borders?.bar || attrs?.shading || attrs?.spacingExplicit?.before || attrs?.spacingExplicit?.after || attrs?.pageBreakBefore || attrs?.renderedPageBreakBefore);
 }
-function suppressTerminalEmptyParagraphsAfterTable(blocks) {
+function suppressFinalEmptyParagraphAfterTable(blocks) {
 	let suffixStart = blocks.length;
 	while (suffixStart > 0 && isPaintlessTerminalParagraph(blocks[suffixStart - 1])) suffixStart -= 1;
 	if (suffixStart === blocks.length || suffixStart === 0 || blocks[suffixStart - 1]?.kind !== "table") return;
-	for (let index = suffixStart; index < blocks.length; index += 1) {
-		const block = blocks[index];
-		if (isPaintlessTerminalParagraph(block)) block.attrs = {
-			...block.attrs,
-			suppressEmptyParagraphHeight: true
-		};
-	}
+	const finalBlock = blocks.at(-1);
+	if (isPaintlessTerminalParagraph(finalBlock)) finalBlock.attrs = {
+		...finalBlock.attrs,
+		suppressEmptyParagraphHeight: true
+	};
 }
 function suppressFinalParagraphInRepeatedEmptySuffix(blocks) {
 	let suffixStart = blocks.length;
@@ -1477,7 +1528,7 @@ function toFlowBlocks(doc, options = {}) {
 		visit(node, offset + nodeOffset);
 	});
 	reserveLeadingEmptyOutlineHeight(blocks);
-	suppressTerminalEmptyParagraphsAfterTable(blocks);
+	suppressFinalEmptyParagraphAfterTable(blocks);
 	suppressFinalParagraphInRepeatedEmptySuffix(blocks);
 	return groupParagraphFrames(applySectionDocumentGrid(mergeRunInParagraphs(blocks), opts.finalSectionDocumentGridLinePitchTwips), nextBlockId);
 }

package/dist/layout-engine/footnoteColumnReflow.d.ts

@@ -0,0 +1,15 @@
+import { Layout } from "./types.js";
+//#region src/layout-engine/footnoteColumnReflow.d.ts
+type ReflowFootnoteColumnsOptions = {
+  initialLayout: Layout;
+  initialReserveFloors?: ReadonlyMap<number, number>;
+  runLayout: (reserveFloors: Map<number, number>) => Layout;
+};
+/**
+ * Re-run a multi-column layout until its body flow clears the shared footnote band.
+ * Reserve floors follow the current page-to-footnote assignment. Repeated floor states
+ * and persistent overlap use a stable fallback layout instead of throwing.
+ */
+declare function reflowFootnoteColumns({ initialLayout, initialReserveFloors, runLayout }: ReflowFootnoteColumnsOptions): Layout;
+//#endregion
+export { reflowFootnoteColumns };