@stll/folio-core 0.14.1 → 0.15.1

package/dist/docx/streamingXmlParser.js

@@ -0,0 +1,248 @@
+//#region src/docx/streamingXmlParser.ts
+const BUILT_IN_ENTITIES = {
+	amp: "&",
+	apos: "'",
+	gt: ">",
+	lt: "<",
+	quot: "\""
+};
+/**
+* Parse ordinary OOXML into Folio's existing element representation in one
+* pass. Unsupported or malformed constructs return a sentinel so callers can
+* retain the general-purpose parser as a compatibility fallback.
+*/
+const parseStreamingXml = (xml) => {
+	const root = { elements: [] };
+	const stack = [];
+	let cursor = 0;
+	let mergeAdjacentText = false;
+	while (cursor < xml.length) {
+		const open = xml.indexOf("<", cursor);
+		if (open === -1) {
+			if (!appendText(xml.slice(cursor), stack, mergeAdjacentText)) return { status: "unsupported" };
+			break;
+		}
+		if (open > cursor && !appendText(xml.slice(cursor, open), stack, mergeAdjacentText)) return { status: "unsupported" };
+		if (open > cursor) mergeAdjacentText = true;
+		if (xml.startsWith("<!--", open)) {
+			const close = xml.indexOf("-->", open + 4);
+			if (close === -1) return { status: "unsupported" };
+			cursor = close + 3;
+			continue;
+		}
+		if (xml.startsWith("<![CDATA[", open)) {
+			const close = xml.indexOf("]]>", open + 9);
+			if (close === -1 || !appendRawText(normalizeLineEndings(xml.slice(open + 9, close)), stack)) return { status: "unsupported" };
+			cursor = close + 3;
+			mergeAdjacentText = false;
+			continue;
+		}
+		if (xml.startsWith("<?", open)) {
+			const close = xml.indexOf("?>", open + 2);
+			if (close === -1) return { status: "unsupported" };
+			cursor = close + 2;
+			continue;
+		}
+		if (xml.startsWith("<!", open)) return { status: "unsupported" };
+		const close = findTagClose(xml, open + 1);
+		if (close === -1) return { status: "unsupported" };
+		if (xml.charCodeAt(open + 1) === 47) {
+			const name = xml.slice(open + 2, close).trim();
+			const frame = stack.pop();
+			if (!frame || frame.name !== name) return { status: "unsupported" };
+			cursor = close + 1;
+			mergeAdjacentText = false;
+			continue;
+		}
+		const parsedTag = parseOpenTag(xml, open + 1, close);
+		if (parsedTag.status === "unsupported") return parsedTag;
+		const parent = stack.at(-1)?.element ?? root;
+		appendElement(parent, parsedTag.element);
+		if (!parsedTag.selfClosing) {
+			if (stack.length >= 100) return { status: "unsupported" };
+			stack.push({
+				element: parsedTag.element,
+				name: parsedTag.name
+			});
+		}
+		cursor = close + 1;
+		mergeAdjacentText = false;
+	}
+	if (stack.length > 0) return { status: "unsupported" };
+	return {
+		status: "parsed",
+		value: root
+	};
+};
+const parseOpenTag = (xml, start, close) => {
+	let cursor = skipWhitespace(xml, start, close);
+	const nameStart = cursor;
+	cursor = scanName(xml, cursor, close);
+	if (cursor === nameStart) return { status: "unsupported" };
+	const name = xml.slice(nameStart, cursor);
+	if (isUnsafePropertyName(name)) return { status: "unsupported" };
+	let attributes;
+	let selfClosing = false;
+	while (cursor < close) {
+		cursor = skipWhitespace(xml, cursor, close);
+		if (cursor >= close) break;
+		if (xml.charCodeAt(cursor) === 47) {
+			selfClosing = true;
+			cursor = skipWhitespace(xml, cursor + 1, close);
+			if (cursor !== close) return { status: "unsupported" };
+			break;
+		}
+		const attributeNameStart = cursor;
+		cursor = scanName(xml, cursor, close);
+		if (cursor === attributeNameStart) return { status: "unsupported" };
+		const attributeName = xml.slice(attributeNameStart, cursor);
+		if (isUnsafePropertyName(attributeName)) return { status: "unsupported" };
+		cursor = skipWhitespace(xml, cursor, close);
+		if (xml.charCodeAt(cursor) !== 61) return { status: "unsupported" };
+		cursor = skipWhitespace(xml, cursor + 1, close);
+		const quote = xml.charCodeAt(cursor);
+		if (quote !== 34 && quote !== 39) return { status: "unsupported" };
+		const valueStart = cursor + 1;
+		cursor = xml.indexOf(quote === 34 ? "\"" : "'", valueStart);
+		if (cursor === -1 || cursor > close) return { status: "unsupported" };
+		const decoded = decodeXmlEntities(normalizeLineEndings(xml.slice(valueStart, cursor)));
+		if (decoded === null) return { status: "unsupported" };
+		attributes ??= {};
+		attributes[attributeName] = decoded;
+		cursor += 1;
+	}
+	const element = {
+		type: "element",
+		name
+	};
+	if (attributes) element.attributes = attributes;
+	return {
+		status: "parsed",
+		element,
+		name,
+		selfClosing
+	};
+};
+const appendElement = (parent, child) => {
+	parent.elements ??= [];
+	parent.elements.push(child);
+};
+const appendText = (text, stack, mergeAdjacentText) => {
+	if (text.length === 0) return true;
+	const decoded = decodeXmlEntities(normalizeLineEndings(text));
+	if (decoded === null) return false;
+	const parent = stack.at(-1)?.element;
+	const previous = mergeAdjacentText ? parent?.elements?.at(-1) : void 0;
+	if (previous?.type === "text" && typeof previous.text === "string") {
+		previous.text += decoded;
+		return true;
+	}
+	return appendRawText(decoded, stack);
+};
+const appendRawText = (text, stack) => {
+	const parent = stack.at(-1)?.element;
+	if (!parent) return text.trim().length === 0;
+	appendElement(parent, {
+		type: "text",
+		text
+	});
+	return true;
+};
+const findTagClose = (xml, start) => {
+	let quote = 0;
+	for (let cursor = start; cursor < xml.length; cursor += 1) {
+		const code = xml.charCodeAt(cursor);
+		if (quote !== 0) {
+			if (code === quote) quote = 0;
+			continue;
+		}
+		if (code === 34 || code === 39) {
+			quote = code;
+			continue;
+		}
+		if (code === 62) return cursor;
+	}
+	return -1;
+};
+const skipWhitespace = (xml, start, limit) => {
+	let cursor = start;
+	while (cursor < limit) {
+		const code = xml.charCodeAt(cursor);
+		if (code !== 9 && code !== 10 && code !== 13 && code !== 32) break;
+		cursor += 1;
+	}
+	return cursor;
+};
+const scanName = (xml, start, limit) => {
+	let cursor = start;
+	while (cursor < limit) {
+		const code = xml.charCodeAt(cursor);
+		if (code === 9 || code === 10 || code === 13 || code === 32 || code === 47 || code === 61 || code === 62) break;
+		cursor += 1;
+	}
+	return cursor;
+};
+const decodeXmlEntities = (value) => {
+	const firstEntity = value.indexOf("&");
+	if (firstEntity === -1) return value;
+	const parts = [];
+	let cursor = 0;
+	let entityStart = firstEntity;
+	while (entityStart !== -1) {
+		parts.push(value.slice(cursor, entityStart));
+		const entityEnd = value.indexOf(";", entityStart + 1);
+		if (entityEnd === -1) return null;
+		const entity = value.slice(entityStart + 1, entityEnd);
+		const decoded = decodeXmlEntity(entity);
+		if (decoded === null) return null;
+		parts.push(decoded);
+		cursor = entityEnd + 1;
+		entityStart = value.indexOf("&", cursor);
+	}
+	parts.push(value.slice(cursor));
+	return parts.join("");
+};
+const normalizeLineEndings = (value) => {
+	if (!value.includes("\r")) return value;
+	return value.replaceAll("\r\n", "\n").replaceAll("\r", "\n");
+};
+const decodeXmlEntity = (entity) => {
+	switch (entity) {
+		case "amp": return BUILT_IN_ENTITIES.amp;
+		case "apos": return BUILT_IN_ENTITIES.apos;
+		case "gt": return BUILT_IN_ENTITIES.gt;
+		case "lt": return BUILT_IN_ENTITIES.lt;
+		case "quot": return BUILT_IN_ENTITIES.quot;
+	}
+	if (entity.charCodeAt(0) !== 35) return null;
+	const hexadecimal = entity.charCodeAt(1) === 120 || entity.charCodeAt(1) === 88;
+	const digits = entity.slice(hexadecimal ? 2 : 1);
+	if (digits.length === 0 || !hasOnlyDigits(digits, hexadecimal)) return null;
+	const codePoint = Number.parseInt(digits, hexadecimal ? 16 : 10);
+	if (!Number.isInteger(codePoint) || codePoint <= 0 || codePoint > 1114111 || codePoint >= 55296 && codePoint <= 57343) return null;
+	return String.fromCodePoint(codePoint);
+};
+const hasOnlyDigits = (value, hexadecimal) => {
+	for (let index = 0; index < value.length; index += 1) {
+		const code = value.charCodeAt(index);
+		if (!(code >= 48 && code <= 57) && !(hexadecimal && code >= 97 && code <= 102) && !(hexadecimal && code >= 65 && code <= 70)) return false;
+	}
+	return true;
+};
+const isUnsafePropertyName = (name) => {
+	switch (name) {
+		case "__defineGetter__":
+		case "__defineSetter__":
+		case "__lookupGetter__":
+		case "__lookupSetter__":
+		case "__proto__":
+		case "constructor":
+		case "hasOwnProperty":
+		case "prototype":
+		case "toString":
+		case "valueOf": return true;
+		default: return false;
+	}
+};
+//#endregion
+export { parseStreamingXml };

package/dist/docx/styleParser.d.ts

@@ -1,5 +1,4 @@
 import { document_d_exports } from "../types/document.js";
-
 //#region src/docx/styleParser.d.ts
 /**
  * Style map keyed by styleId

package/dist/docx/styleParser.js

@@ -1,5 +1,5 @@
-import { BorderStyleSchema, ConditionalStyleTypeSchema, EmphasisMarkSchema, FontThemeSchema, HighlightColorSchema, LineSpacingRuleSchema, ParagraphAlignmentSchema, ShadingPatternSchema, StyleTypeSchema, TabLeaderSchema, TabStopAlignmentSchema, TableCellTextDirectionSchema, TableRowHeightRuleSchema, TableWidthTypeSchema, TextEffectSchema, ThemeColorSlotSchema, UnderlineStyleSchema, narrowEnum } from "./parserEnums.js";
-import { findChild, findChildren, getAttribute, getLocalName, parseBooleanElement, parseNumericAttribute, parseTableMeasurementValue, parseXmlDocument } from "./xmlParser.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 { findChild, findChildren, getAttribute, getLocalName, parseBooleanElement, parseNumberingLevelAttribute, parseNumericAttribute, parseTableMeasurementValue, parseXmlDocument } from "./xmlParser.js";
 import { mergeTextFormatting } from "../utils/textFormattingMerge.js";
 import { mergeParagraphFormatting } from "../utils/paragraphFormattingMerge.js";
 import { resolveThemeFontRef } from "./themeParser.js";
@@ -76,6 +76,8 @@ function parseRunProperties(rPr, theme) {
 		if (eastAsia) fontFamily.eastAsia = eastAsia;
 		const csFont = getAttribute(rFonts, "w", "cs");
 		if (csFont) fontFamily.cs = csFont;
+		const hint = narrowEnum(getAttribute(rFonts, "w", "hint"), FontHintSchema);
+		if (hint) fontFamily.hint = hint;
 		const asciiTheme = narrowEnum(getAttribute(rFonts, "w", "asciiTheme"), FontThemeSchema);
 		if (asciiTheme) {
 			fontFamily.asciiTheme = asciiTheme;
@@ -244,7 +246,7 @@ function parseTabStops(tabs) {
 				alignment
 			};
 			const leader = narrowEnum(getAttribute(tab, "w", "leader"), TabLeaderSchema);
-			if (leader) tabStop.leader = leader;
+			if (leader && leader !== "none") tabStop.leader = leader;
 			result.push(tabStop);
 		}
 	}
@@ -342,7 +344,7 @@ function parseParagraphProperties(pPr, theme) {
 				if (val !== void 0) formatting.numPr.numId = val;
 			}
 			if (ilvl) {
-				const val = parseNumericAttribute(ilvl, "w", "val");
+				const val = parseNumberingLevelAttribute(ilvl);
 				if (val !== void 0) formatting.numPr.ilvl = val;
 			}
 		}

package/dist/docx/tableParser.d.ts

@@ -2,7 +2,6 @@ import { document_d_exports } from "../types/document.js";
 import { NumberingMap } from "./numberingParser.js";
 import { StyleMap } from "./styleParser.js";
 import { XmlElement } from "./xmlParser.js";
-
 //#region src/docx/tableParser.d.ts
 /**
  * Sanity cap on `w:gridSpan` (and the derived table column count). Word's
@@ -139,7 +138,7 @@ declare function parseTableGrid(tblGridElement: XmlElement | null): number[] | u
 declare function parseTable(tblElement: XmlElement, styles: StyleMap | null, theme: document_d_exports.Theme | null, numbering: NumberingMap | null, rels: document_d_exports.RelationshipMap | null, media: Map<string, document_d_exports.MediaFile> | null, options?: {
   inHeaderFooter?: boolean;
   rootXmlns?: Record<string, string>;
-}): document_d_exports.Table;
+}): document_d_exports.Table | undefined;
 /**
  * Get the number of columns in a table
  *

package/dist/docx/tableParser.js

@@ -1,5 +1,5 @@
 import { BorderStyleSchema, FloatingTableXSpecSchema, FloatingTableYSpecSchema, ShadingPatternSchema, TableCellTextDirectionSchema, ThemeColorSlotSchema, narrowEnum } from "./parserEnums.js";
-import { findChild, findChildByLocalName, findChildren, getAttribute, getChildElements, mergeXmlnsDeclarations, parseBooleanElement, parseNumericAttribute, parseTableMeasurementValue } from "./xmlParser.js";
+import { findChild, findChildByLocalName, findChildren, getAttribute, getChildElements, getLocalName, mergeXmlnsDeclarations, parseBooleanElement, parseNumericAttribute, parseTableMeasurementValue } from "./xmlParser.js";
 import { parseBookmarkEnd, parseBookmarkStart } from "./bookmarkParser.js";
 import { parseParagraph } from "./paragraphParser.js";
 import { appendBookmarkMarkerToLastParagraphInBlocks, appendBookmarkMarkerToLastParagraphInCells, prependBookmarkMarkersToFirstParagraphInBlocks, prependBookmarkMarkersToFirstParagraphInCell } from "./bookmarkPlacement.js";
@@ -536,6 +536,7 @@ function parseCellContent(tcElement, styles, theme, numbering, rels, media, opti
 		}
 		if (localName === "tbl") {
 			const table = parseTable(child, styles, theme, numbering, rels, media, options);
+			if (!table) return;
 			if (prependBookmarkMarkersToFirstParagraphInBlocks([table], pendingBookmarkMarkers)) pendingBookmarkMarkers.length = 0;
 			content.push(table);
 			return;
@@ -635,11 +636,6 @@ function parseTableRow(trElement, styles, theme, numbering, rels, media, options
 	if (pendingBookmarkMarkers.length > 0) appendBookmarkMarkersToLastCell(row, pendingBookmarkMarkers);
 	return row;
 }
-function getLocalName(name) {
-	if (!name) return "";
-	const colonIndex = name.indexOf(":");
-	return colonIndex === -1 ? name : name.slice(colonIndex + 1);
-}
 function parseBookmarkMarker(child, localName) {
 	if (localName === "bookmarkStart") return parseBookmarkStart(child);
 	return parseBookmarkEnd(child);
@@ -751,6 +747,7 @@ function parseTable(tblElement, styles, theme, numbering, rels, media, options)
 		for (const sdtChild of getChildElements(sdtContent)) parseTableChild(sdtChild);
 	};
 	for (const child of getChildElements(tblElement)) parseTableChild(child);
+	if (table.rows.length === 0) return;
 	inferImplicitSingleCellRowSpans(table, rowsWithGridOffsets);
 	return table;
 }

package/dist/docx/textBoxParser.d.ts

@@ -2,7 +2,6 @@ import { document_d_exports } from "../types/document.js";
 import { NumberingMap } from "./numberingParser.js";
 import { StyleMap } from "./styleParser.js";
 import { XmlElement } from "./xmlParser.js";
-
 //#region src/docx/textBoxParser.d.ts
 /**
  * Extract raw paragraph elements from w:txbxContent
@@ -19,7 +18,7 @@ type ParagraphParserFn = (node: XmlElement, styles: StyleMap | null, theme: docu
 /**
  * Type for the table parser function to avoid circular imports
  */
-type TableParserFn = (node: XmlElement, styles: StyleMap | null, theme: document_d_exports.Theme | null, numbering: NumberingMap | null, rels: document_d_exports.RelationshipMap | null, media: Map<string, document_d_exports.MediaFile> | null) => document_d_exports.Table;
+type TableParserFn = (node: XmlElement, styles: StyleMap | null, theme: document_d_exports.Theme | null, numbering: NumberingMap | null, rels: document_d_exports.RelationshipMap | null, media: Map<string, document_d_exports.MediaFile> | null) => document_d_exports.Table | undefined;
 /**
  * Parse text box content with provided parser functions
  * This avoids circular dependencies by accepting parser functions as parameters

package/dist/docx/textBoxParser.js

@@ -55,7 +55,10 @@ function parseTextBoxContent(txbxContent, parseParagraph, parseTable, styles, th
 		if (localName === "p") {
 			const paragraph = parseParagraph(child, styles, theme, numbering, rels, media);
 			blocks.push(paragraph);
-		} else if (localName === "tbl" && parseTable) blocks.push(parseTable(child, styles, theme, numbering, rels, media));
+		} else if (localName === "tbl" && parseTable) {
+			const table = parseTable(child, styles, theme, numbering, rels, media);
+			if (table) blocks.push(table);
+		}
 	}
 	return blocks;
 }

package/dist/docx/textWhitespace.d.ts

@@ -0,0 +1,4 @@
+//#region src/docx/textWhitespace.d.ts
+declare const requiresXmlSpacePreserve: (text: string) => boolean;
+//#endregion
+export { requiresXmlSpacePreserve };

package/dist/docx/textWhitespace.js

@@ -0,0 +1,4 @@
+//#region src/docx/textWhitespace.ts
+const requiresXmlSpacePreserve = (text) => text.startsWith(" ") || text.endsWith(" ") || text.includes("  ");
+//#endregion
+export { requiresXmlSpacePreserve };

package/dist/docx/themeParser.d.ts

@@ -1,5 +1,4 @@
 import { document_d_exports } from "../types/document.js";
-
 //#region src/docx/themeParser.d.ts
 /**
  * Parse theme1.xml content

package/dist/docx/trackedMoveRangeNormalization.d.ts

@@ -1,5 +1,4 @@
 import { document_d_exports } from "../types/document.js";
-
 //#region src/docx/trackedMoveRangeNormalization.d.ts
 type NormalizeTrackedMoveRangesInput = {
   documentBody: document_d_exports.DocumentBody;
@@ -11,12 +10,6 @@ type NormalizeTrackedMoveRangesInput = {
 type NormalizeTrackedMoveRangesResult = {
   removedUnbalancedMoveRangeMarkers: number;
 };
-declare const normalizeTrackedMoveRanges: ({
-  documentBody,
-  headers,
-  footers,
-  footnotes,
-  endnotes
-}: NormalizeTrackedMoveRangesInput) => NormalizeTrackedMoveRangesResult;
+declare const normalizeTrackedMoveRanges: ({ documentBody, headers, footers, footnotes, endnotes }: NormalizeTrackedMoveRangesInput) => NormalizeTrackedMoveRangesResult;
 //#endregion
 export { normalizeTrackedMoveRanges };

package/dist/docx/unzip.d.ts

@@ -1,5 +1,4 @@
 import JSZip from "jszip";
-
 //#region src/docx/unzip.d.ts
 declare class DocxSecurityError extends Error {
   constructor(message: string);
@@ -14,7 +13,10 @@ type DocxUnzipLimits = {
   allowedMediaMimeTypes: ReadonlySet<string>;
 };
 type DocxUnzipOptions = Partial<Omit<DocxUnzipLimits, "allowedMediaMimeTypes">> & {
-  allowedMediaMimeTypes?: Iterable<string>; /** Password for Agile-encrypted .docx files (Office 2010+). */
+  allowedMediaMimeTypes?: Iterable<string>;
+  /** Extract every XML part into allXml instead of only parts used by the parser. */
+  extractAllXml?: boolean;
+  /** Password for Agile-encrypted .docx files (Office 2010+). */
   password?: string | undefined;
 };
 /**
@@ -46,7 +48,8 @@ type RawDocxContent = {
   allXml: Map<string, string>;
   originalZip: JSZip;
   originalBuffer: ArrayBuffer;
-  warnings: string[]; /** True when the input was a password-protected CFB container. */
+  warnings: string[];
+  /** True when the input was a password-protected CFB container. */
   wasEncrypted: boolean;
 };
 /**

package/dist/docx/unzip.js

@@ -59,6 +59,27 @@ const DEFAULT_UNZIP_LIMITS = {
 	maxTotalUncompressedBytes: 250 * MEBIBYTE,
 	allowedMediaMimeTypes: DEFAULT_ALLOWED_MEDIA_MIME_TYPES
 };
+const PARSED_XML_PARTS = /* @__PURE__ */ new Set([
+	"[content_types].xml",
+	"_rels/.rels",
+	"docprops/app.xml",
+	"docprops/core.xml",
+	"docprops/custom.xml",
+	"word/_rels/document.xml.rels",
+	"word/_rels/fonttable.xml.rels",
+	"word/comments.xml",
+	"word/commentsextended.xml",
+	"word/commentsextensible.xml",
+	"word/document.xml",
+	"word/endnotes.xml",
+	"word/fonttable.xml",
+	"word/footnotes.xml",
+	"word/numbering.xml",
+	"word/settings.xml",
+	"word/styles.xml",
+	"word/theme/theme1.xml"
+]);
+const shouldExtractXmlPart = (path) => PARSED_XML_PARTS.has(path) || path.startsWith("word/");
 const ZIP_END_OF_CENTRAL_DIRECTORY_SIGNATURE = 101010256;
 const ZIP_CENTRAL_DIRECTORY_FILE_HEADER_SIGNATURE = 33639248;
 const ZIP_END_OF_CENTRAL_DIRECTORY_SIZE = 22;
@@ -73,9 +94,10 @@ const ZIP_CENTRAL_DIRECTORY_FILE_HEADER_SIZE = 46;
 async function unzipDocx(buffer, options = {}) {
 	const limits = createUnzipLimits(options);
 	if (buffer.byteLength > limits.maxInputBytes) throw new DocxSecurityError("DOCX file exceeds the maximum allowed size");
-	const zipBuffer = await openDocxBuffer(buffer, { password: options.password });
+	const containerType = detectDocxContainerType(buffer);
+	const zipBuffer = containerType === DOCX_CONTAINER_TYPES.ZIP ? buffer : await openDocxBuffer(buffer, { password: options.password });
 	if (zipBuffer.byteLength > limits.maxInputBytes) throw new DocxSecurityError("DOCX file exceeds the maximum allowed size");
-	const wasEncrypted = detectDocxContainerType(buffer) === DOCX_CONTAINER_TYPES.CFB;
+	const wasEncrypted = containerType === DOCX_CONTAINER_TYPES.CFB;
 	const loaded = await loadDocxZip(zipBuffer, limits.maxFiles);
 	if (loaded.buffer.byteLength > limits.maxInputBytes) throw new DocxSecurityError("DOCX file exceeds the maximum allowed size");
 	const { zip } = loaded;
@@ -111,6 +133,7 @@ async function unzipDocx(buffer, options = {}) {
 		wasEncrypted
 	};
 	let totalUncompressedBytes = 0;
+	const extractionTasks = [];
 	for (const [path, file] of entries) {
 		if (!isSafeDocxPath(path)) throw new DocxSecurityError("DOCX file contains an unsafe entry path");
 		if (!isPreservableDocxEntry(path)) continue;
@@ -122,34 +145,16 @@ async function unzipDocx(buffer, options = {}) {
 		}
 		if (lowerPath.endsWith(".xml") || lowerPath.endsWith(".rels")) {
 			assertEntrySize(path, declaredSize, limits.maxXmlBytes);
-			const xmlContent = await file.async("text");
-			assertExtractedSize(path, xmlContent.length, limits.maxXmlBytes);
-			content.allXml.set(path, xmlContent);
-			if (lowerPath === "word/document.xml") content.documentXml = xmlContent;
-			else if (lowerPath === "word/styles.xml") content.stylesXml = xmlContent;
-			else if (lowerPath === "word/theme/theme1.xml") content.themeXml = xmlContent;
-			else if (lowerPath === "word/numbering.xml") content.numberingXml = xmlContent;
-			else if (lowerPath === "word/fonttable.xml") content.fontTableXml = xmlContent;
-			else if (lowerPath === "word/settings.xml") content.settingsXml = xmlContent;
-			else if (lowerPath === "word/websettings.xml") content.webSettingsXml = xmlContent;
-			else if (lowerPath === "word/footnotes.xml") content.footnotesXml = xmlContent;
-			else if (lowerPath === "word/endnotes.xml") content.endnotesXml = xmlContent;
-			else if (lowerPath === "word/comments.xml") content.commentsXml = xmlContent;
-			else if (lowerPath === "word/commentsextensible.xml") content.commentsExtensibleXml = xmlContent;
-			else if (lowerPath === "word/commentsextended.xml") content.commentsExtendedXml = xmlContent;
-			else if (lowerPath === "word/_rels/document.xml.rels") content.documentRels = xmlContent;
-			else if (lowerPath === "_rels/.rels") content.packageRels = xmlContent;
-			else if (lowerPath === "[content_types].xml") content.contentTypesXml = xmlContent;
-			else if (lowerPath === "docprops/core.xml") content.corePropsXml = xmlContent;
-			else if (lowerPath === "docprops/app.xml") content.appPropsXml = xmlContent;
-			else if (lowerPath === "docprops/custom.xml") content.customPropsXml = xmlContent;
-			else if (/^word\/header[^/]*\.xml$/u.test(lowerPath)) {
-				const filename = path.split("/").pop() || path;
-				content.headers.set(filename, xmlContent);
-			} else if (/^word\/footer[^/]*\.xml$/u.test(lowerPath)) {
-				const filename = path.split("/").pop() || path;
-				content.footers.set(filename, xmlContent);
-			}
+			if (options.extractAllXml === false && !shouldExtractXmlPart(lowerPath)) continue;
+			extractionTasks.push(file.async("text").then((xmlContent) => {
+				assertExtractedSize(path, xmlContent.length, limits.maxXmlBytes);
+				return {
+					type: "xml",
+					path,
+					lowerPath,
+					content: xmlContent
+				};
+			}));
 		} else if (lowerPath.startsWith("word/media/")) {
 			const mimeType = getMediaMimeType(path);
 			if (!limits.allowedMediaMimeTypes.has(mimeType)) continue;
@@ -157,22 +162,73 @@ async function unzipDocx(buffer, options = {}) {
 				content.warnings.push(`Skipped oversized media file: ${path}; original entry preserved for round-trip.`);
 				continue;
 			}
-			const binaryContent = await file.async("arraybuffer");
-			if (binaryContent.byteLength > limits.maxMediaBytes) {
-				content.warnings.push(`Skipped oversized media file: ${path}; original entry preserved for round-trip.`);
-				continue;
-			}
-			if (!isMediaContentAllowed(binaryContent, mimeType)) continue;
-			content.media.set(path, binaryContent);
+			extractionTasks.push(file.async("arraybuffer").then((binaryContent) => {
+				if (binaryContent.byteLength > limits.maxMediaBytes) {
+					content.warnings.push(`Skipped oversized media file: ${path}; original entry preserved for round-trip.`);
+					return null;
+				}
+				if (!isMediaContentAllowed(binaryContent, mimeType)) return null;
+				return {
+					type: "media",
+					path,
+					mimeType,
+					content: binaryContent
+				};
+			}));
 		} else if (lowerPath.startsWith("word/fonts/")) {
 			if (isEntryTooLarge(declaredSize, limits.maxFontBytes)) continue;
-			const binaryContent = await file.async("arraybuffer");
-			if (binaryContent.byteLength > limits.maxFontBytes) continue;
-			content.fonts.set(path, binaryContent);
+			extractionTasks.push(file.async("arraybuffer").then((binaryContent) => {
+				if (binaryContent.byteLength > limits.maxFontBytes) return null;
+				return {
+					type: "font",
+					path,
+					content: binaryContent
+				};
+			}));
 		}
 	}
+	for (const extracted of await Promise.all(extractionTasks)) {
+		if (!extracted) continue;
+		if (extracted.type === "xml") {
+			assignXmlContent(content, extracted);
+			continue;
+		}
+		if (extracted.type === "media") {
+			content.media.set(extracted.path, extracted.content);
+			continue;
+		}
+		content.fonts.set(extracted.path, extracted.content);
+	}
 	return content;
 }
+function assignXmlContent(content, { path, lowerPath, content: xmlContent }) {
+	content.allXml.set(path, xmlContent);
+	if (lowerPath === "word/document.xml") content.documentXml = xmlContent;
+	else if (lowerPath === "word/styles.xml") content.stylesXml = xmlContent;
+	else if (lowerPath === "word/theme/theme1.xml") content.themeXml = xmlContent;
+	else if (lowerPath === "word/numbering.xml") content.numberingXml = xmlContent;
+	else if (lowerPath === "word/fonttable.xml") content.fontTableXml = xmlContent;
+	else if (lowerPath === "word/settings.xml") content.settingsXml = xmlContent;
+	else if (lowerPath === "word/websettings.xml") content.webSettingsXml = xmlContent;
+	else if (lowerPath === "word/footnotes.xml") content.footnotesXml = xmlContent;
+	else if (lowerPath === "word/endnotes.xml") content.endnotesXml = xmlContent;
+	else if (lowerPath === "word/comments.xml") content.commentsXml = xmlContent;
+	else if (lowerPath === "word/commentsextensible.xml") content.commentsExtensibleXml = xmlContent;
+	else if (lowerPath === "word/commentsextended.xml") content.commentsExtendedXml = xmlContent;
+	else if (lowerPath === "word/_rels/document.xml.rels") content.documentRels = xmlContent;
+	else if (lowerPath === "_rels/.rels") content.packageRels = xmlContent;
+	else if (lowerPath === "[content_types].xml") content.contentTypesXml = xmlContent;
+	else if (lowerPath === "docprops/core.xml") content.corePropsXml = xmlContent;
+	else if (lowerPath === "docprops/app.xml") content.appPropsXml = xmlContent;
+	else if (lowerPath === "docprops/custom.xml") content.customPropsXml = xmlContent;
+	else if (/^word\/header[^/]*\.xml$/u.test(lowerPath)) {
+		const filename = path.split("/").pop() || path;
+		content.headers.set(filename, xmlContent);
+	} else if (/^word\/footer[^/]*\.xml$/u.test(lowerPath)) {
+		const filename = path.split("/").pop() || path;
+		content.footers.set(filename, xmlContent);
+	}
+}
 async function loadDocxZip(buffer, maxFiles) {
 	try {
 		return {

package/dist/docx/vmlImageParser.d.ts

@@ -1,6 +1,5 @@
 import { document_d_exports } from "../types/document.js";
 import { XmlElement } from "./xmlParser.js";
-
 //#region src/docx/vmlImageParser.d.ts
 /**
  * Parse a `w:pict` element into an inline image, or null when it carries no

package/dist/docx/vmlImageParser.js

@@ -1,4 +1,4 @@
-import { cloneWithXmlnsDeclarations, elementToXml, findAllDeep, findChild, getAttribute, getChildElements, getLocalName } from "./xmlParser.js";
+import { cloneWithXmlnsDeclarations, elementToXml, findAllDeep, findChild, findDeep, getAttribute, getChildElements, getLocalName } from "./xmlParser.js";
 import { pixelsToEmu } from "../utils/units.js";
 import { sanitizeImageSrc } from "../utils/sanitizeImageSrc.js";
 import { resolveImageData } from "./imageParser.js";
@@ -136,6 +136,7 @@ const previewImage = (pictElement, svg, widthPx, heightPx, style, rootXmlns) =>
 	};
 };
 const parseStandaloneShapePreview = (pictElement, shape, rootXmlns) => {
+	if (findDeep(shape, "v", "textbox")) return null;
 	const style = parseStyleAttr(getAttribute(shape, null, "style"));
 	const widthPx = cssLengthToPx(style["width"]);
 	const heightPx = cssLengthToPx(style["height"]);
@@ -159,6 +160,7 @@ const parseGroupPreview = (pictElement, group, rootXmlns) => {
 	if (size.first <= 0 || size.second <= 0) return null;
 	const content = getChildElements(group).slice(0, MAX_VML_PREVIEW_SHAPES).map((child) => {
 		const localName = getLocalName(child.name ?? "");
+		if (findDeep(child, "v", "textbox")) return "";
 		if (localName === "line") {
 			const from = coordinatePair(getAttribute(child, null, "from"));
 			const to = coordinatePair(getAttribute(child, null, "to"));

package/dist/docx/watermarkParser.d.ts

@@ -1,6 +1,5 @@
 import { document_d_exports } from "../types/document.js";
 import { XmlElement } from "./xmlParser.js";
-
 //#region src/docx/watermarkParser.d.ts
 /**
  * Result of walking a header for its watermark. Carries the modeled