@stll/folio-core 0.11.0 → 0.13.0

package/dist/docx/paragraphParser.js

@@ -1,12 +1,14 @@
-import { elementToXml, findChild, findChildren, getAttribute, getChildElements, mergeXmlnsDeclarations, parseBooleanElement, parseNumericAttribute } from "./xmlParser.js";
-import { parseBookmarkEnd as parseBookmarkEnd$1, parseBookmarkStart as parseBookmarkStart$1 } from "./bookmarkParser.js";
 import { BorderStyleSchema, FrameWrapSchema, FrameXAlignSchema, FrameYAlignSchema, LineSpacingRuleSchema, ParagraphAlignmentSchema, ShadingPatternSchema, TabLeaderSchema, TabStopAlignmentSchema, ThemeColorSlotSchema, narrowEnum } from "./parserEnums.js";
+import { elementToXml, findChild, findChildren, getAttribute, getChildElements, mergeXmlnsDeclarations, parseBooleanElement, parseNumericAttribute } from "./xmlParser.js";
 import { parseRun, parseRunProperties } from "./runParser.js";
 import { parseFieldType } from "./fieldParser.js";
+import { isValidHexId } from "../utils/hexId.js";
+import { parseBookmarkEnd as parseBookmarkEnd$1, parseBookmarkStart as parseBookmarkStart$1 } from "./bookmarkParser.js";
 import { parseHyperlink as parseHyperlink$1 } from "./hyperlinkParser.js";
 import { consolidateParagraphContent } from "./runConsolidator.js";
 import { parseSdtProperties } from "./sdtProperties.js";
 import { parseSectionProperties } from "./sectionParser.js";
+import { normalizeRevisionId } from "@stll/docx-core/model";
 //#region src/docx/paragraphParser.ts
 /**
 * Extract plain text from a math element (recursive text content extraction)
@@ -191,6 +193,9 @@ function collectFirstParagraphPropertyChildren(pPr) {
 			case "shd":
 				children.shd ??= child;
 				break;
+			case "snapToGrid":
+				children.snapToGrid ??= child;
+				break;
 			case "spacing":
 				children.spacing ??= child;
 				break;
@@ -243,6 +248,8 @@ function parseParagraphProperties(pPr, theme, styles) {
 	}
 	const bidi = propertyChildren.bidi;
 	if (bidi) formatting.bidi = parseBooleanElement(bidi);
+	const snapToGrid = propertyChildren.snapToGrid;
+	if (snapToGrid) formatting.snapToGrid = parseBooleanElement(snapToGrid);
 	const spacing = propertyChildren.spacing;
 	if (spacing) {
 		const before = parseNumericAttribute(spacing, "w", "before");
@@ -369,64 +376,64 @@ function getLocalName(name) {
 	const colonIndex = name.indexOf(":");
 	return colonIndex !== -1 ? name.slice(colonIndex + 1) : name;
 }
+const RENDERED_BREAK_INLINE_WRAPPERS = /* @__PURE__ */ new Set([
+	"hyperlink",
+	"smartTag",
+	"sdt",
+	"sdtContent",
+	"fldSimple",
+	"customXml",
+	"ins",
+	"del",
+	"moveFrom",
+	"moveTo"
+]);
+const RENDERED_BREAK_NON_CONTENT_MARKERS = /* @__PURE__ */ new Set([
+	"pPr",
+	"proofErr",
+	"bookmarkStart",
+	"bookmarkEnd",
+	"commentRangeStart",
+	"commentRangeEnd",
+	"commentReference",
+	"permStart",
+	"permEnd",
+	"rsidR"
+]);
+const RENDERED_BREAK_VISIBLE_RUN_CONTENT = /* @__PURE__ */ new Set([
+	"t",
+	"tab",
+	"br",
+	"cr",
+	"sym",
+	"drawing",
+	"pict",
+	"object",
+	"softHyphen",
+	"noBreakHyphen",
+	"fldChar",
+	"instrText",
+	"pgNum",
+	"separator",
+	"continuationSeparator",
+	"footnoteRef",
+	"endnoteRef",
+	"footnoteReference",
+	"endnoteReference",
+	"ptab",
+	"monthShort",
+	"monthLong",
+	"yearShort",
+	"yearLong",
+	"dayShort",
+	"dayLong"
+]);
 function paragraphStartsWithRenderedPageBreak(node) {
-	const inlineWrappers = /* @__PURE__ */ new Set([
-		"hyperlink",
-		"smartTag",
-		"sdt",
-		"sdtContent",
-		"fldSimple",
-		"customXml",
-		"ins",
-		"del",
-		"moveFrom",
-		"moveTo"
-	]);
-	const nonContentMarkers = /* @__PURE__ */ new Set([
-		"pPr",
-		"proofErr",
-		"bookmarkStart",
-		"bookmarkEnd",
-		"commentRangeStart",
-		"commentRangeEnd",
-		"commentReference",
-		"permStart",
-		"permEnd",
-		"rsidR"
-	]);
-	const visibleRunContent = /* @__PURE__ */ new Set([
-		"t",
-		"tab",
-		"br",
-		"cr",
-		"sym",
-		"drawing",
-		"pict",
-		"object",
-		"softHyphen",
-		"noBreakHyphen",
-		"fldChar",
-		"instrText",
-		"pgNum",
-		"separator",
-		"continuationSeparator",
-		"footnoteRef",
-		"endnoteRef",
-		"footnoteReference",
-		"endnoteReference",
-		"ptab",
-		"monthShort",
-		"monthLong",
-		"yearShort",
-		"yearLong",
-		"dayShort",
-		"dayLong"
-	]);
 	let sawRenderedPageBreak = false;
 	const visit = (element) => {
 		for (const child of getChildElements(element)) {
 			const childName = getLocalName(child.name);
-			if (nonContentMarkers.has(childName)) continue;
+			if (RENDERED_BREAK_NON_CONTENT_MARKERS.has(childName)) continue;
 			if (childName === "lastRenderedPageBreak") {
 				sawRenderedPageBreak = true;
 				continue;
@@ -440,11 +447,11 @@ function paragraphStartsWithRenderedPageBreak(node) {
 						continue;
 					}
 					if (runChildName === "br" && getAttribute(runChild, "w", "type") === "page") return "forced";
-					if (visibleRunContent.has(runChildName)) return "visible";
+					if (RENDERED_BREAK_VISIBLE_RUN_CONTENT.has(runChildName)) return "visible";
 				}
 				continue;
 			}
-			if (inlineWrappers.has(childName)) {
+			if (RENDERED_BREAK_INLINE_WRAPPERS.has(childName)) {
 				const result = visit(child);
 				if (result !== "continue") return result;
 				continue;
@@ -481,11 +488,13 @@ function parseTrackedChangeInfo(node) {
 	const rawDate = getAttribute(node, "w", "date");
 	const author = rawAuthor?.trim() ?? "";
 	const date = rawDate?.trim() ?? "";
+	const initials = (getAttribute(node, "w", "initials") ?? "").trim();
 	const info = {
-		id: Number.isInteger(parsedId) && parsedId >= 0 ? parsedId : 0,
+		id: normalizeRevisionId(parsedId),
 		author: author.length > 0 ? author : "Unknown"
 	};
 	if (date.length > 0) info.date = date;
+	if (initials.length > 0) info.initials = initials;
 	return info;
 }
 function parsePropertyChangeInfo(node) {
@@ -535,7 +544,7 @@ function isTrackedChangeWrapperChild(content) {
 	return content.type === "run" || content.type === "hyperlink";
 }
 function isInlineSdtContent(content) {
-	return content.type === "run" || content.type === "hyperlink" || content.type === "simpleField" || content.type === "complexField" || content.type === "inlineSdt" || content.type === "insertion" || content.type === "deletion" || content.type === "mathEquation";
+	return content.type === "run" || content.type === "hyperlink" || content.type === "simpleField" || content.type === "complexField" || content.type === "inlineSdt" || content.type === "insertion" || content.type === "deletion" || content.type === "moveFrom" || content.type === "moveTo" || content.type === "mathEquation";
 }
 function pushTrackedChangeWrapper({ contents, type, info, content, preserveEmpty = false }) {
 	if (content.length === 0 && !preserveEmpty) return;
@@ -934,9 +943,9 @@ function parseParagraph(node, styles, theme, numbering, rels = null, media = nul
 		content: []
 	};
 	const paraId = getAttribute(node, "w14", "paraId") ?? getAttribute(node, "w", "paraId");
-	if (paraId) paragraph.paraId = paraId;
+	if (paraId && isValidHexId(paraId)) paragraph.paraId = paraId;
 	const textId = getAttribute(node, "w14", "textId") ?? getAttribute(node, "w", "textId");
-	if (textId) paragraph.textId = textId;
+	if (textId && isValidHexId(textId)) paragraph.textId = textId;
 	if (!options?.inHeaderFooter && paragraphStartsWithRenderedPageBreak(node)) paragraph.renderedPageBreakBefore = true;
 	const pPr = findChild(node, "w", "pPr");
 	if (pPr) {

package/dist/docx/paragraphTextBoxEnrichment.d.ts

@@ -2,8 +2,9 @@ import { document_d_exports } from "../types/document.js";
 import { NumberingMap } from "./numberingParser.js";
 import { StyleMap } from "./styleParser.js";
 import { XmlElement } from "./xmlParser.js";
+import { TableParserFn } from "./textBoxParser.js";
 
 //#region src/docx/paragraphTextBoxEnrichment.d.ts
-declare const enrichParagraphTextBoxes: (paragraph: document_d_exports.Paragraph, paraXml: 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) => void;
+declare const enrichParagraphTextBoxes: (paragraph: document_d_exports.Paragraph, paraXml: 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, parseTable: TableParserFn) => void;
 //#endregion
 export { enrichParagraphTextBoxes };

package/dist/docx/paragraphTextBoxEnrichment.js

@@ -72,26 +72,80 @@ const vmlWrapType = (positioned, zIndex) => {
 	if (Number.isFinite(zIndex) && zIndex < 0) return "behind";
 	return "inFront";
 };
-const enrichParagraphTextBoxes = (paragraph, paraXml, styles, theme, numbering, rels, media) => {
-	const xmlChildren = getChildElements(paraXml);
+const enrichParagraphTextBoxes = (paragraph, paraXml, styles, theme, numbering, rels, media, parseTable) => {
+	enrichTextBoxRuns({
+		content: paragraph.content,
+		xmlChildren: getChildElements(paraXml),
+		styles,
+		theme,
+		numbering,
+		rels,
+		media,
+		parseTable
+	});
+};
+const trackedChangeTypeFromXml = (localName) => {
+	if (localName === "ins") return "insertion";
+	if (localName === "del") return "deletion";
+	if (localName === "moveFrom" || localName === "moveTo") return localName;
+};
+const enrichTextBoxRuns = ({ content, xmlChildren, styles, theme, numbering, rels, media, parseTable }) => {
 	let parsedIndex = 0;
 	let lastConsumedRun;
 	for (const xmlChild of xmlChildren) {
-		if (getLocalName(xmlChild.name ?? "") !== "r") {
-			if (parsedIndex < paragraph.content.length && paragraph.content[parsedIndex]?.type !== "run") parsedIndex += 1;
+		const localName = getLocalName(xmlChild.name ?? "");
+		if (localName === "pPr") continue;
+		const trackedChangeType = trackedChangeTypeFromXml(localName);
+		const parsedContent = content[parsedIndex];
+		if (trackedChangeType && parsedContent?.type === trackedChangeType) enrichTextBoxRuns({
+			content: parsedContent.content,
+			xmlChildren: getChildElements(xmlChild),
+			styles,
+			theme,
+			numbering,
+			rels,
+			media,
+			parseTable
+		});
+		if (localName === "sdt" && parsedContent?.type === "inlineSdt") {
+			const properties = parsedContent.properties;
+			const nestedContent = [];
+			let lastSegmentIndex = parsedIndex;
+			for (let index = parsedIndex; index < content.length; index += 1) {
+				const candidate = content[index];
+				if (candidate?.type !== "inlineSdt" || candidate.properties !== properties) continue;
+				nestedContent.push(...candidate.content);
+				lastSegmentIndex = index;
+			}
+			const sdtContent = getChildElements(xmlChild).find((child) => getLocalName(child.name ?? "") === "sdtContent");
+			if (sdtContent) enrichTextBoxRuns({
+				content: nestedContent,
+				xmlChildren: getChildElements(sdtContent),
+				styles,
+				theme,
+				numbering,
+				rels,
+				media,
+				parseTable
+			});
+			parsedIndex = lastSegmentIndex + 1;
+			continue;
+		}
+		if (localName !== "r") {
+			if (parsedIndex < content.length && parsedContent?.type !== "run") parsedIndex += 1;
 			continue;
 		}
 		const { textBoxDrawings, vmlTextBoxes, hasNonTextBoxContent } = scanRunForTextBoxDrawings(xmlChild);
-		const parsedContent = paragraph.content[parsedIndex];
 		const parsedRun = parsedContent?.type === "run" ? parsedContent : void 0;
 		const targetRun = parsedRun ?? (hasNonTextBoxContent ? lastConsumedRun : void 0);
+		const targetRunMatchesXml = targetRun !== void 0 && (hasNonTextBoxContent || parsedRun?.content.length === 0);
 		for (const runEl of textBoxDrawings) {
 			const textBox = parseTextBox(runEl);
 			if (!textBox) continue;
 			const wsp = findDeep(runEl, "wps", "wsp");
 			if (wsp) {
 				const txbxContentEl = getTextBoxContentElement(wsp);
-				if (txbxContentEl) textBox.content = parseTextBoxContent(txbxContentEl, parseParagraph, null, styles, theme, numbering, rels ?? void 0, media ?? void 0);
+				if (txbxContentEl) textBox.content = parseTextBoxContent(txbxContentEl, parseParagraph, parseTable, styles, theme, numbering, rels, media);
 			}
 			const shape = {
 				type: "shape",
@@ -112,31 +166,31 @@ const enrichParagraphTextBoxes = (paragraph, paraXml, styles, theme, numbering,
 				type: "shape",
 				shape
 			};
-			if (targetRun && hasNonTextBoxContent) targetRun.content.push(shapeContent);
+			if (targetRunMatchesXml) targetRun.content.push(shapeContent);
 			else {
 				const newRun = {
 					type: "run",
 					content: [shapeContent]
 				};
-				paragraph.content.splice(parsedIndex, 0, newRun);
+				content.splice(parsedIndex, 0, newRun);
 				lastConsumedRun = newRun;
 				parsedIndex += 1;
 			}
 		}
 		for (const pictEl of vmlTextBoxes) {
-			const shape = parseVmlTextBoxShape(pictEl, styles, theme, numbering, rels, media);
+			const shape = parseVmlTextBoxShape(pictEl, styles, theme, numbering, rels, media, parseTable);
 			if (!shape) continue;
 			const shapeContent = {
 				type: "shape",
 				shape
 			};
-			if (targetRun && hasNonTextBoxContent) targetRun.content.push(shapeContent);
+			if (targetRunMatchesXml) targetRun.content.push(shapeContent);
 			else {
 				const newRun = {
 					type: "run",
 					content: [shapeContent]
 				};
-				paragraph.content.splice(parsedIndex, 0, newRun);
+				content.splice(parsedIndex, 0, newRun);
 				lastConsumedRun = newRun;
 				parsedIndex += 1;
 			}
@@ -196,7 +250,7 @@ const scanRunForTextBoxDrawings = (xmlRun) => {
 		hasNonTextBoxContent
 	};
 };
-const parseVmlTextBoxShape = (pictEl, styles, theme, numbering, rels, media) => {
+const parseVmlTextBoxShape = (pictEl, styles, theme, numbering, rels, media, parseTable) => {
 	const shapeEl = findDeep(pictEl, "v", "shape");
 	const textBoxEl = shapeEl ? findDeep(shapeEl, "v", "textbox") : null;
 	const contentEl = textBoxEl ? findDeep(textBoxEl, "w", "txbxContent") : null;
@@ -221,7 +275,7 @@ const parseVmlTextBoxShape = (pictEl, styles, theme, numbering, rels, media) =>
 		...fill === void 0 ? {} : { fill },
 		wrap: { type: vmlWrapType(positioned, zIndex) },
 		textBody: {
-			content: parseTextBoxContent(contentEl, parseParagraph, null, styles, theme, numbering, rels ?? void 0, media ?? void 0),
+			content: parseTextBoxContent(contentEl, parseParagraph, parseTable, styles, theme, numbering, rels, media),
 			...margins === void 0 ? {} : { margins }
 		}
 	};

package/dist/docx/paragraphTraversal.js

@@ -10,7 +10,7 @@ const visitDocxParagraphs = ({ documentBody, headers, footers, footnotes, endnot
 	const visitRun = (run) => {
 		for (const content of run.content) {
 			if (content.type !== "shape" || !content.shape.textBody) continue;
-			for (const paragraph of content.shape.textBody.content) visitParagraph(paragraph);
+			for (const block of content.shape.textBody.content) visitBlock(block);
 		}
 	};
 	const visitHyperlink = (hyperlink) => {

package/dist/docx/parser.js

@@ -1,13 +1,13 @@
+import { parseNumbering } from "./numberingParser.js";
+import { RELATIONSHIP_TYPES, parseRelationships, resolveRelativePath } from "./relsParser.js";
+import { applyThemeFontLang, parseTheme } from "./themeParser.js";
 import { toArrayBuffer } from "../utils/docxInput.js";
 import { loadFontsWithMapping } from "../utils/fontLoader.js";
 import { convertTiffToPngDataUrl, isTiffMimeType } from "../utils/tiffConverter.js";
-import { RELATIONSHIP_TYPES, parseRelationships, resolveRelativePath } from "./relsParser.js";
-import { applyThemeFontLang, parseTheme } from "./themeParser.js";
 import { parseComments } from "./commentParser.js";
 import { normalizeCommentReferences } from "./commentReferenceNormalization.js";
 import { detectDocxConformanceClass } from "./conformance.js";
 import { parseCoreProperties } from "./corePropertiesParser.js";
-import { parseNumbering } from "./numberingParser.js";
 import { extractAllTemplateVariables, parseDocumentBody } from "./documentParser.js";
 import { parseEndnotes, parseFootnotes } from "./footnoteParser.js";
 import { parseFooter, parseHeader } from "./headerFooterParser.js";

package/dist/docx/rezip.js

@@ -1,10 +1,11 @@
-import { RELATIONSHIP_TYPES, parseRelationships, resolveRelativePath } from "./relsParser.js";
 import { parseNumbering } from "./numberingParser.js";
+import { RELATIONSHIP_TYPES, parseRelationships, resolveRelativePath } from "./relsParser.js";
 import { parseEndnotes, parseFootnotes } from "./footnoteParser.js";
 import { assertValidFolioDocumentModel } from "./modelValidation.js";
 import { isPreservableDocxEntry } from "./unzip.js";
 import { applyReplyThreadMarkers } from "./commentReplyMarkers.js";
-import { buildPatchedNoteXml, buildPatchedNumberingXml, collectChangedNumberingDefs, collectParaIds, extractParagraphXml } from "./selectiveXmlPatch.js";
+import { withoutOrphanCommentRanges } from "./commentRangeIntegrity.js";
+import { buildParagraphOffsetIndex, buildPatchedNoteXml, buildPatchedNumberingXml, collectChangedNumberingDefs, collectParaIds } from "./selectiveXmlPatch.js";
 import { escapeXml } from "./serializer/xmlUtils.js";
 import { ensureThreadedCommentParaIds, serializeComments, serializeCommentsExtended } from "./serializer/commentSerializer.js";
 import { serializeDocument } from "./serializer/documentSerializer.js";
@@ -15,6 +16,7 @@ import { serializeFontTableXml } from "./serializer/fontTableSerializer.js";
 import { serializeSettingsXml } from "./serializer/settingsSerializer.js";
 import { serializeStylesXml } from "./serializer/stylesSerializer.js";
 import { serializeThemeXml } from "./serializer/themeSerializer.js";
+import { isAllowedExternalWatermarkImageUrl } from "../watermark/index.js";
 import { panic } from "better-result";
 import JSZip from "jszip";
 //#region src/docx/rezip.ts
@@ -369,7 +371,7 @@ async function processNewHyperlinks(newHyperlinks, zip, compressionLevel) {
 async function repackDocx(doc, options = {}) {
 	if (!doc.originalBuffer) panic("Cannot repack document: no original buffer for round-trip. Use createDocx() for new documents.");
 	const { compressionLevel = 6, updateModifiedDate = true, modifiedBy } = options;
-	const exportDocument = doc;
+	const exportDocument = withoutOrphanCommentRanges(doc);
 	const originalZip = await JSZip.loadAsync(doc.originalBuffer);
 	const newZip = new JSZip();
 	for (const [path, file] of Object.entries(originalZip.files)) {
@@ -431,7 +433,7 @@ async function repackDocx(doc, options = {}) {
 */
 async function repackDocxFromRaw(doc, rawContent, options = {}) {
 	const { compressionLevel = 6, updateModifiedDate = true, modifiedBy } = options;
-	const exportDocument = doc;
+	const exportDocument = withoutOrphanCommentRanges(doc);
 	const newZip = new JSZip();
 	for (const [path, file] of Object.entries(rawContent.originalZip.files)) {
 		if (file.dir) {
@@ -854,10 +856,11 @@ async function rebindWatermarkRelIds(doc, zip, compressionLevel) {
 	const changedPaths = /* @__PURE__ */ new Set();
 	for (const { watermark, relsPath, partPath } of pending) {
 		let canonical;
-		if (watermark.imageTarget !== void 0) canonical = watermark.imageTargetExternal ? {
+		if (watermark.imageTarget !== void 0) if (watermark.imageTargetExternal) canonical = isAllowedExternalWatermarkImageUrl(watermark.imageTarget) ? {
 			mode: "external",
 			url: watermark.imageTarget
-		} : {
+		} : resolveCanonical(watermark.imageRId);
+		else canonical = {
 			mode: "internal",
 			absolute: watermark.imageTarget
 		};
@@ -1000,15 +1003,23 @@ async function patchNotePartIntoZip(conventionalLowerPath, currentXml, baselineF
 * (re-parsed original) serialization — i.e. the ones actually edited. A
 * paragraph is only considered when its `paraId` resolves uniquely in both,
 * so it can be spliced safely.
+*
+* Builds one {@link buildParagraphOffsetIndex} per side (a single linear scan
+* each) instead of calling `extractParagraphXml` per candidate id, which
+* re-scanned the whole XML per id — O(note count * XML size) for a document
+* with many footnotes/endnotes. The index turns each lookup below into O(1).
 */
 function collectChangedNoteParaIds(baselineXml, currentXml) {
 	const changed = /* @__PURE__ */ new Set();
 	const baselineIds = collectParaIds(baselineXml);
+	const baselineOffsets = buildParagraphOffsetIndex(baselineXml);
+	const currentOffsets = buildParagraphOffsetIndex(currentXml);
 	for (const [id, count] of collectParaIds(currentXml)) {
 		if (count !== 1 || baselineIds.get(id) !== 1) continue;
-		const before = extractParagraphXml(baselineXml, id);
-		const after = extractParagraphXml(currentXml, id);
-		if (before !== null && after !== null && before !== after) changed.add(id);
+		const beforeRange = baselineOffsets.get(id);
+		const afterRange = currentOffsets.get(id);
+		if (!beforeRange || !afterRange) continue;
+		if (baselineXml.slice(beforeRange.start, beforeRange.end) !== currentXml.slice(afterRange.start, afterRange.end)) changed.add(id);
 	}
 	return changed;
 }

package/dist/docx/runParser.js

@@ -1,12 +1,21 @@
-import { cloneWithXmlnsDeclarations, elementToXml, findAllDeep, findChild, findChildren, getAttribute, getChildElements, getTextContent, mergeXmlnsDeclarations, parseBooleanElement, parseNumericAttribute } from "./xmlParser.js";
 import { EmphasisMarkSchema, FontThemeSchema, HighlightColorSchema, ShadingPatternSchema, TextEffectSchema, ThemeColorSlotSchema, UnderlineStyleSchema, narrowEnum } from "./parserEnums.js";
+import { cloneWithXmlnsDeclarations, elementToXml, findAllDeep, findChild, findChildren, getAttribute, getChildElements, getTextContent, mergeXmlnsDeclarations, parseBooleanElement, parseNumericAttribute } from "./xmlParser.js";
 import { parseImage } from "./imageParser.js";
 import { parseGroupDrawing } from "./groupDrawingParser.js";
 import { parseShapeFromDrawing, shouldPreserveRawShapeDrawing } from "./shapeParser.js";
 import { parseVmlImageContent } from "./vmlImageParser.js";
 import { resolveThemeFontRef } from "./themeParser.js";
+import { normalizeRevisionId } from "@stll/docx-core/model";
 //#region src/docx/runParser.ts
 /**
+* Sanity cap on `w:lang` `@w:val`/`@w:eastAsia`/`@w:bidi` tag length. BCP-47
+* tags top out well under this; a hostile/corrupt tag here would drive the
+* hyphenation dictionary lookup and segmenter cache keying with an
+* attacker-sized string per run.
+*/
+const MAX_LANGUAGE_TAG_LENGTH = 35;
+const truncateLanguageTag = (value) => value === void 0 ? void 0 : value.slice(0, MAX_LANGUAGE_TAG_LENGTH);
+/**
 * Parse color value from attributes
 */
 function parseColorValue(rgb, themeColor, themeTint, themeShade) {
@@ -267,9 +276,9 @@ function parseRunProperties(rPr, theme, _styles) {
 	}
 	const lang = propertyChildren.lang;
 	if (lang) {
-		const val = getAttribute(lang, "w", "val") || void 0;
-		const eastAsia = getAttribute(lang, "w", "eastAsia") || void 0;
-		const bidi = getAttribute(lang, "w", "bidi") || void 0;
+		const val = truncateLanguageTag(getAttribute(lang, "w", "val") || void 0);
+		const eastAsia = truncateLanguageTag(getAttribute(lang, "w", "eastAsia") || void 0);
+		const bidi = truncateLanguageTag(getAttribute(lang, "w", "bidi") || void 0);
 		if (val || eastAsia || bidi) formatting.language = {
 			...val ? { val } : {},
 			...eastAsia ? { eastAsia } : {},
@@ -332,7 +341,7 @@ function parsePropertyChangeInfo(changeElement) {
 	const date = (getAttribute(changeElement, "w", "date") ?? "").trim();
 	const rsid = (getAttribute(changeElement, "w", "rsid") ?? "").trim();
 	const info = {
-		id: Number.isInteger(parsedId) && parsedId >= 0 ? parsedId : 0,
+		id: normalizeRevisionId(parsedId),
 		author: author.length > 0 ? author : "Unknown"
 	};
 	if (date.length > 0) info.date = date;
@@ -551,7 +560,9 @@ function parseRunContents(runElement, rels, media, rootXmlns = {}) {
 			break;
 		}
 		case "rPr": break;
-		case "lastRenderedPageBreak": break;
+		case "lastRenderedPageBreak":
+			contents.push({ type: "renderedPageBreak" });
+			break;
 		case "cr":
 			contents.push({
 				type: "break",

package/dist/docx/sdtProperties.js

@@ -1,5 +1,5 @@
-import { elementToXml, findChild, getAttributeAnyPrefix, getLocalName, parseBooleanElement } from "./xmlParser.js";
 import { SdtLockSchema, narrowEnum } from "./parserEnums.js";
+import { elementToXml, findChild, getAttributeAnyPrefix, getLocalName, parseBooleanElement } from "./xmlParser.js";
 //#region src/docx/sdtProperties.ts
 function parseListItems(el) {
 	const items = [];

package/dist/docx/sectionParser.js

@@ -1,8 +1,21 @@
-import { findChild, findChildren, getAttribute, getChildElements, getLocalName, parseBooleanElement, parseNumericAttribute } from "./xmlParser.js";
 import { BorderStyleSchema, ThemeColorSlotSchema, narrowEnum } from "./parserEnums.js";
+import { findChild, findChildren, getAttribute, getChildElements, getLocalName, parseBooleanElement, parseNumericAttribute } from "./xmlParser.js";
 import { parseFooterReference, parseHeaderReference } from "./headerFooterRefParser.js";
 import { parseEndnoteProperties, parseFootnoteProperties } from "./notePropertiesParser.js";
+import { normalizeRevisionId } from "@stll/docx-core/model";
 //#region src/docx/sectionParser.ts
+/**
+* Sanity cap on `w:cols/@w:num`. Word's column picker tops out well below
+* this; a hostile/corrupt value here would force the layout engine to
+* generate a proportional band per column.
+*/
+const MAX_SECTION_COLUMNS = 45;
+/**
+* Sanity cap on `w:pgSz` `@w:w`/`@w:h`, in twips (~22in). Matches the
+* existing `w:defaultTabStop` cap (`settingsParser.ts`) and bounds the
+* ruler tick generators, which are sized off the page dimensions.
+*/
+const MAX_PAGE_DIMENSION_TWIPS = 31680;
 const serializedSectionPropertyChildNames = /* @__PURE__ */ new Set([
 	"headerReference",
 	"footerReference",
@@ -55,7 +68,7 @@ function parsePropertyChangeInfo(node) {
 	const date = (getAttribute(node, "w", "date") ?? "").trim();
 	const rsid = (getAttribute(node, "w", "rsid") ?? "").trim();
 	const info = {
-		id: Number.isInteger(parsedId) && parsedId >= 0 ? parsedId : 0,
+		id: normalizeRevisionId(parsedId),
 		author: author.length > 0 ? author : "Unknown"
 	};
 	if (date.length > 0) info.date = date;
@@ -170,9 +183,9 @@ function parseSectionProperties(sectPr) {
 	const pgSz = findChild(sectPr, "w", "pgSz");
 	if (pgSz) {
 		const w = parseNumericAttribute(pgSz, "w", "w");
-		if (w !== void 0) props.pageWidth = w;
+		if (w !== void 0) props.pageWidth = Math.min(w, MAX_PAGE_DIMENSION_TWIPS);
 		const h = parseNumericAttribute(pgSz, "w", "h");
-		if (h !== void 0) props.pageHeight = h;
+		if (h !== void 0) props.pageHeight = Math.min(h, MAX_PAGE_DIMENSION_TWIPS);
 		const orientation = parseOrientation(getAttribute(pgSz, "w", "orient"));
 		if (orientation) props.orientation = orientation;
 	}
@@ -196,7 +209,7 @@ function parseSectionProperties(sectPr) {
 	const cols = findChild(sectPr, "w", "cols");
 	if (cols) {
 		const num = parseNumericAttribute(cols, "w", "num");
-		if (num !== void 0) props.columnCount = num;
+		if (num !== void 0) props.columnCount = Math.min(num, MAX_SECTION_COLUMNS);
 		const space = parseNumericAttribute(cols, "w", "space");
 		if (space !== void 0) props.columnSpace = space;
 		const equalWidth = getAttribute(cols, "w", "equalWidth");

package/dist/docx/selectiveSave.js

@@ -1,8 +1,10 @@
+import { parseNumbering } from "./numberingParser.js";
 import { RELATIONSHIP_TYPES } from "./relsParser.js";
 import { parseCommentsExtended } from "./commentParser.js";
-import { parseNumbering } from "./numberingParser.js";
 import { validateFolioDocumentModel } from "./modelValidation.js";
+import { isPreservableDocxEntry } from "./unzip.js";
 import { hasUnsynthesizedReplyRanges } from "./commentReplyMarkers.js";
+import { withoutOrphanCommentRanges } from "./commentRangeIntegrity.js";
 import { buildPatchedDocumentXml, buildPatchedNoteXml, buildPatchedNumberingXml, collectChangedNumberingDefs, collectParaIds } from "./selectiveXmlPatch.js";
 import { ensureThreadedCommentParaIds, serializeComments, serializeCommentsExtended } from "./serializer/commentSerializer.js";
 import { serializeDocument } from "./serializer/documentSerializer.js";
@@ -182,6 +184,7 @@ async function attemptSelectiveSave(doc, originalBuffer, options) {
 	if (hasNewImagesOrHyperlinks(content)) return null;
 	if (hasUnmaterializedHeaderFooter(doc)) return null;
 	if (hasModelDrivenPictureWatermark(doc)) return null;
+	if (withoutOrphanCommentRanges(doc) !== doc) return null;
 	if (hasUnsynthesizedReplyRanges(doc)) return null;
 	if (!validateFolioDocumentModel(doc).valid) return null;
 	const comments = doc.package.document.comments ?? [];
@@ -189,6 +192,7 @@ async function attemptSelectiveSave(doc, originalBuffer, options) {
 	const headerFooterUpdates = collectHeaderFooterUpdates(doc);
 	try {
 		const zip = await (await import("jszip")).default.loadAsync(originalBuffer);
+		for (const [path, file] of Object.entries(zip.files)) if (!file.dir && !isPreservableDocxEntry(path)) return null;
 		const updates = /* @__PURE__ */ new Map();
 		if (changedParaIds.size > 0) {
 			const docXmlFile = zip.file("word/document.xml");

package/dist/docx/selectiveXmlPatch.d.ts

@@ -33,6 +33,27 @@ declare function findParagraphOffsets(xml: string, paraId: string): {
  * from a fully serialized document.xml string.
  */
 declare function extractParagraphXml(serializedXml: string, paraId: string): string | null;
+type ParagraphOffsets = {
+  start: number;
+  end: number;
+};
+/**
+ * Single linear-scan index of every `<w:p>` element's start/end offsets,
+ * keyed by `w14:paraId`. Generalizes {@link findParagraphOffsets} (one
+ * regex-scan-plus-depth-walk per lookup) to build every paragraph's offsets
+ * in a single pass, so a caller that needs many paragraphs from the same XML
+ * (e.g. `collectChangedNoteParaIds` in rezip.ts, walking every note paraId)
+ * does O(1) map lookups afterward instead of re-scanning the whole XML once
+ * per id — O(ids * XML length) collapses to O(XML length).
+ *
+ * A `<w:p>` nested inside another (e.g. inside `mc:AlternateContent`) is
+ * indexed too via a depth stack, matching every element `findParagraphOffsets`
+ * can resolve. A paraId that appears on more than one element is ambiguous —
+ * mirroring {@link findParagraphOffsets}'s single-match requirement — and is
+ * omitted from the index, so a lookup misses exactly where the per-id
+ * function would return null.
+ */
+declare function buildParagraphOffsetIndex(xml: string): Map<string, ParagraphOffsets>;
 /**
  * Count <w:p> elements in an XML string (top-level paragraph count).
  * Counts opening <w:p tags that are NOT self-closing.
@@ -109,4 +130,4 @@ declare function collectChangedNumberingDefs(baselineXml: string, currentXml: st
  */
 declare function buildPatchedNumberingXml(originalXml: string, currentXml: string, changed: ChangedNumberingDefs): string | null;
 //#endregion
-export { ChangedNumberingDefs, PatchSafetyOptions, PatchValidationResult, buildPatchedDocumentXml, buildPatchedNoteXml, buildPatchedNumberingXml, collectChangedNumberingDefs, collectParaIds, countParagraphElements, extractParagraphXml, findParagraphOffsets, isXmlNameBoundary, validatePatchSafety };
+export { ChangedNumberingDefs, ParagraphOffsets, PatchSafetyOptions, PatchValidationResult, buildParagraphOffsetIndex, buildPatchedDocumentXml, buildPatchedNoteXml, buildPatchedNumberingXml, collectChangedNumberingDefs, collectParaIds, countParagraphElements, extractParagraphXml, findParagraphOffsets, isXmlNameBoundary, validatePatchSafety };