@stll/folio-core 0.32.2 → 0.33.0

package/dist/docx/revisionIdNormalization.js

@@ -0,0 +1,103 @@
+import { rewriteStreamingXmlDecimalAttributes } from "./streamingXmlParser.js";
+import { WORDPROCESSINGML_NAMESPACE_URIS, findAttributeByNamespaceUri, getLocalName, getNamespaceUri } from "./xmlParser.js";
+import { XmlResourceLimitError, assertXmlResourceLimits } from "./xmlResourceLimits.js";
+//#region src/docx/revisionIdNormalization.ts
+const REVISION_ELEMENT_NAMES = /* @__PURE__ */ new Set([
+	"cellDel",
+	"cellIns",
+	"cellMerge",
+	"del",
+	"ins",
+	"moveFrom",
+	"moveTo",
+	"numberingChange",
+	"pPrChange",
+	"rPrChange",
+	"sectPrChange",
+	"tblGridChange",
+	"tblPrChange",
+	"tblPrExChange",
+	"tcPrChange",
+	"trPrChange"
+]);
+const REVISION_ELEMENT_CANDIDATE = new RegExp(`<(?:[^\\s<>/:]+:)?(?:${[...REVISION_ELEMENT_NAMES].join("|")})(?:[\\s/>])`, "u");
+const revisionAttribute = (element) => {
+	if (!element.name || !WORDPROCESSINGML_NAMESPACE_URIS.has(getNamespaceUri(element) ?? "") || !REVISION_ELEMENT_NAMES.has(getLocalName(element.name))) return null;
+	const attribute = findAttributeByNamespaceUri(element, WORDPROCESSINGML_NAMESPACE_URIS, "id");
+	if (!attribute) return null;
+	const id = Number(attribute.value);
+	return Number.isSafeInteger(id) && id >= 0 ? {
+		name: attribute.name,
+		id
+	} : null;
+};
+/**
+* Keep physical tracked-change element ids unique across a package.
+*
+* Live editor marks and operation receipts keep their logical revision IDs.
+* Saving assigns fresh IDs only where one logical change was split into
+* multiple physical OOXML wrappers; reopening therefore exposes the physical
+* wrapper IDs that the file format requires.
+*/
+const normalizeRevisionIdsInXmlParts = (parts) => {
+	const candidates = [...parts].filter(([, xml]) => REVISION_ELEMENT_CANDIDATE.test(xml));
+	const occurrencesByPath = /* @__PURE__ */ new Map();
+	const reserved = /* @__PURE__ */ new Set();
+	for (const [path, xml] of candidates) {
+		assertXmlResourceLimits(xml);
+		const ids = [];
+		if (rewriteStreamingXmlDecimalAttributes(xml, (element) => {
+			const attribute = revisionAttribute(element);
+			if (attribute) {
+				ids.push(attribute.id);
+				reserved.add(attribute.id);
+			}
+			return null;
+		}).status === "unsupported") throw new XmlResourceLimitError({
+			message: `Revision-id normalization could not safely scan ${path}`,
+			limit: "syntax"
+		});
+		occurrencesByPath.set(path, ids);
+	}
+	const repeatedPaths = /* @__PURE__ */ new Set();
+	const firstSeen = /* @__PURE__ */ new Set();
+	for (const [path, ids] of occurrencesByPath) for (const id of ids) if (firstSeen.has(id)) repeatedPaths.add(path);
+	else firstSeen.add(id);
+	let nextId = 0;
+	const allocate = () => {
+		while (reserved.has(nextId)) nextId += 1;
+		const allocated = nextId;
+		reserved.add(allocated);
+		nextId += 1;
+		return allocated;
+	};
+	const seen = /* @__PURE__ */ new Set();
+	const normalized = new Map(parts);
+	for (const [path, xml] of candidates) {
+		const ids = occurrencesByPath.get(path);
+		if (!ids) continue;
+		if (!repeatedPaths.has(path) && ids.every((id) => !seen.has(id))) {
+			for (const id of ids) seen.add(id);
+			continue;
+		}
+		const rewritten = rewriteStreamingXmlDecimalAttributes(xml, (element) => {
+			const attribute = revisionAttribute(element);
+			if (!attribute) return null;
+			if (!seen.has(attribute.id)) {
+				seen.add(attribute.id);
+				return null;
+			}
+			const replacement = allocate();
+			seen.add(replacement);
+			return /* @__PURE__ */ new Map([[attribute.name, String(replacement)]]);
+		});
+		if (rewritten.status === "unsupported") throw new XmlResourceLimitError({
+			message: `Revision-id normalization could not safely rewrite ${path}`,
+			limit: "syntax"
+		});
+		normalized.set(path, rewritten.value);
+	}
+	return normalized;
+};
+//#endregion
+export { REVISION_ELEMENT_NAMES, normalizeRevisionIdsInXmlParts };

package/dist/docx/rezip.js

@@ -5,7 +5,9 @@ import { parseEndnotes, parseFootnotes } from "./footnoteParser.js";
 import { assertValidFolioDocumentModel } from "./modelValidation.js";
 import { isNewDataUrlDrawing } from "./newImage.js";
 import { parseNumbering } from "./numberingParser.js";
+import { isUnsafePackagePath, reconcilePackageReferences, removeUnsafeEntries } from "./packageParts.js";
 import { RELATIONSHIP_TYPES, parseRelationships, resolveRelativePath } from "./relsParser.js";
+import { normalizeRevisionIdsInXmlParts } from "./revisionIdNormalization.js";
 import { appendNumberingDefs, buildPatchedNotePartXml, buildPatchedNumberingXml, collectAddedNumberingDefs, collectChangedNoteParaIds, collectChangedNumberingDefs, collectParaIds } from "./selectiveXmlPatch.js";
 import { ensureThreadedCommentParaIds, serializeComments, serializeCommentsExtended } from "./serializer/commentSerializer.js";
 import { serializeDocument } from "./serializer/documentSerializer.js";
@@ -17,7 +19,6 @@ import { serializeSettingsXml } from "./serializer/settingsSerializer.js";
 import { serializeStyle, serializeStylesXml } from "./serializer/stylesSerializer.js";
 import { serializeThemeXml } from "./serializer/themeSerializer.js";
 import { escapeXml } from "./serializer/xmlUtils.js";
-import { isPreservableDocxEntry } from "./unzip.js";
 import { WORDPROCESSINGML_NAMESPACE_URIS, findChild, getAttribute, getChildElements, getLocalName, getNamespaceUri, matchesName, parseXml, parseXmlDocument } from "./xmlParser.js";
 import { assertXmlResourceLimits } from "./xmlResourceLimits.js";
 import { panic } from "better-result";
@@ -405,20 +406,32 @@ async function processNewHyperlinks(parts, zip, compressionLevel) {
 		});
 	}
 }
-const generateDocxZip = (zip, compressionLevel) => zip.generateAsync({
-	type: "arraybuffer",
-	compression: "DEFLATE",
-	compressionOptions: { level: compressionLevel }
-});
-const parsedZipSources = /* @__PURE__ */ new WeakMap();
-const removeNonPreservableEntries = (zip) => {
-	for (const [path, file] of Object.entries(zip.files)) if (!file.dir && !isPreservableDocxEntry(path)) zip.remove(path);
+/**
+* The single exit for a repacked package. Reconciliation runs here rather than
+* at each caller so no save path can emit a package whose relationships or
+* content types name a part it does not hold.
+*/
+const generateDocxZip = async (zip, compressionLevel) => {
+	await reconcilePackageReferences(zip, compressionLevel);
+	const xmlParts = /* @__PURE__ */ new Map();
+	for (const [path, file] of Object.entries(zip.files)) if (!file.dir && path.startsWith("word/") && path.endsWith(".xml")) xmlParts.set(path, await file.async("text"));
+	const normalizedParts = normalizeRevisionIdsInXmlParts(xmlParts);
+	for (const [path, xml] of normalizedParts) if (xml !== xmlParts.get(path)) zip.file(path, xml, {
+		compression: "DEFLATE",
+		compressionOptions: { level: compressionLevel }
+	});
+	return zip.generateAsync({
+		type: "arraybuffer",
+		compression: "DEFLATE",
+		compressionOptions: { level: compressionLevel }
+	});
 };
+const parsedZipSources = /* @__PURE__ */ new WeakMap();
 const loadParsedZipSource = async (document, buffer) => {
 	const cached = parsedZipSources.get(document);
 	if (cached?.buffer === buffer) return cached;
 	const zip = await JSZip.loadAsync(buffer);
-	removeNonPreservableEntries(zip);
+	removeUnsafeEntries(zip);
 	const [documentXml, corePropertiesXml] = await Promise.all([zip.file("word/document.xml")?.async("text"), zip.file("docProps/core.xml")?.async("text")]);
 	const source = {
 		buffer,
@@ -486,7 +499,7 @@ async function repackDocx(doc, options = {}) {
 	const exportDocument = withoutOrphanCommentRanges(doc);
 	const originalZip = await JSZip.loadAsync(doc.originalBuffer);
 	const [originalDocumentXml, originalCorePropertiesXml] = await Promise.all([originalZip.file("word/document.xml")?.async("text"), originalZip.file("docProps/core.xml")?.async("text")]);
-	removeNonPreservableEntries(originalZip);
+	removeUnsafeEntries(originalZip);
 	const newZip = cloneDocxZip(originalZip);
 	return finishRepack({
 		document: exportDocument,
@@ -517,7 +530,7 @@ async function repackDocxFromRaw(doc, rawContent, options = {}) {
 			newZip.folder(path.replace(/\/$/u, ""));
 			continue;
 		}
-		if (!isPreservableDocxEntry(path)) continue;
+		if (isUnsafePackagePath(path)) continue;
 		const content = await file.async("arraybuffer");
 		newZip.file(path, content, {
 			compression: "DEFLATE",
@@ -559,11 +572,7 @@ async function repackDocxFromRaw(doc, rawContent, options = {}) {
 			compressionOptions: { level: compressionLevel }
 		});
 	}
-	return await newZip.generateAsync({
-		type: "arraybuffer",
-		compression: "DEFLATE",
-		compressionOptions: { level: compressionLevel }
-	});
+	return generateDocxZip(newZip, compressionLevel);
 }
 const COMMENTS_CONTENT_TYPE = "application/vnd.openxmlformats-officedocument.wordprocessingml.comments+xml";
 const COMMENTS_EXTENDED_CONTENT_TYPE = "application/vnd.openxmlformats-officedocument.wordprocessingml.commentsExtended+xml";

package/dist/docx/selectiveSave.js

@@ -4,6 +4,7 @@ import { hasUnsynthesizedReplyRanges } from "./commentReplyMarkers.js";
 import { validateFolioDocumentModel } from "./modelValidation.js";
 import { isNewDataUrlDrawing } from "./newImage.js";
 import { parseNumbering } from "./numberingParser.js";
+import { isUnsafePackagePath } from "./packageParts.js";
 import { RELATIONSHIP_TYPES } from "./relsParser.js";
 import { COMMENTS_CONTENT_TYPE, COMMENTS_EXTENDED_PART_LOWER, addCommentsExtendedOverride, addCommentsExtendedRelationship, applyUpdatesToZip, collectHeaderFooterUpdates, findMaxRId, hasModelDrivenPictureWatermark, hasUnmaterializedHeaderFooter, updateCoreProperties, withoutAttachedTemplate } from "./rezip.js";
 import "./selectiveSaveFlags.js";
@@ -12,7 +13,6 @@ import { ensureThreadedCommentParaIds, serializeComments, serializeCommentsExten
 import { serializeDocument } from "./serializer/documentSerializer.js";
 import { serializeEndnotes, serializeFootnotes } from "./serializer/noteSerializer.js";
 import { serializeNumberingXml } from "./serializer/numberingSerializer.js";
-import { isPreservableDocxEntry } from "./unzip.js";
 //#region src/docx/selectiveSave.ts
 /**
 * Check if document content has new images (data: URL without rId) or
@@ -212,7 +212,7 @@ 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;
+		for (const [path, file] of Object.entries(zip.files)) if (!file.dir && isUnsafePackagePath(path)) return null;
 		const updates = /* @__PURE__ */ new Map();
 		if (changedParaIds.size > 0) {
 			const docXmlFile = zip.file("word/document.xml");

package/dist/docx/serializer/commentSerializer.js

@@ -36,7 +36,12 @@ function serializeComment(comment) {
 	xml += "</w:comment>";
 	return xml;
 }
-const COMMENTS_HEADER = "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"yes\"?>\n<w:comments xmlns:wpc=\"http://schemas.microsoft.com/office/word/2010/wordprocessingCanvas\" xmlns:mc=\"http://schemas.openxmlformats.org/markup-compatibility/2006\" xmlns:o=\"urn:schemas-microsoft-com:office:office\" xmlns:r=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships\" xmlns:m=\"http://schemas.openxmlformats.org/officeDocument/2006/math\" xmlns:v=\"urn:schemas-microsoft-com:vml\" xmlns:wp=\"http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing\" xmlns:w10=\"urn:schemas-microsoft-com:office:word\" xmlns:w=\"http://schemas.openxmlformats.org/wordprocessingml/2006/main\" xmlns:w14=\"http://schemas.microsoft.com/office/word/2010/wordml\" xmlns:wpg=\"http://schemas.microsoft.com/office/word/2010/wordprocessingGroup\" xmlns:wpi=\"http://schemas.microsoft.com/office/word/2010/wordprocessingInk\" xmlns:wne=\"http://schemas.microsoft.com/office/word/2006/wordml\" xmlns:wps=\"http://schemas.microsoft.com/office/word/2010/wordprocessingShape\" mc:Ignorable=\"w14 wp14\">";
+const COMMENT_EXTENSION_NAMESPACES = {
+	w14: "http://schemas.microsoft.com/office/word/2010/wordml",
+	wp14: "http://schemas.microsoft.com/office/word/2010/wordprocessingDrawing"
+};
+const COMMENTS_HEADER = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<w:comments xmlns:wpc="http://schemas.microsoft.com/office/word/2010/wordprocessingCanvas" xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships" xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math" xmlns:v="urn:schemas-microsoft-com:vml" xmlns:wp="http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing" xmlns:w10="urn:schemas-microsoft-com:office:word" xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main" ${Object.entries(COMMENT_EXTENSION_NAMESPACES).map(([prefix, namespace]) => `xmlns:${prefix}="${namespace}"`).join(" ")} xmlns:wpg="http://schemas.microsoft.com/office/word/2010/wordprocessingGroup" xmlns:wpi="http://schemas.microsoft.com/office/word/2010/wordprocessingInk" xmlns:wne="http://schemas.microsoft.com/office/word/2006/wordml" xmlns:wps="http://schemas.microsoft.com/office/word/2010/wordprocessingShape" mc:Ignorable="${Object.keys(COMMENT_EXTENSION_NAMESPACES).join(" ")}">`;
 /**
 * Serialize comments array to comments.xml content. Returns a valid empty
 * `<w:comments/>` document for an empty array so callers can overwrite an

package/dist/docx/serializer/paragraphSerializer.js

@@ -368,6 +368,14 @@ function serializeMoveRangeStart(tag, marker) {
 function rewriteRunTextAsDeleted(xml) {
 	return xml.replace(/<w:t\b/gu, "<w:delText").replace(/<\/w:t>/gu, "</w:delText>").replace(/<w:instrText\b/gu, "<w:delInstrText").replace(/<\/w:instrText>/gu, "</w:delInstrText>");
 }
+function trackedChangeTag(change) {
+	switch (change.type) {
+		case "insertion": return "ins";
+		case "deletion": return "del";
+		case "moveFrom": return "moveFrom";
+		case "moveTo": return "moveTo";
+	}
+}
 function serializeTrackedChange(tag, change) {
 	const info = change.info;
 	const normalizedId = normalizeRevisionId(info.id);
@@ -394,7 +402,13 @@ function serializeTrackedChange(tag, change) {
 			if (tag === "del" || tag === "moveFrom") return serializeDeletedRun(item);
 			return serializeRun(item);
 		}
-		return serializeHyperlink(item);
+		if (item.type === "hyperlink") return serializeHyperlink(item);
+		if (item.type === "simpleField" || item.type === "complexField") {
+			const xml = item.type === "simpleField" ? serializeSimpleField(item) : serializeComplexField(item);
+			return tag === "del" || tag === "moveFrom" ? rewriteRunTextAsDeleted(xml) : xml;
+		}
+		if (item.type === "insertion" || item.type === "deletion" || item.type === "moveFrom" || item.type === "moveTo") return serializeTrackedChange(trackedChangeTag(item), item);
+		return item.type === "bookmarkStart" ? serializeBookmarkStart(item) : serializeBookmarkEnd(item);
 	}).join("");
 	return `<w:${tag} ${attrs.join(" ")}>${contentXml}</w:${tag}>`;
 }
@@ -470,12 +484,27 @@ function serializeParagraph(paragraph) {
 	return `<w:p${attrsStr}>${parts.join("")}</w:p>`;
 }
 function injectRenderedPageBreakIntoFirstRun(xml) {
-	const runOpeningTag = /<w:r(?=[\s>/])[^>]*>/u;
-	const openingTag = runOpeningTag.exec(xml);
+	const openingTag = /<w:r(?=[\s>/])[^>]*>/u.exec(xml);
 	if (!openingTag) return null;
 	const runEnd = xml.indexOf("</w:r>", openingTag.index + openingTag[0].length);
 	if (runEnd !== -1 && xml.slice(openingTag.index + openingTag[0].length, runEnd).includes("<w:lastRenderedPageBreak/>")) return xml;
-	return xml.replace(runOpeningTag, (match) => `${match}<w:lastRenderedPageBreak/>`);
+	const contentStart = openingTag.index + openingTag[0].length;
+	let insertionOffset = contentStart;
+	if (xml.startsWith("<w:rPr", contentStart)) {
+		const propertyTag = /<\/?w:rPr(?=[\s>/])[^>]*>/gu;
+		propertyTag.lastIndex = contentStart;
+		let depth = 0;
+		for (const match of xml.matchAll(propertyTag)) {
+			if (match.index !== contentStart && depth === 0) break;
+			if (match[0].startsWith("</")) depth--;
+			else if (!match[0].endsWith("/>")) depth++;
+			if (depth === 0) {
+				insertionOffset = match.index + match[0].length;
+				break;
+			}
+		}
+	}
+	return `${xml.slice(0, insertionOffset)}<w:lastRenderedPageBreak/>${xml.slice(insertionOffset)}`;
 }
 /**
 * Serialize multiple paragraphs to OOXML XML

package/dist/docx/serializer/runSerializer.js

@@ -92,13 +92,6 @@ function serializeTextFormatting(formatting) {
 		if (formatting.fontFamily.csTheme) fontAttrs.push(`w:cstheme="${escapeXml(formatting.fontFamily.csTheme)}"`);
 		if (fontAttrs.length > 0) parts.push(`<w:rFonts ${fontAttrs.join(" ")}/>`);
 	}
-	if (formatting.language) {
-		const languageAttrs = [];
-		if (formatting.language.val) languageAttrs.push(`w:val="${escapeXml(formatting.language.val)}"`);
-		if (formatting.language.eastAsia) languageAttrs.push(`w:eastAsia="${escapeXml(formatting.language.eastAsia)}"`);
-		if (formatting.language.bidi) languageAttrs.push(`w:bidi="${escapeXml(formatting.language.bidi)}"`);
-		if (languageAttrs.length > 0) parts.push(`<w:lang ${languageAttrs.join(" ")}/>`);
-	}
 	if (formatting.bold === true) parts.push("<w:b/>");
 	else if (formatting.bold === false) parts.push("<w:b w:val=\"0\"/>");
 	if (formatting.boldCs === true) parts.push("<w:bCs/>");
@@ -134,11 +127,12 @@ function serializeTextFormatting(formatting) {
 	if (formatting.position !== void 0) parts.push(`<w:position w:val="${intAttr(formatting.position)}"/>`);
 	if (formatting.fontSize !== void 0) parts.push(`<w:sz w:val="${intAttr(formatting.fontSize)}"/>`);
 	if (formatting.fontSizeCs !== void 0) parts.push(`<w:szCs w:val="${intAttr(formatting.fontSizeCs)}"/>`);
+	let customHighlightShadingXml = "";
 	if (formatting.highlight) {
 		if (VALID_HIGHLIGHT_COLORS.has(formatting.highlight)) parts.push(`<w:highlight w:val="${formatting.highlight}"/>`);
 		else if (!formatting.shading) {
 			const hex = formatting.highlight.replace(/^#/u, "");
-			if (/^[0-9a-fA-F]{6}$/u.test(hex)) parts.push(`<w:shd w:val="clear" w:color="auto" w:fill="${hex}"/>`);
+			if (/^[0-9a-fA-F]{6}$/u.test(hex)) customHighlightShadingXml = `<w:shd w:val="clear" w:color="auto" w:fill="${hex}"/>`;
 		}
 	}
 	if (formatting.underline) {
@@ -152,14 +146,21 @@ function serializeTextFormatting(formatting) {
 		parts.push(`<w:u ${uAttrs.join(" ")}/>`);
 	}
 	if (formatting.effect && formatting.effect !== "none") parts.push(`<w:effect w:val="${formatting.effect}"/>`);
-	if (formatting.emphasisMark) parts.push(`<w:em w:val="${formatting.emphasisMark}"/>`);
-	const shadingXml = serializeShading(formatting.shading);
+	const shadingXml = serializeShading(formatting.shading) || customHighlightShadingXml;
 	if (shadingXml) parts.push(shadingXml);
 	if (formatting.vertAlign) parts.push(`<w:vertAlign w:val="${formatting.vertAlign}"/>`);
 	if (formatting.rtl === true) parts.push("<w:rtl/>");
 	else if (formatting.rtl === false) parts.push("<w:rtl w:val=\"0\"/>");
 	if (formatting.cs === true) parts.push("<w:cs/>");
 	else if (formatting.cs === false) parts.push("<w:cs w:val=\"0\"/>");
+	if (formatting.emphasisMark) parts.push(`<w:em w:val="${formatting.emphasisMark}"/>`);
+	if (formatting.language) {
+		const languageAttrs = [];
+		if (formatting.language.val) languageAttrs.push(`w:val="${escapeXml(formatting.language.val)}"`);
+		if (formatting.language.eastAsia) languageAttrs.push(`w:eastAsia="${escapeXml(formatting.language.eastAsia)}"`);
+		if (formatting.language.bidi) languageAttrs.push(`w:bidi="${escapeXml(formatting.language.bidi)}"`);
+		if (languageAttrs.length > 0) parts.push(`<w:lang ${languageAttrs.join(" ")}/>`);
+	}
 	if (parts.length === 0) return "";
 	return `<w:rPr>${parts.join("")}</w:rPr>`;
 }

package/dist/docx/serializer/tableSerializer.js

@@ -250,7 +250,8 @@ function serializeTableCellFormatting(formatting, propertyChanges, structuralCha
 		if (formatting.textDirection) parts.push(`<w:textDirection w:val="${formatting.textDirection}"/>`);
 		if (formatting.fitText) parts.push("<w:tcFitText/>");
 		if (formatting.verticalAlign) parts.push(`<w:vAlign w:val="${formatting.verticalAlign}"/>`);
-		if (formatting.hideMark !== void 0) parts.push(formatting.hideMark ? "<w:hideMark/>" : "<w:hideMark w:val=\"false\"/>");
+		if (formatting.hideMark === true) parts.push("<w:hideMark/>");
+		else if (formatting.hideMark === false) parts.push("<w:hideMark w:val=\"off\"/>");
 	}
 	if (structuralChange) {
 		if (structuralChange.type === "tableCellInsertion") parts.push(`<w:cellIns ${serializeTrackedChangeAttributes(structuralChange.info)}/>`);

package/dist/docx/server/applyDocxXmlPatchProposal.js

@@ -2,8 +2,8 @@ import { loadDocxArchive } from "./boundedArchive.js";
 import { FOLIO_DOCX_XML_PATCH_PROPOSAL_PROFILE, InvalidFolioDocxXmlPatchProposalError, evaluateDocxXmlPatchProposal, parseFolioDocxXmlPatchProposal } from "./evaluateDocxXmlPatchProposal.js";
 import { FOLIO_DOCX_CONFORMANCE_PROFILE, validateDocxConformance } from "./validateDocxConformance.js";
 import { TaggedError, panic } from "better-result";
-import JSZip from "jszip";
 import { createHash } from "node:crypto";
+import JSZip from "jszip";
 //#region src/docx/server/applyDocxXmlPatchProposal.ts
 const FOLIO_DOCX_XML_PATCH_APPLICATION_VERSION = 1;
 const FOLIO_DOCX_XML_PATCH_APPLICATION_PROFILE = "folio-xml-patch-application-v1";

package/dist/docx/server/createBilingualDocx.js

@@ -1,4 +1,5 @@
 import { FolioDocxReviewer } from "../../ai-edits/headless.js";
+import { isFolioAIContentBlock } from "../../ai-edits/snapshot.js";
 import { toArrayBuffer } from "../../utils/docxInput.js";
 import { ensureParaIds } from "../ensureParaIds.js";
 import { parseDocx } from "../parser.js";
@@ -14,7 +15,7 @@ import { createBilingualDocument, readBilingualDocument } from "./createBilingua
 */
 async function createBilingualDocx(input, options) {
 	const stampedBuffer = await toArrayBuffer((await ensureParaIds(input)).docx);
-	const editableParagraphIds = new Set((await FolioDocxReviewer.fromBuffer(stampedBuffer)).snapshot().blocks.map(({ id }) => id));
+	const editableParagraphIds = new Set((await FolioDocxReviewer.fromBuffer(stampedBuffer)).snapshot().blocks.filter(isFolioAIContentBlock).map(({ id }) => id));
 	const { document, warnings } = createBilingualDocument(await parseDocx(stampedBuffer, { preloadFonts: false }), {
 		...options,
 		editableParagraphIds
@@ -29,7 +30,7 @@ async function createBilingualDocx(input, options) {
 /** Bytes-in form of {@link readBilingualDocument}. */
 async function readBilingualDocx(input) {
 	const buffer = await toArrayBuffer(input);
-	return readBilingualDocument(await parseDocx(buffer, { preloadFonts: false }), new Set((await FolioDocxReviewer.fromBuffer(buffer)).snapshot().blocks.map(({ id }) => id)));
+	return readBilingualDocument(await parseDocx(buffer, { preloadFonts: false }), new Set((await FolioDocxReviewer.fromBuffer(buffer)).snapshot().blocks.filter(isFolioAIContentBlock).map(({ id }) => id)));
 }
 //#endregion
 export { createBilingualDocx, readBilingualDocx };

package/dist/docx/streamingXmlParser.d.ts

@@ -11,6 +11,18 @@ type ParseXmlResult = {
  * pass. Unsupported or malformed constructs return a sentinel so callers can
  * retain the general-purpose parser as a compatibility fallback.
  */
+type AttributeValueSpan = {
+  start: number;
+  end: number;
+};
+type OpenTagVisitor = (element: XmlElement, attributeValueSpans: ReadonlyMap<string, AttributeValueSpan>) => ReadonlyMap<string, string> | null;
 declare const parseStreamingXml: (xml: string) => ParseXmlResult;
+/** Rewrite selected decimal attribute values while preserving every other source byte. */
+declare const rewriteStreamingXmlDecimalAttributes: (xml: string, visitOpenTag: OpenTagVisitor) => {
+  status: "rewritten";
+  value: string;
+} | {
+  status: "unsupported";
+};
 //#endregion
-export { parseStreamingXml };
+export { parseStreamingXml, rewriteStreamingXmlDecimalAttributes };

package/dist/docx/streamingXmlParser.js

@@ -1,3 +1,4 @@
+import { attachXmlNamespaceContext } from "./xmlParser.js";
 import { FOLIO_XML_RESOURCE_LIMITS } from "./xmlResourceLimits.js";
 //#region src/docx/streamingXmlParser.ts
 const BUILT_IN_ENTITIES = {
@@ -7,14 +8,10 @@ const BUILT_IN_ENTITIES = {
 	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 parseStreamingXmlInternal = (xml, visitOpenTag) => {
 	const root = { elements: [] };
 	const stack = [];
+	const replacements = [];
 	let cursor = 0;
 	let mergeAdjacentText = false;
 	while (cursor < xml.length) {
@@ -55,9 +52,19 @@ const parseStreamingXml = (xml) => {
 			mergeAdjacentText = false;
 			continue;
 		}
-		const parsedTag = parseOpenTag(xml, open + 1, close);
+		const parsedTag = parseOpenTag(xml, open + 1, close, visitOpenTag !== void 0);
 		if (parsedTag.status === "unsupported") return parsedTag;
 		const parent = stack.at(-1)?.element ?? root;
+		attachXmlNamespaceContext(parsedTag.element, parent.namespaceScope);
+		const rewritten = visitOpenTag?.(parsedTag.element, parsedTag.attributeValueSpans ?? /* @__PURE__ */ new Map());
+		if (rewritten) for (const [attributeName, value] of rewritten) {
+			const span = parsedTag.attributeValueSpans?.get(attributeName);
+			if (!span) return { status: "unsupported" };
+			replacements.push({
+				...span,
+				value
+			});
+		}
 		appendElement(parent, parsedTag.element);
 		if (!parsedTag.selfClosing) {
 			if (stack.length >= FOLIO_XML_RESOURCE_LIMITS.maxDepth) return { status: "unsupported" };
@@ -72,10 +79,35 @@ const parseStreamingXml = (xml) => {
 	if (stack.length > 0) return { status: "unsupported" };
 	return {
 		status: "parsed",
-		value: root
+		value: root,
+		replacements
+	};
+};
+const parseStreamingXml = (xml) => {
+	const parsed = parseStreamingXmlInternal(xml);
+	return parsed.status === "parsed" ? {
+		status: "parsed",
+		value: parsed.value
+	} : { status: "unsupported" };
+};
+/** Rewrite selected decimal attribute values while preserving every other source byte. */
+const rewriteStreamingXmlDecimalAttributes = (xml, visitOpenTag) => {
+	const parsed = parseStreamingXmlInternal(xml, visitOpenTag);
+	if (parsed.status === "unsupported") return parsed;
+	const chunks = [];
+	let cursor = 0;
+	for (const replacement of parsed.replacements.toSorted((left, right) => left.start - right.start)) {
+		if (!/^\d+$/u.test(replacement.value)) return { status: "unsupported" };
+		chunks.push(xml.slice(cursor, replacement.start), replacement.value);
+		cursor = replacement.end;
+	}
+	chunks.push(xml.slice(cursor));
+	return {
+		status: "rewritten",
+		value: chunks.join("")
 	};
 };
-const parseOpenTag = (xml, start, close) => {
+const parseOpenTag = (xml, start, close, captureAttributeSpans) => {
 	let cursor = skipWhitespace(xml, start, close);
 	const nameStart = cursor;
 	cursor = scanName(xml, cursor, close);
@@ -83,6 +115,7 @@ const parseOpenTag = (xml, start, close) => {
 	const name = xml.slice(nameStart, cursor);
 	if (isUnsafePropertyName(name)) return { status: "unsupported" };
 	let attributes;
+	const attributeValueSpans = captureAttributeSpans ? /* @__PURE__ */ new Map() : void 0;
 	let selfClosing = false;
 	while (cursor < close) {
 		cursor = skipWhitespace(xml, cursor, close);
@@ -110,6 +143,10 @@ const parseOpenTag = (xml, start, close) => {
 		if (decoded === null) return { status: "unsupported" };
 		attributes ??= {};
 		attributes[attributeName] = decoded;
+		attributeValueSpans?.set(attributeName, {
+			start: valueStart,
+			end: cursor
+		});
 		cursor += 1;
 	}
 	const element = {
@@ -121,7 +158,8 @@ const parseOpenTag = (xml, start, close) => {
 		status: "parsed",
 		element,
 		name,
-		selfClosing
+		selfClosing,
+		attributeValueSpans
 	};
 };
 const appendElement = (parent, child) => {
@@ -246,4 +284,4 @@ const isUnsafePropertyName = (name) => {
 	}
 };
 //#endregion
-export { parseStreamingXml };
+export { parseStreamingXml, rewriteStreamingXmlDecimalAttributes };

package/dist/docx/unzip.d.ts

@@ -59,7 +59,6 @@ type RawDocxContent = {
  * @returns Promise resolving to extracted content
  */
 declare function unzipDocx(buffer: ArrayBuffer, options?: DocxUnzipOptions): Promise<RawDocxContent>;
-declare function isPreservableDocxEntry(path: string): boolean;
 /**
  * Get a list of all files in the DOCX
  *
@@ -120,4 +119,4 @@ declare function getContentSummary(content: RawDocxContent): {
   totalFiles: number;
 };
 //#endregion
-export { DocxSecurityError, DocxUnzipLimits, DocxUnzipOptions, RawDocxContent, extractFile, getContentSummary, getFileList, getMediaMimeType, hasFile, isPreservableDocxEntry, mediaToDataUrl, unzipDocx };
+export { DocxSecurityError, DocxUnzipLimits, DocxUnzipOptions, RawDocxContent, extractFile, getContentSummary, getFileList, getMediaMimeType, hasFile, mediaToDataUrl, unzipDocx };

package/dist/docx/unzip.js

@@ -137,13 +137,13 @@ async function unzipDocx(buffer, options = {}) {
 	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;
-		const lowerPath = path.toLowerCase();
 		const declaredSize = getEntryUncompressedSize(file);
 		if (declaredSize !== null) {
 			totalUncompressedBytes += declaredSize;
 			if (totalUncompressedBytes > limits.maxTotalUncompressedBytes) throw new DocxSecurityError("DOCX file expands beyond the maximum allowed size");
 		}
+		if (!isParsedDocxEntry(path)) continue;
+		const lowerPath = path.toLowerCase();
 		if (lowerPath.endsWith(".xml") || lowerPath.endsWith(".rels")) {
 			assertEntrySize(path, declaredSize, limits.maxXmlBytes);
 			if (options.extractAllXml === false && !shouldExtractXmlPart(lowerPath)) continue;
@@ -328,9 +328,28 @@ function isSafeDocxPath(path) {
 	if (!path || path.startsWith("/") || path.includes("\\")) return false;
 	return !path.split("/").some((part) => part === "..");
 }
-function isPreservableDocxEntry(path) {
+/**
+* Parts folio never looks inside. A macro project and an ActiveX control carry
+* code; folio models neither, so it reads neither — not into the document
+* model, not even into the raw XML map. The save path carries them anyway,
+* straight from the source archive, which is what keeps a `.docm` a `.docm`.
+*/
+const UNINTERPRETED_PART_NAMES = /* @__PURE__ */ new Set(["word/vbaproject.bin", "word/vbadata.xml"]);
+const isUninterpretedPart = (lowerPath) => UNINTERPRETED_PART_NAMES.has(lowerPath) || lowerPath.startsWith("word/activex/");
+/**
+* Whether this entry is one the PARSER reads into the document model. It says
+* nothing about what a save keeps: a repack hands back every part of the
+* package (see `packageParts.ts`), including the ones this returns false for.
+*
+* That asymmetry is the point. A macro project, an ActiveX control and an OLE
+* embedding are opaque bytes to folio — never decoded, never interpreted,
+* never executed — and they travel from the source archive to the saved one
+* without any code here looking inside them.
+*/
+function isParsedDocxEntry(path) {
 	if (!isSafeDocxPath(path)) return false;
 	const lowerPath = path.toLowerCase();
+	if (isUninterpretedPart(lowerPath)) return false;
 	if (lowerPath.startsWith("word/media/")) return PRESERVABLE_MEDIA_MIME_TYPES.has(getMediaMimeType(path));
 	if (lowerPath.startsWith("docprops/thumbnail.")) return PRESERVABLE_MEDIA_MIME_TYPES.has(getMediaMimeType(path));
 	if (lowerPath.startsWith("word/fonts/")) return true;
@@ -374,7 +393,7 @@ function isMediaContentAllowed(data, mimeType) {
 */
 function getFileList(content) {
 	const files = [];
-	for (const path of Object.keys(content.originalZip.files)) if (!content.originalZip.files[path]?.dir && isPreservableDocxEntry(path)) files.push(path);
+	for (const path of Object.keys(content.originalZip.files)) if (!content.originalZip.files[path]?.dir && isParsedDocxEntry(path)) files.push(path);
 	return files.toSorted();
 }
 /**
@@ -425,7 +444,7 @@ function mediaToDataUrl(data, mimeType) {
 */
 function extractFile(content, path) {
 	const file = content.originalZip.file(path);
-	if (!file || !isPreservableDocxEntry(path)) return Promise.resolve(null);
+	if (!file || !isParsedDocxEntry(path)) return Promise.resolve(null);
 	const lowerPath = path.toLowerCase();
 	if (lowerPath.endsWith(".xml") || lowerPath.endsWith(".rels")) return file.async("text");
 	return file.async("arraybuffer");
@@ -464,4 +483,4 @@ function getContentSummary(content) {
 	};
 }
 //#endregion
-export { DocxSecurityError, extractFile, getContentSummary, getFileList, getMediaMimeType, hasFile, isPreservableDocxEntry, mediaToDataUrl, unzipDocx };
+export { DocxSecurityError, extractFile, getContentSummary, getFileList, getMediaMimeType, hasFile, mediaToDataUrl, unzipDocx };

package/dist/docx/xmlParser.d.ts

@@ -47,6 +47,8 @@ type XmlNamespaceScope = {
   bindings: ReadonlyMap<string, string>;
   parent?: XmlNamespaceScope;
 };
+/** Attach the element's resolved namespace metadata from its in-scope declarations. */
+declare const attachXmlNamespaceContext: (element: XmlElement, inheritedNamespaceScope?: XmlNamespaceScope) => XmlNamespaceScope;
 /**
  * Common OOXML namespace URIs — re-exported from @stll/docx-utils.
  */
@@ -344,4 +346,4 @@ declare function mergeXmlnsDeclarations(inherited: Record<string, string>, eleme
  */
 declare function cloneWithXmlnsDeclarations(element: XmlElement, xmlnsDecls: Record<string, string>): XmlElement;
 //#endregion
-export { NAMESPACES, WORDPROCESSINGML_NAMESPACE_URIS, XmlAttributeMatch, XmlElement, XmlNamespaceScope, cloneWithXmlnsDeclarations, collectXmlnsDeclarations, elementToXml, findAllDeep, findAttributeByNamespaceUri, findByFullName, findChild, findChildByLocalName, findChildByNamespaceUri, findChildren, findChildrenByLocalName, findDeep, getAttribute, getAttributeAny, getAttributeAnyPrefix, getAttributeByNamespaceUri, getAttributes, getChildElements, getLocalName, getNamespacePrefix, getNamespaceUri, getTextContent, hasChild, hasFlag, matchesName, mergeXmlnsDeclarations, parseBooleanElement, parseColorElement, parseNumberingLevelAttribute, parseNumericAttribute, parseOnOffValue, parseTableMeasurementValue, parseXml, parseXmlDocument };
+export { NAMESPACES, WORDPROCESSINGML_NAMESPACE_URIS, XmlAttributeMatch, XmlElement, XmlNamespaceScope, attachXmlNamespaceContext, cloneWithXmlnsDeclarations, collectXmlnsDeclarations, elementToXml, findAllDeep, findAttributeByNamespaceUri, findByFullName, findChild, findChildByLocalName, findChildByNamespaceUri, findChildren, findChildrenByLocalName, findDeep, getAttribute, getAttributeAny, getAttributeAnyPrefix, getAttributeByNamespaceUri, getAttributes, getChildElements, getLocalName, getNamespacePrefix, getNamespaceUri, getTextContent, hasChild, hasFlag, matchesName, mergeXmlnsDeclarations, parseBooleanElement, parseColorElement, parseNumberingLevelAttribute, parseNumericAttribute, parseOnOffValue, parseTableMeasurementValue, parseXml, parseXmlDocument };