Documents: read long .docx files about a third faster when opening

src/documents/src/engine/diagram-data-model.js

@@ -1,10 +1,11 @@
 import { UnsupportedDocumentError } from './errors.js';
+import { elementChildren } from './docx-xml.js';
 
 const DGM = 'http://schemas.openxmlformats.org/drawingml/2006/diagram';
 const pointTypes = new Set(['doc', 'node', 'asst', 'parTrans', 'sibTrans', 'pres']);
 const connectionTypes = new Set(['parOf', 'presOf', 'presParOf', 'unknownRelationship']);
 const fail = reason => { throw new UnsupportedDocumentError(`diagram data with ${reason}`); };
-const children = node => [...(node?.children || [])];
+const children = node => node?.children ? elementChildren(node) : [];
 function list(root, name, optional = false) {
   const matches = children(root).filter(n => n.namespaceURI === DGM && n.localName === name);
   if (optional && !matches.length) return [];

src/documents/src/engine/docx-backgrounds.js

@@ -1,9 +1,10 @@
 import { checkedHighlight } from './backgrounds.js';
 import { UnsupportedDocumentError } from './errors.js';
+import { childElements } from './docx-xml.js';
 const W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 const fail = why => { throw new UnsupportedDocumentError(why); };
 function single(parent, tag) {
-  const nodes = [...(parent?.children || [])].filter(n => n.namespaceURI === W && n.localName === tag);
+  const nodes = childElements(parent, W, tag);
   if (nodes.length > 1) fail(`duplicate run ${tag}`);
   if (nodes[0]?.children.length) fail(`nested run ${tag}`);
   return nodes[0];

src/documents/src/engine/docx-cell-margins.js

@@ -1,8 +1,9 @@
 import { UnsupportedDocumentError } from './errors.js';
 import { checkedCellMargins } from './cell-margins.js';
+import { elementChildren } from './docx-xml.js';
 const W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 const ORDER = ['top', 'start', 'left', 'bottom', 'end', 'right'];
-const children = node => [...(node?.children || [])];
+const children = node => node?.children ? elementChildren(node) : [];
 const fail = reason => { throw new UnsupportedDocumentError('cell margins with ' + reason); };
 
 export function readCellMargins(parent, table = false) {

src/documents/src/engine/docx-chart-text.js

@@ -1,9 +1,10 @@
 import { UnsupportedDocumentError } from './errors.js';
 import { drawingColor } from './docx-drawing-style.js';
 import { checkedFontName } from './font-families.js';
+import { elementChildren } from './docx-xml.js';
 
 const A = 'http://schemas.openxmlformats.org/drawingml/2006/main';
-const children = n => [...(n?.children || [])];
+const children = n => n?.children ? elementChildren(n) : [];
 const child = (n, name) => children(n).find(c => c.namespaceURI === A && c.localName === name);
 const fail = why => { throw new UnsupportedDocumentError(`chart text with ${why}`); };
 function check(n, names) {

src/documents/src/engine/docx-colors.js

@@ -1,17 +1,18 @@
-import { parseXml } from './docx-xml.js';
+import { childElements, elementChildren, parseXml } from './docx-xml.js';
+import { relationshipEntries } from './docx-relationships.js';
 import { UnsupportedDocumentError } from './errors.js';
 import { normalizeColor, themeColorTransform } from './colors.js';
 
 const W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 const A = 'http://schemas.openxmlformats.org/drawingml/2006/main';
 const REL = 'http://schemas.openxmlformats.org/package/2006/relationships';
 const R = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/';
-const children = n => [...(n?.children || [])];
+const children = n => n?.children ? elementChildren(n) : [];
 const fail = reason => { throw new UnsupportedDocumentError(`text colors with ${reason}`); };
 const names = { dark1: 'dk1', light1: 'lt1', dark2: 'dk2', light2: 'lt2', hyperlink: 'hlink', followedHyperlink: 'folHlink', ...Object.fromEntries(Array.from({ length: 6 }, (_, i) => [`accent${i + 1}`, `accent${i + 1}`])) };
 const aliases = { background1: ['bg1', 'light1'], background2: ['bg2', 'light2'], text1: ['t1', 'dark1'], text2: ['t2', 'dark2'] };
 function only(parent, ns, name) {
-  const nodes = children(parent).filter(n => n.namespaceURI === ns && n.localName === name);
+  const nodes = childElements(parent, ns, name);
   if (nodes.length > 1) fail(`duplicate ${name}`);
   return nodes[0];
 }
@@ -25,16 +26,16 @@ export function colorReader(parts, settings) {
   let theme, mapping, themeLinks;
   const themeRelationships = () => {
     if (!themeLinks) {
-      const rels = parts['word/_rels/document.xml.rels'] && parseXml(parts['word/_rels/document.xml.rels']);
-      themeLinks = children(rels?.documentElement).filter(n => n.namespaceURI === REL && n.getAttribute('Type') === R + 'theme');
+      const rels = parts['word/_rels/document.xml.rels'] && relationshipEntries(parts['word/_rels/document.xml.rels']);
+      themeLinks = (rels || []).filter(n => n.namespaceURI === REL && n.attribute('Type') === R + 'theme');
     }
     return themeLinks;
   };
   const themeBase = name => {
     if (!theme) {
       const links = themeRelationships();
-      if (links.length !== 1 || ![null, 'Internal'].includes(links[0].getAttribute('TargetMode'))) fail('a missing, external or duplicate theme relationship');
-      const target = links[0].getAttribute('Target') || '';
+      if (links.length !== 1 || ![null, 'Internal'].includes(links[0].attribute('TargetMode'))) fail('a missing, external or duplicate theme relationship');
+      const target = links[0].attribute('Target') || '';
       if (/[\\?#%:]/.test(target)) fail('an unsupported theme target');
       const segments = (target.startsWith('/') ? target.slice(1) : 'word/' + target).split('/'), normalized = [];
       for (const segment of segments) {

src/documents/src/engine/docx-compatibility.js

@@ -50,15 +50,52 @@ export function activeDescendants(node) {
 }
 // Prefixes used in attribute VALUES (Requires/Ignorable) are invisible to the
 // XML serializer. Carry in-scope namespace declarations on detached fragments.
+// Reading a document serializes every paragraph this way, so the walks below index the
+// attribute maps and follow siblings rather than copying live collections into arrays.
 export function serializeFragment(node) {
-  const copy = node.cloneNode(true), bound = new Set();
-  for (let at = node; at?.nodeType === 1; at = at.parentElement) for (const a of at.attributes) if (a.namespaceURI === XMLNS && !bound.has(a.name)) {
-    bound.add(a.name); copy.setAttributeNS(XMLNS, a.name, a.value);
+  return serializeCopy(node.cloneNode(true), inScopeDeclarations(node));
+}
+/**
+ * serializeFragment(node) as it is now, serialized only when first asked for. The node and
+ * the declarations it inherits are copied now, so later changes to the tree cannot alter the
+ * text. Opening keeps each body paragraph's source this way: most are never asked for (an
+ * unedited save writes the original part), and serializing them all was a large share of
+ * the time to open a long document.
+ */
+export function deferredFragment(node) {
+  let copy = node.cloneNode(true), declarations = inScopeDeclarations(node), text;
+  return () => {
+    if (text === undefined) { text = serializeCopy(copy, declarations); copy = declarations = null; }
+    return text;
+  };
+}
+// The namespace declarations on `node` and its ancestors, the nearest of each prefix first.
+function inScopeDeclarations(node) {
+  const declarations = [], bound = new Set();
+  for (let at = node; at?.nodeType === 1; at = at.parentElement) {
+    const attributes = at.attributes;
+    for (let i = 0; i < attributes.length; i++) {
+      const a = attributes[i];
+      if (a.namespaceURI === XMLNS && !bound.has(a.name)) { bound.add(a.name); declarations.push([a.name, a.value]); }
+    }
   }
+  return declarations;
+}
+function serializeCopy(copy, declarations) {
+  for (const [name, value] of declarations) copy.setAttributeNS(XMLNS, name, value);
   // Detached shells have no ancestors. Bind namespaced attributes explicitly so
   // DOM serializers cannot invent a prefix and reuse it on an unbound sibling.
-  for (const element of [copy, ...copy.getElementsByTagName('*')]) for (const a of [...element.attributes]) {
-    if (a.prefix && a.namespaceURI && a.namespaceURI !== XMLNS && a.prefix !== 'xml' && element.lookupNamespaceURI(a.prefix) !== a.namespaceURI) element.setAttributeNS(XMLNS, `xmlns:${a.prefix}`, a.namespaceURI);
+  // Each element and its descendants in document order; a declaration added here is
+  // appended to the element's attributes, so the count read first covers its own.
+  for (let element = copy; element;) {
+    const attributes = element.attributes;
+    for (let i = 0, count = attributes.length; i < count; i++) {
+      const a = attributes[i];
+      if (a.prefix && a.namespaceURI && a.namespaceURI !== XMLNS && a.prefix !== 'xml' && element.lookupNamespaceURI(a.prefix) !== a.namespaceURI) element.setAttributeNS(XMLNS, `xmlns:${a.prefix}`, a.namespaceURI);
+    }
+    let next = element.firstElementChild;
+    while (!next && element !== copy) { next = element.nextElementSibling; if (!next) element = element.parentElement; }
+    element = next;
   }
   return new XMLSerializer().serializeToString(copy);
 }

src/documents/src/engine/docx-content-controls.js

@@ -1,7 +1,7 @@
 import { UnsupportedDocumentError } from './errors.js';
 import { serializeFragment } from './docx-compatibility.js';
 import { relatedPartPath } from './docx-relationships.js';
-import { parseXml, wordText } from './docx-xml.js';
+import { elementChildren, parseXml, wordText } from './docx-xml.js';
 import { bookmarkReader } from './docx-bookmarks.js';
 import { bookmarkEntries } from './bookmarks.js';
 import { unresolvedBindingReader } from './docx-data-binding.js';
@@ -68,7 +68,7 @@ function projectRetainedControls(root) {
   }
   projectControlContents(root);
 }
-const children = n => [...(n?.children || [])];
+const children = n => n?.children ? elementChildren(n) : [];
 const fail = reason => { throw new UnsupportedDocumentError(`content controls with ${reason}`); };
 const child = (n, name) => children(n).find(c => c.namespaceURI === W && c.localName === name);
 function attributes(node, names = [], ns = W) {

src/documents/src/engine/docx-contextual-spacing.js

@@ -1,8 +1,9 @@
 import { UnsupportedDocumentError } from './errors.js';
+import { childElements } from './docx-xml.js';
 const W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 
 export function readContextualFlag(parent, tag = 'contextualSpacing') {
-  const nodes = [...(parent?.children || [])].filter(n => n.namespaceURI === W && n.localName === tag);
+  const nodes = childElements(parent, W, tag);
   if (nodes.length > 1) throw new UnsupportedDocumentError(`duplicate ${tag}`);
   const node = nodes[0]; if (!node) return undefined;
   if (node.children.length || node.textContent.trim() || [...node.attributes].some(a => a.namespaceURI === W && a.localName !== 'val')) throw new UnsupportedDocumentError(`unsupported ${tag} content`);

src/documents/src/engine/docx-drawing-metadata.js

@@ -1,12 +1,13 @@
 import { UnsupportedDocumentError } from './errors.js';
 import { INK_EXTENSION_URI, readInkExtension } from './docx-ink-extension.js';
+import { elementChildren } from './docx-xml.js';
 
 const A = 'http://schemas.openxmlformats.org/drawingml/2006/main';
 const DECORATIVE = 'http://schemas.microsoft.com/office/drawing/2017/decorative';
 const URI = '{C183D7F6-B498-43B3-948B-1728B52AA6E4}';
 const CREATION = 'http://schemas.microsoft.com/office/drawing/2014/main';
 const CREATION_URI = '{FF2B5EF4-FFF2-40B4-BE49-F238E27FC236}';
-const children = node => [...(node?.children || [])];
+const children = node => node?.children ? elementChildren(node) : [];
 const fail = why => { throw new UnsupportedDocumentError(`drawing metadata with ${why}`); };
 function content(node, attrs = [], kind = 'decorative') {
   if ([...(node?.childNodes || [])].some(n => [3, 4].includes(n.nodeType) && /[^ \t\r\n]/.test(n.textContent))) fail('unexpected metadata text');

src/documents/src/engine/docx-drawing-preview.js

@@ -1,8 +1,9 @@
 import { validateImage } from './images.js';
+import { elementChildren } from './docx-xml.js';
 
 const WP = 'http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing';
 const A = 'http://schemas.openxmlformats.org/drawingml/2006/main';
-const children = node => [...(node?.children || [])];
+const children = node => node?.children ? elementChildren(node) : [];
 
 // Use only an unambiguous, unrotated inline frame. A floating placement or
 // invalid dimensions cannot safely be approximated as an inline rectangle.

src/documents/src/engine/docx-drawing-style.js

@@ -1,12 +1,12 @@
-import { parseXml } from './docx-xml.js';
+import { elementChildren, parseXml } from './docx-xml.js';
 import { relatedPartPath } from './docx-relationships.js';
 import { UnsupportedDocumentError } from './errors.js';
 import { drawingPresetColors } from './drawing-preset-colors.js';
 import { drawingSystemColors } from './drawing-system-colors.js';
 import { luminanceTransform } from './colors.js';
 
 const A = 'http://schemas.openxmlformats.org/drawingml/2006/main';
-const children = n => [...(n?.children || [])];
+const children = n => n?.children ? elementChildren(n) : [];
 const fail = why => { throw new UnsupportedDocumentError(`drawing styles with ${why}`); };
 const clamp = n => Math.max(0, Math.min(1, n));
 // The input and final color resolve through linear RGB percentage units.

src/documents/src/engine/docx-field-results.js

@@ -1,9 +1,10 @@
 import { MC, alternateBranch } from './docx-compatibility.js';
 import { macroButtonLabelStart } from './docx-macro-buttons.js';
+import { elementChildren } from './docx-xml.js';
 
 const W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 const is = (node, name) => node.namespaceURI === W && node.localName === name;
-const children = node => [...node.children];
+const children = node => elementChildren(node);
 const metadata = new Set(['pPr', 'rPr', 'sectPr', 'tblPr', 'trPr', 'tcPr', 'sdtPr', 'del', 'moveFrom']);
 
 // Read one story in document order, including fields spanning paragraphs and

src/documents/src/engine/docx-fonts.js

@@ -1,15 +1,15 @@
-import { parseXml } from './docx-xml.js';
+import { childElements, elementChildren, parseXml } from './docx-xml.js';
 import { relatedPartPath } from './docx-relationships.js';
 import { FONT_SLOTS, checkedFontName } from './font-families.js';
 import { UnsupportedDocumentError } from './errors.js';
 import { documentEmbeddedFonts } from './embedded-fonts.js';
 
 const W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 const A = 'http://schemas.openxmlformats.org/drawingml/2006/main';
-const children = n => [...(n?.children || [])];
+const children = n => n?.children ? elementChildren(n) : [];
 const fail = why => { throw new UnsupportedDocumentError(`font declarations with ${why}`); };
 function only(n, ns, name) {
-  const nodes = children(n).filter(c => c.namespaceURI === ns && c.localName === name);
+  const nodes = childElements(n, ns, name);
   if (nodes.length > 1) fail(`duplicate ${name}`);
   return nodes[0];
 }

src/documents/src/engine/docx-hyperlinks.js

@@ -1,18 +1,19 @@
 import { strToU8 } from 'fflate';
-import { parseXml } from './docx-xml.js';
+import { elementChildren, parseXml } from './docx-xml.js';
 import { UnsupportedDocumentError } from './errors.js';
 import { checkedHyperlinkTarget, hyperlinkId, internalHyperlink } from './hyperlinks.js';
 import { relationshipPartPath } from './package-part-paths.js';
+import { relationshipEntries, relationshipRoot } from './docx-relationships.js';
 const W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 const R = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships';
 const REL = 'http://schemas.openxmlformats.org/package/2006/relationships';
 const serialize = n => new XMLSerializer().serializeToString(n);
-const children = n => [...n.children];
+const children = n => elementChildren(n);
 const relationships = (parts, path) => parts[path] ? parseXml(parts[path]) : parseXml(`<Relationships xmlns="${REL}"/>`);
 
 export function hyperlinkReader(parts, sourcePart = 'word/document.xml') {
   const path = relationshipPartPath(sourcePart);
-  let rels;
+  let root, byId;
   return node => {
     if ([...node.childNodes].some(n => [3, 4].includes(n.nodeType) && n.textContent.trim())) throw new UnsupportedDocumentError('text outside hyperlink runs');
     for (const attr of [...node.attributes]) {
@@ -28,14 +29,23 @@ export function hyperlinkReader(parts, sourcePart = 'word/document.xml') {
     if (node.hasAttributeNS(W, 'history') && !['true', 'false', '1', '0', 'on', 'off'].includes(history)) throw new UnsupportedDocumentError('invalid hyperlink history');
     const common = { id: hyperlinkId(), shell: serialize(node.cloneNode(false)), tooltip: node.getAttributeNS(W, 'tooltip') || '', history: ['true', '1', 'on'].includes(history) };
     if (!sourceId) return { ...common, anchor: node.hasAttributeNS(W, 'anchor') ? node.getAttributeNS(W, 'anchor') : null };
-    rels ||= relationships(parts, path);
-    if (rels.documentElement.namespaceURI !== REL || rels.documentElement.localName !== 'Relationships') throw new UnsupportedDocumentError('invalid hyperlink relationships part');
-    const matches = children(rels.documentElement).filter(r => r.getAttribute('Id') === sourceId);
+    // The relationships part is read once (relationshipEntries) and indexed: a document with
+    // hundreds of links would otherwise scan every relationship for each link.
+    if (!byId) {
+      const bytes = parts[path] || `<Relationships xmlns="${REL}"/>`;
+      root = relationshipRoot(bytes); byId = new Map();
+      for (const r of relationshipEntries(bytes)) {
+        const id = r.attribute('Id');
+        if (byId.has(id)) byId.get(id).push(r); else byId.set(id, [r]);
+      }
+    }
+    if (root.namespaceURI !== REL || root.localName !== 'Relationships') throw new UnsupportedDocumentError('invalid hyperlink relationships part');
+    const matches = byId.get(sourceId) || [];
     const rel = matches[0];
-    if (matches.length !== 1 || rel.namespaceURI !== REL || rel.localName !== 'Relationship' || rel.getAttribute('Type') !== `${R}/hyperlink` || rel.getAttribute('TargetMode') !== 'External') throw new UnsupportedDocumentError('missing, duplicate or unsupported hyperlink relationship');
+    if (matches.length !== 1 || rel.namespaceURI !== REL || rel.localName !== 'Relationship' || rel.attribute('Type') !== `${R}/hyperlink` || rel.attribute('TargetMode') !== 'External') throw new UnsupportedDocumentError('missing, duplicate or unsupported hyperlink relationship');
     // Part 2 relationships permit any URI reference. A destination the editor
     // cannot open (a relative or file path, unescaped spaces) is kept verbatim.
-    const target = rel.getAttribute('Target');
+    const target = rel.attribute('Target');
     let retainedTarget = false;
     try { checkedHyperlinkTarget(target); } catch { retainedTarget = true; }
     return { ...common, target, sourceId, ...(retainedTarget ? { retainedTarget } : {}) };

src/documents/src/engine/docx-images.js

@@ -11,7 +11,7 @@ import { drawingStyleReader, drawingEffectColor } from './docx-drawing-style.js'
 import { readDrawingMetadata, readDrawingCreationId } from './docx-drawing-metadata.js';
 import { readDrawingHyperlink, drawingHyperlinkWriter } from './docx-drawing-hyperlinks.js';
 import { readShapeExtensions } from './docx-shape-extensions.js';
-import { parseXml, setContentTypeOverride } from './docx-xml.js';
+import { elementChildren, parseXml, setContentTypeOverride } from './docx-xml.js';
 import { UnsupportedDocumentError } from './errors.js';
 import { imageInfo, validateImage, normalizeCrop, normalizedRotation, imageExtents, CROP_SIDES } from './images.js';
 import { strToU8 } from 'fflate';
@@ -39,7 +39,7 @@ export const ASVG = 'http://schemas.microsoft.com/office/drawing/2016/SVG/main';
 const SVG_EXTENSION = '{96dac541-7b7a-43d3-8b79-37d633b846f1}';
 const LOCAL_DPI = '{28a0092b-c50c-407e-a947-70e740481c1c}';
 const serialize = n => new XMLSerializer().serializeToString(n);
-const children = n => n ? [...n.children] : [];
+const children = n => n ? elementChildren(n) : [];
 const child = (n, ns, name) => children(n).find(c => c.namespaceURI === ns && c.localName === name);
 const fail = why => { throw new UnsupportedDocumentError(`pictures with ${why}`); };
 function check(n, ns, names) {