@yinyoudexing/xml2word 0.1.4 → 0.1.6

package/dist/index.cjs

@@ -39,7 +39,38 @@ function ensureXmlDeclaration(xml) {
   return `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
 ${xml}`;
 }
-function wrapBodyXml(bodyXml) {
+function resolvePageSetupTwips(page) {
+  const size = page?.size;
+  const defaultSize = { widthTwips: 11906, heightTwips: 16838 };
+  const resolvedSize = typeof size === "object" && size && "widthTwips" in size && "heightTwips" in size ? { widthTwips: Math.round(size.widthTwips), heightTwips: Math.round(size.heightTwips) } : size === "Letter" ? { widthTwips: 12240, heightTwips: 15840 } : size === "Letter_Landscape" ? { widthTwips: 15840, heightTwips: 12240 } : size === "A4_Landscape" ? { widthTwips: 16838, heightTwips: 11906 } : defaultSize;
+  const defaultMargins = {
+    top: 1440,
+    right: 1440,
+    bottom: 1440,
+    left: 1440,
+    header: 708,
+    footer: 708,
+    gutter: 0
+  };
+  const m = page?.marginTwips;
+  const margins = typeof m === "number" ? { ...defaultMargins, top: m, right: m, bottom: m, left: m } : m ? {
+    ...defaultMargins,
+    top: m.top ?? defaultMargins.top,
+    right: m.right ?? defaultMargins.right,
+    bottom: m.bottom ?? defaultMargins.bottom,
+    left: m.left ?? defaultMargins.left,
+    header: m.header ?? defaultMargins.header,
+    footer: m.footer ?? defaultMargins.footer,
+    gutter: m.gutter ?? defaultMargins.gutter
+  } : defaultMargins;
+  return {
+    pageWidthTwips: Math.max(1, Math.round(resolvedSize.widthTwips)),
+    pageHeightTwips: Math.max(1, Math.round(resolvedSize.heightTwips)),
+    margins
+  };
+}
+function wrapBodyXml(bodyXml, page) {
+  const { pageWidthTwips, pageHeightTwips, margins } = resolvePageSetupTwips(page);
   const xml = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
 <w:document
   xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"
@@ -50,10 +81,9 @@ ${bodyXml}
     <w:sectPr>
       <w:headerReference w:type="default" r:id="rId6"/>
       <w:footerReference w:type="default" r:id="rId7"/>
-      <w:pgSz w:w="12240" w:h="15840"/>
-      <w:pgMar w:top="1440" w:right="1440" w:bottom="1440" w:left="1440" w:header="708" w:footer="708" w:gutter="0"/>
+      <w:pgSz w:w="${pageWidthTwips}" w:h="${pageHeightTwips}"/>
+      <w:pgMar w:top="${margins.top}" w:right="${margins.right}" w:bottom="${margins.bottom}" w:left="${margins.left}" w:header="${margins.header}" w:footer="${margins.footer}" w:gutter="${margins.gutter}"/>
       <w:cols w:space="708"/>
-      <w:docGrid w:linePitch="360"/>
     </w:sectPr>
   </w:body>
 </w:document>
@@ -70,7 +100,50 @@ function ensureWordNamespace(xml) {
     `<w:document xmlns:w="${WORD_MAIN_NS}"`
   );
 }
-function normalizeDocumentXml(xml, inputKind) {
+function applyPageSetupToDocumentXml(xml, page) {
+  if (!page) return xml;
+  const { pageWidthTwips, pageHeightTwips, margins } = resolvePageSetupTwips(page);
+  const pgSzTag = `<w:pgSz w:w="${pageWidthTwips}" w:h="${pageHeightTwips}"/>`;
+  const pgMarTag = `<w:pgMar w:top="${margins.top}" w:right="${margins.right}" w:bottom="${margins.bottom}" w:left="${margins.left}" w:header="${margins.header}" w:footer="${margins.footer}" w:gutter="${margins.gutter}"/>`;
+  const sectPrRegex = /<w:sectPr\b[^>]*>[\s\S]*?<\/w:sectPr>/g;
+  const hasSectPr = sectPrRegex.test(xml);
+  if (!hasSectPr) {
+    const insertAt = xml.lastIndexOf("</w:body>");
+    if (insertAt < 0) return xml;
+    const sectPr = `
+    <w:sectPr>
+      ${pgSzTag}
+      ${pgMarTag}
+      <w:cols w:space="708"/>
+    </w:sectPr>
+`;
+    return xml.slice(0, insertAt) + sectPr + xml.slice(insertAt);
+  }
+  return xml.replace(sectPrRegex, (sect) => {
+    let out = sect;
+    const insertTag = (tag) => {
+      const headerFooterRegex = /<w:(?:headerReference|footerReference)\b[^>]*\/>/g;
+      let last = null;
+      let m;
+      while (m = headerFooterRegex.exec(out)) last = m;
+      const open = out.match(/<w:sectPr\b[^>]*>/);
+      const insertPos = last ? last.index + last[0].length : open ? open[0].length : 0;
+      out = out.slice(0, insertPos) + tag + out.slice(insertPos);
+    };
+    if (/<w:pgSz\b[^>]*\/>/.test(out)) {
+      out = out.replace(/<w:pgSz\b[^>]*\/>/g, pgSzTag);
+    } else {
+      insertTag(pgSzTag);
+    }
+    if (/<w:pgMar\b[^>]*\/>/.test(out)) {
+      out = out.replace(/<w:pgMar\b[^>]*\/>/g, pgMarTag);
+    } else {
+      insertTag(pgMarTag);
+    }
+    return out;
+  });
+}
+function normalizeDocumentXml(xml, inputKind, page) {
   const trimmed = xml.trim();
   if (!trimmed) {
     throw new Error("XML is empty.");
@@ -81,16 +154,16 @@ function normalizeDocumentXml(xml, inputKind) {
     if (!hasWordDocumentRoot(withNs)) {
       throw new Error('inputKind="document" requires a <w:document> root.');
     }
-    return withNs;
+    return applyPageSetupToDocumentXml(withNs, page);
   }
   if (inputKind === "body") {
-    return wrapBodyXml(trimmed);
+    return wrapBodyXml(trimmed, page);
   }
   if (hasWordDocumentRoot(trimmed)) {
     const withDecl = ensureXmlDeclaration(trimmed);
-    return ensureWordNamespace(withDecl);
+    return applyPageSetupToDocumentXml(ensureWordNamespace(withDecl), page);
   }
-  return wrapBodyXml(trimmed);
+  return wrapBodyXml(trimmed, page);
 }
 var WORD_MAIN_NS;
 var init_normalizeDocumentXml = __esm({
@@ -175,7 +248,7 @@ ${relLines}
 `;
 }
 async function createDocxZipUint8Array(xml, options = {}) {
-  const documentXml = normalizeDocumentXml(xml, options.inputKind ?? "auto");
+  const documentXml = normalizeDocumentXml(xml, options.inputKind ?? "auto", options.page);
   validateXmlIfNeeded(documentXml, options.validateXml ?? true);
   const zip = new import_jszip.default();
   zip.file("[Content_Types].xml", buildContentTypesXml([]));
@@ -195,7 +268,7 @@ async function createDocxZipUint8Array(xml, options = {}) {
   return zip.generateAsync({ type: "uint8array" });
 }
 async function createDocxZipWithAssetsUint8Array(xml, options, assets) {
-  const documentXml = normalizeDocumentXml(xml, options.inputKind ?? "auto");
+  const documentXml = normalizeDocumentXml(xml, options.inputKind ?? "auto", options.page);
   validateXmlIfNeeded(documentXml, options.validateXml ?? true);
   const zip = new import_jszip.default();
   zip.file("[Content_Types].xml", buildContentTypesXml(assets));
@@ -627,6 +700,92 @@ function normalizeFontFamily(value) {
   if (!first) return void 0;
   return first.replace(/^["']|["']$/g, "");
 }
+function cssSelectorsHitTarget(selectorsText, targetSelector) {
+  const target = targetSelector.trim().toLowerCase();
+  if (!target) return false;
+  const selectors = selectorsText.replace(/\/\*[\s\S]*?\*\//g, " ").split(",").map((s) => s.replace(/\s+/g, " ").trim().toLowerCase()).filter(Boolean);
+  return selectors.some((s) => {
+    if (s === target) return true;
+    if (!s.startsWith(target)) return false;
+    const rest = s.slice(target.length);
+    return rest.startsWith(".") || rest.startsWith(":") || rest.startsWith("#") || rest.startsWith("[");
+  });
+}
+function parseCssRuleFromHtml(html, selector) {
+  const styleTagRegex = /<style\b[^>]*>([\s\S]*?)<\/style>/gi;
+  const merged = {};
+  let m;
+  while (m = styleTagRegex.exec(html)) {
+    const cssText = m[1] ?? "";
+    const target = selector.trim().toLowerCase();
+    if (!target) continue;
+    let depth = 0;
+    let lastRuleEnd = 0;
+    let ruleStart = -1;
+    let declStart = -1;
+    for (let i = 0; i < cssText.length; i++) {
+      const ch = cssText[i];
+      if (ch === "{") {
+        if (depth === 0) {
+          ruleStart = lastRuleEnd;
+          declStart = i + 1;
+        }
+        depth++;
+        continue;
+      }
+      if (ch === "}") {
+        if (depth > 0) depth--;
+        if (depth === 0 && ruleStart >= 0 && declStart >= 0) {
+          const selectorsText = cssText.slice(ruleStart, declStart - 1).trim().toLowerCase();
+          const decl = cssText.slice(declStart, i).trim();
+          lastRuleEnd = i + 1;
+          ruleStart = -1;
+          declStart = -1;
+          const hit = cssSelectorsHitTarget(selectorsText, target);
+          if (hit) Object.assign(merged, parseStyleAttribute(decl));
+        }
+      }
+    }
+  }
+  return merged;
+}
+function parseCssRuleFromHtmlFirst(html, selector) {
+  const styleTagRegex = /<style\b[^>]*>([\s\S]*?)<\/style>/gi;
+  let m;
+  while (m = styleTagRegex.exec(html)) {
+    const cssText = m[1] ?? "";
+    const target = selector.trim().toLowerCase();
+    if (!target) continue;
+    let depth = 0;
+    let lastRuleEnd = 0;
+    let ruleStart = -1;
+    let declStart = -1;
+    for (let i = 0; i < cssText.length; i++) {
+      const ch = cssText[i];
+      if (ch === "{") {
+        if (depth === 0) {
+          ruleStart = lastRuleEnd;
+          declStart = i + 1;
+        }
+        depth++;
+        continue;
+      }
+      if (ch === "}") {
+        if (depth > 0) depth--;
+        if (depth === 0 && ruleStart >= 0 && declStart >= 0) {
+          const selectorsText = cssText.slice(ruleStart, declStart - 1).trim().toLowerCase();
+          const decl = cssText.slice(declStart, i).trim();
+          lastRuleEnd = i + 1;
+          ruleStart = -1;
+          declStart = -1;
+          const hit = cssSelectorsHitTarget(selectorsText, target);
+          if (hit) return parseStyleAttribute(decl);
+        }
+      }
+    }
+  }
+  return {};
+}
 function mergeTextStyle(base, patch) {
   return {
     bold: patch.bold ?? base.bold,
@@ -765,7 +924,7 @@ function applyMaxBoxPx(size, maxBox) {
   const scale = Math.min(1, maxBox.maxWidthPx / w, maxBox.maxHeightPx / h);
   return { widthPx: Math.max(1, Math.round(w * scale)), heightPx: Math.max(1, Math.round(h * scale)) };
 }
-function computeImageSizePx(node, intrinsic) {
+function computeImageSizePx(node, intrinsic, maxBox) {
   const wAttr = node.attribs?.width ? Number(node.attribs.width) : void 0;
   const hAttr = node.attribs?.height ? Number(node.attribs.height) : void 0;
   const css = parseStyleAttribute(node.attribs?.style);
@@ -777,9 +936,9 @@ function computeImageSizePx(node, intrinsic) {
   const widthPx = typeof wCss === "number" ? wCss : typeof widthAttrPx === "number" ? widthAttrPx : intrinsic?.widthPx ?? 300;
   const heightPx = typeof hCss === "number" ? hCss : typeof heightAttrPx === "number" ? heightAttrPx : intrinsic?.heightPx ?? 150;
   const finalSize = typeof wCss === "number" && typeof hCss !== "number" ? { widthPx, heightPx: Math.max(1, Math.round(widthPx * ratio)) } : typeof hCss === "number" && typeof wCss !== "number" ? { widthPx: Math.max(1, Math.round(heightPx / ratio)), heightPx } : typeof widthAttrPx === "number" && typeof heightAttrPx !== "number" && intrinsic ? { widthPx, heightPx: Math.max(1, Math.round(widthPx * ratio)) } : typeof heightAttrPx === "number" && typeof widthAttrPx !== "number" && intrinsic ? { widthPx: Math.max(1, Math.round(heightPx / ratio)), heightPx } : { widthPx, heightPx };
-  return applyMaxBoxPx(finalSize, { maxWidthPx: 624, maxHeightPx: 864 });
+  return applyMaxBoxPx(finalSize, maxBox ?? { maxWidthPx: 624, maxHeightPx: 864 });
 }
-function collectInlineRuns(node, inherited, out, result) {
+function collectInlineRuns(node, inherited, out, result, ctx) {
   if (node.type === "text") {
     const text = node.data ?? "";
     if (text) out.push({ kind: "text", text, style: inherited });
@@ -797,7 +956,11 @@ function collectInlineRuns(node, inherited, out, result) {
       const parsed = parseImageDataUrl(src);
       if (!parsed) return;
       const intrinsic = parseIntrinsicImageSizePx(parsed.contentType, parsed.data);
-      const { widthPx, heightPx } = computeImageSizePx(node, intrinsic);
+      const { widthPx, heightPx } = computeImageSizePx(
+        node,
+        intrinsic,
+        ctx ? { maxWidthPx: ctx.maxImageWidthPx, maxHeightPx: ctx.maxImageHeightPx } : void 0
+      );
       const id = result.images.length + 1;
       const relationshipId = `rId${id + IMAGE_RELATIONSHIP_ID_OFFSET}`;
       const target = `media/image${id}.${parsed.extension}`;
@@ -820,7 +983,11 @@ function collectInlineRuns(node, inherited, out, result) {
       const bufferW = node.attribs?.width ? Number(node.attribs.width) : void 0;
       const bufferH = node.attribs?.height ? Number(node.attribs.height) : void 0;
       const intrinsic = Number.isFinite(bufferW) && bufferW && Number.isFinite(bufferH) && bufferH ? { widthPx: Math.max(1, Math.round(bufferW)), heightPx: Math.max(1, Math.round(bufferH)) } : parseIntrinsicImageSizePx(parsed.contentType, parsed.data);
-      const { widthPx, heightPx } = computeImageSizePx(node, intrinsic);
+      const { widthPx, heightPx } = computeImageSizePx(
+        node,
+        intrinsic,
+        ctx ? { maxWidthPx: ctx.maxImageWidthPx, maxHeightPx: ctx.maxImageHeightPx } : void 0
+      );
       const id = result.images.length + 1;
       const relationshipId = `rId${id + IMAGE_RELATIONSHIP_ID_OFFSET}`;
       const target = `media/image${id}.${parsed.extension}`;
@@ -837,11 +1004,11 @@ function collectInlineRuns(node, inherited, out, result) {
     }
     const next = mergeTextStyle(inherited, styleFromElement(node));
     const children2 = node.children ?? [];
-    for (const c of children2) collectInlineRuns(c, next, out, result);
+    for (const c of children2) collectInlineRuns(c, next, out, result, ctx);
     return;
   }
   const children = node.children ?? [];
-  for (const c of children) collectInlineRuns(c, inherited, out, result);
+  for (const c of children) collectInlineRuns(c, inherited, out, result, ctx);
 }
 function buildRunXml(style, text) {
   const rPrParts = [];
@@ -903,10 +1070,29 @@ function parseCssLengthToTwips(value, baseFontHalfPoints) {
     const basePt = baseFontHalfPoints / 2;
     return Math.round(Number(em[1]) * basePt * 20);
   }
+  const rem = v.match(/^(-?\d+(?:\.\d+)?)rem$/);
+  if (rem) return Math.round(Number(rem[1]) * 16 * 72 * 20 / 96);
   const num = v.match(/^(-?\d+(?:\.\d+)?)$/);
   if (num) return Math.round(Number(num[1]));
   return void 0;
 }
+function extractMarginBeforeAfterTwips(css, baseFontHalfPoints) {
+  const before = parseCssLengthToTwips(css["margin-top"], baseFontHalfPoints);
+  const after = parseCssLengthToTwips(css["margin-bottom"], baseFontHalfPoints);
+  if (typeof before === "number" || typeof after === "number") {
+    return { beforeTwips: before, afterTwips: after };
+  }
+  const m = css.margin?.trim().toLowerCase();
+  if (!m) return {};
+  const tokens = m.split(/\s+/).filter(Boolean);
+  if (!tokens.length) return {};
+  const topToken = tokens[0];
+  const bottomToken = tokens.length === 1 ? tokens[0] : tokens.length === 2 ? tokens[0] : tokens[2] ?? tokens[0];
+  return {
+    beforeTwips: parseCssLengthToTwips(topToken, baseFontHalfPoints),
+    afterTwips: parseCssLengthToTwips(bottomToken, baseFontHalfPoints)
+  };
+}
 function inferFirstFontSizeHalfPoints(node) {
   const stack = [node];
   while (stack.length) {
@@ -923,8 +1109,10 @@ function inferFirstFontSizeHalfPoints(node) {
   }
   return void 0;
 }
-function buildParagraphPrXml(node, baseFontHalfPoints, extraInd, pStyleId) {
-  const css = parseStyleAttribute(node.attribs?.style);
+function buildParagraphPrXml(node, baseFontHalfPoints, extraInd, pStyleId, defaultCss) {
+  const tag = node.type === "tag" ? node.name?.toLowerCase() : void 0;
+  const inlineCss = parseStyleAttribute(node.attribs?.style);
+  const css = defaultCss ? { ...defaultCss, ...inlineCss } : inlineCss;
   const parts = [];
   if (pStyleId) parts.push(`<w:pStyle w:val="${escapeXmlText(pStyleId)}"/>`);
   const shdHex = extractBackgroundFillHex(css);
@@ -951,10 +1139,17 @@ function buildParagraphPrXml(node, baseFontHalfPoints, extraInd, pStyleId) {
   if (typeof hangingTwips === "number") indAttrs.push(`w:hanging="${hangingTwips}"`);
   if (typeof firstLine === "number") indAttrs.push(`w:firstLine="${firstLine}"`);
   if (indAttrs.length) parts.push(`<w:ind ${indAttrs.join(" ")}/>`);
-  const before = parseCssLengthToTwips(css["margin-top"], baseFontHalfPoints);
-  const after = parseCssLengthToTwips(css["margin-bottom"], baseFontHalfPoints);
+  const hasInlineBefore = inlineCss["margin-top"] != null;
+  const hasInlineAfter = inlineCss["margin-bottom"] != null;
+  const beforeToken = inlineCss["margin-top"] ?? (pStyleId ? void 0 : defaultCss?.["margin-top"]);
+  const afterToken = inlineCss["margin-bottom"] ?? (pStyleId ? void 0 : defaultCss?.["margin-bottom"]);
+  let before = parseCssLengthToTwips(beforeToken, baseFontHalfPoints);
+  let after = parseCssLengthToTwips(afterToken, baseFontHalfPoints);
+  if (tag === "p" && !hasInlineBefore && typeof before === "number") before = Math.min(before, 160);
+  if (tag === "p" && !hasInlineAfter && typeof after === "number") after = Math.min(after, 160);
   const lineHeight = (() => {
-    const lh = css["line-height"]?.trim().toLowerCase();
+    const lhToken = inlineCss["line-height"] ?? (pStyleId ? void 0 : defaultCss?.["line-height"]);
+    const lh = lhToken?.trim().toLowerCase();
     if (!lh || lh === "normal") return void 0;
     const unitless = lh.match(/^(\d+(?:\.\d+)?)$/);
     if (unitless) {
@@ -969,8 +1164,16 @@ function buildParagraphPrXml(node, baseFontHalfPoints, extraInd, pStyleId) {
   })();
   if (typeof before === "number" || typeof after === "number" || typeof lineHeight === "number") {
     const attrs = [];
-    if (typeof before === "number") attrs.push(`w:before="${Math.max(0, before)}"`);
-    if (typeof after === "number") attrs.push(`w:after="${Math.max(0, after)}"`);
+    if (typeof before === "number") {
+      attrs.push(`w:before="${Math.max(0, before)}"`);
+    } else if (typeof lineHeight === "number") {
+      attrs.push('w:before="0"');
+    }
+    if (typeof after === "number") {
+      attrs.push(`w:after="${Math.max(0, after)}"`);
+    } else if (typeof lineHeight === "number") {
+      attrs.push('w:after="160"');
+    }
     if (typeof lineHeight === "number") {
       attrs.push(`w:line="${lineHeight}"`, 'w:lineRule="exact"');
     }
@@ -979,16 +1182,19 @@ function buildParagraphPrXml(node, baseFontHalfPoints, extraInd, pStyleId) {
   if (!parts.length) return "";
   return `<w:pPr>${parts.join("")}</w:pPr>`;
 }
-function buildParagraphXmlFromContainer(node, baseStyle, extraInd, pStyleId, result) {
-  const containerStyle = mergeTextStyle(baseStyle, styleFromElement(node));
+function buildParagraphXmlFromContainer(node, baseStyle, extraInd, pStyleId, result, ctx) {
+  const seededBaseStyle = ctx ? { fontSizeHalfPoints: ctx.defaultBaseFontHalfPoints } : {};
+  const containerStyle = mergeTextStyle(mergeTextStyle(seededBaseStyle, baseStyle), styleFromElement(node));
   const baseFontHalfPoints = containerStyle.fontSizeHalfPoints ?? inferFirstFontSizeHalfPoints(node) ?? 28;
-  const pPrXml = buildParagraphPrXml(node, baseFontHalfPoints, extraInd, pStyleId);
+  const computedBaseFontHalfPoints = containerStyle.fontSizeHalfPoints ?? inferFirstFontSizeHalfPoints(node) ?? ctx?.defaultBaseFontHalfPoints ?? 28;
+  const defaultCss = ctx ? node.type === "tag" && node.name?.toLowerCase() === "p" ? { ...ctx.defaultBodyCss, ...ctx.defaultPCss } : ctx.defaultBodyCss : void 0;
+  const pPrXml = buildParagraphPrXml(node, computedBaseFontHalfPoints, extraInd, pStyleId, defaultCss);
   const runs = [];
   const res = result ?? {
     bodyXml: "",
     images: []
   };
-  for (const c of node.children ?? []) collectInlineRuns(c, containerStyle, runs, res);
+  for (const c of node.children ?? []) collectInlineRuns(c, containerStyle, runs, res, ctx);
   const rXml = [];
   for (const token of runs) {
     if (token.kind === "br") {
@@ -1007,6 +1213,11 @@ function buildParagraphXmlFromContainer(node, baseStyle, extraInd, pStyleId, res
   if (!rXml.length) return "";
   return `<w:p>${pPrXml}${rXml.join("")}</w:p>`;
 }
+function buildSpacerParagraphXml(afterTwips) {
+  const after = Math.max(0, Math.round(afterTwips));
+  if (!after) return "";
+  return `<w:p><w:pPr><w:spacing w:before="0" w:after="${after}" w:line="1" w:lineRule="exact"/></w:pPr><w:r><w:rPr><w:sz w:val="1"/><w:szCs w:val="1"/></w:rPr><w:t></w:t></w:r></w:p>`;
+}
 function isExplicitPageBreak(node) {
   if (node.type !== "tag") return false;
   const tag = node.name?.toLowerCase();
@@ -1015,13 +1226,14 @@ function isExplicitPageBreak(node) {
   const classList = cls ? cls.split(/\s+/) : [];
   if (tag === "hr" && classList.includes("page-break")) return true;
   if (classList.includes("page-break")) return true;
+  if (classList.includes("umo-page-break")) return true;
   if (node.attribs?.["data-page-break"] === "true") return true;
   const after = css["page-break-after"]?.toLowerCase() ?? css["break-after"]?.toLowerCase();
   const before = css["page-break-before"]?.toLowerCase() ?? css["break-before"]?.toLowerCase();
   if (after?.includes("always") || before?.includes("always")) return true;
   return false;
 }
-function buildListBlocks(listNode, ordered, level, result) {
+function buildListBlocks(listNode, ordered, level, result, ctx) {
   const liNodes = (listNode.children ?? []).filter(
     (c) => c.type === "tag" && c.name?.toLowerCase() === "li"
   );
@@ -1043,7 +1255,7 @@ function buildListBlocks(listNode, ordered, level, result) {
           continue;
         }
       }
-      collectInlineRuns(c, baseStyle, runs, result);
+      collectInlineRuns(c, baseStyle, runs, result, ctx);
     }
     const rXml = [];
     for (const token of runs) {
@@ -1061,12 +1273,13 @@ function buildListBlocks(listNode, ordered, level, result) {
       rXml.push(buildRunXml(token.style, text));
     }
     if (rXml.length) {
-      const baseFontHalfPoints = inferFirstFontSizeHalfPoints(li) ?? 28;
+      const baseFontHalfPoints = inferFirstFontSizeHalfPoints(li) ?? ctx?.defaultBaseFontHalfPoints ?? 28;
       const pPrXml = buildParagraphPrXml(
         li,
         baseFontHalfPoints,
         { leftTwips, hangingTwips },
-        void 0
+        void 0,
+        ctx?.defaultBodyCss
       );
       const numPrXml = `<w:numPr><w:ilvl w:val="${ilvl}"/><w:numId w:val="${numId}"/></w:numPr>`;
       const mergedPPrXml = pPrXml ? pPrXml.replace("<w:pPr>", `<w:pPr>${numPrXml}`) : `<w:pPr>${numPrXml}<w:ind w:left="${leftTwips}" w:hanging="${hangingTwips}"/></w:pPr>`;
@@ -1074,7 +1287,7 @@ function buildListBlocks(listNode, ordered, level, result) {
     }
     for (const nested of nestedLists) {
       const nestedOrdered = nested.name?.toLowerCase() === "ol";
-      out.push(...buildListBlocks(nested, nestedOrdered, ilvl + 1, result));
+      out.push(...buildListBlocks(nested, nestedOrdered, ilvl + 1, result, ctx));
     }
   }
   return out;
@@ -1153,7 +1366,7 @@ function injectTableCellParagraphSpacing(pXml) {
   }
   return pXml.replace("<w:p>", `<w:p><w:pPr>${spacingXml}</w:pPr>`);
 }
-function buildTableCellBlocksXml(cell, baseStyle, result) {
+function buildTableCellBlocksXml(cell, baseStyle, result, ctx) {
   const children = cell.children ?? [];
   const hasBlocks = children.some((c) => {
     if (c.type !== "tag") return false;
@@ -1162,7 +1375,7 @@ function buildTableCellBlocksXml(cell, baseStyle, result) {
   });
   const out = [];
   if (!hasBlocks) {
-    const p = buildParagraphXmlFromContainer(cell, baseStyle, void 0, void 0, result);
+    const p = buildParagraphXmlFromContainer(cell, baseStyle, void 0, void 0, result, ctx);
     if (p) out.push(p);
     return out.length ? out.map(injectTableCellParagraphSpacing).join("") : "<w:p/>";
   }
@@ -1170,18 +1383,18 @@ function buildTableCellBlocksXml(cell, baseStyle, result) {
     if (c.type === "tag") {
       const tag = c.name?.toLowerCase();
       if (tag === "p") {
-        const p = buildParagraphXmlFromContainer(c, baseStyle, void 0, void 0, result);
+        const p = buildParagraphXmlFromContainer(c, baseStyle, void 0, void 0, result, ctx);
         if (p) out.push(p);
         continue;
       }
       if (tag && /^h[1-6]$/.test(tag)) {
         const level = Number(tag.slice(1));
-        const p = buildParagraphXmlFromContainer(c, baseStyle, void 0, `Heading${level}`, result);
+        const p = buildParagraphXmlFromContainer(c, baseStyle, void 0, `Heading${level}`, result, ctx);
         if (p) out.push(p);
         continue;
       }
       if (tag === "ul" || tag === "ol") {
-        out.push(...buildListBlocks(c, tag === "ol", 0, result));
+        out.push(...buildListBlocks(c, tag === "ol", 0, result, ctx));
         continue;
       }
       if (tag === "img" || tag === "canvas") {
@@ -1194,7 +1407,7 @@ function buildTableCellBlocksXml(cell, baseStyle, result) {
   if (!out.length) return "<w:p/>";
   return out.map(injectTableCellParagraphSpacing).join("");
 }
-function buildTableXml(tableNode, result) {
+function buildTableXml(tableNode, result, ctx) {
   const rows = [];
   const stack = [...tableNode.children ?? []];
   while (stack.length) {
@@ -1206,9 +1419,19 @@ function buildTableXml(tableNode, result) {
     (tr) => (tr.children ?? []).filter((c) => c.type === "tag" && (c.name === "td" || c.name === "th"))
   );
   const colCount = Math.max(0, ...rowCells.map((cells) => cells.length));
-  const maxTableWidthTwips = 9360;
+  const maxTableWidthTwips = ctx?.maxTableWidthTwips ?? 9360;
+  const colGroup = (tableNode.children ?? []).find((c) => c.type === "tag" && c.name === "colgroup");
+  const colWidthsFromGroup = [];
+  if (colGroup) {
+    const cols = (colGroup.children ?? []).filter((c) => c.type === "tag" && c.name === "col");
+    for (const col of cols) {
+      const css = parseStyleAttribute(col.attribs?.style);
+      const w = parseCssLengthToTwips(css.width ?? css["min-width"], 28);
+      colWidthsFromGroup.push(w);
+    }
+  }
   const estimatedColWidths = new Array(colCount).fill(0).map((_, i) => {
-    let explicit;
+    let explicit = colWidthsFromGroup[i];
     let estimated = 0;
     for (const cells of rowCells) {
       const cell = cells[i];
@@ -1221,7 +1444,7 @@ function buildTableXml(tableNode, result) {
       const wTwips = estimateTextWidthTwips(text, baseFontHalfPoints) + 240;
       estimated = Math.max(estimated, wTwips);
     }
-    const base = typeof explicit === "number" ? explicit : estimated || Math.round(maxTableWidthTwips / Math.max(1, colCount));
+    const base = typeof explicit === "number" ? Math.max(explicit, estimated) : estimated || Math.round(maxTableWidthTwips / Math.max(1, colCount));
     return Math.max(720, Math.min(6e3, Math.round(base)));
   });
   const normalizedColWidths = (() => {
@@ -1246,7 +1469,7 @@ function buildTableXml(tableNode, result) {
       const cell = cells[i];
       const isHeader = cell.name === "th";
       const baseStyle = isHeader ? { bold: true } : {};
-      const paragraphs = buildTableCellBlocksXml(cell, baseStyle, result);
+      const paragraphs = buildTableCellBlocksXml(cell, baseStyle, result, ctx);
       const css = parseStyleAttribute(cell.attribs?.style);
       const widthTwips = parseCellWidthTwips(cell) ?? normalizedColWidths[i];
       const tcW = typeof widthTwips === "number" ? `<w:tcW w:w="${widthTwips}" w:type="dxa"/>` : `<w:tcW w:w="0" w:type="auto"/>`;
@@ -1335,14 +1558,14 @@ function buildTableXml(tableNode, result) {
   const tblPr = `<w:tblPr>${tblW}<w:tblLayout w:type="fixed"/>${tblAlign}${tblBorder}</w:tblPr>`;
   return `<w:tbl>${tblPr}${tblGrid}${rowXml.join("")}</w:tbl>`;
 }
-function buildParagraphXmlFromSingleInlineNode(node, baseStyle, result) {
+function buildParagraphXmlFromSingleInlineNode(node, baseStyle, result, ctx) {
   const wrapper = {
     type: "tag",
     name: "p",
     attribs: { style: "text-align: center;" },
     children: [node]
   };
-  return buildParagraphXmlFromContainer(wrapper, baseStyle, void 0, void 0, result);
+  return buildParagraphXmlFromContainer(wrapper, baseStyle, void 0, void 0, result, ctx);
 }
 function isRecognizedBlockTag(tag) {
   if (!tag) return false;
@@ -1354,7 +1577,22 @@ function isRecognizedBlockTag(tag) {
   if (tag === "pre") return true;
   return false;
 }
-function collectDivBlocks(node, out, result) {
+function subtreeHasRecognizedBlocks(root) {
+  const stack = [root];
+  while (stack.length) {
+    const cur = stack.pop();
+    if (cur.type === "tag") {
+      if (isExplicitPageBreak(cur)) return true;
+      if (isRecognizedBlockTag(cur.name?.toLowerCase())) return true;
+    }
+    const children = cur.children ?? [];
+    for (let i = children.length - 1; i >= 0; i--) {
+      stack.push(children[i]);
+    }
+  }
+  return false;
+}
+function collectDivBlocks(node, out, result, ctx) {
   const parentStyle = node.attribs?.style;
   const inlineBuffer = [];
   const flushInline = () => {
@@ -1364,7 +1602,7 @@ function collectDivBlocks(node, out, result) {
       attribs: { style: parentStyle },
       children: inlineBuffer.splice(0)
     };
-    const pXml = buildParagraphXmlFromContainer(wrapper, {}, void 0, void 0, result);
+    const pXml = buildParagraphXmlFromContainer(wrapper, {}, void 0, void 0, result, ctx);
     if (pXml) out.push(pXml);
   };
   const children = node.children ?? [];
@@ -1378,19 +1616,15 @@ function collectDivBlocks(node, out, result) {
       }
       if (isRecognizedBlockTag(tag)) {
         if (inlineBuffer.length) flushInline();
-        collectBodyBlocks(child, out, result);
+        collectBodyBlocks(child, out, result, ctx);
+        continue;
+      }
+      if (subtreeHasRecognizedBlocks(child)) {
+        if (inlineBuffer.length) flushInline();
+        collectBodyBlocks(child, out, result, ctx);
         continue;
       }
       if (tag === "div") {
-        const childHasRecognizedBlocks = (child.children ?? []).some((gc) => {
-          if (gc.type !== "tag") return false;
-          return isRecognizedBlockTag(gc.name?.toLowerCase());
-        });
-        if (childHasRecognizedBlocks) {
-          if (inlineBuffer.length) flushInline();
-          collectBodyBlocks(child, out, result);
-          continue;
-        }
         if (inlineBuffer.length) flushInline();
         const mergedStyle = [parentStyle, child.attribs?.style].filter(Boolean).join(";");
         const wrapper = {
@@ -1399,7 +1633,7 @@ function collectDivBlocks(node, out, result) {
           attribs: { style: mergedStyle || void 0 },
           children: child.children ?? []
         };
-        const pXml = buildParagraphXmlFromContainer(wrapper, {}, void 0, void 0, result);
+        const pXml = buildParagraphXmlFromContainer(wrapper, {}, void 0, void 0, result, ctx);
         if (pXml) out.push(pXml);
         continue;
       }
@@ -1408,7 +1642,7 @@ function collectDivBlocks(node, out, result) {
   }
   if (inlineBuffer.length) flushInline();
 }
-function collectBodyBlocks(node, out, result) {
+function collectBodyBlocks(node, out, result, ctx) {
   if (isSkippableSubtree(node)) return;
   if (node.type === "tag") {
     const tag = node.name?.toLowerCase();
@@ -1417,36 +1651,51 @@ function collectBodyBlocks(node, out, result) {
       return;
     }
     if (tag === "p") {
-      const pXml = buildParagraphXmlFromContainer(node, {}, void 0, void 0, result);
+      const pXml = buildParagraphXmlFromContainer(node, {}, void 0, void 0, result, ctx);
       if (pXml) out.push(pXml);
       return;
     }
     if (tag === "img" || tag === "canvas") {
-      const pXml = buildParagraphXmlFromSingleInlineNode(node, {}, result);
+      const pXml = buildParagraphXmlFromSingleInlineNode(node, {}, result, ctx);
       if (pXml) out.push(pXml);
       return;
     }
     if (tag && /^h[1-6]$/.test(tag)) {
       const level = Number(tag.slice(1));
-      const hXml = buildParagraphXmlFromContainer(node, {}, void 0, `Heading${level}`, result);
+      const hXml = buildParagraphXmlFromContainer(node, {}, void 0, `Heading${level}`, result, ctx);
       if (hXml) out.push(hXml);
       return;
     }
     if (tag === "table") {
-      const tblXml = buildTableXml(node, result);
+      const tblXml = buildTableXml(node, result, ctx);
       if (tblXml) out.push(tblXml);
       return;
     }
     if (tag === "ul" || tag === "ol") {
-      out.push(...buildListBlocks(node, tag === "ol", 0, result));
+      out.push(...buildListBlocks(node, tag === "ol", 0, result, ctx));
       return;
     }
     if (tag === "div") {
-      collectDivBlocks(node, out, result);
+      if (hasClass(node, "tableWrapper")) {
+        const display = ctx.tableWrapperCss.display?.trim().toLowerCase();
+        if (display !== "contents") {
+          const baseFontHalfPoints = inferFirstFontSizeHalfPoints(node) ?? ctx.defaultBaseFontHalfPoints;
+          const { beforeTwips, afterTwips } = extractMarginBeforeAfterTwips(ctx.tableWrapperCss, baseFontHalfPoints);
+          const beforeXml = typeof beforeTwips === "number" && beforeTwips > 0 ? buildSpacerParagraphXml(beforeTwips) : "";
+          const afterXml = typeof afterTwips === "number" && afterTwips > 0 ? buildSpacerParagraphXml(afterTwips) : "";
+          if (beforeXml) out.push(beforeXml);
+          collectDivBlocks(node, out, result, ctx);
+          if (afterXml) out.push(afterXml);
+          return;
+        }
+        collectDivBlocks(node, out, result, ctx);
+        return;
+      }
+      collectDivBlocks(node, out, result, ctx);
       return;
     }
   }
-  for (const c of node.children ?? []) collectBodyBlocks(c, out, result);
+  for (const c of node.children ?? []) collectBodyBlocks(c, out, result, ctx);
 }
 function textToWordBodyXml(text) {
   const normalized = text.replace(/\r\n/g, "\n").replace(/\r/g, "\n");
@@ -1464,8 +1713,22 @@ function textToWordBodyXml(text) {
   }
   return out.join("");
 }
-function htmlToWordBody(html) {
+function htmlToWordBody(html, layout) {
   const normalized = html.replace(/\r\n/g, "\n").replace(/\r/g, "\n");
+  const bodyCss = parseCssRuleFromHtmlFirst(normalized, "body");
+  const pCss = parseCssRuleFromHtmlFirst(normalized, "p");
+  const defaultBaseFontHalfPoints = parseFontSizeToHalfPoints(bodyCss["font-size"]) ?? 28;
+  const defaultBodyCss = {};
+  if (bodyCss["line-height"]) defaultBodyCss["line-height"] = bodyCss["line-height"];
+  const defaultPCss = {};
+  if (pCss["line-height"]) defaultPCss["line-height"] = pCss["line-height"];
+  if (pCss["margin-top"]) defaultPCss["margin-top"] = pCss["margin-top"];
+  if (pCss["margin-bottom"]) defaultPCss["margin-bottom"] = pCss["margin-bottom"];
+  if (pCss["text-align"]) defaultPCss["text-align"] = pCss["text-align"];
+  const tableWrapperCss = {
+    ...parseCssRuleFromHtml(normalized, ".tableWrapper"),
+    ...parseCssRuleFromHtml(normalized, ".tiptap .tableWrapper")
+  };
   const doc = (0, import_htmlparser2.parseDocument)(normalized, {
     lowerCaseAttributeNames: true,
     lowerCaseTags: true,
@@ -1473,12 +1736,21 @@ function htmlToWordBody(html) {
   });
   const result = { bodyXml: "", images: [] };
   const out = [];
-  collectBodyBlocks(doc, out, result);
+  const ctx = {
+    defaultBaseFontHalfPoints,
+    defaultBodyCss,
+    defaultPCss,
+    tableWrapperCss,
+    maxTableWidthTwips: layout?.maxTableWidthTwips ?? 9360,
+    maxImageWidthPx: layout?.maxImageWidthPx ?? 624,
+    maxImageHeightPx: layout?.maxImageHeightPx ?? 864
+  };
+  collectBodyBlocks(doc, out, result, ctx);
   result.bodyXml = out.join("");
   return result;
 }
 function htmlToWordBodyXml(html) {
-  const { bodyXml } = htmlToWordBody(html);
+  const { bodyXml } = htmlToWordBody(html, void 0);
   if (!bodyXml) {
     const text = getTextContent(
       (0, import_htmlparser2.parseDocument)(html, {
@@ -1491,8 +1763,8 @@ function htmlToWordBodyXml(html) {
   }
   return bodyXml;
 }
-function htmlToWordBodyWithAssets(html) {
-  const result = htmlToWordBody(html);
+function htmlToWordBodyWithAssets(html, layout) {
+  const result = htmlToWordBody(html, layout);
   if (!result.bodyXml) {
     const text = getTextContent(
       (0, import_htmlparser2.parseDocument)(html, {
@@ -1532,6 +1804,47 @@ function looksLikeHtml(input) {
   if (!s.includes("<") || !s.includes(">")) return false;
   return /<\/?[a-zA-Z][\s\S]*?>/.test(s);
 }
+function resolvePageSetupTwips2(page) {
+  const size = page?.size;
+  const defaultSize = { widthTwips: 11906, heightTwips: 16838 };
+  const resolvedSize = typeof size === "object" && size && "widthTwips" in size && "heightTwips" in size ? { widthTwips: Math.round(size.widthTwips), heightTwips: Math.round(size.heightTwips) } : size === "Letter" ? { widthTwips: 12240, heightTwips: 15840 } : size === "Letter_Landscape" ? { widthTwips: 15840, heightTwips: 12240 } : size === "A4_Landscape" ? { widthTwips: 16838, heightTwips: 11906 } : defaultSize;
+  const defaultMargins = {
+    top: 1440,
+    right: 1440,
+    bottom: 1440,
+    left: 1440,
+    header: 708,
+    footer: 708,
+    gutter: 0
+  };
+  const m = page?.marginTwips;
+  const mergedMargins = typeof m === "number" ? { ...defaultMargins, top: m, right: m, bottom: m, left: m } : m ? {
+    ...defaultMargins,
+    top: m.top ?? defaultMargins.top,
+    right: m.right ?? defaultMargins.right,
+    bottom: m.bottom ?? defaultMargins.bottom,
+    left: m.left ?? defaultMargins.left,
+    header: m.header ?? defaultMargins.header,
+    footer: m.footer ?? defaultMargins.footer,
+    gutter: m.gutter ?? defaultMargins.gutter
+  } : defaultMargins;
+  return {
+    pageWidthTwips: Math.max(1, Math.round(resolvedSize.widthTwips)),
+    pageHeightTwips: Math.max(1, Math.round(resolvedSize.heightTwips)),
+    margins: mergedMargins
+  };
+}
+function computeHtmlConversionLayoutFromPage(page) {
+  const { pageWidthTwips, pageHeightTwips, margins } = resolvePageSetupTwips2(page);
+  const contentWidthTwips = Math.max(1, pageWidthTwips - margins.left - margins.right);
+  const contentHeightTwips = Math.max(1, pageHeightTwips - margins.top - margins.bottom);
+  const twipsToPx = (twips) => Math.max(1, Math.round(twips / 15));
+  return {
+    maxTableWidthTwips: contentWidthTwips,
+    maxImageWidthPx: twipsToPx(contentWidthTwips),
+    maxImageHeightPx: twipsToPx(contentHeightTwips)
+  };
+}
 async function xmlToDocxUint8Array(xml, options = {}) {
   const { createDocxZipUint8Array: createDocxZipUint8Array2 } = await Promise.resolve().then(() => (init_createDocxZip(), createDocxZip_exports));
   return createDocxZipUint8Array2(xml, options);
@@ -1560,14 +1873,17 @@ async function htmlToDocxUint8Array(html, options = {}) {
     const bodyXml2 = textToWordBodyXml2(html);
     return xmlToDocxUint8Array(bodyXml2, {
       inputKind: "body",
-      validateXml: options.validateXml
+      validateXml: options.validateXml,
+      page: options.page
     });
   }
-  const { bodyXml, images } = htmlToWordBodyWithAssets2(html);
+  const layout = options.page ? computeHtmlConversionLayoutFromPage(options.page) : void 0;
+  const { bodyXml, images } = htmlToWordBodyWithAssets2(html, layout);
   if (!images.length) {
     return xmlToDocxUint8Array(bodyXml || htmlToWordBodyXml2(html), {
       inputKind: "body",
-      validateXml: options.validateXml
+      validateXml: options.validateXml,
+      page: options.page
     });
   }
   const assets = images.map((img) => ({
@@ -1579,7 +1895,7 @@ async function htmlToDocxUint8Array(html, options = {}) {
   }));
   return createDocxZipWithAssetsUint8Array2(
     bodyXml,
-    { inputKind: "body", validateXml: options.validateXml },
+    { inputKind: "body", validateXml: options.validateXml, page: options.page },
     assets
   );
 }