@coding01/docsjs 0.1.3 → 0.1.6

package/dist/index.cjs

@@ -34,7 +34,9 @@ __export(index_exports, {
   calculateFidelityScore: () => calculateFidelityScore,
   collectSemanticStatsFromDocument: () => collectSemanticStatsFromDocument,
   collectSemanticStatsFromHtml: () => collectSemanticStatsFromHtml,
-  defineDocsWordElement: () => defineDocsWordElement
+  defineDocsWordElement: () => defineDocsWordElement,
+  parseDocxToHtmlSnapshot: () => parseDocxToHtmlSnapshot,
+  parseDocxToHtmlSnapshotWithReport: () => parseDocxToHtmlSnapshotWithReport
 });
 module.exports = __toCommonJS(index_exports);
 
@@ -57,6 +59,21 @@ function buildHtmlSnapshot(rawHtml) {
 
 // src/lib/docxHtml.ts
 var import_jszip = __toESM(require("jszip"), 1);
+function createEmptyFeatureCounts() {
+  return {
+    hyperlinkCount: 0,
+    anchorImageCount: 0,
+    chartCount: 0,
+    smartArtCount: 0,
+    ommlCount: 0,
+    tableCount: 0,
+    footnoteRefCount: 0,
+    endnoteRefCount: 0,
+    commentRefCount: 0,
+    revisionCount: 0,
+    pageBreakCount: 0
+  };
+}
 function escapeHtml(text) {
   return text.replaceAll("&", "&amp;").replaceAll("<", "&lt;").replaceAll(">", "&gt;").replaceAll('"', "&quot;");
 }
@@ -80,6 +97,9 @@ function getAttr(node, name) {
 function emuToPx(emu) {
   return emu * 96 / 914400;
 }
+function twipToPx(twip) {
+  return twip * 96 / 1440;
+}
 function parseDrawingSizePx(drawing) {
   const extentNode = queryAllByLocalName(drawing, "extent").find((node) => {
     const parent = node.parentElement;
@@ -108,9 +128,7 @@ function imageDimensionAttributes(sizePx) {
   }
   return attrs.length > 0 ? ` ${attrs.join(" ")}` : "";
 }
-function parseAnchorPositionPx(drawing) {
-  const anchor = directChildrenByLocalName(drawing, "anchor")[0] ?? null;
-  if (!anchor) return { leftPx: null, topPx: null };
+function parseAnchorPositionPx(anchor) {
   let leftPx = null;
   let topPx = null;
   const positionH = directChildrenByLocalName(anchor, "positionH")[0] ?? null;
@@ -125,34 +143,80 @@ function parseAnchorPositionPx(drawing) {
   if (Number.isFinite(top)) topPx = emuToPx(top);
   return { leftPx, topPx };
 }
-function parseAnchorWrapMode(drawing) {
-  const anchor = directChildrenByLocalName(drawing, "anchor")[0] ?? null;
-  if (!anchor) return null;
+function parseAnchorWrapMode(anchor) {
   if (directChildrenByLocalName(anchor, "wrapSquare")[0]) return "square";
   if (directChildrenByLocalName(anchor, "wrapTight")[0]) return "tight";
   if (directChildrenByLocalName(anchor, "wrapTopAndBottom")[0]) return "topAndBottom";
   if (directChildrenByLocalName(anchor, "wrapNone")[0]) return "none";
   return null;
 }
-function mergeImageStyle(baseAttrs, anchorPos, wrapMode) {
-  if (anchorPos.leftPx === null && anchorPos.topPx === null) return baseAttrs;
+function parseAnchorMeta(drawing) {
+  const anchor = directChildrenByLocalName(drawing, "anchor")[0] ?? null;
+  if (!anchor) return null;
+  const positionH = directChildrenByLocalName(anchor, "positionH")[0] ?? null;
+  const positionV = directChildrenByLocalName(anchor, "positionV")[0] ?? null;
+  const relativeFromH = getAttr(positionH, "relativeFrom");
+  const relativeFromV = getAttr(positionV, "relativeFrom");
+  const parseDistPx = (name) => {
+    const raw = getAttr(anchor, name);
+    const emu = raw ? Number.parseInt(raw, 10) : Number.NaN;
+    return Number.isFinite(emu) && emu >= 0 ? emuToPx(emu) : null;
+  };
+  const rawHeight = getAttr(anchor, "relativeHeight");
+  const parsedHeight = rawHeight ? Number.parseInt(rawHeight, 10) : Number.NaN;
+  const boolAttr = (name, fallback) => {
+    const raw = (getAttr(anchor, name) ?? "").toLowerCase();
+    if (raw === "1" || raw === "true" || raw === "on") return true;
+    if (raw === "0" || raw === "false" || raw === "off") return false;
+    return fallback;
+  };
+  return {
+    position: parseAnchorPositionPx(anchor),
+    wrapMode: parseAnchorWrapMode(anchor),
+    distTPx: parseDistPx("distT"),
+    distBPx: parseDistPx("distB"),
+    distLPx: parseDistPx("distL"),
+    distRPx: parseDistPx("distR"),
+    relativeFromH,
+    relativeFromV,
+    behindDoc: boolAttr("behindDoc", false),
+    allowOverlap: boolAttr("allowOverlap", true),
+    layoutInCell: boolAttr("layoutInCell", true),
+    relativeHeight: Number.isFinite(parsedHeight) ? parsedHeight : null
+  };
+}
+function mergeImageStyle(baseAttrs, anchorMeta) {
+  if (!anchorMeta) return baseAttrs;
+  const { position, wrapMode } = anchorMeta;
+  if (position.leftPx === null && position.topPx === null) return baseAttrs;
   const styleParts = [
     "position:absolute",
-    anchorPos.leftPx !== null ? `left:${anchorPos.leftPx.toFixed(2)}px` : "",
-    anchorPos.topPx !== null ? `top:${anchorPos.topPx.toFixed(2)}px` : "",
-    "z-index:3"
+    position.leftPx !== null ? `left:${position.leftPx.toFixed(2)}px` : "",
+    position.topPx !== null ? `top:${position.topPx.toFixed(2)}px` : "",
+    `z-index:${anchorMeta.behindDoc ? 0 : anchorMeta.relativeHeight ?? 3}`,
+    anchorMeta.distTPx !== null ? `margin-top:${anchorMeta.distTPx.toFixed(2)}px` : "",
+    anchorMeta.distBPx !== null ? `margin-bottom:${anchorMeta.distBPx.toFixed(2)}px` : "",
+    anchorMeta.distLPx !== null ? `margin-left:${anchorMeta.distLPx.toFixed(2)}px` : "",
+    anchorMeta.distRPx !== null ? `margin-right:${anchorMeta.distRPx.toFixed(2)}px` : ""
   ].filter((x) => x.length > 0);
   if (wrapMode === "topAndBottom") {
-    styleParts.push("display:block");
-  }
+    styleParts.push("display:block", "clear:both");
+  }
+  const anchorAttrs = [
+    `data-word-anchor="1"`,
+    wrapMode ? `data-word-wrap="${wrapMode}"` : "",
+    anchorMeta.relativeFromH ? `data-word-anchor-relh="${escapeHtml(anchorMeta.relativeFromH)}"` : "",
+    anchorMeta.relativeFromV ? `data-word-anchor-relv="${escapeHtml(anchorMeta.relativeFromV)}"` : "",
+    anchorMeta.behindDoc ? `data-word-anchor-behind="1"` : `data-word-anchor-behind="0"`,
+    anchorMeta.allowOverlap ? `data-word-anchor-overlap="1"` : `data-word-anchor-overlap="0"`,
+    anchorMeta.layoutInCell ? `data-word-anchor-layout-cell="1"` : `data-word-anchor-layout-cell="0"`
+  ].filter((x) => x.length > 0).join(" ");
   if (!baseAttrs.includes("style=")) {
-    const wrapAttr = wrapMode ? ` data-word-wrap="${wrapMode}"` : "";
-    return `${baseAttrs} style="${styleParts.join(";")}" data-word-anchor="1"${wrapAttr}`;
+    return `${baseAttrs} style="${styleParts.join(";")}" ${anchorAttrs}`;
   }
   return baseAttrs.replace(/style="([^"]*)"/, (_m, styleText) => {
     const merged = [styleText, ...styleParts].filter((x) => x.length > 0).join(";");
-    const wrapAttr = wrapMode ? ` data-word-wrap="${wrapMode}"` : "";
-    return `style="${merged}" data-word-anchor="1"${wrapAttr}`;
+    return `style="${merged}" ${anchorAttrs}`;
   });
 }
 function parseDocRelsMap(relsXmlText) {
@@ -178,11 +242,29 @@ function extToMime(ext) {
   if (lower === "svg") return "image/svg+xml";
   return "application/octet-stream";
 }
+function normalizeWordPath(relTarget) {
+  const normalized = relTarget.replace(/\\/g, "/").replace(/^\/+/, "");
+  if (normalized.startsWith("word/")) return normalized;
+  if (normalized.startsWith("../")) return `word/${normalized.replace(/^(\.\.\/)+/, "")}`;
+  return `word/${normalized}`;
+}
+function resolveHyperlinkHref(relMap, rid, anchor) {
+  if (anchor && anchor.trim()) return `#${encodeURIComponent(anchor.trim())}`;
+  if (!rid) return null;
+  const relTarget = relMap[rid];
+  if (!relTarget) return null;
+  const trimmed = relTarget.trim();
+  if (!trimmed) return null;
+  const lower = trimmed.toLowerCase();
+  if (lower.startsWith("http://") || lower.startsWith("https://") || lower.startsWith("mailto:") || lower.startsWith("tel:")) {
+    return trimmed;
+  }
+  return trimmed.startsWith("#") ? trimmed : `#${encodeURIComponent(trimmed)}`;
+}
 async function imageRidToDataUrl(zip, relMap, rid) {
   const relTarget = relMap[rid];
   if (!relTarget) return null;
-  const normalized = relTarget.replace(/^\/+/, "");
-  const path = normalized.startsWith("word/") ? normalized : `word/${normalized}`;
+  const path = normalizeWordPath(relTarget);
   const file = zip.file(path);
   if (!file) return null;
   const base64 = await file.async("base64");
@@ -190,6 +272,55 @@ async function imageRidToDataUrl(zip, relMap, rid) {
   const mime = extToMime(ext);
   return `data:${mime};base64,${base64}`;
 }
+async function readXmlByRid(zip, relMap, rid) {
+  const relTarget = relMap[rid];
+  if (!relTarget) return null;
+  const path = normalizeWordPath(relTarget);
+  const file = zip.file(path);
+  return file ? file.async("string") : null;
+}
+function parseChartType(chartDoc) {
+  const known = ["barChart", "lineChart", "pieChart", "areaChart", "scatterChart", "radarChart", "doughnutChart"];
+  for (const type of known) {
+    if (queryByLocalName(chartDoc, type)) return type.replace(/Chart$/, "");
+  }
+  return "unknown";
+}
+function parseChartSummary(chartXmlText) {
+  const chartDoc = parseXml(chartXmlText);
+  const title = queryAllByLocalName(chartDoc, "t").map((n) => (n.textContent ?? "").trim()).find((v) => v.length > 0) ?? "Chart";
+  const seriesCount = queryAllByLocalName(chartDoc, "ser").length;
+  const pointCount = queryAllByLocalName(chartDoc, "pt").length;
+  const type = parseChartType(chartDoc);
+  return { title, type, seriesCount, pointCount };
+}
+function extractSmartArtText(diagramXmlText) {
+  const diagramDoc = parseXml(diagramXmlText);
+  return queryAllByLocalName(diagramDoc, "t").map((n) => (n.textContent ?? "").trim()).filter((v) => v.length > 0).slice(0, 12);
+}
+function ommlNodeToText(node) {
+  if (node.localName === "t") return node.textContent ?? "";
+  if (node.localName === "f") {
+    const num = queryByLocalName(node, "num");
+    const den = queryByLocalName(node, "den");
+    return `(${num ? ommlNodeToText(num) : "?"})/(${den ? ommlNodeToText(den) : "?"})`;
+  }
+  if (node.localName === "sSup") {
+    const e = queryByLocalName(node, "e");
+    const sup = queryByLocalName(node, "sup");
+    return `${e ? ommlNodeToText(e) : ""}^(${sup ? ommlNodeToText(sup) : ""})`;
+  }
+  if (node.localName === "sSub") {
+    const e = queryByLocalName(node, "e");
+    const sub = queryByLocalName(node, "sub");
+    return `${e ? ommlNodeToText(e) : ""}_(${sub ? ommlNodeToText(sub) : ""})`;
+  }
+  if (node.localName === "rad") {
+    const e = queryByLocalName(node, "e");
+    return `sqrt(${e ? ommlNodeToText(e) : ""})`;
+  }
+  return Array.from(node.children).map((child) => ommlNodeToText(child)).join("");
+}
 function runStyleToCss(rPr) {
   if (!rPr) return "";
   const declarations = [];
@@ -304,48 +435,73 @@ function renderEndnotesSection(usedIds, endnotesMap) {
   const items = uniq.map((id) => `<li id="word-endnote-${id}" data-word-endnote-id="${id}">${endnotesMap[id]}</li>`).join("");
   return `<section data-word-endnotes="1"><hr/><ol>${items}</ol></section>`;
 }
-async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotesMap, usedFootnoteIds, endnotesMap, usedEndnoteIds, commentsMap, usedCommentIds) {
+async function paragraphToHtml(zip, relMap, context, paragraph, paragraphIndex, footnotesMap, usedFootnoteIds, endnotesMap, usedEndnoteIds, commentsMap, usedCommentIds) {
   const tag = paragraphTag(paragraph);
   const alignStyle = paragraphAlignStyle(paragraph);
   const dataAttr = paragraphDataAttr(paragraphIndex);
-  const runs = queryAllByLocalName(paragraph, "r");
-  if (runs.length === 0) {
+  const hasRenderableNode = queryAllByLocalName(paragraph, "r").length > 0 || queryAllByLocalName(paragraph, "oMath").length > 0 || queryAllByLocalName(paragraph, "oMathPara").length > 0;
+  if (!hasRenderableNode) {
     return `<${tag}${dataAttr}${alignStyle ? ` style="${alignStyle}"` : ""}><br/></${tag}>`;
   }
-  const parts = [];
-  const renderedPageBreakCount = queryAllByLocalName(paragraph, "lastRenderedPageBreak").length;
-  for (let i = 0; i < renderedPageBreakCount; i += 1) {
-    parts.push(`<span data-word-page-break="1" style="display:block;break-before:page"></span>`);
+  function parseRevisionMeta(node, type) {
+    return {
+      type,
+      id: getAttr(node, "w:id") ?? getAttr(node, "id"),
+      author: getAttr(node, "w:author") ?? getAttr(node, "author"),
+      date: getAttr(node, "w:date") ?? getAttr(node, "date")
+    };
+  }
+  function inferRevisionMeta(run, fallback) {
+    if (fallback) return fallback;
+    let cursor = run;
+    while (cursor) {
+      if (cursor.localName === "ins") return parseRevisionMeta(cursor, "ins");
+      if (cursor.localName === "del") return parseRevisionMeta(cursor, "del");
+      if (cursor.localName === "p") break;
+      cursor = cursor.parentElement;
+    }
+    return null;
+  }
+  function revisionMetaAttrs(meta) {
+    const attrs = [`data-word-revision="${meta.type}"`];
+    if (meta.id) attrs.push(`data-word-revision-id="${escapeHtml(meta.id)}"`);
+    if (meta.author) attrs.push(`data-word-revision-author="${escapeHtml(meta.author)}"`);
+    if (meta.date) attrs.push(`data-word-revision-date="${escapeHtml(meta.date)}"`);
+    return attrs.join(" ");
   }
-  for (const run of runs) {
+  async function runToHtml(run, revisionFallback) {
+    const result = [];
     const rPr = queryByLocalName(run, "rPr");
     const css = runStyleToCss(rPr);
     const footnoteRef = queryByLocalName(run, "footnoteReference");
     const footnoteId = getAttr(footnoteRef, "w:id") ?? getAttr(footnoteRef, "id");
     if (footnoteId && footnotesMap[footnoteId]) {
+      context.features.footnoteRefCount += 1;
       usedFootnoteIds.push(footnoteId);
-      parts.push(
+      result.push(
         `<sup data-word-footnote-ref="${footnoteId}"><a href="#word-footnote-${footnoteId}">[${footnoteId}]</a></sup>`
       );
-      continue;
+      return result;
     }
     const endnoteRef = queryByLocalName(run, "endnoteReference");
     const endnoteId = getAttr(endnoteRef, "w:id") ?? getAttr(endnoteRef, "id");
     if (endnoteId && endnotesMap[endnoteId]) {
+      context.features.endnoteRefCount += 1;
       usedEndnoteIds.push(endnoteId);
-      parts.push(
+      result.push(
         `<sup data-word-endnote-ref="${endnoteId}"><a href="#word-endnote-${endnoteId}">[${endnoteId}]</a></sup>`
       );
-      continue;
+      return result;
     }
     const commentRef = queryByLocalName(run, "commentReference");
     const commentId = getAttr(commentRef, "w:id") ?? getAttr(commentRef, "id");
     if (commentId && commentsMap[commentId]) {
+      context.features.commentRefCount += 1;
       usedCommentIds.push(commentId);
-      parts.push(
+      result.push(
         `<sup data-word-comment-ref="${commentId}"><a href="#word-comment-${commentId}">[c${commentId}]</a></sup>`
       );
-      continue;
+      return result;
     }
     const drawing = queryByLocalName(run, "drawing");
     if (drawing) {
@@ -356,13 +512,38 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
         if (src) {
           const imageSize = parseDrawingSizePx(drawing);
           const dimensionAttrs = imageDimensionAttributes(imageSize);
-          const anchorPos = parseAnchorPositionPx(drawing);
-          const wrapMode = parseAnchorWrapMode(drawing);
-          const attrs = mergeImageStyle(dimensionAttrs, anchorPos, wrapMode);
-          parts.push(`<img src="${src}" alt="word-image"${attrs}/>`);
-          continue;
+          const anchorMeta = parseAnchorMeta(drawing);
+          const attrs = mergeImageStyle(dimensionAttrs, anchorMeta);
+          if (anchorMeta) context.features.anchorImageCount += 1;
+          result.push(`<img src="${src}" alt="word-image"${attrs}/>`);
+          return result;
+        }
+      }
+      const chartRef = queryByLocalName(drawing, "chart");
+      const chartRid = getAttr(chartRef, "r:id") ?? getAttr(chartRef, "id");
+      if (chartRid) {
+        const chartXmlText = await readXmlByRid(zip, relMap, chartRid);
+        if (chartXmlText) {
+          const summary = parseChartSummary(chartXmlText);
+          context.features.chartCount += 1;
+          result.push(
+            `<figure data-word-chart="1" data-word-chart-type="${summary.type}" data-word-chart-series="${summary.seriesCount}" data-word-chart-points="${summary.pointCount}"><figcaption>${escapeHtml(summary.title)}</figcaption><div>Chart(${escapeHtml(summary.type)}): series=${summary.seriesCount}, points=${summary.pointCount}</div></figure>`
+          );
+          return result;
         }
       }
+      const smartArtRef = queryByLocalName(drawing, "relIds");
+      const smartArtRid = getAttr(smartArtRef, "r:dm") ?? getAttr(smartArtRef, "dm");
+      if (smartArtRid) {
+        const diagramXmlText = await readXmlByRid(zip, relMap, smartArtRid);
+        const textItems = diagramXmlText ? extractSmartArtText(diagramXmlText) : [];
+        context.features.smartArtCount += 1;
+        const preview = textItems.length > 0 ? `: ${escapeHtml(textItems.join(" / "))}` : "";
+        result.push(
+          `<figure data-word-smartart="1" data-word-smartart-items="${textItems.length}"><figcaption>SmartArt fallback${preview}</figcaption></figure>`
+        );
+        return result;
+      }
     }
     const texts = queryAllByLocalName(run, "t").map((t) => t.textContent ?? "").join("");
     const delTexts = queryAllByLocalName(run, "delText").map((t) => t.textContent ?? "").join("");
@@ -373,40 +554,86 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
     }).length;
     const lineBreakCount = Math.max(0, brNodes.length - pageBreakCount);
     const runText2 = `${escapeHtml(texts || delTexts)}${"<br/>".repeat(lineBreakCount)}`;
-    if (!runText2) continue;
-    let revisionType = null;
-    let cursor = run;
-    while (cursor) {
-      if (cursor.localName === "ins") {
-        revisionType = "ins";
-        break;
-      }
-      if (cursor.localName === "del") {
-        revisionType = "del";
-        break;
+    if (runText2) {
+      const revisionMeta = inferRevisionMeta(run, revisionFallback);
+      if (css) {
+        const span = `<span style="${css}">${runText2}</span>`;
+        if (revisionMeta) {
+          context.features.revisionCount += 1;
+          const tagName = revisionMeta.type === "ins" ? "ins" : "del";
+          result.push(`<${tagName} ${revisionMetaAttrs(revisionMeta)}>${span}</${tagName}>`);
+        } else {
+          result.push(span);
+        }
+      } else if (revisionMeta) {
+        context.features.revisionCount += 1;
+        const tagName = revisionMeta.type === "ins" ? "ins" : "del";
+        result.push(`<${tagName} ${revisionMetaAttrs(revisionMeta)}>${runText2}</${tagName}>`);
+      } else {
+        result.push(runText2);
       }
-      if (cursor.localName === "p") break;
-      cursor = cursor.parentElement;
     }
-    if (css) {
-      const span = `<span style="${css}">${runText2}</span>`;
-      if (revisionType) {
-        const tag2 = revisionType === "ins" ? "ins" : "del";
-        parts.push(`<${tag2} data-word-revision="${revisionType}">${span}</${tag2}>`);
-      } else {
-        parts.push(span);
+    for (let i = 0; i < pageBreakCount; i += 1) {
+      context.features.pageBreakCount += 1;
+      result.push(`<span data-word-page-break="1" style="display:block;break-before:page"></span>`);
+    }
+    return result;
+  }
+  async function nodeToHtml(node, revisionFallback) {
+    if (node.localName === "commentRangeStart") {
+      const id = getAttr(node, "w:id") ?? getAttr(node, "id");
+      return id ? [`<span data-word-comment-range-start="${id}"></span>`] : [];
+    }
+    if (node.localName === "commentRangeEnd") {
+      const id = getAttr(node, "w:id") ?? getAttr(node, "id");
+      return id ? [`<span data-word-comment-range-end="${id}"></span>`] : [];
+    }
+    if (node.localName === "r") {
+      return runToHtml(node, revisionFallback);
+    }
+    if (node.localName === "hyperlink") {
+      const rid = getAttr(node, "r:id") ?? getAttr(node, "id");
+      const anchor = getAttr(node, "w:anchor") ?? getAttr(node, "anchor");
+      const href = resolveHyperlinkHref(relMap, rid, anchor);
+      const nested2 = [];
+      for (const child of Array.from(node.children)) {
+        nested2.push(...await nodeToHtml(child, revisionFallback));
       }
-    } else {
-      if (revisionType) {
-        const tag2 = revisionType === "ins" ? "ins" : "del";
-        parts.push(`<${tag2} data-word-revision="${revisionType}">${runText2}</${tag2}>`);
-      } else {
-        parts.push(runText2);
+      const content2 = nested2.join("") || escapeHtml(node.textContent ?? "");
+      if (!href) return content2 ? [content2] : [];
+      context.features.hyperlinkCount += 1;
+      return [
+        `<a data-word-hyperlink="1" href="${escapeHtml(href)}" rel="noreferrer noopener" target="_blank">${content2}</a>`
+      ];
+    }
+    if (node.localName === "oMath" || node.localName === "oMathPara") {
+      const linear = ommlNodeToText(node).trim();
+      if (!linear) return [];
+      context.features.ommlCount += 1;
+      return [`<span data-word-omml="1">${escapeHtml(linear)}</span>`];
+    }
+    if (node.localName === "ins" || node.localName === "del") {
+      const scopedMeta = parseRevisionMeta(node, node.localName === "ins" ? "ins" : "del");
+      const nested2 = [];
+      for (const child of Array.from(node.children)) {
+        nested2.push(...await nodeToHtml(child, scopedMeta));
       }
+      return nested2;
     }
-    for (let i = 0; i < pageBreakCount; i += 1) {
-      parts.push(`<span data-word-page-break="1" style="display:block;break-before:page"></span>`);
+    const nested = [];
+    for (const child of Array.from(node.children)) {
+      nested.push(...await nodeToHtml(child, revisionFallback));
     }
+    return nested;
+  }
+  const parts = [];
+  const renderedPageBreakCount = queryAllByLocalName(paragraph, "lastRenderedPageBreak").length;
+  for (let i = 0; i < renderedPageBreakCount; i += 1) {
+    context.features.pageBreakCount += 1;
+    parts.push(`<span data-word-page-break="1" style="display:block;break-before:page"></span>`);
+  }
+  for (const child of Array.from(paragraph.children)) {
+    parts.push(...await nodeToHtml(child, null));
   }
   const content = parts.join("") || "<br/>";
   return `<${tag}${dataAttr}${alignStyle ? ` style="${alignStyle}"` : ""}>${content}</${tag}>`;
@@ -436,7 +663,102 @@ function parseTcVMerge(tc) {
   const rawVal = (getAttr(vMerge, "w:val") ?? getAttr(vMerge, "val") ?? "continue").toLowerCase();
   return rawVal === "restart" ? "restart" : "continue";
 }
-function tableCellHtml(cell, paragraphIndexMap) {
+function parseTblGridWidthsPx(table) {
+  const grid = directChildrenByLocalName(table, "tblGrid")[0] ?? null;
+  if (!grid) return [];
+  return directChildrenByLocalName(grid, "gridCol").map((col) => {
+    const raw = getAttr(col, "w:w") ?? getAttr(col, "w");
+    const twip = raw ? Number.parseInt(raw, 10) : Number.NaN;
+    return Number.isFinite(twip) && twip > 0 ? twipToPx(twip) : 0;
+  }).filter((px) => px > 0);
+}
+function borderSizeToPx(size) {
+  return size / 6;
+}
+function parseBorderCss(borderNode) {
+  if (!borderNode) return null;
+  const val = (getAttr(borderNode, "w:val") ?? getAttr(borderNode, "val") ?? "").toLowerCase();
+  if (!val || val === "nil" || val === "none") return "none";
+  const color = (getAttr(borderNode, "w:color") ?? getAttr(borderNode, "color") ?? "222222").replace(/^#/, "");
+  const rawSize = getAttr(borderNode, "w:sz") ?? getAttr(borderNode, "sz");
+  const size = rawSize ? Number.parseInt(rawSize, 10) : Number.NaN;
+  const px = Number.isFinite(size) && size > 0 ? borderSizeToPx(size) : 1;
+  const style = val === "single" ? "solid" : val;
+  return `${px.toFixed(2)}px ${style} #${color}`;
+}
+function parseTableStyleProfile(table) {
+  const tblPr = directChildrenByLocalName(table, "tblPr")[0] ?? null;
+  const tblBorders = tblPr ? directChildrenByLocalName(tblPr, "tblBorders")[0] ?? null : null;
+  const layout = tblPr ? directChildrenByLocalName(tblPr, "tblLayout")[0] ?? null : null;
+  const spacing = tblPr ? directChildrenByLocalName(tblPr, "tblCellSpacing")[0] ?? null : null;
+  const spacingType = (getAttr(spacing, "w:type") ?? getAttr(spacing, "type") ?? "dxa").toLowerCase();
+  const spacingRaw = getAttr(spacing, "w:w") ?? getAttr(spacing, "w");
+  const spacingVal = spacingRaw ? Number.parseFloat(spacingRaw) : Number.NaN;
+  const borderSpacingPx = spacingType === "dxa" && Number.isFinite(spacingVal) && spacingVal > 0 ? twipToPx(spacingVal) : 0;
+  const borderCollapse = borderSpacingPx > 0 ? "separate" : "collapse";
+  const tableLayout = (getAttr(layout, "w:type") ?? getAttr(layout, "type") ?? "").toLowerCase() === "autofit" ? "auto" : "fixed";
+  const top = parseBorderCss(tblBorders ? directChildrenByLocalName(tblBorders, "top")[0] ?? null : null);
+  const bottom = parseBorderCss(tblBorders ? directChildrenByLocalName(tblBorders, "bottom")[0] ?? null : null);
+  const left = parseBorderCss(tblBorders ? directChildrenByLocalName(tblBorders, "left")[0] ?? null : null);
+  const right = parseBorderCss(tblBorders ? directChildrenByLocalName(tblBorders, "right")[0] ?? null : null);
+  const insideH = parseBorderCss(tblBorders ? directChildrenByLocalName(tblBorders, "insideH")[0] ?? null : null);
+  const insideV = parseBorderCss(tblBorders ? directChildrenByLocalName(tblBorders, "insideV")[0] ?? null : null);
+  const borderCss = top ?? right ?? bottom ?? left ?? "1px solid #222";
+  return {
+    tableLayout,
+    borderCollapse,
+    borderSpacingPx,
+    borderCss,
+    insideHCss: insideH,
+    insideVCss: insideV
+  };
+}
+function parseTableWidthStyle(table, gridWidthsPx) {
+  const tblPr = directChildrenByLocalName(table, "tblPr")[0] ?? null;
+  const tblW = tblPr ? directChildrenByLocalName(tblPr, "tblW")[0] ?? null : null;
+  const type = (getAttr(tblW, "w:type") ?? getAttr(tblW, "type") ?? "").toLowerCase();
+  const rawVal = getAttr(tblW, "w:w") ?? getAttr(tblW, "w");
+  const numericVal = rawVal ? Number.parseFloat(rawVal) : Number.NaN;
+  if (type === "dxa" && Number.isFinite(numericVal) && numericVal > 0) {
+    return `width:${twipToPx(numericVal).toFixed(2)}px`;
+  }
+  if (type === "pct" && Number.isFinite(numericVal) && numericVal > 0) {
+    return `width:${(numericVal / 50).toFixed(2)}%`;
+  }
+  const gridTotal = gridWidthsPx.reduce((sum, item) => sum + item, 0);
+  if (gridTotal > 0) return `width:${gridTotal.toFixed(2)}px;max-width:100%`;
+  return "width:100%";
+}
+function parseCellWidthStyle(cell, colCursor, colSpan, gridWidthsPx) {
+  const tcPr = directChildrenByLocalName(cell, "tcPr")[0] ?? null;
+  const tcW = tcPr ? directChildrenByLocalName(tcPr, "tcW")[0] ?? null : null;
+  const type = (getAttr(tcW, "w:type") ?? getAttr(tcW, "type") ?? "").toLowerCase();
+  const rawVal = getAttr(tcW, "w:w") ?? getAttr(tcW, "w");
+  const numericVal = rawVal ? Number.parseFloat(rawVal) : Number.NaN;
+  if (type === "dxa" && Number.isFinite(numericVal) && numericVal > 0) {
+    return `width:${twipToPx(numericVal).toFixed(2)}px`;
+  }
+  if (type === "pct" && Number.isFinite(numericVal) && numericVal > 0) {
+    return `width:${(numericVal / 50).toFixed(2)}%`;
+  }
+  const width = gridWidthsPx.slice(colCursor, colCursor + colSpan).reduce((sum, item) => sum + item, 0);
+  if (width > 0) return `width:${width.toFixed(2)}px`;
+  return "";
+}
+function parseCellBorderStyle(cell, tableStyle) {
+  const tcPr = directChildrenByLocalName(cell, "tcPr")[0] ?? null;
+  const tcBorders = tcPr ? directChildrenByLocalName(tcPr, "tcBorders")[0] ?? null : null;
+  if (!tcBorders) {
+    const fallback = tableStyle.insideHCss ?? tableStyle.insideVCss ?? tableStyle.borderCss;
+    return `border:${fallback}`;
+  }
+  const top = parseBorderCss(directChildrenByLocalName(tcBorders, "top")[0] ?? null) ?? tableStyle.insideHCss ?? tableStyle.borderCss;
+  const right = parseBorderCss(directChildrenByLocalName(tcBorders, "right")[0] ?? null) ?? tableStyle.insideVCss ?? tableStyle.borderCss;
+  const bottom = parseBorderCss(directChildrenByLocalName(tcBorders, "bottom")[0] ?? null) ?? tableStyle.insideHCss ?? tableStyle.borderCss;
+  const left = parseBorderCss(directChildrenByLocalName(tcBorders, "left")[0] ?? null) ?? tableStyle.insideVCss ?? tableStyle.borderCss;
+  return `border-top:${top};border-right:${right};border-bottom:${bottom};border-left:${left}`;
+}
+function tableCellHtml(cell, paragraphIndexMap, context) {
   const blocks = [];
   for (const child of Array.from(cell.children)) {
     if (child.localName === "tcPr") continue;
@@ -446,7 +768,7 @@ function tableCellHtml(cell, paragraphIndexMap) {
       continue;
     }
     if (child.localName === "tbl") {
-      blocks.push(tableToHtml(child, paragraphIndexMap));
+      blocks.push(tableToHtml(child, paragraphIndexMap, context));
       continue;
     }
   }
@@ -454,8 +776,11 @@ function tableCellHtml(cell, paragraphIndexMap) {
   const text = queryAllByLocalName(cell, "t").map((t) => t.textContent ?? "").join("").trim();
   return escapeHtml(text) || "<br/>";
 }
-function tableToHtml(table, paragraphIndexMap) {
+function tableToHtml(table, paragraphIndexMap, context) {
+  context.features.tableCount += 1;
   const rows = directChildrenByLocalName(table, "tr");
+  const gridWidthsPx = parseTblGridWidthsPx(table);
+  const tableStyle = parseTableStyleProfile(table);
   const activeByCol = /* @__PURE__ */ new Map();
   const allOrigins = [];
   let nextOriginId = 1;
@@ -480,8 +805,10 @@ function tableToHtml(table, paragraphIndexMap) {
       while (activeByCol.has(colCursor)) {
         colCursor += 1;
       }
-      const html = tableCellHtml(cell, paragraphIndexMap);
+      const html = tableCellHtml(cell, paragraphIndexMap, context);
       const attrs = [];
+      const widthStyle = parseCellWidthStyle(cell, colCursor, colSpan, gridWidthsPx);
+      const borderStyle = parseCellBorderStyle(cell, tableStyle);
       if (vMerge === "restart") {
         const origin = {
           id: `m${nextOriginId}`,
@@ -499,7 +826,7 @@ function tableToHtml(table, paragraphIndexMap) {
       }
       if (colSpan > 1) attrs.push(`colspan="${colSpan}"`);
       emittedCells.push(
-        `<td${attrs.length > 0 ? ` ${attrs.join(" ")}` : ""} style="border:1px solid #222;vertical-align:top;">${html}</td>`
+        `<td${attrs.length > 0 ? ` ${attrs.join(" ")}` : ""} style="${borderStyle};vertical-align:top;${widthStyle}">${html}</td>`
       );
       colCursor += colSpan;
     }
@@ -518,9 +845,13 @@ function tableToHtml(table, paragraphIndexMap) {
     const replacement = origin.rowSpan > 1 ? `rowspan="${origin.rowSpan}"` : "";
     merged = merged.replace(marker, replacement).replace(/\s{2,}/g, " ");
   }
-  return `<table style="border-collapse:collapse;table-layout:fixed;width:100%;border:1px solid #222;">${merged}</table>`;
+  const tableWidthStyle = parseTableWidthStyle(table, gridWidthsPx);
+  const spacing = tableStyle.borderSpacingPx > 0 ? `border-spacing:${tableStyle.borderSpacingPx.toFixed(2)}px;` : "";
+  return `<table style="border-collapse:${tableStyle.borderCollapse};${spacing}table-layout:${tableStyle.tableLayout};${tableWidthStyle};border:${tableStyle.borderCss};">${merged}</table>`;
 }
-async function parseDocxToHtmlSnapshot(file) {
+async function parseDocxToHtmlSnapshotWithReport(file) {
+  const startedAt = Date.now();
+  const context = { features: createEmptyFeatureCounts() };
   const maybeArrayBuffer = file.arrayBuffer;
   const buffer = maybeArrayBuffer ? await maybeArrayBuffer.call(file) : await new Response(file).arrayBuffer();
   const zip = await import_jszip.default.loadAsync(buffer);
@@ -557,6 +888,7 @@ async function parseDocxToHtmlSnapshot(file) {
         await paragraphToHtml(
           zip,
           relMap,
+          context,
           child,
           paragraphIndex,
           footnotesMap,
@@ -570,14 +902,24 @@ async function parseDocxToHtmlSnapshot(file) {
       continue;
     }
     if (child.localName === "tbl") {
-      blockHtml.push(tableToHtml(child, paragraphIndexMap));
+      blockHtml.push(tableToHtml(child, paragraphIndexMap, context));
       continue;
     }
   }
   blockHtml.push(renderFootnotesSection(usedFootnoteIds, footnotesMap));
   blockHtml.push(renderEndnotesSection(usedEndnoteIds, endnotesMap));
   blockHtml.push(renderCommentsSection(usedCommentIds, commentsMap));
-  return buildHtmlSnapshot(blockHtml.join("\n"));
+  return {
+    htmlSnapshot: buildHtmlSnapshot(blockHtml.join("\n")),
+    report: {
+      elapsedMs: Date.now() - startedAt,
+      features: context.features
+    }
+  };
+}
+async function parseDocxToHtmlSnapshot(file) {
+  const result = await parseDocxToHtmlSnapshotWithReport(file);
+  return result.htmlSnapshot;
 }
 
 // src/lib/pastePipeline.ts
@@ -808,7 +1150,7 @@ function createFallbackWordStyleProfile(sourceFileName = "snapshot") {
     paragraphProfiles: []
   };
 }
-function twipToPx(twip) {
+function twipToPx2(twip) {
   return twip / 15;
 }
 function getAttr2(node, attr) {
@@ -852,10 +1194,10 @@ function parsePageGeometry(documentXml) {
   const top = getTwipAttr(pgMar, "w:top") ?? getTwipAttr(pgMar, "top") ?? null;
   const bottom = getTwipAttr(pgMar, "w:bottom") ?? getTwipAttr(pgMar, "bottom") ?? null;
   return {
-    contentWidthPx: pageW === null ? null : twipToPx(pageW - left - right),
-    pageHeightPx: pageH === null ? null : twipToPx(pageH),
-    marginTopPx: top === null ? null : twipToPx(top),
-    marginBottomPx: bottom === null ? null : twipToPx(bottom)
+    contentWidthPx: pageW === null ? null : twipToPx2(pageW - left - right),
+    pageHeightPx: pageH === null ? null : twipToPx2(pageH),
+    marginTopPx: top === null ? null : twipToPx2(top),
+    marginBottomPx: bottom === null ? null : twipToPx2(bottom)
   };
 }
 function parseHeadingAlignFromDocument(documentXml) {
@@ -1028,15 +1370,15 @@ function parseParagraphProfiles(documentXml, numberingMap) {
       text,
       isEmpty: text.length === 0,
       align: parseParagraphAlign(paragraph),
-      beforePx: before === null ? null : twipToPx(before),
-      afterPx: after === null ? null : twipToPx(after),
+      beforePx: before === null ? null : twipToPx2(before),
+      afterPx: after === null ? null : twipToPx2(after),
       lineHeightRatio: line === null || lineHeightRule !== "auto" ? null : line / 240,
-      lineHeightPx: line === null || lineHeightRule === "auto" ? null : twipToPx(line),
+      lineHeightPx: line === null || lineHeightRule === "auto" ? null : twipToPx2(line),
       lineHeightRule,
-      indentLeftPx: left === null ? null : twipToPx(left),
-      indentRightPx: right === null ? null : twipToPx(right),
-      firstLinePx: firstLine === null ? null : twipToPx(firstLine),
-      hangingPx: hanging === null ? null : twipToPx(hanging),
+      indentLeftPx: left === null ? null : twipToPx2(left),
+      indentRightPx: right === null ? null : twipToPx2(right),
+      firstLinePx: firstLine === null ? null : twipToPx2(firstLine),
+      hangingPx: hanging === null ? null : twipToPx2(hanging),
       listNumId,
       listLevel,
       listFormat: listSpec?.numFmt ?? null,
@@ -1071,19 +1413,19 @@ function parseTableDefaults(stylesXml) {
   return {
     topPx: (() => {
       const v = getTwipAttr(top, "w:w") ?? getTwipAttr(top, "w") ?? null;
-      return v === null ? null : twipToPx(v);
+      return v === null ? null : twipToPx2(v);
     })(),
     leftPx: (() => {
       const v = getTwipAttr(left, "w:w") ?? getTwipAttr(left, "w") ?? null;
-      return v === null ? null : twipToPx(v);
+      return v === null ? null : twipToPx2(v);
     })(),
     bottomPx: (() => {
       const v = getTwipAttr(bottom, "w:w") ?? getTwipAttr(bottom, "w") ?? null;
-      return v === null ? null : twipToPx(v);
+      return v === null ? null : twipToPx2(v);
     })(),
     rightPx: (() => {
       const v = getTwipAttr(right, "w:w") ?? getTwipAttr(right, "w") ?? null;
-      return v === null ? null : twipToPx(v);
+      return v === null ? null : twipToPx2(v);
     })()
   };
 }
@@ -1181,9 +1523,9 @@ function parseDefaults(stylesXml) {
   const rawLineRule = (getAttr2(spacing, "w:lineRule") ?? getAttr2(spacing, "lineRule") ?? "auto").toLowerCase();
   const bodyLineHeightRule = rawLineRule === "exact" ? "exact" : rawLineRule === "atleast" ? "atLeast" : "auto";
   const bodyLineHeightRatio = line === null || bodyLineHeightRule !== "auto" ? null : line / 240;
-  const bodyLineHeightPx = line === null || bodyLineHeightRule === "auto" ? null : twipToPx(line);
+  const bodyLineHeightPx = line === null || bodyLineHeightRule === "auto" ? null : twipToPx2(line);
   const after = getTwipAttr(spacing, "w:after") ?? getTwipAttr(spacing, "after") ?? null;
-  const paragraphAfterPx = after === null ? null : twipToPx(after);
+  const paragraphAfterPx = after === null ? null : twipToPx2(after);
   return { bodyFontPx, bodyLineHeightRatio, bodyLineHeightPx, bodyLineHeightRule, paragraphAfterPx };
 }
 function parseHeading1Style(stylesXml) {
@@ -1708,7 +2050,7 @@ function applyWordRenderModel({ doc, styleProfile, showFormattingMarks }) {
 }
 
 // src/core/DocsWordElement.ts
-var VERSION = "0.1.2";
+var VERSION = "0.1.5";
 var MESSAGES = {
   zh: {
     readClipboard: "\u4ECE\u7CFB\u7EDF\u526A\u8D34\u677F\u8BFB\u53D6",
@@ -1851,15 +2193,15 @@ var DocsWordElement = class extends HTMLElement {
   }
   async applyDocx(file) {
     try {
-      const [snapshot, profile] = await Promise.all([
-        parseDocxToHtmlSnapshot(file),
+      const [parseResult, profile] = await Promise.all([
+        parseDocxToHtmlSnapshotWithReport(file),
         parseDocxStyleProfile(file)
       ]);
       this.styleProfile = profile;
-      this.htmlSnapshot = snapshot;
+      this.htmlSnapshot = parseResult.htmlSnapshot;
       this.renderSnapshot();
       this.setHint(MESSAGES[this.locale].loadedWord(profile.sourceFileName));
-      this.emitChange("upload", profile.sourceFileName);
+      this.emitChange("upload", profile.sourceFileName, parseResult.report);
     } catch (error) {
       this.emitError(error instanceof Error ? error.message : MESSAGES[this.locale].parseFailed);
     }
@@ -1919,8 +2261,10 @@ var DocsWordElement = class extends HTMLElement {
   renderSnapshot() {
     this.frame.srcdoc = this.htmlSnapshot;
   }
-  emitChange(source, fileName) {
-    this.dispatchEvent(new CustomEvent("docsjs-change", { detail: { htmlSnapshot: this.htmlSnapshot, source, fileName } }));
+  emitChange(source, fileName, parseReport) {
+    this.dispatchEvent(
+      new CustomEvent("docsjs-change", { detail: { htmlSnapshot: this.htmlSnapshot, source, fileName, parseReport } })
+    );
   }
   emitError(message) {
     this.dispatchEvent(new CustomEvent("docsjs-error", { detail: { message } }));
@@ -1976,6 +2320,9 @@ function collectSemanticStatsFromDocument(doc) {
     imageCount: countElements(doc, "img"),
     anchorImageCount: countElements(doc, 'img[data-word-anchor="1"]'),
     wrappedImageCount: countElements(doc, "img[data-word-wrap]"),
+    ommlCount: countElements(doc, "[data-word-omml]"),
+    chartCount: countElements(doc, "[data-word-chart]"),
+    smartArtCount: countElements(doc, "[data-word-smartart]"),
     listParagraphCount,
     commentRefCount: countElements(doc, "[data-word-comment-ref]"),
     revisionInsCount: countElements(doc, '[data-word-revision="ins"]'),
@@ -2006,7 +2353,7 @@ function clamp01(v) {
 }
 function calculateFidelityScore(expected, actual) {
   const structure = clamp01(
-    (ratioScore(actual.paragraphCount, expected.paragraphCount) + ratioScore(actual.headingCount, expected.headingCount) + ratioScore(actual.tableCount, expected.tableCount) + ratioScore(actual.tableCellCount, expected.tableCellCount) + ratioScore(actual.imageCount, expected.imageCount) + ratioScore(actual.listParagraphCount, expected.listParagraphCount)) / 6
+    (ratioScore(actual.paragraphCount, expected.paragraphCount) + ratioScore(actual.headingCount, expected.headingCount) + ratioScore(actual.tableCount, expected.tableCount) + ratioScore(actual.tableCellCount, expected.tableCellCount) + ratioScore(actual.imageCount, expected.imageCount) + ratioScore(actual.ommlCount, expected.ommlCount) + ratioScore(actual.chartCount, expected.chartCount) + ratioScore(actual.smartArtCount, expected.smartArtCount) + ratioScore(actual.listParagraphCount, expected.listParagraphCount)) / 9
   );
   const styleProxy = clamp01(ratioScore(actual.textCharCount, expected.textCharCount));
   const pagination = clamp01(ratioScore(actual.pageSpacerCount, expected.pageSpacerCount));
@@ -2019,6 +2366,8 @@ function calculateFidelityScore(expected, actual) {
   calculateFidelityScore,
   collectSemanticStatsFromDocument,
   collectSemanticStatsFromHtml,
-  defineDocsWordElement
+  defineDocsWordElement,
+  parseDocxToHtmlSnapshot,
+  parseDocxToHtmlSnapshotWithReport
 });
 //# sourceMappingURL=index.cjs.map