@coding01/docsjs 0.1.2 → 0.1.3

package/dist/index.cjs

@@ -31,8 +31,9 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var index_exports = {};
 __export(index_exports, {
   DocsWordElement: () => DocsWordElement,
-  WordFidelityEditorReact: () => WordFidelityEditorReact,
-  WordFidelityEditorVue: () => WordFidelityEditorVue,
+  calculateFidelityScore: () => calculateFidelityScore,
+  collectSemanticStatsFromDocument: () => collectSemanticStatsFromDocument,
+  collectSemanticStatsFromHtml: () => collectSemanticStatsFromHtml,
   defineDocsWordElement: () => defineDocsWordElement
 });
 module.exports = __toCommonJS(index_exports);
@@ -69,6 +70,9 @@ function queryAllByLocalName(root, localName) {
 function queryByLocalName(root, localName) {
   return queryAllByLocalName(root, localName)[0] ?? null;
 }
+function directChildrenByLocalName(node, localName) {
+  return Array.from(node.children).filter((child) => child.localName === localName);
+}
 function getAttr(node, name) {
   if (!node) return null;
   return node.getAttribute(name);
@@ -104,6 +108,53 @@ function imageDimensionAttributes(sizePx) {
   }
   return attrs.length > 0 ? ` ${attrs.join(" ")}` : "";
 }
+function parseAnchorPositionPx(drawing) {
+  const anchor = directChildrenByLocalName(drawing, "anchor")[0] ?? null;
+  if (!anchor) return { leftPx: null, topPx: null };
+  let leftPx = null;
+  let topPx = null;
+  const positionH = directChildrenByLocalName(anchor, "positionH")[0] ?? null;
+  const positionV = directChildrenByLocalName(anchor, "positionV")[0] ?? null;
+  const posH = positionH ? directChildrenByLocalName(positionH, "posOffset")[0] ?? null : null;
+  const posV = positionV ? directChildrenByLocalName(positionV, "posOffset")[0] ?? null : null;
+  const rawLeft = posH?.textContent?.trim() ?? "";
+  const rawTop = posV?.textContent?.trim() ?? "";
+  const left = rawLeft ? Number.parseFloat(rawLeft) : Number.NaN;
+  const top = rawTop ? Number.parseFloat(rawTop) : Number.NaN;
+  if (Number.isFinite(left)) leftPx = emuToPx(left);
+  if (Number.isFinite(top)) topPx = emuToPx(top);
+  return { leftPx, topPx };
+}
+function parseAnchorWrapMode(drawing) {
+  const anchor = directChildrenByLocalName(drawing, "anchor")[0] ?? null;
+  if (!anchor) return null;
+  if (directChildrenByLocalName(anchor, "wrapSquare")[0]) return "square";
+  if (directChildrenByLocalName(anchor, "wrapTight")[0]) return "tight";
+  if (directChildrenByLocalName(anchor, "wrapTopAndBottom")[0]) return "topAndBottom";
+  if (directChildrenByLocalName(anchor, "wrapNone")[0]) return "none";
+  return null;
+}
+function mergeImageStyle(baseAttrs, anchorPos, wrapMode) {
+  if (anchorPos.leftPx === null && anchorPos.topPx === null) return baseAttrs;
+  const styleParts = [
+    "position:absolute",
+    anchorPos.leftPx !== null ? `left:${anchorPos.leftPx.toFixed(2)}px` : "",
+    anchorPos.topPx !== null ? `top:${anchorPos.topPx.toFixed(2)}px` : "",
+    "z-index:3"
+  ].filter((x) => x.length > 0);
+  if (wrapMode === "topAndBottom") {
+    styleParts.push("display:block");
+  }
+  if (!baseAttrs.includes("style=")) {
+    const wrapAttr = wrapMode ? ` data-word-wrap="${wrapMode}"` : "";
+    return `${baseAttrs} style="${styleParts.join(";")}" data-word-anchor="1"${wrapAttr}`;
+  }
+  return baseAttrs.replace(/style="([^"]*)"/, (_m, styleText) => {
+    const merged = [styleText, ...styleParts].filter((x) => x.length > 0).join(";");
+    const wrapAttr = wrapMode ? ` data-word-wrap="${wrapMode}"` : "";
+    return `style="${merged}" data-word-anchor="1"${wrapAttr}`;
+  });
+}
 function parseDocRelsMap(relsXmlText) {
   if (!relsXmlText) return {};
   const rels = parseXml(relsXmlText);
@@ -179,7 +230,81 @@ function paragraphAlignStyle(paragraph) {
 function paragraphDataAttr(paragraphIndex) {
   return paragraphIndex === null ? "" : ` data-word-p-index="${paragraphIndex}"`;
 }
-async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex) {
+function parseFootnotesMap(footnotesXmlText) {
+  if (!footnotesXmlText) return {};
+  const footnotesDoc = parseXml(footnotesXmlText);
+  const map = {};
+  const footnotes = queryAllByLocalName(footnotesDoc, "footnote");
+  for (const footnote of footnotes) {
+    const idRaw = getAttr(footnote, "w:id") ?? getAttr(footnote, "id");
+    const idNum = idRaw ? Number.parseInt(idRaw, 10) : Number.NaN;
+    if (!Number.isFinite(idNum) || idNum <= 0) continue;
+    const paragraphs = queryAllByLocalName(footnote, "p");
+    const text = paragraphs.map((p) => paragraphText(p)).join("<br/>").trim();
+    if (!text) continue;
+    map[String(idNum)] = text;
+  }
+  return map;
+}
+function parseCommentsMap(commentsXmlText) {
+  if (!commentsXmlText) return {};
+  const commentsDoc = parseXml(commentsXmlText);
+  const map = {};
+  const comments = queryAllByLocalName(commentsDoc, "comment");
+  for (const comment of comments) {
+    const idRaw = getAttr(comment, "w:id") ?? getAttr(comment, "id");
+    if (!idRaw) continue;
+    const paragraphs = queryAllByLocalName(comment, "p");
+    const text = paragraphs.map((p) => paragraphText(p)).join("<br/>").trim();
+    if (!text) continue;
+    map[idRaw] = {
+      author: getAttr(comment, "w:author") ?? getAttr(comment, "author"),
+      date: getAttr(comment, "w:date") ?? getAttr(comment, "date"),
+      text
+    };
+  }
+  return map;
+}
+function parseEndnotesMap(endnotesXmlText) {
+  if (!endnotesXmlText) return {};
+  const endnotesDoc = parseXml(endnotesXmlText);
+  const map = {};
+  const endnotes = queryAllByLocalName(endnotesDoc, "endnote");
+  for (const endnote of endnotes) {
+    const idRaw = getAttr(endnote, "w:id") ?? getAttr(endnote, "id");
+    const idNum = idRaw ? Number.parseInt(idRaw, 10) : Number.NaN;
+    if (!Number.isFinite(idNum) || idNum <= 0) continue;
+    const paragraphs = queryAllByLocalName(endnote, "p");
+    const text = paragraphs.map((p) => paragraphText(p)).join("<br/>").trim();
+    if (!text) continue;
+    map[String(idNum)] = text;
+  }
+  return map;
+}
+function renderFootnotesSection(usedIds, footnotesMap) {
+  const uniq = [...new Set(usedIds)].filter((id) => footnotesMap[id]);
+  if (uniq.length === 0) return "";
+  const items = uniq.map((id) => `<li id="word-footnote-${id}" data-word-footnote-id="${id}">${footnotesMap[id]}</li>`).join("");
+  return `<section data-word-footnotes="1"><hr/><ol>${items}</ol></section>`;
+}
+function renderCommentsSection(usedIds, commentsMap) {
+  const uniq = [...new Set(usedIds)].filter((id) => commentsMap[id]);
+  if (uniq.length === 0) return "";
+  const items = uniq.map((id) => {
+    const item = commentsMap[id];
+    const meta = [item.author ?? "", item.date ?? ""].filter((x) => x.length > 0).join(" \xB7 ");
+    const metaHtml = meta ? `<div data-word-comment-meta="1">${escapeHtml(meta)}</div>` : "";
+    return `<li id="word-comment-${id}" data-word-comment-id="${id}">${metaHtml}<div>${item.text}</div></li>`;
+  }).join("");
+  return `<section data-word-comments="1"><hr/><ol>${items}</ol></section>`;
+}
+function renderEndnotesSection(usedIds, endnotesMap) {
+  const uniq = [...new Set(usedIds)].filter((id) => endnotesMap[id]);
+  if (uniq.length === 0) return "";
+  const items = uniq.map((id) => `<li id="word-endnote-${id}" data-word-endnote-id="${id}">${endnotesMap[id]}</li>`).join("");
+  return `<section data-word-endnotes="1"><hr/><ol>${items}</ol></section>`;
+}
+async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotesMap, usedFootnoteIds, endnotesMap, usedEndnoteIds, commentsMap, usedCommentIds) {
   const tag = paragraphTag(paragraph);
   const alignStyle = paragraphAlignStyle(paragraph);
   const dataAttr = paragraphDataAttr(paragraphIndex);
@@ -188,9 +313,40 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex) {
     return `<${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>`);
+  }
   for (const run of runs) {
     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]) {
+      usedFootnoteIds.push(footnoteId);
+      parts.push(
+        `<sup data-word-footnote-ref="${footnoteId}"><a href="#word-footnote-${footnoteId}">[${footnoteId}]</a></sup>`
+      );
+      continue;
+    }
+    const endnoteRef = queryByLocalName(run, "endnoteReference");
+    const endnoteId = getAttr(endnoteRef, "w:id") ?? getAttr(endnoteRef, "id");
+    if (endnoteId && endnotesMap[endnoteId]) {
+      usedEndnoteIds.push(endnoteId);
+      parts.push(
+        `<sup data-word-endnote-ref="${endnoteId}"><a href="#word-endnote-${endnoteId}">[${endnoteId}]</a></sup>`
+      );
+      continue;
+    }
+    const commentRef = queryByLocalName(run, "commentReference");
+    const commentId = getAttr(commentRef, "w:id") ?? getAttr(commentRef, "id");
+    if (commentId && commentsMap[commentId]) {
+      usedCommentIds.push(commentId);
+      parts.push(
+        `<sup data-word-comment-ref="${commentId}"><a href="#word-comment-${commentId}">[c${commentId}]</a></sup>`
+      );
+      continue;
+    }
     const drawing = queryByLocalName(run, "drawing");
     if (drawing) {
       const blip = queryByLocalName(drawing, "blip");
@@ -200,19 +356,56 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex) {
         if (src) {
           const imageSize = parseDrawingSizePx(drawing);
           const dimensionAttrs = imageDimensionAttributes(imageSize);
-          parts.push(`<img src="${src}" alt="word-image"${dimensionAttrs}/>`);
+          const anchorPos = parseAnchorPositionPx(drawing);
+          const wrapMode = parseAnchorWrapMode(drawing);
+          const attrs = mergeImageStyle(dimensionAttrs, anchorPos, wrapMode);
+          parts.push(`<img src="${src}" alt="word-image"${attrs}/>`);
           continue;
         }
       }
     }
     const texts = queryAllByLocalName(run, "t").map((t) => t.textContent ?? "").join("");
-    const brs = queryAllByLocalName(run, "br").length;
-    const runText2 = `${escapeHtml(texts)}${"<br/>".repeat(brs)}`;
+    const delTexts = queryAllByLocalName(run, "delText").map((t) => t.textContent ?? "").join("");
+    const brNodes = queryAllByLocalName(run, "br");
+    const pageBreakCount = brNodes.filter((node) => {
+      const type = (getAttr(node, "w:type") ?? getAttr(node, "type") ?? "").toLowerCase();
+      return type === "page";
+    }).length;
+    const lineBreakCount = Math.max(0, brNodes.length - pageBreakCount);
+    const runText2 = `${escapeHtml(texts || delTexts)}${"<br/>".repeat(lineBreakCount)}`;
     if (!runText2) continue;
+    let revisionType = null;
+    let cursor = run;
+    while (cursor) {
+      if (cursor.localName === "ins") {
+        revisionType = "ins";
+        break;
+      }
+      if (cursor.localName === "del") {
+        revisionType = "del";
+        break;
+      }
+      if (cursor.localName === "p") break;
+      cursor = cursor.parentElement;
+    }
     if (css) {
-      parts.push(`<span style="${css}">${runText2}</span>`);
+      const span = `<span style="${css}">${runText2}</span>`;
+      if (revisionType) {
+        const tag2 = revisionType === "ins" ? "ins" : "del";
+        parts.push(`<${tag2} data-word-revision="${revisionType}">${span}</${tag2}>`);
+      } else {
+        parts.push(span);
+      }
     } else {
-      parts.push(runText2);
+      if (revisionType) {
+        const tag2 = revisionType === "ins" ? "ins" : "del";
+        parts.push(`<${tag2} data-word-revision="${revisionType}">${runText2}</${tag2}>`);
+      } else {
+        parts.push(runText2);
+      }
+    }
+    for (let i = 0; i < pageBreakCount; i += 1) {
+      parts.push(`<span data-word-page-break="1" style="display:block;break-before:page"></span>`);
     }
   }
   const content = parts.join("") || "<br/>";
@@ -220,33 +413,112 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex) {
 }
 function runText(run) {
   const text = queryAllByLocalName(run, "t").map((t) => t.textContent ?? "").join("");
+  const delText = queryAllByLocalName(run, "delText").map((t) => t.textContent ?? "").join("");
   const brCount = queryAllByLocalName(run, "br").length;
-  return `${escapeHtml(text)}${"<br/>".repeat(brCount)}`;
+  return `${escapeHtml(text || delText)}${"<br/>".repeat(brCount)}`;
 }
 function paragraphText(paragraph) {
   const runs = queryAllByLocalName(paragraph, "r");
   const content = runs.map((run) => runText(run)).join("");
   return content || "<br/>";
 }
+function parseTcGridSpan(tc) {
+  const tcPr = directChildrenByLocalName(tc, "tcPr")[0] ?? null;
+  const gridSpan = tcPr ? directChildrenByLocalName(tcPr, "gridSpan")[0] ?? null : null;
+  const rawVal = getAttr(gridSpan, "w:val") ?? getAttr(gridSpan, "val");
+  const parsed = rawVal ? Number.parseInt(rawVal, 10) : Number.NaN;
+  return Number.isFinite(parsed) && parsed > 0 ? parsed : 1;
+}
+function parseTcVMerge(tc) {
+  const tcPr = directChildrenByLocalName(tc, "tcPr")[0] ?? null;
+  const vMerge = tcPr ? directChildrenByLocalName(tcPr, "vMerge")[0] ?? null : null;
+  if (!vMerge) return "none";
+  const rawVal = (getAttr(vMerge, "w:val") ?? getAttr(vMerge, "val") ?? "continue").toLowerCase();
+  return rawVal === "restart" ? "restart" : "continue";
+}
 function tableCellHtml(cell, paragraphIndexMap) {
-  const paragraphs = queryAllByLocalName(cell, "p");
-  if (paragraphs.length === 0) {
-    const text = queryAllByLocalName(cell, "t").map((t) => t.textContent ?? "").join("").trim();
-    return escapeHtml(text) || "<br/>";
+  const blocks = [];
+  for (const child of Array.from(cell.children)) {
+    if (child.localName === "tcPr") continue;
+    if (child.localName === "p") {
+      const paragraphIndex = paragraphIndexMap.get(child) ?? null;
+      blocks.push(`<p${paragraphDataAttr(paragraphIndex)}>${paragraphText(child)}</p>`);
+      continue;
+    }
+    if (child.localName === "tbl") {
+      blocks.push(tableToHtml(child, paragraphIndexMap));
+      continue;
+    }
   }
-  return paragraphs.map((p) => {
-    const paragraphIndex = paragraphIndexMap.get(p) ?? null;
-    return `<p${paragraphDataAttr(paragraphIndex)}>${paragraphText(p)}</p>`;
-  }).join("");
+  if (blocks.length > 0) return blocks.join("");
+  const text = queryAllByLocalName(cell, "t").map((t) => t.textContent ?? "").join("").trim();
+  return escapeHtml(text) || "<br/>";
 }
 function tableToHtml(table, paragraphIndexMap) {
-  const rows = queryAllByLocalName(table, "tr");
-  const htmlRows = rows.map((row) => {
-    const cells = queryAllByLocalName(row, "tc");
-    const htmlCells = cells.map((cell) => `<td style="border:1px solid #222;vertical-align:top;">${tableCellHtml(cell, paragraphIndexMap)}</td>`).join("");
-    return `<tr>${htmlCells}</tr>`;
+  const rows = directChildrenByLocalName(table, "tr");
+  const activeByCol = /* @__PURE__ */ new Map();
+  const allOrigins = [];
+  let nextOriginId = 1;
+  const htmlRows = rows.map((row, rowIndex) => {
+    const directCells = directChildrenByLocalName(row, "tc");
+    const continued = /* @__PURE__ */ new Set();
+    const emittedCells = [];
+    let colCursor = 0;
+    for (const cell of directCells) {
+      const colSpan = parseTcGridSpan(cell);
+      const vMerge = parseTcVMerge(cell);
+      if (vMerge === "continue") {
+        const activeOrigins = Array.from(new Set(activeByCol.values())).filter((origin2) => !continued.has(origin2)).sort((a, b) => a.startCol - b.startCol);
+        const origin = activeOrigins.find((item) => item.startCol >= colCursor) ?? activeOrigins[0] ?? null;
+        if (origin) {
+          origin.rowSpan += 1;
+          continued.add(origin);
+          colCursor = origin.startCol + origin.colSpan;
+        }
+        continue;
+      }
+      while (activeByCol.has(colCursor)) {
+        colCursor += 1;
+      }
+      const html = tableCellHtml(cell, paragraphIndexMap);
+      const attrs = [];
+      if (vMerge === "restart") {
+        const origin = {
+          id: `m${nextOriginId}`,
+          startCol: colCursor,
+          colSpan,
+          rowSpan: 1,
+          startedRow: rowIndex
+        };
+        nextOriginId += 1;
+        allOrigins.push(origin);
+        for (let i = 0; i < colSpan; i += 1) {
+          activeByCol.set(colCursor + i, origin);
+        }
+        attrs.push(`data-word-merge-id="${origin.id}"`);
+      }
+      if (colSpan > 1) attrs.push(`colspan="${colSpan}"`);
+      emittedCells.push(
+        `<td${attrs.length > 0 ? ` ${attrs.join(" ")}` : ""} style="border:1px solid #222;vertical-align:top;">${html}</td>`
+      );
+      colCursor += colSpan;
+    }
+    for (const origin of Array.from(new Set(activeByCol.values()))) {
+      if (origin.startedRow < rowIndex && !continued.has(origin)) {
+        for (let i = 0; i < origin.colSpan; i += 1) {
+          activeByCol.delete(origin.startCol + i);
+        }
+      }
+    }
+    return `<tr>${emittedCells.join("")}</tr>`;
   });
-  return `<table style="border-collapse:collapse;table-layout:fixed;width:100%;border:1px solid #222;">${htmlRows.join("")}</table>`;
+  let merged = htmlRows.join("");
+  for (const origin of allOrigins) {
+    const marker = `data-word-merge-id="${origin.id}"`;
+    const replacement = origin.rowSpan > 1 ? `rowspan="${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>`;
 }
 async function parseDocxToHtmlSnapshot(file) {
   const maybeArrayBuffer = file.arrayBuffer;
@@ -257,7 +529,16 @@ async function parseDocxToHtmlSnapshot(file) {
     throw new Error("DOCX missing document.xml");
   }
   const relsText = await zip.file("word/_rels/document.xml.rels")?.async("string");
+  const footnotesText = await zip.file("word/footnotes.xml")?.async("string");
+  const endnotesText = await zip.file("word/endnotes.xml")?.async("string");
+  const commentsText = await zip.file("word/comments.xml")?.async("string");
   const relMap = parseDocRelsMap(relsText ?? null);
+  const footnotesMap = parseFootnotesMap(footnotesText ?? null);
+  const endnotesMap = parseEndnotesMap(endnotesText ?? null);
+  const commentsMap = parseCommentsMap(commentsText ?? null);
+  const usedFootnoteIds = [];
+  const usedEndnoteIds = [];
+  const usedCommentIds = [];
   const documentXml = parseXml(documentXmlText);
   const body = queryByLocalName(documentXml, "body");
   if (!body) {
@@ -272,7 +553,20 @@ async function parseDocxToHtmlSnapshot(file) {
     if (child.localName === "sectPr") continue;
     if (child.localName === "p") {
       const paragraphIndex = paragraphIndexMap.get(child) ?? null;
-      blockHtml.push(await paragraphToHtml(zip, relMap, child, paragraphIndex));
+      blockHtml.push(
+        await paragraphToHtml(
+          zip,
+          relMap,
+          child,
+          paragraphIndex,
+          footnotesMap,
+          usedFootnoteIds,
+          endnotesMap,
+          usedEndnoteIds,
+          commentsMap,
+          usedCommentIds
+        )
+      );
       continue;
     }
     if (child.localName === "tbl") {
@@ -280,6 +574,9 @@ async function parseDocxToHtmlSnapshot(file) {
       continue;
     }
   }
+  blockHtml.push(renderFootnotesSection(usedFootnoteIds, footnotesMap));
+  blockHtml.push(renderEndnotesSection(usedEndnoteIds, endnotesMap));
+  blockHtml.push(renderCommentsSection(usedCommentIds, commentsMap));
   return buildHtmlSnapshot(blockHtml.join("\n"));
 }
 
@@ -672,7 +969,28 @@ function parseNumberingMap(numberingXml) {
     const lvlMap = abstractMap.get(absId);
     if (!lvlMap) continue;
     for (const [lvl, spec] of lvlMap.entries()) {
-      levelMap.set(`${numId}:${lvl}`, spec);
+      levelMap.set(`${numId}:${lvl}`, { ...spec });
+    }
+    const lvlOverrides = queryAllByLocalName2(num, "lvlOverride");
+    for (const override of lvlOverrides) {
+      const ilvl = toInt(getAttr2(override, "w:ilvl") ?? getAttr2(override, "ilvl"));
+      if (ilvl === null) continue;
+      const key = `${numId}:${ilvl}`;
+      const base = levelMap.get(key) ?? { numFmt: null, lvlText: null, startAt: 1 };
+      const overrideStart = toInt(
+        getAttr2(queryAllByLocalName2(override, "startOverride")[0] ?? null, "w:val") ?? getAttr2(queryAllByLocalName2(override, "startOverride")[0] ?? null, "val")
+      );
+      const overrideLvl = queryAllByLocalName2(override, "lvl")[0] ?? null;
+      const overrideNumFmtNode = overrideLvl ? queryAllByLocalName2(overrideLvl, "numFmt")[0] ?? null : null;
+      const overrideLvlTextNode = overrideLvl ? queryAllByLocalName2(overrideLvl, "lvlText")[0] ?? null : null;
+      const overrideLvlStart = toInt(
+        getAttr2(queryAllByLocalName2(overrideLvl ?? override, "start")[0] ?? null, "w:val") ?? getAttr2(queryAllByLocalName2(overrideLvl ?? override, "start")[0] ?? null, "val")
+      );
+      levelMap.set(key, {
+        numFmt: getAttr2(overrideNumFmtNode, "w:val") ?? getAttr2(overrideNumFmtNode, "val") ?? base.numFmt,
+        lvlText: getAttr2(overrideLvlTextNode, "w:val") ?? getAttr2(overrideLvlTextNode, "val") ?? base.lvlText,
+        startAt: overrideStart ?? overrideLvlStart ?? base.startAt
+      });
     }
   }
   return levelMap;
@@ -918,7 +1236,8 @@ function inferTitleFontFamily(families) {
   return FALLBACK_PROFILE.titleFontFamily;
 }
 async function parseDocxStyleProfile(file) {
-  const buffer = await file.arrayBuffer();
+  const maybeArrayBuffer = file.arrayBuffer;
+  const buffer = maybeArrayBuffer ? await maybeArrayBuffer.call(file) : await new Response(file).arrayBuffer();
   const zip = await import_jszip2.default.loadAsync(buffer);
   const documentXmlText = await zip.file("word/document.xml")?.async("string");
   const stylesXmlText = await zip.file("word/styles.xml")?.async("string");
@@ -1280,6 +1599,7 @@ function applyParagraphProfiles(doc, styleProfile) {
       if (!alreadyHasMarker) {
         const marker = doc.createElement("span");
         marker.className = "__word-list-marker";
+        marker.setAttribute("data-word-list-marker", "1");
         marker.textContent = `${markerText} `;
         marker.style.display = "inline-block";
         marker.style.minWidth = "1.8em";
@@ -1331,7 +1651,7 @@ function applyKeepPagination(doc, styleProfile, paragraphs) {
   for (let i = 0; i < count; i += 1) {
     const p = paragraphs[i];
     const profile = styleProfile.paragraphProfiles[i];
-    const h2 = paragraphHeightPx(p);
+    const h = paragraphHeightPx(p);
     const forceBreak = profile.pageBreakBefore;
     if (forceBreak && used > 0) {
       insertPageSpacerBefore(doc, p, contentHeight - used);
@@ -1342,11 +1662,11 @@ function applyKeepPagination(doc, styleProfile, paragraphs) {
       insertPageSpacerBefore(doc, p, contentHeight - used);
       used = 0;
     }
-    if (used > 0 && used + h2 > contentHeight) {
+    if (used > 0 && used + h > contentHeight) {
       insertPageSpacerBefore(doc, p, contentHeight - used);
       used = 0;
     }
-    used += h2;
+    used += h;
     if (used >= contentHeight) {
       used = used % contentHeight;
     }
@@ -1388,6 +1708,41 @@ function applyWordRenderModel({ doc, styleProfile, showFormattingMarks }) {
 }
 
 // src/core/DocsWordElement.ts
+var VERSION = "0.1.2";
+var MESSAGES = {
+  zh: {
+    readClipboard: "\u4ECE\u7CFB\u7EDF\u526A\u8D34\u677F\u8BFB\u53D6",
+    uploadWord: "\u4E0A\u4F20 Word",
+    clear: "\u6E05\u7A7A",
+    pastePlaceholder: "\u5728\u6B64\u5904\u7C98\u8D34 Word/WPS/Google Docs \u5185\u5BB9\uFF08Ctrl/Cmd+V\uFF09",
+    waitImport: "\u7B49\u5F85\u5185\u5BB9\u5BFC\u5165",
+    loadedHtml: "\u5DF2\u52A0\u8F7D HTML \u5FEB\u7167",
+    cleared: "\u6587\u6863\u5DF2\u6E05\u7A7A",
+    loadedWord: (name) => `\u5DF2\u52A0\u8F7D Word \u6587\u4EF6: ${name}`,
+    importedClipboard: "\u5DF2\u5BFC\u5165\u526A\u8D34\u677F\u5185\u5BB9",
+    noContent: "\u672A\u68C0\u6D4B\u5230\u53EF\u5BFC\u5165\u5185\u5BB9",
+    noClipboardRead: "\u5F53\u524D\u6D4F\u89C8\u5668\u4E0D\u652F\u6301 clipboard.read",
+    parseFailed: "Word \u89E3\u6790\u5931\u8D25",
+    clipboardReadFailed: "\u8BFB\u53D6\u526A\u8D34\u677F\u5931\u8D25",
+    errorPrefix: "\u9519\u8BEF: "
+  },
+  en: {
+    readClipboard: "Read clipboard",
+    uploadWord: "Upload Word",
+    clear: "Clear",
+    pastePlaceholder: "Paste Word/WPS/Google Docs content here (Ctrl/Cmd+V)",
+    waitImport: "Waiting for input",
+    loadedHtml: "HTML snapshot loaded",
+    cleared: "Document cleared",
+    loadedWord: (name) => `Word file loaded: ${name}`,
+    importedClipboard: "Clipboard content imported",
+    noContent: "No importable content detected",
+    noClipboardRead: "navigator.clipboard.read is not supported in this browser",
+    parseFailed: "Word parse failed",
+    clipboardReadFailed: "Failed to read clipboard",
+    errorPrefix: "Error: "
+  }
+};
 var BASE_CSS = `
 :host{display:block;border:1px solid #d8deea;border-radius:12px;background:#fff;overflow:hidden;font-family:ui-sans-serif,system-ui,-apple-system,Segoe UI,Roboto}
 .toolbar{display:flex;gap:8px;flex-wrap:wrap;padding:10px;border-bottom:1px solid #e8edf6;background:#f8faff}
@@ -1398,6 +1753,10 @@ iframe{width:100%;min-height:760px;border:0}
 `;
 var DocsWordElement = class extends HTMLElement {
   rootRef;
+  toolbar;
+  btnRead;
+  btnUpload;
+  btnClear;
   frame;
   pasteArea;
   fileInput;
@@ -1405,64 +1764,92 @@ var DocsWordElement = class extends HTMLElement {
   htmlSnapshot;
   styleProfile = null;
   frameHeight = 0;
+  locale = "zh";
   constructor() {
     super();
     this.rootRef = this.attachShadow({ mode: "open" });
+    this.locale = this.parseLocale(this.getAttribute("lang"));
     this.htmlSnapshot = buildHtmlSnapshot("<p><br/></p>");
     const style = document.createElement("style");
     style.textContent = BASE_CSS;
-    const toolbar = document.createElement("div");
-    toolbar.className = "toolbar";
-    const btnRead = document.createElement("button");
-    btnRead.textContent = "\u4ECE\u7CFB\u7EDF\u526A\u8D34\u677F\u8BFB\u53D6";
-    btnRead.onclick = () => void this.readClipboard();
-    const btnUpload = document.createElement("button");
-    btnUpload.textContent = "\u4E0A\u4F20 Word";
-    btnUpload.onclick = () => this.fileInput.click();
-    const btnClear = document.createElement("button");
-    btnClear.textContent = "\u6E05\u7A7A";
-    btnClear.onclick = () => this.clear();
+    this.toolbar = document.createElement("div");
+    this.toolbar.className = "toolbar";
+    this.btnRead = document.createElement("button");
+    this.btnRead.onclick = () => void this.loadClipboard();
+    this.btnUpload = document.createElement("button");
+    this.btnUpload.onclick = () => this.fileInput.click();
+    this.btnClear = document.createElement("button");
+    this.btnClear.onclick = () => this.clear();
     this.fileInput = document.createElement("input");
     this.fileInput.type = "file";
     this.fileInput.accept = ".docx";
     this.fileInput.style.display = "none";
     this.fileInput.onchange = () => void this.onUpload();
-    toolbar.append(btnRead, btnUpload, btnClear, this.fileInput);
+    this.toolbar.append(this.btnRead, this.btnUpload, this.btnClear, this.fileInput);
     this.pasteArea = document.createElement("textarea");
     this.pasteArea.className = "paste";
-    this.pasteArea.placeholder = "\u5728\u6B64\u5904\u7C98\u8D34 Word/WPS/Google Docs \u5185\u5BB9\uFF08Ctrl/Cmd+V\uFF09";
+    this.pasteArea.placeholder = "";
     this.pasteArea.onpaste = (event) => {
       event.preventDefault();
       void this.applyFromClipboardData(event.clipboardData);
     };
     this.hint = document.createElement("span");
     this.hint.className = "hint";
-    this.hint.textContent = "\u7B49\u5F85\u5185\u5BB9\u5BFC\u5165";
+    this.hint.textContent = "";
     this.frame = document.createElement("iframe");
     this.frame.sandbox.add("allow-same-origin", "allow-scripts");
     this.frame.onload = () => this.onFrameLoad();
-    this.rootRef.append(style, toolbar, this.pasteArea, this.hint, this.frame);
+    this.rootRef.append(style, this.toolbar, this.pasteArea, this.hint, this.frame);
+    this.syncLocaleText();
+    this.syncToolbarVisibility();
+  }
+  static get observedAttributes() {
+    return ["lang", "show-toolbar"];
+  }
+  attributeChangedCallback(name, _, newValue) {
+    if (name === "lang") {
+      this.locale = this.parseLocale(newValue);
+      this.syncLocaleText();
+      return;
+    }
+    if (name === "show-toolbar") {
+      this.syncToolbarVisibility();
+    }
   }
   connectedCallback() {
     this.renderSnapshot();
+    this.dispatchEvent(new CustomEvent("docsjs-ready", { detail: { version: VERSION } }));
   }
   setSnapshot(rawHtml) {
+    this.loadHtml(rawHtml);
+  }
+  loadHtml(rawHtml) {
     this.styleProfile = null;
     this.htmlSnapshot = buildHtmlSnapshot(rawHtml);
     this.renderSnapshot();
-    this.hint.textContent = "\u5DF2\u52A0\u8F7D HTML \u5FEB\u7167";
-    this.emitChange();
+    this.setHint(MESSAGES[this.locale].loadedHtml);
+    this.emitChange("api");
+  }
+  getSnapshot() {
+    return this.htmlSnapshot;
   }
   clear() {
     this.styleProfile = null;
     this.htmlSnapshot = buildHtmlSnapshot("<p><br/></p>");
     this.renderSnapshot();
-    this.hint.textContent = "\u6587\u6863\u5DF2\u6E05\u7A7A";
-    this.emitChange();
+    this.setHint(MESSAGES[this.locale].cleared);
+    this.emitChange("clear");
+  }
+  async loadDocx(file) {
+    await this.applyDocx(file);
   }
   async onUpload() {
     const file = this.fileInput.files?.[0];
     if (!file) return;
+    await this.applyDocx(file);
+    this.fileInput.value = "";
+  }
+  async applyDocx(file) {
     try {
       const [snapshot, profile] = await Promise.all([
         parseDocxToHtmlSnapshot(file),
@@ -1471,17 +1858,15 @@ var DocsWordElement = class extends HTMLElement {
       this.styleProfile = profile;
       this.htmlSnapshot = snapshot;
       this.renderSnapshot();
-      this.hint.textContent = `\u5DF2\u52A0\u8F7D Word \u6587\u4EF6: ${profile.sourceFileName}`;
-      this.emitChange();
+      this.setHint(MESSAGES[this.locale].loadedWord(profile.sourceFileName));
+      this.emitChange("upload", profile.sourceFileName);
     } catch (error) {
-      this.emitError(error instanceof Error ? error.message : "Word \u89E3\u6790\u5931\u8D25");
-    } finally {
-      this.fileInput.value = "";
+      this.emitError(error instanceof Error ? error.message : MESSAGES[this.locale].parseFailed);
     }
   }
-  async readClipboard() {
+  async loadClipboard() {
     if (!navigator.clipboard?.read) {
-      this.emitError("\u5F53\u524D\u6D4F\u89C8\u5668\u4E0D\u652F\u6301 clipboard.read");
+      this.emitError(MESSAGES[this.locale].noClipboardRead);
       return;
     }
     try {
@@ -1489,7 +1874,7 @@ var DocsWordElement = class extends HTMLElement {
       const payload = await extractFromClipboardItems(items);
       this.applyPayload(payload.html, payload.text);
     } catch (error) {
-      this.emitError(error instanceof Error ? error.message : "\u8BFB\u53D6\u526A\u8D34\u677F\u5931\u8D25");
+      this.emitError(error instanceof Error ? error.message : MESSAGES[this.locale].clipboardReadFailed);
     }
   }
   async applyFromClipboardData(data) {
@@ -1504,12 +1889,12 @@ var DocsWordElement = class extends HTMLElement {
     } else if (text.trim()) {
       this.htmlSnapshot = buildHtmlSnapshot(`<p>${text.replaceAll("&", "&amp;").replaceAll("<", "&lt;").replaceAll(">", "&gt;")}</p>`);
     } else {
-      this.hint.textContent = "\u672A\u68C0\u6D4B\u5230\u53EF\u5BFC\u5165\u5185\u5BB9";
+      this.setHint(MESSAGES[this.locale].noContent);
       return;
     }
     this.renderSnapshot();
-    this.hint.textContent = "\u5DF2\u5BFC\u5165\u526A\u8D34\u677F\u5185\u5BB9";
-    this.emitChange();
+    this.setHint(MESSAGES[this.locale].importedClipboard);
+    this.emitChange("paste");
   }
   onFrameLoad() {
     const doc = this.frame.contentDocument;
@@ -1534,12 +1919,33 @@ var DocsWordElement = class extends HTMLElement {
   renderSnapshot() {
     this.frame.srcdoc = this.htmlSnapshot;
   }
-  emitChange() {
-    this.dispatchEvent(new CustomEvent("docsjs-change", { detail: { htmlSnapshot: this.htmlSnapshot } }));
+  emitChange(source, fileName) {
+    this.dispatchEvent(new CustomEvent("docsjs-change", { detail: { htmlSnapshot: this.htmlSnapshot, source, fileName } }));
   }
   emitError(message) {
     this.dispatchEvent(new CustomEvent("docsjs-error", { detail: { message } }));
-    this.hint.textContent = `\u9519\u8BEF: ${message}`;
+    this.setHint(`${MESSAGES[this.locale].errorPrefix}${message}`);
+  }
+  setHint(text) {
+    this.hint.textContent = text;
+  }
+  parseLocale(value) {
+    return value?.toLowerCase() === "en" ? "en" : "zh";
+  }
+  syncToolbarVisibility() {
+    const raw = this.getAttribute("show-toolbar");
+    const show = raw === null || raw === "" || raw === "1" || raw.toLowerCase() === "true";
+    this.toolbar.style.display = show ? "flex" : "none";
+  }
+  syncLocaleText() {
+    const t = MESSAGES[this.locale];
+    this.btnRead.textContent = t.readClipboard;
+    this.btnUpload.textContent = t.uploadWord;
+    this.btnClear.textContent = t.clear;
+    this.pasteArea.placeholder = t.pastePlaceholder;
+    if (!this.hint.textContent || this.hint.textContent === MESSAGES.en.waitImport || this.hint.textContent === MESSAGES.zh.waitImport) {
+      this.hint.textContent = t.waitImport;
+    }
   }
 };
 function defineDocsWordElement() {
@@ -1548,62 +1954,71 @@ function defineDocsWordElement() {
   }
 }
 
-// src/react/WordFidelityEditorReact.tsx
-var import_react = __toESM(require("react"), 1);
-defineDocsWordElement();
-function WordFidelityEditorReact({ onChange, onError }) {
-  const ref2 = (0, import_react.useRef)(null);
-  (0, import_react.useEffect)(() => {
-    const node = ref2.current;
-    if (!node) return;
-    const onChangeEvent = (event) => {
-      const detail = event.detail;
-      onChange?.(detail);
-    };
-    const onErrorEvent = (event) => {
-      const detail = event.detail;
-      onError?.(detail);
-    };
-    node.addEventListener("docsjs-change", onChangeEvent);
-    node.addEventListener("docsjs-error", onErrorEvent);
-    return () => {
-      node.removeEventListener("docsjs-change", onChangeEvent);
-      node.removeEventListener("docsjs-error", onErrorEvent);
-    };
-  }, [onChange, onError]);
-  return import_react.default.createElement("docs-word-editor", { ref: ref2 });
+// src/lib/semanticStats.ts
+function countElements(root, selector) {
+  return root.querySelectorAll(selector).length;
+}
+function isListLikeParagraph(p) {
+  if (p.hasAttribute("data-word-list")) return true;
+  if (p.querySelector("span.__word-list-marker")) return true;
+  const style = (p.getAttribute("style") ?? "").toLowerCase();
+  return style.includes("mso-list");
+}
+function collectSemanticStatsFromDocument(doc) {
+  const paragraphs = Array.from(doc.querySelectorAll("p"));
+  const listParagraphCount = paragraphs.filter((p) => isListLikeParagraph(p)).length;
+  const textCharCount = (doc.body.textContent ?? "").replace(/\s+/g, "").length;
+  return {
+    paragraphCount: paragraphs.length,
+    headingCount: countElements(doc, "h1,h2,h3,h4,h5,h6"),
+    tableCount: countElements(doc, "table"),
+    tableCellCount: countElements(doc, "td,th"),
+    imageCount: countElements(doc, "img"),
+    anchorImageCount: countElements(doc, 'img[data-word-anchor="1"]'),
+    wrappedImageCount: countElements(doc, "img[data-word-wrap]"),
+    listParagraphCount,
+    commentRefCount: countElements(doc, "[data-word-comment-ref]"),
+    revisionInsCount: countElements(doc, '[data-word-revision="ins"]'),
+    revisionDelCount: countElements(doc, '[data-word-revision="del"]'),
+    pageBreakCount: countElements(doc, "[data-word-page-break='1']"),
+    pageSpacerCount: countElements(doc, "[data-word-page-spacer='1']"),
+    textCharCount
+  };
+}
+function collectSemanticStatsFromHtml(rawHtml) {
+  const parser = new DOMParser();
+  const doc = parser.parseFromString(rawHtml, "text/html");
+  return collectSemanticStatsFromDocument(doc);
 }
 
-// src/vue/WordFidelityEditorVue.ts
-var import_vue = require("vue");
-defineDocsWordElement();
-var WordFidelityEditorVue = (0, import_vue.defineComponent)({
-  name: "WordFidelityEditorVue",
-  emits: ["change", "error"],
-  setup(_, { emit }) {
-    const elRef = (0, import_vue.ref)(null);
-    const onChange = (event) => {
-      emit("change", event.detail);
-    };
-    const onError = (event) => {
-      emit("error", event.detail);
-    };
-    (0, import_vue.onMounted)(() => {
-      elRef.value?.addEventListener("docsjs-change", onChange);
-      elRef.value?.addEventListener("docsjs-error", onError);
-    });
-    (0, import_vue.onBeforeUnmount)(() => {
-      elRef.value?.removeEventListener("docsjs-change", onChange);
-      elRef.value?.removeEventListener("docsjs-error", onError);
-    });
-    return () => (0, import_vue.h)("docs-word-editor", { ref: elRef });
-  }
-});
+// src/lib/fidelityScore.ts
+function ratioScore(actual, expected) {
+  if (expected <= 0 && actual <= 0) return 1;
+  if (expected <= 0 || actual < 0) return 0;
+  const delta = Math.abs(actual - expected);
+  const penalty = delta / expected;
+  return Math.max(0, 1 - penalty);
+}
+function clamp01(v) {
+  if (v < 0) return 0;
+  if (v > 1) return 1;
+  return 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
+  );
+  const styleProxy = clamp01(ratioScore(actual.textCharCount, expected.textCharCount));
+  const pagination = clamp01(ratioScore(actual.pageSpacerCount, expected.pageSpacerCount));
+  const overall = clamp01(structure * 0.6 + styleProxy * 0.25 + pagination * 0.15);
+  return { structure, styleProxy, pagination, overall };
+}
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   DocsWordElement,
-  WordFidelityEditorReact,
-  WordFidelityEditorVue,
+  calculateFidelityScore,
+  collectSemanticStatsFromDocument,
+  collectSemanticStatsFromHtml,
   defineDocsWordElement
 });
 //# sourceMappingURL=index.cjs.map