@coding01/docsjs 0.1.5 → 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;");
 }
@@ -231,6 +248,19 @@ function normalizeWordPath(relTarget) {
   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;
@@ -405,7 +435,7 @@ 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);
@@ -446,6 +476,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
     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);
       result.push(
         `<sup data-word-footnote-ref="${footnoteId}"><a href="#word-footnote-${footnoteId}">[${footnoteId}]</a></sup>`
@@ -455,6 +486,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
     const endnoteRef = queryByLocalName(run, "endnoteReference");
     const endnoteId = getAttr(endnoteRef, "w:id") ?? getAttr(endnoteRef, "id");
     if (endnoteId && endnotesMap[endnoteId]) {
+      context.features.endnoteRefCount += 1;
       usedEndnoteIds.push(endnoteId);
       result.push(
         `<sup data-word-endnote-ref="${endnoteId}"><a href="#word-endnote-${endnoteId}">[${endnoteId}]</a></sup>`
@@ -464,6 +496,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
     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);
       result.push(
         `<sup data-word-comment-ref="${commentId}"><a href="#word-comment-${commentId}">[c${commentId}]</a></sup>`
@@ -481,6 +514,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
           const dimensionAttrs = imageDimensionAttributes(imageSize);
           const anchorMeta = parseAnchorMeta(drawing);
           const attrs = mergeImageStyle(dimensionAttrs, anchorMeta);
+          if (anchorMeta) context.features.anchorImageCount += 1;
           result.push(`<img src="${src}" alt="word-image"${attrs}/>`);
           return result;
         }
@@ -491,6 +525,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
         const chartXmlText = await readXmlByRid(zip, relMap, chartRid);
         if (chartXmlText) {
           const summary = parseChartSummary(chartXmlText);
+          context.features.chartCount += 1;
           result.push(
             `<figure data-word-chart="1" data-word-chart-type="${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>`
           );
@@ -502,6 +537,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
       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${preview}</figcaption></figure>`
@@ -523,12 +559,14 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
       if (css) {
         const span = `<span style="${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}</${tagName}>`);
       } else {
@@ -536,6 +574,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
       }
     }
     for (let i = 0; i < pageBreakCount; i += 1) {
+      context.features.pageBreakCount += 1;
       result.push(`<span data-word-page-break="1" style="display:block;break-before:page"></span>`);
     }
     return result;
@@ -552,9 +591,25 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
     if (node.localName === "r") {
       return runToHtml(node, revisionFallback);
     }
+    if (node.localName === "hyperlink") {
+      const rid = getAttr(node, "r:id") ?? getAttr(node, "id");
+      const anchor = getAttr(node, "w:anchor") ?? getAttr(node, "anchor");
+      const href = resolveHyperlinkHref(relMap, rid, anchor);
+      const nested2 = [];
+      for (const child of Array.from(node.children)) {
+        nested2.push(...await nodeToHtml(child, revisionFallback));
+      }
+      const content2 = nested2.join("") || escapeHtml(node.textContent ?? "");
+      if (!href) return content2 ? [content2] : [];
+      context.features.hyperlinkCount += 1;
+      return [
+        `<a data-word-hyperlink="1" href="${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") {
@@ -574,6 +629,7 @@ async function paragraphToHtml(zip, relMap, paragraph, paragraphIndex, footnotes
   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)) {
@@ -702,7 +758,7 @@ function parseCellBorderStyle(cell, tableStyle) {
   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) {
+function tableCellHtml(cell, paragraphIndexMap, context) {
   const blocks = [];
   for (const child of Array.from(cell.children)) {
     if (child.localName === "tcPr") continue;
@@ -712,7 +768,7 @@ function tableCellHtml(cell, paragraphIndexMap) {
       continue;
     }
     if (child.localName === "tbl") {
-      blocks.push(tableToHtml(child, paragraphIndexMap));
+      blocks.push(tableToHtml(child, paragraphIndexMap, context));
       continue;
     }
   }
@@ -720,7 +776,8 @@ 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);
@@ -748,7 +805,7 @@ 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);
@@ -792,7 +849,9 @@ function tableToHtml(table, paragraphIndexMap) {
   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);
@@ -829,6 +888,7 @@ async function parseDocxToHtmlSnapshot(file) {
         await paragraphToHtml(
           zip,
           relMap,
+          context,
           child,
           paragraphIndex,
           footnotesMap,
@@ -842,14 +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
@@ -1980,7 +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",
@@ -2123,15 +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);
     }
@@ -2191,8 +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 } }));
@@ -2294,6 +2366,8 @@ function calculateFidelityScore(expected, actual) {
   calculateFidelityScore,
   collectSemanticStatsFromDocument,
   collectSemanticStatsFromHtml,
-  defineDocsWordElement
+  defineDocsWordElement,
+  parseDocxToHtmlSnapshot,
+  parseDocxToHtmlSnapshotWithReport
 });
 //# sourceMappingURL=index.cjs.map