kordoc 4.0.5 → 4.0.6

package/dist/index.js

@@ -20,7 +20,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-FKK4K53F.js";
+} from "./chunk-HOGJ5ALY.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
@@ -2229,8 +2229,15 @@ function collectSubListText(el, ctx, depth = 0) {
     if (tag === "p" || tag === "para") {
       const t = extractParagraphInfo(ch, ctx.styleMap, ctx).text;
       if (t) parts.push(t);
+      const tbls = [];
+      findTopLevelTbls(ch, tbls);
+      for (const tbl2 of tbls) {
+        const flat = flattenSubListTable(tbl2, ctx, depth);
+        if (flat) parts.push(flat);
+      }
     } else if (tag === "tbl") {
-      continue;
+      const flat = flattenSubListTable(ch, ctx, depth);
+      if (flat) parts.push(flat);
     } else {
       const t = collectSubListText(ch, ctx, depth + 1);
       if (t) parts.push(t);
@@ -2238,6 +2245,29 @@ function collectSubListText(el, ctx, depth = 0) {
   }
   return parts.join("\n").trim();
 }
+function flattenSubListTable(el, ctx, depth) {
+  const sink = [];
+  const st = { rows: [], currentRow: [], cell: null };
+  walkSection(el, sink, st, [], ctx, depth + 1);
+  let flat = convertTableToText(st.rows);
+  if (st.caption) flat = st.caption + (flat ? "\n" + flat : "");
+  return flat;
+}
+function findTopLevelTbls(el, out, depth = 0) {
+  if (depth > MAX_XML_DEPTH) return;
+  const kids2 = el.childNodes;
+  if (!kids2) return;
+  for (let i = 0; i < kids2.length; i++) {
+    const ch = kids2[i];
+    if (ch.nodeType !== 1) continue;
+    const tag = (ch.tagName || ch.localName || "").replace(/^[^:]+:/, "");
+    if (tag === "tbl") {
+      out.push(ch);
+      continue;
+    }
+    findTopLevelTbls(ch, out, depth + 1);
+  }
+}
 function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
   if (depth > MAX_XML_DEPTH) return tableCtx;
   const children = node.childNodes;
@@ -29784,7 +29814,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-YSHZC5FQ.js");
+    const mod = await import("./parser-DMKHIV4E.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {