kordoc 2.2.5 → 2.3.0

package/dist/index.js

@@ -16,7 +16,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-UU2O6D3R.js";
+} from "./chunk-JFTFC2BB.js";
 import {
   parsePageRange
 } from "./chunk-SBVRCJFH.js";
@@ -44,11 +44,17 @@ function isPdfFile(buffer) {
   const b = magicBytes(buffer);
   return b[0] === 37 && b[1] === 80 && b[2] === 68 && b[3] === 70;
 }
+function isHwpmlFile(buffer) {
+  const bytes = new Uint8Array(buffer, 0, Math.min(512, buffer.byteLength));
+  const head = new TextDecoder("utf-8", { fatal: false }).decode(bytes).replace(/^\uFEFF/, "");
+  return head.trimStart().startsWith("<?xml") && head.includes("<HWPML");
+}
 function detectFormat(buffer) {
   if (buffer.byteLength < 4) return "unknown";
   if (isZipFile(buffer)) return "hwpx";
   if (isOldHwpFile(buffer)) return "hwp";
   if (isPdfFile(buffer)) return "pdf";
+  if (isHwpmlFile(buffer)) return "hwpml";
   return "unknown";
 }
 async function detectZipFormat(buffer) {
@@ -184,6 +190,7 @@ async function parseHwpxDocument(buffer, options) {
   const pageFilter = options?.pages ? parsePageRange(options.pages, sectionPaths.length) : null;
   const totalTarget = pageFilter ? pageFilter.size : sectionPaths.length;
   const blocks = [];
+  const nestedTableCounter = { count: 0 };
   let parsedSections = 0;
   for (let si = 0; si < sectionPaths.length; si++) {
     if (pageFilter && !pageFilter.has(si + 1)) continue;
@@ -193,7 +200,7 @@ async function parseHwpxDocument(buffer, options) {
       const xml = await file.async("text");
       decompressed.total += xml.length * 2;
       if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-      blocks.push(...parseSectionXml(xml, styleMap, warnings, si + 1));
+      blocks.push(...parseSectionXml(xml, styleMap, warnings, si + 1, nestedTableCounter));
       parsedSections++;
       options?.onProgress?.(parsedSections, totalTarget);
     } catch (secErr) {
@@ -254,8 +261,20 @@ async function extractImagesFromZip(zip, blocks, decompressed, warnings) {
       ref
       // 절대 경로일 수도 있음
     ];
+    let resolvedPath = null;
+    if (!ref.includes(".")) {
+      const prefixes = [`BinData/${ref}`, `Contents/BinData/${ref}`];
+      for (const prefix of prefixes) {
+        const match = zip.file(new RegExp(`^${prefix.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}\\.[a-zA-Z0-9]+$`));
+        if (match.length > 0) {
+          resolvedPath = match[0].name;
+          break;
+        }
+      }
+    }
     let found = false;
-    for (const path of candidates) {
+    const allCandidates = resolvedPath ? [resolvedPath, ...candidates] : candidates;
+    for (const path of allCandidates) {
       if (isPathTraversal(path)) continue;
       const file = zip.file(path);
       if (!file) continue;
@@ -263,7 +282,8 @@ async function extractImagesFromZip(zip, blocks, decompressed, warnings) {
         const data = await file.async("uint8array");
         decompressed.total += data.length;
         if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new KordocError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-        const ext = ref.includes(".") ? ref.split(".").pop() || "png" : "png";
+        const actualPath = path;
+        const ext = actualPath.includes(".") ? actualPath.split(".").pop() || "png" : "png";
         const mimeType = imageExtToMime(ext);
         imageIndex++;
         const filename = `image_${String(imageIndex).padStart(3, "0")}.${mimeToExt(mimeType)}`;
@@ -336,6 +356,7 @@ function extractFromBrokenZip(buffer) {
   let totalDecompressed = 0;
   let entryCount = 0;
   let sectionNum = 0;
+  const nestedTableCounter = { count: 0 };
   while (pos < data.length - 30) {
     if (data[pos] !== 80 || data[pos + 1] !== 75 || data[pos + 2] !== 3 || data[pos + 3] !== 4) {
       pos++;
@@ -382,7 +403,7 @@ function extractFromBrokenZip(buffer) {
       totalDecompressed += content.length * 2;
       if (totalDecompressed > MAX_DECOMPRESS_SIZE) throw new KordocError("\uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC");
       sectionNum++;
-      blocks.push(...parseSectionXml(content, void 0, warnings, sectionNum));
+      blocks.push(...parseSectionXml(content, void 0, warnings, sectionNum, nestedTableCounter));
     } catch {
       continue;
     }
@@ -467,12 +488,40 @@ function detectHwpxHeadings(blocks, styleMap) {
     }
   }
 }
-function parseSectionXml(xml, styleMap, warnings, sectionNum) {
+function makeNestedTableMarker(counter, rows) {
+  counter.count++;
+  const firstRow = rows[0] ?? [];
+  const hint = firstRow.map((c) => c.text.trim().replace(/\n/g, " ")).filter(Boolean).join(" | ");
+  const hintChars = [...hint];
+  const truncated = hintChars.length > 60 ? hintChars.slice(0, 60).join("") + "\u2026" : hint;
+  return truncated ? `[\uC911\uCCA9 \uD14C\uC774\uBE14 #${counter.count}: ${truncated}]` : `[\uC911\uCCA9 \uD14C\uC774\uBE14 #${counter.count}]`;
+}
+function handleNestedTable(newTable, tableStack, blocks, ctx) {
+  const parentTable = tableStack.pop();
+  let nestedCols = 0;
+  for (const r of newTable.rows) if (r.length > nestedCols) nestedCols = r.length;
+  if (newTable.rows.length >= 3 && nestedCols >= 2) {
+    blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: ctx.sectionNum });
+    if (parentTable.cell) {
+      const marker = ctx.counter ? makeNestedTableMarker(ctx.counter, newTable.rows) : "[\uC911\uCCA9 \uD14C\uC774\uBE14]";
+      parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + marker;
+    }
+  } else {
+    const nestedText = convertTableToText(newTable.rows);
+    if (parentTable.cell) {
+      const marker = ctx.counter ? makeNestedTableMarker(ctx.counter, newTable.rows) : "[\uC911\uCCA9 \uD14C\uC774\uBE14]";
+      parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + marker + "\n" + nestedText;
+    }
+  }
+  return parentTable;
+}
+function parseSectionXml(xml, styleMap, warnings, sectionNum, counter) {
   const parser = createXmlParser(warnings);
   const doc = parser.parseFromString(stripDtd(xml), "text/xml");
   if (!doc.documentElement) return [];
   const blocks = [];
-  walkSection(doc.documentElement, blocks, null, [], styleMap, warnings, sectionNum);
+  const ctx = { styleMap, warnings, sectionNum, counter };
+  walkSection(doc.documentElement, blocks, null, [], ctx);
   return blocks;
 }
 function extractImageRef(el) {
@@ -493,7 +542,7 @@ function extractImageRef(el) {
   if (directRef) return directRef;
   return null;
 }
-function walkSection(node, blocks, tableCtx, tableStack, styleMap, warnings, sectionNum, depth = 0) {
+function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
   if (depth > MAX_XML_DEPTH) return;
   const children = node.childNodes;
   if (!children) return;
@@ -506,23 +555,12 @@ function walkSection(node, blocks, tableCtx, tableStack, styleMap, warnings, sec
       case "tbl": {
         if (tableCtx) tableStack.push(tableCtx);
         const newTable = { rows: [], currentRow: [], cell: null };
-        walkSection(el, blocks, newTable, tableStack, styleMap, warnings, sectionNum, depth + 1);
+        walkSection(el, blocks, newTable, tableStack, ctx, depth + 1);
         if (newTable.rows.length > 0) {
           if (tableStack.length > 0) {
-            const parentTable = tableStack.pop();
-            let nestedCols = 0;
-            for (const r of newTable.rows) if (r.length > nestedCols) nestedCols = r.length;
-            if (newTable.rows.length >= 3 && nestedCols >= 2) {
-              blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: sectionNum });
-            } else {
-              const nestedText = convertTableToText(newTable.rows);
-              if (parentTable.cell) {
-                parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + nestedText;
-              }
-            }
-            tableCtx = parentTable;
+            tableCtx = handleNestedTable(newTable, tableStack, blocks, ctx);
           } else {
-            blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: sectionNum });
+            blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: ctx.sectionNum });
             tableCtx = null;
           }
         } else {
@@ -533,7 +571,7 @@ function walkSection(node, blocks, tableCtx, tableStack, styleMap, warnings, sec
       case "tr":
         if (tableCtx) {
           tableCtx.currentRow = [];
-          walkSection(el, blocks, tableCtx, tableStack, styleMap, warnings, sectionNum, depth + 1);
+          walkSection(el, blocks, tableCtx, tableStack, ctx, depth + 1);
           if (tableCtx.currentRow.length > 0) tableCtx.rows.push(tableCtx.currentRow);
           tableCtx.currentRow = [];
         }
@@ -541,7 +579,7 @@ function walkSection(node, blocks, tableCtx, tableStack, styleMap, warnings, sec
       case "tc":
         if (tableCtx) {
           tableCtx.cell = { text: "", colSpan: 1, rowSpan: 1 };
-          walkSection(el, blocks, tableCtx, tableStack, styleMap, warnings, sectionNum, depth + 1);
+          walkSection(el, blocks, tableCtx, tableStack, ctx, depth + 1);
           if (tableCtx.cell) {
             tableCtx.currentRow.push(tableCtx.cell);
             tableCtx.cell = null;
@@ -567,19 +605,19 @@ function walkSection(node, blocks, tableCtx, tableStack, styleMap, warnings, sec
         }
         break;
       case "p": {
-        const { text, href, footnote, style } = extractParagraphInfo(el, styleMap);
+        const { text, href, footnote, style } = extractParagraphInfo(el, ctx.styleMap);
         if (text) {
           if (tableCtx?.cell) {
             tableCtx.cell.text += (tableCtx.cell.text ? "\n" : "") + text;
           } else if (!tableCtx) {
-            const block = { type: "paragraph", text, pageNumber: sectionNum };
+            const block = { type: "paragraph", text, pageNumber: ctx.sectionNum };
             if (style) block.style = style;
             if (href) block.href = href;
             if (footnote) block.footnoteText = footnote;
             blocks.push(block);
           }
         }
-        tableCtx = walkParagraphChildren(el, blocks, tableCtx, tableStack, styleMap, warnings, sectionNum, depth + 1);
+        tableCtx = walkParagraphChildren(el, blocks, tableCtx, tableStack, ctx, depth + 1);
         break;
       }
       // 이미지/그림 — 경로 추출 또는 경고
@@ -588,19 +626,19 @@ function walkSection(node, blocks, tableCtx, tableStack, styleMap, warnings, sec
       case "drawingObject": {
         const imgRef = extractImageRef(el);
         if (imgRef) {
-          blocks.push({ type: "image", text: imgRef, pageNumber: sectionNum });
-        } else if (warnings && sectionNum) {
-          warnings.push({ page: sectionNum, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
+          blocks.push({ type: "image", text: imgRef, pageNumber: ctx.sectionNum });
+        } else if (ctx.warnings && ctx.sectionNum) {
+          ctx.warnings.push({ page: ctx.sectionNum, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
         }
         break;
       }
       default:
-        walkSection(el, blocks, tableCtx, tableStack, styleMap, warnings, sectionNum, depth + 1);
+        walkSection(el, blocks, tableCtx, tableStack, ctx, depth + 1);
         break;
     }
   }
 }
-function walkParagraphChildren(node, blocks, tableCtx, tableStack, styleMap, warnings, sectionNum, depth = 0) {
+function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
   if (depth > MAX_XML_DEPTH) return tableCtx;
   const children = node.childNodes;
   if (!children) return tableCtx;
@@ -616,23 +654,12 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, styleMap, war
       if (localTag === "tbl") {
         if (tableCtx) tableStack.push(tableCtx);
         const newTable = { rows: [], currentRow: [], cell: null };
-        walkSection(el, blocks, newTable, tableStack, styleMap, warnings, sectionNum, d + 1);
+        walkSection(el, blocks, newTable, tableStack, ctx, d + 1);
         if (newTable.rows.length > 0) {
           if (tableStack.length > 0) {
-            const parentTable = tableStack.pop();
-            let nestedCols = 0;
-            for (const r of newTable.rows) if (r.length > nestedCols) nestedCols = r.length;
-            if (newTable.rows.length >= 3 && nestedCols >= 2) {
-              blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: sectionNum });
-            } else {
-              const nestedText = convertTableToText(newTable.rows);
-              if (parentTable.cell) {
-                parentTable.cell.text += (parentTable.cell.text ? "\n" : "") + nestedText;
-              }
-            }
-            tableCtx = parentTable;
+            tableCtx = handleNestedTable(newTable, tableStack, blocks, ctx);
           } else {
-            blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: sectionNum });
+            blocks.push({ type: "table", table: buildTable(newTable.rows), pageNumber: ctx.sectionNum });
             tableCtx = null;
           }
         } else {
@@ -641,21 +668,21 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, styleMap, war
       } else if (localTag === "pic" || localTag === "shape" || localTag === "drawingObject") {
         const drawTextChild = findDescendant(el, "drawText");
         if (drawTextChild) {
-          extractDrawTextBlocks(drawTextChild, blocks, styleMap, sectionNum);
+          extractDrawTextBlocks(drawTextChild, blocks, ctx.styleMap, ctx.sectionNum);
         } else {
           const imgRef = extractImageRef(el);
           if (imgRef) {
-            blocks.push({ type: "image", text: imgRef, pageNumber: sectionNum });
-          } else if (warnings && sectionNum) {
-            warnings.push({ page: sectionNum, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
+            blocks.push({ type: "image", text: imgRef, pageNumber: ctx.sectionNum });
+          } else if (ctx.warnings && ctx.sectionNum) {
+            ctx.warnings.push({ page: ctx.sectionNum, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
           }
         }
       } else if (localTag === "drawText") {
-        extractDrawTextBlocks(el, blocks, styleMap, sectionNum);
+        extractDrawTextBlocks(el, blocks, ctx.styleMap, ctx.sectionNum);
       } else if (localTag === "r" || localTag === "run" || localTag === "ctrl" || localTag === "rect" || localTag === "ellipse" || localTag === "polygon" || localTag === "line" || localTag === "arc" || localTag === "curve" || localTag === "connectLine" || localTag === "container") {
         walkChildren(el, d + 1);
       } else if (localTag === "run") {
-        tableCtx = walkParagraphChildren(el, blocks, tableCtx, tableStack, styleMap, warnings, sectionNum, depth + 1);
+        tableCtx = walkParagraphChildren(el, blocks, tableCtx, tableStack, ctx, depth + 1);
       }
     }
   };
@@ -1928,6 +1955,7 @@ function parseHwp5Document(buffer, options) {
   const pageFilter = options?.pages ? parsePageRange(options.pages, sections.length) : null;
   const totalTarget = pageFilter ? pageFilter.size : sections.length;
   const blocks = [];
+  const nestedTableCounter = { count: 0 };
   let totalDecompressed = 0;
   let parsedSections = 0;
   for (let si = 0; si < sections.length; si++) {
@@ -1938,7 +1966,7 @@ function parseHwp5Document(buffer, options) {
       totalDecompressed += data.length;
       if (totalDecompressed > MAX_TOTAL_DECOMPRESS) throw new KordocError("\uCD1D \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (decompression bomb \uC758\uC2EC)");
       const records = readRecords(data);
-      const sectionBlocks = parseSection(records, docInfo, warnings, si + 1);
+      const sectionBlocks = parseSection(records, docInfo, warnings, si + 1, nestedTableCounter);
       blocks.push(...sectionBlocks);
       parsedSections++;
       options?.onProgress?.(parsedSections, totalTarget);
@@ -2258,13 +2286,13 @@ function extractHwp5ImagesLenient(lcfb, blocks, compressed, warnings) {
   }
   return images;
 }
-function parseSection(records, docInfo, warnings, sectionNum) {
+function parseSection(records, docInfo, warnings, sectionNum, counter) {
   const blocks = [];
   let i = 0;
   while (i < records.length) {
     const rec = records[i];
     if (rec.tagId === TAG_PARA_HEADER && rec.level === 0) {
-      const { paragraph, tables, nextIdx, charShapeIds, paraShapeId } = parseParagraphWithTables(records, i);
+      const { paragraph, tables, nextIdx, charShapeIds, paraShapeId } = parseParagraphWithTables(records, i, counter);
       if (paragraph) {
         const block = { type: "paragraph", text: paragraph, pageNumber: sectionNum };
         if (docInfo && charShapeIds.length > 0) {
@@ -2287,7 +2315,7 @@ function parseSection(records, docInfo, warnings, sectionNum) {
     if (rec.tagId === TAG_CTRL_HEADER && rec.level <= 1 && rec.data.length >= 4) {
       const ctrlId = rec.data.subarray(0, 4).toString("ascii");
       if (ctrlId === " lbt" || ctrlId === "tbl ") {
-        const { table, nextIdx } = parseTableBlock(records, i);
+        const { table, nextIdx } = parseTableBlock(records, i, counter);
         if (table) blocks.push({ type: "table", table, pageNumber: sectionNum });
         i = nextIdx;
         continue;
@@ -2392,7 +2420,7 @@ function resolveCharStyle(charShapeIds, docInfo) {
   if (cs.attrFlags & 2) style.bold = true;
   return style.fontSize || style.bold || style.italic ? style : void 0;
 }
-function parseParagraphWithTables(records, startIdx) {
+function parseParagraphWithTables(records, startIdx, counter) {
   const startLevel = records[startIdx].level;
   let text = "";
   const tables = [];
@@ -2414,7 +2442,7 @@ function parseParagraphWithTables(records, startIdx) {
     if (rec.tagId === TAG_CTRL_HEADER && rec.data.length >= 4) {
       const ctrlId = rec.data.subarray(0, 4).toString("ascii");
       if (ctrlId === " lbt" || ctrlId === "tbl ") {
-        const { table, nextIdx } = parseTableBlock(records, i);
+        const { table, nextIdx } = parseTableBlock(records, i, counter);
         if (table) tables.push(table);
         i = nextIdx;
         continue;
@@ -2425,7 +2453,7 @@ function parseParagraphWithTables(records, startIdx) {
   const trimmed = text.trim();
   return { paragraph: trimmed || null, tables, nextIdx: i, charShapeIds, paraShapeId };
 }
-function parseTableBlock(records, startIdx) {
+function parseTableBlock(records, startIdx, counter) {
   const tableLevel = records[startIdx].level;
   let i = startIdx + 1;
   let rows = 0, cols = 0;
@@ -2439,7 +2467,7 @@ function parseTableBlock(records, startIdx) {
       cols = Math.min(rec.data.readUInt16LE(6), MAX_COLS);
     }
     if (rec.tagId === TAG_LIST_HEADER) {
-      const { cell, nextIdx } = parseCellBlock(records, i, tableLevel);
+      const { cell, nextIdx } = parseCellBlock(records, i, tableLevel, counter);
       if (cell) cells.push(cell);
       i = nextIdx;
       continue;
@@ -2460,7 +2488,7 @@ function parseTableBlock(records, startIdx) {
   const cellRows = arrangeCells(rows, cols, cells);
   return { table: buildTable(cellRows), nextIdx: i };
 }
-function parseCellBlock(records, startIdx, tableLevel) {
+function parseCellBlock(records, startIdx, tableLevel, counter) {
   const rec = records[startIdx];
   const cellLevel = rec.level;
   const texts = [];
@@ -2485,6 +2513,17 @@ function parseCellBlock(records, startIdx, tableLevel) {
       const t = extractText(r.data).trim();
       if (t) texts.push(t);
     }
+    if (r.tagId === TAG_CTRL_HEADER && r.data.length >= 4) {
+      const ctrlId = r.data.subarray(0, 4).toString("ascii");
+      if (ctrlId === " lbt" || ctrlId === "tbl ") {
+        if (counter) {
+          counter.count++;
+          texts.push(`[\uC911\uCCA9 \uD14C\uC774\uBE14 #${counter.count}]`);
+        } else {
+          texts.push("[\uC911\uCCA9 \uD14C\uC774\uBE14]");
+        }
+      }
+    }
     i++;
   }
   return { cell: { text: texts.join("\n"), colSpan, rowSpan, colAddr, rowAddr }, nextIdx: i };
@@ -2811,21 +2850,21 @@ async function parseXlsxDocument(buffer, options) {
 import JSZip4 from "jszip";
 import { DOMParser as DOMParser3 } from "@xmldom/xmldom";
 var MAX_DECOMPRESS_SIZE4 = 100 * 1024 * 1024;
-function getChildElements(parent, localName2) {
+function getChildElements(parent, localName3) {
   const result = [];
   const children = parent.childNodes;
   for (let i = 0; i < children.length; i++) {
     const node = children[i];
     if (node.nodeType === 1) {
       const el = node;
-      if (el.localName === localName2 || el.tagName?.endsWith(`:${localName2}`)) {
+      if (el.localName === localName3 || el.tagName?.endsWith(`:${localName3}`)) {
         result.push(el);
       }
     }
   }
   return result;
 }
-function findElements(parent, localName2) {
+function findElements(parent, localName3) {
   const result = [];
   const walk = (node) => {
     const children = node.childNodes;
@@ -2833,7 +2872,7 @@ function findElements(parent, localName2) {
       const child = children[i];
       if (child.nodeType === 1) {
         const el = child;
-        if (el.localName === localName2 || el.tagName?.endsWith(`:${localName2}`)) {
+        if (el.localName === localName3 || el.tagName?.endsWith(`:${localName3}`)) {
           result.push(el);
         }
         walk(el);
@@ -2843,11 +2882,11 @@ function findElements(parent, localName2) {
   walk(parent);
   return result;
 }
-function getAttr(el, localName2) {
+function getAttr(el, localName3) {
   const attrs = el.attributes;
   for (let i = 0; i < attrs.length; i++) {
     const attr = attrs[i];
-    if (attr.localName === localName2 || attr.name === localName2) return attr.value;
+    if (attr.localName === localName3 || attr.name === localName3) return attr.value;
   }
   return null;
 }
@@ -3194,11 +3233,11 @@ async function parseDocxDocument(buffer, options) {
     const node = children[i];
     if (node.nodeType !== 1) continue;
     const el = node;
-    const localName2 = el.localName ?? el.tagName?.split(":").pop();
-    if (localName2 === "p") {
+    const localName3 = el.localName ?? el.tagName?.split(":").pop();
+    if (localName3 === "p") {
       const block = parseParagraph(el, styles, numbering, footnotes, rels);
       if (block) blocks.push(block);
-    } else if (localName2 === "tbl") {
+    } else if (localName3 === "tbl") {
       const block = parseTable(el, styles, numbering, footnotes, rels);
       if (block) blocks.push(block);
     }
@@ -3236,6 +3275,259 @@ async function parseDocxDocument(buffer, options) {
   };
 }
 
+// src/hwpml/parser.ts
+import { DOMParser as DOMParser4 } from "@xmldom/xmldom";
+var MAX_XML_DEPTH2 = 200;
+var MAX_TABLE_ROWS = 5e3;
+var MAX_TABLE_COLS = 500;
+var MAX_HWPML_BYTES = 50 * 1024 * 1024;
+function parseHwpmlDocument(buffer, options) {
+  if (buffer.byteLength > MAX_HWPML_BYTES) {
+    throw new Error(`HWPML \uD30C\uC77C \uD06C\uAE30 \uCD08\uACFC (${(buffer.byteLength / 1024 / 1024).toFixed(1)}MB > 50MB)`);
+  }
+  const text = new TextDecoder("utf-8").decode(buffer).replace(/^\uFEFF/, "");
+  const normalized = text.replace(/&nbsp;/g, "&#160;");
+  const xml = stripDtd(normalized);
+  const warnings = [];
+  const parser = new DOMParser4({
+    onError: (_level, msg) => {
+      warnings.push({ message: `HWPML XML \uD30C\uC2F1 \uACBD\uACE0: ${msg}`, code: "MALFORMED_XML" });
+    }
+  });
+  const doc = parser.parseFromString(xml, "text/xml");
+  if (!doc.documentElement) {
+    return { markdown: "", blocks: [], warnings };
+  }
+  const root = doc.documentElement;
+  const metadata = {};
+  const docSummary = findChild(root, "DOCSUMMARY");
+  if (docSummary) {
+    const title = findChild(docSummary, "TITLE");
+    const author = findChild(docSummary, "AUTHOR");
+    const date = findChild(docSummary, "DATE");
+    if (title) metadata.title = textContent(title).trim();
+    if (author) metadata.author = textContent(author).trim();
+    if (date) metadata.createdAt = textContent(date).trim() || void 0;
+  }
+  const paraShapeMap = buildParaShapeMap(root);
+  const body = findChild(root, "BODY");
+  if (!body) {
+    return { markdown: "", blocks: [], metadata, warnings };
+  }
+  const blocks = [];
+  const pageFilter = options?.pages ? parsePageRange(options.pages, countSections(body)) : null;
+  let sectionIdx = 0;
+  const children = body.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const el = children[i];
+    if (el.nodeType !== 1) continue;
+    if (localName(el) !== "SECTION") continue;
+    sectionIdx++;
+    if (pageFilter && !pageFilter.has(sectionIdx)) continue;
+    parseSection2(el, blocks, paraShapeMap, sectionIdx, warnings);
+  }
+  const outline = blocks.filter((b) => b.type === "heading" && b.text).map((b) => ({ level: b.level ?? 1, text: b.text, pageNumber: b.pageNumber }));
+  const markdown = blocksToMarkdown(blocks);
+  return {
+    markdown,
+    blocks,
+    metadata: Object.keys(metadata).length > 0 ? metadata : void 0,
+    outline: outline.length > 0 ? outline : void 0,
+    warnings: warnings.length > 0 ? warnings : void 0
+  };
+}
+function buildParaShapeMap(root) {
+  const map = /* @__PURE__ */ new Map();
+  const head = findChild(root, "HEAD");
+  if (!head) return map;
+  const mappingTable = findChild(head, "MAPPINGTABLE");
+  if (!mappingTable) return map;
+  const paraShapeList = findChild(mappingTable, "PARASHAPELIST");
+  if (!paraShapeList) return map;
+  const children = paraShapeList.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const el = children[i];
+    if (el.nodeType !== 1 || localName(el) !== "PARASHAPE") continue;
+    const id = el.getAttribute("Id") ?? "";
+    const headingType = el.getAttribute("HeadingType") ?? "None";
+    const level = parseInt(el.getAttribute("Level") ?? "0", 10);
+    let headingLevel = null;
+    if (headingType === "Outline") {
+      const safeLevel = isNaN(level) ? 0 : Math.max(0, level);
+      headingLevel = Math.min(safeLevel + 1, 6);
+    }
+    map.set(id, { headingLevel });
+  }
+  return map;
+}
+function parseSection2(section, blocks, paraShapeMap, sectionNum, warnings) {
+  walkContent(section, blocks, paraShapeMap, sectionNum, warnings, false);
+}
+function walkContent(node, blocks, paraShapeMap, sectionNum, warnings, inHeaderFooter, depth = 0) {
+  if (depth > MAX_XML_DEPTH2) return;
+  const children = node.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const el = children[i];
+    if (el.nodeType !== 1) continue;
+    const tag = localName(el);
+    if (tag === "HEADER" || tag === "FOOTER") {
+      continue;
+    }
+    if (tag === "P") {
+      if (!inHeaderFooter) {
+        parseParagraph2(el, blocks, paraShapeMap, sectionNum);
+      }
+      continue;
+    }
+    if (tag === "TABLE") {
+      if (!inHeaderFooter) {
+        parseTable2(el, blocks, paraShapeMap, sectionNum, warnings);
+      }
+      continue;
+    }
+    if (tag === "PARALIST" || tag === "SECTION" || tag === "COLDEF") {
+      walkContent(el, blocks, paraShapeMap, sectionNum, warnings, inHeaderFooter, depth + 1);
+      continue;
+    }
+    walkContent(el, blocks, paraShapeMap, sectionNum, warnings, inHeaderFooter, depth + 1);
+  }
+}
+function parseParagraph2(el, blocks, paraShapeMap, sectionNum) {
+  const paraShapeId = el.getAttribute("ParaShape") ?? "";
+  const shapeInfo = paraShapeMap.get(paraShapeId);
+  const text = extractParagraphText(el);
+  if (!text) return;
+  if (shapeInfo?.headingLevel != null) {
+    blocks.push({ type: "heading", text, level: shapeInfo.headingLevel, pageNumber: sectionNum });
+  } else {
+    blocks.push({ type: "paragraph", text, pageNumber: sectionNum });
+  }
+}
+function extractParagraphText(p) {
+  const parts = [];
+  collectCharText(p, parts);
+  return parts.join("").trim();
+}
+function collectCharText(node, parts, depth = 0) {
+  if (depth > MAX_XML_DEPTH2) return;
+  const children = node.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const el = children[i];
+    if (el.nodeType !== 1) continue;
+    const tag = localName(el);
+    if (tag === "CHAR") {
+      const t = textContent(el);
+      if (t) parts.push(t);
+    } else if (tag === "TABLE" || tag === "PICTURE" || tag === "SHAPEOBJECT") {
+    } else if (tag === "AUTONUM") {
+    } else {
+      collectCharText(el, parts, depth + 1);
+    }
+  }
+}
+function parseTable2(el, blocks, paraShapeMap, sectionNum, warnings) {
+  const cells = [];
+  const rowCount = parseInt(el.getAttribute("RowCount") ?? "0", 10);
+  const colCount = parseInt(el.getAttribute("ColCount") ?? "0", 10);
+  if (isNaN(rowCount) || isNaN(colCount) || rowCount === 0 || colCount === 0) return;
+  if (rowCount > MAX_TABLE_ROWS || colCount > MAX_TABLE_COLS) {
+    warnings.push({ message: `\uD14C\uC774\uBE14 \uD06C\uAE30 \uCD08\uACFC (${rowCount}x${colCount}) \u2014 \uC2A4\uD0B5`, code: "TRUNCATED_TABLE" });
+    return;
+  }
+  const children = el.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const rowEl = children[i];
+    if (rowEl.nodeType !== 1 || localName(rowEl) !== "ROW") continue;
+    const rowCells = rowEl.childNodes;
+    for (let j = 0; j < rowCells.length; j++) {
+      const cellEl = rowCells[j];
+      if (cellEl.nodeType !== 1 || localName(cellEl) !== "CELL") continue;
+      const colAddr = parseInt(cellEl.getAttribute("ColAddr") ?? "0", 10);
+      const rowAddr = parseInt(cellEl.getAttribute("RowAddr") ?? "0", 10);
+      const colSpan = Math.min(Math.max(1, parseInt(cellEl.getAttribute("ColSpan") ?? "1", 10) || 1), MAX_TABLE_COLS);
+      const rowSpan = Math.min(Math.max(1, parseInt(cellEl.getAttribute("RowSpan") ?? "1", 10) || 1), MAX_TABLE_ROWS);
+      const cellText = extractCellText(cellEl);
+      cells.push({ text: cellText, colSpan, rowSpan, colAddr, rowAddr });
+    }
+  }
+  if (cells.length === 0) return;
+  const grid = Array.from({ length: rowCount }, () => Array(colCount).fill(null));
+  for (const cell of cells) {
+    const r = cell.rowAddr ?? 0;
+    const c = cell.colAddr ?? 0;
+    if (isNaN(r) || isNaN(c) || r >= rowCount || c >= colCount) continue;
+    grid[r][c] = cell;
+    for (let dr = 0; dr < cell.rowSpan; dr++) {
+      for (let dc = 0; dc < cell.colSpan; dc++) {
+        if (dr === 0 && dc === 0) continue;
+        if (r + dr < rowCount && c + dc < colCount) {
+          grid[r + dr][c + dc] = { text: "", colSpan: 1, rowSpan: 1 };
+        }
+      }
+    }
+  }
+  const cellRows = grid.map(
+    (row) => row.map((cell) => cell ?? { text: "", colSpan: 1, rowSpan: 1 })
+  );
+  const table = buildTable(cellRows);
+  blocks.push({ type: "table", table, pageNumber: sectionNum });
+}
+function extractCellText(cellEl) {
+  const textParts = [];
+  collectCellText(cellEl, textParts, 0);
+  return textParts.filter(Boolean).join("\n").trim();
+}
+function collectCellText(node, parts, depth) {
+  if (depth > 20) return;
+  const children = node.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const el = children[i];
+    if (el.nodeType !== 1) continue;
+    const tag = localName(el);
+    if (tag === "P") {
+      const t = extractParagraphText(el);
+      if (t) parts.push(t);
+    } else if (tag === "TABLE") {
+      parts.push("[\uC911\uCCA9 \uD14C\uC774\uBE14]");
+    } else {
+      collectCellText(el, parts, depth + 1);
+    }
+  }
+}
+function localName(el) {
+  return (el.tagName || el.localName || "").replace(/^[^:]+:/, "");
+}
+function findChild(parent, tag) {
+  const children = parent.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const el = children[i];
+    if (el.nodeType === 1 && localName(el) === tag) return el;
+  }
+  return null;
+}
+function textContent(el) {
+  const children = el.childNodes;
+  const parts = [];
+  for (let i = 0; i < children.length; i++) {
+    const node = children[i];
+    if (node.nodeType === 3) {
+      parts.push(node.nodeValue || "");
+    } else if (node.nodeType === 1) {
+      parts.push(textContent(node));
+    }
+  }
+  return parts.join("");
+}
+function countSections(body) {
+  let count = 0;
+  const children = body.childNodes;
+  for (let i = 0; i < children.length; i++) {
+    const el = children[i];
+    if (el.nodeType === 1 && localName(el) === "SECTION") count++;
+  }
+  return count;
+}
+
 // src/form/recognize.ts
 var LABEL_KEYWORDS = /* @__PURE__ */ new Set([
   "\uC131\uBA85",
@@ -3570,7 +3862,7 @@ function fillInlineFields(text, values, filled, matchedLabels) {
 
 // src/form/filler-hwpx.ts
 import JSZip5 from "jszip";
-import { DOMParser as DOMParser4, XMLSerializer } from "@xmldom/xmldom";
+import { DOMParser as DOMParser5, XMLSerializer } from "@xmldom/xmldom";
 async function fillHwpx(hwpxBuffer, values) {
   const zip = await JSZip5.loadAsync(hwpxBuffer);
   const filled = [];
@@ -3580,7 +3872,7 @@ async function fillHwpx(hwpxBuffer, values) {
   if (sectionFiles.length === 0) {
     throw new KordocError("HWPX\uC5D0\uC11C \uC139\uC158 \uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
   }
-  const xmlParser = new DOMParser4();
+  const xmlParser = new DOMParser5();
   const xmlSerializer = new XMLSerializer();
   for (const sectionPath of sectionFiles) {
     const zipEntry = zip.file(sectionPath);
@@ -3612,10 +3904,10 @@ async function fillHwpx(hwpxBuffer, values) {
         const trEl = rows[rowIdx];
         const cells = findDirectChildren(trEl, "tc");
         for (let colIdx = 0; colIdx < cells.length - 1; colIdx++) {
-          const labelText = extractCellText(cells[colIdx]);
+          const labelText = extractCellText2(cells[colIdx]);
           if (!isLabelCell(labelText)) continue;
           const valueCell = cells[colIdx + 1];
-          const valueText = extractCellText(valueCell);
+          const valueText = extractCellText2(valueCell);
           if (isKeywordLabel(valueText)) continue;
           const normalizedCellLabel = normalizeLabel(labelText);
           if (!normalizedCellLabel) continue;
@@ -3640,14 +3932,14 @@ async function fillHwpx(hwpxBuffer, values) {
       if (rows.length >= 2) {
         const headerCells = findDirectChildren(rows[0], "tc");
         const allLabels = headerCells.every((cell) => {
-          const t = extractCellText(cell).trim();
+          const t = extractCellText2(cell).trim();
           return t.length > 0 && t.length <= 20 && isLabelCell(t);
         });
         if (allLabels) {
           for (let rowIdx = 1; rowIdx < rows.length; rowIdx++) {
             const dataCells = findDirectChildren(rows[rowIdx], "tc");
             for (let colIdx = 0; colIdx < Math.min(headerCells.length, dataCells.length); colIdx++) {
-              const headerLabel = normalizeLabel(extractCellText(headerCells[colIdx]));
+              const headerLabel = normalizeLabel(extractCellText2(headerCells[colIdx]));
               const matchKey = findMatchingKey(headerLabel, normalizedValues);
               if (matchKey === void 0) continue;
               if (matchedLabels.has(matchKey)) continue;
@@ -3655,7 +3947,7 @@ async function fillHwpx(hwpxBuffer, values) {
               replaceCellText(dataCells[colIdx], newValue);
               matchedLabels.add(matchKey);
               filled.push({
-                label: extractCellText(headerCells[colIdx]).trim(),
+                label: extractCellText2(headerCells[colIdx]).trim(),
                 value: newValue,
                 row: rowIdx,
                 col: colIdx
@@ -3697,7 +3989,7 @@ async function fillHwpx(hwpxBuffer, values) {
   const buffer = await zip.generateAsync({ type: "arraybuffer" });
   return { buffer, filled, unmatched };
 }
-function localName(el) {
+function localName2(el) {
   return (el.tagName || el.localName || "").replace(/^[^:]+:/, "");
 }
 function findAllElements(node, tagLocalName) {
@@ -3708,7 +4000,7 @@ function findAllElements(node, tagLocalName) {
     for (let i = 0; i < children.length; i++) {
       const child = children[i];
       if (child.nodeType !== 1) continue;
-      if (localName(child) === tagLocalName) result.push(child);
+      if (localName2(child) === tagLocalName) result.push(child);
       walk(child);
     }
   };
@@ -3721,7 +4013,7 @@ function findDirectChildren(parent, tagLocalName) {
   if (!children) return result;
   for (let i = 0; i < children.length; i++) {
     const child = children[i];
-    if (child.nodeType === 1 && localName(child) === tagLocalName) {
+    if (child.nodeType === 1 && localName2(child) === tagLocalName) {
       result.push(child);
     }
   }
@@ -3730,12 +4022,12 @@ function findDirectChildren(parent, tagLocalName) {
 function isInsideTable(el) {
   let parent = el.parentNode;
   while (parent) {
-    if (parent.nodeType === 1 && localName(parent) === "tbl") return true;
+    if (parent.nodeType === 1 && localName2(parent) === "tbl") return true;
     parent = parent.parentNode;
   }
   return false;
 }
-function extractCellText(tcEl) {
+function extractCellText2(tcEl) {
   const parts = [];
   const walk = (node) => {
     const children = node.childNodes;
@@ -3745,7 +4037,7 @@ function extractCellText(tcEl) {
       if (child.nodeType === 3) {
         parts.push(child.textContent || "");
       } else if (child.nodeType === 1) {
-        const tag = localName(child);
+        const tag = localName2(child);
         if (tag === "t") walk(child);
         else if (tag === "run" || tag === "r" || tag === "p" || tag === "subList") walk(child);
         else if (tag === "tab") parts.push("	");
@@ -4470,6 +4762,8 @@ async function parse(input, options) {
     }
     case "hwp":
       return parseHwp(buffer, options);
+    case "hwpml":
+      return parseHwpml(buffer, options);
     case "pdf":
       return parsePdf(buffer, options);
     default:
@@ -4495,7 +4789,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-OIRWPKIQ.js");
+    const mod = await import("./parser-25LF2S2J.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -4529,6 +4823,14 @@ async function parseDocx(buffer, options) {
     return { success: false, fileType: "docx", error: err instanceof Error ? err.message : "DOCX \uD30C\uC2F1 \uC2E4\uD328", code: classifyError(err) };
   }
 }
+async function parseHwpml(buffer, options) {
+  try {
+    const { markdown, blocks, metadata, outline, warnings } = parseHwpmlDocument(buffer, options);
+    return { success: true, fileType: "hwpml", markdown, blocks, metadata, outline, warnings };
+  } catch (err) {
+    return { success: false, fileType: "hwpml", error: err instanceof Error ? err.message : "HWPML \uD30C\uC2F1 \uC2E4\uD328", code: classifyError(err) };
+  }
+}
 async function fillForm(input, values, outputFormat = "markdown") {
   let buffer;
   if (typeof input === "string") {
@@ -4588,6 +4890,7 @@ export {
   parse,
   parseDocx,
   parseHwp,
+  parseHwpml,
   parseHwpx,
   parsePdf,
   parseXlsx