kordoc 4.0.6 → 4.0.8

package/dist/{parser-VEY4B5CJ.cjs → parser-ITMU2WN5.cjs}

@@ -6,7 +6,8 @@
 
 
 
-var _chunkDQJTNXVScjs = require('./chunk-DQJTNXVS.cjs');
+
+var _chunk6XAAYAWLcjs = require('./chunk-6XAAYAWL.cjs');
 
 
 var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
@@ -1223,10 +1224,180 @@ function normalizeUndersegmentedTable(originalCells, colXs, items) {
   return rebuilt;
 }
 
+// src/pdf/image-extract.ts
+
+var IMAGE_RESOLVE_TIMEOUT_MS = 5e3;
+function resolveImgData(page, objId) {
+  const store = objId.startsWith("g_") ? page.commonObjs : page.objs;
+  return new Promise((resolve) => {
+    let done = false;
+    const timer = setTimeout(() => {
+      if (!done) {
+        done = true;
+        resolve(null);
+      }
+    }, IMAGE_RESOLVE_TIMEOUT_MS);
+    try {
+      store.get(objId, (data) => {
+        if (!done) {
+          done = true;
+          clearTimeout(timer);
+          resolve(_nullishCoalesce(data, () => ( null)));
+        }
+      });
+    } catch (e2) {
+      if (!done) {
+        done = true;
+        clearTimeout(timer);
+        resolve(null);
+      }
+    }
+  });
+}
+var MIN_DIM = 8;
+var MAX_PIXELS = 36e6;
+var MAX_IMAGES_PER_DOC = 200;
+var MAX_TOTAL_IMAGE_BYTES = 128 * 1024 * 1024;
+function createPdfImageState() {
+  return { imageIndex: 0, totalBytes: 0, seen: /* @__PURE__ */ new Map(), capWarned: false };
+}
+function fnv1a(data) {
+  let h = 2166136261;
+  for (let i = 0; i < data.length; i++) {
+    h ^= data[i];
+    h = Math.imul(h, 16777619);
+  }
+  return h >>> 0;
+}
+function toRgba(img) {
+  const { width: w, height: h, kind, data } = img;
+  if (!data || !w || !h) return null;
+  const rgba = new Uint8Array(w * h * 4);
+  if (kind === _pdfmjs.ImageKind.RGBA_32BPP) {
+    if (data.length < w * h * 4) return null;
+    rgba.set(data.subarray(0, w * h * 4));
+    return rgba;
+  }
+  if (kind === _pdfmjs.ImageKind.RGB_24BPP) {
+    if (data.length < w * h * 3) return null;
+    for (let i = 0, s = 0, d = 0; i < w * h; i++, s += 3, d += 4) {
+      rgba[d] = data[s];
+      rgba[d + 1] = data[s + 1];
+      rgba[d + 2] = data[s + 2];
+      rgba[d + 3] = 255;
+    }
+    return rgba;
+  }
+  if (kind === _pdfmjs.ImageKind.GRAYSCALE_1BPP) {
+    const stride = w + 7 >> 3;
+    if (data.length < stride * h) return null;
+    for (let y = 0; y < h; y++) {
+      for (let x = 0; x < w; x++) {
+        const bit = data[y * stride + (x >> 3)] >> 7 - (x & 7) & 1;
+        const v = bit ? 255 : 0;
+        const d = (y * w + x) * 4;
+        rgba[d] = v;
+        rgba[d + 1] = v;
+        rgba[d + 2] = v;
+        rgba[d + 3] = 255;
+      }
+    }
+    return rgba;
+  }
+  return null;
+}
+async function extractPageImages(page, fnArray, argsArray, pageNumber, state, warnings) {
+  const blocks = [];
+  const images = [];
+  const pageSeenIds = /* @__PURE__ */ new Set();
+  for (let i = 0; i < fnArray.length; i++) {
+    const op = fnArray[i];
+    let imgData = null;
+    let dedupeId = null;
+    if (op === _pdfmjs.OPS.paintImageXObject || op === _pdfmjs.OPS.paintImageXObjectRepeat) {
+      const objId = _optionalChain([argsArray, 'access', _4 => _4[i], 'optionalAccess', _5 => _5[0]]);
+      if (typeof objId !== "string" || pageSeenIds.has(objId)) continue;
+      pageSeenIds.add(objId);
+      dedupeId = objId;
+      imgData = await resolveImgData(page, objId);
+    } else if (op === _pdfmjs.OPS.paintInlineImageXObject) {
+      imgData = _optionalChain([argsArray, 'access', _6 => _6[i], 'optionalAccess', _7 => _7[0]]);
+    } else {
+      continue;
+    }
+    if (!_optionalChain([imgData, 'optionalAccess', _8 => _8.data]) || !imgData.width || !imgData.height) continue;
+    const { width: w, height: h } = imgData;
+    if (w < MIN_DIM || h < MIN_DIM) continue;
+    if (w * h > MAX_PIXELS) {
+      warnings.push({ page: pageNumber, message: `\uC774\uBBF8\uC9C0 \uD06C\uAE30 \uCD08\uACFC\uB85C \uCD94\uCD9C \uC0DD\uB7B5 (${w}\xD7${h})`, code: "SKIPPED_IMAGE" });
+      continue;
+    }
+    const hash = `${w}x${h}k${_nullishCoalesce(imgData.kind, () => ( "?"))}h${fnv1a(imgData.data)}${dedupeId ? "" : "inline"}`;
+    if (state.seen.has(hash)) continue;
+    if (state.imageIndex >= MAX_IMAGES_PER_DOC || state.totalBytes >= MAX_TOTAL_IMAGE_BYTES) {
+      if (!state.capWarned) {
+        state.capWarned = true;
+        warnings.push({ page: pageNumber, message: `\uC774\uBBF8\uC9C0 \uCD94\uCD9C \uC0C1\uD55C \uB3C4\uB2EC (${MAX_IMAGES_PER_DOC}\uAC1C/128MB) \u2014 \uC774\uD6C4 \uC774\uBBF8\uC9C0 \uC0DD\uB7B5`, code: "SKIPPED_IMAGE" });
+      }
+      continue;
+    }
+    const rgba = toRgba(imgData);
+    if (!rgba) continue;
+    const png = _chunk6XAAYAWLcjs.encodePng.call(void 0, w, h, rgba);
+    state.imageIndex++;
+    state.totalBytes += png.length;
+    const filename = `image_${String(state.imageIndex).padStart(3, "0")}.png`;
+    state.seen.set(hash, filename);
+    images.push({ filename, data: png, mimeType: "image/png" });
+    blocks.push({ type: "image", text: filename, pageNumber });
+  }
+  return { blocks, images };
+}
+function injectPageImageBlocks(blocks, pageImages) {
+  if (pageImages.size === 0) return;
+  const lastIndexForPage = /* @__PURE__ */ new Map();
+  for (let i = 0; i < blocks.length; i++) {
+    const p = blocks[i].pageNumber;
+    if (p !== void 0) lastIndexForPage.set(p, i);
+  }
+  const pageAtIndex = /* @__PURE__ */ new Map();
+  for (const [p, last] of lastIndexForPage) pageAtIndex.set(last, p);
+  const result = [];
+  const injected = /* @__PURE__ */ new Set();
+  for (let i = 0; i < blocks.length; i++) {
+    result.push(blocks[i]);
+    const p = pageAtIndex.get(i);
+    if (p !== void 0 && pageImages.has(p)) {
+      result.push(...pageImages.get(p));
+      injected.add(p);
+    }
+  }
+  for (const p of [...pageImages.keys()].sort((a, b) => a - b)) {
+    if (injected.has(p)) continue;
+    let at = result.length;
+    for (let i = 0; i < result.length; i++) {
+      const bp = result[i].pageNumber;
+      if (bp !== void 0 && bp > p) {
+        at = i;
+        break;
+      }
+    }
+    result.splice(at, 0, ...pageImages.get(p));
+  }
+  blocks.length = 0;
+  blocks.push(...result);
+}
+
 // src/pdf/quality.ts
+var RARE_JONG = /* @__PURE__ */ new Set([3, 5, 6, 9, 10, 11, 12, 13, 14, 15, 18, 24, 25, 26]);
+var MOJIBAKE_MIN_HANGUL = 30;
+var MOJIBAKE_MAX_NOBATCHIM = 0.15;
+var MOJIBAKE_MIN_RAREBATCHIM = 0.25;
 function computePageQuality(page, text) {
   let total = 0;
   let hangul = 0;
+  let hangulNoBatchim = 0;
+  let hangulRareBatchim = 0;
   let control = 0;
   let replacement = 0;
   let pua = 0;
@@ -1244,6 +1415,9 @@ function computePageQuality(page, text) {
     }
     if (code >= 44032 && code <= 55203) {
       hangul++;
+      const jong = (code - 44032) % 28;
+      if (jong === 0) hangulNoBatchim++;
+      else if (RARE_JONG.has(jong)) hangulRareBatchim++;
       continue;
     }
     if (code >= 57344 && code <= 63743 || code >= 56192 && code <= 56319) {
@@ -1255,6 +1429,9 @@ function computePageQuality(page, text) {
   const puaRatio = pua / denom;
   const controlCharRatio = control / denom;
   const replacementCharRatio = replacement / denom;
+  const hangulNoBatchimRatio = hangul > 0 ? hangulNoBatchim / hangul : 0;
+  const hangulRareBatchimRatio = hangul > 0 ? hangulRareBatchim / hangul : 0;
+  const garbledHangul = hangul >= MOJIBAKE_MIN_HANGUL && hangulNoBatchimRatio < MOJIBAKE_MAX_NOBATCHIM && hangulRareBatchimRatio >= MOJIBAKE_MIN_RAREBATCHIM;
   let needsOcr = false;
   let ocrReason;
   if (total < LOW_TEXT_THRESHOLD) {
@@ -1269,6 +1446,9 @@ function computePageQuality(page, text) {
   } else if (replacementCharRatio >= HIGH_REPLACEMENT_THRESHOLD) {
     needsOcr = true;
     ocrReason = "high_replacement";
+  } else if (garbledHangul) {
+    needsOcr = true;
+    ocrReason = "garbled_hangul";
   }
   return {
     page,
@@ -1277,6 +1457,8 @@ function computePageQuality(page, text) {
     controlCharRatio,
     replacementCharRatio,
     puaRatio,
+    hangulNoBatchimRatio,
+    hangulRareBatchimRatio,
     needsOcr,
     ocrReason
   };
@@ -1391,7 +1573,7 @@ function dominantStyle(items) {
     }
   }
   if (dominantSize === 0) return void 0;
-  const fontName = _optionalChain([items, 'access', _4 => _4.find, 'call', _5 => _5((i) => i.fontSize === dominantSize), 'optionalAccess', _6 => _6.fontName]) || void 0;
+  const fontName = _optionalChain([items, 'access', _9 => _9.find, 'call', _10 => _10((i) => i.fontSize === dominantSize), 'optionalAccess', _11 => _11.fontName]) || void 0;
   return fontName ? { fontSize: dominantSize, fontName } : { fontSize: dominantSize };
 }
 function normalizeItems(rawItems) {
@@ -1512,7 +1694,7 @@ function mergeSuperscriptLines(lines) {
   return result;
 }
 function mergeLineSimple(items) {
-  if (items.length <= 1) return _optionalChain([items, 'access', _7 => _7[0], 'optionalAccess', _8 => _8.text]) || "";
+  if (items.length <= 1) return _optionalChain([items, 'access', _12 => _12[0], 'optionalAccess', _13 => _13.text]) || "";
   const sorted = [...items].sort((a, b) => a.x - b.x);
   const isEvenSpaced = detectEvenSpacedItems(sorted);
   let result = sorted[0].text;
@@ -1599,12 +1781,12 @@ function isTwoColumnProse(r) {
   if (t.rows < 8) return false;
   const dense = [];
   for (let c = 0; c < t.cols; c++) {
-    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _9 => _9[c], 'optionalAccess', _10 => _10.text, 'access', _11 => _11.trim, 'call', _12 => _12()])).length;
+    const filled = t.cells.filter((row) => _optionalChain([row, 'access', _14 => _14[c], 'optionalAccess', _15 => _15.text, 'access', _16 => _16.trim, 'call', _17 => _17()])).length;
     if (filled / t.rows >= 0.3) dense.push(c);
   }
   if (dense.length !== 2) return false;
   for (const c of dense) {
-    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _13 => _13[c], 'optionalAccess', _14 => _14.text, 'access', _15 => _15.replace, 'call', _16 => _16(/\s+/g, ""), 'access', _17 => _17.length]), () => ( 0))).filter((n) => n > 0);
+    const lens = t.cells.map((row) => _nullishCoalesce(_optionalChain([row, 'access', _18 => _18[c], 'optionalAccess', _19 => _19.text, 'access', _20 => _20.replace, 'call', _21 => _21(/\s+/g, ""), 'access', _22 => _22.length]), () => ( 0))).filter((n) => n > 0);
     const avg = lens.reduce((s, v) => s + v, 0) / (lens.length || 1);
     if (avg < 12) return false;
   }
@@ -2081,7 +2263,7 @@ function buildClusterTable(rows, columns, pageNum) {
     const nonEmptyCols = cells[r].filter((c) => c.text.trim()).length;
     if (nonEmptyCols !== 1) continue;
     if (cells[r][0].text.trim() !== "") continue;
-    const contentText = _optionalChain([cells, 'access', _18 => _18[r], 'access', _19 => _19.find, 'call', _20 => _20((c) => c.text.trim()), 'optionalAccess', _21 => _21.text, 'access', _22 => _22.trim, 'call', _23 => _23()]) || "";
+    const contentText = _optionalChain([cells, 'access', _23 => _23[r], 'access', _24 => _24.find, 'call', _25 => _25((c) => c.text.trim()), 'optionalAccess', _26 => _26.text, 'access', _27 => _27.trim, 'call', _28 => _28()]) || "";
     if (/^[○●▶\-·]/.test(contentText)) continue;
     for (let pr = r - 1; pr >= 0; pr--) {
       if (cells[pr].some((c) => c.text.trim())) {
@@ -2303,14 +2485,14 @@ function isProseSpread(items) {
   for (let i = 1; i < sorted.length; i++) {
     gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
   }
-  const maxGap = _chunkDQJTNXVScjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunk6XAAYAWLcjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   return maxGap < 40 && avgLen < 5;
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunkDQJTNXVScjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkDQJTNXVScjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunk6XAAYAWLcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunk6XAAYAWLcjs.safeMin.call(void 0, allItems.map((i) => i.x));
   if (pageWidth < 100) return null;
   let bigoLineIdx = -1;
   for (let i = 0; i < yLines.length; i++) {
@@ -2526,15 +2708,15 @@ function computeMedianFontSizeFromFreq(freq) {
 }
 function detectHeadings(blocks, medianFontSize) {
   for (const block of blocks) {
-    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _24 => _24.style, 'optionalAccess', _25 => _25.fontSize])) continue;
+    if (block.type !== "paragraph" || !block.text || !_optionalChain([block, 'access', _29 => _29.style, 'optionalAccess', _30 => _30.fontSize])) continue;
     const text = block.text.trim();
     if (text.length === 0 || text.length > 200) continue;
     if (/^\d+$/.test(text)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunkDQJTNXVScjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunkDQJTNXVScjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunkDQJTNXVScjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunk6XAAYAWLcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunk6XAAYAWLcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunk6XAAYAWLcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -2587,7 +2769,7 @@ function detectMarkerHeadings(blocks) {
       block.level = 4;
       continue;
     }
-    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _26 => _26.style, 'optionalAccess', _27 => _27.fontSize])) {
+    if (/^[가-힣]{2,6}$/.test(text) && _optionalChain([block, 'access', _31 => _31.style, 'optionalAccess', _32 => _32.fontSize])) {
       const prev = blocks[i - 1];
       const next = blocks[i + 1];
       const prevIsStructural = !prev || prev.type === "table" || prev.type === "heading" || prev.type === "separator";
@@ -2840,7 +3022,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
   const bottomEntries = [];
   for (let bi = 0; bi < blocks.length; bi++) {
     const b = blocks[bi];
-    if (!b.bbox || !b.pageNumber || !_optionalChain([b, 'access', _28 => _28.text, 'optionalAccess', _29 => _29.trim, 'call', _30 => _30()])) continue;
+    if (!b.bbox || !b.pageNumber || !_optionalChain([b, 'access', _33 => _33.text, 'optionalAccess', _34 => _34.trim, 'call', _35 => _35()])) continue;
     const ph = pageHeights.get(b.bbox.page) || pageHeights.get(b.pageNumber);
     if (!ph) continue;
     const blockTop = ph - (b.bbox.y + b.bbox.height);
@@ -2863,7 +3045,7 @@ function removeHeaderFooterBlocks(blocks, pageHeights, warnings) {
     }
     const repeatedPatterns = /* @__PURE__ */ new Set();
     for (const [p, count] of patternCount) {
-      if (count >= MIN_REPEAT && (_nullishCoalesce(_optionalChain([patternPages, 'access', _31 => _31.get, 'call', _32 => _32(p), 'optionalAccess', _33 => _33.size]), () => ( 0))) >= MIN_REPEAT) {
+      if (count >= MIN_REPEAT && (_nullishCoalesce(_optionalChain([patternPages, 'access', _36 => _36.get, 'call', _37 => _37(p), 'optionalAccess', _38 => _38.size]), () => ( 0))) >= MIN_REPEAT) {
         repeatedPatterns.add(p);
       }
     }
@@ -3111,7 +3293,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     }
     if (remaining.length > 0) {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunkDQJTNXVScjs.safeMax.call(void 0, allY) - _chunkDQJTNXVScjs.safeMin.call(void 0, allY);
+      const pageH = _chunk6XAAYAWLcjs.safeMax.call(void 0, allY) - _chunk6XAAYAWLcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -3287,7 +3469,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false) {
       blocks.push({ type: "paragraph", text: tableText, pageNumber: pageNum, bbox, style: dominantStyle(items) });
     } else {
       const allY = items.map((i) => i.y);
-      const pageHeight = _chunkDQJTNXVScjs.safeMax.call(void 0, allY) - _chunkDQJTNXVScjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunk6XAAYAWLcjs.safeMax.call(void 0, allY) - _chunk6XAAYAWLcjs.safeMin.call(void 0, allY);
       const gapThreshold = Math.max(15, pageHeight * 0.03);
       const orderedGroups = proseCutX !== null ? splitTwoColumnProse(items, proseCutX) : xyCutOrder(items, gapThreshold);
       for (const group of orderedGroups) {
@@ -3523,7 +3705,7 @@ try {
   pdfjsAssets.cMapUrl = _path.join.call(void 0, pkgDir, "cmaps") + "/";
   pdfjsAssets.cMapPacked = true;
   pdfjsAssets.standardFontDataUrl = _path.join.call(void 0, pkgDir, "standard_fonts") + "/";
-} catch (e2) {
+} catch (e3) {
 }
 async function loadPdfWithTimeout(buffer) {
   const loadingTask = _pdfmjs.getDocument.call(void 0, {
@@ -3540,7 +3722,7 @@ async function loadPdfWithTimeout(buffer) {
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunkDQJTNXVScjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunk6XAAYAWLcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -3549,11 +3731,11 @@ async function loadPdfWithTimeout(buffer) {
   }
 }
 async function parsePdfDocument(buffer, options) {
-  const formulaBuffer = _optionalChain([options, 'optionalAccess', _34 => _34.formulaOcr]) ? buffer.slice(0) : null;
+  const formulaBuffer = _optionalChain([options, 'optionalAccess', _39 => _39.formulaOcr]) ? buffer.slice(0) : null;
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunkDQJTNXVScjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunk6XAAYAWLcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -3562,12 +3744,15 @@ async function parsePdfDocument(buffer, options) {
     let totalChars = 0;
     let totalTextBytes = 0;
     const effectivePageCount = Math.min(pageCount, MAX_PAGES);
-    const pageFilter = _optionalChain([options, 'optionalAccess', _35 => _35.pages]) ? _chunkDCZVOIEOcjs.parsePageRange.call(void 0, options.pages, effectivePageCount) : null;
+    const pageFilter = _optionalChain([options, 'optionalAccess', _40 => _40.pages]) ? _chunkDCZVOIEOcjs.parsePageRange.call(void 0, options.pages, effectivePageCount) : null;
     const totalTarget = pageFilter ? pageFilter.size : effectivePageCount;
     const fontSizeFreq = /* @__PURE__ */ new Map();
     const pageHeights = /* @__PURE__ */ new Map();
     const pagesWithLargeImage = /* @__PURE__ */ new Set();
     const skippedImagePages = /* @__PURE__ */ new Map();
+    const imageState = createPdfImageState();
+    const extractedImages = [];
+    const pageImageBlocks = /* @__PURE__ */ new Map();
     let parsedPages = 0;
     for (let i = 1; i <= effectivePageCount; i++) {
       if (pageFilter && !pageFilter.has(i)) continue;
@@ -3605,6 +3790,12 @@ async function parsePdfDocument(buffer, options) {
         }
         const pageBlocks = extractPageBlocksWithLines(visible, i, opList, viewport.width, viewport.height);
         for (const b of pageBlocks) blocks.push(b);
+        try {
+          const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings);
+          if (imgBlocks.length > 0) pageImageBlocks.set(i, imgBlocks);
+          extractedImages.push(...pageImages);
+        } catch (e4) {
+        }
         let pageText = "";
         for (const b of pageBlocks) {
           const t = b.text || "";
@@ -3613,26 +3804,26 @@ async function parsePdfDocument(buffer, options) {
           pageText += pageText ? "\n" + t : t;
         }
         pageQuality.push(computePageQuality(i, pageText));
-        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkDQJTNXVScjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunk6XAAYAWLcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
         parsedPages++;
-        _optionalChain([options, 'optionalAccess', _36 => _36.onProgress, 'optionalCall', _37 => _37(parsedPages, totalTarget)]);
+        _optionalChain([options, 'optionalAccess', _41 => _41.onProgress, 'optionalCall', _42 => _42(parsedPages, totalTarget)]);
       } catch (pageErr) {
-        if (pageErr instanceof _chunkDQJTNXVScjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunk6XAAYAWLcjs.KordocError) throw pageErr;
         warnings.push({ page: i, message: `\uD398\uC774\uC9C0 ${i} \uD30C\uC2F1 \uC2E4\uD328: ${pageErr instanceof Error ? pageErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
       }
     }
     const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount);
     let isImageBased = false;
     if (totalChars / Math.max(parsedPageCount, 1) < 10) {
-      if (_optionalChain([options, 'optionalAccess', _38 => _38.ocr])) {
+      if (_optionalChain([options, 'optionalAccess', _43 => _43.ocr])) {
         try {
-          const { ocrPages } = await Promise.resolve().then(() => _interopRequireWildcard(require("./provider-5K7O3Z2O.cjs")));
+          const { ocrPages } = await Promise.resolve().then(() => _interopRequireWildcard(require("./provider-G4C2V2PD.cjs")));
           const ocrBlocks = await ocrPages(doc, options.ocr, pageFilter, effectivePageCount);
           if (ocrBlocks.length > 0) {
             const ocrMarkdown = ocrBlocks.map((b) => b.text || "").filter(Boolean).join("\n\n");
-            return { markdown: ocrMarkdown, blocks: ocrBlocks, metadata, warnings, isImageBased: true, pageQuality, qualitySummary: summarizeDocumentQuality(pageQuality) };
+            return { markdown: ocrMarkdown, blocks: ocrBlocks, metadata, warnings, isImageBased: true, pageQuality, qualitySummary: summarizeDocumentQuality(pageQuality), images: extractedImages.length > 0 ? extractedImages : void 0 };
           }
-        } catch (e3) {
+        } catch (e5) {
         }
       }
       isImageBased = true;
@@ -3646,7 +3837,8 @@ async function parsePdfDocument(buffer, options) {
         low_text: "\uD14D\uC2A4\uD2B8\uAC00 \uAC70\uC758 \uC5C6\uB294 \uD398\uC774\uC9C0 (\uC2A4\uCE94/\uC774\uBBF8\uC9C0 \uCD94\uC815)",
         high_pua: "\uAE00\uAF34 \uB9E4\uD551 \uC2E4\uD328 (PUA \uBE44\uC728 \uB192\uC74C) \u2014 \uCD94\uCD9C \uD14D\uC2A4\uD2B8 \uC2E0\uB8B0 \uBD88\uAC00",
         high_control: "\uC81C\uC5B4\uBB38\uC790 \uBE44\uC728 \uB192\uC74C \u2014 \uCD94\uCD9C \uD14D\uC2A4\uD2B8 \uC2E0\uB8B0 \uBD88\uAC00",
-        high_replacement: "\uB300\uCCB4\uBB38\uC790(U+FFFD) \uBE44\uC728 \uB192\uC74C \u2014 \uCD94\uCD9C \uD14D\uC2A4\uD2B8 \uC2E0\uB8B0 \uBD88\uAC00"
+        high_replacement: "\uB300\uCCB4\uBB38\uC790(U+FFFD) \uBE44\uC728 \uB192\uC74C \u2014 \uCD94\uCD9C \uD14D\uC2A4\uD2B8 \uC2E0\uB8B0 \uBD88\uAC00",
+        garbled_hangul: "\uAE00\uAF34 \uB9E4\uD551 \uC2E4\uD328 (\uD55C\uAE00 \uC790\uC18C \uBD84\uD3EC \uC774\uC0C1) \u2014 \uCD94\uCD9C \uD14D\uC2A4\uD2B8\uAC00 \uAE68\uC84C\uC744 \uC218 \uC788\uC74C"
       };
       for (const pq of pageQuality) {
         if (!pq.needsOcr || !pq.ocrReason) continue;
@@ -3659,14 +3851,15 @@ async function parsePdfDocument(buffer, options) {
         warnings.push({ page, message: `${count}\uAC1C \uC774\uBBF8\uC9C0 \uC601\uC5ED\uC5D0 \uCD94\uCD9C \uAC00\uB2A5\uD55C \uD14D\uC2A4\uD2B8 \uC5C6\uC74C (\uADF8\uB9BC/\uCC28\uD2B8/\uB3C4\uC7A5 \uB0B4\uC6A9 \uB204\uB77D \uAC00\uB2A5)`, code: "SKIPPED_IMAGE" });
       }
     }
-    if (_optionalChain([options, 'optionalAccess', _39 => _39.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
+    if (_optionalChain([options, 'optionalAccess', _44 => _44.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
       const removed = removeHeaderFooterBlocks(blocks, pageHeights, warnings);
       for (let ri = removed.length - 1; ri >= 0; ri--) {
         blocks.splice(removed[ri], 1);
       }
     }
     mergeCrossPageTables(blocks);
-    if (_optionalChain([options, 'optionalAccess', _40 => _40.formulaOcr]) && formulaBuffer) {
+    injectPageImageBlocks(blocks, pageImageBlocks);
+    if (_optionalChain([options, 'optionalAccess', _45 => _45.formulaOcr]) && formulaBuffer) {
       try {
         await applyFormulaOcr(formulaBuffer, blocks, pageFilter, effectivePageCount, warnings, options.onProgress);
       } catch (e) {
@@ -3685,7 +3878,7 @@ async function parsePdfDocument(buffer, options) {
     detectKoreanListBlocks(blocks);
     const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
     sanitizeBlockControlChars(blocks);
-    let markdown = cleanPdfText(_chunkDQJTNXVScjs.blocksToMarkdown.call(void 0, blocks));
+    let markdown = cleanPdfText(_chunk6XAAYAWLcjs.blocksToMarkdown.call(void 0, blocks));
     return {
       markdown,
       blocks,
@@ -3694,7 +3887,8 @@ async function parsePdfDocument(buffer, options) {
       warnings: warnings.length > 0 ? warnings : void 0,
       isImageBased: isImageBased || void 0,
       pageQuality,
-      qualitySummary: summarizeDocumentQuality(pageQuality)
+      qualitySummary: summarizeDocumentQuality(pageQuality),
+      images: extractedImages.length > 0 ? extractedImages : void 0
     };
   } finally {
     await doc.destroy().catch(() => {
@@ -3704,7 +3898,7 @@ async function parsePdfDocument(buffer, options) {
 async function extractPdfMetadata(doc, metadata) {
   try {
     const result = await doc.getMetadata();
-    if (!_optionalChain([result, 'optionalAccess', _41 => _41.info])) return;
+    if (!_optionalChain([result, 'optionalAccess', _46 => _46.info])) return;
     const info = result.info;
     if (typeof info.Title === "string" && info.Title.trim()) metadata.title = info.Title.trim();
     if (typeof info.Author === "string" && info.Author.trim()) metadata.author = info.Author.trim();
@@ -3715,7 +3909,7 @@ async function extractPdfMetadata(doc, metadata) {
     }
     if (typeof info.CreationDate === "string") metadata.createdAt = parsePdfDate(info.CreationDate);
     if (typeof info.ModDate === "string") metadata.modifiedAt = parsePdfDate(info.ModDate);
-  } catch (e4) {
+  } catch (e6) {
   }
 }
 function parsePdfDate(dateStr) {
@@ -3744,4 +3938,4 @@ async function extractPdfMetadataOnly(buffer) {
 
 
 exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.detectTableCaptions = detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks;
-//# sourceMappingURL=parser-VEY4B5CJ.cjs.map
+//# sourceMappingURL=parser-ITMU2WN5.cjs.map