kordoc 4.17.1 → 4.17.2

package/dist/mcp.js

@@ -3,7 +3,7 @@ import {
   assertWithinRoot,
   getAccessRoot,
   isOfflineMode
-} from "./chunk-N6MPDM55.js";
+} from "./chunk-6OCKAXIS.js";
 import {
   APPROVAL_MAX,
   BODY_FONTS,
@@ -43,25 +43,25 @@ import {
   resolveBuiltinTemplate,
   unknownFontWarnings,
   usesGaejosikMunche
-} from "./chunk-HAPCN7BD.js";
-import "./chunk-OVSCNCT6.js";
-import "./chunk-EULS3PDF.js";
+} from "./chunk-VU26WGE5.js";
+import "./chunk-O4CB7D2Q.js";
+import "./chunk-4HRJISJF.js";
 import "./chunk-3WSNLH4B.js";
-import "./chunk-66YTQ3NQ.js";
+import "./chunk-C4MVZXB5.js";
 import {
   detectFormat,
   detectOle2Format,
   detectZipFormat,
   extractHwp5MetadataOnly
-} from "./chunk-ELNYQRBN.js";
+} from "./chunk-JY2NBFIG.js";
 import "./chunk-7FCNOKOR.js";
 import "./chunk-WL3BWROT.js";
 import "./chunk-VQFA55IC.js";
-import "./chunk-B4PUQ4FP.js";
+import "./chunk-2XHXABCX.js";
 import {
   blocksToMarkdown
-} from "./chunk-BHUYJBSQ.js";
-import "./chunk-AUFO5LVN.js";
+} from "./chunk-WJLICIYV.js";
+import "./chunk-MDCTITUG.js";
 import {
   KordocError,
   VERSION,
@@ -69,7 +69,7 @@ import {
   routeConsoleToStderr,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-QXNQ2HNB.js";
+} from "./chunk-A2W2JQ3S.js";
 
 // src/mcp.ts
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
@@ -355,7 +355,7 @@ ${markdown}`);
             break;
           case "pdf":
             try {
-              const { extractPdfMetadataOnly } = await import("./parser-YIIRSBZS.js");
+              const { extractPdfMetadataOnly } = await import("./parser-M5WHRX25.js");
               metadata = await extractPdfMetadataOnly(buffer);
             } catch {
               metadata = void 0;
@@ -531,7 +531,7 @@ ${tableMarkdown}` }]
         if (!parsed.success) {
           return { content: [{ type: "text", text: `\uD30C\uC2F1 \uC2E4\uD328: ${parsed.error}` }], isError: true };
         }
-        const { blocksToChunks } = await import("./chunks-P5Y7V73K.js");
+        const { blocksToChunks } = await import("./chunks-EBWJR55V.js");
         const chunks = blocksToChunks(parsed.blocks, { granularity, includeTableCells: include_table_cells });
         return { content: [{ type: "text", text: capResponseText(JSON.stringify(chunks, null, 2)) }] };
       } catch (err) {
@@ -776,7 +776,7 @@ ${previewMd}` }]
         }
         const image = new Uint8Array(await readFile2(imgResolved));
         const ext = extname2(imgResolved).slice(1).toLowerCase();
-        const { placeSealHwpx } = await import("./seal-S4KJGZUI.js");
+        const { placeSealHwpx } = await import("./seal-IPKIQHYB.js");
         const result = await placeSealHwpx(buffer, [{
           anchor,
           occurrence,
@@ -893,7 +893,7 @@ ${previewMd}` }]
         if (outPath && await isSameFile(outPath, resolved)) {
           return { content: [{ type: "text", text: "output_path\uAC00 \uC785\uB825 \uD30C\uC77C\uACFC \uAC19\uC2B5\uB2C8\uB2E4 \u2014 \uC6D0\uBCF8\uC744 \uB36E\uC5B4\uC4F0\uC9C0 \uC54A\uC2B5\uB2C8\uB2E4. \uB2E4\uB978 \uACBD\uB85C\uB97C \uC9C0\uC815\uD558\uC138\uC694." }], isError: true };
         }
-        const { redactDocument } = await import("./redact-doc-YTP6AAUP.js");
+        const { redactDocument } = await import("./redact-doc-5MK5L5IQ.js");
         const r = await redactDocument(buffer, { rules, maskChar: mask_char, filePath: file_path, dryRun: dry_run });
         const MAX_LINES = 50;
         const listed = r.fileHits.length > 0 ? r.fileHits.map((h) => `  - [${h.rule}] ${h.masked} (${h.where} @ ${h.part})`) : r.markdownHits.map((h) => `  - [${h.rule}] ${h.masked}`);
@@ -967,7 +967,7 @@ function registerRenderTools(server2) {
         const { buffer, resolved } = await readValidatedFile(file_path, MAX_FILE_SIZE, /* @__PURE__ */ new Set([".hwpx", ".hwp"]));
         const isHwp5 = resolved.toLowerCase().endsWith(".hwp");
         if (!isHwp5 && !pages && (format === "png" || format === "svg")) {
-          const { renderHwpxToSvg } = await import("./render-525LXDMV.js");
+          const { renderHwpxToSvg } = await import("./render-SYZMHYK6.js");
           const result = await renderHwpxToSvg(buffer, { highlights, reflow: true, reflowMode: reflow_mode });
           const summary2 = [
             `\uB80C\uB354 \uC644\uB8CC: ${result.pageCount}\uD398\uC774\uC9C0, ${Math.round(result.width)}x${Math.round(result.height)}pt (\uD14D\uC2A4\uD2B8 ${result.stats.texts}\xB7\uC774\uBBF8\uC9C0 ${result.stats.images}\xB7\uD45C ${result.stats.tables})`,
@@ -979,7 +979,7 @@ function registerRenderTools(server2) {
             summary2.push(`\uC800\uC7A5: ${outPath}`);
             return { content: [{ type: "text", text: summary2.join("\n") }] };
           }
-          const { rasterizeSvg } = await import("./rasterize-OWHFWG6R.js");
+          const { rasterizeSvg } = await import("./rasterize-V2XHHS2I.js");
           const raster = await rasterizeSvg(result.svg, result.width, result.height, max_width_px ? { maxWidthPx: max_width_px } : void 0);
           if (outPath) {
             await mkdir2(dirname3(outPath), { recursive: true });
@@ -994,7 +994,7 @@ function registerRenderTools(server2) {
             ]
           };
         }
-        const { renderDocument } = await import("./render-525LXDMV.js");
+        const { renderDocument } = await import("./render-SYZMHYK6.js");
         const { scene, assets } = await renderDocument(buffer, { format, pages, highlights, reflow: true, reflowMode: reflow_mode, maxWidthPx: max_width_px });
         const summary = [
           `\uB80C\uB354 \uC644\uB8CC: \uBB38\uC11C ${scene.pages.length}\uD398\uC774\uC9C0 \uC911 ${assets.filter((a) => a.page !== void 0).length || 1}\uAC74 (\uD14D\uC2A4\uD2B8 ${scene.stats.texts}\xB7\uC774\uBBF8\uC9C0 ${scene.stats.images}\xB7\uD45C ${scene.stats.tables}\xB7\uB3C4\uD615 ${scene.stats.shapes})`,
@@ -1044,7 +1044,7 @@ function registerRenderTools(server2) {
       try {
         const { buffer } = await readValidatedFile(file_path, MAX_FILE_SIZE, /* @__PURE__ */ new Set([".hwpx", ".hwp"]));
         const dir = safeOutputPath(join2(output_dir, "regions.json"), /* @__PURE__ */ new Set([".json"])).replace(/[\\/]regions\.json$/, "");
-        const { extractRenderedRegions } = await import("./render-525LXDMV.js");
+        const { extractRenderedRegions } = await import("./render-SYZMHYK6.js");
         const regions = await extractRenderedRegions(buffer, { types: target, format, pages, paddingPt: padding_pt, reflow: true });
         await mkdir2(dir, { recursive: true });
         const ext = format === "jpeg" ? "jpg" : "png";
@@ -1080,7 +1080,7 @@ function registerRenderTools(server2) {
         }
         const { buffer } = await readValidatedFile(file_path, MAX_FILE_SIZE, /* @__PURE__ */ new Set([".hwpx", ".hwp"]));
         const dir = output_dir ? safeOutputPath(join2(output_dir, "tables.json"), /* @__PURE__ */ new Set([".json"])).replace(/[\\/]tables\.json$/, "") : void 0;
-        const { extractTables } = await import("./visual-AU3NH6EO.js");
+        const { extractTables } = await import("./visual-RTTGNOAE.js");
         const tables = await extractTables(buffer, { policy: visual, format, paddingPt: padding_pt });
         if (dir) await mkdir2(dir, { recursive: true });
         const ext = format === "jpeg" ? "jpg" : "png";
@@ -1137,7 +1137,7 @@ function registerGenerateTools(server2) {
       try {
         const out = safeOutputPath(output_path, PROFILE_EXTENSIONS);
         const { buffer } = await readValidatedFile(hwpx_path);
-        const { hwpxToProfile } = await import("./-UNMTWAEI.js");
+        const { hwpxToProfile } = await import("./-2UA2JHVA.js");
         const profile = await hwpxToProfile(buffer);
         await mkdir3(dirname4(out), { recursive: true });
         await writeOutputFile(out, JSON.stringify(profile, null, 2));
@@ -1244,7 +1244,7 @@ function registerGenerateTools(server2) {
         }
         let profile;
         if (profile_path) {
-          const { parseFormatProfileJson } = await import("./profile-io-GU45HTYP.js");
+          const { parseFormatProfileJson } = await import("./profile-io-5YB627EO.js");
           profile = parseFormatProfileJson(await readFile3(safePath(profile_path, PROFILE_EXTENSIONS), "utf-8"));
         }
         const out = safeOutputPath(output_path, /* @__PURE__ */ new Set([".hwpx"]));

package/dist/{model-bundle-P2XOZ76Q.js → model-bundle-MA4MMCRJ.js}

@@ -1,14 +1,14 @@
 #!/usr/bin/env node
 import {
   ALL_OCR_MODELS
-} from "./chunk-BQAJHAWR.js";
+} from "./chunk-KQIYC2ZU.js";
 import {
   ALL_FORMULA_MODELS,
   getModelStatusIn,
   getModelsDir
-} from "./chunk-MY6Q2HFW.js";
-import "./chunk-N6MPDM55.js";
-import "./chunk-QXNQ2HNB.js";
+} from "./chunk-XZA6X34R.js";
+import "./chunk-6OCKAXIS.js";
+import "./chunk-A2W2JQ3S.js";
 
 // src/shared/model-bundle.ts
 import { copyFile, mkdir } from "fs/promises";
@@ -101,4 +101,4 @@ export {
   importModels,
   modelCacheStatus
 };
-//# sourceMappingURL=model-bundle-P2XOZ76Q.js.map
+//# sourceMappingURL=model-bundle-MA4MMCRJ.js.map

package/dist/{models-GPHBC2YJ.js → models-SEJ7LNSN.js}

@@ -10,10 +10,10 @@ import {
   ocrModelPath,
   ocrModelsCached,
   parseCharacterDict
-} from "./chunk-BQAJHAWR.js";
-import "./chunk-MY6Q2HFW.js";
-import "./chunk-N6MPDM55.js";
-import "./chunk-QXNQ2HNB.js";
+} from "./chunk-KQIYC2ZU.js";
+import "./chunk-XZA6X34R.js";
+import "./chunk-6OCKAXIS.js";
+import "./chunk-A2W2JQ3S.js";
 export {
   ALL_OCR_MODELS,
   OCR_DET_MODEL,
@@ -26,4 +26,4 @@ export {
   ocrModelsCached,
   parseCharacterDict
 };
-//# sourceMappingURL=models-GPHBC2YJ.js.map
+//# sourceMappingURL=models-SEJ7LNSN.js.map

package/dist/page-range-QHHFTL7X.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/page-range-QHHFTL7X.cjs"],"names":[],"mappings":"AAAA;AACE;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACA;AACF,qIAAC","file":"/Users/mong-e/workspace/kordoc/dist/page-range-QHHFTL7X.cjs"}
+{"version":3,"sources":["/Users/mong-e/workspace/kordoc-quality-99/dist/page-range-QHHFTL7X.cjs"],"names":[],"mappings":"AAAA;AACE;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACA;AACF,qIAAC","file":"/Users/mong-e/workspace/kordoc-quality-99/dist/page-range-QHHFTL7X.cjs"}

package/dist/{parser-YIIRSBZS.js → parser-M5WHRX25.js}

@@ -39,15 +39,15 @@ import {
   startsNewItem,
   trimTrailingEmptyTableCols,
   wrapJoiner
-} from "./chunk-KBIQGGNY.js";
+} from "./chunk-4TSYMJDL.js";
 import {
   ocrModelsCached
-} from "./chunk-BQAJHAWR.js";
-import "./chunk-MY6Q2HFW.js";
-import "./chunk-N6MPDM55.js";
+} from "./chunk-KQIYC2ZU.js";
+import "./chunk-XZA6X34R.js";
+import "./chunk-6OCKAXIS.js";
 import {
   blocksToPages
-} from "./chunk-OVSCNCT6.js";
+} from "./chunk-O4CB7D2Q.js";
 import {
   encodePng
 } from "./chunk-7FCNOKOR.js";
@@ -65,11 +65,11 @@ import {
   WINGDINGS,
   blocksToMarkdown,
   escapeLiteralDollar
-} from "./chunk-BHUYJBSQ.js";
+} from "./chunk-WJLICIYV.js";
 import {
   KordocError,
   sanitizeHref
-} from "./chunk-QXNQ2HNB.js";
+} from "./chunk-A2W2JQ3S.js";
 
 // src/pdf/ocr-region-merge.ts
 function mergeOcrImageRegions(blocks, page, regions, ocrBlocks) {
@@ -689,6 +689,7 @@ function splitContactTables(blocks) {
 }
 
 // src/pdf/table-parts.ts
+var flowText = (text) => text.replace(/<\/?u>|~~/g, "");
 var PART_COL_TOL = 1;
 var PART_COL_SNAP = 0.3;
 function anchorsOf(table) {
@@ -736,7 +737,7 @@ var indexOf = (coords, x) => {
   }
   return bestD <= PART_COL_TOL ? best : -1;
 };
-var rowText = (anchors, r) => anchors.filter((a) => a.r === r).map((a) => a.cell.text.replace(/\s+/g, "")).join("|");
+var rowText = (anchors, r) => anchors.filter((a) => a.r === r).map((a) => flowText(a.cell.text).replace(/\s+/g, "")).join("|");
 var hasText = (t) => t.replace(/\|/g, "") !== "";
 function joinSplitParts(prev, pcx, curr, ccx, dx = 0, prevBottom, lex) {
   if (pcx.length !== prev.cols + 1 || ccx.length !== curr.cols + 1) return null;
@@ -952,8 +953,8 @@ function continuesAcross(u, d, x1, x2) {
 var OUTLINE_HEAD = /^[□■]\s*\S/;
 var OUTLINE_ITEM = /^[ㅇ○◦\-‐–·․‧※*]\s*\S/;
 function outlineContinues(u, d) {
-  const U = u.text.split("\n").map((l) => l.trim()).filter(Boolean);
-  const D = d.text.split("\n").map((l) => l.trim()).filter(Boolean);
+  const U = flowText(u.text).split("\n").map((l) => l.trim()).filter(Boolean);
+  const D = flowText(d.text).split("\n").map((l) => l.trim()).filter(Boolean);
   return U.length >= 2 && D.length >= 1 && U.some((l) => OUTLINE_HEAD.test(l)) && OUTLINE_ITEM.test(D[0]);
 }
 function lineEnded(c, x1, x2, next) {
@@ -965,7 +966,7 @@ function lineEnded(c, x1, x2, next) {
   return x2 - l.r - (l.l - x1) >= Math.max(LABEL_ROOM, firstWordUnits(next?.text ?? "")) * h;
 }
 function firstWordUnits(text) {
-  const word = text.trim().replace(/^([(\[「『<〈])\s+/, "$1").split(/\s+/)[0] ?? "";
+  const word = flowText(text).trim().replace(/^([(\[「『<〈])\s+/, "$1").split(/\s+/)[0] ?? "";
   let units = 0;
   for (const ch of word) units += /[가-힣\u3400-\u9fff]/.test(ch) ? 1 : 0.55;
   return units;
@@ -985,7 +986,7 @@ var batchim = (ch) => {
   return k >= 0 && k < 11172 && k % 28 !== 0;
 };
 function clauseContinues(u, d) {
-  const tail = u.text.trim().split(/\s+/).pop() ?? "", head = d.text.trim();
+  const tail = flowText(u.text).trim().split(/\s+/).pop() ?? "", head = flowText(d.text).trim();
   if (!/^[가-힣]{2,}$/.test(tail) || !/^[가-힣]/.test(head) || startsNewItem(u.text, head)) return false;
   const last = tail[tail.length - 1], before = tail[tail.length - 2];
   return last === "\uC744" && batchim(before) || last === "\uB97C" && !batchim(before) || CLAUSE_OPEN_ENDING.test(tail);
@@ -998,6 +999,7 @@ function wordOverflows(u, d, x1, x2) {
 }
 var KO_ORDER = "\uAC00\uB098\uB2E4\uB77C\uB9C8\uBC14\uC0AC\uC544\uC790\uCC28\uCE74\uD0C0\uD30C\uD558";
 function itemMark(line) {
+  line = flowText(line).trim();
   let m = /^(\d{1,2})([.)])(?!\d)/.exec(line);
   if (m) return { style: "1" + m[2], n: +m[1] };
   if (m = /^\((\d{1,2})\)/.exec(line)) return { style: "(1)", n: +m[1] };
@@ -1049,7 +1051,7 @@ function mergeSplitRow(table, owner, first, colXs, textOnly = false, cut = NO_CU
   const contentPairs = pairs.filter(([u, d]) => hasContent(u.cell) && hasContent(d.cell));
   const carriedSplit = !textOnly && carried && !newCell && (contentPairs.length <= 1 || contentPairs.every(([, d]) => d.rs > 1) || !cut.ruled && contentPairs.some(([u, d]) => wordOverflows(cell(u), cell(d), colXs[u.c], colXs[u.c + u.cs])));
   if (!carriedSplit) {
-    const norm = (c) => c.text.replace(/\s+/g, "");
+    const norm = (c) => flowText(c.text).replace(/\s+/g, "");
     if (pairs.some(([u, d]) => hasContent(d.cell) && norm(cell(d)) !== norm(cell(u)) && lineEnded(cell(u), colXs[u.c], colXs[u.c + u.cs], cell(d)))) return false;
     const flows = ([u, d]) => continuesAcross(cell(u), cell(d), colXs[u.c], colXs[u.c + u.cs]) || clauseContinues(cell(u), cell(d));
     if (!(cut.open && !newCell)) {
@@ -1146,6 +1148,20 @@ function mergeCrossPageTables(blocks, pageHeights, lex) {
       blocks.splice(j, 1);
       continue;
     }
+    if (px && cx && !CLIP_TABLES.has(prev.table) && !CLIP_TABLES.has(curr.table) && currCells.length < curr.table.cells.length && curr.table.cells[0].filter((c) => c.text.trim()).length >= 2 && prev.table.cells.at(-1).concat(currCells[0]).every((c) => c.rowSpan === 1 && c.colSpan === 1 && CELL_LINES.has(c))) {
+      const dx = px[0] - cx[0];
+      const part = joinSplitParts(prev.table, px, curr.table, cx.map((x) => x + dx), dx, void 0, lex);
+      if (part?.split && part.table.rows === prev.table.rows + currCells.length - 1) {
+        const last = prev.table.rows - 1;
+        for (let c = 0; c < part.table.cols; c++) {
+          const U = CELL_LINES.get(prev.table.cells[last][c]), D = CELL_LINES.get(currCells[0][c]);
+          CELL_LINES.set(part.table.cells[last][c], U.concat(D.map((l) => ({ ...l, l: l.l + dx, r: l.r + dx }))));
+        }
+        TABLE_COLXS.set(part.table, part.colXs);
+        placeJoined(blocks, i, j, part.table);
+        continue;
+      }
+    }
     const merged = {
       rows: prev.table.rows + currCells.length,
       cols: prev.table.cols,
@@ -1230,7 +1246,7 @@ function prevLacksLeftCols(prev, px, xs, pageHeights) {
   return xs[0] < px[0] && xs.slice(1, -1).some((x) => Math.abs(x - px[0]) <= CONTINUATION_COL_TOL);
 }
 function headingRow(row) {
-  const heads = row.map((c) => c.text.replace(/\s+/g, "")).filter(Boolean);
+  const heads = row.map((c) => flowText(c.text).replace(/\s+/g, "")).filter(Boolean);
   if (heads.length === 1) return heads[0].length <= 8 || /^[①-⑳❶-❿]/.test(heads[0]);
   return heads.length === 2 && /^(?:\d{1,2}|[ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩ]+|[①-⑳❶-❿])$/.test(heads[0]);
 }
@@ -1241,7 +1257,7 @@ function looksContinued(prev, curr, pageHeights) {
   const firstRow = curr.table.cells[0] ?? [];
   const firstText = firstRow.find((c) => c.text.trim())?.text ?? "";
   if (!ANNEX_HEAD_RE.test(firstText)) return true;
-  const filled = firstRow.map((c) => c.text.replace(/\s+/g, "")).filter(Boolean);
+  const filled = firstRow.map((c) => flowText(c.text).replace(/\s+/g, "")).filter(Boolean);
   return filled.length >= 2 && filled.every((t) => t === filled[0]);
 }
 function pushedLead(prev, curr) {
@@ -1270,7 +1286,7 @@ function tailOf(b) {
 function firstRowSig(t) {
   const x = TABLE_COLXS.get(t);
   if (!x || x.length !== t.cols + 1) return null;
-  return anchorsOf(t).filter((a) => a.r === 0).map((a) => ({ x1: x[a.c], x2: x[a.c + a.cs], t: a.cell.text.replace(/\s+/g, "") }));
+  return anchorsOf(t).filter((a) => a.r === 0).map((a) => ({ x1: x[a.c], x2: x[a.c + a.cs], t: flowText(a.cell.text).replace(/\s+/g, "") }));
 }
 function restartsTable(blocks, i, curr, pageHeights) {
   const cs = firstRowSig(curr);
@@ -1295,7 +1311,7 @@ function restartsTitledForm(head, curr) {
   if (head.rows < 2 || curr.rows < 2 || head.cols !== curr.cols) return false;
   const full = (t) => t.cells[0][0]?.colSpan === t.cols;
   if (!full(head) || !full(curr)) return false;
-  const norm = (c) => (c?.text ?? "").replace(/\s+/g, "");
+  const norm = (c) => flowText(c?.text ?? "").replace(/\s+/g, "");
   const ht = norm(head.cells[0][0]), ct = norm(curr.cells[0][0]);
   if (!ht || !ct || ht === ct) return false;
   const titleLike = (t) => t.length <= TITLE_ROW_MAX_CHARS && !/^[ㅇ○◦•·\-–※*□■▪☞]/.test(t);
@@ -1331,7 +1347,7 @@ function chainHead(blocks, i, pageHeights) {
 }
 function rowTextsEqual(a, b) {
   if (a.length !== b.length) return false;
-  const norm = (t) => t.replace(/\s+/g, "");
+  const norm = (t) => flowText(t).replace(/\s+/g, "");
   for (let i = 0; i < a.length; i++) {
     if (norm(a[i].text) !== norm(b[i].text)) return false;
   }
@@ -2696,7 +2712,7 @@ function mergeLinkRuns(markdown) {
 
 // src/pdf/formula-ocr.ts
 async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
-  const formulaMod = await import("./formula-VH545WNH.js");
+  const formulaMod = await import("./formula-MBLN2DG4.js");
   const { FormulaPipeline, ensureFormulaModels } = formulaMod;
   await ensureFormulaModels((p) => {
     if (p.phase === "download" && p.total) {
@@ -3079,7 +3095,7 @@ async function parsePdfDocument(buffer, options) {
       }
       if (targets.size > 0) {
         try {
-          const { runPdfOcr } = await import("./pdf-ocr-5YBMXW54.js");
+          const { runPdfOcr } = await import("./pdf-ocr-QB2TB5W5.js");
           const mode = typeof options?.ocr === "function" ? options.ocr : "builtin";
           const regionPages = new Map([...ocrRegions].filter(([p]) => options?.ocr !== "force" && !isImageBased && !pageQuality.find((q) => q.page === p)?.needsOcr));
           const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options?.onProgress, options?.tables !== false, vectorPageOps, regionPages);
@@ -3204,7 +3220,8 @@ async function parsePdfDocument(buffer, options) {
       return b.bbox.y > h * 0.1 && b.bbox.y + b.bbox.height < h * 0.9;
     });
     joinLatinCellWraps(outBlocks);
-    if (options?.layoutTables !== "keep" && await isHancomPdf(doc)) {
+    if (options?.layoutTables !== "keep") {
+      const hancom = await isHancomPdf(doc);
       const shown = /* @__PURE__ */ new Map();
       const markContent = (bs) => {
         for (const b of bs ?? []) if (b.table) for (const row of b.table.cells) for (const c of row) {
@@ -3214,6 +3231,7 @@ async function parsePdfDocument(buffer, options) {
       };
       const visual = (bs) => bs.flatMap((b) => {
         if (b.type !== "table") return [b];
+        if (!hancom && b.table && CLIP_TABLES.has(b.table)) return [b];
         let v = shown.get(b);
         if (!v) {
           markContent([b]);
@@ -3309,4 +3327,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-YIIRSBZS.js.map
+//# sourceMappingURL=parser-M5WHRX25.js.map