kordoc 3.12.0 → 3.13.0

package/dist/mcp.js

@@ -10,22 +10,22 @@ import {
   parse,
   patchHwp,
   patchHwpx
-} from "./chunk-RUYSNNYZ.js";
+} from "./chunk-TWACKS3K.js";
 import {
   detectFormat,
   detectZipFormat
 } from "./chunk-MEPHGCPQ.js";
-import "./chunk-ILR7GF3W.js";
+import "./chunk-33CIKNL4.js";
 import {
   blocksToMarkdown
-} from "./chunk-BUDM6ULT.js";
+} from "./chunk-YXOT7P5B.js";
 import "./chunk-MOL7MDBG.js";
 import {
   KordocError,
   VERSION,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-7Z43WEML.js";
+} from "./chunk-Q6UXVJXI.js";
 
 // src/mcp.ts
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
@@ -196,7 +196,7 @@ server.tool(
           break;
         case "pdf":
           try {
-            const { extractPdfMetadataOnly } = await import("./parser-5FSEFR3R.js");
+            const { extractPdfMetadataOnly } = await import("./parser-XUNGOH76.js");
             metadata = await extractPdfMetadataOnly(buffer);
           } catch {
             metadata = void 0;

package/dist/{parser-PCTX7LGN.cjs → parser-DZBI3EA2.cjs}

@@ -6,7 +6,7 @@
 
 
 
-var _chunk2C5GCXPJcjs = require('./chunk-2C5GCXPJ.cjs');
+var _chunkMEXUFQVCcjs = require('./chunk-MEXUFQVC.cjs');
 
 
 var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
@@ -2086,14 +2086,14 @@ 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 = _chunk2C5GCXPJcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkMEXUFQVCcjs.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 = _chunk2C5GCXPJcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunk2C5GCXPJcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkMEXUFQVCcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkMEXUFQVCcjs.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++) {
@@ -2315,9 +2315,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunk2C5GCXPJcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunk2C5GCXPJcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunk2C5GCXPJcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkMEXUFQVCcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkMEXUFQVCcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkMEXUFQVCcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -2721,6 +2721,54 @@ function wrapStrikethroughRuns(items) {
     arr[arr.length - 1].text = arr[arr.length - 1].text + "~~";
   }
 }
+var PROSEBOX_FULLWIDTH_MIN = 0.6;
+var PROSEBOX_LONG_CELL_CHARS = 80;
+var PROSEBOX_LONG_CELL_MIN = 3;
+var PROSEBOX_LONG_CELL_RATIO = 0.4;
+var PROSEBOX_X_TOL = 8;
+function verticalCoverageAt(verticals, x, yMin, yMax) {
+  const tol = PROSEBOX_X_TOL;
+  const spans = [];
+  for (const v of verticals) {
+    if (Math.abs(v.x1 - x) > tol) continue;
+    const lo = Math.max(v.y1, yMin), hi = Math.min(v.y2, yMax);
+    if (hi > lo) spans.push([lo, hi]);
+  }
+  if (spans.length === 0) return 0;
+  spans.sort((a, b) => a[0] - b[0]);
+  let total = 0, s = spans[0][0], e = spans[0][1];
+  for (let i = 1; i < spans.length; i++) {
+    if (spans[i][0] <= e) {
+      if (spans[i][1] > e) e = spans[i][1];
+    } else {
+      total += e - s;
+      s = spans[i][0];
+      e = spans[i][1];
+    }
+  }
+  return total + (e - s);
+}
+function isProseBoxGrid(grid, verticals, table) {
+  const numCols = grid.colXs.length - 1;
+  if (numCols < 2 || grid.rowYs.length < 3) return false;
+  const gyMax = grid.rowYs[0], gyMin = grid.rowYs[grid.rowYs.length - 1];
+  const span = gyMax - gyMin;
+  if (span <= 0) return false;
+  const interior = grid.colXs.slice(1, -1);
+  let fullWidthHeight = 0;
+  for (let r = 0; r < grid.rowYs.length - 1; r++) {
+    const top = grid.rowYs[r], bot = grid.rowYs[r + 1];
+    const h = top - bot;
+    if (h <= 0) continue;
+    const hasDivider = interior.some((cx) => verticalCoverageAt(verticals, cx, bot, top) >= h * 0.5);
+    if (!hasDivider) fullWidthHeight += h;
+  }
+  if (fullWidthHeight < span * PROSEBOX_FULLWIDTH_MIN) return false;
+  const texts = table.cells.flat().map((c) => c.text.trim()).filter(Boolean);
+  const longCells = texts.filter((s) => s.length > PROSEBOX_LONG_CELL_CHARS).length;
+  if (longCells < PROSEBOX_LONG_CELL_MIN || longCells < texts.length * PROSEBOX_LONG_CELL_RATIO) return false;
+  return true;
+}
 function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
   const blocks = [];
   const usedItems = /* @__PURE__ */ new Set();
@@ -2795,6 +2843,10 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     };
     const hasContent = finalGrid.some((row) => row.some((cell) => cell.text.trim() !== ""));
     if (!hasContent) continue;
+    if (isProseBoxGrid(grid, verticals, irTable)) {
+      for (const it of tableItems) usedItems.delete(it);
+      continue;
+    }
     const tableBbox = {
       page: pageNum,
       x: grid.bbox.x1,
@@ -2841,7 +2893,7 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     }
     if (remaining.length > 0) {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunk2C5GCXPJcjs.safeMax.call(void 0, allY) - _chunk2C5GCXPJcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkMEXUFQVCcjs.safeMax.call(void 0, allY) - _chunkMEXUFQVCcjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -3017,7 +3069,7 @@ 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 = _chunk2C5GCXPJcjs.safeMax.call(void 0, allY) - _chunk2C5GCXPJcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkMEXUFQVCcjs.safeMax.call(void 0, allY) - _chunkMEXUFQVCcjs.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) {
@@ -3270,7 +3322,7 @@ async function loadPdfWithTimeout(buffer) {
       new Promise((_, reject) => {
         timer = setTimeout(() => {
           loadingTask.destroy();
-          reject(new (0, _chunk2C5GCXPJcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
+          reject(new (0, _chunkMEXUFQVCcjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
         }, PDF_LOAD_TIMEOUT_MS);
       })
     ]);
@@ -3283,7 +3335,7 @@ async function parsePdfDocument(buffer, options) {
   const doc = await loadPdfWithTimeout(buffer);
   try {
     const pageCount = doc.numPages;
-    if (pageCount === 0) throw new (0, _chunk2C5GCXPJcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
+    if (pageCount === 0) throw new (0, _chunkMEXUFQVCcjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
     const metadata = { pageCount };
     await extractPdfMetadata(doc, metadata);
     const blocks = [];
@@ -3343,11 +3395,11 @@ async function parsePdfDocument(buffer, options) {
           pageText += pageText ? "\n" + t : t;
         }
         pageQuality.push(computePageQuality(i, pageText));
-        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunk2C5GCXPJcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
+        if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunkMEXUFQVCcjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
         parsedPages++;
         _optionalChain([options, 'optionalAccess', _36 => _36.onProgress, 'optionalCall', _37 => _37(parsedPages, totalTarget)]);
       } catch (pageErr) {
-        if (pageErr instanceof _chunk2C5GCXPJcjs.KordocError) throw pageErr;
+        if (pageErr instanceof _chunkMEXUFQVCcjs.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" });
       }
     }
@@ -3415,7 +3467,7 @@ 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(_chunk2C5GCXPJcjs.blocksToMarkdown.call(void 0, blocks));
+    let markdown = cleanPdfText(_chunkMEXUFQVCcjs.blocksToMarkdown.call(void 0, blocks));
     return {
       markdown,
       blocks,
@@ -3474,4 +3526,4 @@ 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-PCTX7LGN.cjs.map
+//# sourceMappingURL=parser-DZBI3EA2.cjs.map