kordoc 3.10.1 → 3.12.0

package/dist/index.js

@@ -19,7 +19,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-I6CX2RK4.js";
+} from "./chunk-NGKIAZ6G.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
@@ -25548,7 +25548,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-664URIIW.js");
+    const mod = await import("./parser-AWEGMHHV.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {

package/dist/mcp.js

@@ -10,22 +10,22 @@ import {
   parse,
   patchHwp,
   patchHwpx
-} from "./chunk-RMGYE4LT.js";
+} from "./chunk-RUYSNNYZ.js";
 import {
   detectFormat,
   detectZipFormat
 } from "./chunk-MEPHGCPQ.js";
-import "./chunk-UR34PYAC.js";
+import "./chunk-ILR7GF3W.js";
 import {
   blocksToMarkdown
-} from "./chunk-YXEE3DEQ.js";
+} from "./chunk-BUDM6ULT.js";
 import "./chunk-MOL7MDBG.js";
 import {
   KordocError,
   VERSION,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-GFS4QWB6.js";
+} from "./chunk-7Z43WEML.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-LJOS7VZM.js");
+            const { extractPdfMetadataOnly } = await import("./parser-5FSEFR3R.js");
             metadata = await extractPdfMetadataOnly(buffer);
           } catch {
             metadata = void 0;

package/dist/{parser-LJOS7VZM.js → parser-5FSEFR3R.js}

@@ -4,7 +4,7 @@ import {
   HEADING_RATIO_H2,
   HEADING_RATIO_H3,
   blocksToMarkdown
-} from "./chunk-YXEE3DEQ.js";
+} from "./chunk-BUDM6ULT.js";
 import {
   parsePageRange
 } from "./chunk-MOL7MDBG.js";
@@ -12,7 +12,7 @@ import {
   KordocError,
   safeMax,
   safeMin
-} from "./chunk-GFS4QWB6.js";
+} from "./chunk-7Z43WEML.js";
 
 // src/pdf/line-extract.ts
 import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
@@ -201,10 +201,120 @@ function classifyAndAdd(seg, lineWidth, horizontals, verticals) {
 function preprocessLines(horizontals, verticals) {
   let h = horizontals.filter((l) => l.lineWidth <= MAX_LINE_WIDTH);
   let v = verticals.filter((l) => l.lineWidth <= MAX_LINE_WIDTH);
+  h = dropShadingStacks(h, "h");
+  v = dropShadingStacks(v, "v");
   h = mergeParallelLines(h, "h");
   v = mergeParallelLines(v, "v");
   return { horizontals: h, verticals: v };
 }
+var STACK_GAP = 2;
+var STACK_MIN_LINES = 6;
+function dropShadingStacks(lines, dir) {
+  if (lines.length < STACK_MIN_LINES) return lines;
+  const groups = /* @__PURE__ */ new Map();
+  for (const l of lines) {
+    const key = dir === "h" ? `${Math.round(l.x1)}:${Math.round(l.x2)}` : `${Math.round(l.y1)}:${Math.round(l.y2)}`;
+    const arr = groups.get(key);
+    if (arr) arr.push(l);
+    else groups.set(key, [l]);
+  }
+  const dropped = /* @__PURE__ */ new Set();
+  for (const group of groups.values()) {
+    if (group.length < STACK_MIN_LINES) continue;
+    group.sort((a, b) => dir === "h" ? a.y1 - b.y1 : a.x1 - b.x1);
+    let runStart = 0;
+    for (let i = 1; i <= group.length; i++) {
+      const gap = i < group.length ? dir === "h" ? group[i].y1 - group[i - 1].y1 : group[i].x1 - group[i - 1].x1 : Infinity;
+      if (gap < STACK_GAP) continue;
+      if (i - runStart >= STACK_MIN_LINES) {
+        for (let j = runStart; j < i; j++) dropped.add(group[j]);
+      }
+      runStart = i;
+    }
+  }
+  return dropped.size ? lines.filter((l) => !dropped.has(l)) : lines;
+}
+var EDGE_ALIGN_TOL = 3;
+var EDGE_MIN_RULES = 3;
+var EDGE_MIN_SPAN = 12;
+var EDGE_INSET = 15;
+var EDGE_NEAR = 10;
+var EDGE_CONNECT_TOL = 5;
+var CHAIN_Y_TOL = 1.5;
+var CHAIN_GAP = 3;
+function chainCollinearRules(horizontals) {
+  if (horizontals.length <= 1) return horizontals;
+  const sorted = [...horizontals].sort((a, b) => a.y1 - b.y1 || a.x1 - b.x1);
+  const rules = [];
+  let bandStart = 0;
+  const flushBand = (end) => {
+    const band = sorted.slice(bandStart, end).sort((a, b) => a.x1 - b.x1);
+    let cur = { ...band[0] };
+    for (let i = 1; i < band.length; i++) {
+      const seg = band[i];
+      if (seg.x1 - cur.x2 <= CHAIN_GAP) {
+        if (seg.x2 > cur.x2) cur.x2 = seg.x2;
+        if (seg.lineWidth > cur.lineWidth) cur.lineWidth = seg.lineWidth;
+      } else {
+        rules.push(cur);
+        cur = { ...seg };
+      }
+    }
+    rules.push(cur);
+  };
+  for (let i = 1; i <= sorted.length; i++) {
+    if (i === sorted.length || sorted[i].y1 - sorted[bandStart].y1 > CHAIN_Y_TOL) {
+      flushBand(i);
+      bandStart = i;
+    }
+  }
+  return rules;
+}
+function closeOpenTableEdges(horizontals, verticals) {
+  if (horizontals.length < EDGE_MIN_RULES) return verticals;
+  const groups = [];
+  for (const hl of chainCollinearRules(horizontals)) {
+    let placed = false;
+    for (const g2 of groups) {
+      if (Math.abs(g2[0].x1 - hl.x1) <= EDGE_ALIGN_TOL && Math.abs(g2[0].x2 - hl.x2) <= EDGE_ALIGN_TOL) {
+        g2.push(hl);
+        placed = true;
+        break;
+      }
+    }
+    if (!placed) groups.push([hl]);
+  }
+  const synthesized = [];
+  for (const g2 of groups) {
+    if (g2.length < EDGE_MIN_RULES) continue;
+    let yMin = Infinity, yMax = -Infinity, x1 = 0, x2 = 0;
+    for (const hl of g2) {
+      if (hl.y1 < yMin) yMin = hl.y1;
+      if (hl.y1 > yMax) yMax = hl.y1;
+      x1 += hl.x1;
+      x2 += hl.x2;
+    }
+    x1 /= g2.length;
+    x2 /= g2.length;
+    if (yMax - yMin < EDGE_MIN_SPAN) continue;
+    const crossCount = (v) => {
+      let n = 0;
+      for (const hl of g2) {
+        if (v.x1 >= hl.x1 - EDGE_CONNECT_TOL && v.x1 <= hl.x2 + EDGE_CONNECT_TOL && hl.y1 >= v.y1 - EDGE_CONNECT_TOL && hl.y1 <= v.y2 + EDGE_CONNECT_TOL) n++;
+      }
+      return n;
+    };
+    const hasInterior = verticals.some((v) => v.x1 > x1 + EDGE_INSET && v.x1 < x2 - EDGE_INSET && crossCount(v) >= 2);
+    if (!hasInterior) continue;
+    for (const edgeX of [x1, x2]) {
+      const closed = verticals.some((v) => Math.abs(v.x1 - edgeX) <= EDGE_NEAR && v.y1 <= yMax + EDGE_CONNECT_TOL && v.y2 >= yMin - EDGE_CONNECT_TOL);
+      if (!closed) {
+        synthesized.push({ x1: edgeX, y1: yMin, x2: edgeX, y2: yMax, lineWidth: 0.5 });
+      }
+    }
+  }
+  return synthesized.length ? [...verticals, ...synthesized] : verticals;
+}
 function mergeParallelLines(lines, dir) {
   if (lines.length <= 1) return lines;
   const sorted = [...lines].sort((a, b) => {
@@ -2218,6 +2328,10 @@ function detectHeadings(blocks, medianFontSize) {
 function shouldDemoteTable(table) {
   const allCells = table.cells.flatMap((row) => row.map((c) => c.text.trim())).filter(Boolean);
   const allText = allCells.join(" ");
+  if (table.rows >= 2 && table.cols >= 2 && table.cells[0].every((c) => {
+    const t = c.text.trim();
+    return t.length > 0 && t.length <= 12 && !/[□■◆○●▶ㅇ<>]/.test(t);
+  }) && table.cells.slice(1).some((row) => row.some((c) => c.text.trim() !== ""))) return false;
   if (table.rows <= 3 && table.cols <= 3) {
     const totalCells2 = table.rows * table.cols;
     const emptyCells2 = totalCells2 - allCells.length;
@@ -2555,6 +2669,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   let { horizontals, verticals } = extractLines(opList.fnArray, opList.argsArray);
   ({ horizontals, verticals } = filterPageBorderLines(horizontals, verticals, pageWidth, pageHeight));
   ({ horizontals, verticals } = preprocessLines(horizontals, verticals));
+  verticals = closeOpenTableEdges(horizontals, verticals);
   markStrikethroughItems(items, horizontals);
   wrapStrikethroughRuns(items);
   const grids = buildTableGrids(horizontals, verticals);
@@ -3359,4 +3474,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-LJOS7VZM.js.map
+//# sourceMappingURL=parser-5FSEFR3R.js.map