kordoc 4.0.5 → 4.0.6

package/dist/{parser-YSHZC5FQ.js → parser-DMKHIV4E.js}

@@ -6,7 +6,7 @@ import {
   blocksToMarkdown,
   safeMax,
   safeMin
-} from "./chunk-FKK4K53F.js";
+} from "./chunk-HOGJ5ALY.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
@@ -427,6 +427,73 @@ function filterPageBorderLines(horizontals, verticals, pageWidth, pageHeight) {
   };
 }
 
+// src/pdf/vertical-bridge.ts
+var BRIDGE_X_TOL = 1.5;
+var BRIDGE_MIN_GAP = 5;
+var BRIDGE_MAX_GAP = 120;
+var BRIDGE_BAND_TOL = 6;
+var BRIDGE_MIN_COLUMNS = 3;
+var BRIDGE_ENDPOINT_TOL = 2;
+var BRIDGE_EDGE_EPS = 2;
+function bridgeSplitColumnVerticals(horizontals, verticals) {
+  if (verticals.length < 4 || horizontals.length === 0) return verticals;
+  const sorted = [...verticals].sort((a, b) => a.x1 - b.x1 || a.y1 - b.y1);
+  const candidates = [];
+  let bandStart = 0;
+  const collectBand = (end) => {
+    const band = sorted.slice(bandStart, end).sort((a, b) => a.y1 - b.y1);
+    for (let i = 1; i < band.length; i++) {
+      const lower = band[i - 1], upper = band[i];
+      const gap = upper.y1 - lower.y2;
+      if (gap <= BRIDGE_MIN_GAP || gap > BRIDGE_MAX_GAP) continue;
+      const x = (lower.x1 + upper.x1) / 2;
+      const hasInteriorH = horizontals.some((h) => h.y1 > lower.y2 + BRIDGE_EDGE_EPS && h.y1 < upper.y1 - BRIDGE_EDGE_EPS && h.x1 <= x + BRIDGE_ENDPOINT_TOL && h.x2 >= x - BRIDGE_ENDPOINT_TOL);
+      if (!hasInteriorH) continue;
+      candidates.push({
+        x,
+        lo: lower.y2,
+        hi: upper.y1,
+        spanLo: lower.y1,
+        spanHi: upper.y2,
+        lineWidth: Math.min(lower.lineWidth, upper.lineWidth)
+      });
+    }
+  };
+  for (let i = 1; i <= sorted.length; i++) {
+    if (i === sorted.length || sorted[i].x1 - sorted[bandStart].x1 > BRIDGE_X_TOL) {
+      collectBand(i);
+      bandStart = i;
+    }
+  }
+  if (candidates.length < BRIDGE_MIN_COLUMNS) return verticals;
+  const clusters = [];
+  for (const c of candidates) {
+    let placed = false;
+    for (const cl of clusters) {
+      if (Math.abs(cl[0].lo - c.lo) <= BRIDGE_BAND_TOL && Math.abs(cl[0].hi - c.hi) <= BRIDGE_BAND_TOL) {
+        cl.push(c);
+        placed = true;
+        break;
+      }
+    }
+    if (!placed) clusters.push([c]);
+  }
+  const synthesized = [];
+  for (const cl of clusters) {
+    if (cl.length < BRIDGE_MIN_COLUMNS) continue;
+    const xs = cl.map((c) => c.x).sort((a, b) => a - b);
+    const interiorXs = xs.slice(1, -1);
+    const bandLo = Math.min(...cl.map((c) => c.lo));
+    const bandHi = Math.max(...cl.map((c) => c.hi));
+    const segmented = horizontals.some((h) => h.y1 > bandLo + BRIDGE_EDGE_EPS && h.y1 < bandHi - BRIDGE_EDGE_EPS && interiorXs.some((ix) => Math.abs(h.x1 - ix) <= BRIDGE_ENDPOINT_TOL || Math.abs(h.x2 - ix) <= BRIDGE_ENDPOINT_TOL));
+    if (!segmented) continue;
+    for (const c of cl) {
+      synthesized.push({ x1: c.x, y1: c.spanLo, x2: c.x, y2: c.spanHi, lineWidth: c.lineWidth });
+    }
+  }
+  return synthesized.length ? [...verticals, ...synthesized] : verticals;
+}
+
 // src/pdf/image-regions.ts
 import { OPS as OPS2 } from "pdfjs-dist/legacy/build/pdf.mjs";
 function multiplyTransform(m, t) {
@@ -2817,6 +2884,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   ({ horizontals, verticals } = filterPageBorderLines(horizontals, verticals, pageWidth, pageHeight));
   ({ horizontals, verticals } = preprocessLines(horizontals, verticals));
   verticals = closeOpenTableEdges(horizontals, verticals);
+  verticals = bridgeSplitColumnVerticals(horizontals, verticals);
   markStrikethroughItems(items, horizontals);
   wrapStrikethroughRuns(items);
   const grids = buildTableGrids(horizontals, verticals);
@@ -3555,7 +3623,7 @@ async function parsePdfDocument(buffer, options) {
     if (totalChars / Math.max(parsedPageCount, 1) < 10) {
       if (options?.ocr) {
         try {
-          const { ocrPages } = await import("./provider-4ZJKV3DC.js");
+          const { ocrPages } = await import("./provider-H4WWSPOV.js");
           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");
@@ -3673,4 +3741,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-YSHZC5FQ.js.map
+//# sourceMappingURL=parser-DMKHIV4E.js.map