kordoc 4.0.2 → 4.0.4

package/dist/{parser-3HM2VYKK.js → parser-46JGSTPR.js}

@@ -4,7 +4,7 @@ import {
   HEADING_RATIO_H2,
   HEADING_RATIO_H3,
   blocksToMarkdown
-} from "./chunk-6YIYFWO2.js";
+} from "./chunk-ZJXR4PWY.js";
 import {
   parsePageRange
 } from "./chunk-MOL7MDBG.js";
@@ -12,7 +12,7 @@ import {
   KordocError,
   safeMax,
   safeMin
-} from "./chunk-STMLZDKR.js";
+} from "./chunk-KA5AN73R.js";
 
 // src/pdf/line-extract.ts
 import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
@@ -490,6 +490,14 @@ var CONNECT_TOL = 5;
 var MIN_COL_WIDTH = 15;
 var MIN_ROW_HEIGHT = 6;
 var MIN_COORD_MERGE_TOL = 8;
+var CUT_FULLWIDTH_RATIO = 0.9;
+var CUT_CROSS_EPS = 2;
+var CUT_MIN_SIDE_VERTICALS = 2;
+var CUT_EDGE_MARGIN = 12;
+var CUT_INTERIOR_MATCH_TOL = 8;
+var CUT_MAX_INTERIOR_OVERLAP = 0.5;
+var CUT_VCHAIN_X_TOL = 1.5;
+var CUT_VCHAIN_GAP = 1;
 function buildVertices(horizontals, verticals) {
   const vertices = [];
   const tol = CONNECT_TOL;
@@ -537,9 +545,13 @@ function buildTableGrids(horizontals, verticals) {
     ...horizontals.map((l, i) => ({ ...l, type: "h", id: i })),
     ...verticals.map((l, i) => ({ ...l, type: "v", id: i + horizontals.length }))
   ];
-  const groups = groupConnectedLines(allLines);
+  const groups = [];
+  for (const g2 of groupConnectedLines(allLines)) {
+    const bands = splitStackedGroup(g2);
+    for (const b of bands) groups.push({ lines: b, fromSplit: bands.length > 1 });
+  }
   const grids = [];
-  for (const group of groups) {
+  for (const { lines: group, fromSplit } of groups) {
     const hLines = group.filter((l) => l.type === "h");
     const vLines = group.filter((l) => l.type === "v");
     if (hLines.length < 2 || vLines.length < 2) continue;
@@ -558,7 +570,7 @@ function buildTableGrids(horizontals, verticals) {
       x2: gx2 + CONNECT_TOL,
       y2: gy2 + CONNECT_TOL
     };
-    const groupVertices = vertices.filter(
+    const groupVertices = fromSplit ? mergeVertices(buildVertices(hLines, vLines)) : vertices.filter(
       (v) => v.x >= groupBbox.x1 && v.x <= groupBbox.x2 && v.y >= groupBbox.y1 && v.y <= groupBbox.y2
     );
     const groupRadius = groupVertices.length > 0 ? groupVertices.reduce((max, v) => Math.max(max, v.radius), 1) : globalRadius;
@@ -658,6 +670,83 @@ function clusterCoordinates(values, tolerance) {
   }
   return clusters.map((c) => c.sum / c.count);
 }
+function chainVerticals(vs) {
+  if (vs.length <= 1) return vs.map((v) => ({ y1: v.y1, y2: v.y2 }));
+  const sorted = [...vs].sort((a, b) => a.x1 - b.x1 || a.y1 - b.y1);
+  const rules = [];
+  let bandStart = 0;
+  const flushBand = (end) => {
+    const band = sorted.slice(bandStart, end).sort((a, b) => a.y1 - b.y1);
+    let cur = { y1: band[0].y1, y2: band[0].y2 };
+    for (let i = 1; i < band.length; i++) {
+      const seg = band[i];
+      if (seg.y1 - cur.y2 <= CUT_VCHAIN_GAP) {
+        if (seg.y2 > cur.y2) cur.y2 = seg.y2;
+      } else {
+        rules.push(cur);
+        cur = { y1: seg.y1, y2: seg.y2 };
+      }
+    }
+    rules.push(cur);
+  };
+  for (let i = 1; i <= sorted.length; i++) {
+    if (i === sorted.length || sorted[i].x1 - sorted[bandStart].x1 > CUT_VCHAIN_X_TOL) {
+      flushBand(i);
+      bandStart = i;
+    }
+  }
+  return rules;
+}
+function splitStackedGroup(group) {
+  const hs = group.filter((l) => l.type === "h");
+  const vs = group.filter((l) => l.type === "v");
+  if (hs.length < 3 || vs.length < 4) return [group];
+  let gx1 = Infinity, gx2 = -Infinity;
+  for (const l of group) {
+    if (l.x1 < gx1) gx1 = l.x1;
+    if (l.x2 > gx2) gx2 = l.x2;
+  }
+  const groupW = gx2 - gx1;
+  if (groupW <= 0) return [group];
+  const isInterior = (v) => v.x1 > gx1 + CUT_EDGE_MARGIN && v.x1 < gx2 - CUT_EDGE_MARGIN;
+  const chained = chainVerticals(vs);
+  const cuts = [];
+  for (const h of hs) {
+    const y = h.y1;
+    if (h.x2 - h.x1 < groupW * CUT_FULLWIDTH_RATIO) continue;
+    if (cuts.some((c) => Math.abs(c - y) <= CUT_CROSS_EPS)) continue;
+    if (chained.some((v) => v.y1 < y - CUT_CROSS_EPS && v.y2 > y + CUT_CROSS_EPS)) continue;
+    const above = vs.filter((v) => v.y1 >= y - CUT_CROSS_EPS);
+    const below = vs.filter((v) => v.y2 <= y + CUT_CROSS_EPS);
+    if (above.length < CUT_MIN_SIDE_VERTICALS || below.length < CUT_MIN_SIDE_VERTICALS) continue;
+    const ia = above.filter(isInterior);
+    const ib = below.filter(isInterior);
+    if (ia.length === 0 || ib.length === 0) continue;
+    let matched = 0;
+    for (const a of ia) if (ib.some((b) => Math.abs(a.x1 - b.x1) <= CUT_INTERIOR_MATCH_TOL)) matched++;
+    if (matched / Math.min(ia.length, ib.length) > CUT_MAX_INTERIOR_OVERLAP) continue;
+    cuts.push(y);
+  }
+  if (cuts.length === 0) return [group];
+  cuts.sort((a, b) => b - a);
+  const bandOf = (y) => {
+    let k = 0;
+    while (k < cuts.length && y < cuts[k]) k++;
+    return k;
+  };
+  const bands = Array.from({ length: cuts.length + 1 }, () => []);
+  for (const v of vs) bands[bandOf((v.y1 + v.y2) / 2)].push(v);
+  for (const h of hs) {
+    const atCut = cuts.findIndex((c) => Math.abs(h.y1 - c) <= CUT_CROSS_EPS);
+    if (atCut >= 0) {
+      bands[atCut].push(h);
+      bands[atCut + 1].push(h);
+    } else {
+      bands[bandOf(h.y1)].push(h);
+    }
+  }
+  return bands.filter((b) => b.length > 0);
+}
 function groupConnectedLines(lines) {
   const parent = lines.map((_, i) => i);
   function find(x) {
@@ -1236,7 +1325,7 @@ function dominantStyle(items) {
   }
   if (dominantSize === 0) return void 0;
   const fontName = items.find((i) => i.fontSize === dominantSize)?.fontName || void 0;
-  return { fontSize: dominantSize, fontName };
+  return fontName ? { fontSize: dominantSize, fontName } : { fontSize: dominantSize };
 }
 function normalizeItems(rawItems) {
   const items = [];
@@ -3469,7 +3558,7 @@ async function parsePdfDocument(buffer, options) {
     if (totalChars / Math.max(parsedPageCount, 1) < 10) {
       if (options?.ocr) {
         try {
-          const { ocrPages } = await import("./provider-7H4CPZYS.js");
+          const { ocrPages } = await import("./provider-AKROB7WQ.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");
@@ -3587,4 +3676,4 @@ export {
   parsePdfDocument,
   removeHeaderFooterBlocks
 };
-//# sourceMappingURL=parser-3HM2VYKK.js.map
+//# sourceMappingURL=parser-46JGSTPR.js.map