kordoc 4.5.0 → 4.6.1

package/dist/{chunk-XYSTZIIL.js → chunk-BCG2PKQU.js}

@@ -1,6 +1,6 @@
 import {
   sanitizeHref
-} from "./chunk-EYLOFSA5.js";
+} from "./chunk-YUVKU5CW.js";
 
 // src/shared/pua.ts
 var BMP_SYMBOL_MAP = {
@@ -772,4 +772,4 @@ export {
   encodePng,
   inlineImagesIntoMarkdown
 };
-//# sourceMappingURL=chunk-XYSTZIIL.js.map
+//# sourceMappingURL=chunk-BCG2PKQU.js.map

package/dist/{chunk-2CF2WXBG.js → chunk-BGSGUONP.js}

@@ -3,11 +3,11 @@ import {
   HEADING_RATIO_H1,
   HEADING_RATIO_H2,
   HEADING_RATIO_H3
-} from "./chunk-UEWW3DH2.js";
+} from "./chunk-MVTX6ODZ.js";
 import {
   safeMax,
   safeMin
-} from "./chunk-TTEHX4FN.js";
+} from "./chunk-ZYVULRYF.js";
 
 // src/pdf/image-regions.ts
 import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
@@ -2684,6 +2684,149 @@ function findVerticalCut(items) {
   return { position, gap: largestGap };
 }
 
+// src/pdf/two-column.ts
+var GUTTER_SCAN_LO = 0.3;
+var GUTTER_SCAN_HI = 0.7;
+var GUTTER_MAX_SAMPLES = 400;
+var WIDE_RECT_RATIO = 0.55;
+var GUTTER_MAX_COVER_RATIO = 0.06;
+var SIDE_MIN_COUNT = 5;
+var SIDE_MIN_HEIGHT_RATIO = 0.3;
+var SIDE_MIN_VSPAN_RATIO = 0.45;
+var SIDE_MIN_WIDTH_SYMMETRY = 0.5;
+var MIN_CONTENT_SPAN = 300;
+var TEXT_RECT_MAX_H = 20;
+var ROW_Y_TOL = 2;
+var SIDE_MIN_TEXT_ROWS = 12;
+var CROSS_ROW_TINY_GAP = 12;
+var MAX_TINY_GAP_ROWS = 1;
+var MAX_BOTH_SIDE_ROW_RATIO = 0.65;
+var BLOCK_DOMINANT_TEXT_RECTS = 10;
+function detectColumnGutter(rects) {
+  if (rects.length < 8) return null;
+  let minX = Infinity;
+  let maxX = -Infinity;
+  for (const r of rects) {
+    if (r.x < minX) minX = r.x;
+    if (r.x + r.w > maxX) maxX = r.x + r.w;
+  }
+  const span = maxX - minX;
+  if (!Number.isFinite(span) || span < MIN_CONTENT_SPAN) return null;
+  const narrow = rects.filter((r) => r.w < span * WIDE_RECT_RATIO && r.w > 0 && r.h > 0);
+  if (narrow.length < SIDE_MIN_COUNT * 2) return null;
+  let totalH = 0;
+  for (const r of narrow) totalH += r.h;
+  if (totalH <= 0) return null;
+  const lo = minX + span * GUTTER_SCAN_LO;
+  const hi = minX + span * GUTTER_SCAN_HI;
+  const step = Math.max(2, (hi - lo) / GUTTER_MAX_SAMPLES);
+  let gutterX = 0;
+  let bestCover = Infinity;
+  for (let x = lo; x <= hi; x += step) {
+    let cover = 0;
+    for (const r of narrow) {
+      if (r.x < x && r.x + r.w > x) cover += r.h;
+    }
+    if (cover < bestCover) {
+      bestCover = cover;
+      gutterX = x;
+    }
+  }
+  if (bestCover > totalH * GUTTER_MAX_COVER_RATIO) return null;
+  const left = [];
+  const right = [];
+  for (const r of narrow) {
+    if (r.x + r.w <= gutterX) left.push(r);
+    else if (r.x >= gutterX) right.push(r);
+  }
+  if (left.length < SIDE_MIN_COUNT || right.length < SIDE_MIN_COUNT) return null;
+  const sideStats = (side) => {
+    let sMinX = Infinity, sMaxR = -Infinity, sMinY = Infinity, sMaxT = -Infinity, hSum = 0;
+    for (const r of side) {
+      if (r.x < sMinX) sMinX = r.x;
+      if (r.x + r.w > sMaxR) sMaxR = r.x + r.w;
+      if (r.y < sMinY) sMinY = r.y;
+      if (r.y + r.h > sMaxT) sMaxT = r.y + r.h;
+      hSum += r.h;
+    }
+    return { width: sMaxR - sMinX, vspan: sMaxT - sMinY, minY: sMinY, maxT: sMaxT, hSum };
+  };
+  const L = sideStats(left);
+  const R = sideStats(right);
+  if (Math.min(L.hSum, R.hSum) / Math.max(L.hSum, R.hSum) < SIDE_MIN_HEIGHT_RATIO) return null;
+  const unionVspan = Math.max(L.maxT, R.maxT) - Math.min(L.minY, R.minY);
+  if (unionVspan <= 0) return null;
+  if (L.vspan / unionVspan < SIDE_MIN_VSPAN_RATIO) return null;
+  if (R.vspan / unionVspan < SIDE_MIN_VSPAN_RATIO) return null;
+  if (Math.min(L.width, R.width) / Math.max(L.width, R.width) < SIDE_MIN_WIDTH_SYMMETRY) return null;
+  const textRects = (side) => side.filter((r) => r.h <= TEXT_RECT_MAX_H);
+  const lText = textRects(left);
+  const rText = textRects(right);
+  if (lText.length + rText.length >= BLOCK_DOMINANT_TEXT_RECTS) {
+    const rows = /* @__PURE__ */ new Map();
+    const rowKey = (y) => Math.round(y / ROW_Y_TOL);
+    for (const r of lText) {
+      const k = rowKey(r.y);
+      const e = rows.get(k) ?? { lEnd: -Infinity, rStart: Infinity };
+      if (r.x + r.w > e.lEnd) e.lEnd = r.x + r.w;
+      rows.set(k, e);
+    }
+    const lRowCount = rows.size;
+    for (const r of rText) {
+      const k = rowKey(r.y);
+      const e = rows.get(k) ?? { lEnd: -Infinity, rStart: Infinity };
+      if (r.x < e.rStart) e.rStart = r.x;
+      rows.set(k, e);
+    }
+    let rRowCount = 0;
+    let bothRows = 0;
+    let tinyGapRows = 0;
+    for (const e of rows.values()) {
+      if (e.rStart < Infinity) rRowCount++;
+      if (e.lEnd > -Infinity && e.rStart < Infinity) {
+        bothRows++;
+        if (e.rStart - e.lEnd < CROSS_ROW_TINY_GAP) tinyGapRows++;
+      }
+    }
+    if (lRowCount < SIDE_MIN_TEXT_ROWS || rRowCount < SIDE_MIN_TEXT_ROWS) return null;
+    if (tinyGapRows > MAX_TINY_GAP_ROWS) return null;
+    if (bothRows / Math.min(lRowCount, rRowCount) > MAX_BOTH_SIDE_ROW_RATIO) return null;
+  }
+  return gutterX;
+}
+function orderByGutter(units, rectOf, gutterX) {
+  const left = [];
+  const right = [];
+  const cross = [];
+  for (const u of units) {
+    const r = rectOf(u);
+    const top = r.y + r.h;
+    if (r.x < gutterX && r.x + r.w > gutterX) cross.push({ u, top });
+    else if (r.x + r.w <= gutterX) left.push({ u, top });
+    else right.push({ u, top });
+  }
+  const byTopDesc = (a, b) => b.top - a.top;
+  cross.sort(byTopDesc);
+  left.sort(byTopDesc);
+  right.sort(byTopDesc);
+  const bandOf = (top) => {
+    let k = 0;
+    while (k < cross.length && cross[k].top > top) k++;
+    return k;
+  };
+  const ordered = [];
+  for (let k = 0; k <= cross.length; k++) {
+    for (const t of left) {
+      if (bandOf(t.top) === k) ordered.push(t.u);
+    }
+    for (const t of right) {
+      if (bandOf(t.top) === k) ordered.push(t.u);
+    }
+    if (k < cross.length) ordered.push(cross[k].u);
+  }
+  return ordered;
+}
+
 // src/pdf/columns.ts
 function isProseSpread(items) {
   if (items.length < 4) return false;
@@ -2901,7 +3044,7 @@ function buildGridTable(lines, columns) {
 }
 
 // src/pdf/page-blocks.ts
-function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeight, extraLines) {
+function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeight, extraLines, detectTables = true) {
   if (items.length === 0) return [];
   let { horizontals, verticals } = extractLines(opList.fnArray, opList.argsArray);
   if (extraLines) {
@@ -2915,11 +3058,11 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   verticals = bridgeSplitColumnVerticals(horizontals, verticals);
   markStrikethroughItems(items, horizontals);
   wrapStrikethroughRuns(items);
-  const grids = buildTableGrids(horizontals, verticals);
+  const grids = detectTables ? buildTableGrids(horizontals, verticals) : [];
   if (grids.length > 0) {
     return extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals);
   }
-  return extractPageBlocksFallback(items, pageNum, true);
+  return extractPageBlocksFallback(items, pageNum, true, detectTables);
 }
 var STRIKE_MAX_THICKNESS = 2;
 var STRIKE_MAX_THICKNESS_RATIO = 0.25;
@@ -3117,6 +3260,9 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
     blocks.push({ type: "table", table: irTable, pageNumber: pageNum, bbox: tableBbox });
   }
   let remaining = items.filter((i) => !usedItems.has(i));
+  const groupSizes = [];
+  let finalTextBlocks = [];
+  let gutterX = null;
   if (remaining.length > 0) {
     remaining.sort((a, b) => b.y - a.y || a.x - b.x);
     const clusterItems = remaining.map((i) => ({
@@ -3143,9 +3289,37 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
       }
       remaining = remaining.filter((_, idx) => !usedClusterIndices.has(idx));
     }
-    const groupSizes = [];
-    let finalTextBlocks = [];
-    if (remaining.length > 0) {
+  }
+  {
+    const rects = remaining.map((i) => ({ x: i.x, y: i.y, w: i.w, h: i.h > 0 ? i.h : i.fontSize }));
+    for (const b of blocks) {
+      if (b.bbox) rects.push({ x: b.bbox.x, y: b.bbox.y, w: b.bbox.width, h: b.bbox.height });
+    }
+    gutterX = detectColumnGutter(rects);
+  }
+  if (remaining.length > 0) {
+    if (gutterX !== null) {
+      const gx = gutterX;
+      const allY = remaining.map((i) => i.y);
+      const pageH = safeMax(allY) - safeMin(allY);
+      const gapThreshold = Math.max(15, pageH * 0.03);
+      const sides = [
+        remaining.filter((i) => i.x + i.w <= gx),
+        remaining.filter((i) => i.x < gx && i.x + i.w > gx),
+        remaining.filter((i) => i.x >= gx)
+      ];
+      const textBlocks = [];
+      for (const side of sides) {
+        if (side.length === 0) continue;
+        for (const group of xyCutOrder(side, gapThreshold)) {
+          if (group.length === 0) continue;
+          const groupBlocks = extractPageBlocksFallback(group, pageNum);
+          for (const b of groupBlocks) textBlocks.push(b);
+          groupSizes.push(groupBlocks.length);
+        }
+      }
+      finalTextBlocks = detectListBlocks(textBlocks);
+    } else {
       const allY = remaining.map((i) => i.y);
       const pageH = safeMax(allY) - safeMin(allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
@@ -3158,26 +3332,43 @@ function extractBlocksWithGrids(items, pageNum, grids, horizontals, verticals) {
       }
       finalTextBlocks = detectListBlocks(textBlocks);
     }
-    const units = blocks.map((b) => [b]);
-    let off = 0;
-    for (const size of groupSizes) {
-      const unit = finalTextBlocks.slice(off, off + size);
-      off += size;
-      if (unit.length > 0) units.push(unit);
+  }
+  const units = blocks.map((b) => [b]);
+  let off = 0;
+  for (const size of groupSizes) {
+    const unit = finalTextBlocks.slice(off, off + size);
+    off += size;
+    if (unit.length > 0) units.push(unit);
+  }
+  const unitTopY = (u) => {
+    let top = 0;
+    for (const b of u) {
+      if (b.bbox && b.bbox.y + b.bbox.height > top) top = b.bbox.y + b.bbox.height;
     }
-    const unitTopY = (u) => {
-      let top = 0;
+    return top;
+  };
+  if (gutterX !== null && units.length > 1) {
+    const gx = gutterX;
+    const unitRect = (u) => {
+      let minX = Infinity, minY = Infinity, maxR = -Infinity, maxT = -Infinity;
       for (const b of u) {
-        if (b.bbox && b.bbox.y + b.bbox.height > top) top = b.bbox.y + b.bbox.height;
-      }
-      return top;
+        if (!b.bbox) continue;
+        if (b.bbox.x < minX) minX = b.bbox.x;
+        if (b.bbox.y < minY) minY = b.bbox.y;
+        if (b.bbox.x + b.bbox.width > maxR) maxR = b.bbox.x + b.bbox.width;
+        if (b.bbox.y + b.bbox.height > maxT) maxT = b.bbox.y + b.bbox.height;
+      }
+      if (!Number.isFinite(minX)) return { x: gx - 1, y: 0, w: 2, h: 0 };
+      return { x: minX, y: minY, w: maxR - minX, h: maxT - minY };
     };
-    units.sort((a, b) => unitTopY(b) - unitTopY(a));
-    const ordered = [];
-    for (const u of units) for (const b of u) ordered.push(b);
-    return mergeAdjacentTableBlocks(ordered);
+    const ordered2 = [];
+    for (const u of orderByGutter(units, unitRect, gx)) for (const b of u) ordered2.push(b);
+    return mergeAdjacentTableBlocks(ordered2);
   }
-  return mergeAdjacentTableBlocks(blocks);
+  units.sort((a, b) => unitTopY(b) - unitTopY(a));
+  const ordered = [];
+  for (const u of units) for (const b of u) ordered.push(b);
+  return mergeAdjacentTableBlocks(ordered);
 }
 var NEIGHBOR_TABLE_EPSILON = 0.2;
 function mergeCrossPageTables(blocks) {
@@ -3285,7 +3476,7 @@ function splitTwoColumnProse(items, cutX) {
   }
   return groups;
 }
-function extractPageBlocksFallback(items, pageNum, fullPage = false) {
+function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTables = true) {
   if (items.length === 0) return [];
   const blocks = [];
   const clusterItems = items.map((i) => ({
@@ -3298,7 +3489,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false) {
     fontName: i.fontName,
     hasSpaceBefore: i.hasSpaceBefore
   }));
-  const clusterResults = detectClusterTables(clusterItems, pageNum);
+  const clusterResults = detectTables ? detectClusterTables(clusterItems, pageNum) : [];
   if (clusterResults.length > 0) {
     const ciToIdx = /* @__PURE__ */ new Map();
     for (let ci = 0; ci < clusterItems.length; ci++) ciToIdx.set(clusterItems[ci], ci);
@@ -3326,9 +3517,12 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false) {
       return by - ay;
     });
   } else {
-    const proseCutX = fullPage ? findTwoColumnProseCutX(items) : null;
+    let proseCutX = fullPage ? findTwoColumnProseCutX(items) : null;
+    if (proseCutX === null && fullPage) {
+      proseCutX = detectColumnGutter(items.map((i) => ({ x: i.x, y: i.y, w: i.w, h: i.h > 0 ? i.h : i.fontSize })));
+    }
     const allYLines = mergeSuperscriptLines(groupByY(items));
-    const columns = proseCutX !== null ? null : detectColumns(allYLines);
+    const columns = proseCutX !== null || !detectTables ? null : detectColumns(allYLines);
     if (columns && columns.length >= 3) {
       const tableText = extractWithColumns(allYLines, columns);
       const bbox = computeBBox(items, pageNum);
@@ -3341,7 +3535,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false) {
       for (const group of orderedGroups) {
         if (group.length === 0) continue;
         const yLines = mergeSuperscriptLines(groupByY(group));
-        const groupColumns = detectColumns(yLines);
+        const groupColumns = detectTables ? detectColumns(yLines) : null;
         if (groupColumns && groupColumns.length >= 3) {
           const tableText = extractWithColumns(yLines, groupColumns);
           const bbox = computeBBox(group, pageNum);
@@ -3357,7 +3551,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false) {
       }
     }
   }
-  return detectSpecialKoreanTables(blocks);
+  return detectTables ? detectSpecialKoreanTables(blocks) : blocks;
 }
 
 export {
@@ -3374,4 +3568,4 @@ export {
   extractPageBlocksWithLines,
   mergeCrossPageTables
 };
-//# sourceMappingURL=chunk-2CF2WXBG.js.map
+//# sourceMappingURL=chunk-BGSGUONP.js.map