kordoc 4.18.4 → 4.18.5

package/dist/{chunk-274ZPMHK.js → chunk-N7GNBY2F.js}

@@ -1,6 +1,6 @@
 import {
   blocksToMarkdown
-} from "./chunk-XG4SYOZL.js";
+} from "./chunk-PQ2ER3VQ.js";
 
 // src/page-markdown.ts
 function blocksToPages(blocks, render = blocksToMarkdown) {
@@ -167,4 +167,4 @@ export {
   inlineImagesIntoMarkdown,
   blocksToPages
 };
-//# sourceMappingURL=chunk-274ZPMHK.js.map
+//# sourceMappingURL=chunk-N7GNBY2F.js.map

package/dist/{chunk-T2BJPT6D.js → chunk-OFSMOXS4.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   KordocError
-} from "./chunk-F5YVCJVF.js";
+} from "./chunk-QAE2NVVO.js";
 
 // src/hwpx/profile-io.ts
 import { z } from "zod";
@@ -89,4 +89,4 @@ export {
   formatProfileSchema,
   parseFormatProfileJson
 };
-//# sourceMappingURL=chunk-T2BJPT6D.js.map
+//# sourceMappingURL=chunk-OFSMOXS4.js.map

package/dist/{chunk-TMJVD6FW.js → chunk-OLCBIT3I.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   visibleText
-} from "./chunk-VTVK2JMI.js";
+} from "./chunk-XSCQQ74A.js";
 
 // src/script-tags.ts
 var LITERAL_TAG = /<(?=\/?(?:u|sup|sub)>)/g;
@@ -274,4 +274,4 @@ export {
   HEADING_RATIO_H2,
   HEADING_RATIO_H3
 };
-//# sourceMappingURL=chunk-TMJVD6FW.js.map
+//# sourceMappingURL=chunk-OLCBIT3I.js.map

package/dist/{chunk-XG4SYOZL.js → chunk-PQ2ER3VQ.js}

@@ -1,7 +1,7 @@
 import {
   escapeHtml,
   sanitizeHref
-} from "./chunk-7GR2GCJF.js";
+} from "./chunk-ZH5Y3JT6.js";
 
 // src/script-tags.ts
 var LITERAL_TAG = /<(?=\/?(?:u|sup|sub)>)/g;
@@ -1242,4 +1242,4 @@ export {
   HEADING_RATIO_H2,
   HEADING_RATIO_H3
 };
-//# sourceMappingURL=chunk-XG4SYOZL.js.map
+//# sourceMappingURL=chunk-PQ2ER3VQ.js.map

package/dist/{chunk-F5YVCJVF.js → chunk-QAE2NVVO.js}

@@ -2,7 +2,7 @@
 
 // src/utils.ts
 import { format } from "util";
-var VERSION = true ? "4.18.4" : "0.0.0-dev";
+var VERSION = true ? "4.18.5" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -170,4 +170,4 @@ export {
   classifyError,
   partExtension
 };
-//# sourceMappingURL=chunk-F5YVCJVF.js.map
+//# sourceMappingURL=chunk-QAE2NVVO.js.map

package/dist/{chunk-YCY2R4FM.js → chunk-U4SBX3B5.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   blocksToMarkdown
-} from "./chunk-VTVK2JMI.js";
+} from "./chunk-XSCQQ74A.js";
 
 // src/chunks.ts
 var LIST_MARKER_RE = /^(?:[□■◇◆○◎●◦ㅇ•▪▸▶※-]|\d{1,3}[.)]|[가나다라마바사아자차카타파하][.)]|\([가나다라마바사아자차카타파하0-9]{1,3}\)|[①-⑳㉮-㉻㈎-㈛])\s/;
@@ -79,4 +79,4 @@ function blocksToChunks(blocks, options) {
 export {
   blocksToChunks
 };
-//# sourceMappingURL=chunk-YCY2R4FM.js.map
+//# sourceMappingURL=chunk-U4SBX3B5.js.map

package/dist/{chunk-E2644XCO.cjs → chunk-XGIUYCXI.cjs}

@@ -4,15 +4,15 @@
 
 
 
-var _chunk3WLZOJIEcjs = require('./chunk-3WLZOJIE.cjs');
+var _chunkMU7XW5KOcjs = require('./chunk-MU7XW5KO.cjs');
 
 
 
-var _chunkAVIZ55NQcjs = require('./chunk-AVIZ55NQ.cjs');
+var _chunkH4NV5KRMcjs = require('./chunk-H4NV5KRM.cjs');
 
 
 
-var _chunk3XSP7V5Pcjs = require('./chunk-3XSP7V5P.cjs');
+var _chunkEFVBB45Ecjs = require('./chunk-EFVBB45E.cjs');
 
 // src/ocr/models.ts
 var _path = require('path');
@@ -40,10 +40,10 @@ var OCR_REC_DICT = {
 };
 var ALL_OCR_MODELS = [OCR_DET_MODEL, OCR_REC_MODEL, OCR_REC_DICT];
 function getOcrModelsDir() {
-  return _chunkAVIZ55NQcjs.getModelsDir.call(void 0, "ppocr");
+  return _chunkH4NV5KRMcjs.getModelsDir.call(void 0, "ppocr");
 }
 async function ensureOcrModels(onProgress) {
-  return _chunkAVIZ55NQcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
+  return _chunkH4NV5KRMcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
 }
 async function ocrModelsCached() {
   const dir = getOcrModelsDir();
@@ -2091,7 +2091,7 @@ function tagScripts(text, lines) {
     const [s, e, kind] = marks[m];
     out = out.slice(0, s) + `<${kind}>` + out.slice(s, e) + `</${kind}>` + out.slice(e);
   }
-  return _chunk3WLZOJIEcjs.tidyScriptTags.call(void 0, out);
+  return _chunkMU7XW5KOcjs.tidyScriptTags.call(void 0, out);
 }
 
 // src/pdf/line-wrap.ts
@@ -3644,9 +3644,9 @@ function detectHeadings(blocks, medianFontSize) {
     if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
     const ratio = block.style.fontSize / medianFontSize;
     let level = 0;
-    if (ratio >= _chunk3WLZOJIEcjs.HEADING_RATIO_H1) level = 1;
-    else if (ratio >= _chunk3WLZOJIEcjs.HEADING_RATIO_H2) level = 2;
-    else if (ratio >= _chunk3WLZOJIEcjs.HEADING_RATIO_H3) level = 3;
+    if (ratio >= _chunkMU7XW5KOcjs.HEADING_RATIO_H1) level = 1;
+    else if (ratio >= _chunkMU7XW5KOcjs.HEADING_RATIO_H2) level = 2;
+    else if (ratio >= _chunkMU7XW5KOcjs.HEADING_RATIO_H3) level = 3;
     if (level > 0) {
       block.type = "heading";
       block.level = level;
@@ -3659,7 +3659,7 @@ function isChapterNumber(block, next, medianFontSize) {
   const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _96 => _96.style, 'optionalAccess', _97 => _97.fontSize]), () => ( 0));
   const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _98 => _98.text, 'optionalAccess', _99 => _99.trim, 'call', _100 => _100()]), () => ( ""));
   const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _101 => _101.bbox]);
-  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _102 => _102.text, 'optionalAccess', _103 => _103.trim, 'call', _104 => _104()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunk3WLZOJIEcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
+  return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _102 => _102.text, 'optionalAccess', _103 => _103.trim, 'call', _104 => _104()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkMU7XW5KOcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
 }
 function mergeStackedHeadingLines(blocks, medianFontSize) {
   for (let i = 0; i < blocks.length - 1; ) {
@@ -5959,16 +5959,16 @@ 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 = _chunk3XSP7V5Pcjs.safeMax.call(void 0, gaps);
+  const maxGap = _chunkEFVBB45Ecjs.safeMax.call(void 0, gaps);
   const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
   if (items.length >= 4 && maxGap < 40 && avgLen < 5) return true;
-  const minGap = _chunk3XSP7V5Pcjs.safeMin.call(void 0, gaps);
+  const minGap = _chunkEFVBB45Ecjs.safeMin.call(void 0, gaps);
   return minGap > 0 && maxGap - minGap <= Math.max(2, maxGap * 0.15) && items.filter((i) => /^\P{L}*\p{Ll}/u.test(i.text)).length * 2 >= items.length;
 }
 function detectColumns(yLines) {
   const allItems = yLines.flat();
   if (allItems.length === 0) return null;
-  const pageWidth = _chunk3XSP7V5Pcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunk3XSP7V5Pcjs.safeMin.call(void 0, allItems.map((i) => i.x));
+  const pageWidth = _chunkEFVBB45Ecjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkEFVBB45Ecjs.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++) {
@@ -6438,12 +6438,12 @@ function recordClipCellEdges(grids, horizontals, verticals, nonRules) {
 }
 function takeClipCellEdges(from, to) {
   const e = CLIP_CELL_EDGES.get(from);
-  if (e) _chunk3WLZOJIEcjs.CELL_EDGES.set(to, e);
+  if (e) _chunkMU7XW5KOcjs.CELL_EDGES.set(to, e);
 }
 function joinCellEdges(a, b) {
-  const u = _chunk3WLZOJIEcjs.CELL_EDGES.get(a), d = _chunk3WLZOJIEcjs.CELL_EDGES.get(b);
+  const u = _chunkMU7XW5KOcjs.CELL_EDGES.get(a), d = _chunkMU7XW5KOcjs.CELL_EDGES.get(b);
   if (!u || !d) return;
-  _chunk3WLZOJIEcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
+  _chunkMU7XW5KOcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
 }
 
 // src/pdf/table-unit-row.ts
@@ -6459,7 +6459,7 @@ function rebuildUnitLine(items, grid) {
 function prependUnitRow(cells, cols, line, usedItems) {
   const text = cleanCellText(line.map((item) => item.text).join(""));
   cells.unshift(Array.from({ length: cols }, (_, c) => ({ text: c === 0 ? text : "", colSpan: c === 0 ? cols : 1, rowSpan: 1 })));
-  for (const cell of cells[0]) _chunk3WLZOJIEcjs.CELL_EDGES.set(cell, NO_EDGES);
+  for (const cell of cells[0]) _chunkMU7XW5KOcjs.CELL_EDGES.set(cell, NO_EDGES);
   for (const item of line) usedItems.add(item);
 }
 function attachUnitRow(items, grid, cells, cols, usedItems) {
@@ -6901,12 +6901,61 @@ function frameLayoutBoxes(blocks) {
   }
   return boxes;
 }
+function takeFrameSpanningText(items, cutX, pageNum, lex) {
+  let minX = Infinity, maxR = -Infinity;
+  for (const item of items) {
+    minX = Math.min(minX, item.x);
+    maxR = Math.max(maxR, item.x + item.w);
+  }
+  const span = maxR - minX;
+  const rows = groupByY([...items].sort((a, b) => b.y - a.y || a.x - b.x)).map((line) => {
+    const runs = [];
+    for (const item of [...line].sort((a, b) => a.x - b.x)) {
+      const run = runs[runs.length - 1], prev = _optionalChain([run, 'optionalAccess', _315 => _315[run.length - 1]]);
+      if (prev && item.x - (prev.x + prev.w) <= Math.max(prev.fontSize, item.fontSize) * 0.6) run.push(item);
+      else runs.push([item]);
+    }
+    return runs;
+  });
+  const taken = /* @__PURE__ */ new Set(), blocks = [];
+  for (let r = 0; r < rows.length; r++) for (const run of rows[r]) {
+    if (taken.has(run[0]) || run[0].x >= cutX || Math.max(...run.map((i) => i.x + i.w)) <= cutX) continue;
+    const lines = [run], fs = Math.max(...run.map((i) => i.fontSize));
+    const left = run[0].x, right = Math.max(...run.map((i) => i.x + i.w));
+    if (right - left < span * 0.75) continue;
+    let prevY = run[0].y;
+    for (let next = r + 1; next < rows.length && rows[next].length === 1; next++) {
+      const tail = rows[next][0], y = tail[0].y;
+      if (taken.has(tail[0]) || prevY - y > fs * 1.8 || tail[0].x < left - 1 || tail[0].x > left + fs * 2 || tail.some((i) => Math.abs(i.fontSize - fs) > fs * 0.15) || Math.max(...tail.map((i) => i.x + i.w)) > right + 1) break;
+      lines.push(tail);
+      prevY = y;
+    }
+    for (const line of lines) for (const item of line) taken.add(item);
+    pushLineParagraphs(blocks, mergeSuperscriptLines(lines), pageNum, lex);
+  }
+  let keep = 0;
+  for (const item of items) if (!taken.has(item)) items[keep++] = item;
+  items.length = keep;
+  return blocks;
+}
+function frameColumnTextBands(items, spanning, cutX) {
+  const tops = spanning.filter((b) => b.bbox).map((b) => b.bbox.y + b.bbox.height).sort((a, b) => b - a);
+  const bands = /* @__PURE__ */ new Map();
+  for (const item of items) {
+    let band = 0;
+    while (band < tops.length && tops[band] > item.y + item.h) band++;
+    const sides = _nullishCoalesce(bands.get(band), () => ( [[], [], []]));
+    sides[item.x + item.w <= cutX ? 0 : item.x >= cutX ? 2 : 1].push(item);
+    bands.set(band, sides);
+  }
+  return [...bands].sort((a, b) => a[0] - b[0]).flatMap(([, sides]) => sides.filter((side) => side.length));
+}
 function groupFrameParagraphUnits(units, frames) {
   const source = /* @__PURE__ */ new Map();
   for (const unit of units) for (const block of unit) source.set(block, unit);
   const members = /* @__PURE__ */ new Map(), emitted = /* @__PURE__ */ new Set();
   for (const frame of frames) {
-    if (frame.length < 2 || frame.some((block) => block.type !== "paragraph" || _optionalChain([source, 'access', _315 => _315.get, 'call', _316 => _316(block), 'optionalAccess', _317 => _317.length]) !== 1)) continue;
+    if (frame.length < 2 || frame.some((block) => block.type !== "paragraph" || _optionalChain([source, 'access', _316 => _316.get, 'call', _317 => _317(block), 'optionalAccess', _318 => _318.length]) !== 1)) continue;
     for (const block of frame) members.set(block, frame);
   }
   const out = [];
@@ -6919,7 +6968,7 @@ function groupFrameParagraphUnits(units, frames) {
     }
   }
   for (const unit of out) {
-    if (unit.length < 2 || _optionalChain([FRAME_READING_UNITS, 'access', _318 => _318.get, 'call', _319 => _319(unit[0]), 'optionalAccess', _320 => _320.blocks]) === unit) continue;
+    if (unit.length < 2 || _optionalChain([FRAME_READING_UNITS, 'access', _319 => _319.get, 'call', _320 => _320(unit[0]), 'optionalAccess', _321 => _321.blocks]) === unit) continue;
     const bs = unit.map((b) => b.bbox).filter((b) => !!b);
     if (bs.length !== unit.length) continue;
     const x = Math.min(...bs.map((b) => b.x)), y = Math.min(...bs.map((b) => b.y));
@@ -6936,14 +6985,14 @@ function groupFrameParagraphUnits(units, frames) {
 function recordFrameTitle(table, source, pageNum) {
   if (!CLIP_TABLES.has(table) || table.rows !== 1 || table.cols !== 3 || source.length < 2) return;
   const cells = table.cells[0];
-  if (cells.length !== 3 || cells.some((c) => c.colSpan !== 1 || c.rowSpan !== 1 || _optionalChain([c, 'access', _321 => _321.blocks, 'optionalAccess', _322 => _322.length]) || c.text.includes("\n"))) return;
+  if (cells.length !== 3 || cells.some((c) => c.colSpan !== 1 || c.rowSpan !== 1 || _optionalChain([c, 'access', _322 => _322.blocks, 'optionalAccess', _323 => _323.length]) || c.text.includes("\n"))) return;
   const [label, spacer, title] = cells;
   if (!/^(?:붙임|별첨|첨부|부록)\s*\d{1,3}$/.test(label.text.trim()) || spacer.text.trim() || !title.text.trim()) return;
   if (cells.some((c) => {
-    const e = _chunk3WLZOJIEcjs.CELL_EDGES.get(c);
+    const e = _chunkMU7XW5KOcjs.CELL_EDGES.get(c);
     return !e || e.t || e.b || e.l || e.r;
   })) return;
-  const xs = TABLE_COLXS.get(table), style = dominantStyle(source), fs = _nullishCoalesce(_optionalChain([style, 'optionalAccess', _323 => _323.fontSize]), () => ( 0));
+  const xs = TABLE_COLXS.get(table), style = dominantStyle(source), fs = _nullishCoalesce(_optionalChain([style, 'optionalAccess', _324 => _324.fontSize]), () => ( 0));
   if (!xs || xs.length !== 4 || fs <= 0) return;
   const width = xs[3] - xs[0], gap = xs[2] - xs[1];
   if (width < fs * 12 || gap <= 0 || gap > fs * 1.5 || xs[1] - xs[0] > width * 0.25 || xs[3] - xs[2] < width * 0.6) return;
@@ -7001,7 +7050,7 @@ function extendNestedShadedHeaders(clipGrids, lineGrids, horizontals, verticals,
   const extended = [];
   for (let i = 0; i < clipGrids.length; i++) {
     const c = clipGrids[i], parent = c.clipParent;
-    if (!parent || c.rowYs.length !== 2 || c.colXs.length < 3 || !_optionalChain([c, 'access', _324 => _324.cells, 'optionalAccess', _325 => _325.length]) || c.cells.some((cell) => cell.filler || cell.rowSpan !== 1 || cell.colSpan !== 1)) continue;
+    if (!parent || c.rowYs.length !== 2 || c.colXs.length < 3 || !_optionalChain([c, 'access', _325 => _325.cells, 'optionalAccess', _326 => _326.length]) || c.cells.some((cell) => cell.filler || cell.rowSpan !== 1 || cell.colSpan !== 1)) continue;
     if (!clipGrids.some((g) => g.rowYs.length === 2 && g.colXs.length === 2 && near(g.bbox.x1, parent.x1) && near(g.bbox.x2, parent.x2) && near(g.bbox.y1, parent.y1) && near(g.bbox.y2, parent.y2))) continue;
     if (!c.cells.every((cell) => fills.some((f) => near(f.x1, cell.bbox.x1) && near(f.x2, cell.bbox.x2) && near(f.y1, cell.bbox.y1) && near(f.y2, cell.bbox.y2)))) continue;
     const host = lineGrids.find((l) => l.rowYs.length > 2 && l.colXs.length === c.colXs.length && c.colXs.every((x, k) => near(x, l.colXs[k])) && c.rowYs.every((y, k) => near(y, l.rowYs[k])) && contains3(parent, l.bbox) && l.rowYs.every((y) => horizontals.some((h) => near(h.y1, y) && near(h.x1, l.bbox.x1) && near(h.x2, l.bbox.x2))) && l.colXs.every((x) => verticals.some((v) => near(v.x1, x) && v.y1 <= l.bbox.y1 + 1.5 && v.y2 >= l.bbox.y2 - 1.5)) && !clipGrids.some((o) => o !== c && !contains3(o.bbox, l.bbox) && Math.min(o.bbox.x2, l.bbox.x2) > Math.max(o.bbox.x1, l.bbox.x1) && Math.min(o.bbox.y2, l.bbox.y2) > Math.max(o.bbox.y1, l.bbox.y1)));
@@ -7189,7 +7238,7 @@ function stackedTableBands(items) {
   const caption = (start, end) => lines.slice(start, end).some((line) => /^Table\s+\d+\s*[:.]/i.test(mergeLineSimple(line).trim()));
   while (runs.length > 0 && !caption(runs[runs.length - 1].end, lines.length)) runs.pop();
   if (runs.length < 2) return null;
-  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _326 => _326[i + 1], 'optionalAccess', _327 => _327.start]), () => ( lines.length))))) return null;
+  if (runs.some((run, i) => !caption(run.end, _nullishCoalesce(_optionalChain([runs, 'access', _327 => _327[i + 1], 'optionalAccess', _328 => _328.start]), () => ( lines.length))))) return null;
   const tables = runs.map((run) => lines.slice(run.start, run.end).flat());
   const between = runs.slice(0, -1).map((run, i) => lines.slice(run.end, runs[i + 1].start).flat());
   const tail = lines.slice(runs[runs.length - 1].end);
@@ -7358,7 +7407,7 @@ function extractPageBlocksWithLines(items, pageNum, opList, pageWidth, pageHeigh
   const filled = fillBlanks(items, horizontals, verticals);
   items = filled.items;
   horizontals = filled.horizontals;
-  const prevPage = _optionalChain([carry, 'optionalAccess', _328 => _328.page]) === pageNum - 1 ? carry.clip : void 0;
+  const prevPage = _optionalChain([carry, 'optionalAccess', _329 => _329.page]) === pageNum - 1 ? carry.clip : void 0;
   const clipResult = detectTables ? buildClipCellGrids(extracted.clipRects, horizontals, verticals, pageWidth, pageHeight, items.map((it) => ({ x: it.x + it.w / 2, y: it.y + it.h / 2 })), extracted.fillRects, prevPage) : { grids: [], containers: [], page: void 0 };
   if (carry) {
     carry.page = pageNum;
@@ -7626,14 +7675,14 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
         if (!assignedItems.has(textItems[ti])) usedItems.delete(tableItems[ti]);
       }
     }
-    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _329 => _329[0], 'optionalAccess', _330 => _330[0], 'optionalAccess', _331 => _331.text]), () => ( "")))) {
+    if (unitLine.length > 0 && rebuiltUsed && !/^\s*\(\s*단위\s*[:：]/.test(_nullishCoalesce(_optionalChain([finalGrid, 'access', _330 => _330[0], 'optionalAccess', _331 => _331[0], 'optionalAccess', _332 => _332.text]), () => ( "")))) {
       prependUnitRow(finalGrid, numCols, unitLine, usedItems);
       finalRows++;
     }
     if (!rebuiltUsed && attachUnitRow(items, grid, finalGrid, numCols, usedItems)) finalRows++;
     let semanticOneColumn = false;
     if (!grid.cells && !nestedAttached && numCols === 1 && numGridRows >= 2) {
-      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _332 => _332[0], 'optionalAccess', _333 => _333.text, 'access', _334 => _334.trim, 'call', _335 => _335()]));
+      const populatedRows = finalGrid.filter((row) => _optionalChain([row, 'access', _333 => _333[0], 'optionalAccess', _334 => _334.text, 'access', _335 => _335.trim, 'call', _336 => _336()]));
       if (populatedRows.length >= 4) {
         const fontSizes = tableItems.map((item) => item.fontSize).filter((size) => size > 0).sort((a, b) => a - b);
         const medianFont = _nullishCoalesce(fontSizes[Math.floor(fontSizes.length / 2)], () => ( 0));
@@ -7719,7 +7768,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       height: grid.bbox.y2 - grid.bbox.y1
     };
     if (numRows === 1 && numCols === 1 && gridW < pageWidth * 0.35) {
-      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _336 => _336[0], 'optionalAccess', _337 => _337[0], 'optionalAccess', _338 => _338.text]), () => ( ""));
+      const prose = _nullishCoalesce(_optionalChain([finalGrid, 'access', _337 => _337[0], 'optionalAccess', _338 => _338[0], 'optionalAccess', _339 => _339.text]), () => ( ""));
       const besideLines = (side) => new Set(side.map((it) => Math.round(it.y / 3))).size;
       const rightLines = items.filter((it) => it.x >= grid.bbox.x2 + 3 && it.y >= grid.bbox.y1 && it.y <= grid.bbox.y2);
       const leftLines = items.filter((it) => it.x + it.w <= grid.bbox.x1 - 3 && it.y >= grid.bbox.y1 && it.y <= grid.bbox.y2);
@@ -7738,8 +7787,8 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
       }
     }
     if (!grid.cells && !nestedAttached && shouldDemoteTable(irTable)) {
-      const paragraphs = numRows === 1 && numCols === 1 ? _optionalChain([irTable, 'access', _339 => _339.cells, 'access', _340 => _340[0], 'optionalAccess', _341 => _341[0], 'optionalAccess', _342 => _342.blocks]) : void 0;
-      if (_optionalChain([paragraphs, 'optionalAccess', _343 => _343.length])) {
+      const paragraphs = numRows === 1 && numCols === 1 ? _optionalChain([irTable, 'access', _340 => _340.cells, 'access', _341 => _341[0], 'optionalAccess', _342 => _342[0], 'optionalAccess', _343 => _343.blocks]) : void 0;
+      if (_optionalChain([paragraphs, 'optionalAccess', _344 => _344.length])) {
         blocks.push(...paragraphs);
         frameParagraphUnits.push(paragraphs);
         recordFrameReadingUnit(paragraphs, tableBbox);
@@ -7780,7 +7829,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
     const proseColumns = findTwoColumnProseCutX(clusterItems) !== null || detectPersistentColumnGutter(remaining.map((i) => ({ x: i.x, y: i.y, w: i.w, h: i.h > 0 ? i.h : i.fontSize }))) !== null;
     const clusterResults = (proseColumns ? [] : detectClusterTables(clusterItems, pageNum)).filter((cr) => {
       const b = cr.bbox;
-      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (_nullishCoalesce(_optionalChain([x, 'access', _344 => _344.text, 'optionalAccess', _345 => _345.match, 'call', _346 => _346(/\p{L}/gu), 'optionalAccess', _347 => _347.length]), () => ( 0))) >= 10 && Math.min(x.bbox.x + x.bbox.width, b.x + b.width) - Math.max(x.bbox.x, b.x) >= 0.8 * b.width && x.bbox.y >= b.y - 2 && x.bbox.y + x.bbox.height <= b.y + b.height + 2);
+      return !blocks.some((x) => x.type === "paragraph" && x.bbox && (_nullishCoalesce(_optionalChain([x, 'access', _345 => _345.text, 'optionalAccess', _346 => _346.match, 'call', _347 => _347(/\p{L}/gu), 'optionalAccess', _348 => _348.length]), () => ( 0))) >= 10 && Math.min(x.bbox.x + x.bbox.width, b.x + b.width) - Math.max(x.bbox.x, b.x) >= 0.8 * b.width && x.bbox.y >= b.y - 2 && x.bbox.y + x.bbox.height <= b.y + b.height + 2);
     });
     if (clusterResults.length > 0) {
       const ciToIdx = /* @__PURE__ */ new Map();
@@ -7811,14 +7860,15 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
   if (remaining.length > 0) {
     if (gutterX !== null) {
       const gx = gutterX;
+      const spanning = detectListBlocks(takeFrameSpanningText(remaining, gx, pageNum, lex));
+      for (const block of spanning) {
+        finalTextBlocks.push(block);
+        groupSizes.push(1);
+      }
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunk3XSP7V5Pcjs.safeMax.call(void 0, allY) - _chunk3XSP7V5Pcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkEFVBB45Ecjs.safeMax.call(void 0, allY) - _chunkEFVBB45Ecjs.safeMin.call(void 0, 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 sides = frameColumnTextBands(remaining, spanning, gx);
       const textBlocks = [];
       for (const side of sides) {
         if (side.length === 0) continue;
@@ -7829,10 +7879,10 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
           groupSizes.push(groupBlocks.length);
         }
       }
-      finalTextBlocks = detectListBlocks(textBlocks);
+      finalTextBlocks.push(...detectListBlocks(textBlocks));
     } else {
       const allY = remaining.map((i) => i.y);
-      const pageH = _chunk3XSP7V5Pcjs.safeMax.call(void 0, allY) - _chunk3XSP7V5Pcjs.safeMin.call(void 0, allY);
+      const pageH = _chunkEFVBB45Ecjs.safeMax.call(void 0, allY) - _chunkEFVBB45Ecjs.safeMin.call(void 0, allY);
       const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
       const textBlocks = [];
       for (const group of groups) {
@@ -8157,7 +8207,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
       blocks.push(...columnTextToBlocks(tableText, pageNum, bbox, dominantStyle(items)));
     } else {
       const allY = items.map((i) => i.y);
-      const pageHeight = _chunk3XSP7V5Pcjs.safeMax.call(void 0, allY) - _chunk3XSP7V5Pcjs.safeMin.call(void 0, allY);
+      const pageHeight = _chunkEFVBB45Ecjs.safeMax.call(void 0, allY) - _chunkEFVBB45Ecjs.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) {
@@ -8235,4 +8285,4 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
 
 
 exports.OCR_DET_MODEL = OCR_DET_MODEL; exports.OCR_REC_MODEL = OCR_REC_MODEL; exports.OCR_REC_DICT = OCR_REC_DICT; exports.getOcrModelsDir = getOcrModelsDir; exports.ensureOcrModels = ensureOcrModels; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.computeBBox = computeBBox; exports.dominantStyle = dominantStyle; exports.normalizeItems = normalizeItems; exports.groupByY = groupByY; exports.mergeLineSimple = mergeLineSimple; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.latinSoftWrap = latinSoftWrap; exports.TOC_BLOCKS = TOC_BLOCKS; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.mergeStackedHeadingLines = mergeStackedHeadingLines; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectDocumentStyleHeadings = detectDocumentStyleHeadings; exports.detectSiblingStyleHeadings = detectSiblingStyleHeadings; exports.detectRepeatedPageLabels = detectRepeatedPageLabels; exports.detectPageLeadHeadings = detectPageLeadHeadings; exports.refineDocumentStyleHeadings = refineDocumentStyleHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.FRAME_TITLE_BLOCKS = FRAME_TITLE_BLOCKS; exports.TABLE_COLXS = TABLE_COLXS; exports.TABLE_ROWYS = TABLE_ROWYS; exports.TABLE_TAIL = TABLE_TAIL; exports.PART_COLXS = PART_COLXS; exports.EMPTY_PARTS = EMPTY_PARTS; exports.IMAGE_CELLS = IMAGE_CELLS; exports.FILLER_CELLS = FILLER_CELLS; exports.CONT_PARTS = CONT_PARTS; exports.CELL_LINES = CELL_LINES; exports.ROW_RULES = ROW_RULES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.NO_EDGES = NO_EDGES; exports.joinCellEdges = joinCellEdges; exports.FRAME_READING_UNITS = FRAME_READING_UNITS; exports.recordFrameReadingUnit = recordFrameReadingUnit; exports.restoreImageBullets = restoreImageBullets; exports.detectRightArrowRegions = detectRightArrowRegions; exports.removeSideTabs = removeSideTabs; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
-//# sourceMappingURL=chunk-E2644XCO.cjs.map
+//# sourceMappingURL=chunk-XGIUYCXI.cjs.map