kordoc 4.24.19 → 4.24.21

package/dist/index.js

@@ -2,7 +2,7 @@ import {
   blocksToPages,
   bmpToPng,
   inlineImagesIntoMarkdown
-} from "./chunk-MRFBERH2.js";
+} from "./chunk-CPVXUEKF.js";
 import {
   parsePageRange
 } from "./chunk-V4BH5SRQ.js";
@@ -44,7 +44,7 @@ import {
   validateOcrOptions,
   visibleText,
   wrapScript
-} from "./chunk-JQJYIYMJ.js";
+} from "./chunk-PDOWSDNM.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT,
@@ -62,7 +62,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-RQCGVE3R.js";
+} from "./chunk-UNYH6MTO.js";
 
 // src/plain-markdown.ts
 function toPlainMarkdown(md2) {
@@ -24645,7 +24645,7 @@ var MIN_SIDE = 24;
 async function ocrEmbeddedImages(blocks, options, warnings) {
   const ocr = options?.ocr;
   if (ocr !== true && ocr !== "force" && typeof ocr !== "function") return false;
-  const { parseImageDocument } = await import("./image-ocr-Y75JNXPZ.js");
+  const { parseImageDocument } = await import("./image-ocr-5YWVBRY6.js");
   let inserted = false, unreadable = 0;
   let failed = null;
   const walk = async (list, depth) => {
@@ -53524,9 +53524,10 @@ function solveRowHeights(cells, rowCount, maxTotal = Number.POSITIVE_INFINITY, s
   }
   for (let k = 0; k < spans.length; k++) if (!done.has(k)) solve(k, false);
   let total = h.reduce((s, x) => s + x, 0);
+  const overSz = total > maxTotal;
   for (const { c, rows } of [...spans].sort((a, b) => a.rows[a.rows.length - 1] - b.rows[b.rows.length - 1] || b.rows[0] - a.rows[0])) {
     const known = rows.reduce((s, r) => s + h[r], 0);
-    const add = Math.min(c.height - known, maxTotal - total);
+    const add = Math.min(c.height - known, overSz && c.contentH != null && c.contentH >= c.height ? Number.POSITIVE_INFINITY : maxTotal - total);
     if (rows.length && add > 0) {
       h[rows[rows.length - 1]] += add;
       total += add;
@@ -55596,12 +55597,24 @@ function footNoteStories(root, paraPages, geom) {
   const above = num2(sp, "aboveLine", 850), below = num2(sp, "belowLine", 567), between = num2(sp, "betweenNotes", 283);
   const byPage = /* @__PURE__ */ new Map();
   for (const p of elements(root)) if (ln(p) === "p") {
-    const page = paraPages.get(p)?.[0] ?? 0;
+    const pages = paraPages.get(p) ?? [];
+    const m = pages.length > 1 ? buildPara(p) : null;
     for (const run of elements(p)) if (ln(run) === "run") {
       for (const ctrl of elements(run)) if (ln(ctrl) === "ctrl") {
         const fn = findChildByLocalName(ctrl, "footNote");
         const sub = fn && !isDeletedControl(fn) ? findChildByLocalName(fn, "subList") : null;
         if (!sub) continue;
+        let page = pages[0] ?? 0;
+        if (m) {
+          const at = m.chars.findIndex((c) => c.note === fn);
+          if (at >= 0) {
+            let k = 0;
+            m.segs.forEach((sg, i) => {
+              if (sg.textpos <= at) k = i;
+            });
+            page = pages[k] ?? page;
+          }
+        }
         let h = 0;
         for (const q of elements(sub)) if (ln(q) === "p") {
           const lsa = findChildByLocalName(q, "linesegarray");
@@ -55713,18 +55726,30 @@ function stackedFloats(c) {
   }
   return out;
 }
-function cellLineBoxes(c) {
+function cellLineBoxes(c, bodyH) {
   const out = [];
   let shift = 0, prevV = Number.NEGATIVE_INFINITY, prevH = 0, prevSavedEnd = 0, prevEnd = 0, part = 0;
   const cols = cellColumns(c.sub);
+  let floatBottom = Number.NEGATIVE_INFINITY;
+  let nestedCell = false;
+  for (let a = c.el.parentNode; a; a = a.parentNode) if (a !== c.el && a.nodeType === 1 && ln(a) === "tc") {
+    nestedCell = true;
+    break;
+  }
   for (const p of c.sub ? elements(c.sub) : []) if (ln(p) === "p") {
     const lsa = findChildByLocalName(p, "linesegarray");
+    let first = true, pTop = 0;
     for (const sg of lsa ? elements(lsa) : []) if (ln(sg) === "lineseg") {
       const v = num2(sg, "vertpos"), hz = num2(sg, "horzpos");
       if ((v < prevV || v === prevV && hz <= prevH && prevSavedEnd > v) && !cols) {
         shift = prevEnd - v;
         part++;
+      } else if (!nestedCell && bodyH !== void 0 && floatBottom > bodyH && first && !cols && v >= prevV && v < floatBottom - 1e3) {
+        shift = prevEnd - v;
+        part++;
       }
+      if (first) pTop = prevSavedEnd > 0 && prevSavedEnd <= v ? prevSavedEnd : v;
+      first = false;
       prevV = v;
       prevH = hz;
       prevSavedEnd = v + num2(sg, "vertsize") + num2(sg, "spacing");
@@ -55732,17 +55757,26 @@ function cellLineBoxes(c) {
       prevEnd = bottom + num2(sg, "spacing");
       out.push({ seg: sg, top, bottom, part, origin: shift });
     }
+    floatBottom = Number.NEGATIVE_INFINITY;
+    for (const run of elements(p)) if (ln(run) === "run") for (const o of elements(run)) {
+      if (ln(o) !== "tbl" || (o.getAttribute("textWrap") ?? "TOP_AND_BOTTOM") !== "TOP_AND_BOTTOM") continue;
+      const pos = findChildByLocalName(o, "pos");
+      if (!pos || pos.getAttribute("treatAsChar") === "1" || (pos.getAttribute("vertRelTo") ?? "PARA") !== "PARA") continue;
+      const om = findChildByLocalName(o, "outMargin"), vo = num2(pos, "vertOffset"), sz = num2(findChildByLocalName(o, "sz"), "height"), full = measureTableHeight(o);
+      const top = pTop + (vo < 2 ** 31 ? vo : 0) + num2(om, "top");
+      if (full > sz + 1e3 && (bodyH === void 0 || top + (measureTableRowHeights(o)[0] ?? 0) <= bodyH)) floatBottom = Math.max(floatBottom, top + full + num2(om, "bottom"));
+    }
   }
   return out;
 }
-function cellLines(c) {
-  return cellLineBoxes(c).map((b) => [b.top, b.bottom]);
+function cellLines(c, bodyH) {
+  return cellLineBoxes(c, bodyH).map((b) => [b.top, b.bottom]);
 }
-function spanPartRows(cells, rowY) {
+function spanPartRows(cells, rowY, bodyH) {
   const out = /* @__PURE__ */ new Set();
   for (const c of cells) {
     if (c.rs < 2) continue;
-    const bs = cellLineBoxes(c);
+    const bs = cellLineBoxes(c, bodyH);
     const parts = bs.length ? bs[bs.length - 1].part : 0;
     let s = c.ra;
     for (let k = 0; k < parts; k++) {
@@ -55786,12 +55820,12 @@ function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan,
     const st = stackedFloats(rc[0]);
     return st.length >= 2 && st.reduce((a, f) => a + f.h, 0) > h;
   });
-  if (slack > 0 && ty + rowY[nRows] > pageBottom && ty + rowY[nRows] <= pageBottom + slack && cells.some((c) => cellLineBoxes(c).some((b) => b.part > 0))) slack = 0;
+  if (slack > 0 && ty + rowY[nRows] > pageBottom && ty + rowY[nRows] <= pageBottom + slack && cells.some((c) => cellLineBoxes(c, BODY_H).some((b) => b.part > 0))) slack = 0;
   if (slack > 0 && ty + rowY[nRows] > pageBottom && ty + rowY[nRows] <= pageBottom + slack) {
     const declared = num2(findChildByLocalName(tbl2, "sz"), "height");
     if (rowY.some((v, k) => k > 0 && k < nRows && Math.abs(v - declared) <= 2)) slack = 0;
   }
-  const savedParts = tbl2.getAttribute("pageBreak") === "CELL" && cells.some((c) => c.h > rowY[nRows] && cellLineBoxes(c).some((b) => b.part > 0));
+  const savedParts = tbl2.getAttribute("pageBreak") === "CELL" && cells.some((c) => c.h > rowY[nRows] && cellLineBoxes(c, BODY_H).some((b) => b.part > 0));
   if (BODY_H <= 0 || ty + rowY[nRows] <= pageBottom + slack && !stackOver && !savedParts) return plain4();
   if (findChildByLocalName(tbl2, "pos")?.getAttribute("treatAsChar") === "1") {
     const declared = num2(findChildByLocalName(tbl2, "sz"), "height") || rowY[nRows];
@@ -55806,11 +55840,15 @@ function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan,
   const out = [];
   const flows = (findChildByLocalName(tbl2, "pos")?.getAttribute("vertRelTo") ?? "PARA") === "PARA";
   let page = 0, y = ty, first = !flows || ty - MT < 2e3;
+  if (!first && (tbl2.getAttribute("pageBreak") ?? "NONE") === "NONE") {
+    const outT = num2(findChildByLocalName(tbl2, "outMargin"), "top");
+    if (outT + rowY[nRows] + outB <= BODY_H) return rowH.map((h, r) => [{ page: 1, y: contMT + outT + rowY[r], h }]);
+  }
   if (!first && coAnchoredFollower(tbl2)) {
     const outT = num2(findChildByLocalName(tbl2, "outMargin"), "top");
     if (outT + rowY[nRows] + outB <= BODY_H) return rowH.map((h, r) => [{ page: 1, y: contMT + outT + rowY[r], h }]);
   }
-  const spanStarts = cellSplit ? spanPartRows(cells, rowY) : /* @__PURE__ */ new Set();
+  const spanStarts = cellSplit ? spanPartRows(cells, rowY, BODY_H) : /* @__PURE__ */ new Set();
   for (let r = 0; r < nRows; r++) {
     let h = rowH[r];
     const pieces = [];
@@ -55849,7 +55887,7 @@ function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan,
         continue;
       }
     }
-    const boxes = splittable ? new Map(rowCells.map((c) => [c.el, cellLineBoxes(c)])) : void 0;
+    const boxes = splittable ? new Map(rowCells.map((c) => [c.el, cellLineBoxes(c, BODY_H)])) : void 0;
     const hParts = boxes ? Math.max(0, ...[...boxes.values()].flatMap((bs) => bs.map((b) => b.part))) : 0;
     if (boxes && hParts > 0) {
       if (!first && rowCells.some((c) => {
@@ -55959,9 +55997,9 @@ function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan,
     }
     if (splittable && y + h + outB > pageBottom) {
       const from = /* @__PURE__ */ new Map();
-      const lines = new Map(rowCells.map((c) => [c.el, cellLines(c)]));
+      const lines = new Map(rowCells.map((c) => [c.el, cellLines(c, BODY_H)]));
       const sp0 = new Map(rowCells.map((c) => {
-        const b = cellLineBoxes(c)[0];
+        const b = cellLineBoxes(c, BODY_H)[0];
         return [c.el, b ? num2(b.seg, "spacing") : 0];
       }));
       while (y + h + outB > pageBottom && BODY_H > rowY[hdrRows]) {
@@ -55990,7 +56028,7 @@ function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan,
         if (allFit) {
           const spanCells = cells.filter((c) => c.ra === r && c.rs > 1);
           const spanExt = (c) => {
-            const bs = cellLineBoxes(c);
+            const bs = cellLineBoxes(c, BODY_H);
             return bs.some((b) => b.part > 0) ? Math.max(...bs.filter((b) => b.part === 0).map((b) => b.bottom)) : cellContentExtent(c);
           };
           if (!first && h <= pageBottom - contTop && [...rowCells.map((c) => cellContentExtent(c) + c.marginT + c.marginB), ...spanCells.map((c) => spanExt(c) + c.marginT + c.marginB)].some((e) => e > avail)) {
@@ -57140,7 +57178,7 @@ function drawTable(tbl2, tx, ty, ctx, depth) {
       const span2 = plan.slice(c.ra, Math.min(c.ra + c.rs, nRows)).flat();
       const pages = [...new Set(span2.map((pc) => pc.page))];
       if (pages.length > 1) {
-        const bs = cellLineBoxes(c);
+        const bs = cellLineBoxes(c, ctx.geom.BODY_H);
         const partTop = (k) => bs.find((b) => b.part === k);
         const groups = pages.map((pg, k) => {
           const on = span2.filter((pc) => pc.page === pg);
@@ -57173,7 +57211,7 @@ function drawTable(tbl2, tx, ty, ctx, depth) {
     ctx.page = g3.page;
     if (g3.pieces) {
       const base = g3.pieces[0].y + c.marginT;
-      const boxes = cellLineBoxes(c);
+      const boxes = cellLineBoxes(c, ctx.geom.BODY_H);
       const contV = new Map(boxes.map((b) => [b.seg, b.top]));
       const pieceOf = (v) => Math.max(0, g3.pieces.findIndex((pc) => {
         const r = pc.range?.get(c.el);
@@ -57376,6 +57414,19 @@ function staleTopCache(root, BODY_H) {
   }
   return false;
 }
+function overrunTopCache(root, BODY_H) {
+  if (BODY_H <= 0) return false;
+  let over = 0;
+  for (const p of elements(root)) {
+    if (ln(p) !== "p") continue;
+    const lsa = findChildByLocalName(p, "linesegarray");
+    for (const sg of lsa ? elements(lsa) : []) if (ln(sg) === "lineseg") {
+      over = num2(sg, "vertpos") > BODY_H ? over + 1 : 0;
+      if (over >= 3) return true;
+    }
+  }
+  return false;
+}
 function absoluteTopCache(root, ML, BODY_W) {
   let tot = 0, over = 0;
   for (const p of elements(root)) {
@@ -57393,7 +57444,7 @@ function absoluteTopCache(root, ML, BODY_W) {
 function renderSectionToPages(root, geom, ctxBase, doReflow, reflowMode, masters = [], hasDeletionMarkers = true, lastSection = true) {
   const { PW, PH, ML, MT, BODY_W, BODY_H } = geom;
   if (hasDeletionMarkers) prepareDeletedRanges(root);
-  const degenerate = staleTopCache(root, BODY_H);
+  const degenerate = staleTopCache(root, BODY_H) || overrunTopCache(root, BODY_H);
   if (degenerate) {
     for (const p of elements(root)) if (ln(p) === "p") {
       const lsa = findChildByLocalName(p, "linesegarray");
@@ -57401,6 +57452,19 @@ function renderSectionToPages(root, geom, ctxBase, doReflow, reflowMode, masters
     }
     ctxBase.warnings.push("\uC870\uD310 \uCE90\uC2DC\uAC00 \uB9DE\uC9C0 \uC54A\uC544(\uCD5C\uC0C1\uC704 \uBB38\uB2E8\uC774 \uBAA8\uB450 \uAC19\uC740 \uC790\uB9AC) \uB2E4\uC2DC \uC870\uD310\uD574 \uB80C\uB354");
   }
+  for (const tc2 of Array.from(root.getElementsByTagName("*")).filter((e) => ln(e) === "tc")) {
+    const sub = findChildByLocalName(tc2, "subList");
+    const paras = sub ? elements(sub).filter((q) => ln(q) === "p") : [];
+    const vs = [];
+    for (const q of paras) {
+      const lsa = findChildByLocalName(q, "linesegarray");
+      for (const sg of lsa ? elements(lsa) : []) if (ln(sg) === "lineseg") vs.push(num2(sg, "vertpos"));
+    }
+    if (vs.length >= 3 && (vs.every((v) => v === vs[0]) || vs.some((v, i) => i >= 2 && v === vs[i - 1] && v === vs[i - 2]))) for (const q of paras) {
+      const lsa = findChildByLocalName(q, "linesegarray");
+      if (lsa) q.removeChild(lsa);
+    }
+  }
   const absolute = !degenerate && absoluteTopCache(root, ML, BODY_W);
   if (absolute) {
     for (const lsa of Array.from(root.getElementsByTagName("*")).filter((e) => ln(e) === "linesegarray")) lsa.parentNode?.removeChild(lsa);
@@ -57434,7 +57498,7 @@ function renderSectionToPages(root, geom, ctxBase, doReflow, reflowMode, masters
     let prevH = Number.NEGATIVE_INFINITY;
     let prevEnd = Number.NEGATIVE_INFINITY;
     let cur = 0, col = 0;
-    let pendingEnd = 0, pendingEndY = 0;
+    let pendingEnd = 0, pendingEndY = 0, pendingTopY = -1;
     let lastEnd = 0, lastPage = -1;
     let lastSA = 0;
     let prevSeg;
@@ -57459,7 +57523,7 @@ function renderSectionToPages(root, geom, ctxBase, doReflow, reflowMode, masters
           }
           if (col > 0) s.setAttribute("horzpos", String(h + Math.round(col * (colW + colGap))));
         } else {
-          if (paraFirst && pendingEnd > cur) {
+          if (paraFirst && pendingEnd > cur && !(v > prevV && v < pendingTopY - 100)) {
             cur = v < pendingEndY - 100 ? pendingEnd + 1 : pendingEnd;
             pendingEnd = 0;
           } else {
@@ -57531,9 +57595,10 @@ function renderSectionToPages(root, geom, ctxBase, doReflow, reflowMode, masters
             lastPage = pagesOf[pagesOf.length - 1];
             continue;
           }
-          if (span2.pages > 0 && (pagesOf[0] ?? cur) + span2.pages > pendingEnd) {
+          if (span2.pages > 0 && o.el.getAttribute("pageBreak") !== "NONE" && (pagesOf[0] ?? cur) + span2.pages > pendingEnd) {
             pendingEnd = (pagesOf[0] ?? cur) + span2.pages;
             pendingEndY = span2.endY - MT;
+            pendingTopY = below && voP > 0 && ty - MT < BODY_H ? ty - MT : -1;
           }
         }
       }
@@ -57768,7 +57833,7 @@ function renderSectionRoots(sections, input) {
     return pm.objs.length > 0 || pm.chars.some((c) => c.ch.trim() !== "" || !!c.ctrlText);
   });
   let lastIdx = sections.length - 1;
-  while (lastIdx > 0 && !hasContent(sections[lastIdx].root)) lastIdx--;
+  while (lastIdx > 0 && !hasContent(sections[lastIdx].root) && !sections[lastIdx].masterPages?.length) lastIdx--;
   for (const [k, { root, index, masterPages, hasDeletionMarkers }] of sections.entries()) {
     if (k > lastIdx) break;
     const geom = readSectionGeom(root);
@@ -58920,7 +58985,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-Y75JNXPZ.js");
+    const { parseImageDocument } = await import("./image-ocr-5YWVBRY6.js");
     const { blocks, warnings, lines } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -58984,7 +59049,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-4NF5BGSC.js");
+    const mod = await import("./parser-G5QTRKII.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {