kordoc 4.24.2 → 4.24.3

package/dist/index.js

@@ -2,7 +2,7 @@ import {
   blocksToPages,
   bmpToPng,
   inlineImagesIntoMarkdown
-} from "./chunk-3HN65YAL.js";
+} from "./chunk-W6O7KBOD.js";
 import {
   parsePageRange
 } from "./chunk-V4BH5SRQ.js";
@@ -43,7 +43,7 @@ import {
   validateOcrOptions,
   visibleText,
   wrapScript
-} from "./chunk-TW3V6FSH.js";
+} from "./chunk-EEXIFCXC.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT,
@@ -61,7 +61,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-WJAFE3Z6.js";
+} from "./chunk-I2QETSZN.js";
 
 // src/plain-markdown.ts
 function toPlainMarkdown(md2) {
@@ -24620,7 +24620,7 @@ 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-J54Q3YWP.js");
+  const { parseImageDocument } = await import("./image-ocr-JQJC5HWK.js");
   let inserted = false, unreadable = 0;
   let failed = null;
   const walk = async (list, depth) => {
@@ -52495,6 +52495,12 @@ function solveRowHeights(cells, rowCount, maxTotal = Number.POSITIVE_INFINITY, s
     for (const r of missing) h[r] = each;
     return true;
   };
+  for (let r = 0; r < rowCount; r++) {
+    if (h[r] !== 0) continue;
+    let need = 0;
+    for (const { c, rows } of spans) if (rows[rows.length - 1] === r && rows.slice(0, -1).every((q) => h[q] > 0)) need = Math.max(need, c.height - rows.slice(0, -1).reduce((t, q) => t + h[q], 0));
+    if (need > 0) h[r] = need;
+  }
   for (let again = true; again; ) {
     again = false;
     for (let k = 0; k < spans.length; k++) if (!done.has(k) && solve(k, true)) again = true;
@@ -52973,6 +52979,7 @@ function splitTableRows(tbl2, top, lim, bodyH, styles, notes) {
   const fl = () => notes && nN > 0 ? Math.min(floor, bodyH - notes.area((moved ? 0 : notes.n0) + nN, (moved ? 0 : notes.s0) + nS)) : floor;
   const sp = num2(tbl2, "cellSpacing");
   const cells = mode === "CELL" ? collectCells(tbl2) : [];
+  const tableCells = mode === "TABLE" ? collectCells(tbl2) : [];
   const record = mode === "CELL" ? /* @__PURE__ */ new Map() : void 0;
   const carries = /* @__PURE__ */ new Map();
   for (let r = 0; r < rows.length; r++) {
@@ -53001,7 +53008,20 @@ function splitTableRows(tbl2, top, lim, bodyH, styles, notes) {
       y += h;
       continue;
     }
-    if (y + h + outB > fl() && !(moved && y === outT + headH)) {
+    let need = h;
+    if (!tableCells.some((c) => c.ra < r && c.ra + c.rs > r)) {
+      let end = r + 1;
+      for (let grown = true; grown; ) {
+        grown = false;
+        for (const c of tableCells) if (c.ra >= r && c.ra < end && c.ra + c.rs > end) {
+          end = c.ra + c.rs;
+          grown = true;
+        }
+      }
+      const groupH = rows.slice(r, end).reduce((a, b) => a + b, 0);
+      if (end > r + 1 && groupH <= bodyH - outT - outB - headH) need = groupH;
+    }
+    if (y + need + outB > fl() && !(moved && y === outT + headH)) {
       if (!moved && r === 0) splitWhole = true;
       y = outT + headH;
       floor = bodyH;
@@ -54609,6 +54629,26 @@ function cellLineBoxes(c) {
 function cellLines(c) {
   return cellLineBoxes(c).map((b) => [b.top, b.bottom]);
 }
+function spanPartRows(cells, rowY) {
+  const out = /* @__PURE__ */ new Set();
+  for (const c of cells) {
+    if (c.rs < 2) continue;
+    const bs = cellLineBoxes(c);
+    const parts = bs.length ? bs[bs.length - 1].part : 0;
+    let s = c.ra;
+    for (let k = 0; k < parts; k++) {
+      const inK = bs.filter((b) => b.part === k);
+      if (!inK.length) break;
+      const ext = Math.max(...inK.map((b) => b.bottom)) - inK[0].origin + c.marginT + c.marginB;
+      let r = s + 1;
+      while (r < c.ra + c.rs && rowY[r] - rowY[s] < ext - 100) r++;
+      if (r >= c.ra + c.rs) break;
+      out.add(r);
+      s = r;
+    }
+  }
+  return out;
+}
 function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan, nested = false, contMT = MT) {
   if (nested) slack = 0;
   const nRows = rowH.length;
@@ -54641,9 +54681,15 @@ 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;
+  const spanStarts = cellSplit ? spanPartRows(cells, rowY) : /* @__PURE__ */ new Set();
   for (let r = 0; r < nRows; r++) {
     let h = rowH[r];
     const pieces = [];
+    if (!first && r >= hdrRows && spanStarts.has(r)) {
+      page++;
+      y = contTop;
+      first = true;
+    }
     const rowCells = cells.filter((c) => c.ra === r && c.rs === 1);
     const splittable = cellSplit && r >= hdrRows && rowCells.length > 0;
     const stack = splittable && rowCells.length === 1 ? stackedFloats(rowCells[0]) : [];
@@ -54858,6 +54904,21 @@ function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan,
       page++;
       y = contTop;
       first = true;
+    } else if (!first && tbl2.getAttribute("pageBreak") === "TABLE" && r >= hdrRows && !cells.some((c) => c.ra < r && c.ra + c.rs > r)) {
+      let end = r + 1;
+      for (let grown = true; grown; ) {
+        grown = false;
+        for (const c of cells) if (c.ra >= r && c.ra < end && c.ra + c.rs > end) {
+          end = c.ra + c.rs;
+          grown = true;
+        }
+      }
+      const groupH = rowY[Math.min(end, nRows)] - rowY[r];
+      if (end > r + 1 && y + groupH + outB > pageBottom && groupH <= pageBottom - contTop) {
+        page++;
+        y = contTop;
+        first = true;
+      }
     }
     out.push([{ page, y, h }]);
     y += h;
@@ -54900,12 +54961,27 @@ function renderingMatrix(el) {
   const info = findChildByLocalName(el, "renderingInfo");
   if (!info) return null;
   let m = null;
+  let pendingS = null;
+  const apply = (n) => {
+    m = m ? mulAffine(m, n) : n;
+  };
   for (const c of elements(info)) {
     const tag = ln(c);
     if (tag !== "transMatrix" && tag !== "scaMatrix" && tag !== "rotMatrix") continue;
     const n = [num2(c, "e1", 1), num2(c, "e4"), num2(c, "e2"), num2(c, "e5", 1), num2(c, "e3"), num2(c, "e6")];
-    m = m ? mulAffine(m, n) : n;
+    if (tag === "scaMatrix") {
+      if (pendingS) apply(pendingS);
+      pendingS = n;
+      continue;
+    }
+    if (tag === "rotMatrix" && pendingS) {
+      apply(mulAffine(n, pendingS));
+      pendingS = null;
+      continue;
+    }
+    apply(n);
   }
+  if (pendingS) apply(pendingS);
   return m;
 }
 function objectGeometry(el, x, y, w, h, boxW = w, boxH = h) {
@@ -55012,7 +55088,7 @@ function linePushedBy(o, lineV, paraTop) {
   const raw = num2(pos, "vertOffset"), vo = (pos?.getAttribute("vertRelTo") ?? "PARA") === "PARA" ? Math.max(0, raw >= 2 ** 31 ? raw - 2 ** 32 : raw) : 0;
   return Math.abs(lineV - (vo + num2(om, "top") + o.height + num2(om, "bottom")) - paraTop) <= 10;
 }
-function anchorObject(o, ox, oy, baseV, areaW, ctx, anchorLineV, stayLine = false, paraStartV, lineH, lineSB = 0) {
+function anchorObject(o, ox, oy, baseV, areaW, ctx, anchorLineV, stayLine = false, paraStartV, lineH, lineSB = 0, anchorAt = "end") {
   const { PW, PH, ML, MT, BODY_W, BODY_H } = ctx.geom;
   const pos = findChildByLocalName(o.el, "pos");
   const om = findChildByLocalName(o.el, "outMargin");
@@ -55040,8 +55116,9 @@ function anchorObject(o, ox, oy, baseV, areaW, ctx, anchorLineV, stayLine = fals
     y = oy + anchor + omT + vo;
   } else {
     const origin = vrel === "PARA" && wrap === "SQUARE" ? paraStartV ?? baseV : baseV;
-    const later = vrel === "PARA" && wrap === "SQUARE" && anchorLineV !== void 0 && anchorLineV > baseV;
-    y = later ? oy + anchorLineV + vo + omT : oy + Math.max(origin + vo, wrap === "SQUARE" && anchorLineV !== void 0 ? anchorLineV : -Infinity) + omT;
+    const later = vrel === "PARA" && wrap === "SQUARE" && anchorAt === "end" && anchorLineV !== void 0 && anchorLineV > baseV;
+    const midLater = vrel === "PARA" && wrap === "SQUARE" && anchorAt === "top" && anchorLineV !== void 0 && anchorLineV > baseV;
+    y = later ? oy + anchorLineV + vo + omT : oy + Math.max(origin + vo, wrap === "SQUARE" && !midLater && anchorLineV !== void 0 ? anchorLineV : -Infinity) + omT;
   }
   let x;
   const omL = num2(om, "left"), omR = num2(om, "right");
@@ -55571,7 +55648,8 @@ function drawPara(p, ox, oy, areaW, ctx, depth, segPages) {
       const plan = plans[planIdx];
       if (segPages && segPages[planIdx] !== void 0) ctx.page = segPages[planIdx];
       const x = ox + plan.seg.horzpos + plan.xoff + advanceTo(m, ctx.styles, plan, o.index) + o.omL;
-      const eqAsc = o.tag === "equation" && Number(o.el.getAttribute("baseLine")) > 0 ? Math.round(o.height * Number(o.el.getAttribute("baseLine")) / 100) : o.tag === "pic" ? Math.round(o.height * 0.85) : o.height;
+      const shadowed = o.tag === "pic" && !!findChildByLocalName(findChildByLocalName(o.el, "effects") ?? o.el, "shadow") && plan.seg.textheight > o.height && Math.abs(plan.seg.baseline - Math.round(plan.seg.textheight * 0.85)) <= 2;
+      const eqAsc = o.tag === "equation" && Number(o.el.getAttribute("baseLine")) > 0 ? Math.round(o.height * Number(o.el.getAttribute("baseLine")) / 100) : shadowed ? Math.round(plan.seg.textheight * 0.85) : o.tag === "pic" ? Math.round(o.height * 0.85) : o.height;
       const iom = findChildByLocalName(o.el, "outMargin"), iomT = num2(iom, "top"), iomB = num2(iom, "bottom");
       const yTop = oy + plan.seg.vertpos + Math.max(iomT, plan.seg.baseline - eqAsc - iomB - captionExtra(o.el));
       drawObject(o, x, yTop, baseV, areaW, ctx, depth);
@@ -55585,9 +55663,11 @@ function drawPara(p, ox, oy, areaW, ctx, depth, segPages) {
       }
       let anchorLineV;
       let anchorLineH = 0;
+      let anchorSeg;
       for (const pl of plans) if (pl.start <= o.index) {
         anchorLineV = pl.seg.vertpos;
         anchorLineH = pl.seg.textheight;
+        anchorSeg = pl.seg;
       }
       let objBaseV = baseV;
       const pTop = depth === 0 ? ctx.paraTops?.get(p) : void 0;
@@ -55598,6 +55678,9 @@ function drawPara(p, ox, oy, areaW, ctx, depth, segPages) {
         if (baseV > start + 100) {
           objBaseV = start;
           if (anchorLineV === baseV) anchorLineV = start;
+        } else if (depth > 0 && start > top0 && Math.abs(baseV - start) <= 100) {
+          objBaseV = top0;
+          if (anchorLineV === baseV) anchorLineV = top0;
         }
       }
       const tbFloats = m.objs.filter((q) => !q.inline && (q.el.getAttribute("textWrap") ?? "TOP_AND_BOTTOM") === "TOP_AND_BOTTOM" && (findChildByLocalName(q.el, "pos")?.getAttribute("vertRelTo") ?? "PARA") === "PARA").length;
@@ -55605,7 +55688,21 @@ function drawPara(p, ox, oy, areaW, ctx, depth, segPages) {
       const sbA = m.paraPrId != null ? ctx.styles.paraGeom.get(m.paraPrId)?.spaceBefore ?? 0 : 0;
       if (stay && wrapA === "TOP_AND_BOTTOM" && linePushedBy(o, anchorLineV ?? baseV, pTop ?? (depth > 0 ? cellParaTop(p, ctx.styles.paraGeom) : void 0))) stay = false;
       const firstLine = anchorLineV === void 0 || anchorLineV === baseV;
-      let { x, y } = anchorObject(o, ax, oy, objBaseV, aw, ctx, anchorLineV, stay ? sbA : false, (firstLine || wrapA === "SQUARE") && pTop !== void 0 && pTop < baseV ? pTop : void 0, anchorLineH, firstLine ? sbA : 0);
+      let { x, y } = anchorObject(
+        o,
+        ax,
+        oy,
+        objBaseV,
+        aw,
+        ctx,
+        anchorLineV,
+        stay ? sbA : false,
+        (firstLine || wrapA === "SQUARE") && pTop !== void 0 && pTop < baseV ? pTop : void 0,
+        anchorLineH,
+        firstLine ? sbA : 0,
+        // 문단 중간 글자 앵커인데 첫 줄부터 앵커 줄만큼 좁게 저장됐으면 개체는 문단 위에 있다(같은 판 한컴 PDF: 둘째 줄 앵커 사진, 네 줄 모두 34602)
+        m.chars.slice(o.index).some((c) => c.ch.trim() !== "") && !!anchorSeg && plans.length > 0 && plans[0].seg !== anchorSeg && Math.abs(plans[0].seg.horzsize - anchorSeg.horzsize) <= 100 && anchorSeg.horzpos + anchorSeg.horzsize < aw - 1e3 ? "top" : m.chars.slice(o.index).some((c) => c.ch.trim() !== "") ? "mid" : "end"
+      );
       let stackTop = depth === 0 ? pTop ?? 0 : cellParaTop(p, ctx.styles.paraGeom);
       if (tbFloats >= 2 && wrapA === "TOP_AND_BOTTOM" && stackTop !== void 0) {
         if (!tbStack) {
@@ -55860,10 +55957,28 @@ function drawTable(tbl2, tx, ty, ctx, depth) {
       pcs.forEach((pc, k) => geom.push({ c, page: startPage + pc.page, ...xw(c), y: pc.y, h: pc.h, pieces: k === 0 ? pcs : [] }));
       continue;
     }
+    if (c.rs > 1) {
+      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 partTop = (k) => bs.find((b) => b.part === k);
+        const groups = pages.map((pg, k) => {
+          const on = span2.filter((pc) => pc.page === pg);
+          const y0 = Math.min(...on.map((pc) => pc.y)), y1 = Math.max(...on.map((pc) => pc.y + pc.h));
+          const from = k === 0 ? 0 : partTop(k)?.origin ?? Number.POSITIVE_INFINITY;
+          const to = k === pages.length - 1 ? Number.POSITIVE_INFINITY : partTop(k + 1)?.top ?? Number.POSITIVE_INFINITY;
+          return { page: pg, y: y0, h: y1 - y0, range: /* @__PURE__ */ new Map([[c.el, [from, to]]]) };
+        });
+        groups.forEach((pc, k) => geom.push({ c, page: startPage + pc.page, ...xw(c), y: pc.y, h: pc.h, pieces: k === 0 ? groups : [] }));
+        continue;
+      }
+    }
     const h = c.rs === 1 ? pcs[0].h : rowY[Math.min(c.ra + c.rs, nRows)] - rowY[c.ra];
     geom.push({ c, page: startPage + pcs[0].page, ...xw(c), y: pcs[0].y, h });
   }
-  for (const page of contPages) for (const c of cells) if (c.ra < hdrRows) geom.push({ c, page, ...xw(c), y: ctx.geom.MT + num2(findChildByLocalName(tbl2, "outMargin"), "top") + rowY[c.ra], h: rowY[Math.min(c.ra + c.rs, nRows)] - rowY[c.ra] });
+  const contBase = depth !== 1 && ctx.splitContMT !== void 0 ? ctx.splitContMT : ctx.geom.MT;
+  for (const page of contPages) for (const c of cells) if (c.ra < hdrRows) geom.push({ c, page, ...xw(c), y: contBase + num2(findChildByLocalName(tbl2, "outMargin"), "top") + rowY[c.ra], h: rowY[Math.min(c.ra + c.rs, nRows)] - rowY[c.ra] });
   for (const g3 of geom) {
     ctx.page = g3.page;
     const bf = g3.c.bfId != null ? ctx.styles.borderFill.get(g3.c.bfId) : void 0;
@@ -55908,7 +56023,7 @@ function drawTable(tbl2, tx, ty, ctx, depth) {
         });
         ctx.splitCell = splitHere;
         const contMTSaved = ctx.splitContMT;
-        ctx.splitContMT = ctx.geom.MT + num2(findChildByLocalName(tbl2, "outMargin"), "top") + rowY[hdrRows] + c.marginT;
+        ctx.splitContMT = (depth > 1 && contMTSaved !== void 0 ? contMTSaved : ctx.geom.MT) + num2(findChildByLocalName(tbl2, "outMargin"), "top") + rowY[hdrRows] + c.marginT;
         drawPara(p, g3.x + c.marginL, base, g3.w - c.marginL - c.marginR, ctx, depth + 1, pages);
         ctx.splitContMT = contMTSaved;
         ctx.splitCell = false;
@@ -56555,6 +56670,7 @@ async function renderHwpxInternal(input, options) {
   }
   const roots = [];
   let noCacheSkipped = false;
+  const lastParity = /* @__PURE__ */ new Map();
   for (let si = 0; si < secXmls.length; si++) {
     const secXml = secXmls[si];
     const hasCache = /<(?:[A-Za-z][\w.-]*:)?linesegarray[\s/>]/.test(secXml);
@@ -56571,6 +56687,14 @@ async function renderHwpxInternal(input, options) {
     }
     const secPr = findFirst(root, "secPr");
     const referenced = secPr ? elements(secPr).filter((el) => ln(el) === "masterPage").map((el) => masterPages.get(el.getAttribute("idRef") || "")).filter((el) => !!el) : [];
+    const typeOf = (m) => m.getAttribute("type") || "BOTH";
+    const has3 = (t) => referenced.some((m) => typeOf(m) === t);
+    if (!has3("BOTH")) for (const [t, other] of [["EVEN", "ODD"], ["ODD", "EVEN"]]) {
+      const prev = lastParity.get(t);
+      if (prev && !has3(t) && has3(other)) referenced.unshift(prev);
+    }
+    for (const m of referenced) if (typeOf(m) === "EVEN" || typeOf(m) === "ODD") lastParity.set(typeOf(m), m);
+    if (has3("BOTH")) lastParity.clear();
     roots.push({ root, index: si, masterPages: referenced, hasDeletionMarkers: secXml.includes("deleteBegin") || secXml.includes("deleteEnd") });
   }
   if (roots.length === 0) {
@@ -57605,7 +57729,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-J54Q3YWP.js");
+    const { parseImageDocument } = await import("./image-ocr-JQJC5HWK.js");
     const { blocks, warnings, lines } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -57669,7 +57793,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-OBGWEWSU.js");
+    const mod = await import("./parser-ZBP5GOQK.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {