kordoc 4.24.3 → 4.24.4

package/dist/index.js

@@ -2,7 +2,7 @@ import {
   blocksToPages,
   bmpToPng,
   inlineImagesIntoMarkdown
-} from "./chunk-W6O7KBOD.js";
+} from "./chunk-MC4DPT4I.js";
 import {
   parsePageRange
 } from "./chunk-V4BH5SRQ.js";
@@ -43,7 +43,7 @@ import {
   validateOcrOptions,
   visibleText,
   wrapScript
-} from "./chunk-EEXIFCXC.js";
+} from "./chunk-VBY3JQL2.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT,
@@ -61,7 +61,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-I2QETSZN.js";
+} from "./chunk-FK3JPTHI.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-JQJC5HWK.js");
+  const { parseImageDocument } = await import("./image-ocr-THA5BASF.js");
   let inserted = false, unreadable = 0;
   let failed = null;
   const walk = async (list, depth) => {
@@ -44217,7 +44217,7 @@ function hwpFaceToCssStack(face) {
   const head = FONT_ALIASES[trimmed] ?? cssQuote(trimmed);
   return `${head},${generic}`;
 }
-function collectHangulFonts(root, lang = "HANGUL", hftIds) {
+function collectHangulFonts(root, lang = "HANGUL", hftIds, substKoPub = false) {
   const map = /* @__PURE__ */ new Map();
   const findFaces = (el, depth) => {
     if (depth > 24) return null;
@@ -44253,7 +44253,7 @@ function collectHangulFonts(root, lang = "HANGUL", hftIds) {
     const id = e.getAttribute("id");
     let face = e.getAttribute("face");
     let type = e.getAttribute("type");
-    if (face && type === "TTF" && /^한양/.test(face)) {
+    if (face && type === "TTF" && (/^한양/.test(face) || substKoPub && /^KoPub/.test(face))) {
       const sub = Array.from(e.childNodes).find((c) => c.nodeType === 1 && (c.tagName || "").replace(/^[^:]+:/, "") === "substFont");
       const sf = sub?.getAttribute("face");
       if (sf) {
@@ -44344,10 +44344,11 @@ function parseRenderStyles(headXml, opts) {
   const hv = Number.parseFloat(root.getAttribute("version") ?? "");
   const oldUnit = Number.isFinite(hv) && hv < 1.4;
   const hftHangul = /* @__PURE__ */ new Set(), hftLatin = /* @__PURE__ */ new Set();
-  const hangulFonts = collectHangulFonts(root, "HANGUL", hftHangul);
-  const latinFonts = collectHangulFonts(root, "LATIN", hftLatin);
+  const sk = !!opts?.substKoPub;
+  const hangulFonts = collectHangulFonts(root, "HANGUL", hftHangul, sk);
+  const latinFonts = collectHangulFonts(root, "LATIN", hftLatin, sk);
   const hftHanja = /* @__PURE__ */ new Set(), hftJapanese = /* @__PURE__ */ new Set(), hftSymbol = /* @__PURE__ */ new Set();
-  const langFonts = { hanja: [collectHangulFonts(root, "HANJA", hftHanja), hftHanja], japanese: [collectHangulFonts(root, "JAPANESE", hftJapanese), hftJapanese], symbol: [collectHangulFonts(root, "SYMBOL", hftSymbol), hftSymbol] };
+  const langFonts = { hanja: [collectHangulFonts(root, "HANJA", hftHanja, sk), hftHanja], japanese: [collectHangulFonts(root, "JAPANESE", hftJapanese, sk), hftJapanese], symbol: [collectHangulFonts(root, "SYMBOL", hftSymbol, sk), hftSymbol] };
   const autoTabPr = /* @__PURE__ */ new Set();
   const paraTabRef = /* @__PURE__ */ new Map();
   const paraBullet = /* @__PURE__ */ new Map();
@@ -52517,7 +52518,8 @@ function solveRowHeights(cells, rowCount, maxTotal = Number.POSITIVE_INFINITY, s
   }
   const last = rowCount - 1, over = total - maxTotal;
   const lastCells = cells.filter((c) => c.rowAddr + c.rowSpan - 1 === last);
-  if (shrinkLast && over > 0 && last >= 0 && lastCells.length && lastCells.every((c) => c.rowSpan === 1 && c.contentH != null && c.contentH <= h[last] - over)) h[last] -= over;
+  const spanH = (c) => h.slice(c.rowAddr, c.rowAddr + c.rowSpan).reduce((t, x) => t + x, 0);
+  if (shrinkLast && over > 0 && last >= 0 && lastCells.length && lastCells.every((c) => c.contentH != null && c.contentH <= (c.rowSpan === 1 ? h[last] : spanH(c)) - over)) h[last] -= over;
   return h;
 }
 
@@ -52785,7 +52787,7 @@ function cellContentExtent(cell2, memo) {
       const pushed = baseV - (Math.max(0, vo) + num2(om, "top") + o.height + num2(om, "bottom"));
       const anchor = pushed >= (paraStart > 0 ? paraStart - 100 : -100) ? pushed : baseV - (memo?.paraGeom?.get(p.getAttribute("paraPrIDRef") ?? "")?.spaceBefore ?? 0);
       const oh = o.tag === "tbl" ? Math.max(o.height, measureTableHeight(o.el, memo)) : o.height;
-      ext = Math.max(ext, anchor + num2(om, "top") + Math.max(0, vo) + oh);
+      ext = Math.max(ext, anchor + num2(om, "top") + Math.max(0, vo) + oh + num2(om, "bottom"));
     }
   }
   memo?.cell.set(cell2.el, ext);
@@ -52852,7 +52854,7 @@ function measureTableRowHeights(tbl2, memo) {
   const rows = solveRowHeights(
     // 여러 행 칸은 내용(+ 칸 여백)이 저장 높이보다 크면 내용 높이로 — 솔버가 모자란 만큼 그 칸 마지막 행을 키운다
     // (같은 판 PDF: 두 줄 든 2행 병합 칸 내용 2882 > 저장 2564 → 둘째 행이 318 자라고 첫 행은 그대로)
-    cells.map((c) => ({ rowAddr: c.ra, rowSpan: c.rs, height: (c.rs > 1 ? Math.max(c.h, cellContentExtent(c, memo) + c.marginT + c.marginB) : c.h) - sp * (c.rs - 1), contentH: c.rs === 1 ? cellContentExtent(c, memo) + c.marginT + c.marginB : void 0 })),
+    cells.map((c) => ({ rowAddr: c.ra, rowSpan: c.rs, height: (c.rs > 1 ? Math.max(c.h, cellContentExtent(c, memo) + c.marginT + c.marginB) : c.h) - sp * (c.rs - 1), contentH: cellContentExtent(c, memo) + c.marginT + c.marginB })),
     nRows,
     num2(findChildByLocalName(tbl2, "sz"), "height") - (sp > 0 ? sp * (nRows + 1) : 0) || Number.POSITIVE_INFINITY,
     !inCell(tbl2) && !cells.some((c) => c.ra + c.rs === nRows && splitSaved(c))
@@ -54401,6 +54403,53 @@ function flowEndNotes(root, bodyH) {
   });
 }
 
+// src/render/font-env.ts
+function hasKoPubSubst(headXml) {
+  return /<hh:font\b[^>]*\bface="KoPub[^"]*"[^>]*\btype="TTF"[^>]*>\s*<hh:substFont\b/.test(headXml);
+}
+function lineStarts(root) {
+  const out = [];
+  for (const p of Array.from(root.getElementsByTagName("hp:p"))) {
+    const lsa = elements(p).find((e) => ln(e) === "linesegarray");
+    const segs = lsa ? elements(lsa).filter((e) => ln(e) === "lineseg") : [];
+    out.push(segs.length >= 2 ? segs.map((s) => s.getAttribute("textpos") ?? "0").join(",") : null);
+  }
+  return out;
+}
+function matchCount(xml, saved, styles) {
+  const root = createXmlParser().parseFromString(xml, "text/xml").documentElement;
+  const pagePr = root.getElementsByTagName("hp:pagePr")[0];
+  if (!pagePr) return 0;
+  const margin = pagePr.getElementsByTagName("hp:margin")[0];
+  let PW = num2(pagePr, "width", 59528), PH = num2(pagePr, "height", 84188);
+  if (pagePr.getAttribute("landscape") === "NARROWLY" && PW < PH) [PW, PH] = [PH, PW];
+  const gutterTB = pagePr.getAttribute("gutterType") === "TOP_BOTTOM", gutter = num2(margin ?? null, "gutter", 0);
+  const BODY_W = PW - num2(margin ?? null, "left", 8504) - num2(margin ?? null, "right", 8504) - (gutterTB ? 0 : gutter);
+  const BODY_H = PH - num2(margin ?? null, "top", 5668) - num2(margin ?? null, "header", 0) - num2(margin ?? null, "bottom", 4252) - num2(margin ?? null, "footer", 0) - (gutterTB ? gutter : 0);
+  for (const p of Array.from(root.getElementsByTagName("hp:p"))) {
+    const lsa = elements(p).find((e) => ln(e) === "linesegarray");
+    if (lsa) p.removeChild(lsa);
+  }
+  reflowSection(root, styles, { BODY_W, BODY_H });
+  const got = lineStarts(root);
+  let n = 0;
+  saved.forEach((s, i) => {
+    if (s !== null && s === got[i]) n++;
+  });
+  return n;
+}
+function kopubSubstituted(secXmls, base, alt2) {
+  let a = 0, b = 0, seen = 0;
+  for (const xml of secXmls) {
+    const saved = lineStarts(createXmlParser().parseFromString(xml, "text/xml").documentElement);
+    a += matchCount(xml, saved, base);
+    b += matchCount(xml, saved, alt2);
+    seen += saved.filter((x) => x !== null).length;
+    if (seen >= 60) break;
+  }
+  return b >= a + 5 && b > a * 1.1;
+}
+
 // src/render/scene.ts
 function regionId(type, seq) {
   return `${type}-${String(seq).padStart(6, "0")}`;
@@ -54828,11 +54877,11 @@ function planTableRows(tbl2, cells, rowH, ty, MT, BODY_H, slack = 0, floatSpan,
         continue;
       }
     }
-    if (splittable && y + h > pageBottom) {
+    if (splittable && y + h + outB > pageBottom) {
       const from = /* @__PURE__ */ new Map();
       const lines = new Map(rowCells.map((c) => [c.el, cellLines(c)]));
-      while (y + h > pageBottom && BODY_H > rowY[hdrRows]) {
-        const avail = pageBottom - y;
+      while (y + h + outB > pageBottom && BODY_H > rowY[hdrRows]) {
+        const avail = pageBottom - y - outB;
         let rem = 0, placed = false, allFit = true, blocked = false;
         const cuts = [];
         for (const c of rowCells) {
@@ -55441,30 +55490,53 @@ function planLines(m, styles, numText = "") {
       }
     }
     const stretch = align === "JUSTIFY" || align === "DISTRIBUTE" || align === "DISTRIBUTE_SPACE";
-    let spaceAdd = 0, spaceFrom = fixedTo;
+    let spaceAdd = 0, spaceFrom = fixedTo, charAdd = 0;
+    const squeeze = (spFrom) => {
+      const short = nat.text - lastSp - avail;
+      let nSp = 0, cap = 0, nG = 0;
+      for (let k = spFrom; k < textEnd; k++) {
+        const c = m.chars[k];
+        if (c.ch === " ") {
+          nSp++;
+          cap += charW(c, styles, m.chars[k - 1]) * 0.2;
+        } else if (c.ch && !c.tab) nG++;
+      }
+      if (nSp > 0) {
+        spaceAdd = -Math.min(short, cap) / nSp;
+        spaceFrom = spFrom;
+      }
+      if (short > cap && nG > 0) charAdd = -(short - (nSp > 0 ? cap : 0)) / nG;
+    };
     if (nat.text > 0 && stretch && (!isLast || align === "DISTRIBUTE" || align === "DISTRIBUTE_SPACE")) {
       let nSp = 0, spFrom = fixedTo;
-      while (spFrom < textEnd && (m.chars[spFrom].ch === " " || m.chars[spFrom].ch === "")) spFrom++;
+      while (spFrom < textEnd && (m.chars[spFrom].ch === " " || m.chars[spFrom].ch === "") && !m.chars[spFrom].ctrl) spFrom++;
       for (let k = spFrom; k < textEnd; k++) if (m.chars[k].ch === " ") nSp++;
       if (nSp > 0 && align !== "DISTRIBUTE") {
         spaceAdd = (avail - nat.text + lastSp) / nSp;
         spaceFrom = spFrom;
+        if (spaceAdd < 0) squeeze(spFrom);
       } else scale2 = avail > 0 ? avail / nat.text : 1;
     } else if (nat.text + nat.obj > 0 && (isLast || !stretch)) {
       const w = nat.text + nat.obj - lastSp;
       if (align === "CENTER") xoff = Math.max(0, (lineW - w) / 2);
       else if (align === "RIGHT") xoff = Math.max(0, lineW - w);
     }
+    if (spaceAdd === 0 && charAdd === 0 && scale2 === 1 && nat.text - lastSp > avail) {
+      let spFrom = fixedTo;
+      while (spFrom < textEnd && (m.chars[spFrom].ch === " " || m.chars[spFrom].ch === "") && !m.chars[spFrom].ctrl) spFrom++;
+      squeeze(spFrom);
+    }
     if (!Number.isFinite(scale2) || scale2 <= 0) scale2 = 1;
     scale2 = Math.min(4, Math.max(0.25, scale2));
-    plans.push({ seg, xoff: xoff + lead, scale: scale2, spaceAdd, spFrom: spaceFrom, start, end, tabs, fixedTo, textEnd, ...i === 0 && bu ? { bullet: { ch: bu.ch, w: bulletW, prId: bu.charPrId ?? firstPr } } : {} });
+    plans.push({ seg, xoff: xoff + lead, scale: scale2, spaceAdd, spFrom: spaceFrom, ...charAdd ? { charAdd } : {}, start, end, tabs, fixedTo, textEnd, ...i === 0 && bu ? { bullet: { ch: bu.ch, w: bulletW, prId: bu.charPrId ?? firstPr } } : {} });
   }
   return plans;
 }
 function advanceTo(m, styles, plan, upto) {
   let x = 0;
   for (let i = plan.start; i < upto && i < m.chars.length; i++) {
-    x += m.chars[i].tab ? plan.tabs.get(i) ?? 0 : charW(m.chars[i], styles, m.chars[i - 1]) * (i < plan.fixedTo ? 1 : plan.scale) + (i >= plan.spFrom && i < plan.textEnd && m.chars[i].ch === " " ? plan.spaceAdd : 0);
+    const c = m.chars[i], inStretch = i >= plan.spFrom && i < plan.textEnd;
+    x += c.tab ? plan.tabs.get(i) ?? 0 : charW(c, styles, m.chars[i - 1]) * (i < plan.fixedTo ? 1 : plan.scale) + (inStretch && c.ch === " " ? plan.spaceAdd : 0) + (inStretch && plan.charAdd && c.ch && c.ch !== " " ? plan.charAdd : 0);
   }
   for (const o of m.objs) if (o.inline && o.index >= plan.start && o.index < upto) x += o.omL + o.width + o.omR;
   return x;
@@ -55582,7 +55654,7 @@ function drawPara(p, ox, oy, areaW, ctx, depth, segPages) {
         return n;
       };
       const renderSeg = (text, cx, hit) => {
-        const sw = measure(text) * scale2 + (plan.spaceAdd && i >= plan.fixedTo ? plan.spaceAdd * spacesIn(text) : 0);
+        const sw = measure(text) * scale2 + (plan.spaceAdd && i >= plan.fixedTo ? plan.spaceAdd * spacesIn(text) : 0) + (plan.charAdd && i >= plan.fixedTo ? plan.charAdd * [...text].filter((ch) => ch !== " ").length : 0);
         if (st.shade) emit(ctx, `<rect x="${pt2(cx)}" y="${pt2(glyphY(st.height) - Math.round(st.height * 0.85))}" width="${pt2(sw)}" height="${pt2(st.height)}" fill="${escapeXml7(st.shade)}"/>`);
         if (hit) {
           emit(ctx, `<rect x="${pt2(cx)}" y="${pt2(oy + seg.vertpos)}" width="${pt2(sw)}" height="${pt2(seg.textheight)}" fill="#ffd54f" fill-opacity="0.45"/>`);
@@ -55664,11 +55736,15 @@ 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 anchorK = -1;
+      plans.forEach((pl, k) => {
+        if (pl.start <= o.index) {
+          anchorLineV = pl.seg.vertpos;
+          anchorLineH = pl.seg.textheight;
+          anchorSeg = pl.seg;
+          anchorK = k;
+        }
+      });
       let objBaseV = baseV;
       const pTop = depth === 0 ? ctx.paraTops?.get(p) : void 0;
       const wrapA = o.el.getAttribute("textWrap") ?? "TOP_AND_BOTTOM";
@@ -55687,6 +55763,11 @@ function drawPara(p, ox, oy, areaW, ctx, depth, segPages) {
       let stay = tbFloats < 2 && !m.objs.some((q) => q.inline) && !m.chars.some((c) => c.ch !== "" && !c.nb && c.ch.trim() !== "");
       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;
+      let carried;
+      if (segPages && anchorK > 0 && segPages[anchorK] !== segPages[0] && segPages[anchorK - 1] === segPages[0] && wrapA === "TOP_AND_BOTTOM" && (findChildByLocalName(o.el, "pos")?.getAttribute("vertRelTo") ?? "PARA") === "PARA") {
+        const s0 = plans[anchorK - 1].seg;
+        carried = s0.vertpos + s0.textheight + (s0.spacing ?? 0);
+      }
       const firstLine = anchorLineV === void 0 || anchorLineV === baseV;
       let { x, y } = anchorObject(
         o,
@@ -55703,6 +55784,10 @@ function drawPara(p, ox, oy, areaW, ctx, depth, segPages) {
         // 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"
       );
+      if (carried !== void 0) {
+        const posC = findChildByLocalName(o.el, "pos"), rawC = num2(posC, "vertOffset");
+        y = oy + carried + num2(findChildByLocalName(o.el, "outMargin"), "top") + Math.max(0, rawC >= 2 ** 31 ? rawC - 2 ** 32 : rawC);
+      }
       let stackTop = depth === 0 ? pTop ?? 0 : cellParaTop(p, ctx.styles.paraGeom);
       if (tbFloats >= 2 && wrapA === "TOP_AND_BOTTOM" && stackTop !== void 0) {
         if (!tbStack) {
@@ -56611,7 +56696,8 @@ async function renderHwpxInternal(input, options) {
   const verFile = zip.file("version.xml");
   const verXml = verFile ? await readZipEntry(verFile, "text") : void 0;
   const appMajor = hwpxAppMajor(verXml), appBuild = hwpxAppBuild(verXml);
-  const styles = headFile ? parseRenderStyles(await readZipEntry(headFile, "text"), { appMajor, appBuild }) : { charPr: /* @__PURE__ */ new Map(), paraAlign: /* @__PURE__ */ new Map(), paraGeom: /* @__PURE__ */ new Map(), borderFill: /* @__PURE__ */ new Map() };
+  const headXml = headFile ? await readZipEntry(headFile, "text") : void 0;
+  let styles = headXml ? parseRenderStyles(headXml, { appMajor, appBuild }) : { charPr: /* @__PURE__ */ new Map(), paraAlign: /* @__PURE__ */ new Map(), paraGeom: /* @__PURE__ */ new Map(), borderFill: /* @__PURE__ */ new Map() };
   if (!headFile) warnings.push("header.xml \uC5C6\uC74C \u2014 \uAE30\uBCF8 \uC2A4\uD0C0\uC77C\uB85C \uB80C\uB354");
   const secXmls = [];
   for (const f of secFiles) {
@@ -56619,6 +56705,10 @@ async function renderHwpxInternal(input, options) {
     if (xml.length > MAX_DECOMPRESS_SIZE) throw new KordocError("\uC139\uC158 XML\uC774 \uD5C8\uC6A9 \uD06C\uAE30\uB97C \uCD08\uACFC");
     secXmls.push(xml);
   }
+  if (headXml && hasKoPubSubst(headXml)) {
+    const alt2 = parseRenderStyles(headXml, { appMajor, appBuild, substKoPub: true });
+    if (kopubSubstituted(secXmls, styles, alt2)) styles = alt2;
+  }
   const masterPages = /* @__PURE__ */ new Map();
   const masterXmls = [];
   for (const file of zip.file(/Contents\/masterpage[^/]*\.xml$/i)) {
@@ -57729,7 +57819,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-JQJC5HWK.js");
+    const { parseImageDocument } = await import("./image-ocr-THA5BASF.js");
     const { blocks, warnings, lines } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -57793,7 +57883,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-ZBP5GOQK.js");
+    const mod = await import("./parser-MNNJSFEZ.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {