kordoc 4.7.1 → 4.7.3

package/dist/index.js

@@ -8,14 +8,16 @@ import {
   flattenLayoutTables,
   inlineImagesIntoMarkdown,
   mapPuaText
-} from "./chunk-IMEMTJ76.js";
+} from "./chunk-P663PYWV.js";
 import {
   parsePageRange
 } from "./chunk-GE43BE46.js";
 import {
   HEADING_RATIO_H1,
   HEADING_RATIO_H2,
-  HEADING_RATIO_H3,
+  HEADING_RATIO_H3
+} from "./chunk-KSATSTC5.js";
+import {
   KordocError,
   VERSION,
   classifyError,
@@ -27,7 +29,7 @@ import {
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-IZEUWEEY.js";
+} from "./chunk-46F5RMU7.js";
 
 // src/index.ts
 import { readFile } from "fs/promises";
@@ -393,7 +395,7 @@ function comResultToParseResult(pages, pageCount, warnings) {
     }
   }
   const markdown = lines.join("\n\n");
-  const metadata = { pageCount };
+  const metadata = { pageCount, pageMode: "layout" };
   warnings.push({
     message: "DRM \uBB38\uC11C: \uD55C\uCEF4 COM API\uB85C \uD14D\uC2A4\uD2B8 \uCD94\uCD9C (\uC11C\uC2DD/\uD45C \uC815\uBCF4 \uC81C\uD55C\uC801)",
     code: "DRM_COM_FALLBACK"
@@ -489,7 +491,12 @@ function clampSpan(val, max) {
   return Math.max(1, Math.min(val, max));
 }
 function createSectionShared() {
-  return { numState: /* @__PURE__ */ new Map(), pageText: { headers: [], footers: [] }, track: { deleteDepth: 0, warned: false } };
+  return {
+    numState: /* @__PURE__ */ new Map(),
+    pageText: { headers: [], footers: [] },
+    track: { deleteDepth: 0, warned: false },
+    pageState: { base: 0, allUsable: true }
+  };
 }
 function createXmlParser(warnings) {
   return new DOMParser({
@@ -1480,11 +1487,11 @@ function buildTableWithCellMeta(state, keepAnchoredEmptyCols, keepEmptyParagraph
 function completeTable(newTable, tableStack, blocks, ctx) {
   const parentTable = tableStack.length > 0 ? tableStack.pop() : null;
   if (newTable.rows.length === 0) {
-    if (newTable.caption) blocks.push({ type: "paragraph", text: newTable.caption, pageNumber: ctx.sectionNum });
+    if (newTable.caption) blocks.push({ type: "paragraph", text: newTable.caption, pageNumber: ctx.page });
     return parentTable;
   }
   const ir = buildTableWithCellMeta(newTable, ctx.shared.keepTrailingEmptyCols, ctx.shared.keepEmptyParagraphs);
-  const block = { type: "table", table: ir, pageNumber: ctx.sectionNum };
+  const block = { type: "table", table: ir, pageNumber: ctx.page };
   if (parentTable?.cell) {
     const cell2 = parentTable.cell;
     (cell2.blocks ??= []).push(block);
@@ -1502,6 +1509,133 @@ function completeTable(newTable, tableStack, blocks, ctx) {
   return parentTable;
 }
 
+// src/hwpx/page-boundary.ts
+var MIDPAGE_V = 2e3;
+var localName2 = (el) => {
+  const n = el.nodeName;
+  const i = n.indexOf(":");
+  return i >= 0 ? n.slice(i + 1) : n;
+};
+var childElements = (parent) => {
+  const out = [];
+  for (let c = parent.firstChild; c; c = c.nextSibling) {
+    if (c.nodeType === 1) out.push(c);
+  }
+  return out;
+};
+var numAttr = (el, name) => {
+  const v = el.getAttribute(name);
+  const n = v == null ? NaN : Number(v);
+  return Number.isFinite(n) ? n : 0;
+};
+var findDescendant = (root, local, maxDepth = 8) => {
+  if (maxDepth < 0) return null;
+  for (const c of childElements(root)) {
+    if (localName2(c) === local) return c;
+    const found = findDescendant(c, local, maxDepth - 1);
+    if (found) return found;
+  }
+  return null;
+};
+function linesegsOf(p) {
+  const lsa = childElements(p).find((c) => localName2(c) === "linesegarray");
+  return lsa ? childElements(lsa).filter((s) => localName2(s) === "lineseg") : [];
+}
+function cellFlowResets(subList) {
+  let prevV = Number.NEGATIVE_INFINITY;
+  let resets = 0;
+  for (const p of childElements(subList)) {
+    if (localName2(p) !== "p") continue;
+    for (const s of linesegsOf(p)) {
+      const v = numAttr(s, "vertpos");
+      if (v < prevV) resets++;
+      prevV = v;
+    }
+  }
+  return resets;
+}
+function tableIntraBreaks(tbl2) {
+  const rowMax = /* @__PURE__ */ new Map();
+  for (const tr of childElements(tbl2)) {
+    if (localName2(tr) !== "tr") continue;
+    for (const tc2 of childElements(tr)) {
+      if (localName2(tc2) !== "tc") continue;
+      const addr = childElements(tc2).find((e) => localName2(e) === "cellAddr");
+      const row = addr ? numAttr(addr, "rowAddr") : 0;
+      const sub = childElements(tc2).find((e) => localName2(e) === "subList");
+      if (!sub) continue;
+      const r = cellFlowResets(sub);
+      rowMax.set(row, Math.max(rowMax.get(row) ?? 0, r));
+    }
+  }
+  let sum = 0;
+  for (const v of rowMax.values()) sum += v;
+  return sum;
+}
+function detectHwpxSectionPages(root) {
+  const colPr = findDescendant(root, "colPr");
+  const multiCol = colPr ? numAttr(colPr, "colCount") > 1 : false;
+  const paraPage = /* @__PURE__ */ new Map();
+  let cur = 0;
+  let prevV = Number.NEGATIVE_INFINITY;
+  let prevH = Number.NEGATIVE_INFINITY;
+  let first = true;
+  let suppressMidReset = false;
+  let topParas = 0;
+  let parasWithSegs = 0;
+  for (const p of childElements(root)) {
+    if (localName2(p) !== "p") continue;
+    topParas++;
+    const explicit = p.getAttribute("pageBreak") === "1";
+    let brokeByExplicit = false;
+    if (explicit && !first) {
+      cur++;
+      brokeByExplicit = true;
+      suppressMidReset = false;
+    }
+    const segs = linesegsOf(p);
+    if (segs.length > 0) parasWithSegs++;
+    let paraFirst = true;
+    let startPage = cur;
+    for (const s of segs) {
+      const v = numAttr(s, "vertpos");
+      const h = numAttr(s, "horzpos");
+      const brk = v < prevV ? !multiCol || h <= prevH : paraFirst && v === prevV && h <= prevH;
+      if (brk && !(paraFirst && brokeByExplicit)) {
+        if (paraFirst && suppressMidReset && v >= MIDPAGE_V) {
+        } else {
+          cur++;
+        }
+      }
+      if (paraFirst) {
+        suppressMidReset = false;
+        startPage = cur;
+      }
+      paraFirst = false;
+      prevV = v;
+      prevH = h;
+    }
+    paraPage.set(p, startPage);
+    first = false;
+    let add = 0;
+    for (const r of childElements(p)) {
+      if (localName2(r) !== "run") continue;
+      for (const t of childElements(r)) {
+        if (localName2(t) === "tbl") add += tableIntraBreaks(t);
+      }
+    }
+    if (add > 0) {
+      cur += add;
+      suppressMidReset = true;
+    }
+  }
+  return {
+    paraPage,
+    pages: cur + 1,
+    usable: topParas > 0 && parasWithSegs === topParas
+  };
+}
+
 // src/hwpx/section-walker.ts
 function parseSectionXml(xml, styleMap, warnings, sectionNum, shared) {
   const parser = createXmlParser(warnings);
@@ -1509,6 +1643,13 @@ function parseSectionXml(xml, styleMap, warnings, sectionNum, shared) {
   if (!doc.documentElement) return [];
   const ctx = { styleMap, warnings, sectionNum, shared: shared ?? createSectionShared() };
   ctx.shared.track.deleteDepth = 0;
+  const detect = detectHwpxSectionPages(doc.documentElement);
+  const ps = ctx.shared.pageState;
+  if (!detect.usable) ps.allUsable = false;
+  ctx.paraPage = detect.paraPage;
+  ctx.pageBase = ps.base;
+  ctx.page = ps.base + 1;
+  ps.base += detect.pages;
   for (const tagName of ["hp:secPr", "secPr"]) {
     const els = doc.getElementsByTagName(tagName);
     if (els.length > 0) {
@@ -1555,7 +1696,7 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
           const chan = kordocTableChannel(el, ctx);
           if (chan) {
             if (chan.kind === "heading" && chan.text) {
-              blocks.push({ type: "heading", level: chan.level, text: chan.text, pageNumber: ctx.sectionNum });
+              blocks.push({ type: "heading", level: chan.level, text: chan.text, pageNumber: ctx.page });
             }
             break;
           }
@@ -1574,7 +1715,7 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
             tableCtx.caption = (tableCtx.caption ? tableCtx.caption + "\n" : "") + cap.text;
             if (cap.hasStructure) (tableCtx.captionBlocks ??= []).push(...cap.blocks);
           } else {
-            blocks.push({ type: "paragraph", text: cap.text, pageNumber: ctx.sectionNum });
+            blocks.push({ type: "paragraph", text: cap.text, pageNumber: ctx.page });
           }
         }
         break;
@@ -1617,6 +1758,8 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
         }
         break;
       case "p": {
+        const paraPg = ctx.paraPage?.get(el);
+        if (paraPg !== void 0 && ctx.pageBase !== void 0) ctx.page = ctx.pageBase + paraPg + 1;
         const { text: rawText, href, footnote, style, segments } = extractParagraphInfo(el, ctx.styleMap, ctx);
         let text = rawText;
         let headingLevel;
@@ -1638,7 +1781,7 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
           const flush = () => {
             const s = segs[segIdx++];
             if (!s) return;
-            const block = { type: "paragraph", text: s, pageNumber: ctx.sectionNum };
+            const block = { type: "paragraph", text: s, pageNumber: ctx.page };
             if (first && !cell2) {
               first = false;
               if (style) block.style = style;
@@ -1662,7 +1805,7 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
             if (footnote) text += ` (\uC8FC: ${footnote})`;
             cell2.text += (cell2.text || cell2.paraSeen ? "\n" : "") + text;
             cell2.paraSeen = true;
-            const cellBlock = { type: "paragraph", text, pageNumber: ctx.sectionNum };
+            const cellBlock = { type: "paragraph", text, pageNumber: ctx.page };
             const cellSpanMode = spanModeOf(ctx.shared.kordocLayout);
             if (cellSpanMode && !ph?.prefix) {
               const spans = extractRunSpans(el, ctx, cellSpanMode, cellSpanMode !== "kordoc");
@@ -1674,11 +1817,11 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
             (cell2.blocks ??= []).push(cellBlock);
           } else if (!tableCtx) {
             if (ctx.shared.kordocLayout && /^─{10,}$/.test(text)) {
-              blocks.push({ type: "separator", pageNumber: ctx.sectionNum });
+              blocks.push({ type: "separator", pageNumber: ctx.page });
               tableCtx = walkParagraphChildren(el, blocks, tableCtx, tableStack, ctx, depth + 1);
               break;
             }
-            const block = { type: headingLevel ? "heading" : "paragraph", text, pageNumber: ctx.sectionNum };
+            const block = { type: headingLevel ? "heading" : "paragraph", text, pageNumber: ctx.page };
             if (headingLevel) block.level = headingLevel;
             if (style) block.style = style;
             if (href) block.href = href;
@@ -1704,7 +1847,7 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
             }
             blocks.push(block);
           } else {
-            blocks.push({ type: "paragraph", text, pageNumber: ctx.sectionNum });
+            blocks.push({ type: "paragraph", text, pageNumber: ctx.page });
           }
         } else if (ctx.shared.keepEmptyParagraphs && !hasObjectDescendant(el)) {
           if (tableCtx?.cell) {
@@ -1712,9 +1855,9 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
             if (cell2.text || cell2.paraSeen) cell2.text += "\n";
             cell2.paraSeen = true;
             cell2.lineOpen = false;
-            (cell2.blocks ??= []).push({ type: "paragraph", text: "", pageNumber: ctx.sectionNum });
+            (cell2.blocks ??= []).push({ type: "paragraph", text: "", pageNumber: ctx.page });
           } else if (!tableCtx) {
-            blocks.push({ type: "paragraph", text: "", pageNumber: ctx.sectionNum });
+            blocks.push({ type: "paragraph", text: "", pageNumber: ctx.page });
           }
         }
         tableCtx = walkParagraphChildren(el, blocks, tableCtx, tableStack, ctx, depth + 1);
@@ -1737,7 +1880,7 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
       case "memogroup":
       case "memo": {
         if (ctx.warnings && extractTextFromNode(el)) {
-          ctx.warnings.push({ page: ctx.sectionNum, message: "\uBA54\uBAA8 \uD14D\uC2A4\uD2B8 \uBCF8\uBB38 \uC81C\uC678: memogroup", code: "HIDDEN_TEXT_FILTERED" });
+          ctx.warnings.push({ page: ctx.page, message: "\uBA54\uBAA8 \uD14D\uC2A4\uD2B8 \uBCF8\uBB38 \uC81C\uC678: memogroup", code: "HIDDEN_TEXT_FILTERED" });
         }
         break;
       }
@@ -1749,9 +1892,9 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
 }
 function handleShape(el, sink, ctx) {
   const imgRef = extractImageRef(el);
-  const drawTextChild = findDescendant(el, "drawText");
+  const drawTextChild = findDescendant2(el, "drawText");
   if (imgRef) {
-    const block = { type: "image", text: imgRef, pageNumber: ctx.sectionNum };
+    const block = { type: "image", text: imgRef, pageNumber: ctx.page };
     const alt = userShapeComment(el);
     if (alt) block.footnoteText = alt;
     sink.push(block);
@@ -1762,11 +1905,11 @@ function handleShape(el, sink, ctx) {
   const capEl = findChildByLocalName(el, "caption");
   if (capEl) {
     const capText = collectSubListText(capEl, ctx);
-    if (capText) sink.push({ type: "paragraph", text: capText, pageNumber: ctx.sectionNum });
+    if (capText) sink.push({ type: "paragraph", text: capText, pageNumber: ctx.page });
   }
   if (!imgRef && !drawTextChild && ctx.warnings && ctx.sectionNum) {
     const localTag = (el.tagName || el.localName || "").replace(/^[^:]+:/, "");
-    ctx.warnings.push({ page: ctx.sectionNum, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
+    ctx.warnings.push({ page: ctx.page, message: `\uC2A4\uD0B5\uB41C \uC694\uC18C: ${localTag}`, code: "SKIPPED_IMAGE" });
   }
 }
 function userShapeComment(el) {
@@ -1851,7 +1994,7 @@ function collectSubListContent(el, ctx, depth = 0) {
       const t = extractParagraphInfo(ch, ctx.styleMap, ctx).text;
       if (t) {
         parts.push(t);
-        out.blocks.push({ type: "paragraph", text: t, pageNumber: ctx.sectionNum });
+        out.blocks.push({ type: "paragraph", text: t, pageNumber: ctx.page });
       }
       const tbls = [];
       findTopLevelTbls(ch, tbls);
@@ -1924,7 +2067,7 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth =
           const chan = kordocTableChannel(el, ctx);
           if (chan) {
             if (chan.kind === "heading" && chan.text) {
-              blocks.push({ type: "heading", level: chan.level, text: chan.text, pageNumber: ctx.sectionNum });
+              blocks.push({ type: "heading", level: chan.level, text: chan.text, pageNumber: ctx.page });
             }
             continue;
           }
@@ -1936,11 +2079,11 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth =
       } else if (localTag === "caption" && !inShape) {
         const cap = collectSubListContent(el, ctx);
         if (cap.text) {
-          if (tableCtx?.cell) mergeBlocksIntoCell(tableCtx.cell, [{ type: "paragraph", text: cap.text, pageNumber: ctx.sectionNum }]);
+          if (tableCtx?.cell) mergeBlocksIntoCell(tableCtx.cell, [{ type: "paragraph", text: cap.text, pageNumber: ctx.page }]);
           else if (tableCtx) {
             tableCtx.caption = (tableCtx.caption ? tableCtx.caption + "\n" : "") + cap.text;
             if (cap.hasStructure) (tableCtx.captionBlocks ??= []).push(...cap.blocks);
-          } else blocks.push({ type: "paragraph", text: cap.text, pageNumber: ctx.sectionNum });
+          } else blocks.push({ type: "paragraph", text: cap.text, pageNumber: ctx.page });
         }
       } else if (localTag === "pic" || localTag === "shape" || localTag === "drawingObject") {
         if (tableCtx?.cell) {
@@ -1985,7 +2128,7 @@ function hasObjectDescendant(node, depth = 0) {
 function isInlineTbl(tbl2) {
   return findChildByLocalName(tbl2, "pos")?.getAttribute("treatAsChar") === "1";
 }
-function findDescendant(node, targetTag, depth = 0) {
+function findDescendant2(node, targetTag, depth = 0) {
   if (depth > 5) return null;
   const children = node.childNodes;
   if (!children) return null;
@@ -1994,7 +2137,7 @@ function findDescendant(node, targetTag, depth = 0) {
     if (child.nodeType !== 1) continue;
     const tag = (child.tagName || child.localName || "").replace(/^[^:]+:/, "");
     if (tag === targetTag) return child;
-    const found = findDescendant(child, targetTag, depth + 1);
+    const found = findDescendant2(child, targetTag, depth + 1);
     if (found) return found;
   }
   return null;
@@ -2015,7 +2158,7 @@ function extractDrawTextBlocks(drawTextNode, blocks, ctx) {
         const ph = resolveParaHeading(child, ctx);
         if (text) {
           if (ph?.prefix) text = ph.prefix + " " + text;
-          const block = { type: "paragraph", text, style: info.style ?? void 0, pageNumber: ctx.sectionNum };
+          const block = { type: "paragraph", text, style: info.style ?? void 0, pageNumber: ctx.page };
           if (info.href) block.href = info.href;
           if (info.footnote) block.footnoteText = info.footnote;
           blocks.push(block);
@@ -2118,7 +2261,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         // 숨은 설명 — 본문 혼입 차단
         case "hiddenComment": {
           if (ctx?.warnings && extractTextFromNode(k)) {
-            ctx.warnings.push({ page: ctx.sectionNum, message: "\uC228\uC740 \uC124\uBA85 \uD14D\uC2A4\uD2B8 \uC81C\uC678: hiddenComment", code: "HIDDEN_TEXT_FILTERED" });
+            ctx.warnings.push({ page: ctx.page, message: "\uC228\uC740 \uC124\uBA85 \uD14D\uC2A4\uD2B8 \uC81C\uC678: hiddenComment", code: "HIDDEN_TEXT_FILTERED" });
           }
           break;
         }
@@ -2138,7 +2281,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         // 미지원 요소 — 텍스트를 가졌으면 무음 손실 대신 경고
         default: {
           if (ctx?.warnings && extractTextFromNode(k)) {
-            ctx.warnings.push({ page: ctx.sectionNum, message: `\uBBF8\uC9C0\uC6D0 \uC81C\uC5B4 \uC694\uC18C\uC758 \uD14D\uC2A4\uD2B8 \uC190\uC2E4: ${ktag}`, code: "UNSUPPORTED_ELEMENT" });
+            ctx.warnings.push({ page: ctx.page, message: `\uBBF8\uC9C0\uC6D0 \uC81C\uC5B4 \uC694\uC18C\uC758 \uD14D\uC2A4\uD2B8 \uC190\uC2E4: ${ktag}`, code: "UNSUPPORTED_ELEMENT" });
           }
         }
       }
@@ -2155,7 +2298,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         if (isInDeletedRange(ctx)) {
           if (t && ctx && !ctx.shared.track.warned) {
             ctx.shared.track.warned = true;
-            ctx.warnings?.push({ page: ctx.sectionNum, message: "\uBCC0\uACBD\uCD94\uC801 \uC0AD\uC81C \uD14D\uC2A4\uD2B8 \uCD9C\uB825 \uC81C\uC678", code: "HIDDEN_TEXT_FILTERED" });
+            ctx.warnings?.push({ page: ctx.page, message: "\uBCC0\uACBD\uCD94\uC801 \uC0AD\uC81C \uD14D\uC2A4\uD2B8 \uCD9C\uB825 \uC81C\uC678", code: "HIDDEN_TEXT_FILTERED" });
           }
         } else {
           text += t.replace(/\x1E/g, "");
@@ -2267,7 +2410,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
             } catch {
               if (ctx?.warnings) {
                 ctx.warnings.push({
-                  page: ctx.sectionNum,
+                  page: ctx.page,
                   message: `\uC218\uC2DD LaTeX \uBCC0\uD658 \uC2E4\uD328 \u2014 \uC218\uC2DD \uD14D\uC2A4\uD2B8 \uC81C\uC678: ${raw.trim().slice(0, 40)}`,
                   code: "PARTIAL_PARSE"
                 });
@@ -2424,6 +2567,7 @@ import { inflateRawSync as inflateRawSync2, inflateSync } from "zlib";
 var TAG_PARA_HEADER = 66;
 var TAG_PARA_TEXT = 67;
 var TAG_CHAR_SHAPE = 68;
+var TAG_PARA_LINE_SEG = 69;
 var TAG_CTRL_HEADER = 71;
 var TAG_LIST_HEADER = 72;
 var TAG_TABLE = 77;
@@ -3058,7 +3202,9 @@ function extractFromBrokenZip(buffer) {
       totalDecompressed += content.length * 2;
       if (totalDecompressed > MAX_DECOMPRESS_SIZE) throw new KordocError("\uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC");
       sectionNum++;
-      blocks.push(...parseSectionXml(content, void 0, warnings, sectionNum, shared));
+      const secBlocks = parseSectionXml(content, void 0, warnings, sectionNum, shared);
+      for (const b of secBlocks) b.pageNumber = sectionNum;
+      blocks.push(...secBlocks);
     } catch {
       continue;
     }
@@ -3197,33 +3343,48 @@ async function parseHwpxDocument(buffer, options) {
   const warnings = [];
   const sectionPaths = await resolveSectionPaths(zip);
   if (sectionPaths.length === 0) throw new KordocError("HWPX\uC5D0\uC11C \uC139\uC158 \uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
-  metadata.pageCount = sectionPaths.length;
-  const pageFilter = options?.pages ? parsePageRange(options.pages, sectionPaths.length) : null;
-  const totalTarget = pageFilter ? pageFilter.size : sectionPaths.length;
-  const blocks = [];
+  let allBlocks = [];
   const shared = createSectionShared();
   shared.kordocLayout = await readKordocLayout(zip);
   shared.keepTrailingEmptyCols = options?.keepTrailingEmptyCols;
   shared.keepEmptyParagraphs = options?.keepEmptyParagraphs;
+  const sectionRanges = [];
   let parsedSections = 0;
   for (let si = 0; si < sectionPaths.length; si++) {
-    if (pageFilter && !pageFilter.has(si + 1)) continue;
     const file = zip.file(sectionPaths[si]);
     if (!file) continue;
     try {
       const xml = await file.async("text");
       decompressed.total += xml.length * 2;
       if (decompressed.total > MAX_DECOMPRESS_SIZE) throw new ZipBombError("ZIP \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (ZIP bomb \uC758\uC2EC)");
-      blocks.push(...parseSectionXml(xml, styleMap, warnings, si + 1, shared));
+      const start = allBlocks.length;
+      allBlocks.push(...parseSectionXml(xml, styleMap, warnings, si + 1, shared));
+      sectionRanges.push({ sectionNum: si + 1, start, end: allBlocks.length });
       parsedSections++;
-      options?.onProgress?.(parsedSections, totalTarget);
+      options?.onProgress?.(parsedSections, sectionPaths.length);
     } catch (secErr) {
       if (secErr instanceof ZipBombError) throw secErr;
       warnings.push({ page: si + 1, message: `\uC139\uC158 ${si + 1} \uD30C\uC2F1 \uC2E4\uD328: ${secErr instanceof Error ? secErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
     }
   }
+  const layoutPages = shared.pageState.allUsable && parsedSections > 0;
+  metadata.pageMode = layoutPages ? "layout" : "section";
+  metadata.pageCount = layoutPages ? Math.max(shared.pageState.base, 1) : sectionPaths.length;
+  if (!layoutPages) {
+    for (const r of sectionRanges) {
+      for (let b = r.start; b < r.end; b++) allBlocks[b].pageNumber = r.sectionNum;
+    }
+  }
+  if (options?.pages) {
+    const pageFilter = parsePageRange(options.pages, metadata.pageCount);
+    allBlocks = allBlocks.filter((b) => b.pageNumber != null && pageFilter.has(b.pageNumber));
+    if (!layoutPages) {
+      warnings.push({ code: "PAGE_BOUNDARY_APPROXIMATE", message: "\uC870\uD310 \uCE90\uC2DC\uAC00 \uC5C6\uC5B4 pages \uD544\uD130\uB97C \uC139\uC158 \uB2E8\uC704 \uADFC\uC0AC\uB85C \uC801\uC6A9\uD588\uC2B5\uB2C8\uB2E4" });
+    }
+  }
+  const blocks = allBlocks;
   applyPageText(blocks, shared);
-  const images = await extractImagesFromZip(zip, blocks, decompressed, warnings, !pageFilter);
+  const images = await extractImagesFromZip(zip, blocks, decompressed, warnings, !options?.pages);
   detectHwpxHeadings(blocks, styleMap);
   const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
   const markdown = blocksToMarkdown(blocks);
@@ -4244,6 +4405,112 @@ function assertSupportedEncryptVersion(version) {
   }
 }
 
+// src/hwp5/page-boundary.ts
+var MIDPAGE_V2 = 2e3;
+var CTRL_ID_TBL = 1952607264;
+function tableIntraBreaks2(records, start, end) {
+  const rowMax = /* @__PURE__ */ new Map();
+  let curRow = -1;
+  let prevV = Number.NEGATIVE_INFINITY;
+  let resets = 0;
+  const flush = () => {
+    if (curRow >= 0) rowMax.set(curRow, Math.max(rowMax.get(curRow) ?? 0, resets));
+    resets = 0;
+    prevV = Number.NEGATIVE_INFINITY;
+  };
+  for (let i = start; i < end; i++) {
+    const rec = records[i];
+    if (rec.tagId === TAG_LIST_HEADER && rec.level === 2 && rec.data.length >= 12) {
+      flush();
+      curRow = rec.data.readUInt16LE(10);
+    } else if (rec.tagId === TAG_PARA_LINE_SEG && rec.level === 3 && curRow >= 0) {
+      for (let off = 0; off + 36 <= rec.data.length; off += 36) {
+        const v = rec.data.readInt32LE(off + 4);
+        if (v < prevV) resets++;
+        prevV = v;
+      }
+    }
+  }
+  flush();
+  let sum = 0;
+  for (const v of rowMax.values()) sum += v;
+  return sum;
+}
+function detectHwp5SectionPages(records) {
+  const pageAtPara = [];
+  let cur = 0;
+  let prevV = Number.NEGATIVE_INFINITY;
+  let prevH = Number.NEGATIVE_INFINITY;
+  let first = true;
+  let suppressMidReset = false;
+  let parasWithSegs = 0;
+  let i = 0;
+  while (i < records.length) {
+    const rec = records[i];
+    if (rec.tagId !== TAG_PARA_HEADER || rec.level !== 0) {
+      i++;
+      continue;
+    }
+    let j = i + 1;
+    while (j < records.length && records[j].level > 0) j++;
+    const divideSort = rec.data.length >= 12 ? rec.data.readUInt8(11) : 0;
+    const explicit = (divideSort & 4) !== 0;
+    let brokeByExplicit = false;
+    if (explicit && !first) {
+      cur++;
+      brokeByExplicit = true;
+      suppressMidReset = false;
+    }
+    let paraFirst = true;
+    let startPage = cur;
+    let hadSegs = false;
+    let intraAdd = 0;
+    for (let k = i + 1; k < j; k++) {
+      const r = records[k];
+      if (r.tagId === TAG_PARA_LINE_SEG && r.level === 1) {
+        hadSegs = true;
+        for (let off = 0; off + 36 <= r.data.length; off += 36) {
+          const v = r.data.readInt32LE(off + 4);
+          const h = r.data.readInt32LE(off + 24);
+          const brk = v < prevV || paraFirst && v === prevV && h <= prevH;
+          if (brk && !(paraFirst && brokeByExplicit)) {
+            if (paraFirst && suppressMidReset && v >= MIDPAGE_V2) {
+            } else {
+              cur++;
+            }
+          }
+          if (paraFirst) {
+            suppressMidReset = false;
+            startPage = cur;
+          }
+          paraFirst = false;
+          prevV = v;
+          prevH = h;
+        }
+      }
+      if (r.tagId === TAG_CTRL_HEADER && r.level === 1 && r.data.length >= 4 && r.data.readUInt32LE(0) === CTRL_ID_TBL) {
+        let ce = k + 1;
+        while (ce < j && records[ce].level > 1) ce++;
+        intraAdd += tableIntraBreaks2(records, k + 1, ce);
+        k = ce - 1;
+      }
+    }
+    if (hadSegs) parasWithSegs++;
+    pageAtPara.push(startPage);
+    first = false;
+    if (intraAdd > 0) {
+      cur += intraAdd;
+      suppressMidReset = true;
+    }
+    i = j;
+  }
+  return {
+    pageAtPara,
+    pages: cur + 1,
+    usable: pageAtPara.length > 0 && parasWithSegs === pageAtPara.length
+  };
+}
+
 // src/hwp5/parser.ts
 import { createRequire } from "module";
 var require2 = createRequire(import.meta.url);
@@ -4369,31 +4636,63 @@ function parseHwp5Document(buffer, options) {
   const docInfo = cfb && !encrypted ? parseDocInfoStream(cfb, compressed) : parseDocInfoFromStream(findStream("/DocInfo"), compressed);
   const sections = distribution ? cfb ? findViewTextSections(cfb, compressed) : findViewTextSectionsLenient(lenientCfb, compressed) : encrypted ? findSectionsVia(findStream) : cfb ? findSections(cfb) : findSectionsLenient(lenientCfb, compressed);
   if (sections.length === 0) throw new KordocError("\uC139\uC158 \uC2A4\uD2B8\uB9BC\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4");
-  metadata.pageCount = sections.length;
-  const pageFilter = options?.pages ? parsePageRange(options.pages, sections.length) : null;
-  const totalTarget = pageFilter ? pageFilter.size : sections.length;
-  const bodyBlocks = [];
-  const doc = createHwp5DocState();
-  doc.keepTrailingEmptyCols = options?.keepTrailingEmptyCols;
+  const sectionRecords = [];
+  const sectionDetects = [];
   let totalDecompressed = 0;
-  let parsedSections = 0;
   for (let si = 0; si < sections.length; si++) {
-    if (pageFilter && !pageFilter.has(si + 1)) continue;
     try {
-      const sectionData = sections[si];
-      const data = !distribution && compressed ? decompressStream(Buffer.from(sectionData)) : Buffer.from(sectionData);
+      const data = !distribution && compressed ? decompressStream(Buffer.from(sections[si])) : Buffer.from(sections[si]);
       totalDecompressed += data.length;
       if (totalDecompressed > MAX_TOTAL_DECOMPRESS) throw new KordocError("\uCD1D \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (decompression bomb \uC758\uC2EC)");
       const records = readRecords(data);
-      const sectionBlocks = parseSection(records, docInfo, warnings, si + 1, doc);
+      sectionRecords.push(records);
+      sectionDetects.push(detectHwp5SectionPages(records));
+    } catch (secErr) {
+      if (secErr instanceof KordocError) throw secErr;
+      sectionRecords.push(null);
+      sectionDetects.push(null);
+      warnings.push({ page: si + 1, message: `\uC139\uC158 ${si + 1} \uD30C\uC2F1 \uC2E4\uD328: ${secErr instanceof Error ? secErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
+    }
+  }
+  const layoutPages = sectionDetects.length > 0 && sectionDetects.every((d) => d != null && d.usable);
+  metadata.pageMode = layoutPages ? "layout" : "section";
+  metadata.pageCount = layoutPages ? Math.max(sectionDetects.reduce((sum, d) => sum + (d?.pages ?? 0), 0), 1) : sections.length;
+  const pageFilter = options?.pages ? parsePageRange(options.pages, metadata.pageCount) : null;
+  if (pageFilter && !layoutPages) {
+    warnings.push({ code: "PAGE_BOUNDARY_APPROXIMATE", message: "\uC870\uD310 \uCE90\uC2DC\uAC00 \uC5C6\uC5B4 pages \uD544\uD130\uB97C \uC139\uC158 \uB2E8\uC704 \uADFC\uC0AC\uB85C \uC801\uC6A9\uD588\uC2B5\uB2C8\uB2E4" });
+  }
+  let bodyBlocks = [];
+  const doc = createHwp5DocState();
+  doc.keepTrailingEmptyCols = options?.keepTrailingEmptyCols;
+  let parsedSections = 0;
+  let pageBase = 0;
+  for (let si = 0; si < sections.length; si++) {
+    const records = sectionRecords[si];
+    if (!records) continue;
+    const detect = sectionDetects[si];
+    if (!layoutPages && pageFilter && !pageFilter.has(si + 1)) continue;
+    try {
+      const sectionBlocks = parseSection(
+        records,
+        docInfo,
+        warnings,
+        si + 1,
+        doc,
+        layoutPages ? { base: pageBase, pageAtPara: detect.pageAtPara } : void 0
+      );
       bodyBlocks.push(...sectionBlocks);
       parsedSections++;
-      options?.onProgress?.(parsedSections, totalTarget);
+      options?.onProgress?.(parsedSections, sections.length);
     } catch (secErr) {
       if (secErr instanceof KordocError) throw secErr;
       warnings.push({ page: si + 1, message: `\uC139\uC158 ${si + 1} \uD30C\uC2F1 \uC2E4\uD328: ${secErr instanceof Error ? secErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
+    } finally {
+      pageBase += detect.pages;
     }
   }
+  if (pageFilter && layoutPages) {
+    bodyBlocks = bodyBlocks.filter((b) => b.pageNumber != null && pageFilter.has(b.pageNumber));
+  }
   const blocks = [...doc.headerBlocks, ...bodyBlocks, ...doc.footerBlocks];
   const images = cfb ? extractHwp5Images(cfb.FileIndex, blocks, warnings, !pageFilter) : extractHwp5ImagesLenient(lenientCfb, blocks, warnings, !pageFilter);
   let flatBlocks = flattenLayoutTables(blocks);
@@ -4630,8 +4929,17 @@ function createHwp5DocState() {
     footerBlocks: []
   };
 }
-function parseSection(records, docInfo, warnings, sectionNum, doc) {
-  const ctx = { docInfo, warnings, sectionNum, doc: doc ?? createHwp5DocState(), depth: 0 };
+function parseSection(records, docInfo, warnings, sectionNum, doc, pageMap) {
+  const ctx = {
+    docInfo,
+    warnings,
+    sectionNum,
+    doc: doc ?? createHwp5DocState(),
+    depth: 0,
+    page: pageMap ? pageMap.base + 1 : sectionNum,
+    pageMap,
+    topOrdinal: 0
+  };
   return parseParagraphList(records, 0, records.length, ctx);
 }
 function parseParagraphList(records, start, end, ctx) {
@@ -4640,6 +4948,11 @@ function parseParagraphList(records, start, end, ctx) {
   while (i < end) {
     if (records[i].tagId === TAG_PARA_HEADER) {
       const baseLevel = records[i].level;
+      if (baseLevel === 0 && ctx.depth === 0 && ctx.pageMap) {
+        const pg = ctx.pageMap.pageAtPara[ctx.topOrdinal];
+        if (pg !== void 0) ctx.page = ctx.pageMap.base + pg + 1;
+        ctx.topOrdinal++;
+      }
       let j = i + 1;
       while (j < end && records[j].level > baseLevel) j++;
       blocks.push(...parseParagraph(records, i, j, ctx));
@@ -4740,7 +5053,7 @@ function parseParagraph(records, start, end, ctx) {
     const block = {
       type: headingLevel > 0 ? "heading" : "paragraph",
       text: headMarker ? `${headMarker} ${trimmed}` : trimmed,
-      pageNumber: ctx.sectionNum
+      pageNumber: ctx.page
     };
     if (headingLevel > 0) block.level = headingLevel;
     if (ctx.docInfo && charShapeIds.length > 0) {
@@ -4757,7 +5070,7 @@ function parseParagraph(records, start, end, ctx) {
     if (footnotes.length > 0) block.footnoteText = footnotes.join("; ");
     blocks.push(block);
   } else if (footnotes.length > 0) {
-    blocks.push({ type: "paragraph", text: `(\uC8FC: ${footnotes.join("; ")})`, pageNumber: ctx.sectionNum });
+    blocks.push({ type: "paragraph", text: `(\uC8FC: ${footnotes.join("; ")})`, pageNumber: ctx.page });
   }
   for (const ctrl of ctrls) {
     if (ctrl.afterBlocks) blocks.push(...ctrl.afterBlocks);
@@ -4768,7 +5081,7 @@ function applyCtrlEffect(ctrl, records, ctx) {
   switch (ctrl.id) {
     case CTRL_TBL: {
       const table2 = parseTableControl(ctrl, records, ctx);
-      if (table2) ctrl.afterBlocks = [{ type: "table", table: table2, pageNumber: ctx.sectionNum }];
+      if (table2) ctrl.afterBlocks = [{ type: "table", table: table2, pageNumber: ctx.page }];
       return;
     }
     case CTRL_GSO: {
@@ -4820,7 +5133,7 @@ function applyCtrlEffect(ctrl, records, ctx) {
       return;
     }
     case CTRL_OLE: {
-      ctx.warnings.push({ page: ctx.sectionNum, message: "\uC2A4\uD0B5\uB41C OLE \uAC1C\uCCB4", code: "SKIPPED_OLE" });
+      ctx.warnings.push({ page: ctx.page, message: "\uC2A4\uD0B5\uB41C OLE \uAC1C\uCCB4", code: "SKIPPED_OLE" });
       return;
     }
     // 숨은 설명/단 정의/쪽번호 위치/감추기/찾아보기/책갈피/글자겹침/덧말 — 본문 텍스트 없음 또는 의도적 스킵
@@ -4897,7 +5210,7 @@ function applyHeaderFooterEffect(ctrl, records, ctx, isHeader) {
   const key = (isHeader ? "h:" : "f:") + text;
   if (ctx.doc.headerTexts.has(key)) return;
   ctx.doc.headerTexts.add(key);
-  const block = { type: "paragraph", text, pageNumber: ctx.sectionNum };
+  const block = { type: "paragraph", text, pageNumber: ctx.page };
   if (isHeader) ctx.doc.headerBlocks.push(block);
   else ctx.doc.footerBlocks.push(block);
 }
@@ -5119,11 +5432,11 @@ function pictureToImageBlock(data, ctx) {
   if (binDataId === 0) return null;
   const item = ctx.docInfo?.binData[binDataId - 1];
   if (item?.kind === "link") {
-    ctx.warnings.push({ page: ctx.sectionNum, message: `\uC678\uBD80 \uC5F0\uACB0 \uC774\uBBF8\uC9C0 (binDataId ${binDataId})`, code: "SKIPPED_IMAGE" });
+    ctx.warnings.push({ page: ctx.page, message: `\uC678\uBD80 \uC5F0\uACB0 \uC774\uBBF8\uC9C0 (binDataId ${binDataId})`, code: "SKIPPED_IMAGE" });
     return null;
   }
   const storageId = item && item.storageId > 0 ? item.storageId : binDataId;
-  return { type: "image", text: String(storageId), pageNumber: ctx.sectionNum };
+  return { type: "image", text: String(storageId), pageNumber: ctx.page };
 }
 function resolveCharStyle(charShapeIds, docInfo) {
   if (charShapeIds.length === 0 || docInfo.charShapes.length === 0) return void 0;
@@ -19014,8 +19327,8 @@ function isDisplayMath(el) {
 
 // src/docx/parser.ts
 var MAX_DECOMPRESS_SIZE5 = 100 * 1024 * 1024;
-function matchesLocal(el, localName2) {
-  return el.localName === localName2 || (el.tagName?.endsWith(`:${localName2}`) ?? false);
+function matchesLocal(el, localName3) {
+  return el.localName === localName3 || (el.tagName?.endsWith(`:${localName3}`) ?? false);
 }
 function effectiveChildElements(parent) {
   const result = [];
@@ -19039,10 +19352,10 @@ function effectiveChildElements(parent) {
   }
   return result;
 }
-function getChildElements(parent, localName2) {
-  return effectiveChildElements(parent).filter((el) => matchesLocal(el, localName2));
+function getChildElements(parent, localName3) {
+  return effectiveChildElements(parent).filter((el) => matchesLocal(el, localName3));
 }
-function findElements(parent, localName2) {
+function findElements(parent, localName3) {
   const result = [];
   const walk = (node) => {
     const children = node.childNodes;
@@ -19050,7 +19363,7 @@ function findElements(parent, localName2) {
       const child = children[i];
       if (child.nodeType === 1) {
         const el = child;
-        if (el.localName === localName2 || el.tagName?.endsWith(`:${localName2}`)) {
+        if (el.localName === localName3 || el.tagName?.endsWith(`:${localName3}`)) {
           result.push(el);
         }
         walk(el);
@@ -19060,11 +19373,11 @@ function findElements(parent, localName2) {
   walk(parent);
   return result;
 }
-function getAttr(el, localName2) {
+function getAttr(el, localName3) {
   const attrs = el.attributes;
   for (let i = 0; i < attrs.length; i++) {
     const attr = attrs[i];
-    if (attr.localName === localName2 || attr.name === localName2) return attr.value;
+    if (attr.localName === localName3 || attr.name === localName3) return attr.value;
   }
   return null;
 }
@@ -19583,8 +19896,8 @@ async function parseDocxDocument(buffer, options) {
   const bodyEl = body[0];
   const topLevel = effectiveChildElements(bodyEl);
   for (const el of topLevel) {
-    const localName2 = el.localName ?? el.tagName?.split(":").pop();
-    if (localName2 === "p") {
+    const localName3 = el.localName ?? el.tagName?.split(":").pop();
+    if (localName3 === "p") {
       const block = parseParagraph2(el, styles, numbering, footnotes, rels);
       if (block) blocks.push(block);
       if (imageMap.size > 0) emitParagraphImages(el, imageMap, linkedImages, blocks);
@@ -19593,7 +19906,7 @@ async function parseDocxDocument(buffer, options) {
         if (tb) blocks.push(tb);
         if (imageMap.size > 0) emitParagraphImages(tp, imageMap, linkedImages, blocks);
       }
-    } else if (localName2 === "tbl") {
+    } else if (localName3 === "tbl") {
       const block = parseTable(el, styles, numbering, footnotes, rels, options?.keepTrailingEmptyCols);
       if (block) blocks.push(block);
     }
@@ -28676,7 +28989,7 @@ var Surgeon = class {
 // src/roundtrip/hwp5-patch.ts
 var require3 = createRequire2(import.meta.url);
 var CFB2 = require3("cfb");
-var TAG_PARA_LINE_SEG = 69;
+var TAG_PARA_LINE_SEG2 = 69;
 function cid2(s) {
   return (s.charCodeAt(0) << 24 | s.charCodeAt(1) << 16 | s.charCodeAt(2) << 8 | s.charCodeAt(3)) >>> 0;
 }
@@ -28757,7 +29070,7 @@ function scanSection(stream, sectionIndex, compressed) {
         textIdx = textIdx === -1 ? j : -2;
         appendParaText(state, records[j].data);
       } else if (t === TAG_CHAR_SHAPE && charShapeIdx === -1) charShapeIdx = j;
-      else if (t === TAG_PARA_LINE_SEG && lineSegIdx === -1) lineSegIdx = j;
+      else if (t === TAG_PARA_LINE_SEG2 && lineSegIdx === -1) lineSegIdx = j;
     }
     let ctrlSeen = false, nonGso = false;
     for (let a = parent[i]; a >= 0; a = parent[a]) {
@@ -31324,7 +31637,7 @@ async function parse(input, options) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-TAJHRZ5T.js");
+    const { parseImageDocument } = await import("./image-ocr-XHBZONFZ.js");
     const { blocks, warnings } = await parseImageDocument(buffer, options);
     return {
       success: true,
@@ -31383,7 +31696,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-RCEHB2EV.js");
+    const mod = await import("./parser-Z36MZUHS.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {