kordoc 4.18.2 → 4.18.4

package/dist/index.js

@@ -1,7 +1,7 @@
 import {
   blocksToPages,
   inlineImagesIntoMarkdown
-} from "./chunk-O2DI4GHD.js";
+} from "./chunk-274ZPMHK.js";
 import {
   parsePageRange
 } from "./chunk-V4BH5SRQ.js";
@@ -31,7 +31,7 @@ import {
   tidyScriptTags,
   unframeLayoutTables,
   wrapScript
-} from "./chunk-B2E6C2LH.js";
+} from "./chunk-XG4SYOZL.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT,
@@ -41,6 +41,7 @@ import {
   escapeHtml,
   isPathTraversal,
   normalizeSectionHref,
+  partExtension,
   precheckZipSize,
   sanitizeError,
   sanitizeHref,
@@ -48,7 +49,7 @@ import {
   toArrayBuffer,
   unescapeHtml,
   unzipLimitBytes
-} from "./chunk-N5DQFA7L.js";
+} from "./chunk-7GR2GCJF.js";
 
 // src/plain-markdown.ts
 function toPlainMarkdown(md2) {
@@ -585,7 +586,8 @@ function createSectionShared() {
     numState: /* @__PURE__ */ new Map(),
     pageText: { headers: [], footers: [] },
     track: { deleteDepth: 0, warned: false, deletedObjects: /* @__PURE__ */ new WeakSet(), deletedImageRefs: /* @__PURE__ */ new Set() },
-    pageState: { base: 0, allUsable: true }
+    pageState: { base: 0, allUsable: true },
+    tocTexts: /* @__PURE__ */ new Set()
   };
 }
 function createXmlParser(warnings) {
@@ -937,7 +939,7 @@ function parseParaHeadings(doc, map) {
     if (map.size > 0) break;
   }
 }
-function detectHwpxHeadings(blocks, styleMap) {
+function detectHwpxHeadings(blocks, styleMap, tocTexts) {
   if (blocks.some((b) => b.type === "heading")) return;
   let baseFontSize = 0;
   const sizeFreq = /* @__PURE__ */ new Map();
@@ -954,7 +956,7 @@ function detectHwpxHeadings(blocks, styleMap) {
     }
   }
   for (const block of blocks) {
-    if (block.type !== "paragraph" || !block.text || TOC_ENTRY_BLOCKS.has(block)) continue;
+    if (block.type !== "paragraph" || !block.text || TOC_ENTRY_BLOCKS.has(block) || tocTexts?.has(block.text.trim())) continue;
     const text = block.text.trim();
     if (text.length === 0 || text.length > 200 || /^\d+$/.test(text)) continue;
     let level = 0;
@@ -2121,13 +2123,13 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
         }
         if (tableCtx) tableStack.push(tableCtx);
         const newTable = { rows: [], currentRow: [], cell: null, sourceId: el.getAttribute("id") ?? void 0 };
-        walkSection(el, blocks, newTable, tableStack, ctx, depth + 1);
+        inIndependentDeletionStory(ctx.shared.track, () => walkSection(el, blocks, newTable, tableStack, ctx, depth + 1));
         tableCtx = completeTable(newTable, tableStack, blocks, ctx);
         break;
       }
       // 표/도표 캡션 — IRTable.caption으로 보존 (v3.0, 기존 무음 드롭 수정)
       case "caption": {
-        const cap = collectSubListContent(el, ctx);
+        const cap = inIndependentDeletionStory(ctx.shared.track, () => collectSubListContent(el, ctx));
         if (cap.text) {
           if (tableCtx) {
             tableCtx.caption = (tableCtx.caption ? tableCtx.caption + "\n" : "") + cap.text;
@@ -2187,6 +2189,9 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
         if (text) {
           if (ph?.prefix) text = ph.prefix + " " + text;
           headingLevel = tocEntry ? void 0 : ph?.headingLevel;
+          if (tocEntry) {
+            for (const line of text.split("\n")) if (line.trim()) ctx.shared.tocTexts.add(line.trim());
+          }
         }
         if (segments) {
           const cell2 = tableCtx?.cell ?? null;
@@ -2272,6 +2277,15 @@ function walkSection(node, blocks, tableCtx, tableStack, ctx, depth = 0) {
               if (spanMode !== "foreign" && el.getAttribute("paraPrIDRef") === KORDOC_PARA_QUOTE) block.quote = true;
             }
             if (!headingLevel && !block.spans && placeholderSpans) block.spans = withPrefixSpan(placeholderSpans, ph?.prefix);
+            if (headingLevel && placeholderSpans) {
+              const shown = withPrefixSpan(placeholderSpans.filter((s) => !s.placeholder), ph?.prefix).map((s) => s.text).join("").trim();
+              if (shown) block.text = shown;
+              else {
+                block.type = "paragraph";
+                delete block.level;
+                block.spans = withPrefixSpan(placeholderSpans, ph?.prefix);
+              }
+            }
             blocks.push(block);
           } else {
             blocks.push({ type: "paragraph", text, pageNumber: ctx.page });
@@ -2513,10 +2527,10 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, ctx, depth =
         }
         if (tableCtx) tableStack.push(tableCtx);
         const newTable = { rows: [], currentRow: [], cell: null, inline: isInlineTbl(el), sourceId: el.getAttribute("id") ?? void 0 };
-        walkSection(el, blocks, newTable, tableStack, ctx, d + 1);
+        inIndependentDeletionStory(ctx.shared.track, () => walkSection(el, blocks, newTable, tableStack, ctx, d + 1));
         tableCtx = completeTable(newTable, tableStack, blocks, ctx);
       } else if (localTag === "caption" && !inShape) {
-        const cap = collectSubListContent(el, ctx);
+        const cap = inIndependentDeletionStory(ctx.shared.track, () => collectSubListContent(el, ctx));
         if (cap.text) {
           if (tableCtx?.cell) mergeBlocksIntoCell(tableCtx.cell, [{ type: "paragraph", text: cap.text, pageNumber: ctx.page }]);
           else if (tableCtx) {
@@ -2712,7 +2726,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         // 아니므로 방출하지 않는다
         case "header":
         case "footer": {
-          if (!ctx) break;
+          if (!ctx || isInDeletedRange(ctx)) break;
           const t = collectSubListText(k, ctx);
           if (t && hasPageAutoNum(k) && !/\p{L}/u.test(t)) break;
           if (t) {
@@ -2834,9 +2848,10 @@ function extractParagraphInfo(para2, styleMap, ctx) {
           if (isInlineTbl(child)) text += "";
           break;
         // 양식 선택 상자·라디오 단추 — 한컴이 그리는 상자 기호와 캡션 글 (종전엔 통째로 빠졌다: form-002 "원천기술형"·서울 결재 "부분공개")
+        // 지운 구간의 양식 단추·수식은 글처럼 내지 않는다 (최종본)
         case "checkBtn":
         case "radioBtn":
-          text += formButtonText(child, tag === "radioBtn");
+          if (!isInDeletedRange(ctx)) text += formButtonText(child, tag === "radioBtn");
           break;
         // 하이퍼링크
         case "hyperlink": {
@@ -2906,6 +2921,7 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         // 스크립트가 담겨 있음. hml-equation-parser 로 LaTeX 변환 후 `$...$`
         // 로 래핑. 실패/빈 스크립트면 무시 (대체 텍스트 누출 방지).
         case "equation": {
+          if (isInDeletedRange(ctx)) break;
           const script = findChildByLocalName(child, "script");
           const raw = script ? extractTextFromNode(script) : "";
           if (raw.trim()) {
@@ -3453,7 +3469,7 @@ function resolveImageBlocks(binDataMap, blocks, warnings, sweepUnreferenced) {
   resolveImageSentinels(blocks, renamed);
   return images;
 }
-function collectHwp5BinData(fileIndex) {
+function collectHwp5BinData(fileIndex, readStream) {
   const binDataMap = /* @__PURE__ */ new Map();
   if (fileIndex) {
     for (const entry of fileIndex) {
@@ -3461,35 +3477,37 @@ function collectHwp5BinData(fileIndex) {
       const match = entry.name.match(BIN_ENTRY_RE);
       if (!match) continue;
       const idx = parseInt(match[1], 16);
-      const data = normalizeBinPayload(Buffer.from(entry.content));
+      const raw = readStream ? readStream(`/BinData/${entry.name}`) : Buffer.from(entry.content);
+      if (!raw) continue;
+      const data = normalizeBinPayload(raw);
       binDataMap.set(idx, { data, name: entry.name });
     }
   }
   return binDataMap;
 }
-function collectHwp5BinDataLenient(lcfb) {
+function collectHwp5BinDataLenient(lcfb, readStream = lcfb.findStream) {
   const binDataMap = /* @__PURE__ */ new Map();
   const binRe = /^BIN([0-9A-Fa-f]{4,8})(?:\.|$)/;
   for (const e of lcfb.entries()) {
     const match = e.name.match(binRe);
     if (!match) continue;
     const idx = parseInt(match[1], 16);
-    const raw = lcfb.findStream(e.name);
+    const raw = readStream(e.name);
     if (!raw) continue;
     binDataMap.set(idx, { data: normalizeBinPayload(raw), name: e.name });
   }
   return binDataMap;
 }
-function extractHwp5Images(fileIndex, blocks, warnings, sweepUnreferenced) {
-  const binDataMap = collectHwp5BinData(fileIndex);
+function extractHwp5Images(fileIndex, blocks, warnings, sweepUnreferenced, readStream) {
+  const binDataMap = collectHwp5BinData(fileIndex, readStream);
   if (binDataMap.size === 0) {
     resolveImageSentinels(blocks, /* @__PURE__ */ new Map());
     return [];
   }
   return resolveImageBlocks(binDataMap, blocks, warnings, sweepUnreferenced);
 }
-function extractHwp5ImagesLenient(lcfb, blocks, warnings, sweepUnreferenced) {
-  const binDataMap = collectHwp5BinDataLenient(lcfb);
+function extractHwp5ImagesLenient(lcfb, blocks, warnings, sweepUnreferenced, readStream) {
+  const binDataMap = collectHwp5BinDataLenient(lcfb, readStream);
   if (binDataMap.size === 0) {
     resolveImageSentinels(blocks, /* @__PURE__ */ new Map());
     return [];
@@ -3746,8 +3764,11 @@ async function resolveSectionPaths(zip) {
 function parseSectionPathsFromManifest(xml) {
   const parser = createXmlParser();
   const doc = parser.parseFromString(stripDtd(xml), "text/xml");
-  const items = doc.getElementsByTagName("opf:item");
-  const spine = doc.getElementsByTagName("opf:itemref");
+  const opf = "http://www.idpf.org/2007/opf/";
+  const namespacedItems = doc.getElementsByTagNameNS(opf, "item");
+  const namespacedSpine = doc.getElementsByTagNameNS(opf, "itemref");
+  const items = namespacedItems.length ? namespacedItems : doc.getElementsByTagName("opf:item");
+  const spine = namespacedSpine.length ? namespacedSpine : doc.getElementsByTagName("opf:itemref");
   const idToHref = /* @__PURE__ */ new Map();
   for (let i = 0; i < items.length; i++) {
     const item = items[i];
@@ -3954,7 +3975,7 @@ async function parseHwpxDocument(buffer, options) {
   applyPageText(blocks, shared);
   const images = await extractImagesFromZip(zip, blocks, decompressed, warnings, !options?.pages, shared.track.deletedImageRefs);
   const shown = options?.layoutTables === "keep" ? blocks : unframeLayoutTables(blocks, !!options?.keepTrailingEmptyCols);
-  detectHwpxHeadings(shown, styleMap);
+  detectHwpxHeadings(shown, styleMap, shared.tocTexts);
   const outline = shown.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
   const markdown = blocksToMarkdown(shown);
   return { markdown, blocks: shown, metadata, outline: outline.length > 0 ? outline : void 0, warnings: warnings.length > 0 ? warnings : void 0, images: images.length > 0 ? images : void 0 };
@@ -5361,6 +5382,7 @@ function createHwp5DocState() {
     outlineNumberingId: 0,
     autoCounters: /* @__PURE__ */ new Map(),
     headerTexts: /* @__PURE__ */ new Set(),
+    tocTexts: /* @__PURE__ */ new Set(),
     headerBlocks: [],
     footerBlocks: []
   };
@@ -5506,6 +5528,12 @@ function parseParagraph(records, start, end, ctx) {
       if (bullet && bullet.char !== "\uFFFF") headMarker = bullet.char;
     }
   }
+  if (toc.tocEntry) {
+    for (const line of text.split("\n")) if (line.trim()) {
+      ctx.doc.tocTexts.add(line.trim());
+      if (headMarker) ctx.doc.tocTexts.add(`${headMarker} ${line.trim()}`);
+    }
+  }
   return emitParagraphBlocks({
     text,
     headMarker,
@@ -5970,7 +5998,7 @@ function readHwp5DocInfoRecords(c) {
   }
 }
 function readHwp5SectionStreams(c) {
-  return c.distribution ? c.cfb ? findViewTextSections(c.cfb, c.compressed) : findViewTextSectionsLenient(c.lenientCfb, c.compressed) : c.encrypted ? findSectionsVia(c.findStream) : c.cfb ? findSections(c.cfb) : findSectionsLenient(c.lenientCfb, c.compressed);
+  return c.distribution ? c.cfb ? findViewTextSections(c.cfb, c.compressed) : findViewTextSectionsLenient(c.lenientCfb, c.compressed) : c.encrypted ? findSectionsVia(c.findStream) : c.cfb ? findSections(c.cfb) : findSectionsLenient(c.lenientCfb);
 }
 function readHwp5SectionRecords(c, sections, warnings) {
   const out = [];
@@ -5990,7 +6018,8 @@ function readHwp5SectionRecords(c, sections, warnings) {
   return out;
 }
 function readHwp5BinData(c) {
-  return c.cfb ? collectHwp5BinData(c.cfb.FileIndex) : collectHwp5BinDataLenient(c.lenientCfb);
+  const readStream = c.encrypted ? c.findStream : void 0;
+  return c.cfb ? collectHwp5BinData(c.cfb.FileIndex, readStream) : collectHwp5BinDataLenient(c.lenientCfb, readStream);
 }
 function parseHwp5Document(buffer, options) {
   const c = openHwp5Container(buffer, options);
@@ -6055,7 +6084,7 @@ function parseHwp5Document(buffer, options) {
     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);
+  const images = cfb ? extractHwp5Images(cfb.FileIndex, blocks, warnings, !pageFilter, encrypted ? findStream : void 0) : extractHwp5ImagesLenient(lenientCfb, blocks, warnings, !pageFilter, encrypted ? findStream : void 0);
   let flatBlocks = flattenLayoutTables(blocks);
   if (options?.layoutTables !== "keep") flatBlocks = unframeLayoutTables(flatBlocks, !!options?.keepTrailingEmptyCols);
   if (options?.dedupeRunningHeaders) {
@@ -6065,7 +6094,7 @@ function parseHwp5Document(buffer, options) {
     flatBlocks = deduped;
   }
   if (docInfo) {
-    detectHwp5Headings(flatBlocks, docInfo);
+    detectHwp5Headings(flatBlocks, docInfo, doc.tocTexts);
   }
   const outline = flatBlocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
   let markdown = blocksToMarkdown(flatBlocks);
@@ -6098,7 +6127,7 @@ function parseDocInfoFromStream(raw, compressed) {
     return null;
   }
 }
-function detectHwp5Headings(blocks, docInfo) {
+function detectHwp5Headings(blocks, docInfo, tocTexts) {
   let baseFontSize = 0;
   const sizeFreq = /* @__PURE__ */ new Map();
   for (const b of blocks) {
@@ -6128,7 +6157,7 @@ function detectHwp5Headings(blocks, docInfo) {
   if (baseFontSize <= 0) return;
   for (const block of blocks) {
     if (block.type === "heading") continue;
-    if (block.type !== "paragraph" || !block.text || TOC_ENTRY_BLOCKS.has(block)) continue;
+    if (block.type !== "paragraph" || !block.text || TOC_ENTRY_BLOCKS.has(block) || tocTexts?.has(block.text.trim())) continue;
     const text = block.text.trim();
     if (text.length === 0 || text.length > 200) continue;
     if (/^\d+$/.test(text)) continue;
@@ -6262,16 +6291,15 @@ function findSections(cfb) {
   }
   return sections.sort((a, b) => a.idx - b.idx).map((s) => s.content);
 }
-function findSectionsLenient(lcfb, compressed) {
+function findSectionsLenient(lcfb) {
   const sections = [];
-  let totalDecompressed = 0;
+  let totalBytes = 0;
   for (let i = 0; i < MAX_SECTIONS; i++) {
     const raw = lcfb.findStream(`/BodyText/Section${i}`) ?? lcfb.findStream(`Section${i}`);
     if (!raw) break;
-    const content = compressed ? decompressStream(raw) : raw;
-    totalDecompressed += content.length;
-    if (totalDecompressed > MAX_TOTAL_DECOMPRESS) throw new KordocError("\uCD1D \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (decompression bomb \uC758\uC2EC)");
-    sections.push({ idx: i, content });
+    totalBytes += raw.length;
+    if (totalBytes > MAX_TOTAL_DECOMPRESS) throw new KordocError("\uCD1D \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (decompression bomb \uC758\uC2EC)");
+    sections.push({ idx: i, content: raw });
   }
   if (sections.length === 0) {
     for (const e of lcfb.entries()) {
@@ -6280,10 +6308,9 @@ function findSectionsLenient(lcfb, compressed) {
         const idx = parseInt(e.name.replace("Section", ""), 10) || 0;
         const raw = lcfb.findStream(e.name);
         if (raw) {
-          const content = compressed ? decompressStream(raw) : raw;
-          totalDecompressed += content.length;
-          if (totalDecompressed > MAX_TOTAL_DECOMPRESS) throw new KordocError("\uCD1D \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (decompression bomb \uC758\uC2EC)");
-          sections.push({ idx, content });
+          totalBytes += raw.length;
+          if (totalBytes > MAX_TOTAL_DECOMPRESS) throw new KordocError("\uCD1D \uC555\uCD95 \uD574\uC81C \uD06C\uAE30 \uCD08\uACFC (decompression bomb \uC758\uC2EC)");
+          sections.push({ idx, content: raw });
         }
       }
     }
@@ -19387,7 +19414,7 @@ function collectRichText(root) {
       const el = children[i];
       const local = el.localName || el.tagName?.replace(/^[^:]+:/, "") || "";
       if (local === "rPh") continue;
-      if (local === "t") out += escapeLiteralTags(el.textContent ?? "");
+      if (local === "t") out += escapeLiteralTags(escapeLiteralDollar(el.textContent ?? ""));
       else walk(el);
     }
   };
@@ -20011,14 +20038,14 @@ function extractSheetCells(records, bofIndex, sst, convertNum) {
         const h = readCellHeader(rec.data);
         if (h && rec.data.length >= 10) {
           const isst = rec.data.readUInt32LE(6);
-          cells.push({ row: h.row, col: h.col, value: escapeLiteralTags(sst[isst] ?? "") });
+          cells.push({ row: h.row, col: h.col, value: escapeLiteralTags(escapeLiteralDollar(sst[isst] ?? "")) });
         }
         break;
       }
       case OP_LABEL: {
         const h = readCellHeader(rec.data);
         if (h) {
-          cells.push({ row: h.row, col: h.col, value: escapeLiteralTags(decodeLabelString(rec.data)) });
+          cells.push({ row: h.row, col: h.col, value: escapeLiteralTags(escapeLiteralDollar(decodeLabelString(rec.data))) });
         }
         break;
       }
@@ -20034,7 +20061,7 @@ function extractSheetCells(records, bofIndex, sst, convertNum) {
               cells.push({
                 row: h.row,
                 col: h.col,
-                value: escapeLiteralTags(decodeFormulaStringRecord(next.data))
+                value: escapeLiteralTags(escapeLiteralDollar(decodeFormulaStringRecord(next.data)))
               });
               i = j;
             } else {
@@ -20746,6 +20773,7 @@ function isDisplayMath(el) {
 // src/docx/parser.ts
 var MAX_DECOMPRESS_SIZE5 = unzipLimitBytes(100 * 1024 * 1024);
 var MEDIA_PART_RE = /^word\/(?:media|embeddings)\//;
+var NEVER_READ_RE = /^word\/(?:embeddings\/.*|media\/.+\.(?:mp4|m4v|mov|avi|wmv|mpe?g|mkv|webm|mp3|wav|m4a|wma|aac|ogg|flac|mid|bin))$/i;
 function matchesLocal(el, localName3) {
   return el.localName === localName3 || (el.tagName?.endsWith(`:${localName3}`) ?? false);
 }
@@ -20937,7 +20965,7 @@ function parseFootnotes(xml) {
       const runs = findElements(p, "r");
       for (const r of runs) {
         const tElements = getChildElements(r, "t");
-        for (const t of tElements) texts.push(escapeLiteralTags(t.textContent ?? ""));
+        for (const t of tElements) texts.push(escapeLiteralTags(escapeLiteralDollar(t.textContent ?? "")));
       }
     }
     notes.set(id, texts.join("").trim());
@@ -20968,7 +20996,7 @@ function collectOmmlRoots(p) {
 function extractRun(r) {
   let text = "";
   for (const el of effectiveChildElements(r)) {
-    if (matchesLocal(el, "t")) text += escapeLiteralTags(el.textContent ?? "");
+    if (matchesLocal(el, "t")) text += escapeLiteralTags(escapeLiteralDollar(el.textContent ?? ""));
     else if (matchesLocal(el, "br") || matchesLocal(el, "cr")) text += "\n";
     else if (matchesLocal(el, "tab")) text += " ";
     else if (matchesLocal(el, "sym")) text += symbolChar(getAttr(el, "font"), getAttr(el, "char"));
@@ -21264,7 +21292,7 @@ async function buildImageMap(zip, rels, doc, warnings, readBytes) {
       let data = null;
       if (readBytes) data = await imgFile.async("uint8array");
       imgIdx++;
-      const ext = imgPath.split(".").pop()?.toLowerCase() ?? "png";
+      const ext = partExtension(imgPath);
       const mimeMap = {
         png: "image/png",
         jpg: "image/jpeg",
@@ -21363,7 +21391,7 @@ async function parseHeaderFooterBlocks(zip, doc, rels, kind, styles, numbering,
 }
 async function parseDocxDocument(buffer, options) {
   const readImages = options?.images !== false;
-  precheckZipSize(buffer, MAX_DECOMPRESS_SIZE5, void 0, { re: MEDIA_PART_RE, skip: !readImages });
+  precheckZipSize(buffer, MAX_DECOMPRESS_SIZE5, void 0, { re: MEDIA_PART_RE, skip: !readImages, never: NEVER_READ_RE });
   const zip = await JSZip5.loadAsync(buffer);
   const warnings = [];
   const docFile = zip.file("word/document.xml");
@@ -21484,6 +21512,7 @@ import JSZip6 from "jszip";
 import { DOMParser as DOMParser4 } from "@xmldom/xmldom";
 var MAX_DECOMPRESS_SIZE6 = unzipLimitBytes(100 * 1024 * 1024);
 var MEDIA_PART_RE2 = /^ppt\/(?:media|embeddings)\//;
+var NEVER_READ_RE2 = /^ppt\/(?:embeddings\/.*|media\/.+\.(?:mp4|m4v|mov|avi|wmv|mpe?g|mkv|webm|mp3|wav|m4a|wma|aac|ogg|flac|mid|bin))$/i;
 var MAX_ZIP_ENTRIES2 = 1e4;
 var MAX_SLIDES = 2e3;
 var SKIP_PLACEHOLDERS = /* @__PURE__ */ new Set(["sldNum", "dt", "ftr", "hdr", "sldImg"]);
@@ -21580,12 +21609,14 @@ function paragraphText(p) {
     const name = localName(child);
     if (name === "r" || name === "fld") {
       const t = findChildByLocalName(child, "t");
-      if (t) text += escapeLiteralTags(escapeLiteralDollar(rawTextContent(t)));
+      const rPr = findChildByLocalName(child, "rPr");
+      const baseline = Number(rPr ? getAttr2(rPr, "baseline") ?? 0 : 0);
+      if (t) text += wrapScript(escapeLiteralTags(escapeLiteralDollar(rawTextContent(t))), baseline > 0 ? "sup" : baseline < 0 ? "sub" : null);
     } else if (name === "br") {
       text += " ";
     }
   }
-  return text.replace(/\s+/g, " ").trim();
+  return tidyScriptTags(text.replace(/\s+/g, " ").trim());
 }
 function bodyParagraphs(txBody) {
   const out = [];
@@ -21659,7 +21690,7 @@ async function pictureBlock(pic, ctx) {
   if (!filename) {
     const file = ctx.zip.file(rel.target);
     if (!file) return null;
-    const ext = rel.target.split(".").pop()?.toLowerCase() ?? "png";
+    const ext = partExtension(rel.target);
     filename = `image_${String(ctx.imageNames.size + 1).padStart(3, "0")}.${ext}`;
     try {
       if (ctx.options?.images !== false) {
@@ -21737,7 +21768,7 @@ async function notesText(zip, rels) {
   return texts.join(" ");
 }
 async function parsePptxDocument(buffer, options) {
-  precheckZipSize(buffer, MAX_DECOMPRESS_SIZE6, MAX_ZIP_ENTRIES2, { re: MEDIA_PART_RE2, skip: options?.images === false });
+  precheckZipSize(buffer, MAX_DECOMPRESS_SIZE6, MAX_ZIP_ENTRIES2, { re: MEDIA_PART_RE2, skip: options?.images === false, never: NEVER_READ_RE2 });
   const zip = await JSZip6.loadAsync(buffer);
   if (!zip.file("ppt/presentation.xml")) {
     throw new KordocError("\uC720\uD6A8\uD558\uC9C0 \uC54A\uC740 PPTX \uD30C\uC77C: ppt/presentation.xml\uC774 \uC5C6\uC2B5\uB2C8\uB2E4");
@@ -21747,8 +21778,10 @@ async function parsePptxDocument(buffer, options) {
   const imageNames = /* @__PURE__ */ new Map();
   const images = [];
   const slides = await slideParts(zip);
+  const pageFilter = options?.pages ? parsePageRange(options.pages, slides.length) : null;
   for (let i = 0; i < slides.length; i++) {
     const page = i + 1;
+    if (pageFilter && !pageFilter.has(page)) continue;
     const root = parseXml3(await zip.file(slides[i]).async("text")).documentElement;
     if (getAttr2(root, "show") === "0") {
       warnings.push({ page, code: "HIDDEN_TEXT_FILTERED", message: `PPTX \uC228\uAE34 \uC2AC\uB77C\uC774\uB4DC ${page}\uCABD\uC744 \uAC74\uB108\uB701\uB2C8\uB2E4` });
@@ -28647,7 +28680,7 @@ function sanitizeText(text) {
   return result;
 }
 function normForMatch(text) {
-  return sanitizeText(text.replace(/\\\$/g, "$")).replace(/\s+/g, " ").trim();
+  return sanitizeText(text.replace(/\\\$/g, "$").replace(/\\(?=<\/?(?:u|sup|sub)>)/g, "")).replace(/\s+/g, " ").trim();
 }
 function unescapeGfm(text) {
   return text.replace(/\\([~*_`|#<$\\])/g, "$1");
@@ -29279,6 +29312,7 @@ var ImageRegistry = class {
   supplied;
   byUrl = /* @__PURE__ */ new Map();
   ids = /* @__PURE__ */ new Set();
+  names = /* @__PURE__ */ new Set();
   parts = [];
   picSeq = 0;
   binSeq = 0;
@@ -29299,8 +29333,12 @@ var ImageRegistry = class {
       let n = 1;
       while (this.ids.has(itemId)) itemId = `${m[1].replace(/[^A-Za-z0-9_]/g, "_")}_${n++}`;
       this.ids.add(itemId);
+      let name = `BinData/${url}`;
+      let suffix = 1;
+      while (this.names.has(name)) name = `BinData/${m[1]}_${suffix++}.${m[2]}`;
+      this.names.add(name);
       part = {
-        name: `BinData/${url}`,
+        name,
         itemId,
         mime,
         data: m[2].toLowerCase() === "bmp" ? PLACEHOLDER_BMP : PLACEHOLDER_PNG
@@ -29329,8 +29367,9 @@ var ImageRegistry = class {
     const base = safe && !url.includes("..") ? safe[1] : `image${++this.binSeq}`;
     let itemId = base.replace(/[^A-Za-z0-9_]/g, "_");
     let n = 1;
-    while (this.ids.has(itemId)) itemId = `${base.replace(/[^A-Za-z0-9_]/g, "_")}_${n++}`;
+    while (this.ids.has(itemId) || this.names.has(`BinData/${itemId}.${sniffed.ext}`)) itemId = `${base.replace(/[^A-Za-z0-9_]/g, "_")}_${n++}`;
     this.ids.add(itemId);
+    this.names.add(`BinData/${itemId}.${sniffed.ext}`);
     const dim = probeImageSize(bytes);
     let wHU, hHU;
     if (dim && dim.w > 0 && dim.h > 0) {
@@ -34306,30 +34345,11 @@ async function resolveSectionEntryNames(zip) {
     const f = zip.file(mp);
     if (!f) continue;
     const xml = await f.async("text");
-    const paths = sectionPathsFromManifest(xml).filter((p) => zip.file(p) !== null);
+    const paths = parseSectionPathsFromManifest(xml).filter((p) => zip.file(p) !== null);
     if (paths.length > 0) return paths;
   }
   return Object.keys(zip.files).filter((n) => /[Ss]ection\d+\.xml$/.test(n)).sort(compareSectionPaths);
 }
-function sectionPathsFromManifest(xml) {
-  const attr = (tag, name) => {
-    const m = tag.match(new RegExp(`(?:^|\\s)${name}\\s*=\\s*(?:"([^"]*)"|'([^']*)')`));
-    return m ? m[1] ?? m[2] : "";
-  };
-  const idToHref = /* @__PURE__ */ new Map();
-  for (const m of xml.matchAll(/<opf:item(\s(?:"[^"]*"|'[^']*'|[^>"'])*?)\/?>/g)) {
-    const id = attr(m[1], "id");
-    const href = normalizeSectionHref(attr(m[1], "href"));
-    if (id && href) idToHref.set(id, href);
-  }
-  const ordered = [];
-  for (const m of xml.matchAll(/<opf:itemref(\s(?:"[^"]*"|'[^']*'|[^>"'])*?)\/?>/g)) {
-    const href = idToHref.get(attr(m[1], "idref"));
-    if (href) ordered.push(href);
-  }
-  if (ordered.length > 0) return ordered;
-  return Array.from(idToHref.values()).sort(compareSectionPaths);
-}
 
 // src/roundtrip/hwp5-patch.ts
 import { deflateRawSync as deflateRawSync2 } from "zlib";
@@ -35555,6 +35575,10 @@ async function patchHwpx(original, editedMarkdown, options) {
     const parsed = await parseHwpxDocument(u8ToArrayBuffer(original), { layoutTables: "keep" });
     origBlocks = parsed.blocks;
     const visual = await parseHwpxDocument(u8ToArrayBuffer(original));
+    if (parsed.markdown !== visual.markdown && editedMarkdown === visual.markdown) {
+      const verification2 = options?.verify === false ? void 0 : diffUnitLists(splitMarkdownUnits(visual.markdown), splitMarkdownUnits(editedMarkdown));
+      return { success: true, data: new Uint8Array(original), applied: 0, skipped, verification: verification2 };
+    }
     if (editedFromVisual(parsed.markdown, visual.markdown, editedMarkdown)) return { success: false, applied: 0, skipped, error: LAYOUT_MODE_MISMATCH };
   } catch (err) {
     return { success: false, applied: 0, skipped, error: `\uC6D0\uBCF8 HWPX \uD30C\uC2F1 \uC2E4\uD328: ${err instanceof Error ? err.message : String(err)}` };
@@ -39192,7 +39216,7 @@ var CHAR_CTRL_1SLOT = /* @__PURE__ */ new Set(["lineBreak", "hyphen", "nbSpace",
 function pushFillers(chars, n, prId) {
   for (let i = 0; i < n; i++) chars.push({ ch: "", prId });
 }
-function pushTextSlots(t, chars, prId, depth) {
+function pushTextSlots(t, chars, prId, depth, del = { deleted: 0 }) {
   if (depth > 32) return;
   const kids = t.childNodes;
   if (!kids) return;
@@ -39200,19 +39224,24 @@ function pushTextSlots(t, chars, prId, depth) {
     const c = kids[i];
     if (c.nodeType === 3 || c.nodeType === 4) {
       for (const cp of c.textContent ?? "") {
-        chars.push({ ch: cp, prId });
+        chars.push({ ch: del.deleted ? "" : cp, prId });
         if (cp.length === 2) chars.push({ ch: "", prId });
       }
     } else if (c.nodeType === 1) {
       const el = c;
       const tag = ln(el);
-      if (tag === "tab") {
-        chars.push({ ch: "", prId, tab: true, tabW: num2(el, "width") });
-        pushFillers(chars, 7, prId);
+      if (tag === "deleteBegin" || tag === "deleteEnd") {
+        del.deleted = deletionDelta(el, del.deleted);
+      } else if (tag === "tab") {
+        if (del.deleted) pushFillers(chars, 8, prId);
+        else {
+          chars.push({ ch: "", prId, tab: true, tabW: num2(el, "width") });
+          pushFillers(chars, 7, prId);
+        }
       } else if (CHAR_CTRL_1SLOT.has(tag)) {
-        chars.push(tag === "nbSpace" ? { ch: " ", prId, nb: true } : { ch: tag === "fwSpace" ? " " : "", prId });
+        chars.push(del.deleted ? { ch: "", prId } : tag === "nbSpace" ? { ch: " ", prId, nb: true } : { ch: tag === "fwSpace" ? " " : "", prId });
       } else {
-        pushTextSlots(el, chars, prId, depth + 1);
+        pushTextSlots(el, chars, prId, depth + 1, del);
       }
     }
   }
@@ -39237,7 +39266,7 @@ function prepareDeletedRanges(root, depth = 0) {
     deletedStarts.set(p, deleted);
     const scan = (el, xmlDepth) => {
       if (xmlDepth > 64) return;
-      if (ln(el) === "ctrl") for (const ch of elements(el)) scan(ch, xmlDepth + 1);
+      if (ln(el) === "ctrl" || ln(el) === "t") for (const ch of elements(el)) scan(ch, xmlDepth + 1);
       else {
         if (ln(el) === "deleteBegin" || ln(el) === "deleteEnd") deleted = deletionDelta(el, deleted);
         deletedControls.set(el, deleted > 0);
@@ -39269,9 +39298,9 @@ function buildPara(p) {
       for (const ch of elements(runEl)) {
         const cn = ln(ch);
         if (cn === "t") {
-          const start = chars.length;
-          pushTextSlots(ch, chars, prId, 0);
-          if (deleted) for (let i = start; i < chars.length; i++) chars[i] = { ch: "", prId };
+          const del = { deleted };
+          pushTextSlots(ch, chars, prId, 0, del);
+          deleted = del.deleted;
         } else if (OBJ_TAGS.has(cn)) {
           const sz = findChildByLocalName(ch, "sz");
           const pos = findChildByLocalName(ch, "pos");
@@ -39793,14 +39822,15 @@ function shapeGeometry(o, boxX, boxY, defs, images, warn, local = false) {
   const color = esc(line?.getAttribute("color") || "#000000");
   const strokeW = style === "NONE" ? 0 : Math.max(0.2, num2(line, "width", 33) / 100 * 2.834645);
   const dash = /DASH|DOT/.test(style) ? ` stroke-dasharray="${style.includes("DOT") ? "1,1.5" : "3,1.5"}"` : "";
-  const stroke = strokeW ? ` stroke="${color}" stroke-width="${strokeW.toFixed(2)}"${dash}` : "";
+  const keepWidth = local ? ` vector-effect="non-scaling-stroke"` : "";
+  const stroke = strokeW ? ` stroke="${color}" stroke-width="${strokeW.toFixed(2)}"${dash}${keepWidth}` : "";
   const fill = ` fill="${shapeFill(el, defs, images, warn)}"`;
   let svg = "";
   if (o.tag === "rect") svg = `<rect x="${pt(x)}" y="${pt(y)}" width="${pt(w)}" height="${pt(h)}"${fill}${stroke}/>`;
   else if (o.tag === "ellipse") svg = `<ellipse cx="${pt(x + w / 2)}" cy="${pt(y + h / 2)}" rx="${pt(w / 2)}" ry="${pt(h / 2)}"${fill}${stroke}/>`;
   else if (o.tag === "line") {
     const s = findChildByLocalName(el, "startPt"), e = findChildByLocalName(el, "endPt");
-    svg = `<line x1="${pt(x + num2(s, "x") * sx)}" y1="${pt(y + num2(s, "y") * sy)}" x2="${pt(x + num2(e, "x") * sx)}" y2="${pt(y + num2(e, "y") * sy)}" stroke="${color}" stroke-width="${(strokeW || 0.3).toFixed(2)}"${dash}/>`;
+    svg = `<line x1="${pt(x + num2(s, "x") * sx)}" y1="${pt(y + num2(s, "y") * sy)}" x2="${pt(x + num2(e, "x") * sx)}" y2="${pt(y + num2(e, "y") * sy)}" stroke="${color}" stroke-width="${(strokeW || 0.3).toFixed(2)}"${dash}${keepWidth}/>`;
   } else if (o.tag === "polygon" || o.tag === "curv") {
     const pts = elements(el).filter((c) => ln(c) === "pt").map((c) => `${pt(x + num2(c, "x") * sx)},${pt(y + num2(c, "y") * sy)}`);
     if (pts.length >= 2) svg = `<polygon points="${pts.join(" ")}"${fill}${stroke}/>`;
@@ -40156,15 +40186,15 @@ function drawGroupByMatrix(o, x, y, baseV, areaW, ctx, depth) {
     emit(ctx, `<g transform="matrix(${m.slice(0, 4).map((v) => Math.round(v * 1e6) / 1e6).join(" ")} ${pt2(m[4])} ${pt2(m[5])})">`);
     const sub = { el: leaf, tag, index: 0, inline: true, width: w, height: h, omL: 0, omR: 0 };
     if (tag === "pic") drawPicFrame(leaf, 0, 0, w, h, "", ctx);
-    else if (SHAPE_TAGS.has(tag)) drawShape(sub, 0, 0, ctx, depth + 1, true);
+    else if (SHAPE_TAGS.has(tag)) drawShape(sub, 0, 0, ctx, depth + 1, [Math.hypot(m[0], m[1]) || 1, Math.hypot(m[2], m[3]) || 1]);
     else drawObject(sub, 0, 0, baseV, areaW, ctx, depth + 1);
     emit(ctx, "</g>");
     ctx.regionRotations.pop();
   }
   return true;
 }
-function drawShape(o, x, y, ctx, depth, local = false) {
-  const geometry = shapeGeometry(o, x, y, ctx.defs, ctx.images, (key, msg2) => warnOnce(ctx, key, msg2), local);
+function drawShape(o, x, y, ctx, depth, groupScale) {
+  const geometry = shapeGeometry(o, x, y, ctx.defs, ctx.images, (key, msg2) => warnOnce(ctx, key, msg2), !!groupScale);
   const b = geometry;
   if (geometry.rotation) ctx.regionRotations.push(geometry.rotation);
   ctx.stats.shapes++;
@@ -40176,7 +40206,17 @@ function drawShape(o, x, y, ctx, depth, local = false) {
   const dt = findChildByLocalName(o.el, "drawText");
   const sub = dt ? findChildByLocalName(dt, "subList") : null;
   if (sub) {
-    for (const p of elements(sub)) if (ln(p) === "p") drawPara(p, x, y, geometry.w, ctx, depth + 1);
+    const [sx, sy] = groupScale ?? [1, 1];
+    const unscale = sx !== 1 || sy !== 1;
+    if (unscale) {
+      emit(ctx, `<g transform="scale(${1 / sx} ${1 / sy})">`);
+      ctx.regionRotations.push([1 / sx, 0, 0, 1 / sy, 0, 0]);
+    }
+    for (const p of elements(sub)) if (ln(p) === "p") drawPara(p, geometry.x * sx, geometry.y * sy, geometry.w * sx, ctx, depth + 1);
+    if (unscale) {
+      ctx.regionRotations.pop();
+      emit(ctx, "</g>");
+    }
   }
   if (geometry.rotation) ctx.regionRotations.pop();
   if (geometry.transform) emit(ctx, "</g>");
@@ -41440,7 +41480,7 @@ async function parse(input, options) {
       buffer = toArrayBuffer(buf);
     } catch (err) {
       const msg2 = err instanceof Error && "code" in err && err.code === "ENOENT" ? `\uD30C\uC77C\uC744 \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4: ${input}` : `\uD30C\uC77C \uC77D\uAE30 \uC2E4\uD328: ${input}`;
-      return { success: false, fileType: "unknown", error: msg2, code: "PARSE_ERROR" };
+      return { success: false, fileType: "unknown", error: msg2, code: classifyError(err) };
     }
   } else if (Buffer.isBuffer(input)) {
     buffer = toArrayBuffer(input);
@@ -41458,9 +41498,12 @@ async function parse(input, options) {
   }
   if (result.success && opts?.classifyTables) classifyTableTree(result.blocks);
   let out = result;
-  if (result.success && !result.pages) {
-    const pages = blocksToPages(result.blocks);
-    if (pages) out = { ...result, pages };
+  if (result.success && opts?.pages && result.pageCount && parsePageRange(opts.pages, result.pageCount).size === 0) {
+    out = { ...result, warnings: [...result.warnings ?? [], { code: "PARTIAL_PARSE", message: `\uC694\uCCAD\uD55C \uCABD \uBC94\uC704(${opts.pages})\uAC00 \uBB38\uC11C \uCABD(1~${result.pageCount})\uACFC \uACB9\uCE58\uC9C0 \uC54A\uC2B5\uB2C8\uB2E4` }] };
+  }
+  if (out.success && !out.pages) {
+    const pages = blocksToPages(out.blocks);
+    if (pages) out = { ...out, pages };
   }
   if (out.success && opts?.plain) {
     out = { ...out, markdown: toPlainMarkdown(out.markdown), ...out.pages ? { pages: out.pages.map((p) => ({ ...p, markdown: toPlainMarkdown(p.markdown) })) } : {} };
@@ -41512,7 +41555,7 @@ function scriptsOff(r, off) {
 }
 async function parseImage(buffer, options) {
   try {
-    const { parseImageDocument } = await import("./image-ocr-IMYXOHHV.js");
+    const { parseImageDocument } = await import("./image-ocr-4BGX2EJJ.js");
     const { blocks, warnings } = await parseImageDocument(buffer, options);
     stripScriptTags({ blocks });
     return {
@@ -41572,7 +41615,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-M3T2EWUN.js");
+    const mod = await import("./parser-GFDGAVLV.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {
@@ -41617,7 +41660,7 @@ async function parseDocx(buffer, options) {
 async function parsePptx(buffer, options) {
   try {
     const { markdown, blocks, metadata, outline, warnings, images } = await parsePptxDocument(buffer, options);
-    return { success: true, fileType: "pptx", markdown, blocks, metadata, outline, warnings, images: images?.length ? images : void 0, pageCount: metadata?.pageCount };
+    return scriptsOff({ success: true, fileType: "pptx", markdown, blocks, metadata, outline, warnings, images: images?.length ? images : void 0, pageCount: metadata?.pageCount }, options?.scriptTags === false);
   } catch (err) {
     return { success: false, fileType: "pptx", error: sanitizeError(err), code: classifyError(err) };
   }