kordoc 4.1.0 → 4.2.0

package/dist/index.d.cts

@@ -134,8 +134,12 @@ interface ParseOptions {
      * PDF: 정확한 페이지 단위. HWP/HWPX: 섹션 단위 근사치.
      */
     pages?: number[] | string;
-    /** 이미지 기반 PDF용 OCR 프로바이더 (선택) */
-    ocr?: OcrProvider;
+    /** 이미지 기반 PDF OCR (선택).
+     *  - `true`: 내장 엔진(PP-OCRv5 korean, ~18MB 자동 다운로드)으로 OCR 필요 판정
+     *    페이지만 인식 (스캔 페이지·글꼴 매핑 깨진 페이지). 정상 페이지는 파싱 결과 유지.
+     *  - `"force"`: 전 페이지를 내장 엔진으로 강제 OCR.
+     *  - 함수: 사용자 제공 OcrProvider (Claude Vision·Tesseract 등) — 판정은 `true`와 동일. */
+    ocr?: boolean | "force" | OcrProvider;
     /** 진행률 콜백 — current: 현재 페이지/섹션, total: 전체 수 */
     onProgress?: (current: number, total: number) => void;
     /** PDF 머리글/바닥글 자동 제거 */
@@ -190,7 +194,7 @@ interface ParseWarning {
     /** 구조화된 경고 코드 */
     code: WarningCode;
 }
-type WarningCode = "SKIPPED_IMAGE" | "SKIPPED_OLE" | "TRUNCATED_TABLE" | "OCR_FALLBACK" | "UNSUPPORTED_ELEMENT" | "BROKEN_ZIP_RECOVERY" | "HIDDEN_TEXT_FILTERED" | "MALFORMED_XML" | "PARTIAL_PARSE" | "LENIENT_CFB_RECOVERY" | "NEEDS_OCR" | "COM_EMPTY" | "DRM_COM_FALLBACK";
+type WarningCode = "SKIPPED_IMAGE" | "SKIPPED_OLE" | "TRUNCATED_TABLE" | "OCR_FALLBACK" | "UNSUPPORTED_ELEMENT" | "BROKEN_ZIP_RECOVERY" | "HIDDEN_TEXT_FILTERED" | "MALFORMED_XML" | "PARTIAL_PARSE" | "LENIENT_CFB_RECOVERY" | "NEEDS_OCR" | "OCR_FAILED" | "OCR_APPLIED" | "COM_EMPTY" | "DRM_COM_FALLBACK";
 /** 문서 구조 (헤딩 트리) */
 interface OutlineItem {
     level: number;

package/dist/index.d.ts

@@ -134,8 +134,12 @@ interface ParseOptions {
      * PDF: 정확한 페이지 단위. HWP/HWPX: 섹션 단위 근사치.
      */
     pages?: number[] | string;
-    /** 이미지 기반 PDF용 OCR 프로바이더 (선택) */
-    ocr?: OcrProvider;
+    /** 이미지 기반 PDF OCR (선택).
+     *  - `true`: 내장 엔진(PP-OCRv5 korean, ~18MB 자동 다운로드)으로 OCR 필요 판정
+     *    페이지만 인식 (스캔 페이지·글꼴 매핑 깨진 페이지). 정상 페이지는 파싱 결과 유지.
+     *  - `"force"`: 전 페이지를 내장 엔진으로 강제 OCR.
+     *  - 함수: 사용자 제공 OcrProvider (Claude Vision·Tesseract 등) — 판정은 `true`와 동일. */
+    ocr?: boolean | "force" | OcrProvider;
     /** 진행률 콜백 — current: 현재 페이지/섹션, total: 전체 수 */
     onProgress?: (current: number, total: number) => void;
     /** PDF 머리글/바닥글 자동 제거 */
@@ -190,7 +194,7 @@ interface ParseWarning {
     /** 구조화된 경고 코드 */
     code: WarningCode;
 }
-type WarningCode = "SKIPPED_IMAGE" | "SKIPPED_OLE" | "TRUNCATED_TABLE" | "OCR_FALLBACK" | "UNSUPPORTED_ELEMENT" | "BROKEN_ZIP_RECOVERY" | "HIDDEN_TEXT_FILTERED" | "MALFORMED_XML" | "PARTIAL_PARSE" | "LENIENT_CFB_RECOVERY" | "NEEDS_OCR" | "COM_EMPTY" | "DRM_COM_FALLBACK";
+type WarningCode = "SKIPPED_IMAGE" | "SKIPPED_OLE" | "TRUNCATED_TABLE" | "OCR_FALLBACK" | "UNSUPPORTED_ELEMENT" | "BROKEN_ZIP_RECOVERY" | "HIDDEN_TEXT_FILTERED" | "MALFORMED_XML" | "PARTIAL_PARSE" | "LENIENT_CFB_RECOVERY" | "NEEDS_OCR" | "OCR_FAILED" | "OCR_APPLIED" | "COM_EMPTY" | "DRM_COM_FALLBACK";
 /** 문서 구조 (헤딩 트리) */
 interface OutlineItem {
     level: number;

package/dist/index.js

@@ -1,31 +1,33 @@
 import {
-  HEADING_RATIO_H1,
-  HEADING_RATIO_H2,
-  HEADING_RATIO_H3,
-  KordocError,
   MAX_COLS,
   MAX_ROWS,
-  VERSION,
   blocksToMarkdown,
   buildTable,
-  classifyError,
-  compareSectionPaths,
   convertTableToText,
   dedupeRunningHeaders,
   flattenLayoutTables,
   inlineImagesIntoMarkdown,
+  mapPuaText
+} from "./chunk-WF2E5P2G.js";
+import {
+  parsePageRange
+} from "./chunk-GE43BE46.js";
+import {
+  HEADING_RATIO_H1,
+  HEADING_RATIO_H2,
+  HEADING_RATIO_H3,
+  KordocError,
+  VERSION,
+  classifyError,
+  compareSectionPaths,
   isPathTraversal,
-  mapPuaText,
   normalizeSectionHref,
   precheckZipSize,
   sanitizeError,
   sanitizeHref,
   stripDtd,
   toArrayBuffer
-} from "./chunk-4KI23VHA.js";
-import {
-  parsePageRange
-} from "./chunk-GE43BE46.js";
+} from "./chunk-JP5YH3Z4.js";
 
 // src/index.ts
 import { readFile } from "fs/promises";
@@ -2232,6 +2234,19 @@ function findKordocMarkedCell(el, depth) {
   }
   return null;
 }
+function hasPageAutoNum(el, depth = 0) {
+  if (depth > 10) return false;
+  const children = el.childNodes;
+  if (!children) return false;
+  for (let i = 0; i < children.length; i++) {
+    const ch = children[i];
+    if (ch.nodeType !== 1) continue;
+    const tag = (ch.tagName || ch.localName || "").replace(/^[^:]+:/, "");
+    if (tag === "autoNum" && ch.getAttribute?.("numType") === "PAGE") return true;
+    if (hasPageAutoNum(ch, depth + 1)) return true;
+  }
+  return false;
+}
 function collectSubListText(el, ctx, depth = 0) {
   if (depth > 10) return "";
   const parts = [];
@@ -2410,6 +2425,22 @@ function extractParagraphInfo(para2, styleMap, ctx) {
   let href;
   let footnote;
   let charPrId;
+  const linkRanges = [];
+  const openFields = [];
+  const onFieldBegin = (el) => {
+    const url = extractHyperlinkHref(el);
+    if (url) {
+      linkRanges.push({ url, start: text.length });
+      openFields.push({ rangeIdx: linkRanges.length - 1 });
+      if (!href) href = url;
+    } else {
+      openFields.push({});
+    }
+  };
+  const onFieldEnd = () => {
+    const open = openFields.pop();
+    if (open?.rangeIdx !== void 0) linkRanges[open.rangeIdx].end = text.length;
+  };
   const handleCtrl = (ctrlEl) => {
     const kids2 = ctrlEl.childNodes;
     if (!kids2) return;
@@ -2419,10 +2450,13 @@ function extractParagraphInfo(para2, styleMap, ctx) {
       const ktag = (k.tagName || k.localName || "").replace(/^[^:]+:/, "");
       switch (ktag) {
         // 머리말/꼬리말 — 문서당 1회 수집, 본문 앞/뒤 배치
+        // 페이지 번호 크롬(autoNum PAGE + 문자 없는 잔여 텍스트, 예: "- 1 -")은 본문 정보가
+        // 아니므로 방출하지 않는다
         case "header":
         case "footer": {
           if (!ctx) break;
           const t = collectSubListText(k, ctx);
+          if (t && hasPageAutoNum(k) && !/\p{L}/u.test(t)) break;
           if (t) {
             const bucket = ktag === "header" ? ctx.shared.pageText.headers : ctx.shared.pageText.footers;
             if (!bucket.includes(t)) bucket.push(t);
@@ -2436,13 +2470,12 @@ function extractParagraphInfo(para2, styleMap, ctx) {
           if (noteText) footnote = (footnote ? footnote + "; " : "") + noteText;
           break;
         }
-        // 하이퍼링크 — fieldBegin type=HYPERLINK의 Path 파라미터
-        case "fieldBegin": {
-          const url = extractHyperlinkHref(k);
-          if (url && !href) href = url;
+        // 하이퍼링크 — fieldBegin type=HYPERLINK의 Path 파라미터 (extent 오프셋 추적)
+        case "fieldBegin":
+          onFieldBegin(k);
           break;
-        }
         case "fieldEnd":
+          onFieldEnd();
           break;
         // 변경추적 — 삭제 구간(deleteBegin~End)의 텍스트는 출력 제외 (최종본 상태 재현)
         case "deleteBegin":
@@ -2555,12 +2588,10 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         case "ctrl":
           handleCtrl(child);
           break;
-        // run 직계 fieldBegin (비표준 경로) — 하이퍼링크 URL만 추출
-        case "fieldBegin": {
-          const url = extractHyperlinkHref(child);
-          if (url && !href) href = url;
+        // run 직계 fieldBegin (비표준 경로) — 하이퍼링크 URL·extent 추적
+        case "fieldBegin":
+          onFieldBegin(child);
           break;
-        }
         // run 직계 변경추적 마커 (비표준 경로)
         case "deleteBegin":
           if (ctx) ctx.shared.track.deleteDepth++;
@@ -2572,6 +2603,8 @@ function extractParagraphInfo(para2, styleMap, ctx) {
         case "insertEnd":
           break;
         case "fieldEnd":
+          onFieldEnd();
+          break;
         case "parameters":
         case "stringParam":
         case "integerParam":
@@ -2620,6 +2653,18 @@ function extractParagraphInfo(para2, styleMap, ctx) {
     }
   };
   walk(para2);
+  {
+    const applied = [];
+    const closed = linkRanges.filter((r) => r.end !== void 0 && r.end > r.start).sort((a, b) => b.start - a.start);
+    for (const r of closed) {
+      if (applied.some(([s, e]) => r.start < e && r.end > s)) continue;
+      const anchor = text.slice(r.start, r.end);
+      if (!anchor.trim() || /[\n\x1F\[\]]/.test(anchor)) continue;
+      text = text.slice(0, r.start) + `[${anchor}](${r.url})` + text.slice(r.end);
+      applied.push([r.start, r.end]);
+    }
+    if (applied.length) href = void 0;
+  }
   const leaderIdx = text.indexOf("");
   if (leaderIdx >= 0) text = text.substring(0, leaderIdx);
   let cleanText = text.replace(/[ \t]+/g, " ").trim();
@@ -17413,8 +17458,50 @@ function decodeJohab(ch) {
     }
     const hit = lookupSymbol(ch);
     if (hit !== null) return hit;
+    return JOHAB_UNMAPPED;
+  }
+  return decodeHwp3Extra(ch);
+}
+function decodeHwp3Extra(ch) {
+  if (ch >= 13712 && ch <= 13721) return 8544 + (ch - 13712);
+  if (ch >= 14055 && ch <= 14064) return 9312 + (ch - 14055);
+  switch (ch) {
+    case 129:
+      return 8220;
+    // 왼쪽 큰따옴표
+    case 130:
+      return 8221;
+    // 오른쪽 큰따옴표
+    case 12316:
+      return 9473;
+    // ━ 굵은 가로선 (rhwp: U+F080F, 표시값 직행)
+    case 12349:
+      return 9632;
+    // ■ (rhwp: U+F0827, 표시값 직행)
+    case 13158:
+      return 9633;
+    // □ 글머리 (rhwp: U+F03C5, 한컴 표시값 직행)
+    case 13316:
+      return 8228;
+    // 한 점 리더
+    case 13377:
+      return 9632;
+    // ■
+    case 13382:
+      return 8594;
+    // → 오른쪽 화살표
+    case 13793:
+      return 9472;
+    // ─ 상자 그리기 가로선
+    case 13433:
+      return 9655;
+    // ▷
+    case 13434:
+      return 9654;
+    // ▶
+    default:
+      return JOHAB_UNMAPPED;
   }
-  return JOHAB_UNMAPPED;
 }
 function decodeHcharString(bytes) {
   let out = "";
@@ -17534,7 +17621,7 @@ var PARA_SHAPE_SIZE = 187;
 var LINE_INFO_SIZE = 14;
 var INLINE_CHAR_SHAPE_SIZE = 31;
 var SIMPLE_CTRL = /* @__PURE__ */ new Map([
-  [9, { extraBytes: 0, extraHchar: 0, emit: "	" }],
+  [9, { extraBytes: 6, extraHchar: 3, emit: "	" }],
   [7, { extraBytes: 6, extraHchar: 3, emit: "\uFFFC" }],
   [8, { extraBytes: 6, extraHchar: 3, emit: "\uFFFC" }],
   [18, { extraBytes: 6, extraHchar: 3, emit: " " }],
@@ -17710,6 +17797,12 @@ function parseCharStream(reader, charCount, ctx) {
         parseParagraphList2(reader, ctx);
         break;
       }
+      case 5:
+        if (headerVal1 > 0 && headerVal1 < 1e6) reader.skip(headerVal1);
+        break;
+      case 6:
+        reader.skip(34);
+        break;
       case 29:
         if (headerVal1 < 1e6) reader.skip(headerVal1);
         break;
@@ -17987,7 +18080,7 @@ function parseWorksheet(xml, sharedStrings, dateXfs, date1904) {
   }
   return { grid, merges, maxRow, maxCol };
 }
-function sheetToBlocks(sheetName, grid, merges, maxRow, maxCol, sheetIndex) {
+function sheetToBlocks(sheetName, grid, merges, maxRow, maxCol, sheetIndex, keepAnchoredEmptyCols) {
   const blocks = [];
   if (sheetName) {
     blocks.push({
@@ -18039,7 +18132,7 @@ function sheetToBlocks(sheetName, grid, merges, maxRow, maxCol, sheetIndex) {
     cellRows.push(row);
   }
   if (cellRows.length > 0) {
-    const table2 = buildTable(cellRows);
+    const table2 = buildTable(cellRows, { keepAnchoredEmptyCols });
     if (table2.rows > 0) {
       blocks.push({ type: "table", table: table2, pageNumber: sheetIndex + 1 });
     }
@@ -18109,7 +18202,7 @@ async function parseXlsxDocument(buffer, options) {
     try {
       const sheetXml = await sheetFile.async("text");
       const { grid, merges, maxRow, maxCol } = parseWorksheet(sheetXml, sharedStrings, dateXfs, date1904);
-      const sheetBlocks = sheetToBlocks(sheet.name, grid, merges, maxRow, maxCol, i);
+      const sheetBlocks = sheetToBlocks(sheet.name, grid, merges, maxRow, maxCol, i, options?.keepTrailingEmptyCols);
       blocks.push(...sheetBlocks);
     } catch (err) {
       warnings.push({
@@ -18658,7 +18751,7 @@ function cellValueToText(v) {
   if (typeof v === "boolean") return v ? "TRUE" : "FALSE";
   return v;
 }
-function sheetToBlocks2(sheetName, sheet, sheetIndex) {
+function sheetToBlocks2(sheetName, sheet, sheetIndex, keepAnchoredEmptyCols) {
   const blocks = [];
   if (sheetName) {
     blocks.push({
@@ -18731,7 +18824,7 @@ function sheetToBlocks2(sheetName, sheet, sheetIndex) {
     cellRows.push(row);
   }
   if (cellRows.length > 0) {
-    const table2 = buildTable(cellRows);
+    const table2 = buildTable(cellRows, { keepAnchoredEmptyCols });
     if (table2.rows > 0) {
       blocks.push({ type: "table", table: table2, pageNumber: sheetIndex + 1 });
     }
@@ -18812,7 +18905,7 @@ async function parseXlsDocument(buffer, options) {
     }
     try {
       const { sheet } = extractSheetCells(records, bofIdx, globals.sst, convertNum);
-      const blocks = sheetToBlocks2(meta.name, sheet, i);
+      const blocks = sheetToBlocks2(meta.name, sheet, i, options?.keepTrailingEmptyCols);
       allBlocks.push(...blocks);
     } catch (e) {
       warnings.push({
@@ -20033,8 +20126,8 @@ function parseTable2(el, blocks, paraShapeMap, sectionNum, warnings, keep) {
     for (let j = 0; j < rowCells.length; j++) {
       const cellEl = rowCells[j];
       if (cellEl.nodeType !== 1 || localName(cellEl) !== "CELL") continue;
-      const colAddr = parseInt(cellEl.getAttribute("ColAddr") ?? "0", 10);
-      const rowAddr = parseInt(cellEl.getAttribute("RowAddr") ?? "0", 10);
+      const colAddr = parseInt(cellEl.getAttribute("ColAddr") ?? "0", 10) || 0;
+      const rowAddr = parseInt(cellEl.getAttribute("RowAddr") ?? "0", 10) || 0;
       const colSpan = Math.min(Math.max(1, parseInt(cellEl.getAttribute("ColSpan") ?? "1", 10) || 1), MAX_TABLE_COLS);
       const rowSpan = Math.min(Math.max(1, parseInt(cellEl.getAttribute("RowSpan") ?? "1", 10) || 1), MAX_TABLE_ROWS);
       const cellText = extractCellText(cellEl);
@@ -20042,24 +20135,11 @@ function parseTable2(el, blocks, paraShapeMap, sectionNum, warnings, keep) {
     }
   }
   if (cells.length === 0) return;
-  const grid = Array.from({ length: rowCount }, () => Array(colCount).fill(null));
+  const cellRows = Array.from({ length: rowCount }, () => []);
   for (const cell2 of cells) {
-    const r = cell2.rowAddr ?? 0;
-    const c = cell2.colAddr ?? 0;
-    if (isNaN(r) || isNaN(c) || r >= rowCount || c >= colCount) continue;
-    grid[r][c] = cell2;
-    for (let dr = 0; dr < cell2.rowSpan; dr++) {
-      for (let dc = 0; dc < cell2.colSpan; dc++) {
-        if (dr === 0 && dc === 0) continue;
-        if (r + dr < rowCount && c + dc < colCount) {
-          grid[r + dr][c + dc] = { text: "", colSpan: 1, rowSpan: 1 };
-        }
-      }
-    }
+    const r = Math.min(Math.max(cell2.rowAddr ?? 0, 0), rowCount - 1);
+    cellRows[r].push(cell2);
   }
-  const cellRows = grid.map(
-    (row) => row.map((cell2) => cell2 ?? { text: "", colSpan: 1, rowSpan: 1 })
-  );
   const table2 = buildTable(cellRows, { keepAnchoredEmptyCols: keep });
   const caption = extractShapeCaption(el);
   if (caption.text && caption.before) {
@@ -30414,7 +30494,7 @@ async function parseHwp(buffer, options) {
 async function parsePdf(buffer, options) {
   let parsePdfDocument;
   try {
-    const mod = await import("./parser-XX4QTDFQ.js");
+    const mod = await import("./parser-ND2YGQZR.js");
     parsePdfDocument = mod.parsePdfDocument;
   } catch {
     return {