kordoc 4.1.0 → 4.2.0

package/dist/pdf-ocr-VYE4DHD4.js

@@ -0,0 +1,386 @@
+#!/usr/bin/env node
+import {
+  OCR_DET_MODEL,
+  OCR_REC_DICT,
+  OCR_REC_MODEL,
+  ensureOcrModels,
+  getOcrModelsDir,
+  parseCharacterDict
+} from "./chunk-GEPINOL6.js";
+import "./chunk-NNDVFNMJ.js";
+import {
+  extractPageBlocksWithLines
+} from "./chunk-26AJJCFF.js";
+import "./chunk-4VNLZ2GZ.js";
+import "./chunk-FSXLSHCW.js";
+
+// src/ocr/engine.ts
+import { readFile } from "fs/promises";
+import { join } from "path";
+var DET_LONG_SIDE = 960;
+var DET_THRESH = 0.3;
+var DET_BOX_THRESH = 0.6;
+var DET_UNCLIP_RATIO = 1.5;
+var DET_MIN_SIZE = 3;
+var DET_MAX_BOXES = 1e3;
+var REC_HEIGHT = 48;
+var REC_MIN_WIDTH = 320;
+var REC_MAX_WIDTH = 3200;
+var TEXT_SCORE = 0.5;
+var DET_MEAN = [0.485, 0.456, 0.406];
+var DET_STD = [0.229, 0.224, 0.225];
+var OcrEngine = class _OcrEngine {
+  det;
+  rec;
+  dict;
+  ort;
+  sharp;
+  constructor(parts) {
+    this.det = parts.det;
+    this.rec = parts.rec;
+    this.dict = parts.dict;
+    this.ort = parts.ort;
+    this.sharp = parts.sharp;
+  }
+  static async create() {
+    const [ortMod, sharpModRaw] = await Promise.all([
+      tryImport("onnxruntime-node", () => import("onnxruntime-node")),
+      tryImport(
+        "sharp",
+        () => import("sharp")
+      )
+    ]);
+    const sharpAny = sharpModRaw;
+    const sharpMod = typeof sharpAny === "function" ? sharpAny : sharpAny.default ?? sharpAny;
+    const dir = getOcrModelsDir();
+    const sessionOpts = {
+      graphOptimizationLevel: "all",
+      executionProviders: ["cpu"],
+      logSeverityLevel: 3
+      // paddle2onnx 변환 잔여물 W 로그 폭주 억제
+    };
+    const [det, rec, dictYml] = await Promise.all([
+      ortMod.InferenceSession.create(join(dir, OCR_DET_MODEL.filename), sessionOpts),
+      ortMod.InferenceSession.create(join(dir, OCR_REC_MODEL.filename), sessionOpts),
+      readFile(join(dir, OCR_REC_DICT.filename), "utf-8")
+    ]);
+    const dict = parseCharacterDict(dictYml);
+    if (dict.length === 0) throw new Error("OCR \uC0AC\uC804 \uD30C\uC2F1 \uC2E4\uD328 \u2014 \uBAA8\uB378 \uCE90\uC2DC\uB97C \uC0AD\uC81C \uD6C4 \uC7AC\uB2E4\uC6B4\uB85C\uB4DC\uD558\uC138\uC694");
+    return new _OcrEngine({ det, rec, dict, ort: ortMod, sharp: sharpMod });
+  }
+  /** onnxruntime-node 1.14+ InferenceSession.release() — 구버전은 무시 */
+  async destroy() {
+    for (const s of [this.det, this.rec]) {
+      const rel = s.release;
+      if (typeof rel === "function") {
+        try {
+          await rel.call(s);
+        } catch {
+        }
+      }
+    }
+  }
+  /**
+   * 페이지 RGBA 픽셀 → 텍스트 라인 인식.
+   * 반환 좌표는 입력 픽셀 기준. 라인은 위→아래, 좌→우 정렬.
+   */
+  async recognizePage(rgba, width, height) {
+    if (width < DET_MIN_SIZE || height < DET_MIN_SIZE) return [];
+    const boxes = await this.detect(rgba, width, height);
+    const items = [];
+    for (const b of boxes) {
+      const r = await this.recognizeLine(rgba, width, height, b);
+      if (r && r.confidence >= TEXT_SCORE && r.text.trim()) items.push(r);
+    }
+    items.sort((a, b) => a.y - b.y || a.x - b.x);
+    return items;
+  }
+  // ─── det ─────────────────────────────────────────────
+  async detect(rgba, width, height) {
+    const ratio = DET_LONG_SIDE / Math.max(width, height);
+    const dw = Math.max(32, Math.round(width * ratio / 32) * 32);
+    const dh = Math.max(32, Math.round(height * ratio / 32) * 32);
+    const rgb = await this.sharp(rgba, { raw: { width, height, channels: 4 } }).resize(dw, dh, { fit: "fill" }).removeAlpha().raw().toBuffer();
+    const plane = dw * dh;
+    const input = new Float32Array(3 * plane);
+    for (let i = 0; i < plane; i++) {
+      const r = rgb[i * 3] / 255;
+      const g = rgb[i * 3 + 1] / 255;
+      const b = rgb[i * 3 + 2] / 255;
+      input[i] = (b - DET_MEAN[0]) / DET_STD[0];
+      input[plane + i] = (g - DET_MEAN[1]) / DET_STD[1];
+      input[2 * plane + i] = (r - DET_MEAN[2]) / DET_STD[2];
+    }
+    const tensor = new this.ort.Tensor("float32", input, [1, 3, dh, dw]);
+    const out = await this.det.run({ [this.det.inputNames[0]]: tensor });
+    const probMap = out[this.det.outputNames[0]].data;
+    const rawBoxes = componentBoxes(probMap, dw, dh);
+    const sx = width / dw;
+    const sy = height / dh;
+    const boxes = [];
+    for (const rb of rawBoxes.slice(0, DET_MAX_BOXES)) {
+      const bw = rb.x2 - rb.x1 + 1;
+      const bh = rb.y2 - rb.y1 + 1;
+      const delta = bw * bh * DET_UNCLIP_RATIO / (2 * (bw + bh));
+      const x1 = Math.max(0, Math.floor((rb.x1 - delta) * sx));
+      const y1 = Math.max(0, Math.floor((rb.y1 - delta) * sy));
+      const x2 = Math.min(width, Math.ceil((rb.x2 + 1 + delta) * sx));
+      const y2 = Math.min(height, Math.ceil((rb.y2 + 1 + delta) * sy));
+      if (x2 - x1 < DET_MIN_SIZE || y2 - y1 < DET_MIN_SIZE) continue;
+      boxes.push({ x: x1, y: y1, w: x2 - x1, h: y2 - y1 });
+    }
+    return boxes;
+  }
+  // ─── rec ─────────────────────────────────────────────
+  async recognizeLine(rgba, width, height, box) {
+    const rw = Math.min(REC_MAX_WIDTH, Math.max(16, Math.round(box.w * REC_HEIGHT / box.h)));
+    const padded = Math.max(REC_MIN_WIDTH, rw);
+    const rgb = await this.sharp(rgba, { raw: { width, height, channels: 4 } }).extract({ left: box.x, top: box.y, width: box.w, height: box.h }).resize(rw, REC_HEIGHT, { fit: "fill" }).removeAlpha().raw().toBuffer();
+    const plane = padded * REC_HEIGHT;
+    const input = new Float32Array(3 * plane);
+    for (let y = 0; y < REC_HEIGHT; y++) {
+      for (let x = 0; x < rw; x++) {
+        const src = (y * rw + x) * 3;
+        const dst = y * padded + x;
+        input[dst] = rgb[src + 2] / 255 / 0.5 - 1;
+        input[plane + dst] = rgb[src + 1] / 255 / 0.5 - 1;
+        input[2 * plane + dst] = rgb[src] / 255 / 0.5 - 1;
+      }
+    }
+    const tensor = new this.ort.Tensor("float32", input, [1, 3, REC_HEIGHT, padded]);
+    const out = await this.rec.run({ [this.rec.inputNames[0]]: tensor });
+    const logits = out[this.rec.outputNames[0]];
+    const [, T, C] = logits.dims;
+    const data = logits.data;
+    const decoded = ctcDecode(data, T, C, this.dict);
+    if (!decoded) return null;
+    return { text: decoded.text, x: box.x, y: box.y, w: box.w, h: box.h, confidence: decoded.confidence };
+  }
+};
+function ctcDecode(data, T, C, dict) {
+  let text = "";
+  let confSum = 0;
+  let confCount = 0;
+  let prev = -1;
+  for (let t = 0; t < T; t++) {
+    const off = t * C;
+    let best = 0;
+    let bestV = data[off];
+    for (let c = 1; c < C; c++) {
+      const v = data[off + c];
+      if (v > bestV) {
+        bestV = v;
+        best = c;
+      }
+    }
+    const repeat = best === prev;
+    prev = best;
+    if (best === 0 || repeat) continue;
+    let p = bestV;
+    if (p > 1.0001 || p < 0) {
+      let denom = 0;
+      for (let c = 0; c < C; c++) denom += Math.exp(data[off + c] - bestV);
+      p = 1 / denom;
+    }
+    confSum += p;
+    confCount++;
+    if (best >= 1 && best <= dict.length) text += dict[best - 1];
+    else if (best === dict.length + 1) text += " ";
+  }
+  if (!text) return null;
+  return { text, confidence: confCount > 0 ? confSum / confCount : 0 };
+}
+function componentBoxes(prob, w, h) {
+  const visited = new Uint8Array(w * h);
+  const boxes = [];
+  const stack = [];
+  for (let start = 0; start < w * h; start++) {
+    if (visited[start] || prob[start] <= DET_THRESH) continue;
+    let x1 = start % w, x2 = x1, y1 = start / w | 0, y2 = y1;
+    let sum = 0;
+    let count = 0;
+    stack.length = 0;
+    stack.push(start);
+    visited[start] = 1;
+    while (stack.length) {
+      const p = stack.pop();
+      const px = p % w;
+      const py = p / w | 0;
+      sum += prob[p];
+      count++;
+      if (px < x1) x1 = px;
+      if (px > x2) x2 = px;
+      if (py < y1) y1 = py;
+      if (py > y2) y2 = py;
+      if (px > 0 && !visited[p - 1] && prob[p - 1] > DET_THRESH) {
+        visited[p - 1] = 1;
+        stack.push(p - 1);
+      }
+      if (px < w - 1 && !visited[p + 1] && prob[p + 1] > DET_THRESH) {
+        visited[p + 1] = 1;
+        stack.push(p + 1);
+      }
+      if (py > 0 && !visited[p - w] && prob[p - w] > DET_THRESH) {
+        visited[p - w] = 1;
+        stack.push(p - w);
+      }
+      if (py < h - 1 && !visited[p + w] && prob[p + w] > DET_THRESH) {
+        visited[p + w] = 1;
+        stack.push(p + w);
+      }
+    }
+    if (x2 - x1 + 1 < DET_MIN_SIZE && y2 - y1 + 1 < DET_MIN_SIZE) continue;
+    boxes.push({ x1, y1, x2, y2, score: sum / count });
+  }
+  return boxes.filter((b) => b.score >= DET_BOX_THRESH).sort((a, b) => a.y1 - b.y1 || a.x1 - b.x1);
+}
+async function tryImport(name, loader) {
+  try {
+    return await loader();
+  } catch (e) {
+    throw new Error(
+      `\uB0B4\uC7A5 OCR \uC744 \uC0AC\uC6A9\uD558\uB824\uBA74 optional dependency '${name}' \uC774 \uD544\uC694\uD569\uB2C8\uB2E4. \`npm install ${name}\` \uD6C4 \uB2E4\uC2DC \uC2E4\uD589\uD558\uC138\uC694. \uC6D0\uC778: ${e.message}`
+    );
+  }
+}
+var enginePromise = null;
+function getOcrEngine() {
+  if (!enginePromise) {
+    enginePromise = OcrEngine.create().catch((err) => {
+      enginePromise = null;
+      throw err;
+    });
+  }
+  return enginePromise;
+}
+
+// src/ocr/pdf-ocr.ts
+var OCR_RENDER_SCALE = 3;
+var PAGE_TIMEOUT_MS = 12e4;
+async function runPdfOcr(buffer, targets, mode, warnings, onProgress) {
+  const result = /* @__PURE__ */ new Map();
+  if (targets.size === 0) return result;
+  const pdfiumMod = await tryImport2(
+    "@hyzyla/pdfium",
+    () => import("@hyzyla/pdfium")
+  );
+  if (mode === "builtin") {
+    await ensureOcrModels((p) => {
+      if (p.phase === "download" && p.downloaded === 0) {
+        process.stderr.write(`[kordoc-ocr] ${p.spec.name} \uB2E4\uC6B4\uB85C\uB4DC \uC911 (~${p.spec.sizeMb}MB)...
+`);
+      }
+    });
+  }
+  const engine = mode === "builtin" ? await getOcrEngine() : null;
+  const pdfium = await pdfiumMod.PDFiumLibrary.init();
+  const doc = await pdfium.loadDocument(new Uint8Array(buffer));
+  try {
+    let done = 0;
+    for (const page of doc.pages()) {
+      const pageNo = page.number + 1;
+      if (!targets.has(pageNo)) continue;
+      onProgress?.(++done, targets.size);
+      try {
+        const blocks = await withTimeout(
+          ocrOnePage(page, pageNo, mode, engine, warnings),
+          PAGE_TIMEOUT_MS,
+          `OCR \uD398\uC774\uC9C0 ${pageNo} \uD0C0\uC784\uC544\uC6C3 (${PAGE_TIMEOUT_MS / 1e3}\uCD08)`
+        );
+        result.set(pageNo, blocks);
+      } catch (e) {
+        warnings.push({
+          page: pageNo,
+          message: `\uD398\uC774\uC9C0 ${pageNo} OCR \uC2E4\uD328: ${e instanceof Error ? e.message : String(e)}`,
+          code: "OCR_FAILED"
+        });
+      }
+    }
+  } finally {
+    doc.destroy();
+    pdfium.destroy();
+  }
+  return result;
+}
+async function ocrOnePage(page, pageNo, mode, engine, warnings) {
+  const { originalWidth: pdfW, originalHeight: pdfH } = page.getOriginalSize();
+  const rendered = await page.render({
+    scale: OCR_RENDER_SCALE,
+    render: async ({ data }) => data
+  });
+  const { data: bgra, width: rw, height: rh } = rendered;
+  const rgba = bgraToRgba(bgra);
+  if (mode === "builtin") {
+    const items = await engine.recognizePage(rgba, rw, rh);
+    return ocrItemsToBlocks(items, pageNo, pdfW, pdfH, rh / pdfH);
+  }
+  const sharpModRaw = await tryImport2(
+    "sharp",
+    () => import("sharp")
+  );
+  const sharpAny = sharpModRaw;
+  const sharp = typeof sharpAny === "function" ? sharpAny : sharpAny.default ?? sharpAny;
+  const png = await sharp(rgba, { raw: { width: rw, height: rh, channels: 4 } }).png().toBuffer();
+  const text = await mode(new Uint8Array(png), pageNo, "image/png");
+  if (!text.trim()) {
+    warnings.push({ page: pageNo, message: `\uD398\uC774\uC9C0 ${pageNo} OCR \uACB0\uACFC \uC5C6\uC74C`, code: "OCR_FAILED" });
+    return [];
+  }
+  return [{ type: "paragraph", text: text.trim(), pageNumber: pageNo }];
+}
+function ocrItemsToBlocks(items, pageNumber, pdfW, pdfH, scale) {
+  const norm = items.map((it) => {
+    const h = it.h / scale;
+    return {
+      text: it.text,
+      x: Math.round(it.x / scale),
+      // NormItem.y 는 pdfjs transform[5] = 베이스라인 (bottom-up) — 박스 하단으로 근사
+      y: Math.round(pdfH - (it.y + it.h) / scale),
+      w: Math.round(it.w / scale),
+      h: Math.round(h),
+      fontSize: Math.max(1, Math.round(h * 0.8)),
+      // 박스높이는 어센더+디센더 포함 — 글자크기 근사
+      fontName: "ocr",
+      isHidden: false
+    };
+  });
+  return extractPageBlocksWithLines(norm, pageNumber, { fnArray: [], argsArray: [] }, pdfW, pdfH);
+}
+async function tryImport2(name, loader) {
+  try {
+    return await loader();
+  } catch (e) {
+    throw new Error(
+      `OCR \uC744 \uC0AC\uC6A9\uD558\uB824\uBA74 optional dependency '${name}' \uC774 \uD544\uC694\uD569\uB2C8\uB2E4. \`npm install ${name}\` \uD6C4 \uB2E4\uC2DC \uC2E4\uD589\uD558\uC138\uC694. \uC6D0\uC778: ${e.message}`
+    );
+  }
+}
+async function withTimeout(promise, ms, msg) {
+  promise.catch(() => {
+  });
+  let timer;
+  try {
+    return await Promise.race([
+      promise,
+      new Promise((_, reject) => {
+        timer = setTimeout(() => reject(new Error(msg)), ms);
+      })
+    ]);
+  } finally {
+    if (timer) clearTimeout(timer);
+  }
+}
+function bgraToRgba(bgra) {
+  const out = new Uint8Array(bgra.length);
+  for (let i = 0; i < bgra.length; i += 4) {
+    out[i] = bgra[i + 2];
+    out[i + 1] = bgra[i + 1];
+    out[i + 2] = bgra[i];
+    out[i + 3] = bgra[i + 3];
+  }
+  return out;
+}
+export {
+  runPdfOcr
+};
+//# sourceMappingURL=pdf-ocr-VYE4DHD4.js.map

package/dist/pdf-ocr-VYE4DHD4.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/engine.ts","../src/ocr/pdf-ocr.ts"],"sourcesContent":["/**\n * 내장 텍스트 OCR 엔진 — PP-OCRv5 korean (det DBNet + rec SVTR/CTC) ONNX 추론.\n *\n * 파이프라인: 페이지 RGBA → det(선 검출) → 라인 crop → rec(CTC 인식) → OcrItem[]\n * (좌표는 입력 픽셀 기준 top-left origin — 호출자가 PDF 좌표계로 변환).\n *\n * 전·후처리는 공식 inference.yml 스펙 그대로:\n *  - det: BGR, 긴 변 960 리사이즈(32 배수), mean/std [0.485,0.456,0.406]/[0.229,0.224,0.225],\n *         DBPostProcess thresh 0.3 / box_thresh 0.6 / unclip_ratio 1.5\n *  - rec: BGR, 높이 48 고정 비율 리사이즈 + 우측 zero-pad, (x/255-0.5)/0.5,\n *         CTC 디코드 (blank=0, 1..N=사전, N+1=공백), text_score 0.5\n * DB 후처리의 contour+minAreaRect 는 축정렬 connected-component bbox 로 근사\n * (공문서 스캔은 수평 텍스트가 지배적 — 회전 텍스트는 v1 범위 밖).\n *\n * 의존성(onnxruntime-node, sharp)은 optional — 미설치 시 create()가 명확한 에러.\n * 모델 미다운로드 시에도 즉시 실패 — 호출자가 ensureOcrModels() 먼저.\n */\n\nimport type { InferenceSession } from \"onnxruntime-node\"\nimport { readFile } from \"fs/promises\"\nimport { join } from \"path\"\nimport {\n  OCR_DET_MODEL,\n  OCR_REC_MODEL,\n  OCR_REC_DICT,\n  getOcrModelsDir,\n  parseCharacterDict,\n} from \"./models.js\"\n\n/** OCR 인식 결과 한 줄 — 좌표는 입력 이미지 픽셀 (top-left origin, y down) */\nexport interface OcrItem {\n  text: string\n  x: number\n  y: number\n  w: number\n  h: number\n  /** CTC 평균 신뢰도 0~1 */\n  confidence: number\n}\n\n/** det 입력 긴 변 (공식 resize_long) */\nconst DET_LONG_SIDE = 960\nconst DET_THRESH = 0.3\nconst DET_BOX_THRESH = 0.6\nconst DET_UNCLIP_RATIO = 1.5\nconst DET_MIN_SIZE = 3\nconst DET_MAX_BOXES = 1000\nconst REC_HEIGHT = 48\nconst REC_MIN_WIDTH = 320\nconst REC_MAX_WIDTH = 3200\nconst TEXT_SCORE = 0.5\n\n// det: BGR 채널 순서에 yml 기재 순서 그대로 적용 (mean[0]→B)\nconst DET_MEAN = [0.485, 0.456, 0.406]\nconst DET_STD = [0.229, 0.224, 0.225]\n\ntype SharpFactory = (\n  input: Uint8Array | Buffer,\n  options?: { raw?: { width: number; height: number; channels: number } },\n) => SharpChain\ninterface SharpChain {\n  extract(region: { left: number; top: number; width: number; height: number }): SharpChain\n  resize(w: number, h: number, opts?: { fit?: string }): SharpChain\n  removeAlpha(): SharpChain\n  raw(): { toBuffer(): Promise<Buffer> }\n}\n\nexport class OcrEngine {\n  private det: InferenceSession\n  private rec: InferenceSession\n  private dict: string[]\n  private ort: typeof import(\"onnxruntime-node\")\n  private sharp: SharpFactory\n\n  private constructor(parts: {\n    det: InferenceSession\n    rec: InferenceSession\n    dict: string[]\n    ort: typeof import(\"onnxruntime-node\")\n    sharp: SharpFactory\n  }) {\n    this.det = parts.det\n    this.rec = parts.rec\n    this.dict = parts.dict\n    this.ort = parts.ort\n    this.sharp = parts.sharp\n  }\n\n  static async create(): Promise<OcrEngine> {\n    const [ortMod, sharpModRaw] = await Promise.all([\n      tryImport<typeof import(\"onnxruntime-node\")>(\"onnxruntime-node\", () => import(\"onnxruntime-node\")),\n      tryImport<{ default?: SharpFactory } & SharpFactory>(\n        \"sharp\",\n        () => import(\"sharp\") as unknown as Promise<{ default?: SharpFactory } & SharpFactory>,\n      ),\n    ])\n    const sharpAny = sharpModRaw as { default?: SharpFactory } | SharpFactory\n    const sharpMod: SharpFactory =\n      typeof sharpAny === \"function\" ? sharpAny : (sharpAny.default ?? (sharpAny as unknown as SharpFactory))\n\n    const dir = getOcrModelsDir()\n    const sessionOpts: import(\"onnxruntime-node\").InferenceSession.SessionOptions = {\n      graphOptimizationLevel: \"all\",\n      executionProviders: [\"cpu\"],\n      logSeverityLevel: 3, // paddle2onnx 변환 잔여물 W 로그 폭주 억제\n    }\n    const [det, rec, dictYml] = await Promise.all([\n      ortMod.InferenceSession.create(join(dir, OCR_DET_MODEL.filename), sessionOpts),\n      ortMod.InferenceSession.create(join(dir, OCR_REC_MODEL.filename), sessionOpts),\n      readFile(join(dir, OCR_REC_DICT.filename), \"utf-8\"),\n    ])\n    const dict = parseCharacterDict(dictYml)\n    if (dict.length === 0) throw new Error(\"OCR 사전 파싱 실패 — 모델 캐시를 삭제 후 재다운로드하세요\")\n\n    return new OcrEngine({ det, rec, dict, ort: ortMod, sharp: sharpMod })\n  }\n\n  /** onnxruntime-node 1.14+ InferenceSession.release() — 구버전은 무시 */\n  async destroy(): Promise<void> {\n    for (const s of [this.det, this.rec]) {\n      const rel = (s as unknown as { release?: () => Promise<void> }).release\n      if (typeof rel === \"function\") {\n        try { await rel.call(s) } catch { /* ignore */ }\n      }\n    }\n  }\n\n  /**\n   * 페이지 RGBA 픽셀 → 텍스트 라인 인식.\n   * 반환 좌표는 입력 픽셀 기준. 라인은 위→아래, 좌→우 정렬.\n   */\n  async recognizePage(rgba: Uint8Array, width: number, height: number): Promise<OcrItem[]> {\n    if (width < DET_MIN_SIZE || height < DET_MIN_SIZE) return []\n    const boxes = await this.detect(rgba, width, height)\n    const items: OcrItem[] = []\n    for (const b of boxes) {\n      const r = await this.recognizeLine(rgba, width, height, b)\n      if (r && r.confidence >= TEXT_SCORE && r.text.trim()) items.push(r)\n    }\n    items.sort((a, b) => (a.y - b.y) || (a.x - b.x))\n    return items\n  }\n\n  // ─── det ─────────────────────────────────────────────\n\n  private async detect(\n    rgba: Uint8Array,\n    width: number,\n    height: number,\n  ): Promise<Array<{ x: number; y: number; w: number; h: number }>> {\n    const ratio = DET_LONG_SIDE / Math.max(width, height)\n    const dw = Math.max(32, Math.round((width * ratio) / 32) * 32)\n    const dh = Math.max(32, Math.round((height * ratio) / 32) * 32)\n\n    const rgb = await this.sharp(rgba, { raw: { width, height, channels: 4 } })\n      .resize(dw, dh, { fit: \"fill\" })\n      .removeAlpha()\n      .raw()\n      .toBuffer()\n\n    // HWC RGB → CHW BGR float32 정규화\n    const plane = dw * dh\n    const input = new Float32Array(3 * plane)\n    for (let i = 0; i < plane; i++) {\n      const r = rgb[i * 3] / 255\n      const g = rgb[i * 3 + 1] / 255\n      const b = rgb[i * 3 + 2] / 255\n      input[i] = (b - DET_MEAN[0]) / DET_STD[0]\n      input[plane + i] = (g - DET_MEAN[1]) / DET_STD[1]\n      input[2 * plane + i] = (r - DET_MEAN[2]) / DET_STD[2]\n    }\n\n    const tensor = new this.ort.Tensor(\"float32\", input, [1, 3, dh, dw])\n    const out = await this.det.run({ [this.det.inputNames[0]]: tensor })\n    const probMap = out[this.det.outputNames[0]].data as Float32Array\n\n    const rawBoxes = componentBoxes(probMap, dw, dh)\n    const sx = width / dw\n    const sy = height / dh\n    const boxes: Array<{ x: number; y: number; w: number; h: number }> = []\n    for (const rb of rawBoxes.slice(0, DET_MAX_BOXES)) {\n      // unclip: DB 는 학습 시 텍스트 영역을 수축시키므로 검출 박스를 되팽창\n      const bw = rb.x2 - rb.x1 + 1\n      const bh = rb.y2 - rb.y1 + 1\n      const delta = (bw * bh * DET_UNCLIP_RATIO) / (2 * (bw + bh))\n      const x1 = Math.max(0, Math.floor((rb.x1 - delta) * sx))\n      const y1 = Math.max(0, Math.floor((rb.y1 - delta) * sy))\n      const x2 = Math.min(width, Math.ceil((rb.x2 + 1 + delta) * sx))\n      const y2 = Math.min(height, Math.ceil((rb.y2 + 1 + delta) * sy))\n      if (x2 - x1 < DET_MIN_SIZE || y2 - y1 < DET_MIN_SIZE) continue\n      boxes.push({ x: x1, y: y1, w: x2 - x1, h: y2 - y1 })\n    }\n    return boxes\n  }\n\n  // ─── rec ─────────────────────────────────────────────\n\n  private async recognizeLine(\n    rgba: Uint8Array,\n    width: number,\n    height: number,\n    box: { x: number; y: number; w: number; h: number },\n  ): Promise<OcrItem | null> {\n    const rw = Math.min(REC_MAX_WIDTH, Math.max(16, Math.round((box.w * REC_HEIGHT) / box.h)))\n    const padded = Math.max(REC_MIN_WIDTH, rw)\n\n    const rgb = await this.sharp(rgba, { raw: { width, height, channels: 4 } })\n      .extract({ left: box.x, top: box.y, width: box.w, height: box.h })\n      .resize(rw, REC_HEIGHT, { fit: \"fill\" })\n      .removeAlpha()\n      .raw()\n      .toBuffer()\n\n    // HWC RGB → CHW BGR, (x/255-0.5)/0.5, 우측 zero-pad\n    const plane = padded * REC_HEIGHT\n    const input = new Float32Array(3 * plane) // pad 영역은 0 (=정규화 후 회색 아님 주의: PP 도 0 pad)\n    for (let y = 0; y < REC_HEIGHT; y++) {\n      for (let x = 0; x < rw; x++) {\n        const src = (y * rw + x) * 3\n        const dst = y * padded + x\n        input[dst] = rgb[src + 2] / 255 / 0.5 - 1 // B\n        input[plane + dst] = rgb[src + 1] / 255 / 0.5 - 1 // G\n        input[2 * plane + dst] = rgb[src] / 255 / 0.5 - 1 // R\n      }\n    }\n\n    const tensor = new this.ort.Tensor(\"float32\", input, [1, 3, REC_HEIGHT, padded])\n    const out = await this.rec.run({ [this.rec.inputNames[0]]: tensor })\n    const logits = out[this.rec.outputNames[0]]\n    const [, T, C] = logits.dims as number[]\n    const data = logits.data as Float32Array\n\n    const decoded = ctcDecode(data, T, C, this.dict)\n    if (!decoded) return null\n    return { text: decoded.text, x: box.x, y: box.y, w: box.w, h: box.h, confidence: decoded.confidence }\n  }\n}\n\n/** CTC greedy 디코드 — 연속 중복 붕괴 → blank(0) 제거 → 사전 매핑 (테스트용 export) */\nexport function ctcDecode(\n  data: Float32Array,\n  T: number,\n  C: number,\n  dict: string[],\n): { text: string; confidence: number } | null {\n  let text = \"\"\n  let confSum = 0\n  let confCount = 0\n  let prev = -1\n  for (let t = 0; t < T; t++) {\n    const off = t * C\n    let best = 0\n    let bestV = data[off]\n    for (let c = 1; c < C; c++) {\n      const v = data[off + c]\n      if (v > bestV) { bestV = v; best = c }\n    }\n    const repeat = best === prev\n    prev = best\n    if (best === 0 || repeat) continue\n    // 모델 출력이 softmax 확률이 아니면 (>1) 해당 스텝만 정규화\n    let p = bestV\n    if (p > 1.0001 || p < 0) {\n      let denom = 0\n      for (let c = 0; c < C; c++) denom += Math.exp(data[off + c] - bestV)\n      p = 1 / denom\n    }\n    confSum += p\n    confCount++\n    if (best >= 1 && best <= dict.length) text += dict[best - 1]\n    else if (best === dict.length + 1) text += \" \"\n  }\n  if (!text) return null\n  return { text, confidence: confCount > 0 ? confSum / confCount : 0 }\n}\n\n/** 이진화 확률맵의 4-연결 성분 bbox (score = 성분 평균 확률, 테스트용 export) */\nexport function componentBoxes(\n  prob: Float32Array,\n  w: number,\n  h: number,\n): Array<{ x1: number; y1: number; x2: number; y2: number }> {\n  const visited = new Uint8Array(w * h)\n  const boxes: Array<{ x1: number; y1: number; x2: number; y2: number; score: number }> = []\n  const stack: number[] = []\n\n  for (let start = 0; start < w * h; start++) {\n    if (visited[start] || prob[start] <= DET_THRESH) continue\n    let x1 = start % w, x2 = x1, y1 = (start / w) | 0, y2 = y1\n    let sum = 0\n    let count = 0\n    stack.length = 0\n    stack.push(start)\n    visited[start] = 1\n    while (stack.length) {\n      const p = stack.pop()!\n      const px = p % w\n      const py = (p / w) | 0\n      sum += prob[p]\n      count++\n      if (px < x1) x1 = px\n      if (px > x2) x2 = px\n      if (py < y1) y1 = py\n      if (py > y2) y2 = py\n      // 4-이웃\n      if (px > 0 && !visited[p - 1] && prob[p - 1] > DET_THRESH) { visited[p - 1] = 1; stack.push(p - 1) }\n      if (px < w - 1 && !visited[p + 1] && prob[p + 1] > DET_THRESH) { visited[p + 1] = 1; stack.push(p + 1) }\n      if (py > 0 && !visited[p - w] && prob[p - w] > DET_THRESH) { visited[p - w] = 1; stack.push(p - w) }\n      if (py < h - 1 && !visited[p + w] && prob[p + w] > DET_THRESH) { visited[p + w] = 1; stack.push(p + w) }\n    }\n    if (x2 - x1 + 1 < DET_MIN_SIZE && y2 - y1 + 1 < DET_MIN_SIZE) continue\n    boxes.push({ x1, y1, x2, y2, score: sum / count })\n  }\n\n  return boxes\n    .filter(b => b.score >= DET_BOX_THRESH)\n    .sort((a, b) => (a.y1 - b.y1) || (a.x1 - b.x1))\n}\n\nasync function tryImport<T>(name: string, loader: () => Promise<T>): Promise<T> {\n  try {\n    return await loader()\n  } catch (e) {\n    throw new Error(\n      `내장 OCR 을 사용하려면 optional dependency '${name}' 이 필요합니다. ` +\n        `\\`npm install ${name}\\` 후 다시 실행하세요. 원인: ${(e as Error).message}`,\n    )\n  }\n}\n\n// ─── 엔진 싱글턴 (watch/서버 장기 실행에서 세션 재사용) ───\nlet enginePromise: Promise<OcrEngine> | null = null\n\nexport function getOcrEngine(): Promise<OcrEngine> {\n  if (!enginePromise) {\n    enginePromise = OcrEngine.create().catch(err => {\n      enginePromise = null // 실패는 캐시하지 않음 — 모델 설치 후 재시도 가능\n      throw err\n    })\n  }\n  return enginePromise\n}\n","/**\n * PDF OCR 브릿지 — 대상 페이지를 pdfium 으로 래스터해 내장 엔진 또는\n * 사용자 OcrProvider 로 인식하고, 페이지별 IRBlock[] 을 돌려준다.\n *\n * - 내장 엔진: OcrItem 좌표를 PDF 포인트로 환산해 기존 블록 파이프라인\n *   (extractPageBlocksWithLines — xy-cut 읽기순서·클러스터 표 감지)에 태운다.\n *   스캔 문서도 표 구조 복원이 가능한 이유.\n * - 사용자 프로바이더: 페이지 PNG → 텍스트 → 페이지당 paragraph 블록 (종전 계약).\n *\n * 에러 계약 (v4.1.0 리뷰 F6/F7):\n * - 환경 오류(의존성·모델 미설치, PDF 오픈 실패)는 throw — 호출자가 NEEDS_OCR 폴백.\n * - 페이지 단위 실패는 본문 오염 대신 warnings(OCR_FAILED) 로 기록하고 계속.\n */\n\nimport type { IRBlock, OcrProvider, ParseWarning } from \"../types.js\"\nimport type { NormItem } from \"../pdf/text-line.js\"\nimport { extractPageBlocksWithLines } from \"../pdf/page-blocks.js\"\nimport { getOcrEngine, type OcrItem } from \"./engine.js\"\nimport { ensureOcrModels } from \"./models.js\"\n\n/** OCR 렌더 스케일 (72dpi × 3 = 216dpi) — 10pt 본문이 rec 입력 높이(48px)에 근접 */\nconst OCR_RENDER_SCALE = 3\n/** 페이지 하나당 OCR 타임아웃 — det+rec 수십 라인 기준 넉넉히 */\nconst PAGE_TIMEOUT_MS = 120_000\n\nexport type OcrMode = \"builtin\" | OcrProvider\n\n/**\n * 대상 페이지들을 OCR 해 페이지별 블록 맵 반환.\n * @param buffer 원본 PDF (pdfjs 가 detach 하기 전에 clone 해 둔 것)\n * @param targets 1-based 페이지 번호 집합\n */\nexport async function runPdfOcr(\n  buffer: ArrayBuffer,\n  targets: Set<number>,\n  mode: OcrMode,\n  warnings: ParseWarning[],\n  onProgress?: (current: number, total: number) => void,\n): Promise<Map<number, IRBlock[]>> {\n  const result = new Map<number, IRBlock[]>()\n  if (targets.size === 0) return result\n\n  // 환경 준비 — 실패는 그대로 throw (호출자가 NEEDS_OCR 폴백)\n  const pdfiumMod = await tryImport<typeof import(\"@hyzyla/pdfium\")>(\n    \"@hyzyla/pdfium\",\n    () => import(\"@hyzyla/pdfium\"),\n  )\n  if (mode === \"builtin\") {\n    await ensureOcrModels(p => {\n      if (p.phase === \"download\" && p.downloaded === 0) {\n        process.stderr.write(`[kordoc-ocr] ${p.spec.name} 다운로드 중 (~${p.spec.sizeMb}MB)...\\n`)\n      }\n    })\n  }\n  const engine = mode === \"builtin\" ? await getOcrEngine() : null\n\n  const pdfium = await pdfiumMod.PDFiumLibrary.init()\n  const doc = await pdfium.loadDocument(new Uint8Array(buffer))\n  try {\n    let done = 0\n    for (const page of doc.pages()) {\n      // pdfium page.number 는 0-based pageIndex — 대외 계약(1-based)으로 환산\n      const pageNo = page.number + 1\n      if (!targets.has(pageNo)) continue\n      onProgress?.(++done, targets.size)\n      try {\n        const blocks = await withTimeout(\n          ocrOnePage(page, pageNo, mode, engine, warnings),\n          PAGE_TIMEOUT_MS,\n          `OCR 페이지 ${pageNo} 타임아웃 (${PAGE_TIMEOUT_MS / 1000}초)`,\n        )\n        result.set(pageNo, blocks)\n      } catch (e) {\n        warnings.push({\n          page: pageNo,\n          message: `페이지 ${pageNo} OCR 실패: ${e instanceof Error ? e.message : String(e)}`,\n          code: \"OCR_FAILED\",\n        })\n      }\n    }\n  } finally {\n    doc.destroy()\n    pdfium.destroy()\n  }\n  return result\n}\n\nasync function ocrOnePage(\n  page: import(\"@hyzyla/pdfium\").PDFiumPage,\n  pageNo: number,\n  mode: OcrMode,\n  engine: Awaited<ReturnType<typeof getOcrEngine>> | null,\n  warnings: ParseWarning[],\n): Promise<IRBlock[]> {\n  const { originalWidth: pdfW, originalHeight: pdfH } = page.getOriginalSize()\n  const rendered = await page.render({\n    scale: OCR_RENDER_SCALE,\n    render: async ({ data }) => data,\n  })\n  const { data: bgra, width: rw, height: rh } = rendered\n  const rgba = bgraToRgba(bgra)\n\n  if (mode === \"builtin\") {\n    const items = await engine!.recognizePage(rgba, rw, rh)\n    return ocrItemsToBlocks(items, pageNo, pdfW, pdfH, rh / pdfH)\n  }\n\n  // 사용자 프로바이더 — PNG 인코딩 후 호출, 페이지당 paragraph (종전 계약)\n  type SharpPngFactory = (\n    input: Uint8Array,\n    opts: { raw: { width: number; height: number; channels: number } },\n  ) => { png(): { toBuffer(): Promise<Buffer> } }\n  const sharpModRaw = await tryImport<{ default?: SharpPngFactory } & SharpPngFactory>(\n    \"sharp\",\n    () => import(\"sharp\") as unknown as Promise<{ default?: SharpPngFactory } & SharpPngFactory>,\n  )\n  const sharpAny = sharpModRaw as { default?: SharpPngFactory } | SharpPngFactory\n  const sharp: SharpPngFactory =\n    typeof sharpAny === \"function\" ? sharpAny : (sharpAny.default ?? (sharpAny as unknown as SharpPngFactory))\n  const png = await sharp(rgba, { raw: { width: rw, height: rh, channels: 4 } })\n    .png()\n    .toBuffer()\n\n  const text = await mode(new Uint8Array(png), pageNo, \"image/png\")\n  if (!text.trim()) {\n    warnings.push({ page: pageNo, message: `페이지 ${pageNo} OCR 결과 없음`, code: \"OCR_FAILED\" })\n    return []\n  }\n  return [{ type: \"paragraph\", text: text.trim(), pageNumber: pageNo }]\n}\n\n/**\n * OcrItem(렌더 픽셀, top-left origin) → NormItem(PDF pt, bottom-up baseline)\n * 으로 환산해 기존 블록 파이프라인에 태운다. 선(괘선) 정보는 없으므로\n * 표 감지는 클러스터 감지기 몫이다.\n */\nfunction ocrItemsToBlocks(\n  items: OcrItem[],\n  pageNumber: number,\n  pdfW: number,\n  pdfH: number,\n  scale: number,\n): IRBlock[] {\n  const norm: NormItem[] = items.map(it => {\n    const h = it.h / scale\n    return {\n      text: it.text,\n      x: Math.round(it.x / scale),\n      // NormItem.y 는 pdfjs transform[5] = 베이스라인 (bottom-up) — 박스 하단으로 근사\n      y: Math.round(pdfH - (it.y + it.h) / scale),\n      w: Math.round(it.w / scale),\n      h: Math.round(h),\n      fontSize: Math.max(1, Math.round(h * 0.8)), // 박스높이는 어센더+디센더 포함 — 글자크기 근사\n      fontName: \"ocr\",\n      isHidden: false,\n    }\n  })\n  return extractPageBlocksWithLines(norm, pageNumber, { fnArray: [], argsArray: [] }, pdfW, pdfH)\n}\n\nasync function tryImport<T>(name: string, loader: () => Promise<T>): Promise<T> {\n  try {\n    return await loader()\n  } catch (e) {\n    throw new Error(\n      `OCR 을 사용하려면 optional dependency '${name}' 이 필요합니다. ` +\n        `\\`npm install ${name}\\` 후 다시 실행하세요. 원인: ${(e as Error).message}`,\n    )\n  }\n}\n\nasync function withTimeout<T>(promise: Promise<T>, ms: number, msg: string): Promise<T> {\n  // 타임아웃 패배 후 원 promise 의 사후 reject 가 unhandled rejection 이 되지 않게 흡수\n  promise.catch(() => {})\n  let timer: NodeJS.Timeout | undefined\n  try {\n    return await Promise.race([\n      promise,\n      new Promise<never>((_, reject) => {\n        timer = setTimeout(() => reject(new Error(msg)), ms)\n      }),\n    ])\n  } finally {\n    if (timer) clearTimeout(timer)\n  }\n}\n\n/** pdfium BGRA → RGBA */\nfunction bgraToRgba(bgra: Uint8Array): Uint8Array {\n  const out = new Uint8Array(bgra.length)\n  for (let i = 0; i < bgra.length; i += 4) {\n    out[i] = bgra[i + 2]\n    out[i + 1] = bgra[i + 1]\n    out[i + 2] = bgra[i]\n    out[i + 3] = bgra[i + 3]\n  }\n  return out\n}\n"],"mappings":";;;;;;;;;;;;;;;;;AAmBA,SAAS,gBAAgB;AACzB,SAAS,YAAY;AAqBrB,IAAM,gBAAgB;AACtB,IAAM,aAAa;AACnB,IAAM,iBAAiB;AACvB,IAAM,mBAAmB;AACzB,IAAM,eAAe;AACrB,IAAM,gBAAgB;AACtB,IAAM,aAAa;AACnB,IAAM,gBAAgB;AACtB,IAAM,gBAAgB;AACtB,IAAM,aAAa;AAGnB,IAAM,WAAW,CAAC,OAAO,OAAO,KAAK;AACrC,IAAM,UAAU,CAAC,OAAO,OAAO,KAAK;AAa7B,IAAM,YAAN,MAAM,WAAU;AAAA,EACb;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YAAY,OAMjB;AACD,SAAK,MAAM,MAAM;AACjB,SAAK,MAAM,MAAM;AACjB,SAAK,OAAO,MAAM;AAClB,SAAK,MAAM,MAAM;AACjB,SAAK,QAAQ,MAAM;AAAA,EACrB;AAAA,EAEA,aAAa,SAA6B;AACxC,UAAM,CAAC,QAAQ,WAAW,IAAI,MAAM,QAAQ,IAAI;AAAA,MAC9C,UAA6C,oBAAoB,MAAM,OAAO,kBAAkB,CAAC;AAAA,MACjG;AAAA,QACE;AAAA,QACA,MAAM,OAAO,OAAO;AAAA,MACtB;AAAA,IACF,CAAC;AACD,UAAM,WAAW;AACjB,UAAM,WACJ,OAAO,aAAa,aAAa,WAAY,SAAS,WAAY;AAEpE,UAAM,MAAM,gBAAgB;AAC5B,UAAM,cAA0E;AAAA,MAC9E,wBAAwB;AAAA,MACxB,oBAAoB,CAAC,KAAK;AAAA,MAC1B,kBAAkB;AAAA;AAAA,IACpB;AACA,UAAM,CAAC,KAAK,KAAK,OAAO,IAAI,MAAM,QAAQ,IAAI;AAAA,MAC5C,OAAO,iBAAiB,OAAO,KAAK,KAAK,cAAc,QAAQ,GAAG,WAAW;AAAA,MAC7E,OAAO,iBAAiB,OAAO,KAAK,KAAK,cAAc,QAAQ,GAAG,WAAW;AAAA,MAC7E,SAAS,KAAK,KAAK,aAAa,QAAQ,GAAG,OAAO;AAAA,IACpD,CAAC;AACD,UAAM,OAAO,mBAAmB,OAAO;AACvC,QAAI,KAAK,WAAW,EAAG,OAAM,IAAI,MAAM,wJAAqC;AAE5E,WAAO,IAAI,WAAU,EAAE,KAAK,KAAK,MAAM,KAAK,QAAQ,OAAO,SAAS,CAAC;AAAA,EACvE;AAAA;AAAA,EAGA,MAAM,UAAyB;AAC7B,eAAW,KAAK,CAAC,KAAK,KAAK,KAAK,GAAG,GAAG;AACpC,YAAM,MAAO,EAAmD;AAChE,UAAI,OAAO,QAAQ,YAAY;AAC7B,YAAI;AAAE,gBAAM,IAAI,KAAK,CAAC;AAAA,QAAE,QAAQ;AAAA,QAAe;AAAA,MACjD;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,MAAM,cAAc,MAAkB,OAAe,QAAoC;AACvF,QAAI,QAAQ,gBAAgB,SAAS,aAAc,QAAO,CAAC;AAC3D,UAAM,QAAQ,MAAM,KAAK,OAAO,MAAM,OAAO,MAAM;AACnD,UAAM,QAAmB,CAAC;AAC1B,eAAW,KAAK,OAAO;AACrB,YAAM,IAAI,MAAM,KAAK,cAAc,MAAM,OAAO,QAAQ,CAAC;AACzD,UAAI,KAAK,EAAE,cAAc,cAAc,EAAE,KAAK,KAAK,EAAG,OAAM,KAAK,CAAC;AAAA,IACpE;AACA,UAAM,KAAK,CAAC,GAAG,MAAO,EAAE,IAAI,EAAE,KAAO,EAAE,IAAI,EAAE,CAAE;AAC/C,WAAO;AAAA,EACT;AAAA;AAAA,EAIA,MAAc,OACZ,MACA,OACA,QACgE;AAChE,UAAM,QAAQ,gBAAgB,KAAK,IAAI,OAAO,MAAM;AACpD,UAAM,KAAK,KAAK,IAAI,IAAI,KAAK,MAAO,QAAQ,QAAS,EAAE,IAAI,EAAE;AAC7D,UAAM,KAAK,KAAK,IAAI,IAAI,KAAK,MAAO,SAAS,QAAS,EAAE,IAAI,EAAE;AAE9D,UAAM,MAAM,MAAM,KAAK,MAAM,MAAM,EAAE,KAAK,EAAE,OAAO,QAAQ,UAAU,EAAE,EAAE,CAAC,EACvE,OAAO,IAAI,IAAI,EAAE,KAAK,OAAO,CAAC,EAC9B,YAAY,EACZ,IAAI,EACJ,SAAS;AAGZ,UAAM,QAAQ,KAAK;AACnB,UAAM,QAAQ,IAAI,aAAa,IAAI,KAAK;AACxC,aAAS,IAAI,GAAG,IAAI,OAAO,KAAK;AAC9B,YAAM,IAAI,IAAI,IAAI,CAAC,IAAI;AACvB,YAAM,IAAI,IAAI,IAAI,IAAI,CAAC,IAAI;AAC3B,YAAM,IAAI,IAAI,IAAI,IAAI,CAAC,IAAI;AAC3B,YAAM,CAAC,KAAK,IAAI,SAAS,CAAC,KAAK,QAAQ,CAAC;AACxC,YAAM,QAAQ,CAAC,KAAK,IAAI,SAAS,CAAC,KAAK,QAAQ,CAAC;AAChD,YAAM,IAAI,QAAQ,CAAC,KAAK,IAAI,SAAS,CAAC,KAAK,QAAQ,CAAC;AAAA,IACtD;AAEA,UAAM,SAAS,IAAI,KAAK,IAAI,OAAO,WAAW,OAAO,CAAC,GAAG,GAAG,IAAI,EAAE,CAAC;AACnE,UAAM,MAAM,MAAM,KAAK,IAAI,IAAI,EAAE,CAAC,KAAK,IAAI,WAAW,CAAC,CAAC,GAAG,OAAO,CAAC;AACnE,UAAM,UAAU,IAAI,KAAK,IAAI,YAAY,CAAC,CAAC,EAAE;AAE7C,UAAM,WAAW,eAAe,SAAS,IAAI,EAAE;AAC/C,UAAM,KAAK,QAAQ;AACnB,UAAM,KAAK,SAAS;AACpB,UAAM,QAA+D,CAAC;AACtE,eAAW,MAAM,SAAS,MAAM,GAAG,aAAa,GAAG;AAEjD,YAAM,KAAK,GAAG,KAAK,GAAG,KAAK;AAC3B,YAAM,KAAK,GAAG,KAAK,GAAG,KAAK;AAC3B,YAAM,QAAS,KAAK,KAAK,oBAAqB,KAAK,KAAK;AACxD,YAAM,KAAK,KAAK,IAAI,GAAG,KAAK,OAAO,GAAG,KAAK,SAAS,EAAE,CAAC;AACvD,YAAM,KAAK,KAAK,IAAI,GAAG,KAAK,OAAO,GAAG,KAAK,SAAS,EAAE,CAAC;AACvD,YAAM,KAAK,KAAK,IAAI,OAAO,KAAK,MAAM,GAAG,KAAK,IAAI,SAAS,EAAE,CAAC;AAC9D,YAAM,KAAK,KAAK,IAAI,QAAQ,KAAK,MAAM,GAAG,KAAK,IAAI,SAAS,EAAE,CAAC;AAC/D,UAAI,KAAK,KAAK,gBAAgB,KAAK,KAAK,aAAc;AACtD,YAAM,KAAK,EAAE,GAAG,IAAI,GAAG,IAAI,GAAG,KAAK,IAAI,GAAG,KAAK,GAAG,CAAC;AAAA,IACrD;AACA,WAAO;AAAA,EACT;AAAA;AAAA,EAIA,MAAc,cACZ,MACA,OACA,QACA,KACyB;AACzB,UAAM,KAAK,KAAK,IAAI,eAAe,KAAK,IAAI,IAAI,KAAK,MAAO,IAAI,IAAI,aAAc,IAAI,CAAC,CAAC,CAAC;AACzF,UAAM,SAAS,KAAK,IAAI,eAAe,EAAE;AAEzC,UAAM,MAAM,MAAM,KAAK,MAAM,MAAM,EAAE,KAAK,EAAE,OAAO,QAAQ,UAAU,EAAE,EAAE,CAAC,EACvE,QAAQ,EAAE,MAAM,IAAI,GAAG,KAAK,IAAI,GAAG,OAAO,IAAI,GAAG,QAAQ,IAAI,EAAE,CAAC,EAChE,OAAO,IAAI,YAAY,EAAE,KAAK,OAAO,CAAC,EACtC,YAAY,EACZ,IAAI,EACJ,SAAS;AAGZ,UAAM,QAAQ,SAAS;AACvB,UAAM,QAAQ,IAAI,aAAa,IAAI,KAAK;AACxC,aAAS,IAAI,GAAG,IAAI,YAAY,KAAK;AACnC,eAAS,IAAI,GAAG,IAAI,IAAI,KAAK;AAC3B,cAAM,OAAO,IAAI,KAAK,KAAK;AAC3B,cAAM,MAAM,IAAI,SAAS;AACzB,cAAM,GAAG,IAAI,IAAI,MAAM,CAAC,IAAI,MAAM,MAAM;AACxC,cAAM,QAAQ,GAAG,IAAI,IAAI,MAAM,CAAC,IAAI,MAAM,MAAM;AAChD,cAAM,IAAI,QAAQ,GAAG,IAAI,IAAI,GAAG,IAAI,MAAM,MAAM;AAAA,MAClD;AAAA,IACF;AAEA,UAAM,SAAS,IAAI,KAAK,IAAI,OAAO,WAAW,OAAO,CAAC,GAAG,GAAG,YAAY,MAAM,CAAC;AAC/E,UAAM,MAAM,MAAM,KAAK,IAAI,IAAI,EAAE,CAAC,KAAK,IAAI,WAAW,CAAC,CAAC,GAAG,OAAO,CAAC;AACnE,UAAM,SAAS,IAAI,KAAK,IAAI,YAAY,CAAC,CAAC;AAC1C,UAAM,CAAC,EAAE,GAAG,CAAC,IAAI,OAAO;AACxB,UAAM,OAAO,OAAO;AAEpB,UAAM,UAAU,UAAU,MAAM,GAAG,GAAG,KAAK,IAAI;AAC/C,QAAI,CAAC,QAAS,QAAO;AACrB,WAAO,EAAE,MAAM,QAAQ,MAAM,GAAG,IAAI,GAAG,GAAG,IAAI,GAAG,GAAG,IAAI,GAAG,GAAG,IAAI,GAAG,YAAY,QAAQ,WAAW;AAAA,EACtG;AACF;AAGO,SAAS,UACd,MACA,GACA,GACA,MAC6C;AAC7C,MAAI,OAAO;AACX,MAAI,UAAU;AACd,MAAI,YAAY;AAChB,MAAI,OAAO;AACX,WAAS,IAAI,GAAG,IAAI,GAAG,KAAK;AAC1B,UAAM,MAAM,IAAI;AAChB,QAAI,OAAO;AACX,QAAI,QAAQ,KAAK,GAAG;AACpB,aAAS,IAAI,GAAG,IAAI,GAAG,KAAK;AAC1B,YAAM,IAAI,KAAK,MAAM,CAAC;AACtB,UAAI,IAAI,OAAO;AAAE,gBAAQ;AAAG,eAAO;AAAA,MAAE;AAAA,IACvC;AACA,UAAM,SAAS,SAAS;AACxB,WAAO;AACP,QAAI,SAAS,KAAK,OAAQ;AAE1B,QAAI,IAAI;AACR,QAAI,IAAI,UAAU,IAAI,GAAG;AACvB,UAAI,QAAQ;AACZ,eAAS,IAAI,GAAG,IAAI,GAAG,IAAK,UAAS,KAAK,IAAI,KAAK,MAAM,CAAC,IAAI,KAAK;AACnE,UAAI,IAAI;AAAA,IACV;AACA,eAAW;AACX;AACA,QAAI,QAAQ,KAAK,QAAQ,KAAK,OAAQ,SAAQ,KAAK,OAAO,CAAC;AAAA,aAClD,SAAS,KAAK,SAAS,EAAG,SAAQ;AAAA,EAC7C;AACA,MAAI,CAAC,KAAM,QAAO;AAClB,SAAO,EAAE,MAAM,YAAY,YAAY,IAAI,UAAU,YAAY,EAAE;AACrE;AAGO,SAAS,eACd,MACA,GACA,GAC2D;AAC3D,QAAM,UAAU,IAAI,WAAW,IAAI,CAAC;AACpC,QAAM,QAAkF,CAAC;AACzF,QAAM,QAAkB,CAAC;AAEzB,WAAS,QAAQ,GAAG,QAAQ,IAAI,GAAG,SAAS;AAC1C,QAAI,QAAQ,KAAK,KAAK,KAAK,KAAK,KAAK,WAAY;AACjD,QAAI,KAAK,QAAQ,GAAG,KAAK,IAAI,KAAM,QAAQ,IAAK,GAAG,KAAK;AACxD,QAAI,MAAM;AACV,QAAI,QAAQ;AACZ,UAAM,SAAS;AACf,UAAM,KAAK,KAAK;AAChB,YAAQ,KAAK,IAAI;AACjB,WAAO,MAAM,QAAQ;AACnB,YAAM,IAAI,MAAM,IAAI;AACpB,YAAM,KAAK,IAAI;AACf,YAAM,KAAM,IAAI,IAAK;AACrB,aAAO,KAAK,CAAC;AACb;AACA,UAAI,KAAK,GAAI,MAAK;AAClB,UAAI,KAAK,GAAI,MAAK;AAClB,UAAI,KAAK,GAAI,MAAK;AAClB,UAAI,KAAK,GAAI,MAAK;AAElB,UAAI,KAAK,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,IAAI,YAAY;AAAE,gBAAQ,IAAI,CAAC,IAAI;AAAG,cAAM,KAAK,IAAI,CAAC;AAAA,MAAE;AACnG,UAAI,KAAK,IAAI,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,IAAI,YAAY;AAAE,gBAAQ,IAAI,CAAC,IAAI;AAAG,cAAM,KAAK,IAAI,CAAC;AAAA,MAAE;AACvG,UAAI,KAAK,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,IAAI,YAAY;AAAE,gBAAQ,IAAI,CAAC,IAAI;AAAG,cAAM,KAAK,IAAI,CAAC;AAAA,MAAE;AACnG,UAAI,KAAK,IAAI,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,IAAI,YAAY;AAAE,gBAAQ,IAAI,CAAC,IAAI;AAAG,cAAM,KAAK,IAAI,CAAC;AAAA,MAAE;AAAA,IACzG;AACA,QAAI,KAAK,KAAK,IAAI,gBAAgB,KAAK,KAAK,IAAI,aAAc;AAC9D,UAAM,KAAK,EAAE,IAAI,IAAI,IAAI,IAAI,OAAO,MAAM,MAAM,CAAC;AAAA,EACnD;AAEA,SAAO,MACJ,OAAO,OAAK,EAAE,SAAS,cAAc,EACrC,KAAK,CAAC,GAAG,MAAO,EAAE,KAAK,EAAE,MAAQ,EAAE,KAAK,EAAE,EAAG;AAClD;AAEA,eAAe,UAAa,MAAc,QAAsC;AAC9E,MAAI;AACF,WAAO,MAAM,OAAO;AAAA,EACtB,SAAS,GAAG;AACV,UAAM,IAAI;AAAA,MACR,+EAAuC,IAAI,0DACxB,IAAI,wEAAuB,EAAY,OAAO;AAAA,IACnE;AAAA,EACF;AACF;AAGA,IAAI,gBAA2C;AAExC,SAAS,eAAmC;AACjD,MAAI,CAAC,eAAe;AAClB,oBAAgB,UAAU,OAAO,EAAE,MAAM,SAAO;AAC9C,sBAAgB;AAChB,YAAM;AAAA,IACR,CAAC;AAAA,EACH;AACA,SAAO;AACT;;;AChUA,IAAM,mBAAmB;AAEzB,IAAM,kBAAkB;AASxB,eAAsB,UACpB,QACA,SACA,MACA,UACA,YACiC;AACjC,QAAM,SAAS,oBAAI,IAAuB;AAC1C,MAAI,QAAQ,SAAS,EAAG,QAAO;AAG/B,QAAM,YAAY,MAAMA;AAAA,IACtB;AAAA,IACA,MAAM,OAAO,gBAAgB;AAAA,EAC/B;AACA,MAAI,SAAS,WAAW;AACtB,UAAM,gBAAgB,OAAK;AACzB,UAAI,EAAE,UAAU,cAAc,EAAE,eAAe,GAAG;AAChD,gBAAQ,OAAO,MAAM,gBAAgB,EAAE,KAAK,IAAI,sCAAa,EAAE,KAAK,MAAM;AAAA,CAAU;AAAA,MACtF;AAAA,IACF,CAAC;AAAA,EACH;AACA,QAAM,SAAS,SAAS,YAAY,MAAM,aAAa,IAAI;AAE3D,QAAM,SAAS,MAAM,UAAU,cAAc,KAAK;AAClD,QAAM,MAAM,MAAM,OAAO,aAAa,IAAI,WAAW,MAAM,CAAC;AAC5D,MAAI;AACF,QAAI,OAAO;AACX,eAAW,QAAQ,IAAI,MAAM,GAAG;AAE9B,YAAM,SAAS,KAAK,SAAS;AAC7B,UAAI,CAAC,QAAQ,IAAI,MAAM,EAAG;AAC1B,mBAAa,EAAE,MAAM,QAAQ,IAAI;AACjC,UAAI;AACF,cAAM,SAAS,MAAM;AAAA,UACnB,WAAW,MAAM,QAAQ,MAAM,QAAQ,QAAQ;AAAA,UAC/C;AAAA,UACA,0BAAW,MAAM,8BAAU,kBAAkB,GAAI;AAAA,QACnD;AACA,eAAO,IAAI,QAAQ,MAAM;AAAA,MAC3B,SAAS,GAAG;AACV,iBAAS,KAAK;AAAA,UACZ,MAAM;AAAA,UACN,SAAS,sBAAO,MAAM,sBAAY,aAAa,QAAQ,EAAE,UAAU,OAAO,CAAC,CAAC;AAAA,UAC5E,MAAM;AAAA,QACR,CAAC;AAAA,MACH;AAAA,IACF;AAAA,EACF,UAAE;AACA,QAAI,QAAQ;AACZ,WAAO,QAAQ;AAAA,EACjB;AACA,SAAO;AACT;AAEA,eAAe,WACb,MACA,QACA,MACA,QACA,UACoB;AACpB,QAAM,EAAE,eAAe,MAAM,gBAAgB,KAAK,IAAI,KAAK,gBAAgB;AAC3E,QAAM,WAAW,MAAM,KAAK,OAAO;AAAA,IACjC,OAAO;AAAA,IACP,QAAQ,OAAO,EAAE,KAAK,MAAM;AAAA,EAC9B,CAAC;AACD,QAAM,EAAE,MAAM,MAAM,OAAO,IAAI,QAAQ,GAAG,IAAI;AAC9C,QAAM,OAAO,WAAW,IAAI;AAE5B,MAAI,SAAS,WAAW;AACtB,UAAM,QAAQ,MAAM,OAAQ,cAAc,MAAM,IAAI,EAAE;AACtD,WAAO,iBAAiB,OAAO,QAAQ,MAAM,MAAM,KAAK,IAAI;AAAA,EAC9D;AAOA,QAAM,cAAc,MAAMA;AAAA,IACxB;AAAA,IACA,MAAM,OAAO,OAAO;AAAA,EACtB;AACA,QAAM,WAAW;AACjB,QAAM,QACJ,OAAO,aAAa,aAAa,WAAY,SAAS,WAAY;AACpE,QAAM,MAAM,MAAM,MAAM,MAAM,EAAE,KAAK,EAAE,OAAO,IAAI,QAAQ,IAAI,UAAU,EAAE,EAAE,CAAC,EAC1E,IAAI,EACJ,SAAS;AAEZ,QAAM,OAAO,MAAM,KAAK,IAAI,WAAW,GAAG,GAAG,QAAQ,WAAW;AAChE,MAAI,CAAC,KAAK,KAAK,GAAG;AAChB,aAAS,KAAK,EAAE,MAAM,QAAQ,SAAS,sBAAO,MAAM,kCAAc,MAAM,aAAa,CAAC;AACtF,WAAO,CAAC;AAAA,EACV;AACA,SAAO,CAAC,EAAE,MAAM,aAAa,MAAM,KAAK,KAAK,GAAG,YAAY,OAAO,CAAC;AACtE;AAOA,SAAS,iBACP,OACA,YACA,MACA,MACA,OACW;AACX,QAAM,OAAmB,MAAM,IAAI,QAAM;AACvC,UAAM,IAAI,GAAG,IAAI;AACjB,WAAO;AAAA,MACL,MAAM,GAAG;AAAA,MACT,GAAG,KAAK,MAAM,GAAG,IAAI,KAAK;AAAA;AAAA,MAE1B,GAAG,KAAK,MAAM,QAAQ,GAAG,IAAI,GAAG,KAAK,KAAK;AAAA,MAC1C,GAAG,KAAK,MAAM,GAAG,IAAI,KAAK;AAAA,MAC1B,GAAG,KAAK,MAAM,CAAC;AAAA,MACf,UAAU,KAAK,IAAI,GAAG,KAAK,MAAM,IAAI,GAAG,CAAC;AAAA;AAAA,MACzC,UAAU;AAAA,MACV,UAAU;AAAA,IACZ;AAAA,EACF,CAAC;AACD,SAAO,2BAA2B,MAAM,YAAY,EAAE,SAAS,CAAC,GAAG,WAAW,CAAC,EAAE,GAAG,MAAM,IAAI;AAChG;AAEA,eAAeA,WAAa,MAAc,QAAsC;AAC9E,MAAI;AACF,WAAO,MAAM,OAAO;AAAA,EACtB,SAAS,GAAG;AACV,UAAM,IAAI;AAAA,MACR,kEAAoC,IAAI,0DACrB,IAAI,wEAAuB,EAAY,OAAO;AAAA,IACnE;AAAA,EACF;AACF;AAEA,eAAe,YAAe,SAAqB,IAAY,KAAyB;AAEtF,UAAQ,MAAM,MAAM;AAAA,EAAC,CAAC;AACtB,MAAI;AACJ,MAAI;AACF,WAAO,MAAM,QAAQ,KAAK;AAAA,MACxB;AAAA,MACA,IAAI,QAAe,CAAC,GAAG,WAAW;AAChC,gBAAQ,WAAW,MAAM,OAAO,IAAI,MAAM,GAAG,CAAC,GAAG,EAAE;AAAA,MACrD,CAAC;AAAA,IACH,CAAC;AAAA,EACH,UAAE;AACA,QAAI,MAAO,cAAa,KAAK;AAAA,EAC/B;AACF;AAGA,SAAS,WAAW,MAA8B;AAChD,QAAM,MAAM,IAAI,WAAW,KAAK,MAAM;AACtC,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK,GAAG;AACvC,QAAI,CAAC,IAAI,KAAK,IAAI,CAAC;AACnB,QAAI,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC;AACvB,QAAI,IAAI,CAAC,IAAI,KAAK,CAAC;AACnB,QAAI,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC;AAAA,EACzB;AACA,SAAO;AACT;","names":["tryImport"]}

package/dist/{profile-io-PLDQFPJA.js → profile-io-TWHDIVPL.js}

@@ -2,10 +2,10 @@
 import {
   formatProfileSchema,
   parseFormatProfileJson
-} from "./chunk-WO52HVQJ.js";
-import "./chunk-O5VS7RIR.js";
+} from "./chunk-PLIY23LY.js";
+import "./chunk-FSXLSHCW.js";
 export {
   formatProfileSchema,
   parseFormatProfileJson
 };
-//# sourceMappingURL=profile-io-PLDQFPJA.js.map
+//# sourceMappingURL=profile-io-TWHDIVPL.js.map

package/dist/{rasterize-WWNQLKYW.js → rasterize-W2NEZUSW.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   KordocError
-} from "./chunk-O5VS7RIR.js";
+} from "./chunk-FSXLSHCW.js";
 
 // src/render/rasterize.ts
 async function rasterizeSvg(svg, widthPt, heightPt, options) {
@@ -37,4 +37,4 @@ async function rasterizeSvg(svg, widthPt, heightPt, options) {
 export {
   rasterizeSvg
 };
-//# sourceMappingURL=rasterize-WWNQLKYW.js.map
+//# sourceMappingURL=rasterize-W2NEZUSW.js.map

package/dist/render-4MG5QUDG.js

@@ -0,0 +1,10 @@
+#!/usr/bin/env node
+import {
+  renderHwpxToSvg
+} from "./chunk-6SHO3BNV.js";
+import "./chunk-V6U6YS5B.js";
+import "./chunk-FSXLSHCW.js";
+export {
+  renderHwpxToSvg
+};
+//# sourceMappingURL=render-4MG5QUDG.js.map

package/dist/seal-D2QNXJVO.js

@@ -0,0 +1,10 @@
+#!/usr/bin/env node
+import {
+  placeSealHwpx
+} from "./chunk-WQLHPOMJ.js";
+import "./chunk-V6U6YS5B.js";
+import "./chunk-FSXLSHCW.js";
+export {
+  placeSealHwpx
+};
+//# sourceMappingURL=seal-D2QNXJVO.js.map

package/dist/seal-D2QNXJVO.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/{watch-SFHXZALC.js → watch-VWXJZT2R.js}

@@ -1,21 +1,22 @@
 #!/usr/bin/env node
 import {
   parse
-} from "./chunk-L2XQQTOY.js";
-import "./chunk-UOO7LSDG.js";
+} from "./chunk-IGGRLWT5.js";
+import "./chunk-7FCNOKOR.js";
+import "./chunk-4VNLZ2GZ.js";
 import {
   detectFormat
 } from "./chunk-YOO6ET6M.js";
 import "./chunk-MOL7MDBG.js";
-import "./chunk-AX2R5Q2N.js";
+import "./chunk-WQLHPOMJ.js";
 import "./chunk-DZIXKL7E.js";
-import "./chunk-F2ZU3IZG.js";
-import "./chunk-SSTK6IKK.js";
-import "./chunk-37HJHCPA.js";
-import "./chunk-GQVSHGG4.js";
+import "./chunk-ELOV2GIJ.js";
+import "./chunk-CMN3SYZ3.js";
+import "./chunk-6SHO3BNV.js";
+import "./chunk-V6U6YS5B.js";
 import {
   toArrayBuffer
-} from "./chunk-O5VS7RIR.js";
+} from "./chunk-FSXLSHCW.js";
 
 // src/watch.ts
 import { watch, readFileSync, writeFileSync, mkdirSync, statSync, existsSync, realpathSync } from "fs";
@@ -57,7 +58,6 @@ function createTaskQueue(maxConcurrent, run) {
 async function watchDirectory(options) {
   const { dir, outDir, webhook, format = "markdown", pages, silent } = options;
   if (!existsSync(dir)) throw new Error(`\uB514\uB809\uD1A0\uB9AC\uB97C \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4: ${dir}`);
-  assertRecursiveWatchSupport();
   if (webhook) validateWebhookUrl(webhook);
   if (outDir) mkdirSync(outDir, { recursive: true });
   const log = silent ? () => {
@@ -65,6 +65,7 @@ async function watchDirectory(options) {
   log(`[kordoc watch] \uAC10\uC2DC \uC2DC\uC791: ${resolve(dir)}`);
   if (outDir) log(`[kordoc watch] \uCD9C\uB825: ${resolve(outDir)}`);
   if (webhook) log(`[kordoc watch] \uC6F9\uD6C5: ${webhook}`);
+  assertRecursiveWatchSupport();
   const pending = /* @__PURE__ */ new Map();
   const MAX_CONCURRENT = 3;
   const waitForStableSize = async (absPath) => {
@@ -205,4 +206,4 @@ export {
   isPrivateIp,
   watchDirectory
 };
-//# sourceMappingURL=watch-SFHXZALC.js.map
+//# sourceMappingURL=watch-VWXJZT2R.js.map

package/dist/watch-VWXJZT2R.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/watch.ts"],"sourcesContent":["/** 디렉토리 감시 모드 — 새 문서 자동 변환 + Webhook 알림 */\n\nimport { watch, readFileSync, writeFileSync, mkdirSync, statSync, existsSync, realpathSync } from \"fs\"\nimport { lookup } from \"dns/promises\"\nimport { basename, resolve, extname, sep } from \"path\"\nimport { parse, detectFormat } from \"./index.js\"\nimport { toArrayBuffer } from \"./utils.js\"\nimport type { WatchOptions, ParseOptions } from \"./types.js\"\n\nexport const SUPPORTED_EXTENSIONS = new Set([\".hwp\", \".hwpx\", \".hml\", \".pdf\", \".xls\", \".xlsx\", \".docx\"])\nconst DEBOUNCE_MS = 1000\n/** 파일 쓰기 완료 판정: 연속 2회 동일 크기 확인 간격 */\nconst STABLE_CHECK_MS = 300\nconst MAX_FILE_SIZE = 500 * 1024 * 1024\n\n/**\n * Linux fs.watch recursive는 Node 19.1+ 전용 — 미지원 조합이면 크립틱\n * ERR_FEATURE_UNAVAILABLE 대신 명확한 안내로 즉시 실패시킨다.\n */\nexport function assertRecursiveWatchSupport(\n  platform: NodeJS.Platform = process.platform,\n  nodeVersion: string = process.versions.node,\n): void {\n  if (platform !== \"linux\") return\n  const [major = 0, minor = 0] = nodeVersion.split(\".\").map(Number)\n  if (major >= 20 || (major === 19 && minor >= 1)) return\n  throw new Error(\n    `Linux에서 디렉토리 감시(fs.watch recursive)는 Node 19.1 이상이 필요합니다 (현재 ${nodeVersion}) — Node 20+ 업그레이드를 권장합니다`,\n  )\n}\n\n/**\n * 동시 처리 상한 게이트 — 상한 초과분은 버리지 않고 대기 큐에 넣어 처리 완료 시\n * 순서대로 drain (기존: 상한 초과 return으로 새 파일이 무음 유실되던 것 방지).\n */\nexport function createTaskQueue(maxConcurrent: number, run: (key: string) => Promise<void>): (key: string) => void {\n  let active = 0\n  const inProgress = new Set<string>()\n  const waiting: string[] = []\n  const enqueue = (key: string): void => {\n    if (inProgress.has(key)) return // 동일 파일 동시 처리 방지\n    if (active >= maxConcurrent) {\n      if (!waiting.includes(key)) waiting.push(key)\n      return\n    }\n    inProgress.add(key)\n    active++\n    void run(key).finally(() => {\n      inProgress.delete(key)\n      active--\n      const next = waiting.shift()\n      if (next) enqueue(next)\n    })\n  }\n  return enqueue\n}\n\n/**\n * 디렉토리를 감시하여 새 문서 파일을 자동 변환.\n *\n * @example\n * ```bash\n * kordoc watch ./incoming -d ./output --webhook https://api.example.com/docs\n * ```\n */\nexport async function watchDirectory(options: WatchOptions): Promise<void> {\n  const { dir, outDir, webhook, format = \"markdown\", pages, silent } = options\n\n  if (!existsSync(dir)) throw new Error(`디렉토리를 찾을 수 없습니다: ${dir}`)\n  if (webhook) validateWebhookUrl(webhook)\n  if (outDir) mkdirSync(outDir, { recursive: true })\n\n  const log = silent ? () => {} : (msg: string) => process.stderr.write(msg + \"\\n\")\n  log(`[kordoc watch] 감시 시작: ${resolve(dir)}`)\n  if (outDir) log(`[kordoc watch] 출력: ${resolve(outDir)}`)\n  if (webhook) log(`[kordoc watch] 웹훅: ${webhook}`)\n\n  // 설정 로그 뒤·감시 시작 전에 검사 — 무엇을 하려다 막혔는지 보이고,\n  // 옵션 파싱 계약 테스트(plugin-5)도 Linux+Node<19.1 에서 로그를 관측할 수 있다\n  assertRecursiveWatchSupport()\n\n  // 디바운스 맵\n  const pending = new Map<string, ReturnType<typeof setTimeout>>()\n  // 동시 처리 제한 — 메모리 폭주 방지 (초과분은 대기 큐로, createTaskQueue)\n  const MAX_CONCURRENT = 3\n\n  /** 파일 크기가 안정화될 때까지 대기 (쓰기 완료 감지) */\n  const waitForStableSize = async (absPath: string): Promise<number> => {\n    let prevSize = statSync(absPath).size\n    await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\n    if (!existsSync(absPath)) return 0\n    const currSize = statSync(absPath).size\n    if (currSize !== prevSize) {\n      // 크기가 변했으면 한 번 더 대기\n      await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\n      if (!existsSync(absPath)) return 0\n      return statSync(absPath).size\n    }\n    return currSize\n  }\n\n  const processFile = async (filePath: string) => {\n    const ext = extname(filePath).toLowerCase()\n    if (!SUPPORTED_EXTENSIONS.has(ext)) return\n\n    const fileName = basename(filePath)\n    try {\n      const rawPath = resolve(dir, filePath)\n      if (!existsSync(rawPath)) return\n      // 심볼릭 링크 해석 후 감시 디렉토리 외부 파일 차단\n      let absPath: string\n      try { absPath = realpathSync(rawPath) } catch { return }\n      const realDir = realpathSync(resolve(dir))\n      if (!absPath.startsWith(realDir + sep) && absPath !== realDir) return\n\n      const fileSize = await waitForStableSize(absPath)\n      if (fileSize > MAX_FILE_SIZE || fileSize === 0) return\n\n      log(`[kordoc watch] 변환 중: ${fileName}`)\n\n      const buffer = readFileSync(absPath)\n      const arrayBuffer = toArrayBuffer(buffer)\n      // filePath 전달 — 배포용 HWP의 COM fallback에 필요 (CLI와 동일)\n      const parseOptions: ParseOptions = { filePath: absPath }\n      if (pages) parseOptions.pages = pages\n      const result = await parse(arrayBuffer, parseOptions)\n\n      if (!result.success) {\n        log(`[kordoc watch] 실패: ${fileName} — ${result.error}`)\n        await sendWebhook(webhook, { file: fileName, format: detectFormat(arrayBuffer), success: false, error: result.error })\n        return\n      }\n\n      const output = format === \"json\" ? JSON.stringify(result, null, 2) : result.markdown\n\n      if (outDir) {\n        const outExt = format === \"json\" ? \".json\" : \".md\"\n        const outPath = resolve(outDir, fileName.replace(/\\.[^.]+$/, outExt))\n        writeFileSync(outPath, output, \"utf-8\")\n        log(`[kordoc watch] 완료: ${fileName} → ${basename(outPath)}`)\n      } else {\n        process.stdout.write(output + \"\\n\")\n      }\n\n      await sendWebhook(webhook, {\n        file: fileName,\n        format: result.fileType,\n        success: true,\n        markdown: format === \"markdown\" ? output.substring(0, 1000) : undefined,\n      })\n    } catch (err) {\n      log(`[kordoc watch] 에러: ${fileName} — ${err instanceof Error ? err.message : err}`)\n    }\n  }\n\n  const enqueue = createTaskQueue(MAX_CONCURRENT, (filePath) =>\n    processFile(filePath).catch((err) => {\n      process.stderr.write(`[kordoc watch] 처리 실패: ${filePath} — ${err instanceof Error ? err.message : String(err)}\\n`)\n    }),\n  )\n\n  // fs.watch recursive (Node 18+ Windows/macOS, Node 19.1+ Linux — 위에서 사전 검증)\n  const watcher = watch(dir, { recursive: true }, (event, filename) => {\n    if (!filename) return\n    const filePath = filename.toString()\n\n    // 디바운스\n    const existing = pending.get(filePath)\n    if (existing) clearTimeout(existing)\n    pending.set(filePath, setTimeout(() => {\n      pending.delete(filePath)\n      enqueue(filePath)\n    }, DEBOUNCE_MS))\n  })\n  // 감시 자체의 런타임 에러(EMFILE 등)로 데몬이 죽지 않게 — 로그만 남기고 유지\n  watcher.on(\"error\", (err) => {\n    process.stderr.write(`[kordoc watch] 감시 오류: ${err instanceof Error ? err.message : String(err)}\\n`)\n  })\n\n  // 프로세스 종료 방지 (Ctrl+C로 종료)\n  return new Promise(() => {})\n}\n\n/** Webhook URL 검증 — SSRF 방지: http/https만 허용, localhost/private IP 차단 */\nfunction validateWebhookUrl(url: string): void {\n  let parsed: URL\n  try {\n    parsed = new URL(url)\n  } catch {\n    throw new Error(`유효하지 않은 webhook URL: ${url}`)\n  }\n  if (parsed.protocol !== \"http:\" && parsed.protocol !== \"https:\") {\n    throw new Error(`허용되지 않는 webhook 프로토콜: ${parsed.protocol}`)\n  }\n  const hostname = parsed.hostname.toLowerCase()\n  if (\n    hostname === \"localhost\" ||\n    hostname === \"[::1]\" ||\n    hostname.startsWith(\"127.\") ||\n    hostname.startsWith(\"10.\") ||\n    hostname.startsWith(\"192.168.\") ||\n    /^172\\.(1[6-9]|2\\d|3[01])\\./.test(hostname) ||\n    hostname === \"0.0.0.0\" ||\n    hostname.startsWith(\"169.254.\") ||\n    hostname.endsWith(\".local\") ||\n    // IPv6 사설 대역\n    hostname.startsWith(\"[fc\") ||\n    hostname.startsWith(\"[fd\") ||\n    hostname.startsWith(\"[fe80:\") ||\n    hostname === \"[::0]\" ||\n    hostname === \"[::]\" ||\n    // 클라우드 메타데이터 엔드포인트\n    hostname === \"metadata.google.internal\" ||\n    hostname === \"metadata.google\" ||\n    // 16진수/8진수/10진수 정수 IP 인코딩 우회 방지\n    /^0x[0-9a-f]+$/i.test(hostname) ||\n    /^0[0-7]+$/.test(hostname) ||\n    /^\\d+$/.test(hostname)\n  ) {\n    throw new Error(`내부 네트워크 대상 webhook은 허용되지 않습니다: ${hostname}`)\n  }\n}\n\n/** IP 리터럴의 사설/내부 대역 여부 — DNS 해석 결과 재검증용 (IPv4-mapped IPv6 포함) */\nexport function isPrivateIp(ip: string): boolean {\n  let addr = ip.toLowerCase()\n  // IPv4-mapped IPv6 (::ffff:10.0.0.1) → 내장 IPv4로 재검사\n  const mapped = /^::ffff:(\\d+\\.\\d+\\.\\d+\\.\\d+)$/.exec(addr)\n  if (mapped) addr = mapped[1]\n  if (/^\\d+\\.\\d+\\.\\d+\\.\\d+$/.test(addr)) {\n    const [a, b] = addr.split(\".\").map(Number)\n    return (\n      a === 0 || a === 10 || a === 127 ||\n      (a === 100 && b >= 64 && b <= 127) || // CGNAT 100.64/10\n      (a === 169 && b === 254) ||\n      (a === 172 && b >= 16 && b <= 31) ||\n      (a === 192 && b === 168)\n    )\n  }\n  // IPv6 — loopback/미지정, ULA fc00::/7, link-local fe80::/10\n  return addr === \"::\" || addr === \"::1\" || /^f[cd]/.test(addr) || /^fe[89ab]/.test(addr)\n}\n\nasync function sendWebhook(url: string | undefined, payload: Record<string, unknown>): Promise<void> {\n  if (!url) return\n  try {\n    validateWebhookUrl(url)\n    // DNS 해석 후 사설 IP 재검증 — 문자열 검사만으로는 내부망 도메인을 못 거른다\n    const hostname = new URL(url).hostname.replace(/^\\[|\\]$/g, \"\")\n    const addrs = await lookup(hostname, { all: true, verbatim: true })\n    const bad = addrs.find((a) => isPrivateIp(a.address))\n    if (bad) throw new Error(`webhook 대상이 내부 네트워크로 해석됩니다: ${hostname} → ${bad.address}`)\n    await fetch(url, {\n      method: \"POST\",\n      headers: { \"Content-Type\": \"application/json\" },\n      body: JSON.stringify({ ...payload, timestamp: new Date().toISOString() }),\n      redirect: \"error\",\n    })\n  } catch (err) {\n    process.stderr.write(`[kordoc watch] webhook 전송 실패: ${err instanceof Error ? err.message : String(err)}\\n`)\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;AAEA,SAAS,OAAO,cAAc,eAAe,WAAW,UAAU,YAAY,oBAAoB;AAClG,SAAS,cAAc;AACvB,SAAS,UAAU,SAAS,SAAS,WAAW;AAKzC,IAAM,uBAAuB,oBAAI,IAAI,CAAC,QAAQ,SAAS,QAAQ,QAAQ,QAAQ,SAAS,OAAO,CAAC;AACvG,IAAM,cAAc;AAEpB,IAAM,kBAAkB;AACxB,IAAM,gBAAgB,MAAM,OAAO;AAM5B,SAAS,4BACd,WAA4B,QAAQ,UACpC,cAAsB,QAAQ,SAAS,MACjC;AACN,MAAI,aAAa,QAAS;AAC1B,QAAM,CAAC,QAAQ,GAAG,QAAQ,CAAC,IAAI,YAAY,MAAM,GAAG,EAAE,IAAI,MAAM;AAChE,MAAI,SAAS,MAAO,UAAU,MAAM,SAAS,EAAI;AACjD,QAAM,IAAI;AAAA,IACR,+JAAgE,WAAW;AAAA,EAC7E;AACF;AAMO,SAAS,gBAAgB,eAAuB,KAA4D;AACjH,MAAI,SAAS;AACb,QAAM,aAAa,oBAAI,IAAY;AACnC,QAAM,UAAoB,CAAC;AAC3B,QAAM,UAAU,CAAC,QAAsB;AACrC,QAAI,WAAW,IAAI,GAAG,EAAG;AACzB,QAAI,UAAU,eAAe;AAC3B,UAAI,CAAC,QAAQ,SAAS,GAAG,EAAG,SAAQ,KAAK,GAAG;AAC5C;AAAA,IACF;AACA,eAAW,IAAI,GAAG;AAClB;AACA,SAAK,IAAI,GAAG,EAAE,QAAQ,MAAM;AAC1B,iBAAW,OAAO,GAAG;AACrB;AACA,YAAM,OAAO,QAAQ,MAAM;AAC3B,UAAI,KAAM,SAAQ,IAAI;AAAA,IACxB,CAAC;AAAA,EACH;AACA,SAAO;AACT;AAUA,eAAsB,eAAe,SAAsC;AACzE,QAAM,EAAE,KAAK,QAAQ,SAAS,SAAS,YAAY,OAAO,OAAO,IAAI;AAErE,MAAI,CAAC,WAAW,GAAG,EAAG,OAAM,IAAI,MAAM,gFAAoB,GAAG,EAAE;AAC/D,MAAI,QAAS,oBAAmB,OAAO;AACvC,MAAI,OAAQ,WAAU,QAAQ,EAAE,WAAW,KAAK,CAAC;AAEjD,QAAM,MAAM,SAAS,MAAM;AAAA,EAAC,IAAI,CAAC,QAAgB,QAAQ,OAAO,MAAM,MAAM,IAAI;AAChF,MAAI,6CAAyB,QAAQ,GAAG,CAAC,EAAE;AAC3C,MAAI,OAAQ,KAAI,gCAAsB,QAAQ,MAAM,CAAC,EAAE;AACvD,MAAI,QAAS,KAAI,gCAAsB,OAAO,EAAE;AAIhD,8BAA4B;AAG5B,QAAM,UAAU,oBAAI,IAA2C;AAE/D,QAAM,iBAAiB;AAGvB,QAAM,oBAAoB,OAAO,YAAqC;AACpE,QAAI,WAAW,SAAS,OAAO,EAAE;AACjC,UAAM,IAAI,QAAQ,OAAK,WAAW,GAAG,eAAe,CAAC;AACrD,QAAI,CAAC,WAAW,OAAO,EAAG,QAAO;AACjC,UAAM,WAAW,SAAS,OAAO,EAAE;AACnC,QAAI,aAAa,UAAU;AAEzB,YAAM,IAAI,QAAQ,OAAK,WAAW,GAAG,eAAe,CAAC;AACrD,UAAI,CAAC,WAAW,OAAO,EAAG,QAAO;AACjC,aAAO,SAAS,OAAO,EAAE;AAAA,IAC3B;AACA,WAAO;AAAA,EACT;AAEA,QAAM,cAAc,OAAO,aAAqB;AAC9C,UAAM,MAAM,QAAQ,QAAQ,EAAE,YAAY;AAC1C,QAAI,CAAC,qBAAqB,IAAI,GAAG,EAAG;AAEpC,UAAM,WAAW,SAAS,QAAQ;AAClC,QAAI;AACF,YAAM,UAAU,QAAQ,KAAK,QAAQ;AACrC,UAAI,CAAC,WAAW,OAAO,EAAG;AAE1B,UAAI;AACJ,UAAI;AAAE,kBAAU,aAAa,OAAO;AAAA,MAAE,QAAQ;AAAE;AAAA,MAAO;AACvD,YAAM,UAAU,aAAa,QAAQ,GAAG,CAAC;AACzC,UAAI,CAAC,QAAQ,WAAW,UAAU,GAAG,KAAK,YAAY,QAAS;AAE/D,YAAM,WAAW,MAAM,kBAAkB,OAAO;AAChD,UAAI,WAAW,iBAAiB,aAAa,EAAG;AAEhD,UAAI,uCAAwB,QAAQ,EAAE;AAEtC,YAAM,SAAS,aAAa,OAAO;AACnC,YAAM,cAAc,cAAc,MAAM;AAExC,YAAM,eAA6B,EAAE,UAAU,QAAQ;AACvD,UAAI,MAAO,cAAa,QAAQ;AAChC,YAAM,SAAS,MAAM,MAAM,aAAa,YAAY;AAEpD,UAAI,CAAC,OAAO,SAAS;AACnB,YAAI,gCAAsB,QAAQ,WAAM,OAAO,KAAK,EAAE;AACtD,cAAM,YAAY,SAAS,EAAE,MAAM,UAAU,QAAQ,aAAa,WAAW,GAAG,SAAS,OAAO,OAAO,OAAO,MAAM,CAAC;AACrH;AAAA,MACF;AAEA,YAAM,SAAS,WAAW,SAAS,KAAK,UAAU,QAAQ,MAAM,CAAC,IAAI,OAAO;AAE5E,UAAI,QAAQ;AACV,cAAM,SAAS,WAAW,SAAS,UAAU;AAC7C,cAAM,UAAU,QAAQ,QAAQ,SAAS,QAAQ,YAAY,MAAM,CAAC;AACpE,sBAAc,SAAS,QAAQ,OAAO;AACtC,YAAI,gCAAsB,QAAQ,WAAM,SAAS,OAAO,CAAC,EAAE;AAAA,MAC7D,OAAO;AACL,gBAAQ,OAAO,MAAM,SAAS,IAAI;AAAA,MACpC;AAEA,YAAM,YAAY,SAAS;AAAA,QACzB,MAAM;AAAA,QACN,QAAQ,OAAO;AAAA,QACf,SAAS;AAAA,QACT,UAAU,WAAW,aAAa,OAAO,UAAU,GAAG,GAAI,IAAI;AAAA,MAChE,CAAC;AAAA,IACH,SAAS,KAAK;AACZ,UAAI,gCAAsB,QAAQ,WAAM,eAAe,QAAQ,IAAI,UAAU,GAAG,EAAE;AAAA,IACpF;AAAA,EACF;AAEA,QAAM,UAAU;AAAA,IAAgB;AAAA,IAAgB,CAAC,aAC/C,YAAY,QAAQ,EAAE,MAAM,CAAC,QAAQ;AACnC,cAAQ,OAAO,MAAM,6CAAyB,QAAQ,WAAM,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC;AAAA,CAAI;AAAA,IAClH,CAAC;AAAA,EACH;AAGA,QAAM,UAAU,MAAM,KAAK,EAAE,WAAW,KAAK,GAAG,CAAC,OAAO,aAAa;AACnE,QAAI,CAAC,SAAU;AACf,UAAM,WAAW,SAAS,SAAS;AAGnC,UAAM,WAAW,QAAQ,IAAI,QAAQ;AACrC,QAAI,SAAU,cAAa,QAAQ;AACnC,YAAQ,IAAI,UAAU,WAAW,MAAM;AACrC,cAAQ,OAAO,QAAQ;AACvB,cAAQ,QAAQ;AAAA,IAClB,GAAG,WAAW,CAAC;AAAA,EACjB,CAAC;AAED,UAAQ,GAAG,SAAS,CAAC,QAAQ;AAC3B,YAAQ,OAAO,MAAM,6CAAyB,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC;AAAA,CAAI;AAAA,EACpG,CAAC;AAGD,SAAO,IAAI,QAAQ,MAAM;AAAA,EAAC,CAAC;AAC7B;AAGA,SAAS,mBAAmB,KAAmB;AAC7C,MAAI;AACJ,MAAI;AACF,aAAS,IAAI,IAAI,GAAG;AAAA,EACtB,QAAQ;AACN,UAAM,IAAI,MAAM,sDAAwB,GAAG,EAAE;AAAA,EAC/C;AACA,MAAI,OAAO,aAAa,WAAW,OAAO,aAAa,UAAU;AAC/D,UAAM,IAAI,MAAM,2EAAyB,OAAO,QAAQ,EAAE;AAAA,EAC5D;AACA,QAAM,WAAW,OAAO,SAAS,YAAY;AAC7C,MACE,aAAa,eACb,aAAa,WACb,SAAS,WAAW,MAAM,KAC1B,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,UAAU,KAC9B,6BAA6B,KAAK,QAAQ,KAC1C,aAAa,aACb,SAAS,WAAW,UAAU,KAC9B,SAAS,SAAS,QAAQ;AAAA,EAE1B,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,KAAK,KACzB,SAAS,WAAW,QAAQ,KAC5B,aAAa,WACb,aAAa;AAAA,EAEb,aAAa,8BACb,aAAa;AAAA,EAEb,iBAAiB,KAAK,QAAQ,KAC9B,YAAY,KAAK,QAAQ,KACzB,QAAQ,KAAK,QAAQ,GACrB;AACA,UAAM,IAAI,MAAM,uHAAkC,QAAQ,EAAE;AAAA,EAC9D;AACF;AAGO,SAAS,YAAY,IAAqB;AAC/C,MAAI,OAAO,GAAG,YAAY;AAE1B,QAAM,SAAS,gCAAgC,KAAK,IAAI;AACxD,MAAI,OAAQ,QAAO,OAAO,CAAC;AAC3B,MAAI,uBAAuB,KAAK,IAAI,GAAG;AACrC,UAAM,CAAC,GAAG,CAAC,IAAI,KAAK,MAAM,GAAG,EAAE,IAAI,MAAM;AACzC,WACE,MAAM,KAAK,MAAM,MAAM,MAAM,OAC5B,MAAM,OAAO,KAAK,MAAM,KAAK;AAAA,IAC7B,MAAM,OAAO,MAAM,OACnB,MAAM,OAAO,KAAK,MAAM,KAAK,MAC7B,MAAM,OAAO,MAAM;AAAA,EAExB;AAEA,SAAO,SAAS,QAAQ,SAAS,SAAS,SAAS,KAAK,IAAI,KAAK,YAAY,KAAK,IAAI;AACxF;AAEA,eAAe,YAAY,KAAyB,SAAiD;AACnG,MAAI,CAAC,IAAK;AACV,MAAI;AACF,uBAAmB,GAAG;AAEtB,UAAM,WAAW,IAAI,IAAI,GAAG,EAAE,SAAS,QAAQ,YAAY,EAAE;AAC7D,UAAM,QAAQ,MAAM,OAAO,UAAU,EAAE,KAAK,MAAM,UAAU,KAAK,CAAC;AAClE,UAAM,MAAM,MAAM,KAAK,CAAC,MAAM,YAAY,EAAE,OAAO,CAAC;AACpD,QAAI,IAAK,OAAM,IAAI,MAAM,0GAA+B,QAAQ,WAAM,IAAI,OAAO,EAAE;AACnF,UAAM,MAAM,KAAK;AAAA,MACf,QAAQ;AAAA,MACR,SAAS,EAAE,gBAAgB,mBAAmB;AAAA,MAC9C,MAAM,KAAK,UAAU,EAAE,GAAG,SAAS,YAAW,oBAAI,KAAK,GAAE,YAAY,EAAE,CAAC;AAAA,MACxE,UAAU;AAAA,IACZ,CAAC;AAAA,EACH,SAAS,KAAK;AACZ,YAAQ,OAAO,MAAM,qDAAiC,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC;AAAA,CAAI;AAAA,EAC5G;AACF;","names":[]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "kordoc",
-  "version": "4.1.0",
+  "version": "4.2.0",
   "description": "Parse Korean documents (HWP3-5, HWPX, HWPML, PDF, XLS(X), DOCX) to Markdown — CLI + MCP server with lossless patch/roundtrip, form-filler, document diff, print & layout-preserving SVG renderer",
   "type": "module",
   "exports": {
@@ -43,6 +43,7 @@
     "docx",
     "markdown",
     "converter",
+    "ocr",
     "mcp",
     "model-context-protocol",
     "print",