kordoc 4.18.12 → 4.18.13

package/dist/{image-ocr-YLTHT7IE.cjs → image-ocr-AZ25IOC7.cjs}

@@ -5,17 +5,17 @@
 
 
 
-var _chunkGGX5AA2Lcjs = require('./chunk-GGX5AA2L.cjs');
+var _chunkGGMIPTM5cjs = require('./chunk-GGMIPTM5.cjs');
 require('./chunk-TK3FDAEE.cjs');
 
 
-var _chunkOB3GPKKZcjs = require('./chunk-OB3GPKKZ.cjs');
-require('./chunk-5WLN2JNN.cjs');
-require('./chunk-A7VXXIRT.cjs');
+var _chunk3FGHQWJPcjs = require('./chunk-3FGHQWJP.cjs');
+require('./chunk-6FW3AZWN.cjs');
+require('./chunk-6CQZMVXW.cjs');
 
 
 
-var _chunk2CMRR7SHcjs = require('./chunk-2CMRR7SH.cjs');
+var _chunkK6WSMTVUcjs = require('./chunk-K6WSMTVU.cjs');
 require('./chunk-GS7T56RP.cjs');
 
 // src/ocr/image-ocr.ts
@@ -43,14 +43,14 @@ async function parseImageDocument(buffer, options) {
   const { data: decoded, width, height, density, shrink } = await decodeToRgba(buffer);
   if (shrink < 1) warnings.push({ page: 1, code: "PARTIAL_PARSE", message: "OCR \uB798\uC2A4\uD130 \uD53D\uC140 \uC0C1\uD55C\uC73C\uB85C \uC774\uBBF8\uC9C0 \uD574\uC0C1\uB3C4\uB97C \uCD95\uC18C\uD588\uC2B5\uB2C8\uB2E4 (\uC791\uC740 \uAE00\uC790 \uC778\uC2DD \uACB0\uC190 \uAC00\uB2A5)" });
   const scale = imageScale(width / shrink, height / shrink, density) * shrink;
-  const { rgba: data } = _chunkGGX5AA2Lcjs.deskewPage.call(void 0, decoded, width, height);
-  await _chunkOB3GPKKZcjs.ensureOcrModels.call(void 0, (p) => {
+  const { rgba: data } = _chunkGGMIPTM5cjs.deskewPage.call(void 0, decoded, width, height);
+  await _chunk3FGHQWJPcjs.ensureOcrModels.call(void 0, (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 = await _chunkGGX5AA2Lcjs.getOcrEngine.call(void 0, );
+  const engine = await _chunkGGMIPTM5cjs.getOcrEngine.call(void 0, );
   const stats = { droppedLowConf: 0 };
   const items = await engine.recognizePage(data, width, height, stats);
   if (stats.droppedLowConf > 0) {
@@ -69,9 +69,9 @@ async function parseImageDocument(buffer, options) {
   }
   const pdfW = width / scale;
   const pdfH = height / scale;
-  const ruling = _chunkGGX5AA2Lcjs.detectRulingLines.call(void 0, data, width, height, scale);
-  const extraLines = _chunkGGX5AA2Lcjs.rulingToPdfLines.call(void 0, ruling, scale, pdfH);
-  return { blocks: _chunkGGX5AA2Lcjs.ocrItemsToBlocks.call(void 0, items, 1, pdfW, pdfH, scale, extraLines, _optionalChain([options, 'optionalAccess', _2 => _2.tables]) !== false), warnings };
+  const ruling = _chunkGGMIPTM5cjs.detectRulingLines.call(void 0, data, width, height, scale);
+  const extraLines = _chunkGGMIPTM5cjs.rulingToPdfLines.call(void 0, ruling, scale, pdfH);
+  return { blocks: _chunkGGMIPTM5cjs.ocrItemsToBlocks.call(void 0, items, 1, pdfW, pdfH, scale, extraLines, _optionalChain([options, 'optionalAccess', _2 => _2.tables]) !== false), warnings };
 }
 async function decodeToRgba(buffer) {
   let sharp;
@@ -79,14 +79,14 @@ async function decodeToRgba(buffer) {
     const mod = await Promise.resolve().then(() => _interopRequireWildcard(require("sharp")));
     sharp = typeof mod === "function" ? mod : _nullishCoalesce(mod.default, () => ( mod));
   } catch (e) {
-    throw new (0, _chunk2CMRR7SHcjs.KordocError)(
-      `\uC774\uBBF8\uC9C0 \uD30C\uC2F1\uC5D0\uB294 optional dependency 'sharp' \uAC00 \uD544\uC694\uD569\uB2C8\uB2E4. \`npm install sharp\` \uD6C4 \uB2E4\uC2DC \uC2E4\uD589\uD558\uC138\uC694.${_chunk2CMRR7SHcjs.OPTIONAL_DEP_INSTALL_HINT} \uC6D0\uC778: ${e.message}`
+    throw new (0, _chunkK6WSMTVUcjs.KordocError)(
+      `\uC774\uBBF8\uC9C0 \uD30C\uC2F1\uC5D0\uB294 optional dependency 'sharp' \uAC00 \uD544\uC694\uD569\uB2C8\uB2E4. \`npm install sharp\` \uD6C4 \uB2E4\uC2DC \uC2E4\uD589\uD558\uC138\uC694.${_chunkK6WSMTVUcjs.OPTIONAL_DEP_INSTALL_HINT} \uC6D0\uC778: ${e.message}`
     );
   }
   const input = Buffer.from(buffer);
   const meta = await sharp(input).metadata().catch(() => ({}));
   const [w0, h0] = (_nullishCoalesce(meta.orientation, () => ( 1))) >= 5 ? [_nullishCoalesce(meta.height, () => ( 0)), _nullishCoalesce(meta.width, () => ( 0))] : [_nullishCoalesce(meta.width, () => ( 0)), _nullishCoalesce(meta.height, () => ( 0))];
-  const shrink = w0 * h0 > _chunkGGX5AA2Lcjs.MAX_OCR_PIXELS ? Math.sqrt(_chunkGGX5AA2Lcjs.MAX_OCR_PIXELS / (w0 * h0)) : 1;
+  const shrink = w0 * h0 > _chunkGGMIPTM5cjs.MAX_OCR_PIXELS ? Math.sqrt(_chunkGGMIPTM5cjs.MAX_OCR_PIXELS / (w0 * h0)) : 1;
   let chain = sharp(input).rotate();
   if (shrink < 1) chain = chain.resize(Math.max(1, Math.floor(w0 * shrink)), Math.max(1, Math.floor(h0 * shrink)), { fit: "fill" });
   const { data, info } = await chain.ensureAlpha().raw().toBuffer({ resolveWithObject: true });
@@ -109,4 +109,4 @@ function detectImageMime(buffer) {
 
 
 exports.decodeToRgba = decodeToRgba; exports.imageScale = imageScale; exports.parseImageDocument = parseImageDocument;
-//# sourceMappingURL=image-ocr-YLTHT7IE.cjs.map
+//# sourceMappingURL=image-ocr-AZ25IOC7.cjs.map

package/dist/{image-ocr-YLTHT7IE.cjs.map → image-ocr-AZ25IOC7.cjs.map}

@@ -1 +1 @@
-{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/image-ocr-YLTHT7IE.cjs","../src/ocr/image-ocr.ts"],"names":[],"mappings":"AAAA;AACE;AACA;AACA;AACA;AACA;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACA;ACCA,IAAM,YAAA,EAAc,CAAA;AASb,SAAS,UAAA,CAAW,KAAA,EAAe,MAAA,EAAgB,UAAA,EAA6B;AACrF,EAAA,GAAA,CAAI,WAAA,GAAc,WAAA,GAAc,IAAA,GAAO,WAAA,GAAc,IAAA,EAAM,OAAO,WAAA,EAAa,EAAA;AAC/E,EAAA,MAAM,KAAA,EAAO,IAAA,CAAK,GAAA,CAAI,KAAA,EAAO,MAAM,CAAA,EAAG,MAAA,EAAQ,IAAA,CAAK,GAAA,CAAI,KAAA,EAAO,MAAM,CAAA;AACpE,EAAA,GAAA,CAAI,MAAA,EAAQ,CAAA,EAAG;AACb,IAAA,MAAM,EAAA,EAAI,KAAA,EAAO,KAAA;AACjB,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,EAAA,EAAI,IAAA,CAAK,MAAA,EAAQ,CAAC,EAAA,GAAK,IAAA,EAAM,OAAO,KAAA,EAAO,GAAA;AACxD,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,EAAA,EAAA,CAAK,GAAA,EAAK,GAAA,EAAA,EAAO,CAAC,EAAA,GAAK,IAAA,EAAM,OAAO,KAAA,EAAO,GAAA;AAAA,EAC1D;AACA,EAAA,OAAO,WAAA;AACT;AAQA,MAAA,SAAsB,kBAAA,CACpB,MAAA,EACA,OAAA,EACyB;AACzB,EAAA,MAAM,SAAA,EAA2B,CAAC,CAAA;AAGlC,EAAA,GAAA,CAAI,uBAAO,OAAA,2BAAS,MAAA,IAAQ,UAAA,EAAY;AACtC,IAAA,MAAM,KAAA,EAAO,MAAM,OAAA,CAAQ,GAAA,CAAI,IAAI,UAAA,CAAW,MAAM,CAAA,EAAG,CAAA,EAAG,eAAA,CAAgB,MAAM,CAAC,CAAA;AACjF,IAAA,GAAA,CAAI,CAAC,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG;AAChB,MAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,+BAAA,EAAa,IAAA,EAAM,aAAa,CAAC,CAAA;AACnE,MAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,IAChC;AACA,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,EAAE,IAAA,EAAM,WAAA,EAAa,IAAA,EAAM,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG,UAAA,EAAY,EAAE,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EACvF;AAEA,EAAA,MAAM,EAAE,IAAA,EAAM,OAAA,EAAS,KAAA,EAAO,MAAA,EAAQ,OAAA,EAAS,OAAO,EAAA,EAAI,MAAM,YAAA,CAAa,MAAM,CAAA;AACnF,EAAA,GAAA,CAAI,OAAA,EAAS,CAAA,EAAG,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,IAAA,EAAM,eAAA,EAAiB,OAAA,EAAS,mNAAmD,CAAC,CAAA;AAE7H,EAAA,MAAM,MAAA,EAAQ,UAAA,CAAW,MAAA,EAAQ,MAAA,EAAQ,OAAA,EAAS,MAAA,EAAQ,OAAO,EAAA,EAAI,MAAA;AAErE,EAAA,MAAM,EAAE,IAAA,EAAM,KAAK,EAAA,EAAI,0CAAA,OAAW,EAAS,KAAA,EAAO,MAAM,CAAA;AAExD,EAAA,MAAM,+CAAA,CAAgB,CAAA,EAAA,GAAK;AACzB,IAAA,GAAA,CAAI,CAAA,CAAE,MAAA,IAAU,WAAA,GAAc,CAAA,CAAE,WAAA,IAAe,CAAA,EAAG;AAChD,MAAA,OAAA,CAAQ,MAAA,CAAO,KAAA,CAAM,CAAA,aAAA,EAAgB,CAAA,CAAE,IAAA,CAAK,IAAI,CAAA,mCAAA,EAAa,CAAA,CAAE,IAAA,CAAK,MAAM,CAAA;AAAA,CAAU,CAAA;AAAA,IACtF;AAAA,EACF,CAAC,CAAA;AACD,EAAA,MAAM,OAAA,EAAS,MAAM,4CAAA,CAAa;AAClC,EAAA,MAAM,MAAA,EAAsB,EAAE,cAAA,EAAgB,EAAE,CAAA;AAChD,EAAA,MAAM,MAAA,EAAQ,MAAM,MAAA,CAAO,aAAA,CAAc,IAAA,EAAM,KAAA,EAAO,MAAA,EAAQ,KAAK,CAAA;AACnE,EAAA,GAAA,CAAI,KAAA,CAAM,eAAA,EAAiB,CAAA,EAAG;AAC5B,IAAA,QAAA,CAAS,IAAA,CAAK;AAAA,MACZ,IAAA,EAAM,CAAA;AAAA,MACN,OAAA,EAAS,CAAA,oCAAA,EAAc,KAAA,CAAM,cAAc,CAAA,4DAAA,CAAA;AAAA,MAC3C,IAAA,EAAM;AAAA,IACR,CAAC,CAAA;AAAA,EACH;AACA,EAAA,GAAA,CAAI,KAAA,CAAM,cAAA,EAAgB;AACxB,IAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,CAAA,8DAAA,EAAoB,KAAA,CAAM,cAAc,CAAA,qFAAA,CAAA,EAAwB,IAAA,EAAM,gBAAgB,CAAC,CAAA;AAAA,EAC3H;AACA,EAAA,GAAA,CAAI,KAAA,CAAM,OAAA,IAAW,CAAA,EAAG;AACtB,IAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,iHAAA,EAAyB,IAAA,EAAM,aAAa,CAAC,CAAA;AAC/E,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EAChC;AAEA,EAAA,MAAM,KAAA,EAAO,MAAA,EAAQ,KAAA;AACrB,EAAA,MAAM,KAAA,EAAO,OAAA,EAAS,KAAA;AACtB,EAAA,MAAM,OAAA,EAAS,iDAAA,IAAkB,EAAM,KAAA,EAAO,MAAA,EAAQ,KAAK,CAAA;AAC3D,EAAA,MAAM,WAAA,EAAa,gDAAA,MAAiB,EAAQ,KAAA,EAAO,IAAI,CAAA;AACvD,EAAA,OAAO,EAAE,MAAA,EAAQ,gDAAA,KAAiB,EAAO,CAAA,EAAG,IAAA,EAAM,IAAA,EAAM,KAAA,EAAO,UAAA,kBAAY,OAAA,6BAAS,SAAA,IAAW,KAAK,CAAA,EAAG,SAAS,CAAA;AAClH;AAQA,MAAA,SAAsB,YAAA,CACpB,MAAA,EACgG;AAQhG,EAAA,IAAI,KAAA;AACJ,EAAA,IAAI;AACF,IAAA,MAAM,IAAA,EAAO,MAAM,4DAAA,CAAO,OAAO,GAAA;AACjC,IAAA,MAAA,EAAQ,OAAO,IAAA,IAAQ,WAAA,EAAa,IAAA,mBAAO,GAAA,CAAI,OAAA,UAAY,KAAA;AAAA,EAC7D,EAAA,MAAA,CAAS,CAAA,EAAG;AAEV,IAAA,MAAM,IAAI,kCAAA;AAAA,MACR,CAAA,wLAAA,EACsC,2CAAyB,CAAA,eAAA,EAAS,CAAA,CAAY,OAAO,CAAA;AAAA,IAAA;AAC7F,EAAA;AAEF,EAAA;AACA,EAAA;AAEA,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AAAO,IAAA;AAC6D,IAAA;AAAe,IAAA;AAAoB,IAAA;AAAsB,IAAA;AAC5F,EAAA;AAEnC;AAGA;AACE,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACF;ADzCA;AACA;AACA;AACA;AACA","file":"/Users/mong-e/workspace/kordoc/dist/image-ocr-YLTHT7IE.cjs","sourcesContent":[null,"/**\n * 이미지 직접 입력 OCR — PNG/JPEG/WebP 를 PDF 래핑 없이 바로 파싱한다.\n *\n * 이미지는 텍스트층이 없으므로 OCR 이 항상 필요하다 — `--ocr` 플래그 없이도\n * 내장 엔진(PP-OCRv5 korean)을 기본 적용하고, OcrProvider 가 주어지면 위임한다.\n * 좌표는 이미지 해상도(imageScale — 메타데이터 DPI, 없으면 A4·Letter 판형 추정, 그것도\n * 아니면 216dpi)로 PDF pt 공간에 환산해 기존 블록 파이프라인\n * (괘선 감지 → 선 기반 표 → 클러스터 → xy-cut)을 그대로 태운다.\n *\n * 디코딩은 optional dependency `sharp` — 미설치면 MISSING_DEPENDENCY 안내.\n */\n\nimport type { IRBlock, ParseOptions, ParseWarning } from \"../types.js\"\nimport { KordocError, OPTIONAL_DEP_INSTALL_HINT } from \"../utils.js\"\nimport { getOcrEngine, type OcrPageStats } from \"./engine.js\"\nimport { ensureOcrModels } from \"./models.js\"\nimport { detectRulingLines, rulingToPdfLines } from \"./ruling-lines.js\"\nimport { MAX_OCR_PIXELS, ocrItemsToBlocks } from \"./pdf-ocr.js\"\nimport { deskewPage } from \"./deskew.js\"\n\n/** 좌표 환산 기본 스케일 (px/pt) — PDF OCR 렌더(216dpi)와 동일 기준 */\nconst IMAGE_SCALE = 3\n\n/**\n * 이미지 px/pt 스케일 추정. 블록 파이프라인의 문턱(줄 묶음 3pt·괘선 최소 길이 20pt·표 간격)은\n * pt 기준이라, 150dpi 스캔을 216dpi 로 가정하면 페이지가 0.69배로 쪼그라든 좌표가 된다.\n *  1) 메타데이터 DPI(JFIF·pHYs·EXIF) 100~1200 — 72/96 은 기본값·화면값이라 불신\n *  2) 판형: 긴 변/짧은 변이 √2(A판)·11/8.5(Letter) ±2% 면 전면 스캔 — A4 842pt·Letter 792pt 긴 변\n *  3) 그 밖: 216dpi 가정 (종전 동작)\n */\nexport function imageScale(width: number, height: number, densityDpi?: number): number {\n  if (densityDpi && densityDpi >= 100 && densityDpi <= 1200) return densityDpi / 72\n  const long = Math.max(width, height), short = Math.min(width, height)\n  if (short > 0) {\n    const r = long / short\n    if (Math.abs(r / Math.SQRT2 - 1) <= 0.02) return long / 842\n    if (Math.abs(r / (11 / 8.5) - 1) <= 0.02) return long / 792\n  }\n  return IMAGE_SCALE\n}\n\nexport interface ImageOcrResult {\n  blocks: IRBlock[]\n  warnings: ParseWarning[]\n}\n\n/** 이미지 버퍼 → OCR → IRBlock[] (표 괘선 감지 포함) */\nexport async function parseImageDocument(\n  buffer: ArrayBuffer,\n  options?: ParseOptions,\n): Promise<ImageOcrResult> {\n  const warnings: ParseWarning[] = []\n\n  // 사용자 OcrProvider — 원본 바이트 그대로 위임 (종전 PDF 경로와 같은 계약)\n  if (typeof options?.ocr === \"function\") {\n    const text = await options.ocr(new Uint8Array(buffer), 1, detectImageMime(buffer))\n    if (!text.trim()) {\n      warnings.push({ page: 1, message: \"OCR 결과 없음\", code: \"OCR_FAILED\" })\n      return { blocks: [], warnings }\n    }\n    return { blocks: [{ type: \"paragraph\", text: text.trim(), pageNumber: 1 }], warnings }\n  }\n\n  const { data: decoded, width, height, density, shrink } = await decodeToRgba(buffer)\n  if (shrink < 1) warnings.push({ page: 1, code: \"PARTIAL_PARSE\", message: \"OCR 래스터 픽셀 상한으로 이미지 해상도를 축소했습니다 (작은 글자 인식 결손 가능)\" })\n  // 픽셀 상한으로 줄인 이미지는 같은 지면을 낮은 해상도로 본 것 — px/pt 도 같은 비율로\n  const scale = imageScale(width / shrink, height / shrink, density) * shrink\n  // 스캔·사진 기울기 보정 — 인식과 괘선 감지가 같은(바로 선) 래스터를 본다\n  const { rgba: data } = deskewPage(decoded, width, height)\n\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  const engine = await getOcrEngine()\n  const stats: OcrPageStats = { droppedLowConf: 0 }\n  const items = await engine.recognizePage(data, width, height, stats)\n  if (stats.droppedLowConf > 0) {\n    warnings.push({\n      page: 1,\n      message: `저신뢰 OCR 라인 ${stats.droppedLowConf}개 폐기 (인식 결손 가능)`,\n      code: \"OCR_LOW_CONF\",\n    })\n  }\n  if (stats.truncatedBoxes) {\n    warnings.push({ page: 1, message: `OCR 검출 상자가 너무 많아 ${stats.truncatedBoxes}개는 인식하지 않음 (일부 글 결손)`, code: \"PARTIAL_PARSE\" })\n  }\n  if (items.length === 0) {\n    warnings.push({ page: 1, message: \"이미지에서 텍스트를 인식하지 못했습니다\", code: \"OCR_FAILED\" })\n    return { blocks: [], warnings }\n  }\n\n  const pdfW = width / scale\n  const pdfH = height / scale\n  const ruling = detectRulingLines(data, width, height, scale)\n  const extraLines = rulingToPdfLines(ruling, scale, pdfH)\n  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, scale, extraLines, options?.tables !== false), warnings }\n}\n\n/**\n * sharp 로 RGBA 디코딩 — 미설치는 MISSING_DEPENDENCY 로 분류되도록 안내 메시지 throw.\n * EXIF 방향대로 세운다(`rotate()`): 폰으로 세로 촬영한 JPEG 은 픽셀을 눕혀 저장하고 Orientation 6 으로 표시 방향만 적는다 —\n * 무시하면 누운 글을 읽어 한글 131자 쪽이 쓰레기 57자가 됐다. 픽셀이 MAX_OCR_PIXELS 를 넘으면 그 안으로 줄인다(shrink = 줄인 배율).\n * (테스트용 export)\n */\nexport async function decodeToRgba(\n  buffer: ArrayBuffer,\n): Promise<{ data: Uint8Array; width: number; height: number; density?: number; shrink: number }> {\n  type Chain = {\n    rotate(): Chain\n    resize(w: number, h: number, opts: { fit: \"fill\" }): Chain\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: number } }> } }\n    metadata(): Promise<{ density?: number; width?: number; height?: number; orientation?: number }>\n  }\n  type SharpFactory = (input: Buffer) => Chain\n  let sharp: SharpFactory\n  try {\n    const mod = (await import(\"sharp\")) as unknown as SharpFactory | { default?: SharpFactory }\n    sharp = typeof mod === \"function\" ? mod : (mod.default ?? (mod as unknown as SharpFactory))\n  } catch (e) {\n    // KordocError + \"optional dependency\" 문구 → sanitizeError 메시지 보존 + MISSING_DEPENDENCY 분류\n    throw new KordocError(\n      \"이미지 파싱에는 optional dependency 'sharp' 가 필요합니다. \" +\n        `\\`npm install sharp\\` 후 다시 실행하세요.${OPTIONAL_DEP_INSTALL_HINT} 원인: ${(e as Error).message}`,\n    )\n  }\n  const input = Buffer.from(buffer)\n  const meta: { density?: number; width?: number; height?: number; orientation?: number } = await sharp(input).metadata().catch(() => ({}))\n  // 5~8 은 90° 돌린 방향 — 세운 뒤 가로·세로가 바뀐다\n  const [w0, h0] = (meta.orientation ?? 1) >= 5 ? [meta.height ?? 0, meta.width ?? 0] : [meta.width ?? 0, meta.height ?? 0]\n  const shrink = w0 * h0 > MAX_OCR_PIXELS ? Math.sqrt(MAX_OCR_PIXELS / (w0 * h0)) : 1\n  let chain = sharp(input).rotate()\n  if (shrink < 1) chain = chain.resize(Math.max(1, Math.floor(w0 * shrink)), Math.max(1, Math.floor(h0 * shrink)), { fit: \"fill\" })\n  const { data, info } = await chain.ensureAlpha().raw().toBuffer({ resolveWithObject: true })\n  return {\n    data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height, density: meta.density,\n    shrink: w0 ? info.width / w0 : 1,\n  }\n}\n\n/** OcrProvider 계약용 mime 판별 */\nfunction detectImageMime(buffer: ArrayBuffer): \"image/png\" | \"image/jpeg\" | \"image/webp\" {\n  const b = new Uint8Array(buffer, 0, Math.min(12, buffer.byteLength))\n  if (b[0] === 0xff && b[1] === 0xd8) return \"image/jpeg\"\n  if (b[0] === 0x52 && b[1] === 0x49) return \"image/webp\"\n  return \"image/png\"\n}\n"]}
+{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/image-ocr-AZ25IOC7.cjs","../src/ocr/image-ocr.ts"],"names":[],"mappings":"AAAA;AACE;AACA;AACA;AACA;AACA;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACA;ACCA,IAAM,YAAA,EAAc,CAAA;AASb,SAAS,UAAA,CAAW,KAAA,EAAe,MAAA,EAAgB,UAAA,EAA6B;AACrF,EAAA,GAAA,CAAI,WAAA,GAAc,WAAA,GAAc,IAAA,GAAO,WAAA,GAAc,IAAA,EAAM,OAAO,WAAA,EAAa,EAAA;AAC/E,EAAA,MAAM,KAAA,EAAO,IAAA,CAAK,GAAA,CAAI,KAAA,EAAO,MAAM,CAAA,EAAG,MAAA,EAAQ,IAAA,CAAK,GAAA,CAAI,KAAA,EAAO,MAAM,CAAA;AACpE,EAAA,GAAA,CAAI,MAAA,EAAQ,CAAA,EAAG;AACb,IAAA,MAAM,EAAA,EAAI,KAAA,EAAO,KAAA;AACjB,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,EAAA,EAAI,IAAA,CAAK,MAAA,EAAQ,CAAC,EAAA,GAAK,IAAA,EAAM,OAAO,KAAA,EAAO,GAAA;AACxD,IAAA,GAAA,CAAI,IAAA,CAAK,GAAA,CAAI,EAAA,EAAA,CAAK,GAAA,EAAK,GAAA,EAAA,EAAO,CAAC,EAAA,GAAK,IAAA,EAAM,OAAO,KAAA,EAAO,GAAA;AAAA,EAC1D;AACA,EAAA,OAAO,WAAA;AACT;AAQA,MAAA,SAAsB,kBAAA,CACpB,MAAA,EACA,OAAA,EACyB;AACzB,EAAA,MAAM,SAAA,EAA2B,CAAC,CAAA;AAGlC,EAAA,GAAA,CAAI,uBAAO,OAAA,2BAAS,MAAA,IAAQ,UAAA,EAAY;AACtC,IAAA,MAAM,KAAA,EAAO,MAAM,OAAA,CAAQ,GAAA,CAAI,IAAI,UAAA,CAAW,MAAM,CAAA,EAAG,CAAA,EAAG,eAAA,CAAgB,MAAM,CAAC,CAAA;AACjF,IAAA,GAAA,CAAI,CAAC,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG;AAChB,MAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,+BAAA,EAAa,IAAA,EAAM,aAAa,CAAC,CAAA;AACnE,MAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,IAChC;AACA,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,EAAE,IAAA,EAAM,WAAA,EAAa,IAAA,EAAM,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG,UAAA,EAAY,EAAE,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EACvF;AAEA,EAAA,MAAM,EAAE,IAAA,EAAM,OAAA,EAAS,KAAA,EAAO,MAAA,EAAQ,OAAA,EAAS,OAAO,EAAA,EAAI,MAAM,YAAA,CAAa,MAAM,CAAA;AACnF,EAAA,GAAA,CAAI,OAAA,EAAS,CAAA,EAAG,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,IAAA,EAAM,eAAA,EAAiB,OAAA,EAAS,mNAAmD,CAAC,CAAA;AAE7H,EAAA,MAAM,MAAA,EAAQ,UAAA,CAAW,MAAA,EAAQ,MAAA,EAAQ,OAAA,EAAS,MAAA,EAAQ,OAAO,EAAA,EAAI,MAAA;AAErE,EAAA,MAAM,EAAE,IAAA,EAAM,KAAK,EAAA,EAAI,0CAAA,OAAW,EAAS,KAAA,EAAO,MAAM,CAAA;AAExD,EAAA,MAAM,+CAAA,CAAgB,CAAA,EAAA,GAAK;AACzB,IAAA,GAAA,CAAI,CAAA,CAAE,MAAA,IAAU,WAAA,GAAc,CAAA,CAAE,WAAA,IAAe,CAAA,EAAG;AAChD,MAAA,OAAA,CAAQ,MAAA,CAAO,KAAA,CAAM,CAAA,aAAA,EAAgB,CAAA,CAAE,IAAA,CAAK,IAAI,CAAA,mCAAA,EAAa,CAAA,CAAE,IAAA,CAAK,MAAM,CAAA;AAAA,CAAU,CAAA;AAAA,IACtF;AAAA,EACF,CAAC,CAAA;AACD,EAAA,MAAM,OAAA,EAAS,MAAM,4CAAA,CAAa;AAClC,EAAA,MAAM,MAAA,EAAsB,EAAE,cAAA,EAAgB,EAAE,CAAA;AAChD,EAAA,MAAM,MAAA,EAAQ,MAAM,MAAA,CAAO,aAAA,CAAc,IAAA,EAAM,KAAA,EAAO,MAAA,EAAQ,KAAK,CAAA;AACnE,EAAA,GAAA,CAAI,KAAA,CAAM,eAAA,EAAiB,CAAA,EAAG;AAC5B,IAAA,QAAA,CAAS,IAAA,CAAK;AAAA,MACZ,IAAA,EAAM,CAAA;AAAA,MACN,OAAA,EAAS,CAAA,oCAAA,EAAc,KAAA,CAAM,cAAc,CAAA,4DAAA,CAAA;AAAA,MAC3C,IAAA,EAAM;AAAA,IACR,CAAC,CAAA;AAAA,EACH;AACA,EAAA,GAAA,CAAI,KAAA,CAAM,cAAA,EAAgB;AACxB,IAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,CAAA,8DAAA,EAAoB,KAAA,CAAM,cAAc,CAAA,qFAAA,CAAA,EAAwB,IAAA,EAAM,gBAAgB,CAAC,CAAA;AAAA,EAC3H;AACA,EAAA,GAAA,CAAI,KAAA,CAAM,OAAA,IAAW,CAAA,EAAG;AACtB,IAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,iHAAA,EAAyB,IAAA,EAAM,aAAa,CAAC,CAAA;AAC/E,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EAChC;AAEA,EAAA,MAAM,KAAA,EAAO,MAAA,EAAQ,KAAA;AACrB,EAAA,MAAM,KAAA,EAAO,OAAA,EAAS,KAAA;AACtB,EAAA,MAAM,OAAA,EAAS,iDAAA,IAAkB,EAAM,KAAA,EAAO,MAAA,EAAQ,KAAK,CAAA;AAC3D,EAAA,MAAM,WAAA,EAAa,gDAAA,MAAiB,EAAQ,KAAA,EAAO,IAAI,CAAA;AACvD,EAAA,OAAO,EAAE,MAAA,EAAQ,gDAAA,KAAiB,EAAO,CAAA,EAAG,IAAA,EAAM,IAAA,EAAM,KAAA,EAAO,UAAA,kBAAY,OAAA,6BAAS,SAAA,IAAW,KAAK,CAAA,EAAG,SAAS,CAAA;AAClH;AAQA,MAAA,SAAsB,YAAA,CACpB,MAAA,EACgG;AAQhG,EAAA,IAAI,KAAA;AACJ,EAAA,IAAI;AACF,IAAA,MAAM,IAAA,EAAO,MAAM,4DAAA,CAAO,OAAO,GAAA;AACjC,IAAA,MAAA,EAAQ,OAAO,IAAA,IAAQ,WAAA,EAAa,IAAA,mBAAO,GAAA,CAAI,OAAA,UAAY,KAAA;AAAA,EAC7D,EAAA,MAAA,CAAS,CAAA,EAAG;AAEV,IAAA,MAAM,IAAI,kCAAA;AAAA,MACR,CAAA,wLAAA,EACsC,2CAAyB,CAAA,eAAA,EAAS,CAAA,CAAY,OAAO,CAAA;AAAA,IAAA;AAC7F,EAAA;AAEF,EAAA;AACA,EAAA;AAEA,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AAAO,IAAA;AAC6D,IAAA;AAAe,IAAA;AAAoB,IAAA;AAAsB,IAAA;AAC5F,EAAA;AAEnC;AAGA;AACE,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACF;ADzCA;AACA;AACA;AACA;AACA","file":"/Users/mong-e/workspace/kordoc/dist/image-ocr-AZ25IOC7.cjs","sourcesContent":[null,"/**\n * 이미지 직접 입력 OCR — PNG/JPEG/WebP 를 PDF 래핑 없이 바로 파싱한다.\n *\n * 이미지는 텍스트층이 없으므로 OCR 이 항상 필요하다 — `--ocr` 플래그 없이도\n * 내장 엔진(PP-OCRv5 korean)을 기본 적용하고, OcrProvider 가 주어지면 위임한다.\n * 좌표는 이미지 해상도(imageScale — 메타데이터 DPI, 없으면 A4·Letter 판형 추정, 그것도\n * 아니면 216dpi)로 PDF pt 공간에 환산해 기존 블록 파이프라인\n * (괘선 감지 → 선 기반 표 → 클러스터 → xy-cut)을 그대로 태운다.\n *\n * 디코딩은 optional dependency `sharp` — 미설치면 MISSING_DEPENDENCY 안내.\n */\n\nimport type { IRBlock, ParseOptions, ParseWarning } from \"../types.js\"\nimport { KordocError, OPTIONAL_DEP_INSTALL_HINT } from \"../utils.js\"\nimport { getOcrEngine, type OcrPageStats } from \"./engine.js\"\nimport { ensureOcrModels } from \"./models.js\"\nimport { detectRulingLines, rulingToPdfLines } from \"./ruling-lines.js\"\nimport { MAX_OCR_PIXELS, ocrItemsToBlocks } from \"./pdf-ocr.js\"\nimport { deskewPage } from \"./deskew.js\"\n\n/** 좌표 환산 기본 스케일 (px/pt) — PDF OCR 렌더(216dpi)와 동일 기준 */\nconst IMAGE_SCALE = 3\n\n/**\n * 이미지 px/pt 스케일 추정. 블록 파이프라인의 문턱(줄 묶음 3pt·괘선 최소 길이 20pt·표 간격)은\n * pt 기준이라, 150dpi 스캔을 216dpi 로 가정하면 페이지가 0.69배로 쪼그라든 좌표가 된다.\n *  1) 메타데이터 DPI(JFIF·pHYs·EXIF) 100~1200 — 72/96 은 기본값·화면값이라 불신\n *  2) 판형: 긴 변/짧은 변이 √2(A판)·11/8.5(Letter) ±2% 면 전면 스캔 — A4 842pt·Letter 792pt 긴 변\n *  3) 그 밖: 216dpi 가정 (종전 동작)\n */\nexport function imageScale(width: number, height: number, densityDpi?: number): number {\n  if (densityDpi && densityDpi >= 100 && densityDpi <= 1200) return densityDpi / 72\n  const long = Math.max(width, height), short = Math.min(width, height)\n  if (short > 0) {\n    const r = long / short\n    if (Math.abs(r / Math.SQRT2 - 1) <= 0.02) return long / 842\n    if (Math.abs(r / (11 / 8.5) - 1) <= 0.02) return long / 792\n  }\n  return IMAGE_SCALE\n}\n\nexport interface ImageOcrResult {\n  blocks: IRBlock[]\n  warnings: ParseWarning[]\n}\n\n/** 이미지 버퍼 → OCR → IRBlock[] (표 괘선 감지 포함) */\nexport async function parseImageDocument(\n  buffer: ArrayBuffer,\n  options?: ParseOptions,\n): Promise<ImageOcrResult> {\n  const warnings: ParseWarning[] = []\n\n  // 사용자 OcrProvider — 원본 바이트 그대로 위임 (종전 PDF 경로와 같은 계약)\n  if (typeof options?.ocr === \"function\") {\n    const text = await options.ocr(new Uint8Array(buffer), 1, detectImageMime(buffer))\n    if (!text.trim()) {\n      warnings.push({ page: 1, message: \"OCR 결과 없음\", code: \"OCR_FAILED\" })\n      return { blocks: [], warnings }\n    }\n    return { blocks: [{ type: \"paragraph\", text: text.trim(), pageNumber: 1 }], warnings }\n  }\n\n  const { data: decoded, width, height, density, shrink } = await decodeToRgba(buffer)\n  if (shrink < 1) warnings.push({ page: 1, code: \"PARTIAL_PARSE\", message: \"OCR 래스터 픽셀 상한으로 이미지 해상도를 축소했습니다 (작은 글자 인식 결손 가능)\" })\n  // 픽셀 상한으로 줄인 이미지는 같은 지면을 낮은 해상도로 본 것 — px/pt 도 같은 비율로\n  const scale = imageScale(width / shrink, height / shrink, density) * shrink\n  // 스캔·사진 기울기 보정 — 인식과 괘선 감지가 같은(바로 선) 래스터를 본다\n  const { rgba: data } = deskewPage(decoded, width, height)\n\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  const engine = await getOcrEngine()\n  const stats: OcrPageStats = { droppedLowConf: 0 }\n  const items = await engine.recognizePage(data, width, height, stats)\n  if (stats.droppedLowConf > 0) {\n    warnings.push({\n      page: 1,\n      message: `저신뢰 OCR 라인 ${stats.droppedLowConf}개 폐기 (인식 결손 가능)`,\n      code: \"OCR_LOW_CONF\",\n    })\n  }\n  if (stats.truncatedBoxes) {\n    warnings.push({ page: 1, message: `OCR 검출 상자가 너무 많아 ${stats.truncatedBoxes}개는 인식하지 않음 (일부 글 결손)`, code: \"PARTIAL_PARSE\" })\n  }\n  if (items.length === 0) {\n    warnings.push({ page: 1, message: \"이미지에서 텍스트를 인식하지 못했습니다\", code: \"OCR_FAILED\" })\n    return { blocks: [], warnings }\n  }\n\n  const pdfW = width / scale\n  const pdfH = height / scale\n  const ruling = detectRulingLines(data, width, height, scale)\n  const extraLines = rulingToPdfLines(ruling, scale, pdfH)\n  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, scale, extraLines, options?.tables !== false), warnings }\n}\n\n/**\n * sharp 로 RGBA 디코딩 — 미설치는 MISSING_DEPENDENCY 로 분류되도록 안내 메시지 throw.\n * EXIF 방향대로 세운다(`rotate()`): 폰으로 세로 촬영한 JPEG 은 픽셀을 눕혀 저장하고 Orientation 6 으로 표시 방향만 적는다 —\n * 무시하면 누운 글을 읽어 한글 131자 쪽이 쓰레기 57자가 됐다. 픽셀이 MAX_OCR_PIXELS 를 넘으면 그 안으로 줄인다(shrink = 줄인 배율).\n * (테스트용 export)\n */\nexport async function decodeToRgba(\n  buffer: ArrayBuffer,\n): Promise<{ data: Uint8Array; width: number; height: number; density?: number; shrink: number }> {\n  type Chain = {\n    rotate(): Chain\n    resize(w: number, h: number, opts: { fit: \"fill\" }): Chain\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: number } }> } }\n    metadata(): Promise<{ density?: number; width?: number; height?: number; orientation?: number }>\n  }\n  type SharpFactory = (input: Buffer) => Chain\n  let sharp: SharpFactory\n  try {\n    const mod = (await import(\"sharp\")) as unknown as SharpFactory | { default?: SharpFactory }\n    sharp = typeof mod === \"function\" ? mod : (mod.default ?? (mod as unknown as SharpFactory))\n  } catch (e) {\n    // KordocError + \"optional dependency\" 문구 → sanitizeError 메시지 보존 + MISSING_DEPENDENCY 분류\n    throw new KordocError(\n      \"이미지 파싱에는 optional dependency 'sharp' 가 필요합니다. \" +\n        `\\`npm install sharp\\` 후 다시 실행하세요.${OPTIONAL_DEP_INSTALL_HINT} 원인: ${(e as Error).message}`,\n    )\n  }\n  const input = Buffer.from(buffer)\n  const meta: { density?: number; width?: number; height?: number; orientation?: number } = await sharp(input).metadata().catch(() => ({}))\n  // 5~8 은 90° 돌린 방향 — 세운 뒤 가로·세로가 바뀐다\n  const [w0, h0] = (meta.orientation ?? 1) >= 5 ? [meta.height ?? 0, meta.width ?? 0] : [meta.width ?? 0, meta.height ?? 0]\n  const shrink = w0 * h0 > MAX_OCR_PIXELS ? Math.sqrt(MAX_OCR_PIXELS / (w0 * h0)) : 1\n  let chain = sharp(input).rotate()\n  if (shrink < 1) chain = chain.resize(Math.max(1, Math.floor(w0 * shrink)), Math.max(1, Math.floor(h0 * shrink)), { fit: \"fill\" })\n  const { data, info } = await chain.ensureAlpha().raw().toBuffer({ resolveWithObject: true })\n  return {\n    data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height, density: meta.density,\n    shrink: w0 ? info.width / w0 : 1,\n  }\n}\n\n/** OcrProvider 계약용 mime 판별 */\nfunction detectImageMime(buffer: ArrayBuffer): \"image/png\" | \"image/jpeg\" | \"image/webp\" {\n  const b = new Uint8Array(buffer, 0, Math.min(12, buffer.byteLength))\n  if (b[0] === 0xff && b[1] === 0xd8) return \"image/jpeg\"\n  if (b[0] === 0x52 && b[1] === 0x49) return \"image/webp\"\n  return \"image/png\"\n}\n"]}

package/dist/{image-ocr-UOAHUR4Y.js → image-ocr-PAF76DAW.js}

@@ -6,20 +6,20 @@ import {
   getOcrEngine,
   ocrItemsToBlocks,
   rulingToPdfLines
-} from "./chunk-A72XCFET.js";
+} from "./chunk-7LOD4IGO.js";
 import "./chunk-6H25SXDI.js";
-import "./chunk-G2PGKJUH.js";
+import "./chunk-INESPVEV.js";
 import {
   ensureOcrModels
-} from "./chunk-MXCQIXGJ.js";
-import "./chunk-QYT3XW2J.js";
-import "./chunk-CXNMHMPS.js";
-import "./chunk-R2NGHNS3.js";
-import "./chunk-MFA3TJWM.js";
+} from "./chunk-LKA7Z2NH.js";
+import "./chunk-JWTZ3OVZ.js";
+import "./chunk-ISC2TH24.js";
+import "./chunk-XVB3PUPJ.js";
+import "./chunk-JKHU7HID.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT
-} from "./chunk-5V4XXKGX.js";
+} from "./chunk-PN5WWW56.js";
 
 // src/ocr/image-ocr.ts
 var IMAGE_SCALE = 3;
@@ -112,4 +112,4 @@ export {
   imageScale,
   parseImageDocument
 };
-//# sourceMappingURL=image-ocr-UOAHUR4Y.js.map
+//# sourceMappingURL=image-ocr-PAF76DAW.js.map

package/dist/{image-ocr-LFMD7GFW.js → image-ocr-VH54SDZ7.js}

@@ -5,17 +5,17 @@ import {
   getOcrEngine,
   ocrItemsToBlocks,
   rulingToPdfLines
-} from "./chunk-GZAYV2SV.js";
+} from "./chunk-32ZUIHSY.js";
 import "./chunk-DC6CTQTK.js";
 import {
   ensureOcrModels
-} from "./chunk-IRTDNELH.js";
-import "./chunk-FWLSXLDC.js";
-import "./chunk-A2B5AYXD.js";
+} from "./chunk-ZNY5BRV6.js";
+import "./chunk-JPSLMPFI.js";
+import "./chunk-PHMWSWHJ.js";
 import {
   KordocError,
   OPTIONAL_DEP_INSTALL_HINT
-} from "./chunk-APPUKIPQ.js";
+} from "./chunk-EBXJ45VZ.js";
 
 // src/ocr/image-ocr.ts
 var IMAGE_SCALE = 3;
@@ -108,4 +108,4 @@ export {
   imageScale,
   parseImageDocument
 };
-//# sourceMappingURL=image-ocr-LFMD7GFW.js.map
+//# sourceMappingURL=image-ocr-VH54SDZ7.js.map