kordoc 4.14.2 → 4.14.4

package/dist/{image-ocr-ZTYAGV3W.cjs → image-ocr-NYBJEOH2.cjs}

@@ -4,17 +4,28 @@
 
 
 
-var _chunk422427XJcjs = require('./chunk-422427XJ.cjs');
-require('./chunk-EQZI77PP.cjs');
-require('./chunk-247ANM72.cjs');
-require('./chunk-L7DMSMAG.cjs');
 
+var _chunkIMAAFSAXcjs = require('./chunk-IMAAFSAX.cjs');
+require('./chunk-QEJUY6SK.cjs');
+require('./chunk-USHQCF5T.cjs');
+require('./chunk-VWUYVEFN.cjs');
 
-var _chunkHR2XNERBcjs = require('./chunk-HR2XNERB.cjs');
+
+var _chunkWJHAU73Bcjs = require('./chunk-WJHAU73B.cjs');
 require('./chunk-GS7T56RP.cjs');
 
 // src/ocr/image-ocr.ts
 var IMAGE_SCALE = 3;
+function imageScale(width, height, densityDpi) {
+  if (densityDpi && densityDpi >= 100 && densityDpi <= 1200) return densityDpi / 72;
+  const long = Math.max(width, height), short = Math.min(width, height);
+  if (short > 0) {
+    const r = long / short;
+    if (Math.abs(r / Math.SQRT2 - 1) <= 0.02) return long / 842;
+    if (Math.abs(r / (11 / 8.5) - 1) <= 0.02) return long / 792;
+  }
+  return IMAGE_SCALE;
+}
 async function parseImageDocument(buffer, options) {
   const warnings = [];
   if (typeof _optionalChain([options, 'optionalAccess', _ => _.ocr]) === "function") {
@@ -25,14 +36,16 @@ async function parseImageDocument(buffer, options) {
     }
     return { blocks: [{ type: "paragraph", text: text.trim(), pageNumber: 1 }], warnings };
   }
-  const { data, width, height } = await decodeToRgba(buffer);
-  await _chunk422427XJcjs.ensureOcrModels.call(void 0, (p) => {
+  const { data: decoded, width, height, density } = await decodeToRgba(buffer);
+  const scale = imageScale(width, height, density);
+  const { rgba: data } = _chunkIMAAFSAXcjs.deskewPage.call(void 0, decoded, width, height);
+  await _chunkIMAAFSAXcjs.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 _chunk422427XJcjs.getOcrEngine.call(void 0, );
+  const engine = await _chunkIMAAFSAXcjs.getOcrEngine.call(void 0, );
   const stats = { droppedLowConf: 0 };
   const items = await engine.recognizePage(data, width, height, stats);
   if (stats.droppedLowConf > 0) {
@@ -46,11 +59,11 @@ async function parseImageDocument(buffer, options) {
     warnings.push({ page: 1, message: "\uC774\uBBF8\uC9C0\uC5D0\uC11C \uD14D\uC2A4\uD2B8\uB97C \uC778\uC2DD\uD558\uC9C0 \uBABB\uD588\uC2B5\uB2C8\uB2E4", code: "OCR_FAILED" });
     return { blocks: [], warnings };
   }
-  const pdfW = width / IMAGE_SCALE;
-  const pdfH = height / IMAGE_SCALE;
-  const ruling = _chunk422427XJcjs.detectRulingLines.call(void 0, data, width, height, IMAGE_SCALE);
-  const extraLines = _chunk422427XJcjs.rulingToPdfLines.call(void 0, ruling, IMAGE_SCALE, pdfH);
-  return { blocks: _chunk422427XJcjs.ocrItemsToBlocks.call(void 0, items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines, _optionalChain([options, 'optionalAccess', _2 => _2.tables]) !== false), warnings };
+  const pdfW = width / scale;
+  const pdfH = height / scale;
+  const ruling = _chunkIMAAFSAXcjs.detectRulingLines.call(void 0, data, width, height, scale);
+  const extraLines = _chunkIMAAFSAXcjs.rulingToPdfLines.call(void 0, ruling, scale, pdfH);
+  return { blocks: _chunkIMAAFSAXcjs.ocrItemsToBlocks.call(void 0, items, 1, pdfW, pdfH, scale, extraLines, _optionalChain([options, 'optionalAccess', _2 => _2.tables]) !== false), warnings };
 }
 async function decodeToRgba(buffer) {
   let sharp;
@@ -58,12 +71,16 @@ 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, _chunkHR2XNERBcjs.KordocError)(
+    throw new (0, _chunkWJHAU73Bcjs.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. \uC6D0\uC778: ${e.message}`
     );
   }
-  const { data, info } = await sharp(Buffer.from(buffer)).ensureAlpha().raw().toBuffer({ resolveWithObject: true });
-  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height };
+  const input = Buffer.from(buffer);
+  const [{ data, info }, meta] = await Promise.all([
+    sharp(input).ensureAlpha().raw().toBuffer({ resolveWithObject: true }),
+    sharp(input).metadata().catch(() => ({ density: void 0 }))
+  ]);
+  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height, density: meta.density };
 }
 function detectImageMime(buffer) {
   const b = new Uint8Array(buffer, 0, Math.min(12, buffer.byteLength));
@@ -73,5 +90,6 @@ function detectImageMime(buffer) {
 }
 
 
-exports.parseImageDocument = parseImageDocument;
-//# sourceMappingURL=image-ocr-ZTYAGV3W.cjs.map
+
+exports.imageScale = imageScale; exports.parseImageDocument = parseImageDocument;
+//# sourceMappingURL=image-ocr-NYBJEOH2.cjs.map

package/dist/image-ocr-NYBJEOH2.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["/Users/chris_gomdori/workspace/kordoc/dist/image-ocr-NYBJEOH2.cjs","../src/ocr/image-ocr.ts"],"names":[],"mappings":"AAAA;AACE;AACA;AACA;AACA;AACA;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACA;ACKA,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,QAAQ,EAAA,EAAI,MAAM,YAAA,CAAa,MAAM,CAAA;AAC3E,EAAA,MAAM,MAAA,EAAQ,UAAA,CAAW,KAAA,EAAO,MAAA,EAAQ,OAAO,CAAA;AAE/C,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,EAAQ,EAAE,cAAA,EAAgB,EAAE,CAAA;AAClC,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,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;AAGA,MAAA,SAAe,YAAA,CACb,MAAA,EACgF;AAKhF,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,uMAAA,EAC4C,CAAA,CAAY,OAAO,CAAA;AAAA,IAAA;AACjE,EAAA;AAEF,EAAA;AACA,EAAA;AAAiD,IAAA;AACsB,IAAA;AACT,EAAA;AAE9D,EAAA;AACF;AAGA;AACE,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACF;ADtCA;AACA;AACA;AACA","file":"/Users/chris_gomdori/workspace/kordoc/dist/image-ocr-NYBJEOH2.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 } from \"../utils.js\"\nimport { getOcrEngine } from \"./engine.js\"\nimport { ensureOcrModels } from \"./models.js\"\nimport { detectRulingLines, rulingToPdfLines } from \"./ruling-lines.js\"\nimport { 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 } = await decodeToRgba(buffer)\n  const scale = imageScale(width, height, density)\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 = { 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 (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/** sharp 로 RGBA 디코딩 — 미설치는 MISSING_DEPENDENCY 로 분류되도록 안내 메시지 throw */\nasync function decodeToRgba(\n  buffer: ArrayBuffer,\n): Promise<{ data: Uint8Array; width: number; height: number; density?: number }> {\n  type SharpFactory = (input: Buffer) => {\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: number } }> } }\n    metadata(): Promise<{ density?: number }>\n  }\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. 원인: ${(e as Error).message}`,\n    )\n  }\n  const input = Buffer.from(buffer)\n  const [{ data, info }, meta] = await Promise.all([\n    sharp(input).ensureAlpha().raw().toBuffer({ resolveWithObject: true }),\n    sharp(input).metadata().catch(() => ({ density: undefined })),\n  ])\n  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height, density: meta.density }\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-FLIL7MGN.js → image-ocr-ZV6PDBUQ.js}

@@ -1,19 +1,30 @@
 import {
+  deskewPage,
   detectRulingLines,
   ensureOcrModels,
   getOcrEngine,
   ocrItemsToBlocks,
   rulingToPdfLines
-} from "./chunk-EP7KOI2V.js";
-import "./chunk-DOGPOY4B.js";
-import "./chunk-A7QC2ZVP.js";
-import "./chunk-FOJZNMMK.js";
+} from "./chunk-LFSQAMSM.js";
+import "./chunk-VDN7LMPL.js";
+import "./chunk-YT3GC62E.js";
+import "./chunk-6CUIXJOC.js";
 import {
   KordocError
-} from "./chunk-JQEFP45L.js";
+} from "./chunk-GUJ4Q73I.js";
 
 // src/ocr/image-ocr.ts
 var IMAGE_SCALE = 3;
+function imageScale(width, height, densityDpi) {
+  if (densityDpi && densityDpi >= 100 && densityDpi <= 1200) return densityDpi / 72;
+  const long = Math.max(width, height), short = Math.min(width, height);
+  if (short > 0) {
+    const r = long / short;
+    if (Math.abs(r / Math.SQRT2 - 1) <= 0.02) return long / 842;
+    if (Math.abs(r / (11 / 8.5) - 1) <= 0.02) return long / 792;
+  }
+  return IMAGE_SCALE;
+}
 async function parseImageDocument(buffer, options) {
   const warnings = [];
   if (typeof options?.ocr === "function") {
@@ -24,7 +35,9 @@ async function parseImageDocument(buffer, options) {
     }
     return { blocks: [{ type: "paragraph", text: text.trim(), pageNumber: 1 }], warnings };
   }
-  const { data, width, height } = await decodeToRgba(buffer);
+  const { data: decoded, width, height, density } = await decodeToRgba(buffer);
+  const scale = imageScale(width, height, density);
+  const { rgba: data } = deskewPage(decoded, width, height);
   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)...
@@ -45,11 +58,11 @@ async function parseImageDocument(buffer, options) {
     warnings.push({ page: 1, message: "\uC774\uBBF8\uC9C0\uC5D0\uC11C \uD14D\uC2A4\uD2B8\uB97C \uC778\uC2DD\uD558\uC9C0 \uBABB\uD588\uC2B5\uB2C8\uB2E4", code: "OCR_FAILED" });
     return { blocks: [], warnings };
   }
-  const pdfW = width / IMAGE_SCALE;
-  const pdfH = height / IMAGE_SCALE;
-  const ruling = detectRulingLines(data, width, height, IMAGE_SCALE);
-  const extraLines = rulingToPdfLines(ruling, IMAGE_SCALE, pdfH);
-  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines, options?.tables !== false), warnings };
+  const pdfW = width / scale;
+  const pdfH = height / scale;
+  const ruling = detectRulingLines(data, width, height, scale);
+  const extraLines = rulingToPdfLines(ruling, scale, pdfH);
+  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, scale, extraLines, options?.tables !== false), warnings };
 }
 async function decodeToRgba(buffer) {
   let sharp;
@@ -61,8 +74,12 @@ async function decodeToRgba(buffer) {
       `\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. \uC6D0\uC778: ${e.message}`
     );
   }
-  const { data, info } = await sharp(Buffer.from(buffer)).ensureAlpha().raw().toBuffer({ resolveWithObject: true });
-  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height };
+  const input = Buffer.from(buffer);
+  const [{ data, info }, meta] = await Promise.all([
+    sharp(input).ensureAlpha().raw().toBuffer({ resolveWithObject: true }),
+    sharp(input).metadata().catch(() => ({ density: void 0 }))
+  ]);
+  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height, density: meta.density };
 }
 function detectImageMime(buffer) {
   const b = new Uint8Array(buffer, 0, Math.min(12, buffer.byteLength));
@@ -71,6 +88,7 @@ function detectImageMime(buffer) {
   return "image/png";
 }
 export {
+  imageScale,
   parseImageDocument
 };
-//# sourceMappingURL=image-ocr-FLIL7MGN.js.map
+//# sourceMappingURL=image-ocr-ZV6PDBUQ.js.map

package/dist/image-ocr-ZV6PDBUQ.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/image-ocr.ts"],"sourcesContent":["/**\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 } from \"../utils.js\"\nimport { getOcrEngine } from \"./engine.js\"\nimport { ensureOcrModels } from \"./models.js\"\nimport { detectRulingLines, rulingToPdfLines } from \"./ruling-lines.js\"\nimport { 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 } = await decodeToRgba(buffer)\n  const scale = imageScale(width, height, density)\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 = { 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 (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/** sharp 로 RGBA 디코딩 — 미설치는 MISSING_DEPENDENCY 로 분류되도록 안내 메시지 throw */\nasync function decodeToRgba(\n  buffer: ArrayBuffer,\n): Promise<{ data: Uint8Array; width: number; height: number; density?: number }> {\n  type SharpFactory = (input: Buffer) => {\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: number } }> } }\n    metadata(): Promise<{ density?: number }>\n  }\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. 원인: ${(e as Error).message}`,\n    )\n  }\n  const input = Buffer.from(buffer)\n  const [{ data, info }, meta] = await Promise.all([\n    sharp(input).ensureAlpha().raw().toBuffer({ resolveWithObject: true }),\n    sharp(input).metadata().catch(() => ({ density: undefined })),\n  ])\n  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height, density: meta.density }\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"],"mappings":";;;;;;;;;;;;;;;;AAqBA,IAAM,cAAc;AASb,SAAS,WAAW,OAAe,QAAgB,YAA6B;AACrF,MAAI,cAAc,cAAc,OAAO,cAAc,KAAM,QAAO,aAAa;AAC/E,QAAM,OAAO,KAAK,IAAI,OAAO,MAAM,GAAG,QAAQ,KAAK,IAAI,OAAO,MAAM;AACpE,MAAI,QAAQ,GAAG;AACb,UAAM,IAAI,OAAO;AACjB,QAAI,KAAK,IAAI,IAAI,KAAK,QAAQ,CAAC,KAAK,KAAM,QAAO,OAAO;AACxD,QAAI,KAAK,IAAI,KAAK,KAAK,OAAO,CAAC,KAAK,KAAM,QAAO,OAAO;AAAA,EAC1D;AACA,SAAO;AACT;AAQA,eAAsB,mBACpB,QACA,SACyB;AACzB,QAAM,WAA2B,CAAC;AAGlC,MAAI,OAAO,SAAS,QAAQ,YAAY;AACtC,UAAM,OAAO,MAAM,QAAQ,IAAI,IAAI,WAAW,MAAM,GAAG,GAAG,gBAAgB,MAAM,CAAC;AACjF,QAAI,CAAC,KAAK,KAAK,GAAG;AAChB,eAAS,KAAK,EAAE,MAAM,GAAG,SAAS,iCAAa,MAAM,aAAa,CAAC;AACnE,aAAO,EAAE,QAAQ,CAAC,GAAG,SAAS;AAAA,IAChC;AACA,WAAO,EAAE,QAAQ,CAAC,EAAE,MAAM,aAAa,MAAM,KAAK,KAAK,GAAG,YAAY,EAAE,CAAC,GAAG,SAAS;AAAA,EACvF;AAEA,QAAM,EAAE,MAAM,SAAS,OAAO,QAAQ,QAAQ,IAAI,MAAM,aAAa,MAAM;AAC3E,QAAM,QAAQ,WAAW,OAAO,QAAQ,OAAO;AAE/C,QAAM,EAAE,MAAM,KAAK,IAAI,WAAW,SAAS,OAAO,MAAM;AAExD,QAAM,gBAAgB,OAAK;AACzB,QAAI,EAAE,UAAU,cAAc,EAAE,eAAe,GAAG;AAChD,cAAQ,OAAO,MAAM,gBAAgB,EAAE,KAAK,IAAI,sCAAa,EAAE,KAAK,MAAM;AAAA,CAAU;AAAA,IACtF;AAAA,EACF,CAAC;AACD,QAAM,SAAS,MAAM,aAAa;AAClC,QAAM,QAAQ,EAAE,gBAAgB,EAAE;AAClC,QAAM,QAAQ,MAAM,OAAO,cAAc,MAAM,OAAO,QAAQ,KAAK;AACnE,MAAI,MAAM,iBAAiB,GAAG;AAC5B,aAAS,KAAK;AAAA,MACZ,MAAM;AAAA,MACN,SAAS,uCAAc,MAAM,cAAc;AAAA,MAC3C,MAAM;AAAA,IACR,CAAC;AAAA,EACH;AACA,MAAI,MAAM,WAAW,GAAG;AACtB,aAAS,KAAK,EAAE,MAAM,GAAG,SAAS,mHAAyB,MAAM,aAAa,CAAC;AAC/E,WAAO,EAAE,QAAQ,CAAC,GAAG,SAAS;AAAA,EAChC;AAEA,QAAM,OAAO,QAAQ;AACrB,QAAM,OAAO,SAAS;AACtB,QAAM,SAAS,kBAAkB,MAAM,OAAO,QAAQ,KAAK;AAC3D,QAAM,aAAa,iBAAiB,QAAQ,OAAO,IAAI;AACvD,SAAO,EAAE,QAAQ,iBAAiB,OAAO,GAAG,MAAM,MAAM,OAAO,YAAY,SAAS,WAAW,KAAK,GAAG,SAAS;AAClH;AAGA,eAAe,aACb,QACgF;AAKhF,MAAI;AACJ,MAAI;AACF,UAAM,MAAO,MAAM,OAAO,OAAO;AACjC,YAAQ,OAAO,QAAQ,aAAa,MAAO,IAAI,WAAY;AAAA,EAC7D,SAAS,GAAG;AAEV,UAAM,IAAI;AAAA,MACR,0MAC4C,EAAY,OAAO;AAAA,IACjE;AAAA,EACF;AACA,QAAM,QAAQ,OAAO,KAAK,MAAM;AAChC,QAAM,CAAC,EAAE,MAAM,KAAK,GAAG,IAAI,IAAI,MAAM,QAAQ,IAAI;AAAA,IAC/C,MAAM,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,SAAS,EAAE,mBAAmB,KAAK,CAAC;AAAA,IACrE,MAAM,KAAK,EAAE,SAAS,EAAE,MAAM,OAAO,EAAE,SAAS,OAAU,EAAE;AAAA,EAC9D,CAAC;AACD,SAAO,EAAE,MAAM,IAAI,WAAW,KAAK,QAAQ,KAAK,YAAY,KAAK,UAAU,GAAG,OAAO,KAAK,OAAO,QAAQ,KAAK,QAAQ,SAAS,KAAK,QAAQ;AAC9I;AAGA,SAAS,gBAAgB,QAAgE;AACvF,QAAM,IAAI,IAAI,WAAW,QAAQ,GAAG,KAAK,IAAI,IAAI,OAAO,UAAU,CAAC;AACnE,MAAI,EAAE,CAAC,MAAM,OAAQ,EAAE,CAAC,MAAM,IAAM,QAAO;AAC3C,MAAI,EAAE,CAAC,MAAM,MAAQ,EAAE,CAAC,MAAM,GAAM,QAAO;AAC3C,SAAO;AACT;","names":[]}