kordoc 4.3.1 → 4.4.0

package/dist/{image-ocr-6MO4FXJK.js → image-ocr-U2DRRFL3.js}

@@ -4,12 +4,12 @@ import {
   getOcrEngine,
   ocrItemsToBlocks,
   rulingToPdfLines
-} from "./chunk-PV3NVBGL.js";
-import "./chunk-3FFPHTXD.js";
-import "./chunk-CD5X6UG5.js";
+} from "./chunk-YVYU267E.js";
+import "./chunk-3H5X63BQ.js";
+import "./chunk-GHGZU2LQ.js";
 import {
   KordocError
-} from "./chunk-LEC2HG7A.js";
+} from "./chunk-ONS7JKVA.js";
 
 // src/ocr/image-ocr.ts
 var IMAGE_SCALE = 3;
@@ -64,4 +64,4 @@ function detectImageMime(buffer) {
 export {
   parseImageDocument
 };
-//# sourceMappingURL=image-ocr-6MO4FXJK.js.map
+//# sourceMappingURL=image-ocr-U2DRRFL3.js.map

package/dist/image-ocr-U2DRRFL3.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 216dpi(스케일 3) 가정으로 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\"\n\n/** 좌표 환산 스케일 (px/pt) — PDF OCR 렌더(216dpi)와 동일 기준 */\nconst IMAGE_SCALE = 3\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, width, height } = await decodeToRgba(buffer)\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 items = await engine.recognizePage(data, width, height)\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 / IMAGE_SCALE\n  const pdfH = height / IMAGE_SCALE\n  const ruling = detectRulingLines(data, width, height, IMAGE_SCALE)\n  const extraLines = rulingToPdfLines(ruling, IMAGE_SCALE, pdfH)\n  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines), 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 }> {\n  type SharpFactory = (input: Buffer) => {\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: 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 { data, info } = await sharp(Buffer.from(buffer)).ensureAlpha().raw().toBuffer({ resolveWithObject: true })\n  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height }\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":";;;;;;;;;;;;;;AAmBA,IAAM,cAAc;AAQpB,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,OAAO,OAAO,IAAI,MAAM,aAAa,MAAM;AAEzD,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,MAAM,OAAO,cAAc,MAAM,OAAO,MAAM;AAC5D,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,WAAW;AACjE,QAAM,aAAa,iBAAiB,QAAQ,aAAa,IAAI;AAC7D,SAAO,EAAE,QAAQ,iBAAiB,OAAO,GAAG,MAAM,MAAM,aAAa,UAAU,GAAG,SAAS;AAC7F;AAGA,eAAe,aACb,QAC8D;AAI9D,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,EAAE,MAAM,KAAK,IAAI,MAAM,MAAM,OAAO,KAAK,MAAM,CAAC,EAAE,YAAY,EAAE,IAAI,EAAE,SAAS,EAAE,mBAAmB,KAAK,CAAC;AAChH,SAAO,EAAE,MAAM,IAAI,WAAW,KAAK,QAAQ,KAAK,YAAY,KAAK,UAAU,GAAG,OAAO,KAAK,OAAO,QAAQ,KAAK,OAAO;AACvH;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":[]}

package/dist/{image-ocr-TWUU2RNG.js → image-ocr-XC52AQDV.js}

@@ -4,16 +4,16 @@ import {
   getOcrEngine,
   ocrItemsToBlocks,
   rulingToPdfLines
-} from "./chunk-YDCD4AV4.js";
-import "./chunk-HPRK67YB.js";
-import "./chunk-ZSL6MQVE.js";
+} from "./chunk-HFJF477F.js";
+import "./chunk-W6CUD7V3.js";
+import "./chunk-A73RMJ6B.js";
 import {
   KordocError
-} from "./chunk-ZJ75YDW7.js";
+} from "./chunk-P7YPCK7C.js";
 import {
   ensureOcrModels
-} from "./chunk-6XPXIG57.js";
-import "./chunk-YNJDNP44.js";
+} from "./chunk-GZDOWYJM.js";
+import "./chunk-UAFSPWU7.js";
 
 // src/ocr/image-ocr.ts
 var IMAGE_SCALE = 3;
@@ -68,4 +68,4 @@ function detectImageMime(buffer) {
 export {
   parseImageDocument
 };
-//# sourceMappingURL=image-ocr-TWUU2RNG.js.map
+//# sourceMappingURL=image-ocr-XC52AQDV.js.map

package/dist/image-ocr-XC52AQDV.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 216dpi(스케일 3) 가정으로 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\"\n\n/** 좌표 환산 스케일 (px/pt) — PDF OCR 렌더(216dpi)와 동일 기준 */\nconst IMAGE_SCALE = 3\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, width, height } = await decodeToRgba(buffer)\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 items = await engine.recognizePage(data, width, height)\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 / IMAGE_SCALE\n  const pdfH = height / IMAGE_SCALE\n  const ruling = detectRulingLines(data, width, height, IMAGE_SCALE)\n  const extraLines = rulingToPdfLines(ruling, IMAGE_SCALE, pdfH)\n  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines), 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 }> {\n  type SharpFactory = (input: Buffer) => {\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: 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 { data, info } = await sharp(Buffer.from(buffer)).ensureAlpha().raw().toBuffer({ resolveWithObject: true })\n  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height }\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":";;;;;;;;;;;;;;;;;;AAmBA,IAAM,cAAc;AAQpB,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,OAAO,OAAO,IAAI,MAAM,aAAa,MAAM;AAEzD,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,MAAM,OAAO,cAAc,MAAM,OAAO,MAAM;AAC5D,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,WAAW;AACjE,QAAM,aAAa,iBAAiB,QAAQ,aAAa,IAAI;AAC7D,SAAO,EAAE,QAAQ,iBAAiB,OAAO,GAAG,MAAM,MAAM,aAAa,UAAU,GAAG,SAAS;AAC7F;AAGA,eAAe,aACb,QAC8D;AAI9D,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,EAAE,MAAM,KAAK,IAAI,MAAM,MAAM,OAAO,KAAK,MAAM,CAAC,EAAE,YAAY,EAAE,IAAI,EAAE,SAAS,EAAE,mBAAmB,KAAK,CAAC;AAChH,SAAO,EAAE,MAAM,IAAI,WAAW,KAAK,QAAQ,KAAK,YAAY,KAAK,UAAU,GAAG,OAAO,KAAK,OAAO,QAAQ,KAAK,OAAO;AACvH;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":[]}