kordoc 2.2.5 → 2.3.0

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "kordoc",
-  "version": "2.2.5",
-  "description": "Parse Korean documents (HWP, HWPX, PDF, XLSX, DOCX) to Markdown",
+  "version": "2.3.0",
+  "description": "Parse Korean documents (HWP, HWPX, HWPML, PDF, XLSX, DOCX) to Markdown",
   "type": "module",
   "exports": {
     ".": {

package/dist/chunk-5Y2Q3BRW.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../src/detect.ts"],"sourcesContent":["/** 매직 바이트 기반 파일 포맷 감지 */\n\nimport JSZip from \"jszip\"\nimport type { FileType } from \"./types.js\"\n\n/** 매직 바이트 뷰 생성 (복사 없이 view) */\nfunction magicBytes(buffer: ArrayBuffer): Uint8Array {\n  return new Uint8Array(buffer, 0, Math.min(4, buffer.byteLength))\n}\n\n/** ZIP 파일 여부: PK\\x03\\x04 */\nexport function isZipFile(buffer: ArrayBuffer): boolean {\n  const b = magicBytes(buffer)\n  return b[0] === 0x50 && b[1] === 0x4b && b[2] === 0x03 && b[3] === 0x04\n}\n\n/** HWPX (ZIP 기반 한컴 문서): PK\\x03\\x04 — 하위 호환용 */\nexport function isHwpxFile(buffer: ArrayBuffer): boolean {\n  return isZipFile(buffer)\n}\n\n/** HWP 5.x (OLE2 바이너리 한컴 문서): \\xD0\\xCF\\x11\\xE0 */\nexport function isOldHwpFile(buffer: ArrayBuffer): boolean {\n  const b = magicBytes(buffer)\n  return b[0] === 0xd0 && b[1] === 0xcf && b[2] === 0x11 && b[3] === 0xe0\n}\n\n/** PDF 문서: %PDF */\nexport function isPdfFile(buffer: ArrayBuffer): boolean {\n  const b = magicBytes(buffer)\n  return b[0] === 0x25 && b[1] === 0x50 && b[2] === 0x44 && b[3] === 0x46\n}\n\n/** 동기 포맷 감지 — ZIP은 모두 \"hwpx\"로 반환 (하위 호환) */\nexport function detectFormat(buffer: ArrayBuffer): FileType {\n  if (buffer.byteLength < 4) return \"unknown\"\n  if (isZipFile(buffer)) return \"hwpx\"\n  if (isOldHwpFile(buffer)) return \"hwp\"\n  if (isPdfFile(buffer)) return \"pdf\"\n  return \"unknown\"\n}\n\n/**\n * ZIP 내부 구조 기반 포맷 세분화.\n * HWPX, XLSX, DOCX 모두 ZIP이므로 내부 파일로 구분.\n */\nexport async function detectZipFormat(buffer: ArrayBuffer): Promise<\"hwpx\" | \"xlsx\" | \"docx\" | \"unknown\"> {\n  try {\n    const zip = await JSZip.loadAsync(buffer)\n    // XLSX: xl/workbook.xml\n    if (zip.file(\"xl/workbook.xml\")) return \"xlsx\"\n    // DOCX: word/document.xml\n    if (zip.file(\"word/document.xml\")) return \"docx\"\n    // HWPX: Contents/ 또는 content.hpf 또는 mimetype\n    if (zip.file(\"Contents/content.hpf\") || zip.file(\"mimetype\")) return \"hwpx\"\n    // 기타 ZIP 내에 section 파일이 있으면 HWPX로 추정\n    const hasSection = Object.keys(zip.files).some(f => f.startsWith(\"Contents/\"))\n    if (hasSection) return \"hwpx\"\n    return \"unknown\"\n  } catch {\n    return \"unknown\"\n  }\n}\n"],"mappings":";;;AAEA,OAAO,WAAW;AAIlB,SAAS,WAAW,QAAiC;AACnD,SAAO,IAAI,WAAW,QAAQ,GAAG,KAAK,IAAI,GAAG,OAAO,UAAU,CAAC;AACjE;AAGO,SAAS,UAAU,QAA8B;AACtD,QAAM,IAAI,WAAW,MAAM;AAC3B,SAAO,EAAE,CAAC,MAAM,MAAQ,EAAE,CAAC,MAAM,MAAQ,EAAE,CAAC,MAAM,KAAQ,EAAE,CAAC,MAAM;AACrE;AAGO,SAAS,WAAW,QAA8B;AACvD,SAAO,UAAU,MAAM;AACzB;AAGO,SAAS,aAAa,QAA8B;AACzD,QAAM,IAAI,WAAW,MAAM;AAC3B,SAAO,EAAE,CAAC,MAAM,OAAQ,EAAE,CAAC,MAAM,OAAQ,EAAE,CAAC,MAAM,MAAQ,EAAE,CAAC,MAAM;AACrE;AAGO,SAAS,UAAU,QAA8B;AACtD,QAAM,IAAI,WAAW,MAAM;AAC3B,SAAO,EAAE,CAAC,MAAM,MAAQ,EAAE,CAAC,MAAM,MAAQ,EAAE,CAAC,MAAM,MAAQ,EAAE,CAAC,MAAM;AACrE;AAGO,SAAS,aAAa,QAA+B;AAC1D,MAAI,OAAO,aAAa,EAAG,QAAO;AAClC,MAAI,UAAU,MAAM,EAAG,QAAO;AAC9B,MAAI,aAAa,MAAM,EAAG,QAAO;AACjC,MAAI,UAAU,MAAM,EAAG,QAAO;AAC9B,SAAO;AACT;AAMA,eAAsB,gBAAgB,QAAoE;AACxG,MAAI;AACF,UAAM,MAAM,MAAM,MAAM,UAAU,MAAM;AAExC,QAAI,IAAI,KAAK,iBAAiB,EAAG,QAAO;AAExC,QAAI,IAAI,KAAK,mBAAmB,EAAG,QAAO;AAE1C,QAAI,IAAI,KAAK,sBAAsB,KAAK,IAAI,KAAK,UAAU,EAAG,QAAO;AAErE,UAAM,aAAa,OAAO,KAAK,IAAI,KAAK,EAAE,KAAK,OAAK,EAAE,WAAW,WAAW,CAAC;AAC7E,QAAI,WAAY,QAAO;AACvB,WAAO;AAAA,EACT,QAAQ;AACN,WAAO;AAAA,EACT;AACF;","names":[]}