kordoc 4.3.1 → 4.4.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (168) hide show
  1. package/NOTICE +176 -0
  2. package/README.md +936 -922
  3. package/THIRD_PARTY/apache-2.0.LICENSE +201 -0
  4. package/THIRD_PARTY/claw-hwp.LICENSE +21 -0
  5. package/THIRD_PARTY/hml-equation-parser.LICENSE +201 -0
  6. package/THIRD_PARTY/hml-equation-parser.txt +35 -0
  7. package/THIRD_PARTY/opendataloader-pdf.txt +32 -0
  8. package/THIRD_PARTY/pix2text.txt +17 -0
  9. package/THIRD_PARTY/rhwp-forms.txt +35 -0
  10. package/dist/{-RH6BEVTT.js → -R6NUMPRV.js} +11 -11
  11. package/dist/{chunk-3FFPHTXD.js → chunk-3H5X63BQ.js} +1 -1
  12. package/dist/chunk-3H5X63BQ.js.map +1 -0
  13. package/dist/{chunk-HZU7Y22R.js → chunk-5KLE3U23.js} +7 -4
  14. package/dist/chunk-5KLE3U23.js.map +1 -0
  15. package/dist/{chunk-6GBLFQMA.js → chunk-5RPYBC2Q.js} +1 -1
  16. package/dist/chunk-5RPYBC2Q.js.map +1 -0
  17. package/dist/{chunk-SA22NK56.js → chunk-65DYCXGC.js} +4 -4
  18. package/dist/chunk-65DYCXGC.js.map +1 -0
  19. package/dist/{chunk-W34QSJFF.js → chunk-7FCNOKOR.js} +1 -1
  20. package/dist/chunk-7FCNOKOR.js.map +1 -0
  21. package/dist/{chunk-ZSL6MQVE.js → chunk-A73RMJ6B.js} +1 -1
  22. package/dist/chunk-A73RMJ6B.js.map +1 -0
  23. package/dist/{chunk-5F5UCACZ.cjs → chunk-AB322K3O.cjs} +6 -6
  24. package/dist/chunk-AB322K3O.cjs.map +1 -0
  25. package/dist/chunk-DCZVOIEO.cjs.map +1 -1
  26. package/dist/{chunk-O5EK52TG.js → chunk-DZIXKL7E.js} +1 -1
  27. package/dist/chunk-DZIXKL7E.js.map +1 -0
  28. package/dist/{chunk-CD5X6UG5.js → chunk-GHGZU2LQ.js} +2 -2
  29. package/dist/chunk-GHGZU2LQ.js.map +1 -0
  30. package/dist/chunk-GS7T56RP.cjs.map +1 -1
  31. package/dist/{chunk-6XPXIG57.js → chunk-GZDOWYJM.js} +2 -2
  32. package/dist/chunk-GZDOWYJM.js.map +1 -0
  33. package/dist/{chunk-VQB5P5KJ.cjs → chunk-H2KMJOB6.cjs} +2 -2
  34. package/dist/chunk-H2KMJOB6.cjs.map +1 -0
  35. package/dist/{chunk-YDCD4AV4.js → chunk-HFJF477F.js} +3 -3
  36. package/dist/chunk-HFJF477F.js.map +1 -0
  37. package/dist/{chunk-ZYYTU7NY.js → chunk-HNRULYUH.js} +2 -2
  38. package/dist/chunk-HNRULYUH.js.map +1 -0
  39. package/dist/chunk-MOL7MDBG.js +0 -0
  40. package/dist/{chunk-LEC2HG7A.js → chunk-ONS7JKVA.js} +2 -2
  41. package/dist/chunk-ONS7JKVA.js.map +1 -0
  42. package/dist/{chunk-ZJ75YDW7.js → chunk-P7YPCK7C.js} +2 -2
  43. package/dist/chunk-P7YPCK7C.js.map +1 -0
  44. package/dist/{chunk-BPJSTJBN.js → chunk-QXYB74QF.js} +24 -3
  45. package/dist/chunk-QXYB74QF.js.map +1 -0
  46. package/dist/{chunk-BYEQDLZS.js → chunk-RZURSA4N.js} +6 -4
  47. package/dist/chunk-RZURSA4N.js.map +1 -0
  48. package/dist/{chunk-7DZPPUQD.cjs → chunk-SKM6KZUH.cjs} +9 -9
  49. package/dist/chunk-SKM6KZUH.cjs.map +1 -0
  50. package/dist/{chunk-YNJDNP44.js → chunk-UAFSPWU7.js} +1 -1
  51. package/dist/chunk-UAFSPWU7.js.map +1 -0
  52. package/dist/{chunk-ZMKTKJZN.cjs → chunk-UDGPM5R5.cjs} +26 -5
  53. package/dist/chunk-UDGPM5R5.cjs.map +1 -0
  54. package/dist/{chunk-GTPEKBN6.js → chunk-V35DRUJC.js} +2 -2
  55. package/dist/chunk-V35DRUJC.js.map +1 -0
  56. package/dist/{chunk-HPRK67YB.js → chunk-W6CUD7V3.js} +3 -3
  57. package/dist/chunk-W6CUD7V3.js.map +1 -0
  58. package/dist/{chunk-FUXDD36A.js → chunk-XCIC7U6Y.js} +24 -3
  59. package/dist/chunk-XCIC7U6Y.js.map +1 -0
  60. package/dist/{chunk-RO4Y4BFI.js → chunk-YOZ7ZRCD.js} +366 -51
  61. package/dist/chunk-YOZ7ZRCD.js.map +1 -0
  62. package/dist/{chunk-PV3NVBGL.js → chunk-YVYU267E.js} +3 -3
  63. package/dist/chunk-YVYU267E.js.map +1 -0
  64. package/dist/{chunk-MGGU3ZRK.cjs → chunk-ZV5OKQ3Z.cjs} +1 -1
  65. package/dist/chunk-ZV5OKQ3Z.cjs.map +1 -0
  66. package/dist/chunks-QMGCI5DG.js +10 -0
  67. package/dist/cli.js +28 -24
  68. package/dist/cli.js.map +1 -1
  69. package/dist/{formula-IWV5IJ3Y.cjs → formula-CG4S7TWO.cjs} +8 -8
  70. package/dist/formula-CG4S7TWO.cjs.map +1 -0
  71. package/dist/{formula-7ZPC54JO.js → formula-JLYS73QL.js} +2 -2
  72. package/dist/formula-JLYS73QL.js.map +1 -0
  73. package/dist/{formula-EQRHSMYW.js → formula-LIVQSCMM.js} +2 -2
  74. package/dist/formula-LIVQSCMM.js.map +1 -0
  75. package/dist/{image-ocr-MGVXT6XD.cjs → image-ocr-F2BQA26V.cjs} +11 -11
  76. package/dist/image-ocr-F2BQA26V.cjs.map +1 -0
  77. package/dist/{image-ocr-6MO4FXJK.js → image-ocr-U2DRRFL3.js} +5 -5
  78. package/dist/image-ocr-U2DRRFL3.js.map +1 -0
  79. package/dist/{image-ocr-TWUU2RNG.js → image-ocr-XC52AQDV.js} +7 -7
  80. package/dist/image-ocr-XC52AQDV.js.map +1 -0
  81. package/dist/index.cjs +858 -537
  82. package/dist/index.cjs.map +1 -1
  83. package/dist/index.d.cts +7 -0
  84. package/dist/index.d.ts +7 -0
  85. package/dist/index.js +373 -52
  86. package/dist/index.js.map +1 -1
  87. package/dist/mcp.js +24 -22
  88. package/dist/mcp.js.map +1 -1
  89. package/dist/{models-6VMAVZR6.js → models-6SIKLKRH.js} +3 -3
  90. package/dist/page-range-737B4EZW.js +0 -0
  91. package/dist/page-range-P7SDW6LR.cjs.map +1 -1
  92. package/dist/{parser-VR2BTUT7.js → parser-ESGFWWCT.js} +8 -8
  93. package/dist/parser-ESGFWWCT.js.map +1 -0
  94. package/dist/{parser-JFGLN4DA.cjs → parser-FFQN67O6.cjs} +25 -25
  95. package/dist/parser-FFQN67O6.cjs.map +1 -0
  96. package/dist/{parser-YA7HSSPK.js → parser-TFPGQWCC.js} +6 -6
  97. package/dist/parser-TFPGQWCC.js.map +1 -0
  98. package/dist/pdf-ocr-4R646OO6.js +12 -0
  99. package/dist/pdf-ocr-UQBZH3HE.cjs +13 -0
  100. package/dist/pdf-ocr-UQBZH3HE.cjs.map +1 -0
  101. package/dist/pdf-ocr-WE5TMRFU.js +15 -0
  102. package/dist/{profile-io-QPIMGVO3.js → profile-io-O6BDCUJN.js} +3 -3
  103. package/dist/{rasterize-AFYGNKXG.js → rasterize-V22VCCE7.js} +2 -2
  104. package/dist/rasterize-V22VCCE7.js.map +1 -0
  105. package/dist/{redact-W5K3ILQH.js → redact-7ILEAUTS.js} +2 -2
  106. package/dist/render-3BSVTTZE.js +10 -0
  107. package/dist/seal-6ZSEJVOD.js +10 -0
  108. package/dist/{setup-I4I75QJU.js → setup-GUQT5ELJ.js} +1 -1
  109. package/dist/setup-GUQT5ELJ.js.map +1 -0
  110. package/dist/{watch-D3JJEOAB.js → watch-HGKBICED.js} +11 -11
  111. package/dist/watch-HGKBICED.js.map +1 -0
  112. package/package.json +103 -101
  113. package/templates//352/260/204/354/235/264/352/270/260/354/225/210/353/254/270_/354/230/210/354/213/234/352/260/222.json +9 -9
  114. package/templates//354/235/274/353/260/230/352/270/260/354/225/210/353/254/270_/354/230/210/354/213/234/352/260/222.json +24 -24
  115. package/dist/chunk-3FFPHTXD.js.map +0 -1
  116. package/dist/chunk-5F5UCACZ.cjs.map +0 -1
  117. package/dist/chunk-6GBLFQMA.js.map +0 -1
  118. package/dist/chunk-6XPXIG57.js.map +0 -1
  119. package/dist/chunk-7DZPPUQD.cjs.map +0 -1
  120. package/dist/chunk-BPJSTJBN.js.map +0 -1
  121. package/dist/chunk-BYEQDLZS.js.map +0 -1
  122. package/dist/chunk-CD5X6UG5.js.map +0 -1
  123. package/dist/chunk-FUXDD36A.js.map +0 -1
  124. package/dist/chunk-GTPEKBN6.js.map +0 -1
  125. package/dist/chunk-HPRK67YB.js.map +0 -1
  126. package/dist/chunk-HZU7Y22R.js.map +0 -1
  127. package/dist/chunk-LEC2HG7A.js.map +0 -1
  128. package/dist/chunk-MGGU3ZRK.cjs.map +0 -1
  129. package/dist/chunk-O5EK52TG.js.map +0 -1
  130. package/dist/chunk-PV3NVBGL.js.map +0 -1
  131. package/dist/chunk-RO4Y4BFI.js.map +0 -1
  132. package/dist/chunk-SA22NK56.js.map +0 -1
  133. package/dist/chunk-VQB5P5KJ.cjs.map +0 -1
  134. package/dist/chunk-W34QSJFF.js.map +0 -1
  135. package/dist/chunk-YDCD4AV4.js.map +0 -1
  136. package/dist/chunk-YNJDNP44.js.map +0 -1
  137. package/dist/chunk-ZJ75YDW7.js.map +0 -1
  138. package/dist/chunk-ZMKTKJZN.cjs.map +0 -1
  139. package/dist/chunk-ZSL6MQVE.js.map +0 -1
  140. package/dist/chunk-ZYYTU7NY.js.map +0 -1
  141. package/dist/chunks-JEYQWXIZ.js +0 -10
  142. package/dist/formula-7ZPC54JO.js.map +0 -1
  143. package/dist/formula-EQRHSMYW.js.map +0 -1
  144. package/dist/formula-IWV5IJ3Y.cjs.map +0 -1
  145. package/dist/image-ocr-6MO4FXJK.js.map +0 -1
  146. package/dist/image-ocr-MGVXT6XD.cjs.map +0 -1
  147. package/dist/image-ocr-TWUU2RNG.js.map +0 -1
  148. package/dist/parser-JFGLN4DA.cjs.map +0 -1
  149. package/dist/parser-VR2BTUT7.js.map +0 -1
  150. package/dist/parser-YA7HSSPK.js.map +0 -1
  151. package/dist/pdf-ocr-4LISPZEM.cjs +0 -13
  152. package/dist/pdf-ocr-4LISPZEM.cjs.map +0 -1
  153. package/dist/pdf-ocr-AIWB2QCS.js +0 -12
  154. package/dist/pdf-ocr-QMOKUMC2.js +0 -15
  155. package/dist/rasterize-AFYGNKXG.js.map +0 -1
  156. package/dist/render-PSPTMZIZ.js +0 -10
  157. package/dist/seal-4XBJ7ISE.js +0 -10
  158. package/dist/setup-I4I75QJU.js.map +0 -1
  159. package/dist/watch-D3JJEOAB.js.map +0 -1
  160. /package/dist/{-RH6BEVTT.js.map → -R6NUMPRV.js.map} +0 -0
  161. /package/dist/{chunks-JEYQWXIZ.js.map → chunks-QMGCI5DG.js.map} +0 -0
  162. /package/dist/{models-6VMAVZR6.js.map → models-6SIKLKRH.js.map} +0 -0
  163. /package/dist/{pdf-ocr-AIWB2QCS.js.map → pdf-ocr-4R646OO6.js.map} +0 -0
  164. /package/dist/{pdf-ocr-QMOKUMC2.js.map → pdf-ocr-WE5TMRFU.js.map} +0 -0
  165. /package/dist/{profile-io-QPIMGVO3.js.map → profile-io-O6BDCUJN.js.map} +0 -0
  166. /package/dist/{redact-W5K3ILQH.js.map → redact-7ILEAUTS.js.map} +0 -0
  167. /package/dist/{render-PSPTMZIZ.js.map → render-3BSVTTZE.js.map} +0 -0
  168. /package/dist/{seal-4XBJ7ISE.js.map → seal-6ZSEJVOD.js.map} +0 -0
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/pdf/image-extract.ts","../src/pdf/quality.ts","../src/pdf/text-clean.ts","../src/pdf/formula-ocr.ts","../src/pdf/polyfill.ts","../src/pdf/parser.ts"],"sourcesContent":["/**\r\n * PDF 이미지 XObject 바이트 추출 — 페이지 operatorList의 paint 이미지를\r\n * page.objs/commonObjs에서 디코딩된 픽셀로 받아 PNG로 인코딩한다.\r\n *\r\n * getOperatorList()가 이미 디코딩 비용을 지불하므로(선 감지용으로 항상 호출)\r\n * 여기서는 RGBA 변환 + PNG deflate만 추가된다. 순수 JS (node:zlib) 전용.\r\n */\r\n\r\nimport { OPS, ImageKind } from \"pdfjs-dist/legacy/build/pdf.mjs\"\r\nimport { encodePng } from \"../image/transcode.js\"\r\nimport type { ExtractedImage, IRBlock, ParseWarning } from \"../types.js\"\r\n\r\n/** 디코딩된 이미지 픽셀 (pdfjs page.objs 반환 형태의 최소 부분) */\r\ninterface PdfImgData {\r\n width: number\r\n height: number\r\n kind?: number\r\n data?: Uint8Array | Uint8ClampedArray\r\n}\r\n\r\n/** 페이지 프록시의 최소 형태 — objs/commonObjs 만 사용 */\r\ninterface PdfObjects {\r\n get(id: string, callback?: (data: unknown) => void): unknown\r\n}\r\ninterface PdfPageObjs {\r\n objs: PdfObjects\r\n commonObjs: PdfObjects\r\n}\r\n\r\n/** 개별 이미지 디코딩 대기 상한 — 실패 시에도 null 로 resolve 되므로 안전망 성격 */\r\nconst IMAGE_RESOLVE_TIMEOUT_MS = 5_000\r\n\r\n/**\r\n * 이미지 객체 대기 해제 — 디코딩은 getOperatorList() resolve 후에도 워커에서\r\n * 비동기로 진행되므로(pdfjs buildPaintImageXObject 는 await 하지 않음) 동기\r\n * has()/get() 은 이미 완료된 것만 잡힌다. 콜백형 get()으로 완료를 기다리며,\r\n * 디코딩 실패는 pdfjs 가 null 로 resolve 하므로 무한 대기는 없다.\r\n */\r\nfunction resolveImgData(page: PdfPageObjs, objId: string): Promise<PdfImgData | null> {\r\n // canvas.js 규약과 동일 — \"g_\" 접두사는 문서 공용 객체\r\n const store = objId.startsWith(\"g_\") ? page.commonObjs : page.objs\r\n return new Promise(resolve => {\r\n let done = false\r\n const timer = setTimeout(() => { if (!done) { done = true; resolve(null) } }, IMAGE_RESOLVE_TIMEOUT_MS)\r\n try {\r\n store.get(objId, (data: unknown) => {\r\n if (!done) { done = true; clearTimeout(timer); resolve((data ?? null) as PdfImgData | null) }\r\n })\r\n } catch {\r\n if (!done) { done = true; clearTimeout(timer); resolve(null) }\r\n }\r\n })\r\n}\r\n\r\n/** 폭·높이 하한 — 괘선/불릿 등 장식 조각 제외 */\r\nconst MIN_DIM = 8\r\n/** 총 픽셀 상한 — transcode.ts MAX_PIXELS와 동일 근거 (deflateSync 동기 블로킹 방지) */\r\nconst MAX_PIXELS = 36_000_000\r\n/** 문서당 추출 이미지 수 상한 */\r\nconst MAX_IMAGES_PER_DOC = 200\r\n/** 문서당 PNG 누적 바이트 상한 (128MB) */\r\nconst MAX_TOTAL_IMAGE_BYTES = 128 * 1024 * 1024\r\n\r\n/** 문서 단위 추출 상태 — 페이지 간 중복(로고·워터마크) 억제 + 상한 추적 */\r\nexport interface PdfImageState {\r\n imageIndex: number\r\n totalBytes: number\r\n /** 내용 해시 → 파일명 (이전 페이지에서 이미 추출된 이미지) */\r\n seen: Map<string, string>\r\n capWarned: boolean\r\n}\r\n\r\nexport function createPdfImageState(): PdfImageState {\r\n return { imageIndex: 0, totalBytes: 0, seen: new Map(), capWarned: false }\r\n}\r\n\r\n/** FNV-1a 32bit — 픽셀 버퍼 내용 해시 (crypto 불필요, 충돌은 dims 결합으로 완화) */\r\nfunction fnv1a(data: Uint8Array | Uint8ClampedArray): number {\r\n let h = 0x811c9dc5\r\n for (let i = 0; i < data.length; i++) {\r\n h ^= data[i]\r\n h = Math.imul(h, 0x01000193)\r\n }\r\n return h >>> 0\r\n}\r\n\r\n/** kind별 픽셀 → 8bit RGBA. 미지원/불일치 형태는 null. */\r\nfunction toRgba(img: PdfImgData): Uint8Array | null {\r\n const { width: w, height: h, kind, data } = img\r\n if (!data || !w || !h) return null\r\n const rgba = new Uint8Array(w * h * 4)\r\n\r\n if (kind === ImageKind.RGBA_32BPP) {\r\n if (data.length < w * h * 4) return null\r\n rgba.set(data.subarray(0, w * h * 4))\r\n return rgba\r\n }\r\n if (kind === ImageKind.RGB_24BPP) {\r\n if (data.length < w * h * 3) return null\r\n for (let i = 0, s = 0, d = 0; i < w * h; i++, s += 3, d += 4) {\r\n rgba[d] = data[s]; rgba[d + 1] = data[s + 1]; rgba[d + 2] = data[s + 2]; rgba[d + 3] = 255\r\n }\r\n return rgba\r\n }\r\n if (kind === ImageKind.GRAYSCALE_1BPP) {\r\n // 행 단위 비트 패킹 (stride = ceil(w/8)), 1=백 0=흑\r\n const stride = (w + 7) >> 3\r\n if (data.length < stride * h) return null\r\n for (let y = 0; y < h; y++) {\r\n for (let x = 0; x < w; x++) {\r\n const bit = (data[y * stride + (x >> 3)] >> (7 - (x & 7))) & 1\r\n const v = bit ? 255 : 0\r\n const d = (y * w + x) * 4\r\n rgba[d] = v; rgba[d + 1] = v; rgba[d + 2] = v; rgba[d + 3] = 255\r\n }\r\n }\r\n return rgba\r\n }\r\n return null\r\n}\r\n\r\n/**\r\n * 한 페이지의 operatorList에서 이미지를 추출해 image 블록·바이너리로 반환.\r\n * 같은 페이지 내 반복 paint(타일링)는 1회만, 이전 페이지와 동일 내용(로고·\r\n * 워터마크)은 블록·바이너리 모두 재방출하지 않는다.\r\n */\r\nexport async function extractPageImages(\r\n page: PdfPageObjs,\r\n fnArray: Uint32Array | number[],\r\n argsArray: unknown[][],\r\n pageNumber: number,\r\n state: PdfImageState,\r\n warnings: ParseWarning[],\r\n): Promise<{ blocks: IRBlock[]; images: ExtractedImage[] }> {\r\n const blocks: IRBlock[] = []\r\n const images: ExtractedImage[] = []\r\n const pageSeenIds = new Set<string>()\r\n\r\n for (let i = 0; i < fnArray.length; i++) {\r\n const op = fnArray[i]\r\n let imgData: PdfImgData | null = null\r\n let dedupeId: string | null = null\r\n\r\n if (op === OPS.paintImageXObject || op === OPS.paintImageXObjectRepeat) {\r\n const objId = argsArray[i]?.[0]\r\n if (typeof objId !== \"string\" || pageSeenIds.has(objId)) continue\r\n pageSeenIds.add(objId)\r\n dedupeId = objId\r\n } else if (op !== OPS.paintInlineImageXObject) {\r\n continue\r\n }\r\n\r\n // 상한 도달 검사 — resolveImgData(이미지당 최대 5초 대기) 앞에서 중단해야\r\n // 상한 이후 이미지들이 페이지마다 디코딩 대기 시간만 소모하지 않는다\r\n if (state.imageIndex >= MAX_IMAGES_PER_DOC || state.totalBytes >= MAX_TOTAL_IMAGE_BYTES) {\r\n if (!state.capWarned) {\r\n state.capWarned = true\r\n warnings.push({ page: pageNumber, message: `이미지 추출 상한 도달 (${MAX_IMAGES_PER_DOC}개/128MB) — 이후 이미지 생략`, code: \"SKIPPED_IMAGE\" })\r\n }\r\n continue\r\n }\r\n\r\n if (dedupeId) {\r\n imgData = await resolveImgData(page, dedupeId)\r\n } else {\r\n imgData = argsArray[i]?.[0] as PdfImgData\r\n }\r\n\r\n if (!imgData?.data || !imgData.width || !imgData.height) continue\r\n const { width: w, height: h } = imgData\r\n if (w < MIN_DIM || h < MIN_DIM) continue // 괘선/불릿 장식\r\n if (w * h > MAX_PIXELS) {\r\n warnings.push({ page: pageNumber, message: `이미지 크기 초과로 추출 생략 (${w}×${h})`, code: \"SKIPPED_IMAGE\" })\r\n continue\r\n }\r\n\r\n // 페이지 간 내용 중복 — 로고·워터마크 재추출 방지\r\n const hash = `${w}x${h}k${imgData.kind ?? \"?\"}h${fnv1a(imgData.data)}${dedupeId ? \"\" : \"inline\"}`\r\n if (state.seen.has(hash)) continue\r\n\r\n const rgba = toRgba(imgData)\r\n if (!rgba) continue // 미지원 픽셀 형태 (bitmap 전용 등)\r\n\r\n const png = encodePng(w, h, rgba)\r\n state.imageIndex++\r\n state.totalBytes += png.length\r\n const filename = `image_${String(state.imageIndex).padStart(3, \"0\")}.png`\r\n state.seen.set(hash, filename)\r\n images.push({ filename, data: png, mimeType: \"image/png\" })\r\n blocks.push({ type: \"image\", text: filename, pageNumber })\r\n }\r\n\r\n return { blocks, images }\r\n}\r\n\r\n/**\r\n * 페이지별 image 블록을 본문 블록 열에 주입 — 각 페이지의 마지막 블록 뒤.\r\n * 페이지 루프 중 바로 끼워 넣으면 페이지 경계에서 표 블록 인접성이 깨져\r\n * mergeCrossPageTables가 무산되므로, 병합이 끝난 뒤 호출해야 한다.\r\n */\r\nexport function injectPageImageBlocks(blocks: IRBlock[], pageImages: Map<number, IRBlock[]>): void {\r\n if (pageImages.size === 0) return\r\n\r\n // 페이지별 마지막 블록 위치 → 위치별 페이지 역인덱스\r\n const lastIndexForPage = new Map<number, number>()\r\n for (let i = 0; i < blocks.length; i++) {\r\n const p = blocks[i].pageNumber\r\n if (p !== undefined) lastIndexForPage.set(p, i)\r\n }\r\n const pageAtIndex = new Map<number, number>()\r\n for (const [p, last] of lastIndexForPage) pageAtIndex.set(last, p)\r\n\r\n const result: IRBlock[] = []\r\n const injected = new Set<number>()\r\n for (let i = 0; i < blocks.length; i++) {\r\n result.push(blocks[i])\r\n const p = pageAtIndex.get(i)\r\n if (p !== undefined && pageImages.has(p)) {\r\n result.push(...pageImages.get(p)!)\r\n injected.add(p)\r\n }\r\n }\r\n // 텍스트 블록이 하나도 없는 페이지(전면 이미지 등) — 페이지 순서 위치에 삽입\r\n for (const p of [...pageImages.keys()].sort((a, b) => a - b)) {\r\n if (injected.has(p)) continue\r\n let at = result.length\r\n for (let i = 0; i < result.length; i++) {\r\n const bp = result[i].pageNumber\r\n if (bp !== undefined && bp > p) { at = i; break }\r\n }\r\n result.splice(at, 0, ...pageImages.get(p)!)\r\n }\r\n\r\n blocks.length = 0\r\n blocks.push(...result)\r\n}\r\n","/**\r\n * PDF 페이지별 텍스트 품질 신호 수집.\r\n *\r\n * pdfjs가 텍스트층을 추출했더라도, 폰트의 ToUnicode/CMap이 불완전하면\r\n * 한글이 PUA(Private Use Area) 글리프 코드로 그대로 떨어진다. 또 일부 PDF는\r\n * NUL/제어문자가 본문에 섞여 있다. 본 모듈은 그런 신호를 페이지 단위로 수집해\r\n * \"OCR 검토 필요\" 판정에 사용한다.\r\n *\r\n * 더 까다로운 케이스: ToUnicode가 \"잘못\" 매핑되면 글리프가 PUA/FFFD가 아니라\r\n * 정상 한글 음절 영역(0xAC00-0xD7A3)의 엉뚱한 글자로 떨어진다(\"GPU쭒컫 많핂슪\").\r\n * 이건 PUA/제어/대체문자 신호로 못 잡는다. 대신 자소를 분해해 종성(받침) 분포를\r\n * 본다 — CID 스크램블은 받침을 균등하게 흩뿌리므로, 자연 한국어(받침없음 다수 +\r\n * 겹받침 희소)와 통계적으로 갈린다. 이것이 garbled_hangul 신호다.\r\n */\r\n\r\nexport interface PageQuality {\r\n /** 1-based 페이지 번호 */\r\n page: number\r\n /** 공백 제외 문자 수 */\r\n textChars: number\r\n /** 한글 음절(0xAC00-0xD7A3) 비율 (0~1) */\r\n hangulRatio: number\r\n /** C0/C1 제어문자 비율 (tab/newline/CR 제외) */\r\n controlCharRatio: number\r\n /** U+FFFD replacement character 비율 */\r\n replacementCharRatio: number\r\n /** PUA 비율 — 글꼴 매핑 실패의 핵심 신호 */\r\n puaRatio: number\r\n /** 한글 음절 중 받침 없는 음절 비율 (자연 한국어 ~0.5, mojibake ~0.04) */\r\n hangulNoBatchimRatio: number\r\n /** 한글 음절 중 겹받침/희귀 받침 비율 (자연 한국어 <0.06, mojibake ~0.5) */\r\n hangulRareBatchimRatio: number\r\n /** OCR 검토 권장 여부 (pdfjs 텍스트층 기준 원신호 — OCR 적용 후에도 보존) */\r\n needsOcr: boolean\r\n /** needsOcr=true일 때 사유 (단일 신호로 충분, 가장 강한 신호 선택) */\r\n ocrReason?: \"low_text\" | \"high_pua\" | \"high_control\" | \"high_replacement\" | \"garbled_hangul\"\r\n /** 이 페이지에 OCR 이 실제 적용되어 본문이 OCR 결과로 대체됨 */\r\n ocrApplied?: boolean\r\n}\r\n\r\n/**\r\n * 희귀 종성(받침) 인덱스 집합. 종성 인덱스 = (code - 0xAC00) % 28,\r\n * 0 = 받침 없음. 겹받침(ㄳㄵㄶㄺㄻㄼㄽㄾㄿㅀㅄ) + 거의 안 쓰는 홑받침(ㅋㅌㅍ).\r\n * 자연 한국어에서 이들의 합산 비율은 매우 낮다(<6%).\r\n */\r\nconst RARE_JONG = new Set<number>([3, 5, 6, 9, 10, 11, 12, 13, 14, 15, 18, 24, 25, 26])\r\n\r\n/** garbled_hangul 판정 최소 한글 음절 수 — 통계적 유의성 확보 */\r\nconst MOJIBAKE_MIN_HANGUL = 30\r\n/** 받침 없는 음절 비율이 이 값 미만이면 비정상 (자연 한국어는 훨씬 높음) */\r\nconst MOJIBAKE_MAX_NOBATCHIM = 0.15\r\n/** 희귀 받침 비율이 이 값 이상이면 비정상 (자연 한국어는 훨씬 낮음) */\r\nconst MOJIBAKE_MIN_RAREBATCHIM = 0.25\r\n\r\n/** 페이지 텍스트에서 품질 메트릭을 계산한다. */\r\nexport function computePageQuality(page: number, text: string): PageQuality {\r\n let total = 0\r\n let hangul = 0\r\n let hangulNoBatchim = 0\r\n let hangulRareBatchim = 0\r\n let control = 0\r\n let replacement = 0\r\n let pua = 0\r\n\r\n for (let i = 0; i < text.length; i++) {\r\n const code = text.charCodeAt(i)\r\n // 공백류는 분모에서 제외\r\n if (code === 0x20 || code === 0x09 || code === 0x0a || code === 0x0d) continue\r\n total++\r\n\r\n // C0 제어문자 (0x00-0x1F 중 tab/lf/cr 제외) + DEL(0x7F) + C1(0x80-0x9F)\r\n if ((code < 0x20) || code === 0x7f || (code >= 0x80 && code <= 0x9f)) {\r\n control++\r\n continue\r\n }\r\n if (code === 0xfffd) {\r\n replacement++\r\n continue\r\n }\r\n // 한글 음절 — 종성(받침) 분포로 mojibake 판정\r\n if (code >= 0xac00 && code <= 0xd7a3) {\r\n hangul++\r\n const jong = (code - 0xac00) % 28\r\n if (jong === 0) hangulNoBatchim++\r\n else if (RARE_JONG.has(jong)) hangulRareBatchim++\r\n continue\r\n }\r\n // PUA: BMP(E000-F8FF) + SPUA-A(F0000-FFFFD) + SPUA-B(100000-10FFFD)\r\n // 서로게이트 페어는 high만 검사해도 충분 (Plane 15/16 high surrogate 범위 DB80-DBFF)\r\n if ((code >= 0xe000 && code <= 0xf8ff) || (code >= 0xdb80 && code <= 0xdbff)) {\r\n pua++\r\n continue\r\n }\r\n }\r\n\r\n const denom = total || 1\r\n const puaRatio = pua / denom\r\n const controlCharRatio = control / denom\r\n const replacementCharRatio = replacement / denom\r\n const hangulNoBatchimRatio = hangul > 0 ? hangulNoBatchim / hangul : 0\r\n const hangulRareBatchimRatio = hangul > 0 ? hangulRareBatchim / hangul : 0\r\n\r\n // 받침 분포 이상 = ToUnicode 오매핑 mojibake. 받침없음 희소 + 희귀받침 과다를\r\n // 둘 다(AND) 만족해야 발화 — 자연 한국어(받침없음 다수)에서 오탐 방지.\r\n const garbledHangul =\r\n hangul >= MOJIBAKE_MIN_HANGUL &&\r\n hangulNoBatchimRatio < MOJIBAKE_MAX_NOBATCHIM &&\r\n hangulRareBatchimRatio >= MOJIBAKE_MIN_RAREBATCHIM\r\n\r\n let needsOcr = false\r\n let ocrReason: PageQuality[\"ocrReason\"] | undefined\r\n // 우선순위: low_text > high_pua > high_control > high_replacement > garbled_hangul\r\n if (total < LOW_TEXT_THRESHOLD) { needsOcr = true; ocrReason = \"low_text\" }\r\n else if (puaRatio >= HIGH_PUA_THRESHOLD) { needsOcr = true; ocrReason = \"high_pua\" }\r\n else if (controlCharRatio >= HIGH_CONTROL_THRESHOLD) { needsOcr = true; ocrReason = \"high_control\" }\r\n else if (replacementCharRatio >= HIGH_REPLACEMENT_THRESHOLD) { needsOcr = true; ocrReason = \"high_replacement\" }\r\n else if (garbledHangul) { needsOcr = true; ocrReason = \"garbled_hangul\" }\r\n\r\n return {\r\n page,\r\n textChars: total,\r\n hangulRatio: hangul / denom,\r\n controlCharRatio,\r\n replacementCharRatio,\r\n puaRatio,\r\n hangulNoBatchimRatio,\r\n hangulRareBatchimRatio,\r\n needsOcr,\r\n ocrReason,\r\n }\r\n}\r\n\r\n/** 페이지별 품질에서 문서 단위 요약을 계산한다. */\r\nexport interface DocumentQualitySummary {\r\n totalPages: number\r\n totalTextChars: number\r\n avgHangulRatio: number\r\n avgControlCharRatio: number\r\n avgReplacementCharRatio: number\r\n avgPuaRatio: number\r\n /** textChars 매우 적은 페이지 수 (이미지/빈 페이지 후보) */\r\n lowTextPageCount: number\r\n /** PUA 비율 임계 초과 페이지 수 (글꼴 매핑 깨짐 후보) */\r\n highPuaPageCount: number\r\n /** 문서 전체에 OCR 검토가 권장되는지 — needsOcr 페이지 비율 또는 평균 신호 기반 */\r\n needsOcr: boolean\r\n /** needsOcr=true인 페이지 번호 목록 */\r\n ocrCandidatePages: number[]\r\n}\r\n\r\nconst LOW_TEXT_THRESHOLD = 20\r\nconst HIGH_PUA_THRESHOLD = 0.2\r\nconst HIGH_CONTROL_THRESHOLD = 0.05\r\nconst HIGH_REPLACEMENT_THRESHOLD = 0.05\r\n/** 페이지 중 needsOcr 비율이 이 값 이상이면 문서 전체에 OCR 권장 */\r\nconst DOC_NEEDS_OCR_PAGE_RATIO = 0.3\r\n\r\n/**\r\n * 비표시 제어문자를 제거한다. C0(NUL 등, tab/lf/cr 제외) + DEL + C1.\r\n * PUA는 사용자가 글꼴 매핑 실패를 시각적으로 확인할 수 있도록 보존.\r\n */\r\nexport function stripControlChars(text: string): string {\r\n // eslint-disable-next-line no-control-regex\r\n return text.replace(/[\\x00-\\x08\\x0B\\x0C\\x0E-\\x1F\\x7F\\x80-\\x9F]/g, \"\")\r\n}\r\n\r\nexport function summarizeDocumentQuality(pages: PageQuality[]): DocumentQualitySummary {\r\n if (pages.length === 0) {\r\n return {\r\n totalPages: 0,\r\n totalTextChars: 0,\r\n avgHangulRatio: 0,\r\n avgControlCharRatio: 0,\r\n avgReplacementCharRatio: 0,\r\n avgPuaRatio: 0,\r\n lowTextPageCount: 0,\r\n highPuaPageCount: 0,\r\n needsOcr: false,\r\n ocrCandidatePages: [],\r\n }\r\n }\r\n\r\n let textChars = 0\r\n let hangul = 0\r\n let control = 0\r\n let replacement = 0\r\n let pua = 0\r\n let lowText = 0\r\n let highPua = 0\r\n const ocrCandidatePages: number[] = []\r\n\r\n for (const p of pages) {\r\n textChars += p.textChars\r\n hangul += p.hangulRatio\r\n control += p.controlCharRatio\r\n replacement += p.replacementCharRatio\r\n pua += p.puaRatio\r\n if (p.textChars < LOW_TEXT_THRESHOLD) lowText++\r\n if (p.puaRatio >= HIGH_PUA_THRESHOLD) highPua++\r\n // OCR 이 이미 적용된 페이지는 후보에서 제외 — \"OCR 했는데 또 하라\" 자기모순 방지\r\n if (p.needsOcr && !p.ocrApplied) ocrCandidatePages.push(p.page)\r\n }\r\n\r\n const n = pages.length\r\n return {\r\n totalPages: n,\r\n totalTextChars: textChars,\r\n avgHangulRatio: hangul / n,\r\n avgControlCharRatio: control / n,\r\n avgReplacementCharRatio: replacement / n,\r\n avgPuaRatio: pua / n,\r\n lowTextPageCount: lowText,\r\n highPuaPageCount: highPua,\r\n needsOcr: ocrCandidatePages.length / n >= DOC_NEEDS_OCR_PAGE_RATIO,\r\n ocrCandidatePages,\r\n }\r\n}\r\n","/**\r\n * PDF 마크다운 최종 정리.\r\n *\r\n * 페이지 번호 제거, 균등배분 후처리, 취소선 복원, 한글 줄바꿈 병합.\r\n * blocksToMarkdown 이후의 문자열 수준 후처리를 담당한다.\r\n */\r\n\r\nimport type { IRBlock } from \"../types.js\"\r\nimport { stripControlChars } from \"./quality.js\"\r\nimport { collapseEvenSpacing } from \"./text-line.js\"\r\n\r\n/** 블록 트리의 텍스트에서 비표시 제어문자를 in-place로 제거한다. */\r\nexport function sanitizeBlockControlChars(blocks: IRBlock[]): void {\r\n for (const b of blocks) {\r\n if (b.text) b.text = stripControlChars(b.text)\r\n if (b.table) {\r\n for (const row of b.table.cells) {\r\n for (const cell of row) {\r\n if (cell.text) cell.text = stripControlChars(cell.text)\r\n }\r\n }\r\n }\r\n if (b.children) sanitizeBlockControlChars(b.children)\r\n }\r\n}\r\n\r\nexport function cleanPdfText(text: string): string {\r\n return mergeKoreanLines(\r\n stripControlChars(text)\r\n // 문서 시작 단독 페이지 번호\r\n .replace(/^\\d{1,4}\\n/, \"\")\r\n // \"- 2 -\" 스타일 페이지 번호 (독립 라인 및 목록 항목 형태 포함)\r\n .replace(/^[\\s]*[-–—]\\s*[-–—]?\\d+[-–—]?[\\s]*[-–—]?[\\s]*$/gm, \"\")\r\n // \"1 / 5\" 스타일 페이지 번호\r\n .replace(/^\\s*\\d+\\s*\\/\\s*\\d+\\s*$/gm, \"\")\r\n // 단독 페이지 번호 (줄 끝에 혼자 있는 숫자)\r\n .replace(/\\n\\d{1,4}\\n/g, \"\\n\")\r\n // 문서 마지막 단독 페이지 번호\r\n .replace(/\\n\\d{1,4}$/, \"\")\r\n // 단독 숫자 헤딩 제거 (\"# 6\\n재무과\" → \"\\n재무과\")\r\n .replace(/^#{1,6}\\s*\\d{1,4}\\s*$/gm, \"\")\r\n )\r\n // 균등배분 문자열 후처리 (pdfjs가 합친 TextItem + buildGridTable 셀 텍스트)\r\n // LaTeX 수식 라인 ($...$ / $$...$$) 은 공백이 토큰 구분자라 collapse 시 `\\cdot d` → `\\cdotd` 로 망가짐 — skip\r\n .replace(/^(?!\\| ---).*$/gm, line => {\r\n if (/^\\s*\\${1,2}.+\\${1,2}\\s*$/.test(line)) return line\r\n return collapseEvenSpacing(line)\r\n })\r\n // 마커 뒤 2글자 균등배분 합침 (\"□ 일 시\" → \"□ 일시\", \"□ 장 소\" → \"□ 장소\")\r\n .replace(/([□■◆○●▶ㅇ])\\s+([가-힣])\\s+([가-힣])/g, \"$1 $2$3\")\r\n // 취소선 복원: builder escapeGfm이 ~를 \\~로 이스케이프 — 쌍(~~)만 되살림\r\n .replace(/\\\\~\\\\~/g, \"~~\")\r\n // 인접 취소선 run이 붙어 생긴 빈 마크(~~~~) 정리\r\n .replace(/~~~~/g, \"\")\r\n .replace(/\\n{3,}/g, \"\\n\\n\")\r\n .trim()\r\n}\r\n\r\nfunction startsWithMarker(line: string): boolean {\r\n const t = line.trimStart()\r\n return /^[가-힣ㄱ-ㅎ][.)]/.test(t) || /^\\d+[.)]/.test(t) || /^\\([가-힣ㄱ-ㅎ\\d]+\\)/.test(t) ||\r\n /^[○●※▶▷◆◇■□★☆\\-·]\\s/.test(t) || /^제\\d+[조항호장절]/.test(t)\r\n}\r\n\r\nfunction isStandaloneHeader(line: string): boolean {\r\n return /^제\\d+[조항호장절](\\([^)]*\\))?(\\s+\\S+){0,7}$/.test(line.trim())\r\n}\r\n\r\nfunction mergeKoreanLines(text: string): string {\r\n if (!text) return \"\"\r\n const lines = text.split(\"\\n\")\r\n if (lines.length <= 1) return text\r\n const result: string[] = [lines[0]]\r\n\r\n for (let i = 1; i < lines.length; i++) {\r\n const prev = result[result.length - 1]\r\n const curr = lines[i]\r\n const currTrimmed = curr.trim()\r\n // 마크다운 헤딩/테이블/구분선은 병합하지 않음\r\n if (/^#{1,6}\\s/.test(prev) || /^#{1,6}\\s/.test(curr) || /^\\|/.test(currTrimmed) || /^---/.test(currTrimmed)) {\r\n result.push(curr)\r\n continue\r\n }\r\n // 쉼표로 끝나는 줄 + 다음 줄 = 연속 문장\r\n if (/,$/.test(prev.trim()) && currTrimmed.length > 0) {\r\n result[result.length - 1] = prev + \"\\n\" + curr\r\n continue\r\n }\r\n // (※ 로 시작하는 줄 = 이전 줄의 부연설명\r\n if (/^\\(※/.test(currTrimmed)) {\r\n result[result.length - 1] = prev + \" \" + currTrimmed\r\n continue\r\n }\r\n // 한글 줄바꿈 병합 — 마커(○, □ 등)로 시작하는 이전 줄은 합치지 않음\r\n if (/[가-힣·,\\-]$/.test(prev) && /^[가-힣(]/.test(curr) &&\r\n !startsWithMarker(curr) && !isStandaloneHeader(prev) &&\r\n !startsWithMarker(prev)) {\r\n result[result.length - 1] = prev + \" \" + curr\r\n } else {\r\n result.push(curr)\r\n }\r\n }\r\n return result.join(\"\\n\")\r\n}\r\n","/**\r\n * 수식 OCR 통합 (optional)\r\n *\r\n * 페이지 이미지 렌더 후 수식만 검출/인식해 blocks 에 formula paragraph 를 삽입.\r\n * 모델은 최초 실행 시 자동 다운로드 (./formula/index.js 에 위임).\r\n */\r\n\r\nimport type { IRBlock, ParseWarning } from \"../types.js\"\r\n\r\n/**\r\n * 수식 OCR 을 적용하여 blocks 에 formula paragraph 를 삽입한다.\r\n *\r\n * 좌표 매핑:\r\n * - pdfium 픽셀 bbox (top-left origin) → PDF 포인트 (bottom-left origin) 변환\r\n * - 수식 bbox 의 y center 와 같은 페이지 내 pdfjs block 의 y center 비교로 삽입 위치 결정\r\n * - pdfjs 가 이미 뽑은 수식 흔적(block) 과 겹치면 해당 block 제거 (중복 방지)\r\n *\r\n * 실패/trivial 수식(latex === \"\") 은 삽입하지 않는다.\r\n */\r\nexport async function applyFormulaOcr(\r\n buffer: ArrayBuffer,\r\n blocks: IRBlock[],\r\n pageFilter: Set<number> | null,\r\n effectivePageCount: number,\r\n warnings: ParseWarning[],\r\n _onProgress?: (current: number, total: number) => void,\r\n): Promise<void> {\r\n const formulaMod = await import(\"./formula/index.js\")\r\n const { FormulaPipeline, ensureFormulaModels } = formulaMod\r\n\r\n // 모델 준비 — 없으면 자동 다운로드. 진행률은 stderr 로 출력.\r\n await ensureFormulaModels((p) => {\r\n if (p.phase === \"download\" && p.total) {\r\n const pct = Math.floor((p.downloaded / p.total) * 100)\r\n process.stderr.write(`\\r[kordoc-formula] ${p.spec.name} ${pct}% (${formatMb(p.downloaded)}/${formatMb(p.total)})`)\r\n if (p.downloaded >= p.total) process.stderr.write(\"\\n\")\r\n } else if (p.phase === \"verify\") {\r\n process.stderr.write(`[kordoc-formula] ${p.spec.name} SHA-256 검증 중...\\n`)\r\n } else if (p.phase === \"done\") {\r\n process.stderr.write(`[kordoc-formula] ${p.spec.name} 준비 완료\\n`)\r\n } else if (p.phase === \"skip\") {\r\n // 조용히 스킵\r\n }\r\n })\r\n\r\n const pipeline = await FormulaPipeline.create()\r\n try {\r\n // MAX_PAGES 상한을 수식 OCR 에도 적용 — 필터 없으면 텍스트 추출과 같은 범위만\r\n // (없으면 대형 PDF 에서 상한 밖 페이지까지 ONNX 추론 + 결과가 문서 끝에 오배치)\r\n const effectiveFilter = pageFilter\r\n ?? new Set(Array.from({ length: effectivePageCount }, (_, i) => i + 1))\r\n const pagesResult = await pipeline.runOnBuffer(buffer, effectiveFilter)\r\n\r\n if (pagesResult.length === 0) return\r\n\r\n let insertedCount = 0\r\n let removedDupCount = 0\r\n\r\n for (const page of pagesResult) {\r\n const pageNumber = page.pageNumber\r\n const pdfHeight = page.pdfHeight\r\n const scaleX = page.renderedWidth > 0 ? page.pdfWidth / page.renderedWidth : 0.5\r\n const scaleY = page.renderedHeight > 0 ? page.pdfHeight / page.renderedHeight : 0.5\r\n\r\n // 1) 수식 → (PDF 포인트 bbox, latex) 정규화 + trivial 제외\r\n interface FormulaCandidate {\r\n block: IRBlock\r\n pdfBbox: { x1: number; x2: number; yTop: number; yBottom: number }\r\n centerY: number // PDF bottom-up\r\n }\r\n const candidates: FormulaCandidate[] = []\r\n for (const r of page.regions) {\r\n if (!r.latex || !r.latex.trim()) continue\r\n const wrapped = r.kind === \"display\" ? `$$${r.latex}$$` : `$${r.latex}$`\r\n\r\n const x1 = r.bbox.x1 * scaleX\r\n const x2 = r.bbox.x2 * scaleX\r\n // pdfium 픽셀 y → PDF bottom-up\r\n const yTop = pdfHeight - r.bbox.y1 * scaleY\r\n const yBottom = pdfHeight - r.bbox.y2 * scaleY\r\n const centerY = (yTop + yBottom) / 2\r\n const width = x2 - x1\r\n const height = yTop - yBottom\r\n\r\n candidates.push({\r\n block: {\r\n type: \"paragraph\",\r\n text: wrapped,\r\n pageNumber,\r\n bbox: { page: pageNumber, x: x1, y: yBottom, width, height },\r\n },\r\n pdfBbox: { x1, x2, yTop, yBottom },\r\n centerY,\r\n })\r\n }\r\n if (candidates.length === 0) continue\r\n\r\n // 2) 같은 페이지의 pdfjs block 중 수식 bbox 와 크게 겹치는 것 제거\r\n // (pdfjs 가 수식을 텍스트로 파편 추출한 경우 — overlap ratio ≥ 0.6 이면 중복으로 간주)\r\n const OVERLAP_THRESHOLD = 0.6\r\n const indicesToRemove = new Set<number>()\r\n for (let i = 0; i < blocks.length; i++) {\r\n const b = blocks[i]\r\n if (b.pageNumber !== pageNumber) continue\r\n if (b.type === \"table\") continue // 표는 건드리지 않음\r\n if (!b.bbox || b.bbox.width <= 0 || b.bbox.height <= 0) continue\r\n const blockArea = b.bbox.width * b.bbox.height\r\n if (blockArea <= 0) continue\r\n\r\n for (const c of candidates) {\r\n const ox1 = Math.max(b.bbox.x, c.pdfBbox.x1)\r\n const ox2 = Math.min(b.bbox.x + b.bbox.width, c.pdfBbox.x2)\r\n const oy1 = Math.max(b.bbox.y, c.pdfBbox.yBottom)\r\n const oy2 = Math.min(b.bbox.y + b.bbox.height, c.pdfBbox.yTop)\r\n const interArea = Math.max(0, ox2 - ox1) * Math.max(0, oy2 - oy1)\r\n if (interArea / blockArea >= OVERLAP_THRESHOLD) {\r\n indicesToRemove.add(i)\r\n break\r\n }\r\n }\r\n }\r\n\r\n if (indicesToRemove.size > 0) {\r\n // 내림차순으로 제거해야 인덱스가 밀리지 않음\r\n const sorted = [...indicesToRemove].sort((a, b) => b - a)\r\n for (const idx of sorted) blocks.splice(idx, 1)\r\n removedDupCount += indicesToRemove.size\r\n }\r\n\r\n // 3) 각 수식을 y 좌표 기준 적절한 위치에 삽입\r\n // 수식들을 위→아래(centerY 큰 것부터) 정렬 후, 각 수식마다 현재 blocks 에서\r\n // \"center y < 수식 center y\" 인 첫 블록(= 수식보다 아래) 앞에 삽입.\r\n candidates.sort((a, b) => b.centerY - a.centerY)\r\n\r\n for (const c of candidates) {\r\n let insertIdx = -1\r\n let pageFirstIdx = -1\r\n let pageLastIdx = -1\r\n for (let i = 0; i < blocks.length; i++) {\r\n const b = blocks[i]\r\n if (b.pageNumber !== pageNumber) continue\r\n if (pageFirstIdx === -1) pageFirstIdx = i\r\n pageLastIdx = i\r\n if (!b.bbox) continue\r\n const blockCenter = b.bbox.y + b.bbox.height / 2\r\n if (blockCenter < c.centerY) {\r\n insertIdx = i\r\n break\r\n }\r\n }\r\n\r\n if (insertIdx !== -1) {\r\n blocks.splice(insertIdx, 0, c.block)\r\n } else if (pageLastIdx !== -1) {\r\n blocks.splice(pageLastIdx + 1, 0, c.block)\r\n } else {\r\n // 해당 페이지에 텍스트 블록 없음 — 맨 끝에 추가\r\n blocks.push(c.block)\r\n }\r\n insertedCount++\r\n }\r\n }\r\n\r\n if (insertedCount > 0 || removedDupCount > 0) {\r\n process.stderr.write(\r\n `[kordoc-formula] ${insertedCount}개 수식 삽입, ${removedDupCount}개 중복 block 제거 (${pagesResult.length}개 페이지)\\n`,\r\n )\r\n }\r\n } finally {\r\n await pipeline.destroy().catch(() => {})\r\n }\r\n}\r\n\r\nfunction formatMb(bytes: number): string {\r\n return `${(bytes / 1024 / 1024).toFixed(1)}MB`\r\n}\r\n","/**\n * 서버리스/Node.js 환경에서 pdfjs-dist가 요구하는 브라우저 API polyfill.\n * pdfjs-dist import보다 먼저 실행되어야 함 (ES 모듈 호이스팅 대응으로 별도 파일 분리).\n *\n * 1. DOMMatrix / Path2D polyfill — pdfjs-dist가 참조하지만 Node.js에 없음\n * 2. pdfjsWorker 사전 주입 — fake worker의 동적 import를 우회\n */\n\n// eslint-disable-next-line @typescript-eslint/no-explicit-any\nconst g = globalThis as any\n\nif (typeof g.DOMMatrix === \"undefined\") {\n g.DOMMatrix = class DOMMatrix {\n m: number[] = [1, 0, 0, 1, 0, 0]\n constructor(init?: number[]) { if (init) this.m = init }\n }\n}\n\nif (typeof g.Path2D === \"undefined\") {\n g.Path2D = class Path2D {}\n}\n\n// worker 모듈 static import → fake worker가 동적 import를 건너뜀\n// @ts-expect-error pdfjs-dist worker has no type declarations\nimport * as pdfjsWorker from \"pdfjs-dist/legacy/build/pdf.worker.mjs\"\ng.pdfjsWorker = pdfjsWorker\n","/**\r\n * PDF 텍스트 추출 (pdfjs-dist static import 기반)\r\n *\r\n * polyfill을 먼저 import해야 DOMMatrix/Path2D/pdfjsWorker가 주입됨.\r\n * ES 모듈 호이스팅 때문에 별도 파일로 분리되어 있음.\r\n *\r\n * 이 파일은 엔트리(파이프라인 오케스트레이션)와 메타데이터 추출만 담당한다.\r\n * 세부 단계는 모듈로 분리: text-line(아이템/줄), xy-cut(읽기 순서),\r\n * columns(열 감지), page-blocks(블록 추출), block-detect(후처리 감지),\r\n * text-clean(마크다운 정리), formula-ocr(수식).\r\n */\r\n\r\nimport type { InternalParseResult, IRBlock, DocumentMetadata, ExtractedImage, ParseOptions, ParseWarning, OutlineItem } from \"../types.js\"\r\nimport { KordocError } from \"../utils.js\"\r\nimport { parsePageRange } from \"../page-range.js\"\r\nimport { blocksToMarkdown } from \"../table/builder.js\"\r\nimport { extractImageRegions } from \"./line-detector.js\"\r\nimport { createPdfImageState, extractPageImages, injectPageImageBlocks } from \"./image-extract.js\"\r\nimport { computePageQuality, summarizeDocumentQuality, type PageQuality } from \"./quality.js\"\r\nimport { type PdfTextItem, normalizeItems, filterHiddenText } from \"./text-line.js\"\r\nimport { extractPageBlocksWithLines, mergeCrossPageTables } from \"./page-blocks.js\"\r\nimport { computeMedianFontSizeFromFreq, detectHeadings, detectMarkerHeadings, detectTableCaptions, detectKoreanListBlocks, removeHeaderFooterBlocks } from \"./block-detect.js\"\r\nimport { sanitizeBlockControlChars, cleanPdfText } from \"./text-clean.js\"\r\nimport { applyFormulaOcr } from \"./formula-ocr.js\"\r\n// polyfill 먼저 (ES 모듈 호이스팅되므로 별도 파일 필수)\r\nimport \"./polyfill.js\"\r\nimport { getDocument, GlobalWorkerOptions } from \"pdfjs-dist/legacy/build/pdf.mjs\"\r\nimport { createRequire } from \"node:module\"\r\nimport { dirname, join } from \"node:path\"\r\n\r\n// 기존 공개 API 경로 유지 — 이동된 함수의 re-export\r\nexport { mergeCrossPageTables }\r\nexport { cleanPdfText }\r\nexport { detectTableCaptions, detectKoreanListBlocks, removeHeaderFooterBlocks }\r\n\r\n// worker 비활성화 (polyfill에서 pdfjsWorker를 이미 주입했으므로)\r\nGlobalWorkerOptions.workerSrc = \"\"\r\n\r\n// ─── 안전 한계값 (구조적 파싱과 무관) ────────────────\r\nconst MAX_PAGES = 5000\r\nconst MAX_TOTAL_TEXT = 100 * 1024 * 1024 // 100MB\r\n/** PDF 로딩 타임아웃 (30초) — 악성/대용량 PDF 무한 대기 방지 */\r\nconst PDF_LOAD_TIMEOUT_MS = 30_000\r\n\r\n// CID 폰트 자산(cmaps/standard_fonts) 경로 — 미지정이면 CMap 필요 폰트의 텍스트가\r\n// \"loadFont failed: cMapUrl 필요\" 경고와 함께 통째로 소실된다 (성과계획서류 숫자 전멸).\r\n// pdfjs-dist 패키지 위치에서 해석하고, 실패 시 미지정(기존 동작) 유지.\r\nconst pdfjsAssets: { cMapUrl?: string; cMapPacked?: boolean; standardFontDataUrl?: string } = {}\r\ntry {\r\n const _require = createRequire(import.meta.url)\r\n const pkgDir = dirname(_require.resolve(\"pdfjs-dist/package.json\"))\r\n pdfjsAssets.cMapUrl = join(pkgDir, \"cmaps\") + \"/\"\r\n pdfjsAssets.cMapPacked = true\r\n pdfjsAssets.standardFontDataUrl = join(pkgDir, \"standard_fonts\") + \"/\"\r\n} catch { /* optional dep — 경로 해석 실패 시 cMap 없이 진행 */ }\r\n\r\n/** getDocument + 타임아웃 래퍼 */\r\nasync function loadPdfWithTimeout(buffer: ArrayBuffer) {\r\n const loadingTask = getDocument({\r\n data: new Uint8Array(buffer),\r\n useSystemFonts: true,\r\n disableFontFace: true,\r\n isEvalSupported: false,\r\n ...pdfjsAssets,\r\n })\r\n let timer: ReturnType<typeof setTimeout> | undefined\r\n try {\r\n return await Promise.race([\r\n loadingTask.promise,\r\n new Promise<never>((_, reject) => {\r\n timer = setTimeout(() => { loadingTask.destroy(); reject(new KordocError(\"PDF 로딩 타임아웃 (30초 초과)\")) }, PDF_LOAD_TIMEOUT_MS)\r\n }),\r\n ])\r\n } finally {\r\n if (timer !== undefined) clearTimeout(timer)\r\n }\r\n}\r\n\r\nexport async function parsePdfDocument(buffer: ArrayBuffer, options?: ParseOptions): Promise<InternalParseResult> {\r\n // pdfjs-dist 는 전달받은 buffer 의 underlying storage 를 detach 할 수 있다.\r\n // 수식/텍스트 OCR 은 같은 버퍼를 pdfium 으로 재사용해야 하므로 옵션 on 일 때만 clone 을 보관.\r\n const formulaBuffer: ArrayBuffer | null = options?.formulaOcr ? buffer.slice(0) : null\r\n const ocrBuffer: ArrayBuffer | null = options?.ocr ? buffer.slice(0) : null\r\n const doc = await loadPdfWithTimeout(buffer)\r\n\r\n try {\r\n const pageCount = doc.numPages\r\n if (pageCount === 0) throw new KordocError(\"PDF에 페이지가 없습니다.\")\r\n\r\n // 메타데이터 추출 (best-effort)\r\n const metadata: DocumentMetadata = { pageCount }\r\n await extractPdfMetadata(doc, metadata)\r\n\r\n const blocks: IRBlock[] = []\r\n const warnings: ParseWarning[] = []\r\n const pageQuality: PageQuality[] = []\r\n let totalChars = 0\r\n let totalTextBytes = 0\r\n const effectivePageCount = Math.min(pageCount, MAX_PAGES)\r\n\r\n // 페이지 범위 필터링\r\n const pageFilter = options?.pages ? parsePageRange(options.pages, effectivePageCount) : null\r\n const totalTarget = pageFilter ? pageFilter.size : effectivePageCount\r\n\r\n // 전체 문서의 폰트 크기 빈도 수집 (헤딩 감지용) — 빈도 Map으로 메모리 절약\r\n const fontSizeFreq = new Map<number, number>()\r\n const pageHeights = new Map<number, number>()\r\n // 큰 이미지가 있는 페이지 (needsOcr 경고 노이즈 필터 + SKIPPED_IMAGE)\r\n const pagesWithLargeImage = new Set<number>()\r\n // 텍스트 없는 큰 이미지 영역: page → count\r\n const skippedImagePages = new Map<number, number>()\r\n // 이미지 XObject 바이트 추출 상태 (문서 단위 중복 억제·상한).\r\n // image 블록은 페이지 경계 표 병합(mergeCrossPageTables)의 인접성을 깨지 않도록\r\n // 페이지별로 모아뒀다가 병합 후 주입한다.\r\n const imageState = createPdfImageState()\r\n const extractedImages: ExtractedImage[] = []\r\n const pageImageBlocks = new Map<number, IRBlock[]>()\r\n\r\n let parsedPages = 0\r\n for (let i = 1; i <= effectivePageCount; i++) {\r\n if (pageFilter && !pageFilter.has(i)) continue\r\n try {\r\n const page = await doc.getPage(i)\r\n const tc = await page.getTextContent()\r\n const viewport = page.getViewport({ scale: 1 })\r\n pageHeights.set(i, viewport.height)\r\n const rawItems = tc.items as PdfTextItem[]\r\n const items = normalizeItems(rawItems)\r\n\r\n // hidden text 필터링 + 경고 수집\r\n const { visible, hiddenCount } = filterHiddenText(items, viewport.width, viewport.height)\r\n if (hiddenCount > 0) {\r\n warnings.push({ page: i, message: `${hiddenCount}개 숨겨진 텍스트 요소 필터링됨`, code: \"HIDDEN_TEXT_FILTERED\" })\r\n }\r\n\r\n // 폰트 크기 빈도 수집\r\n for (const item of visible) {\r\n if (item.fontSize > 0) fontSizeFreq.set(item.fontSize, (fontSizeFreq.get(item.fontSize) || 0) + 1)\r\n }\r\n\r\n // 선 기반 테이블 감지를 위한 operatorList\r\n const opList = await page.getOperatorList()\r\n\r\n // 이미지 영역 감지 — 텍스트 없는 큰 이미지는 무음 정보손실이므로 가시화 (ODL 아이디어)\r\n const pageArea = viewport.width * viewport.height\r\n if (pageArea > 0) {\r\n const imageRegions = extractImageRegions(opList.fnArray, opList.argsArray)\r\n let uncovered = 0\r\n for (const r of imageRegions) {\r\n const area = (r.x2 - r.x1) * (r.y2 - r.y1)\r\n if (area < pageArea * 0.05) continue // 작은 장식 이미지 무시\r\n pagesWithLargeImage.add(i)\r\n const hasText = visible.some(it => {\r\n const cx = it.x + it.w / 2\r\n const cy = it.y + (it.h || it.fontSize) / 2\r\n return cx >= r.x1 && cx <= r.x2 && cy >= r.y1 && cy <= r.y2\r\n })\r\n if (!hasText) uncovered++\r\n }\r\n if (uncovered > 0) skippedImagePages.set(i, uncovered)\r\n }\r\n\r\n const pageBlocks = extractPageBlocksWithLines(visible, i, opList, viewport.width, viewport.height)\r\n for (const b of pageBlocks) blocks.push(b)\r\n\r\n // 이미지 XObject 바이트 추출 — 블록 주입은 표 병합 후(injectPageImageBlocks)\r\n try {\r\n const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings)\r\n if (imgBlocks.length > 0) pageImageBlocks.set(i, imgBlocks)\r\n extractedImages.push(...pageImages)\r\n } catch { /* 이미지 추출 실패가 텍스트 파싱을 막지 않도록 */ }\r\n\r\n // 이미지 기반 PDF 감지 + 크기 제한용 문자 수 집계 + 페이지 품질 신호\r\n let pageText = \"\"\r\n for (const b of pageBlocks) {\r\n let t = b.text || \"\"\r\n // 표 블록은 text 없이 table만 가지므로 셀 텍스트도 집계 — 괘선 양식(표-전용)\r\n // 문서가 totalChars=0으로 이미지 기반 오판되어 정상 파싱 표가 OCR로\r\n // 대체되는 것 방지. 페이지 품질(computePageQuality) 집계도 같은 뿌리.\r\n if (b.type === \"table\" && b.table) {\r\n const cellText = b.table.cells.map(row => row.map(c => c.text).join(\" \")).join(\"\\n\")\r\n t = t ? t + \"\\n\" + cellText : cellText\r\n }\r\n totalChars += t.replace(/\\s/g, \"\").length\r\n totalTextBytes += t.length * 2\r\n pageText += pageText ? \"\\n\" + t : t\r\n }\r\n pageQuality.push(computePageQuality(i, pageText))\r\n if (totalTextBytes > MAX_TOTAL_TEXT) throw new KordocError(\"텍스트 추출 크기 초과\")\r\n parsedPages++\r\n options?.onProgress?.(parsedPages, totalTarget)\r\n } catch (pageErr) {\r\n // 크기 초과는 전체 중단\r\n if (pageErr instanceof KordocError) throw pageErr\r\n warnings.push({ page: i, message: `페이지 ${i} 파싱 실패: ${pageErr instanceof Error ? pageErr.message : \"알 수 없는 오류\"}`, code: \"PARTIAL_PARSE\" })\r\n }\r\n }\r\n\r\n const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount)\r\n // 문서 단위 이미지 기반 판정 (평균 10자/페이지 미만 = 텍스트층 부재)\r\n const isImageBased = totalChars / Math.max(parsedPageCount, 1) < 10\r\n\r\n // ── OCR 실행 (옵션) — 페이지 단위 선정·병합 ──\r\n // 대상: \"force\"=전 페이지 / 문서가 이미지 기반=전 페이지 /\r\n // 그 외=품질 신호가 OCR 을 권하는 페이지만 (깨진 텍스트층 포함 — F1,\r\n // 혼합 문서의 스캔 페이지 포함 — F2). 정상 페이지 파싱 결과는 유지 (F3).\r\n const ocrDone = new Set<number>()\r\n if (options?.ocr && ocrBuffer) {\r\n const inScope = (p: number) => !pageFilter || pageFilter.has(p)\r\n const targets = new Set<number>()\r\n if (options.ocr === \"force\" || isImageBased) {\r\n for (let i = 1; i <= effectivePageCount; i++) if (inScope(i)) targets.add(i)\r\n } else {\r\n for (const pq of pageQuality) {\r\n if (!pq.needsOcr) continue\r\n // low_text 는 빈 페이지(표지/간지)일 수 있으므로 큰 이미지가 있는 페이지만\r\n if (pq.ocrReason === \"low_text\" && !pagesWithLargeImage.has(pq.page)) continue\r\n targets.add(pq.page)\r\n }\r\n }\r\n if (targets.size > 0) {\r\n try {\r\n const { runPdfOcr } = await import(\"../ocr/pdf-ocr.js\")\r\n const mode = typeof options.ocr === \"function\" ? options.ocr : (\"builtin\" as const)\r\n const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options.onProgress)\r\n if (ocrPageBlocks.size > 0) {\r\n // 페이지 단위 교체: OCR 성공 페이지의 기존(깨진/빈) 블록 제거 후 병합\r\n for (const p of ocrPageBlocks.keys()) ocrDone.add(p)\r\n const merged = blocks.filter(b => !(b.pageNumber && ocrDone.has(b.pageNumber)))\r\n for (const obs of ocrPageBlocks.values()) merged.push(...obs)\r\n merged.sort((a, b) => (a.pageNumber ?? 0) - (b.pageNumber ?? 0)) // stable — 페이지 내 순서 보존\r\n blocks.length = 0\r\n blocks.push(...merged)\r\n for (const pq of pageQuality) if (ocrDone.has(pq.page)) pq.ocrApplied = true\r\n warnings.push({\r\n message: `${ocrDone.size}개 페이지에 OCR 적용 (${mode === \"builtin\" ? \"내장 PP-OCRv5\" : \"사용자 프로바이더\"})`,\r\n code: \"OCR_APPLIED\",\r\n })\r\n }\r\n } catch (e) {\r\n // 환경 오류(의존성·모델) — 아래 NEEDS_OCR 폴백이 가시화, 원인은 별도 경고로 보존 (F6)\r\n warnings.push({\r\n message: `OCR 실행 불가: ${e instanceof Error ? e.message : String(e)}`,\r\n code: \"OCR_FAILED\",\r\n })\r\n }\r\n }\r\n }\r\n\r\n if (isImageBased && ocrDone.size === 0) {\r\n // OCR 미설정/실패 — 빈 출력을 무경고로 내보내지 않고 경고 + 플래그로 가시화 (v3.0)\r\n warnings.push({\r\n message: `이미지 기반 PDF (${pageCount}페이지, 텍스트 ${totalChars}자) — 텍스트 레이어가 없어 OCR이 필요합니다`,\r\n code: \"NEEDS_OCR\",\r\n })\r\n }\r\n\r\n // 페이지 단위 needsOcr 경고 — 텍스트+스캔 혼합 문서에서 스캔 페이지 무음 손실 방지.\r\n // low_text는 빈 페이지(표지/간지)일 수 있으므로 큰 이미지가 있는 페이지만 경고.\r\n // OCR 이 적용된 페이지는 해소된 것이므로 제외.\r\n if (!isImageBased) {\r\n const OCR_REASON_MESSAGES: Record<string, string> = {\r\n low_text: \"텍스트가 거의 없는 페이지 (스캔/이미지 추정)\",\r\n high_pua: \"글꼴 매핑 실패 (PUA 비율 높음) — 추출 텍스트 신뢰 불가\",\r\n high_control: \"제어문자 비율 높음 — 추출 텍스트 신뢰 불가\",\r\n high_replacement: \"대체문자(U+FFFD) 비율 높음 — 추출 텍스트 신뢰 불가\",\r\n garbled_hangul: \"글꼴 매핑 실패 (한글 자소 분포 이상) — 추출 텍스트가 깨졌을 수 있음\",\r\n }\r\n for (const pq of pageQuality) {\r\n if (!pq.needsOcr || !pq.ocrReason || pq.ocrApplied) continue\r\n if (pq.ocrReason === \"low_text\" && !pagesWithLargeImage.has(pq.page)) continue\r\n warnings.push({ page: pq.page, message: `${OCR_REASON_MESSAGES[pq.ocrReason]} — OCR 검토 필요`, code: \"NEEDS_OCR\" })\r\n }\r\n }\r\n\r\n // 텍스트 없는 큰 이미지 영역 경고 — 그림/차트/도장 무음 누락 가시화\r\n // (문서 전체가 이미지 기반이면 위의 NEEDS_OCR 단일 경고로 충분)\r\n if (!isImageBased) {\r\n for (const [page, count] of [...skippedImagePages.entries()].sort((a, b) => a[0] - b[0])) {\r\n warnings.push({ page, message: `${count}개 이미지 영역에 추출 가능한 텍스트 없음 (그림/차트/도장 내용 누락 가능)`, code: \"SKIPPED_IMAGE\" })\r\n }\r\n }\r\n\r\n // 머리글/바닥글 필터링 (기본 ON — 명시적 false일 때만 비활성화)\r\n if (options?.removeHeaderFooter !== false && parsedPageCount >= 3) {\r\n const removed = removeHeaderFooterBlocks(blocks, pageHeights, warnings)\r\n // 필터링된 블록 제거 (뒤에서부터 삭제)\r\n for (let ri = removed.length - 1; ri >= 0; ri--) {\r\n blocks.splice(removed[ri], 1)\r\n }\r\n }\r\n\r\n // 페이지 걸친 표 병합 — 머리글/바닥글 제거 후 인접해진 표를 하나로\r\n // (ODL TableBorderProcessor.checkNeighborTables 포팅)\r\n mergeCrossPageTables(blocks)\r\n\r\n // 추출 이미지 참조를 페이지 말미 위치에 주입 (표 병합 뒤 — 인접성 보존)\r\n injectPageImageBlocks(blocks, pageImageBlocks)\r\n\r\n // 수식 OCR (선택) — 기본 텍스트 추출과 별개로 페이지 이미지 렌더 후 수식만 검출/인식.\r\n // 실패 시 경고만 기록하고 일반 텍스트 추출 결과는 그대로 반환한다.\r\n if (options?.formulaOcr && formulaBuffer) {\r\n try {\r\n await applyFormulaOcr(formulaBuffer, blocks, pageFilter, effectivePageCount, warnings, options.onProgress)\r\n } catch (e) {\r\n warnings.push({\r\n message: `수식 OCR 실패: ${e instanceof Error ? e.message : String(e)}`,\r\n code: \"PARTIAL_PARSE\",\r\n })\r\n }\r\n }\r\n\r\n // 헤딩 감지: 폰트 크기 기반\r\n const medianFontSize = computeMedianFontSizeFromFreq(fontSizeFreq)\r\n if (medianFontSize > 0) {\r\n detectHeadings(blocks, medianFontSize)\r\n }\r\n\r\n // □/■ 마커 기반 서브헤딩 감지 (ODL 패턴)\r\n detectMarkerHeadings(blocks)\r\n\r\n // 표 캡션 감지 — 표 직전/직후 '표 N./그림 N' 패턴 텍스트를 IRTable.caption으로\r\n detectTableCaptions(blocks)\r\n\r\n // 한국어 리스트 감지 — 공문서 계층 라벨(1.→가.→1)→가)→①) 시퀀스 검증\r\n detectKoreanListBlocks(blocks)\r\n\r\n // outline 구축\r\n const outline: OutlineItem[] = blocks\r\n .filter(b => b.type === \"heading\" && b.level && b.text)\r\n .map(b => ({ level: b.level!, text: b.text!, pageNumber: b.pageNumber }))\r\n\r\n // 메트릭 수집 끝났으니 블록 텍스트의 C0/C1 제어문자(NUL 등) 정리\r\n sanitizeBlockControlChars(blocks)\r\n\r\n // blocksToMarkdown로 통일 — 헤딩 마크다운 반영 (HWP5/HWPX와 일관성)\r\n let markdown = cleanPdfText(blocksToMarkdown(blocks))\r\n\r\n return {\r\n markdown,\r\n blocks,\r\n metadata,\r\n outline: outline.length > 0 ? outline : undefined,\r\n warnings: warnings.length > 0 ? warnings : undefined,\r\n isImageBased: isImageBased || undefined,\r\n pageQuality,\r\n qualitySummary: summarizeDocumentQuality(pageQuality),\r\n images: extractedImages.length > 0 ? extractedImages : undefined,\r\n }\r\n } finally {\r\n await doc.destroy().catch(() => {})\r\n }\r\n}\r\n\r\n// ─── PDF 메타데이터 추출 ────────────────────────────\r\n\r\nasync function extractPdfMetadata(doc: { getMetadata(): Promise<unknown> }, metadata: DocumentMetadata): Promise<void> {\r\n try {\r\n const result = await doc.getMetadata() as { info?: Record<string, unknown> } | null\r\n if (!result?.info) return\r\n const info = result.info\r\n\r\n if (typeof info.Title === \"string\" && info.Title.trim()) metadata.title = info.Title.trim()\r\n if (typeof info.Author === \"string\" && info.Author.trim()) metadata.author = info.Author.trim()\r\n if (typeof info.Creator === \"string\" && info.Creator.trim()) metadata.creator = info.Creator.trim()\r\n if (typeof info.Subject === \"string\" && info.Subject.trim()) metadata.description = info.Subject.trim()\r\n if (typeof info.Keywords === \"string\" && info.Keywords.trim()) {\r\n metadata.keywords = info.Keywords.split(/[,;]/).map((k: string) => k.trim()).filter(Boolean)\r\n }\r\n if (typeof info.CreationDate === \"string\") metadata.createdAt = parsePdfDate(info.CreationDate)\r\n if (typeof info.ModDate === \"string\") metadata.modifiedAt = parsePdfDate(info.ModDate)\r\n } catch {\r\n // best-effort\r\n }\r\n}\r\n\r\n/** PDF 날짜 형식 (D:YYYYMMDDHHmmSS) → ISO 8601 변환 */\r\nfunction parsePdfDate(dateStr: string): string | undefined {\r\n const m = dateStr.match(/D:(\\d{4})(\\d{2})?(\\d{2})?(\\d{2})?(\\d{2})?(\\d{2})?/)\r\n if (!m) return undefined\r\n const [, year, month = \"01\", day = \"01\", hour = \"00\", min = \"00\", sec = \"00\"] = m\r\n return `${year}-${month}-${day}T${hour}:${min}:${sec}`\r\n}\r\n\r\n/** 메타데이터만 추출 (전체 파싱 없이) — MCP parse_metadata용 */\r\nexport async function extractPdfMetadataOnly(buffer: ArrayBuffer): Promise<DocumentMetadata> {\r\n const doc = await loadPdfWithTimeout(buffer)\r\n\r\n try {\r\n const metadata: DocumentMetadata = { pageCount: doc.numPages }\r\n await extractPdfMetadata(doc, metadata)\r\n return metadata\r\n } finally {\r\n await doc.destroy().catch(() => {})\r\n }\r\n}\r\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAQA,SAAS,KAAK,iBAAiB;AAsB/B,IAAM,2BAA2B;AAQjC,SAAS,eAAe,MAAmB,OAA2C;AAEpF,QAAM,QAAQ,MAAM,WAAW,IAAI,IAAI,KAAK,aAAa,KAAK;AAC9D,SAAO,IAAI,QAAQ,aAAW;AAC5B,QAAI,OAAO;AACX,UAAM,QAAQ,WAAW,MAAM;AAAE,UAAI,CAAC,MAAM;AAAE,eAAO;AAAM,gBAAQ,IAAI;AAAA,MAAE;AAAA,IAAE,GAAG,wBAAwB;AACtG,QAAI;AACF,YAAM,IAAI,OAAO,CAAC,SAAkB;AAClC,YAAI,CAAC,MAAM;AAAE,iBAAO;AAAM,uBAAa,KAAK;AAAG,kBAAS,QAAQ,IAA0B;AAAA,QAAE;AAAA,MAC9F,CAAC;AAAA,IACH,QAAQ;AACN,UAAI,CAAC,MAAM;AAAE,eAAO;AAAM,qBAAa,KAAK;AAAG,gBAAQ,IAAI;AAAA,MAAE;AAAA,IAC/D;AAAA,EACF,CAAC;AACH;AAGA,IAAM,UAAU;AAEhB,IAAM,aAAa;AAEnB,IAAM,qBAAqB;AAE3B,IAAM,wBAAwB,MAAM,OAAO;AAWpC,SAAS,sBAAqC;AACnD,SAAO,EAAE,YAAY,GAAG,YAAY,GAAG,MAAM,oBAAI,IAAI,GAAG,WAAW,MAAM;AAC3E;AAGA,SAAS,MAAM,MAA8C;AAC3D,MAAI,IAAI;AACR,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,SAAK,KAAK,CAAC;AACX,QAAI,KAAK,KAAK,GAAG,QAAU;AAAA,EAC7B;AACA,SAAO,MAAM;AACf;AAGA,SAAS,OAAO,KAAoC;AAClD,QAAM,EAAE,OAAO,GAAG,QAAQ,GAAG,MAAM,KAAK,IAAI;AAC5C,MAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAG,QAAO;AAC9B,QAAM,OAAO,IAAI,WAAW,IAAI,IAAI,CAAC;AAErC,MAAI,SAAS,UAAU,YAAY;AACjC,QAAI,KAAK,SAAS,IAAI,IAAI,EAAG,QAAO;AACpC,SAAK,IAAI,KAAK,SAAS,GAAG,IAAI,IAAI,CAAC,CAAC;AACpC,WAAO;AAAA,EACT;AACA,MAAI,SAAS,UAAU,WAAW;AAChC,QAAI,KAAK,SAAS,IAAI,IAAI,EAAG,QAAO;AACpC,aAAS,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,IAAI,GAAG,KAAK,KAAK,GAAG,KAAK,GAAG;AAC5D,WAAK,CAAC,IAAI,KAAK,CAAC;AAAG,WAAK,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC;AAAG,WAAK,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC;AAAG,WAAK,IAAI,CAAC,IAAI;AAAA,IACzF;AACA,WAAO;AAAA,EACT;AACA,MAAI,SAAS,UAAU,gBAAgB;AAErC,UAAM,SAAU,IAAI,KAAM;AAC1B,QAAI,KAAK,SAAS,SAAS,EAAG,QAAO;AACrC,aAAS,IAAI,GAAG,IAAI,GAAG,KAAK;AAC1B,eAAS,IAAI,GAAG,IAAI,GAAG,KAAK;AAC1B,cAAM,MAAO,KAAK,IAAI,UAAU,KAAK,EAAE,KAAM,KAAK,IAAI,KAAO;AAC7D,cAAM,IAAI,MAAM,MAAM;AACtB,cAAM,KAAK,IAAI,IAAI,KAAK;AACxB,aAAK,CAAC,IAAI;AAAG,aAAK,IAAI,CAAC,IAAI;AAAG,aAAK,IAAI,CAAC,IAAI;AAAG,aAAK,IAAI,CAAC,IAAI;AAAA,MAC/D;AAAA,IACF;AACA,WAAO;AAAA,EACT;AACA,SAAO;AACT;AAOA,eAAsB,kBACpB,MACA,SACA,WACA,YACA,OACA,UAC0D;AAC1D,QAAM,SAAoB,CAAC;AAC3B,QAAM,SAA2B,CAAC;AAClC,QAAM,cAAc,oBAAI,IAAY;AAEpC,WAAS,IAAI,GAAG,IAAI,QAAQ,QAAQ,KAAK;AACvC,UAAM,KAAK,QAAQ,CAAC;AACpB,QAAI,UAA6B;AACjC,QAAI,WAA0B;AAE9B,QAAI,OAAO,IAAI,qBAAqB,OAAO,IAAI,yBAAyB;AACtE,YAAM,QAAQ,UAAU,CAAC,IAAI,CAAC;AAC9B,UAAI,OAAO,UAAU,YAAY,YAAY,IAAI,KAAK,EAAG;AACzD,kBAAY,IAAI,KAAK;AACrB,iBAAW;AAAA,IACb,WAAW,OAAO,IAAI,yBAAyB;AAC7C;AAAA,IACF;AAIA,QAAI,MAAM,cAAc,sBAAsB,MAAM,cAAc,uBAAuB;AACvF,UAAI,CAAC,MAAM,WAAW;AACpB,cAAM,YAAY;AAClB,iBAAS,KAAK,EAAE,MAAM,YAAY,SAAS,8DAAiB,kBAAkB,qEAAwB,MAAM,gBAAgB,CAAC;AAAA,MAC/H;AACA;AAAA,IACF;AAEA,QAAI,UAAU;AACZ,gBAAU,MAAM,eAAe,MAAM,QAAQ;AAAA,IAC/C,OAAO;AACL,gBAAU,UAAU,CAAC,IAAI,CAAC;AAAA,IAC5B;AAEA,QAAI,CAAC,SAAS,QAAQ,CAAC,QAAQ,SAAS,CAAC,QAAQ,OAAQ;AACzD,UAAM,EAAE,OAAO,GAAG,QAAQ,EAAE,IAAI;AAChC,QAAI,IAAI,WAAW,IAAI,QAAS;AAChC,QAAI,IAAI,IAAI,YAAY;AACtB,eAAS,KAAK,EAAE,MAAM,YAAY,SAAS,iFAAqB,CAAC,OAAI,CAAC,KAAK,MAAM,gBAAgB,CAAC;AAClG;AAAA,IACF;AAGA,UAAM,OAAO,GAAG,CAAC,IAAI,CAAC,IAAI,QAAQ,QAAQ,GAAG,IAAI,MAAM,QAAQ,IAAI,CAAC,GAAG,WAAW,KAAK,QAAQ;AAC/F,QAAI,MAAM,KAAK,IAAI,IAAI,EAAG;AAE1B,UAAM,OAAO,OAAO,OAAO;AAC3B,QAAI,CAAC,KAAM;AAEX,UAAM,MAAM,UAAU,GAAG,GAAG,IAAI;AAChC,UAAM;AACN,UAAM,cAAc,IAAI;AACxB,UAAM,WAAW,SAAS,OAAO,MAAM,UAAU,EAAE,SAAS,GAAG,GAAG,CAAC;AACnE,UAAM,KAAK,IAAI,MAAM,QAAQ;AAC7B,WAAO,KAAK,EAAE,UAAU,MAAM,KAAK,UAAU,YAAY,CAAC;AAC1D,WAAO,KAAK,EAAE,MAAM,SAAS,MAAM,UAAU,WAAW,CAAC;AAAA,EAC3D;AAEA,SAAO,EAAE,QAAQ,OAAO;AAC1B;AAOO,SAAS,sBAAsB,QAAmB,YAA0C;AACjG,MAAI,WAAW,SAAS,EAAG;AAG3B,QAAM,mBAAmB,oBAAI,IAAoB;AACjD,WAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,UAAM,IAAI,OAAO,CAAC,EAAE;AACpB,QAAI,MAAM,OAAW,kBAAiB,IAAI,GAAG,CAAC;AAAA,EAChD;AACA,QAAM,cAAc,oBAAI,IAAoB;AAC5C,aAAW,CAAC,GAAG,IAAI,KAAK,iBAAkB,aAAY,IAAI,MAAM,CAAC;AAEjE,QAAM,SAAoB,CAAC;AAC3B,QAAM,WAAW,oBAAI,IAAY;AACjC,WAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,WAAO,KAAK,OAAO,CAAC,CAAC;AACrB,UAAM,IAAI,YAAY,IAAI,CAAC;AAC3B,QAAI,MAAM,UAAa,WAAW,IAAI,CAAC,GAAG;AACxC,aAAO,KAAK,GAAG,WAAW,IAAI,CAAC,CAAE;AACjC,eAAS,IAAI,CAAC;AAAA,IAChB;AAAA,EACF;AAEA,aAAW,KAAK,CAAC,GAAG,WAAW,KAAK,CAAC,EAAE,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC,GAAG;AAC5D,QAAI,SAAS,IAAI,CAAC,EAAG;AACrB,QAAI,KAAK,OAAO;AAChB,aAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,YAAM,KAAK,OAAO,CAAC,EAAE;AACrB,UAAI,OAAO,UAAa,KAAK,GAAG;AAAE,aAAK;AAAG;AAAA,MAAM;AAAA,IAClD;AACA,WAAO,OAAO,IAAI,GAAG,GAAG,WAAW,IAAI,CAAC,CAAE;AAAA,EAC5C;AAEA,SAAO,SAAS;AAChB,SAAO,KAAK,GAAG,MAAM;AACvB;;;AC9LA,IAAM,YAAY,oBAAI,IAAY,CAAC,GAAG,GAAG,GAAG,GAAG,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,EAAE,CAAC;AAGtF,IAAM,sBAAsB;AAE5B,IAAM,yBAAyB;AAE/B,IAAM,2BAA2B;AAG1B,SAAS,mBAAmB,MAAc,MAA2B;AAC1E,MAAI,QAAQ;AACZ,MAAI,SAAS;AACb,MAAI,kBAAkB;AACtB,MAAI,oBAAoB;AACxB,MAAI,UAAU;AACd,MAAI,cAAc;AAClB,MAAI,MAAM;AAEV,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,UAAM,OAAO,KAAK,WAAW,CAAC;AAE9B,QAAI,SAAS,MAAQ,SAAS,KAAQ,SAAS,MAAQ,SAAS,GAAM;AACtE;AAGA,QAAK,OAAO,MAAS,SAAS,OAAS,QAAQ,OAAQ,QAAQ,KAAO;AACpE;AACA;AAAA,IACF;AACA,QAAI,SAAS,OAAQ;AACnB;AACA;AAAA,IACF;AAEA,QAAI,QAAQ,SAAU,QAAQ,OAAQ;AACpC;AACA,YAAM,QAAQ,OAAO,SAAU;AAC/B,UAAI,SAAS,EAAG;AAAA,eACP,UAAU,IAAI,IAAI,EAAG;AAC9B;AAAA,IACF;AAGA,QAAK,QAAQ,SAAU,QAAQ,SAAY,QAAQ,SAAU,QAAQ,OAAS;AAC5E;AACA;AAAA,IACF;AAAA,EACF;AAEA,QAAM,QAAQ,SAAS;AACvB,QAAM,WAAW,MAAM;AACvB,QAAM,mBAAmB,UAAU;AACnC,QAAM,uBAAuB,cAAc;AAC3C,QAAM,uBAAuB,SAAS,IAAI,kBAAkB,SAAS;AACrE,QAAM,yBAAyB,SAAS,IAAI,oBAAoB,SAAS;AAIzE,QAAM,gBACJ,UAAU,uBACV,uBAAuB,0BACvB,0BAA0B;AAE5B,MAAI,WAAW;AACf,MAAI;AAEJ,MAAI,QAAQ,oBAAoB;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAW,WACjE,YAAY,oBAAoB;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAW,WAC1E,oBAAoB,wBAAwB;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAe,WAC1F,wBAAwB,4BAA4B;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAmB,WACtG,eAAe;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAiB;AAExE,SAAO;AAAA,IACL;AAAA,IACA,WAAW;AAAA,IACX,aAAa,SAAS;AAAA,IACtB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACF;AAoBA,IAAM,qBAAqB;AAC3B,IAAM,qBAAqB;AAC3B,IAAM,yBAAyB;AAC/B,IAAM,6BAA6B;AAEnC,IAAM,2BAA2B;AAM1B,SAAS,kBAAkB,MAAsB;AAEtD,SAAO,KAAK,QAAQ,8CAA8C,EAAE;AACtE;AAEO,SAAS,yBAAyB,OAA8C;AACrF,MAAI,MAAM,WAAW,GAAG;AACtB,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,gBAAgB;AAAA,MAChB,gBAAgB;AAAA,MAChB,qBAAqB;AAAA,MACrB,yBAAyB;AAAA,MACzB,aAAa;AAAA,MACb,kBAAkB;AAAA,MAClB,kBAAkB;AAAA,MAClB,UAAU;AAAA,MACV,mBAAmB,CAAC;AAAA,IACtB;AAAA,EACF;AAEA,MAAI,YAAY;AAChB,MAAI,SAAS;AACb,MAAI,UAAU;AACd,MAAI,cAAc;AAClB,MAAI,MAAM;AACV,MAAI,UAAU;AACd,MAAI,UAAU;AACd,QAAM,oBAA8B,CAAC;AAErC,aAAW,KAAK,OAAO;AACrB,iBAAa,EAAE;AACf,cAAU,EAAE;AACZ,eAAW,EAAE;AACb,mBAAe,EAAE;AACjB,WAAO,EAAE;AACT,QAAI,EAAE,YAAY,mBAAoB;AACtC,QAAI,EAAE,YAAY,mBAAoB;AAEtC,QAAI,EAAE,YAAY,CAAC,EAAE,WAAY,mBAAkB,KAAK,EAAE,IAAI;AAAA,EAChE;AAEA,QAAM,IAAI,MAAM;AAChB,SAAO;AAAA,IACL,YAAY;AAAA,IACZ,gBAAgB;AAAA,IAChB,gBAAgB,SAAS;AAAA,IACzB,qBAAqB,UAAU;AAAA,IAC/B,yBAAyB,cAAc;AAAA,IACvC,aAAa,MAAM;AAAA,IACnB,kBAAkB;AAAA,IAClB,kBAAkB;AAAA,IAClB,UAAU,kBAAkB,SAAS,KAAK;AAAA,IAC1C;AAAA,EACF;AACF;;;AC5MO,SAAS,0BAA0B,QAAyB;AACjE,aAAW,KAAK,QAAQ;AACtB,QAAI,EAAE,KAAM,GAAE,OAAO,kBAAkB,EAAE,IAAI;AAC7C,QAAI,EAAE,OAAO;AACX,iBAAW,OAAO,EAAE,MAAM,OAAO;AAC/B,mBAAW,QAAQ,KAAK;AACtB,cAAI,KAAK,KAAM,MAAK,OAAO,kBAAkB,KAAK,IAAI;AAAA,QACxD;AAAA,MACF;AAAA,IACF;AACA,QAAI,EAAE,SAAU,2BAA0B,EAAE,QAAQ;AAAA,EACtD;AACF;AAEO,SAAS,aAAa,MAAsB;AACjD,SAAO;AAAA,IACL,kBAAkB,IAAI,EAEnB,QAAQ,cAAc,EAAE,EAExB,QAAQ,oDAAoD,EAAE,EAE9D,QAAQ,4BAA4B,EAAE,EAEtC,QAAQ,gBAAgB,IAAI,EAE5B,QAAQ,cAAc,EAAE,EAExB,QAAQ,2BAA2B,EAAE;AAAA,EAC1C,EAGG,QAAQ,oBAAoB,UAAQ;AACnC,QAAI,2BAA2B,KAAK,IAAI,EAAG,QAAO;AAClD,WAAO,oBAAoB,IAAI;AAAA,EACjC,CAAC,EAEA,QAAQ,oCAAoC,SAAS,EAErD,QAAQ,WAAW,IAAI,EAEvB,QAAQ,SAAS,EAAE,EACnB,QAAQ,WAAW,MAAM,EACzB,KAAK;AACV;AAEA,SAAS,iBAAiB,MAAuB;AAC/C,QAAM,IAAI,KAAK,UAAU;AACzB,SAAO,gBAAgB,KAAK,CAAC,KAAK,WAAW,KAAK,CAAC,KAAK,mBAAmB,KAAK,CAAC,KAC/E,sBAAsB,KAAK,CAAC,KAAK,eAAe,KAAK,CAAC;AAC1D;AAEA,SAAS,mBAAmB,MAAuB;AACjD,SAAO,yCAAyC,KAAK,KAAK,KAAK,CAAC;AAClE;AAEA,SAAS,iBAAiB,MAAsB;AAC9C,MAAI,CAAC,KAAM,QAAO;AAClB,QAAM,QAAQ,KAAK,MAAM,IAAI;AAC7B,MAAI,MAAM,UAAU,EAAG,QAAO;AAC9B,QAAM,SAAmB,CAAC,MAAM,CAAC,CAAC;AAElC,WAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;AACrC,UAAM,OAAO,OAAO,OAAO,SAAS,CAAC;AACrC,UAAM,OAAO,MAAM,CAAC;AACpB,UAAM,cAAc,KAAK,KAAK;AAE9B,QAAI,YAAY,KAAK,IAAI,KAAK,YAAY,KAAK,IAAI,KAAK,MAAM,KAAK,WAAW,KAAK,OAAO,KAAK,WAAW,GAAG;AAC3G,aAAO,KAAK,IAAI;AAChB;AAAA,IACF;AAEA,QAAI,KAAK,KAAK,KAAK,KAAK,CAAC,KAAK,YAAY,SAAS,GAAG;AACpD,aAAO,OAAO,SAAS,CAAC,IAAI,OAAO,OAAO;AAC1C;AAAA,IACF;AAEA,QAAI,OAAO,KAAK,WAAW,GAAG;AAC5B,aAAO,OAAO,SAAS,CAAC,IAAI,OAAO,MAAM;AACzC;AAAA,IACF;AAEA,QAAI,aAAa,KAAK,IAAI,KAAK,UAAU,KAAK,IAAI,KAC9C,CAAC,iBAAiB,IAAI,KAAK,CAAC,mBAAmB,IAAI,KACnD,CAAC,iBAAiB,IAAI,GAAG;AAC3B,aAAO,OAAO,SAAS,CAAC,IAAI,OAAO,MAAM;AAAA,IAC3C,OAAO;AACL,aAAO,KAAK,IAAI;AAAA,IAClB;AAAA,EACF;AACA,SAAO,OAAO,KAAK,IAAI;AACzB;;;ACpFA,eAAsB,gBACpB,QACA,QACA,YACA,oBACA,UACA,aACe;AACf,QAAM,aAAa,MAAM,OAAO,uBAAoB;AACpD,QAAM,EAAE,iBAAiB,oBAAoB,IAAI;AAGjD,QAAM,oBAAoB,CAAC,MAAM;AAC/B,QAAI,EAAE,UAAU,cAAc,EAAE,OAAO;AACrC,YAAM,MAAM,KAAK,MAAO,EAAE,aAAa,EAAE,QAAS,GAAG;AACrD,cAAQ,OAAO,MAAM,sBAAsB,EAAE,KAAK,IAAI,IAAI,GAAG,MAAM,SAAS,EAAE,UAAU,CAAC,IAAI,SAAS,EAAE,KAAK,CAAC,GAAG;AACjH,UAAI,EAAE,cAAc,EAAE,MAAO,SAAQ,OAAO,MAAM,IAAI;AAAA,IACxD,WAAW,EAAE,UAAU,UAAU;AAC/B,cAAQ,OAAO,MAAM,oBAAoB,EAAE,KAAK,IAAI;AAAA,CAAoB;AAAA,IAC1E,WAAW,EAAE,UAAU,QAAQ;AAC7B,cAAQ,OAAO,MAAM,oBAAoB,EAAE,KAAK,IAAI;AAAA,CAAU;AAAA,IAChE,WAAW,EAAE,UAAU,QAAQ;AAAA,IAE/B;AAAA,EACF,CAAC;AAED,QAAM,WAAW,MAAM,gBAAgB,OAAO;AAC9C,MAAI;AAGF,UAAM,kBAAkB,cACnB,IAAI,IAAI,MAAM,KAAK,EAAE,QAAQ,mBAAmB,GAAG,CAAC,GAAG,MAAM,IAAI,CAAC,CAAC;AACxE,UAAM,cAAc,MAAM,SAAS,YAAY,QAAQ,eAAe;AAEtE,QAAI,YAAY,WAAW,EAAG;AAE9B,QAAI,gBAAgB;AACpB,QAAI,kBAAkB;AAEtB,eAAW,QAAQ,aAAa;AAC9B,YAAM,aAAa,KAAK;AACxB,YAAM,YAAY,KAAK;AACvB,YAAM,SAAS,KAAK,gBAAgB,IAAI,KAAK,WAAW,KAAK,gBAAgB;AAC7E,YAAM,SAAS,KAAK,iBAAiB,IAAI,KAAK,YAAY,KAAK,iBAAiB;AAQhF,YAAM,aAAiC,CAAC;AACxC,iBAAW,KAAK,KAAK,SAAS;AAC5B,YAAI,CAAC,EAAE,SAAS,CAAC,EAAE,MAAM,KAAK,EAAG;AACjC,cAAM,UAAU,EAAE,SAAS,YAAY,KAAK,EAAE,KAAK,OAAO,IAAI,EAAE,KAAK;AAErE,cAAM,KAAK,EAAE,KAAK,KAAK;AACvB,cAAM,KAAK,EAAE,KAAK,KAAK;AAEvB,cAAM,OAAO,YAAY,EAAE,KAAK,KAAK;AACrC,cAAM,UAAU,YAAY,EAAE,KAAK,KAAK;AACxC,cAAM,WAAW,OAAO,WAAW;AACnC,cAAM,QAAQ,KAAK;AACnB,cAAM,SAAS,OAAO;AAEtB,mBAAW,KAAK;AAAA,UACd,OAAO;AAAA,YACL,MAAM;AAAA,YACN,MAAM;AAAA,YACN;AAAA,YACA,MAAM,EAAE,MAAM,YAAY,GAAG,IAAI,GAAG,SAAS,OAAO,OAAO;AAAA,UAC7D;AAAA,UACA,SAAS,EAAE,IAAI,IAAI,MAAM,QAAQ;AAAA,UACjC;AAAA,QACF,CAAC;AAAA,MACH;AACA,UAAI,WAAW,WAAW,EAAG;AAI7B,YAAM,oBAAoB;AAC1B,YAAM,kBAAkB,oBAAI,IAAY;AACxC,eAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,cAAM,IAAI,OAAO,CAAC;AAClB,YAAI,EAAE,eAAe,WAAY;AACjC,YAAI,EAAE,SAAS,QAAS;AACxB,YAAI,CAAC,EAAE,QAAQ,EAAE,KAAK,SAAS,KAAK,EAAE,KAAK,UAAU,EAAG;AACxD,cAAM,YAAY,EAAE,KAAK,QAAQ,EAAE,KAAK;AACxC,YAAI,aAAa,EAAG;AAEpB,mBAAW,KAAK,YAAY;AAC1B,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,GAAG,EAAE,QAAQ,EAAE;AAC3C,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,IAAI,EAAE,KAAK,OAAO,EAAE,QAAQ,EAAE;AAC1D,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,GAAG,EAAE,QAAQ,OAAO;AAChD,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,IAAI,EAAE,KAAK,QAAQ,EAAE,QAAQ,IAAI;AAC7D,gBAAM,YAAY,KAAK,IAAI,GAAG,MAAM,GAAG,IAAI,KAAK,IAAI,GAAG,MAAM,GAAG;AAChE,cAAI,YAAY,aAAa,mBAAmB;AAC9C,4BAAgB,IAAI,CAAC;AACrB;AAAA,UACF;AAAA,QACF;AAAA,MACF;AAEA,UAAI,gBAAgB,OAAO,GAAG;AAE5B,cAAM,SAAS,CAAC,GAAG,eAAe,EAAE,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC;AACxD,mBAAW,OAAO,OAAQ,QAAO,OAAO,KAAK,CAAC;AAC9C,2BAAmB,gBAAgB;AAAA,MACrC;AAKA,iBAAW,KAAK,CAAC,GAAG,MAAM,EAAE,UAAU,EAAE,OAAO;AAE/C,iBAAW,KAAK,YAAY;AAC1B,YAAI,YAAY;AAChB,YAAI,eAAe;AACnB,YAAI,cAAc;AAClB,iBAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,gBAAM,IAAI,OAAO,CAAC;AAClB,cAAI,EAAE,eAAe,WAAY;AACjC,cAAI,iBAAiB,GAAI,gBAAe;AACxC,wBAAc;AACd,cAAI,CAAC,EAAE,KAAM;AACb,gBAAM,cAAc,EAAE,KAAK,IAAI,EAAE,KAAK,SAAS;AAC/C,cAAI,cAAc,EAAE,SAAS;AAC3B,wBAAY;AACZ;AAAA,UACF;AAAA,QACF;AAEA,YAAI,cAAc,IAAI;AACpB,iBAAO,OAAO,WAAW,GAAG,EAAE,KAAK;AAAA,QACrC,WAAW,gBAAgB,IAAI;AAC7B,iBAAO,OAAO,cAAc,GAAG,GAAG,EAAE,KAAK;AAAA,QAC3C,OAAO;AAEL,iBAAO,KAAK,EAAE,KAAK;AAAA,QACrB;AACA;AAAA,MACF;AAAA,IACF;AAEA,QAAI,gBAAgB,KAAK,kBAAkB,GAAG;AAC5C,cAAQ,OAAO;AAAA,QACb,oBAAoB,aAAa,qCAAY,eAAe,2CAAkB,YAAY,MAAM;AAAA;AAAA,MAClG;AAAA,IACF;AAAA,EACF,UAAE;AACA,UAAM,SAAS,QAAQ,EAAE,MAAM,MAAM;AAAA,IAAC,CAAC;AAAA,EACzC;AACF;AAEA,SAAS,SAAS,OAAuB;AACvC,SAAO,IAAI,QAAQ,OAAO,MAAM,QAAQ,CAAC,CAAC;AAC5C;;;ACvJA,YAAY,iBAAiB;AAf7B,IAAM,IAAI;AAEV,IAAI,OAAO,EAAE,cAAc,aAAa;AACtC,IAAE,YAAY,MAAM,UAAU;AAAA,IAC5B,IAAc,CAAC,GAAG,GAAG,GAAG,GAAG,GAAG,CAAC;AAAA,IAC/B,YAAY,MAAiB;AAAE,UAAI,KAAM,MAAK,IAAI;AAAA,IAAK;AAAA,EACzD;AACF;AAEA,IAAI,OAAO,EAAE,WAAW,aAAa;AACnC,IAAE,SAAS,MAAM,OAAO;AAAA,EAAC;AAC3B;AAKA,EAAE,cAAc;;;ACChB,SAAS,aAAa,2BAA2B;AACjD,SAAS,qBAAqB;AAC9B,SAAS,SAAS,YAAY;AAQ9B,oBAAoB,YAAY;AAGhC,IAAM,YAAY;AAClB,IAAM,iBAAiB,MAAM,OAAO;AAEpC,IAAM,sBAAsB;AAK5B,IAAM,cAAwF,CAAC;AAC/F,IAAI;AACF,QAAM,WAAW,cAAc,YAAY,GAAG;AAC9C,QAAM,SAAS,QAAQ,SAAS,QAAQ,yBAAyB,CAAC;AAClE,cAAY,UAAU,KAAK,QAAQ,OAAO,IAAI;AAC9C,cAAY,aAAa;AACzB,cAAY,sBAAsB,KAAK,QAAQ,gBAAgB,IAAI;AACrE,QAAQ;AAA6C;AAGrD,eAAe,mBAAmB,QAAqB;AACrD,QAAM,cAAc,YAAY;AAAA,IAC9B,MAAM,IAAI,WAAW,MAAM;AAAA,IAC3B,gBAAgB;AAAA,IAChB,iBAAiB;AAAA,IACjB,iBAAiB;AAAA,IACjB,GAAG;AAAA,EACL,CAAC;AACD,MAAI;AACJ,MAAI;AACF,WAAO,MAAM,QAAQ,KAAK;AAAA,MACxB,YAAY;AAAA,MACZ,IAAI,QAAe,CAAC,GAAG,WAAW;AAChC,gBAAQ,WAAW,MAAM;AAAE,sBAAY,QAAQ;AAAG,iBAAO,IAAI,YAAY,mEAAsB,CAAC;AAAA,QAAE,GAAG,mBAAmB;AAAA,MAC1H,CAAC;AAAA,IACH,CAAC;AAAA,EACH,UAAE;AACA,QAAI,UAAU,OAAW,cAAa,KAAK;AAAA,EAC7C;AACF;AAEA,eAAsB,iBAAiB,QAAqB,SAAsD;AAGhH,QAAM,gBAAoC,SAAS,aAAa,OAAO,MAAM,CAAC,IAAI;AAClF,QAAM,YAAgC,SAAS,MAAM,OAAO,MAAM,CAAC,IAAI;AACvE,QAAM,MAAM,MAAM,mBAAmB,MAAM;AAE3C,MAAI;AACF,UAAM,YAAY,IAAI;AACtB,QAAI,cAAc,EAAG,OAAM,IAAI,YAAY,8DAAiB;AAG5D,UAAM,WAA6B,EAAE,UAAU;AAC/C,UAAM,mBAAmB,KAAK,QAAQ;AAEtC,UAAM,SAAoB,CAAC;AAC3B,UAAM,WAA2B,CAAC;AAClC,UAAM,cAA6B,CAAC;AACpC,QAAI,aAAa;AACjB,QAAI,iBAAiB;AACrB,UAAM,qBAAqB,KAAK,IAAI,WAAW,SAAS;AAGxD,UAAM,aAAa,SAAS,QAAQ,eAAe,QAAQ,OAAO,kBAAkB,IAAI;AACxF,UAAM,cAAc,aAAa,WAAW,OAAO;AAGnD,UAAM,eAAe,oBAAI,IAAoB;AAC7C,UAAM,cAAc,oBAAI,IAAoB;AAE5C,UAAM,sBAAsB,oBAAI,IAAY;AAE5C,UAAM,oBAAoB,oBAAI,IAAoB;AAIlD,UAAM,aAAa,oBAAoB;AACvC,UAAM,kBAAoC,CAAC;AAC3C,UAAM,kBAAkB,oBAAI,IAAuB;AAEnD,QAAI,cAAc;AAClB,aAAS,IAAI,GAAG,KAAK,oBAAoB,KAAK;AAC5C,UAAI,cAAc,CAAC,WAAW,IAAI,CAAC,EAAG;AACtC,UAAI;AACF,cAAM,OAAO,MAAM,IAAI,QAAQ,CAAC;AAChC,cAAM,KAAK,MAAM,KAAK,eAAe;AACrC,cAAM,WAAW,KAAK,YAAY,EAAE,OAAO,EAAE,CAAC;AAC9C,oBAAY,IAAI,GAAG,SAAS,MAAM;AAClC,cAAM,WAAW,GAAG;AACpB,cAAM,QAAQ,eAAe,QAAQ;AAGrC,cAAM,EAAE,SAAS,YAAY,IAAI,iBAAiB,OAAO,SAAS,OAAO,SAAS,MAAM;AACxF,YAAI,cAAc,GAAG;AACnB,mBAAS,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,WAAW,sFAAqB,MAAM,uBAAuB,CAAC;AAAA,QACrG;AAGA,mBAAW,QAAQ,SAAS;AAC1B,cAAI,KAAK,WAAW,EAAG,cAAa,IAAI,KAAK,WAAW,aAAa,IAAI,KAAK,QAAQ,KAAK,KAAK,CAAC;AAAA,QACnG;AAGA,cAAM,SAAS,MAAM,KAAK,gBAAgB;AAG1C,cAAM,WAAW,SAAS,QAAQ,SAAS;AAC3C,YAAI,WAAW,GAAG;AAChB,gBAAM,eAAe,oBAAoB,OAAO,SAAS,OAAO,SAAS;AACzE,cAAI,YAAY;AAChB,qBAAW,KAAK,cAAc;AAC5B,kBAAM,QAAQ,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE;AACvC,gBAAI,OAAO,WAAW,KAAM;AAC5B,gCAAoB,IAAI,CAAC;AACzB,kBAAM,UAAU,QAAQ,KAAK,QAAM;AACjC,oBAAM,KAAK,GAAG,IAAI,GAAG,IAAI;AACzB,oBAAM,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,YAAY;AAC1C,qBAAO,MAAM,EAAE,MAAM,MAAM,EAAE,MAAM,MAAM,EAAE,MAAM,MAAM,EAAE;AAAA,YAC3D,CAAC;AACD,gBAAI,CAAC,QAAS;AAAA,UAChB;AACA,cAAI,YAAY,EAAG,mBAAkB,IAAI,GAAG,SAAS;AAAA,QACvD;AAEA,cAAM,aAAa,2BAA2B,SAAS,GAAG,QAAQ,SAAS,OAAO,SAAS,MAAM;AACjG,mBAAW,KAAK,WAAY,QAAO,KAAK,CAAC;AAGzC,YAAI;AACF,gBAAM,EAAE,QAAQ,WAAW,QAAQ,WAAW,IAAI,MAAM,kBAAkB,MAAM,OAAO,SAAS,OAAO,WAAW,GAAG,YAAY,QAAQ;AACzI,cAAI,UAAU,SAAS,EAAG,iBAAgB,IAAI,GAAG,SAAS;AAC1D,0BAAgB,KAAK,GAAG,UAAU;AAAA,QACpC,QAAQ;AAAA,QAAkC;AAG1C,YAAI,WAAW;AACf,mBAAW,KAAK,YAAY;AAC1B,cAAI,IAAI,EAAE,QAAQ;AAIlB,cAAI,EAAE,SAAS,WAAW,EAAE,OAAO;AACjC,kBAAM,WAAW,EAAE,MAAM,MAAM,IAAI,SAAO,IAAI,IAAI,OAAK,EAAE,IAAI,EAAE,KAAK,GAAG,CAAC,EAAE,KAAK,IAAI;AACnF,gBAAI,IAAI,IAAI,OAAO,WAAW;AAAA,UAChC;AACA,wBAAc,EAAE,QAAQ,OAAO,EAAE,EAAE;AACnC,4BAAkB,EAAE,SAAS;AAC7B,sBAAY,WAAW,OAAO,IAAI;AAAA,QACpC;AACA,oBAAY,KAAK,mBAAmB,GAAG,QAAQ,CAAC;AAChD,YAAI,iBAAiB,eAAgB,OAAM,IAAI,YAAY,2DAAc;AACzE;AACA,iBAAS,aAAa,aAAa,WAAW;AAAA,MAChD,SAAS,SAAS;AAEhB,YAAI,mBAAmB,YAAa,OAAM;AAC1C,iBAAS,KAAK,EAAE,MAAM,GAAG,SAAS,sBAAO,CAAC,+BAAW,mBAAmB,QAAQ,QAAQ,UAAU,yCAAW,IAAI,MAAM,gBAAgB,CAAC;AAAA,MAC1I;AAAA,IACF;AAEA,UAAM,kBAAkB,gBAAgB,aAAa,WAAW,OAAO;AAEvE,UAAM,eAAe,aAAa,KAAK,IAAI,iBAAiB,CAAC,IAAI;AAMjE,UAAM,UAAU,oBAAI,IAAY;AAChC,QAAI,SAAS,OAAO,WAAW;AAC7B,YAAM,UAAU,CAAC,MAAc,CAAC,cAAc,WAAW,IAAI,CAAC;AAC9D,YAAM,UAAU,oBAAI,IAAY;AAChC,UAAI,QAAQ,QAAQ,WAAW,cAAc;AAC3C,iBAAS,IAAI,GAAG,KAAK,oBAAoB,IAAK,KAAI,QAAQ,CAAC,EAAG,SAAQ,IAAI,CAAC;AAAA,MAC7E,OAAO;AACL,mBAAW,MAAM,aAAa;AAC5B,cAAI,CAAC,GAAG,SAAU;AAElB,cAAI,GAAG,cAAc,cAAc,CAAC,oBAAoB,IAAI,GAAG,IAAI,EAAG;AACtE,kBAAQ,IAAI,GAAG,IAAI;AAAA,QACrB;AAAA,MACF;AACA,UAAI,QAAQ,OAAO,GAAG;AACpB,YAAI;AACF,gBAAM,EAAE,UAAU,IAAI,MAAM,OAAO,uBAAmB;AACtD,gBAAM,OAAO,OAAO,QAAQ,QAAQ,aAAa,QAAQ,MAAO;AAChE,gBAAM,gBAAgB,MAAM,UAAU,WAAW,SAAS,MAAM,UAAU,QAAQ,UAAU;AAC5F,cAAI,cAAc,OAAO,GAAG;AAE1B,uBAAW,KAAK,cAAc,KAAK,EAAG,SAAQ,IAAI,CAAC;AACnD,kBAAM,SAAS,OAAO,OAAO,OAAK,EAAE,EAAE,cAAc,QAAQ,IAAI,EAAE,UAAU,EAAE;AAC9E,uBAAW,OAAO,cAAc,OAAO,EAAG,QAAO,KAAK,GAAG,GAAG;AAC5D,mBAAO,KAAK,CAAC,GAAG,OAAO,EAAE,cAAc,MAAM,EAAE,cAAc,EAAE;AAC/D,mBAAO,SAAS;AAChB,mBAAO,KAAK,GAAG,MAAM;AACrB,uBAAW,MAAM,YAAa,KAAI,QAAQ,IAAI,GAAG,IAAI,EAAG,IAAG,aAAa;AACxE,qBAAS,KAAK;AAAA,cACZ,SAAS,GAAG,QAAQ,IAAI,qDAAkB,SAAS,YAAY,0BAAgB,mDAAW;AAAA,cAC1F,MAAM;AAAA,YACR,CAAC;AAAA,UACH;AAAA,QACF,SAAS,GAAG;AAEV,mBAAS,KAAK;AAAA,YACZ,SAAS,kCAAc,aAAa,QAAQ,EAAE,UAAU,OAAO,CAAC,CAAC;AAAA,YACjE,MAAM;AAAA,UACR,CAAC;AAAA,QACH;AAAA,MACF;AAAA,IACF;AAEA,QAAI,gBAAgB,QAAQ,SAAS,GAAG;AAEtC,eAAS,KAAK;AAAA,QACZ,SAAS,wCAAe,SAAS,0CAAY,UAAU;AAAA,QACvD,MAAM;AAAA,MACR,CAAC;AAAA,IACH;AAKA,QAAI,CAAC,cAAc;AACjB,YAAM,sBAA8C;AAAA,QAClD,UAAU;AAAA,QACV,UAAU;AAAA,QACV,cAAc;AAAA,QACd,kBAAkB;AAAA,QAClB,gBAAgB;AAAA,MAClB;AACA,iBAAW,MAAM,aAAa;AAC5B,YAAI,CAAC,GAAG,YAAY,CAAC,GAAG,aAAa,GAAG,WAAY;AACpD,YAAI,GAAG,cAAc,cAAc,CAAC,oBAAoB,IAAI,GAAG,IAAI,EAAG;AACtE,iBAAS,KAAK,EAAE,MAAM,GAAG,MAAM,SAAS,GAAG,oBAAoB,GAAG,SAAS,CAAC,yCAAgB,MAAM,YAAY,CAAC;AAAA,MACjH;AAAA,IACF;AAIA,QAAI,CAAC,cAAc;AACjB,iBAAW,CAAC,MAAM,KAAK,KAAK,CAAC,GAAG,kBAAkB,QAAQ,CAAC,EAAE,KAAK,CAAC,GAAG,MAAM,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG;AACxF,iBAAS,KAAK,EAAE,MAAM,SAAS,GAAG,KAAK,gMAA+C,MAAM,gBAAgB,CAAC;AAAA,MAC/G;AAAA,IACF;AAGA,QAAI,SAAS,uBAAuB,SAAS,mBAAmB,GAAG;AACjE,YAAM,UAAU,yBAAyB,QAAQ,aAAa,QAAQ;AAEtE,eAAS,KAAK,QAAQ,SAAS,GAAG,MAAM,GAAG,MAAM;AAC/C,eAAO,OAAO,QAAQ,EAAE,GAAG,CAAC;AAAA,MAC9B;AAAA,IACF;AAIA,yBAAqB,MAAM;AAG3B,0BAAsB,QAAQ,eAAe;AAI7C,QAAI,SAAS,cAAc,eAAe;AACxC,UAAI;AACF,cAAM,gBAAgB,eAAe,QAAQ,YAAY,oBAAoB,UAAU,QAAQ,UAAU;AAAA,MAC3G,SAAS,GAAG;AACV,iBAAS,KAAK;AAAA,UACZ,SAAS,kCAAc,aAAa,QAAQ,EAAE,UAAU,OAAO,CAAC,CAAC;AAAA,UACjE,MAAM;AAAA,QACR,CAAC;AAAA,MACH;AAAA,IACF;AAGA,UAAM,iBAAiB,8BAA8B,YAAY;AACjE,QAAI,iBAAiB,GAAG;AACtB,qBAAe,QAAQ,cAAc;AAAA,IACvC;AAGA,yBAAqB,MAAM;AAG3B,wBAAoB,MAAM;AAG1B,2BAAuB,MAAM;AAG7B,UAAM,UAAyB,OAC5B,OAAO,OAAK,EAAE,SAAS,aAAa,EAAE,SAAS,EAAE,IAAI,EACrD,IAAI,QAAM,EAAE,OAAO,EAAE,OAAQ,MAAM,EAAE,MAAO,YAAY,EAAE,WAAW,EAAE;AAG1E,8BAA0B,MAAM;AAGhC,QAAI,WAAW,aAAa,iBAAiB,MAAM,CAAC;AAEpD,WAAO;AAAA,MACL;AAAA,MACA;AAAA,MACA;AAAA,MACA,SAAS,QAAQ,SAAS,IAAI,UAAU;AAAA,MACxC,UAAU,SAAS,SAAS,IAAI,WAAW;AAAA,MAC3C,cAAc,gBAAgB;AAAA,MAC9B;AAAA,MACA,gBAAgB,yBAAyB,WAAW;AAAA,MACpD,QAAQ,gBAAgB,SAAS,IAAI,kBAAkB;AAAA,IACzD;AAAA,EACF,UAAE;AACA,UAAM,IAAI,QAAQ,EAAE,MAAM,MAAM;AAAA,IAAC,CAAC;AAAA,EACpC;AACF;AAIA,eAAe,mBAAmB,KAA0C,UAA2C;AACrH,MAAI;AACF,UAAM,SAAS,MAAM,IAAI,YAAY;AACrC,QAAI,CAAC,QAAQ,KAAM;AACnB,UAAM,OAAO,OAAO;AAEpB,QAAI,OAAO,KAAK,UAAU,YAAY,KAAK,MAAM,KAAK,EAAG,UAAS,QAAQ,KAAK,MAAM,KAAK;AAC1F,QAAI,OAAO,KAAK,WAAW,YAAY,KAAK,OAAO,KAAK,EAAG,UAAS,SAAS,KAAK,OAAO,KAAK;AAC9F,QAAI,OAAO,KAAK,YAAY,YAAY,KAAK,QAAQ,KAAK,EAAG,UAAS,UAAU,KAAK,QAAQ,KAAK;AAClG,QAAI,OAAO,KAAK,YAAY,YAAY,KAAK,QAAQ,KAAK,EAAG,UAAS,cAAc,KAAK,QAAQ,KAAK;AACtG,QAAI,OAAO,KAAK,aAAa,YAAY,KAAK,SAAS,KAAK,GAAG;AAC7D,eAAS,WAAW,KAAK,SAAS,MAAM,MAAM,EAAE,IAAI,CAAC,MAAc,EAAE,KAAK,CAAC,EAAE,OAAO,OAAO;AAAA,IAC7F;AACA,QAAI,OAAO,KAAK,iBAAiB,SAAU,UAAS,YAAY,aAAa,KAAK,YAAY;AAC9F,QAAI,OAAO,KAAK,YAAY,SAAU,UAAS,aAAa,aAAa,KAAK,OAAO;AAAA,EACvF,QAAQ;AAAA,EAER;AACF;AAGA,SAAS,aAAa,SAAqC;AACzD,QAAM,IAAI,QAAQ,MAAM,mDAAmD;AAC3E,MAAI,CAAC,EAAG,QAAO;AACf,QAAM,CAAC,EAAE,MAAM,QAAQ,MAAM,MAAM,MAAM,OAAO,MAAM,MAAM,MAAM,MAAM,IAAI,IAAI;AAChF,SAAO,GAAG,IAAI,IAAI,KAAK,IAAI,GAAG,IAAI,IAAI,IAAI,GAAG,IAAI,GAAG;AACtD;AAGA,eAAsB,uBAAuB,QAAgD;AAC3F,QAAM,MAAM,MAAM,mBAAmB,MAAM;AAE3C,MAAI;AACF,UAAM,WAA6B,EAAE,WAAW,IAAI,SAAS;AAC7D,UAAM,mBAAmB,KAAK,QAAQ;AACtC,WAAO;AAAA,EACT,UAAE;AACA,UAAM,IAAI,QAAQ,EAAE,MAAM,MAAM;AAAA,IAAC,CAAC;AAAA,EACpC;AACF;","names":[]}
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/pdf/image-extract.ts","../src/pdf/quality.ts","../src/pdf/text-clean.ts","../src/pdf/formula-ocr.ts","../src/pdf/polyfill.ts","../src/pdf/parser.ts"],"sourcesContent":["/**\r\n * PDF 이미지 XObject 바이트 추출 — 페이지 operatorList의 paint 이미지를\r\n * page.objs/commonObjs에서 디코딩된 픽셀로 받아 PNG로 인코딩한다.\r\n *\r\n * getOperatorList()가 이미 디코딩 비용을 지불하므로(선 감지용으로 항상 호출)\r\n * 여기서는 RGBA 변환 + PNG deflate만 추가된다. 순수 JS (node:zlib) 전용.\r\n */\r\n\r\nimport { OPS, ImageKind } from \"pdfjs-dist/legacy/build/pdf.mjs\"\r\nimport { encodePng } from \"../image/transcode.js\"\r\nimport type { ExtractedImage, IRBlock, ParseWarning } from \"../types.js\"\r\n\r\n/** 디코딩된 이미지 픽셀 (pdfjs page.objs 반환 형태의 최소 부분) */\r\ninterface PdfImgData {\r\n width: number\r\n height: number\r\n kind?: number\r\n data?: Uint8Array | Uint8ClampedArray\r\n}\r\n\r\n/** 페이지 프록시의 최소 형태 — objs/commonObjs 만 사용 */\r\ninterface PdfObjects {\r\n get(id: string, callback?: (data: unknown) => void): unknown\r\n}\r\ninterface PdfPageObjs {\r\n objs: PdfObjects\r\n commonObjs: PdfObjects\r\n}\r\n\r\n/** 개별 이미지 디코딩 대기 상한 — 실패 시에도 null 로 resolve 되므로 안전망 성격 */\r\nconst IMAGE_RESOLVE_TIMEOUT_MS = 5_000\r\n\r\n/**\r\n * 이미지 객체 대기 해제 — 디코딩은 getOperatorList() resolve 후에도 워커에서\r\n * 비동기로 진행되므로(pdfjs buildPaintImageXObject 는 await 하지 않음) 동기\r\n * has()/get() 은 이미 완료된 것만 잡힌다. 콜백형 get()으로 완료를 기다리며,\r\n * 디코딩 실패는 pdfjs 가 null 로 resolve 하므로 무한 대기는 없다.\r\n */\r\nfunction resolveImgData(page: PdfPageObjs, objId: string): Promise<PdfImgData | null> {\r\n // canvas.js 규약과 동일 — \"g_\" 접두사는 문서 공용 객체\r\n const store = objId.startsWith(\"g_\") ? page.commonObjs : page.objs\r\n return new Promise(resolve => {\r\n let done = false\r\n const timer = setTimeout(() => { if (!done) { done = true; resolve(null) } }, IMAGE_RESOLVE_TIMEOUT_MS)\r\n try {\r\n store.get(objId, (data: unknown) => {\r\n if (!done) { done = true; clearTimeout(timer); resolve((data ?? null) as PdfImgData | null) }\r\n })\r\n } catch {\r\n if (!done) { done = true; clearTimeout(timer); resolve(null) }\r\n }\r\n })\r\n}\r\n\r\n/** 폭·높이 하한 — 괘선/불릿 등 장식 조각 제외 */\r\nconst MIN_DIM = 8\r\n/** 총 픽셀 상한 — transcode.ts MAX_PIXELS와 동일 근거 (deflateSync 동기 블로킹 방지) */\r\nconst MAX_PIXELS = 36_000_000\r\n/** 문서당 추출 이미지 수 상한 */\r\nconst MAX_IMAGES_PER_DOC = 200\r\n/** 문서당 PNG 누적 바이트 상한 (128MB) */\r\nconst MAX_TOTAL_IMAGE_BYTES = 128 * 1024 * 1024\r\n\r\n/** 문서 단위 추출 상태 — 페이지 간 중복(로고·워터마크) 억제 + 상한 추적 */\r\nexport interface PdfImageState {\r\n imageIndex: number\r\n totalBytes: number\r\n /** 내용 해시 → 파일명 (이전 페이지에서 이미 추출된 이미지) */\r\n seen: Map<string, string>\r\n capWarned: boolean\r\n}\r\n\r\nexport function createPdfImageState(): PdfImageState {\r\n return { imageIndex: 0, totalBytes: 0, seen: new Map(), capWarned: false }\r\n}\r\n\r\n/** FNV-1a 32bit — 픽셀 버퍼 내용 해시 (crypto 불필요, 충돌은 dims 결합으로 완화) */\r\nfunction fnv1a(data: Uint8Array | Uint8ClampedArray): number {\r\n let h = 0x811c9dc5\r\n for (let i = 0; i < data.length; i++) {\r\n h ^= data[i]\r\n h = Math.imul(h, 0x01000193)\r\n }\r\n return h >>> 0\r\n}\r\n\r\n/** kind별 픽셀 → 8bit RGBA. 미지원/불일치 형태는 null. */\r\nfunction toRgba(img: PdfImgData): Uint8Array | null {\r\n const { width: w, height: h, kind, data } = img\r\n if (!data || !w || !h) return null\r\n const rgba = new Uint8Array(w * h * 4)\r\n\r\n if (kind === ImageKind.RGBA_32BPP) {\r\n if (data.length < w * h * 4) return null\r\n rgba.set(data.subarray(0, w * h * 4))\r\n return rgba\r\n }\r\n if (kind === ImageKind.RGB_24BPP) {\r\n if (data.length < w * h * 3) return null\r\n for (let i = 0, s = 0, d = 0; i < w * h; i++, s += 3, d += 4) {\r\n rgba[d] = data[s]; rgba[d + 1] = data[s + 1]; rgba[d + 2] = data[s + 2]; rgba[d + 3] = 255\r\n }\r\n return rgba\r\n }\r\n if (kind === ImageKind.GRAYSCALE_1BPP) {\r\n // 행 단위 비트 패킹 (stride = ceil(w/8)), 1=백 0=흑\r\n const stride = (w + 7) >> 3\r\n if (data.length < stride * h) return null\r\n for (let y = 0; y < h; y++) {\r\n for (let x = 0; x < w; x++) {\r\n const bit = (data[y * stride + (x >> 3)] >> (7 - (x & 7))) & 1\r\n const v = bit ? 255 : 0\r\n const d = (y * w + x) * 4\r\n rgba[d] = v; rgba[d + 1] = v; rgba[d + 2] = v; rgba[d + 3] = 255\r\n }\r\n }\r\n return rgba\r\n }\r\n return null\r\n}\r\n\r\n/**\r\n * 한 페이지의 operatorList에서 이미지를 추출해 image 블록·바이너리로 반환.\r\n * 같은 페이지 내 반복 paint(타일링)는 1회만, 이전 페이지와 동일 내용(로고·\r\n * 워터마크)은 블록·바이너리 모두 재방출하지 않는다.\r\n */\r\nexport async function extractPageImages(\r\n page: PdfPageObjs,\r\n fnArray: Uint32Array | number[],\r\n argsArray: unknown[][],\r\n pageNumber: number,\r\n state: PdfImageState,\r\n warnings: ParseWarning[],\r\n): Promise<{ blocks: IRBlock[]; images: ExtractedImage[] }> {\r\n const blocks: IRBlock[] = []\r\n const images: ExtractedImage[] = []\r\n const pageSeenIds = new Set<string>()\r\n\r\n for (let i = 0; i < fnArray.length; i++) {\r\n const op = fnArray[i]\r\n let imgData: PdfImgData | null = null\r\n let dedupeId: string | null = null\r\n\r\n if (op === OPS.paintImageXObject || op === OPS.paintImageXObjectRepeat) {\r\n const objId = argsArray[i]?.[0]\r\n if (typeof objId !== \"string\" || pageSeenIds.has(objId)) continue\r\n pageSeenIds.add(objId)\r\n dedupeId = objId\r\n } else if (op !== OPS.paintInlineImageXObject) {\r\n continue\r\n }\r\n\r\n // 상한 도달 검사 — resolveImgData(이미지당 최대 5초 대기) 앞에서 중단해야\r\n // 상한 이후 이미지들이 페이지마다 디코딩 대기 시간만 소모하지 않는다\r\n if (state.imageIndex >= MAX_IMAGES_PER_DOC || state.totalBytes >= MAX_TOTAL_IMAGE_BYTES) {\r\n if (!state.capWarned) {\r\n state.capWarned = true\r\n warnings.push({ page: pageNumber, message: `이미지 추출 상한 도달 (${MAX_IMAGES_PER_DOC}개/128MB) — 이후 이미지 생략`, code: \"SKIPPED_IMAGE\" })\r\n }\r\n continue\r\n }\r\n\r\n if (dedupeId) {\r\n imgData = await resolveImgData(page, dedupeId)\r\n } else {\r\n imgData = argsArray[i]?.[0] as PdfImgData\r\n }\r\n\r\n if (!imgData?.data || !imgData.width || !imgData.height) continue\r\n const { width: w, height: h } = imgData\r\n if (w < MIN_DIM || h < MIN_DIM) continue // 괘선/불릿 장식\r\n if (w * h > MAX_PIXELS) {\r\n warnings.push({ page: pageNumber, message: `이미지 크기 초과로 추출 생략 (${w}×${h})`, code: \"SKIPPED_IMAGE\" })\r\n continue\r\n }\r\n\r\n // 페이지 간 내용 중복 — 로고·워터마크 재추출 방지\r\n const hash = `${w}x${h}k${imgData.kind ?? \"?\"}h${fnv1a(imgData.data)}${dedupeId ? \"\" : \"inline\"}`\r\n if (state.seen.has(hash)) continue\r\n\r\n const rgba = toRgba(imgData)\r\n if (!rgba) continue // 미지원 픽셀 형태 (bitmap 전용 등)\r\n\r\n const png = encodePng(w, h, rgba)\r\n state.imageIndex++\r\n state.totalBytes += png.length\r\n const filename = `image_${String(state.imageIndex).padStart(3, \"0\")}.png`\r\n state.seen.set(hash, filename)\r\n images.push({ filename, data: png, mimeType: \"image/png\" })\r\n blocks.push({ type: \"image\", text: filename, pageNumber })\r\n }\r\n\r\n return { blocks, images }\r\n}\r\n\r\n/**\r\n * 페이지별 image 블록을 본문 블록 열에 주입 — 각 페이지의 마지막 블록 뒤.\r\n * 페이지 루프 중 바로 끼워 넣으면 페이지 경계에서 표 블록 인접성이 깨져\r\n * mergeCrossPageTables가 무산되므로, 병합이 끝난 뒤 호출해야 한다.\r\n */\r\nexport function injectPageImageBlocks(blocks: IRBlock[], pageImages: Map<number, IRBlock[]>): void {\r\n if (pageImages.size === 0) return\r\n\r\n // 페이지별 마지막 블록 위치 → 위치별 페이지 역인덱스\r\n const lastIndexForPage = new Map<number, number>()\r\n for (let i = 0; i < blocks.length; i++) {\r\n const p = blocks[i].pageNumber\r\n if (p !== undefined) lastIndexForPage.set(p, i)\r\n }\r\n const pageAtIndex = new Map<number, number>()\r\n for (const [p, last] of lastIndexForPage) pageAtIndex.set(last, p)\r\n\r\n const result: IRBlock[] = []\r\n const injected = new Set<number>()\r\n for (let i = 0; i < blocks.length; i++) {\r\n result.push(blocks[i])\r\n const p = pageAtIndex.get(i)\r\n if (p !== undefined && pageImages.has(p)) {\r\n result.push(...pageImages.get(p)!)\r\n injected.add(p)\r\n }\r\n }\r\n // 텍스트 블록이 하나도 없는 페이지(전면 이미지 등) — 페이지 순서 위치에 삽입\r\n for (const p of [...pageImages.keys()].sort((a, b) => a - b)) {\r\n if (injected.has(p)) continue\r\n let at = result.length\r\n for (let i = 0; i < result.length; i++) {\r\n const bp = result[i].pageNumber\r\n if (bp !== undefined && bp > p) { at = i; break }\r\n }\r\n result.splice(at, 0, ...pageImages.get(p)!)\r\n }\r\n\r\n blocks.length = 0\r\n blocks.push(...result)\r\n}\r\n","/**\r\n * PDF 페이지별 텍스트 품질 신호 수집.\r\n *\r\n * pdfjs가 텍스트층을 추출했더라도, 폰트의 ToUnicode/CMap이 불완전하면\r\n * 한글이 PUA(Private Use Area) 글리프 코드로 그대로 떨어진다. 또 일부 PDF는\r\n * NUL/제어문자가 본문에 섞여 있다. 본 모듈은 그런 신호를 페이지 단위로 수집해\r\n * \"OCR 검토 필요\" 판정에 사용한다.\r\n *\r\n * 더 까다로운 케이스: ToUnicode가 \"잘못\" 매핑되면 글리프가 PUA/FFFD가 아니라\r\n * 정상 한글 음절 영역(0xAC00-0xD7A3)의 엉뚱한 글자로 떨어진다(\"GPU쭒컫 많핂슪\").\r\n * 이건 PUA/제어/대체문자 신호로 못 잡는다. 대신 자소를 분해해 종성(받침) 분포를\r\n * 본다 — CID 스크램블은 받침을 균등하게 흩뿌리므로, 자연 한국어(받침없음 다수 +\r\n * 겹받침 희소)와 통계적으로 갈린다. 이것이 garbled_hangul 신호다.\r\n */\r\n\r\nexport interface PageQuality {\r\n /** 1-based 페이지 번호 */\r\n page: number\r\n /** 공백 제외 문자 수 */\r\n textChars: number\r\n /** 한글 음절(0xAC00-0xD7A3) 비율 (0~1) */\r\n hangulRatio: number\r\n /** C0/C1 제어문자 비율 (tab/newline/CR 제외) */\r\n controlCharRatio: number\r\n /** U+FFFD replacement character 비율 */\r\n replacementCharRatio: number\r\n /** PUA 비율 — 글꼴 매핑 실패의 핵심 신호 */\r\n puaRatio: number\r\n /** 한글 음절 중 받침 없는 음절 비율 (자연 한국어 ~0.5, mojibake ~0.04) */\r\n hangulNoBatchimRatio: number\r\n /** 한글 음절 중 겹받침/희귀 받침 비율 (자연 한국어 <0.06, mojibake ~0.5) */\r\n hangulRareBatchimRatio: number\r\n /** OCR 검토 권장 여부 (pdfjs 텍스트층 기준 원신호 — OCR 적용 후에도 보존) */\r\n needsOcr: boolean\r\n /** needsOcr=true일 때 사유 (단일 신호로 충분, 가장 강한 신호 선택) */\r\n ocrReason?: \"low_text\" | \"high_pua\" | \"high_control\" | \"high_replacement\" | \"garbled_hangul\"\r\n /** 이 페이지에 OCR 이 실제 적용되어 본문이 OCR 결과로 대체됨 */\r\n ocrApplied?: boolean\r\n}\r\n\r\n/**\r\n * 희귀 종성(받침) 인덱스 집합. 종성 인덱스 = (code - 0xAC00) % 28,\r\n * 0 = 받침 없음. 겹받침(ㄳㄵㄶㄺㄻㄼㄽㄾㄿㅀㅄ) + 거의 안 쓰는 홑받침(ㅋㅌㅍ).\r\n * 자연 한국어에서 이들의 합산 비율은 매우 낮다(<6%).\r\n */\r\nconst RARE_JONG = new Set<number>([3, 5, 6, 9, 10, 11, 12, 13, 14, 15, 18, 24, 25, 26])\r\n\r\n/** garbled_hangul 판정 최소 한글 음절 수 — 통계적 유의성 확보 */\r\nconst MOJIBAKE_MIN_HANGUL = 30\r\n/** 받침 없는 음절 비율이 이 값 미만이면 비정상 (자연 한국어는 훨씬 높음) */\r\nconst MOJIBAKE_MAX_NOBATCHIM = 0.15\r\n/** 희귀 받침 비율이 이 값 이상이면 비정상 (자연 한국어는 훨씬 낮음) */\r\nconst MOJIBAKE_MIN_RAREBATCHIM = 0.25\r\n\r\n/** 페이지 텍스트에서 품질 메트릭을 계산한다. */\r\nexport function computePageQuality(page: number, text: string): PageQuality {\r\n let total = 0\r\n let hangul = 0\r\n let hangulNoBatchim = 0\r\n let hangulRareBatchim = 0\r\n let control = 0\r\n let replacement = 0\r\n let pua = 0\r\n\r\n for (let i = 0; i < text.length; i++) {\r\n const code = text.charCodeAt(i)\r\n // 공백류는 분모에서 제외\r\n if (code === 0x20 || code === 0x09 || code === 0x0a || code === 0x0d) continue\r\n total++\r\n\r\n // C0 제어문자 (0x00-0x1F 중 tab/lf/cr 제외) + DEL(0x7F) + C1(0x80-0x9F)\r\n if ((code < 0x20) || code === 0x7f || (code >= 0x80 && code <= 0x9f)) {\r\n control++\r\n continue\r\n }\r\n if (code === 0xfffd) {\r\n replacement++\r\n continue\r\n }\r\n // 한글 음절 — 종성(받침) 분포로 mojibake 판정\r\n if (code >= 0xac00 && code <= 0xd7a3) {\r\n hangul++\r\n const jong = (code - 0xac00) % 28\r\n if (jong === 0) hangulNoBatchim++\r\n else if (RARE_JONG.has(jong)) hangulRareBatchim++\r\n continue\r\n }\r\n // PUA: BMP(E000-F8FF) + SPUA-A(F0000-FFFFD) + SPUA-B(100000-10FFFD)\r\n // 서로게이트 페어는 high만 검사해도 충분 (Plane 15/16 high surrogate 범위 DB80-DBFF)\r\n if ((code >= 0xe000 && code <= 0xf8ff) || (code >= 0xdb80 && code <= 0xdbff)) {\r\n pua++\r\n continue\r\n }\r\n }\r\n\r\n const denom = total || 1\r\n const puaRatio = pua / denom\r\n const controlCharRatio = control / denom\r\n const replacementCharRatio = replacement / denom\r\n const hangulNoBatchimRatio = hangul > 0 ? hangulNoBatchim / hangul : 0\r\n const hangulRareBatchimRatio = hangul > 0 ? hangulRareBatchim / hangul : 0\r\n\r\n // 받침 분포 이상 = ToUnicode 오매핑 mojibake. 받침없음 희소 + 희귀받침 과다를\r\n // 둘 다(AND) 만족해야 발화 — 자연 한국어(받침없음 다수)에서 오탐 방지.\r\n const garbledHangul =\r\n hangul >= MOJIBAKE_MIN_HANGUL &&\r\n hangulNoBatchimRatio < MOJIBAKE_MAX_NOBATCHIM &&\r\n hangulRareBatchimRatio >= MOJIBAKE_MIN_RAREBATCHIM\r\n\r\n let needsOcr = false\r\n let ocrReason: PageQuality[\"ocrReason\"] | undefined\r\n // 우선순위: low_text > high_pua > high_control > high_replacement > garbled_hangul\r\n if (total < LOW_TEXT_THRESHOLD) { needsOcr = true; ocrReason = \"low_text\" }\r\n else if (puaRatio >= HIGH_PUA_THRESHOLD) { needsOcr = true; ocrReason = \"high_pua\" }\r\n else if (controlCharRatio >= HIGH_CONTROL_THRESHOLD) { needsOcr = true; ocrReason = \"high_control\" }\r\n else if (replacementCharRatio >= HIGH_REPLACEMENT_THRESHOLD) { needsOcr = true; ocrReason = \"high_replacement\" }\r\n else if (garbledHangul) { needsOcr = true; ocrReason = \"garbled_hangul\" }\r\n\r\n return {\r\n page,\r\n textChars: total,\r\n hangulRatio: hangul / denom,\r\n controlCharRatio,\r\n replacementCharRatio,\r\n puaRatio,\r\n hangulNoBatchimRatio,\r\n hangulRareBatchimRatio,\r\n needsOcr,\r\n ocrReason,\r\n }\r\n}\r\n\r\n/** 페이지별 품질에서 문서 단위 요약을 계산한다. */\r\nexport interface DocumentQualitySummary {\r\n totalPages: number\r\n totalTextChars: number\r\n avgHangulRatio: number\r\n avgControlCharRatio: number\r\n avgReplacementCharRatio: number\r\n avgPuaRatio: number\r\n /** textChars 매우 적은 페이지 수 (이미지/빈 페이지 후보) */\r\n lowTextPageCount: number\r\n /** PUA 비율 임계 초과 페이지 수 (글꼴 매핑 깨짐 후보) */\r\n highPuaPageCount: number\r\n /** 문서 전체에 OCR 검토가 권장되는지 — needsOcr 페이지 비율 또는 평균 신호 기반 */\r\n needsOcr: boolean\r\n /** needsOcr=true인 페이지 번호 목록 */\r\n ocrCandidatePages: number[]\r\n}\r\n\r\nconst LOW_TEXT_THRESHOLD = 20\r\nconst HIGH_PUA_THRESHOLD = 0.2\r\nconst HIGH_CONTROL_THRESHOLD = 0.05\r\nconst HIGH_REPLACEMENT_THRESHOLD = 0.05\r\n/** 페이지 중 needsOcr 비율이 이 값 이상이면 문서 전체에 OCR 권장 */\r\nconst DOC_NEEDS_OCR_PAGE_RATIO = 0.3\r\n\r\n/**\r\n * 비표시 제어문자를 제거한다. C0(NUL 등, tab/lf/cr 제외) + DEL + C1.\r\n * PUA는 사용자가 글꼴 매핑 실패를 시각적으로 확인할 수 있도록 보존.\r\n */\r\nexport function stripControlChars(text: string): string {\r\n // eslint-disable-next-line no-control-regex\r\n return text.replace(/[\\x00-\\x08\\x0B\\x0C\\x0E-\\x1F\\x7F\\x80-\\x9F]/g, \"\")\r\n}\r\n\r\nexport function summarizeDocumentQuality(pages: PageQuality[]): DocumentQualitySummary {\r\n if (pages.length === 0) {\r\n return {\r\n totalPages: 0,\r\n totalTextChars: 0,\r\n avgHangulRatio: 0,\r\n avgControlCharRatio: 0,\r\n avgReplacementCharRatio: 0,\r\n avgPuaRatio: 0,\r\n lowTextPageCount: 0,\r\n highPuaPageCount: 0,\r\n needsOcr: false,\r\n ocrCandidatePages: [],\r\n }\r\n }\r\n\r\n let textChars = 0\r\n let hangul = 0\r\n let control = 0\r\n let replacement = 0\r\n let pua = 0\r\n let lowText = 0\r\n let highPua = 0\r\n const ocrCandidatePages: number[] = []\r\n\r\n for (const p of pages) {\r\n textChars += p.textChars\r\n hangul += p.hangulRatio\r\n control += p.controlCharRatio\r\n replacement += p.replacementCharRatio\r\n pua += p.puaRatio\r\n if (p.textChars < LOW_TEXT_THRESHOLD) lowText++\r\n if (p.puaRatio >= HIGH_PUA_THRESHOLD) highPua++\r\n // OCR 이 이미 적용된 페이지는 후보에서 제외 — \"OCR 했는데 또 하라\" 자기모순 방지\r\n if (p.needsOcr && !p.ocrApplied) ocrCandidatePages.push(p.page)\r\n }\r\n\r\n const n = pages.length\r\n return {\r\n totalPages: n,\r\n totalTextChars: textChars,\r\n avgHangulRatio: hangul / n,\r\n avgControlCharRatio: control / n,\r\n avgReplacementCharRatio: replacement / n,\r\n avgPuaRatio: pua / n,\r\n lowTextPageCount: lowText,\r\n highPuaPageCount: highPua,\r\n needsOcr: ocrCandidatePages.length / n >= DOC_NEEDS_OCR_PAGE_RATIO,\r\n ocrCandidatePages,\r\n }\r\n}\r\n","/**\r\n * PDF 마크다운 최종 정리.\r\n *\r\n * 페이지 번호 제거, 균등배분 후처리, 취소선 복원, 한글 줄바꿈 병합.\r\n * blocksToMarkdown 이후의 문자열 수준 후처리를 담당한다.\r\n */\r\n\r\nimport type { IRBlock } from \"../types.js\"\r\nimport { stripControlChars } from \"./quality.js\"\r\nimport { collapseEvenSpacing } from \"./text-line.js\"\r\n\r\n/** 블록 트리의 텍스트에서 비표시 제어문자를 in-place로 제거한다. */\r\nexport function sanitizeBlockControlChars(blocks: IRBlock[]): void {\r\n for (const b of blocks) {\r\n if (b.text) b.text = stripControlChars(b.text)\r\n if (b.table) {\r\n for (const row of b.table.cells) {\r\n for (const cell of row) {\r\n if (cell.text) cell.text = stripControlChars(cell.text)\r\n }\r\n }\r\n }\r\n if (b.children) sanitizeBlockControlChars(b.children)\r\n }\r\n}\r\n\r\nexport function cleanPdfText(text: string): string {\r\n return mergeKoreanLines(\r\n stripControlChars(text)\r\n // 문서 시작 단독 페이지 번호\r\n .replace(/^\\d{1,4}\\n/, \"\")\r\n // \"- 2 -\" 스타일 페이지 번호 (독립 라인 및 목록 항목 형태 포함)\r\n .replace(/^[\\s]*[-–—]\\s*[-–—]?\\d+[-–—]?[\\s]*[-–—]?[\\s]*$/gm, \"\")\r\n // \"1 / 5\" 스타일 페이지 번호\r\n .replace(/^\\s*\\d+\\s*\\/\\s*\\d+\\s*$/gm, \"\")\r\n // 단독 페이지 번호 (줄 끝에 혼자 있는 숫자)\r\n .replace(/\\n\\d{1,4}\\n/g, \"\\n\")\r\n // 문서 마지막 단독 페이지 번호\r\n .replace(/\\n\\d{1,4}$/, \"\")\r\n // 단독 숫자 헤딩 제거 (\"# 6\\n재무과\" → \"\\n재무과\")\r\n .replace(/^#{1,6}\\s*\\d{1,4}\\s*$/gm, \"\")\r\n )\r\n // 균등배분 문자열 후처리 (pdfjs가 합친 TextItem + buildGridTable 셀 텍스트)\r\n // LaTeX 수식 라인 ($...$ / $$...$$) 은 공백이 토큰 구분자라 collapse 시 `\\cdot d` → `\\cdotd` 로 망가짐 — skip\r\n .replace(/^(?!\\| ---).*$/gm, line => {\r\n if (/^\\s*\\${1,2}.+\\${1,2}\\s*$/.test(line)) return line\r\n return collapseEvenSpacing(line)\r\n })\r\n // 마커 뒤 2글자 균등배분 합침 (\"□ 일 시\" → \"□ 일시\", \"□ 장 소\" → \"□ 장소\")\r\n .replace(/([□■◆○●▶ㅇ])\\s+([가-힣])\\s+([가-힣])/g, \"$1 $2$3\")\r\n // 취소선 복원: builder escapeGfm이 ~를 \\~로 이스케이프 — 쌍(~~)만 되살림\r\n .replace(/\\\\~\\\\~/g, \"~~\")\r\n // 인접 취소선 run이 붙어 생긴 빈 마크(~~~~) 정리\r\n .replace(/~~~~/g, \"\")\r\n .replace(/\\n{3,}/g, \"\\n\\n\")\r\n .trim()\r\n}\r\n\r\nfunction startsWithMarker(line: string): boolean {\r\n const t = line.trimStart()\r\n return /^[가-힣ㄱ-ㅎ][.)]/.test(t) || /^\\d+[.)]/.test(t) || /^\\([가-힣ㄱ-ㅎ\\d]+\\)/.test(t) ||\r\n /^[○●※▶▷◆◇■□★☆\\-·]\\s/.test(t) || /^제\\d+[조항호장절]/.test(t)\r\n}\r\n\r\nfunction isStandaloneHeader(line: string): boolean {\r\n return /^제\\d+[조항호장절](\\([^)]*\\))?(\\s+\\S+){0,7}$/.test(line.trim())\r\n}\r\n\r\nfunction mergeKoreanLines(text: string): string {\r\n if (!text) return \"\"\r\n const lines = text.split(\"\\n\")\r\n if (lines.length <= 1) return text\r\n const result: string[] = [lines[0]]\r\n\r\n for (let i = 1; i < lines.length; i++) {\r\n const prev = result[result.length - 1]\r\n const curr = lines[i]\r\n const currTrimmed = curr.trim()\r\n // 마크다운 헤딩/테이블/구분선은 병합하지 않음\r\n if (/^#{1,6}\\s/.test(prev) || /^#{1,6}\\s/.test(curr) || /^\\|/.test(currTrimmed) || /^---/.test(currTrimmed)) {\r\n result.push(curr)\r\n continue\r\n }\r\n // 쉼표로 끝나는 줄 + 다음 줄 = 연속 문장\r\n if (/,$/.test(prev.trim()) && currTrimmed.length > 0) {\r\n result[result.length - 1] = prev + \"\\n\" + curr\r\n continue\r\n }\r\n // (※ 로 시작하는 줄 = 이전 줄의 부연설명\r\n if (/^\\(※/.test(currTrimmed)) {\r\n result[result.length - 1] = prev + \" \" + currTrimmed\r\n continue\r\n }\r\n // 한글 줄바꿈 병합 — 마커(○, □ 등)로 시작하는 이전 줄은 합치지 않음\r\n if (/[가-힣·,\\-]$/.test(prev) && /^[가-힣(]/.test(curr) &&\r\n !startsWithMarker(curr) && !isStandaloneHeader(prev) &&\r\n !startsWithMarker(prev)) {\r\n result[result.length - 1] = prev + \" \" + curr\r\n } else {\r\n result.push(curr)\r\n }\r\n }\r\n return result.join(\"\\n\")\r\n}\r\n","/**\r\n * 수식 OCR 통합 (optional)\r\n *\r\n * 페이지 이미지 렌더 후 수식만 검출/인식해 blocks 에 formula paragraph 를 삽입.\r\n * 모델은 최초 실행 시 자동 다운로드 (./formula/index.js 에 위임).\r\n */\r\n\r\nimport type { IRBlock, ParseWarning } from \"../types.js\"\r\n\r\n/**\r\n * 수식 OCR 을 적용하여 blocks 에 formula paragraph 를 삽입한다.\r\n *\r\n * 좌표 매핑:\r\n * - pdfium 픽셀 bbox (top-left origin) → PDF 포인트 (bottom-left origin) 변환\r\n * - 수식 bbox 의 y center 와 같은 페이지 내 pdfjs block 의 y center 비교로 삽입 위치 결정\r\n * - pdfjs 가 이미 뽑은 수식 흔적(block) 과 겹치면 해당 block 제거 (중복 방지)\r\n *\r\n * 실패/trivial 수식(latex === \"\") 은 삽입하지 않는다.\r\n */\r\nexport async function applyFormulaOcr(\r\n buffer: ArrayBuffer,\r\n blocks: IRBlock[],\r\n pageFilter: Set<number> | null,\r\n effectivePageCount: number,\r\n warnings: ParseWarning[],\r\n _onProgress?: (current: number, total: number) => void,\r\n): Promise<void> {\r\n const formulaMod = await import(\"./formula/index.js\")\r\n const { FormulaPipeline, ensureFormulaModels } = formulaMod\r\n\r\n // 모델 준비 — 없으면 자동 다운로드. 진행률은 stderr 로 출력.\r\n await ensureFormulaModels((p) => {\r\n if (p.phase === \"download\" && p.total) {\r\n const pct = Math.floor((p.downloaded / p.total) * 100)\r\n process.stderr.write(`\\r[kordoc-formula] ${p.spec.name} ${pct}% (${formatMb(p.downloaded)}/${formatMb(p.total)})`)\r\n if (p.downloaded >= p.total) process.stderr.write(\"\\n\")\r\n } else if (p.phase === \"verify\") {\r\n process.stderr.write(`[kordoc-formula] ${p.spec.name} SHA-256 검증 중...\\n`)\r\n } else if (p.phase === \"done\") {\r\n process.stderr.write(`[kordoc-formula] ${p.spec.name} 준비 완료\\n`)\r\n } else if (p.phase === \"skip\") {\r\n // 조용히 스킵\r\n }\r\n })\r\n\r\n const pipeline = await FormulaPipeline.create()\r\n try {\r\n // MAX_PAGES 상한을 수식 OCR 에도 적용 — 필터 없으면 텍스트 추출과 같은 범위만\r\n // (없으면 대형 PDF 에서 상한 밖 페이지까지 ONNX 추론 + 결과가 문서 끝에 오배치)\r\n const effectiveFilter = pageFilter\r\n ?? new Set(Array.from({ length: effectivePageCount }, (_, i) => i + 1))\r\n const pagesResult = await pipeline.runOnBuffer(buffer, effectiveFilter)\r\n\r\n if (pagesResult.length === 0) return\r\n\r\n let insertedCount = 0\r\n let removedDupCount = 0\r\n\r\n for (const page of pagesResult) {\r\n const pageNumber = page.pageNumber\r\n const pdfHeight = page.pdfHeight\r\n const scaleX = page.renderedWidth > 0 ? page.pdfWidth / page.renderedWidth : 0.5\r\n const scaleY = page.renderedHeight > 0 ? page.pdfHeight / page.renderedHeight : 0.5\r\n\r\n // 1) 수식 → (PDF 포인트 bbox, latex) 정규화 + trivial 제외\r\n interface FormulaCandidate {\r\n block: IRBlock\r\n pdfBbox: { x1: number; x2: number; yTop: number; yBottom: number }\r\n centerY: number // PDF bottom-up\r\n }\r\n const candidates: FormulaCandidate[] = []\r\n for (const r of page.regions) {\r\n if (!r.latex || !r.latex.trim()) continue\r\n const wrapped = r.kind === \"display\" ? `$$${r.latex}$$` : `$${r.latex}$`\r\n\r\n const x1 = r.bbox.x1 * scaleX\r\n const x2 = r.bbox.x2 * scaleX\r\n // pdfium 픽셀 y → PDF bottom-up\r\n const yTop = pdfHeight - r.bbox.y1 * scaleY\r\n const yBottom = pdfHeight - r.bbox.y2 * scaleY\r\n const centerY = (yTop + yBottom) / 2\r\n const width = x2 - x1\r\n const height = yTop - yBottom\r\n\r\n candidates.push({\r\n block: {\r\n type: \"paragraph\",\r\n text: wrapped,\r\n pageNumber,\r\n bbox: { page: pageNumber, x: x1, y: yBottom, width, height },\r\n },\r\n pdfBbox: { x1, x2, yTop, yBottom },\r\n centerY,\r\n })\r\n }\r\n if (candidates.length === 0) continue\r\n\r\n // 2) 같은 페이지의 pdfjs block 중 수식 bbox 와 크게 겹치는 것 제거\r\n // (pdfjs 가 수식을 텍스트로 파편 추출한 경우 — overlap ratio ≥ 0.6 이면 중복으로 간주)\r\n const OVERLAP_THRESHOLD = 0.6\r\n const indicesToRemove = new Set<number>()\r\n for (let i = 0; i < blocks.length; i++) {\r\n const b = blocks[i]\r\n if (b.pageNumber !== pageNumber) continue\r\n if (b.type === \"table\") continue // 표는 건드리지 않음\r\n if (!b.bbox || b.bbox.width <= 0 || b.bbox.height <= 0) continue\r\n const blockArea = b.bbox.width * b.bbox.height\r\n if (blockArea <= 0) continue\r\n\r\n for (const c of candidates) {\r\n const ox1 = Math.max(b.bbox.x, c.pdfBbox.x1)\r\n const ox2 = Math.min(b.bbox.x + b.bbox.width, c.pdfBbox.x2)\r\n const oy1 = Math.max(b.bbox.y, c.pdfBbox.yBottom)\r\n const oy2 = Math.min(b.bbox.y + b.bbox.height, c.pdfBbox.yTop)\r\n const interArea = Math.max(0, ox2 - ox1) * Math.max(0, oy2 - oy1)\r\n if (interArea / blockArea >= OVERLAP_THRESHOLD) {\r\n indicesToRemove.add(i)\r\n break\r\n }\r\n }\r\n }\r\n\r\n if (indicesToRemove.size > 0) {\r\n // 내림차순으로 제거해야 인덱스가 밀리지 않음\r\n const sorted = [...indicesToRemove].sort((a, b) => b - a)\r\n for (const idx of sorted) blocks.splice(idx, 1)\r\n removedDupCount += indicesToRemove.size\r\n }\r\n\r\n // 3) 각 수식을 y 좌표 기준 적절한 위치에 삽입\r\n // 수식들을 위→아래(centerY 큰 것부터) 정렬 후, 각 수식마다 현재 blocks 에서\r\n // \"center y < 수식 center y\" 인 첫 블록(= 수식보다 아래) 앞에 삽입.\r\n candidates.sort((a, b) => b.centerY - a.centerY)\r\n\r\n for (const c of candidates) {\r\n let insertIdx = -1\r\n let pageFirstIdx = -1\r\n let pageLastIdx = -1\r\n for (let i = 0; i < blocks.length; i++) {\r\n const b = blocks[i]\r\n if (b.pageNumber !== pageNumber) continue\r\n if (pageFirstIdx === -1) pageFirstIdx = i\r\n pageLastIdx = i\r\n if (!b.bbox) continue\r\n const blockCenter = b.bbox.y + b.bbox.height / 2\r\n if (blockCenter < c.centerY) {\r\n insertIdx = i\r\n break\r\n }\r\n }\r\n\r\n if (insertIdx !== -1) {\r\n blocks.splice(insertIdx, 0, c.block)\r\n } else if (pageLastIdx !== -1) {\r\n blocks.splice(pageLastIdx + 1, 0, c.block)\r\n } else {\r\n // 해당 페이지에 텍스트 블록 없음 — 맨 끝에 추가\r\n blocks.push(c.block)\r\n }\r\n insertedCount++\r\n }\r\n }\r\n\r\n if (insertedCount > 0 || removedDupCount > 0) {\r\n process.stderr.write(\r\n `[kordoc-formula] ${insertedCount}개 수식 삽입, ${removedDupCount}개 중복 block 제거 (${pagesResult.length}개 페이지)\\n`,\r\n )\r\n }\r\n } finally {\r\n await pipeline.destroy().catch(() => {})\r\n }\r\n}\r\n\r\nfunction formatMb(bytes: number): string {\r\n return `${(bytes / 1024 / 1024).toFixed(1)}MB`\r\n}\r\n","/**\n * 서버리스/Node.js 환경에서 pdfjs-dist가 요구하는 브라우저 API polyfill.\n * pdfjs-dist import보다 먼저 실행되어야 함 (ES 모듈 호이스팅 대응으로 별도 파일 분리).\n *\n * 1. DOMMatrix / Path2D polyfill — pdfjs-dist가 참조하지만 Node.js에 없음\n * 2. pdfjsWorker 사전 주입 — fake worker의 동적 import를 우회\n */\n\n// eslint-disable-next-line @typescript-eslint/no-explicit-any\nconst g = globalThis as any\n\nif (typeof g.DOMMatrix === \"undefined\") {\n g.DOMMatrix = class DOMMatrix {\n m: number[] = [1, 0, 0, 1, 0, 0]\n constructor(init?: number[]) { if (init) this.m = init }\n }\n}\n\nif (typeof g.Path2D === \"undefined\") {\n g.Path2D = class Path2D {}\n}\n\n// worker 모듈 static import → fake worker가 동적 import를 건너뜀\n// @ts-expect-error pdfjs-dist worker has no type declarations\nimport * as pdfjsWorker from \"pdfjs-dist/legacy/build/pdf.worker.mjs\"\ng.pdfjsWorker = pdfjsWorker\n","/**\r\n * PDF 텍스트 추출 (pdfjs-dist static import 기반)\r\n *\r\n * polyfill을 먼저 import해야 DOMMatrix/Path2D/pdfjsWorker가 주입됨.\r\n * ES 모듈 호이스팅 때문에 별도 파일로 분리되어 있음.\r\n *\r\n * 이 파일은 엔트리(파이프라인 오케스트레이션)와 메타데이터 추출만 담당한다.\r\n * 세부 단계는 모듈로 분리: text-line(아이템/줄), xy-cut(읽기 순서),\r\n * columns(열 감지), page-blocks(블록 추출), block-detect(후처리 감지),\r\n * text-clean(마크다운 정리), formula-ocr(수식).\r\n */\r\n\r\nimport type { InternalParseResult, IRBlock, DocumentMetadata, ExtractedImage, ParseOptions, ParseWarning, OutlineItem } from \"../types.js\"\r\nimport { KordocError } from \"../utils.js\"\r\nimport { parsePageRange } from \"../page-range.js\"\r\nimport { blocksToMarkdown } from \"../table/builder.js\"\r\nimport { extractImageRegions } from \"./line-detector.js\"\r\nimport { createPdfImageState, extractPageImages, injectPageImageBlocks } from \"./image-extract.js\"\r\nimport { computePageQuality, summarizeDocumentQuality, type PageQuality } from \"./quality.js\"\r\nimport { type PdfTextItem, normalizeItems, filterHiddenText } from \"./text-line.js\"\r\nimport { extractPageBlocksWithLines, mergeCrossPageTables } from \"./page-blocks.js\"\r\nimport { computeMedianFontSizeFromFreq, detectHeadings, detectMarkerHeadings, detectTableCaptions, detectKoreanListBlocks, removeHeaderFooterBlocks } from \"./block-detect.js\"\r\nimport { sanitizeBlockControlChars, cleanPdfText } from \"./text-clean.js\"\r\nimport { applyFormulaOcr } from \"./formula-ocr.js\"\r\n// polyfill 먼저 (ES 모듈 호이스팅되므로 별도 파일 필수)\r\nimport \"./polyfill.js\"\r\nimport { getDocument, GlobalWorkerOptions } from \"pdfjs-dist/legacy/build/pdf.mjs\"\r\nimport { createRequire } from \"node:module\"\r\nimport { dirname, join } from \"node:path\"\r\n\r\n// 기존 공개 API 경로 유지 — 이동된 함수의 re-export\r\nexport { mergeCrossPageTables }\r\nexport { cleanPdfText }\r\nexport { detectTableCaptions, detectKoreanListBlocks, removeHeaderFooterBlocks }\r\n\r\n// worker 비활성화 (polyfill에서 pdfjsWorker를 이미 주입했으므로)\r\nGlobalWorkerOptions.workerSrc = \"\"\r\n\r\n// ─── 안전 한계값 (구조적 파싱과 무관) ────────────────\r\nconst MAX_PAGES = 5000\r\nconst MAX_TOTAL_TEXT = 100 * 1024 * 1024 // 100MB\r\n/** PDF 로딩 타임아웃 (30초) — 악성/대용량 PDF 무한 대기 방지 */\r\nconst PDF_LOAD_TIMEOUT_MS = 30_000\r\n\r\n// CID 폰트 자산(cmaps/standard_fonts) 경로 — 미지정이면 CMap 필요 폰트의 텍스트가\r\n// \"loadFont failed: cMapUrl 필요\" 경고와 함께 통째로 소실된다 (성과계획서류 숫자 전멸).\r\n// pdfjs-dist 패키지 위치에서 해석하고, 실패 시 미지정(기존 동작) 유지.\r\nconst pdfjsAssets: { cMapUrl?: string; cMapPacked?: boolean; standardFontDataUrl?: string } = {}\r\ntry {\r\n const _require = createRequire(import.meta.url)\r\n const pkgDir = dirname(_require.resolve(\"pdfjs-dist/package.json\"))\r\n pdfjsAssets.cMapUrl = join(pkgDir, \"cmaps\") + \"/\"\r\n pdfjsAssets.cMapPacked = true\r\n pdfjsAssets.standardFontDataUrl = join(pkgDir, \"standard_fonts\") + \"/\"\r\n} catch { /* optional dep — 경로 해석 실패 시 cMap 없이 진행 */ }\r\n\r\n/** getDocument + 타임아웃 래퍼 */\r\nasync function loadPdfWithTimeout(buffer: ArrayBuffer) {\r\n const loadingTask = getDocument({\r\n data: new Uint8Array(buffer),\r\n useSystemFonts: true,\r\n disableFontFace: true,\r\n isEvalSupported: false,\r\n ...pdfjsAssets,\r\n })\r\n let timer: ReturnType<typeof setTimeout> | undefined\r\n try {\r\n return await Promise.race([\r\n loadingTask.promise,\r\n new Promise<never>((_, reject) => {\r\n timer = setTimeout(() => { loadingTask.destroy(); reject(new KordocError(\"PDF 로딩 타임아웃 (30초 초과)\")) }, PDF_LOAD_TIMEOUT_MS)\r\n }),\r\n ])\r\n } finally {\r\n if (timer !== undefined) clearTimeout(timer)\r\n }\r\n}\r\n\r\nexport async function parsePdfDocument(buffer: ArrayBuffer, options?: ParseOptions): Promise<InternalParseResult> {\r\n // pdfjs-dist 는 전달받은 buffer 의 underlying storage 를 detach 할 수 있다.\r\n // 수식/텍스트 OCR 은 같은 버퍼를 pdfium 으로 재사용해야 하므로 옵션 on 일 때만 clone 을 보관.\r\n const formulaBuffer: ArrayBuffer | null = options?.formulaOcr ? buffer.slice(0) : null\r\n const ocrBuffer: ArrayBuffer | null = options?.ocr ? buffer.slice(0) : null\r\n const doc = await loadPdfWithTimeout(buffer)\r\n\r\n try {\r\n const pageCount = doc.numPages\r\n if (pageCount === 0) throw new KordocError(\"PDF에 페이지가 없습니다.\")\r\n\r\n // 메타데이터 추출 (best-effort)\r\n const metadata: DocumentMetadata = { pageCount }\r\n await extractPdfMetadata(doc, metadata)\r\n\r\n const blocks: IRBlock[] = []\r\n const warnings: ParseWarning[] = []\r\n const pageQuality: PageQuality[] = []\r\n let totalChars = 0\r\n let totalTextBytes = 0\r\n const effectivePageCount = Math.min(pageCount, MAX_PAGES)\r\n\r\n // 페이지 범위 필터링\r\n const pageFilter = options?.pages ? parsePageRange(options.pages, effectivePageCount) : null\r\n const totalTarget = pageFilter ? pageFilter.size : effectivePageCount\r\n\r\n // 전체 문서의 폰트 크기 빈도 수집 (헤딩 감지용) — 빈도 Map으로 메모리 절약\r\n const fontSizeFreq = new Map<number, number>()\r\n const pageHeights = new Map<number, number>()\r\n // 큰 이미지가 있는 페이지 (needsOcr 경고 노이즈 필터 + SKIPPED_IMAGE)\r\n const pagesWithLargeImage = new Set<number>()\r\n // 텍스트 없는 큰 이미지 영역: page → count\r\n const skippedImagePages = new Map<number, number>()\r\n // 이미지 XObject 바이트 추출 상태 (문서 단위 중복 억제·상한).\r\n // image 블록은 페이지 경계 표 병합(mergeCrossPageTables)의 인접성을 깨지 않도록\r\n // 페이지별로 모아뒀다가 병합 후 주입한다.\r\n const imageState = createPdfImageState()\r\n const extractedImages: ExtractedImage[] = []\r\n const pageImageBlocks = new Map<number, IRBlock[]>()\r\n\r\n let parsedPages = 0\r\n for (let i = 1; i <= effectivePageCount; i++) {\r\n if (pageFilter && !pageFilter.has(i)) continue\r\n try {\r\n const page = await doc.getPage(i)\r\n const tc = await page.getTextContent()\r\n const viewport = page.getViewport({ scale: 1 })\r\n pageHeights.set(i, viewport.height)\r\n const rawItems = tc.items as PdfTextItem[]\r\n const items = normalizeItems(rawItems)\r\n\r\n // hidden text 필터링 + 경고 수집\r\n const { visible, hiddenCount } = filterHiddenText(items, viewport.width, viewport.height)\r\n if (hiddenCount > 0) {\r\n warnings.push({ page: i, message: `${hiddenCount}개 숨겨진 텍스트 요소 필터링됨`, code: \"HIDDEN_TEXT_FILTERED\" })\r\n }\r\n\r\n // 폰트 크기 빈도 수집\r\n for (const item of visible) {\r\n if (item.fontSize > 0) fontSizeFreq.set(item.fontSize, (fontSizeFreq.get(item.fontSize) || 0) + 1)\r\n }\r\n\r\n // 선 기반 테이블 감지를 위한 operatorList\r\n const opList = await page.getOperatorList()\r\n\r\n // 이미지 영역 감지 — 텍스트 없는 큰 이미지는 무음 정보손실이므로 가시화 (ODL 아이디어)\r\n const pageArea = viewport.width * viewport.height\r\n if (pageArea > 0) {\r\n const imageRegions = extractImageRegions(opList.fnArray, opList.argsArray)\r\n let uncovered = 0\r\n for (const r of imageRegions) {\r\n const area = (r.x2 - r.x1) * (r.y2 - r.y1)\r\n if (area < pageArea * 0.05) continue // 작은 장식 이미지 무시\r\n pagesWithLargeImage.add(i)\r\n const hasText = visible.some(it => {\r\n const cx = it.x + it.w / 2\r\n const cy = it.y + (it.h || it.fontSize) / 2\r\n return cx >= r.x1 && cx <= r.x2 && cy >= r.y1 && cy <= r.y2\r\n })\r\n if (!hasText) uncovered++\r\n }\r\n if (uncovered > 0) skippedImagePages.set(i, uncovered)\r\n }\r\n\r\n const pageBlocks = extractPageBlocksWithLines(visible, i, opList, viewport.width, viewport.height)\r\n for (const b of pageBlocks) blocks.push(b)\r\n\r\n // 이미지 XObject 바이트 추출 — 블록 주입은 표 병합 후(injectPageImageBlocks)\r\n try {\r\n const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings)\r\n if (imgBlocks.length > 0) pageImageBlocks.set(i, imgBlocks)\r\n extractedImages.push(...pageImages)\r\n } catch { /* 이미지 추출 실패가 텍스트 파싱을 막지 않도록 */ }\r\n\r\n // 이미지 기반 PDF 감지 + 크기 제한용 문자 수 집계 + 페이지 품질 신호\r\n let pageText = \"\"\r\n for (const b of pageBlocks) {\r\n let t = b.text || \"\"\r\n // 표 블록은 text 없이 table만 가지므로 셀 텍스트도 집계 — 괘선 양식(표-전용)\r\n // 문서가 totalChars=0으로 이미지 기반 오판되어 정상 파싱 표가 OCR로\r\n // 대체되는 것 방지. 페이지 품질(computePageQuality) 집계도 같은 뿌리.\r\n if (b.type === \"table\" && b.table) {\r\n const cellText = b.table.cells.map(row => row.map(c => c.text).join(\" \")).join(\"\\n\")\r\n t = t ? t + \"\\n\" + cellText : cellText\r\n }\r\n totalChars += t.replace(/\\s/g, \"\").length\r\n totalTextBytes += t.length * 2\r\n pageText += pageText ? \"\\n\" + t : t\r\n }\r\n pageQuality.push(computePageQuality(i, pageText))\r\n if (totalTextBytes > MAX_TOTAL_TEXT) throw new KordocError(\"텍스트 추출 크기 초과\")\r\n parsedPages++\r\n options?.onProgress?.(parsedPages, totalTarget)\r\n } catch (pageErr) {\r\n // 크기 초과는 전체 중단\r\n if (pageErr instanceof KordocError) throw pageErr\r\n warnings.push({ page: i, message: `페이지 ${i} 파싱 실패: ${pageErr instanceof Error ? pageErr.message : \"알 수 없는 오류\"}`, code: \"PARTIAL_PARSE\" })\r\n }\r\n }\r\n\r\n const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount)\r\n // 문서 단위 이미지 기반 판정 (평균 10자/페이지 미만 = 텍스트층 부재)\r\n const isImageBased = totalChars / Math.max(parsedPageCount, 1) < 10\r\n\r\n // ── OCR 실행 (옵션) — 페이지 단위 선정·병합 ──\r\n // 대상: \"force\"=전 페이지 / 문서가 이미지 기반=전 페이지 /\r\n // 그 외=품질 신호가 OCR 을 권하는 페이지만 (깨진 텍스트층 포함 — F1,\r\n // 혼합 문서의 스캔 페이지 포함 — F2). 정상 페이지 파싱 결과는 유지 (F3).\r\n const ocrDone = new Set<number>()\r\n if (options?.ocr && ocrBuffer) {\r\n const inScope = (p: number) => !pageFilter || pageFilter.has(p)\r\n const targets = new Set<number>()\r\n if (options.ocr === \"force\" || isImageBased) {\r\n for (let i = 1; i <= effectivePageCount; i++) if (inScope(i)) targets.add(i)\r\n } else {\r\n for (const pq of pageQuality) {\r\n if (!pq.needsOcr) continue\r\n // low_text 는 빈 페이지(표지/간지)일 수 있으므로 큰 이미지가 있는 페이지만\r\n if (pq.ocrReason === \"low_text\" && !pagesWithLargeImage.has(pq.page)) continue\r\n targets.add(pq.page)\r\n }\r\n }\r\n if (targets.size > 0) {\r\n try {\r\n const { runPdfOcr } = await import(\"../ocr/pdf-ocr.js\")\r\n const mode = typeof options.ocr === \"function\" ? options.ocr : (\"builtin\" as const)\r\n const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options.onProgress)\r\n if (ocrPageBlocks.size > 0) {\r\n // 페이지 단위 교체: OCR 성공 페이지의 기존(깨진/빈) 블록 제거 후 병합\r\n for (const p of ocrPageBlocks.keys()) ocrDone.add(p)\r\n const merged = blocks.filter(b => !(b.pageNumber && ocrDone.has(b.pageNumber)))\r\n for (const obs of ocrPageBlocks.values()) merged.push(...obs)\r\n merged.sort((a, b) => (a.pageNumber ?? 0) - (b.pageNumber ?? 0)) // stable — 페이지 내 순서 보존\r\n blocks.length = 0\r\n blocks.push(...merged)\r\n for (const pq of pageQuality) if (ocrDone.has(pq.page)) pq.ocrApplied = true\r\n warnings.push({\r\n message: `${ocrDone.size}개 페이지에 OCR 적용 (${mode === \"builtin\" ? \"내장 PP-OCRv5\" : \"사용자 프로바이더\"})`,\r\n code: \"OCR_APPLIED\",\r\n })\r\n }\r\n } catch (e) {\r\n // 환경 오류(의존성·모델) — 아래 NEEDS_OCR 폴백이 가시화, 원인은 별도 경고로 보존 (F6)\r\n warnings.push({\r\n message: `OCR 실행 불가: ${e instanceof Error ? e.message : String(e)}`,\r\n code: \"OCR_FAILED\",\r\n })\r\n }\r\n }\r\n }\r\n\r\n if (isImageBased && ocrDone.size === 0) {\r\n // OCR 미설정/실패 — 빈 출력을 무경고로 내보내지 않고 경고 + 플래그로 가시화 (v3.0)\r\n warnings.push({\r\n message: `이미지 기반 PDF (${pageCount}페이지, 텍스트 ${totalChars}자) — 텍스트 레이어가 없어 OCR이 필요합니다`,\r\n code: \"NEEDS_OCR\",\r\n })\r\n }\r\n\r\n // 페이지 단위 needsOcr 경고 — 텍스트+스캔 혼합 문서에서 스캔 페이지 무음 손실 방지.\r\n // low_text는 빈 페이지(표지/간지)일 수 있으므로 큰 이미지가 있는 페이지만 경고.\r\n // OCR 이 적용된 페이지는 해소된 것이므로 제외.\r\n if (!isImageBased) {\r\n const OCR_REASON_MESSAGES: Record<string, string> = {\r\n low_text: \"텍스트가 거의 없는 페이지 (스캔/이미지 추정)\",\r\n high_pua: \"글꼴 매핑 실패 (PUA 비율 높음) — 추출 텍스트 신뢰 불가\",\r\n high_control: \"제어문자 비율 높음 — 추출 텍스트 신뢰 불가\",\r\n high_replacement: \"대체문자(U+FFFD) 비율 높음 — 추출 텍스트 신뢰 불가\",\r\n garbled_hangul: \"글꼴 매핑 실패 (한글 자소 분포 이상) — 추출 텍스트가 깨졌을 수 있음\",\r\n }\r\n for (const pq of pageQuality) {\r\n if (!pq.needsOcr || !pq.ocrReason || pq.ocrApplied) continue\r\n if (pq.ocrReason === \"low_text\" && !pagesWithLargeImage.has(pq.page)) continue\r\n warnings.push({ page: pq.page, message: `${OCR_REASON_MESSAGES[pq.ocrReason]} — OCR 검토 필요`, code: \"NEEDS_OCR\" })\r\n }\r\n }\r\n\r\n // 텍스트 없는 큰 이미지 영역 경고 — 그림/차트/도장 무음 누락 가시화\r\n // (문서 전체가 이미지 기반이면 위의 NEEDS_OCR 단일 경고로 충분)\r\n if (!isImageBased) {\r\n for (const [page, count] of [...skippedImagePages.entries()].sort((a, b) => a[0] - b[0])) {\r\n warnings.push({ page, message: `${count}개 이미지 영역에 추출 가능한 텍스트 없음 (그림/차트/도장 내용 누락 가능)`, code: \"SKIPPED_IMAGE\" })\r\n }\r\n }\r\n\r\n // 머리글/바닥글 필터링 (기본 ON — 명시적 false일 때만 비활성화)\r\n if (options?.removeHeaderFooter !== false && parsedPageCount >= 3) {\r\n const removed = removeHeaderFooterBlocks(blocks, pageHeights, warnings)\r\n // 필터링된 블록 제거 (뒤에서부터 삭제)\r\n for (let ri = removed.length - 1; ri >= 0; ri--) {\r\n blocks.splice(removed[ri], 1)\r\n }\r\n }\r\n\r\n // 페이지 걸친 표 병합 — 머리글/바닥글 제거 후 인접해진 표를 하나로\r\n // (ODL TableBorderProcessor.checkNeighborTables 포팅)\r\n mergeCrossPageTables(blocks)\r\n\r\n // 추출 이미지 참조를 페이지 말미 위치에 주입 (표 병합 뒤 — 인접성 보존)\r\n injectPageImageBlocks(blocks, pageImageBlocks)\r\n\r\n // 수식 OCR (선택) — 기본 텍스트 추출과 별개로 페이지 이미지 렌더 후 수식만 검출/인식.\r\n // 실패 시 경고만 기록하고 일반 텍스트 추출 결과는 그대로 반환한다.\r\n if (options?.formulaOcr && formulaBuffer) {\r\n try {\r\n await applyFormulaOcr(formulaBuffer, blocks, pageFilter, effectivePageCount, warnings, options.onProgress)\r\n } catch (e) {\r\n warnings.push({\r\n message: `수식 OCR 실패: ${e instanceof Error ? e.message : String(e)}`,\r\n code: \"PARTIAL_PARSE\",\r\n })\r\n }\r\n }\r\n\r\n // 헤딩 감지: 폰트 크기 기반\r\n const medianFontSize = computeMedianFontSizeFromFreq(fontSizeFreq)\r\n if (medianFontSize > 0) {\r\n detectHeadings(blocks, medianFontSize)\r\n }\r\n\r\n // □/■ 마커 기반 서브헤딩 감지 (ODL 패턴)\r\n detectMarkerHeadings(blocks)\r\n\r\n // 표 캡션 감지 — 표 직전/직후 '표 N./그림 N' 패턴 텍스트를 IRTable.caption으로\r\n detectTableCaptions(blocks)\r\n\r\n // 한국어 리스트 감지 — 공문서 계층 라벨(1.→가.→1)→가)→①) 시퀀스 검증\r\n detectKoreanListBlocks(blocks)\r\n\r\n // outline 구축\r\n const outline: OutlineItem[] = blocks\r\n .filter(b => b.type === \"heading\" && b.level && b.text)\r\n .map(b => ({ level: b.level!, text: b.text!, pageNumber: b.pageNumber }))\r\n\r\n // 메트릭 수집 끝났으니 블록 텍스트의 C0/C1 제어문자(NUL 등) 정리\r\n sanitizeBlockControlChars(blocks)\r\n\r\n // blocksToMarkdown로 통일 — 헤딩 마크다운 반영 (HWP5/HWPX와 일관성)\r\n let markdown = cleanPdfText(blocksToMarkdown(blocks))\r\n\r\n return {\r\n markdown,\r\n blocks,\r\n metadata,\r\n outline: outline.length > 0 ? outline : undefined,\r\n warnings: warnings.length > 0 ? warnings : undefined,\r\n isImageBased: isImageBased || undefined,\r\n pageQuality,\r\n qualitySummary: summarizeDocumentQuality(pageQuality),\r\n images: extractedImages.length > 0 ? extractedImages : undefined,\r\n }\r\n } finally {\r\n await doc.destroy().catch(() => {})\r\n }\r\n}\r\n\r\n// ─── PDF 메타데이터 추출 ────────────────────────────\r\n\r\nasync function extractPdfMetadata(doc: { getMetadata(): Promise<unknown> }, metadata: DocumentMetadata): Promise<void> {\r\n try {\r\n const result = await doc.getMetadata() as { info?: Record<string, unknown> } | null\r\n if (!result?.info) return\r\n const info = result.info\r\n\r\n if (typeof info.Title === \"string\" && info.Title.trim()) metadata.title = info.Title.trim()\r\n if (typeof info.Author === \"string\" && info.Author.trim()) metadata.author = info.Author.trim()\r\n if (typeof info.Creator === \"string\" && info.Creator.trim()) metadata.creator = info.Creator.trim()\r\n if (typeof info.Subject === \"string\" && info.Subject.trim()) metadata.description = info.Subject.trim()\r\n if (typeof info.Keywords === \"string\" && info.Keywords.trim()) {\r\n metadata.keywords = info.Keywords.split(/[,;]/).map((k: string) => k.trim()).filter(Boolean)\r\n }\r\n if (typeof info.CreationDate === \"string\") metadata.createdAt = parsePdfDate(info.CreationDate)\r\n if (typeof info.ModDate === \"string\") metadata.modifiedAt = parsePdfDate(info.ModDate)\r\n } catch {\r\n // best-effort\r\n }\r\n}\r\n\r\n/** PDF 날짜 형식 (D:YYYYMMDDHHmmSS) → ISO 8601 변환 */\r\nfunction parsePdfDate(dateStr: string): string | undefined {\r\n const m = dateStr.match(/D:(\\d{4})(\\d{2})?(\\d{2})?(\\d{2})?(\\d{2})?(\\d{2})?/)\r\n if (!m) return undefined\r\n const [, year, month = \"01\", day = \"01\", hour = \"00\", min = \"00\", sec = \"00\"] = m\r\n return `${year}-${month}-${day}T${hour}:${min}:${sec}`\r\n}\r\n\r\n/** 메타데이터만 추출 (전체 파싱 없이) — MCP parse_metadata용 */\r\nexport async function extractPdfMetadataOnly(buffer: ArrayBuffer): Promise<DocumentMetadata> {\r\n const doc = await loadPdfWithTimeout(buffer)\r\n\r\n try {\r\n const metadata: DocumentMetadata = { pageCount: doc.numPages }\r\n await extractPdfMetadata(doc, metadata)\r\n return metadata\r\n } finally {\r\n await doc.destroy().catch(() => {})\r\n }\r\n}\r\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;AAQA,SAAS,KAAK,iBAAiB;AAsB/B,IAAM,2BAA2B;AAQjC,SAAS,eAAe,MAAmB,OAA2C;AAEpF,QAAM,QAAQ,MAAM,WAAW,IAAI,IAAI,KAAK,aAAa,KAAK;AAC9D,SAAO,IAAI,QAAQ,aAAW;AAC5B,QAAI,OAAO;AACX,UAAM,QAAQ,WAAW,MAAM;AAAE,UAAI,CAAC,MAAM;AAAE,eAAO;AAAM,gBAAQ,IAAI;AAAA,MAAE;AAAA,IAAE,GAAG,wBAAwB;AACtG,QAAI;AACF,YAAM,IAAI,OAAO,CAAC,SAAkB;AAClC,YAAI,CAAC,MAAM;AAAE,iBAAO;AAAM,uBAAa,KAAK;AAAG,kBAAS,QAAQ,IAA0B;AAAA,QAAE;AAAA,MAC9F,CAAC;AAAA,IACH,QAAQ;AACN,UAAI,CAAC,MAAM;AAAE,eAAO;AAAM,qBAAa,KAAK;AAAG,gBAAQ,IAAI;AAAA,MAAE;AAAA,IAC/D;AAAA,EACF,CAAC;AACH;AAGA,IAAM,UAAU;AAEhB,IAAM,aAAa;AAEnB,IAAM,qBAAqB;AAE3B,IAAM,wBAAwB,MAAM,OAAO;AAWpC,SAAS,sBAAqC;AACnD,SAAO,EAAE,YAAY,GAAG,YAAY,GAAG,MAAM,oBAAI,IAAI,GAAG,WAAW,MAAM;AAC3E;AAGA,SAAS,MAAM,MAA8C;AAC3D,MAAI,IAAI;AACR,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,SAAK,KAAK,CAAC;AACX,QAAI,KAAK,KAAK,GAAG,QAAU;AAAA,EAC7B;AACA,SAAO,MAAM;AACf;AAGA,SAAS,OAAO,KAAoC;AAClD,QAAM,EAAE,OAAO,GAAG,QAAQ,GAAG,MAAM,KAAK,IAAI;AAC5C,MAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAG,QAAO;AAC9B,QAAM,OAAO,IAAI,WAAW,IAAI,IAAI,CAAC;AAErC,MAAI,SAAS,UAAU,YAAY;AACjC,QAAI,KAAK,SAAS,IAAI,IAAI,EAAG,QAAO;AACpC,SAAK,IAAI,KAAK,SAAS,GAAG,IAAI,IAAI,CAAC,CAAC;AACpC,WAAO;AAAA,EACT;AACA,MAAI,SAAS,UAAU,WAAW;AAChC,QAAI,KAAK,SAAS,IAAI,IAAI,EAAG,QAAO;AACpC,aAAS,IAAI,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,IAAI,GAAG,KAAK,KAAK,GAAG,KAAK,GAAG;AAC5D,WAAK,CAAC,IAAI,KAAK,CAAC;AAAG,WAAK,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC;AAAG,WAAK,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC;AAAG,WAAK,IAAI,CAAC,IAAI;AAAA,IACzF;AACA,WAAO;AAAA,EACT;AACA,MAAI,SAAS,UAAU,gBAAgB;AAErC,UAAM,SAAU,IAAI,KAAM;AAC1B,QAAI,KAAK,SAAS,SAAS,EAAG,QAAO;AACrC,aAAS,IAAI,GAAG,IAAI,GAAG,KAAK;AAC1B,eAAS,IAAI,GAAG,IAAI,GAAG,KAAK;AAC1B,cAAM,MAAO,KAAK,IAAI,UAAU,KAAK,EAAE,KAAM,KAAK,IAAI,KAAO;AAC7D,cAAM,IAAI,MAAM,MAAM;AACtB,cAAM,KAAK,IAAI,IAAI,KAAK;AACxB,aAAK,CAAC,IAAI;AAAG,aAAK,IAAI,CAAC,IAAI;AAAG,aAAK,IAAI,CAAC,IAAI;AAAG,aAAK,IAAI,CAAC,IAAI;AAAA,MAC/D;AAAA,IACF;AACA,WAAO;AAAA,EACT;AACA,SAAO;AACT;AAOA,eAAsB,kBACpB,MACA,SACA,WACA,YACA,OACA,UAC0D;AAC1D,QAAM,SAAoB,CAAC;AAC3B,QAAM,SAA2B,CAAC;AAClC,QAAM,cAAc,oBAAI,IAAY;AAEpC,WAAS,IAAI,GAAG,IAAI,QAAQ,QAAQ,KAAK;AACvC,UAAM,KAAK,QAAQ,CAAC;AACpB,QAAI,UAA6B;AACjC,QAAI,WAA0B;AAE9B,QAAI,OAAO,IAAI,qBAAqB,OAAO,IAAI,yBAAyB;AACtE,YAAM,QAAQ,UAAU,CAAC,IAAI,CAAC;AAC9B,UAAI,OAAO,UAAU,YAAY,YAAY,IAAI,KAAK,EAAG;AACzD,kBAAY,IAAI,KAAK;AACrB,iBAAW;AAAA,IACb,WAAW,OAAO,IAAI,yBAAyB;AAC7C;AAAA,IACF;AAIA,QAAI,MAAM,cAAc,sBAAsB,MAAM,cAAc,uBAAuB;AACvF,UAAI,CAAC,MAAM,WAAW;AACpB,cAAM,YAAY;AAClB,iBAAS,KAAK,EAAE,MAAM,YAAY,SAAS,8DAAiB,kBAAkB,qEAAwB,MAAM,gBAAgB,CAAC;AAAA,MAC/H;AACA;AAAA,IACF;AAEA,QAAI,UAAU;AACZ,gBAAU,MAAM,eAAe,MAAM,QAAQ;AAAA,IAC/C,OAAO;AACL,gBAAU,UAAU,CAAC,IAAI,CAAC;AAAA,IAC5B;AAEA,QAAI,CAAC,SAAS,QAAQ,CAAC,QAAQ,SAAS,CAAC,QAAQ,OAAQ;AACzD,UAAM,EAAE,OAAO,GAAG,QAAQ,EAAE,IAAI;AAChC,QAAI,IAAI,WAAW,IAAI,QAAS;AAChC,QAAI,IAAI,IAAI,YAAY;AACtB,eAAS,KAAK,EAAE,MAAM,YAAY,SAAS,iFAAqB,CAAC,OAAI,CAAC,KAAK,MAAM,gBAAgB,CAAC;AAClG;AAAA,IACF;AAGA,UAAM,OAAO,GAAG,CAAC,IAAI,CAAC,IAAI,QAAQ,QAAQ,GAAG,IAAI,MAAM,QAAQ,IAAI,CAAC,GAAG,WAAW,KAAK,QAAQ;AAC/F,QAAI,MAAM,KAAK,IAAI,IAAI,EAAG;AAE1B,UAAM,OAAO,OAAO,OAAO;AAC3B,QAAI,CAAC,KAAM;AAEX,UAAM,MAAM,UAAU,GAAG,GAAG,IAAI;AAChC,UAAM;AACN,UAAM,cAAc,IAAI;AACxB,UAAM,WAAW,SAAS,OAAO,MAAM,UAAU,EAAE,SAAS,GAAG,GAAG,CAAC;AACnE,UAAM,KAAK,IAAI,MAAM,QAAQ;AAC7B,WAAO,KAAK,EAAE,UAAU,MAAM,KAAK,UAAU,YAAY,CAAC;AAC1D,WAAO,KAAK,EAAE,MAAM,SAAS,MAAM,UAAU,WAAW,CAAC;AAAA,EAC3D;AAEA,SAAO,EAAE,QAAQ,OAAO;AAC1B;AAOO,SAAS,sBAAsB,QAAmB,YAA0C;AACjG,MAAI,WAAW,SAAS,EAAG;AAG3B,QAAM,mBAAmB,oBAAI,IAAoB;AACjD,WAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,UAAM,IAAI,OAAO,CAAC,EAAE;AACpB,QAAI,MAAM,OAAW,kBAAiB,IAAI,GAAG,CAAC;AAAA,EAChD;AACA,QAAM,cAAc,oBAAI,IAAoB;AAC5C,aAAW,CAAC,GAAG,IAAI,KAAK,iBAAkB,aAAY,IAAI,MAAM,CAAC;AAEjE,QAAM,SAAoB,CAAC;AAC3B,QAAM,WAAW,oBAAI,IAAY;AACjC,WAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,WAAO,KAAK,OAAO,CAAC,CAAC;AACrB,UAAM,IAAI,YAAY,IAAI,CAAC;AAC3B,QAAI,MAAM,UAAa,WAAW,IAAI,CAAC,GAAG;AACxC,aAAO,KAAK,GAAG,WAAW,IAAI,CAAC,CAAE;AACjC,eAAS,IAAI,CAAC;AAAA,IAChB;AAAA,EACF;AAEA,aAAW,KAAK,CAAC,GAAG,WAAW,KAAK,CAAC,EAAE,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC,GAAG;AAC5D,QAAI,SAAS,IAAI,CAAC,EAAG;AACrB,QAAI,KAAK,OAAO;AAChB,aAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,YAAM,KAAK,OAAO,CAAC,EAAE;AACrB,UAAI,OAAO,UAAa,KAAK,GAAG;AAAE,aAAK;AAAG;AAAA,MAAM;AAAA,IAClD;AACA,WAAO,OAAO,IAAI,GAAG,GAAG,WAAW,IAAI,CAAC,CAAE;AAAA,EAC5C;AAEA,SAAO,SAAS;AAChB,SAAO,KAAK,GAAG,MAAM;AACvB;;;AC9LA,IAAM,YAAY,oBAAI,IAAY,CAAC,GAAG,GAAG,GAAG,GAAG,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,IAAI,EAAE,CAAC;AAGtF,IAAM,sBAAsB;AAE5B,IAAM,yBAAyB;AAE/B,IAAM,2BAA2B;AAG1B,SAAS,mBAAmB,MAAc,MAA2B;AAC1E,MAAI,QAAQ;AACZ,MAAI,SAAS;AACb,MAAI,kBAAkB;AACtB,MAAI,oBAAoB;AACxB,MAAI,UAAU;AACd,MAAI,cAAc;AAClB,MAAI,MAAM;AAEV,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,UAAM,OAAO,KAAK,WAAW,CAAC;AAE9B,QAAI,SAAS,MAAQ,SAAS,KAAQ,SAAS,MAAQ,SAAS,GAAM;AACtE;AAGA,QAAK,OAAO,MAAS,SAAS,OAAS,QAAQ,OAAQ,QAAQ,KAAO;AACpE;AACA;AAAA,IACF;AACA,QAAI,SAAS,OAAQ;AACnB;AACA;AAAA,IACF;AAEA,QAAI,QAAQ,SAAU,QAAQ,OAAQ;AACpC;AACA,YAAM,QAAQ,OAAO,SAAU;AAC/B,UAAI,SAAS,EAAG;AAAA,eACP,UAAU,IAAI,IAAI,EAAG;AAC9B;AAAA,IACF;AAGA,QAAK,QAAQ,SAAU,QAAQ,SAAY,QAAQ,SAAU,QAAQ,OAAS;AAC5E;AACA;AAAA,IACF;AAAA,EACF;AAEA,QAAM,QAAQ,SAAS;AACvB,QAAM,WAAW,MAAM;AACvB,QAAM,mBAAmB,UAAU;AACnC,QAAM,uBAAuB,cAAc;AAC3C,QAAM,uBAAuB,SAAS,IAAI,kBAAkB,SAAS;AACrE,QAAM,yBAAyB,SAAS,IAAI,oBAAoB,SAAS;AAIzE,QAAM,gBACJ,UAAU,uBACV,uBAAuB,0BACvB,0BAA0B;AAE5B,MAAI,WAAW;AACf,MAAI;AAEJ,MAAI,QAAQ,oBAAoB;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAW,WACjE,YAAY,oBAAoB;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAW,WAC1E,oBAAoB,wBAAwB;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAe,WAC1F,wBAAwB,4BAA4B;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAmB,WACtG,eAAe;AAAE,eAAW;AAAM,gBAAY;AAAA,EAAiB;AAExE,SAAO;AAAA,IACL;AAAA,IACA,WAAW;AAAA,IACX,aAAa,SAAS;AAAA,IACtB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACF;AAoBA,IAAM,qBAAqB;AAC3B,IAAM,qBAAqB;AAC3B,IAAM,yBAAyB;AAC/B,IAAM,6BAA6B;AAEnC,IAAM,2BAA2B;AAM1B,SAAS,kBAAkB,MAAsB;AAEtD,SAAO,KAAK,QAAQ,8CAA8C,EAAE;AACtE;AAEO,SAAS,yBAAyB,OAA8C;AACrF,MAAI,MAAM,WAAW,GAAG;AACtB,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,gBAAgB;AAAA,MAChB,gBAAgB;AAAA,MAChB,qBAAqB;AAAA,MACrB,yBAAyB;AAAA,MACzB,aAAa;AAAA,MACb,kBAAkB;AAAA,MAClB,kBAAkB;AAAA,MAClB,UAAU;AAAA,MACV,mBAAmB,CAAC;AAAA,IACtB;AAAA,EACF;AAEA,MAAI,YAAY;AAChB,MAAI,SAAS;AACb,MAAI,UAAU;AACd,MAAI,cAAc;AAClB,MAAI,MAAM;AACV,MAAI,UAAU;AACd,MAAI,UAAU;AACd,QAAM,oBAA8B,CAAC;AAErC,aAAW,KAAK,OAAO;AACrB,iBAAa,EAAE;AACf,cAAU,EAAE;AACZ,eAAW,EAAE;AACb,mBAAe,EAAE;AACjB,WAAO,EAAE;AACT,QAAI,EAAE,YAAY,mBAAoB;AACtC,QAAI,EAAE,YAAY,mBAAoB;AAEtC,QAAI,EAAE,YAAY,CAAC,EAAE,WAAY,mBAAkB,KAAK,EAAE,IAAI;AAAA,EAChE;AAEA,QAAM,IAAI,MAAM;AAChB,SAAO;AAAA,IACL,YAAY;AAAA,IACZ,gBAAgB;AAAA,IAChB,gBAAgB,SAAS;AAAA,IACzB,qBAAqB,UAAU;AAAA,IAC/B,yBAAyB,cAAc;AAAA,IACvC,aAAa,MAAM;AAAA,IACnB,kBAAkB;AAAA,IAClB,kBAAkB;AAAA,IAClB,UAAU,kBAAkB,SAAS,KAAK;AAAA,IAC1C;AAAA,EACF;AACF;;;AC5MO,SAAS,0BAA0B,QAAyB;AACjE,aAAW,KAAK,QAAQ;AACtB,QAAI,EAAE,KAAM,GAAE,OAAO,kBAAkB,EAAE,IAAI;AAC7C,QAAI,EAAE,OAAO;AACX,iBAAW,OAAO,EAAE,MAAM,OAAO;AAC/B,mBAAW,QAAQ,KAAK;AACtB,cAAI,KAAK,KAAM,MAAK,OAAO,kBAAkB,KAAK,IAAI;AAAA,QACxD;AAAA,MACF;AAAA,IACF;AACA,QAAI,EAAE,SAAU,2BAA0B,EAAE,QAAQ;AAAA,EACtD;AACF;AAEO,SAAS,aAAa,MAAsB;AACjD,SAAO;AAAA,IACL,kBAAkB,IAAI,EAEnB,QAAQ,cAAc,EAAE,EAExB,QAAQ,oDAAoD,EAAE,EAE9D,QAAQ,4BAA4B,EAAE,EAEtC,QAAQ,gBAAgB,IAAI,EAE5B,QAAQ,cAAc,EAAE,EAExB,QAAQ,2BAA2B,EAAE;AAAA,EAC1C,EAGG,QAAQ,oBAAoB,UAAQ;AACnC,QAAI,2BAA2B,KAAK,IAAI,EAAG,QAAO;AAClD,WAAO,oBAAoB,IAAI;AAAA,EACjC,CAAC,EAEA,QAAQ,oCAAoC,SAAS,EAErD,QAAQ,WAAW,IAAI,EAEvB,QAAQ,SAAS,EAAE,EACnB,QAAQ,WAAW,MAAM,EACzB,KAAK;AACV;AAEA,SAAS,iBAAiB,MAAuB;AAC/C,QAAM,IAAI,KAAK,UAAU;AACzB,SAAO,gBAAgB,KAAK,CAAC,KAAK,WAAW,KAAK,CAAC,KAAK,mBAAmB,KAAK,CAAC,KAC/E,sBAAsB,KAAK,CAAC,KAAK,eAAe,KAAK,CAAC;AAC1D;AAEA,SAAS,mBAAmB,MAAuB;AACjD,SAAO,yCAAyC,KAAK,KAAK,KAAK,CAAC;AAClE;AAEA,SAAS,iBAAiB,MAAsB;AAC9C,MAAI,CAAC,KAAM,QAAO;AAClB,QAAM,QAAQ,KAAK,MAAM,IAAI;AAC7B,MAAI,MAAM,UAAU,EAAG,QAAO;AAC9B,QAAM,SAAmB,CAAC,MAAM,CAAC,CAAC;AAElC,WAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;AACrC,UAAM,OAAO,OAAO,OAAO,SAAS,CAAC;AACrC,UAAM,OAAO,MAAM,CAAC;AACpB,UAAM,cAAc,KAAK,KAAK;AAE9B,QAAI,YAAY,KAAK,IAAI,KAAK,YAAY,KAAK,IAAI,KAAK,MAAM,KAAK,WAAW,KAAK,OAAO,KAAK,WAAW,GAAG;AAC3G,aAAO,KAAK,IAAI;AAChB;AAAA,IACF;AAEA,QAAI,KAAK,KAAK,KAAK,KAAK,CAAC,KAAK,YAAY,SAAS,GAAG;AACpD,aAAO,OAAO,SAAS,CAAC,IAAI,OAAO,OAAO;AAC1C;AAAA,IACF;AAEA,QAAI,OAAO,KAAK,WAAW,GAAG;AAC5B,aAAO,OAAO,SAAS,CAAC,IAAI,OAAO,MAAM;AACzC;AAAA,IACF;AAEA,QAAI,aAAa,KAAK,IAAI,KAAK,UAAU,KAAK,IAAI,KAC9C,CAAC,iBAAiB,IAAI,KAAK,CAAC,mBAAmB,IAAI,KACnD,CAAC,iBAAiB,IAAI,GAAG;AAC3B,aAAO,OAAO,SAAS,CAAC,IAAI,OAAO,MAAM;AAAA,IAC3C,OAAO;AACL,aAAO,KAAK,IAAI;AAAA,IAClB;AAAA,EACF;AACA,SAAO,OAAO,KAAK,IAAI;AACzB;;;ACpFA,eAAsB,gBACpB,QACA,QACA,YACA,oBACA,UACA,aACe;AACf,QAAM,aAAa,MAAM,OAAO,uBAAoB;AACpD,QAAM,EAAE,iBAAiB,oBAAoB,IAAI;AAGjD,QAAM,oBAAoB,CAAC,MAAM;AAC/B,QAAI,EAAE,UAAU,cAAc,EAAE,OAAO;AACrC,YAAM,MAAM,KAAK,MAAO,EAAE,aAAa,EAAE,QAAS,GAAG;AACrD,cAAQ,OAAO,MAAM,sBAAsB,EAAE,KAAK,IAAI,IAAI,GAAG,MAAM,SAAS,EAAE,UAAU,CAAC,IAAI,SAAS,EAAE,KAAK,CAAC,GAAG;AACjH,UAAI,EAAE,cAAc,EAAE,MAAO,SAAQ,OAAO,MAAM,IAAI;AAAA,IACxD,WAAW,EAAE,UAAU,UAAU;AAC/B,cAAQ,OAAO,MAAM,oBAAoB,EAAE,KAAK,IAAI;AAAA,CAAoB;AAAA,IAC1E,WAAW,EAAE,UAAU,QAAQ;AAC7B,cAAQ,OAAO,MAAM,oBAAoB,EAAE,KAAK,IAAI;AAAA,CAAU;AAAA,IAChE,WAAW,EAAE,UAAU,QAAQ;AAAA,IAE/B;AAAA,EACF,CAAC;AAED,QAAM,WAAW,MAAM,gBAAgB,OAAO;AAC9C,MAAI;AAGF,UAAM,kBAAkB,cACnB,IAAI,IAAI,MAAM,KAAK,EAAE,QAAQ,mBAAmB,GAAG,CAAC,GAAG,MAAM,IAAI,CAAC,CAAC;AACxE,UAAM,cAAc,MAAM,SAAS,YAAY,QAAQ,eAAe;AAEtE,QAAI,YAAY,WAAW,EAAG;AAE9B,QAAI,gBAAgB;AACpB,QAAI,kBAAkB;AAEtB,eAAW,QAAQ,aAAa;AAC9B,YAAM,aAAa,KAAK;AACxB,YAAM,YAAY,KAAK;AACvB,YAAM,SAAS,KAAK,gBAAgB,IAAI,KAAK,WAAW,KAAK,gBAAgB;AAC7E,YAAM,SAAS,KAAK,iBAAiB,IAAI,KAAK,YAAY,KAAK,iBAAiB;AAQhF,YAAM,aAAiC,CAAC;AACxC,iBAAW,KAAK,KAAK,SAAS;AAC5B,YAAI,CAAC,EAAE,SAAS,CAAC,EAAE,MAAM,KAAK,EAAG;AACjC,cAAM,UAAU,EAAE,SAAS,YAAY,KAAK,EAAE,KAAK,OAAO,IAAI,EAAE,KAAK;AAErE,cAAM,KAAK,EAAE,KAAK,KAAK;AACvB,cAAM,KAAK,EAAE,KAAK,KAAK;AAEvB,cAAM,OAAO,YAAY,EAAE,KAAK,KAAK;AACrC,cAAM,UAAU,YAAY,EAAE,KAAK,KAAK;AACxC,cAAM,WAAW,OAAO,WAAW;AACnC,cAAM,QAAQ,KAAK;AACnB,cAAM,SAAS,OAAO;AAEtB,mBAAW,KAAK;AAAA,UACd,OAAO;AAAA,YACL,MAAM;AAAA,YACN,MAAM;AAAA,YACN;AAAA,YACA,MAAM,EAAE,MAAM,YAAY,GAAG,IAAI,GAAG,SAAS,OAAO,OAAO;AAAA,UAC7D;AAAA,UACA,SAAS,EAAE,IAAI,IAAI,MAAM,QAAQ;AAAA,UACjC;AAAA,QACF,CAAC;AAAA,MACH;AACA,UAAI,WAAW,WAAW,EAAG;AAI7B,YAAM,oBAAoB;AAC1B,YAAM,kBAAkB,oBAAI,IAAY;AACxC,eAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,cAAM,IAAI,OAAO,CAAC;AAClB,YAAI,EAAE,eAAe,WAAY;AACjC,YAAI,EAAE,SAAS,QAAS;AACxB,YAAI,CAAC,EAAE,QAAQ,EAAE,KAAK,SAAS,KAAK,EAAE,KAAK,UAAU,EAAG;AACxD,cAAM,YAAY,EAAE,KAAK,QAAQ,EAAE,KAAK;AACxC,YAAI,aAAa,EAAG;AAEpB,mBAAW,KAAK,YAAY;AAC1B,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,GAAG,EAAE,QAAQ,EAAE;AAC3C,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,IAAI,EAAE,KAAK,OAAO,EAAE,QAAQ,EAAE;AAC1D,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,GAAG,EAAE,QAAQ,OAAO;AAChD,gBAAM,MAAM,KAAK,IAAI,EAAE,KAAK,IAAI,EAAE,KAAK,QAAQ,EAAE,QAAQ,IAAI;AAC7D,gBAAM,YAAY,KAAK,IAAI,GAAG,MAAM,GAAG,IAAI,KAAK,IAAI,GAAG,MAAM,GAAG;AAChE,cAAI,YAAY,aAAa,mBAAmB;AAC9C,4BAAgB,IAAI,CAAC;AACrB;AAAA,UACF;AAAA,QACF;AAAA,MACF;AAEA,UAAI,gBAAgB,OAAO,GAAG;AAE5B,cAAM,SAAS,CAAC,GAAG,eAAe,EAAE,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC;AACxD,mBAAW,OAAO,OAAQ,QAAO,OAAO,KAAK,CAAC;AAC9C,2BAAmB,gBAAgB;AAAA,MACrC;AAKA,iBAAW,KAAK,CAAC,GAAG,MAAM,EAAE,UAAU,EAAE,OAAO;AAE/C,iBAAW,KAAK,YAAY;AAC1B,YAAI,YAAY;AAChB,YAAI,eAAe;AACnB,YAAI,cAAc;AAClB,iBAAS,IAAI,GAAG,IAAI,OAAO,QAAQ,KAAK;AACtC,gBAAM,IAAI,OAAO,CAAC;AAClB,cAAI,EAAE,eAAe,WAAY;AACjC,cAAI,iBAAiB,GAAI,gBAAe;AACxC,wBAAc;AACd,cAAI,CAAC,EAAE,KAAM;AACb,gBAAM,cAAc,EAAE,KAAK,IAAI,EAAE,KAAK,SAAS;AAC/C,cAAI,cAAc,EAAE,SAAS;AAC3B,wBAAY;AACZ;AAAA,UACF;AAAA,QACF;AAEA,YAAI,cAAc,IAAI;AACpB,iBAAO,OAAO,WAAW,GAAG,EAAE,KAAK;AAAA,QACrC,WAAW,gBAAgB,IAAI;AAC7B,iBAAO,OAAO,cAAc,GAAG,GAAG,EAAE,KAAK;AAAA,QAC3C,OAAO;AAEL,iBAAO,KAAK,EAAE,KAAK;AAAA,QACrB;AACA;AAAA,MACF;AAAA,IACF;AAEA,QAAI,gBAAgB,KAAK,kBAAkB,GAAG;AAC5C,cAAQ,OAAO;AAAA,QACb,oBAAoB,aAAa,qCAAY,eAAe,2CAAkB,YAAY,MAAM;AAAA;AAAA,MAClG;AAAA,IACF;AAAA,EACF,UAAE;AACA,UAAM,SAAS,QAAQ,EAAE,MAAM,MAAM;AAAA,IAAC,CAAC;AAAA,EACzC;AACF;AAEA,SAAS,SAAS,OAAuB;AACvC,SAAO,IAAI,QAAQ,OAAO,MAAM,QAAQ,CAAC,CAAC;AAC5C;;;ACvJA,YAAY,iBAAiB;AAf7B,IAAM,IAAI;AAEV,IAAI,OAAO,EAAE,cAAc,aAAa;AACtC,IAAE,YAAY,MAAM,UAAU;AAAA,IAC5B,IAAc,CAAC,GAAG,GAAG,GAAG,GAAG,GAAG,CAAC;AAAA,IAC/B,YAAY,MAAiB;AAAE,UAAI,KAAM,MAAK,IAAI;AAAA,IAAK;AAAA,EACzD;AACF;AAEA,IAAI,OAAO,EAAE,WAAW,aAAa;AACnC,IAAE,SAAS,MAAM,OAAO;AAAA,EAAC;AAC3B;AAKA,EAAE,cAAc;;;ACChB,SAAS,aAAa,2BAA2B;AACjD,SAAS,qBAAqB;AAC9B,SAAS,SAAS,YAAY;AAQ9B,oBAAoB,YAAY;AAGhC,IAAM,YAAY;AAClB,IAAM,iBAAiB,MAAM,OAAO;AAEpC,IAAM,sBAAsB;AAK5B,IAAM,cAAwF,CAAC;AAC/F,IAAI;AACF,QAAM,WAAW,cAAc,YAAY,GAAG;AAC9C,QAAM,SAAS,QAAQ,SAAS,QAAQ,yBAAyB,CAAC;AAClE,cAAY,UAAU,KAAK,QAAQ,OAAO,IAAI;AAC9C,cAAY,aAAa;AACzB,cAAY,sBAAsB,KAAK,QAAQ,gBAAgB,IAAI;AACrE,QAAQ;AAA6C;AAGrD,eAAe,mBAAmB,QAAqB;AACrD,QAAM,cAAc,YAAY;AAAA,IAC9B,MAAM,IAAI,WAAW,MAAM;AAAA,IAC3B,gBAAgB;AAAA,IAChB,iBAAiB;AAAA,IACjB,iBAAiB;AAAA,IACjB,GAAG;AAAA,EACL,CAAC;AACD,MAAI;AACJ,MAAI;AACF,WAAO,MAAM,QAAQ,KAAK;AAAA,MACxB,YAAY;AAAA,MACZ,IAAI,QAAe,CAAC,GAAG,WAAW;AAChC,gBAAQ,WAAW,MAAM;AAAE,sBAAY,QAAQ;AAAG,iBAAO,IAAI,YAAY,mEAAsB,CAAC;AAAA,QAAE,GAAG,mBAAmB;AAAA,MAC1H,CAAC;AAAA,IACH,CAAC;AAAA,EACH,UAAE;AACA,QAAI,UAAU,OAAW,cAAa,KAAK;AAAA,EAC7C;AACF;AAEA,eAAsB,iBAAiB,QAAqB,SAAsD;AAGhH,QAAM,gBAAoC,SAAS,aAAa,OAAO,MAAM,CAAC,IAAI;AAClF,QAAM,YAAgC,SAAS,MAAM,OAAO,MAAM,CAAC,IAAI;AACvE,QAAM,MAAM,MAAM,mBAAmB,MAAM;AAE3C,MAAI;AACF,UAAM,YAAY,IAAI;AACtB,QAAI,cAAc,EAAG,OAAM,IAAI,YAAY,8DAAiB;AAG5D,UAAM,WAA6B,EAAE,UAAU;AAC/C,UAAM,mBAAmB,KAAK,QAAQ;AAEtC,UAAM,SAAoB,CAAC;AAC3B,UAAM,WAA2B,CAAC;AAClC,UAAM,cAA6B,CAAC;AACpC,QAAI,aAAa;AACjB,QAAI,iBAAiB;AACrB,UAAM,qBAAqB,KAAK,IAAI,WAAW,SAAS;AAGxD,UAAM,aAAa,SAAS,QAAQ,eAAe,QAAQ,OAAO,kBAAkB,IAAI;AACxF,UAAM,cAAc,aAAa,WAAW,OAAO;AAGnD,UAAM,eAAe,oBAAI,IAAoB;AAC7C,UAAM,cAAc,oBAAI,IAAoB;AAE5C,UAAM,sBAAsB,oBAAI,IAAY;AAE5C,UAAM,oBAAoB,oBAAI,IAAoB;AAIlD,UAAM,aAAa,oBAAoB;AACvC,UAAM,kBAAoC,CAAC;AAC3C,UAAM,kBAAkB,oBAAI,IAAuB;AAEnD,QAAI,cAAc;AAClB,aAAS,IAAI,GAAG,KAAK,oBAAoB,KAAK;AAC5C,UAAI,cAAc,CAAC,WAAW,IAAI,CAAC,EAAG;AACtC,UAAI;AACF,cAAM,OAAO,MAAM,IAAI,QAAQ,CAAC;AAChC,cAAM,KAAK,MAAM,KAAK,eAAe;AACrC,cAAM,WAAW,KAAK,YAAY,EAAE,OAAO,EAAE,CAAC;AAC9C,oBAAY,IAAI,GAAG,SAAS,MAAM;AAClC,cAAM,WAAW,GAAG;AACpB,cAAM,QAAQ,eAAe,QAAQ;AAGrC,cAAM,EAAE,SAAS,YAAY,IAAI,iBAAiB,OAAO,SAAS,OAAO,SAAS,MAAM;AACxF,YAAI,cAAc,GAAG;AACnB,mBAAS,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,WAAW,sFAAqB,MAAM,uBAAuB,CAAC;AAAA,QACrG;AAGA,mBAAW,QAAQ,SAAS;AAC1B,cAAI,KAAK,WAAW,EAAG,cAAa,IAAI,KAAK,WAAW,aAAa,IAAI,KAAK,QAAQ,KAAK,KAAK,CAAC;AAAA,QACnG;AAGA,cAAM,SAAS,MAAM,KAAK,gBAAgB;AAG1C,cAAM,WAAW,SAAS,QAAQ,SAAS;AAC3C,YAAI,WAAW,GAAG;AAChB,gBAAM,eAAe,oBAAoB,OAAO,SAAS,OAAO,SAAS;AACzE,cAAI,YAAY;AAChB,qBAAW,KAAK,cAAc;AAC5B,kBAAM,QAAQ,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE;AACvC,gBAAI,OAAO,WAAW,KAAM;AAC5B,gCAAoB,IAAI,CAAC;AACzB,kBAAM,UAAU,QAAQ,KAAK,QAAM;AACjC,oBAAM,KAAK,GAAG,IAAI,GAAG,IAAI;AACzB,oBAAM,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,YAAY;AAC1C,qBAAO,MAAM,EAAE,MAAM,MAAM,EAAE,MAAM,MAAM,EAAE,MAAM,MAAM,EAAE;AAAA,YAC3D,CAAC;AACD,gBAAI,CAAC,QAAS;AAAA,UAChB;AACA,cAAI,YAAY,EAAG,mBAAkB,IAAI,GAAG,SAAS;AAAA,QACvD;AAEA,cAAM,aAAa,2BAA2B,SAAS,GAAG,QAAQ,SAAS,OAAO,SAAS,MAAM;AACjG,mBAAW,KAAK,WAAY,QAAO,KAAK,CAAC;AAGzC,YAAI;AACF,gBAAM,EAAE,QAAQ,WAAW,QAAQ,WAAW,IAAI,MAAM,kBAAkB,MAAM,OAAO,SAAS,OAAO,WAAW,GAAG,YAAY,QAAQ;AACzI,cAAI,UAAU,SAAS,EAAG,iBAAgB,IAAI,GAAG,SAAS;AAC1D,0BAAgB,KAAK,GAAG,UAAU;AAAA,QACpC,QAAQ;AAAA,QAAkC;AAG1C,YAAI,WAAW;AACf,mBAAW,KAAK,YAAY;AAC1B,cAAI,IAAI,EAAE,QAAQ;AAIlB,cAAI,EAAE,SAAS,WAAW,EAAE,OAAO;AACjC,kBAAM,WAAW,EAAE,MAAM,MAAM,IAAI,SAAO,IAAI,IAAI,OAAK,EAAE,IAAI,EAAE,KAAK,GAAG,CAAC,EAAE,KAAK,IAAI;AACnF,gBAAI,IAAI,IAAI,OAAO,WAAW;AAAA,UAChC;AACA,wBAAc,EAAE,QAAQ,OAAO,EAAE,EAAE;AACnC,4BAAkB,EAAE,SAAS;AAC7B,sBAAY,WAAW,OAAO,IAAI;AAAA,QACpC;AACA,oBAAY,KAAK,mBAAmB,GAAG,QAAQ,CAAC;AAChD,YAAI,iBAAiB,eAAgB,OAAM,IAAI,YAAY,2DAAc;AACzE;AACA,iBAAS,aAAa,aAAa,WAAW;AAAA,MAChD,SAAS,SAAS;AAEhB,YAAI,mBAAmB,YAAa,OAAM;AAC1C,iBAAS,KAAK,EAAE,MAAM,GAAG,SAAS,sBAAO,CAAC,+BAAW,mBAAmB,QAAQ,QAAQ,UAAU,yCAAW,IAAI,MAAM,gBAAgB,CAAC;AAAA,MAC1I;AAAA,IACF;AAEA,UAAM,kBAAkB,gBAAgB,aAAa,WAAW,OAAO;AAEvE,UAAM,eAAe,aAAa,KAAK,IAAI,iBAAiB,CAAC,IAAI;AAMjE,UAAM,UAAU,oBAAI,IAAY;AAChC,QAAI,SAAS,OAAO,WAAW;AAC7B,YAAM,UAAU,CAAC,MAAc,CAAC,cAAc,WAAW,IAAI,CAAC;AAC9D,YAAM,UAAU,oBAAI,IAAY;AAChC,UAAI,QAAQ,QAAQ,WAAW,cAAc;AAC3C,iBAAS,IAAI,GAAG,KAAK,oBAAoB,IAAK,KAAI,QAAQ,CAAC,EAAG,SAAQ,IAAI,CAAC;AAAA,MAC7E,OAAO;AACL,mBAAW,MAAM,aAAa;AAC5B,cAAI,CAAC,GAAG,SAAU;AAElB,cAAI,GAAG,cAAc,cAAc,CAAC,oBAAoB,IAAI,GAAG,IAAI,EAAG;AACtE,kBAAQ,IAAI,GAAG,IAAI;AAAA,QACrB;AAAA,MACF;AACA,UAAI,QAAQ,OAAO,GAAG;AACpB,YAAI;AACF,gBAAM,EAAE,UAAU,IAAI,MAAM,OAAO,uBAAmB;AACtD,gBAAM,OAAO,OAAO,QAAQ,QAAQ,aAAa,QAAQ,MAAO;AAChE,gBAAM,gBAAgB,MAAM,UAAU,WAAW,SAAS,MAAM,UAAU,QAAQ,UAAU;AAC5F,cAAI,cAAc,OAAO,GAAG;AAE1B,uBAAW,KAAK,cAAc,KAAK,EAAG,SAAQ,IAAI,CAAC;AACnD,kBAAM,SAAS,OAAO,OAAO,OAAK,EAAE,EAAE,cAAc,QAAQ,IAAI,EAAE,UAAU,EAAE;AAC9E,uBAAW,OAAO,cAAc,OAAO,EAAG,QAAO,KAAK,GAAG,GAAG;AAC5D,mBAAO,KAAK,CAAC,GAAG,OAAO,EAAE,cAAc,MAAM,EAAE,cAAc,EAAE;AAC/D,mBAAO,SAAS;AAChB,mBAAO,KAAK,GAAG,MAAM;AACrB,uBAAW,MAAM,YAAa,KAAI,QAAQ,IAAI,GAAG,IAAI,EAAG,IAAG,aAAa;AACxE,qBAAS,KAAK;AAAA,cACZ,SAAS,GAAG,QAAQ,IAAI,qDAAkB,SAAS,YAAY,0BAAgB,mDAAW;AAAA,cAC1F,MAAM;AAAA,YACR,CAAC;AAAA,UACH;AAAA,QACF,SAAS,GAAG;AAEV,mBAAS,KAAK;AAAA,YACZ,SAAS,kCAAc,aAAa,QAAQ,EAAE,UAAU,OAAO,CAAC,CAAC;AAAA,YACjE,MAAM;AAAA,UACR,CAAC;AAAA,QACH;AAAA,MACF;AAAA,IACF;AAEA,QAAI,gBAAgB,QAAQ,SAAS,GAAG;AAEtC,eAAS,KAAK;AAAA,QACZ,SAAS,wCAAe,SAAS,0CAAY,UAAU;AAAA,QACvD,MAAM;AAAA,MACR,CAAC;AAAA,IACH;AAKA,QAAI,CAAC,cAAc;AACjB,YAAM,sBAA8C;AAAA,QAClD,UAAU;AAAA,QACV,UAAU;AAAA,QACV,cAAc;AAAA,QACd,kBAAkB;AAAA,QAClB,gBAAgB;AAAA,MAClB;AACA,iBAAW,MAAM,aAAa;AAC5B,YAAI,CAAC,GAAG,YAAY,CAAC,GAAG,aAAa,GAAG,WAAY;AACpD,YAAI,GAAG,cAAc,cAAc,CAAC,oBAAoB,IAAI,GAAG,IAAI,EAAG;AACtE,iBAAS,KAAK,EAAE,MAAM,GAAG,MAAM,SAAS,GAAG,oBAAoB,GAAG,SAAS,CAAC,yCAAgB,MAAM,YAAY,CAAC;AAAA,MACjH;AAAA,IACF;AAIA,QAAI,CAAC,cAAc;AACjB,iBAAW,CAAC,MAAM,KAAK,KAAK,CAAC,GAAG,kBAAkB,QAAQ,CAAC,EAAE,KAAK,CAAC,GAAG,MAAM,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG;AACxF,iBAAS,KAAK,EAAE,MAAM,SAAS,GAAG,KAAK,gMAA+C,MAAM,gBAAgB,CAAC;AAAA,MAC/G;AAAA,IACF;AAGA,QAAI,SAAS,uBAAuB,SAAS,mBAAmB,GAAG;AACjE,YAAM,UAAU,yBAAyB,QAAQ,aAAa,QAAQ;AAEtE,eAAS,KAAK,QAAQ,SAAS,GAAG,MAAM,GAAG,MAAM;AAC/C,eAAO,OAAO,QAAQ,EAAE,GAAG,CAAC;AAAA,MAC9B;AAAA,IACF;AAIA,yBAAqB,MAAM;AAG3B,0BAAsB,QAAQ,eAAe;AAI7C,QAAI,SAAS,cAAc,eAAe;AACxC,UAAI;AACF,cAAM,gBAAgB,eAAe,QAAQ,YAAY,oBAAoB,UAAU,QAAQ,UAAU;AAAA,MAC3G,SAAS,GAAG;AACV,iBAAS,KAAK;AAAA,UACZ,SAAS,kCAAc,aAAa,QAAQ,EAAE,UAAU,OAAO,CAAC,CAAC;AAAA,UACjE,MAAM;AAAA,QACR,CAAC;AAAA,MACH;AAAA,IACF;AAGA,UAAM,iBAAiB,8BAA8B,YAAY;AACjE,QAAI,iBAAiB,GAAG;AACtB,qBAAe,QAAQ,cAAc;AAAA,IACvC;AAGA,yBAAqB,MAAM;AAG3B,wBAAoB,MAAM;AAG1B,2BAAuB,MAAM;AAG7B,UAAM,UAAyB,OAC5B,OAAO,OAAK,EAAE,SAAS,aAAa,EAAE,SAAS,EAAE,IAAI,EACrD,IAAI,QAAM,EAAE,OAAO,EAAE,OAAQ,MAAM,EAAE,MAAO,YAAY,EAAE,WAAW,EAAE;AAG1E,8BAA0B,MAAM;AAGhC,QAAI,WAAW,aAAa,iBAAiB,MAAM,CAAC;AAEpD,WAAO;AAAA,MACL;AAAA,MACA;AAAA,MACA;AAAA,MACA,SAAS,QAAQ,SAAS,IAAI,UAAU;AAAA,MACxC,UAAU,SAAS,SAAS,IAAI,WAAW;AAAA,MAC3C,cAAc,gBAAgB;AAAA,MAC9B;AAAA,MACA,gBAAgB,yBAAyB,WAAW;AAAA,MACpD,QAAQ,gBAAgB,SAAS,IAAI,kBAAkB;AAAA,IACzD;AAAA,EACF,UAAE;AACA,UAAM,IAAI,QAAQ,EAAE,MAAM,MAAM;AAAA,IAAC,CAAC;AAAA,EACpC;AACF;AAIA,eAAe,mBAAmB,KAA0C,UAA2C;AACrH,MAAI;AACF,UAAM,SAAS,MAAM,IAAI,YAAY;AACrC,QAAI,CAAC,QAAQ,KAAM;AACnB,UAAM,OAAO,OAAO;AAEpB,QAAI,OAAO,KAAK,UAAU,YAAY,KAAK,MAAM,KAAK,EAAG,UAAS,QAAQ,KAAK,MAAM,KAAK;AAC1F,QAAI,OAAO,KAAK,WAAW,YAAY,KAAK,OAAO,KAAK,EAAG,UAAS,SAAS,KAAK,OAAO,KAAK;AAC9F,QAAI,OAAO,KAAK,YAAY,YAAY,KAAK,QAAQ,KAAK,EAAG,UAAS,UAAU,KAAK,QAAQ,KAAK;AAClG,QAAI,OAAO,KAAK,YAAY,YAAY,KAAK,QAAQ,KAAK,EAAG,UAAS,cAAc,KAAK,QAAQ,KAAK;AACtG,QAAI,OAAO,KAAK,aAAa,YAAY,KAAK,SAAS,KAAK,GAAG;AAC7D,eAAS,WAAW,KAAK,SAAS,MAAM,MAAM,EAAE,IAAI,CAAC,MAAc,EAAE,KAAK,CAAC,EAAE,OAAO,OAAO;AAAA,IAC7F;AACA,QAAI,OAAO,KAAK,iBAAiB,SAAU,UAAS,YAAY,aAAa,KAAK,YAAY;AAC9F,QAAI,OAAO,KAAK,YAAY,SAAU,UAAS,aAAa,aAAa,KAAK,OAAO;AAAA,EACvF,QAAQ;AAAA,EAER;AACF;AAGA,SAAS,aAAa,SAAqC;AACzD,QAAM,IAAI,QAAQ,MAAM,mDAAmD;AAC3E,MAAI,CAAC,EAAG,QAAO;AACf,QAAM,CAAC,EAAE,MAAM,QAAQ,MAAM,MAAM,MAAM,OAAO,MAAM,MAAM,MAAM,MAAM,IAAI,IAAI;AAChF,SAAO,GAAG,IAAI,IAAI,KAAK,IAAI,GAAG,IAAI,IAAI,IAAI,GAAG,IAAI,GAAG;AACtD;AAGA,eAAsB,uBAAuB,QAAgD;AAC3F,QAAM,MAAM,MAAM,mBAAmB,MAAM;AAE3C,MAAI;AACF,UAAM,WAA6B,EAAE,WAAW,IAAI,SAAS;AAC7D,UAAM,mBAAmB,KAAK,QAAQ;AACtC,WAAO;AAAA,EACT,UAAE;AACA,UAAM,IAAI,QAAQ,EAAE,MAAM,MAAM;AAAA,IAAC,CAAC;AAAA,EACpC;AACF;","names":[]}
@@ -1,13 +0,0 @@
1
- "use strict";Object.defineProperty(exports, "__esModule", {value: true});
2
-
3
-
4
- var _chunk5F5UCACZcjs = require('./chunk-5F5UCACZ.cjs');
5
- require('./chunk-MGGU3ZRK.cjs');
6
- require('./chunk-7DZPPUQD.cjs');
7
- require('./chunk-VQB5P5KJ.cjs');
8
- require('./chunk-GS7T56RP.cjs');
9
-
10
-
11
-
12
- exports.ocrItemsToBlocks = _chunk5F5UCACZcjs.ocrItemsToBlocks; exports.runPdfOcr = _chunk5F5UCACZcjs.runPdfOcr;
13
- //# sourceMappingURL=pdf-ocr-4LISPZEM.cjs.map
@@ -1 +0,0 @@
1
- {"version":3,"sources":["c:\\github_project\\kordoc\\dist\\pdf-ocr-4LISPZEM.cjs"],"names":[],"mappings":"AAAA;AACE;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACA;AACF,+GAAC","file":"C:\\github_project\\kordoc\\dist\\pdf-ocr-4LISPZEM.cjs"}
@@ -1,12 +0,0 @@
1
- import {
2
- ocrItemsToBlocks,
3
- runPdfOcr
4
- } from "./chunk-PV3NVBGL.js";
5
- import "./chunk-3FFPHTXD.js";
6
- import "./chunk-CD5X6UG5.js";
7
- import "./chunk-LEC2HG7A.js";
8
- export {
9
- ocrItemsToBlocks,
10
- runPdfOcr
11
- };
12
- //# sourceMappingURL=pdf-ocr-AIWB2QCS.js.map
@@ -1,15 +0,0 @@
1
- #!/usr/bin/env node
2
- import {
3
- ocrItemsToBlocks,
4
- runPdfOcr
5
- } from "./chunk-YDCD4AV4.js";
6
- import "./chunk-HPRK67YB.js";
7
- import "./chunk-ZSL6MQVE.js";
8
- import "./chunk-ZJ75YDW7.js";
9
- import "./chunk-6XPXIG57.js";
10
- import "./chunk-YNJDNP44.js";
11
- export {
12
- ocrItemsToBlocks,
13
- runPdfOcr
14
- };
15
- //# sourceMappingURL=pdf-ocr-QMOKUMC2.js.map
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/render/rasterize.ts"],"sourcesContent":["/** SVG → PNG 래스터 — sharp optional 의존 (미설치 시 KordocError, SVG 경로는 sharp 불필요) */\r\n\r\nimport { KordocError } from \"../utils.js\"\r\n\r\nexport interface RasterizeOptions {\r\n /** 출력 최대 폭 px (기본 1400) */\r\n maxWidthPx?: number\r\n /** 출력 최대 높이 px (기본 8000 — 멀티페이지 세로 스택 대비) */\r\n maxHeightPx?: number\r\n /** PNG 바이트 상한 (기본 4MB) — 초과 시 절반 스케일로 1회 재시도 */\r\n maxBytes?: number\r\n}\r\n\r\nexport interface RasterizeResult {\r\n png: Buffer\r\n widthPx: number\r\n heightPx: number\r\n /** pt → px 배율 */\r\n scale: number\r\n}\r\n\r\n/** SVG 문자열을 PNG로 래스터. widthPt/heightPt는 SVG 캔버스 크기(pt). */\r\nexport async function rasterizeSvg(\r\n svg: string,\r\n widthPt: number,\r\n heightPt: number,\r\n options?: RasterizeOptions,\r\n): Promise<RasterizeResult> {\r\n let sharp: (input: Buffer, opts: { density: number; limitInputPixels: number }) => { png(): { toBuffer(): Promise<Buffer> } }\r\n try {\r\n const mod: any = await import(\"sharp\")\r\n sharp = mod.default ?? mod\r\n } catch {\r\n throw new KordocError(\r\n 'PNG 래스터에는 sharp가 필요합니다 (npm install sharp). sharp 없이 쓰려면 format: \"svg\" + output_path로 SVG 파일 저장을 사용하세요',\r\n )\r\n }\r\n if (!(widthPt > 0) || !(heightPt > 0)) throw new KordocError(`잘못된 SVG 크기: ${widthPt}x${heightPt}pt`)\r\n const maxW = options?.maxWidthPx ?? 1400\r\n const maxH = options?.maxHeightPx ?? 8000\r\n const maxBytes = options?.maxBytes ?? 4 * 1024 * 1024\r\n\r\n const render = async (scale: number): Promise<RasterizeResult> => {\r\n // sharp의 SVG 기본 밀도 72DPI 기준 — density를 올리면 pt×scale 픽셀로 래스터된다\r\n const png = await sharp(Buffer.from(svg), { density: 72 * scale, limitInputPixels: 268402689 })\r\n .png()\r\n .toBuffer()\r\n return { png, widthPx: Math.round(widthPt * scale), heightPx: Math.round(heightPt * scale), scale }\r\n }\r\n\r\n const scale = Math.min(maxW / widthPt, maxH / heightPt, 2)\r\n let result = await render(scale)\r\n if (result.png.length > maxBytes && scale > 0.25) {\r\n result = await render(scale / 2)\r\n }\r\n if (result.png.length > maxBytes) {\r\n throw new KordocError(\r\n `PNG가 상한(${(maxBytes / 1024 / 1024).toFixed(0)}MB)을 초과합니다 (${(result.png.length / 1024 / 1024).toFixed(1)}MB) — format: \"svg\" + output_path로 파일 저장을 사용하세요`,\r\n )\r\n }\r\n return result\r\n}\r\n"],"mappings":";;;;;;AAsBA,eAAsB,aACpB,KACA,SACA,UACA,SAC0B;AAC1B,MAAI;AACJ,MAAI;AACF,UAAM,MAAW,MAAM,OAAO,OAAO;AACrC,YAAQ,IAAI,WAAW;AAAA,EACzB,QAAQ;AACN,UAAM,IAAI;AAAA,MACR;AAAA,IACF;AAAA,EACF;AACA,MAAI,EAAE,UAAU,MAAM,EAAE,WAAW,GAAI,OAAM,IAAI,YAAY,wCAAe,OAAO,IAAI,QAAQ,IAAI;AACnG,QAAM,OAAO,SAAS,cAAc;AACpC,QAAM,OAAO,SAAS,eAAe;AACrC,QAAM,WAAW,SAAS,YAAY,IAAI,OAAO;AAEjD,QAAM,SAAS,OAAOA,WAA4C;AAEhE,UAAM,MAAM,MAAM,MAAM,OAAO,KAAK,GAAG,GAAG,EAAE,SAAS,KAAKA,QAAO,kBAAkB,UAAU,CAAC,EAC3F,IAAI,EACJ,SAAS;AACZ,WAAO,EAAE,KAAK,SAAS,KAAK,MAAM,UAAUA,MAAK,GAAG,UAAU,KAAK,MAAM,WAAWA,MAAK,GAAG,OAAAA,OAAM;AAAA,EACpG;AAEA,QAAM,QAAQ,KAAK,IAAI,OAAO,SAAS,OAAO,UAAU,CAAC;AACzD,MAAI,SAAS,MAAM,OAAO,KAAK;AAC/B,MAAI,OAAO,IAAI,SAAS,YAAY,QAAQ,MAAM;AAChD,aAAS,MAAM,OAAO,QAAQ,CAAC;AAAA,EACjC;AACA,MAAI,OAAO,IAAI,SAAS,UAAU;AAChC,UAAM,IAAI;AAAA,MACR,2BAAY,WAAW,OAAO,MAAM,QAAQ,CAAC,CAAC,8CAAgB,OAAO,IAAI,SAAS,OAAO,MAAM,QAAQ,CAAC,CAAC;AAAA,IAC3G;AAAA,EACF;AACA,SAAO;AACT;","names":["scale"]}
@@ -1,10 +0,0 @@
1
- #!/usr/bin/env node
2
- import {
3
- renderHwpxToSvg
4
- } from "./chunk-SA22NK56.js";
5
- import "./chunk-HZU7Y22R.js";
6
- import "./chunk-ZJ75YDW7.js";
7
- export {
8
- renderHwpxToSvg
9
- };
10
- //# sourceMappingURL=render-PSPTMZIZ.js.map
@@ -1,10 +0,0 @@
1
- #!/usr/bin/env node
2
- import {
3
- placeSealHwpx
4
- } from "./chunk-BYEQDLZS.js";
5
- import "./chunk-HZU7Y22R.js";
6
- import "./chunk-ZJ75YDW7.js";
7
- export {
8
- placeSealHwpx
9
- };
10
- //# sourceMappingURL=seal-4XBJ7ISE.js.map
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/setup.ts"],"sourcesContent":["/**\r\n * 대화형 설치 마법사 — kordoc MCP\r\n *\r\n * `npx kordoc setup` 으로 실행.\r\n * 선택한 AI 클라이언트 설정 파일에 kordoc-mcp 서버를 자동 등록합니다.\r\n * API 키 불필요, macOS / Linux / Windows 공용.\r\n */\r\n\r\nimport { createInterface } from \"node:readline/promises\"\r\nimport { readFile, writeFile, mkdir } from \"node:fs/promises\"\r\nimport { execFile } from \"node:child_process\"\r\nimport { existsSync } from \"node:fs\"\r\nimport { resolve, dirname, delimiter } from \"node:path\"\r\nimport { homedir, platform } from \"node:os\"\r\nimport { stdin, stdout } from \"node:process\"\r\nimport { promisify } from \"node:util\"\r\n\r\ninterface ClientConfig {\r\n readonly name: string\r\n readonly configPath: string\r\n readonly format: \"mcpServers\" | \"servers\" | \"context_servers\"\r\n readonly setupMethod?: \"codex\"\r\n}\r\n\r\nconst execFileAsync = promisify(execFile)\r\n\r\n/** PATH에 있는 실행 파일을 확인한다. Windows에서는 PATHEXT 확장자도 고려한다. */\r\nfunction commandExists(command: string): boolean {\r\n const pathEntries = (process.env[\"PATH\"] ?? \"\").split(delimiter).filter(Boolean)\r\n const extensions = platform() === \"win32\"\r\n ? (process.env[\"PATHEXT\"] ?? \".EXE;.CMD;.BAT;.COM\").split(\";\")\r\n : [\"\"]\r\n return pathEntries.some((entry) => extensions.some((ext) => existsSync(resolve(entry, `${command}${ext}`))))\r\n}\r\n\r\nfunction detectClients(): readonly ClientConfig[] {\r\n const home = homedir()\r\n const os = platform()\r\n const clients: ClientConfig[] = []\r\n\r\n const claudePaths: Record<string, string> = {\r\n darwin: resolve(home, \"Library/Application Support/Claude/claude_desktop_config.json\"),\r\n win32: resolve(process.env[\"APPDATA\"] ?? resolve(home, \"AppData/Roaming\"), \"Claude/claude_desktop_config.json\"),\r\n linux: resolve(home, \".config/Claude/claude_desktop_config.json\"),\r\n }\r\n const claudePath = claudePaths[os]\r\n if (claudePath) clients.push({ name: \"Claude Desktop\", configPath: claudePath, format: \"mcpServers\" })\r\n\r\n clients.push({ name: \"Claude Code (현재 디렉토리)\", configPath: resolve(process.cwd(), \".mcp.json\"), format: \"mcpServers\" })\r\n clients.push({ name: \"Cursor\", configPath: resolve(home, \".cursor/mcp.json\"), format: \"mcpServers\" })\r\n clients.push({ name: \"VS Code (현재 디렉토리)\", configPath: resolve(process.cwd(), \".vscode/mcp.json\"), format: \"servers\" })\r\n clients.push({ name: \"Windsurf\", configPath: resolve(home, \".codeium/windsurf/mcp_config.json\"), format: \"mcpServers\" })\r\n clients.push({ name: \"Gemini CLI\", configPath: resolve(home, \".gemini/settings.json\"), format: \"mcpServers\" })\r\n\r\n const zedPaths: Record<string, string> = {\r\n darwin: resolve(home, \".zed/settings.json\"),\r\n linux: resolve(home, \".config/zed/settings.json\"),\r\n win32: resolve(home, \".zed/settings.json\"),\r\n }\r\n const zedPath = zedPaths[os]\r\n if (zedPath) clients.push({ name: \"Zed\", configPath: zedPath, format: \"context_servers\" })\r\n\r\n clients.push({ name: \"Antigravity\", configPath: resolve(home, \".gemini/antigravity/mcp_config.json\"), format: \"mcpServers\" })\r\n clients.push({ name: \"Codex\", configPath: resolve(home, \".codex/config.toml\"), format: \"mcpServers\", setupMethod: \"codex\" })\r\n\r\n return clients\r\n}\r\n\r\nasync function readJsonFile(path: string): Promise<Record<string, unknown>> {\r\n if (!existsSync(path)) return {}\r\n const raw = await readFile(path, \"utf-8\")\r\n return JSON.parse(raw) as Record<string, unknown>\r\n}\r\n\r\nasync function writeJsonFile(path: string, data: Record<string, unknown>): Promise<void> {\r\n const dir = dirname(path)\r\n if (!existsSync(dir)) await mkdir(dir, { recursive: true })\r\n await writeFile(path, JSON.stringify(data, null, 2) + \"\\n\", \"utf-8\")\r\n}\r\n\r\n/**\r\n * Windows 에선 npx.cmd 를 Claude Desktop 이 해석 못 해서 `cmd /c` 래핑이 필요하다.\r\n * `kordoc` 은 `npx kordoc mcp` 로 MCP 서버 실행, 또는 `kordoc-mcp` bin 직접 실행.\r\n * npx 래핑이 범용적이므로 `npx -y kordoc mcp` 로 통일.\r\n */\r\nfunction buildServerEntry(): Record<string, unknown> {\r\n if (platform() === \"win32\") {\r\n return { command: \"cmd\", args: [\"/c\", \"npx\", \"-y\", \"kordoc\", \"mcp\"] }\r\n }\r\n return { command: \"npx\", args: [\"-y\", \"kordoc\", \"mcp\"] }\r\n}\r\n\r\nfunction buildZedEntry(): Record<string, unknown> {\r\n const base = platform() === \"win32\"\r\n ? { path: \"cmd\", args: [\"/c\", \"npx\", \"-y\", \"kordoc\", \"mcp\"] }\r\n : { path: \"npx\", args: [\"-y\", \"kordoc\", \"mcp\"] }\r\n return { command: base }\r\n}\r\n\r\n/** Codex의 TOML 설정을 직접 파싱하지 않고, 공식 CLI로 MCP 서버를 등록한다. */\r\nasync function setupCodex(): Promise<void> {\r\n const entry = buildServerEntry()\r\n const command = entry[\"command\"] as string\r\n const args = entry[\"args\"] as string[]\r\n await execFileAsync(\"codex\", [\"mcp\", \"add\", \"kordoc\", \"--\", command, ...args])\r\n}\r\n\r\n// ─── ANSI ─────────────────────────────────────────────────────────────\r\nconst ESC = \"\\x1b[\"\r\nconst c = {\r\n reset: `${ESC}0m`, bold: `${ESC}1m`, dim: `${ESC}2m`,\r\n cyan: `${ESC}36m`, green: `${ESC}32m`, yellow: `${ESC}33m`,\r\n red: `${ESC}31m`, white: `${ESC}37m`,\r\n} as const\r\n\r\nfunction rgb(r: number, g: number, b: number): string {\r\n return `${ESC}38;2;${r};${g};${b}m`\r\n}\r\nfunction sleep(ms: number): Promise<void> {\r\n return new Promise((res) => setTimeout(res, ms))\r\n}\r\nasync function typewrite(text: string, delay = 15): Promise<void> {\r\n for (const ch of text) { process.stdout.write(ch); await sleep(delay) }\r\n console.log()\r\n}\r\n\r\nasync function printBanner(): Promise<void> {\r\n const gradients = [\r\n rgb(255, 120, 80), rgb(255, 140, 80), rgb(255, 160, 80),\r\n rgb(240, 180, 80), rgb(220, 200, 80), rgb(200, 220, 80),\r\n ]\r\n const logo = [\r\n \" _ _ \",\r\n \" | | _____ _ __ __| | ___ ___ \",\r\n \" | |/ / _ \\\\| '__/ _` |/ _ \\\\ / __|\",\r\n \" | < (_) | | | (_| | (_) | (__ \",\r\n \" |_|\\\\_\\\\___/|_| \\\\__,_|\\\\___/ \\\\___|\",\r\n ]\r\n console.log()\r\n for (let i = 0; i < logo.length; i++) {\r\n console.log(`${gradients[i % gradients.length]}${c.bold}${logo[i]}${c.reset}`)\r\n await sleep(60)\r\n }\r\n console.log()\r\n await typewrite(`${c.dim} 모두 파싱해버리겠다 ━━ HWP · HWPX · PDF · XLSX · DOCX → Markdown${c.reset}`, 10)\r\n console.log()\r\n console.log(`${c.cyan} ${\"━\".repeat(60)}${c.reset}`)\r\n console.log()\r\n}\r\n\r\nfunction stepHeader(step: number, total: number, title: string): void {\r\n const dots = `${c.dim}${\"·\".repeat(Math.max(0, 40 - title.length))}${c.reset}`\r\n console.log(` ${c.cyan}${c.bold}[${step}/${total}]${c.reset} ${c.white}${c.bold}${title}${c.reset} ${dots}`)\r\n console.log()\r\n}\r\n\r\nfunction successLine(label: string, detail: string): void {\r\n console.log(` ${c.green}${c.bold}+${c.reset} ${c.white}${label}${c.reset}${c.dim} ${detail}${c.reset}`)\r\n}\r\nfunction failLine(label: string, detail: string): void {\r\n console.log(` ${c.red}${c.bold}x${c.reset} ${c.white}${label}${c.reset}${c.dim} ${detail}${c.reset}`)\r\n}\r\n\r\nasync function printComplete(): Promise<void> {\r\n console.log()\r\n const box = [\r\n ` ${c.green}${c.bold}╔${\"═\".repeat(50)}╗${c.reset}`,\r\n ` ${c.green}${c.bold}║${c.reset}${\" \".repeat(14)}${c.green}${c.bold}Setup Complete!${c.reset}${\" \".repeat(22)}${c.green}${c.bold}║${c.reset}`,\r\n ` ${c.green}${c.bold}╚${\"═\".repeat(50)}╝${c.reset}`,\r\n ]\r\n for (const line of box) { console.log(line); await sleep(40) }\r\n console.log()\r\n console.log(` ${c.dim}클라이언트를 재시작하면 'kordoc' MCP 서버가 활성화됩니다.${c.reset}`)\r\n console.log(` ${c.dim}15개 도구: parse_document / parse_metadata / parse_pages / parse_table${c.reset}`)\r\n console.log(` ${c.dim} detect_format / compare_documents / parse_form / fill_form${c.reset}`)\r\n console.log(` ${c.dim} place_seal / patch_document / extract_profile / generate_document${c.reset}`)\r\n console.log(` ${c.dim} render_document / redact_document / parse_chunks${c.reset}`)\r\n console.log()\r\n}\r\n\r\nexport async function runSetup(): Promise<void> {\r\n const rl = createInterface({ input: stdin, output: stdout })\r\n\r\n try {\r\n await printBanner()\r\n\r\n stepHeader(1, 2, \"MCP 클라이언트 선택\")\r\n const clients = detectClients()\r\n clients.forEach((cl, i) => {\r\n const exists = cl.setupMethod === \"codex\" ? commandExists(\"codex\") : existsSync(cl.configPath)\r\n const badge = exists ? `${c.green} [감지됨]${c.reset}` : \"\"\r\n const num = `${c.cyan}${String(i + 1).padStart(2)}${c.reset}`\r\n console.log(` ${num}) ${c.white}${cl.name}${c.reset}${badge}`)\r\n })\r\n console.log()\r\n const clientInput = (await rl.question(` ${c.cyan}>${c.reset} 번호 (예: 1,3): `)).trim()\r\n\r\n if (!clientInput) {\r\n console.log(`\\n ${c.yellow}선택 없음${c.reset} — 수동 설정 안내:`)\r\n printManualConfig()\r\n return\r\n }\r\n\r\n const indices = clientInput\r\n .split(\",\")\r\n .map((s) => parseInt(s.trim(), 10) - 1)\r\n .filter((i) => i >= 0 && i < clients.length)\r\n\r\n if (indices.length === 0) {\r\n console.log(`\\n ${c.yellow}유효한 선택 없음${c.reset} — 수동 설정 안내:`)\r\n printManualConfig()\r\n return\r\n }\r\n\r\n console.log()\r\n stepHeader(2, 2, \"설정 파일 업데이트\")\r\n const entry = buildServerEntry()\r\n\r\n for (const idx of indices) {\r\n const client = clients[idx]\r\n await sleep(150)\r\n try {\r\n if (client.setupMethod === \"codex\") {\r\n await setupCodex()\r\n successLine(client.name, client.configPath)\r\n continue\r\n }\r\n const config = await readJsonFile(client.configPath)\r\n const key = client.format\r\n const serverEntry = key === \"context_servers\" ? buildZedEntry() : entry\r\n const servers = (config[key] ?? {}) as Record<string, unknown>\r\n servers[\"kordoc\"] = serverEntry\r\n config[key] = servers\r\n await writeJsonFile(client.configPath, config)\r\n successLine(client.name, client.configPath)\r\n } catch (err) {\r\n const msg = err instanceof Error ? err.message : String(err)\r\n failLine(client.name, msg)\r\n }\r\n }\r\n\r\n await printComplete()\r\n } catch (err) {\r\n if ((err as NodeJS.ErrnoException)?.code === \"ERR_USE_AFTER_CLOSE\") return\r\n throw err\r\n } finally {\r\n rl.close()\r\n }\r\n}\r\n\r\nfunction printManualConfig(): void {\r\n const entry = buildServerEntry()\r\n console.log()\r\n console.log(` ${c.dim}아래 JSON을 설정 파일의 mcpServers에 추가하세요:${c.reset}`)\r\n console.log()\r\n console.log(` ${c.cyan}\"kordoc\"${c.reset}: ${JSON.stringify(entry, null, 4)}`)\r\n console.log()\r\n}\r\n"],"mappings":";;;AAQA,SAAS,uBAAuB;AAChC,SAAS,UAAU,WAAW,aAAa;AAC3C,SAAS,gBAAgB;AACzB,SAAS,kBAAkB;AAC3B,SAAS,SAAS,SAAS,iBAAiB;AAC5C,SAAS,SAAS,gBAAgB;AAClC,SAAS,OAAO,cAAc;AAC9B,SAAS,iBAAiB;AAS1B,IAAM,gBAAgB,UAAU,QAAQ;AAGxC,SAAS,cAAc,SAA0B;AAC/C,QAAM,eAAe,QAAQ,IAAI,MAAM,KAAK,IAAI,MAAM,SAAS,EAAE,OAAO,OAAO;AAC/E,QAAM,aAAa,SAAS,MAAM,WAC7B,QAAQ,IAAI,SAAS,KAAK,uBAAuB,MAAM,GAAG,IAC3D,CAAC,EAAE;AACP,SAAO,YAAY,KAAK,CAAC,UAAU,WAAW,KAAK,CAAC,QAAQ,WAAW,QAAQ,OAAO,GAAG,OAAO,GAAG,GAAG,EAAE,CAAC,CAAC,CAAC;AAC7G;AAEA,SAAS,gBAAyC;AAChD,QAAM,OAAO,QAAQ;AACrB,QAAM,KAAK,SAAS;AACpB,QAAM,UAA0B,CAAC;AAEjC,QAAM,cAAsC;AAAA,IAC1C,QAAQ,QAAQ,MAAM,+DAA+D;AAAA,IACrF,OAAO,QAAQ,QAAQ,IAAI,SAAS,KAAK,QAAQ,MAAM,iBAAiB,GAAG,mCAAmC;AAAA,IAC9G,OAAO,QAAQ,MAAM,2CAA2C;AAAA,EAClE;AACA,QAAM,aAAa,YAAY,EAAE;AACjC,MAAI,WAAY,SAAQ,KAAK,EAAE,MAAM,kBAAkB,YAAY,YAAY,QAAQ,aAAa,CAAC;AAErG,UAAQ,KAAK,EAAE,MAAM,uDAAyB,YAAY,QAAQ,QAAQ,IAAI,GAAG,WAAW,GAAG,QAAQ,aAAa,CAAC;AACrH,UAAQ,KAAK,EAAE,MAAM,UAAU,YAAY,QAAQ,MAAM,kBAAkB,GAAG,QAAQ,aAAa,CAAC;AACpG,UAAQ,KAAK,EAAE,MAAM,mDAAqB,YAAY,QAAQ,QAAQ,IAAI,GAAG,kBAAkB,GAAG,QAAQ,UAAU,CAAC;AACrH,UAAQ,KAAK,EAAE,MAAM,YAAY,YAAY,QAAQ,MAAM,mCAAmC,GAAG,QAAQ,aAAa,CAAC;AACvH,UAAQ,KAAK,EAAE,MAAM,cAAc,YAAY,QAAQ,MAAM,uBAAuB,GAAG,QAAQ,aAAa,CAAC;AAE7G,QAAM,WAAmC;AAAA,IACvC,QAAQ,QAAQ,MAAM,oBAAoB;AAAA,IAC1C,OAAO,QAAQ,MAAM,2BAA2B;AAAA,IAChD,OAAO,QAAQ,MAAM,oBAAoB;AAAA,EAC3C;AACA,QAAM,UAAU,SAAS,EAAE;AAC3B,MAAI,QAAS,SAAQ,KAAK,EAAE,MAAM,OAAO,YAAY,SAAS,QAAQ,kBAAkB,CAAC;AAEzF,UAAQ,KAAK,EAAE,MAAM,eAAe,YAAY,QAAQ,MAAM,qCAAqC,GAAG,QAAQ,aAAa,CAAC;AAC5H,UAAQ,KAAK,EAAE,MAAM,SAAS,YAAY,QAAQ,MAAM,oBAAoB,GAAG,QAAQ,cAAc,aAAa,QAAQ,CAAC;AAE3H,SAAO;AACT;AAEA,eAAe,aAAa,MAAgD;AAC1E,MAAI,CAAC,WAAW,IAAI,EAAG,QAAO,CAAC;AAC/B,QAAM,MAAM,MAAM,SAAS,MAAM,OAAO;AACxC,SAAO,KAAK,MAAM,GAAG;AACvB;AAEA,eAAe,cAAc,MAAc,MAA8C;AACvF,QAAM,MAAM,QAAQ,IAAI;AACxB,MAAI,CAAC,WAAW,GAAG,EAAG,OAAM,MAAM,KAAK,EAAE,WAAW,KAAK,CAAC;AAC1D,QAAM,UAAU,MAAM,KAAK,UAAU,MAAM,MAAM,CAAC,IAAI,MAAM,OAAO;AACrE;AAOA,SAAS,mBAA4C;AACnD,MAAI,SAAS,MAAM,SAAS;AAC1B,WAAO,EAAE,SAAS,OAAO,MAAM,CAAC,MAAM,OAAO,MAAM,UAAU,KAAK,EAAE;AAAA,EACtE;AACA,SAAO,EAAE,SAAS,OAAO,MAAM,CAAC,MAAM,UAAU,KAAK,EAAE;AACzD;AAEA,SAAS,gBAAyC;AAChD,QAAM,OAAO,SAAS,MAAM,UACxB,EAAE,MAAM,OAAO,MAAM,CAAC,MAAM,OAAO,MAAM,UAAU,KAAK,EAAE,IAC1D,EAAE,MAAM,OAAO,MAAM,CAAC,MAAM,UAAU,KAAK,EAAE;AACjD,SAAO,EAAE,SAAS,KAAK;AACzB;AAGA,eAAe,aAA4B;AACzC,QAAM,QAAQ,iBAAiB;AAC/B,QAAM,UAAU,MAAM,SAAS;AAC/B,QAAM,OAAO,MAAM,MAAM;AACzB,QAAM,cAAc,SAAS,CAAC,OAAO,OAAO,UAAU,MAAM,SAAS,GAAG,IAAI,CAAC;AAC/E;AAGA,IAAM,MAAM;AACZ,IAAM,IAAI;AAAA,EACR,OAAO,GAAG,GAAG;AAAA,EAAM,MAAM,GAAG,GAAG;AAAA,EAAM,KAAK,GAAG,GAAG;AAAA,EAChD,MAAM,GAAG,GAAG;AAAA,EAAO,OAAO,GAAG,GAAG;AAAA,EAAO,QAAQ,GAAG,GAAG;AAAA,EACrD,KAAK,GAAG,GAAG;AAAA,EAAO,OAAO,GAAG,GAAG;AACjC;AAEA,SAAS,IAAI,GAAW,GAAW,GAAmB;AACpD,SAAO,GAAG,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;AAClC;AACA,SAAS,MAAM,IAA2B;AACxC,SAAO,IAAI,QAAQ,CAAC,QAAQ,WAAW,KAAK,EAAE,CAAC;AACjD;AACA,eAAe,UAAU,MAAc,QAAQ,IAAmB;AAChE,aAAW,MAAM,MAAM;AAAE,YAAQ,OAAO,MAAM,EAAE;AAAG,UAAM,MAAM,KAAK;AAAA,EAAE;AACtE,UAAQ,IAAI;AACd;AAEA,eAAe,cAA6B;AAC1C,QAAM,YAAY;AAAA,IAChB,IAAI,KAAK,KAAK,EAAE;AAAA,IAAG,IAAI,KAAK,KAAK,EAAE;AAAA,IAAG,IAAI,KAAK,KAAK,EAAE;AAAA,IACtD,IAAI,KAAK,KAAK,EAAE;AAAA,IAAG,IAAI,KAAK,KAAK,EAAE;AAAA,IAAG,IAAI,KAAK,KAAK,EAAE;AAAA,EACxD;AACA,QAAM,OAAO;AAAA,IACX;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACA,UAAQ,IAAI;AACZ,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,YAAQ,IAAI,GAAG,UAAU,IAAI,UAAU,MAAM,CAAC,GAAG,EAAE,IAAI,GAAG,KAAK,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE;AAC7E,UAAM,MAAM,EAAE;AAAA,EAChB;AACA,UAAQ,IAAI;AACZ,QAAM,UAAU,GAAG,EAAE,GAAG,sIAA8D,EAAE,KAAK,IAAI,EAAE;AACnG,UAAQ,IAAI;AACZ,UAAQ,IAAI,GAAG,EAAE,IAAI,KAAK,SAAI,OAAO,EAAE,CAAC,GAAG,EAAE,KAAK,EAAE;AACpD,UAAQ,IAAI;AACd;AAEA,SAAS,WAAW,MAAc,OAAe,OAAqB;AACpE,QAAM,OAAO,GAAG,EAAE,GAAG,GAAG,OAAI,OAAO,KAAK,IAAI,GAAG,KAAK,MAAM,MAAM,CAAC,CAAC,GAAG,EAAE,KAAK;AAC5E,UAAQ,IAAI,KAAK,EAAE,IAAI,GAAG,EAAE,IAAI,IAAI,IAAI,IAAI,KAAK,IAAI,EAAE,KAAK,IAAI,EAAE,KAAK,GAAG,EAAE,IAAI,GAAG,KAAK,GAAG,EAAE,KAAK,IAAI,IAAI,EAAE;AAC5G,UAAQ,IAAI;AACd;AAEA,SAAS,YAAY,OAAe,QAAsB;AACxD,UAAQ,IAAI,KAAK,EAAE,KAAK,GAAG,EAAE,IAAI,IAAI,EAAE,KAAK,IAAI,EAAE,KAAK,GAAG,KAAK,GAAG,EAAE,KAAK,GAAG,EAAE,GAAG,IAAI,MAAM,GAAG,EAAE,KAAK,EAAE;AACzG;AACA,SAAS,SAAS,OAAe,QAAsB;AACrD,UAAQ,IAAI,KAAK,EAAE,GAAG,GAAG,EAAE,IAAI,IAAI,EAAE,KAAK,IAAI,EAAE,KAAK,GAAG,KAAK,GAAG,EAAE,KAAK,GAAG,EAAE,GAAG,IAAI,MAAM,GAAG,EAAE,KAAK,EAAE;AACvG;AAEA,eAAe,gBAA+B;AAC5C,UAAQ,IAAI;AACZ,QAAM,MAAM;AAAA,IACV,KAAK,EAAE,KAAK,GAAG,EAAE,IAAI,SAAI,SAAI,OAAO,EAAE,CAAC,SAAI,EAAE,KAAK;AAAA,IAClD,KAAK,EAAE,KAAK,GAAG,EAAE,IAAI,SAAI,EAAE,KAAK,GAAG,IAAI,OAAO,EAAE,CAAC,GAAG,EAAE,KAAK,GAAG,EAAE,IAAI,kBAAkB,EAAE,KAAK,GAAG,IAAI,OAAO,EAAE,CAAC,GAAG,EAAE,KAAK,GAAG,EAAE,IAAI,SAAI,EAAE,KAAK;AAAA,IAC5I,KAAK,EAAE,KAAK,GAAG,EAAE,IAAI,SAAI,SAAI,OAAO,EAAE,CAAC,SAAI,EAAE,KAAK;AAAA,EACpD;AACA,aAAW,QAAQ,KAAK;AAAE,YAAQ,IAAI,IAAI;AAAG,UAAM,MAAM,EAAE;AAAA,EAAE;AAC7D,UAAQ,IAAI;AACZ,UAAQ,IAAI,KAAK,EAAE,GAAG,4IAAwC,EAAE,KAAK,EAAE;AACvE,UAAQ,IAAI,KAAK,EAAE,GAAG,qFAAsE,EAAE,KAAK,EAAE;AACrG,UAAQ,IAAI,KAAK,EAAE,GAAG,uEAAuE,EAAE,KAAK,EAAE;AACtG,UAAQ,IAAI,KAAK,EAAE,GAAG,8EAA8E,EAAE,KAAK,EAAE;AAC7G,UAAQ,IAAI,KAAK,EAAE,GAAG,6DAA6D,EAAE,KAAK,EAAE;AAC5F,UAAQ,IAAI;AACd;AAEA,eAAsB,WAA0B;AAC9C,QAAM,KAAK,gBAAgB,EAAE,OAAO,OAAO,QAAQ,OAAO,CAAC;AAE3D,MAAI;AACF,UAAM,YAAY;AAElB,eAAW,GAAG,GAAG,iDAAc;AAC/B,UAAM,UAAU,cAAc;AAC9B,YAAQ,QAAQ,CAAC,IAAI,MAAM;AACzB,YAAM,SAAS,GAAG,gBAAgB,UAAU,cAAc,OAAO,IAAI,WAAW,GAAG,UAAU;AAC7F,YAAM,QAAQ,SAAS,GAAG,EAAE,KAAK,wBAAS,EAAE,KAAK,KAAK;AACtD,YAAM,MAAM,GAAG,EAAE,IAAI,GAAG,OAAO,IAAI,CAAC,EAAE,SAAS,CAAC,CAAC,GAAG,EAAE,KAAK;AAC3D,cAAQ,IAAI,KAAK,GAAG,KAAK,EAAE,KAAK,GAAG,GAAG,IAAI,GAAG,EAAE,KAAK,GAAG,KAAK,EAAE;AAAA,IAChE,CAAC;AACD,YAAQ,IAAI;AACZ,UAAM,eAAe,MAAM,GAAG,SAAS,KAAK,EAAE,IAAI,IAAI,EAAE,KAAK,+BAAgB,GAAG,KAAK;AAErF,QAAI,CAAC,aAAa;AAChB,cAAQ,IAAI;AAAA,IAAO,EAAE,MAAM,4BAAQ,EAAE,KAAK,iDAAc;AACxD,wBAAkB;AAClB;AAAA,IACF;AAEA,UAAM,UAAU,YACb,MAAM,GAAG,EACT,IAAI,CAAC,MAAM,SAAS,EAAE,KAAK,GAAG,EAAE,IAAI,CAAC,EACrC,OAAO,CAAC,MAAM,KAAK,KAAK,IAAI,QAAQ,MAAM;AAE7C,QAAI,QAAQ,WAAW,GAAG;AACxB,cAAQ,IAAI;AAAA,IAAO,EAAE,MAAM,+CAAY,EAAE,KAAK,iDAAc;AAC5D,wBAAkB;AAClB;AAAA,IACF;AAEA,YAAQ,IAAI;AACZ,eAAW,GAAG,GAAG,oDAAY;AAC7B,UAAM,QAAQ,iBAAiB;AAE/B,eAAW,OAAO,SAAS;AACzB,YAAM,SAAS,QAAQ,GAAG;AAC1B,YAAM,MAAM,GAAG;AACf,UAAI;AACF,YAAI,OAAO,gBAAgB,SAAS;AAClC,gBAAM,WAAW;AACjB,sBAAY,OAAO,MAAM,OAAO,UAAU;AAC1C;AAAA,QACF;AACA,cAAM,SAAS,MAAM,aAAa,OAAO,UAAU;AACnD,cAAM,MAAM,OAAO;AACnB,cAAM,cAAc,QAAQ,oBAAoB,cAAc,IAAI;AAClE,cAAM,UAAW,OAAO,GAAG,KAAK,CAAC;AACjC,gBAAQ,QAAQ,IAAI;AACpB,eAAO,GAAG,IAAI;AACd,cAAM,cAAc,OAAO,YAAY,MAAM;AAC7C,oBAAY,OAAO,MAAM,OAAO,UAAU;AAAA,MAC5C,SAAS,KAAK;AACZ,cAAM,MAAM,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAC3D,iBAAS,OAAO,MAAM,GAAG;AAAA,MAC3B;AAAA,IACF;AAEA,UAAM,cAAc;AAAA,EACtB,SAAS,KAAK;AACZ,QAAK,KAA+B,SAAS,sBAAuB;AACpE,UAAM;AAAA,EACR,UAAE;AACA,OAAG,MAAM;AAAA,EACX;AACF;AAEA,SAAS,oBAA0B;AACjC,QAAM,QAAQ,iBAAiB;AAC/B,UAAQ,IAAI;AACZ,UAAQ,IAAI,KAAK,EAAE,GAAG,2GAAqC,EAAE,KAAK,EAAE;AACpE,UAAQ,IAAI;AACZ,UAAQ,IAAI,KAAK,EAAE,IAAI,WAAW,EAAE,KAAK,KAAK,KAAK,UAAU,OAAO,MAAM,CAAC,CAAC,EAAE;AAC9E,UAAQ,IAAI;AACd;","names":[]}
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/watch.ts"],"sourcesContent":["/** 디렉토리 감시 모드 — 새 문서 자동 변환 + Webhook 알림 */\r\n\r\nimport { watch, readFileSync, writeFileSync, mkdirSync, statSync, existsSync, realpathSync } from \"fs\"\r\nimport { lookup } from \"dns/promises\"\r\nimport { basename, resolve, extname, sep } from \"path\"\r\nimport { parse, detectFormat } from \"./index.js\"\r\nimport { toArrayBuffer } from \"./utils.js\"\r\nimport type { WatchOptions, ParseOptions } from \"./types.js\"\r\n\r\nexport const SUPPORTED_EXTENSIONS = new Set([\".hwp\", \".hwpx\", \".hml\", \".pdf\", \".xls\", \".xlsx\", \".docx\"])\r\nconst DEBOUNCE_MS = 1000\r\n/** 파일 쓰기 완료 판정: 연속 2회 동일 크기 확인 간격 */\r\nconst STABLE_CHECK_MS = 300\r\nconst MAX_FILE_SIZE = 500 * 1024 * 1024\r\n\r\n/**\r\n * Linux fs.watch recursive는 Node 19.1+ 전용 — 미지원 조합이면 크립틱\r\n * ERR_FEATURE_UNAVAILABLE 대신 명확한 안내로 즉시 실패시킨다.\r\n */\r\nexport function assertRecursiveWatchSupport(\r\n platform: NodeJS.Platform = process.platform,\r\n nodeVersion: string = process.versions.node,\r\n): void {\r\n if (platform !== \"linux\") return\r\n const [major = 0, minor = 0] = nodeVersion.split(\".\").map(Number)\r\n if (major >= 20 || (major === 19 && minor >= 1)) return\r\n throw new Error(\r\n `Linux에서 디렉토리 감시(fs.watch recursive)는 Node 19.1 이상이 필요합니다 (현재 ${nodeVersion}) — Node 20+ 업그레이드를 권장합니다`,\r\n )\r\n}\r\n\r\n/**\r\n * 동시 처리 상한 게이트 — 상한 초과분은 버리지 않고 대기 큐에 넣어 처리 완료 시\r\n * 순서대로 drain (기존: 상한 초과 return으로 새 파일이 무음 유실되던 것 방지).\r\n */\r\nexport function createTaskQueue(maxConcurrent: number, run: (key: string) => Promise<void>): (key: string) => void {\r\n let active = 0\r\n const inProgress = new Set<string>()\r\n const waiting: string[] = []\r\n const enqueue = (key: string): void => {\r\n if (inProgress.has(key)) return // 동일 파일 동시 처리 방지\r\n if (active >= maxConcurrent) {\r\n if (!waiting.includes(key)) waiting.push(key)\r\n return\r\n }\r\n inProgress.add(key)\r\n active++\r\n void run(key).finally(() => {\r\n inProgress.delete(key)\r\n active--\r\n const next = waiting.shift()\r\n if (next) enqueue(next)\r\n })\r\n }\r\n return enqueue\r\n}\r\n\r\n/**\r\n * 디렉토리를 감시하여 새 문서 파일을 자동 변환.\r\n *\r\n * @example\r\n * ```bash\r\n * kordoc watch ./incoming -d ./output --webhook https://api.example.com/docs\r\n * ```\r\n */\r\nexport async function watchDirectory(options: WatchOptions): Promise<void> {\r\n const { dir, outDir, webhook, format = \"markdown\", pages, silent } = options\r\n\r\n if (!existsSync(dir)) throw new Error(`디렉토리를 찾을 수 없습니다: ${dir}`)\r\n if (webhook) validateWebhookUrl(webhook)\r\n if (outDir) mkdirSync(outDir, { recursive: true })\r\n\r\n const log = silent ? () => {} : (msg: string) => process.stderr.write(msg + \"\\n\")\r\n log(`[kordoc watch] 감시 시작: ${resolve(dir)}`)\r\n if (outDir) log(`[kordoc watch] 출력: ${resolve(outDir)}`)\r\n if (webhook) log(`[kordoc watch] 웹훅: ${webhook}`)\r\n\r\n // 설정 로그 뒤·감시 시작 전에 검사 — 무엇을 하려다 막혔는지 보이고,\r\n // 옵션 파싱 계약 테스트(plugin-5)도 Linux+Node<19.1 에서 로그를 관측할 수 있다\r\n assertRecursiveWatchSupport()\r\n\r\n // 디바운스 맵\r\n const pending = new Map<string, ReturnType<typeof setTimeout>>()\r\n // 동시 처리 제한 — 메모리 폭주 방지 (초과분은 대기 큐로, createTaskQueue)\r\n const MAX_CONCURRENT = 3\r\n\r\n /** 파일 크기가 안정화될 때까지 대기 (쓰기 완료 감지) */\r\n const waitForStableSize = async (absPath: string): Promise<number> => {\r\n let prevSize = statSync(absPath).size\r\n await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\r\n if (!existsSync(absPath)) return 0\r\n const currSize = statSync(absPath).size\r\n if (currSize !== prevSize) {\r\n // 크기가 변했으면 한 번 더 대기\r\n await new Promise(r => setTimeout(r, STABLE_CHECK_MS))\r\n if (!existsSync(absPath)) return 0\r\n return statSync(absPath).size\r\n }\r\n return currSize\r\n }\r\n\r\n const processFile = async (filePath: string) => {\r\n const ext = extname(filePath).toLowerCase()\r\n if (!SUPPORTED_EXTENSIONS.has(ext)) return\r\n\r\n const fileName = basename(filePath)\r\n try {\r\n const rawPath = resolve(dir, filePath)\r\n if (!existsSync(rawPath)) return\r\n // 심볼릭 링크 해석 후 감시 디렉토리 외부 파일 차단\r\n let absPath: string\r\n try { absPath = realpathSync(rawPath) } catch { return }\r\n const realDir = realpathSync(resolve(dir))\r\n if (!absPath.startsWith(realDir + sep) && absPath !== realDir) return\r\n\r\n const fileSize = await waitForStableSize(absPath)\r\n if (fileSize > MAX_FILE_SIZE || fileSize === 0) return\r\n\r\n log(`[kordoc watch] 변환 중: ${fileName}`)\r\n\r\n const buffer = readFileSync(absPath)\r\n const arrayBuffer = toArrayBuffer(buffer)\r\n // filePath 전달 — 배포용 HWP의 COM fallback에 필요 (CLI와 동일)\r\n const parseOptions: ParseOptions = { filePath: absPath }\r\n if (pages) parseOptions.pages = pages\r\n const result = await parse(arrayBuffer, parseOptions)\r\n\r\n if (!result.success) {\r\n log(`[kordoc watch] 실패: ${fileName} — ${result.error}`)\r\n await sendWebhook(webhook, { file: fileName, format: detectFormat(arrayBuffer), success: false, error: result.error })\r\n return\r\n }\r\n\r\n const output = format === \"json\" ? JSON.stringify(result, null, 2) : result.markdown\r\n\r\n if (outDir) {\r\n const outExt = format === \"json\" ? \".json\" : \".md\"\r\n const outPath = resolve(outDir, fileName.replace(/\\.[^.]+$/, outExt))\r\n writeFileSync(outPath, output, \"utf-8\")\r\n log(`[kordoc watch] 완료: ${fileName} → ${basename(outPath)}`)\r\n } else {\r\n process.stdout.write(output + \"\\n\")\r\n }\r\n\r\n await sendWebhook(webhook, {\r\n file: fileName,\r\n format: result.fileType,\r\n success: true,\r\n markdown: format === \"markdown\" ? output.substring(0, 1000) : undefined,\r\n })\r\n } catch (err) {\r\n log(`[kordoc watch] 에러: ${fileName} — ${err instanceof Error ? err.message : err}`)\r\n }\r\n }\r\n\r\n const enqueue = createTaskQueue(MAX_CONCURRENT, (filePath) =>\r\n processFile(filePath).catch((err) => {\r\n process.stderr.write(`[kordoc watch] 처리 실패: ${filePath} — ${err instanceof Error ? err.message : String(err)}\\n`)\r\n }),\r\n )\r\n\r\n // fs.watch recursive (Node 18+ Windows/macOS, Node 19.1+ Linux — 위에서 사전 검증)\r\n const watcher = watch(dir, { recursive: true }, (event, filename) => {\r\n if (!filename) return\r\n const filePath = filename.toString()\r\n\r\n // 디바운스\r\n const existing = pending.get(filePath)\r\n if (existing) clearTimeout(existing)\r\n pending.set(filePath, setTimeout(() => {\r\n pending.delete(filePath)\r\n enqueue(filePath)\r\n }, DEBOUNCE_MS))\r\n })\r\n // 감시 자체의 런타임 에러(EMFILE 등)로 데몬이 죽지 않게 — 로그만 남기고 유지\r\n watcher.on(\"error\", (err) => {\r\n process.stderr.write(`[kordoc watch] 감시 오류: ${err instanceof Error ? err.message : String(err)}\\n`)\r\n })\r\n\r\n // 프로세스 종료 방지 (Ctrl+C로 종료)\r\n return new Promise(() => {})\r\n}\r\n\r\n/** Webhook URL 검증 — SSRF 방지: http/https만 허용, localhost/private IP 차단 */\r\nfunction validateWebhookUrl(url: string): void {\r\n let parsed: URL\r\n try {\r\n parsed = new URL(url)\r\n } catch {\r\n throw new Error(`유효하지 않은 webhook URL: ${url}`)\r\n }\r\n if (parsed.protocol !== \"http:\" && parsed.protocol !== \"https:\") {\r\n throw new Error(`허용되지 않는 webhook 프로토콜: ${parsed.protocol}`)\r\n }\r\n const hostname = parsed.hostname.toLowerCase()\r\n if (\r\n hostname === \"localhost\" ||\r\n hostname === \"[::1]\" ||\r\n hostname.startsWith(\"127.\") ||\r\n hostname.startsWith(\"10.\") ||\r\n hostname.startsWith(\"192.168.\") ||\r\n /^172\\.(1[6-9]|2\\d|3[01])\\./.test(hostname) ||\r\n hostname === \"0.0.0.0\" ||\r\n hostname.startsWith(\"169.254.\") ||\r\n hostname.endsWith(\".local\") ||\r\n // IPv6 사설 대역\r\n hostname.startsWith(\"[fc\") ||\r\n hostname.startsWith(\"[fd\") ||\r\n hostname.startsWith(\"[fe80:\") ||\r\n hostname === \"[::0]\" ||\r\n hostname === \"[::]\" ||\r\n // 클라우드 메타데이터 엔드포인트\r\n hostname === \"metadata.google.internal\" ||\r\n hostname === \"metadata.google\" ||\r\n // 16진수/8진수/10진수 정수 IP 인코딩 우회 방지\r\n /^0x[0-9a-f]+$/i.test(hostname) ||\r\n /^0[0-7]+$/.test(hostname) ||\r\n /^\\d+$/.test(hostname)\r\n ) {\r\n throw new Error(`내부 네트워크 대상 webhook은 허용되지 않습니다: ${hostname}`)\r\n }\r\n}\r\n\r\n/** IP 리터럴의 사설/내부 대역 여부 — DNS 해석 결과 재검증용 (IPv4-mapped IPv6 포함) */\r\nexport function isPrivateIp(ip: string): boolean {\r\n let addr = ip.toLowerCase()\r\n // IPv4-mapped IPv6 (::ffff:10.0.0.1) → 내장 IPv4로 재검사\r\n const mapped = /^::ffff:(\\d+\\.\\d+\\.\\d+\\.\\d+)$/.exec(addr)\r\n if (mapped) addr = mapped[1]\r\n if (/^\\d+\\.\\d+\\.\\d+\\.\\d+$/.test(addr)) {\r\n const [a, b] = addr.split(\".\").map(Number)\r\n return (\r\n a === 0 || a === 10 || a === 127 ||\r\n (a === 100 && b >= 64 && b <= 127) || // CGNAT 100.64/10\r\n (a === 169 && b === 254) ||\r\n (a === 172 && b >= 16 && b <= 31) ||\r\n (a === 192 && b === 168)\r\n )\r\n }\r\n // IPv6 — loopback/미지정, ULA fc00::/7, link-local fe80::/10\r\n return addr === \"::\" || addr === \"::1\" || /^f[cd]/.test(addr) || /^fe[89ab]/.test(addr)\r\n}\r\n\r\nasync function sendWebhook(url: string | undefined, payload: Record<string, unknown>): Promise<void> {\r\n if (!url) return\r\n try {\r\n validateWebhookUrl(url)\r\n // DNS 해석 후 사설 IP 재검증 — 문자열 검사만으로는 내부망 도메인을 못 거른다\r\n const hostname = new URL(url).hostname.replace(/^\\[|\\]$/g, \"\")\r\n const addrs = await lookup(hostname, { all: true, verbatim: true })\r\n const bad = addrs.find((a) => isPrivateIp(a.address))\r\n if (bad) throw new Error(`webhook 대상이 내부 네트워크로 해석됩니다: ${hostname} → ${bad.address}`)\r\n await fetch(url, {\r\n method: \"POST\",\r\n headers: { \"Content-Type\": \"application/json\" },\r\n body: JSON.stringify({ ...payload, timestamp: new Date().toISOString() }),\r\n redirect: \"error\",\r\n })\r\n } catch (err) {\r\n process.stderr.write(`[kordoc watch] webhook 전송 실패: ${err instanceof Error ? err.message : String(err)}\\n`)\r\n }\r\n}\r\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":[]}
File without changes