kordoc 4.9.1 → 4.10.0
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.
- package/README.md +51 -0
- package/dist/{-KIYHEVUW.js → -AN4MKOOO.js} +9 -9
- package/dist/{chunk-65TL66UD.cjs → chunk-2FNQN23O.cjs} +3 -3
- package/dist/chunk-2FNQN23O.cjs.map +1 -0
- package/dist/{chunk-Z2SX4E7K.js → chunk-34IBS3JN.js} +3 -3
- package/dist/{chunk-JHXAQDXS.js → chunk-4EJLRZ7A.js} +82 -22
- package/dist/chunk-4EJLRZ7A.js.map +1 -0
- package/dist/{chunk-4R73ASZF.cjs → chunk-4SIKWOLQ.cjs} +2 -2
- package/dist/chunk-4SIKWOLQ.cjs.map +1 -0
- package/dist/{chunk-INV3W43Y.js → chunk-7UUQXASL.js} +2 -2
- package/dist/{chunk-INV3W43Y.js.map → chunk-7UUQXASL.js.map} +1 -1
- package/dist/{chunk-SFSB3FAL.js → chunk-AOTA4XTU.js} +2 -2
- package/dist/{chunk-HG7GXBVP.js → chunk-B4MQXOU5.js} +2 -2
- package/dist/chunk-DCZVOIEO.cjs.map +1 -1
- package/dist/{chunk-Q3IVAWOK.js → chunk-EWUFLCRP.js} +3 -3
- package/dist/{chunk-VGRIWMTQ.js → chunk-F3OPGGQ2.js} +2 -2
- package/dist/chunk-GS7T56RP.cjs.map +1 -1
- package/dist/{chunk-L27YEW3T.js → chunk-HO7N6T36.js} +3 -3
- package/dist/{chunk-TC7EDYZD.js → chunk-JAY6BXZ6.js} +2 -2
- package/dist/{chunk-HFZIYAQQ.js → chunk-KEF4UQYI.js} +2 -2
- package/dist/{chunk-XINP35F4.js → chunk-N3DLQ2YH.js} +1 -1
- package/dist/{chunk-BN33O25L.js.map → chunk-N3DLQ2YH.js.map} +1 -1
- package/dist/{chunk-GNRBCV4R.js → chunk-NE7GR65O.js} +3 -3
- package/dist/{chunk-3PY5YCJP.js → chunk-NMXDXHHW.js} +2 -2
- package/dist/{chunk-3PY5YCJP.js.map → chunk-NMXDXHHW.js.map} +1 -1
- package/dist/{chunk-BN33O25L.js → chunk-ONQHGZTS.js} +1 -1
- package/dist/{chunk-XINP35F4.js.map → chunk-ONQHGZTS.js.map} +1 -1
- package/dist/{chunk-D7EMDOX7.cjs → chunk-OQNZCCJI.cjs} +6 -6
- package/dist/chunk-OQNZCCJI.cjs.map +1 -0
- package/dist/{chunk-GU4NCNCT.js → chunk-PDXRLQD6.js} +2 -2
- package/dist/{chunk-UIQWUYOJ.js → chunk-QYXQJPB7.js} +3 -3
- package/dist/{chunk-55Q76S6S.js → chunk-SV2YYPBO.js} +3 -3
- package/dist/{chunk-L5CDTOC3.cjs → chunk-T7GRHAIW.cjs} +1 -1
- package/dist/chunk-T7GRHAIW.cjs.map +1 -0
- package/dist/{chunk-2ULV2DX4.cjs → chunk-V7F5765M.cjs} +4 -4
- package/dist/chunk-V7F5765M.cjs.map +1 -0
- package/dist/{chunk-TN6AALZG.js → chunk-VR3ECURZ.js} +2 -2
- package/dist/{chunk-SGTGSV4A.js → chunk-WLUDCIOK.js} +2 -2
- package/dist/{chunk-5CAKNM2K.js → chunk-YVPYCQOG.js} +2 -2
- package/dist/{chunk-HOSNMNKU.cjs → chunk-ZOLKTMBS.cjs} +11 -11
- package/dist/chunk-ZOLKTMBS.cjs.map +1 -0
- package/dist/chunks-JUBR2TW7.js +10 -0
- package/dist/cli.js +37 -33
- package/dist/cli.js.map +1 -1
- package/dist/{formula-I5D3UNLV.cjs → formula-4QA6ZSFP.cjs} +9 -9
- package/dist/formula-4QA6ZSFP.cjs.map +1 -0
- package/dist/{formula-5DQQ4MFT.js → formula-LFDXPFQH.js} +4 -4
- package/dist/{formula-S4O7SBN3.js → formula-UYS7Z7IH.js} +3 -3
- package/dist/{image-ocr-LIH2E7WC.js → image-ocr-3YP5I2H4.js} +6 -6
- package/dist/{image-ocr-SQXFTNYR.js → image-ocr-NOU6O7ZY.js} +8 -8
- package/dist/{image-ocr-BKUYR4WU.cjs → image-ocr-POPCJGL5.cjs} +12 -12
- package/dist/image-ocr-POPCJGL5.cjs.map +1 -0
- package/dist/index.cjs +254 -195
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.cts +2 -0
- package/dist/index.d.ts +2 -0
- package/dist/index.js +77 -18
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +16 -16
- package/dist/{model-bundle-4JMY5E37.js → model-bundle-AUKJF6GA.js} +5 -5
- package/dist/{models-D7X5YBWX.js → models-QKSQSUWJ.js} +5 -5
- package/dist/page-range-P7SDW6LR.cjs.map +1 -1
- package/dist/{parser-MY7JJISJ.js → parser-6HKVNDZM.js} +7 -7
- package/dist/{parser-ZLHOAZAA.cjs → parser-VAIMONF2.cjs} +27 -27
- package/dist/parser-VAIMONF2.cjs.map +1 -0
- package/dist/{parser-GHRXWFJN.js → parser-VVGAZIAX.js} +7 -7
- package/dist/pdf-ocr-4CWHKKTX.cjs +14 -0
- package/dist/pdf-ocr-4CWHKKTX.cjs.map +1 -0
- package/dist/pdf-ocr-D6HR23OB.js +13 -0
- package/dist/pdf-ocr-WIMNTEVH.js +16 -0
- package/dist/{profile-io-EAPI6M7O.js → profile-io-PYCSBITE.js} +3 -3
- package/dist/{rasterize-UIDZAEWB.js → rasterize-TUBHYALD.js} +2 -2
- package/dist/render-QW5Z4IMW.js +10 -0
- package/dist/seal-OTOP5CLA.js +10 -0
- package/dist/{setup-UHX3HHWY.js → setup-6EUKBDMC.js} +3 -3
- package/dist/{watch-23MV3PBX.js → watch-JVGXPTCI.js} +10 -10
- package/package.json +1 -1
- package/dist/chunk-2ULV2DX4.cjs.map +0 -1
- package/dist/chunk-4R73ASZF.cjs.map +0 -1
- package/dist/chunk-65TL66UD.cjs.map +0 -1
- package/dist/chunk-D7EMDOX7.cjs.map +0 -1
- package/dist/chunk-HOSNMNKU.cjs.map +0 -1
- package/dist/chunk-JHXAQDXS.js.map +0 -1
- package/dist/chunk-L5CDTOC3.cjs.map +0 -1
- package/dist/chunks-FB344JZT.js +0 -10
- package/dist/formula-I5D3UNLV.cjs.map +0 -1
- package/dist/image-ocr-BKUYR4WU.cjs.map +0 -1
- package/dist/parser-ZLHOAZAA.cjs.map +0 -1
- package/dist/pdf-ocr-FESGFQYN.cjs +0 -14
- package/dist/pdf-ocr-FESGFQYN.cjs.map +0 -1
- package/dist/pdf-ocr-IEAXGG3D.js +0 -13
- package/dist/pdf-ocr-Z6CHIYEM.js +0 -16
- package/dist/render-XL7ZVAHK.js +0 -10
- package/dist/seal-CY7EIWDJ.js +0 -10
- /package/dist/{-KIYHEVUW.js.map → -AN4MKOOO.js.map} +0 -0
- /package/dist/{chunk-Z2SX4E7K.js.map → chunk-34IBS3JN.js.map} +0 -0
- /package/dist/{chunk-SFSB3FAL.js.map → chunk-AOTA4XTU.js.map} +0 -0
- /package/dist/{chunk-HG7GXBVP.js.map → chunk-B4MQXOU5.js.map} +0 -0
- /package/dist/{chunk-Q3IVAWOK.js.map → chunk-EWUFLCRP.js.map} +0 -0
- /package/dist/{chunk-VGRIWMTQ.js.map → chunk-F3OPGGQ2.js.map} +0 -0
- /package/dist/{chunk-L27YEW3T.js.map → chunk-HO7N6T36.js.map} +0 -0
- /package/dist/{chunk-TC7EDYZD.js.map → chunk-JAY6BXZ6.js.map} +0 -0
- /package/dist/{chunk-HFZIYAQQ.js.map → chunk-KEF4UQYI.js.map} +0 -0
- /package/dist/{chunk-GNRBCV4R.js.map → chunk-NE7GR65O.js.map} +0 -0
- /package/dist/{chunk-GU4NCNCT.js.map → chunk-PDXRLQD6.js.map} +0 -0
- /package/dist/{chunk-UIQWUYOJ.js.map → chunk-QYXQJPB7.js.map} +0 -0
- /package/dist/{chunk-55Q76S6S.js.map → chunk-SV2YYPBO.js.map} +0 -0
- /package/dist/{chunk-TN6AALZG.js.map → chunk-VR3ECURZ.js.map} +0 -0
- /package/dist/{chunk-SGTGSV4A.js.map → chunk-WLUDCIOK.js.map} +0 -0
- /package/dist/{chunk-5CAKNM2K.js.map → chunk-YVPYCQOG.js.map} +0 -0
- /package/dist/{chunks-FB344JZT.js.map → chunks-JUBR2TW7.js.map} +0 -0
- /package/dist/{formula-5DQQ4MFT.js.map → formula-LFDXPFQH.js.map} +0 -0
- /package/dist/{formula-S4O7SBN3.js.map → formula-UYS7Z7IH.js.map} +0 -0
- /package/dist/{image-ocr-LIH2E7WC.js.map → image-ocr-3YP5I2H4.js.map} +0 -0
- /package/dist/{image-ocr-SQXFTNYR.js.map → image-ocr-NOU6O7ZY.js.map} +0 -0
- /package/dist/{model-bundle-4JMY5E37.js.map → model-bundle-AUKJF6GA.js.map} +0 -0
- /package/dist/{models-D7X5YBWX.js.map → models-QKSQSUWJ.js.map} +0 -0
- /package/dist/{parser-MY7JJISJ.js.map → parser-6HKVNDZM.js.map} +0 -0
- /package/dist/{parser-GHRXWFJN.js.map → parser-VVGAZIAX.js.map} +0 -0
- /package/dist/{pdf-ocr-IEAXGG3D.js.map → pdf-ocr-D6HR23OB.js.map} +0 -0
- /package/dist/{pdf-ocr-Z6CHIYEM.js.map → pdf-ocr-WIMNTEVH.js.map} +0 -0
- /package/dist/{profile-io-EAPI6M7O.js.map → profile-io-PYCSBITE.js.map} +0 -0
- /package/dist/{rasterize-UIDZAEWB.js.map → rasterize-TUBHYALD.js.map} +0 -0
- /package/dist/{render-XL7ZVAHK.js.map → render-QW5Z4IMW.js.map} +0 -0
- /package/dist/{seal-CY7EIWDJ.js.map → seal-OTOP5CLA.js.map} +0 -0
- /package/dist/{setup-UHX3HHWY.js.map → setup-6EUKBDMC.js.map} +0 -0
- /package/dist/{watch-23MV3PBX.js.map → watch-JVGXPTCI.js.map} +0 -0
package/README.md
CHANGED
|
@@ -89,6 +89,19 @@ MCP 등록 대신 스킬(SKILL.md) 형태로 쓰려면:
|
|
|
89
89
|
|
|
90
90
|
---
|
|
91
91
|
|
|
92
|
+
## v4.10.0 변경사항
|
|
93
|
+
|
|
94
|
+
- **📡 실패 계약 전 포맷 확장 (#69)**: 실패 JSON(`success:false` + `code`)이 `--format json` 전용에서 **markdown·chunks 포함 전 포맷**으로 확장됐습니다. 헤드리스 파이프라인이 stderr 한국어 문구 파싱 없이 원인 코드로 분기할 수 있습니다. `code` 값 집합·종료 코드 규칙·안정성 보증을 README [실패 계약](#실패-계약--기계-판독-가능한-실패-json-v4100-69) 표로 문서화했습니다. (@sorbetsharkroundhand 제보)
|
|
95
|
+
- **🖼️ `images/manifest.json` (#70)**: 이미지 저장 시 `[ { name, mimeType, bytes, source } ]` manifest 를 함께 방출합니다. `mimeType` 은 매직바이트 실측 우선이라 확장자 추측이 필요 없습니다. DOCX 이미지의 미지 확장자를 `image/png` 로 단정하던 오폴백도 실측 판별로 교체. (@sorbetsharkroundhand 제보)
|
|
96
|
+
- **🩹 patchHwp soft-wrap 축소 치환 무결성 (#71)**: 자동 줄바꿈으로 LINE_SEG 가 2세그먼트 이상인 문단을 더 짧은 한 줄로 치환하면 잔존 세그먼트의 textpos 가 새 nChars 밖을 가리켜 **한컴이 "문서 손상/변조" 경고로 열기를 거부**하던 버그 수정 — 범위를 벗어난 꼬리 세그먼트만 잘라내고 lineSegCount 를 정합화합니다(유효 세그먼트는 보존해 soft-wrap 렌더 유지). 재작성 압축 스트림에 한컴 실저장본의 8바이트 꼬리(CRC32+비압축 크기, 코퍼스 실측 7/7)도 복원합니다. (@heesun-woodi 제보)
|
|
97
|
+
|
|
98
|
+
## v4.9.2 변경사항
|
|
99
|
+
|
|
100
|
+
`<right>` 출처행이 두 줄을 넘어가는 실제 보고서를 한글로 열어 확인하고 고쳤습니다.
|
|
101
|
+
|
|
102
|
+
- **📌 다줄 출처행 왼쪽정렬 폴백**: `<right>` 문단(출처행·발신일자)이 본문폭을 넘겨 줄바꿈되면 오른쪽정렬 특성상 **왼쪽 끝이 줄마다 들쭉날쭉**해져 들여쓰기가 깨진 듯 보였습니다(한글 실렌더 QA — 3줄 출처행). 이제 조판 폭을 미리 계산해 한 줄에 들어가면 종전대로 우측 배치(실측 GT6/GT7/GT11 관행), 넘치면 왼쪽정렬로 방출합니다.
|
|
103
|
+
- **🔤 출처행 캡션체**: `출처:`·`자료:`로 시작하는 `<right>` 행은 본문과 같은 크기라 캡션으로 읽히지 않았습니다. 전용 charPr을 신설해 **본문보다 3pt 작게**(하한 10pt), 실측 프리셋(개조식·보고서·계획서)은 한양중고딕, 그 외 프리셋은 본문 폰트로 방출합니다. 표 아래 출처를 본문과 구분하는 실무 관행을 따른 것으로, 발신일자 등 다른 `<right>` 행은 본문 크기를 유지합니다.
|
|
104
|
+
|
|
92
105
|
## v4.9.1 변경사항
|
|
93
106
|
|
|
94
107
|
실제 자치구 검토보고서를 뽑아 한컴으로 열어보고, 눈에 걸리는 것을 코퍼스 555건 실측과 대조해 고쳤습니다. 세 결함 모두 **양식 파일 하나의 저장값을 실무 관행보다 앞세운 것**이 원인이었습니다.
|
|
@@ -881,6 +894,44 @@ npx kordoc watch ./문서 --webhook https://api/hook # 웹훅 알림
|
|
|
881
894
|
> 생성물을 검수하려면 원고 마크다운을 넘기거나, `kordoc 문서.hwpx | kordoc lint -` 로 파이프하세요.
|
|
882
895
|
> (공문서 생성 시에는 `generate` 가 같은 13룰 경고를 함께 냅니다.)
|
|
883
896
|
|
|
897
|
+
### 실패 계약 — 기계 판독 가능한 실패 JSON (v4.10.0+, #69)
|
|
898
|
+
|
|
899
|
+
변환 실패는 **모든 `--format`(markdown·json·chunks)에서** stdout 에 동일한 실패 JSON 을 내고 exit 1 로 끝납니다. 호출자는 stderr 문구가 아니라 `code` 로 분기하면 됩니다.
|
|
900
|
+
|
|
901
|
+
```json
|
|
902
|
+
{ "success": false, "fileType": "hwpx", "error": "암호화된 문서입니다 …", "code": "ENCRYPTED" }
|
|
903
|
+
```
|
|
904
|
+
|
|
905
|
+
성공 출력과 충돌하지 않습니다 — `markdown` 성공은 마크다운 텍스트, `chunks` 성공은 JSON **배열**, 실패는 언제나 `success:false` **객체**입니다. `-o`/`-d` 지정 시 실패한 파일의 출력물은 생성되지 않습니다. 여러 파일 입력 시 실패한 파일마다 실패 JSON 이 한 개씩 나옵니다.
|
|
906
|
+
|
|
907
|
+
| `code` | 의미 |
|
|
908
|
+
|---|---|
|
|
909
|
+
| `ENCRYPTED` | 열기 암호 문서 (`--password` 필요 또는 불일치) |
|
|
910
|
+
| `DRM_PROTECTED` | 한컴 문서보안(DRM) — 열 수 없음 |
|
|
911
|
+
| `UNSUPPORTED_FORMAT` | 지원하지 않는 파일 형식 |
|
|
912
|
+
| `CORRUPTED` | 시그니처 불일치·복구 불가 손상 |
|
|
913
|
+
| `IMAGE_BASED_PDF` | 텍스트층 없는 스캔 PDF (`--ocr` 필요) |
|
|
914
|
+
| `ZIP_BOMB` / `DECOMPRESSION_BOMB` | 압축 폭탄 방어 발동 |
|
|
915
|
+
| `NO_SECTIONS` | 본문 섹션 없음 |
|
|
916
|
+
| `OUTPUT_TOO_LARGE` | 출력 직렬화가 런타임 문자열 한계 초과 (#65) |
|
|
917
|
+
| `MISSING_DEPENDENCY` | 선택 의존성 미설치 (pdfjs-dist 등) |
|
|
918
|
+
| `EMPTY_INPUT` | 빈 입력 |
|
|
919
|
+
| `PARSE_ERROR` | 그 외 파싱 실패 |
|
|
920
|
+
|
|
921
|
+
**안정성 보증**: 종료 코드 규칙(성공 0 / 실패 1)과 실패 JSON 의 필드(`success`·`fileType`·`error`·`code`)는 유지되고, `code` 값 집합은 **추가만** 됩니다(기존 값 변경·제거 없음). `error` 문구는 사람용이라 계약이 아닙니다.
|
|
922
|
+
|
|
923
|
+
### 이미지 번들 — `images/manifest.json` (v4.10.0+, #70)
|
|
924
|
+
|
|
925
|
+
`-o`/`-d` 로 저장할 때 추출 이미지는 `images/` 에 저장되고, 같은 폴더에 `manifest.json` 이 함께 나옵니다. 소비자는 확장자·매직바이트 추측 없이 manifest 로 형식을 분기하면 됩니다.
|
|
926
|
+
|
|
927
|
+
```json
|
|
928
|
+
[ { "name": "image_001.png", "mimeType": "image/png", "bytes": 68, "source": "BinData/image1.png" } ]
|
|
929
|
+
```
|
|
930
|
+
|
|
931
|
+
- `mimeType` 은 **매직바이트 실측 우선**(PNG/JPEG/GIF/BMP/WMF/EMF 판별) — 확장자 유래 선언값과 다르면 실측이 이깁니다. 실측 불가 형식(TIFF·SVG 등)은 선언값을 그대로 둡니다.
|
|
932
|
+
- `source` 는 원본 컨테이너 항목명(HWPX/DOCX ZIP 경로, HWP5 BinData 스토리지명). PDF 처럼 재인코딩으로 합성된 이미지에는 없습니다.
|
|
933
|
+
- 나올 수 있는 확장자 집합: **PDF 는 항상 `png`**(순수 JS 재인코딩) · HWP5 는 스니프 유래 `png/jpg/gif/bmp`(WMF/EMF 는 `bin`) · HWPX 는 원본 확장자 유래 `png/jpg/gif/bmp/tif/wmf/emf/svg`(미지 확장자는 `bin`) · DOCX 는 원본 확장자 유지. 이미지 재인코딩은 하지 않습니다(원형 유지) — 형식 판정은 manifest 를 믿으세요.
|
|
934
|
+
|
|
884
935
|
## MCP 서버 (Claude / Cursor / Windsurf / Codex)
|
|
885
936
|
|
|
886
937
|
**자동 설치 (추천)**:
|
|
@@ -56,9 +56,9 @@ import {
|
|
|
56
56
|
unknownFontWarnings,
|
|
57
57
|
usesGaejosikMunche,
|
|
58
58
|
validateHwpx
|
|
59
|
-
} from "./chunk-
|
|
59
|
+
} from "./chunk-4EJLRZ7A.js";
|
|
60
60
|
import "./chunk-7FCNOKOR.js";
|
|
61
|
-
import "./chunk-
|
|
61
|
+
import "./chunk-N3DLQ2YH.js";
|
|
62
62
|
import "./chunk-MOL7MDBG.js";
|
|
63
63
|
import {
|
|
64
64
|
applySplices,
|
|
@@ -66,7 +66,7 @@ import {
|
|
|
66
66
|
buildRangeSplices,
|
|
67
67
|
placeSealHwpx,
|
|
68
68
|
scanSectionXml
|
|
69
|
-
} from "./chunk-
|
|
69
|
+
} from "./chunk-EWUFLCRP.js";
|
|
70
70
|
import {
|
|
71
71
|
DEFAULT_REDACT_RULES,
|
|
72
72
|
redactMarkdown,
|
|
@@ -74,10 +74,10 @@ import {
|
|
|
74
74
|
} from "./chunk-DZIXKL7E.js";
|
|
75
75
|
import {
|
|
76
76
|
blocksToChunks
|
|
77
|
-
} from "./chunk-
|
|
77
|
+
} from "./chunk-F3OPGGQ2.js";
|
|
78
78
|
import {
|
|
79
79
|
blocksToMarkdown
|
|
80
|
-
} from "./chunk-
|
|
80
|
+
} from "./chunk-B4MQXOU5.js";
|
|
81
81
|
import {
|
|
82
82
|
SPACE_EM_FIXED,
|
|
83
83
|
SPACE_EM_FONT,
|
|
@@ -87,11 +87,11 @@ import {
|
|
|
87
87
|
renderHwpxToSvg,
|
|
88
88
|
simulateWrap,
|
|
89
89
|
simulateWrapKeepWord
|
|
90
|
-
} from "./chunk-
|
|
91
|
-
import "./chunk-
|
|
90
|
+
} from "./chunk-QYXQJPB7.js";
|
|
91
|
+
import "./chunk-JAY6BXZ6.js";
|
|
92
92
|
import {
|
|
93
93
|
VERSION
|
|
94
|
-
} from "./chunk-
|
|
94
|
+
} from "./chunk-NMXDXHHW.js";
|
|
95
95
|
export {
|
|
96
96
|
BUILTIN_TEMPLATES,
|
|
97
97
|
DEFAULT_REDACT_RULES,
|
|
@@ -169,4 +169,4 @@ export {
|
|
|
169
169
|
usesGaejosikMunche,
|
|
170
170
|
validateHwpx
|
|
171
171
|
};
|
|
172
|
-
//# sourceMappingURL=-
|
|
172
|
+
//# sourceMappingURL=-AN4MKOOO.js.map
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
"use strict";Object.defineProperty(exports, "__esModule", {value: true}); function _optionalChain(ops) { let lastAccessLHS = undefined; let value = ops[0]; let i = 1; while (i < ops.length) { const op = ops[i]; const fn = ops[i + 1]; i += 2; if ((op === 'optionalAccess' || op === 'optionalCall') && value == null) { return undefined; } if (op === 'access' || op === 'optionalAccess') { lastAccessLHS = value; value = fn(value); } else if (op === 'call' || op === 'optionalCall') { value = fn((...args) => value.call(lastAccessLHS, ...args)); lastAccessLHS = undefined; } } return value; }
|
|
2
2
|
|
|
3
|
-
var
|
|
3
|
+
var _chunk4SIKWOLQcjs = require('./chunk-4SIKWOLQ.cjs');
|
|
4
4
|
|
|
5
5
|
// src/pdf/formula/models.ts
|
|
6
6
|
var _crypto = require('crypto');
|
|
@@ -23,7 +23,7 @@ function isOfflineMode() {
|
|
|
23
23
|
}
|
|
24
24
|
function assertNetworkAllowed(what, hint) {
|
|
25
25
|
if (!isOfflineMode()) return;
|
|
26
|
-
throw new (0,
|
|
26
|
+
throw new (0, _chunk4SIKWOLQcjs.KordocError)(
|
|
27
27
|
`\uD3D0\uC1C4\uB9DD \uBAA8\uB4DC(KORDOC_OFFLINE)\uC5D0\uC11C \uCC28\uB2E8\uB428: ${what}` + (hint ? ` \u2014 ${hint}` : "")
|
|
28
28
|
);
|
|
29
29
|
}
|
|
@@ -266,4 +266,4 @@ async function sha256OfFile(p) {
|
|
|
266
266
|
|
|
267
267
|
|
|
268
268
|
exports.MFD_MODEL = MFD_MODEL; exports.MFR_ENCODER_MODEL = MFR_ENCODER_MODEL; exports.MFR_DECODER_MODEL = MFR_DECODER_MODEL; exports.MFR_TOKENIZER = MFR_TOKENIZER; exports.ALL_FORMULA_MODELS = ALL_FORMULA_MODELS; exports.getModelsDir = getModelsDir; exports.getFormulaModelsDir = getFormulaModelsDir; exports.getFormulaModelStatus = getFormulaModelStatus; exports.ensureModelsIn = ensureModelsIn; exports.ensureFormulaModels = ensureFormulaModels; exports.ensureSingleModel = ensureSingleModel;
|
|
269
|
-
//# sourceMappingURL=chunk-
|
|
269
|
+
//# sourceMappingURL=chunk-2FNQN23O.cjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/chunk-2FNQN23O.cjs","../src/pdf/formula/models.ts","../src/shared/offline.ts"],"names":[],"mappings":"AAAA;AACE;AACF,wDAA6B;AAC7B;AACA;ACMA,gCAA2B;AAC3B,wBAAiC;AACjC,uCAA4C;AAC5C;AACA,wBAAwB;AACxB,4BAA8B;AAC9B,4CAAyB;AACzB,gCAAyB;ADJzB;AACA;AEKA,IAAM,OAAA,kBAAS,IAAI,GAAA,CAAI,CAAC,GAAA,EAAK,MAAA,EAAQ,KAAA,EAAO,IAAI,CAAC,CAAA;AAEjD,SAAS,OAAA,CAAQ,IAAA,EAAuB;AACtC,EAAA,MAAM,EAAA,EAAI,OAAA,CAAQ,GAAA,CAAI,IAAI,CAAA;AAC1B,EAAA,OAAO,CAAC,CAAC,EAAA,GAAK,MAAA,CAAO,GAAA,CAAI,CAAA,CAAE,IAAA,CAAK,CAAA,CAAE,WAAA,CAAY,CAAC,CAAA;AACjD;AAGO,SAAS,aAAA,CAAA,EAAyB;AACvC,EAAA,OAAO,OAAA,CAAQ,gBAAgB,CAAA;AACjC;AAOO,SAAS,oBAAA,CAAqB,IAAA,EAAc,IAAA,EAAqB;AACtE,EAAA,GAAA,CAAI,CAAC,aAAA,CAAc,CAAA,EAAG,MAAA;AACtB,EAAA,MAAM,IAAI,kCAAA;AAAA,IACR,CAAA,gFAAA,EAAiC,IAAI,CAAA,EAAA;AACvC,EAAA;AACF;AFZ4F;AACA;ACDxD;AAC5B,EAAA;AACI,EAAA;AACL,EAAA;AACG,EAAA;AACA,EAAA;AACV;AAE4C;AACpC,EAAA;AACI,EAAA;AACL,EAAA;AACG,EAAA;AACA,EAAA;AACV;AAE4C;AACpC,EAAA;AACI,EAAA;AACL,EAAA;AACG,EAAA;AACA,EAAA;AACV;AAEwC;AAChC,EAAA;AACI,EAAA;AACL,EAAA;AACG,EAAA;AACA,EAAA;AACV;AAE4D;AAC1D,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACF;AAOqD;AACtB,EAAA;AACI,EAAA;AACH,IAAA;AAC9B,EAAA;AAC2D,EAAA;AAC7D;AAE8C;AACd,EAAA;AAChC;AAa6G;AAC5E,EAAA;AACL,EAAA;AACiB,IAAA;AAC5B,IAAA;AACT,IAAA;AAC4B,MAAA;AACE,MAAA;AAC1B,IAAA;AACG,MAAA;AACX,IAAA;AACa,IAAA;AACoD,MAAA;AAC/D,MAAA;AACF,IAAA;AACI,IAAA;AACyC,MAAA;AACf,MAAA;AACmC,QAAA;AACxD,MAAA;AACO,QAAA;AACV,UAAA;AACA,UAAA;AACQ,UAAA;AACE,UAAA;AAC4D,UAAA;AACvE,QAAA;AACH,MAAA;AACU,IAAA;AACE,MAAA;AACV,QAAA;AACA,QAAA;AACQ,QAAA;AACE,QAAA;AACgD,QAAA;AAC3D,MAAA;AACH,IAAA;AACF,EAAA;AACO,EAAA;AACT;AAGsE;AACH,EAAA;AACnE;AAoBgI;AAC1F,EAAA;AAEV,EAAA;AACiB,IAAA;AAEU,IAAA;AACpC,sBAAA;AACX,QAAA;AACY,QAAA;AACL,QAAA;AACA,QAAA;AACE,QAAA;AACV,MAAA;AACD,MAAA;AACF,IAAA;AAGI,IAAA;AACoB,MAAA;AAChB,IAAA;AAER,IAAA;AAEgD,IAAA;AAClD,EAAA;AACF;AAGuF;AACV,EAAA;AAC7E;AAGsG;AACpE,EAAA;AACI,EAAA;AACK,EAAA;AACU,EAAA;AACe,oBAAA;AAChE,IAAA;AACF,EAAA;AACI,EAAA;AACoB,IAAA;AAChB,EAAA;AAER,EAAA;AACgD,EAAA;AAClD;AAE4F;AACtF,EAAA;AAC4B,IAAA;AACU,IAAA;AAClC,EAAA;AACC,IAAA;AACT,EAAA;AACI,EAAA;AACyC,IAAA;AACzB,IAAA;AACZ,EAAA;AACC,IAAA;AACT,EAAA;AACF;AAMiB;AAEf,EAAA;AACc,IAAA;AACZ,IAAA;AACF,EAAA;AAG6B,EAAA;AACsB,EAAA;AAEhB,EAAA;AACxB,IAAA;AAAA;AAEO,MAAA;AAChB,IAAA;AACD,EAAA;AAC2B,EAAA;AAChB,IAAA;AAC0C,MAAA;AACpD,IAAA;AACF,EAAA;AAEmD,EAAA;AACQ,EAAA;AAC1C,EAAA;AAEoB,EAAA;AACjC,EAAA;AACyF,IAAA;AACzC,IAAA;AAC5B,MAAA;AACP,sBAAA;AACX,QAAA;AACA,QAAA;AACA,QAAA;AACO,QAAA;AACR,MAAA;AACF,IAAA;AACwB,IAAA;AACf,EAAA;AACN,IAAA;AACmB,MAAA;AACf,IAAA;AAER,IAAA;AAC+D,IAAA;AACjE,EAAA;AAEa,kBAAA;AACX,IAAA;AACA,IAAA;AACA,IAAA;AACO,IAAA;AACR,EAAA;AAGG,EAAA;AACA,EAAA;AACkC,IAAA;AAC1B,EAAA;AACN,IAAA;AACmB,MAAA;AACf,IAAA;AAER,IAAA;AACiE,IAAA;AACnE,EAAA;AAE4B,EAAA;AACtB,IAAA;AACmB,MAAA;AACf,IAAA;AAER,IAAA;AACU,IAAA;AAC4D,MAAA;AACtE,IAAA;AACF,EAAA;AAEgC,EAAA;AACnB,kBAAA;AACX,IAAA;AACA,IAAA;AACA,IAAA;AACO,IAAA;AACR,EAAA;AACH;AAEwD;AACzB,EAAA;AACI,EAAA;AACY,EAAA;AACZ,IAAA;AACf,MAAA;AAEhB,IAAA;AACD,EAAA;AACoB,EAAA;AACvB;AD1E4F;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA","file":"/Users/mong-e/workspace/kordoc/dist/chunk-2FNQN23O.cjs","sourcesContent":[null,"/**\n * 수식 OCR 모델 자동 다운로드 + SHA-256 검증\n *\n * 캐시 위치: `~/.cache/kordoc/models/pix2text/`\n * 총 ~155MB (MFD 44MB + MFR encoder 87MB + MFR decoder 30MB + tokenizer 40KB)\n *\n * HF URL + SHA 는 Docufinder archive/rust-formula-ocr-wip 에서 검증 완료됨.\n * 모델 버전 갱신 시 URL + SHA 를 함께 갱신해야 한다.\n */\n\nimport { createHash } from \"crypto\"\nimport { createReadStream } from \"fs\"\nimport { mkdir, stat, unlink, rename } from \"fs/promises\"\nimport { createWriteStream } from \"fs\"\nimport { homedir } from \"os\"\nimport { join, dirname } from \"path\"\nimport { pipeline } from \"stream/promises\"\nimport { Readable } from \"stream\"\nimport { assertNetworkAllowed } from \"../../shared/offline.js\"\n\nexport interface ModelSpec {\n name: string\n filename: string\n url: string\n sha256: string\n sizeMb: number\n}\n\n/** 모델 스펙 (SHA-256 은 실제 다운로드로 검증됨 — 변경 금지) */\nexport const MFD_MODEL: ModelSpec = {\n name: \"Pix2Text MFD\",\n filename: \"mfd.onnx\",\n url: \"https://huggingface.co/breezedeus/pix2text-mfd/resolve/main/mfd-v20240618.onnx\",\n sha256: \"51a8854743b17ae654729af8db82a630c1ccfa06debf4856c8b28055f87d02c1\",\n sizeMb: 42,\n}\n\nexport const MFR_ENCODER_MODEL: ModelSpec = {\n name: \"Pix2Text MFR encoder\",\n filename: \"encoder_model.onnx\",\n url: \"https://huggingface.co/breezedeus/pix2text-mfr/resolve/main/encoder_model.onnx\",\n sha256: \"bd8d5c322792e9ec45793af5569e9748f82a3d728a9e00213dbfc56c1486f37d\",\n sizeMb: 87,\n}\n\nexport const MFR_DECODER_MODEL: ModelSpec = {\n name: \"Pix2Text MFR decoder\",\n filename: \"decoder_model.onnx\",\n url: \"https://huggingface.co/breezedeus/pix2text-mfr/resolve/main/decoder_model.onnx\",\n sha256: \"fd0f92d7a012f3dae41e1ac79421aea0ea888b5a66cb3f9a004e424f82f3daed\",\n sizeMb: 30,\n}\n\nexport const MFR_TOKENIZER: ModelSpec = {\n name: \"Pix2Text MFR tokenizer\",\n filename: \"tokenizer.json\",\n url: \"https://huggingface.co/breezedeus/pix2text-mfr/resolve/main/tokenizer.json\",\n sha256: \"3e2ab757277d22639bec28c9d7972e352d3d1dba223051fa674002dc5ab64df3\",\n sizeMb: 1,\n}\n\nexport const ALL_FORMULA_MODELS: ReadonlyArray<ModelSpec> = [\n MFD_MODEL,\n MFR_ENCODER_MODEL,\n MFR_DECODER_MODEL,\n MFR_TOKENIZER,\n]\n\n/**\n * 모델 캐시 하위 디렉토리 경로. 환경 변수 `KORDOC_MODEL_CACHE` 가 설정되면\n * 해당 경로를 우선 사용, 없으면 `~/.cache/kordoc/models/<subdir>/`.\n * (수식 OCR 외 텍스트 OCR 등 다른 모델군도 공용 — subdir 로 격리)\n */\nexport function getModelsDir(subdir: string): string {\n const override = process.env.KORDOC_MODEL_CACHE\n if (override && override.trim()) {\n return join(override, subdir)\n }\n return join(homedir(), \".cache\", \"kordoc\", \"models\", subdir)\n}\n\nexport function getFormulaModelsDir(): string {\n return getModelsDir(\"pix2text\")\n}\n\nexport interface ModelStatus {\n spec: ModelSpec\n localPath: string\n exists: boolean\n /** SHA 검증까지 끝났는가 (exists && valid) */\n verified: boolean\n /** 검증 실패 시 이유 */\n invalidReason?: string\n}\n\n/** 지정 모델군의 존재/유효성 상태 반환 (다운로드 없이 확인만) */\nexport async function getModelStatusIn(dir: string, specs: ReadonlyArray<ModelSpec>): Promise<ModelStatus[]> {\n const result: ModelStatus[] = []\n for (const spec of specs) {\n const localPath = join(dir, spec.filename)\n let exists = false\n try {\n const s = await stat(localPath)\n exists = s.isFile() && s.size > 0\n } catch {\n exists = false\n }\n if (!exists) {\n result.push({ spec, localPath, exists: false, verified: false })\n continue\n }\n try {\n const actual = await sha256OfFile(localPath)\n if (actual === spec.sha256) {\n result.push({ spec, localPath, exists: true, verified: true })\n } else {\n result.push({\n spec,\n localPath,\n exists: true,\n verified: false,\n invalidReason: `SHA256 mismatch: expected ${spec.sha256}, got ${actual}`,\n })\n }\n } catch (e) {\n result.push({\n spec,\n localPath,\n exists: true,\n verified: false,\n invalidReason: `SHA compute failed: ${(e as Error).message}`,\n })\n }\n }\n return result\n}\n\n/** 모든 수식 모델의 존재/유효성 상태 반환 (다운로드 없이 확인만) */\nexport async function getFormulaModelStatus(): Promise<ModelStatus[]> {\n return getModelStatusIn(getFormulaModelsDir(), ALL_FORMULA_MODELS)\n}\n\nexport interface DownloadProgress {\n spec: ModelSpec\n /** 현재까지 다운로드한 바이트 */\n downloaded: number\n /** 전체 바이트 (알 수 있으면) */\n total: number | null\n /** \"download\" | \"verify\" | \"done\" | \"skip\" */\n phase: \"download\" | \"verify\" | \"done\" | \"skip\" | \"error\"\n message?: string\n}\n\nexport type ProgressHandler = (p: DownloadProgress) => void\n\n/**\n * 지정 모델군을 다운로드/검증한다.\n * - 이미 있고 SHA 일치 → skip\n * - 없음 → 다운로드 후 SHA 검증. 실패 시 파일 삭제.\n */\nexport async function ensureModelsIn(dir: string, specs: ReadonlyArray<ModelSpec>, onProgress?: ProgressHandler): Promise<void> {\n await mkdir(dir, { recursive: true })\n\n for (const spec of specs) {\n const localPath = join(dir, spec.filename)\n\n if (await isExistingValid(localPath, spec.sha256)) {\n onProgress?.({\n spec,\n downloaded: 0,\n total: null,\n phase: \"skip\",\n message: \"이미 존재 + SHA 일치\",\n })\n continue\n }\n\n // 기존 파일 있지만 SHA 불일치 → 삭제\n try {\n await unlink(localPath)\n } catch {\n // 없을 수 있음\n }\n\n await downloadToFile(spec, localPath, onProgress)\n }\n}\n\n/** 필요한 모든 수식 모델을 다운로드/검증한다. */\nexport async function ensureFormulaModels(onProgress?: ProgressHandler): Promise<void> {\n return ensureModelsIn(getFormulaModelsDir(), ALL_FORMULA_MODELS, onProgress)\n}\n\n/** 단일 모델만 확인/다운로드 (진단 UI 에서 개별 상태 새로고침용) */\nexport async function ensureSingleModel(spec: ModelSpec, onProgress?: ProgressHandler): Promise<void> {\n const dir = getFormulaModelsDir()\n await mkdir(dir, { recursive: true })\n const localPath = join(dir, spec.filename)\n if (await isExistingValid(localPath, spec.sha256)) {\n onProgress?.({ spec, downloaded: 0, total: null, phase: \"skip\" })\n return\n }\n try {\n await unlink(localPath)\n } catch {\n // 재다운로드 전 손상/잔존 파일 정리 best-effort — 애초에 없으면 실패해도 무방\n }\n await downloadToFile(spec, localPath, onProgress)\n}\n\nasync function isExistingValid(localPath: string, sha256Expected: string): Promise<boolean> {\n try {\n const s = await stat(localPath)\n if (!s.isFile() || s.size === 0) return false\n } catch {\n return false\n }\n try {\n const actual = await sha256OfFile(localPath)\n return actual === sha256Expected\n } catch {\n return false\n }\n}\n\nasync function downloadToFile(\n spec: ModelSpec,\n localPath: string,\n onProgress?: ProgressHandler,\n): Promise<void> {\n // 폐쇄망 모드에서는 요청 자체를 보내지 않는다 (사이드로드 안내로 대체)\n assertNetworkAllowed(\n `${spec.name} 모델 다운로드`,\n \"온라인 PC에서 `kordoc models --export <디렉토리>` 로 내보낸 뒤 이 PC에서 `kordoc models --import <디렉토리>` 하거나, KORDOC_MODEL_CACHE 로 모델 캐시 경로를 지정하세요\",\n )\n\n // 먼저 .part 로 받고 검증 후 rename — 중단된 다운로드가 \"정상 파일\"로 오인되는 걸 방지\n const partPath = `${localPath}.part`\n await mkdir(dirname(localPath), { recursive: true })\n\n const resp = await fetch(spec.url, {\n headers: {\n // HF CDN 은 UA 없으면 가끔 403 을 뱉는다\n \"User-Agent\": \"kordoc-formula-ocr/1.0 (+https://github.com/chrisryugj/kordoc)\",\n },\n })\n if (!resp.ok || !resp.body) {\n throw new Error(\n `${spec.name} 다운로드 실패: HTTP ${resp.status} ${resp.statusText} (${spec.url})`,\n )\n }\n\n const lenHeader = resp.headers.get(\"content-length\")\n const total = lenHeader ? Number.parseInt(lenHeader, 10) : null\n let downloaded = 0\n\n const ws = createWriteStream(partPath)\n try {\n const reader = Readable.fromWeb(resp.body as unknown as import(\"stream/web\").ReadableStream)\n reader.on(\"data\", (chunk: Buffer | Uint8Array) => {\n downloaded += chunk.length\n onProgress?.({\n spec,\n downloaded,\n total,\n phase: \"download\",\n })\n })\n await pipeline(reader, ws)\n } catch (e) {\n try {\n await unlink(partPath)\n } catch {\n // .part 임시파일 정리 best-effort — 본 에러(스트리밍 실패)를 가리지 않도록 무시\n }\n throw new Error(`${spec.name} 스트리밍 실패: ${(e as Error).message}`)\n }\n\n onProgress?.({\n spec,\n downloaded,\n total,\n phase: \"verify\",\n })\n\n // SHA 검증\n let actual: string\n try {\n actual = await sha256OfFile(partPath)\n } catch (e) {\n try {\n await unlink(partPath)\n } catch {\n // .part 임시파일 정리 best-effort — 본 에러(SHA 계산 실패)를 가리지 않도록 무시\n }\n throw new Error(`${spec.name} SHA 계산 실패: ${(e as Error).message}`)\n }\n\n if (actual !== spec.sha256) {\n try {\n await unlink(partPath)\n } catch {\n // 오염된 .part 정리 best-effort — 본 에러(SHA mismatch)를 가리지 않도록 무시\n }\n throw new Error(\n `${spec.name} SHA256 mismatch: expected ${spec.sha256}, got ${actual} — 모델 URL 이 오염되었거나 전송 중 손상되었습니다.`,\n )\n }\n\n await rename(partPath, localPath)\n onProgress?.({\n spec,\n downloaded,\n total,\n phase: \"done\",\n })\n}\n\nasync function sha256OfFile(p: string): Promise<string> {\n const h = createHash(\"sha256\")\n const stream = createReadStream(p)\n await pipeline(stream, async function* (src) {\n for await (const chunk of src) {\n h.update(chunk)\n // yield 안 함 (소비만)\n }\n })\n return h.digest(\"hex\")\n}\n","/**\n * 폐쇄망(내부망) 배포용 아웃바운드 통신 킬스위치 + 파일 접근 루트 제한.\n *\n * kordoc 자체는 문서 파서라 상시 외부 통신이 없지만, 두 경로만 나간다:\n * 1. OCR/수식 모델 최초 1회 다운로드 (huggingface.co)\n * 2. `kordoc watch --webhook` 사용자가 명시한 URL\n *\n * `KORDOC_OFFLINE=1` 이면 위 경로를 **시도 전에** 차단한다. 보안성 검토에서\n * \"외부 통신 없음\"을 런타임으로 증명하기 위한 단일 관문 — 새 아웃바운드 경로를\n * 추가하면 반드시 `assertNetworkAllowed()` 를 먼저 통과시킬 것.\n *\n * `KORDOC_ROOT` 는 파일 접근을 특정 디렉토리 하위로 제한한다(미설정 시 제한 없음 —\n * 기존 동작 보존). 내부망 배포에서는 두 변수를 함께 고정하는 것을 권장.\n */\n\nimport { realpathSync } from \"fs\"\nimport { isAbsolute, relative, resolve, sep } from \"path\"\nimport { KordocError } from \"../utils.js\"\n\nconst TRUTHY = new Set([\"1\", \"true\", \"yes\", \"on\"])\n\nfunction envFlag(name: string): boolean {\n const v = process.env[name]\n return !!v && TRUTHY.has(v.trim().toLowerCase())\n}\n\n/** 폐쇄망 모드 여부 (`KORDOC_OFFLINE=1|true|yes|on`) */\nexport function isOfflineMode(): boolean {\n return envFlag(\"KORDOC_OFFLINE\")\n}\n\n/**\n * 아웃바운드 통신 관문. 폐쇄망 모드면 요청을 보내기 전에 예외로 끊는다.\n * @param what 차단 대상 설명 (사용자에게 노출됨 — 호스트/URL 은 넣지 말 것)\n * @param hint 대안 안내 (오프라인 사이드로드 방법 등)\n */\nexport function assertNetworkAllowed(what: string, hint?: string): void {\n if (!isOfflineMode()) return\n throw new KordocError(\n `폐쇄망 모드(KORDOC_OFFLINE)에서 차단됨: ${what}` + (hint ? ` — ${hint}` : \"\"),\n )\n}\n\n/** 파일 접근 루트 (`KORDOC_ROOT`). 미설정이면 null = 제한 없음 */\nexport function getAccessRoot(): string | null {\n const raw = process.env.KORDOC_ROOT\n if (!raw || !raw.trim()) return null\n const resolved = resolve(raw.trim())\n try {\n return realpathSync(resolved)\n } catch {\n // 루트가 실재하지 않으면 모든 접근이 막히는 게 안전한 기본값\n return resolved\n }\n}\n\n/** 경로가 루트 하위인지 (루트 자신 포함). 경계 문자열 매칭이 아닌 세그먼트 단위 비교 */\nexport function isWithinRoot(realPath: string, root: string): boolean {\n if (realPath === root) return true\n const rel = relative(root, realPath)\n return rel !== \"\" && !rel.startsWith(\"..\") && !isAbsolute(rel) && !rel.startsWith(`..${sep}`)\n}\n\n/**\n * `KORDOC_ROOT` 가 설정된 경우 경로가 루트 하위인지 검증한다.\n * 인자는 이미 realpath 로 해석된 절대경로여야 한다 (심볼릭 링크 탈출 방지).\n */\nexport function assertWithinRoot(realPath: string): void {\n const root = getAccessRoot()\n if (!root) return\n if (!isWithinRoot(realPath, root)) {\n throw new KordocError(`허용된 작업 디렉토리(KORDOC_ROOT) 밖의 경로입니다: ${realPath}`)\n }\n}\n"]}
|
|
@@ -3,11 +3,11 @@ import {
|
|
|
3
3
|
HEADING_RATIO_H1,
|
|
4
4
|
HEADING_RATIO_H2,
|
|
5
5
|
HEADING_RATIO_H3
|
|
6
|
-
} from "./chunk-
|
|
6
|
+
} from "./chunk-N3DLQ2YH.js";
|
|
7
7
|
import {
|
|
8
8
|
safeMax,
|
|
9
9
|
safeMin
|
|
10
|
-
} from "./chunk-
|
|
10
|
+
} from "./chunk-NMXDXHHW.js";
|
|
11
11
|
|
|
12
12
|
// src/pdf/image-regions.ts
|
|
13
13
|
import { OPS } from "pdfjs-dist/legacy/build/pdf.mjs";
|
|
@@ -3687,4 +3687,4 @@ export {
|
|
|
3687
3687
|
extractPageBlocksWithLines,
|
|
3688
3688
|
mergeCrossPageTables
|
|
3689
3689
|
};
|
|
3690
|
-
//# sourceMappingURL=chunk-
|
|
3690
|
+
//# sourceMappingURL=chunk-34IBS3JN.js.map
|
|
@@ -6,7 +6,7 @@ import {
|
|
|
6
6
|
HEADING_RATIO_H1,
|
|
7
7
|
HEADING_RATIO_H2,
|
|
8
8
|
HEADING_RATIO_H3
|
|
9
|
-
} from "./chunk-
|
|
9
|
+
} from "./chunk-N3DLQ2YH.js";
|
|
10
10
|
import {
|
|
11
11
|
parsePageRange
|
|
12
12
|
} from "./chunk-MOL7MDBG.js";
|
|
@@ -22,7 +22,7 @@ import {
|
|
|
22
22
|
paraTextPureT,
|
|
23
23
|
patchZipEntries,
|
|
24
24
|
scanSectionXml
|
|
25
|
-
} from "./chunk-
|
|
25
|
+
} from "./chunk-EWUFLCRP.js";
|
|
26
26
|
import {
|
|
27
27
|
MAX_COLS,
|
|
28
28
|
MAX_ROWS,
|
|
@@ -32,14 +32,14 @@ import {
|
|
|
32
32
|
dedupeRunningHeaders,
|
|
33
33
|
flattenLayoutTables,
|
|
34
34
|
mapPuaText
|
|
35
|
-
} from "./chunk-
|
|
35
|
+
} from "./chunk-B4MQXOU5.js";
|
|
36
36
|
import {
|
|
37
37
|
SPACE_EM_FIXED,
|
|
38
38
|
charWidthEm1000,
|
|
39
39
|
fitRatioForFewerLines,
|
|
40
40
|
measureTextWidth,
|
|
41
41
|
simulateWrap
|
|
42
|
-
} from "./chunk-
|
|
42
|
+
} from "./chunk-QYXQJPB7.js";
|
|
43
43
|
import {
|
|
44
44
|
MAX_DECOMPRESS_SIZE,
|
|
45
45
|
MAX_XML_DEPTH,
|
|
@@ -55,7 +55,7 @@ import {
|
|
|
55
55
|
findChildByLocalName,
|
|
56
56
|
localName,
|
|
57
57
|
rawTextContent
|
|
58
|
-
} from "./chunk-
|
|
58
|
+
} from "./chunk-JAY6BXZ6.js";
|
|
59
59
|
import {
|
|
60
60
|
KordocError,
|
|
61
61
|
classifyError,
|
|
@@ -67,7 +67,7 @@ import {
|
|
|
67
67
|
sanitizeHref,
|
|
68
68
|
stripDtd,
|
|
69
69
|
toArrayBuffer
|
|
70
|
-
} from "./chunk-
|
|
70
|
+
} from "./chunk-NMXDXHHW.js";
|
|
71
71
|
|
|
72
72
|
// src/index.ts
|
|
73
73
|
import { readFile } from "fs/promises";
|
|
@@ -2859,7 +2859,7 @@ function resolveImageBlocks(binDataMap, blocks, warnings, sweepUnreferenced) {
|
|
|
2859
2859
|
const ext = mime.includes("jpeg") ? "jpg" : mime.includes("png") ? "png" : mime.includes("gif") ? "gif" : mime.includes("bmp") ? "bmp" : "bin";
|
|
2860
2860
|
img = { filename: `image_${String(imageIndex).padStart(3, "0")}.${ext}`, data: new Uint8Array(bin.data), mime };
|
|
2861
2861
|
resolved.set(storageId, img);
|
|
2862
|
-
images.push({ filename: img.filename, data: img.data, mimeType: img.mime });
|
|
2862
|
+
images.push({ filename: img.filename, data: img.data, mimeType: img.mime, source: bin.name });
|
|
2863
2863
|
renamed.set(storageId, img.filename);
|
|
2864
2864
|
}
|
|
2865
2865
|
}
|
|
@@ -2883,7 +2883,7 @@ function resolveImageBlocks(binDataMap, blocks, warnings, sweepUnreferenced) {
|
|
|
2883
2883
|
const ext = mime.includes("jpeg") ? "jpg" : mime.includes("png") ? "png" : mime.includes("gif") ? "gif" : mime.includes("bmp") ? "bmp" : "bin";
|
|
2884
2884
|
const filename = `image_${String(imageIndex).padStart(3, "0")}.${ext}`;
|
|
2885
2885
|
const data = new Uint8Array(bin.data);
|
|
2886
|
-
images.push({ filename, data, mimeType: mime });
|
|
2886
|
+
images.push({ filename, data, mimeType: mime, source: bin.name });
|
|
2887
2887
|
blocks.push({ type: "image", text: filename, imageData: { data, mimeType: mime, filename: bin.name } });
|
|
2888
2888
|
}
|
|
2889
2889
|
}
|
|
@@ -3020,7 +3020,7 @@ async function extractImagesFromZip(zip, blocks, decompressed, warnings, sweepUn
|
|
|
3020
3020
|
const mimeType = imageExtToMime(ext);
|
|
3021
3021
|
imageIndex++;
|
|
3022
3022
|
const filename2 = `image_${String(imageIndex).padStart(3, "0")}.${mimeToExt(mimeType)}`;
|
|
3023
|
-
img = { filename: filename2, data, mimeType };
|
|
3023
|
+
img = { filename: filename2, data, mimeType, source: path };
|
|
3024
3024
|
images.push(img);
|
|
3025
3025
|
usedPaths.add(path);
|
|
3026
3026
|
break;
|
|
@@ -3059,7 +3059,7 @@ async function extractImagesFromZip(zip, blocks, decompressed, warnings, sweepUn
|
|
|
3059
3059
|
}
|
|
3060
3060
|
imageIndex++;
|
|
3061
3061
|
const filename = `image_${String(imageIndex).padStart(3, "0")}.${mimeToExt(mimeType)}`;
|
|
3062
|
-
const img = { filename, data, mimeType };
|
|
3062
|
+
const img = { filename, data, mimeType, source: file.name };
|
|
3063
3063
|
images.push(img);
|
|
3064
3064
|
blocks.push({ type: "image", text: filename, imageData: { data, mimeType, filename: file.name } });
|
|
3065
3065
|
} catch (err) {
|
|
@@ -19749,10 +19749,14 @@ async function buildImageMap(zip, rels, doc, warnings) {
|
|
|
19749
19749
|
gif: "image/gif",
|
|
19750
19750
|
bmp: "image/bmp",
|
|
19751
19751
|
wmf: "image/wmf",
|
|
19752
|
-
emf: "image/emf"
|
|
19752
|
+
emf: "image/emf",
|
|
19753
|
+
tif: "image/tiff",
|
|
19754
|
+
tiff: "image/tiff",
|
|
19755
|
+
svg: "image/svg+xml"
|
|
19753
19756
|
};
|
|
19754
19757
|
const filename = `image_${String(imgIdx).padStart(3, "0")}.${ext}`;
|
|
19755
|
-
|
|
19758
|
+
const mimeType = mimeMap[ext] ?? detectImageMime(data) ?? "application/octet-stream";
|
|
19759
|
+
images.push({ filename, data, mimeType, source: imgPath });
|
|
19756
19760
|
imageMap.set(embedId, filename);
|
|
19757
19761
|
} catch (err) {
|
|
19758
19762
|
warnings.push({
|
|
@@ -21934,7 +21938,9 @@ var GJ_CHAR_BAR = 24;
|
|
|
21934
21938
|
var GJ_CHAR_BODY_TITLE = 25;
|
|
21935
21939
|
var GJ_CHAR_TITLE_BAR = 26;
|
|
21936
21940
|
var GJ_CHAR_APPROVAL = 27;
|
|
21937
|
-
var
|
|
21941
|
+
var GJ_CHAR_SRC = 28;
|
|
21942
|
+
var GJ_CHAR_SRC_BOLD = 29;
|
|
21943
|
+
var GJ_CHAR_COUNT = GJ_CHAR_SRC_BOLD - GJ_CHAR_DAE + 1;
|
|
21938
21944
|
var GJ_PARA_CHAM = 20;
|
|
21939
21945
|
var GJ_PARA_COVER = 21;
|
|
21940
21946
|
var GJ_PARA_TOC_ITEM = 22;
|
|
@@ -21942,7 +21948,8 @@ var GJ_PARA_CHAPTER = 23;
|
|
|
21942
21948
|
var GJ_PARA_BAR = 24;
|
|
21943
21949
|
var GONGMUN_LIST_PLAIN_BASE = 25;
|
|
21944
21950
|
var GONGMUN_PARA_APPROVAL = GONGMUN_LIST_PLAIN_BASE + GONGMUN_LIST_LEVELS;
|
|
21945
|
-
var
|
|
21951
|
+
var GONGMUN_SRC_LEFT = GONGMUN_PARA_APPROVAL + 1;
|
|
21952
|
+
var GONGMUN_LIST_VARIANT_BASE = GONGMUN_SRC_LEFT + 1;
|
|
21946
21953
|
var GJ_BF_CHAPTER_NUM = 3;
|
|
21947
21954
|
var GJ_BF_CHAPTER_GAP = 4;
|
|
21948
21955
|
var GJ_BF_CHAPTER_TITLE = 5;
|
|
@@ -21986,10 +21993,15 @@ var GONGMUN_TBL_CHAR_BOLD = 12;
|
|
|
21986
21993
|
var GONGMUN_TBL_PT = 1200;
|
|
21987
21994
|
var GONGMUN_TITLE_BAR_CHAR = 13;
|
|
21988
21995
|
var GONGMUN_APPROVAL_CHAR = 14;
|
|
21996
|
+
var GONGMUN_SRC_CHAR = 15;
|
|
21997
|
+
var GONGMUN_SRC_CHAR_BOLD = 16;
|
|
21998
|
+
function srcCaptionPt(bodyHeight) {
|
|
21999
|
+
return Math.max(bodyHeight - 300, 1e3);
|
|
22000
|
+
}
|
|
21989
22001
|
var CHAR_VARIANT_BASE = CHAR_QUOTE + 1;
|
|
21990
22002
|
function charVariantBase(richAssets, isGongmun = true) {
|
|
21991
22003
|
if (!isGongmun) return CHAR_VARIANT_BASE;
|
|
21992
|
-
return richAssets ? CHAR_VARIANT_BASE + GJ_CHAR_COUNT : CHAR_VARIANT_BASE + (
|
|
22004
|
+
return richAssets ? CHAR_VARIANT_BASE + GJ_CHAR_COUNT : CHAR_VARIANT_BASE + (GONGMUN_SRC_CHAR_BOLD - GONGMUN_TBL_CHAR + 1);
|
|
21993
22005
|
}
|
|
21994
22006
|
var GONGMUN_BODY_RATIO = 95;
|
|
21995
22007
|
|
|
@@ -22491,7 +22503,10 @@ function buildCharProperties(theme, gongmun, ratioVariants = [], extraCharPrXmls
|
|
|
22491
22503
|
charPr(11, GONGMUN_TBL_PT, false, false, bodyFont, theme.body),
|
|
22492
22504
|
charPr(12, GONGMUN_TBL_PT, true, false, bodyFont, theme.body),
|
|
22493
22505
|
charPr(GONGMUN_TITLE_BAR_CHAR, 100, false, false, bodyFont, theme.body),
|
|
22494
|
-
charPr(GONGMUN_APPROVAL_CHAR, 1e3, false, false, bodyFont, theme.body)
|
|
22506
|
+
charPr(GONGMUN_APPROVAL_CHAR, 1e3, false, false, bodyFont, theme.body),
|
|
22507
|
+
// 출처행 캡션 — 본문 폰트 -3pt (실무 보고서: 표 아래 출처는 본문과 구분되는 작은 글씨)
|
|
22508
|
+
charPr(GONGMUN_SRC_CHAR, srcCaptionPt(body), false, false, bodyFont, theme.body),
|
|
22509
|
+
charPr(GONGMUN_SRC_CHAR_BOLD, srcCaptionPt(body), true, false, bodyFont, theme.body)
|
|
22495
22510
|
);
|
|
22496
22511
|
}
|
|
22497
22512
|
if (richAssets) {
|
|
@@ -22529,8 +22544,12 @@ function buildCharProperties(theme, gongmun, ratioVariants = [], extraCharPrXmls
|
|
|
22529
22544
|
// 본문 제목박스(HY헤드라인M 22pt, 실측 GT3 표④)
|
|
22530
22545
|
charPr(GJ_CHAR_TITLE_BAR, 100, false, false, 4),
|
|
22531
22546
|
// 1페이지형 제목박스 바(1pt)
|
|
22532
|
-
charPr(GJ_CHAR_APPROVAL, 1e3, false, false, 7)
|
|
22547
|
+
charPr(GJ_CHAR_APPROVAL, 1e3, false, false, 7),
|
|
22533
22548
|
// 결재란 라벨(맑은 고딕 10pt)
|
|
22549
|
+
charPr(GJ_CHAR_SRC, srcCaptionPt(body), false, false, 5),
|
|
22550
|
+
// 출처행 캡션(한양중고딕 본문-3pt)
|
|
22551
|
+
charPr(GJ_CHAR_SRC_BOLD, srcCaptionPt(body), true, false, 5)
|
|
22552
|
+
// 출처행 캡션 bold
|
|
22534
22553
|
);
|
|
22535
22554
|
}
|
|
22536
22555
|
const expectedBase = charVariantBase(richAssets, !!gongmun);
|
|
@@ -22618,6 +22637,7 @@ ${base2.join("\n")}
|
|
|
22618
22637
|
base.push(paraPr(GONGMUN_LIST_PLAIN_BASE + d, { align: listAlign, lineSpacing: ls, left, indent: 0, spaceBefore: sectionGap, keepWord: true }));
|
|
22619
22638
|
}
|
|
22620
22639
|
base.push(paraPr(GONGMUN_PARA_APPROVAL, { align: "CENTER", lineSpacing: 100, keepWord: true }));
|
|
22640
|
+
base.push(paraPr(GONGMUN_SRC_LEFT, { align: "LEFT", lineSpacing: ls, keepWord: true }));
|
|
22621
22641
|
for (let vi = 0; vi < listIndentVariants.length; vi++) {
|
|
22622
22642
|
const { depth, widthHu } = listIndentVariants[vi];
|
|
22623
22643
|
const { left } = levelIndent(depth, gongmun.bodyHeight, gongmun.numbering, gongmun.sizes, gongmun.bullet2, usesAsteriskThird(gongmun.preset));
|
|
@@ -25193,7 +25213,17 @@ function renderParagraph(block, blockIdx, ctx) {
|
|
|
25193
25213
|
const ctr = gongmun && /^<center>([\s\S]*)<\/center>$/i.exec((block.text || "").trim());
|
|
25194
25214
|
const rgt = gongmun && /^<right>([\s\S]*)<\/right>$/i.exec((block.text || "").trim());
|
|
25195
25215
|
if (ctr) return generateParagraph(ctr[1].trim(), GONGMUN_CENTER);
|
|
25196
|
-
if (rgt)
|
|
25216
|
+
if (rgt) {
|
|
25217
|
+
const t = rgt[1].trim();
|
|
25218
|
+
const isSrc = /^(출처|자료)\s*[::]/.test(t);
|
|
25219
|
+
const srcChar = ctx.richAssets ? GJ_CHAR_SRC : GONGMUN_SRC_CHAR;
|
|
25220
|
+
const srcBold = ctx.richAssets ? GJ_CHAR_SRC_BOLD : GONGMUN_SRC_CHAR_BOLD;
|
|
25221
|
+
const h = isSrc ? srcCaptionPt(gongmun.bodyHeight) : gongmun.bodyHeight;
|
|
25222
|
+
const fits = measureTextWidth(t, h, isSrc ? 100 : GONGMUN_BODY_RATIO) <= ctx.gjBodyW + 0.5;
|
|
25223
|
+
const paraId = fits ? GONGMUN_RIGHT : GONGMUN_SRC_LEFT;
|
|
25224
|
+
if (!isSrc) return generateParagraph(t, paraId);
|
|
25225
|
+
return generateParagraph(t, paraId, srcChar, (id) => id === CHAR_BOLD ? srcBold : id);
|
|
25226
|
+
}
|
|
25197
25227
|
return generateParagraph(block.text || "", PARA_NORMAL, CHAR_NORMAL, fit ? variantMapper(fit, blockIdx, vBase) : void 0);
|
|
25198
25228
|
}
|
|
25199
25229
|
function renderCodeBlock(block, blockIdx, ctx) {
|
|
@@ -27986,7 +28016,7 @@ async function patchHwp(original, editedMarkdown, options) {
|
|
|
27986
28016
|
for (let i = 0; i < scans.length; i++) {
|
|
27987
28017
|
if (scans[i].repl.size === 0 && scans[i].inserts.size === 0) continue;
|
|
27988
28018
|
const newStream = serializeRecords(scans[i].records, scans[i].repl, scans[i].inserts);
|
|
27989
|
-
const content = compressed ?
|
|
28019
|
+
const content = compressed ? compressWithTail(newStream) : newStream;
|
|
27990
28020
|
out = replaceOleStream(out, sectionPaths[i], content);
|
|
27991
28021
|
}
|
|
27992
28022
|
data = new Uint8Array(out);
|
|
@@ -28469,6 +28499,26 @@ function rebuildCharShape(csData, coreStartUnit) {
|
|
|
28469
28499
|
});
|
|
28470
28500
|
return { buf, count: kept.length };
|
|
28471
28501
|
}
|
|
28502
|
+
var CRC_TABLE = (() => {
|
|
28503
|
+
const t = new Uint32Array(256);
|
|
28504
|
+
for (let n = 0; n < 256; n++) {
|
|
28505
|
+
let c = n;
|
|
28506
|
+
for (let k = 0; k < 8; k++) c = c & 1 ? 3988292384 ^ c >>> 1 : c >>> 1;
|
|
28507
|
+
t[n] = c >>> 0;
|
|
28508
|
+
}
|
|
28509
|
+
return t;
|
|
28510
|
+
})();
|
|
28511
|
+
function crc32(buf) {
|
|
28512
|
+
let c = 4294967295;
|
|
28513
|
+
for (let i = 0; i < buf.length; i++) c = CRC_TABLE[(c ^ buf[i]) & 255] ^ c >>> 8;
|
|
28514
|
+
return (c ^ 4294967295) >>> 0;
|
|
28515
|
+
}
|
|
28516
|
+
function compressWithTail(raw) {
|
|
28517
|
+
const tail = Buffer.alloc(8);
|
|
28518
|
+
tail.writeUInt32LE(crc32(raw), 0);
|
|
28519
|
+
tail.writeUInt32LE(raw.length >>> 0, 4);
|
|
28520
|
+
return Buffer.concat([deflateRawSync(raw), tail]);
|
|
28521
|
+
}
|
|
28472
28522
|
function synthesizeLineSegs(lineSegData, newRaw, startUnits) {
|
|
28473
28523
|
if (lineSegData.length < 36) return null;
|
|
28474
28524
|
const seg0 = lineSegData.subarray(0, 36);
|
|
@@ -28546,6 +28596,15 @@ function stageParaPatch(scan, para2, newPlain, skip) {
|
|
|
28546
28596
|
scan.repl.set(para2.lineSegIdx, synth.buf);
|
|
28547
28597
|
newHeader.writeUInt16LE(synth.count, 16);
|
|
28548
28598
|
}
|
|
28599
|
+
} else {
|
|
28600
|
+
const lineSegData = records[para2.lineSegIdx].data;
|
|
28601
|
+
const segCount = Math.floor(lineSegData.length / 36);
|
|
28602
|
+
let keep = 1;
|
|
28603
|
+
while (keep < segCount && lineSegData.readUInt32LE(keep * 36) < nChars) keep++;
|
|
28604
|
+
if (keep < segCount) {
|
|
28605
|
+
scan.repl.set(para2.lineSegIdx, Buffer.from(lineSegData.subarray(0, keep * 36)));
|
|
28606
|
+
newHeader.writeUInt16LE(keep, 16);
|
|
28607
|
+
}
|
|
28549
28608
|
}
|
|
28550
28609
|
scan.repl.set(para2.headerIdx, newHeader);
|
|
28551
28610
|
return 1;
|
|
@@ -29190,7 +29249,7 @@ async function dispatch(format, buffer, opts) {
|
|
|
29190
29249
|
}
|
|
29191
29250
|
async function parseImage(buffer, options) {
|
|
29192
29251
|
try {
|
|
29193
|
-
const { parseImageDocument } = await import("./image-ocr-
|
|
29252
|
+
const { parseImageDocument } = await import("./image-ocr-NOU6O7ZY.js");
|
|
29194
29253
|
const { blocks, warnings } = await parseImageDocument(buffer, options);
|
|
29195
29254
|
return {
|
|
29196
29255
|
success: true,
|
|
@@ -29249,7 +29308,7 @@ async function parseHwp(buffer, options) {
|
|
|
29249
29308
|
async function parsePdf(buffer, options) {
|
|
29250
29309
|
let parsePdfDocument;
|
|
29251
29310
|
try {
|
|
29252
|
-
const mod = await import("./parser-
|
|
29311
|
+
const mod = await import("./parser-VVGAZIAX.js");
|
|
29253
29312
|
parsePdfDocument = mod.parsePdfDocument;
|
|
29254
29313
|
} catch {
|
|
29255
29314
|
return {
|
|
@@ -29347,6 +29406,7 @@ export {
|
|
|
29347
29406
|
detectFormat,
|
|
29348
29407
|
detectOle2Format,
|
|
29349
29408
|
detectZipFormat,
|
|
29409
|
+
detectImageMime,
|
|
29350
29410
|
extractHwpxMetadataOnly,
|
|
29351
29411
|
extractHwp5MetadataOnly,
|
|
29352
29412
|
formatFillValue,
|
|
@@ -29399,4 +29459,4 @@ export {
|
|
|
29399
29459
|
parseHwpml,
|
|
29400
29460
|
fillForm
|
|
29401
29461
|
};
|
|
29402
|
-
//# sourceMappingURL=chunk-
|
|
29462
|
+
//# sourceMappingURL=chunk-4EJLRZ7A.js.map
|