kordoc 4.24.18 → 4.24.20
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 +5 -5
- package/dist/{-X3WSJ2ZN.js → -3RL7OPC7.js} +11 -11
- package/dist/{chunk-7HLAVRQ7.js → chunk-23GHEDMW.js} +2 -2
- package/dist/{chunk-5QY4UMRR.js → chunk-3EHQKGTL.js} +4 -4
- package/dist/{chunk-REOGJMES.cjs → chunk-CSWE6FQ4.cjs} +3 -3
- package/dist/{chunk-REOGJMES.cjs.map → chunk-CSWE6FQ4.cjs.map} +1 -1
- package/dist/{chunk-PUZUGFR2.cjs → chunk-DY7IZPA5.cjs} +25 -25
- package/dist/{chunk-PUZUGFR2.cjs.map → chunk-DY7IZPA5.cjs.map} +1 -1
- package/dist/{chunk-KNJ3TDMZ.js → chunk-G5WFGGUL.js} +2 -2
- package/dist/{chunk-UZYI4IDT.js → chunk-GAHHZ32A.js} +3 -3
- package/dist/{chunk-T3K74N6U.cjs → chunk-GKZPJDLN.cjs} +3 -3
- package/dist/{chunk-T3K74N6U.cjs.map → chunk-GKZPJDLN.cjs.map} +1 -1
- package/dist/{chunk-TRXBFZJO.js → chunk-HRZC7ITK.js} +2 -2
- package/dist/{chunk-JOWQTHXJ.js → chunk-IBXG4UPB.js} +4 -4
- package/dist/{chunk-X6RNJ2AX.js → chunk-IR6ZXRAK.js} +2 -2
- package/dist/{chunk-S6KLQX2K.js → chunk-JCRJGCYA.js} +4 -4
- package/dist/{chunk-3IYSX7EA.js → chunk-KQULVTQH.js} +2 -2
- package/dist/{chunk-VQED4B7M.js → chunk-NBQDGYCQ.js} +2 -2
- package/dist/{chunk-WDIHTINW.js → chunk-NWL7D4F7.js} +2 -2
- package/dist/{chunk-2WNVFE4V.cjs → chunk-OJB7QZHP.cjs} +6 -6
- package/dist/{chunk-2WNVFE4V.cjs.map → chunk-OJB7QZHP.cjs.map} +1 -1
- package/dist/{chunk-P74R5MK7.js → chunk-PKX6CKAI.js} +2 -2
- package/dist/{chunk-ZDTFPB5Q.cjs → chunk-PUHS26TX.cjs} +2 -2
- package/dist/{chunk-ZDTFPB5Q.cjs.map → chunk-PUHS26TX.cjs.map} +1 -1
- package/dist/{chunk-LPDFNN44.js → chunk-PWGE67XI.js} +2 -2
- package/dist/{chunk-K5S5SZC6.js → chunk-PZT5RCSM.js} +2 -2
- package/dist/{chunk-P65KKO2X.js → chunk-QRUTGCC2.js} +2 -2
- package/dist/{chunk-3HFH3SBY.js → chunk-S2LEIH33.js} +11 -11
- package/dist/{chunk-FSGDOD2E.js → chunk-SSTOQ3ZS.js} +2 -2
- package/dist/{chunk-4JAHUMTT.js → chunk-VDG4G44B.js} +2 -2
- package/dist/{chunk-GCAE6YQT.js → chunk-VO27MPHI.js} +2 -2
- package/dist/{chunk-Q6KTIFXA.js → chunk-VYWZASYT.js} +2 -2
- package/dist/{chunk-3642GYL7.js → chunk-XAUJF2GZ.js} +3 -3
- package/dist/{chunk-X6X45X3R.cjs → chunk-XIDZJIB4.cjs} +79 -79
- package/dist/{chunk-X6X45X3R.cjs.map → chunk-XIDZJIB4.cjs.map} +1 -1
- package/dist/{chunk-JHPLKR4V.js → chunk-YUEPDCVZ.js} +82 -25
- package/dist/chunk-YUEPDCVZ.js.map +1 -0
- package/dist/{chunk-R6VHZYGR.js → chunk-ZGXXR3DH.js} +3 -3
- package/dist/chunks-ODK2CKBM.js +10 -0
- package/dist/cli.js +30 -30
- package/dist/{formula-4RRREDF6.cjs → formula-7C6O7KHQ.cjs} +9 -9
- package/dist/{formula-4RRREDF6.cjs.map → formula-7C6O7KHQ.cjs.map} +1 -1
- package/dist/{formula-U6PMTA5U.js → formula-CHBBFZGG.js} +4 -4
- package/dist/{formula-6SGEN5JG.js → formula-GQMRKV2P.js} +3 -3
- package/dist/{image-ocr-3AJ54ZFX.cjs → image-ocr-2VND3EWX.cjs} +17 -17
- package/dist/{image-ocr-3AJ54ZFX.cjs.map → image-ocr-2VND3EWX.cjs.map} +1 -1
- package/dist/{image-ocr-EY54VBAM.js → image-ocr-6DPYV3Z2.js} +9 -9
- package/dist/{image-ocr-XVU5MXQB.js → image-ocr-SAW7A4YW.js} +6 -6
- package/dist/index.cjs +516 -460
- package/dist/index.cjs.map +1 -1
- package/dist/index.js +81 -25
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +23 -23
- package/dist/{model-bundle-BEAQMPMH.js → model-bundle-7K554JAW.js} +5 -5
- package/dist/{models-AFAT3YQI.js → models-VU65CWCI.js} +5 -5
- package/dist/{parse-worker-v2-MJOBLVHV.js → parse-worker-v2-435QCDJP.js} +12 -12
- package/dist/{parser-D7W5Q6Y3.js → parser-5LUP4EF5.js} +12 -12
- package/dist/{parser-UVFRWRGQ.js → parser-BD4VNEUA.js} +9 -9
- package/dist/{parser-NN4DPHHD.cjs → parser-YRGM2WD3.cjs} +159 -159
- package/dist/{parser-NN4DPHHD.cjs.map → parser-YRGM2WD3.cjs.map} +1 -1
- package/dist/{pdf-ocr-DKRVXK2X.js → pdf-ocr-2TNMUGTU.js} +9 -9
- package/dist/pdf-ocr-EYW5Q7DX.cjs +35 -0
- package/dist/{pdf-ocr-NXO3VLF4.cjs.map → pdf-ocr-EYW5Q7DX.cjs.map} +1 -1
- package/dist/{pdf-ocr-ONGESN2A.js → pdf-ocr-GJDRJACB.js} +6 -6
- package/dist/{profile-io-LLIOYHEN.js → profile-io-6XPZGJHS.js} +3 -3
- package/dist/{rasterize-FIX7WWYA.js → rasterize-PERRMQDZ.js} +3 -3
- package/dist/{redact-doc-VS5YBG7O.js → redact-doc-EA2QBZG6.js} +11 -11
- package/dist/{render-5UIF2HOQ.js → render-YLCMDE5W.js} +7 -7
- package/dist/seal-HZX7BXWG.js +10 -0
- package/dist/{setup-VJ2C3MHC.js → setup-IJSICTFK.js} +3 -3
- package/dist/visual-GFFZPSRH.js +21 -0
- package/dist/{watch-ZETJYMXE.js → watch-Z6PMJO7G.js} +12 -12
- package/package.json +1 -1
- package/dist/chunk-JHPLKR4V.js.map +0 -1
- package/dist/chunks-A6ALUARW.js +0 -10
- package/dist/pdf-ocr-NXO3VLF4.cjs +0 -35
- package/dist/seal-LBSLEGXI.js +0 -10
- package/dist/visual-UU4MYIFZ.js +0 -21
- /package/dist/{-X3WSJ2ZN.js.map → -3RL7OPC7.js.map} +0 -0
- /package/dist/{chunk-7HLAVRQ7.js.map → chunk-23GHEDMW.js.map} +0 -0
- /package/dist/{chunk-5QY4UMRR.js.map → chunk-3EHQKGTL.js.map} +0 -0
- /package/dist/{chunk-KNJ3TDMZ.js.map → chunk-G5WFGGUL.js.map} +0 -0
- /package/dist/{chunk-UZYI4IDT.js.map → chunk-GAHHZ32A.js.map} +0 -0
- /package/dist/{chunk-TRXBFZJO.js.map → chunk-HRZC7ITK.js.map} +0 -0
- /package/dist/{chunk-JOWQTHXJ.js.map → chunk-IBXG4UPB.js.map} +0 -0
- /package/dist/{chunk-X6RNJ2AX.js.map → chunk-IR6ZXRAK.js.map} +0 -0
- /package/dist/{chunk-S6KLQX2K.js.map → chunk-JCRJGCYA.js.map} +0 -0
- /package/dist/{chunk-3IYSX7EA.js.map → chunk-KQULVTQH.js.map} +0 -0
- /package/dist/{chunk-VQED4B7M.js.map → chunk-NBQDGYCQ.js.map} +0 -0
- /package/dist/{chunk-WDIHTINW.js.map → chunk-NWL7D4F7.js.map} +0 -0
- /package/dist/{chunk-P74R5MK7.js.map → chunk-PKX6CKAI.js.map} +0 -0
- /package/dist/{chunk-LPDFNN44.js.map → chunk-PWGE67XI.js.map} +0 -0
- /package/dist/{chunk-K5S5SZC6.js.map → chunk-PZT5RCSM.js.map} +0 -0
- /package/dist/{chunk-P65KKO2X.js.map → chunk-QRUTGCC2.js.map} +0 -0
- /package/dist/{chunk-3HFH3SBY.js.map → chunk-S2LEIH33.js.map} +0 -0
- /package/dist/{chunk-FSGDOD2E.js.map → chunk-SSTOQ3ZS.js.map} +0 -0
- /package/dist/{chunk-4JAHUMTT.js.map → chunk-VDG4G44B.js.map} +0 -0
- /package/dist/{chunk-GCAE6YQT.js.map → chunk-VO27MPHI.js.map} +0 -0
- /package/dist/{chunk-Q6KTIFXA.js.map → chunk-VYWZASYT.js.map} +0 -0
- /package/dist/{chunk-3642GYL7.js.map → chunk-XAUJF2GZ.js.map} +0 -0
- /package/dist/{chunk-R6VHZYGR.js.map → chunk-ZGXXR3DH.js.map} +0 -0
- /package/dist/{chunks-A6ALUARW.js.map → chunks-ODK2CKBM.js.map} +0 -0
- /package/dist/{formula-U6PMTA5U.js.map → formula-CHBBFZGG.js.map} +0 -0
- /package/dist/{formula-6SGEN5JG.js.map → formula-GQMRKV2P.js.map} +0 -0
- /package/dist/{image-ocr-EY54VBAM.js.map → image-ocr-6DPYV3Z2.js.map} +0 -0
- /package/dist/{image-ocr-XVU5MXQB.js.map → image-ocr-SAW7A4YW.js.map} +0 -0
- /package/dist/{model-bundle-BEAQMPMH.js.map → model-bundle-7K554JAW.js.map} +0 -0
- /package/dist/{models-AFAT3YQI.js.map → models-VU65CWCI.js.map} +0 -0
- /package/dist/{parse-worker-v2-MJOBLVHV.js.map → parse-worker-v2-435QCDJP.js.map} +0 -0
- /package/dist/{parser-D7W5Q6Y3.js.map → parser-5LUP4EF5.js.map} +0 -0
- /package/dist/{parser-UVFRWRGQ.js.map → parser-BD4VNEUA.js.map} +0 -0
- /package/dist/{pdf-ocr-DKRVXK2X.js.map → pdf-ocr-2TNMUGTU.js.map} +0 -0
- /package/dist/{pdf-ocr-ONGESN2A.js.map → pdf-ocr-GJDRJACB.js.map} +0 -0
- /package/dist/{profile-io-LLIOYHEN.js.map → profile-io-6XPZGJHS.js.map} +0 -0
- /package/dist/{rasterize-FIX7WWYA.js.map → rasterize-PERRMQDZ.js.map} +0 -0
- /package/dist/{redact-doc-VS5YBG7O.js.map → redact-doc-EA2QBZG6.js.map} +0 -0
- /package/dist/{render-5UIF2HOQ.js.map → render-YLCMDE5W.js.map} +0 -0
- /package/dist/{seal-LBSLEGXI.js.map → seal-HZX7BXWG.js.map} +0 -0
- /package/dist/{setup-VJ2C3MHC.js.map → setup-IJSICTFK.js.map} +0 -0
- /package/dist/{visual-UU4MYIFZ.js.map → visual-GFFZPSRH.js.map} +0 -0
- /package/dist/{watch-ZETJYMXE.js.map → watch-Z6PMJO7G.js.map} +0 -0
package/README.md
CHANGED
|
@@ -108,7 +108,7 @@ await writeFile("보고서.hwpx", Buffer.from(hwpx))
|
|
|
108
108
|
|
|
109
109
|
## 검증 결과
|
|
110
110
|
|
|
111
|
-
**4.22.14 · 2026-10-19 검증**(렌더 배치·PDF 글·표·OCR 은 4.24.
|
|
111
|
+
**4.22.14 · 2026-10-19 검증**(렌더 배치·PDF 글·표·OCR 은 4.24.20 · 2026-10-29). 고정 코퍼스와 참조 기준으로 측정한 결과입니다.
|
|
112
112
|
|
|
113
113
|
| 대상 | 규모 | 결과 |
|
|
114
114
|
| --- | --- | --- |
|
|
@@ -122,21 +122,21 @@ await writeFile("보고서.hwpx", Buffer.from(hwpx))
|
|
|
122
122
|
| PDF 전체 | 1,911문서 중 1,729건 채점 | 글 커버리지 **99.80%** |
|
|
123
123
|
| OCR | 53문서 · 102쪽 · 67표 | CER **0.0244** · 글자 재현율 **99.32%** · 표 구조 일치 **26/67** |
|
|
124
124
|
| opendataloader-bench | PDF 200문서 | 종합 기본값 **0.964** · OCR 끔 **0.939** |
|
|
125
|
-
| HWPX 렌더 배치 | 한컴 PDF 짝 1,565문서 · 글 조각
|
|
125
|
+
| HWPX 렌더 배치 | 한컴 PDF 짝 1,565문서(쪽 수 일치 1,505) · 글 조각 105,145 · 그림 4,240 | 글 위치(±1.5pt) **100,019 · 95.1%** · 그림 위치(±2pt) **4,149 · 97.9%** |
|
|
126
126
|
|
|
127
127
|
표 구조 점수와 셀 내용·화면 재현은 별도 지표입니다. 중첩표 100%는 위 101표에 한정하며 전체 PDF·조판의 완전 일치를 뜻하지 않습니다. 채점 범위·제외 기준·옵션별 수치·재현 방법은 [벤치마크 상세](docs/benchmarks.md)에 있습니다.
|
|
128
128
|
|
|
129
129
|
## 최근 업데이트
|
|
130
130
|
|
|
131
|
-
최신 배포: **[4.24.
|
|
131
|
+
최신 배포: **[4.24.20](https://github.com/chrisryugj/kordoc/releases/tag/v4.24.20)** · 2026-10-29
|
|
132
132
|
|
|
133
133
|
| 버전 | 주요 변경 | 수치 변화 |
|
|
134
134
|
| --- | --- | --- |
|
|
135
|
+
| 4.24.20 | 렌더: 칸 안 여러 쪽 떠 있는 표 뒤 문단의 쪽, 줄이 나아가지 않게 저장된 칸 캐시, 쪽마다 되돌아가지 않는 캐시(쪽 밖 글) | 렌더 쪽 수 일치 1,504 → **1,505**문서 · 글 위치 100,015 → **100,019** · 규제영향분석서 116쪽 → 66쪽(PDF 64) · 나머지 지표 그대로 |
|
|
136
|
+
| 4.24.19 | 렌더: 쪽 나눔으로 저장한 칸·병합 칸의 쪽 넘김(겹쳐 그리기·빈 쪽 하나 더), 쪽 바닥에서 갈리는 행의 첫 줄 간격, BOM 붙은 HWPX 렌더 실패 | 렌더 쪽 수 일치 1,492 → **1,504**문서 · 렌더 실패 17 → **14** · 글 위치 97,156 → **100,015**(기존 문서 3건 상승·하락 0, 새로 잰 문서 8건) · 나머지 지표 그대로 |
|
|
135
137
|
| 4.24.18 | 렌더: 칸 내용을 지운 표의 낡은 높이, 한 줄에 나란한 인라인 표의 기준선, 쪽 바닥에서 갈린 표 행의 남는 높이, 같은 문단에 함께 앵커한 표의 다음 쪽 넘김, 줄 정보 없는 문단 뒤 저장 위치 | 렌더 글 위치 96,993 → **97,156**(한컴 PDF 짝 1,565문서, 12문서 상승·하락 0) · 그림 위치 4,148 → **4,149** · 나머지 지표 그대로 |
|
|
136
138
|
| 4.24.17 | OCR: 검출기가 놓친 목차 줄 머리 장 번호·리더 뒤 쪽 번호, 두 줄로 잡혀 버려진 표 칸, 기울인 판의 그림자 글머리 ❐·○와 목차 ◎, 괄호 짝·거북등 괄호 〔〕·평행 기호 ∥, 리더가 끊겨 갈린 목차 줄 · 선택 검출 모델 `ocrDet: "medium"`(#176) | OCR CER 0.0250 → **0.0244** · 글자 재현율 99.27% → **99.32%** · 정밀도 99.54% → **99.56%** · 열화 입력 9종 모두 개선(원본 0.0394 → **0.0386**, 흐림 0.0550 → **0.0522**, 스캔 0.0693 → **0.0668**) |
|
|
137
139
|
| 4.24.16 | OCR: 짧은 목차 리더 ··, 과제 번호의 Ⅲ(파·피로 읽음)·네모 번호 앞 [ ·붙임표가 붙은 원문자 ①, 한글 뒤 로마 숫자 획 수(Ⅰ↔Ⅱ), 칸 테두리 앞·빈칸 긴 서식 칸 끝 닫는 괄호 | OCR CER 0.0255 → **0.0250** · 글자 재현율 99.23% → **99.27%** · 정밀도 99.53% → **99.54%** · 열화 입력 8종 개선(원본 0.0397 → **0.0394**, 스캔만 0.0692 → 0.0693) |
|
|
138
|
-
| 4.24.15 | OCR: 검출기가 놓친 서식 빈칸 네모 □□-□□□·목차 줄 머리 대시·줄 끝 홑 쪽 번호, 큰 글머리 ■, 리터 ℓ, 작은 고리 。, 윗줄에 묻힌 줄 머리 가운뎃점, 빗금 칸 환각 · PDF: 줄 간격 넓은 중·일문 두 줄 묶음 문단 잇기 | OCR CER 0.0260 → **0.0255** · 글자 재현율 99.19% → **99.23%** · 열화 입력 8종 개선(원본 0.0405 → **0.0397**, 잡음만 0.0485 → 0.0486) · PDF 어절 F1 99.066% → **99.067%** |
|
|
139
|
-
| 4.24.14 | 렌더: 칸 간격(`cellSpacing`) 표의 칸 자리 · OCR: 공백·따옴표로 읽은 홑낫표 「, 검출기가 놓친 쪽 번호, 라틴 약어 앞·C 로 읽은 줄 머리 ○, 고리만 든 칸 ○, 마름모 글머리 ◆·◇·◈, 서식 칸 끝 닫는 괄호 | 렌더 글 위치 87,577 → **87,643** · OCR CER 0.0263 → **0.0260** · 열화 입력 9종 모두 개선(원본 0.0409 → **0.0405**) |
|
|
140
140
|
|
|
141
141
|
이전 버전과 상세 내용은 [변경 이력](CHANGELOG.md)에 있습니다.
|
|
142
142
|
|
|
@@ -64,10 +64,10 @@ import {
|
|
|
64
64
|
usesGaejosikMunche,
|
|
65
65
|
usesLogicStructure,
|
|
66
66
|
validateHwpx
|
|
67
|
-
} from "./chunk-
|
|
67
|
+
} from "./chunk-S2LEIH33.js";
|
|
68
68
|
import {
|
|
69
69
|
blocksToChunks
|
|
70
|
-
} from "./chunk-
|
|
70
|
+
} from "./chunk-PWGE67XI.js";
|
|
71
71
|
import {
|
|
72
72
|
SPACE_EM_FIXED,
|
|
73
73
|
SPACE_EM_FONT,
|
|
@@ -92,20 +92,20 @@ import {
|
|
|
92
92
|
renderSceneToHtml,
|
|
93
93
|
simulateWrap,
|
|
94
94
|
simulateWrapKeepWord
|
|
95
|
-
} from "./chunk-
|
|
96
|
-
import "./chunk-
|
|
95
|
+
} from "./chunk-YUEPDCVZ.js";
|
|
96
|
+
import "./chunk-PKX6CKAI.js";
|
|
97
97
|
import {
|
|
98
98
|
blocksToPages
|
|
99
|
-
} from "./chunk-
|
|
99
|
+
} from "./chunk-VYWZASYT.js";
|
|
100
100
|
import "./chunk-VKVDMH7X.js";
|
|
101
101
|
import "./chunk-VQFA55IC.js";
|
|
102
102
|
import "./chunk-IQJKLIJV.js";
|
|
103
|
-
import "./chunk-
|
|
103
|
+
import "./chunk-VO27MPHI.js";
|
|
104
104
|
import {
|
|
105
105
|
blocksToMarkdown,
|
|
106
106
|
flattenLayoutTables,
|
|
107
107
|
hasStructuredCellContent
|
|
108
|
-
} from "./chunk-
|
|
108
|
+
} from "./chunk-KQULVTQH.js";
|
|
109
109
|
import {
|
|
110
110
|
ValueCursor,
|
|
111
111
|
applySplices,
|
|
@@ -119,11 +119,11 @@ import {
|
|
|
119
119
|
isLabelCell,
|
|
120
120
|
placeSealHwpx,
|
|
121
121
|
scanSectionXml
|
|
122
|
-
} from "./chunk-
|
|
123
|
-
import "./chunk-
|
|
122
|
+
} from "./chunk-ZGXXR3DH.js";
|
|
123
|
+
import "./chunk-NWL7D4F7.js";
|
|
124
124
|
import {
|
|
125
125
|
VERSION
|
|
126
|
-
} from "./chunk-
|
|
126
|
+
} from "./chunk-VDG4G44B.js";
|
|
127
127
|
import {
|
|
128
128
|
DEFAULT_REDACT_RULES,
|
|
129
129
|
redactMarkdown,
|
|
@@ -239,4 +239,4 @@ export {
|
|
|
239
239
|
usesLogicStructure,
|
|
240
240
|
validateHwpx
|
|
241
241
|
};
|
|
242
|
-
//# sourceMappingURL=-
|
|
242
|
+
//# sourceMappingURL=-3RL7OPC7.js.map
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
#!/usr/bin/env node
|
|
2
2
|
import {
|
|
3
3
|
KordocError
|
|
4
|
-
} from "./chunk-
|
|
4
|
+
} from "./chunk-VDG4G44B.js";
|
|
5
5
|
|
|
6
6
|
// src/shared/offline.ts
|
|
7
7
|
import { realpathSync } from "fs";
|
|
@@ -49,4 +49,4 @@ export {
|
|
|
49
49
|
getAccessRoot,
|
|
50
50
|
assertWithinRoot
|
|
51
51
|
};
|
|
52
|
-
//# sourceMappingURL=chunk-
|
|
52
|
+
//# sourceMappingURL=chunk-23GHEDMW.js.map
|
|
@@ -31,7 +31,7 @@ import {
|
|
|
31
31
|
splitRowBands,
|
|
32
32
|
starRun,
|
|
33
33
|
tallInkCount
|
|
34
|
-
} from "./chunk-
|
|
34
|
+
} from "./chunk-GAHHZ32A.js";
|
|
35
35
|
import {
|
|
36
36
|
validateOcrOptions
|
|
37
37
|
} from "./chunk-IQJKLIJV.js";
|
|
@@ -42,11 +42,11 @@ import {
|
|
|
42
42
|
ocrSourcePath,
|
|
43
43
|
parseCharacterDict,
|
|
44
44
|
resolveOcrModels
|
|
45
|
-
} from "./chunk-
|
|
45
|
+
} from "./chunk-QRUTGCC2.js";
|
|
46
46
|
import {
|
|
47
47
|
KordocError,
|
|
48
48
|
OPTIONAL_DEP_INSTALL_HINT
|
|
49
|
-
} from "./chunk-
|
|
49
|
+
} from "./chunk-VDG4G44B.js";
|
|
50
50
|
|
|
51
51
|
// src/ocr/ruling-lines.ts
|
|
52
52
|
var INK_LUMA_MAX = 205;
|
|
@@ -2259,4 +2259,4 @@ export {
|
|
|
2259
2259
|
groupOcrLines,
|
|
2260
2260
|
ocrItemsToBlocks
|
|
2261
2261
|
};
|
|
2262
|
-
//# sourceMappingURL=chunk-
|
|
2262
|
+
//# sourceMappingURL=chunk-3EHQKGTL.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 _chunkOJB7QZHPcjs = require('./chunk-OJB7QZHP.cjs');
|
|
4
4
|
|
|
5
5
|
// src/image/transcode.ts
|
|
6
6
|
var _zlib = require('zlib');
|
|
@@ -140,7 +140,7 @@ function inlineImagesIntoMarkdown(markdown, images, opts) {
|
|
|
140
140
|
}
|
|
141
141
|
|
|
142
142
|
// src/page-markdown.ts
|
|
143
|
-
function blocksToPages(blocks, render =
|
|
143
|
+
function blocksToPages(blocks, render = _chunkOJB7QZHPcjs.blocksToMarkdown) {
|
|
144
144
|
const firstNumbered = _optionalChain([blocks, 'access', _2 => _2.find, 'call', _3 => _3((b) => typeof b.pageNumber === "number"), 'optionalAccess', _4 => _4.pageNumber]);
|
|
145
145
|
if (firstNumbered === void 0) return void 0;
|
|
146
146
|
const byPage = /* @__PURE__ */ new Map();
|
|
@@ -168,4 +168,4 @@ function blocksToPages(blocks, render = _chunk2WNVFE4Vcjs.blocksToMarkdown) {
|
|
|
168
168
|
|
|
169
169
|
|
|
170
170
|
exports.bmpToPng = bmpToPng; exports.encodePng = encodePng; exports.inlineImagesIntoMarkdown = inlineImagesIntoMarkdown; exports.blocksToPages = blocksToPages;
|
|
171
|
-
//# sourceMappingURL=chunk-
|
|
171
|
+
//# sourceMappingURL=chunk-CSWE6FQ4.cjs.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/chunk-REOGJMES.cjs","../src/image/transcode.ts","../src/page-markdown.ts"],"names":[],"mappings":"AAAA;AACE;AACF,wDAA6B;AAC7B;AACA;ACMA,4BAA4B;AAI5B,IAAM,UAAA,EAAA,CAAa,CAAA,EAAA,GAAM;AACvB,EAAA,MAAM,MAAA,EAAQ,IAAI,WAAA,CAAY,GAAG,CAAA;AACjC,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,GAAA,EAAK,CAAA,EAAA,EAAK;AAC5B,IAAA,IAAI,EAAA,EAAI,CAAA;AACR,IAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,CAAA,EAAG,CAAA,EAAA,EAAK;AAC1B,MAAA,EAAA,EAAK,EAAA,EAAI,EAAA,EAAM,WAAA,EAAc,EAAA,IAAM,EAAA,EAAO,EAAA,IAAM,CAAA;AAAA,IAClD;AACA,IAAA,KAAA,CAAM,CAAC,EAAA,EAAI,EAAA,IAAM,CAAA;AAAA,EACnB;AACA,EAAA,OAAO,KAAA;AACT,CAAA,CAAA,CAAG,CAAA;AAGH,SAAS,KAAA,CAAM,KAAA,EAA2B;AACxC,EAAA,IAAI,EAAA,EAAI,UAAA;AACR,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,KAAA,CAAM,MAAA,EAAQ,CAAA,EAAA,EAAK;AACrC,IAAA,EAAA,EAAI,SAAA,CAAA,CAAW,EAAA,EAAI,KAAA,CAAM,CAAC,CAAA,EAAA,EAAK,GAAI,EAAA,EAAK,EAAA,IAAM,CAAA;AAAA,EAChD;AACA,EAAA,OAAA,CAAQ,EAAA,EAAI,UAAA,EAAA,IAAgB,CAAA;AAC9B;AAIA,IAAM,OAAA,EAAS,CAAA;AAEf,IAAM,QAAA,EAAU,KAAA;AAOhB,IAAM,WAAA,EAAa,IAAA;AAUZ,SAAS,QAAA,CAAS,GAAA,EAAoC;AAE3D,EAAA,GAAA,CAAI,GAAA,CAAI,OAAA,EAAS,EAAA,EAAI,OAAO,IAAA;AAC5B,EAAA,GAAA,CAAI,GAAA,CAAI,CAAC,EAAA,IAAM,GAAA,GAAQ,GAAA,CAAI,CAAC,EAAA,IAAM,EAAA,EAAM,OAAO,IAAA;AAE/C,EAAA,MAAM,GAAA,EAAK,IAAI,QAAA,CAAS,GAAA,CAAI,MAAA,EAAQ,GAAA,CAAI,UAAA,EAAY,GAAA,CAAI,UAAU,CAAA;AAClE,EAAA,MAAM,WAAA,EAAa,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AACxC,EAAA,MAAM,WAAA,EAAa,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AACxC,EAAA,GAAA,CAAI,WAAA,EAAa,EAAA,EAAI,OAAO,IAAA;AAE5B,EAAA,MAAM,MAAA,EAAQ,EAAA,CAAG,QAAA,CAAS,EAAA,EAAI,IAAI,CAAA;AAClC,EAAA,MAAM,UAAA,EAAY,EAAA,CAAG,QAAA,CAAS,EAAA,EAAI,IAAI,CAAA;AACtC,EAAA,MAAM,SAAA,EAAW,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AACtC,EAAA,MAAM,YAAA,EAAc,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AAEzC,EAAA,GAAA,CAAI,YAAA,IAAgB,MAAA,EAAQ,OAAO,IAAA;AACnC,EAAA,GAAA,CAAI,SAAA,IAAa,GAAA,GAAM,SAAA,IAAa,EAAA,EAAI,OAAO,IAAA;AAC/C,EAAA,GAAA,CAAI,MAAA,GAAS,EAAA,GAAK,UAAA,IAAc,CAAA,EAAG,OAAO,IAAA;AAC1C,EAAA,GAAA,CAAI,MAAA,EAAQ,QAAA,GAAW,IAAA,CAAK,GAAA,CAAI,SAAS,EAAA,EAAI,OAAA,EAAS,OAAO,IAAA;AAE7D,EAAA,MAAM,QAAA,EAAU,UAAA,EAAY,CAAA;AAC5B,EAAA,MAAM,OAAA,EAAS,IAAA,CAAK,GAAA,CAAI,SAAS,CAAA;AAEjC,EAAA,GAAA,CAAI,MAAA,EAAQ,OAAA,EAAS,UAAA,EAAY,OAAO,IAAA;AACxC,EAAA,MAAM,cAAA,EAAgB,SAAA,GAAY,CAAA;AAElC,EAAA,MAAM,UAAA,EAAa,MAAA,EAAQ,cAAA,EAAgB,EAAA,EAAK,CAAC,CAAA;AACjD,EAAA,GAAA,CAAI,WAAA,EAAa,UAAA,EAAY,OAAA,EAAS,GAAA,CAAI,MAAA,EAAQ,OAAO,IAAA;AAEzD,EAAA,MAAM,KAAA,EAAO,IAAI,UAAA,CAAW,MAAA,EAAQ,OAAA,EAAS,CAAC,CAAA;AAC9C,EAAA,IAAI,SAAA,EAAW,CAAA;AAEf,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,MAAA,EAAQ,CAAA,EAAA,EAAK;AAE/B,IAAA,MAAM,OAAA,EAAS,QAAA,EAAU,EAAA,EAAI,OAAA,EAAS,EAAA,EAAI,CAAA;AAC1C,IAAA,IAAI,IAAA,EAAM,WAAA,EAAa,OAAA,EAAS,SAAA;AAChC,IAAA,IAAI,IAAA,EAAM,EAAA,EAAI,MAAA,EAAQ,CAAA;AACtB,IAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,KAAA,EAAO,CAAA,EAAA,EAAK;AAE9B,MAAA,IAAA,CAAK,GAAG,EAAA,EAAI,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AACvB,MAAA,IAAA,CAAK,IAAA,EAAM,CAAC,EAAA,EAAI,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AAC3B,MAAA,IAAA,CAAK,IAAA,EAAM,CAAC,EAAA,EAAI,GAAA,CAAI,GAAG,CAAA;AACvB,MAAA,MAAM,EAAA,EAAI,SAAA,IAAa,GAAA,EAAK,GAAA,CAAI,IAAA,EAAM,CAAC,EAAA,EAAI,GAAA;AAC3C,MAAA,IAAA,CAAK,IAAA,EAAM,CAAC,EAAA,EAAI,CAAA;AAChB,MAAA,SAAA,GAAY,CAAA;AACZ,MAAA,IAAA,GAAO,aAAA;AACP,MAAA,IAAA,GAAO,CAAA;AAAA,IACT;AAAA,EACF;AAIA,EAAA,GAAA,CAAI,SAAA,IAAa,GAAA,GAAM,SAAA,IAAa,CAAA,EAAG;AACrC,IAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,IAAA,CAAK,MAAA,EAAQ,EAAA,GAAK,CAAA,EAAG,IAAA,CAAK,CAAC,EAAA,EAAI,GAAA;AAAA,EACrD;AAEA,EAAA,OAAO,SAAA,CAAU,KAAA,EAAO,MAAA,EAAQ,IAAI,CAAA;AACtC;AAIA,IAAM,cAAA,EAAgB,UAAA,CAAW,EAAA,CAAG,GAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAI,CAAA;AAGlF,SAAS,KAAA,CAAM,IAAA,EAAc,IAAA,EAA8B;AACzD,EAAA,MAAM,KAAA,EAAO,IAAI,UAAA,CAAW,EAAA,EAAI,IAAA,CAAK,MAAM,CAAA;AAC3C,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AAEhB,EAAA,MAAM,IAAA,EAAM,IAAI,UAAA,CAAW,EAAA,EAAI,IAAA,CAAK,OAAA,EAAS,CAAC,CAAA;AAC9C,EAAA,MAAM,GAAA,EAAK,IAAI,QAAA,CAAS,GAAA,CAAI,MAAM,CAAA;AAClC,EAAA,EAAA,CAAG,SAAA,CAAU,CAAA,EAAG,IAAA,CAAK,MAAA,EAAQ,KAAK,CAAA;AAClC,EAAA,GAAA,CAAI,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AACf,EAAA,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAA,CAAK,MAAA,EAAQ,KAAA,CAAM,IAAI,CAAA,EAAG,KAAK,CAAA;AAChD,EAAA,OAAO,GAAA;AACT;AAGO,SAAS,SAAA,CAAU,KAAA,EAAe,MAAA,EAAgB,IAAA,EAA8B;AAErF,EAAA,MAAM,KAAA,EAAO,IAAI,UAAA,CAAW,EAAE,CAAA;AAC9B,EAAA,MAAM,GAAA,EAAK,IAAI,QAAA,CAAS,IAAA,CAAK,MAAM,CAAA;AACnC,EAAA,EAAA,CAAG,SAAA,CAAU,CAAA,EAAG,KAAA,EAAO,KAAK,CAAA;AAC5B,EAAA,EAAA,CAAG,SAAA,CAAU,CAAA,EAAG,MAAA,EAAQ,KAAK,CAAA;AAC7B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,CAAA;AACV,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,CAAA;AAGV,EAAA,MAAM,OAAA,EAAS,MAAA,EAAQ,CAAA;AACvB,EAAA,MAAM,IAAA,EAAM,IAAI,UAAA,CAAA,CAAY,OAAA,EAAS,CAAA,EAAA,EAAK,MAAM,CAAA;AAChD,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,MAAA,EAAQ,CAAA,EAAA,EAAK;AAC/B,IAAA,MAAM,SAAA,EAAW,EAAA,EAAA,CAAK,OAAA,EAAS,CAAA,CAAA;AAC/B,IAAA,GAAA,CAAI,QAAQ,EAAA,EAAI,CAAA;AAChB,IAAA,GAAA,CAAI,GAAA,CAAI,IAAA,CAAK,QAAA,CAAS,EAAA,EAAI,MAAA,EAAQ,EAAA,EAAI,OAAA,EAAS,MAAM,CAAA,EAAG,SAAA,EAAW,CAAC,CAAA;AAAA,EACtE;AACA,EAAA,MAAM,KAAA,EAAO,+BAAA,GAAe,CAAA;AAE5B,EAAA,MAAM,UAAA,EAAY,KAAA,CAAM,MAAA,EAAQ,IAAI,CAAA;AACpC,EAAA,MAAM,UAAA,EAAY,KAAA,CAAM,MAAA,EAAQ,IAAI,CAAA;AACpC,EAAA,MAAM,UAAA,EAAY,KAAA,CAAM,MAAA,EAAQ,IAAI,UAAA,CAAW,CAAC,CAAC,CAAA;AAEjD,EAAA,MAAM,IAAA,EAAM,IAAI,UAAA,CAAW,aAAA,CAAc,OAAA,EAAS,SAAA,CAAU,OAAA,EAAS,SAAA,CAAU,OAAA,EAAS,SAAA,CAAU,MAAM,CAAA;AACxG,EAAA,IAAI,EAAA,EAAI,CAAA;AACR,EAAA,GAAA,CAAI,GAAA,CAAI,aAAA,EAAe,CAAC,CAAA;AAAG,EAAA,EAAA,GAAK,aAAA,CAAc,MAAA;AAC9C,EAAA,GAAA,CAAI,GAAA,CAAI,SAAA,EAAW,CAAC,CAAA;AAAG,EAAA,EAAA,GAAK,SAAA,CAAU,MAAA;AACtC,EAAA,GAAA,CAAI,GAAA,CAAI,SAAA,EAAW,CAAC,CAAA;AAAG,EAAA,EAAA,GAAK,SAAA,CAAU,MAAA;AACtC,EAAA,GAAA,CAAI,GAAA,CAAI,SAAA,EAAW,CAAC,CAAA;AACpB,EAAA,OAAO,GAAA;AACT;AAqBO,SAAS,wBAAA,CACd,QAAA,EACA,MAAA,EACA,IAAA,EACQ;AACR,EAAA,MAAM,SAAA,kBAAW,IAAA,2BAAM,WAAA,IAAa,KAAA;AAGpC,EAAA,MAAM,KAAA,kBAAO,IAAI,GAAA,CAAoB,CAAA;AACrC,EAAA,IAAA,CAAA,MAAW,IAAA,GAAO,MAAA,EAAQ;AACxB,IAAA,IAAI,MAAA,EAAQ,GAAA,CAAI,IAAA;AAChB,IAAA,IAAI,KAAA,EAAO,GAAA,CAAI,QAAA;AACf,IAAA,GAAA,CAAI,SAAA,GAAY,KAAA,IAAS,WAAA,EAAa;AACpC,MAAA,MAAM,IAAA,EAAM,QAAA,CAAS,GAAA,CAAI,IAAI,CAAA;AAC7B,MAAA,GAAA,CAAI,GAAA,EAAK;AACP,QAAA,MAAA,EAAQ,GAAA;AACR,QAAA,KAAA,EAAO,WAAA;AAAA,MACT;AAAA,IACF;AACA,IAAA,IAAA,CAAK,GAAA,CAAI,GAAA,CAAI,QAAA,EAAU,CAAA,KAAA,EAAQ,IAAI,CAAA,QAAA,EAAW,MAAA,CAAO,IAAA,CAAK,KAAK,CAAA,CAAE,QAAA,CAAS,QAAQ,CAAC,CAAA,CAAA;AACrF,EAAA;AAC4B,EAAA;AAE8D,EAAA;AAC/D,IAAA;AACuB,IAAA;AACjD,EAAA;AACgG,EAAA;AACtE,IAAA;AAC0B,IAAA;AACpD,EAAA;AACM,EAAA;AACT;ADhFwF;AACA;AEnH+C;AAC3D,EAAA;AAClC,EAAA;AAEE,EAAA;AAC5B,EAAA;AACc,EAAA;AACgC,IAAA;AACzB,IAAA;AACJ,IAAA;AACG,IAAA;AAClC,EAAA;AAEiC,EAAA;AACF,EAAA;AACA,EAAA;AAEA,EAAA;AAC6B,EAAA;AAClB,IAAA;AACiC,IAAA;AAC3E,EAAA;AACO,EAAA;AACT;AFkHwF;AACA;AACA;AACA;AACA;AACA;AACA","file":"/Users/mong-e/workspace/kordoc/dist/chunk-REOGJMES.cjs","sourcesContent":[null,"/**\n * 순수 JS BMP → PNG 트랜스코더 + 마크다운 이미지 인라이너.\n *\n * HWP5 BinData 임베드 이미지는 비압축 BMP(수 MB)인 경우가 많다. AI 에이전트가\n * 자체 완결형 마크다운을 받도록, BMP 를 PNG(무손실 압축)로 변환한 뒤 base64\n * data URI 로 마크다운에 인라인한다.\n *\n * 의존성 최소화 원칙 — Node 내장(node:zlib, Buffer)만 사용한다.\n */\n\nimport { deflateSync } from \"node:zlib\"\n\n// ─── CRC32 (PNG 청크용) ──────────────────────────────\n\nconst CRC_TABLE = (() => {\n const table = new Uint32Array(256)\n for (let n = 0; n < 256; n++) {\n let c = n\n for (let k = 0; k < 8; k++) {\n c = (c & 1) ? (0xedb88320 ^ (c >>> 1)) : (c >>> 1)\n }\n table[n] = c >>> 0\n }\n return table\n})()\n\n/** 표준 PNG CRC32 (다항식 0xEDB88320) */\nfunction crc32(bytes: Uint8Array): number {\n let c = 0xffffffff\n for (let i = 0; i < bytes.length; i++) {\n c = CRC_TABLE[(c ^ bytes[i]) & 0xff] ^ (c >>> 8)\n }\n return (c ^ 0xffffffff) >>> 0\n}\n\n// ─── BMP 디코드 ──────────────────────────────────────\n\nconst BI_RGB = 0\n/** 폭·높이 상한 — 비정상 헤더에 의한 과대 할당 방지 */\nconst MAX_DIM = 0x7fff\n/**\n * 총 픽셀(W×H) 상한 — MAX_DIM 은 각 변만 제한하므로 곱의 폭주(과대 할당 + 긴 deflateSync\n * 동기 블로킹)를 별도로 차단한다. raw 저장 BMP 는 상위 inflate 100MB 캡을 우회하므로 필요.\n * 36MP = A4 전면 600dpi 스캔(≈35MP)까지 허용 — 그 이상은 rgba+raw 로 300MB 를 넘겨\n * MCP 서버 이벤트루프를 수 초 세우므로 원본 바이트 폴백으로 돌린다.\n */\nconst MAX_PIXELS = 36_000_000\n\n/**\n * 비압축 BMP(24/32bpp, BI_RGB) → PNG(8bit RGBA).\n *\n * 지원: BITMAPFILEHEADER + BITMAPINFOHEADER(biSize>=40), biBitCount 24/32,\n * biCompression BI_RGB, bottom-up(양수 biHeight)/top-down(음수), 24bpp 4바이트 행 패딩.\n * 미지원 변형(팔레트/8bpp·RLE·BITFIELDS·OS/2 헤더 등)은 null 을 반환하여\n * 호출부가 원본 바이트로 폴백하게 한다.\n */\nexport function bmpToPng(bmp: Uint8Array): Uint8Array | null {\n // BITMAPFILEHEADER(14) + BITMAPINFOHEADER(40)\n if (bmp.length < 54) return null\n if (bmp[0] !== 0x42 || bmp[1] !== 0x4d) return null // \"BM\"\n\n const dv = new DataView(bmp.buffer, bmp.byteOffset, bmp.byteLength)\n const dataOffset = dv.getUint32(10, true)\n const headerSize = dv.getUint32(14, true)\n if (headerSize < 40) return null // OS/2 BITMAPCOREHEADER 등 미지원\n\n const width = dv.getInt32(18, true)\n const rawHeight = dv.getInt32(22, true)\n const bitCount = dv.getUint16(28, true)\n const compression = dv.getUint32(30, true)\n\n if (compression !== BI_RGB) return null // RLE / BITFIELDS 미지원\n if (bitCount !== 24 && bitCount !== 32) return null // 팔레트 / 8bpp 등 미지원\n if (width <= 0 || rawHeight === 0) return null\n if (width > MAX_DIM || Math.abs(rawHeight) > MAX_DIM) return null\n\n const topDown = rawHeight < 0\n const height = Math.abs(rawHeight)\n // 총 픽셀 상한 검사 — rgba/raw 할당·deflate 전에 조기 차단 (각 변은 MAX_DIM 이하여도 곱은 폭주 가능)\n if (width * height > MAX_PIXELS) return null\n const bytesPerPixel = bitCount >> 3\n // 행 스트라이드 — 4바이트 경계 패딩 (24bpp 에서 유효)\n const rowStride = (width * bytesPerPixel + 3) & ~3\n if (dataOffset + rowStride * height > bmp.length) return null // 잘린 파일\n\n const rgba = new Uint8Array(width * height * 4)\n let anyAlpha = 0\n\n for (let y = 0; y < height; y++) {\n // bottom-up 은 첫 행이 이미지 맨 아래 → 뒤집어 읽는다\n const srcRow = topDown ? y : height - 1 - y\n let src = dataOffset + srcRow * rowStride\n let dst = y * width * 4\n for (let x = 0; x < width; x++) {\n // BMP 픽셀은 BGR(A) 순\n rgba[dst] = bmp[src + 2] // R\n rgba[dst + 1] = bmp[src + 1] // G\n rgba[dst + 2] = bmp[src] // B\n const a = bitCount === 32 ? bmp[src + 3] : 255\n rgba[dst + 3] = a\n anyAlpha |= a\n src += bytesPerPixel\n dst += 4\n }\n }\n\n // 32bpp BI_RGB 의 4번째 바이트는 \"미정의\" — 전부 0 이면 전면 투명으로 오해되므로\n // 불투명(255)으로 승격한다. 실제 알파가 있으면(anyAlpha!=0) 그대로 보존.\n if (bitCount === 32 && anyAlpha === 0) {\n for (let i = 3; i < rgba.length; i += 4) rgba[i] = 255\n }\n\n return encodePng(width, height, rgba)\n}\n\n// ─── PNG 인코드 (8bit RGBA, color type 6) ────────────\n\nconst PNG_SIGNATURE = Uint8Array.of(0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a)\n\n/** [4B 길이][4B 타입][데이터][4B CRC] 청크 조립. CRC 는 (타입+데이터) 기준. */\nfunction chunk(type: string, data: Uint8Array): Uint8Array {\n const body = new Uint8Array(4 + data.length)\n body[0] = type.charCodeAt(0)\n body[1] = type.charCodeAt(1)\n body[2] = type.charCodeAt(2)\n body[3] = type.charCodeAt(3)\n body.set(data, 4)\n\n const out = new Uint8Array(8 + data.length + 4)\n const dv = new DataView(out.buffer)\n dv.setUint32(0, data.length, false) // 길이(데이터만, big-endian)\n out.set(body, 4)\n dv.setUint32(8 + data.length, crc32(body), false)\n return out\n}\n\n/** 8bit RGBA 픽셀 버퍼 → PNG (PDF 이미지 XObject 추출에서도 사용) */\nexport function encodePng(width: number, height: number, rgba: Uint8Array): Uint8Array {\n // IHDR: width, height, bit depth 8, color type 6(RGBA), compression 0, filter 0, interlace 0\n const ihdr = new Uint8Array(13)\n const iv = new DataView(ihdr.buffer)\n iv.setUint32(0, width, false)\n iv.setUint32(4, height, false)\n ihdr[8] = 8\n ihdr[9] = 6\n\n // IDAT: 각 스캔라인 앞에 필터 바이트 0(None) 을 붙인 raw 데이터를 zlib deflate\n const stride = width * 4\n const raw = new Uint8Array((stride + 1) * height)\n for (let y = 0; y < height; y++) {\n const rowStart = y * (stride + 1)\n raw[rowStart] = 0 // filter: None\n raw.set(rgba.subarray(y * stride, y * stride + stride), rowStart + 1)\n }\n const idat = deflateSync(raw)\n\n const ihdrChunk = chunk(\"IHDR\", ihdr)\n const idatChunk = chunk(\"IDAT\", idat)\n const iendChunk = chunk(\"IEND\", new Uint8Array(0))\n\n const out = new Uint8Array(PNG_SIGNATURE.length + ihdrChunk.length + idatChunk.length + iendChunk.length)\n let o = 0\n out.set(PNG_SIGNATURE, o); o += PNG_SIGNATURE.length\n out.set(ihdrChunk, o); o += ihdrChunk.length\n out.set(idatChunk, o); o += idatChunk.length\n out.set(iendChunk, o)\n return out\n}\n\n// ─── 마크다운 이미지 인라이너 ────────────────────────\n\ninterface InlineImage {\n filename: string\n data: Uint8Array\n mimeType: string\n}\n\n/**\n * 마크다운의 이미지 참조를 base64 data URI 로 치환해 자체 완결형 마크다운을 만든다.\n * 두 방출 형태를 모두 커버한다:\n * - `` / `` — 일반 문단 (CLI 접두사 포함)\n * - `<img src=\"FILENAME\">` — 병합/중첩 표 셀 (table/builder.ts HTML 표 경로).\n * 이걸 빼면 CLI `--inline-images` 가 이미지 저장을 생략하면서 표 셀 이미지만\n * dangling 참조로 유실된다.\n *\n * @param opts.compress 기본 true. `image/bmp` 는 PNG 로 트랜스코딩 후 인라인하며,\n * 트랜스코딩 실패 시 원본 바이트+원본 MIME 으로 폴백한다.\n */\nexport function inlineImagesIntoMarkdown(\n markdown: string,\n images: ReadonlyArray<InlineImage>,\n opts?: { compress?: boolean },\n): string {\n const compress = opts?.compress !== false\n // 파일명 → data URI 사전 구축 후 단일 패스 치환 — 이미지 k개마다 전문 재스캔하던\n // O(k²) 제거. 맵에 없는 파일명 참조는 원문 그대로 둔다 (기존 계약 동일).\n const uris = new Map<string, string>()\n for (const img of images) {\n let bytes = img.data\n let mime = img.mimeType\n if (compress && mime === \"image/bmp\") {\n const png = bmpToPng(img.data)\n if (png) {\n bytes = png\n mime = \"image/png\"\n }\n }\n uris.set(img.filename, `data:${mime};base64,${Buffer.from(bytes).toString(\"base64\")}`)\n }\n if (uris.size === 0) return markdown\n // FILENAME 과 images/FILENAME 접두사를 모두 정확한 파일명 기준으로 치환\n let out = markdown.replace(/!\\[image\\]\\((?:images\\/)?([^()\\s]+)\\)/g, (m, name: string) => {\n const uri = uris.get(name)\n return uri !== undefined ? `` : m\n })\n out = out.replace(/(<img\\b[^>]*\\bsrc=\")(?:images\\/)?([^\"]*)(\")/g, (m, pre: string, name: string, post: string) => {\n const uri = uris.get(name)\n return uri !== undefined ? `${pre}${uri}${post}` : m\n })\n return out\n}\n","/**\n * 페이지별 마크다운 사영 (#68)\n *\n * 문서를 다시 파싱하지 않고 이미 만들어진 `IRBlock.pageNumber` 로만 블록을 갈라\n * 페이지마다 `blocksToMarkdown()` 을 돌린다. 페이지 번호의 신뢰도는 파서가\n * 정하는 것이고(`metadata.pageMode` — \"layout\" 은 실제 페이지, \"section\" 은\n * 섹션 근사), 여기서는 그 값을 그대로 존중한다.\n */\n\nimport type { IRBlock, PageMarkdown } from \"./types.js\"\nimport { blocksToMarkdown } from \"./table/builder.js\"\n\n/**\n * 블록을 페이지 번호로 묶어 페이지별 마크다운을 만든다.\n *\n * - 어떤 블록에도 페이지 번호가 없으면 `undefined` — 페이지 개념이 없는 포맷에\n * \"전부 1페이지\" 같은 없는 사실을 만들어 내지 않는다.\n * - 번호가 없는 블록은 **직전 블록의 페이지로 이어붙인다.** 블록은 읽기 순서라\n * 이게 원문 위치에 가장 가깝고, 무엇보다 어느 페이지에도 안 실려 조용히\n * 사라지는 블록이 생기지 않는다. 선두의 번호 없는 블록은 첫 실번호 페이지로.\n * - 관측된 최소~최대 페이지 사이의 빈 페이지도 항목으로 낸다. 여러 페이지에\n * 걸친 표는 시작 페이지 한 블록이라 중간 페이지가 실제로 비어 있고, 이때\n * 항목을 빼면 소비자가 배열 길이로 페이지 수를 셀 수 없다.\n * - `render` 는 쪽 블록 → 마크다운. 문서 마크다운에 포맷 고유 마무리가 붙는 파서(PDF: 쪽번호 제거·균등배분 정리)는\n * 같은 마무리를 넘겨 쪽 마크다운이 문서 마크다운과 어긋나지 않게 한다.\n */\nexport function blocksToPages(blocks: IRBlock[], render: (blocks: IRBlock[]) => string = blocksToMarkdown): PageMarkdown[] | undefined {\n const firstNumbered = blocks.find(b => typeof b.pageNumber === \"number\")?.pageNumber\n if (firstNumbered === undefined) return undefined\n\n const byPage = new Map<number, IRBlock[]>()\n let current = firstNumbered\n for (const block of blocks) {\n if (typeof block.pageNumber === \"number\") current = block.pageNumber\n const bucket = byPage.get(current)\n if (bucket) bucket.push(block)\n else byPage.set(current, [block])\n }\n\n const numbers = [...byPage.keys()]\n const min = Math.min(...numbers)\n const max = Math.max(...numbers)\n\n const pages: PageMarkdown[] = []\n for (let pageNumber = min; pageNumber <= max; pageNumber++) {\n const pageBlocks = byPage.get(pageNumber)\n pages.push({ pageNumber, markdown: pageBlocks ? render(pageBlocks) : \"\" })\n }\n return pages\n}\n"]}
|
|
1
|
+
{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/chunk-CSWE6FQ4.cjs","../src/image/transcode.ts","../src/page-markdown.ts"],"names":[],"mappings":"AAAA;AACE;AACF,wDAA6B;AAC7B;AACA;ACMA,4BAA4B;AAI5B,IAAM,UAAA,EAAA,CAAa,CAAA,EAAA,GAAM;AACvB,EAAA,MAAM,MAAA,EAAQ,IAAI,WAAA,CAAY,GAAG,CAAA;AACjC,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,GAAA,EAAK,CAAA,EAAA,EAAK;AAC5B,IAAA,IAAI,EAAA,EAAI,CAAA;AACR,IAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,CAAA,EAAG,CAAA,EAAA,EAAK;AAC1B,MAAA,EAAA,EAAK,EAAA,EAAI,EAAA,EAAM,WAAA,EAAc,EAAA,IAAM,EAAA,EAAO,EAAA,IAAM,CAAA;AAAA,IAClD;AACA,IAAA,KAAA,CAAM,CAAC,EAAA,EAAI,EAAA,IAAM,CAAA;AAAA,EACnB;AACA,EAAA,OAAO,KAAA;AACT,CAAA,CAAA,CAAG,CAAA;AAGH,SAAS,KAAA,CAAM,KAAA,EAA2B;AACxC,EAAA,IAAI,EAAA,EAAI,UAAA;AACR,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,KAAA,CAAM,MAAA,EAAQ,CAAA,EAAA,EAAK;AACrC,IAAA,EAAA,EAAI,SAAA,CAAA,CAAW,EAAA,EAAI,KAAA,CAAM,CAAC,CAAA,EAAA,EAAK,GAAI,EAAA,EAAK,EAAA,IAAM,CAAA;AAAA,EAChD;AACA,EAAA,OAAA,CAAQ,EAAA,EAAI,UAAA,EAAA,IAAgB,CAAA;AAC9B;AAIA,IAAM,OAAA,EAAS,CAAA;AAEf,IAAM,QAAA,EAAU,KAAA;AAOhB,IAAM,WAAA,EAAa,IAAA;AAUZ,SAAS,QAAA,CAAS,GAAA,EAAoC;AAE3D,EAAA,GAAA,CAAI,GAAA,CAAI,OAAA,EAAS,EAAA,EAAI,OAAO,IAAA;AAC5B,EAAA,GAAA,CAAI,GAAA,CAAI,CAAC,EAAA,IAAM,GAAA,GAAQ,GAAA,CAAI,CAAC,EAAA,IAAM,EAAA,EAAM,OAAO,IAAA;AAE/C,EAAA,MAAM,GAAA,EAAK,IAAI,QAAA,CAAS,GAAA,CAAI,MAAA,EAAQ,GAAA,CAAI,UAAA,EAAY,GAAA,CAAI,UAAU,CAAA;AAClE,EAAA,MAAM,WAAA,EAAa,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AACxC,EAAA,MAAM,WAAA,EAAa,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AACxC,EAAA,GAAA,CAAI,WAAA,EAAa,EAAA,EAAI,OAAO,IAAA;AAE5B,EAAA,MAAM,MAAA,EAAQ,EAAA,CAAG,QAAA,CAAS,EAAA,EAAI,IAAI,CAAA;AAClC,EAAA,MAAM,UAAA,EAAY,EAAA,CAAG,QAAA,CAAS,EAAA,EAAI,IAAI,CAAA;AACtC,EAAA,MAAM,SAAA,EAAW,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AACtC,EAAA,MAAM,YAAA,EAAc,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAI,CAAA;AAEzC,EAAA,GAAA,CAAI,YAAA,IAAgB,MAAA,EAAQ,OAAO,IAAA;AACnC,EAAA,GAAA,CAAI,SAAA,IAAa,GAAA,GAAM,SAAA,IAAa,EAAA,EAAI,OAAO,IAAA;AAC/C,EAAA,GAAA,CAAI,MAAA,GAAS,EAAA,GAAK,UAAA,IAAc,CAAA,EAAG,OAAO,IAAA;AAC1C,EAAA,GAAA,CAAI,MAAA,EAAQ,QAAA,GAAW,IAAA,CAAK,GAAA,CAAI,SAAS,EAAA,EAAI,OAAA,EAAS,OAAO,IAAA;AAE7D,EAAA,MAAM,QAAA,EAAU,UAAA,EAAY,CAAA;AAC5B,EAAA,MAAM,OAAA,EAAS,IAAA,CAAK,GAAA,CAAI,SAAS,CAAA;AAEjC,EAAA,GAAA,CAAI,MAAA,EAAQ,OAAA,EAAS,UAAA,EAAY,OAAO,IAAA;AACxC,EAAA,MAAM,cAAA,EAAgB,SAAA,GAAY,CAAA;AAElC,EAAA,MAAM,UAAA,EAAa,MAAA,EAAQ,cAAA,EAAgB,EAAA,EAAK,CAAC,CAAA;AACjD,EAAA,GAAA,CAAI,WAAA,EAAa,UAAA,EAAY,OAAA,EAAS,GAAA,CAAI,MAAA,EAAQ,OAAO,IAAA;AAEzD,EAAA,MAAM,KAAA,EAAO,IAAI,UAAA,CAAW,MAAA,EAAQ,OAAA,EAAS,CAAC,CAAA;AAC9C,EAAA,IAAI,SAAA,EAAW,CAAA;AAEf,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,MAAA,EAAQ,CAAA,EAAA,EAAK;AAE/B,IAAA,MAAM,OAAA,EAAS,QAAA,EAAU,EAAA,EAAI,OAAA,EAAS,EAAA,EAAI,CAAA;AAC1C,IAAA,IAAI,IAAA,EAAM,WAAA,EAAa,OAAA,EAAS,SAAA;AAChC,IAAA,IAAI,IAAA,EAAM,EAAA,EAAI,MAAA,EAAQ,CAAA;AACtB,IAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,KAAA,EAAO,CAAA,EAAA,EAAK;AAE9B,MAAA,IAAA,CAAK,GAAG,EAAA,EAAI,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AACvB,MAAA,IAAA,CAAK,IAAA,EAAM,CAAC,EAAA,EAAI,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AAC3B,MAAA,IAAA,CAAK,IAAA,EAAM,CAAC,EAAA,EAAI,GAAA,CAAI,GAAG,CAAA;AACvB,MAAA,MAAM,EAAA,EAAI,SAAA,IAAa,GAAA,EAAK,GAAA,CAAI,IAAA,EAAM,CAAC,EAAA,EAAI,GAAA;AAC3C,MAAA,IAAA,CAAK,IAAA,EAAM,CAAC,EAAA,EAAI,CAAA;AAChB,MAAA,SAAA,GAAY,CAAA;AACZ,MAAA,IAAA,GAAO,aAAA;AACP,MAAA,IAAA,GAAO,CAAA;AAAA,IACT;AAAA,EACF;AAIA,EAAA,GAAA,CAAI,SAAA,IAAa,GAAA,GAAM,SAAA,IAAa,CAAA,EAAG;AACrC,IAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,IAAA,CAAK,MAAA,EAAQ,EAAA,GAAK,CAAA,EAAG,IAAA,CAAK,CAAC,EAAA,EAAI,GAAA;AAAA,EACrD;AAEA,EAAA,OAAO,SAAA,CAAU,KAAA,EAAO,MAAA,EAAQ,IAAI,CAAA;AACtC;AAIA,IAAM,cAAA,EAAgB,UAAA,CAAW,EAAA,CAAG,GAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAA,EAAM,EAAI,CAAA;AAGlF,SAAS,KAAA,CAAM,IAAA,EAAc,IAAA,EAA8B;AACzD,EAAA,MAAM,KAAA,EAAO,IAAI,UAAA,CAAW,EAAA,EAAI,IAAA,CAAK,MAAM,CAAA;AAC3C,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,IAAA,CAAK,UAAA,CAAW,CAAC,CAAA;AAC3B,EAAA,IAAA,CAAK,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AAEhB,EAAA,MAAM,IAAA,EAAM,IAAI,UAAA,CAAW,EAAA,EAAI,IAAA,CAAK,OAAA,EAAS,CAAC,CAAA;AAC9C,EAAA,MAAM,GAAA,EAAK,IAAI,QAAA,CAAS,GAAA,CAAI,MAAM,CAAA;AAClC,EAAA,EAAA,CAAG,SAAA,CAAU,CAAA,EAAG,IAAA,CAAK,MAAA,EAAQ,KAAK,CAAA;AAClC,EAAA,GAAA,CAAI,GAAA,CAAI,IAAA,EAAM,CAAC,CAAA;AACf,EAAA,EAAA,CAAG,SAAA,CAAU,EAAA,EAAI,IAAA,CAAK,MAAA,EAAQ,KAAA,CAAM,IAAI,CAAA,EAAG,KAAK,CAAA;AAChD,EAAA,OAAO,GAAA;AACT;AAGO,SAAS,SAAA,CAAU,KAAA,EAAe,MAAA,EAAgB,IAAA,EAA8B;AAErF,EAAA,MAAM,KAAA,EAAO,IAAI,UAAA,CAAW,EAAE,CAAA;AAC9B,EAAA,MAAM,GAAA,EAAK,IAAI,QAAA,CAAS,IAAA,CAAK,MAAM,CAAA;AACnC,EAAA,EAAA,CAAG,SAAA,CAAU,CAAA,EAAG,KAAA,EAAO,KAAK,CAAA;AAC5B,EAAA,EAAA,CAAG,SAAA,CAAU,CAAA,EAAG,MAAA,EAAQ,KAAK,CAAA;AAC7B,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,CAAA;AACV,EAAA,IAAA,CAAK,CAAC,EAAA,EAAI,CAAA;AAGV,EAAA,MAAM,OAAA,EAAS,MAAA,EAAQ,CAAA;AACvB,EAAA,MAAM,IAAA,EAAM,IAAI,UAAA,CAAA,CAAY,OAAA,EAAS,CAAA,EAAA,EAAK,MAAM,CAAA;AAChD,EAAA,IAAA,CAAA,IAAS,EAAA,EAAI,CAAA,EAAG,EAAA,EAAI,MAAA,EAAQ,CAAA,EAAA,EAAK;AAC/B,IAAA,MAAM,SAAA,EAAW,EAAA,EAAA,CAAK,OAAA,EAAS,CAAA,CAAA;AAC/B,IAAA,GAAA,CAAI,QAAQ,EAAA,EAAI,CAAA;AAChB,IAAA,GAAA,CAAI,GAAA,CAAI,IAAA,CAAK,QAAA,CAAS,EAAA,EAAI,MAAA,EAAQ,EAAA,EAAI,OAAA,EAAS,MAAM,CAAA,EAAG,SAAA,EAAW,CAAC,CAAA;AAAA,EACtE;AACA,EAAA,MAAM,KAAA,EAAO,+BAAA,GAAe,CAAA;AAE5B,EAAA,MAAM,UAAA,EAAY,KAAA,CAAM,MAAA,EAAQ,IAAI,CAAA;AACpC,EAAA,MAAM,UAAA,EAAY,KAAA,CAAM,MAAA,EAAQ,IAAI,CAAA;AACpC,EAAA,MAAM,UAAA,EAAY,KAAA,CAAM,MAAA,EAAQ,IAAI,UAAA,CAAW,CAAC,CAAC,CAAA;AAEjD,EAAA,MAAM,IAAA,EAAM,IAAI,UAAA,CAAW,aAAA,CAAc,OAAA,EAAS,SAAA,CAAU,OAAA,EAAS,SAAA,CAAU,OAAA,EAAS,SAAA,CAAU,MAAM,CAAA;AACxG,EAAA,IAAI,EAAA,EAAI,CAAA;AACR,EAAA,GAAA,CAAI,GAAA,CAAI,aAAA,EAAe,CAAC,CAAA;AAAG,EAAA,EAAA,GAAK,aAAA,CAAc,MAAA;AAC9C,EAAA,GAAA,CAAI,GAAA,CAAI,SAAA,EAAW,CAAC,CAAA;AAAG,EAAA,EAAA,GAAK,SAAA,CAAU,MAAA;AACtC,EAAA,GAAA,CAAI,GAAA,CAAI,SAAA,EAAW,CAAC,CAAA;AAAG,EAAA,EAAA,GAAK,SAAA,CAAU,MAAA;AACtC,EAAA,GAAA,CAAI,GAAA,CAAI,SAAA,EAAW,CAAC,CAAA;AACpB,EAAA,OAAO,GAAA;AACT;AAqBO,SAAS,wBAAA,CACd,QAAA,EACA,MAAA,EACA,IAAA,EACQ;AACR,EAAA,MAAM,SAAA,kBAAW,IAAA,2BAAM,WAAA,IAAa,KAAA;AAGpC,EAAA,MAAM,KAAA,kBAAO,IAAI,GAAA,CAAoB,CAAA;AACrC,EAAA,IAAA,CAAA,MAAW,IAAA,GAAO,MAAA,EAAQ;AACxB,IAAA,IAAI,MAAA,EAAQ,GAAA,CAAI,IAAA;AAChB,IAAA,IAAI,KAAA,EAAO,GAAA,CAAI,QAAA;AACf,IAAA,GAAA,CAAI,SAAA,GAAY,KAAA,IAAS,WAAA,EAAa;AACpC,MAAA,MAAM,IAAA,EAAM,QAAA,CAAS,GAAA,CAAI,IAAI,CAAA;AAC7B,MAAA,GAAA,CAAI,GAAA,EAAK;AACP,QAAA,MAAA,EAAQ,GAAA;AACR,QAAA,KAAA,EAAO,WAAA;AAAA,MACT;AAAA,IACF;AACA,IAAA,IAAA,CAAK,GAAA,CAAI,GAAA,CAAI,QAAA,EAAU,CAAA,KAAA,EAAQ,IAAI,CAAA,QAAA,EAAW,MAAA,CAAO,IAAA,CAAK,KAAK,CAAA,CAAE,QAAA,CAAS,QAAQ,CAAC,CAAA,CAAA;AACrF,EAAA;AAC4B,EAAA;AAE8D,EAAA;AAC/D,IAAA;AACuB,IAAA;AACjD,EAAA;AACgG,EAAA;AACtE,IAAA;AAC0B,IAAA;AACpD,EAAA;AACM,EAAA;AACT;ADhFwF;AACA;AEnH+C;AAC3D,EAAA;AAClC,EAAA;AAEE,EAAA;AAC5B,EAAA;AACc,EAAA;AACgC,IAAA;AACzB,IAAA;AACJ,IAAA;AACG,IAAA;AAClC,EAAA;AAEiC,EAAA;AACF,EAAA;AACA,EAAA;AAEA,EAAA;AAC6B,EAAA;AAClB,IAAA;AACiC,IAAA;AAC3E,EAAA;AACO,EAAA;AACT;AFkHwF;AACA;AACA;AACA;AACA;AACA;AACA","file":"/Users/mong-e/workspace/kordoc/dist/chunk-CSWE6FQ4.cjs","sourcesContent":[null,"/**\n * 순수 JS BMP → PNG 트랜스코더 + 마크다운 이미지 인라이너.\n *\n * HWP5 BinData 임베드 이미지는 비압축 BMP(수 MB)인 경우가 많다. AI 에이전트가\n * 자체 완결형 마크다운을 받도록, BMP 를 PNG(무손실 압축)로 변환한 뒤 base64\n * data URI 로 마크다운에 인라인한다.\n *\n * 의존성 최소화 원칙 — Node 내장(node:zlib, Buffer)만 사용한다.\n */\n\nimport { deflateSync } from \"node:zlib\"\n\n// ─── CRC32 (PNG 청크용) ──────────────────────────────\n\nconst CRC_TABLE = (() => {\n const table = new Uint32Array(256)\n for (let n = 0; n < 256; n++) {\n let c = n\n for (let k = 0; k < 8; k++) {\n c = (c & 1) ? (0xedb88320 ^ (c >>> 1)) : (c >>> 1)\n }\n table[n] = c >>> 0\n }\n return table\n})()\n\n/** 표준 PNG CRC32 (다항식 0xEDB88320) */\nfunction crc32(bytes: Uint8Array): number {\n let c = 0xffffffff\n for (let i = 0; i < bytes.length; i++) {\n c = CRC_TABLE[(c ^ bytes[i]) & 0xff] ^ (c >>> 8)\n }\n return (c ^ 0xffffffff) >>> 0\n}\n\n// ─── BMP 디코드 ──────────────────────────────────────\n\nconst BI_RGB = 0\n/** 폭·높이 상한 — 비정상 헤더에 의한 과대 할당 방지 */\nconst MAX_DIM = 0x7fff\n/**\n * 총 픽셀(W×H) 상한 — MAX_DIM 은 각 변만 제한하므로 곱의 폭주(과대 할당 + 긴 deflateSync\n * 동기 블로킹)를 별도로 차단한다. raw 저장 BMP 는 상위 inflate 100MB 캡을 우회하므로 필요.\n * 36MP = A4 전면 600dpi 스캔(≈35MP)까지 허용 — 그 이상은 rgba+raw 로 300MB 를 넘겨\n * MCP 서버 이벤트루프를 수 초 세우므로 원본 바이트 폴백으로 돌린다.\n */\nconst MAX_PIXELS = 36_000_000\n\n/**\n * 비압축 BMP(24/32bpp, BI_RGB) → PNG(8bit RGBA).\n *\n * 지원: BITMAPFILEHEADER + BITMAPINFOHEADER(biSize>=40), biBitCount 24/32,\n * biCompression BI_RGB, bottom-up(양수 biHeight)/top-down(음수), 24bpp 4바이트 행 패딩.\n * 미지원 변형(팔레트/8bpp·RLE·BITFIELDS·OS/2 헤더 등)은 null 을 반환하여\n * 호출부가 원본 바이트로 폴백하게 한다.\n */\nexport function bmpToPng(bmp: Uint8Array): Uint8Array | null {\n // BITMAPFILEHEADER(14) + BITMAPINFOHEADER(40)\n if (bmp.length < 54) return null\n if (bmp[0] !== 0x42 || bmp[1] !== 0x4d) return null // \"BM\"\n\n const dv = new DataView(bmp.buffer, bmp.byteOffset, bmp.byteLength)\n const dataOffset = dv.getUint32(10, true)\n const headerSize = dv.getUint32(14, true)\n if (headerSize < 40) return null // OS/2 BITMAPCOREHEADER 등 미지원\n\n const width = dv.getInt32(18, true)\n const rawHeight = dv.getInt32(22, true)\n const bitCount = dv.getUint16(28, true)\n const compression = dv.getUint32(30, true)\n\n if (compression !== BI_RGB) return null // RLE / BITFIELDS 미지원\n if (bitCount !== 24 && bitCount !== 32) return null // 팔레트 / 8bpp 등 미지원\n if (width <= 0 || rawHeight === 0) return null\n if (width > MAX_DIM || Math.abs(rawHeight) > MAX_DIM) return null\n\n const topDown = rawHeight < 0\n const height = Math.abs(rawHeight)\n // 총 픽셀 상한 검사 — rgba/raw 할당·deflate 전에 조기 차단 (각 변은 MAX_DIM 이하여도 곱은 폭주 가능)\n if (width * height > MAX_PIXELS) return null\n const bytesPerPixel = bitCount >> 3\n // 행 스트라이드 — 4바이트 경계 패딩 (24bpp 에서 유효)\n const rowStride = (width * bytesPerPixel + 3) & ~3\n if (dataOffset + rowStride * height > bmp.length) return null // 잘린 파일\n\n const rgba = new Uint8Array(width * height * 4)\n let anyAlpha = 0\n\n for (let y = 0; y < height; y++) {\n // bottom-up 은 첫 행이 이미지 맨 아래 → 뒤집어 읽는다\n const srcRow = topDown ? y : height - 1 - y\n let src = dataOffset + srcRow * rowStride\n let dst = y * width * 4\n for (let x = 0; x < width; x++) {\n // BMP 픽셀은 BGR(A) 순\n rgba[dst] = bmp[src + 2] // R\n rgba[dst + 1] = bmp[src + 1] // G\n rgba[dst + 2] = bmp[src] // B\n const a = bitCount === 32 ? bmp[src + 3] : 255\n rgba[dst + 3] = a\n anyAlpha |= a\n src += bytesPerPixel\n dst += 4\n }\n }\n\n // 32bpp BI_RGB 의 4번째 바이트는 \"미정의\" — 전부 0 이면 전면 투명으로 오해되므로\n // 불투명(255)으로 승격한다. 실제 알파가 있으면(anyAlpha!=0) 그대로 보존.\n if (bitCount === 32 && anyAlpha === 0) {\n for (let i = 3; i < rgba.length; i += 4) rgba[i] = 255\n }\n\n return encodePng(width, height, rgba)\n}\n\n// ─── PNG 인코드 (8bit RGBA, color type 6) ────────────\n\nconst PNG_SIGNATURE = Uint8Array.of(0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a)\n\n/** [4B 길이][4B 타입][데이터][4B CRC] 청크 조립. CRC 는 (타입+데이터) 기준. */\nfunction chunk(type: string, data: Uint8Array): Uint8Array {\n const body = new Uint8Array(4 + data.length)\n body[0] = type.charCodeAt(0)\n body[1] = type.charCodeAt(1)\n body[2] = type.charCodeAt(2)\n body[3] = type.charCodeAt(3)\n body.set(data, 4)\n\n const out = new Uint8Array(8 + data.length + 4)\n const dv = new DataView(out.buffer)\n dv.setUint32(0, data.length, false) // 길이(데이터만, big-endian)\n out.set(body, 4)\n dv.setUint32(8 + data.length, crc32(body), false)\n return out\n}\n\n/** 8bit RGBA 픽셀 버퍼 → PNG (PDF 이미지 XObject 추출에서도 사용) */\nexport function encodePng(width: number, height: number, rgba: Uint8Array): Uint8Array {\n // IHDR: width, height, bit depth 8, color type 6(RGBA), compression 0, filter 0, interlace 0\n const ihdr = new Uint8Array(13)\n const iv = new DataView(ihdr.buffer)\n iv.setUint32(0, width, false)\n iv.setUint32(4, height, false)\n ihdr[8] = 8\n ihdr[9] = 6\n\n // IDAT: 각 스캔라인 앞에 필터 바이트 0(None) 을 붙인 raw 데이터를 zlib deflate\n const stride = width * 4\n const raw = new Uint8Array((stride + 1) * height)\n for (let y = 0; y < height; y++) {\n const rowStart = y * (stride + 1)\n raw[rowStart] = 0 // filter: None\n raw.set(rgba.subarray(y * stride, y * stride + stride), rowStart + 1)\n }\n const idat = deflateSync(raw)\n\n const ihdrChunk = chunk(\"IHDR\", ihdr)\n const idatChunk = chunk(\"IDAT\", idat)\n const iendChunk = chunk(\"IEND\", new Uint8Array(0))\n\n const out = new Uint8Array(PNG_SIGNATURE.length + ihdrChunk.length + idatChunk.length + iendChunk.length)\n let o = 0\n out.set(PNG_SIGNATURE, o); o += PNG_SIGNATURE.length\n out.set(ihdrChunk, o); o += ihdrChunk.length\n out.set(idatChunk, o); o += idatChunk.length\n out.set(iendChunk, o)\n return out\n}\n\n// ─── 마크다운 이미지 인라이너 ────────────────────────\n\ninterface InlineImage {\n filename: string\n data: Uint8Array\n mimeType: string\n}\n\n/**\n * 마크다운의 이미지 참조를 base64 data URI 로 치환해 자체 완결형 마크다운을 만든다.\n * 두 방출 형태를 모두 커버한다:\n * - `` / `` — 일반 문단 (CLI 접두사 포함)\n * - `<img src=\"FILENAME\">` — 병합/중첩 표 셀 (table/builder.ts HTML 표 경로).\n * 이걸 빼면 CLI `--inline-images` 가 이미지 저장을 생략하면서 표 셀 이미지만\n * dangling 참조로 유실된다.\n *\n * @param opts.compress 기본 true. `image/bmp` 는 PNG 로 트랜스코딩 후 인라인하며,\n * 트랜스코딩 실패 시 원본 바이트+원본 MIME 으로 폴백한다.\n */\nexport function inlineImagesIntoMarkdown(\n markdown: string,\n images: ReadonlyArray<InlineImage>,\n opts?: { compress?: boolean },\n): string {\n const compress = opts?.compress !== false\n // 파일명 → data URI 사전 구축 후 단일 패스 치환 — 이미지 k개마다 전문 재스캔하던\n // O(k²) 제거. 맵에 없는 파일명 참조는 원문 그대로 둔다 (기존 계약 동일).\n const uris = new Map<string, string>()\n for (const img of images) {\n let bytes = img.data\n let mime = img.mimeType\n if (compress && mime === \"image/bmp\") {\n const png = bmpToPng(img.data)\n if (png) {\n bytes = png\n mime = \"image/png\"\n }\n }\n uris.set(img.filename, `data:${mime};base64,${Buffer.from(bytes).toString(\"base64\")}`)\n }\n if (uris.size === 0) return markdown\n // FILENAME 과 images/FILENAME 접두사를 모두 정확한 파일명 기준으로 치환\n let out = markdown.replace(/!\\[image\\]\\((?:images\\/)?([^()\\s]+)\\)/g, (m, name: string) => {\n const uri = uris.get(name)\n return uri !== undefined ? `` : m\n })\n out = out.replace(/(<img\\b[^>]*\\bsrc=\")(?:images\\/)?([^\"]*)(\")/g, (m, pre: string, name: string, post: string) => {\n const uri = uris.get(name)\n return uri !== undefined ? `${pre}${uri}${post}` : m\n })\n return out\n}\n","/**\n * 페이지별 마크다운 사영 (#68)\n *\n * 문서를 다시 파싱하지 않고 이미 만들어진 `IRBlock.pageNumber` 로만 블록을 갈라\n * 페이지마다 `blocksToMarkdown()` 을 돌린다. 페이지 번호의 신뢰도는 파서가\n * 정하는 것이고(`metadata.pageMode` — \"layout\" 은 실제 페이지, \"section\" 은\n * 섹션 근사), 여기서는 그 값을 그대로 존중한다.\n */\n\nimport type { IRBlock, PageMarkdown } from \"./types.js\"\nimport { blocksToMarkdown } from \"./table/builder.js\"\n\n/**\n * 블록을 페이지 번호로 묶어 페이지별 마크다운을 만든다.\n *\n * - 어떤 블록에도 페이지 번호가 없으면 `undefined` — 페이지 개념이 없는 포맷에\n * \"전부 1페이지\" 같은 없는 사실을 만들어 내지 않는다.\n * - 번호가 없는 블록은 **직전 블록의 페이지로 이어붙인다.** 블록은 읽기 순서라\n * 이게 원문 위치에 가장 가깝고, 무엇보다 어느 페이지에도 안 실려 조용히\n * 사라지는 블록이 생기지 않는다. 선두의 번호 없는 블록은 첫 실번호 페이지로.\n * - 관측된 최소~최대 페이지 사이의 빈 페이지도 항목으로 낸다. 여러 페이지에\n * 걸친 표는 시작 페이지 한 블록이라 중간 페이지가 실제로 비어 있고, 이때\n * 항목을 빼면 소비자가 배열 길이로 페이지 수를 셀 수 없다.\n * - `render` 는 쪽 블록 → 마크다운. 문서 마크다운에 포맷 고유 마무리가 붙는 파서(PDF: 쪽번호 제거·균등배분 정리)는\n * 같은 마무리를 넘겨 쪽 마크다운이 문서 마크다운과 어긋나지 않게 한다.\n */\nexport function blocksToPages(blocks: IRBlock[], render: (blocks: IRBlock[]) => string = blocksToMarkdown): PageMarkdown[] | undefined {\n const firstNumbered = blocks.find(b => typeof b.pageNumber === \"number\")?.pageNumber\n if (firstNumbered === undefined) return undefined\n\n const byPage = new Map<number, IRBlock[]>()\n let current = firstNumbered\n for (const block of blocks) {\n if (typeof block.pageNumber === \"number\") current = block.pageNumber\n const bucket = byPage.get(current)\n if (bucket) bucket.push(block)\n else byPage.set(current, [block])\n }\n\n const numbers = [...byPage.keys()]\n const min = Math.min(...numbers)\n const max = Math.max(...numbers)\n\n const pages: PageMarkdown[] = []\n for (let pageNumber = min; pageNumber <= max; pageNumber++) {\n const pageBlocks = byPage.get(pageNumber)\n pages.push({ pageNumber, markdown: pageBlocks ? render(pageBlocks) : \"\" })\n }\n return pages\n}\n"]}
|
|
@@ -5,15 +5,15 @@
|
|
|
5
5
|
|
|
6
6
|
|
|
7
7
|
|
|
8
|
-
var
|
|
8
|
+
var _chunkOJB7QZHPcjs = require('./chunk-OJB7QZHP.cjs');
|
|
9
9
|
|
|
10
10
|
|
|
11
11
|
|
|
12
|
-
var
|
|
12
|
+
var _chunkGKZPJDLNcjs = require('./chunk-GKZPJDLN.cjs');
|
|
13
13
|
|
|
14
14
|
|
|
15
15
|
|
|
16
|
-
var
|
|
16
|
+
var _chunkPUHS26TXcjs = require('./chunk-PUHS26TX.cjs');
|
|
17
17
|
|
|
18
18
|
// src/ocr/models.ts
|
|
19
19
|
var _path = require('path');
|
|
@@ -65,7 +65,7 @@ var OCR_REC_DICT = {
|
|
|
65
65
|
sizeMb: 1
|
|
66
66
|
};
|
|
67
67
|
function getOcrModelsDir() {
|
|
68
|
-
return
|
|
68
|
+
return _chunkGKZPJDLNcjs.getModelsDir.call(void 0, "ppocr");
|
|
69
69
|
}
|
|
70
70
|
var DEFAULT_OCR_MODELS = Object.freeze({
|
|
71
71
|
det: { spec: OCR_DET_MODEL },
|
|
@@ -107,7 +107,7 @@ async function ensureOcrModelsFor(sel, onProgress) {
|
|
|
107
107
|
if (!stamp || verifiedInProcess.get(path) !== stamp) todo.push(s.spec);
|
|
108
108
|
}
|
|
109
109
|
if (!todo.length) return;
|
|
110
|
-
await
|
|
110
|
+
await _chunkGKZPJDLNcjs.ensureModelsIn.call(void 0, dir, todo, onProgress);
|
|
111
111
|
for (const spec of todo) {
|
|
112
112
|
const path = _path.join.call(void 0, dir, spec.filename), stamp = await fileStamp(path);
|
|
113
113
|
if (stamp) verifiedInProcess.set(path, stamp);
|
|
@@ -3154,16 +3154,16 @@ function recordOpenEndCellEdges(grid, cells, horizontals) {
|
|
|
3154
3154
|
}
|
|
3155
3155
|
function takeClipCellEdges(from, to) {
|
|
3156
3156
|
const e = CLIP_CELL_EDGES.get(from);
|
|
3157
|
-
if (e)
|
|
3157
|
+
if (e) _chunkOJB7QZHPcjs.CELL_EDGES.set(to, e);
|
|
3158
3158
|
}
|
|
3159
3159
|
function hasOpenCellSeam(a, b) {
|
|
3160
|
-
const u =
|
|
3160
|
+
const u = _chunkOJB7QZHPcjs.CELL_EDGES.get(a), d = _chunkOJB7QZHPcjs.CELL_EDGES.get(b);
|
|
3161
3161
|
return !!u && !!d && u.t && !u.b && !d.t && d.b && u.l && u.r && d.l && d.r;
|
|
3162
3162
|
}
|
|
3163
3163
|
function joinCellEdges(a, b) {
|
|
3164
|
-
const u =
|
|
3164
|
+
const u = _chunkOJB7QZHPcjs.CELL_EDGES.get(a), d = _chunkOJB7QZHPcjs.CELL_EDGES.get(b);
|
|
3165
3165
|
if (!u || !d) return;
|
|
3166
|
-
|
|
3166
|
+
_chunkOJB7QZHPcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
|
|
3167
3167
|
}
|
|
3168
3168
|
|
|
3169
3169
|
// src/pdf/cell-extract.ts
|
|
@@ -3339,7 +3339,7 @@ function tagScripts(text, lines) {
|
|
|
3339
3339
|
const [s, e, kind] = marks[m];
|
|
3340
3340
|
out = out.slice(0, s) + `<${kind}>` + out.slice(s, e) + `</${kind}>` + out.slice(e);
|
|
3341
3341
|
}
|
|
3342
|
-
return
|
|
3342
|
+
return _chunkOJB7QZHPcjs.tidyScriptTags.call(void 0, out);
|
|
3343
3343
|
}
|
|
3344
3344
|
|
|
3345
3345
|
// src/pdf/line-wrap.ts
|
|
@@ -6598,9 +6598,9 @@ function detectHeadings(blocks, medianFontSize) {
|
|
|
6598
6598
|
if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
|
|
6599
6599
|
const ratio = block.style.fontSize / medianFontSize;
|
|
6600
6600
|
let level = 0;
|
|
6601
|
-
if (ratio >=
|
|
6602
|
-
else if (ratio >=
|
|
6603
|
-
else if (ratio >=
|
|
6601
|
+
if (ratio >= _chunkOJB7QZHPcjs.HEADING_RATIO_H1) level = 1;
|
|
6602
|
+
else if (ratio >= _chunkOJB7QZHPcjs.HEADING_RATIO_H2) level = 2;
|
|
6603
|
+
else if (ratio >= _chunkOJB7QZHPcjs.HEADING_RATIO_H3) level = 3;
|
|
6604
6604
|
if (level > 0) {
|
|
6605
6605
|
block.type = "heading";
|
|
6606
6606
|
block.level = level;
|
|
@@ -6613,7 +6613,7 @@ function isChapterNumber(block, next, medianFontSize) {
|
|
|
6613
6613
|
const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _174 => _174.style, 'optionalAccess', _175 => _175.fontSize]), () => ( 0));
|
|
6614
6614
|
const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _176 => _176.text, 'optionalAccess', _177 => _177.trim, 'call', _178 => _178()]), () => ( ""));
|
|
6615
6615
|
const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _179 => _179.bbox]);
|
|
6616
|
-
return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _180 => _180.text, 'optionalAccess', _181 => _181.trim, 'call', _182 => _182()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize *
|
|
6616
|
+
return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _180 => _180.text, 'optionalAccess', _181 => _181.trim, 'call', _182 => _182()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkOJB7QZHPcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
|
|
6617
6617
|
}
|
|
6618
6618
|
function mergeStackedHeadingLines(blocks, medianFontSize) {
|
|
6619
6619
|
for (let i = 0; i < blocks.length - 1; ) {
|
|
@@ -8151,16 +8151,16 @@ function isProseSpread(items) {
|
|
|
8151
8151
|
for (let i = 1; i < sorted.length; i++) {
|
|
8152
8152
|
gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
|
|
8153
8153
|
}
|
|
8154
|
-
const maxGap =
|
|
8154
|
+
const maxGap = _chunkPUHS26TXcjs.safeMax.call(void 0, gaps);
|
|
8155
8155
|
const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
|
|
8156
8156
|
if (items.length >= 4 && maxGap < 40 && avgLen < 5) return true;
|
|
8157
|
-
const minGap =
|
|
8157
|
+
const minGap = _chunkPUHS26TXcjs.safeMin.call(void 0, gaps);
|
|
8158
8158
|
return minGap > 0 && maxGap - minGap <= Math.max(2, maxGap * 0.15) && (items.filter((i) => /^\P{L}*\p{Ll}/u.test(i.text)).length * 2 >= items.length || items.some((i) => /\p{L}[,.;:]["”’)]?$/u.test(i.text)));
|
|
8159
8159
|
}
|
|
8160
8160
|
function detectColumns(yLines) {
|
|
8161
8161
|
const allItems = yLines.flat();
|
|
8162
8162
|
if (allItems.length === 0) return null;
|
|
8163
|
-
const pageWidth =
|
|
8163
|
+
const pageWidth = _chunkPUHS26TXcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkPUHS26TXcjs.safeMin.call(void 0, allItems.map((i) => i.x));
|
|
8164
8164
|
if (pageWidth < 100) return null;
|
|
8165
8165
|
let bigoLineIdx = -1;
|
|
8166
8166
|
for (let i = 0; i < yLines.length; i++) {
|
|
@@ -8558,13 +8558,13 @@ function partTex(group) {
|
|
|
8558
8558
|
const base = Math.min(...g.filter((i) => i.fontSize >= size * SCRIPT_SIZE_RATIO).map((i) => i.y));
|
|
8559
8559
|
let s = "", run = "";
|
|
8560
8560
|
const flush = () => {
|
|
8561
|
-
if (run) s +=
|
|
8561
|
+
if (run) s += _chunkOJB7QZHPcjs.toTex.call(void 0, run);
|
|
8562
8562
|
run = "";
|
|
8563
8563
|
};
|
|
8564
8564
|
g.forEach((it, k) => {
|
|
8565
8565
|
if (it.fontSize < size * SCRIPT_SIZE_RATIO && it.y - base >= size * SCRIPT_RISE_EM) {
|
|
8566
8566
|
flush();
|
|
8567
|
-
s += `^{${
|
|
8567
|
+
s += `^{${_chunkOJB7QZHPcjs.toTex.call(void 0, it.text)}}`;
|
|
8568
8568
|
return;
|
|
8569
8569
|
}
|
|
8570
8570
|
if (k && it.x - (g[k - 1].x + g[k - 1].w) > spaceGapThreshold(it.fontSize)) run += " ";
|
|
@@ -8736,7 +8736,7 @@ function rebuildUnitLine(items, grid) {
|
|
|
8736
8736
|
function prependUnitRow(cells, cols, line, usedItems) {
|
|
8737
8737
|
const text = cleanCellText(line.map((item) => item.text).join(""));
|
|
8738
8738
|
cells.unshift(Array.from({ length: cols }, (_, c) => ({ text: c === 0 ? text : "", colSpan: c === 0 ? cols : 1, rowSpan: 1 })));
|
|
8739
|
-
for (const cell of cells[0])
|
|
8739
|
+
for (const cell of cells[0]) _chunkOJB7QZHPcjs.CELL_EDGES.set(cell, NO_EDGES);
|
|
8740
8740
|
for (const item of line) usedItems.add(item);
|
|
8741
8741
|
}
|
|
8742
8742
|
function attachUnitRow(items, grid, cells, cols, usedItems) {
|
|
@@ -9318,7 +9318,7 @@ function recordFrameTitle(table, source, pageNum) {
|
|
|
9318
9318
|
const [label, spacer, title] = cells;
|
|
9319
9319
|
if (!/^(?:붙임|별첨|첨부|부록)\s*\d{1,3}$/.test(label.text.trim()) || spacer.text.trim() || !title.text.trim()) return;
|
|
9320
9320
|
if (cells.some((c) => {
|
|
9321
|
-
const e =
|
|
9321
|
+
const e = _chunkOJB7QZHPcjs.CELL_EDGES.get(c);
|
|
9322
9322
|
return !e || e.t || e.b || e.l || e.r;
|
|
9323
9323
|
})) return;
|
|
9324
9324
|
const xs = TABLE_COLXS.get(table), style = dominantStyle(source), fs = _nullishCoalesce(_optionalChain([style, 'optionalAccess', _373 => _373.fontSize]), () => ( 0));
|
|
@@ -9920,7 +9920,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
|
|
|
9920
9920
|
groupSizes.push(1);
|
|
9921
9921
|
}
|
|
9922
9922
|
const allY = remaining.map((i) => i.y);
|
|
9923
|
-
const pageH =
|
|
9923
|
+
const pageH = _chunkPUHS26TXcjs.safeMax.call(void 0, allY) - _chunkPUHS26TXcjs.safeMin.call(void 0, allY);
|
|
9924
9924
|
const gapThreshold = Math.max(15, pageH * 0.03);
|
|
9925
9925
|
const sides = frameColumnTextBands(remaining, spanning, gx);
|
|
9926
9926
|
const textBlocks = [];
|
|
@@ -9936,7 +9936,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
|
|
|
9936
9936
|
finalTextBlocks.push(...detectListBlocks(textBlocks));
|
|
9937
9937
|
} else {
|
|
9938
9938
|
const allY = remaining.map((i) => i.y);
|
|
9939
|
-
const pageH =
|
|
9939
|
+
const pageH = _chunkPUHS26TXcjs.safeMax.call(void 0, allY) - _chunkPUHS26TXcjs.safeMin.call(void 0, allY);
|
|
9940
9940
|
const gapThreshold = Math.max(15, pageH * 0.03);
|
|
9941
9941
|
const groups = xyCutOrder(remaining, gapThreshold);
|
|
9942
9942
|
const textBlocks = [];
|
|
@@ -10268,7 +10268,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
|
|
|
10268
10268
|
blocks.push(...columnTextToBlocks(tableText, pageNum, bbox, dominantStyle(items)));
|
|
10269
10269
|
} else {
|
|
10270
10270
|
const allY = items.map((i) => i.y);
|
|
10271
|
-
const pageHeight =
|
|
10271
|
+
const pageHeight = _chunkPUHS26TXcjs.safeMax.call(void 0, allY) - _chunkPUHS26TXcjs.safeMin.call(void 0, allY);
|
|
10272
10272
|
const gapThreshold = Math.max(15, pageHeight * 0.03);
|
|
10273
10273
|
const orderedGroups = proseCutX !== null ? splitTwoColumnProse(items, proseCutX) : xyCutOrder(items, gapThreshold, 0, void 0, minGap);
|
|
10274
10274
|
for (const group of orderedGroups) {
|
|
@@ -10510,4 +10510,4 @@ function rowParagraphs(block) {
|
|
|
10510
10510
|
|
|
10511
10511
|
|
|
10512
10512
|
exports.DEFAULT_OCR_MODELS = DEFAULT_OCR_MODELS; exports.resolveOcrModels = resolveOcrModels; exports.ocrSelectionKey = ocrSelectionKey; exports.ocrSourcePath = ocrSourcePath; exports.ensureOcrModelsFor = ensureOcrModelsFor; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.grayCrop = grayCrop; exports.inkStats = inkStats; exports.splitRowBands = splitRowBands; exports.leadingTriangle = leadingTriangle; exports.arrowGlyph = arrowGlyph; exports.leaderRuns = leaderRuns; exports.inkBounds = inkBounds; exports.bracketFeatures = bracketFeatures; exports.slantedPair = slantedPair; exports.romanStems = romanStems; exports.circledAt = circledAt; exports.quoteHead = quoteHead; exports.edgeTrim = edgeTrim; exports.tallInkCount = tallInkCount; exports.gapGlyphs = gapGlyphs; exports.ringBullet = ringBullet; exports.dashRatio = dashRatio; exports.leadingRing = leadingRing; exports.discAt = discAt; exports.serifOne = serifOne; exports.starRun = starRun; exports.bracketShape = bracketShape; exports.leadingBullet = leadingBullet; exports.rememberOcrTextGaps = rememberOcrTextGaps; exports.copyOcrTextGaps = copyOcrTextGaps; exports.findOcrTextGaps = findOcrTextGaps; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.NO_EDGES = NO_EDGES; exports.hasOpenCellSeam = hasOpenCellSeam; exports.joinCellEdges = joinCellEdges; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.computeBBox = computeBBox; exports.dominantStyle = dominantStyle; exports.normalizeItems = normalizeItems; exports.groupByY = groupByY; exports.mergeLineSimple = mergeLineSimple; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.splitPageBreakWraps = splitPageBreakWraps; exports.latinSoftWrap = latinSoftWrap; exports.TOC_BLOCKS = TOC_BLOCKS; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.mergeStackedHeadingLines = mergeStackedHeadingLines; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectDocumentStyleHeadings = detectDocumentStyleHeadings; exports.detectSiblingStyleHeadings = detectSiblingStyleHeadings; exports.detectRepeatedPageLabels = detectRepeatedPageLabels; exports.detectPageLeadHeadings = detectPageLeadHeadings; exports.refineDocumentStyleHeadings = refineDocumentStyleHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.runningLinePatterns = runningLinePatterns; exports.stripAbsorbedRunningLines = stripAbsorbedRunningLines; exports.CLIP_TABLES = CLIP_TABLES; exports.FRAME_TITLE_BLOCKS = FRAME_TITLE_BLOCKS; exports.TABLE_COLXS = TABLE_COLXS; exports.TABLE_ROWYS = TABLE_ROWYS; exports.TABLE_TAIL = TABLE_TAIL; exports.PART_COLXS = PART_COLXS; exports.EMPTY_PARTS = EMPTY_PARTS; exports.IMAGE_CELLS = IMAGE_CELLS; exports.FILLER_CELLS = FILLER_CELLS; exports.CONT_PARTS = CONT_PARTS; exports.CELL_LINES = CELL_LINES; exports.ROW_RULES = ROW_RULES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.restoreImageBullets = restoreImageBullets; exports.detectRightArrowRegions = detectRightArrowRegions; exports.removeSideTabs = removeSideTabs; exports.extractPageBlocksWithLines = extractPageBlocksWithLines; exports.UNRULED_REGION_TABLES = UNRULED_REGION_TABLES; exports.mergeOcrImageRegions = mergeOcrImageRegions;
|
|
10513
|
-
//# sourceMappingURL=chunk-
|
|
10513
|
+
//# sourceMappingURL=chunk-DY7IZPA5.cjs.map
|