kordoc 4.24.13 → 4.24.14
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/{-NBCZTPBD.js → -MG25ST6Y.js} +11 -11
- package/dist/{chunk-UVQOPGOA.js → chunk-3LFNTOL7.js} +14 -9
- package/dist/chunk-3LFNTOL7.js.map +1 -0
- package/dist/{chunk-BC4LVU4D.js → chunk-7VNYQHJV.js} +2 -2
- package/dist/{chunk-CKOCIQU6.js → chunk-AM6TECXD.js} +2 -2
- package/dist/{chunk-X2E5QCKO.js → chunk-AULD3QAB.js} +2 -2
- package/dist/{chunk-XGESPSWE.cjs → chunk-BUGXM2H5.cjs} +3 -3
- package/dist/{chunk-XGESPSWE.cjs.map → chunk-BUGXM2H5.cjs.map} +1 -1
- package/dist/{chunk-ZUMOEVMV.js → chunk-DIFJPJGN.js} +3 -3
- package/dist/{chunk-XWWQGSEP.js → chunk-ERYRF3EA.js} +2 -2
- package/dist/{chunk-QAGPSWGH.js → chunk-FGERVDQZ.js} +2 -2
- package/dist/{chunk-ZSCDH3NU.js → chunk-FUC3V2FR.js} +2 -2
- package/dist/{chunk-U2VAABMK.cjs → chunk-GV7ECKOK.cjs} +44 -27
- package/dist/chunk-GV7ECKOK.cjs.map +1 -0
- package/dist/{chunk-2H2VYPM2.cjs → chunk-HVILFVAT.cjs} +2 -2
- package/dist/{chunk-2H2VYPM2.cjs.map → chunk-HVILFVAT.cjs.map} +1 -1
- package/dist/{chunk-OVQWOOS3.js → chunk-J3OJWUON.js} +2 -2
- package/dist/{chunk-QVEAR6IR.js → chunk-LMQT2JFS.js} +22 -5
- package/dist/chunk-LMQT2JFS.js.map +1 -0
- package/dist/{chunk-37MTSMUF.js → chunk-LSFVBS2I.js} +2 -2
- package/dist/{chunk-IQGXY63Y.js → chunk-MACV3MZF.js} +23 -6
- package/dist/chunk-MACV3MZF.js.map +1 -0
- package/dist/{chunk-QW2H2HHS.js → chunk-MIJYCKJA.js} +115 -6
- package/dist/chunk-MIJYCKJA.js.map +1 -0
- package/dist/{chunk-DMPX4KLC.js → chunk-NIPFEMJJ.js} +2 -2
- package/dist/{chunk-WOVJRXIA.js → chunk-NVD24SLL.js} +2 -2
- package/dist/{chunk-A232MW5F.js → chunk-PVAJ4DBW.js} +2 -2
- package/dist/{chunk-BJBPBU5B.js → chunk-SSDHVKLM.js} +2 -2
- package/dist/{chunk-4I2CKMKU.js → chunk-TPZPICDM.js} +3 -3
- package/dist/{chunk-KTXDK5A6.js → chunk-UTLHDVC5.js} +2 -2
- package/dist/{chunk-PUHBCAXI.js → chunk-UUNCLJB7.js} +11 -11
- package/dist/{chunk-W52B6GOE.cjs → chunk-V26UPU2L.cjs} +3 -3
- package/dist/{chunk-W52B6GOE.cjs.map → chunk-V26UPU2L.cjs.map} +1 -1
- package/dist/{chunk-A4CYMI55.js → chunk-VBOFP27O.js} +2 -2
- package/dist/{chunk-FJ3IKKDV.js → chunk-WFOPOJHN.js} +115 -6
- package/dist/chunk-WFOPOJHN.js.map +1 -0
- package/dist/{chunk-VPKZB6Y7.cjs → chunk-ZBHS7KLW.cjs} +6 -6
- package/dist/{chunk-VPKZB6Y7.cjs.map → chunk-ZBHS7KLW.cjs.map} +1 -1
- package/dist/{chunk-WAWREGAC.js → chunk-ZIGWQ3IS.js} +2 -2
- package/dist/{chunk-QZ2H4ZTO.cjs → chunk-ZX66BXB5.cjs} +190 -81
- package/dist/chunk-ZX66BXB5.cjs.map +1 -0
- package/dist/chunks-JGSWNVLF.js +10 -0
- package/dist/cli.js +30 -30
- package/dist/{formula-M4FM364X.js → formula-4KFXQSEG.js} +4 -4
- package/dist/{formula-ZKZGV6JV.cjs → formula-N6GWFPHK.cjs} +9 -9
- package/dist/{formula-ZKZGV6JV.cjs.map → formula-N6GWFPHK.cjs.map} +1 -1
- package/dist/{formula-WFL6JRCR.js → formula-WAYIKS54.js} +3 -3
- package/dist/{image-ocr-2V3M3ZEA.cjs → image-ocr-GY7SZAHN.cjs} +17 -17
- package/dist/{image-ocr-2V3M3ZEA.cjs.map → image-ocr-GY7SZAHN.cjs.map} +1 -1
- package/dist/{image-ocr-YJISJPKE.js → image-ocr-P5S2QDKW.js} +6 -6
- package/dist/{image-ocr-47CBNIS5.js → image-ocr-QBCR7ZD6.js} +9 -9
- package/dist/index.cjs +368 -363
- package/dist/index.cjs.map +1 -1
- package/dist/index.js +14 -9
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +23 -23
- package/dist/{model-bundle-7W344HDI.js → model-bundle-QAM24SYG.js} +5 -5
- package/dist/{models-TBHMNI7M.js → models-YHC236MP.js} +5 -5
- package/dist/{parse-worker-v2-JUYUEXCY.js → parse-worker-v2-LP7NKQBP.js} +12 -12
- package/dist/{parser-EKQCZLML.js → parser-26YUZU55.js} +12 -12
- package/dist/{parser-RNJ4KSDF.js → parser-DFJZ3CCE.js} +9 -9
- package/dist/{parser-R44KTBZX.cjs → parser-JWU6HYZZ.cjs} +159 -159
- package/dist/{parser-R44KTBZX.cjs.map → parser-JWU6HYZZ.cjs.map} +1 -1
- package/dist/{pdf-ocr-EE5KGFVX.js → pdf-ocr-BO6ACAGL.js} +9 -9
- package/dist/{pdf-ocr-ZAF4P74R.js → pdf-ocr-EEWQFS7P.js} +6 -6
- package/dist/pdf-ocr-Z4PBLB75.cjs +35 -0
- package/dist/{pdf-ocr-74BOCXAT.cjs.map → pdf-ocr-Z4PBLB75.cjs.map} +1 -1
- package/dist/{profile-io-RU3RVIYV.js → profile-io-PQXYWUBL.js} +3 -3
- package/dist/{rasterize-UNCPM26D.js → rasterize-CS7WZUGF.js} +3 -3
- package/dist/{redact-doc-BFE4UZCY.js → redact-doc-PI4JNHV6.js} +11 -11
- package/dist/{render-3DXTLKIA.js → render-M2CIEOF4.js} +7 -7
- package/dist/seal-4755WLFB.js +10 -0
- package/dist/{setup-7POQAKDG.js → setup-HGPUR27M.js} +3 -3
- package/dist/visual-JWWGGTKQ.js +21 -0
- package/dist/{watch-37YCXJWW.js → watch-B5ISBY3N.js} +12 -12
- package/package.json +1 -1
- package/dist/chunk-FJ3IKKDV.js.map +0 -1
- package/dist/chunk-IQGXY63Y.js.map +0 -1
- package/dist/chunk-QVEAR6IR.js.map +0 -1
- package/dist/chunk-QW2H2HHS.js.map +0 -1
- package/dist/chunk-QZ2H4ZTO.cjs.map +0 -1
- package/dist/chunk-U2VAABMK.cjs.map +0 -1
- package/dist/chunk-UVQOPGOA.js.map +0 -1
- package/dist/chunks-IXFBIZPW.js +0 -10
- package/dist/pdf-ocr-74BOCXAT.cjs +0 -35
- package/dist/seal-3LBTETDG.js +0 -10
- package/dist/visual-7SVIGIJC.js +0 -21
- /package/dist/{-NBCZTPBD.js.map → -MG25ST6Y.js.map} +0 -0
- /package/dist/{chunk-BC4LVU4D.js.map → chunk-7VNYQHJV.js.map} +0 -0
- /package/dist/{chunk-CKOCIQU6.js.map → chunk-AM6TECXD.js.map} +0 -0
- /package/dist/{chunk-X2E5QCKO.js.map → chunk-AULD3QAB.js.map} +0 -0
- /package/dist/{chunk-ZUMOEVMV.js.map → chunk-DIFJPJGN.js.map} +0 -0
- /package/dist/{chunk-XWWQGSEP.js.map → chunk-ERYRF3EA.js.map} +0 -0
- /package/dist/{chunk-QAGPSWGH.js.map → chunk-FGERVDQZ.js.map} +0 -0
- /package/dist/{chunk-ZSCDH3NU.js.map → chunk-FUC3V2FR.js.map} +0 -0
- /package/dist/{chunk-OVQWOOS3.js.map → chunk-J3OJWUON.js.map} +0 -0
- /package/dist/{chunk-37MTSMUF.js.map → chunk-LSFVBS2I.js.map} +0 -0
- /package/dist/{chunk-DMPX4KLC.js.map → chunk-NIPFEMJJ.js.map} +0 -0
- /package/dist/{chunk-WOVJRXIA.js.map → chunk-NVD24SLL.js.map} +0 -0
- /package/dist/{chunk-A232MW5F.js.map → chunk-PVAJ4DBW.js.map} +0 -0
- /package/dist/{chunk-BJBPBU5B.js.map → chunk-SSDHVKLM.js.map} +0 -0
- /package/dist/{chunk-4I2CKMKU.js.map → chunk-TPZPICDM.js.map} +0 -0
- /package/dist/{chunk-KTXDK5A6.js.map → chunk-UTLHDVC5.js.map} +0 -0
- /package/dist/{chunk-PUHBCAXI.js.map → chunk-UUNCLJB7.js.map} +0 -0
- /package/dist/{chunk-A4CYMI55.js.map → chunk-VBOFP27O.js.map} +0 -0
- /package/dist/{chunk-WAWREGAC.js.map → chunk-ZIGWQ3IS.js.map} +0 -0
- /package/dist/{chunks-IXFBIZPW.js.map → chunks-JGSWNVLF.js.map} +0 -0
- /package/dist/{formula-M4FM364X.js.map → formula-4KFXQSEG.js.map} +0 -0
- /package/dist/{formula-WFL6JRCR.js.map → formula-WAYIKS54.js.map} +0 -0
- /package/dist/{image-ocr-YJISJPKE.js.map → image-ocr-P5S2QDKW.js.map} +0 -0
- /package/dist/{image-ocr-47CBNIS5.js.map → image-ocr-QBCR7ZD6.js.map} +0 -0
- /package/dist/{model-bundle-7W344HDI.js.map → model-bundle-QAM24SYG.js.map} +0 -0
- /package/dist/{models-TBHMNI7M.js.map → models-YHC236MP.js.map} +0 -0
- /package/dist/{parse-worker-v2-JUYUEXCY.js.map → parse-worker-v2-LP7NKQBP.js.map} +0 -0
- /package/dist/{parser-EKQCZLML.js.map → parser-26YUZU55.js.map} +0 -0
- /package/dist/{parser-RNJ4KSDF.js.map → parser-DFJZ3CCE.js.map} +0 -0
- /package/dist/{pdf-ocr-EE5KGFVX.js.map → pdf-ocr-BO6ACAGL.js.map} +0 -0
- /package/dist/{pdf-ocr-ZAF4P74R.js.map → pdf-ocr-EEWQFS7P.js.map} +0 -0
- /package/dist/{profile-io-RU3RVIYV.js.map → profile-io-PQXYWUBL.js.map} +0 -0
- /package/dist/{rasterize-UNCPM26D.js.map → rasterize-CS7WZUGF.js.map} +0 -0
- /package/dist/{redact-doc-BFE4UZCY.js.map → redact-doc-PI4JNHV6.js.map} +0 -0
- /package/dist/{render-3DXTLKIA.js.map → render-M2CIEOF4.js.map} +0 -0
- /package/dist/{seal-3LBTETDG.js.map → seal-4755WLFB.js.map} +0 -0
- /package/dist/{setup-7POQAKDG.js.map → setup-HGPUR27M.js.map} +0 -0
- /package/dist/{visual-7SVIGIJC.js.map → visual-JWWGGTKQ.js.map} +0 -0
- /package/dist/{watch-37YCXJWW.js.map → watch-B5ISBY3N.js.map} +0 -0
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import {
|
|
2
2
|
escapeHtml,
|
|
3
3
|
sanitizeHref
|
|
4
|
-
} from "./chunk-
|
|
4
|
+
} from "./chunk-J3OJWUON.js";
|
|
5
5
|
|
|
6
6
|
// src/script-tags.ts
|
|
7
7
|
var LITERAL_TAG = /<(?=\/?(?:u|sup|sub)>)/g;
|
|
@@ -1551,4 +1551,4 @@ export {
|
|
|
1551
1551
|
HEADING_RATIO_H3,
|
|
1552
1552
|
validateOcrOptions
|
|
1553
1553
|
};
|
|
1554
|
-
//# sourceMappingURL=chunk-
|
|
1554
|
+
//# sourceMappingURL=chunk-SSDHVKLM.js.map
|
|
@@ -1,12 +1,12 @@
|
|
|
1
1
|
#!/usr/bin/env node
|
|
2
2
|
import {
|
|
3
3
|
parse
|
|
4
|
-
} from "./chunk-
|
|
4
|
+
} from "./chunk-UUNCLJB7.js";
|
|
5
5
|
import {
|
|
6
6
|
classifyError,
|
|
7
7
|
sanitizeError,
|
|
8
8
|
toArrayBuffer
|
|
9
|
-
} from "./chunk-
|
|
9
|
+
} from "./chunk-ZIGWQ3IS.js";
|
|
10
10
|
|
|
11
11
|
// src/cli/worker-parse.ts
|
|
12
12
|
import { readFileSync, statSync } from "fs";
|
|
@@ -26,4 +26,4 @@ async function parseWorkerFile(absPath, options) {
|
|
|
26
26
|
export {
|
|
27
27
|
parseWorkerFile
|
|
28
28
|
};
|
|
29
|
-
//# sourceMappingURL=chunk-
|
|
29
|
+
//# sourceMappingURL=chunk-TPZPICDM.js.map
|
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
import {
|
|
3
3
|
KordocError,
|
|
4
4
|
OPTIONAL_DEP_INSTALL_HINT
|
|
5
|
-
} from "./chunk-
|
|
5
|
+
} from "./chunk-ZIGWQ3IS.js";
|
|
6
6
|
|
|
7
7
|
// src/render/rasterize.ts
|
|
8
8
|
var densityExponent = null;
|
|
@@ -88,4 +88,4 @@ export {
|
|
|
88
88
|
rasterizeSvg,
|
|
89
89
|
rasterizePageSvg
|
|
90
90
|
};
|
|
91
|
-
//# sourceMappingURL=chunk-
|
|
91
|
+
//# sourceMappingURL=chunk-UTLHDVC5.js.map
|
|
@@ -51,10 +51,10 @@ import {
|
|
|
51
51
|
resolveParaHeading,
|
|
52
52
|
simulateWrap,
|
|
53
53
|
takeTocLeaders
|
|
54
|
-
} from "./chunk-
|
|
54
|
+
} from "./chunk-3LFNTOL7.js";
|
|
55
55
|
import {
|
|
56
56
|
blocksToPages
|
|
57
|
-
} from "./chunk-
|
|
57
|
+
} from "./chunk-LSFVBS2I.js";
|
|
58
58
|
import {
|
|
59
59
|
bmpToPng,
|
|
60
60
|
inlineImagesIntoMarkdown
|
|
@@ -76,7 +76,7 @@ import {
|
|
|
76
76
|
tidyScriptTags,
|
|
77
77
|
unframeLayoutTables,
|
|
78
78
|
wrapScript
|
|
79
|
-
} from "./chunk-
|
|
79
|
+
} from "./chunk-AM6TECXD.js";
|
|
80
80
|
import {
|
|
81
81
|
MAX_COLS,
|
|
82
82
|
MAX_ROWS,
|
|
@@ -100,7 +100,7 @@ import {
|
|
|
100
100
|
noteTableHtml,
|
|
101
101
|
sanitizeText,
|
|
102
102
|
visibleText
|
|
103
|
-
} from "./chunk-
|
|
103
|
+
} from "./chunk-NIPFEMJJ.js";
|
|
104
104
|
import {
|
|
105
105
|
ANNOTATION_BLANK_RE,
|
|
106
106
|
ValueCursor,
|
|
@@ -133,7 +133,7 @@ import {
|
|
|
133
133
|
resolveUnmatched,
|
|
134
134
|
scanInlineSegments,
|
|
135
135
|
scanSectionXml
|
|
136
|
-
} from "./chunk-
|
|
136
|
+
} from "./chunk-DIFJPJGN.js";
|
|
137
137
|
import {
|
|
138
138
|
MAX_DECOMPRESS_SIZE,
|
|
139
139
|
MAX_XML_DEPTH,
|
|
@@ -150,7 +150,7 @@ import {
|
|
|
150
150
|
localName,
|
|
151
151
|
rawTextContent,
|
|
152
152
|
readZipEntry
|
|
153
|
-
} from "./chunk-
|
|
153
|
+
} from "./chunk-ERYRF3EA.js";
|
|
154
154
|
import {
|
|
155
155
|
KordocError,
|
|
156
156
|
VERSION,
|
|
@@ -167,7 +167,7 @@ import {
|
|
|
167
167
|
toArrayBuffer,
|
|
168
168
|
unescapeHtml,
|
|
169
169
|
unzipLimitBytes
|
|
170
|
-
} from "./chunk-
|
|
170
|
+
} from "./chunk-ZIGWQ3IS.js";
|
|
171
171
|
|
|
172
172
|
// src/table/visual.ts
|
|
173
173
|
import { readFile as readFile2 } from "fs/promises";
|
|
@@ -19351,7 +19351,7 @@ var MIN_SIDE = 24;
|
|
|
19351
19351
|
async function ocrEmbeddedImages(blocks, options, warnings) {
|
|
19352
19352
|
const ocr = options?.ocr;
|
|
19353
19353
|
if (ocr !== true && ocr !== "force" && typeof ocr !== "function") return false;
|
|
19354
|
-
const { parseImageDocument } = await import("./image-ocr-
|
|
19354
|
+
const { parseImageDocument } = await import("./image-ocr-QBCR7ZD6.js");
|
|
19355
19355
|
let inserted = false, unreadable = 0;
|
|
19356
19356
|
let failed = null;
|
|
19357
19357
|
const walk = async (list, depth) => {
|
|
@@ -34072,7 +34072,7 @@ function scriptsOff(r, off) {
|
|
|
34072
34072
|
}
|
|
34073
34073
|
async function parseImage(buffer, options) {
|
|
34074
34074
|
try {
|
|
34075
|
-
const { parseImageDocument } = await import("./image-ocr-
|
|
34075
|
+
const { parseImageDocument } = await import("./image-ocr-QBCR7ZD6.js");
|
|
34076
34076
|
const { blocks, warnings, lines } = await parseImageDocument(buffer, options);
|
|
34077
34077
|
stripScriptTags({ blocks });
|
|
34078
34078
|
return {
|
|
@@ -34136,7 +34136,7 @@ async function parseHwp(buffer, options) {
|
|
|
34136
34136
|
async function parsePdf(buffer, options) {
|
|
34137
34137
|
let parsePdfDocument;
|
|
34138
34138
|
try {
|
|
34139
|
-
const mod = await import("./parser-
|
|
34139
|
+
const mod = await import("./parser-26YUZU55.js");
|
|
34140
34140
|
parsePdfDocument = mod.parsePdfDocument;
|
|
34141
34141
|
} catch {
|
|
34142
34142
|
return {
|
|
@@ -34375,4 +34375,4 @@ export {
|
|
|
34375
34375
|
parseHwpml,
|
|
34376
34376
|
fillForm
|
|
34377
34377
|
};
|
|
34378
|
-
//# sourceMappingURL=chunk-
|
|
34378
|
+
//# sourceMappingURL=chunk-UUNCLJB7.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 _chunkZBHS7KLWcjs = require('./chunk-ZBHS7KLW.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 = _chunkZBHS7KLWcjs.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 = _chunkVPKZB6Y7cjs.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-V26UPU2L.cjs.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/chunk-W52B6GOE.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-W52B6GOE.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-V26UPU2L.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-V26UPU2L.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"]}
|
|
@@ -3,7 +3,7 @@ import {
|
|
|
3
3
|
ensureModelsIn,
|
|
4
4
|
getModelStatusIn,
|
|
5
5
|
getModelsDir
|
|
6
|
-
} from "./chunk-
|
|
6
|
+
} from "./chunk-7VNYQHJV.js";
|
|
7
7
|
|
|
8
8
|
// src/ocr/models.ts
|
|
9
9
|
import { join } from "path";
|
|
@@ -162,4 +162,4 @@ export {
|
|
|
162
162
|
parseCharacterDict,
|
|
163
163
|
ocrModelPath
|
|
164
164
|
};
|
|
165
|
-
//# sourceMappingURL=chunk-
|
|
165
|
+
//# sourceMappingURL=chunk-VBOFP27O.js.map
|
|
@@ -35,14 +35,14 @@ import {
|
|
|
35
35
|
splitRowBands,
|
|
36
36
|
starRun,
|
|
37
37
|
tallInkCount
|
|
38
|
-
} from "./chunk-
|
|
38
|
+
} from "./chunk-MACV3MZF.js";
|
|
39
39
|
import {
|
|
40
40
|
validateOcrOptions
|
|
41
|
-
} from "./chunk-
|
|
41
|
+
} from "./chunk-SSDHVKLM.js";
|
|
42
42
|
import {
|
|
43
43
|
KordocError,
|
|
44
44
|
OPTIONAL_DEP_INSTALL_HINT
|
|
45
|
-
} from "./chunk-
|
|
45
|
+
} from "./chunk-J3OJWUON.js";
|
|
46
46
|
|
|
47
47
|
// src/ocr/ruling-lines.ts
|
|
48
48
|
var INK_LUMA_MAX = 205;
|
|
@@ -392,6 +392,7 @@ function restoreGlyphs(rgba, pageW, box, text, steps, stepPx, note = {}) {
|
|
|
392
392
|
const gray = grayCrop(rgba, pageW, box);
|
|
393
393
|
const c = { chars, cx: (i) => (steps[i] + 0.5) * stepPx, gray, w: box.w, h: box.h, ink: inkStats(gray) };
|
|
394
394
|
if (/[[\]]/.test(text)) brackets(c);
|
|
395
|
+
if (text.includes("'")) cornerQuotes(c);
|
|
395
396
|
if (/['"]/.test(text)) quotes(c);
|
|
396
397
|
if (/(^|\D)[1-9](\D|$)/.test(text)) circled(c);
|
|
397
398
|
if (text.includes("*")) stars(c);
|
|
@@ -445,6 +446,19 @@ function cornerCommas({ chars, cx, gray, w, h, ink }) {
|
|
|
445
446
|
}
|
|
446
447
|
}
|
|
447
448
|
}
|
|
449
|
+
function cornerQuotes({ chars, cx, gray, w, h, ink }) {
|
|
450
|
+
for (let i = 0; i < chars.length; i++) {
|
|
451
|
+
if (chars[i] !== "'" || !/^[가-힣]/.test(chars[i + 1] ?? "")) continue;
|
|
452
|
+
const f = bracketFeatures(gray, w, h, ink, cx(i));
|
|
453
|
+
if (!f || f.ch < f.charH * 0.45 || bracketShape(f, false) !== "\u300C") continue;
|
|
454
|
+
const rest = chars.slice(i + 1), end = rest.findIndex((c) => c === "'" || c === "\u300D" || c === ",");
|
|
455
|
+
const pair = end >= 0 && rest[end] !== "'" && (rest[end] === "\u300D" || (() => {
|
|
456
|
+
const g = bracketFeatures(gray, w, h, ink, cx(i + 1 + end));
|
|
457
|
+
return !!g && bracketShape(g, true) === "\u300D";
|
|
458
|
+
})());
|
|
459
|
+
if (pair) chars[i] = "\u300C";
|
|
460
|
+
}
|
|
461
|
+
}
|
|
448
462
|
function brackets({ chars, cx, gray, w, h, ink }) {
|
|
449
463
|
for (let i = 0; i < chars.length; i++) {
|
|
450
464
|
if (chars[i] !== "[" && chars[i] !== "]") continue;
|
|
@@ -546,6 +560,14 @@ function gaps({ chars, cx, gray, w, h, ink }) {
|
|
|
546
560
|
if (loose.some((o) => o !== g && Math.abs((o.x0 + o.x1) / 2 - mid) <= pitch * 1.5)) continue;
|
|
547
561
|
if (/[\u2026.\u00b7]$/.test(prev) || /^[\u2026.\u00b7]/.test(next)) continue;
|
|
548
562
|
}
|
|
563
|
+
if (g.mark === "\u300C") {
|
|
564
|
+
let sp = chars.findIndex((c, i) => c === " " && near(i));
|
|
565
|
+
if (sp >= 0) while (chars[sp + 1] === " ") sp++;
|
|
566
|
+
if (sp >= 0) {
|
|
567
|
+
chars[sp] = g.mark;
|
|
568
|
+
continue;
|
|
569
|
+
}
|
|
570
|
+
}
|
|
549
571
|
if (g.mark === "\xB7") {
|
|
550
572
|
if (/^\d/.test(next)) continue;
|
|
551
573
|
const sp = chars.findIndex((c, i) => c === " " && near(i));
|
|
@@ -559,10 +581,23 @@ function gaps({ chars, cx, gray, w, h, ink }) {
|
|
|
559
581
|
}
|
|
560
582
|
function ring({ chars, cx, gray, w, h, ink }) {
|
|
561
583
|
const k = chars.findIndex((c) => c.trim());
|
|
562
|
-
if (k < 0 || !/^[\u3147Oo0\
|
|
584
|
+
if (k < 0 || !/^[\u3147Oo0\u25cbCc]$/.test(chars[k])) return false;
|
|
563
585
|
const rest = chars.slice(k + 1).join("").trimStart();
|
|
586
|
+
const live = chars.map((c, i) => i).filter((i) => chars[i].trim());
|
|
587
|
+
if (live.length <= 3 && live.every((i) => /^[Oo]$/.test(chars[i]))) {
|
|
588
|
+
if (leadingBullet(gray, w, h, ink, cx(live[0]))?.mark === "\u25CE") return false;
|
|
589
|
+
if (!live.every((i) => ringBullet(gray, w, h, ink, cx(i)) === "\u25CB")) return false;
|
|
590
|
+
for (const i of live) chars[i] = "\u25CB";
|
|
591
|
+
return true;
|
|
592
|
+
}
|
|
593
|
+
if (/^[A-Z]{2,}/.test(rest) && /^[Oo0]$/.test(chars[k])) {
|
|
594
|
+
if (ringBullet(gray, w, h, ink, cx(k)) !== "\u25CB") return false;
|
|
595
|
+
chars[k] = chars[k + 1] === " " ? "\u25CB" : "\u25CB ";
|
|
596
|
+
return true;
|
|
597
|
+
}
|
|
564
598
|
if (!/^[(\[「【<]?\d*[가-힣]/.test(rest)) return false;
|
|
565
599
|
const r = ringBullet(gray, w, h, ink, cx(k));
|
|
600
|
+
if (/^[Cc]$/.test(chars[k]) && r !== "\u25CB") return false;
|
|
566
601
|
if (r) chars[k] = r;
|
|
567
602
|
return r !== null;
|
|
568
603
|
}
|
|
@@ -869,6 +904,71 @@ var MIN_INK_RATIO = 0.01;
|
|
|
869
904
|
var MAX_LOCAL_MODEL_BYTES = 1024 * 1024 * 1024;
|
|
870
905
|
var DET_MEAN = [0.485, 0.456, 0.406];
|
|
871
906
|
var DET_STD = [0.229, 0.224, 0.225];
|
|
907
|
+
function footerInkBox(rgba, width, height, boxes) {
|
|
908
|
+
const y0 = Math.floor(height * 0.88), x0 = Math.floor(width * 0.3), x1 = Math.ceil(width * 0.7);
|
|
909
|
+
let minX = Infinity, maxX = -1, minY = Infinity, maxY = -1;
|
|
910
|
+
for (let y = y0; y < height; y++) for (let x = x0; x < x1; x++) {
|
|
911
|
+
const p = (y * width + x) * 4;
|
|
912
|
+
if (rgba[p] * 77 + rgba[p + 1] * 150 + rgba[p + 2] * 29 >> 8 >= 128) continue;
|
|
913
|
+
if (boxes.some((b) => x >= b.x && x < b.x + b.w && y >= b.y && y < b.y + b.h)) continue;
|
|
914
|
+
if (x < minX) minX = x;
|
|
915
|
+
if (x > maxX) maxX = x;
|
|
916
|
+
if (y < minY) minY = y;
|
|
917
|
+
if (y > maxY) maxY = y;
|
|
918
|
+
}
|
|
919
|
+
if (maxX < 0) return null;
|
|
920
|
+
const bh = maxY - minY + 1;
|
|
921
|
+
if (bh < height * 5e-3 || bh > height * 0.025 || maxX - minX + 1 > width * 0.2) return null;
|
|
922
|
+
const pad = 4, bx = Math.max(0, minX - pad), by = Math.max(0, minY - pad);
|
|
923
|
+
return { x: bx, y: by, w: Math.min(width, maxX + pad + 1) - bx, h: Math.min(height, maxY + pad + 1) - by };
|
|
924
|
+
}
|
|
925
|
+
function closingParens(rgba, width, items) {
|
|
926
|
+
const out = [];
|
|
927
|
+
const dark = (x, y) => {
|
|
928
|
+
const p = (y * width + x) * 4;
|
|
929
|
+
return rgba[p] * 77 + rgba[p + 1] * 150 + rgba[p + 2] * 29 >> 8 < 128;
|
|
930
|
+
};
|
|
931
|
+
for (const it of items) {
|
|
932
|
+
const t = it.text.replace(/\s+$/, "");
|
|
933
|
+
if ((t.match(/\(/g)?.length ?? 0) <= (t.match(/\)/g)?.length ?? 0) || !/\([^()]*$/.test(t)) continue;
|
|
934
|
+
const x0 = it.x + it.w + 2;
|
|
935
|
+
const next = items.filter((o) => o !== it && o.x >= x0 && o.y < it.y + it.h && o.y + o.h > it.y).reduce((m, o) => Math.min(m, o.x), Infinity);
|
|
936
|
+
const x1 = Math.min(width, next, x0 + it.h * 8);
|
|
937
|
+
let minX = Infinity, maxX = -1, minY = Infinity, maxY = -1, cols = 0, prevOn = false, runs = 0;
|
|
938
|
+
for (let x = x0; x < x1; x++) {
|
|
939
|
+
let on = false;
|
|
940
|
+
for (let y = it.y; y < it.y + it.h; y++) if (dark(x, y)) {
|
|
941
|
+
on = true;
|
|
942
|
+
if (y < minY) minY = y;
|
|
943
|
+
if (y > maxY) maxY = y;
|
|
944
|
+
}
|
|
945
|
+
if (on) {
|
|
946
|
+
cols++;
|
|
947
|
+
if (!prevOn) runs++;
|
|
948
|
+
if (x < minX) minX = x;
|
|
949
|
+
maxX = x;
|
|
950
|
+
}
|
|
951
|
+
prevOn = on;
|
|
952
|
+
}
|
|
953
|
+
if (runs !== 1 || maxX - minX + 1 > it.h * 0.5 || maxY - minY + 1 < it.h * 0.6) continue;
|
|
954
|
+
let rule = false;
|
|
955
|
+
for (let x = minX; x <= maxX && !rule; x++) rule = minY > 1 && dark(x, minY - 2) || dark(x, Math.min(maxY + 2, Math.floor(rgba.length / 4 / width) - 1));
|
|
956
|
+
if (rule) continue;
|
|
957
|
+
const meanX = (ya, yb) => {
|
|
958
|
+
let sx = 0, k = 0;
|
|
959
|
+
for (let y = ya; y < yb; y++) for (let x = minX; x <= maxX; x++) if (dark(x, y)) {
|
|
960
|
+
sx += x;
|
|
961
|
+
k++;
|
|
962
|
+
}
|
|
963
|
+
return k ? sx / k : NaN;
|
|
964
|
+
};
|
|
965
|
+
const ch = maxY - minY + 1, band = Math.max(1, Math.round(ch * 0.2));
|
|
966
|
+
const ends = (meanX(minY, minY + band) + meanX(maxY + 1 - band, maxY + 1)) / 2, mid = meanX(minY + Math.floor(ch * 0.4), minY + Math.ceil(ch * 0.6));
|
|
967
|
+
if (!(mid - ends >= 0.8)) continue;
|
|
968
|
+
out.push({ text: ")", x: minX, y: minY, w: maxX - minX + 1, h: maxY - minY + 1, confidence: it.confidence });
|
|
969
|
+
}
|
|
970
|
+
return out;
|
|
971
|
+
}
|
|
872
972
|
var OcrEngine = class _OcrEngine {
|
|
873
973
|
det;
|
|
874
974
|
rec;
|
|
@@ -958,6 +1058,7 @@ var OcrEngine = class _OcrEngine {
|
|
|
958
1058
|
const detected = await this.detect(rgba, width, height, tuning, stats);
|
|
959
1059
|
const rows = rowRules.length ? detected.flatMap((b) => splitBoxAtRowRules(b, rowRules)) : detected;
|
|
960
1060
|
const boxes = cellRules.length ? rows.flatMap((b) => splitBoxAtCellRules(b, cellRules)) : rows;
|
|
1061
|
+
const footer = tuning.postprocess ? footerInkBox(rgba, width, height, boxes) : null;
|
|
961
1062
|
const jobs = [];
|
|
962
1063
|
const joins = [];
|
|
963
1064
|
let group = 0;
|
|
@@ -1001,6 +1102,7 @@ var OcrEngine = class _OcrEngine {
|
|
|
1001
1102
|
if (keep) jobs.push({ box: b, rot: 0, group, keep });
|
|
1002
1103
|
jobs.push({ box: b, rot: 0, group: group++ });
|
|
1003
1104
|
}
|
|
1105
|
+
if (footer) jobs.push({ box: footer, rot: 0, group: group++, footer: true });
|
|
1004
1106
|
const results = await this.recognizeJobs(rgba, width, jobs, tuning.recBatch);
|
|
1005
1107
|
const best = /* @__PURE__ */ new Map();
|
|
1006
1108
|
const score = (job, confidence) => confidence - (job.keep ? KEEP_MARGIN : 0);
|
|
@@ -1018,7 +1120,7 @@ var OcrEngine = class _OcrEngine {
|
|
|
1018
1120
|
const raw = tuning.postprocess && job.rot === 0 ? restoreGlyphs(rgba, width, job.box, read, steps, stepPx, note) : read;
|
|
1019
1121
|
let text = tuning.postprocess ? restoreSymbols(raw.trim(), note.ringLead) : raw;
|
|
1020
1122
|
if (!text.trim()) continue;
|
|
1021
|
-
if (tuning.postprocess && job.rot === 0 && !/^[
|
|
1123
|
+
if (tuning.postprocess && job.rot === 0 && !/^[◎●▪□■▣❐❏○ㅇ◆◇◈]/.test(text)) {
|
|
1022
1124
|
const chars = [...read], k = chars.findIndex((c) => c.trim());
|
|
1023
1125
|
if (k >= 0 && steps.length === chars.length) {
|
|
1024
1126
|
const g = grayCrop(rgba, width, job.box);
|
|
@@ -1028,6 +1130,8 @@ var OcrEngine = class _OcrEngine {
|
|
|
1028
1130
|
if (b?.mark === "\u25CE" && b.covers && /^[Oo0○]/.test(text)) text = b.mark + " " + text.slice(1).trimStart();
|
|
1029
1131
|
else if (b?.mark === "\u25A1" || b?.mark === "\u25A3") {
|
|
1030
1132
|
if (!b.covers && body) text = b.mark + " " + text;
|
|
1133
|
+
} else if (b?.mark === "\u25C6" || b?.mark === "\u25C7" || b?.mark === "\u25C8") {
|
|
1134
|
+
if (b.covers ? /^[\p{S}\p{P}]/u.test(text) : body) text = b.covers ? b.mark + " " + text.slice(1).trimStart() : b.mark + " " + text;
|
|
1031
1135
|
} else if (b?.mark === "\u2023") {
|
|
1032
1136
|
if (b.covers ? text.startsWith("\u2192") : body) text = b.covers ? b.mark + " " + text.slice(1).trimStart() : b.mark + " " + text;
|
|
1033
1137
|
} else if (b?.mark === "\u2750" || b?.mark === "\u274F") {
|
|
@@ -1045,6 +1149,10 @@ var OcrEngine = class _OcrEngine {
|
|
|
1045
1149
|
const g = grayCrop(rgba, width, job.box);
|
|
1046
1150
|
text = arrowGlyph(g, job.box.w, job.box.h, inkStats(g)) ?? text;
|
|
1047
1151
|
}
|
|
1152
|
+
if (job.footer) {
|
|
1153
|
+
if (!/^[-\u2013\u2014]?\s*(?:\d{1,4}|[ivxIVXⅰ-ⅹⅠ-Ⅹ]{1,4})\s*[-\u2013\u2014]?$/.test(text.trim())) continue;
|
|
1154
|
+
text = text.replace(/[\u2013\u2014]/g, "-");
|
|
1155
|
+
}
|
|
1048
1156
|
if (tuning.postprocess && isDotFragment(text)) continue;
|
|
1049
1157
|
if (confidence < tuning.textScore) {
|
|
1050
1158
|
if (stats) {
|
|
@@ -1083,6 +1191,7 @@ var OcrEngine = class _OcrEngine {
|
|
|
1083
1191
|
if (leaderEnds.size) items = joinLeaderItems(items, leaderEnds);
|
|
1084
1192
|
if (tuning.postprocess) items = dropNestedFragments(items);
|
|
1085
1193
|
if (tuning.postprocess) for (const it of items) this.outerSquare(rgba, width, it, items);
|
|
1194
|
+
if (tuning.postprocess) items.push(...closingParens(rgba, width, items));
|
|
1086
1195
|
if (tuning.postprocess) restoreBulletItems(items);
|
|
1087
1196
|
items.sort((a, b) => a.y - b.y || a.x - b.x);
|
|
1088
1197
|
return items;
|
|
@@ -1752,4 +1861,4 @@ export {
|
|
|
1752
1861
|
groupOcrLines,
|
|
1753
1862
|
ocrItemsToBlocks
|
|
1754
1863
|
};
|
|
1755
|
-
//# sourceMappingURL=chunk-
|
|
1864
|
+
//# sourceMappingURL=chunk-WFOPOJHN.js.map
|