kordoc 4.12.0 → 4.12.2
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 +26 -5
- package/dist/{-B3EUUBPV.js → -EZH6NKTE.js} +8 -8
- package/dist/{chunk-HGUNWAXK.js → chunk-2B7OIEJJ.js} +9 -5
- package/dist/chunk-2B7OIEJJ.js.map +1 -0
- package/dist/{chunk-QKFCGJJY.js → chunk-2HBYWA5X.js} +2 -2
- package/dist/{chunk-HFJWCAJ2.cjs → chunk-2LKYJ4G3.cjs} +6 -6
- package/dist/chunk-2LKYJ4G3.cjs.map +1 -0
- package/dist/{chunk-COGFPMEB.js → chunk-2LO525TL.js} +2 -2
- package/dist/{chunk-TROD36DK.cjs → chunk-2LPBIW2B.cjs} +3 -3
- package/dist/chunk-2LPBIW2B.cjs.map +1 -0
- package/dist/{chunk-BQSWSMRG.cjs → chunk-62IU4NV3.cjs} +2 -2
- package/dist/chunk-62IU4NV3.cjs.map +1 -0
- package/dist/{chunk-SMGSCP57.js → chunk-7XUXSDH5.js} +308 -53
- package/dist/chunk-7XUXSDH5.js.map +1 -0
- package/dist/{chunk-YASFVKDT.js → chunk-DBGEUDBY.js} +2 -2
- package/dist/chunk-DCZVOIEO.cjs.map +1 -1
- package/dist/{chunk-4PKNWBXJ.js → chunk-FBLMVHDK.js} +2 -2
- package/dist/{chunk-I2OETEEL.js → chunk-FVBR7SC3.js} +2 -2
- package/dist/chunk-GS7T56RP.cjs.map +1 -1
- package/dist/{chunk-46UDI56N.js → chunk-IOJX4Q63.js} +3 -3
- package/dist/{chunk-BQE7AMFS.js → chunk-JCMBQC5N.js} +2 -2
- package/dist/{chunk-TF5DS74F.js → chunk-KX7IYCCZ.js} +3 -3
- package/dist/{chunk-PFUW4I4I.js → chunk-LD2K4K5V.js} +170 -18
- package/dist/chunk-LD2K4K5V.js.map +1 -0
- package/dist/{chunk-MJPRWCI5.js → chunk-MUYU4JI5.js} +2 -2
- package/dist/{chunk-SKGYPM6L.js → chunk-OEYECZGV.js} +308 -53
- package/dist/chunk-OEYECZGV.js.map +1 -0
- package/dist/{chunk-AT23GFSK.cjs → chunk-ONNAHSPV.cjs} +323 -68
- package/dist/chunk-ONNAHSPV.cjs.map +1 -0
- package/dist/{chunk-GHVXLV52.js → chunk-Q6ETEJFK.js} +9 -5
- package/dist/chunk-Q6ETEJFK.js.map +1 -0
- package/dist/{chunk-2IR3PEZU.js → chunk-QBCMFNHS.js} +2 -2
- package/dist/{chunk-DRLKMFDV.js → chunk-QM2RSJ2S.js} +3 -3
- package/dist/chunk-T7GRHAIW.cjs.map +1 -1
- package/dist/{chunk-TJYZY2O6.js → chunk-UL2MDA5C.js} +3 -3
- package/dist/{chunk-TYYJSOZS.cjs → chunk-V5BUPZ6I.cjs} +23 -19
- package/dist/chunk-V5BUPZ6I.cjs.map +1 -0
- package/dist/{chunk-7JQ2ITT2.js → chunk-XXDZ2KTW.js} +2 -2
- package/dist/chunks-TT7ZE5QQ.js +10 -0
- package/dist/cli.js +21 -21
- package/dist/cli.js.map +1 -1
- package/dist/{formula-WGKKR5OQ.js → formula-6RQX4NMJ.js} +4 -4
- package/dist/{formula-YMGXXRBQ.cjs → formula-NBEO6CFC.cjs} +9 -9
- package/dist/formula-NBEO6CFC.cjs.map +1 -0
- package/dist/{formula-G4CD5KQ6.js → formula-NW33S4TA.js} +3 -3
- package/dist/{image-ocr-MBWZK3ZT.js → image-ocr-CAVU2TNZ.js} +5 -5
- package/dist/{image-ocr-M4RP4INO.js → image-ocr-SO56VETS.js} +7 -7
- package/dist/{image-ocr-BE2HHMVN.cjs → image-ocr-V4KY373B.cjs} +11 -11
- package/dist/image-ocr-V4KY373B.cjs.map +1 -0
- package/dist/index.cjs +547 -395
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.cts +5 -2
- package/dist/index.d.ts +5 -2
- package/dist/index.js +166 -14
- package/dist/index.js.map +1 -1
- package/dist/mcp.js +15 -15
- package/dist/{model-bundle-BIMP5DIH.js → model-bundle-EIJF34DR.js} +5 -5
- package/dist/{models-QPOEGHVA.js → models-6V4CPQXC.js} +5 -5
- package/dist/page-range-P7SDW6LR.cjs.map +1 -1
- package/dist/{parser-CMNHBVTL.js → parser-GIREMEPD.js} +307 -6
- package/dist/parser-GIREMEPD.js.map +1 -0
- package/dist/{parser-NR3XO4SM.js → parser-OBHNXEQM.js} +307 -6
- package/dist/parser-OBHNXEQM.js.map +1 -0
- package/dist/{parser-QJY5W47V.cjs → parser-OMT25WH2.cjs} +334 -33
- package/dist/parser-OMT25WH2.cjs.map +1 -0
- package/dist/pdf-ocr-IRXQ5GBM.js +13 -0
- package/dist/pdf-ocr-SL4GCOGS.js +16 -0
- package/dist/pdf-ocr-TFYVK4FI.cjs +14 -0
- package/dist/pdf-ocr-TFYVK4FI.cjs.map +1 -0
- package/dist/{profile-io-UZ43CLWX.js → profile-io-PZVMM66V.js} +3 -3
- package/dist/{rasterize-JC2JVVOF.js → rasterize-NRWP2C6L.js} +2 -2
- package/dist/render-V5ZPTZQ7.js +10 -0
- package/dist/seal-LVG2DGAB.js +10 -0
- package/dist/{setup-WDUVPFX3.js → setup-V55AJP6H.js} +3 -3
- package/dist/{watch-UI7OXIES.js → watch-7CQSMDDP.js} +9 -9
- package/package.json +1 -1
- package/dist/chunk-AT23GFSK.cjs.map +0 -1
- package/dist/chunk-BQSWSMRG.cjs.map +0 -1
- package/dist/chunk-GHVXLV52.js.map +0 -1
- package/dist/chunk-HFJWCAJ2.cjs.map +0 -1
- package/dist/chunk-HGUNWAXK.js.map +0 -1
- package/dist/chunk-PFUW4I4I.js.map +0 -1
- package/dist/chunk-SKGYPM6L.js.map +0 -1
- package/dist/chunk-SMGSCP57.js.map +0 -1
- package/dist/chunk-TROD36DK.cjs.map +0 -1
- package/dist/chunk-TYYJSOZS.cjs.map +0 -1
- package/dist/chunks-Q4GCPRRU.js +0 -10
- package/dist/formula-YMGXXRBQ.cjs.map +0 -1
- package/dist/image-ocr-BE2HHMVN.cjs.map +0 -1
- package/dist/parser-CMNHBVTL.js.map +0 -1
- package/dist/parser-NR3XO4SM.js.map +0 -1
- package/dist/parser-QJY5W47V.cjs.map +0 -1
- package/dist/pdf-ocr-DF3OTBKH.js +0 -13
- package/dist/pdf-ocr-R4DIKF4Q.js +0 -16
- package/dist/pdf-ocr-T3RHIZKY.cjs +0 -14
- package/dist/pdf-ocr-T3RHIZKY.cjs.map +0 -1
- package/dist/render-KSQ6UGCP.js +0 -10
- package/dist/seal-EOFFQ4JR.js +0 -10
- /package/dist/{-B3EUUBPV.js.map → -EZH6NKTE.js.map} +0 -0
- /package/dist/{chunk-QKFCGJJY.js.map → chunk-2HBYWA5X.js.map} +0 -0
- /package/dist/{chunk-COGFPMEB.js.map → chunk-2LO525TL.js.map} +0 -0
- /package/dist/{chunk-YASFVKDT.js.map → chunk-DBGEUDBY.js.map} +0 -0
- /package/dist/{chunk-4PKNWBXJ.js.map → chunk-FBLMVHDK.js.map} +0 -0
- /package/dist/{chunk-I2OETEEL.js.map → chunk-FVBR7SC3.js.map} +0 -0
- /package/dist/{chunk-46UDI56N.js.map → chunk-IOJX4Q63.js.map} +0 -0
- /package/dist/{chunk-BQE7AMFS.js.map → chunk-JCMBQC5N.js.map} +0 -0
- /package/dist/{chunk-TF5DS74F.js.map → chunk-KX7IYCCZ.js.map} +0 -0
- /package/dist/{chunk-MJPRWCI5.js.map → chunk-MUYU4JI5.js.map} +0 -0
- /package/dist/{chunk-2IR3PEZU.js.map → chunk-QBCMFNHS.js.map} +0 -0
- /package/dist/{chunk-DRLKMFDV.js.map → chunk-QM2RSJ2S.js.map} +0 -0
- /package/dist/{chunk-TJYZY2O6.js.map → chunk-UL2MDA5C.js.map} +0 -0
- /package/dist/{chunk-7JQ2ITT2.js.map → chunk-XXDZ2KTW.js.map} +0 -0
- /package/dist/{chunks-Q4GCPRRU.js.map → chunks-TT7ZE5QQ.js.map} +0 -0
- /package/dist/{formula-WGKKR5OQ.js.map → formula-6RQX4NMJ.js.map} +0 -0
- /package/dist/{formula-G4CD5KQ6.js.map → formula-NW33S4TA.js.map} +0 -0
- /package/dist/{image-ocr-MBWZK3ZT.js.map → image-ocr-CAVU2TNZ.js.map} +0 -0
- /package/dist/{image-ocr-M4RP4INO.js.map → image-ocr-SO56VETS.js.map} +0 -0
- /package/dist/{model-bundle-BIMP5DIH.js.map → model-bundle-EIJF34DR.js.map} +0 -0
- /package/dist/{models-QPOEGHVA.js.map → models-6V4CPQXC.js.map} +0 -0
- /package/dist/{pdf-ocr-DF3OTBKH.js.map → pdf-ocr-IRXQ5GBM.js.map} +0 -0
- /package/dist/{pdf-ocr-R4DIKF4Q.js.map → pdf-ocr-SL4GCOGS.js.map} +0 -0
- /package/dist/{profile-io-UZ43CLWX.js.map → profile-io-PZVMM66V.js.map} +0 -0
- /package/dist/{rasterize-JC2JVVOF.js.map → rasterize-NRWP2C6L.js.map} +0 -0
- /package/dist/{render-KSQ6UGCP.js.map → render-V5ZPTZQ7.js.map} +0 -0
- /package/dist/{seal-EOFFQ4JR.js.map → seal-LVG2DGAB.js.map} +0 -0
- /package/dist/{setup-WDUVPFX3.js.map → setup-V55AJP6H.js.map} +0 -0
- /package/dist/{watch-UI7OXIES.js.map → watch-7CQSMDDP.js.map} +0 -0
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
"use strict";Object.defineProperty(exports, "__esModule", {value: true}); function _interopRequireWildcard(obj) { if (obj && obj.__esModule) { return obj; } else { var newObj = {}; if (obj != null) { for (var key in obj) { if (Object.prototype.hasOwnProperty.call(obj, key)) { newObj[key] = obj[key]; } } } newObj.default = obj; return newObj; } } function _nullishCoalesce(lhs, rhsFn) { if (lhs != null) { return lhs; } else { return rhsFn(); } } 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; } var _class;
|
|
2
2
|
|
|
3
3
|
|
|
4
|
-
var
|
|
4
|
+
var _chunkV5BUPZ6Icjs = require('./chunk-V5BUPZ6I.cjs');
|
|
5
5
|
|
|
6
6
|
|
|
7
7
|
var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
|
|
@@ -18,12 +18,12 @@ var _chunkDCZVOIEOcjs = require('./chunk-DCZVOIEO.cjs');
|
|
|
18
18
|
|
|
19
19
|
|
|
20
20
|
|
|
21
|
-
var
|
|
21
|
+
var _chunkONNAHSPVcjs = require('./chunk-ONNAHSPV.cjs');
|
|
22
22
|
require('./chunk-T7GRHAIW.cjs');
|
|
23
23
|
|
|
24
24
|
|
|
25
25
|
|
|
26
|
-
var
|
|
26
|
+
var _chunk62IU4NV3cjs = require('./chunk-62IU4NV3.cjs');
|
|
27
27
|
|
|
28
28
|
|
|
29
29
|
var _chunkGS7T56RPcjs = require('./chunk-GS7T56RP.cjs');
|
|
@@ -149,7 +149,7 @@ async function extractPageImages(page, fnArray, argsArray, pageNumber, state, wa
|
|
|
149
149
|
if (state.seen.has(hash)) continue;
|
|
150
150
|
const rgba = toRgba(imgData);
|
|
151
151
|
if (!rgba) continue;
|
|
152
|
-
const png =
|
|
152
|
+
const png = _chunkV5BUPZ6Icjs.encodePng.call(void 0, w, h, rgba);
|
|
153
153
|
state.imageIndex++;
|
|
154
154
|
state.totalBytes += png.length;
|
|
155
155
|
const filename = `image_${String(state.imageIndex).padStart(3, "0")}.png`;
|
|
@@ -325,6 +325,300 @@ function summarizeDocumentQuality(pages) {
|
|
|
325
325
|
};
|
|
326
326
|
}
|
|
327
327
|
|
|
328
|
+
// src/pdf/symbol-fonts.ts
|
|
329
|
+
var WINGDINGS = [
|
|
330
|
+
"\u{1F589}",
|
|
331
|
+
"\u2702",
|
|
332
|
+
"\u2701",
|
|
333
|
+
"\u{1F453}",
|
|
334
|
+
"\u{1F56D}",
|
|
335
|
+
"\u{1F56E}",
|
|
336
|
+
"\u{1F56F}",
|
|
337
|
+
"\u{1F57F}",
|
|
338
|
+
"\u2706",
|
|
339
|
+
"\u{1F582}",
|
|
340
|
+
"\u{1F583}",
|
|
341
|
+
"\u{1F4EA}",
|
|
342
|
+
"\u{1F4EB}",
|
|
343
|
+
"\u{1F4EC}",
|
|
344
|
+
"\u{1F4ED}",
|
|
345
|
+
"\u{1F4C1}",
|
|
346
|
+
"\u{1F4C2}",
|
|
347
|
+
"\u{1F4C4}",
|
|
348
|
+
"\u{1F5CF}",
|
|
349
|
+
"\u{1F5D0}",
|
|
350
|
+
"\u{1F5C4}",
|
|
351
|
+
"\u231B",
|
|
352
|
+
"\u{1F5AE}",
|
|
353
|
+
"\u{1F5B0}",
|
|
354
|
+
"\u{1F5B2}",
|
|
355
|
+
"\u{1F5B3}",
|
|
356
|
+
"\u{1F5B4}",
|
|
357
|
+
"\u{1F5AB}",
|
|
358
|
+
"\u{1F5AC}",
|
|
359
|
+
"\u2707",
|
|
360
|
+
"\u270D",
|
|
361
|
+
"\u{1F58E}",
|
|
362
|
+
"\u270C",
|
|
363
|
+
"\u{1F44C}",
|
|
364
|
+
"\u{1F44D}",
|
|
365
|
+
"\u{1F44E}",
|
|
366
|
+
"\u261C",
|
|
367
|
+
"\u261E",
|
|
368
|
+
"\u261D",
|
|
369
|
+
"\u261F",
|
|
370
|
+
"\u{1F590}",
|
|
371
|
+
"\u263A",
|
|
372
|
+
"\u{1F610}",
|
|
373
|
+
"\u2639",
|
|
374
|
+
"\u{1F4A3}",
|
|
375
|
+
"\u2620",
|
|
376
|
+
"\u{1F3F3}",
|
|
377
|
+
"\u{1F3F1}",
|
|
378
|
+
"\u2708",
|
|
379
|
+
"\u263C",
|
|
380
|
+
"\u{1F4A7}",
|
|
381
|
+
"\u2744",
|
|
382
|
+
"\u{1F546}",
|
|
383
|
+
"\u271E",
|
|
384
|
+
"\u{1F548}",
|
|
385
|
+
"\u2720",
|
|
386
|
+
"\u2721",
|
|
387
|
+
"\u262A",
|
|
388
|
+
"\u262F",
|
|
389
|
+
"\u0950",
|
|
390
|
+
"\u2638",
|
|
391
|
+
"\u2648",
|
|
392
|
+
"\u2649",
|
|
393
|
+
"\u264A",
|
|
394
|
+
"\u264B",
|
|
395
|
+
"\u264C",
|
|
396
|
+
"\u264D",
|
|
397
|
+
"\u264E",
|
|
398
|
+
"\u264F",
|
|
399
|
+
"\u2650",
|
|
400
|
+
"\u2651",
|
|
401
|
+
"\u2652",
|
|
402
|
+
"\u2653",
|
|
403
|
+
"\u{1F670}",
|
|
404
|
+
"\u{1F675}",
|
|
405
|
+
"\u25CF",
|
|
406
|
+
"\u25CB",
|
|
407
|
+
"\u25A0",
|
|
408
|
+
"\u25A1",
|
|
409
|
+
"\u25A1",
|
|
410
|
+
"\u2751",
|
|
411
|
+
"\u2752",
|
|
412
|
+
"\u2B27",
|
|
413
|
+
"\u29EB",
|
|
414
|
+
"\u25C6",
|
|
415
|
+
"\u2756",
|
|
416
|
+
"\u2B25",
|
|
417
|
+
"\u2327",
|
|
418
|
+
"\u2BB9",
|
|
419
|
+
"\u2318",
|
|
420
|
+
"\u{1F3F5}",
|
|
421
|
+
"\u{1F3F6}",
|
|
422
|
+
"\u{1F676}",
|
|
423
|
+
"\u{1F677}",
|
|
424
|
+
"",
|
|
425
|
+
"\u24EA",
|
|
426
|
+
"\u2460",
|
|
427
|
+
"\u2461",
|
|
428
|
+
"\u2462",
|
|
429
|
+
"\u2463",
|
|
430
|
+
"\u2464",
|
|
431
|
+
"\u2465",
|
|
432
|
+
"\u2466",
|
|
433
|
+
"\u2467",
|
|
434
|
+
"\u2468",
|
|
435
|
+
"\u2469",
|
|
436
|
+
"\u24FF",
|
|
437
|
+
"\u2776",
|
|
438
|
+
"\u2777",
|
|
439
|
+
"\u2778",
|
|
440
|
+
"\u2779",
|
|
441
|
+
"\u277A",
|
|
442
|
+
"\u277B",
|
|
443
|
+
"\u277C",
|
|
444
|
+
"\u277D",
|
|
445
|
+
"\u277E",
|
|
446
|
+
"\u277F",
|
|
447
|
+
"\u{1F662}",
|
|
448
|
+
"\u{1F660}",
|
|
449
|
+
"\u{1F661}",
|
|
450
|
+
"\u{1F663}",
|
|
451
|
+
"\u{1F65E}",
|
|
452
|
+
"\u{1F65C}",
|
|
453
|
+
"\u{1F65D}",
|
|
454
|
+
"\u{1F65F}",
|
|
455
|
+
"\xB7",
|
|
456
|
+
"\u2022",
|
|
457
|
+
"\u25AA",
|
|
458
|
+
"\u26AA",
|
|
459
|
+
"\u25CB",
|
|
460
|
+
"\u25EF",
|
|
461
|
+
"\u25C9",
|
|
462
|
+
"\u25CE",
|
|
463
|
+
"\u{1F53F}",
|
|
464
|
+
"\u25AA",
|
|
465
|
+
"\u25FB",
|
|
466
|
+
"\u{1F7C2}",
|
|
467
|
+
"\u2726",
|
|
468
|
+
"\u2605",
|
|
469
|
+
"\u2736",
|
|
470
|
+
"\u2734",
|
|
471
|
+
"\u2739",
|
|
472
|
+
"\u2735",
|
|
473
|
+
"\u2BD0",
|
|
474
|
+
"\u2316",
|
|
475
|
+
"\u27E1",
|
|
476
|
+
"\u2311",
|
|
477
|
+
"\u2BD1",
|
|
478
|
+
"\u272A",
|
|
479
|
+
"\u2730",
|
|
480
|
+
"\u{1F550}",
|
|
481
|
+
"\u{1F551}",
|
|
482
|
+
"\u{1F552}",
|
|
483
|
+
"\u{1F553}",
|
|
484
|
+
"\u{1F554}",
|
|
485
|
+
"\u{1F555}",
|
|
486
|
+
"\u{1F556}",
|
|
487
|
+
"\u{1F557}",
|
|
488
|
+
"\u{1F558}",
|
|
489
|
+
"\u{1F559}",
|
|
490
|
+
"\u{1F55A}",
|
|
491
|
+
"\u{1F55B}",
|
|
492
|
+
"\u2BB0",
|
|
493
|
+
"\u2BB1",
|
|
494
|
+
"\u2BB2",
|
|
495
|
+
"\u2BB3",
|
|
496
|
+
"\u2BB4",
|
|
497
|
+
"\u2BB5",
|
|
498
|
+
"\u2BB6",
|
|
499
|
+
"\u2BB7",
|
|
500
|
+
"\u{1F66A}",
|
|
501
|
+
"\u{1F66B}",
|
|
502
|
+
"\u{1F655}",
|
|
503
|
+
"\u{1F654}",
|
|
504
|
+
"\u{1F657}",
|
|
505
|
+
"\u{1F656}",
|
|
506
|
+
"\u{1F650}",
|
|
507
|
+
"\u{1F651}",
|
|
508
|
+
"\u{1F652}",
|
|
509
|
+
"\u{1F653}",
|
|
510
|
+
"\u232B",
|
|
511
|
+
"\u2326",
|
|
512
|
+
"\u2B98",
|
|
513
|
+
"\u2B9A",
|
|
514
|
+
"\u2B99",
|
|
515
|
+
"\u2B9B",
|
|
516
|
+
"\u2B88",
|
|
517
|
+
"\u2B8A",
|
|
518
|
+
"\u2B89",
|
|
519
|
+
"\u2B8B",
|
|
520
|
+
"\u2190",
|
|
521
|
+
"\u2192",
|
|
522
|
+
"\u2191",
|
|
523
|
+
"\u2193",
|
|
524
|
+
"\u2196",
|
|
525
|
+
"\u2197",
|
|
526
|
+
"\u2199",
|
|
527
|
+
"\u2198",
|
|
528
|
+
"\u2B05",
|
|
529
|
+
"\u2794",
|
|
530
|
+
"\u2B06",
|
|
531
|
+
"\u2B07",
|
|
532
|
+
"\u2B09",
|
|
533
|
+
"\u2B08",
|
|
534
|
+
"\u2B0B",
|
|
535
|
+
"\u2B0A",
|
|
536
|
+
"\u21E6",
|
|
537
|
+
"\u21E8",
|
|
538
|
+
"\u21E7",
|
|
539
|
+
"\u21E9",
|
|
540
|
+
"\u2B04",
|
|
541
|
+
"\u21F3",
|
|
542
|
+
"\u2B00",
|
|
543
|
+
"\u2B01",
|
|
544
|
+
"\u2B03",
|
|
545
|
+
"\u2B02",
|
|
546
|
+
"\u{1F8AC}",
|
|
547
|
+
"\u{1F8AD}",
|
|
548
|
+
"\u2717",
|
|
549
|
+
"\u2714",
|
|
550
|
+
"\u2612",
|
|
551
|
+
"\u2611",
|
|
552
|
+
""
|
|
553
|
+
];
|
|
554
|
+
var WINANSI_REVERSE = {
|
|
555
|
+
8364: 128,
|
|
556
|
+
8218: 130,
|
|
557
|
+
402: 131,
|
|
558
|
+
8222: 132,
|
|
559
|
+
8230: 133,
|
|
560
|
+
8224: 134,
|
|
561
|
+
8225: 135,
|
|
562
|
+
710: 136,
|
|
563
|
+
8240: 137,
|
|
564
|
+
352: 138,
|
|
565
|
+
8249: 139,
|
|
566
|
+
338: 140,
|
|
567
|
+
381: 142,
|
|
568
|
+
8216: 145,
|
|
569
|
+
8217: 146,
|
|
570
|
+
8220: 147,
|
|
571
|
+
8221: 148,
|
|
572
|
+
8226: 149,
|
|
573
|
+
8211: 150,
|
|
574
|
+
8212: 151,
|
|
575
|
+
732: 152,
|
|
576
|
+
8482: 153,
|
|
577
|
+
353: 154,
|
|
578
|
+
8250: 155,
|
|
579
|
+
339: 156,
|
|
580
|
+
382: 158,
|
|
581
|
+
376: 159
|
|
582
|
+
};
|
|
583
|
+
function symbolFontTable(fontName) {
|
|
584
|
+
if (!fontName) return void 0;
|
|
585
|
+
if (/wingdings(?![\s-]*[23])/i.test(fontName)) return WINGDINGS;
|
|
586
|
+
return void 0;
|
|
587
|
+
}
|
|
588
|
+
function remapSymbolText(text, table) {
|
|
589
|
+
let out = "";
|
|
590
|
+
for (const ch of text) {
|
|
591
|
+
let code = ch.codePointAt(0);
|
|
592
|
+
if (code >= 128 && WINANSI_REVERSE[code] !== void 0) code = WINANSI_REVERSE[code];
|
|
593
|
+
if (code >= 33 && code <= 255) {
|
|
594
|
+
const mapped = table[code - 33];
|
|
595
|
+
out += mapped !== void 0 && mapped !== "" ? mapped : ch;
|
|
596
|
+
} else {
|
|
597
|
+
out += ch;
|
|
598
|
+
}
|
|
599
|
+
}
|
|
600
|
+
return out;
|
|
601
|
+
}
|
|
602
|
+
function remapSymbolFontItems(items, resolveFontName) {
|
|
603
|
+
const cache = /* @__PURE__ */ new Map();
|
|
604
|
+
let changed = 0;
|
|
605
|
+
for (const it of items) {
|
|
606
|
+
if (!it.fontName || !it.text) continue;
|
|
607
|
+
let table = cache.get(it.fontName);
|
|
608
|
+
if (!cache.has(it.fontName)) {
|
|
609
|
+
table = symbolFontTable(resolveFontName(it.fontName));
|
|
610
|
+
cache.set(it.fontName, table);
|
|
611
|
+
}
|
|
612
|
+
if (!table) continue;
|
|
613
|
+
const mapped = remapSymbolText(it.text, table);
|
|
614
|
+
if (mapped !== it.text) {
|
|
615
|
+
it.text = mapped;
|
|
616
|
+
changed++;
|
|
617
|
+
}
|
|
618
|
+
}
|
|
619
|
+
return changed;
|
|
620
|
+
}
|
|
621
|
+
|
|
328
622
|
// src/pdf/text-clean.ts
|
|
329
623
|
function sanitizeBlockControlChars(blocks) {
|
|
330
624
|
for (const b of blocks) {
|
|
@@ -344,7 +638,7 @@ function cleanPdfText(text) {
|
|
|
344
638
|
stripControlChars(text).replace(/^\d{1,4}\n/, "").replace(/^[\s]*[-–—]\s*[-–—]?\d+[-–—]?[\s]*[-–—]?[\s]*$/gm, "").replace(/^\s*\d+\s*\/\s*\d+\s*$/gm, "").replace(/\n\d{1,4}\n/g, "\n").replace(/\n\d{1,4}$/, "").replace(/^#{1,6}\s*\d{1,4}\s*$/gm, "")
|
|
345
639
|
).replace(/^(?!\| ---).*$/gm, (line) => {
|
|
346
640
|
if (/^\s*\${1,2}.+\${1,2}\s*$/.test(line)) return line;
|
|
347
|
-
return
|
|
641
|
+
return _chunkONNAHSPVcjs.collapseEvenSpacing.call(void 0, line);
|
|
348
642
|
}).replace(/([□■◆○●▶ㅇ])\s+([가-힣])\s+([가-힣])/g, "$1 $2$3").replace(/\\~\\~/g, "~~").replace(/~~~~/g, "").replace(/<u>\s*<\/u>/g, "").replace(/\n{3,}/g, "\n\n").trim();
|
|
349
643
|
}
|
|
350
644
|
function startsWithMarker(line) {
|
|
@@ -392,7 +686,7 @@ function applyLinkAnnotations(items, annots) {
|
|
|
392
686
|
const wrapped = /* @__PURE__ */ new Set();
|
|
393
687
|
for (const a of annots) {
|
|
394
688
|
if (a.subtype !== "Link" || !a.url || !a.rect || a.rect.length < 4) continue;
|
|
395
|
-
const url =
|
|
689
|
+
const url = _chunk62IU4NV3cjs.sanitizeHref.call(void 0, a.url);
|
|
396
690
|
if (!url) continue;
|
|
397
691
|
const x1 = Math.min(a.rect[0], a.rect[2]), x2 = Math.max(a.rect[0], a.rect[2]);
|
|
398
692
|
const y1 = Math.min(a.rect[1], a.rect[3]), y2 = Math.max(a.rect[1], a.rect[3]);
|
|
@@ -423,7 +717,7 @@ function applyLinkAnnotations(items, annots) {
|
|
|
423
717
|
|
|
424
718
|
// src/pdf/formula-ocr.ts
|
|
425
719
|
async function applyFormulaOcr(buffer, blocks, pageFilter, effectivePageCount, warnings, _onProgress) {
|
|
426
|
-
const formulaMod = await Promise.resolve().then(() => _interopRequireWildcard(require("./formula-
|
|
720
|
+
const formulaMod = await Promise.resolve().then(() => _interopRequireWildcard(require("./formula-NBEO6CFC.cjs")));
|
|
427
721
|
const { FormulaPipeline, ensureFormulaModels } = formulaMod;
|
|
428
722
|
await ensureFormulaModels((p) => {
|
|
429
723
|
if (p.phase === "download" && p.total) {
|
|
@@ -591,7 +885,7 @@ async function loadPdfWithTimeout(buffer) {
|
|
|
591
885
|
new Promise((_, reject) => {
|
|
592
886
|
timer = setTimeout(() => {
|
|
593
887
|
loadingTask.destroy();
|
|
594
|
-
reject(new (0,
|
|
888
|
+
reject(new (0, _chunk62IU4NV3cjs.KordocError)("PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (30\uCD08 \uCD08\uACFC)"));
|
|
595
889
|
}, PDF_LOAD_TIMEOUT_MS);
|
|
596
890
|
})
|
|
597
891
|
]);
|
|
@@ -605,7 +899,7 @@ async function parsePdfDocument(buffer, options) {
|
|
|
605
899
|
const doc = await loadPdfWithTimeout(buffer);
|
|
606
900
|
try {
|
|
607
901
|
const pageCount = doc.numPages;
|
|
608
|
-
if (pageCount === 0) throw new (0,
|
|
902
|
+
if (pageCount === 0) throw new (0, _chunk62IU4NV3cjs.KordocError)("PDF\uC5D0 \uD398\uC774\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
|
|
609
903
|
const metadata = { pageCount, pageMode: "layout" };
|
|
610
904
|
await extractPdfMetadata(doc, metadata);
|
|
611
905
|
const blocks = [];
|
|
@@ -632,8 +926,8 @@ async function parsePdfDocument(buffer, options) {
|
|
|
632
926
|
const viewport = page.getViewport({ scale: 1 });
|
|
633
927
|
pageHeights.set(i, viewport.height);
|
|
634
928
|
const rawItems = tc.items;
|
|
635
|
-
const items =
|
|
636
|
-
const { visible, hiddenCount } =
|
|
929
|
+
const items = _chunkONNAHSPVcjs.normalizeItems.call(void 0, rawItems);
|
|
930
|
+
const { visible, hiddenCount } = _chunkONNAHSPVcjs.filterHiddenText.call(void 0, items, viewport.width, viewport.height);
|
|
637
931
|
if (hiddenCount > 0) {
|
|
638
932
|
warnings.push({ page: i, message: `${hiddenCount}\uAC1C \uC228\uACA8\uC9C4 \uD14D\uC2A4\uD2B8 \uC694\uC18C \uD544\uD130\uB9C1\uB428`, code: "HIDDEN_TEXT_FILTERED" });
|
|
639
933
|
}
|
|
@@ -646,9 +940,16 @@ async function parsePdfDocument(buffer, options) {
|
|
|
646
940
|
} catch (e4) {
|
|
647
941
|
}
|
|
648
942
|
const opList = await page.getOperatorList();
|
|
943
|
+
remapSymbolFontItems(visible, (loadedName) => {
|
|
944
|
+
try {
|
|
945
|
+
return page.commonObjs.has(loadedName) ? _optionalChain([page, 'access', _10 => _10.commonObjs, 'access', _11 => _11.get, 'call', _12 => _12(loadedName), 'optionalAccess', _13 => _13.name]) : void 0;
|
|
946
|
+
} catch (e5) {
|
|
947
|
+
return void 0;
|
|
948
|
+
}
|
|
949
|
+
});
|
|
649
950
|
const pageArea = viewport.width * viewport.height;
|
|
650
951
|
if (pageArea > 0) {
|
|
651
|
-
const imageRegions =
|
|
952
|
+
const imageRegions = _chunkONNAHSPVcjs.extractImageRegions.call(void 0, opList.fnArray, opList.argsArray);
|
|
652
953
|
let uncovered = 0;
|
|
653
954
|
for (const r of imageRegions) {
|
|
654
955
|
const area = (r.x2 - r.x1) * (r.y2 - r.y1);
|
|
@@ -663,13 +964,13 @@ async function parsePdfDocument(buffer, options) {
|
|
|
663
964
|
}
|
|
664
965
|
if (uncovered > 0) skippedImagePages.set(i, uncovered);
|
|
665
966
|
}
|
|
666
|
-
const pageBlocks =
|
|
967
|
+
const pageBlocks = _chunkONNAHSPVcjs.extractPageBlocksWithLines.call(void 0, visible, i, opList, viewport.width, viewport.height, void 0, _optionalChain([options, 'optionalAccess', _14 => _14.tables]) !== false);
|
|
667
968
|
for (const b of pageBlocks) blocks.push(b);
|
|
668
969
|
try {
|
|
669
970
|
const { blocks: imgBlocks, images: pageImages } = await extractPageImages(page, opList.fnArray, opList.argsArray, i, imageState, warnings);
|
|
670
971
|
if (imgBlocks.length > 0) pageImageBlocks.set(i, imgBlocks);
|
|
671
972
|
extractedImages.push(...pageImages);
|
|
672
|
-
} catch (
|
|
973
|
+
} catch (e6) {
|
|
673
974
|
}
|
|
674
975
|
let pageText = "";
|
|
675
976
|
for (const b of pageBlocks) {
|
|
@@ -683,18 +984,18 @@ async function parsePdfDocument(buffer, options) {
|
|
|
683
984
|
pageText += pageText ? "\n" + t : t;
|
|
684
985
|
}
|
|
685
986
|
pageQuality.push(computePageQuality(i, pageText));
|
|
686
|
-
if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0,
|
|
987
|
+
if (totalTextBytes > MAX_TOTAL_TEXT) throw new (0, _chunk62IU4NV3cjs.KordocError)("\uD14D\uC2A4\uD2B8 \uCD94\uCD9C \uD06C\uAE30 \uCD08\uACFC");
|
|
687
988
|
parsedPages++;
|
|
688
|
-
_optionalChain([options, 'optionalAccess',
|
|
989
|
+
_optionalChain([options, 'optionalAccess', _15 => _15.onProgress, 'optionalCall', _16 => _16(parsedPages, totalTarget)]);
|
|
689
990
|
} catch (pageErr) {
|
|
690
|
-
if (pageErr instanceof
|
|
991
|
+
if (pageErr instanceof _chunk62IU4NV3cjs.KordocError) throw pageErr;
|
|
691
992
|
warnings.push({ page: i, message: `\uD398\uC774\uC9C0 ${i} \uD30C\uC2F1 \uC2E4\uD328: ${pageErr instanceof Error ? pageErr.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`, code: "PARTIAL_PARSE" });
|
|
692
993
|
}
|
|
693
994
|
}
|
|
694
995
|
const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount);
|
|
695
996
|
const isImageBased = totalChars / Math.max(parsedPageCount, 1) < 10;
|
|
696
997
|
const ocrDone = /* @__PURE__ */ new Set();
|
|
697
|
-
if (_optionalChain([options, 'optionalAccess',
|
|
998
|
+
if (_optionalChain([options, 'optionalAccess', _17 => _17.ocr]) && ocrBuffer) {
|
|
698
999
|
const inScope = (p) => !pageFilter || pageFilter.has(p);
|
|
699
1000
|
const targets = /* @__PURE__ */ new Set();
|
|
700
1001
|
if (options.ocr === "force" || isImageBased) {
|
|
@@ -708,7 +1009,7 @@ async function parsePdfDocument(buffer, options) {
|
|
|
708
1009
|
}
|
|
709
1010
|
if (targets.size > 0) {
|
|
710
1011
|
try {
|
|
711
|
-
const { runPdfOcr } = await Promise.resolve().then(() => _interopRequireWildcard(require("./pdf-ocr-
|
|
1012
|
+
const { runPdfOcr } = await Promise.resolve().then(() => _interopRequireWildcard(require("./pdf-ocr-TFYVK4FI.cjs")));
|
|
712
1013
|
const mode = typeof options.ocr === "function" ? options.ocr : "builtin";
|
|
713
1014
|
const ocrPageBlocks = await runPdfOcr(ocrBuffer, targets, mode, warnings, options.onProgress, options.tables !== false);
|
|
714
1015
|
if (ocrPageBlocks.size > 0) {
|
|
@@ -757,15 +1058,15 @@ async function parsePdfDocument(buffer, options) {
|
|
|
757
1058
|
warnings.push({ page, message: `${count}\uAC1C \uC774\uBBF8\uC9C0 \uC601\uC5ED\uC5D0 \uCD94\uCD9C \uAC00\uB2A5\uD55C \uD14D\uC2A4\uD2B8 \uC5C6\uC74C (\uADF8\uB9BC/\uCC28\uD2B8/\uB3C4\uC7A5 \uB0B4\uC6A9 \uB204\uB77D \uAC00\uB2A5)`, code: "SKIPPED_IMAGE" });
|
|
758
1059
|
}
|
|
759
1060
|
}
|
|
760
|
-
if (_optionalChain([options, 'optionalAccess',
|
|
761
|
-
const removed =
|
|
1061
|
+
if (_optionalChain([options, 'optionalAccess', _18 => _18.removeHeaderFooter]) !== false && parsedPageCount >= 3) {
|
|
1062
|
+
const removed = _chunkONNAHSPVcjs.removeHeaderFooterBlocks.call(void 0, blocks, pageHeights, warnings);
|
|
762
1063
|
for (let ri = removed.length - 1; ri >= 0; ri--) {
|
|
763
1064
|
blocks.splice(removed[ri], 1);
|
|
764
1065
|
}
|
|
765
1066
|
}
|
|
766
|
-
|
|
1067
|
+
_chunkONNAHSPVcjs.mergeCrossPageTables.call(void 0, blocks);
|
|
767
1068
|
injectPageImageBlocks(blocks, pageImageBlocks);
|
|
768
|
-
if (_optionalChain([options, 'optionalAccess',
|
|
1069
|
+
if (_optionalChain([options, 'optionalAccess', _19 => _19.formulaOcr]) && formulaBuffer) {
|
|
769
1070
|
try {
|
|
770
1071
|
await applyFormulaOcr(formulaBuffer, blocks, pageFilter, effectivePageCount, warnings, options.onProgress);
|
|
771
1072
|
} catch (e) {
|
|
@@ -775,16 +1076,16 @@ async function parsePdfDocument(buffer, options) {
|
|
|
775
1076
|
});
|
|
776
1077
|
}
|
|
777
1078
|
}
|
|
778
|
-
const medianFontSize =
|
|
1079
|
+
const medianFontSize = _chunkONNAHSPVcjs.computeMedianFontSizeFromFreq.call(void 0, fontSizeFreq);
|
|
779
1080
|
if (medianFontSize > 0) {
|
|
780
|
-
|
|
1081
|
+
_chunkONNAHSPVcjs.detectHeadings.call(void 0, blocks, medianFontSize);
|
|
781
1082
|
}
|
|
782
|
-
|
|
783
|
-
|
|
784
|
-
|
|
1083
|
+
_chunkONNAHSPVcjs.detectMarkerHeadings.call(void 0, blocks);
|
|
1084
|
+
_chunkONNAHSPVcjs.detectTableCaptions.call(void 0, blocks);
|
|
1085
|
+
_chunkONNAHSPVcjs.detectKoreanListBlocks.call(void 0, blocks);
|
|
785
1086
|
const outline = blocks.filter((b) => b.type === "heading" && b.level && b.text).map((b) => ({ level: b.level, text: b.text, pageNumber: b.pageNumber }));
|
|
786
1087
|
sanitizeBlockControlChars(blocks);
|
|
787
|
-
let markdown = cleanPdfText(
|
|
1088
|
+
let markdown = cleanPdfText(_chunkV5BUPZ6Icjs.blocksToMarkdown.call(void 0, blocks));
|
|
788
1089
|
return {
|
|
789
1090
|
markdown,
|
|
790
1091
|
blocks,
|
|
@@ -804,7 +1105,7 @@ async function parsePdfDocument(buffer, options) {
|
|
|
804
1105
|
async function extractPdfMetadata(doc, metadata) {
|
|
805
1106
|
try {
|
|
806
1107
|
const result = await doc.getMetadata();
|
|
807
|
-
if (!_optionalChain([result, 'optionalAccess',
|
|
1108
|
+
if (!_optionalChain([result, 'optionalAccess', _20 => _20.info])) return;
|
|
808
1109
|
const info = result.info;
|
|
809
1110
|
if (typeof info.Title === "string" && info.Title.trim()) metadata.title = info.Title.trim();
|
|
810
1111
|
if (typeof info.Author === "string" && info.Author.trim()) metadata.author = info.Author.trim();
|
|
@@ -815,7 +1116,7 @@ async function extractPdfMetadata(doc, metadata) {
|
|
|
815
1116
|
}
|
|
816
1117
|
if (typeof info.CreationDate === "string") metadata.createdAt = parsePdfDate(info.CreationDate);
|
|
817
1118
|
if (typeof info.ModDate === "string") metadata.modifiedAt = parsePdfDate(info.ModDate);
|
|
818
|
-
} catch (
|
|
1119
|
+
} catch (e7) {
|
|
819
1120
|
}
|
|
820
1121
|
}
|
|
821
1122
|
function parsePdfDate(dateStr) {
|
|
@@ -843,5 +1144,5 @@ async function extractPdfMetadataOnly(buffer) {
|
|
|
843
1144
|
|
|
844
1145
|
|
|
845
1146
|
|
|
846
|
-
exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks =
|
|
847
|
-
//# sourceMappingURL=parser-
|
|
1147
|
+
exports.cleanPdfText = cleanPdfText; exports.detectKoreanListBlocks = _chunkONNAHSPVcjs.detectKoreanListBlocks; exports.detectTableCaptions = _chunkONNAHSPVcjs.detectTableCaptions; exports.extractPdfMetadataOnly = extractPdfMetadataOnly; exports.mergeCrossPageTables = _chunkONNAHSPVcjs.mergeCrossPageTables; exports.parsePdfDocument = parsePdfDocument; exports.removeHeaderFooterBlocks = _chunkONNAHSPVcjs.removeHeaderFooterBlocks;
|
|
1148
|
+
//# sourceMappingURL=parser-OMT25WH2.cjs.map
|