@clazic/kordoc 2.5.0 → 2.5.2

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.
Files changed (44) hide show
  1. package/dist/batch-provider-XRF6F26E.js +234 -0
  2. package/dist/batch-provider-XRF6F26E.js.map +1 -0
  3. package/dist/{chunk-NKUNXGWI.js → chunk-25ZYYLVP.js} +8 -6
  4. package/dist/chunk-25ZYYLVP.js.map +1 -0
  5. package/dist/{chunk-IJGNPAK2.js → chunk-5CILZHRW.js} +2 -2
  6. package/dist/chunk-S7BHLD2V.js +200 -0
  7. package/dist/{chunk-Y4WFKJ5P.js.map → chunk-S7BHLD2V.js.map} +1 -1
  8. package/dist/cli.js +6 -38
  9. package/dist/cli.js.map +1 -1
  10. package/dist/index.cjs +473 -2816
  11. package/dist/index.cjs.map +1 -1
  12. package/dist/index.d.cts +4 -1
  13. package/dist/index.d.ts +4 -1
  14. package/dist/index.js +482 -2826
  15. package/dist/index.js.map +1 -1
  16. package/dist/mcp.js +3 -6
  17. package/dist/mcp.js.map +1 -1
  18. package/dist/{resolve-XWYJYKKH.js → resolve-ZSUEJK3E.js} +4 -4
  19. package/dist/{utils-RBXHHCLI.js → utils-H2BL5GNR.js} +2 -2
  20. package/dist/{watch-FRLS6FKE.js → watch-D6ODQLPJ.js} +4 -7
  21. package/dist/{watch-FRLS6FKE.js.map → watch-D6ODQLPJ.js.map} +1 -1
  22. package/package.json +3 -7
  23. package/dist/batch-provider-5BFJRKAZ.js +0 -190
  24. package/dist/batch-provider-5BFJRKAZ.js.map +0 -1
  25. package/dist/chunk-AEWWERJ5.js +0 -35
  26. package/dist/chunk-AEWWERJ5.js.map +0 -1
  27. package/dist/chunk-CPTOBJJD.js +0 -125
  28. package/dist/chunk-CPTOBJJD.js.map +0 -1
  29. package/dist/chunk-NKUNXGWI.js.map +0 -1
  30. package/dist/chunk-THBLCND6.js +0 -33
  31. package/dist/chunk-THBLCND6.js.map +0 -1
  32. package/dist/chunk-Y4WFKJ5P.js +0 -167
  33. package/dist/doctor-SJ7NYSXC.js +0 -126
  34. package/dist/doctor-SJ7NYSXC.js.map +0 -1
  35. package/dist/install-commands-P2KTEXQ4.js +0 -11
  36. package/dist/pm-7KGLH6MX.js +0 -9
  37. package/dist/pm-7KGLH6MX.js.map +0 -1
  38. package/dist/setup/doctor.cjs +0 -308
  39. package/dist/setup/doctor.js +0 -288
  40. package/dist/utils-RBXHHCLI.js.map +0 -1
  41. package/scripts/postinstall.cjs +0 -27
  42. /package/dist/{chunk-IJGNPAK2.js.map → chunk-5CILZHRW.js.map} +0 -0
  43. /package/dist/{resolve-XWYJYKKH.js.map → resolve-ZSUEJK3E.js.map} +0 -0
  44. /package/dist/{install-commands-P2KTEXQ4.js.map → utils-H2BL5GNR.js.map} +0 -0
package/dist/index.js CHANGED
@@ -36,7 +36,6 @@ var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__ge
36
36
  isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
37
37
  mod
38
38
  ));
39
- var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
40
39
 
41
40
  // src/page-range.ts
42
41
  var page_range_exports = {};
@@ -127,9 +126,9 @@ function sanitizeMeta(meta, cfg) {
127
126
  }
128
127
  return out;
129
128
  }
130
- function parsePositiveInt(input, fallback2) {
129
+ function parsePositiveInt(input, fallback) {
131
130
  const n = Number(input);
132
- if (!Number.isFinite(n) || n < 0) return fallback2;
131
+ if (!Number.isFinite(n) || n < 0) return fallback;
133
132
  return Math.floor(n);
134
133
  }
135
134
  function limitText(input, maxLen) {
@@ -294,9 +293,9 @@ var init_logger = __esm({
294
293
  }
295
294
  });
296
295
 
297
- // ../../../node_modules/cfb/cfb.js
296
+ // node_modules/cfb/cfb.js
298
297
  var require_cfb = __commonJS({
299
- "../../../node_modules/cfb/cfb.js"(exports, module) {
298
+ "node_modules/cfb/cfb.js"(exports, module) {
300
299
  "use strict";
301
300
  var Base64_map = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=";
302
301
  function Base64_encode(input) {
@@ -2399,15 +2398,48 @@ var OCR_PROMPT, _tempDir;
2399
2398
  var init_cli_provider = __esm({
2400
2399
  "src/ocr/cli-provider.ts"() {
2401
2400
  "use strict";
2402
- OCR_PROMPT = `\uC774 PDF \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0\uC5D0\uC11C \uD14D\uC2A4\uD2B8\uC640 \uD14C\uC774\uBE14\uC744 \uCD94\uCD9C\uD558\uC5EC \uC21C\uC218 Markdown\uC73C\uB85C \uBCC0\uD658\uD558\uC138\uC694.
2403
- \uADDC\uCE59:
2404
- - \uD14C\uC774\uBE14\uC740 Markdown \uD14C\uC774\uBE14 \uBB38\uBC95 \uC0AC\uC6A9 (| \uAD6C\uBD84, |---|---| \uD5E4\uB354 \uAD6C\uBD84\uC120 \uD3EC\uD568)
2405
- - \uBCD1\uD569\uB41C \uC140\uC740 \uD574\uB2F9 \uC704\uCE58\uC5D0 \uB0B4\uC6A9 \uAE30\uC7AC
2406
- - \uD5E4\uB529\uC740 \uAE00\uC790 \uD06C\uAE30\uC5D0 \uB530\uB77C ## ~ ###### \uC0AC\uC6A9
2407
- - \uB9AC\uC2A4\uD2B8\uB294 - \uB610\uB294 1. \uC0AC\uC6A9
2408
- - \uC774\uBBF8\uC9C0, \uB3C4\uD615 \uB4F1 \uBE44\uD14D\uC2A4\uD2B8 \uC694\uC18C\uB294 \uBB34\uC2DC
2409
- - \uC6D0\uBB38\uC758 \uC77D\uAE30 \uC21C\uC11C\uC640 \uAD6C\uC870\uB97C \uC720\uC9C0
2410
- - \`\`\`\uB85C \uAC10\uC2F8\uC9C0 \uB9D0\uACE0 \uC21C\uC218 Markdown\uB9CC \uCD9C\uB825`;
2401
+ OCR_PROMPT = `\uC774 PDF \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0\uC5D0\uC11C \uD14D\uC2A4\uD2B8\uC640 \uD45C\uB97C \uCD94\uCD9C\uD558\uC5EC \uC21C\uC218 Markdown\uC73C\uB85C \uBCC0\uD658\uD558\uACE0, \uBA85\uBC31\uD55C OCR \uC624\uC778\uC2DD\uACFC \uD45C \uAD6C\uC870 \uAE68\uC9D0\uB9CC \uAD50\uC815\uD558\uC5EC \uCD5C\uC885 \uACB0\uACFC\uB97C \uCD9C\uB825\uD558\uC138\uC694.
2402
+
2403
+ [\uAE30\uBCF8 \uCD94\uCD9C \uADDC\uCE59]
2404
+ - \uBCF8\uBB38, \uD45C, \uC81C\uBAA9, \uB9AC\uC2A4\uD2B8, \uCEA1\uC158\uC744 \uC6D0\uBB38 \uAD6C\uC870 \uADF8\uB300\uB85C Markdown\uC73C\uB85C \uBCC0\uD658
2405
+ - \uD5E4\uB529\uC740 \uC2DC\uAC01\uC801 \uD06C\uAE30\xB7\uAD75\uAE30\uC5D0 \uB530\uB77C # ~ ###### \uC0AC\uC6A9
2406
+ - \uB9AC\uC2A4\uD2B8\uB294 -, 1. \uC0AC\uC6A9 (\uC6D0\uBB38 \uBC88\uD638 \uCCB4\uACC4\uAC00 \u2460, \uAC00., 1) \uB4F1\uC774\uBA74 \uADF8 \uD45C\uAE30 \uC720\uC9C0)
2407
+ - \uC774\uBBF8\uC9C0\xB7\uB3C4\uD615\xB7\uB85C\uACE0\xB7\uD398\uC774\uC9C0\uBC88\uD638\xB7\uBA38\uB9AC\uAE00/\uBC14\uB2E5\uAE00\uC740 \uBB34\uC2DC
2408
+ - \uC6D0\uBB38\uC758 \uC77D\uAE30 \uC21C\uC11C(\uC88C\u2192\uC6B0, \uC704\u2192\uC544\uB798, \uB2E4\uB2E8\uC774\uBA74 \uB2E8\uBCC4\uB85C)\uB97C \uC720\uC9C0
2409
+
2410
+ [\uD45C \uCD94\uCD9C \uADDC\uCE59 \u2014 \uAC00\uC7A5 \uC911\uC694]
2411
+ - \uD45C\uB294 \uBC18\uB4DC\uC2DC Markdown \uD30C\uC774\uD504 \uD14C\uC774\uBE14 \uBB38\uBC95 \uC0AC\uC6A9: \uD5E4\uB354 \uD589 + |---| \uAD6C\uBD84\uC120 + \uB370\uC774\uD130 \uD589
2412
+ - \uC2DC\uAC01\uC801\uC73C\uB85C \uAD75\uAC70\uB098 \uC74C\uC601(\uD68C\uC0C9/\uC0C9\uC0C1) \uBC30\uACBD\uC774 \uC788\uB294 \uD589\uC744 \uD5E4\uB354\uB85C \uC2DD\uBCC4. \uD5E4\uB354\uAC00 \uC5C6\uB294 \uD45C\uB77C\uB3C4 \uCCAB \uD589\uC744 \uD5E4\uB354\uB85C \uB450\uACE0 |---| \uAD6C\uBD84\uC120 \uCD94\uAC00
2413
+ - \uBAA8\uB4E0 \uD589\uC758 \uD30C\uC774\uD504(|) \uAC1C\uC218\uB97C \uD5E4\uB354\uC640 \uB3D9\uC77C\uD558\uAC8C \uB9DE\uCD9C \uAC83 \u2014 \uBE48 \uC140\uC740 \uACF5\uBC31\uC73C\uB85C \uCC44\uC6B0\uACE0 \uC808\uB300 \uC140\uC744 \uC0DD\uB7B5\uD558\uC9C0 \uB9D0 \uAC83
2414
+ - \uC140 \uC548\uC758 \uC904\uBC14\uAFC8\uC740 <br>\uB85C \uD45C\uAE30 (\uC2E4\uC81C \uAC1C\uD589 \uBB38\uC790 \uC0AC\uC6A9 \uAE08\uC9C0 \u2014 \uD45C\uAC00 \uAE68\uC9D0)
2415
+ - \uBCD1\uD569 \uC140 \uCC98\uB9AC:
2416
+ - \uAC00\uB85C \uBCD1\uD569(colspan): \uBCD1\uD569\uB41C \uC140 \uB0B4\uC6A9\uC744 \uCCAB \uCE78\uC5D0 \uC4F0\uACE0 \uB098\uBA38\uC9C0 \uCE78\uC740 \uBE48 \uCE78\uC73C\uB85C \uB458 \uAC83
2417
+ - \uC138\uB85C \uBCD1\uD569(rowspan): \uBCD1\uD569\uB41C \uBAA8\uB4E0 \uD589\uC758 \uD574\uB2F9 \uCE78\uC5D0 \uB3D9\uC77C\uD55C \uB0B4\uC6A9\uC744 \uBC18\uBCF5 \uAE30\uC7AC\uD560 \uAC83
2418
+ - 2\uB2E8 \uD5E4\uB354(\uD5E4\uB354\uAC00 \uB450 \uC904\uC778 \uACBD\uC6B0): \uC0C1\uC704/\uD558\uC704 \uD5E4\uB354\uB97C " / "\uB85C \uD569\uCCD0 \uB2E8\uC77C \uD5E4\uB354 \uD589\uC73C\uB85C \uB9CC\uB4E4 \uAC83 (\uC608: "2024 / 1\uBD84\uAE30")
2419
+ - \uD45C \uC704\xB7\uC544\uB798\uC758 \uCEA1\uC158(\uC608: "<\uD45C 1> ...", "[\uD45C 2-1]")\uC740 \uD45C \uBC14\uB85C \uC704\uC5D0 \uBCC4\uB3C4 \uC904\uB85C \uBCF4\uC874
2420
+ - \uD55C \uD398\uC774\uC9C0\uC5D0\uC11C \uD45C\uAC00 \uC911\uAC04\uC5D0 \uB04A\uC5B4\uC838 \uBCF4\uC774\uB354\uB77C\uB3C4, \uD5E4\uB354\uAC00 \uB3D9\uC77C\uD558\uBA74 \uD558\uB098\uC758 \uD45C\uB85C \uC774\uC5B4\uBD99\uC77C \uAC83
2421
+ - \uC88C\uCE21 \uCCAB \uCEEC\uB7FC\uC774 \uC138\uB85C\uC4F0\uAE30 \uB77C\uBCA8(\uAD6C\uBD84/\uBD84\uB958/\uC5F0\uB3C4 \uB4F1)\uC774\uB77C\uB3C4 \uC77C\uBC18 \uC140\uB85C \uBCC0\uD658\uD558\uC5EC \uB204\uB77D \uAE08\uC9C0
2422
+
2423
+ [OCR \uC624\uC778\uC2DD\xB7\uAE68\uC9D0 \uAD50\uC815 \u2014 \uD5C8\uC6A9 \uBC94\uC704\uB9CC]
2424
+ - \uC22B\uC790 \uCE78\uC5D0\uC11C 'O'\u2192'0', 'l/I'\u2192'1' \uB4F1 \uB9E5\uB77D\uC0C1 \uBA85\uBC31\uD55C \uC624\uC778\uC2DD\uB9CC \uAD50\uC815
2425
+ - \uB2E8\uC5B4 \uC911\uAC04\uC5D0 \uC798\uBABB \uC0BD\uC785\uB41C \uACF5\uBC31 \uC81C\uAC70 (\uC608: "\uC8FC\uAC70 \uC885 \uD569 \uACC4\uD68D" \u2192 "\uC8FC\uAC70\uC885\uD569\uACC4\uD68D")
2426
+ - \uC904\uBC14\uAFC8 \uC624\uB958\uB85C \uB04A\uAE34 \uBB38\uC7A5\uC740 \uC758\uBBF8 \uB2E8\uC704\uB85C \uBCD1\uD569 (\uB2E8, \uD45C \uC140 \uB0B4\uBD80\uB294 <br> \uC720\uC9C0)
2427
+ - \uD45C\uC5D0\uC11C \uD589 \uC5B4\uAE0B\uB0A8\uC774 \uC758\uC2EC\uB418\uBA74(\uCEEC\uB7FC \uC218 \uBD88\uC77C\uCE58) \uBE48 \uC140\uB85C \uD328\uB529\uD558\uC5EC \uCEEC\uB7FC \uC815\uD569\uC131\uC744 \uC6B0\uC120 \uD655\uBCF4
2428
+
2429
+ [\uC808\uB300 \uAE08\uC9C0 \uC0AC\uD56D]
2430
+ - \uBB38\uC7A5\xB7\uB2E8\uB77D\xB7\uD56D\uBAA9\xB7\uD45C \uD589/\uC5F4\uC744 \uCD94\uAC00\uD558\uAC70\uB098 \uC0AD\uC81C\uD558\uC9C0 \uB9D0 \uAC83
2431
+ - \uC22B\uC790, \uD37C\uC13C\uD2B8, \uB0A0\uC9DC, \uB2E8\uC704, \uAE08\uC561\uC744 \uC808\uB300 \uBCC0\uACBD\uD558\uC9C0 \uB9D0 \uAC83
2432
+ - \uACE0\uC720\uBA85\uC0AC, \uAE30\uAD00\uBA85, \uBC95\uB839\uBA85, \uC9C0\uBA85, \uC778\uBA85\uC744 \uBCC0\uACBD\uD558\uC9C0 \uB9D0 \uAC83
2433
+ - \uD45C\uC758 \uD5E4\uB354 \uD14D\uC2A4\uD2B8, \uCEA1\uC158, \uD589/\uC5F4 \uAC1C\uC218\uB97C \uC784\uC758\uB85C \uBC14\uAFB8\uC9C0 \uB9D0 \uAC83
2434
+ - \uCCB4\uD06C\uBC15\uC2A4(\u2611/\u2610/\u25A0/\u25A1), \uD2B9\uC218\uBB38\uC790(\u203B, \u2460\u2461\u2462, \u3260, \uAC00., \uB098.)\uB97C \uC784\uC758\uB85C \uBC14\uAFB8\uC9C0 \uB9D0 \uAC83
2435
+ - \uC6D0\uBB38\uC5D0 \uC5C6\uB294 \uB0B4\uC6A9\uC744 \uC694\uC57D\xB7\uBCF4\uC644\xB7\uCD94\uB860\uD558\uC9C0 \uB9D0 \uAC83
2436
+ - \`\`\`, \`\`\`markdown \uAC19\uC740 \uCF54\uB4DC\uD39C\uC2A4\uB85C \uAC10\uC2F8\uC9C0 \uB9D0 \uAC83
2437
+ - \uC124\uBA85\xB7\uC8FC\uC11D\xB7\uBA54\uD0C0 \uBB38\uC7A5 \uCD94\uAC00 \uAE08\uC9C0
2438
+
2439
+ [\uBD88\uD655\uC2E4\uD560 \uB54C]
2440
+ - \uAE00\uC790\uAC00 \uD750\uB9BF\uD558\uAC70\uB098 \uD310\uB3C5 \uBD88\uAC00\uD558\uBA74 \uBCF4\uC774\uB294 \uADF8\uB300\uB85C \uB450\uAC70\uB098, \uBD88\uAC00\uD53C\uD558\uBA74 \uD55C \uAE00\uC790\uB9CC ?\uB85C \uD45C\uC2DC
2441
+ - \uD45C \uAD6C\uC870\uAC00 \uBAA8\uD638\uD558\uBA74 \uCEEC\uB7FC \uC218 \uC77C\uCE58 + \uBE48 \uC140 \uD328\uB529 \uD615\uD0DC\uB85C \uCD9C\uB825
2442
+ - \uCD94\uCE21\uBCF4\uB2E4 \uC6D0\uBB38 \uBCF4\uC874\uC744 \uD56D\uC0C1 \uC6B0\uC120`;
2411
2443
  _tempDir = null;
2412
2444
  }
2413
2445
  });
@@ -2476,7 +2508,7 @@ function createBatchCliProvider(mode, batchSize) {
2476
2508
  };
2477
2509
  }
2478
2510
  function spawnAsync(cmd, args, opts) {
2479
- return new Promise((resolve4, reject2) => {
2511
+ return new Promise((resolve4, reject) => {
2480
2512
  const child = spawn(cmd, args, {
2481
2513
  cwd: opts.cwd,
2482
2514
  env: process.env,
@@ -2510,14 +2542,14 @@ function spawnAsync(cmd, args, opts) {
2510
2542
  child.on("close", (code) => {
2511
2543
  clearTimeout(timer);
2512
2544
  if (killed) {
2513
- reject2(new Error(`\uD0C0\uC784\uC544\uC6C3 (${Math.round(opts.timeoutMs / 1e3)}\uCD08)`));
2545
+ reject(new Error(`\uD0C0\uC784\uC544\uC6C3 (${Math.round(opts.timeoutMs / 1e3)}\uCD08)`));
2514
2546
  } else {
2515
2547
  resolve4({ stdout, stderr, exitCode: code ?? 1 });
2516
2548
  }
2517
2549
  });
2518
2550
  child.on("error", (err) => {
2519
2551
  clearTimeout(timer);
2520
- reject2(err);
2552
+ reject(err);
2521
2553
  });
2522
2554
  });
2523
2555
  }
@@ -2596,7 +2628,51 @@ var BATCH_OCR_PROMPT, DEFAULT_BATCH_SIZES, _batchTempDir;
2596
2628
  var init_batch_provider = __esm({
2597
2629
  "src/ocr/batch-provider.ts"() {
2598
2630
  "use strict";
2599
- BATCH_OCR_PROMPT = "\uB2E4\uC74C \uBB38\uC11C \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0\uB4E4\uC744 OCR\uD558\uC5EC \uC21C\uC218 Markdown\uC73C\uB85C \uBCC0\uD658\uD558\uC138\uC694.\n\n\uADDC\uCE59:\n- \uAC01 \uD398\uC774\uC9C0 \uACB0\uACFC \uC0AC\uC774\uC5D0 \uBC18\uB4DC\uC2DC \uC774 \uAD6C\uBD84\uC790\uB97C \uC0BD\uC785: <!-- PAGE_BREAK -->\n- \uD14C\uC774\uBE14\uC740 Markdown \uD14C\uC774\uBE14 \uBB38\uBC95 \uC0AC\uC6A9 (| \uAD6C\uBD84, |---|---| \uD5E4\uB354 \uAD6C\uBD84\uC120 \uD3EC\uD568)\n- \uBCD1\uD569\uB41C \uC140\uC740 \uD574\uB2F9 \uC704\uCE58\uC5D0 \uB0B4\uC6A9 \uAE30\uC7AC\n- \uD5E4\uB529\uC740 \uAE00\uC790 \uD06C\uAE30\uC5D0 \uB530\uB77C ## ~ ###### \uC0AC\uC6A9\n- \uB9AC\uC2A4\uD2B8\uB294 - \uB610\uB294 1. \uC0AC\uC6A9\n- \uC774\uBBF8\uC9C0, \uB3C4\uD615 \uB4F1 \uBE44\uD14D\uC2A4\uD2B8 \uC694\uC18C\uB294 \uBB34\uC2DC\n- \uC6D0\uBB38\uC758 \uC77D\uAE30 \uC21C\uC11C\uC640 \uAD6C\uC870\uB97C \uC720\uC9C0\n- ```\uB85C \uAC10\uC2F8\uC9C0 \uB9D0\uACE0 \uC21C\uC218 Markdown\uB9CC \uCD9C\uB825";
2631
+ BATCH_OCR_PROMPT = `\uB2E4\uC74C \uBB38\uC11C \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0\uB4E4\uC744 OCR\uD558\uC5EC \uC21C\uC218 Markdown\uC73C\uB85C \uBCC0\uD658\uD558\uACE0, \uBA85\uBC31\uD55C OCR \uC624\uC778\uC2DD\uACFC \uD45C \uAD6C\uC870 \uAE68\uC9D0\uB9CC \uAD50\uC815\uD558\uC5EC \uCD5C\uC885 \uACB0\uACFC\uB97C \uCD9C\uB825\uD558\uC138\uC694.
2632
+
2633
+ [\uD398\uC774\uC9C0 \uAD6C\uBD84 \u2014 \uD544\uC218]
2634
+ - \uAC01 \uD398\uC774\uC9C0 \uACB0\uACFC \uC0AC\uC774\uC5D0 \uBC18\uB4DC\uC2DC \uC774 \uAD6C\uBD84\uC790\uB97C \uC0BD\uC785: <!-- PAGE_BREAK -->
2635
+
2636
+ [\uAE30\uBCF8 \uCD94\uCD9C \uADDC\uCE59]
2637
+ - \uBCF8\uBB38, \uD45C, \uC81C\uBAA9, \uB9AC\uC2A4\uD2B8, \uCEA1\uC158\uC744 \uC6D0\uBB38 \uAD6C\uC870 \uADF8\uB300\uB85C Markdown\uC73C\uB85C \uBCC0\uD658
2638
+ - \uD5E4\uB529\uC740 \uC2DC\uAC01\uC801 \uD06C\uAE30\xB7\uAD75\uAE30\uC5D0 \uB530\uB77C # ~ ###### \uC0AC\uC6A9
2639
+ - \uB9AC\uC2A4\uD2B8\uB294 -, 1. \uC0AC\uC6A9 (\uC6D0\uBB38 \uBC88\uD638 \uCCB4\uACC4\uAC00 \u2460, \uAC00., 1) \uB4F1\uC774\uBA74 \uADF8 \uD45C\uAE30 \uC720\uC9C0)
2640
+ - \uC774\uBBF8\uC9C0\xB7\uB3C4\uD615\xB7\uB85C\uACE0\xB7\uD398\uC774\uC9C0\uBC88\uD638\xB7\uBA38\uB9AC\uAE00/\uBC14\uB2E5\uAE00\uC740 \uBB34\uC2DC
2641
+ - \uC6D0\uBB38\uC758 \uC77D\uAE30 \uC21C\uC11C(\uC88C\u2192\uC6B0, \uC704\u2192\uC544\uB798, \uB2E4\uB2E8\uC774\uBA74 \uB2E8\uBCC4\uB85C)\uB97C \uC720\uC9C0
2642
+
2643
+ [\uD45C \uCD94\uCD9C \uADDC\uCE59 \u2014 \uAC00\uC7A5 \uC911\uC694]
2644
+ - \uD45C\uB294 \uBC18\uB4DC\uC2DC Markdown \uD30C\uC774\uD504 \uD14C\uC774\uBE14 \uBB38\uBC95 \uC0AC\uC6A9: \uD5E4\uB354 \uD589 + |---| \uAD6C\uBD84\uC120 + \uB370\uC774\uD130 \uD589
2645
+ - \uC2DC\uAC01\uC801\uC73C\uB85C \uAD75\uAC70\uB098 \uC74C\uC601(\uD68C\uC0C9/\uC0C9\uC0C1) \uBC30\uACBD\uC774 \uC788\uB294 \uD589\uC744 \uD5E4\uB354\uB85C \uC2DD\uBCC4. \uD5E4\uB354\uAC00 \uC5C6\uB294 \uD45C\uB77C\uB3C4 \uCCAB \uD589\uC744 \uD5E4\uB354\uB85C \uB450\uACE0 |---| \uAD6C\uBD84\uC120 \uCD94\uAC00
2646
+ - \uBAA8\uB4E0 \uD589\uC758 \uD30C\uC774\uD504(|) \uAC1C\uC218\uB97C \uD5E4\uB354\uC640 \uB3D9\uC77C\uD558\uAC8C \uB9DE\uCD9C \uAC83 \u2014 \uBE48 \uC140\uC740 \uACF5\uBC31\uC73C\uB85C \uCC44\uC6B0\uACE0 \uC808\uB300 \uC140\uC744 \uC0DD\uB7B5\uD558\uC9C0 \uB9D0 \uAC83
2647
+ - \uC140 \uC548\uC758 \uC904\uBC14\uAFC8\uC740 <br>\uB85C \uD45C\uAE30 (\uC2E4\uC81C \uAC1C\uD589 \uBB38\uC790 \uC0AC\uC6A9 \uAE08\uC9C0 \u2014 \uD45C\uAC00 \uAE68\uC9D0)
2648
+ - \uBCD1\uD569 \uC140 \uCC98\uB9AC:
2649
+ - \uAC00\uB85C \uBCD1\uD569(colspan): \uBCD1\uD569\uB41C \uC140 \uB0B4\uC6A9\uC744 \uCCAB \uCE78\uC5D0 \uC4F0\uACE0 \uB098\uBA38\uC9C0 \uCE78\uC740 \uBE48 \uCE78\uC73C\uB85C \uB458 \uAC83
2650
+ - \uC138\uB85C \uBCD1\uD569(rowspan): \uBCD1\uD569\uB41C \uBAA8\uB4E0 \uD589\uC758 \uD574\uB2F9 \uCE78\uC5D0 \uB3D9\uC77C\uD55C \uB0B4\uC6A9\uC744 \uBC18\uBCF5 \uAE30\uC7AC\uD560 \uAC83
2651
+ - 2\uB2E8 \uD5E4\uB354(\uD5E4\uB354\uAC00 \uB450 \uC904\uC778 \uACBD\uC6B0): \uC0C1\uC704/\uD558\uC704 \uD5E4\uB354\uB97C " / "\uB85C \uD569\uCCD0 \uB2E8\uC77C \uD5E4\uB354 \uD589\uC73C\uB85C \uB9CC\uB4E4 \uAC83 (\uC608: "2024 / 1\uBD84\uAE30")
2652
+ - \uD45C \uC704\xB7\uC544\uB798\uC758 \uCEA1\uC158(\uC608: "<\uD45C 1> ...", "[\uD45C 2-1]")\uC740 \uD45C \uBC14\uB85C \uC704\uC5D0 \uBCC4\uB3C4 \uC904\uB85C \uBCF4\uC874
2653
+ - \uD55C \uD398\uC774\uC9C0\uC5D0\uC11C \uD45C\uAC00 \uC911\uAC04\uC5D0 \uB04A\uC5B4\uC838 \uBCF4\uC774\uB354\uB77C\uB3C4, \uD5E4\uB354\uAC00 \uB3D9\uC77C\uD558\uBA74 \uD558\uB098\uC758 \uD45C\uB85C \uC774\uC5B4\uBD99\uC77C \uAC83
2654
+ - \uC88C\uCE21 \uCCAB \uCEEC\uB7FC\uC774 \uC138\uB85C\uC4F0\uAE30 \uB77C\uBCA8(\uAD6C\uBD84/\uBD84\uB958/\uC5F0\uB3C4 \uB4F1)\uC774\uB77C\uB3C4 \uC77C\uBC18 \uC140\uB85C \uBCC0\uD658\uD558\uC5EC \uB204\uB77D \uAE08\uC9C0
2655
+
2656
+ [OCR \uC624\uC778\uC2DD\xB7\uAE68\uC9D0 \uAD50\uC815 \u2014 \uD5C8\uC6A9 \uBC94\uC704\uB9CC]
2657
+ - \uC22B\uC790 \uCE78\uC5D0\uC11C 'O'\u2192'0', 'l/I'\u2192'1' \uB4F1 \uB9E5\uB77D\uC0C1 \uBA85\uBC31\uD55C \uC624\uC778\uC2DD\uB9CC \uAD50\uC815
2658
+ - \uB2E8\uC5B4 \uC911\uAC04\uC5D0 \uC798\uBABB \uC0BD\uC785\uB41C \uACF5\uBC31 \uC81C\uAC70 (\uC608: "\uC8FC\uAC70 \uC885 \uD569 \uACC4\uD68D" \u2192 "\uC8FC\uAC70\uC885\uD569\uACC4\uD68D")
2659
+ - \uC904\uBC14\uAFC8 \uC624\uB958\uB85C \uB04A\uAE34 \uBB38\uC7A5\uC740 \uC758\uBBF8 \uB2E8\uC704\uB85C \uBCD1\uD569 (\uB2E8, \uD45C \uC140 \uB0B4\uBD80\uB294 <br> \uC720\uC9C0)
2660
+ - \uD45C\uC5D0\uC11C \uD589 \uC5B4\uAE0B\uB0A8\uC774 \uC758\uC2EC\uB418\uBA74(\uCEEC\uB7FC \uC218 \uBD88\uC77C\uCE58) \uBE48 \uC140\uB85C \uD328\uB529\uD558\uC5EC \uCEEC\uB7FC \uC815\uD569\uC131\uC744 \uC6B0\uC120 \uD655\uBCF4
2661
+
2662
+ [\uC808\uB300 \uAE08\uC9C0 \uC0AC\uD56D]
2663
+ - \uBB38\uC7A5\xB7\uB2E8\uB77D\xB7\uD56D\uBAA9\xB7\uD45C \uD589/\uC5F4\uC744 \uCD94\uAC00\uD558\uAC70\uB098 \uC0AD\uC81C\uD558\uC9C0 \uB9D0 \uAC83
2664
+ - \uC22B\uC790, \uD37C\uC13C\uD2B8, \uB0A0\uC9DC, \uB2E8\uC704, \uAE08\uC561\uC744 \uC808\uB300 \uBCC0\uACBD\uD558\uC9C0 \uB9D0 \uAC83
2665
+ - \uACE0\uC720\uBA85\uC0AC, \uAE30\uAD00\uBA85, \uBC95\uB839\uBA85, \uC9C0\uBA85, \uC778\uBA85\uC744 \uBCC0\uACBD\uD558\uC9C0 \uB9D0 \uAC83
2666
+ - \uD45C\uC758 \uD5E4\uB354 \uD14D\uC2A4\uD2B8, \uCEA1\uC158, \uD589/\uC5F4 \uAC1C\uC218\uB97C \uC784\uC758\uB85C \uBC14\uAFB8\uC9C0 \uB9D0 \uAC83
2667
+ - \uCCB4\uD06C\uBC15\uC2A4(\u2611/\u2610/\u25A0/\u25A1), \uD2B9\uC218\uBB38\uC790(\u203B, \u2460\u2461\u2462, \u3260, \uAC00., \uB098.)\uB97C \uC784\uC758\uB85C \uBC14\uAFB8\uC9C0 \uB9D0 \uAC83
2668
+ - \uC6D0\uBB38\uC5D0 \uC5C6\uB294 \uB0B4\uC6A9\uC744 \uC694\uC57D\xB7\uBCF4\uC644\xB7\uCD94\uB860\uD558\uC9C0 \uB9D0 \uAC83
2669
+ - \`\`\`, \`\`\`markdown \uAC19\uC740 \uCF54\uB4DC\uD39C\uC2A4\uB85C \uAC10\uC2F8\uC9C0 \uB9D0 \uAC83
2670
+ - \uC124\uBA85\xB7\uC8FC\uC11D\xB7\uBA54\uD0C0 \uBB38\uC7A5 \uCD94\uAC00 \uAE08\uC9C0
2671
+
2672
+ [\uBD88\uD655\uC2E4\uD560 \uB54C]
2673
+ - \uAE00\uC790\uAC00 \uD750\uB9BF\uD558\uAC70\uB098 \uD310\uB3C5 \uBD88\uAC00\uD558\uBA74 \uBCF4\uC774\uB294 \uADF8\uB300\uB85C \uB450\uAC70\uB098, \uBD88\uAC00\uD53C\uD558\uBA74 \uD55C \uAE00\uC790\uB9CC ?\uB85C \uD45C\uC2DC
2674
+ - \uD45C \uAD6C\uC870\uAC00 \uBAA8\uD638\uD558\uBA74 \uCEEC\uB7FC \uC218 \uC77C\uCE58 + \uBE48 \uC140 \uD328\uB529 \uD615\uD0DC\uB85C \uCD9C\uB825
2675
+ - \uCD94\uCE21\uBCF4\uB2E4 \uC6D0\uBB38 \uBCF4\uC874\uC744 \uD56D\uC0C1 \uC6B0\uC120`;
2600
2676
  DEFAULT_BATCH_SIZES = {
2601
2677
  gemini: 5,
2602
2678
  claude: 5,
@@ -2842,2457 +2918,143 @@ async function ocrPages(doc, provider, pageFilter, effectivePageCount, warnings,
2842
2918
  if (pageFilter && !pageFilter.has(i)) continue;
2843
2919
  const page = await doc.getPage(i);
2844
2920
  try {
2845
- const imageData = await renderPageToPng(page);
2846
- const result = await provider(imageData, i, "image/png");
2847
- for (const b of ocrResultToBlocks(result, i)) blocks.push(b);
2848
- } catch (err) {
2849
- warnings?.push({
2850
- page: i,
2851
- message: `\uD398\uC774\uC9C0 ${i} OCR \uC2E4\uD328: ${err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`,
2852
- code: "OCR_PAGE_FAILED"
2853
- });
2854
- logger.log({
2855
- level: "warn",
2856
- event: "progress",
2857
- message: "\uD398\uC774\uC9C0 OCR \uC2E4\uD328",
2858
- meta: { page: i },
2859
- error: { message: err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958", name: err instanceof Error ? err.name : "Error" }
2860
- });
2861
- }
2862
- }
2863
- logger.log({ level: "info", event: "done", message: "OCR \uC21C\uCC28 \uCC98\uB9AC \uC644\uB8CC", meta: { blocks: blocks.length } });
2864
- return blocks;
2865
- }
2866
- const pageNumbers = [];
2867
- for (let i = 1; i <= effectivePageCount; i++) {
2868
- if (pageFilter && !pageFilter.has(i)) continue;
2869
- pageNumbers.push(i);
2870
- }
2871
- const tasks = pageNumbers.map((pageNum) => async () => {
2872
- try {
2873
- const page = await doc.getPage(pageNum);
2874
- const imageData = await renderPageToPng(page);
2875
- const result = await provider(imageData, pageNum, "image/png");
2876
- return { pageNum, pageBlocks: ocrResultToBlocks(result, pageNum) };
2877
- } catch (err) {
2878
- warnings?.push({
2879
- page: pageNum,
2880
- message: `\uD398\uC774\uC9C0 ${pageNum} OCR \uC2E4\uD328: ${err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`,
2881
- code: "OCR_PAGE_FAILED"
2882
- });
2883
- logger.log({
2884
- level: "warn",
2885
- event: "progress",
2886
- message: "\uD398\uC774\uC9C0 OCR \uC2E4\uD328(\uBCD1\uB82C)",
2887
- meta: { page: pageNum },
2888
- error: { message: err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958", name: err instanceof Error ? err.name : "Error" }
2889
- });
2890
- return null;
2891
- }
2892
- });
2893
- const taskResults = await runWithConcurrency(tasks, concurrency);
2894
- for (const item of taskResults) {
2895
- if (!item) continue;
2896
- for (const b of item.pageBlocks) blocks.push(b);
2897
- }
2898
- logger.log({ level: "info", event: "done", message: "OCR \uBCD1\uB82C \uCC98\uB9AC \uC644\uB8CC", meta: { blocks: blocks.length, pages: pageNumbers.length } });
2899
- return blocks;
2900
- }
2901
- async function ocrPagesBatch(doc, provider, pageFilter, effectivePageCount, warnings, concurrency = 1, onProgress) {
2902
- const pageNumbers = [];
2903
- for (let i = 1; i <= effectivePageCount; i++) {
2904
- if (pageFilter && !pageFilter.has(i)) continue;
2905
- pageNumbers.push(i);
2906
- }
2907
- const pageBatches = [];
2908
- for (let i = 0; i < pageNumbers.length; i += provider.batchSize) {
2909
- pageBatches.push(pageNumbers.slice(i, i + provider.batchSize));
2910
- }
2911
- let processed = 0;
2912
- const batchTasks = pageBatches.map((batchPageNums, batchIdx) => async () => {
2913
- const pageBlocks = [];
2914
- const batchImages = [];
2915
- try {
2916
- for (const pageNum of batchPageNums) {
2917
- const page = await doc.getPage(pageNum);
2918
- const image = await renderPageToPng(page);
2919
- batchImages.push({ image, pageNum });
2920
- }
2921
- } catch (renderErr) {
2922
- const rendered = new Set(batchImages.map((b) => b.pageNum));
2923
- for (const pageNum of batchPageNums) {
2924
- if (!rendered.has(pageNum)) pageBlocks.push({ pageNum, blocks: [] });
2925
- }
2926
- }
2927
- if (batchImages.length > 0) {
2928
- try {
2929
- const results = await provider.processBatch(batchImages);
2930
- for (const { pageNum } of batchImages) {
2931
- const result = results.get(pageNum);
2932
- pageBlocks.push({
2933
- pageNum,
2934
- blocks: result ? ocrResultToBlocks(result, pageNum) : []
2935
- });
2936
- }
2937
- } catch (err) {
2938
- const range2 = `${batchPageNums[0]}-${batchPageNums[batchPageNums.length - 1]}`;
2939
- warnings?.push({
2940
- message: `\uBC30\uCE58 OCR \uC2E4\uD328 (\uD398\uC774\uC9C0 ${range2}): ${err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"} \u2014 \uB2E8\uC77C \uD398\uC774\uC9C0\uB85C \uC7AC\uC2DC\uB3C4`,
2941
- code: "OCR_PAGE_FAILED"
2942
- });
2943
- for (const { image, pageNum } of batchImages) {
2944
- try {
2945
- const singleResult = await provider.processBatch([{ image, pageNum }]);
2946
- const r = singleResult.get(pageNum);
2947
- pageBlocks.push({ pageNum, blocks: r ? ocrResultToBlocks(r, pageNum) : [] });
2948
- } catch {
2949
- pageBlocks.push({ pageNum, blocks: [] });
2950
- }
2951
- }
2952
- }
2953
- }
2954
- processed += batchPageNums.length;
2955
- onProgress?.(processed, pageNumbers.length);
2956
- return { batchIdx, pageBlocks };
2957
- });
2958
- const effectiveConcurrency = Math.max(1, concurrency);
2959
- const batchResults = await runWithConcurrency(batchTasks, effectiveConcurrency);
2960
- const blocks = [];
2961
- for (const result of batchResults) {
2962
- for (const { blocks: pageBlks } of result.pageBlocks) {
2963
- for (const b of pageBlks) blocks.push(b);
2964
- }
2965
- }
2966
- return blocks;
2967
- }
2968
- async function renderPageToPng(page) {
2969
- const { createCanvas } = await import("@napi-rs/canvas");
2970
- const scale = 2;
2971
- const viewport = page.getViewport({ scale });
2972
- const canvas = createCanvas(Math.floor(viewport.width), Math.floor(viewport.height));
2973
- const ctx = canvas.getContext("2d");
2974
- await page.render({ canvasContext: ctx, viewport }).promise;
2975
- return new Uint8Array(canvas.toBuffer("image/png"));
2976
- }
2977
- var init_provider = __esm({
2978
- "src/ocr/provider.ts"() {
2979
- "use strict";
2980
- init_markdown_to_blocks();
2981
- init_logger();
2982
- }
2983
- });
2984
-
2985
- // ../../../node_modules/async/dist/async.mjs
2986
- var async_exports = {};
2987
- __export(async_exports, {
2988
- all: () => every$1,
2989
- allLimit: () => everyLimit$1,
2990
- allSeries: () => everySeries$1,
2991
- any: () => some$1,
2992
- anyLimit: () => someLimit$1,
2993
- anySeries: () => someSeries$1,
2994
- apply: () => apply,
2995
- applyEach: () => applyEach,
2996
- applyEachSeries: () => applyEachSeries,
2997
- asyncify: () => asyncify,
2998
- auto: () => auto,
2999
- autoInject: () => autoInject,
3000
- cargo: () => cargo$1,
3001
- cargoQueue: () => cargo,
3002
- compose: () => compose,
3003
- concat: () => concat$1,
3004
- concatLimit: () => concatLimit$1,
3005
- concatSeries: () => concatSeries$1,
3006
- constant: () => constant$1,
3007
- default: () => index,
3008
- detect: () => detect$1,
3009
- detectLimit: () => detectLimit$1,
3010
- detectSeries: () => detectSeries$1,
3011
- dir: () => dir,
3012
- doDuring: () => doWhilst$1,
3013
- doUntil: () => doUntil,
3014
- doWhilst: () => doWhilst$1,
3015
- during: () => whilst$1,
3016
- each: () => each,
3017
- eachLimit: () => eachLimit$1,
3018
- eachOf: () => eachOf$1,
3019
- eachOfLimit: () => eachOfLimit$1,
3020
- eachOfSeries: () => eachOfSeries$1,
3021
- eachSeries: () => eachSeries$1,
3022
- ensureAsync: () => ensureAsync,
3023
- every: () => every$1,
3024
- everyLimit: () => everyLimit$1,
3025
- everySeries: () => everySeries$1,
3026
- filter: () => filter$1,
3027
- filterLimit: () => filterLimit$1,
3028
- filterSeries: () => filterSeries$1,
3029
- find: () => detect$1,
3030
- findLimit: () => detectLimit$1,
3031
- findSeries: () => detectSeries$1,
3032
- flatMap: () => concat$1,
3033
- flatMapLimit: () => concatLimit$1,
3034
- flatMapSeries: () => concatSeries$1,
3035
- foldl: () => reduce$1,
3036
- foldr: () => reduceRight,
3037
- forEach: () => each,
3038
- forEachLimit: () => eachLimit$1,
3039
- forEachOf: () => eachOf$1,
3040
- forEachOfLimit: () => eachOfLimit$1,
3041
- forEachOfSeries: () => eachOfSeries$1,
3042
- forEachSeries: () => eachSeries$1,
3043
- forever: () => forever$1,
3044
- groupBy: () => groupBy,
3045
- groupByLimit: () => groupByLimit$1,
3046
- groupBySeries: () => groupBySeries,
3047
- inject: () => reduce$1,
3048
- log: () => log,
3049
- map: () => map$1,
3050
- mapLimit: () => mapLimit$1,
3051
- mapSeries: () => mapSeries$1,
3052
- mapValues: () => mapValues,
3053
- mapValuesLimit: () => mapValuesLimit$1,
3054
- mapValuesSeries: () => mapValuesSeries,
3055
- memoize: () => memoize,
3056
- nextTick: () => nextTick,
3057
- parallel: () => parallel,
3058
- parallelLimit: () => parallelLimit,
3059
- priorityQueue: () => priorityQueue,
3060
- queue: () => queue,
3061
- race: () => race$1,
3062
- reduce: () => reduce$1,
3063
- reduceRight: () => reduceRight,
3064
- reflect: () => reflect,
3065
- reflectAll: () => reflectAll,
3066
- reject: () => reject$1,
3067
- rejectLimit: () => rejectLimit$1,
3068
- rejectSeries: () => rejectSeries$1,
3069
- retry: () => retry,
3070
- retryable: () => retryable,
3071
- select: () => filter$1,
3072
- selectLimit: () => filterLimit$1,
3073
- selectSeries: () => filterSeries$1,
3074
- seq: () => seq,
3075
- series: () => series,
3076
- setImmediate: () => setImmediate$1,
3077
- some: () => some$1,
3078
- someLimit: () => someLimit$1,
3079
- someSeries: () => someSeries$1,
3080
- sortBy: () => sortBy$1,
3081
- timeout: () => timeout,
3082
- times: () => times,
3083
- timesLimit: () => timesLimit,
3084
- timesSeries: () => timesSeries,
3085
- transform: () => transform,
3086
- tryEach: () => tryEach$1,
3087
- unmemoize: () => unmemoize,
3088
- until: () => until,
3089
- waterfall: () => waterfall$1,
3090
- whilst: () => whilst$1,
3091
- wrapSync: () => asyncify
3092
- });
3093
- function apply(fn, ...args) {
3094
- return (...callArgs) => fn(...args, ...callArgs);
3095
- }
3096
- function initialParams(fn) {
3097
- return function(...args) {
3098
- var callback = args.pop();
3099
- return fn.call(this, args, callback);
3100
- };
3101
- }
3102
- function fallback(fn) {
3103
- setTimeout(fn, 0);
3104
- }
3105
- function wrap(defer) {
3106
- return (fn, ...args) => defer(() => fn(...args));
3107
- }
3108
- function asyncify(func) {
3109
- if (isAsync(func)) {
3110
- return function(...args) {
3111
- const callback = args.pop();
3112
- const promise = func.apply(this, args);
3113
- return handlePromise(promise, callback);
3114
- };
3115
- }
3116
- return initialParams(function(args, callback) {
3117
- var result;
3118
- try {
3119
- result = func.apply(this, args);
3120
- } catch (e) {
3121
- return callback(e);
3122
- }
3123
- if (result && typeof result.then === "function") {
3124
- return handlePromise(result, callback);
3125
- } else {
3126
- callback(null, result);
3127
- }
3128
- });
3129
- }
3130
- function handlePromise(promise, callback) {
3131
- return promise.then((value) => {
3132
- invokeCallback(callback, null, value);
3133
- }, (err) => {
3134
- invokeCallback(callback, err && (err instanceof Error || err.message) ? err : new Error(err));
3135
- });
3136
- }
3137
- function invokeCallback(callback, error, value) {
3138
- try {
3139
- callback(error, value);
3140
- } catch (err) {
3141
- setImmediate$1((e) => {
3142
- throw e;
3143
- }, err);
3144
- }
3145
- }
3146
- function isAsync(fn) {
3147
- return fn[Symbol.toStringTag] === "AsyncFunction";
3148
- }
3149
- function isAsyncGenerator(fn) {
3150
- return fn[Symbol.toStringTag] === "AsyncGenerator";
3151
- }
3152
- function isAsyncIterable(obj) {
3153
- return typeof obj[Symbol.asyncIterator] === "function";
3154
- }
3155
- function wrapAsync(asyncFn) {
3156
- if (typeof asyncFn !== "function") throw new Error("expected a function");
3157
- return isAsync(asyncFn) ? asyncify(asyncFn) : asyncFn;
3158
- }
3159
- function awaitify(asyncFn, arity) {
3160
- if (!arity) arity = asyncFn.length;
3161
- if (!arity) throw new Error("arity is undefined");
3162
- function awaitable(...args) {
3163
- if (typeof args[arity - 1] === "function") {
3164
- return asyncFn.apply(this, args);
3165
- }
3166
- return new Promise((resolve4, reject2) => {
3167
- args[arity - 1] = (err, ...cbArgs) => {
3168
- if (err) return reject2(err);
3169
- resolve4(cbArgs.length > 1 ? cbArgs : cbArgs[0]);
3170
- };
3171
- asyncFn.apply(this, args);
3172
- });
3173
- }
3174
- return awaitable;
3175
- }
3176
- function applyEach$1(eachfn) {
3177
- return function applyEach2(fns, ...callArgs) {
3178
- const go = awaitify(function(callback) {
3179
- var that = this;
3180
- return eachfn(fns, (fn, cb) => {
3181
- wrapAsync(fn).apply(that, callArgs.concat(cb));
3182
- }, callback);
3183
- });
3184
- return go;
3185
- };
3186
- }
3187
- function _asyncMap(eachfn, arr, iteratee, callback) {
3188
- arr = arr || [];
3189
- var results = [];
3190
- var counter = 0;
3191
- var _iteratee = wrapAsync(iteratee);
3192
- return eachfn(arr, (value, _, iterCb) => {
3193
- var index2 = counter++;
3194
- _iteratee(value, (err, v) => {
3195
- results[index2] = v;
3196
- iterCb(err);
3197
- });
3198
- }, (err) => {
3199
- callback(err, results);
3200
- });
3201
- }
3202
- function isArrayLike(value) {
3203
- return value && typeof value.length === "number" && value.length >= 0 && value.length % 1 === 0;
3204
- }
3205
- function once(fn) {
3206
- function wrapper(...args) {
3207
- if (fn === null) return;
3208
- var callFn = fn;
3209
- fn = null;
3210
- callFn.apply(this, args);
3211
- }
3212
- Object.assign(wrapper, fn);
3213
- return wrapper;
3214
- }
3215
- function getIterator(coll) {
3216
- return coll[Symbol.iterator] && coll[Symbol.iterator]();
3217
- }
3218
- function createArrayIterator(coll) {
3219
- var i = -1;
3220
- var len = coll.length;
3221
- return function next() {
3222
- return ++i < len ? { value: coll[i], key: i } : null;
3223
- };
3224
- }
3225
- function createES2015Iterator(iterator) {
3226
- var i = -1;
3227
- return function next() {
3228
- var item = iterator.next();
3229
- if (item.done)
3230
- return null;
3231
- i++;
3232
- return { value: item.value, key: i };
3233
- };
3234
- }
3235
- function createObjectIterator(obj) {
3236
- var okeys = obj ? Object.keys(obj) : [];
3237
- var i = -1;
3238
- var len = okeys.length;
3239
- return function next() {
3240
- var key = okeys[++i];
3241
- if (key === "__proto__") {
3242
- return next();
3243
- }
3244
- return i < len ? { value: obj[key], key } : null;
3245
- };
3246
- }
3247
- function createIterator(coll) {
3248
- if (isArrayLike(coll)) {
3249
- return createArrayIterator(coll);
3250
- }
3251
- var iterator = getIterator(coll);
3252
- return iterator ? createES2015Iterator(iterator) : createObjectIterator(coll);
3253
- }
3254
- function onlyOnce(fn) {
3255
- return function(...args) {
3256
- if (fn === null) throw new Error("Callback was already called.");
3257
- var callFn = fn;
3258
- fn = null;
3259
- callFn.apply(this, args);
3260
- };
3261
- }
3262
- function asyncEachOfLimit(generator, limit, iteratee, callback) {
3263
- let done = false;
3264
- let canceled = false;
3265
- let awaiting = false;
3266
- let running = 0;
3267
- let idx = 0;
3268
- function replenish() {
3269
- if (running >= limit || awaiting || done) return;
3270
- awaiting = true;
3271
- generator.next().then(({ value, done: iterDone }) => {
3272
- if (canceled || done) return;
3273
- awaiting = false;
3274
- if (iterDone) {
3275
- done = true;
3276
- if (running <= 0) {
3277
- callback(null);
3278
- }
3279
- return;
3280
- }
3281
- running++;
3282
- iteratee(value, idx, iterateeCallback);
3283
- idx++;
3284
- replenish();
3285
- }).catch(handleError);
3286
- }
3287
- function iterateeCallback(err, result) {
3288
- running -= 1;
3289
- if (canceled) return;
3290
- if (err) return handleError(err);
3291
- if (err === false) {
3292
- done = true;
3293
- canceled = true;
3294
- return;
3295
- }
3296
- if (result === breakLoop || done && running <= 0) {
3297
- done = true;
3298
- return callback(null);
3299
- }
3300
- replenish();
3301
- }
3302
- function handleError(err) {
3303
- if (canceled) return;
3304
- awaiting = false;
3305
- done = true;
3306
- callback(err);
3307
- }
3308
- replenish();
3309
- }
3310
- function eachOfLimit(coll, limit, iteratee, callback) {
3311
- return eachOfLimit$2(limit)(coll, wrapAsync(iteratee), callback);
3312
- }
3313
- function eachOfArrayLike(coll, iteratee, callback) {
3314
- callback = once(callback);
3315
- var index2 = 0, completed = 0, { length } = coll, canceled = false;
3316
- if (length === 0) {
3317
- callback(null);
3318
- }
3319
- function iteratorCallback(err, value) {
3320
- if (err === false) {
3321
- canceled = true;
3322
- }
3323
- if (canceled === true) return;
3324
- if (err) {
3325
- callback(err);
3326
- } else if (++completed === length || value === breakLoop) {
3327
- callback(null);
3328
- }
3329
- }
3330
- for (; index2 < length; index2++) {
3331
- iteratee(coll[index2], index2, onlyOnce(iteratorCallback));
3332
- }
3333
- }
3334
- function eachOfGeneric(coll, iteratee, callback) {
3335
- return eachOfLimit$1(coll, Infinity, iteratee, callback);
3336
- }
3337
- function eachOf(coll, iteratee, callback) {
3338
- var eachOfImplementation = isArrayLike(coll) ? eachOfArrayLike : eachOfGeneric;
3339
- return eachOfImplementation(coll, wrapAsync(iteratee), callback);
3340
- }
3341
- function map(coll, iteratee, callback) {
3342
- return _asyncMap(eachOf$1, coll, iteratee, callback);
3343
- }
3344
- function eachOfSeries(coll, iteratee, callback) {
3345
- return eachOfLimit$1(coll, 1, iteratee, callback);
3346
- }
3347
- function mapSeries(coll, iteratee, callback) {
3348
- return _asyncMap(eachOfSeries$1, coll, iteratee, callback);
3349
- }
3350
- function promiseCallback() {
3351
- let resolve4, reject2;
3352
- function callback(err, ...args) {
3353
- if (err) return reject2(err);
3354
- resolve4(args.length > 1 ? args : args[0]);
3355
- }
3356
- callback[PROMISE_SYMBOL] = new Promise((res, rej) => {
3357
- resolve4 = res, reject2 = rej;
3358
- });
3359
- return callback;
3360
- }
3361
- function auto(tasks, concurrency, callback) {
3362
- if (typeof concurrency !== "number") {
3363
- callback = concurrency;
3364
- concurrency = null;
3365
- }
3366
- callback = once(callback || promiseCallback());
3367
- var numTasks = Object.keys(tasks).length;
3368
- if (!numTasks) {
3369
- return callback(null);
3370
- }
3371
- if (!concurrency) {
3372
- concurrency = numTasks;
3373
- }
3374
- var results = {};
3375
- var runningTasks = 0;
3376
- var canceled = false;
3377
- var hasError = false;
3378
- var listeners = /* @__PURE__ */ Object.create(null);
3379
- var readyTasks = [];
3380
- var readyToCheck = [];
3381
- var uncheckedDependencies = {};
3382
- Object.keys(tasks).forEach((key) => {
3383
- var task = tasks[key];
3384
- if (!Array.isArray(task)) {
3385
- enqueueTask(key, [task]);
3386
- readyToCheck.push(key);
3387
- return;
3388
- }
3389
- var dependencies = task.slice(0, task.length - 1);
3390
- var remainingDependencies = dependencies.length;
3391
- if (remainingDependencies === 0) {
3392
- enqueueTask(key, task);
3393
- readyToCheck.push(key);
3394
- return;
3395
- }
3396
- uncheckedDependencies[key] = remainingDependencies;
3397
- dependencies.forEach((dependencyName) => {
3398
- if (!tasks[dependencyName]) {
3399
- throw new Error("async.auto task `" + key + "` has a non-existent dependency `" + dependencyName + "` in " + dependencies.join(", "));
3400
- }
3401
- addListener(dependencyName, () => {
3402
- remainingDependencies--;
3403
- if (remainingDependencies === 0) {
3404
- enqueueTask(key, task);
3405
- }
3406
- });
3407
- });
3408
- });
3409
- checkForDeadlocks();
3410
- processQueue();
3411
- function enqueueTask(key, task) {
3412
- readyTasks.push(() => runTask(key, task));
3413
- }
3414
- function processQueue() {
3415
- if (canceled) return;
3416
- if (readyTasks.length === 0 && runningTasks === 0) {
3417
- return callback(null, results);
3418
- }
3419
- while (readyTasks.length && runningTasks < concurrency) {
3420
- var run = readyTasks.shift();
3421
- run();
3422
- }
3423
- }
3424
- function addListener(taskName, fn) {
3425
- var taskListeners = listeners[taskName];
3426
- if (!taskListeners) {
3427
- taskListeners = listeners[taskName] = [];
3428
- }
3429
- taskListeners.push(fn);
3430
- }
3431
- function taskComplete(taskName) {
3432
- var taskListeners = listeners[taskName] || [];
3433
- taskListeners.forEach((fn) => fn());
3434
- processQueue();
3435
- }
3436
- function runTask(key, task) {
3437
- if (hasError) return;
3438
- var taskCallback = onlyOnce((err, ...result) => {
3439
- runningTasks--;
3440
- if (err === false) {
3441
- canceled = true;
3442
- return;
3443
- }
3444
- if (result.length < 2) {
3445
- [result] = result;
3446
- }
3447
- if (err) {
3448
- var safeResults = {};
3449
- Object.keys(results).forEach((rkey) => {
3450
- safeResults[rkey] = results[rkey];
3451
- });
3452
- safeResults[key] = result;
3453
- hasError = true;
3454
- listeners = /* @__PURE__ */ Object.create(null);
3455
- if (canceled) return;
3456
- callback(err, safeResults);
3457
- } else {
3458
- results[key] = result;
3459
- taskComplete(key);
3460
- }
3461
- });
3462
- runningTasks++;
3463
- var taskFn = wrapAsync(task[task.length - 1]);
3464
- if (task.length > 1) {
3465
- taskFn(results, taskCallback);
3466
- } else {
3467
- taskFn(taskCallback);
3468
- }
3469
- }
3470
- function checkForDeadlocks() {
3471
- var currentTask;
3472
- var counter = 0;
3473
- while (readyToCheck.length) {
3474
- currentTask = readyToCheck.pop();
3475
- counter++;
3476
- getDependents(currentTask).forEach((dependent) => {
3477
- if (--uncheckedDependencies[dependent] === 0) {
3478
- readyToCheck.push(dependent);
3479
- }
3480
- });
3481
- }
3482
- if (counter !== numTasks) {
3483
- throw new Error(
3484
- "async.auto cannot execute tasks due to a recursive dependency"
3485
- );
3486
- }
3487
- }
3488
- function getDependents(taskName) {
3489
- var result = [];
3490
- Object.keys(tasks).forEach((key) => {
3491
- const task = tasks[key];
3492
- if (Array.isArray(task) && task.indexOf(taskName) >= 0) {
3493
- result.push(key);
3494
- }
3495
- });
3496
- return result;
3497
- }
3498
- return callback[PROMISE_SYMBOL];
3499
- }
3500
- function stripComments(string) {
3501
- let stripped = "";
3502
- let index2 = 0;
3503
- let endBlockComment = string.indexOf("*/");
3504
- while (index2 < string.length) {
3505
- if (string[index2] === "/" && string[index2 + 1] === "/") {
3506
- let endIndex = string.indexOf("\n", index2);
3507
- index2 = endIndex === -1 ? string.length : endIndex;
3508
- } else if (endBlockComment !== -1 && string[index2] === "/" && string[index2 + 1] === "*") {
3509
- let endIndex = string.indexOf("*/", index2);
3510
- if (endIndex !== -1) {
3511
- index2 = endIndex + 2;
3512
- endBlockComment = string.indexOf("*/", index2);
3513
- } else {
3514
- stripped += string[index2];
3515
- index2++;
3516
- }
3517
- } else {
3518
- stripped += string[index2];
3519
- index2++;
3520
- }
3521
- }
3522
- return stripped;
3523
- }
3524
- function parseParams(func) {
3525
- const src = stripComments(func.toString());
3526
- let match = src.match(FN_ARGS);
3527
- if (!match) {
3528
- match = src.match(ARROW_FN_ARGS);
3529
- }
3530
- if (!match) throw new Error("could not parse args in autoInject\nSource:\n" + src);
3531
- let [, args] = match;
3532
- return args.replace(/\s/g, "").split(FN_ARG_SPLIT).map((arg) => arg.replace(FN_ARG, "").trim());
3533
- }
3534
- function autoInject(tasks, callback) {
3535
- var newTasks = {};
3536
- Object.keys(tasks).forEach((key) => {
3537
- var taskFn = tasks[key];
3538
- var params;
3539
- var fnIsAsync = isAsync(taskFn);
3540
- var hasNoDeps = !fnIsAsync && taskFn.length === 1 || fnIsAsync && taskFn.length === 0;
3541
- if (Array.isArray(taskFn)) {
3542
- params = [...taskFn];
3543
- taskFn = params.pop();
3544
- newTasks[key] = params.concat(params.length > 0 ? newTask : taskFn);
3545
- } else if (hasNoDeps) {
3546
- newTasks[key] = taskFn;
3547
- } else {
3548
- params = parseParams(taskFn);
3549
- if (taskFn.length === 0 && !fnIsAsync && params.length === 0) {
3550
- throw new Error("autoInject task functions require explicit parameters.");
3551
- }
3552
- if (!fnIsAsync) params.pop();
3553
- newTasks[key] = params.concat(newTask);
3554
- }
3555
- function newTask(results, taskCb) {
3556
- var newArgs = params.map((name) => results[name]);
3557
- newArgs.push(taskCb);
3558
- wrapAsync(taskFn)(...newArgs);
3559
- }
3560
- });
3561
- return auto(newTasks, callback);
3562
- }
3563
- function setInitial(dll, node) {
3564
- dll.length = 1;
3565
- dll.head = dll.tail = node;
3566
- }
3567
- function queue$1(worker, concurrency, payload) {
3568
- if (concurrency == null) {
3569
- concurrency = 1;
3570
- } else if (concurrency === 0) {
3571
- throw new RangeError("Concurrency must not be zero");
3572
- }
3573
- var _worker = wrapAsync(worker);
3574
- var numRunning = 0;
3575
- var workersList = [];
3576
- const events = {
3577
- error: [],
3578
- drain: [],
3579
- saturated: [],
3580
- unsaturated: [],
3581
- empty: []
3582
- };
3583
- function on(event, handler) {
3584
- events[event].push(handler);
3585
- }
3586
- function once2(event, handler) {
3587
- const handleAndRemove = (...args) => {
3588
- off(event, handleAndRemove);
3589
- handler(...args);
3590
- };
3591
- events[event].push(handleAndRemove);
3592
- }
3593
- function off(event, handler) {
3594
- if (!event) return Object.keys(events).forEach((ev) => events[ev] = []);
3595
- if (!handler) return events[event] = [];
3596
- events[event] = events[event].filter((ev) => ev !== handler);
3597
- }
3598
- function trigger(event, ...args) {
3599
- events[event].forEach((handler) => handler(...args));
3600
- }
3601
- var processingScheduled = false;
3602
- function _insert(data, insertAtFront, rejectOnError, callback) {
3603
- if (callback != null && typeof callback !== "function") {
3604
- throw new Error("task callback must be a function");
3605
- }
3606
- q.started = true;
3607
- var res, rej;
3608
- function promiseCallback2(err, ...args) {
3609
- if (err) return rejectOnError ? rej(err) : res();
3610
- if (args.length <= 1) return res(args[0]);
3611
- res(args);
3612
- }
3613
- var item = q._createTaskItem(
3614
- data,
3615
- rejectOnError ? promiseCallback2 : callback || promiseCallback2
3616
- );
3617
- if (insertAtFront) {
3618
- q._tasks.unshift(item);
3619
- } else {
3620
- q._tasks.push(item);
3621
- }
3622
- if (!processingScheduled) {
3623
- processingScheduled = true;
3624
- setImmediate$1(() => {
3625
- processingScheduled = false;
3626
- q.process();
3627
- });
3628
- }
3629
- if (rejectOnError || !callback) {
3630
- return new Promise((resolve4, reject2) => {
3631
- res = resolve4;
3632
- rej = reject2;
3633
- });
3634
- }
3635
- }
3636
- function _createCB(tasks) {
3637
- return function(err, ...args) {
3638
- numRunning -= 1;
3639
- for (var i = 0, l = tasks.length; i < l; i++) {
3640
- var task = tasks[i];
3641
- var index2 = workersList.indexOf(task);
3642
- if (index2 === 0) {
3643
- workersList.shift();
3644
- } else if (index2 > 0) {
3645
- workersList.splice(index2, 1);
3646
- }
3647
- task.callback(err, ...args);
3648
- if (err != null) {
3649
- trigger("error", err, task.data);
3650
- }
3651
- }
3652
- if (numRunning <= q.concurrency - q.buffer) {
3653
- trigger("unsaturated");
3654
- }
3655
- if (q.idle()) {
3656
- trigger("drain");
3657
- }
3658
- q.process();
3659
- };
3660
- }
3661
- function _maybeDrain(data) {
3662
- if (data.length === 0 && q.idle()) {
3663
- setImmediate$1(() => trigger("drain"));
3664
- return true;
3665
- }
3666
- return false;
3667
- }
3668
- const eventMethod = (name) => (handler) => {
3669
- if (!handler) {
3670
- return new Promise((resolve4, reject2) => {
3671
- once2(name, (err, data) => {
3672
- if (err) return reject2(err);
3673
- resolve4(data);
3674
- });
3675
- });
3676
- }
3677
- off(name);
3678
- on(name, handler);
3679
- };
3680
- var isProcessing = false;
3681
- var q = {
3682
- _tasks: new DLL(),
3683
- _createTaskItem(data, callback) {
3684
- return {
3685
- data,
3686
- callback
3687
- };
3688
- },
3689
- *[Symbol.iterator]() {
3690
- yield* q._tasks[Symbol.iterator]();
3691
- },
3692
- concurrency,
3693
- payload,
3694
- buffer: concurrency / 4,
3695
- started: false,
3696
- paused: false,
3697
- push(data, callback) {
3698
- if (Array.isArray(data)) {
3699
- if (_maybeDrain(data)) return;
3700
- return data.map((datum) => _insert(datum, false, false, callback));
3701
- }
3702
- return _insert(data, false, false, callback);
3703
- },
3704
- pushAsync(data, callback) {
3705
- if (Array.isArray(data)) {
3706
- if (_maybeDrain(data)) return;
3707
- return data.map((datum) => _insert(datum, false, true, callback));
3708
- }
3709
- return _insert(data, false, true, callback);
3710
- },
3711
- kill() {
3712
- off();
3713
- q._tasks.empty();
3714
- },
3715
- unshift(data, callback) {
3716
- if (Array.isArray(data)) {
3717
- if (_maybeDrain(data)) return;
3718
- return data.map((datum) => _insert(datum, true, false, callback));
3719
- }
3720
- return _insert(data, true, false, callback);
3721
- },
3722
- unshiftAsync(data, callback) {
3723
- if (Array.isArray(data)) {
3724
- if (_maybeDrain(data)) return;
3725
- return data.map((datum) => _insert(datum, true, true, callback));
3726
- }
3727
- return _insert(data, true, true, callback);
3728
- },
3729
- remove(testFn) {
3730
- q._tasks.remove(testFn);
3731
- },
3732
- process() {
3733
- if (isProcessing) {
3734
- return;
3735
- }
3736
- isProcessing = true;
3737
- while (!q.paused && numRunning < q.concurrency && q._tasks.length) {
3738
- var tasks = [], data = [];
3739
- var l = q._tasks.length;
3740
- if (q.payload) l = Math.min(l, q.payload);
3741
- for (var i = 0; i < l; i++) {
3742
- var node = q._tasks.shift();
3743
- tasks.push(node);
3744
- workersList.push(node);
3745
- data.push(node.data);
3746
- }
3747
- numRunning += 1;
3748
- if (q._tasks.length === 0) {
3749
- trigger("empty");
3750
- }
3751
- if (numRunning === q.concurrency) {
3752
- trigger("saturated");
3753
- }
3754
- var cb = onlyOnce(_createCB(tasks));
3755
- _worker(data, cb);
3756
- }
3757
- isProcessing = false;
3758
- },
3759
- length() {
3760
- return q._tasks.length;
3761
- },
3762
- running() {
3763
- return numRunning;
3764
- },
3765
- workersList() {
3766
- return workersList;
3767
- },
3768
- idle() {
3769
- return q._tasks.length + numRunning === 0;
3770
- },
3771
- pause() {
3772
- q.paused = true;
3773
- },
3774
- resume() {
3775
- if (q.paused === false) {
3776
- return;
3777
- }
3778
- q.paused = false;
3779
- setImmediate$1(q.process);
3780
- }
3781
- };
3782
- Object.defineProperties(q, {
3783
- saturated: {
3784
- writable: false,
3785
- value: eventMethod("saturated")
3786
- },
3787
- unsaturated: {
3788
- writable: false,
3789
- value: eventMethod("unsaturated")
3790
- },
3791
- empty: {
3792
- writable: false,
3793
- value: eventMethod("empty")
3794
- },
3795
- drain: {
3796
- writable: false,
3797
- value: eventMethod("drain")
3798
- },
3799
- error: {
3800
- writable: false,
3801
- value: eventMethod("error")
3802
- }
3803
- });
3804
- return q;
3805
- }
3806
- function cargo$1(worker, payload) {
3807
- return queue$1(worker, 1, payload);
3808
- }
3809
- function cargo(worker, concurrency, payload) {
3810
- return queue$1(worker, concurrency, payload);
3811
- }
3812
- function reduce(coll, memo, iteratee, callback) {
3813
- callback = once(callback);
3814
- var _iteratee = wrapAsync(iteratee);
3815
- return eachOfSeries$1(coll, (x, i, iterCb) => {
3816
- _iteratee(memo, x, (err, v) => {
3817
- memo = v;
3818
- iterCb(err);
3819
- });
3820
- }, (err) => callback(err, memo));
3821
- }
3822
- function seq(...functions) {
3823
- var _functions = functions.map(wrapAsync);
3824
- return function(...args) {
3825
- var that = this;
3826
- var cb = args[args.length - 1];
3827
- if (typeof cb == "function") {
3828
- args.pop();
3829
- } else {
3830
- cb = promiseCallback();
3831
- }
3832
- reduce$1(
3833
- _functions,
3834
- args,
3835
- (newargs, fn, iterCb) => {
3836
- fn.apply(that, newargs.concat((err, ...nextargs) => {
3837
- iterCb(err, nextargs);
3838
- }));
3839
- },
3840
- (err, results) => cb(err, ...results)
3841
- );
3842
- return cb[PROMISE_SYMBOL];
3843
- };
3844
- }
3845
- function compose(...args) {
3846
- return seq(...args.reverse());
3847
- }
3848
- function mapLimit(coll, limit, iteratee, callback) {
3849
- return _asyncMap(eachOfLimit$2(limit), coll, iteratee, callback);
3850
- }
3851
- function concatLimit(coll, limit, iteratee, callback) {
3852
- var _iteratee = wrapAsync(iteratee);
3853
- return mapLimit$1(coll, limit, (val, iterCb) => {
3854
- _iteratee(val, (err, ...args) => {
3855
- if (err) return iterCb(err);
3856
- return iterCb(err, args);
3857
- });
3858
- }, (err, mapResults) => {
3859
- var result = [];
3860
- for (var i = 0; i < mapResults.length; i++) {
3861
- if (mapResults[i]) {
3862
- result = result.concat(...mapResults[i]);
3863
- }
3864
- }
3865
- return callback(err, result);
3866
- });
3867
- }
3868
- function concat(coll, iteratee, callback) {
3869
- return concatLimit$1(coll, Infinity, iteratee, callback);
3870
- }
3871
- function concatSeries(coll, iteratee, callback) {
3872
- return concatLimit$1(coll, 1, iteratee, callback);
3873
- }
3874
- function constant$1(...args) {
3875
- return function(...ignoredArgs) {
3876
- var callback = ignoredArgs.pop();
3877
- return callback(null, ...args);
3878
- };
3879
- }
3880
- function _createTester(check, getResult) {
3881
- return (eachfn, arr, _iteratee, cb) => {
3882
- var testPassed = false;
3883
- var testResult;
3884
- const iteratee = wrapAsync(_iteratee);
3885
- eachfn(arr, (value, _, callback) => {
3886
- iteratee(value, (err, result) => {
3887
- if (err || err === false) return callback(err);
3888
- if (check(result) && !testResult) {
3889
- testPassed = true;
3890
- testResult = getResult(true, value);
3891
- return callback(null, breakLoop);
3892
- }
3893
- callback();
3894
- });
3895
- }, (err) => {
3896
- if (err) return cb(err);
3897
- cb(null, testPassed ? testResult : getResult(false));
3898
- });
3899
- };
3900
- }
3901
- function detect(coll, iteratee, callback) {
3902
- return _createTester((bool) => bool, (res, item) => item)(eachOf$1, coll, iteratee, callback);
3903
- }
3904
- function detectLimit(coll, limit, iteratee, callback) {
3905
- return _createTester((bool) => bool, (res, item) => item)(eachOfLimit$2(limit), coll, iteratee, callback);
3906
- }
3907
- function detectSeries(coll, iteratee, callback) {
3908
- return _createTester((bool) => bool, (res, item) => item)(eachOfLimit$2(1), coll, iteratee, callback);
3909
- }
3910
- function consoleFunc(name) {
3911
- return (fn, ...args) => wrapAsync(fn)(...args, (err, ...resultArgs) => {
3912
- if (typeof console === "object") {
3913
- if (err) {
3914
- if (console.error) {
3915
- console.error(err);
3916
- }
3917
- } else if (console[name]) {
3918
- resultArgs.forEach((x) => console[name](x));
3919
- }
3920
- }
3921
- });
3922
- }
3923
- function doWhilst(iteratee, test, callback) {
3924
- callback = onlyOnce(callback);
3925
- var _fn = wrapAsync(iteratee);
3926
- var _test = wrapAsync(test);
3927
- var results;
3928
- function next(err, ...args) {
3929
- if (err) return callback(err);
3930
- if (err === false) return;
3931
- results = args;
3932
- _test(...args, check);
3933
- }
3934
- function check(err, truth) {
3935
- if (err) return callback(err);
3936
- if (err === false) return;
3937
- if (!truth) return callback(null, ...results);
3938
- _fn(next);
3939
- }
3940
- return check(null, true);
3941
- }
3942
- function doUntil(iteratee, test, callback) {
3943
- const _test = wrapAsync(test);
3944
- return doWhilst$1(iteratee, (...args) => {
3945
- const cb = args.pop();
3946
- _test(...args, (err, truth) => cb(err, !truth));
3947
- }, callback);
3948
- }
3949
- function _withoutIndex(iteratee) {
3950
- return (value, index2, callback) => iteratee(value, callback);
3951
- }
3952
- function eachLimit$2(coll, iteratee, callback) {
3953
- return eachOf$1(coll, _withoutIndex(wrapAsync(iteratee)), callback);
3954
- }
3955
- function eachLimit(coll, limit, iteratee, callback) {
3956
- return eachOfLimit$2(limit)(coll, _withoutIndex(wrapAsync(iteratee)), callback);
3957
- }
3958
- function eachSeries(coll, iteratee, callback) {
3959
- return eachLimit$1(coll, 1, iteratee, callback);
3960
- }
3961
- function ensureAsync(fn) {
3962
- if (isAsync(fn)) return fn;
3963
- return function(...args) {
3964
- var callback = args.pop();
3965
- var sync = true;
3966
- args.push((...innerArgs) => {
3967
- if (sync) {
3968
- setImmediate$1(() => callback(...innerArgs));
3969
- } else {
3970
- callback(...innerArgs);
3971
- }
3972
- });
3973
- fn.apply(this, args);
3974
- sync = false;
3975
- };
3976
- }
3977
- function every(coll, iteratee, callback) {
3978
- return _createTester((bool) => !bool, (res) => !res)(eachOf$1, coll, iteratee, callback);
3979
- }
3980
- function everyLimit(coll, limit, iteratee, callback) {
3981
- return _createTester((bool) => !bool, (res) => !res)(eachOfLimit$2(limit), coll, iteratee, callback);
3982
- }
3983
- function everySeries(coll, iteratee, callback) {
3984
- return _createTester((bool) => !bool, (res) => !res)(eachOfSeries$1, coll, iteratee, callback);
3985
- }
3986
- function filterArray(eachfn, arr, iteratee, callback) {
3987
- var truthValues = new Array(arr.length);
3988
- eachfn(arr, (x, index2, iterCb) => {
3989
- iteratee(x, (err, v) => {
3990
- truthValues[index2] = !!v;
3991
- iterCb(err);
3992
- });
3993
- }, (err) => {
3994
- if (err) return callback(err);
3995
- var results = [];
3996
- for (var i = 0; i < arr.length; i++) {
3997
- if (truthValues[i]) results.push(arr[i]);
3998
- }
3999
- callback(null, results);
4000
- });
4001
- }
4002
- function filterGeneric(eachfn, coll, iteratee, callback) {
4003
- var results = [];
4004
- eachfn(coll, (x, index2, iterCb) => {
4005
- iteratee(x, (err, v) => {
4006
- if (err) return iterCb(err);
4007
- if (v) {
4008
- results.push({ index: index2, value: x });
4009
- }
4010
- iterCb(err);
4011
- });
4012
- }, (err) => {
4013
- if (err) return callback(err);
4014
- callback(null, results.sort((a, b) => a.index - b.index).map((v) => v.value));
4015
- });
4016
- }
4017
- function _filter(eachfn, coll, iteratee, callback) {
4018
- var filter2 = isArrayLike(coll) ? filterArray : filterGeneric;
4019
- return filter2(eachfn, coll, wrapAsync(iteratee), callback);
4020
- }
4021
- function filter(coll, iteratee, callback) {
4022
- return _filter(eachOf$1, coll, iteratee, callback);
4023
- }
4024
- function filterLimit(coll, limit, iteratee, callback) {
4025
- return _filter(eachOfLimit$2(limit), coll, iteratee, callback);
4026
- }
4027
- function filterSeries(coll, iteratee, callback) {
4028
- return _filter(eachOfSeries$1, coll, iteratee, callback);
4029
- }
4030
- function forever(fn, errback) {
4031
- var done = onlyOnce(errback);
4032
- var task = wrapAsync(ensureAsync(fn));
4033
- function next(err) {
4034
- if (err) return done(err);
4035
- if (err === false) return;
4036
- task(next);
4037
- }
4038
- return next();
4039
- }
4040
- function groupByLimit(coll, limit, iteratee, callback) {
4041
- var _iteratee = wrapAsync(iteratee);
4042
- return mapLimit$1(coll, limit, (val, iterCb) => {
4043
- _iteratee(val, (err, key) => {
4044
- if (err) return iterCb(err);
4045
- return iterCb(err, { key, val });
4046
- });
4047
- }, (err, mapResults) => {
4048
- var result = {};
4049
- var { hasOwnProperty } = Object.prototype;
4050
- for (var i = 0; i < mapResults.length; i++) {
4051
- if (mapResults[i]) {
4052
- var { key } = mapResults[i];
4053
- var { val } = mapResults[i];
4054
- if (hasOwnProperty.call(result, key)) {
4055
- result[key].push(val);
4056
- } else {
4057
- result[key] = [val];
4058
- }
4059
- }
4060
- }
4061
- return callback(err, result);
4062
- });
4063
- }
4064
- function groupBy(coll, iteratee, callback) {
4065
- return groupByLimit$1(coll, Infinity, iteratee, callback);
4066
- }
4067
- function groupBySeries(coll, iteratee, callback) {
4068
- return groupByLimit$1(coll, 1, iteratee, callback);
4069
- }
4070
- function mapValuesLimit(obj, limit, iteratee, callback) {
4071
- callback = once(callback);
4072
- var newObj = {};
4073
- var _iteratee = wrapAsync(iteratee);
4074
- return eachOfLimit$2(limit)(obj, (val, key, next) => {
4075
- _iteratee(val, key, (err, result) => {
4076
- if (err) return next(err);
4077
- newObj[key] = result;
4078
- next(err);
4079
- });
4080
- }, (err) => callback(err, newObj));
4081
- }
4082
- function mapValues(obj, iteratee, callback) {
4083
- return mapValuesLimit$1(obj, Infinity, iteratee, callback);
4084
- }
4085
- function mapValuesSeries(obj, iteratee, callback) {
4086
- return mapValuesLimit$1(obj, 1, iteratee, callback);
4087
- }
4088
- function memoize(fn, hasher = (v) => v) {
4089
- var memo = /* @__PURE__ */ Object.create(null);
4090
- var queues = /* @__PURE__ */ Object.create(null);
4091
- var _fn = wrapAsync(fn);
4092
- var memoized = initialParams((args, callback) => {
4093
- var key = hasher(...args);
4094
- if (key in memo) {
4095
- setImmediate$1(() => callback(null, ...memo[key]));
4096
- } else if (key in queues) {
4097
- queues[key].push(callback);
4098
- } else {
4099
- queues[key] = [callback];
4100
- _fn(...args, (err, ...resultArgs) => {
4101
- if (!err) {
4102
- memo[key] = resultArgs;
4103
- }
4104
- var q = queues[key];
4105
- delete queues[key];
4106
- for (var i = 0, l = q.length; i < l; i++) {
4107
- q[i](err, ...resultArgs);
4108
- }
4109
- });
4110
- }
4111
- });
4112
- memoized.memo = memo;
4113
- memoized.unmemoized = fn;
4114
- return memoized;
4115
- }
4116
- function parallel(tasks, callback) {
4117
- return _parallel(eachOf$1, tasks, callback);
4118
- }
4119
- function parallelLimit(tasks, limit, callback) {
4120
- return _parallel(eachOfLimit$2(limit), tasks, callback);
4121
- }
4122
- function queue(worker, concurrency) {
4123
- var _worker = wrapAsync(worker);
4124
- return queue$1((items, cb) => {
4125
- _worker(items[0], cb);
4126
- }, concurrency, 1);
4127
- }
4128
- function leftChi(i) {
4129
- return (i << 1) + 1;
4130
- }
4131
- function parent(i) {
4132
- return (i + 1 >> 1) - 1;
4133
- }
4134
- function smaller(x, y) {
4135
- if (x.priority !== y.priority) {
4136
- return x.priority < y.priority;
4137
- } else {
4138
- return x.pushCount < y.pushCount;
4139
- }
4140
- }
4141
- function priorityQueue(worker, concurrency) {
4142
- var q = queue(worker, concurrency);
4143
- var {
4144
- push,
4145
- pushAsync
4146
- } = q;
4147
- q._tasks = new Heap();
4148
- q._createTaskItem = ({ data, priority }, callback) => {
4149
- return {
4150
- data,
4151
- priority,
4152
- callback
4153
- };
4154
- };
4155
- function createDataItems(tasks, priority) {
4156
- if (!Array.isArray(tasks)) {
4157
- return { data: tasks, priority };
4158
- }
4159
- return tasks.map((data) => {
4160
- return { data, priority };
4161
- });
4162
- }
4163
- q.push = function(data, priority = 0, callback) {
4164
- return push(createDataItems(data, priority), callback);
4165
- };
4166
- q.pushAsync = function(data, priority = 0, callback) {
4167
- return pushAsync(createDataItems(data, priority), callback);
4168
- };
4169
- delete q.unshift;
4170
- delete q.unshiftAsync;
4171
- return q;
4172
- }
4173
- function race(tasks, callback) {
4174
- callback = once(callback);
4175
- if (!Array.isArray(tasks)) return callback(new TypeError("First argument to race must be an array of functions"));
4176
- if (!tasks.length) return callback();
4177
- for (var i = 0, l = tasks.length; i < l; i++) {
4178
- wrapAsync(tasks[i])(callback);
4179
- }
4180
- }
4181
- function reduceRight(array, memo, iteratee, callback) {
4182
- var reversed = [...array].reverse();
4183
- return reduce$1(reversed, memo, iteratee, callback);
4184
- }
4185
- function reflect(fn) {
4186
- var _fn = wrapAsync(fn);
4187
- return initialParams(function reflectOn(args, reflectCallback) {
4188
- args.push((error, ...cbArgs) => {
4189
- let retVal = {};
4190
- if (error) {
4191
- retVal.error = error;
4192
- }
4193
- if (cbArgs.length > 0) {
4194
- var value = cbArgs;
4195
- if (cbArgs.length <= 1) {
4196
- [value] = cbArgs;
4197
- }
4198
- retVal.value = value;
4199
- }
4200
- reflectCallback(null, retVal);
4201
- });
4202
- return _fn.apply(this, args);
4203
- });
4204
- }
4205
- function reflectAll(tasks) {
4206
- var results;
4207
- if (Array.isArray(tasks)) {
4208
- results = tasks.map(reflect);
4209
- } else {
4210
- results = {};
4211
- Object.keys(tasks).forEach((key) => {
4212
- results[key] = reflect.call(this, tasks[key]);
4213
- });
4214
- }
4215
- return results;
4216
- }
4217
- function reject$2(eachfn, arr, _iteratee, callback) {
4218
- const iteratee = wrapAsync(_iteratee);
4219
- return _filter(eachfn, arr, (value, cb) => {
4220
- iteratee(value, (err, v) => {
4221
- cb(err, !v);
4222
- });
4223
- }, callback);
4224
- }
4225
- function reject(coll, iteratee, callback) {
4226
- return reject$2(eachOf$1, coll, iteratee, callback);
4227
- }
4228
- function rejectLimit(coll, limit, iteratee, callback) {
4229
- return reject$2(eachOfLimit$2(limit), coll, iteratee, callback);
4230
- }
4231
- function rejectSeries(coll, iteratee, callback) {
4232
- return reject$2(eachOfSeries$1, coll, iteratee, callback);
4233
- }
4234
- function constant(value) {
4235
- return function() {
4236
- return value;
4237
- };
4238
- }
4239
- function retry(opts, task, callback) {
4240
- var options = {
4241
- times: DEFAULT_TIMES,
4242
- intervalFunc: constant(DEFAULT_INTERVAL)
4243
- };
4244
- if (arguments.length < 3 && typeof opts === "function") {
4245
- callback = task || promiseCallback();
4246
- task = opts;
4247
- } else {
4248
- parseTimes(options, opts);
4249
- callback = callback || promiseCallback();
4250
- }
4251
- if (typeof task !== "function") {
4252
- throw new Error("Invalid arguments for async.retry");
4253
- }
4254
- var _task = wrapAsync(task);
4255
- var attempt = 1;
4256
- function retryAttempt() {
4257
- _task((err, ...args) => {
4258
- if (err === false) return;
4259
- if (err && attempt++ < options.times && (typeof options.errorFilter != "function" || options.errorFilter(err))) {
4260
- setTimeout(retryAttempt, options.intervalFunc(attempt - 1));
4261
- } else {
4262
- callback(err, ...args);
4263
- }
4264
- });
4265
- }
4266
- retryAttempt();
4267
- return callback[PROMISE_SYMBOL];
4268
- }
4269
- function parseTimes(acc, t) {
4270
- if (typeof t === "object") {
4271
- acc.times = +t.times || DEFAULT_TIMES;
4272
- acc.intervalFunc = typeof t.interval === "function" ? t.interval : constant(+t.interval || DEFAULT_INTERVAL);
4273
- acc.errorFilter = t.errorFilter;
4274
- } else if (typeof t === "number" || typeof t === "string") {
4275
- acc.times = +t || DEFAULT_TIMES;
4276
- } else {
4277
- throw new Error("Invalid arguments for async.retry");
4278
- }
4279
- }
4280
- function retryable(opts, task) {
4281
- if (!task) {
4282
- task = opts;
4283
- opts = null;
4284
- }
4285
- let arity = opts && opts.arity || task.length;
4286
- if (isAsync(task)) {
4287
- arity += 1;
4288
- }
4289
- var _task = wrapAsync(task);
4290
- return initialParams((args, callback) => {
4291
- if (args.length < arity - 1 || callback == null) {
4292
- args.push(callback);
4293
- callback = promiseCallback();
4294
- }
4295
- function taskFn(cb) {
4296
- _task(...args, cb);
4297
- }
4298
- if (opts) retry(opts, taskFn, callback);
4299
- else retry(taskFn, callback);
4300
- return callback[PROMISE_SYMBOL];
4301
- });
4302
- }
4303
- function series(tasks, callback) {
4304
- return _parallel(eachOfSeries$1, tasks, callback);
4305
- }
4306
- function some(coll, iteratee, callback) {
4307
- return _createTester(Boolean, (res) => res)(eachOf$1, coll, iteratee, callback);
4308
- }
4309
- function someLimit(coll, limit, iteratee, callback) {
4310
- return _createTester(Boolean, (res) => res)(eachOfLimit$2(limit), coll, iteratee, callback);
4311
- }
4312
- function someSeries(coll, iteratee, callback) {
4313
- return _createTester(Boolean, (res) => res)(eachOfSeries$1, coll, iteratee, callback);
4314
- }
4315
- function sortBy(coll, iteratee, callback) {
4316
- var _iteratee = wrapAsync(iteratee);
4317
- return map$1(coll, (x, iterCb) => {
4318
- _iteratee(x, (err, criteria) => {
4319
- if (err) return iterCb(err);
4320
- iterCb(err, { value: x, criteria });
4321
- });
4322
- }, (err, results) => {
4323
- if (err) return callback(err);
4324
- callback(null, results.sort(comparator).map((v) => v.value));
4325
- });
4326
- function comparator(left, right) {
4327
- var a = left.criteria, b = right.criteria;
4328
- return a < b ? -1 : a > b ? 1 : 0;
4329
- }
4330
- }
4331
- function timeout(asyncFn, milliseconds, info) {
4332
- var fn = wrapAsync(asyncFn);
4333
- return initialParams((args, callback) => {
4334
- var timedOut = false;
4335
- var timer;
4336
- function timeoutCallback() {
4337
- var name = asyncFn.name || "anonymous";
4338
- var error = new Error('Callback function "' + name + '" timed out.');
4339
- error.code = "ETIMEDOUT";
4340
- if (info) {
4341
- error.info = info;
4342
- }
4343
- timedOut = true;
4344
- callback(error);
4345
- }
4346
- args.push((...cbArgs) => {
4347
- if (!timedOut) {
4348
- callback(...cbArgs);
4349
- clearTimeout(timer);
4350
- }
4351
- });
4352
- timer = setTimeout(timeoutCallback, milliseconds);
4353
- fn(...args);
4354
- });
4355
- }
4356
- function range(size) {
4357
- var result = Array(size);
4358
- while (size--) {
4359
- result[size] = size;
4360
- }
4361
- return result;
4362
- }
4363
- function timesLimit(count, limit, iteratee, callback) {
4364
- var _iteratee = wrapAsync(iteratee);
4365
- return mapLimit$1(range(count), limit, _iteratee, callback);
4366
- }
4367
- function times(n, iteratee, callback) {
4368
- return timesLimit(n, Infinity, iteratee, callback);
4369
- }
4370
- function timesSeries(n, iteratee, callback) {
4371
- return timesLimit(n, 1, iteratee, callback);
4372
- }
4373
- function transform(coll, accumulator, iteratee, callback) {
4374
- if (arguments.length <= 3 && typeof accumulator === "function") {
4375
- callback = iteratee;
4376
- iteratee = accumulator;
4377
- accumulator = Array.isArray(coll) ? [] : {};
4378
- }
4379
- callback = once(callback || promiseCallback());
4380
- var _iteratee = wrapAsync(iteratee);
4381
- eachOf$1(coll, (v, k, cb) => {
4382
- _iteratee(accumulator, v, k, cb);
4383
- }, (err) => callback(err, accumulator));
4384
- return callback[PROMISE_SYMBOL];
4385
- }
4386
- function tryEach(tasks, callback) {
4387
- var error = null;
4388
- var result;
4389
- return eachSeries$1(tasks, (task, taskCb) => {
4390
- wrapAsync(task)((err, ...args) => {
4391
- if (err === false) return taskCb(err);
4392
- if (args.length < 2) {
4393
- [result] = args;
4394
- } else {
4395
- result = args;
4396
- }
4397
- error = err;
4398
- taskCb(err ? null : {});
4399
- });
4400
- }, () => callback(error, result));
4401
- }
4402
- function unmemoize(fn) {
4403
- return (...args) => {
4404
- return (fn.unmemoized || fn)(...args);
4405
- };
4406
- }
4407
- function whilst(test, iteratee, callback) {
4408
- callback = onlyOnce(callback);
4409
- var _fn = wrapAsync(iteratee);
4410
- var _test = wrapAsync(test);
4411
- var results = [];
4412
- function next(err, ...rest) {
4413
- if (err) return callback(err);
4414
- results = rest;
4415
- if (err === false) return;
4416
- _test(check);
4417
- }
4418
- function check(err, truth) {
4419
- if (err) return callback(err);
4420
- if (err === false) return;
4421
- if (!truth) return callback(null, ...results);
4422
- _fn(next);
4423
- }
4424
- return _test(check);
4425
- }
4426
- function until(test, iteratee, callback) {
4427
- const _test = wrapAsync(test);
4428
- return whilst$1((cb) => _test((err, truth) => cb(err, !truth)), iteratee, callback);
4429
- }
4430
- function waterfall(tasks, callback) {
4431
- callback = once(callback);
4432
- if (!Array.isArray(tasks)) return callback(new Error("First argument to waterfall must be an array of functions"));
4433
- if (!tasks.length) return callback();
4434
- var taskIndex = 0;
4435
- function nextTask(args) {
4436
- var task = wrapAsync(tasks[taskIndex++]);
4437
- task(...args, onlyOnce(next));
4438
- }
4439
- function next(err, ...args) {
4440
- if (err === false) return;
4441
- if (err || taskIndex === tasks.length) {
4442
- return callback(err, ...args);
4443
- }
4444
- nextTask(args);
4445
- }
4446
- nextTask([]);
4447
- }
4448
- var hasQueueMicrotask, hasSetImmediate, hasNextTick, _defer$1, setImmediate$1, breakLoop, eachOfLimit$2, eachOfLimit$1, eachOf$1, map$1, applyEach, eachOfSeries$1, mapSeries$1, applyEachSeries, PROMISE_SYMBOL, FN_ARGS, ARROW_FN_ARGS, FN_ARG_SPLIT, FN_ARG, DLL, reduce$1, mapLimit$1, concatLimit$1, concat$1, concatSeries$1, detect$1, detectLimit$1, detectSeries$1, dir, doWhilst$1, each, eachLimit$1, eachSeries$1, every$1, everyLimit$1, everySeries$1, filter$1, filterLimit$1, filterSeries$1, forever$1, groupByLimit$1, log, mapValuesLimit$1, _defer, nextTick, _parallel, Heap, race$1, reject$1, rejectLimit$1, rejectSeries$1, DEFAULT_TIMES, DEFAULT_INTERVAL, some$1, someLimit$1, someSeries$1, sortBy$1, tryEach$1, whilst$1, waterfall$1, index;
4449
- var init_async = __esm({
4450
- "../../../node_modules/async/dist/async.mjs"() {
4451
- "use strict";
4452
- hasQueueMicrotask = typeof queueMicrotask === "function" && queueMicrotask;
4453
- hasSetImmediate = typeof setImmediate === "function" && setImmediate;
4454
- hasNextTick = typeof process === "object" && typeof process.nextTick === "function";
4455
- if (hasQueueMicrotask) {
4456
- _defer$1 = queueMicrotask;
4457
- } else if (hasSetImmediate) {
4458
- _defer$1 = setImmediate;
4459
- } else if (hasNextTick) {
4460
- _defer$1 = process.nextTick;
4461
- } else {
4462
- _defer$1 = fallback;
4463
- }
4464
- setImmediate$1 = wrap(_defer$1);
4465
- breakLoop = {};
4466
- eachOfLimit$2 = (limit) => {
4467
- return (obj, iteratee, callback) => {
4468
- callback = once(callback);
4469
- if (limit <= 0) {
4470
- throw new RangeError("concurrency limit cannot be less than 1");
4471
- }
4472
- if (!obj) {
4473
- return callback(null);
4474
- }
4475
- if (isAsyncGenerator(obj)) {
4476
- return asyncEachOfLimit(obj, limit, iteratee, callback);
4477
- }
4478
- if (isAsyncIterable(obj)) {
4479
- return asyncEachOfLimit(obj[Symbol.asyncIterator](), limit, iteratee, callback);
4480
- }
4481
- var nextElem = createIterator(obj);
4482
- var done = false;
4483
- var canceled = false;
4484
- var running = 0;
4485
- var looping = false;
4486
- function iterateeCallback(err, value) {
4487
- if (canceled) return;
4488
- running -= 1;
4489
- if (err) {
4490
- done = true;
4491
- callback(err);
4492
- } else if (err === false) {
4493
- done = true;
4494
- canceled = true;
4495
- } else if (value === breakLoop || done && running <= 0) {
4496
- done = true;
4497
- return callback(null);
4498
- } else if (!looping) {
4499
- replenish();
4500
- }
4501
- }
4502
- function replenish() {
4503
- looping = true;
4504
- while (running < limit && !done) {
4505
- var elem = nextElem();
4506
- if (elem === null) {
4507
- done = true;
4508
- if (running <= 0) {
4509
- callback(null);
4510
- }
4511
- return;
4512
- }
4513
- running += 1;
4514
- iteratee(elem.value, elem.key, onlyOnce(iterateeCallback));
4515
- }
4516
- looping = false;
4517
- }
4518
- replenish();
4519
- };
4520
- };
4521
- eachOfLimit$1 = awaitify(eachOfLimit, 4);
4522
- eachOf$1 = awaitify(eachOf, 3);
4523
- map$1 = awaitify(map, 3);
4524
- applyEach = applyEach$1(map$1);
4525
- eachOfSeries$1 = awaitify(eachOfSeries, 3);
4526
- mapSeries$1 = awaitify(mapSeries, 3);
4527
- applyEachSeries = applyEach$1(mapSeries$1);
4528
- PROMISE_SYMBOL = /* @__PURE__ */ Symbol("promiseCallback");
4529
- FN_ARGS = /^(?:async\s)?(?:function)?\s*(?:\w+\s*)?\(([^)]+)\)(?:\s*{)/;
4530
- ARROW_FN_ARGS = /^(?:async\s)?\s*(?:\(\s*)?((?:[^)=\s]\s*)*)(?:\)\s*)?=>/;
4531
- FN_ARG_SPLIT = /,/;
4532
- FN_ARG = /(=.+)?(\s*)$/;
4533
- DLL = class {
4534
- constructor() {
4535
- this.head = this.tail = null;
4536
- this.length = 0;
4537
- }
4538
- removeLink(node) {
4539
- if (node.prev) node.prev.next = node.next;
4540
- else this.head = node.next;
4541
- if (node.next) node.next.prev = node.prev;
4542
- else this.tail = node.prev;
4543
- node.prev = node.next = null;
4544
- this.length -= 1;
4545
- return node;
4546
- }
4547
- empty() {
4548
- while (this.head) this.shift();
4549
- return this;
4550
- }
4551
- insertAfter(node, newNode) {
4552
- newNode.prev = node;
4553
- newNode.next = node.next;
4554
- if (node.next) node.next.prev = newNode;
4555
- else this.tail = newNode;
4556
- node.next = newNode;
4557
- this.length += 1;
4558
- }
4559
- insertBefore(node, newNode) {
4560
- newNode.prev = node.prev;
4561
- newNode.next = node;
4562
- if (node.prev) node.prev.next = newNode;
4563
- else this.head = newNode;
4564
- node.prev = newNode;
4565
- this.length += 1;
4566
- }
4567
- unshift(node) {
4568
- if (this.head) this.insertBefore(this.head, node);
4569
- else setInitial(this, node);
4570
- }
4571
- push(node) {
4572
- if (this.tail) this.insertAfter(this.tail, node);
4573
- else setInitial(this, node);
4574
- }
4575
- shift() {
4576
- return this.head && this.removeLink(this.head);
4577
- }
4578
- pop() {
4579
- return this.tail && this.removeLink(this.tail);
4580
- }
4581
- toArray() {
4582
- return [...this];
4583
- }
4584
- *[Symbol.iterator]() {
4585
- var cur = this.head;
4586
- while (cur) {
4587
- yield cur.data;
4588
- cur = cur.next;
4589
- }
4590
- }
4591
- remove(testFn) {
4592
- var curr = this.head;
4593
- while (curr) {
4594
- var { next } = curr;
4595
- if (testFn(curr)) {
4596
- this.removeLink(curr);
4597
- }
4598
- curr = next;
4599
- }
4600
- return this;
4601
- }
4602
- };
4603
- reduce$1 = awaitify(reduce, 4);
4604
- mapLimit$1 = awaitify(mapLimit, 4);
4605
- concatLimit$1 = awaitify(concatLimit, 4);
4606
- concat$1 = awaitify(concat, 3);
4607
- concatSeries$1 = awaitify(concatSeries, 3);
4608
- detect$1 = awaitify(detect, 3);
4609
- detectLimit$1 = awaitify(detectLimit, 4);
4610
- detectSeries$1 = awaitify(detectSeries, 3);
4611
- dir = consoleFunc("dir");
4612
- doWhilst$1 = awaitify(doWhilst, 3);
4613
- each = awaitify(eachLimit$2, 3);
4614
- eachLimit$1 = awaitify(eachLimit, 4);
4615
- eachSeries$1 = awaitify(eachSeries, 3);
4616
- every$1 = awaitify(every, 3);
4617
- everyLimit$1 = awaitify(everyLimit, 4);
4618
- everySeries$1 = awaitify(everySeries, 3);
4619
- filter$1 = awaitify(filter, 3);
4620
- filterLimit$1 = awaitify(filterLimit, 4);
4621
- filterSeries$1 = awaitify(filterSeries, 3);
4622
- forever$1 = awaitify(forever, 2);
4623
- groupByLimit$1 = awaitify(groupByLimit, 4);
4624
- log = consoleFunc("log");
4625
- mapValuesLimit$1 = awaitify(mapValuesLimit, 4);
4626
- if (hasNextTick) {
4627
- _defer = process.nextTick;
4628
- } else if (hasSetImmediate) {
4629
- _defer = setImmediate;
4630
- } else {
4631
- _defer = fallback;
4632
- }
4633
- nextTick = wrap(_defer);
4634
- _parallel = awaitify((eachfn, tasks, callback) => {
4635
- var results = isArrayLike(tasks) ? [] : {};
4636
- eachfn(tasks, (task, key, taskCb) => {
4637
- wrapAsync(task)((err, ...result) => {
4638
- if (result.length < 2) {
4639
- [result] = result;
4640
- }
4641
- results[key] = result;
4642
- taskCb(err);
4643
- });
4644
- }, (err) => callback(err, results));
4645
- }, 3);
4646
- Heap = class {
4647
- constructor() {
4648
- this.heap = [];
4649
- this.pushCount = Number.MIN_SAFE_INTEGER;
4650
- }
4651
- get length() {
4652
- return this.heap.length;
4653
- }
4654
- empty() {
4655
- this.heap = [];
4656
- return this;
4657
- }
4658
- percUp(index2) {
4659
- let p;
4660
- while (index2 > 0 && smaller(this.heap[index2], this.heap[p = parent(index2)])) {
4661
- let t = this.heap[index2];
4662
- this.heap[index2] = this.heap[p];
4663
- this.heap[p] = t;
4664
- index2 = p;
4665
- }
4666
- }
4667
- percDown(index2) {
4668
- let l;
4669
- while ((l = leftChi(index2)) < this.heap.length) {
4670
- if (l + 1 < this.heap.length && smaller(this.heap[l + 1], this.heap[l])) {
4671
- l = l + 1;
4672
- }
4673
- if (smaller(this.heap[index2], this.heap[l])) {
4674
- break;
4675
- }
4676
- let t = this.heap[index2];
4677
- this.heap[index2] = this.heap[l];
4678
- this.heap[l] = t;
4679
- index2 = l;
4680
- }
4681
- }
4682
- push(node) {
4683
- node.pushCount = ++this.pushCount;
4684
- this.heap.push(node);
4685
- this.percUp(this.heap.length - 1);
4686
- }
4687
- unshift(node) {
4688
- return this.heap.push(node);
4689
- }
4690
- shift() {
4691
- let [top] = this.heap;
4692
- this.heap[0] = this.heap[this.heap.length - 1];
4693
- this.heap.pop();
4694
- this.percDown(0);
4695
- return top;
4696
- }
4697
- toArray() {
4698
- return [...this];
4699
- }
4700
- *[Symbol.iterator]() {
4701
- for (let i = 0; i < this.heap.length; i++) {
4702
- yield this.heap[i].data;
4703
- }
4704
- }
4705
- remove(testFn) {
4706
- let j = 0;
4707
- for (let i = 0; i < this.heap.length; i++) {
4708
- if (!testFn(this.heap[i])) {
4709
- this.heap[j] = this.heap[i];
4710
- j++;
4711
- }
4712
- }
4713
- this.heap.splice(j);
4714
- for (let i = parent(this.heap.length - 1); i >= 0; i--) {
4715
- this.percDown(i);
4716
- }
4717
- return this;
4718
- }
4719
- };
4720
- race$1 = awaitify(race, 2);
4721
- reject$1 = awaitify(reject, 3);
4722
- rejectLimit$1 = awaitify(rejectLimit, 4);
4723
- rejectSeries$1 = awaitify(rejectSeries, 3);
4724
- DEFAULT_TIMES = 5;
4725
- DEFAULT_INTERVAL = 0;
4726
- some$1 = awaitify(some, 3);
4727
- someLimit$1 = awaitify(someLimit, 4);
4728
- someSeries$1 = awaitify(someSeries, 3);
4729
- sortBy$1 = awaitify(sortBy, 3);
4730
- tryEach$1 = awaitify(tryEach);
4731
- whilst$1 = awaitify(whilst, 3);
4732
- waterfall$1 = awaitify(waterfall);
4733
- index = {
4734
- apply,
4735
- applyEach,
4736
- applyEachSeries,
4737
- asyncify,
4738
- auto,
4739
- autoInject,
4740
- cargo: cargo$1,
4741
- cargoQueue: cargo,
4742
- compose,
4743
- concat: concat$1,
4744
- concatLimit: concatLimit$1,
4745
- concatSeries: concatSeries$1,
4746
- constant: constant$1,
4747
- detect: detect$1,
4748
- detectLimit: detectLimit$1,
4749
- detectSeries: detectSeries$1,
4750
- dir,
4751
- doUntil,
4752
- doWhilst: doWhilst$1,
4753
- each,
4754
- eachLimit: eachLimit$1,
4755
- eachOf: eachOf$1,
4756
- eachOfLimit: eachOfLimit$1,
4757
- eachOfSeries: eachOfSeries$1,
4758
- eachSeries: eachSeries$1,
4759
- ensureAsync,
4760
- every: every$1,
4761
- everyLimit: everyLimit$1,
4762
- everySeries: everySeries$1,
4763
- filter: filter$1,
4764
- filterLimit: filterLimit$1,
4765
- filterSeries: filterSeries$1,
4766
- forever: forever$1,
4767
- groupBy,
4768
- groupByLimit: groupByLimit$1,
4769
- groupBySeries,
4770
- log,
4771
- map: map$1,
4772
- mapLimit: mapLimit$1,
4773
- mapSeries: mapSeries$1,
4774
- mapValues,
4775
- mapValuesLimit: mapValuesLimit$1,
4776
- mapValuesSeries,
4777
- memoize,
4778
- nextTick,
4779
- parallel,
4780
- parallelLimit,
4781
- priorityQueue,
4782
- queue,
4783
- race: race$1,
4784
- reduce: reduce$1,
4785
- reduceRight,
4786
- reflect,
4787
- reflectAll,
4788
- reject: reject$1,
4789
- rejectLimit: rejectLimit$1,
4790
- rejectSeries: rejectSeries$1,
4791
- retry,
4792
- retryable,
4793
- seq,
4794
- series,
4795
- setImmediate: setImmediate$1,
4796
- some: some$1,
4797
- someLimit: someLimit$1,
4798
- someSeries: someSeries$1,
4799
- sortBy: sortBy$1,
4800
- timeout,
4801
- times,
4802
- timesLimit,
4803
- timesSeries,
4804
- transform,
4805
- tryEach: tryEach$1,
4806
- unmemoize,
4807
- until,
4808
- waterfall: waterfall$1,
4809
- whilst: whilst$1,
4810
- // aliases
4811
- all: every$1,
4812
- allLimit: everyLimit$1,
4813
- allSeries: everySeries$1,
4814
- any: some$1,
4815
- anyLimit: someLimit$1,
4816
- anySeries: someSeries$1,
4817
- find: detect$1,
4818
- findLimit: detectLimit$1,
4819
- findSeries: detectSeries$1,
4820
- flatMap: concat$1,
4821
- flatMapLimit: concatLimit$1,
4822
- flatMapSeries: concatSeries$1,
4823
- forEach: each,
4824
- forEachSeries: eachSeries$1,
4825
- forEachLimit: eachLimit$1,
4826
- forEachOf: eachOf$1,
4827
- forEachOfSeries: eachOfSeries$1,
4828
- forEachOfLimit: eachOfLimit$1,
4829
- inject: reduce$1,
4830
- foldl: reduce$1,
4831
- foldr: reduceRight,
4832
- select: filter$1,
4833
- selectLimit: filterLimit$1,
4834
- selectSeries: filterSeries$1,
4835
- wrapSync: asyncify,
4836
- during: whilst$1,
4837
- doDuring: doWhilst$1
4838
- };
4839
- }
4840
- });
4841
-
4842
- // ../../../node_modules/tmp/lib/tmp.js
4843
- var require_tmp = __commonJS({
4844
- "../../../node_modules/tmp/lib/tmp.js"(exports, module) {
4845
- "use strict";
4846
- var fs = __require("fs");
4847
- var os = __require("os");
4848
- var path = __require("path");
4849
- var crypto = __require("crypto");
4850
- var _c = { fs: fs.constants, os: os.constants };
4851
- var RANDOM_CHARS = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
4852
- var TEMPLATE_PATTERN = /XXXXXX/;
4853
- var DEFAULT_TRIES = 3;
4854
- var CREATE_FLAGS = (_c.O_CREAT || _c.fs.O_CREAT) | (_c.O_EXCL || _c.fs.O_EXCL) | (_c.O_RDWR || _c.fs.O_RDWR);
4855
- var IS_WIN32 = os.platform() === "win32";
4856
- var EBADF = _c.EBADF || _c.os.errno.EBADF;
4857
- var ENOENT = _c.ENOENT || _c.os.errno.ENOENT;
4858
- var DIR_MODE = 448;
4859
- var FILE_MODE = 384;
4860
- var EXIT = "exit";
4861
- var _removeObjects = [];
4862
- var FN_RMDIR_SYNC = fs.rmdirSync.bind(fs);
4863
- var _gracefulCleanup = false;
4864
- function rimraf(dirPath, callback) {
4865
- return fs.rm(dirPath, { recursive: true }, callback);
4866
- }
4867
- function FN_RIMRAF_SYNC(dirPath) {
4868
- return fs.rmSync(dirPath, { recursive: true });
4869
- }
4870
- function tmpName(options, callback) {
4871
- const args = _parseArguments(options, callback), opts = args[0], cb = args[1];
4872
- _assertAndSanitizeOptions(opts, function(err, sanitizedOptions) {
4873
- if (err) return cb(err);
4874
- let tries = sanitizedOptions.tries;
4875
- (function _getUniqueName() {
4876
- try {
4877
- const name = _generateTmpName(sanitizedOptions);
4878
- fs.stat(name, function(err2) {
4879
- if (!err2) {
4880
- if (tries-- > 0) return _getUniqueName();
4881
- return cb(new Error("Could not get a unique tmp filename, max tries reached " + name));
4882
- }
4883
- cb(null, name);
4884
- });
4885
- } catch (err2) {
4886
- cb(err2);
4887
- }
4888
- })();
4889
- });
4890
- }
4891
- function tmpNameSync(options) {
4892
- const args = _parseArguments(options), opts = args[0];
4893
- const sanitizedOptions = _assertAndSanitizeOptionsSync(opts);
4894
- let tries = sanitizedOptions.tries;
4895
- do {
4896
- const name = _generateTmpName(sanitizedOptions);
4897
- try {
4898
- fs.statSync(name);
4899
- } catch (e) {
4900
- return name;
4901
- }
4902
- } while (tries-- > 0);
4903
- throw new Error("Could not get a unique tmp filename, max tries reached");
4904
- }
4905
- function file(options, callback) {
4906
- const args = _parseArguments(options, callback), opts = args[0], cb = args[1];
4907
- tmpName(opts, function _tmpNameCreated(err, name) {
4908
- if (err) return cb(err);
4909
- fs.open(name, CREATE_FLAGS, opts.mode || FILE_MODE, function _fileCreated(err2, fd) {
4910
- if (err2) return cb(err2);
4911
- if (opts.discardDescriptor) {
4912
- return fs.close(fd, function _discardCallback(possibleErr) {
4913
- return cb(possibleErr, name, void 0, _prepareTmpFileRemoveCallback(name, -1, opts, false));
4914
- });
4915
- } else {
4916
- const discardOrDetachDescriptor = opts.discardDescriptor || opts.detachDescriptor;
4917
- cb(null, name, fd, _prepareTmpFileRemoveCallback(name, discardOrDetachDescriptor ? -1 : fd, opts, false));
4918
- }
4919
- });
4920
- });
4921
- }
4922
- function fileSync(options) {
4923
- const args = _parseArguments(options), opts = args[0];
4924
- const discardOrDetachDescriptor = opts.discardDescriptor || opts.detachDescriptor;
4925
- const name = tmpNameSync(opts);
4926
- let fd = fs.openSync(name, CREATE_FLAGS, opts.mode || FILE_MODE);
4927
- if (opts.discardDescriptor) {
4928
- fs.closeSync(fd);
4929
- fd = void 0;
4930
- }
4931
- return {
4932
- name,
4933
- fd,
4934
- removeCallback: _prepareTmpFileRemoveCallback(name, discardOrDetachDescriptor ? -1 : fd, opts, true)
4935
- };
4936
- }
4937
- function dir2(options, callback) {
4938
- const args = _parseArguments(options, callback), opts = args[0], cb = args[1];
4939
- tmpName(opts, function _tmpNameCreated(err, name) {
4940
- if (err) return cb(err);
4941
- fs.mkdir(name, opts.mode || DIR_MODE, function _dirCreated(err2) {
4942
- if (err2) return cb(err2);
4943
- cb(null, name, _prepareTmpDirRemoveCallback(name, opts, false));
2921
+ const imageData = await renderPageToPng(page);
2922
+ const result = await provider(imageData, i, "image/png");
2923
+ for (const b of ocrResultToBlocks(result, i)) blocks.push(b);
2924
+ } catch (err) {
2925
+ warnings?.push({
2926
+ page: i,
2927
+ message: `\uD398\uC774\uC9C0 ${i} OCR \uC2E4\uD328: ${err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`,
2928
+ code: "OCR_PAGE_FAILED"
4944
2929
  });
4945
- });
4946
- }
4947
- function dirSync(options) {
4948
- const args = _parseArguments(options), opts = args[0];
4949
- const name = tmpNameSync(opts);
4950
- fs.mkdirSync(name, opts.mode || DIR_MODE);
4951
- return {
4952
- name,
4953
- removeCallback: _prepareTmpDirRemoveCallback(name, opts, true)
4954
- };
4955
- }
4956
- function _removeFileAsync(fdPath, next) {
4957
- const _handler = function(err) {
4958
- if (err && !_isENOENT(err)) {
4959
- return next(err);
4960
- }
4961
- next();
4962
- };
4963
- if (0 <= fdPath[0])
4964
- fs.close(fdPath[0], function() {
4965
- fs.unlink(fdPath[1], _handler);
2930
+ logger.log({
2931
+ level: "warn",
2932
+ event: "progress",
2933
+ message: "\uD398\uC774\uC9C0 OCR \uC2E4\uD328",
2934
+ meta: { page: i },
2935
+ error: { message: err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958", name: err instanceof Error ? err.name : "Error" }
4966
2936
  });
4967
- else fs.unlink(fdPath[1], _handler);
4968
- }
4969
- function _removeFileSync(fdPath) {
4970
- let rethrownException = null;
4971
- try {
4972
- if (0 <= fdPath[0]) fs.closeSync(fdPath[0]);
4973
- } catch (e) {
4974
- if (!_isEBADF(e) && !_isENOENT(e)) throw e;
4975
- } finally {
4976
- try {
4977
- fs.unlinkSync(fdPath[1]);
4978
- } catch (e) {
4979
- if (!_isENOENT(e)) rethrownException = e;
4980
- }
4981
- }
4982
- if (rethrownException !== null) {
4983
- throw rethrownException;
4984
- }
4985
- }
4986
- function _prepareTmpFileRemoveCallback(name, fd, opts, sync) {
4987
- const removeCallbackSync = _prepareRemoveCallback(_removeFileSync, [fd, name], sync);
4988
- const removeCallback = _prepareRemoveCallback(_removeFileAsync, [fd, name], sync, removeCallbackSync);
4989
- if (!opts.keep) _removeObjects.unshift(removeCallbackSync);
4990
- return sync ? removeCallbackSync : removeCallback;
4991
- }
4992
- function _prepareTmpDirRemoveCallback(name, opts, sync) {
4993
- const removeFunction = opts.unsafeCleanup ? rimraf : fs.rmdir.bind(fs);
4994
- const removeFunctionSync = opts.unsafeCleanup ? FN_RIMRAF_SYNC : FN_RMDIR_SYNC;
4995
- const removeCallbackSync = _prepareRemoveCallback(removeFunctionSync, name, sync);
4996
- const removeCallback = _prepareRemoveCallback(removeFunction, name, sync, removeCallbackSync);
4997
- if (!opts.keep) _removeObjects.unshift(removeCallbackSync);
4998
- return sync ? removeCallbackSync : removeCallback;
4999
- }
5000
- function _prepareRemoveCallback(removeFunction, fileOrDirName, sync, cleanupCallbackSync) {
5001
- let called = false;
5002
- return function _cleanupCallback(next) {
5003
- if (!called) {
5004
- const toRemove = cleanupCallbackSync || _cleanupCallback;
5005
- const index2 = _removeObjects.indexOf(toRemove);
5006
- if (index2 >= 0) _removeObjects.splice(index2, 1);
5007
- called = true;
5008
- if (sync || removeFunction === FN_RMDIR_SYNC || removeFunction === FN_RIMRAF_SYNC) {
5009
- return removeFunction(fileOrDirName);
5010
- } else {
5011
- return removeFunction(fileOrDirName, next || function() {
5012
- });
5013
- }
5014
- }
5015
- };
5016
- }
5017
- function _garbageCollector() {
5018
- if (!_gracefulCleanup) return;
5019
- while (_removeObjects.length) {
5020
- try {
5021
- _removeObjects[0]();
5022
- } catch (e) {
5023
- }
5024
- }
5025
- }
5026
- function _randomChars(howMany) {
5027
- let value = [], rnd = null;
5028
- try {
5029
- rnd = crypto.randomBytes(howMany);
5030
- } catch (e) {
5031
- rnd = crypto.pseudoRandomBytes(howMany);
5032
- }
5033
- for (let i = 0; i < howMany; i++) {
5034
- value.push(RANDOM_CHARS[rnd[i] % RANDOM_CHARS.length]);
5035
- }
5036
- return value.join("");
5037
- }
5038
- function _isUndefined(obj) {
5039
- return typeof obj === "undefined";
5040
- }
5041
- function _parseArguments(options, callback) {
5042
- if (typeof options === "function") {
5043
- return [{}, options];
5044
- }
5045
- if (_isUndefined(options)) {
5046
- return [{}, callback];
5047
2937
  }
5048
- const actualOptions = {};
5049
- for (const key of Object.getOwnPropertyNames(options)) {
5050
- actualOptions[key] = options[key];
5051
- }
5052
- return [actualOptions, callback];
5053
2938
  }
5054
- function _resolvePath(name, tmpDir, cb) {
5055
- const pathToResolve = path.isAbsolute(name) ? name : path.join(tmpDir, name);
5056
- fs.stat(pathToResolve, function(err) {
5057
- if (err) {
5058
- fs.realpath(path.dirname(pathToResolve), function(err2, parentDir) {
5059
- if (err2) return cb(err2);
5060
- cb(null, path.join(parentDir, path.basename(pathToResolve)));
5061
- });
5062
- } else {
5063
- fs.realpath(pathToResolve, cb);
5064
- }
2939
+ logger.log({ level: "info", event: "done", message: "OCR \uC21C\uCC28 \uCC98\uB9AC \uC644\uB8CC", meta: { blocks: blocks.length } });
2940
+ return blocks;
2941
+ }
2942
+ const pageNumbers = [];
2943
+ for (let i = 1; i <= effectivePageCount; i++) {
2944
+ if (pageFilter && !pageFilter.has(i)) continue;
2945
+ pageNumbers.push(i);
2946
+ }
2947
+ const tasks = pageNumbers.map((pageNum) => async () => {
2948
+ try {
2949
+ const page = await doc.getPage(pageNum);
2950
+ const imageData = await renderPageToPng(page);
2951
+ const result = await provider(imageData, pageNum, "image/png");
2952
+ return { pageNum, pageBlocks: ocrResultToBlocks(result, pageNum) };
2953
+ } catch (err) {
2954
+ warnings?.push({
2955
+ page: pageNum,
2956
+ message: `\uD398\uC774\uC9C0 ${pageNum} OCR \uC2E4\uD328: ${err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"}`,
2957
+ code: "OCR_PAGE_FAILED"
5065
2958
  });
5066
- }
5067
- function _resolvePathSync(name, tmpDir) {
5068
- const pathToResolve = path.isAbsolute(name) ? name : path.join(tmpDir, name);
5069
- try {
5070
- fs.statSync(pathToResolve);
5071
- return fs.realpathSync(pathToResolve);
5072
- } catch (_err) {
5073
- const parentDir = fs.realpathSync(path.dirname(pathToResolve));
5074
- return path.join(parentDir, path.basename(pathToResolve));
5075
- }
5076
- }
5077
- function _generateTmpName(opts) {
5078
- const tmpDir = opts.tmpdir;
5079
- if (!_isUndefined(opts.name)) {
5080
- return path.join(tmpDir, opts.dir, opts.name);
5081
- }
5082
- if (!_isUndefined(opts.template)) {
5083
- return path.join(tmpDir, opts.dir, opts.template).replace(TEMPLATE_PATTERN, _randomChars(6));
5084
- }
5085
- const name = [
5086
- opts.prefix ? opts.prefix : "tmp",
5087
- "-",
5088
- process.pid,
5089
- "-",
5090
- _randomChars(12),
5091
- opts.postfix ? "-" + opts.postfix : ""
5092
- ].join("");
5093
- return path.join(tmpDir, opts.dir, name);
5094
- }
5095
- function _assertOptionsBase(options) {
5096
- if (!_isUndefined(options.name)) {
5097
- const name = options.name;
5098
- if (path.isAbsolute(name)) throw new Error(`name option must not contain an absolute path, found "${name}".`);
5099
- const basename3 = path.basename(name);
5100
- if (basename3 === ".." || basename3 === "." || basename3 !== name)
5101
- throw new Error(`name option must not contain a path, found "${name}".`);
5102
- }
5103
- if (!_isUndefined(options.template) && !options.template.match(TEMPLATE_PATTERN)) {
5104
- throw new Error(`Invalid template, found "${options.template}".`);
5105
- }
5106
- if (!_isUndefined(options.tries) && isNaN(options.tries) || options.tries < 0) {
5107
- throw new Error(`Invalid tries, found "${options.tries}".`);
5108
- }
5109
- options.tries = _isUndefined(options.name) ? options.tries || DEFAULT_TRIES : 1;
5110
- options.keep = !!options.keep;
5111
- options.detachDescriptor = !!options.detachDescriptor;
5112
- options.discardDescriptor = !!options.discardDescriptor;
5113
- options.unsafeCleanup = !!options.unsafeCleanup;
5114
- options.prefix = _isUndefined(options.prefix) ? "" : options.prefix;
5115
- options.postfix = _isUndefined(options.postfix) ? "" : options.postfix;
5116
- }
5117
- function _getRelativePath(option, name, tmpDir, cb) {
5118
- if (_isUndefined(name)) return cb(null);
5119
- _resolvePath(name, tmpDir, function(err, resolvedPath) {
5120
- if (err) return cb(err);
5121
- const relativePath = path.relative(tmpDir, resolvedPath);
5122
- if (!resolvedPath.startsWith(tmpDir)) {
5123
- return cb(new Error(`${option} option must be relative to "${tmpDir}", found "${relativePath}".`));
5124
- }
5125
- cb(null, relativePath);
2959
+ logger.log({
2960
+ level: "warn",
2961
+ event: "progress",
2962
+ message: "\uD398\uC774\uC9C0 OCR \uC2E4\uD328(\uBCD1\uB82C)",
2963
+ meta: { page: pageNum },
2964
+ error: { message: err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958", name: err instanceof Error ? err.name : "Error" }
5126
2965
  });
2966
+ return null;
5127
2967
  }
5128
- function _getRelativePathSync(option, name, tmpDir) {
5129
- if (_isUndefined(name)) return;
5130
- const resolvedPath = _resolvePathSync(name, tmpDir);
5131
- const relativePath = path.relative(tmpDir, resolvedPath);
5132
- if (!resolvedPath.startsWith(tmpDir)) {
5133
- throw new Error(`${option} option must be relative to "${tmpDir}", found "${relativePath}".`);
2968
+ });
2969
+ const taskResults = await runWithConcurrency(tasks, concurrency);
2970
+ for (const item of taskResults) {
2971
+ if (!item) continue;
2972
+ for (const b of item.pageBlocks) blocks.push(b);
2973
+ }
2974
+ logger.log({ level: "info", event: "done", message: "OCR \uBCD1\uB82C \uCC98\uB9AC \uC644\uB8CC", meta: { blocks: blocks.length, pages: pageNumbers.length } });
2975
+ return blocks;
2976
+ }
2977
+ async function ocrPagesBatch(doc, provider, pageFilter, effectivePageCount, warnings, concurrency = 1, onProgress) {
2978
+ const pageNumbers = [];
2979
+ for (let i = 1; i <= effectivePageCount; i++) {
2980
+ if (pageFilter && !pageFilter.has(i)) continue;
2981
+ pageNumbers.push(i);
2982
+ }
2983
+ const pageBatches = [];
2984
+ for (let i = 0; i < pageNumbers.length; i += provider.batchSize) {
2985
+ pageBatches.push(pageNumbers.slice(i, i + provider.batchSize));
2986
+ }
2987
+ let processed = 0;
2988
+ const batchTasks = pageBatches.map((batchPageNums, batchIdx) => async () => {
2989
+ const pageBlocks = [];
2990
+ const batchImages = [];
2991
+ try {
2992
+ for (const pageNum of batchPageNums) {
2993
+ const page = await doc.getPage(pageNum);
2994
+ const image = await renderPageToPng(page);
2995
+ batchImages.push({ image, pageNum });
2996
+ }
2997
+ } catch (renderErr) {
2998
+ const rendered = new Set(batchImages.map((b) => b.pageNum));
2999
+ for (const pageNum of batchPageNums) {
3000
+ if (!rendered.has(pageNum)) pageBlocks.push({ pageNum, blocks: [] });
5134
3001
  }
5135
- return relativePath;
5136
3002
  }
5137
- function _assertAndSanitizeOptions(options, cb) {
5138
- _getTmpDir(options, function(err, tmpDir) {
5139
- if (err) return cb(err);
5140
- options.tmpdir = tmpDir;
5141
- try {
5142
- _assertOptionsBase(options, tmpDir);
5143
- } catch (err2) {
5144
- return cb(err2);
5145
- }
5146
- _getRelativePath("dir", options.dir, tmpDir, function(err2, dir3) {
5147
- if (err2) return cb(err2);
5148
- options.dir = _isUndefined(dir3) ? "" : dir3;
5149
- _getRelativePath("template", options.template, tmpDir, function(err3, template) {
5150
- if (err3) return cb(err3);
5151
- options.template = template;
5152
- cb(null, options);
3003
+ if (batchImages.length > 0) {
3004
+ try {
3005
+ const results = await provider.processBatch(batchImages);
3006
+ for (const { pageNum } of batchImages) {
3007
+ const result = results.get(pageNum);
3008
+ pageBlocks.push({
3009
+ pageNum,
3010
+ blocks: result ? ocrResultToBlocks(result, pageNum) : []
5153
3011
  });
3012
+ }
3013
+ } catch (err) {
3014
+ const range = `${batchPageNums[0]}-${batchPageNums[batchPageNums.length - 1]}`;
3015
+ warnings?.push({
3016
+ message: `\uBC30\uCE58 OCR \uC2E4\uD328 (\uD398\uC774\uC9C0 ${range}): ${err instanceof Error ? err.message : "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958"} \u2014 \uB2E8\uC77C \uD398\uC774\uC9C0\uB85C \uC7AC\uC2DC\uB3C4`,
3017
+ code: "OCR_PAGE_FAILED"
5154
3018
  });
5155
- });
5156
- }
5157
- function _assertAndSanitizeOptionsSync(options) {
5158
- const tmpDir = options.tmpdir = _getTmpDirSync(options);
5159
- _assertOptionsBase(options, tmpDir);
5160
- const dir3 = _getRelativePathSync("dir", options.dir, tmpDir);
5161
- options.dir = _isUndefined(dir3) ? "" : dir3;
5162
- options.template = _getRelativePathSync("template", options.template, tmpDir);
5163
- return options;
5164
- }
5165
- function _isEBADF(error) {
5166
- return _isExpectedError(error, -EBADF, "EBADF");
5167
- }
5168
- function _isENOENT(error) {
5169
- return _isExpectedError(error, -ENOENT, "ENOENT");
5170
- }
5171
- function _isExpectedError(error, errno, code) {
5172
- return IS_WIN32 ? error.code === code : error.code === code && error.errno === errno;
5173
- }
5174
- function setGracefulCleanup() {
5175
- _gracefulCleanup = true;
5176
- }
5177
- function _getTmpDir(options, cb) {
5178
- return fs.realpath(options && options.tmpdir || os.tmpdir(), cb);
3019
+ for (const { image, pageNum } of batchImages) {
3020
+ try {
3021
+ const singleResult = await provider.processBatch([{ image, pageNum }]);
3022
+ const r = singleResult.get(pageNum);
3023
+ pageBlocks.push({ pageNum, blocks: r ? ocrResultToBlocks(r, pageNum) : [] });
3024
+ } catch {
3025
+ pageBlocks.push({ pageNum, blocks: [] });
3026
+ }
3027
+ }
3028
+ }
5179
3029
  }
5180
- function _getTmpDirSync(options) {
5181
- return fs.realpathSync(options && options.tmpdir || os.tmpdir());
3030
+ processed += batchPageNums.length;
3031
+ onProgress?.(processed, pageNumbers.length);
3032
+ return { batchIdx, pageBlocks };
3033
+ });
3034
+ const effectiveConcurrency = Math.max(1, concurrency);
3035
+ const batchResults = await runWithConcurrency(batchTasks, effectiveConcurrency);
3036
+ const blocks = [];
3037
+ for (const result of batchResults) {
3038
+ for (const { blocks: pageBlks } of result.pageBlocks) {
3039
+ for (const b of pageBlks) blocks.push(b);
5182
3040
  }
5183
- process.addListener(EXIT, _garbageCollector);
5184
- Object.defineProperty(module.exports, "tmpdir", {
5185
- enumerable: true,
5186
- configurable: false,
5187
- get: function() {
5188
- return _getTmpDirSync();
5189
- }
5190
- });
5191
- module.exports.dir = dir2;
5192
- module.exports.dirSync = dirSync;
5193
- module.exports.file = file;
5194
- module.exports.fileSync = fileSync;
5195
- module.exports.tmpName = tmpName;
5196
- module.exports.tmpNameSync = tmpNameSync;
5197
- module.exports.setGracefulCleanup = setGracefulCleanup;
5198
3041
  }
5199
- });
5200
-
5201
- // ../../../node_modules/libreoffice-convert/index.js
5202
- var require_libreoffice_convert = __commonJS({
5203
- "../../../node_modules/libreoffice-convert/index.js"(exports, module) {
3042
+ return blocks;
3043
+ }
3044
+ async function renderPageToPng(page) {
3045
+ const { createCanvas } = await import("@napi-rs/canvas");
3046
+ const scale = 2;
3047
+ const viewport = page.getViewport({ scale });
3048
+ const canvas = createCanvas(Math.floor(viewport.width), Math.floor(viewport.height));
3049
+ const ctx = canvas.getContext("2d");
3050
+ await page.render({ canvasContext: ctx, viewport }).promise;
3051
+ return new Uint8Array(canvas.toBuffer("image/png"));
3052
+ }
3053
+ var init_provider = __esm({
3054
+ "src/ocr/provider.ts"() {
5204
3055
  "use strict";
5205
- var fs = __require("fs");
5206
- var path = __require("path");
5207
- var url = __require("url");
5208
- var async = (init_async(), __toCommonJS(async_exports));
5209
- var tmp = require_tmp();
5210
- var { execFile } = __require("child_process");
5211
- var convertWithOptions = (document, format, filter2, options, callback) => {
5212
- const tmpOptions = (options || {}).tmpOptions || {};
5213
- const asyncOptions = (options || {}).asyncOptions || {};
5214
- const execOptions = (options || {}).execOptions || {};
5215
- const fileName = (options || {}).fileName || "source";
5216
- const sofficeAdditionalArgs = (options || {}).sofficeAdditionalArgs || [];
5217
- const tempDir = tmp.dirSync({ prefix: "libreofficeConvert_", unsafeCleanup: true, ...tmpOptions });
5218
- const installDir = tmp.dirSync({ prefix: "soffice", unsafeCleanup: true, ...tmpOptions });
5219
- return async.auto({
5220
- soffice: (callback2) => {
5221
- let paths = (options || {}).sofficeBinaryPaths || [];
5222
- switch (process.platform) {
5223
- case "darwin":
5224
- paths = [...paths, "/Applications/LibreOffice.app/Contents/MacOS/soffice"];
5225
- break;
5226
- case "linux":
5227
- paths = [...paths, "/usr/bin/libreoffice", "/usr/bin/soffice", "/snap/bin/libreoffice", "/opt/libreoffice/program/soffice", "/opt/libreoffice7.6/program/soffice"];
5228
- break;
5229
- case "win32":
5230
- paths = [
5231
- ...paths,
5232
- path.join(process.env["PROGRAMFILES(X86)"] || "", "LIBREO~1/program/soffice.exe"),
5233
- path.join(process.env["PROGRAMFILES(X86)"] || "", "LibreOffice/program/soffice.exe"),
5234
- path.join(process.env.PROGRAMFILES_X86 || "", "LibreOffice/program/soffice.exe"),
5235
- path.join(process.env.PROGRAMFILES || "", "LibreOffice/program/soffice.exe"),
5236
- process.env.LIBRE_OFFICE_EXE || "",
5237
- "C:/Program Files/LibreOffice/program/soffice.exe"
5238
- ];
5239
- break;
5240
- default:
5241
- return callback2(new Error(`Operating system not yet supported: ${process.platform}`));
5242
- }
5243
- return async.filter(
5244
- paths,
5245
- (filePath, callback3) => fs.access(filePath, (err) => callback3(null, !err)),
5246
- (err, res) => {
5247
- if (res.length === 0) {
5248
- return callback2(new Error("Could not find soffice binary"));
5249
- }
5250
- return callback2(null, res[0]);
5251
- }
5252
- );
5253
- },
5254
- saveSource: (callback2) => fs.writeFile(path.join(tempDir.name, fileName), document, callback2),
5255
- convert: ["soffice", "saveSource", (results, callback2) => {
5256
- let filterParam = filter2?.length ? `:${filter2}` : "";
5257
- let fmt = !(filter2 ?? "").includes(" ") ? `${format}${filterParam}` : `"${format}${filterParam}"`;
5258
- let args = [];
5259
- args.push(`-env:UserInstallation=${url.pathToFileURL(installDir.name)}`);
5260
- args.push("--headless");
5261
- args.push("--convert-to");
5262
- args.push(fmt);
5263
- args.push("--outdir");
5264
- args.push(tempDir.name);
5265
- args.push(...sofficeAdditionalArgs);
5266
- args.push(path.join(tempDir.name, fileName));
5267
- return execFile(results.soffice, args, execOptions, (err, stdout, stderr) => {
5268
- if (stderr && stderr.toLowerCase().includes("error"))
5269
- callback2(new Error("Error calling soffice: " + stderr));
5270
- else callback2(err, stdout, stderr);
5271
- });
5272
- }],
5273
- loadDestination: [
5274
- "convert",
5275
- (results, callback2) => async.retry({
5276
- times: asyncOptions.times || 3,
5277
- interval: asyncOptions.interval || 200
5278
- }, (callback3) => fs.readFile(path.join(tempDir.name, `${fileName.slice(0, fileName.length - path.extname(fileName).length)}.${format.split(":")[0]}`), callback3), callback2)
5279
- ]
5280
- }).then((res) => {
5281
- return callback(null, res.loadDestination);
5282
- }).catch((err) => {
5283
- return callback(err);
5284
- }).finally(() => {
5285
- tempDir.removeCallback();
5286
- installDir.removeCallback();
5287
- });
5288
- };
5289
- var convert = (document, format, filter2, callback) => {
5290
- return convertWithOptions(document, format, filter2, {}, callback);
5291
- };
5292
- module.exports = {
5293
- convert,
5294
- convertWithOptions
5295
- };
3056
+ init_markdown_to_blocks();
3057
+ init_logger();
5296
3058
  }
5297
3059
  });
5298
3060
 
@@ -5350,7 +3112,7 @@ import JSZip2 from "jszip";
5350
3112
  import { DOMParser } from "@xmldom/xmldom";
5351
3113
 
5352
3114
  // src/utils.ts
5353
- var VERSION = true ? "2.5.0" : "0.0.0-dev";
3115
+ var VERSION = true ? "2.5.1" : "0.0.0-dev";
5354
3116
  function toArrayBuffer(buf) {
5355
3117
  if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
5356
3118
  return buf.buffer;
@@ -5743,7 +3505,7 @@ async function extractHwpxStyles(zip, decompressed) {
5743
3505
  }
5744
3506
  return result;
5745
3507
  }
5746
- function parseCharProperties(doc, map2) {
3508
+ function parseCharProperties(doc, map) {
5747
3509
  const tagNames = ["hh:charPr", "charPr", "hp:charPr"];
5748
3510
  for (const tagName of tagNames) {
5749
3511
  const elements = doc.getElementsByTagName(tagName);
@@ -5770,11 +3532,11 @@ function parseCharProperties(doc, map2) {
5770
3532
  }
5771
3533
  }
5772
3534
  }
5773
- map2.set(id, prop);
3535
+ map.set(id, prop);
5774
3536
  }
5775
3537
  }
5776
3538
  }
5777
- function parseStyleElements(doc, map2) {
3539
+ function parseStyleElements(doc, map) {
5778
3540
  const tagNames = ["hh:style", "style", "hp:style"];
5779
3541
  for (const tagName of tagNames) {
5780
3542
  const elements = doc.getElementsByTagName(tagName);
@@ -5784,7 +3546,7 @@ function parseStyleElements(doc, map2) {
5784
3546
  const name = el.getAttribute("name") || el.getAttribute("engName") || "";
5785
3547
  const charPrId = el.getAttribute("charPrIDRef") || void 0;
5786
3548
  const paraPrId = el.getAttribute("paraPrIDRef") || void 0;
5787
- map2.set(id, { name, charPrId, paraPrId });
3549
+ map.set(id, { name, charPrId, paraPrId });
5788
3550
  }
5789
3551
  }
5790
3552
  }
@@ -6310,9 +4072,9 @@ function walkParagraphChildren(node, blocks, tableCtx, tableStack, styleMap, war
6310
4072
  if (depth > MAX_XML_DEPTH) return tableCtx;
6311
4073
  const children = node.childNodes;
6312
4074
  if (!children) return tableCtx;
6313
- const walkChildren = (parent2, d) => {
4075
+ const walkChildren = (parent, d) => {
6314
4076
  if (d > MAX_XML_DEPTH) return;
6315
- const kids = parent2.childNodes;
4077
+ const kids = parent.childNodes;
6316
4078
  if (!kids) return;
6317
4079
  for (let i = 0; i < kids.length; i++) {
6318
4080
  const el = kids[i];
@@ -8525,17 +6287,17 @@ function lineOverlapRatio(a1, a2, b1, b2) {
8525
6287
  return overlap / target;
8526
6288
  }
8527
6289
  function groupConnectedLines(lines) {
8528
- const parent2 = lines.map((_, i) => i);
6290
+ const parent = lines.map((_, i) => i);
8529
6291
  function find2(x) {
8530
- while (parent2[x] !== x) {
8531
- parent2[x] = parent2[parent2[x]];
8532
- x = parent2[x];
6292
+ while (parent[x] !== x) {
6293
+ parent[x] = parent[parent[x]];
6294
+ x = parent[x];
8533
6295
  }
8534
6296
  return x;
8535
6297
  }
8536
6298
  function union(a, b) {
8537
6299
  const ra = find2(a), rb = find2(b);
8538
- if (ra !== rb) parent2[ra] = rb;
6300
+ if (ra !== rb) parent[ra] = rb;
8539
6301
  }
8540
6302
  for (let i = 0; i < lines.length; i++) {
8541
6303
  for (let j = i + 1; j < lines.length; j++) {
@@ -8956,10 +6718,10 @@ async function loadPdfWithTimeout(buffer) {
8956
6718
  try {
8957
6719
  return await Promise.race([
8958
6720
  loadingTask.promise,
8959
- new Promise((_, reject2) => {
6721
+ new Promise((_, reject) => {
8960
6722
  timer = setTimeout(() => {
8961
6723
  loadingTask.destroy();
8962
- reject2(new KordocError(`PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (${timeoutSec}\uCD08 \uCD08\uACFC)`));
6724
+ reject(new KordocError(`PDF \uB85C\uB529 \uD0C0\uC784\uC544\uC6C3 (${timeoutSec}\uCD08 \uCD08\uACFC)`));
8963
6725
  }, timeoutMs);
8964
6726
  })
8965
6727
  ]);
@@ -9148,10 +6910,10 @@ async function parsePdfDocument(buffer, options) {
9148
6910
  }
9149
6911
  const { resolveOcrProvider: resolveOcrProvider2 } = await Promise.resolve().then(() => (init_resolve(), resolve_exports));
9150
6912
  const { ocrPages: ocrPages2 } = await Promise.resolve().then(() => (init_provider(), provider_exports));
9151
- const tryProvider = async (provider, filter2) => {
6913
+ const tryProvider = async (provider, filter) => {
9152
6914
  try {
9153
- logger.log({ level: "debug", stage: "ocr", event: "progress", message: "OCR provider \uC2E4\uD589", meta: { filteredPages: filter2?.size } });
9154
- return await ocrPages2(doc, provider, filter2, effectivePageCount, warnings, concurrency, options?.onProgress);
6915
+ logger.log({ level: "debug", stage: "ocr", event: "progress", message: "OCR provider \uC2E4\uD589", meta: { filteredPages: filter?.size } });
6916
+ return await ocrPages2(doc, provider, filter, effectivePageCount, warnings, concurrency, options?.onProgress);
9155
6917
  } catch {
9156
6918
  logger.log({ level: "warn", stage: "ocr", event: "progress", message: "OCR provider \uC2E4\uD589 \uC2E4\uD328(\uBE48 \uACB0\uACFC\uB85C \uCC98\uB9AC)" });
9157
6919
  return [];
@@ -10166,8 +7928,8 @@ function parseMergeRef(ref) {
10166
7928
  if (!start || !end) return null;
10167
7929
  return { startCol: start.col, startRow: start.row, endCol: end.col, endRow: end.row };
10168
7930
  }
10169
- function getElements(parent2, tagName) {
10170
- const nodes = parent2.getElementsByTagName(tagName);
7931
+ function getElements(parent, tagName) {
7932
+ const nodes = parent.getElementsByTagName(tagName);
10171
7933
  const result = [];
10172
7934
  for (let i = 0; i < nodes.length; i++) result.push(nodes[i]);
10173
7935
  return result;
@@ -10203,14 +7965,14 @@ function parseWorkbook(xml) {
10203
7965
  }
10204
7966
  function parseRels(xml) {
10205
7967
  const doc = parseXml(xml);
10206
- const map2 = /* @__PURE__ */ new Map();
7968
+ const map = /* @__PURE__ */ new Map();
10207
7969
  const rels = getElements(doc.documentElement, "Relationship");
10208
7970
  for (const rel of rels) {
10209
7971
  const id = rel.getAttribute("Id");
10210
7972
  const target = rel.getAttribute("Target");
10211
- if (id && target) map2.set(id, target);
7973
+ if (id && target) map.set(id, target);
10212
7974
  }
10213
- return map2;
7975
+ return map;
10214
7976
  }
10215
7977
  function parseWorksheet(xml, sharedStrings) {
10216
7978
  const doc = parseXml(xml);
@@ -10468,9 +8230,9 @@ import JSZip4 from "jszip";
10468
8230
  import { DOMParser as DOMParser3 } from "@xmldom/xmldom";
10469
8231
  init_logger();
10470
8232
  var MAX_DECOMPRESS_SIZE4 = 500 * 1024 * 1024;
10471
- function getChildElements(parent2, localName) {
8233
+ function getChildElements(parent, localName) {
10472
8234
  const result = [];
10473
- const children = parent2.childNodes;
8235
+ const children = parent.childNodes;
10474
8236
  for (let i = 0; i < children.length; i++) {
10475
8237
  const node = children[i];
10476
8238
  if (node.nodeType === 1) {
@@ -10482,7 +8244,7 @@ function getChildElements(parent2, localName) {
10482
8244
  }
10483
8245
  return result;
10484
8246
  }
10485
- function findElements(parent2, localName) {
8247
+ function findElements(parent, localName) {
10486
8248
  const result = [];
10487
8249
  const walk = (node) => {
10488
8250
  const children = node.childNodes;
@@ -10497,7 +8259,7 @@ function findElements(parent2, localName) {
10497
8259
  }
10498
8260
  }
10499
8261
  };
10500
- walk(parent2);
8262
+ walk(parent);
10501
8263
  return result;
10502
8264
  }
10503
8265
  function getAttr(el, localName) {
@@ -10512,7 +8274,7 @@ function parseXml2(text) {
10512
8274
  return new DOMParser3().parseFromString(text, "text/xml");
10513
8275
  }
10514
8276
  function buildElementIndex(root) {
10515
- const index2 = /* @__PURE__ */ new Map();
8277
+ const index = /* @__PURE__ */ new Map();
10516
8278
  const walk = (node) => {
10517
8279
  const children = node.childNodes;
10518
8280
  for (let i = 0; i < children.length; i++) {
@@ -10521,10 +8283,10 @@ function buildElementIndex(root) {
10521
8283
  const el = child;
10522
8284
  const name = el.localName ?? "";
10523
8285
  if (name) {
10524
- let list = index2.get(name);
8286
+ let list = index.get(name);
10525
8287
  if (!list) {
10526
8288
  list = [];
10527
- index2.set(name, list);
8289
+ index.set(name, list);
10528
8290
  }
10529
8291
  list.push(el);
10530
8292
  }
@@ -10533,7 +8295,7 @@ function buildElementIndex(root) {
10533
8295
  }
10534
8296
  };
10535
8297
  walk(root);
10536
- return index2;
8298
+ return index;
10537
8299
  }
10538
8300
  function parseStyles(xml) {
10539
8301
  const doc = parseXml2(xml);
@@ -10599,14 +8361,14 @@ function parseNumbering(xml) {
10599
8361
  }
10600
8362
  function parseRels2(xml) {
10601
8363
  const doc = parseXml2(xml);
10602
- const map2 = /* @__PURE__ */ new Map();
8364
+ const map = /* @__PURE__ */ new Map();
10603
8365
  const rels = findElements(doc, "Relationship");
10604
8366
  for (const rel of rels) {
10605
8367
  const id = getAttr(rel, "Id");
10606
8368
  const target = getAttr(rel, "Target");
10607
- if (id && target) map2.set(id, target);
8369
+ if (id && target) map.set(id, target);
10608
8370
  }
10609
- return map2;
8371
+ return map;
10610
8372
  }
10611
8373
  function parseFootnotes(xml) {
10612
8374
  const doc = parseXml2(xml);
@@ -12449,22 +10211,22 @@ function findCharPrByCharacteristic(charPrMap, characteristic) {
12449
10211
  italic: { minHeight: 800, maxHeight: 1100, needBold: false },
12450
10212
  normal: { minHeight: 900, maxHeight: 1100, needBold: false }
12451
10213
  };
12452
- const range2 = ranges[characteristic];
12453
- if (!range2) return null;
10214
+ const range = ranges[characteristic];
10215
+ if (!range) return null;
12454
10216
  let bestMatch = null;
12455
10217
  let bestScore = 0;
12456
10218
  for (const [id, prop] of charPrMap) {
12457
10219
  let score = 0;
12458
10220
  if (prop.height) {
12459
- if (prop.height >= range2.minHeight && prop.height <= range2.maxHeight) {
10221
+ if (prop.height >= range.minHeight && prop.height <= range.maxHeight) {
12460
10222
  score += 50;
12461
- } else if (Math.abs(prop.height - (range2.minHeight + range2.maxHeight) / 2) < 100) {
10223
+ } else if (Math.abs(prop.height - (range.minHeight + range.maxHeight) / 2) < 100) {
12462
10224
  score += 20;
12463
10225
  }
12464
10226
  }
12465
- if (range2.needBold !== void 0) {
12466
- if (range2.needBold && prop.bold) score += 30;
12467
- else if (!range2.needBold && !prop.bold) score += 20;
10227
+ if (range.needBold !== void 0) {
10228
+ if (range.needBold && prop.bold) score += 30;
10229
+ else if (!range.needBold && !prop.bold) score += 20;
12468
10230
  }
12469
10231
  if (characteristic === "italic" && prop.italic) score += 20;
12470
10232
  if (score > bestScore) {
@@ -13303,10 +11065,10 @@ function mimeToExtension(mimeType) {
13303
11065
  return "png";
13304
11066
  }
13305
11067
  function buildImageMap(images) {
13306
- const map2 = /* @__PURE__ */ new Map();
13307
- if (!images) return map2;
13308
- for (const img of images) map2.set(img.filename, img);
13309
- return map2;
11068
+ const map = /* @__PURE__ */ new Map();
11069
+ if (!images) return map;
11070
+ for (const img of images) map.set(img.filename, img);
11071
+ return map;
13310
11072
  }
13311
11073
  async function fillBodySheet(sheet, workbook, blocks, imageMap) {
13312
11074
  sheet.getColumn(1).width = 80;
@@ -13508,8 +11270,8 @@ var ApiKeyRotationPool = class _ApiKeyRotationPool {
13508
11270
  const s = this.find(keyId);
13509
11271
  s.failureCount++;
13510
11272
  s.consecutiveFailures++;
13511
- const retryable2 = this.isRetryableFailure(opts.status, opts.timeout);
13512
- if (!retryable2) return;
11273
+ const retryable = this.isRetryableFailure(opts.status, opts.timeout);
11274
+ if (!retryable) return;
13513
11275
  const exp = Math.max(0, s.consecutiveFailures - 1);
13514
11276
  const backoff = Math.min(this.baseCooldownMs * 2 ** exp, this.maxCooldownMs);
13515
11277
  const cooldown = Math.max(backoff, opts.retryAfterMs ?? 0);
@@ -13526,8 +11288,8 @@ var ApiKeyRotationPool = class _ApiKeyRotationPool {
13526
11288
  cooldownUntil: s.cooldownUntil
13527
11289
  }));
13528
11290
  }
13529
- isRetryableFailure(status, timeout2) {
13530
- if (timeout2) return true;
11291
+ isRetryableFailure(status, timeout) {
11292
+ if (timeout) return true;
13531
11293
  if (status === 429) return true;
13532
11294
  if (typeof status === "number" && status >= 500) return true;
13533
11295
  return false;
@@ -13541,172 +11303,72 @@ var ApiKeyRotationPool = class _ApiKeyRotationPool {
13541
11303
 
13542
11304
  // src/pipeline/unified-ocr.ts
13543
11305
  import { mkdir, readdir, readFile, stat, writeFile } from "fs/promises";
13544
- import { basename as basename2, dirname as dirname3, extname, join as join5, resolve as resolve3 } from "path";
11306
+ import { basename as basename2, dirname as dirname3, extname, join as join4, resolve as resolve3 } from "path";
13545
11307
  import { spawn as spawn2 } from "child_process";
13546
11308
  import { performance } from "perf_hooks";
11309
+ import libre from "libreoffice-convert";
13547
11310
  init_logger();
13548
11311
 
13549
- // src/setup/detect.ts
13550
- import { spawnSync as spawnSync2 } from "child_process";
13551
- import { existsSync } from "fs";
13552
- import { join as join4 } from "path";
13553
- var isWin = process.platform === "win32";
13554
- function which(cmd) {
13555
- const finder = isWin ? "where" : "which";
13556
- const r = spawnSync2(finder, [cmd], { encoding: "utf8", shell: isWin });
13557
- if (r.status !== 0) return void 0;
13558
- return r.stdout.split(/\r?\n/).find(Boolean)?.trim();
13559
- }
13560
- function resolveCmd(cmd, envKey) {
13561
- const override = process.env[envKey];
13562
- if (override) return override;
13563
- const found = which(cmd);
13564
- if (found) return found;
13565
- if (isWin) return findWinCandidate(cmd);
13566
- return void 0;
13567
- }
13568
- function findWinCandidate(cmd) {
13569
- const exe = cmd.endsWith(".exe") ? cmd : `${cmd}.exe`;
13570
- const userProfile = process.env["USERPROFILE"] ?? "C:\\Users\\Default";
13571
- const localAppData = process.env["LOCALAPPDATA"] ?? join4(userProfile, "AppData", "Local");
13572
- const programFiles = process.env["ProgramFiles"] ?? "C:\\Program Files";
13573
- const programFilesX86 = process.env["ProgramFiles(x86)"] ?? "C:\\Program Files (x86)";
13574
- const isPopplerExe = ["pdftoppm.exe", "pdfinfo.exe", "pdftotext.exe"].includes(exe);
13575
- const isSoffice = exe === "soffice.exe";
13576
- if (isPopplerExe) return findPopplerWin(exe, { userProfile, localAppData, programFiles, programFilesX86 });
13577
- if (isSoffice) return findSofficeWin({ programFiles, programFilesX86, localAppData });
13578
- return void 0;
13579
- }
13580
- function findPopplerWin(exe, p) {
13581
- const scoopPath = join4(p.userProfile, "scoop", "apps", "poppler", "current", "bin", exe);
13582
- if (existsSync(scoopPath)) return scoopPath;
13583
- const wingetBase = join4(p.localAppData, "Microsoft", "WinGet", "Packages");
13584
- const wingetResult = globFirst(wingetBase, ["oschwartz10612.Poppler_*"], ["poppler-*", "Library", "bin", exe]);
13585
- if (wingetResult) return wingetResult;
13586
- const chocoBase = join4("C:", "ProgramData", "chocolatey", "lib", "poppler", "tools");
13587
- const chocoResult = globFirst(chocoBase, ["poppler-*"], ["bin", exe]);
13588
- if (chocoResult) return chocoResult;
13589
- for (const base of [
13590
- join4("C:", "poppler", "bin", exe),
13591
- join4("C:", "tools", "poppler", "bin", exe),
13592
- join4(p.programFiles, "poppler", "bin", exe),
13593
- join4(p.programFilesX86, "poppler", "bin", exe),
13594
- join4(p.localAppData, "Programs", "poppler", "bin", exe)
13595
- ]) {
13596
- if (existsSync(base)) return base;
13597
- }
13598
- return void 0;
13599
- }
13600
- function findSofficeWin(p) {
13601
- for (const candidate of [
13602
- join4(p.programFiles, "LibreOffice", "program", "soffice.exe"),
13603
- join4(p.programFilesX86, "LibreOffice", "program", "soffice.exe"),
13604
- join4(p.localAppData, "Programs", "LibreOffice", "program", "soffice.exe"),
13605
- // winget 설치는 Program Files와 동일
13606
- join4(p.programFiles, "LibreOffice 24", "program", "soffice.exe"),
13607
- join4(p.programFiles, "LibreOffice 25", "program", "soffice.exe")
13608
- ]) {
13609
- if (existsSync(candidate)) return candidate;
13610
- }
13611
- return void 0;
13612
- }
13613
- function globFirst(base, wildcardSegments, rest) {
13614
- if (!existsSync(base)) return void 0;
13615
- let current = base;
13616
- for (const seg of wildcardSegments) {
13617
- if (!seg.includes("*")) {
13618
- current = join4(current, seg);
13619
- if (!existsSync(current)) return void 0;
13620
- continue;
11312
+ // src/pipeline/bounded-queue.ts
11313
+ var QUEUE_DONE = /* @__PURE__ */ Symbol("QUEUE_DONE");
11314
+ var BoundedQueue = class {
11315
+ buffer = [];
11316
+ capacity;
11317
+ closed = false;
11318
+ producerWaiters = [];
11319
+ consumerWaiters = [];
11320
+ constructor(capacity) {
11321
+ if (capacity < 1) throw new RangeError("BoundedQueue capacity must be >= 1");
11322
+ this.capacity = capacity;
11323
+ }
11324
+ async enqueue(item) {
11325
+ if (this.closed) throw new Error("BoundedQueue: cannot enqueue after close()");
11326
+ if (this.consumerWaiters.length > 0) {
11327
+ const resolve4 = this.consumerWaiters.shift();
11328
+ resolve4(item);
11329
+ return;
13621
11330
  }
13622
- try {
13623
- const { readdirSync } = __require("fs");
13624
- const entries = readdirSync(current);
13625
- const pattern = seg.replace(/\*/g, ".*");
13626
- const re = new RegExp(`^${pattern}$`, "i");
13627
- const match = entries.find((e) => re.test(e));
13628
- if (!match) return void 0;
13629
- current = join4(current, match);
13630
- } catch {
13631
- return void 0;
11331
+ while (this.buffer.length >= this.capacity) {
11332
+ await new Promise((resolve4) => this.producerWaiters.push(resolve4));
11333
+ if (this.closed) throw new Error("BoundedQueue: closed while waiting to enqueue");
11334
+ }
11335
+ this.buffer.push(item);
11336
+ }
11337
+ async dequeue() {
11338
+ if (this.buffer.length > 0) {
11339
+ const item = this.buffer.shift();
11340
+ this._wakeProducer();
11341
+ return item;
11342
+ }
11343
+ if (this.closed) return QUEUE_DONE;
11344
+ return new Promise((resolve4) => {
11345
+ this.consumerWaiters.push(resolve4);
11346
+ });
11347
+ }
11348
+ close() {
11349
+ if (this.closed) return;
11350
+ this.closed = true;
11351
+ for (const resolve4 of this.consumerWaiters) {
11352
+ resolve4(QUEUE_DONE);
11353
+ }
11354
+ this.consumerWaiters = [];
11355
+ for (const wake of this.producerWaiters) {
11356
+ wake();
11357
+ }
11358
+ this.producerWaiters = [];
11359
+ }
11360
+ get size() {
11361
+ return this.buffer.length;
11362
+ }
11363
+ _wakeProducer() {
11364
+ if (this.producerWaiters.length > 0) {
11365
+ this.producerWaiters.shift()();
13632
11366
  }
13633
11367
  }
13634
- const final = join4(current, ...rest);
13635
- return existsSync(final) ? final : void 0;
13636
- }
13637
- function probe(cmd, envKey, versionArgs = ["--version"]) {
13638
- const resolvedPath = resolveCmd(cmd, envKey);
13639
- if (!resolvedPath) return { name: cmd, found: false };
13640
- const v = spawnSync2(resolvedPath, versionArgs, { encoding: "utf8", shell: isWin });
13641
- const versionLine = (v.stdout || v.stderr).split(/\r?\n/)[0]?.trim();
13642
- return { name: cmd, found: true, path: resolvedPath, version: versionLine };
13643
- }
13644
- function probeOcrToolchain() {
13645
- return {
13646
- // poppler는 --version이 비표준이므로 -v 사용
13647
- pdftoppm: probe("pdftoppm", "KORDOC_PDFTOPPM_PATH", ["-v"]),
13648
- pdfinfo: probe("pdfinfo", "KORDOC_PDFINFO_PATH", ["-v"]),
13649
- pdftotext: probe("pdftotext", "KORDOC_PDFTOTEXT_PATH", ["-v"]),
13650
- soffice: probe("soffice", "KORDOC_SOFFICE_PATH", ["--version"])
13651
- };
13652
- }
13653
-
13654
- // src/setup/pm.ts
13655
- import { spawnSync as spawnSync3 } from "child_process";
13656
- var isWin2 = process.platform === "win32";
13657
- function which2(cmd) {
13658
- const finder = isWin2 ? "where" : "which";
13659
- const r = spawnSync3(finder, [cmd], { encoding: "utf8", shell: isWin2 });
13660
- return r.status === 0;
13661
- }
13662
- function detectPackageManager() {
13663
- if (process.platform === "darwin") {
13664
- return which2("brew") ? "brew" : "unknown";
13665
- }
13666
- if (isWin2) {
13667
- if (which2("winget")) return "winget";
13668
- if (which2("scoop")) return "scoop";
13669
- if (which2("choco")) return "choco";
13670
- return "unknown";
13671
- }
13672
- if (which2("apt-get") || which2("apt")) return "apt";
13673
- if (which2("dnf")) return "dnf";
13674
- if (which2("yum")) return "yum";
13675
- if (which2("pacman")) return "pacman";
13676
- if (which2("zypper")) return "zypper";
13677
- if (which2("apk")) return "apk";
13678
- return "unknown";
13679
- }
13680
-
13681
- // src/setup/install-commands.ts
13682
- var POPPLER_RECIPES = {
13683
- brew: { pm: "brew", cmd: "brew install poppler", needsSudo: false },
13684
- winget: { pm: "winget", cmd: "winget install --id oschwartz10612.Poppler", needsSudo: false, notes: "\uC2E4\uC874 \uD655\uC778\uB428 (winget-pkgs 25.07.0-0)" },
13685
- scoop: { pm: "scoop", cmd: "scoop install poppler", needsSudo: false, notes: "main \uBC84\uD0B7 \uD3EC\uD568, PATH \uC790\uB3D9 \uB4F1\uB85D" },
13686
- choco: { pm: "choco", cmd: "choco install poppler -y", needsSudo: true, notes: "\uAD00\uB9AC\uC790 PowerShell \uD544\uC694" },
13687
- apt: { pm: "apt", cmd: "sudo apt-get install -y poppler-utils", needsSudo: true },
13688
- dnf: { pm: "dnf", cmd: "sudo dnf install -y poppler-utils", needsSudo: true },
13689
- yum: { pm: "yum", cmd: "sudo yum install -y poppler-utils", needsSudo: true },
13690
- pacman: { pm: "pacman", cmd: "sudo pacman -S --noconfirm poppler", needsSudo: true },
13691
- zypper: { pm: "zypper", cmd: "sudo zypper install -y poppler-tools", needsSudo: true },
13692
- apk: { pm: "apk", cmd: "apk add --no-cache poppler-utils", needsSudo: false, notes: "Alpine/Docker \uCE5C\uD654" },
13693
- unknown: null
13694
- };
13695
- var LIBREOFFICE_RECIPES = {
13696
- brew: { pm: "brew", cmd: "brew install --cask libreoffice", needsSudo: false },
13697
- winget: { pm: "winget", cmd: "winget install --id TheDocumentFoundation.LibreOffice", needsSudo: false, notes: "\uC2E4\uC874 \uD655\uC778\uB428" },
13698
- scoop: { pm: "scoop", cmd: "scoop bucket add extras && scoop install libreoffice", needsSudo: false },
13699
- choco: { pm: "choco", cmd: "choco install libreoffice-fresh -y", needsSudo: true, notes: "\uAD00\uB9AC\uC790 PowerShell \uD544\uC694" },
13700
- apt: { pm: "apt", cmd: "sudo apt-get install -y libreoffice", needsSudo: true },
13701
- dnf: { pm: "dnf", cmd: "sudo dnf install -y libreoffice", needsSudo: true },
13702
- yum: { pm: "yum", cmd: "sudo yum install -y libreoffice", needsSudo: true },
13703
- pacman: { pm: "pacman", cmd: "sudo pacman -S --noconfirm libreoffice-still", needsSudo: true },
13704
- zypper: { pm: "zypper", cmd: "sudo zypper install -y libreoffice", needsSudo: true },
13705
- apk: { pm: "apk", cmd: "apk add --no-cache libreoffice", needsSudo: false, notes: "Alpine \uBBF8\uC9C0\uC6D0 \uAC00\uB2A5 \u2014 node:20-slim(Debian) \uAD8C\uC7A5" },
13706
- unknown: null
13707
11368
  };
13708
11369
 
13709
11370
  // src/pipeline/unified-ocr.ts
11371
+ var libreConvert = libre.convert;
13710
11372
  var UnifiedOcrError = class extends Error {
13711
11373
  code;
13712
11374
  stage;
@@ -13737,11 +11399,11 @@ var DEFAULT_MODEL_MAX_TOKENS = {
13737
11399
  };
13738
11400
  var DEFAULT_STAGE_WEIGHTS = {
13739
11401
  convert: 15,
13740
- render: 20,
11402
+ render: 15,
13741
11403
  probe: 5,
13742
- ocr: 45,
11404
+ ocr: 55,
13743
11405
  proofread: 0,
13744
- merge: 5
11406
+ merge: 10
13745
11407
  };
13746
11408
  var OCR_PROMPT2 = [
13747
11409
  "\uC774 PDF \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0\uC5D0\uC11C \uD14D\uC2A4\uD2B8\uC640 \uD45C\uB97C \uCD94\uCD9C\uD558\uC5EC Markdown\uC73C\uB85C \uBCC0\uD658\uD558\uACE0, OCR \uC624\uC778\uC2DD \uC624\uB958\uB97C \uC989\uC2DC \uAD50\uC815\uD558\uC5EC \uCD5C\uC885 \uACB0\uACFC\uBB3C\uC744 \uCD9C\uB825\uD558\uC138\uC694.",
@@ -13780,13 +11442,13 @@ function elapsedMs(startAt) {
13780
11442
  async function runUnifiedOcrPipeline(inputPath, options = {}) {
13781
11443
  const absInput = resolve3(inputPath);
13782
11444
  const stem = basename2(absInput, extname(absInput));
13783
- const workspaceDir = resolve3(options.workspaceDir ?? join5(dirname3(absInput), `${stem}_ocr_workspace`));
13784
- const imagesDir = join5(workspaceDir, "images");
13785
- const rawDir = join5(workspaceDir, "ocr", "raw");
13786
- const diffDir = join5(workspaceDir, "ocr", "diff");
13787
- const outputPath = resolve3(options.outputPath ?? join5(dirname3(absInput), `${stem}.md`));
13788
- const reportPath = join5(workspaceDir, "run-report.json");
13789
- const modelCachePath = join5(dirname3(absInput), ".kordoc-model-cache.json");
11445
+ const workspaceDir = resolve3(options.workspaceDir ?? join4(dirname3(absInput), `${stem}_ocr_workspace`));
11446
+ const imagesDir = join4(workspaceDir, "images");
11447
+ const rawDir = join4(workspaceDir, "ocr", "raw");
11448
+ const diffDir = join4(workspaceDir, "ocr", "diff");
11449
+ const outputPath = resolve3(options.outputPath ?? join4(dirname3(absInput), `${stem}.md`));
11450
+ const reportPath = join4(workspaceDir, "run-report.json");
11451
+ const modelCachePath = join4(dirname3(absInput), ".kordoc-model-cache.json");
13790
11452
  const baseUrl = options.baseUrl ?? "https://integrate.api.nvidia.com/v1/chat/completions";
13791
11453
  const timeoutMs = options.timeoutMs ?? 6e4;
13792
11454
  const maxRetriesPerPage = options.maxRetriesPerPage ?? 5;
@@ -13813,15 +11475,14 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
13813
11475
  };
13814
11476
  try {
13815
11477
  ensureSupportedInput(absInput);
13816
- const isPdfInput = extname(absInput).toLowerCase() === ".pdf";
13817
- await assertOcrToolchain(isPdfInput);
13818
11478
  let workingPdfPath = absInput;
13819
11479
  const convertStart = performance.now();
13820
11480
  currentStage = "convert";
13821
11481
  markStageStart("convert", "\uBB38\uC11C\uB97C PDF\uB85C \uBCC0\uD658 \uC911");
13822
11482
  logStage("info", "convert", "start", "\uBB38\uC11C\uB97C PDF\uB85C \uBCC0\uD658 \uC2DC\uC791", { input: absInput });
13823
- if (!isPdfInput) {
13824
- workingPdfPath = join5(workspaceDir, `${stem}.pdf`);
11483
+ if (extname(absInput).toLowerCase() !== ".pdf") {
11484
+ await assertSofficeAvailable();
11485
+ workingPdfPath = join4(workspaceDir, `${stem}.pdf`);
13825
11486
  const inputBuffer = await readFile(absInput);
13826
11487
  const out = await convertWithLibreOffice(inputBuffer, ".pdf");
13827
11488
  await writeFile(workingPdfPath, out);
@@ -13831,39 +11492,19 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
13831
11492
  logStage("info", "convert", "done", "PDF \uBCC0\uD658 \uC644\uB8CC", { elapsedMs: timingsMs.convert });
13832
11493
  const renderStart = performance.now();
13833
11494
  currentStage = "render";
11495
+ const totalPages = await getPdfPageCount(workingPdfPath).catch(() => 0);
11496
+ if (totalPages === 0) throw new UnifiedOcrError("RENDER_FAILED", "render", "\uD398\uC774\uC9C0 \uC218\uB97C \uD655\uC778\uD560 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4.");
13834
11497
  markStageStart("render", "PDF \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC911");
13835
- logStage("info", "render", "start", "PDF \uD398\uC774\uC9C0 \uB80C\uB354\uB9C1 \uC2DC\uC791", { pdf: workingPdfPath, dpi });
13836
- const renderWithProgress = await renderPdfToPngWithProgress(
13837
- workingPdfPath,
13838
- join5(imagesDir, "page"),
13839
- dpi,
13840
- (current, total) => {
13841
- markStageProgress(
13842
- "render",
13843
- Math.round(current / total * 100),
13844
- current,
13845
- total,
13846
- `\uD398\uC774\uC9C0 ${current}/${total} \uB80C\uB354\uB9C1`
13847
- );
13848
- }
13849
- );
13850
- const images = await listPageImages(imagesDir);
13851
- if (images.length === 0) throw new UnifiedOcrError("RENDER_FAILED", "render", "\uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC2E4\uD328: \uACB0\uACFC \uC774\uBBF8\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
13852
- if (!renderWithProgress.emittedPerPageProgress) {
13853
- markStageProgress("render", 100, images.length, images.length, `\uD398\uC774\uC9C0 ${images.length}\uC7A5 \uC0DD\uC131`);
13854
- }
13855
- timingsMs.render = elapsedMs(renderStart);
13856
- markStageDone("render", "\uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC644\uB8CC");
13857
- logStage("info", "render", "done", "\uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC644\uB8CC", {
13858
- pages: images.length,
13859
- elapsedMs: timingsMs.render,
13860
- pageCountSource: renderWithProgress.pageCountSource
13861
- });
11498
+ logStage("info", "render", "start", "PDF \uD398\uC774\uC9C0 \uB80C\uB354\uB9C1 \uC2DC\uC791", { pdf: workingPdfPath, dpi, totalPages });
11499
+ await runCommand("pdftoppm", ["-png", "-r", String(dpi), "-f", "1", "-l", "1", workingPdfPath, join4(imagesDir, "page")]);
11500
+ const firstFiles = (await readdir(imagesDir)).filter((f) => f.endsWith(".png")).sort((a, b) => naturalPageSort(a, b));
11501
+ if (firstFiles.length === 0) throw new UnifiedOcrError("RENDER_FAILED", "render", "\uCCAB \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC2E4\uD328");
11502
+ const probeImage = join4(imagesDir, firstFiles[0]);
11503
+ markStageProgress("render", Math.round(1 / totalPages * 100), 1, totalPages, `\uD398\uC774\uC9C0 1/${totalPages} \uB80C\uB354\uB9C1`);
13862
11504
  const probeStart = performance.now();
13863
11505
  currentStage = "probe";
13864
11506
  markStageStart("probe", "\uBAA8\uB378 \uC18D\uB3C4 \uD504\uB85C\uBE0C \uC218\uD589 \uC911");
13865
11507
  logStage("info", "probe", "start", "\uBAA8\uB378 \uC18D\uB3C4 \uD504\uB85C\uBE0C \uC2DC\uC791", { models, probeConcurrency });
13866
- const probeImage = await pickRepresentativeImage(images);
13867
11508
  let probeDone = 0;
13868
11509
  const probeRuns = startParallelProbeRuns({
13869
11510
  models,
@@ -13874,10 +11515,10 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
13874
11515
  keyPool,
13875
11516
  timeoutMs,
13876
11517
  logger,
13877
- onProbeResult: ({ index: index2, model, result }) => {
11518
+ onProbeResult: ({ index, model, result }) => {
13878
11519
  probeDone += 1;
13879
11520
  markStageProgress("probe", Math.round(probeDone / models.length * 100), probeDone, models.length, `\uBAA8\uB378 \uD504\uB85C\uBE0C ${probeDone}/${models.length}`);
13880
- logStage("debug", "probe", "progress", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC9C4\uD589", { index: index2 + 1, total: models.length, model, result });
11521
+ logStage("debug", "probe", "progress", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC9C4\uD589", { index: index + 1, total: models.length, model, result });
13881
11522
  }
13882
11523
  });
13883
11524
  const selected = await probeRuns.firstSuccess;
@@ -13890,33 +11531,74 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
13890
11531
  await updateModelCache(modelCachePath, results);
13891
11532
  return results;
13892
11533
  });
11534
+ const concurrencyPerKey = Math.max(1, options.concurrencyPerKey ?? 2);
11535
+ const keyCount = keyPool.snapshot().length;
11536
+ const workerCount = Math.max(1, keyCount * concurrencyPerKey);
11537
+ const queueCapacity = workerCount * 2;
11538
+ const queue = new BoundedQueue(queueCapacity);
13893
11539
  const ocrStart = performance.now();
13894
11540
  currentStage = "ocr";
13895
- markStageStart("ocr", `OCR \uC9C4\uD589 \uC911 (${selectedModel})`);
13896
- logStage("info", "ocr", "start", "\uD398\uC774\uC9C0 OCR \uC2DC\uC791", { selectedModel, pageCount: images.length });
13897
- const rawPagePaths = [];
13898
- for (let i = 0; i < images.length; i++) {
13899
- const imagePath = images[i];
13900
- const markdown = await ocrImageWithFallback({
13901
- imagePath,
13902
- prompt: OCR_PROMPT2,
13903
- models: fallbackModelOrder,
13904
- modelMaxTokens,
13905
- baseUrl,
13906
- keyPool,
13907
- timeoutMs,
13908
- maxRetriesPerPage,
11541
+ markStageStart("ocr", `OCR \uC9C4\uD589 \uC911 (\uC6CC\uCEE4 ${workerCount}\uAC1C)`);
11542
+ logStage("info", "ocr", "start", "\uD398\uC774\uC9C0 OCR \uC2DC\uC791", { workerCount, keyCount, pageCount: totalPages });
11543
+ let renderDone = 1;
11544
+ const renderProducer = (async () => {
11545
+ try {
11546
+ await queue.enqueue({ pageNumber: 1, imagePath: probeImage });
11547
+ if (totalPages > 1) {
11548
+ for await (const item of renderPdfToPngStream(workingPdfPath, join4(imagesDir, "page"), dpi, totalPages, 2)) {
11549
+ await queue.enqueue(item);
11550
+ renderDone++;
11551
+ markStageProgress("render", Math.round(renderDone / totalPages * 100), renderDone, totalPages, `\uD398\uC774\uC9C0 ${renderDone}/${totalPages} \uB80C\uB354\uB9C1`);
11552
+ logStage("debug", "render", "progress", "\uD398\uC774\uC9C0 \uB80C\uB354 \uC644\uB8CC", { page: item.pageNumber });
11553
+ }
11554
+ }
11555
+ } finally {
11556
+ queue.close();
11557
+ timingsMs.render = elapsedMs(renderStart);
11558
+ markStageDone("render", "\uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC644\uB8CC");
11559
+ logStage("info", "render", "done", "\uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC644\uB8CC", { pages: renderDone, elapsedMs: timingsMs.render });
11560
+ }
11561
+ })();
11562
+ const [, pageResultsMap] = await Promise.all([
11563
+ renderProducer,
11564
+ ocrWorkerPool({
11565
+ queue,
11566
+ workerCount,
11567
+ totalPages,
11568
+ ocrInput: {
11569
+ prompt: OCR_PROMPT2,
11570
+ models: fallbackModelOrder,
11571
+ modelMaxTokens,
11572
+ baseUrl,
11573
+ keyPool,
11574
+ timeoutMs,
11575
+ maxRetriesPerPage,
11576
+ logger
11577
+ },
11578
+ onPageDone: (pageNumber, completedCount, model) => {
11579
+ markStageProgress(
11580
+ "ocr",
11581
+ Math.round(completedCount / totalPages * 100),
11582
+ completedCount,
11583
+ totalPages,
11584
+ `OCR ${completedCount}/${totalPages}`,
11585
+ model || void 0
11586
+ );
11587
+ logStage("debug", "ocr", "progress", "\uD398\uC774\uC9C0 OCR \uC644\uB8CC", { page: pageNumber, total: totalPages, model });
11588
+ },
13909
11589
  logger
13910
- });
13911
- const pagePath = join5(rawDir, `page_${String(i + 1).padStart(4, "0")}.md`);
13912
- await writeFile(pagePath, markdown, "utf-8");
13913
- rawPagePaths.push(pagePath);
13914
- markStageProgress("ocr", Math.round((i + 1) / images.length * 100), i + 1, images.length, `OCR ${i + 1}/${images.length}`, selectedModel);
13915
- logStage("debug", "ocr", "progress", "\uD398\uC774\uC9C0 OCR \uC644\uB8CC", { page: i + 1, total: images.length });
13916
- }
11590
+ })
11591
+ ]);
13917
11592
  timingsMs.ocr = elapsedMs(ocrStart);
13918
11593
  markStageDone("ocr", "OCR \uC644\uB8CC");
13919
11594
  logStage("info", "ocr", "done", "\uD398\uC774\uC9C0 OCR \uC644\uB8CC", { elapsedMs: timingsMs.ocr });
11595
+ const sortedEntries = Array.from(pageResultsMap.entries()).sort((a, b) => a[0] - b[0]);
11596
+ const rawPagePaths = [];
11597
+ for (const [pageNum, markdown] of sortedEntries) {
11598
+ const pagePath = join4(rawDir, `page_${String(pageNum).padStart(4, "0")}.md`);
11599
+ await writeFile(pagePath, markdown, "utf-8");
11600
+ rawPagePaths.push(pagePath);
11601
+ }
13920
11602
  const mergeStart = performance.now();
13921
11603
  currentStage = "merge";
13922
11604
  markStageStart("merge", "\uCD5C\uC885 Markdown \uBCD1\uD569 \uC911");
@@ -13933,7 +11615,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
13933
11615
  selectedModel,
13934
11616
  probeImage,
13935
11617
  probeResults: await probeResultsPromise,
13936
- pageCount: images.length,
11618
+ pageCount: totalPages,
13937
11619
  keyHealth: keyPool.snapshot(),
13938
11620
  timingsMs,
13939
11621
  modelCachePath
@@ -14000,37 +11682,17 @@ function emitProgress(cb, stage, stagePercent, weights, extra) {
14000
11682
  model: extra.model
14001
11683
  });
14002
11684
  }
14003
- async function loadLibreConvert() {
14004
- try {
14005
- const m = await Promise.resolve().then(() => __toESM(require_libreoffice_convert(), 1));
14006
- return (m.default ?? m).convert;
14007
- } catch {
14008
- throw new UnifiedOcrError(
14009
- "LIBRE_NOT_INSTALLED",
14010
- "convert",
14011
- "libreoffice-convert npm \uD328\uD0A4\uC9C0\uAC00 \uC124\uCE58\uB418\uC5B4 \uC788\uC9C0 \uC54A\uC2B5\uB2C8\uB2E4.\n npm install libreoffice-convert\n\uB610\uB294 npx kordoc setup \uC744 \uC2E4\uD589\uD558\uC138\uC694."
14012
- );
14013
- }
14014
- }
14015
11685
  async function convertWithLibreOffice(buffer, ext) {
14016
- const libreConvert = await loadLibreConvert();
14017
- return await new Promise((resolvePromise, reject2) => {
11686
+ return await new Promise((resolvePromise, reject) => {
14018
11687
  libreConvert(buffer, ext, void 0, (err, done) => {
14019
11688
  if (err || !done) {
14020
- reject2(new UnifiedOcrError("CONVERT_FAILED", "convert", err?.message ?? "LibreOffice \uBCC0\uD658 \uC2E4\uD328"));
11689
+ reject(new UnifiedOcrError("CONVERT_FAILED", "convert", err?.message ?? "LibreOffice \uBCC0\uD658 \uC2E4\uD328"));
14021
11690
  return;
14022
11691
  }
14023
11692
  resolvePromise(done);
14024
11693
  });
14025
11694
  });
14026
11695
  }
14027
- async function renderPdfToPng(pdfPath, prefixPath, dpi) {
14028
- try {
14029
- await runCommand("pdftoppm", ["-png", "-r", String(dpi), pdfPath, prefixPath]);
14030
- } catch (err) {
14031
- throw new UnifiedOcrError("RENDER_FAILED", "render", err instanceof Error ? err.message : String(err));
14032
- }
14033
- }
14034
11696
  async function getPdfPageCount(pdfPath) {
14035
11697
  const stdout = await runCommandWithStdout("pdfinfo", [pdfPath]);
14036
11698
  const m = stdout.match(/^\s*Pages:\s*(\d+)\s*$/mi);
@@ -14043,53 +11705,50 @@ async function getPdfPageCount(pdfPath) {
14043
11705
  }
14044
11706
  return n;
14045
11707
  }
14046
- async function renderPdfToPngWithProgress(pdfPath, prefixPath, dpi, onPageDone) {
14047
- let totalPages = 0;
14048
- try {
14049
- totalPages = await getPdfPageCount(pdfPath);
14050
- } catch {
14051
- totalPages = 0;
14052
- }
14053
- if (totalPages > 0) {
11708
+ async function* renderPdfToPngStream(pdfPath, prefixPath, dpi, totalPages, startPage = 1) {
11709
+ const imagesDir = dirname3(prefixPath);
11710
+ for (let page = startPage; page <= totalPages; page++) {
14054
11711
  try {
14055
- for (let page = 1; page <= totalPages; page++) {
14056
- await runCommand("pdftoppm", [
14057
- "-png",
14058
- "-r",
14059
- String(dpi),
14060
- "-f",
14061
- String(page),
14062
- "-l",
14063
- String(page),
14064
- pdfPath,
14065
- prefixPath
14066
- ]);
14067
- onPageDone(page, totalPages);
14068
- }
14069
- return { emittedPerPageProgress: true, pageCountSource: "pdfinfo" };
11712
+ await runCommand("pdftoppm", [
11713
+ "-png",
11714
+ "-r",
11715
+ String(dpi),
11716
+ "-f",
11717
+ String(page),
11718
+ "-l",
11719
+ String(page),
11720
+ pdfPath,
11721
+ prefixPath
11722
+ ]);
11723
+ const files = await readdir(imagesDir);
11724
+ const pageFiles = files.filter((f) => f.endsWith(".png")).sort((a, b) => naturalPageSort(a, b));
11725
+ const imagePath = join4(imagesDir, pageFiles[pageFiles.length - 1]);
11726
+ yield { pageNumber: page, imagePath };
14070
11727
  } catch (err) {
14071
- throw new UnifiedOcrError("RENDER_FAILED", "render", err instanceof Error ? err.message : String(err));
11728
+ yield {
11729
+ pageNumber: page,
11730
+ imagePath: null,
11731
+ error: err instanceof Error ? err : new Error(String(err))
11732
+ };
14072
11733
  }
14073
11734
  }
14074
- await renderPdfToPng(pdfPath, prefixPath, dpi);
14075
- return { emittedPerPageProgress: false, pageCountSource: "fallback" };
14076
11735
  }
14077
11736
  async function runCommand(cmd, args) {
14078
- await new Promise((resolvePromise, reject2) => {
11737
+ await new Promise((resolvePromise, reject) => {
14079
11738
  const child = spawn2(cmd, args, { stdio: "pipe" });
14080
11739
  let stderr = "";
14081
11740
  child.stderr.on("data", (d) => {
14082
11741
  stderr += String(d);
14083
11742
  });
14084
- child.on("error", reject2);
11743
+ child.on("error", reject);
14085
11744
  child.on("close", (code) => {
14086
11745
  if (code === 0) resolvePromise();
14087
- else reject2(new Error(`${cmd} \uC2E4\uD328 (code=${code}): ${stderr.trim()}`));
11746
+ else reject(new Error(`${cmd} \uC2E4\uD328 (code=${code}): ${stderr.trim()}`));
14088
11747
  });
14089
11748
  });
14090
11749
  }
14091
11750
  async function runCommandWithStdout(cmd, args) {
14092
- return await new Promise((resolvePromise, reject2) => {
11751
+ return await new Promise((resolvePromise, reject) => {
14093
11752
  const child = spawn2(cmd, args, { stdio: "pipe" });
14094
11753
  let stdout = "";
14095
11754
  let stderr = "";
@@ -14099,52 +11758,25 @@ async function runCommandWithStdout(cmd, args) {
14099
11758
  child.stderr.on("data", (d) => {
14100
11759
  stderr += String(d);
14101
11760
  });
14102
- child.on("error", reject2);
11761
+ child.on("error", reject);
14103
11762
  child.on("close", (code) => {
14104
11763
  if (code === 0) resolvePromise(stdout);
14105
- else reject2(new Error(`${cmd} \uC2E4\uD328 (code=${code}): ${stderr.trim()}`));
11764
+ else reject(new Error(`${cmd} \uC2E4\uD328 (code=${code}): ${stderr.trim()}`));
14106
11765
  });
14107
11766
  });
14108
11767
  }
14109
- async function assertOcrToolchain(isPdfInput = false) {
14110
- const t = probeOcrToolchain();
14111
- const pm = detectPackageManager();
14112
- const missing = [];
14113
- if (!t.pdftoppm.found || !t.pdfinfo.found) missing.push("poppler");
14114
- if (!isPdfInput && !t.soffice.found) missing.push("libreoffice");
14115
- if (missing.length === 0) return;
14116
- const lines = ["OCR \uD30C\uC774\uD504\uB77C\uC778\uC5D0 \uD544\uC694\uD55C \uC2DC\uC2A4\uD15C \uB3C4\uAD6C\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4."];
14117
- if (missing.includes("poppler")) {
14118
- const r = POPPLER_RECIPES[pm];
14119
- lines.push(` - poppler (pdftoppm/pdfinfo): ${r ? r.cmd : "OS\uBCC4 \uC124\uCE58 \uC548\uB0B4\uB294 README \uCC38\uACE0"}`);
14120
- }
14121
- if (missing.includes("libreoffice")) {
14122
- const r = LIBREOFFICE_RECIPES[pm];
14123
- lines.push(` - LibreOffice (soffice): ${r ? r.cmd : "OS\uBCC4 \uC124\uCE58 \uC548\uB0B4\uB294 README \uCC38\uACE0"}`);
14124
- }
14125
- lines.push("\uC790\uC138\uD55C \uC810\uAC80: npx kordoc doctor");
14126
- throw new UnifiedOcrError("TOOLCHAIN_MISSING", "convert", lines.join("\n"));
14127
- }
14128
- async function listPageImages(imagesDir) {
14129
- const files = await readdir(imagesDir);
14130
- return files.filter((f) => f.endsWith(".png")).sort((a, b) => naturalPageSort(a, b)).map((f) => join5(imagesDir, f));
11768
+ async function assertSofficeAvailable() {
11769
+ try {
11770
+ await runCommand("soffice", ["--version"]);
11771
+ } catch {
11772
+ throw new UnifiedOcrError("SOFFICE_NOT_FOUND", "convert", "soffice\uB97C \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4. LibreOffice\uB97C \uC124\uCE58\uD574 \uC8FC\uC138\uC694.");
11773
+ }
14131
11774
  }
14132
11775
  function naturalPageSort(a, b) {
14133
11776
  const na = Number((a.match(/\d+/g) || ["0"]).at(-1) || 0);
14134
11777
  const nb = Number((b.match(/\d+/g) || ["0"]).at(-1) || 0);
14135
11778
  return na - nb;
14136
11779
  }
14137
- async function pickRepresentativeImage(images) {
14138
- const sample = images.slice(0, Math.min(images.length, 8));
14139
- const weighted = [];
14140
- for (const p of sample) {
14141
- const st = await stat(p);
14142
- if (st.size > 8 * 1024) weighted.push({ path: p, size: st.size });
14143
- }
14144
- const use = weighted.length > 0 ? weighted : await Promise.all(sample.map(async (p) => ({ path: p, size: (await stat(p)).size })));
14145
- use.sort((a, b) => a.size - b.size);
14146
- return use[Math.floor(use.length / 2)].path;
14147
- }
14148
11780
  async function mapWithConcurrency(items, concurrency, mapper) {
14149
11781
  const results = new Array(items.length);
14150
11782
  let nextIndex = 0;
@@ -14165,12 +11797,12 @@ function startParallelProbeRuns(input) {
14165
11797
  let doneCount = 0;
14166
11798
  let resolveFirst;
14167
11799
  let rejectFirst;
14168
- const firstSuccess = new Promise((resolve4, reject2) => {
11800
+ const firstSuccess = new Promise((resolve4, reject) => {
14169
11801
  resolveFirst = resolve4;
14170
- rejectFirst = reject2;
11802
+ rejectFirst = reject;
14171
11803
  });
14172
11804
  let lastErr = "\uC18D\uB3C4 \uD504\uB85C\uBE0C \uC2E4\uD328: \uC0AC\uC6A9 \uAC00\uB2A5\uD55C OCR \uBAA8\uB378\uC774 \uC5C6\uC2B5\uB2C8\uB2E4.";
14173
- const allResults = mapWithConcurrency(input.models, input.probeConcurrency, async (model, index2) => {
11805
+ const allResults = mapWithConcurrency(input.models, input.probeConcurrency, async (model, index) => {
14174
11806
  const t0 = performance.now();
14175
11807
  try {
14176
11808
  await ocrImageViaNim({
@@ -14186,7 +11818,7 @@ function startParallelProbeRuns(input) {
14186
11818
  stage: "probe"
14187
11819
  });
14188
11820
  const result = { model, durationMs: elapsedMs(t0), success: true };
14189
- input.onProbeResult?.({ index: index2, model, result });
11821
+ input.onProbeResult?.({ index, model, result });
14190
11822
  if (!firstResolved) {
14191
11823
  firstResolved = true;
14192
11824
  resolveFirst?.({ selectedModel: model, firstDurationMs: result.durationMs });
@@ -14200,7 +11832,7 @@ function startParallelProbeRuns(input) {
14200
11832
  error: err instanceof Error ? err.message : String(err)
14201
11833
  };
14202
11834
  lastErr = result.error ?? lastErr;
14203
- input.onProbeResult?.({ index: index2, model, result });
11835
+ input.onProbeResult?.({ index, model, result });
14204
11836
  return result;
14205
11837
  } finally {
14206
11838
  doneCount += 1;
@@ -14246,11 +11878,43 @@ async function updateModelCache(path, probes) {
14246
11878
  current.updatedAt = (/* @__PURE__ */ new Date()).toISOString();
14247
11879
  await writeFile(path, JSON.stringify(current, null, 2), "utf-8");
14248
11880
  }
11881
+ async function ocrWorkerPool(input) {
11882
+ const { queue, workerCount, ocrInput, onPageDone } = input;
11883
+ const results = /* @__PURE__ */ new Map();
11884
+ let completedCount = 0;
11885
+ async function worker() {
11886
+ while (true) {
11887
+ const item = await queue.dequeue();
11888
+ if (item === QUEUE_DONE) break;
11889
+ const { pageNumber, imagePath, error } = item;
11890
+ if (imagePath === null) {
11891
+ input.logger?.log({
11892
+ level: "warn",
11893
+ stage: "ocr",
11894
+ event: "message",
11895
+ message: `\uD398\uC774\uC9C0 ${pageNumber} \uB80C\uB354 \uC2E4\uD328 \u2014 \uBE48 \uD398\uC774\uC9C0\uB85C \uCC98\uB9AC`,
11896
+ meta: { error: String(error) }
11897
+ });
11898
+ results.set(pageNumber, "");
11899
+ completedCount++;
11900
+ onPageDone(pageNumber, completedCount, "");
11901
+ continue;
11902
+ }
11903
+ const { markdown, model } = await ocrImageWithFallback({ ...ocrInput, imagePath });
11904
+ results.set(pageNumber, markdown);
11905
+ completedCount++;
11906
+ onPageDone(pageNumber, completedCount, model);
11907
+ }
11908
+ }
11909
+ const workers = Array.from({ length: workerCount }, () => worker());
11910
+ await Promise.all(workers);
11911
+ return results;
11912
+ }
14249
11913
  async function ocrImageWithFallback(input) {
14250
11914
  let lastErr = "\uC54C \uC218 \uC5C6\uB294 \uC624\uB958";
14251
11915
  for (const model of input.models) {
14252
11916
  try {
14253
- return await ocrImageViaNim({
11917
+ const markdown = await ocrImageViaNim({
14254
11918
  imagePath: input.imagePath,
14255
11919
  prompt: input.prompt,
14256
11920
  model,
@@ -14262,6 +11926,7 @@ async function ocrImageWithFallback(input) {
14262
11926
  logger: input.logger,
14263
11927
  stage: "ocr"
14264
11928
  });
11929
+ return { markdown, model };
14265
11930
  } catch (err) {
14266
11931
  lastErr = err instanceof Error ? err.message : String(err);
14267
11932
  }
@@ -14607,14 +12272,5 @@ export {
14607
12272
 
14608
12273
  cfb/cfb.js:
14609
12274
  (*! crc32.js (C) 2014-present SheetJS -- http://sheetjs.com *)
14610
-
14611
- tmp/lib/tmp.js:
14612
- (*!
14613
- * Tmp
14614
- *
14615
- * Copyright (c) 2011-2017 KARASZI Istvan <github@spam.raszi.hu>
14616
- *
14617
- * MIT Licensed
14618
- *)
14619
12275
  */
14620
12276
  //# sourceMappingURL=index.js.map