koztv-blog-tools 1.2.4 → 1.2.6

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 (3) hide show
  1. package/dist/index.js +91 -27
  2. package/dist/index.mjs +91 -27
  3. package/package.json +1 -1
package/dist/index.js CHANGED
@@ -374,6 +374,42 @@ var import_sessions = require("telegram/sessions");
374
374
  var fs = __toESM(require("fs"));
375
375
  var path = __toESM(require("path"));
376
376
  var readline = __toESM(require("readline"));
377
+ function entitiesToMarkdown(text, entities) {
378
+ if (!entities || entities.length === 0) return text;
379
+ const sorted = [...entities].sort((a, b) => b.offset - a.offset);
380
+ let result = text;
381
+ for (const entity of sorted) {
382
+ const start = entity.offset;
383
+ const end = entity.offset + entity.length;
384
+ const content = result.substring(start, end);
385
+ let replacement = content;
386
+ if (entity instanceof import_telegram.Api.MessageEntityBold) {
387
+ replacement = `**${content}**`;
388
+ } else if (entity instanceof import_telegram.Api.MessageEntityItalic) {
389
+ replacement = `*${content}*`;
390
+ } else if (entity instanceof import_telegram.Api.MessageEntityCode) {
391
+ replacement = `\`${content}\``;
392
+ } else if (entity instanceof import_telegram.Api.MessageEntityPre) {
393
+ replacement = `\`\`\`
394
+ ${content}
395
+ \`\`\``;
396
+ } else if (entity instanceof import_telegram.Api.MessageEntityStrike) {
397
+ replacement = `~~${content}~~`;
398
+ } else if (entity instanceof import_telegram.Api.MessageEntityUnderline) {
399
+ replacement = `**${content}**`;
400
+ } else if (entity instanceof import_telegram.Api.MessageEntityTextUrl) {
401
+ replacement = `[${content}](${entity.url})`;
402
+ } else if (entity instanceof import_telegram.Api.MessageEntityUrl) {
403
+ replacement = content;
404
+ } else if (entity instanceof import_telegram.Api.MessageEntityMention) {
405
+ replacement = content;
406
+ } else if (entity instanceof import_telegram.Api.MessageEntityHashtag) {
407
+ replacement = content;
408
+ }
409
+ result = result.substring(0, start) + replacement + result.substring(end);
410
+ }
411
+ return result;
412
+ }
377
413
  async function defaultReadline(prompt) {
378
414
  const rl = readline.createInterface({
379
415
  input: process.stdin,
@@ -497,40 +533,47 @@ async function exportTelegramChannel(options) {
497
533
  const postMediaDir = path.join(mediaDir, paddedId);
498
534
  const mediaFiles = [];
499
535
  if (downloadMedia && message.media) {
500
- fs.mkdirSync(postMediaDir, { recursive: true });
501
- try {
502
- const buffer = await client.downloadMedia(message.media, {});
503
- if (buffer) {
504
- let ext = ".bin";
505
- if (message.media instanceof import_telegram.Api.MessageMediaPhoto) {
506
- ext = ".jpg";
507
- } else if (message.media instanceof import_telegram.Api.MessageMediaDocument) {
508
- const doc = message.media.document;
509
- if (doc instanceof import_telegram.Api.Document) {
510
- const mimeExt = doc.mimeType?.split("/")[1];
511
- if (mimeExt) {
512
- ext = "." + mimeExt.replace("jpeg", "jpg");
513
- }
514
- for (const attr of doc.attributes) {
515
- if (attr instanceof import_telegram.Api.DocumentAttributeVideo) {
516
- ext = ".mp4";
536
+ const existingFiles = fs.existsSync(postMediaDir) ? fs.readdirSync(postMediaDir).filter((f) => f.startsWith("media.")) : [];
537
+ if (existingFiles.length > 0) {
538
+ for (const f of existingFiles) {
539
+ mediaFiles.push(`media/${paddedId}/${f}`);
540
+ }
541
+ } else {
542
+ fs.mkdirSync(postMediaDir, { recursive: true });
543
+ try {
544
+ const buffer = await client.downloadMedia(message.media, {});
545
+ if (buffer) {
546
+ let ext = ".bin";
547
+ if (message.media instanceof import_telegram.Api.MessageMediaPhoto) {
548
+ ext = ".jpg";
549
+ } else if (message.media instanceof import_telegram.Api.MessageMediaDocument) {
550
+ const doc = message.media.document;
551
+ if (doc instanceof import_telegram.Api.Document) {
552
+ const mimeExt = doc.mimeType?.split("/")[1];
553
+ if (mimeExt) {
554
+ ext = "." + mimeExt.replace("jpeg", "jpg");
517
555
  }
518
- if (attr instanceof import_telegram.Api.DocumentAttributeFilename) {
519
- ext = path.extname(attr.fileName) || ext;
556
+ for (const attr of doc.attributes) {
557
+ if (attr instanceof import_telegram.Api.DocumentAttributeVideo) {
558
+ ext = ".mp4";
559
+ }
560
+ if (attr instanceof import_telegram.Api.DocumentAttributeFilename) {
561
+ ext = path.extname(attr.fileName) || ext;
562
+ }
520
563
  }
521
564
  }
522
565
  }
566
+ const mediaFileName = `media${ext}`;
567
+ const mediaPath = path.join(postMediaDir, mediaFileName);
568
+ fs.writeFileSync(mediaPath, buffer);
569
+ mediaFiles.push(`media/${paddedId}/${mediaFileName}`);
523
570
  }
524
- const mediaFileName = `media${ext}`;
525
- const mediaPath = path.join(postMediaDir, mediaFileName);
526
- fs.writeFileSync(mediaPath, buffer);
527
- mediaFiles.push(`media/${paddedId}/${mediaFileName}`);
571
+ } catch (e) {
572
+ console.error(`Error downloading media for message ${msgId}:`, e);
528
573
  }
529
- } catch (e) {
530
- console.error(`Error downloading media for message ${msgId}:`, e);
531
574
  }
532
575
  }
533
- const content = message.message || "";
576
+ const content = entitiesToMarkdown(message.message || "", message.entities);
534
577
  const link = channelMeta.username ? `https://t.me/${channelMeta.username}/${msgId}` : "";
535
578
  const post = {
536
579
  msgId,
@@ -664,9 +707,26 @@ async function processPost(post, options, exportDir) {
664
707
  const { title: originalTitle, body: originalBody } = extractTitleAndBody(cleanedContent);
665
708
  const date = post.date.toISOString().split("T")[0];
666
709
  const languages = [];
710
+ const translationExists = (lang) => {
711
+ const langDir = path2.join(outputDir, lang);
712
+ if (!fs2.existsSync(langDir)) return false;
713
+ try {
714
+ for (const slug of fs2.readdirSync(langDir)) {
715
+ const indexPath = path2.join(langDir, slug, "index.md");
716
+ if (fs2.existsSync(indexPath)) {
717
+ const content = fs2.readFileSync(indexPath, "utf-8");
718
+ if (content.includes(`original_link: "${post.link}"`)) {
719
+ return true;
720
+ }
721
+ }
722
+ }
723
+ } catch {
724
+ }
725
+ return false;
726
+ };
667
727
  if (translate && translate.targetLangs.length > 0) {
668
728
  const sourceLang = translate.sourceLang || "ru";
669
- if (translate.keepOriginal) {
729
+ if (translate.keepOriginal && !translationExists(sourceLang)) {
670
730
  languages.push({
671
731
  lang: sourceLang,
672
732
  title: originalTitle,
@@ -676,6 +736,10 @@ async function processPost(post, options, exportDir) {
676
736
  }
677
737
  for (const targetLang of translate.targetLangs) {
678
738
  if (targetLang === sourceLang) continue;
739
+ if (translationExists(targetLang)) {
740
+ onProgress?.(` ${targetLang}: already exists, skipping`);
741
+ continue;
742
+ }
679
743
  onProgress?.(` Translating to ${targetLang}...`);
680
744
  const translateOpts = {
681
745
  apiKey: translate.apiKey,
package/dist/index.mjs CHANGED
@@ -317,6 +317,42 @@ import { StringSession } from "telegram/sessions";
317
317
  import * as fs from "fs";
318
318
  import * as path from "path";
319
319
  import * as readline from "readline";
320
+ function entitiesToMarkdown(text, entities) {
321
+ if (!entities || entities.length === 0) return text;
322
+ const sorted = [...entities].sort((a, b) => b.offset - a.offset);
323
+ let result = text;
324
+ for (const entity of sorted) {
325
+ const start = entity.offset;
326
+ const end = entity.offset + entity.length;
327
+ const content = result.substring(start, end);
328
+ let replacement = content;
329
+ if (entity instanceof Api.MessageEntityBold) {
330
+ replacement = `**${content}**`;
331
+ } else if (entity instanceof Api.MessageEntityItalic) {
332
+ replacement = `*${content}*`;
333
+ } else if (entity instanceof Api.MessageEntityCode) {
334
+ replacement = `\`${content}\``;
335
+ } else if (entity instanceof Api.MessageEntityPre) {
336
+ replacement = `\`\`\`
337
+ ${content}
338
+ \`\`\``;
339
+ } else if (entity instanceof Api.MessageEntityStrike) {
340
+ replacement = `~~${content}~~`;
341
+ } else if (entity instanceof Api.MessageEntityUnderline) {
342
+ replacement = `**${content}**`;
343
+ } else if (entity instanceof Api.MessageEntityTextUrl) {
344
+ replacement = `[${content}](${entity.url})`;
345
+ } else if (entity instanceof Api.MessageEntityUrl) {
346
+ replacement = content;
347
+ } else if (entity instanceof Api.MessageEntityMention) {
348
+ replacement = content;
349
+ } else if (entity instanceof Api.MessageEntityHashtag) {
350
+ replacement = content;
351
+ }
352
+ result = result.substring(0, start) + replacement + result.substring(end);
353
+ }
354
+ return result;
355
+ }
320
356
  async function defaultReadline(prompt) {
321
357
  const rl = readline.createInterface({
322
358
  input: process.stdin,
@@ -440,40 +476,47 @@ async function exportTelegramChannel(options) {
440
476
  const postMediaDir = path.join(mediaDir, paddedId);
441
477
  const mediaFiles = [];
442
478
  if (downloadMedia && message.media) {
443
- fs.mkdirSync(postMediaDir, { recursive: true });
444
- try {
445
- const buffer = await client.downloadMedia(message.media, {});
446
- if (buffer) {
447
- let ext = ".bin";
448
- if (message.media instanceof Api.MessageMediaPhoto) {
449
- ext = ".jpg";
450
- } else if (message.media instanceof Api.MessageMediaDocument) {
451
- const doc = message.media.document;
452
- if (doc instanceof Api.Document) {
453
- const mimeExt = doc.mimeType?.split("/")[1];
454
- if (mimeExt) {
455
- ext = "." + mimeExt.replace("jpeg", "jpg");
456
- }
457
- for (const attr of doc.attributes) {
458
- if (attr instanceof Api.DocumentAttributeVideo) {
459
- ext = ".mp4";
479
+ const existingFiles = fs.existsSync(postMediaDir) ? fs.readdirSync(postMediaDir).filter((f) => f.startsWith("media.")) : [];
480
+ if (existingFiles.length > 0) {
481
+ for (const f of existingFiles) {
482
+ mediaFiles.push(`media/${paddedId}/${f}`);
483
+ }
484
+ } else {
485
+ fs.mkdirSync(postMediaDir, { recursive: true });
486
+ try {
487
+ const buffer = await client.downloadMedia(message.media, {});
488
+ if (buffer) {
489
+ let ext = ".bin";
490
+ if (message.media instanceof Api.MessageMediaPhoto) {
491
+ ext = ".jpg";
492
+ } else if (message.media instanceof Api.MessageMediaDocument) {
493
+ const doc = message.media.document;
494
+ if (doc instanceof Api.Document) {
495
+ const mimeExt = doc.mimeType?.split("/")[1];
496
+ if (mimeExt) {
497
+ ext = "." + mimeExt.replace("jpeg", "jpg");
460
498
  }
461
- if (attr instanceof Api.DocumentAttributeFilename) {
462
- ext = path.extname(attr.fileName) || ext;
499
+ for (const attr of doc.attributes) {
500
+ if (attr instanceof Api.DocumentAttributeVideo) {
501
+ ext = ".mp4";
502
+ }
503
+ if (attr instanceof Api.DocumentAttributeFilename) {
504
+ ext = path.extname(attr.fileName) || ext;
505
+ }
463
506
  }
464
507
  }
465
508
  }
509
+ const mediaFileName = `media${ext}`;
510
+ const mediaPath = path.join(postMediaDir, mediaFileName);
511
+ fs.writeFileSync(mediaPath, buffer);
512
+ mediaFiles.push(`media/${paddedId}/${mediaFileName}`);
466
513
  }
467
- const mediaFileName = `media${ext}`;
468
- const mediaPath = path.join(postMediaDir, mediaFileName);
469
- fs.writeFileSync(mediaPath, buffer);
470
- mediaFiles.push(`media/${paddedId}/${mediaFileName}`);
514
+ } catch (e) {
515
+ console.error(`Error downloading media for message ${msgId}:`, e);
471
516
  }
472
- } catch (e) {
473
- console.error(`Error downloading media for message ${msgId}:`, e);
474
517
  }
475
518
  }
476
- const content = message.message || "";
519
+ const content = entitiesToMarkdown(message.message || "", message.entities);
477
520
  const link = channelMeta.username ? `https://t.me/${channelMeta.username}/${msgId}` : "";
478
521
  const post = {
479
522
  msgId,
@@ -607,9 +650,26 @@ async function processPost(post, options, exportDir) {
607
650
  const { title: originalTitle, body: originalBody } = extractTitleAndBody(cleanedContent);
608
651
  const date = post.date.toISOString().split("T")[0];
609
652
  const languages = [];
653
+ const translationExists = (lang) => {
654
+ const langDir = path2.join(outputDir, lang);
655
+ if (!fs2.existsSync(langDir)) return false;
656
+ try {
657
+ for (const slug of fs2.readdirSync(langDir)) {
658
+ const indexPath = path2.join(langDir, slug, "index.md");
659
+ if (fs2.existsSync(indexPath)) {
660
+ const content = fs2.readFileSync(indexPath, "utf-8");
661
+ if (content.includes(`original_link: "${post.link}"`)) {
662
+ return true;
663
+ }
664
+ }
665
+ }
666
+ } catch {
667
+ }
668
+ return false;
669
+ };
610
670
  if (translate && translate.targetLangs.length > 0) {
611
671
  const sourceLang = translate.sourceLang || "ru";
612
- if (translate.keepOriginal) {
672
+ if (translate.keepOriginal && !translationExists(sourceLang)) {
613
673
  languages.push({
614
674
  lang: sourceLang,
615
675
  title: originalTitle,
@@ -619,6 +679,10 @@ async function processPost(post, options, exportDir) {
619
679
  }
620
680
  for (const targetLang of translate.targetLangs) {
621
681
  if (targetLang === sourceLang) continue;
682
+ if (translationExists(targetLang)) {
683
+ onProgress?.(` ${targetLang}: already exists, skipping`);
684
+ continue;
685
+ }
622
686
  onProgress?.(` Translating to ${targetLang}...`);
623
687
  const translateOpts = {
624
688
  apiKey: translate.apiKey,
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "koztv-blog-tools",
3
- "version": "1.2.4",
3
+ "version": "1.2.6",
4
4
  "description": "Shared utilities for Telegram-based blog sites",
5
5
  "main": "dist/index.js",
6
6
  "module": "dist/index.mjs",