this repo has no description
feat: add URL normalization to bsky.app format in Bluesky scraper
+22
-4
+22
-4
scrapers/blueskyScraper.js
+22
-4
scrapers/blueskyScraper.js
···
110
110
}
111
111
112
112
/**
113
+
* Normalize any ATProto frontend URL to bsky.app format
114
+
* @param {string} url - ATProto URL from any frontend (bsky.app, deer.social, witchsky.app, etc.)
115
+
* @returns {string} - Normalized bsky.app URL
116
+
*/
117
+
normalizeToBskyApp(url) {
118
+
try {
119
+
const { repo, rkey } = this.parseBlueskyUrl(url);
120
+
return `https://bsky.app/profile/${repo}/post/${rkey}`;
121
+
} catch (error) {
122
+
console.error(`Failed to normalize URL to bsky.app: ${error.message}`);
123
+
return url; // Return original URL if normalization fails
124
+
}
125
+
}
126
+
127
+
/**
113
128
* Get user's display name from their handle or DID
114
129
* @param {string} identifier - The user's Bluesky handle or DID
115
130
* @returns {Promise<string>} - User's display name or original identifier if not found
···
431
446
*/
432
447
async extract(url) {
433
448
try {
449
+
// Normalize URL to bsky.app format for consistent output
450
+
const normalizedUrl = this.normalizeToBskyApp(url);
451
+
434
452
// 1. Parse the URL to extract repo (handle) and rkey (post ID)
435
453
const { repo, rkey } = this.parseBlueskyUrl(url);
436
454
console.log(`Fetching Bluesky post: ${repo}/${rkey}`);
···
534
552
return {
535
553
imageUrl: imagePaths[0], // First image as primary
536
554
imageUrls: imagePaths, // All images
537
-
sourceUrl: url,
555
+
sourceUrl: normalizedUrl,
538
556
title: `Bluesky Image${imagePaths.length > 1 ? 's' : ''} by ${displayName}`,
539
557
siteName: 'Bluesky',
540
558
isVideo: false,
···
567
585
return {
568
586
imageUrl: imagePaths[0], // First image as primary
569
587
imageUrls: imagePaths, // All images
570
-
sourceUrl: url,
588
+
sourceUrl: normalizedUrl,
571
589
title: `Bluesky Image${imagePaths.length > 1 ? 's' : ''} by ${displayName}`,
572
590
siteName: 'Bluesky',
573
591
isVideo: false,
···
620
638
imageUrls: [thumbnailProcessed.localPath], // Single thumbnail
621
639
videoUrl: videoProcessed.localPath, // Cached and transcoded video
622
640
isVideo: true,
623
-
sourceUrl: url,
641
+
sourceUrl: normalizedUrl,
624
642
title: `Bluesky Video by ${displayName}`,
625
643
siteName: 'Bluesky',
626
644
originalImageUrl: thumbUrl,
···
665
683
imageUrls: [thumbnailProcessed.localPath], // Single thumbnail
666
684
videoUrl: videoProcessed.localPath,
667
685
isVideo: true,
668
-
sourceUrl: url,
686
+
sourceUrl: normalizedUrl,
669
687
title: `Bluesky Video by ${displayName}`,
670
688
siteName: 'Bluesky',
671
689
originalImageUrl: thumbUrl,