···

       
408
       
408
       
 
       
  "links": {

     

       
409
       
409
       
 
       
    "https://example.com/post/123": {

     

       
410
       
410
       
 
       
      "referencing_entries": ["https://blog.user.com/entry/456"],

     

       
411
       
411
       
-
       
      "is_tracked_post": true,

     

       
412
       
411
       
 
       
      "target_username": "user2"

     

       
413
       
412
       
 
       
    },

     

       
414
       
413
       
 
       
    "https://external-site.com/article": {

     

       
415
       
415
       
-
       
      "referencing_entries": ["https://blog.user.com/entry/789"],

     

       
416
       
416
       
-
       
      "is_tracked_post": false

     

       
414
       
414
       
+
       
      "referencing_entries": ["https://blog.user.com/entry/789"]

     

       
417
       
415
       
 
       
    }

     

       
418
       
416
       
 
       
  },

     

       
419
       
417
       
 
       
  "reverse_mapping": {

     
···

       
438
       
436
       
 
       
```

     

       
439
       
437
       
 
       


     

       
440
       
438
       
 
       
This unified structure eliminates duplication by:

     

       
441
       
441
       
-
       
- Storing each URL only once with metadata flags

     

       
439
       
439
       
+
       
- Storing each URL only once with minimal metadata

     

       
442
       
440
       
 
       
- Including all link data, reference data, and mappings in one file

     

       
443
       
443
       
-
       
- Using `is_tracked_post` to identify internal vs external links

     

       
441
       
441
       
+
       
- Using presence of `target_username` to identify tracked vs external links

     

       
444
       
442
       
 
       
- Providing bidirectional mappings for efficient queries

     

       
445
       
443
       
 
       


     

       
446
       
444
       
 
       
### Unified Structure Benefits

     

···

       
320
       
320
       
 
       
        reverse_mapping = {}

     

       
321
       
321
       
 
       
        

     

       
322
       
322
       
 
       
        for url, entry_ids in link_dict.items():

     

       
323
       
323
       
-
       
            is_tracked = url in registered_urls

     

       
324
       
324
       
-
       
            target_username = None

     

       
323
       
323
       
+
       
            unified_links[url] = {

     

       
324
       
324
       
+
       
                "referencing_entries": entry_ids

     

       
325
       
325
       
+
       
            }

     

       
325
       
326
       
 
       
            

     

       
326
       
327
       
 
       
            # Find target username if this is a tracked post

     

       
327
       
327
       
-
       
            if is_tracked:

     

       
328
       
328
       
+
       
            if url in registered_urls:

     

       
328
       
329
       
 
       
                for username in users:

     

       
329
       
330
       
 
       
                    user_domains_set = {domain for domain in user_domains.get(username, [])}

     

       
330
       
331
       
 
       
                    if any(domain in url for domain in user_domains_set):

     

       
331
       
331
       
-
       
                        target_username = username

     

       
332
       
332
       
+
       
                        unified_links[url]["target_username"] = username

     

       
332
       
333
       
 
       
                        break

     

       
333
       
334
       
 
       
            

     

       
334
       
334
       
-
       
            unified_links[url] = {

     

       
335
       
335
       
-
       
                "referencing_entries": entry_ids,

     

       
336
       
336
       
-
       
                "is_tracked_post": is_tracked

     

       
337
       
337
       
-
       
            }

     

       
338
       
338
       
-
       
            

     

       
339
       
339
       
-
       
            if target_username:

     

       
340
       
340
       
-
       
                unified_links[url]["target_username"] = target_username

     

       
341
       
341
       
-
       
            

     

       
342
       
335
       
 
       
            # Build reverse mapping

     

       
343
       
336
       
 
       
            for entry_id in entry_ids:

     

       
344
       
337
       
 
       
                if entry_id not in reverse_mapping:

     
···

       
355
       
348
       
 
       
        

     

       
356
       
349
       
 
       
        if verbose:

     

       
357
       
350
       
 
       
            console.print(f"Found {len(registered_urls)} registered post URLs")

     

       
358
       
358
       
-
       
            console.print(f"Found {len(link_dict)} total links, {sum(1 for link in unified_links.values() if link['is_tracked_post'])} tracked posts")

     

       
351
       
351
       
+
       
            console.print(f"Found {len(link_dict)} total links, {sum(1 for link in unified_links.values() if 'target_username' in link)} tracked posts")

     

       
359
       
352
       
 
       
        

     

       
360
       
353
       
 
       
        # Save unified data

     

       
361
       
354
       
 
       
        with open(output_path, "w") as f:

     
···

       
373
       
366
       
 
       
            print(f"Unknown\t{len(link_categories['unknown'])}\tLinks to external sites")

     

       
374
       
367
       
 
       
            print(f"Total Extracted\t{len(all_links)}\tAll extracted links")

     

       
375
       
368
       
 
       
            print(f"Saved to Output\t{len(output_data['links'])}\tLinks saved to output file")

     

       
376
       
376
       
-
       
            print(f"Cross-references\t{sum(1 for link in unified_links.values() if link['is_tracked_post'])}\tLinks to registered posts only")

     

       
369
       
369
       
+
       
            print(f"Cross-references\t{sum(1 for link in unified_links.values() if 'target_username' in link)}\tLinks to registered posts only")

     

       
377
       
370
       
 
       
        else:

     

       
378
       
371
       
 
       
            table = Table(title="Links Summary")

     

       
379
       
372
       
 
       
            table.add_column("Category", style="cyan")

     
···

       
385
       
378
       
 
       
            table.add_row("Unknown", str(len(link_categories["unknown"])), "Links to external sites")

     

       
386
       
379
       
 
       
            table.add_row("Total Extracted", str(len(all_links)), "All extracted links")

     

       
387
       
380
       
 
       
            table.add_row("Saved to Output", str(len(output_data['links'])), "Links saved to output file")

     

       
388
       
388
       
-
       
            table.add_row("Cross-references", str(sum(1 for link in unified_links.values() if link['is_tracked_post'])), "Links to registered posts only")

     

       
381
       
381
       
+
       
            table.add_row("Cross-references", str(sum(1 for link in unified_links.values() if 'target_username' in link)), "Links to registered posts only")

     

       
389
       
382
       
 
       
            

     

       
390
       
383
       
 
       
            console.print(table)

     

       
391
       
384