Usually, I expect one e621 post ID to match with one MD5. This appears to not be the case all the time?

There aren't many I have found, here's one example (I will try to find more and edit this post):

Post: https://e621.net/posts/1964489
Previous MD5: 30d7682261fa1df488bb836250adbe6f (https://e621.net/posts?tags=md5%3A30d7682261fa1df488bb836250adbe6f)
Current MD5: aacde09d4889ae1725306be7281efc8a (https://e621.net/posts?tags=md5%3Aaacde09d4889ae1725306be7281efc8a)

I know for certain that e621 had the Previous MD5 assigned to that post ID in the past, I have a copy of the old API JSON data from back when it was the previous MD5. The current API JSON data reports the current MD5.

There appears to be 0 references to the old MD5 left on e621 anymore. I thought what usually happens is the post is flagged for deletion as inferior/duplicate and a new post is created for the new MD5? If this is not the case, what is the actual practice e621 follows, or if it is the case, why might this have happened?

The MD5 changed because the file was replaced.

gattonero2001 said:
The MD5 changed because the file was replaced.

Right, that makes a bit more sense. I see there is a searchable API for that too, which is useful. Is it likely this might become a DB Export in the future?

pokis said:
Right, that makes a bit more sense. I see there is a searchable API for that too, which is useful. Is it likely this might become a DB Export in the future?

For a specific post, you can just add .json to the URL to get the metadata including tags, size, hash, date, etc.

alphamule said:
For a specific post, you can just add .json to the URL to get the metadata including tags, size, hash, date, etc.

I know, but this does not allow you to take a list of updates and run through them all to apply them to your own dataset. You would have to download the entire item list and go through every single item.

https://e621.net/post_replacements.json?search[status]=approved might help you out.

earlopain said:
https://e621.net/post_replacements.json?search[status]=approved might help you out.

I am also aware this API exists, but I also can't use it to retrieve all approved replacements - That'll end up being hundreds of pages. That's why a DB Export for it would be good, then I don't have to fetch page after page and also slam your system with requests (1 per second sure, but over hours...)