Return to Answer

Of course, things like this always happen during my holiday ... not that I take that many, but still ... Anyway, here's what I (read: my script) did: analyze almost 80k of posts containing links to i.imgur.com and try to upload these to the corporate repository.

Here are the results - note that the script doesn't know about the official site names like 'Arqade' and uses the API site parameter (gaming) instead. Some statistics:

• 114376 images have been uploaded
• 1672 images could not be uploaded because they were over 2 MiB
• 78 images could not be uploaded because of the MIME type (the uploader only supports PNG, JPG and GIF)
• Sometimes, images have already been removed, like http://i.imgur.com/uqJeW.png. The script already tries to fetch those from the Wayback Machine, but that doesn't always work because of some rate limiting (I guess? But even when I do analysis by hand, it isn't always working properly.). In this case, the image can be found: https://web.archive.org/web/20130607111838/http://i.imgur.com/uqJeW.png so I'll probably do another run for these 3507 images (which have "FatalException: Wayback Machine can't be reached.")
  • But even if the Wayback Machine saved a snapshot, that may not be reliable. This question refers to an image URL that does not seem to match the Wayback Machine copy, right?

The results have been grouped per site, and the last column in the "Success" table contains the IDs of the posts affected. Because there are so many different links (images are likely to be used in a single post), the mass replacement tool is not an option here. Perhaps Aaron Bertrand can help out like he did here; otherwise, we shouldn't bump too many posts only because they might be broken in the future.

Of course, things like this always happen during my holiday ... not that I take that many, but still ... Anyway, here's what I (read: my script) did: analyze almost 80k of posts containing links to i.imgur.com and almost 35k of posts containing links to imgur.com and try to upload these to the corporate repository.

Here are the results - note that the script doesn't know about the official site names like 'Arqade' and uses the API site parameter (gaming) instead. Some statistics:

• 155k images have been uploaded
• Some images (about 2k) could not be uploaded because they were over 2 MiB, or because of the MIME type (the uploader only supports PNG, JPG and GIF) and they have been archived to the Wayback Machine
• Sometimes, images have already been removed, like http://i.imgur.com/uqJeW.png. The script tries to fetch those from the Wayback Machine and upload them, but the Wayback Machine snapshot may not be reliable. This question refers to an image URL that does not seem to match the Wayback Machine copy, right?

The results have been grouped per site, and the last column contains the IDs of the posts which have that image. Because there are so many different links (images are likely to be used in a single post), the mass replacement tool is not an option here. Perhaps Aaron Bertrand can help out like he did here; otherwise, we shouldn't bump too many posts only because they might be broken in the future.

Of course, things like this always happen during my holiday ... not that I take that many, but still ... Anyway, here's what I (read: my script) did: analyze almost 80k of posts containing links to i.imgur.com and try to upload these to the corporate repository.  Here are the results. Some statistics:

• 114376 images have been uploaded
• 1672 images could not be uploaded because they were over 2 MiB
• 78 images could not be uploaded because of the MIME type (the uploader only supports PNG, JPG and GIF)
• Sometimes, images have already been removed, like http://i.imgur.com/uqJeW.png. The script already tries to fetch those from the Wayback Machine, but that doesn't always work because of some rate limiting (I guess? But even when I do analysis by hand, it isn't always working properly.). In this case, the image can be found: https://web.archive.org/web/20130607111838/http://i.imgur.com/uqJeW.png so I'll probably do another run for these 3507 images (which have "FatalException: Wayback Machine can't be reached.")
  • But even if the Wayback Machine saved a snapshot, that may not be reliable. This question refers to an image URL that does not seem to match the Wayback Machine copy, right?

The results have been grouped per site, and the last column in the "Success" table contains the IDs of the posts affected. Because there are so many different links (images are likely to be used in a single post), the mass replacement tool is not an option here. Perhaps Aaron Bertrand can help out like he did here; otherwise, we shouldn't bump too many posts only because they might be broken in the future.

Of course, things like this always happen during my holiday ... not that I take that many, but still ... Anyway, here's what I (read: my script) did: analyze almost 80k of posts containing links to i.imgur.com and try to upload these to the corporate repository. 

Here are the results - note that the script doesn't know about the official site names like 'Arqade' and uses the API site parameter (gaming) instead. Some statistics:

• 114376 images have been uploaded
• 1672 images could not be uploaded because they were over 2 MiB
• 78 images could not be uploaded because of the MIME type (the uploader only supports PNG, JPG and GIF)
• Sometimes, images have already been removed, like http://i.imgur.com/uqJeW.png. The script already tries to fetch those from the Wayback Machine, but that doesn't always work because of some rate limiting (I guess? But even when I do analysis by hand, it isn't always working properly.). In this case, the image can be found: https://web.archive.org/web/20130607111838/http://i.imgur.com/uqJeW.png so I'll probably do another run for these 3507 images (which have "FatalException: Wayback Machine can't be reached.")
  • But even if the Wayback Machine saved a snapshot, that may not be reliable. This question refers to an image URL that does not seem to match the Wayback Machine copy, right?

The results have been grouped per site, and the last column in the "Success" table contains the IDs of the posts affected. Because there are so many different links (images are likely to be used in a single post), the mass replacement tool is not an option here. Perhaps Aaron Bertrand can help out like he did here; otherwise, we shouldn't bump too many posts only because they might be broken in the future.

Of course, things like this always happen during my holiday ... not that I take that many, but still ... Anyway, here's what I (read: my script) did: analyze almost 80k of posts containing links to i.imgur.com and try to upload these to the corporate repository. Here are the results. Some statistics:

• 114376 images have been uploaded
• 1672 images could not be uploaded because they were over 2 MiB
• 78 images could not be uploaded because of the MIME type (the uploader only supports PNG, JPG and GIF)
• Sometimes, images have already been removed, like http://i.imgur.com/uqJeW.png. The script already tries to fetch those from the Wayback Machine, but that doesn't always work because of some rate limiting (I guess? But even when I do analysis by hand, it isn't always working properly.). In this case, the image can be found: https://web.archive.org/web/20130607111838/http://i.imgur.com/uqJeW.png so I'll probably do another run for these 3507 images (which have "FatalException: Wayback Machine can't be reached.")

The results have been grouped per site, and the last column in the "Success" table contains the IDs of the posts affected. Because there are so many different links (images are likely to be used in a single post), the mass replacement tool is not an option here. Perhaps Aaron Bertrand can help out like he did here; otherwise, we shouldn't bump too many posts only because they might be broken in the future.

Of course, things like this always happen during my holiday ... not that I take that many, but still ... Anyway, here's what I (read: my script) did: analyze almost 80k of posts containing links to i.imgur.com and try to upload these to the corporate repository. Here are the results. Some statistics:

• 114376 images have been uploaded
• 1672 images could not be uploaded because they were over 2 MiB
• 78 images could not be uploaded because of the MIME type (the uploader only supports PNG, JPG and GIF)
• Sometimes, images have already been removed, like http://i.imgur.com/uqJeW.png. The script already tries to fetch those from the Wayback Machine, but that doesn't always work because of some rate limiting (I guess? But even when I do analysis by hand, it isn't always working properly.). In this case, the image can be found: https://web.archive.org/web/20130607111838/http://i.imgur.com/uqJeW.png so I'll probably do another run for these 3507 images (which have "FatalException: Wayback Machine can't be reached.")
  • But even if the Wayback Machine saved a snapshot, that may not be reliable. This question refers to an image URL that does not seem to match the Wayback Machine copy, right?

The results have been grouped per site, and the last column in the "Success" table contains the IDs of the posts affected. Because there are so many different links (images are likely to be used in a single post), the mass replacement tool is not an option here. Perhaps Aaron Bertrand can help out like he did here; otherwise, we shouldn't bump too many posts only because they might be broken in the future.