Kephost.com, a Hungarian image sharing site
|Archiving status||Partially saved continuously|
- Not to be confused with Kephost.hu.
Kephost.com is a Hungarian one-click image sharing service, one of the oldest and most popular ones. People can easily upload files from their computers with a few clicks, and then paste the URLs of the images (or their pages) wherever they want. Here users can
also choose a category to upload to,not any more and even create an account for themselves. Pictures can be commented on through the embedded Facebook comment module.
The service started in 2008. Since then, they have changed design in almost every year, and also have had several different site structures (including accessing images) – however, they seem to have not deleted older pictures since the first, 2010 redesign.
According to the numbers, the popularity of the site is steadily increasing, and as of 2017 – after the disappearance of Kepfeltoltes.hu – it seems to be the most popular Hungarian image hosting site, with nearly 2,000 images uploaded daily.
No information available.
No information available. Entity behind the site is unknown.
As already mentioned, they have had several different site structures. The following applies to their last, which has been in use apparently since 2014.
The size seems to be utilizing v3 of the Chevereto image hosting engine.
Uploaded images can be public, and also probably private (this function needs registration).
Only public pictures appear in the image browser, private ones cannot be discovered. Also, recently some older pictures don't appear in the end of the image browser. Update: as of July 2017, uploaded images can't be browsed any more.
Images are given a page whose URL is http://www.kephost.com/image/<ID> where <ID> seems to be an
incrementing base-62 number (alphabet A..Z, 0..9, a..z (in this order (?)), starting with one character). Letter 'A' acts as a leading zero when being the first character, that is, Aw = AAw = AAAw etc. The location of the actual image is http://kephost.com/images/<YYYY>/<MM>/<DD>/<FILENAME>.<EXT> where YYYY-MM-DD is the date the file was uploaded, FILENAME is the same as the user's file had on their computer, <EXT> is its extension. This original version is always linked in with an
<a> tag, as the download link. A thumbnail version is also generated for every picture, although not shown (just provided among the links); its filename is postfixed with .th, like <FILENAME>.th.<EXT>. Finally, if the image is large enough, an .md (medium) version is also generated (filename extended with .md), and in this case this version, otherwise the original one is shown on the page (within an
- Partly incrementing, partly indefinite. No general rule of handing out tokens has been found yet.
Archiving older images would need web scrape. Some of the older pattern prefixes:
kephost.com/view3.php?filename=– followed by a random alphanumeric/underscore token plus extension
kephost.com/images/– followed by a random alphanumeric/underscore token plus extension
kephost.com/images4/– followed by YYYY/M/D (one or two digit month and day!) then a random alphanumeric/underscore token plus extension, the token often also contains the date in YYYY_M_D format
– used until 2010, images seem to have been deleted
kephost.com/view-NUM_TOKEN.EXT– NUM might be a day counter, TOKEN might be the original filename
|Upload date||Status||Note||Images per day|
|2008 – 2010||Lost||First structure|
|2010 – 2014-05-16||Not saved yet||Second structure|
|2014-05-16 – 2016-03-21||Saved (archive)||291,039 images||431|
|2016-03-21 – 2016-09-04||Saved (archive)||135,604 images||812|
|2016-09-04 – 2017-01-23||Saved (archive)||194,603 images||1380|
|2017-01-23 – 2017-03-31||Saved (archive)||131,347 images||1960|
- Numbers are based on discovery results. Actual number of saved images may be a bit less.
As image browsing has been disabled, no very efficient way has been left to discover images:
- IDs are not incrementing in a regular way
- web scrape finds just a fraction of uploaded images
- there is a "random image" feature on the site, but at the current scale of the website, it would need millions of requests to find most of the new images.
Still, the most suitable way seems to be (3), but it would worth the effort only when there are enough new images (say, when image number doubles, so there are as many new images as many old ones), or, of course, when the site advances into the "Endangered" or "Closing" phase.
Based on this, estimated date of next archiving run: December 2017.