Drawr
Drawr | |
![]() Drawr is one of the biggest online drawing platform (similar to Pixiv Sketch). | |
URL | http://drawr.net/ |
Status | Offline |
Archiving status | Partially saved (nearly complete) |
Archiving type | Unknown |
Project source | drawr-grab |
Project tracker | drawr |
IRC channel | #archiveteam-bs (on hackint) (formerly #drawrnomore (on EFnet)) |
Data[how to use] | drawr.net_201911_part0 drawr.net_201911_part1 drawr.net_201911_part2 drawr.net_201911_part3 drawr.net_201911_part4 drawr.net_201911_part5 drawr.net_failed_201911 job:cvyzq |
Drawr was one of the biggest online drawing platform (similar to Pixiv Sketch). The website shut down on December 2, 2019, 13:00 JST (04:00 UTC)[1].
Site structure
Browsing pages:
Drawings are viewed under http://drawr.net/show.php?id=$ID
with incremental IDs going to about 7.17M as of the shutdown. User profiles are under /user.php?id=$ID
, with the maximum ID being 359451 as of 2019-11-14.
Each drawing may be accompanied by three files, hosted on a img##.drawr.net
domain (##
going from 01 to 12): a .png, a .xml, and/or a .gz, but for most drawings, not all files exist. The .png normally exists and contains a render of the finished drawing, as does the (at least one) .gz file, which contains the actual drawing data in zlib-compressed AMF3 format. The .xml only exists if the drawing was made in multiple sessions, and if so, it contains references to two or more .gz files. Drawings can also be based on a previous drawing, and in this case, the .gz file contains an obfuscated URL to the previous drawing's .png; this can happen for multiple layers. (In a few rare cases for very old drawings from the initial days of the website, the .gz file instead contains some URL that needs to be passed into an API to get that obfuscated URL, but the API is broken as of Nov 2019.)
There is an API referenced under http://drawr.net/api but this no longer worked as of November 2019.
Examples
- Private drawing: http://drawr.net/show.php?id=1806004
- Single .gz drawing on blank canvas: http://drawr.net/show.php?id=6854546
- Single .gz drawing on prior drawing: http://drawr.net/show.php?id=7161353
- Uses .xml: http://drawr.net/show.php?id=7166017 | http://drawr.net/show.php?id=1425823
- Uses chain of .gz: http://drawr.net/show.php?id=7166783
- Thread: http://drawr.net/show.php?id=1386396
- Needs the repost API: http://drawr.net/show.php?id=93
Problems
- Corrupt .gz files; there are numerous of these being corrupted in a variety of ways; examples:
- http://drawr.net/show.php?id=799104
- http://img03.drawr.net/draw/img/63530/4b6f3b78fYDu5Arg.gz from drawing 2474692 (or a previous drawing)
- http://img03.drawr.net/draw/img/73360/4a6dbd90DPJS98qL.gz from 728925 simply cuts off in the middle of the AMF data.
- Empty .gz files, e.g. http://img01.drawr.net/draw/img/16224/4a11732fVv2EDaJ3.gz from drawing 784685. (Normally, even for empty drawings, there is a "drawing started"-like entry.)
- .gz file with an empty '1' value (the one encoding a repost if necessary): http://img01.drawr.net/draw/img/4562/49918048ztxTGgM8.gz from http://drawr.net/show.php?id=337092
- .gz file referencing a previous drawing on the wrong host: http://img04.drawr.net/draw/img/107411/4ba0f2e4nwsg2uSq.gz from http://drawr.net/show.php?id=1277801 references http://cds04.drawr.net/draw/img/100522/4b5ff92dBr3nxda9_tpl.png which moved to cdn04.drawr.net.
- .gz file referencing a .gz instead of (as normal) a .png: http://img07.drawr.net/draw/img/193204/55aaa136XUhwTME5.gz discovered from from http://drawr.net/show.php?id=6289303
- .xml using wrong host: http://img02.drawr.net/draw/img/55226/4cbb1308AZarqsL1.xml from http://drawr.net/show.php?id=1806001 points to img03.drawr.net, but the files are on img02.
- Broken embeds, e.g. on 1277801
- Unknown: http://drawr.net/show.php?id=5218013
- ... and many more...
All of these (except the embed one) cause issues in the playback on the website player, but some can be worked around.
Archival
Nearly all of the drawings were archived by User:JustAnotherArchivist between 2019-11-26 and 2019-12-02 with qwarc. About 7.04 million drawing IDs were covered, and 160k were left to be scanned when the website was shut down.
An ArchiveBot job for the site was running as well (job:cvyzqtvmq3dhxbsn4sux95ju2), but this did not cover the drawing data and many other things.
The warrior project got banned after a short time and was never resumed after that.