This is a fairly simple hack to record any imgur URLs discovered via the
random image facility that match a certain MD5 hash.
A known bug with this commit is that it may crash the bot the first time
the bot tries to add to the database, as the category is undefined. This
can be easily worked around by ensuring that the category is
prepopulated with at least one image before letting the bot add
automatically.
An issue should probably be opened to merge this code into something
reading hashes from a configuration file: it's pretty hardcoded at the
moment.
* There is probably a much nicer way of doing it (perhaps a better
hash)?
* Added a counter to the imgur random page counting the number of
[Facebook Monopoly Men](http://i.imgur.com/3Tt6N.gif).
* Added the md5sum of the current image to the getRandomImage api call.
* Future functionality might be storing every image md5 in an array and
counting duplicates, instead of the current check-against-known-dupes.
* Added config.json to the party
* Moved API key (client ID) to the configuration file
* Added image link to random imgur image page
* Minimised loading message
* Added fade to loading message
* Added count to title attribute of webpage
* Removed font-size increase
* Added placeholders
What imgur calls "title" is possibly better described as a caption. Most titles in imgur's gallery are full sentences, making the current behaviour of "xxx is yyy" fall flat, as the resulting string does not have satisfactory grammar.
This commit changes the functionality from "[foo is bar]" to "[foo][bar]".
Untested.
Specifying that images are non-animated each time isn't really required, and causes the string to be rather long. Stripping this implicit property makes it look a bit nicer.
