a quick challenge to make a website that displays an image search using a colour selected by the user and a random noun and includes some surrounding features.
Installing a local version isn't easy as you have to configure a CSE from Google to run an image search and then also get an APIKEY to go with it.
When you have both create a file called .env in the root of the project and fill it with the following, replacing the stuff in []'s
GOOGLE_KEY=[your API key]
CX=[your CSE's id code]
after this, you should just be able to run rails s in the console and visit it on a local browser at localhost:3000
When you visit the site you get told what your noun is, you can then click a square to pick a colour.
The site then displays the top 5 results below the colours and gives you a new noun.
At this point, you can either search again or click an image to see the full-size version at its original address.
After conducting another search the history box appears below the results box, from here you can rerun searches using the buttons at the side
https://ancient-ocean-25028.herokuapp.com/
split the index view into partials.
manipulate the main view using javaScript and jquery instead of going through redirects to update. could keep the required data in javaScript objects.
check for the thumbnails from google API instead of getting fullsize images.
the single helper was not tested as it's a single line method that returns a random result.
the singleton pattern was used to store data for the current session only, as it was the simplest choice available for quickly storing and retrieving data. this has led to the side effect of sharing the searches across all users - as this isn't against the brief it was kept in.
you can click on any image to see the full version - because it was more simple to allow all than to process the first result separately.
not using dominant colour as a filter- Google search API does not allow the colour red, amongst others. instead, the colour is appended onto the search term.
heavier TDD - the code was mostly spiked because I'd not worked with rails for a while so was refamiliarizing myself with its workings. I was able to spike in a testable and lightweight fashion, but given a bit more familiarity with a domain, I'd prefer good strong TDD.
given the original timeframe, I think the choice to make a site without a database wasn't the worst idea, although if I'd known earlier about the extension I'd have gone with a database, to begin with.
given a more free-range on solutions instead of focussing on Ruby and metrics - I'd definitely go the down the javaScript route for data-storage, it's got the advantage of not clogging up the server with needless data to be cleaned up and makes for a much smoother single-page user experience with jquery.