This is a repository for all work related to making scrolling, dynamically built strings work on an LED array. Then it's about making tweets display on that array, because the internet of things is really about hooking everything up to twitter.
The sparktweet directory is the code that sets up the function on a spark core. It exposes a REST call that can be passed a string and a couple other functions will display and scroll it on an LED array.
It does require the spark neopixel library. The file is buildString.ino and you are looking for line 972 that has the actual program and not just the character setup.
The Spark has a 63 length limit on the args being passed in, so to use the REST calls, you will need to do a series of calls. The calls need to begin with a 0 or 1 to indicate if it's an admin message or not. Of the admin calls, they will need to be either BEGIN to clear out the previous message or END to tell the Spark that you're done appending messages and it can display the whole string.
All non admin messages are simply appended together.
Keep in mind that, with the usage of two extra characters for the admin flag, you will need to split up each message to be 61 characters or less.
An example series of calls looks like the following:
curl https://api.spark.io/v1/devices/deviceID/buildString -d access_token=accessToken -d "args=1,BEGIN"
curl https://api.spark.io/v1/devices/deviceID/buildString -d access_token=accessToken -d "args=0,Message one "
curl https://api.spark.io/v1/devices/deviceID/buildString -d access_token=accessToken -d "args=0,Message two "
curl https://api.spark.io/v1/devices/deviceID/buildString -d access_token=accessToken -d "args=1,END"
Characters outside of the ASCII range will be converted to spaces. Ampersands are a special case because the Spark API doesn't like them - it looks for %26 and converts that into an ampersand. Your bot should convert ampersands into %26.
This directory contains the node program that will look to twitter for tweets and queue them to display, then display them off the queue.
Run npm install in the directory to install all dependencies. Rename sparkConfigTEMP.js and twitterConfigTEMP.js to not have TEMP in the name, and fill in the IDs, keys and tokens from twitter and spark.
You are interested in the bot.js file.
The program gets messages from two sources - mentions to @tweetSkirt and recent uses of #tweetSkirt. Tweets that have not been displayed before go in the queue, and then a seperate process pulls things from the queue. Currently, it looks for new tweets every 3 minutes, and will rotate the display every minute.
For mentions, it grabs the last 50, and finds any that are less than 24 hours old that have not been displayed before. For #tweetSkirt usage, it grabs the last 50 "recent" (recent is up to the twitter search API) and finds any that have not been displayed before.
Any tweets that have been displayed have their data stored in a small database. Tweets don't have their ID stored until after the display was successful.
Special handling of single and double quotes are handled here. Additionally, ampersands are converted to %26 to keep the Spark API happy.
Error handling is handled in this app - There is a variable called sparkErrorThreshhold that will record how many times it will handle a failure to send before giving up and moving it to the displayed tweets (with a message that it errored). If you find this is happening to one tweet consistantly, there is probably some special character I didn't think of that is escaping all the rest of the special character checks.
This directory contains both the API (server.js) and the site to interact with the API to move tweets from the incoming queue (where the bot puts them) to the display queue. Run bower install for the app and npm install for the API.
The configTEMP.js should be whereever the API is hosted (in case you end up breaking this up into two servers). Remove the TEMP on save.
The loginConfigTEMP.js is in two places and both files need different things. The loginConfig in the public directory just needs the salt. The loginConfig outside of it should have whatever your chosen password plus the salt ends up being after being md5 hashed. Remove the TEMP on save on both.
Be sure to host this site under SSL. I haven't gotten that setup yet but I will when I go to host this remotely.
The ASCII Array Building directory contains all the files for turning individual letters into strings of 1s and 0s to show if an LED should be lit or not, and also record the length of each character so we know how much 0 padding we need. Thanks to Scott Lembcke for his ideas and help with this great way of saving space! You shouldn't need to run this again, this is just for reference.