Algorithmia houses the largest marketplace of online algorithms. They make it easy for software developers to create algorithms and host them as microservices that anyone can call. This R package can be used by others to call algorithms hosted on the Algorithmia platform and interact with files through their data API. This package can also be used inside R algorithms. The only difference is how you create the client. More documention and information about algorithms on the Algorithmia platform can be found at http://algorithmia.com.
To do anything with the Algorithmia platform you first need to create a client.
client <- getAlgorithmiaClient("YOUR_ALGORITHMIA_API_KEY") # Not running in Algorithmia
If you are developing an Algorithma algorithm and want to call another algorithm in their platorm, you need to make this call without the API key argument passed in.
client <- getAlgorithmiaClient() # Do this when you are inside an Algorithmia Algorithm
Once you have a client, you can call algorithms in the Algorithmia marketplace. Algorithms are uniquely identified by a string of the form [ALGORITHM AUTHOR]/[ALGORITHM NAME]/[OPTIONAL VERSION]. An example of this would be the algorithm called “Hello” by the user “demo” - the information about this algorithm can be found here: https://algorithmia.com/algorithms/demo/Hello. The unique identifier for this algorithm is “demo/Hello”. By default, if you don't pass a specific version, the latest published version will be called. If you want to peg the version, use the full algorithm identifier “demo/Hello/0.1.1”. Once you know the algorithm you want to call, create the algorithm object and pass the data you want the algorithm to run on. The response is a list with a named member called result that is the output of the algorithm.
algo <- client$algo("demo/Hello") # Creating the algorithm object which can be used for multiple algorithm calls
result <- algo$pipe("sample input")$result # Calling the "demo/Hello" algorithm with the input of the algorithm being the string "sample input"
result == "Hello sample input" # evaluates to TRUE
Writing a file is as simple as creating a directory and putting data into a file. Here is an example that takes an R object, json encodes it and puts it in a file called “data.txt” in your directory called “Rfolder”.
dataDirectory <- client$dir("data://.my/Rfolder") # Get the Algorithma directory object
if (!dataDirectory$exists()) { # Check that the directory exists
dataDirectory$create() # If it doesn't, create it
}
f <- dataDirectory$file("data.txt") # Get the Algorithmia file object
f$putJson(myObject) # Encodes the object as json and writes that to the file, overwriting it if it exists and creating it if it doesn't
To learn more about the Algorithmia platform and all the latest developer news, visit https://algorithmia.com/developers.