mlds-couchdb-demo's Introduction

DeployR - R Analytics Integration

Same DeployR enabled R Script, diffrent client.

This example can be used in conjunction with the mlds-ui-demo to demonstrate how a different DeployR consumer/client can invoke the same R Script /sheri/geo/RUGclusters.R and do something different with the result. For this example, the geoJSON returned from calling /sheri/geo/RUGclusters.R is stored in CouchDB rather than building a UI as we did in mlds-ui-demo.

Prerequisites

Install the latest stable version of Node.js (version 0.10.x).
Install CouchDB

Quick start

Install Node.js
Install CouchDB
Create a new Database named feature from the CouchDB Admin Console
$ git clone https://github.com/mlds-deployr/mlds-couchdb-demo.git
$ cd mlds-couchdb-demo
$ npm install
$ node index.js This runs the demo.
Open the CouchDB Admin Console and view the newly inserted geoJSON returned from executing the R Script /sheri/geo/RUGclusters.R under the feature Database.

var deployr    = require('deployr').configure({ host: 'http://166.78.105.110:7400' }),
    stream     = require('stream')
    JSONStream = require('JSONStream'),
    got        = require('got'),
    nano       = require('nano')('http://localhost:5984'),
    feature    = nano.db.use('feature');

//
// Stream Transform duplex to download DeployR response artifact URLs in 
// geoJSON and store it in CouchDB
//
var geo = new stream.Transform({ objectMode: true });

geo._transform = function(chunk, encoding, done) {
    var data = chunk.toString();

    if (data) {
        got.get(data.replace(/\"/g, ''))
            .pipe(feature.insert(null, 'feature'));
    }

    done();
};

//
// Execute `RUGclusters.R` in DeployR and pipe geoJSON response into CouchDB
//
deployr.script('/sheri/geo/RUGclusters.R')
    .numeric('nclus', 5)
    .pipe(JSONStream.parse('deployr.response.execution.artifacts.*.url'))
    .pipe(JSONStream.stringify(false))
    .pipe(geo);

Recommend Projects