scriptremote

Overview

ScriptRemote is a MEAN-stack project to provide secure remote access to scripted applications from a browser UI. Basically it is a messaging service between scripts running on machines in protected networks and browsers. Using the service it is easy to create web UI's for applications for which developing full-custom web or mobile apps would require too much time or resources.

Features

• Configurable browser UI based on AngularJS templates
• Native script-based utilities (bash, python2.7) for script-to-server communication
• API for adding other script or client interfaces
• Hierarchical message organization
• Message attached files (text, png, svg)
• Message sharing with permissions
• Firewall friendly - uses only http/https and mail protocols
• End-to-end script-browser message encryption
• Notifications

Example

In this simulated IOT example, a script sends a message containing some temperatures obtained from sensors and the current state of a fan switch. It then waits for a reply message, which can leave the fan state unchanged or flip it. The message cycle repeats indefinitely.

The screenshot on the left shows one of the messages viewed under the default UI settings. On the right the settings are changed to use a customized template that better matches the message content.

Here is the bash script:

#!/bin/bash

# Load utilities SR_start, SR_set, SR_send
. ./srio.sh

function read_temp {
    # simulate temperature sensor
    echo $((50 + RANDOM % 50))
}

fan_state=off
function fan_control {
    # simulate fan control
    fan_state="$1"
}

SR_start ${SRUSER} ${SRTOKEN} 'MyProject' 'FanControlJob'

while true
do
    # read some temps
    sensor1=$(read_temp 'sensor1')
    sensor2=$(read_temp 'sensor2')
    sensor3=$(read_temp 'sensor3')

    # prepare temps to send
    SR_set 'Sensor1' "$sensor1" 'false'
    SR_set 'Sensor2' "$sensor2" 'false'
    SR_set 'Sensor3' "$sensor3" 'false'

    # prepare fan state to send, reply allowed
    SR_set 'Fan' "$fan_state" 'true'

    # send message, wait for reply
    SR_send 'CheckTemps' 

    if [ "$SR_status" == 'REPLY_TIMEOUT' ]
    then
        # timed out
        SR_clear
        continue
    elif [ "$SR_status" == 'OK' ]
    then
        # get fan reply
        SR_get 'Fan'
        if [ "$SR_output" != "$fan_state" ]
        then
            # change fan state
            fan_control "$SR_output"
        fi
    else
        # got an error
        echo "$SR_status"
        break
    fi
done

SR_end

These are the customized Angular templates:

The 'ng' attributes provide the links to the Angular framework. A few HTML5 and bootstrap library features are used for styling. The second template is for the reply part of the message. The reply button is handled by the template's parent element so it does not appear here.

<ul class="list-group">
    <li ng-repeat="item in content" class="list-group-item">
        Temperature at {{item.name}}: <meter min="0" max="120" low="0.01" high="90" optimum="78" value="{{item.value}}"></meter><span class="label label-default">{{item.value}}F</span>
    </li>
</ul>


<div class="row">
    <div class="col-sm-1">
        Fan state: 
    </div>
    <div class="col-sm-1">
        <input type="radio" ng-model="replyContent[0].value" value="off">Off
    </div>
    <div class="col-sm-1">
        <input type="radio" ng-model="replyContent[0].value" value="on">On
    </div>
</div>

Additional Documentation

• The Guide describes ScriptRemote features in more detail
• The API Reference defines the server data endpoints available to clients

Demo Server

If you would like to try a few tests with ScriptRemote before installing there is a demo server. To use it:

1. Register

2. Login and get API credentials by selecting Settings in the menu bar and then clicking Generate in the API Credentials section

3. Copy the API credentials to the test system

4. Download a utility script (bash in this example) to the test system:

   >$ wget https://scriptremote.com/dist/srio.sh

   The test system also needs to have curl installed.

5. Start with a simple test case:

   >$ cat > test.sh
   #!/bin/bash 
   . ./srio.sh
   SR_start ${SRUSER} ${SRTOKEN} 'myproject' 'myjob'
   SR_set 'myhello' 'Hello World, how are you?' 'false'
   SR_send 'mysend'
   SR_set 'myreply' 'I am ...' 'true'
   SR_send 'mysend'
   SR_get 'myreply'
   SR_end

6. Export the API credentials and run the test:

   >$ export SRUSER=<youruserid>
   >$ export SRTOKEN=<yourtoken>
   >$ bash test.sh

7. The messages should now be accessible under Projects in the menu bar. Submitting a modified reply to the second message should cause the script to display the new value. If the test system is behind a proxy server you may need to set the HTTPS_PROXY environment variable.

8. There are some limits per user on the amount of data residing on the demo server: 2 projects, up to 10 jobs per project, up to approximately 1MB of message data per job.

Installation

Installation methods vary depending on the platform chosen and level of availability needed for the service. Only simple installation options can be covered here.

A basic platform choice is between public and private servers.

Public server - Has a public IP and domain name, is probably hosted by a cloud provider, implements standard security protections. Browser access is similar to that for any public website. Script connections to the server will traverse firewalls and/or proxies on the client side, and possibly load balancers/reverse proxies on the server side. Those systems only need to allow script-to-server https requests because long-polling is used for server-to-script data transfer. Support for https/tls is necessary in order to secure browser and script connections.

Private server - Is located in a protected network, probably the same one containing the scripted application. Browser access is through a VPN or SSH tunnel client running on the remote device. Script connections might traverse internal firewalls. Use of TLS on the server may not be required if connections through the public net are otherwise protected and there is sufficient confidence in the security of the private network.

For a small site not needing high availability, the server can just be a machine, VPS, or container that can support a MEAN-stack environment and forward email. Another option is a platform-as-a-service vendor that supports MEAN-stack.

Installation on Docker

Installation on Digital Ocean Droplet

Installation on IBM Bluemix

Generic private server installation

Installation for development/contributing

Author

Jim Brandt

License

MIT