Basic usage

1. Start server

   bash SERVE.sh [language] (--silent)

Optionally add a language: spanish, english, or chinese are supported.

Optionally add --silent to the end to suppress printed output (which makes everything run faster).

2. Send requests

   • using wget

     wget --post-data 'I like her because she is nice' 'http://localhost:12345/?properties={"annotators":"tokenize","outputFormat":"json"}' -O -

   • using curl

     curl --data 'It did not however , cover any sort of local taxes or similar measures' 'http://localhost:12345/?properties={%22annotators%22%3A%22tokenize%2Cssplit%2Cdepparse%2Cpos%22%2C%22outputFormat%22%3A%22json%22}' -o -

   • using python

     import requests data = "The quick brown fox jumped over the lazy dog." url = "http://localhost:12345?properties={annotators:'tokenize,ssplit,pos,depparse'}" parse_string = requests.post(url, data=data).text return json.loads(parse_string)["sentences"][0]

   • using javascript

     var properties = {annotators: "tokenize"}; var property_string = JSON.stringify(properties); var properties_for_url = encodeURIComponent(property_string); $.ajax({ type: "POST", url: 'http://' + 'whatever-the-public-DNS-is/' + '?properties=' + properties_for_url, data: 'The quick brown fox jumped over the lazy dog.', success: function (data){ parse = data; console.log(parse); }, error: function (responseData, textStatus, errorThrown) { alert('POST failed.'); } });

Installation

On unix:

bash INSTALL.sh

On windows, just look at the INSTALL.sh file and take the appropriate lines from there...

Background for setting this up on a remote server

On Digital Ocean

On Robert's server, when everything is already installed, open screeni screen or screen -r and run:

bash SERVE.sh

and in another screen (control+a c)

python server.py

Then detach with control+a a control+a d. Use control+a n to switch between screens. Type exit when done.

---

This is for when I want to spin up an NLP server on EC2 for access from a web browser. I need to install CoreNLP and listen for POST requests.

Setting up the EC2 instance

Prerequisites

• set up an AWS account

Set up a new instance

1. Go online to the EC2 Dashboard in the AWS console.

   • Warning: Pay attention to your Region, in the upper right corner of this page. I once had instances running for a couple months on a different region, so I didn't see that they were running and I didn't know how to turn them off.

2. Go to EC2 Instances (in the left panel)

3. Launch Instance

   • Step 1: Choose an Amazon Machine Image (AMI)

     Ubuntu Server 16.04 LTS (HVM), SSD Volume Type - ami-16efb076

   • Step 2: Choose an Instance Type

     t2.small ~ this has 2g memory, which we need to run the server (actually, more sophisticated NLP would require more memory, but this is enough for dependency parsing)

   • skip through the next steps (keep the defaults):

     • Step 3: Configure Instance Details
     • Step 4: Add Storage
     • Step 5: Add Tags

   • Step 6: Configure Security Group

     Add rules for HTTP and HTTPS, accept the default settings. It will warn you about security. I'm ignoring this.

   • Step 7: Review Instance Launch

     Click launch and make a new key pair if you need to. This will be a file that ends with .pem, e.g. key.pem. I put this file in my home .ssh directory ~/.ssh/key.pem.

4. View Instances and get Public DNS

You should see one instance that's being set up. Select that instance and you'll see the properties of that instance on the bottom of the page. Copy the Public DNS (IPv4), which will look something like ec2-54-219-187-44.us-west-1.compute.amazonaws.com . In the rest of this README, I will call this address whatever-the-public-DNS-is.

4. SSH to the instance

   ssh ubuntu@whatever-the-public-DNS-is -i path/to/key.pem

Once you're SSHed into an Ubuntu instance

Clone this repo into your home directory. Then run INSTALL.sh on a new instance. That will set everything up.

When the installation is done, you can start running the server by running SERVE.sh.

Here are some example requests:

wget --post-data 'I like her because she is nice' 'http://localhost:12345/?properties={"annotators":"tokenize","outputFormat":"json"}' -O -

curl --data 'It did not however , cover any sort of local taxes or similar measures' 'http://localhost:12345/?properties={%22annotators%22%3A%22tokenize%2Cssplit%2Cdepparse%2Cpos%22%2C%22outputFormat%22%3A%22json%22}' -o -

And here's an example request from a browser:

var properties = {annotators: "tokenize"};
var property_string = JSON.stringify(properties);
var properties_for_url = encodeURIComponent(property_string);
$.ajax({
  type: "POST",
  url: 'http://' +
    'whatever-the-public-DNS-is/' +
    '?properties=' +
    properties_for_url,
  data: 'The quick brown fox jumped over the lazy dog.',
  success: function (data){
    parse = data;
    console.log(parse);
  },
  error: function (responseData, textStatus, errorThrown) {
    alert('POST failed.');
  }
});

Note that with only 2g of memory, we won't be able to run all of the annotators. But "tokenize,ssplit,pos,depparse" works fine, and that's all I need for many uses.

Multilingual

HTTPS wrapper

Test multithreading

curl --data 'It did not however , cover any sort of local taxes or similar measures' 'https://rxdhawkins.me:8882/?properties={%22annotators%22%3A%22tokenize%2Cssplit%2Cdepparse%2Cpos%22%2C%22outputFormat%22%3A%22json%22}' -o -