[javascript] How to make a JSON call to a url?

I'm looking at the following API:

http://wiki.github.com/soundcloud/api/oembed-api

The example they give is

Call:

http://soundcloud.com/oembed?url=http%3A//soundcloud.com/forss/flickermood&format=json

Response:

{
"html":"<object height=\"81\" ... ",
"user":"Forss",
"permalink":"http:\/\/soundcloud.com\/forss\/flickermood",
"title":"Flickermood",
"type":"rich",
"provider_url":"http:\/\/soundcloud.com",
"description":"From the Soulhack album...",
"version":1.0,
"user_permalink_url":"http:\/\/soundcloud.com\/forss",
"height":81,
"provider_name":"Soundcloud",
"width":0
}

What do I have to do to get this JSON object from just a url?

You make a bog standard HTTP GET Request. You get a bog standard HTTP Response with an application/json content type and a JSON document as the body. You then parse this.

Since you have tagged this 'JavaScript' (I assume you mean "from a web page in a browser"), and I assume this is a third party service, you're stuck. You can't fetch data from remote URI in JavaScript unless explicit workarounds (such as JSONP) are put in place.

Oh wait, reading the documentation you linked to - JSONP is available, but you must say 'js' not 'json' and specify a callback: format=js&callback=foo

Then you can just define the callback function:

function foo(myData) { 
    // do stuff with myData
}

And then load the data:

var script = document.createElement('script');
script.type = 'text/javascript';
script.src = theUrlForTheApi;
document.body.appendChild(script);

In modern-day JS, you can get your JSON data by calling ES6's fetch() on your URL and then using ES7's async/await to "unpack" the Response object from the fetch to get the JSON data like so:

const getJSON = async url => {
  try {
    const response = await fetch(url);
    if(!response.ok) // check if response worked (no 404 errors etc...)
      throw new Error(response.statusText);

    const data = await response.json(); // get JSON from the response
    return data; // returns a promise, which resolves to this data value
  } catch(error) {
    return error;
  }
}

console.log("Fetching data...");
getJSON("https://soundcloud.com/oembed?url=http%3A//soundcloud.com/forss/flickermood&format=json").then(data => {
  console.log(data);
}).catch(error => {
  console.error(error);
});

The above method can be simplified down to a few lines if you ignore the exception/error handling (usually not recommended as this can lead to unwanted errors):

const getJSON = async url => {
  const response = await fetch(url);
  return response.json(); // get JSON from the response 
}

console.log("Fetching data...");
getJSON("https://soundcloud.com/oembed?url=http%3A//soundcloud.com/forss/flickermood&format=json")
  .then(data => console.log(data));

Because the URL isn't on the same domain as your website, you need to use JSONP.

For example: (In jQuery):

$.getJSON(
    'http://soundcloud.com/oembed?url=http%3A//soundcloud.com/forss/flickermood&format=js&callback=?', 
    function(data) { ... }
);

This works by creating a <script> tag like this one:

<script src="http://soundcloud.com/oembed?url=http%3A//soundcloud.com/forss/flickermood&format=js&callback=someFunction" type="text/javascript"></script>

Their server then emits Javascript that calls someFunction with the data to retrieve.
`someFunction is an internal callback generated by jQuery that then calls your callback.

DickFeynman's answer is a workable solution for any circumstance in which JQuery is not a good fit, or isn't otherwise necessary. As ComFreek notes, this requires setting the CORS headers on the server-side. If it's your service, and you have a handle on the bigger question of security, then that's entirely feasible.

Here's a listing of a Flask service, setting the CORS headers, grabbing data from a database, responding with JSON, and working happily with DickFeynman's approach on the client-side:

#!/usr/bin/env python 
from __future__ import unicode_literals
from flask      import Flask, Response, jsonify, redirect, request, url_for
from your_model import *
import os
try:
    import simplejson as json;
except ImportError:
    import json
try:
    from flask.ext.cors import *
except:
    from flask_cors import *

app = Flask(__name__)

@app.before_request
def before_request():
try:
    # Provided by an object in your_model
    app.session = SessionManager.connect()
except:
    print "Database connection failed."

@app.teardown_request
def shutdown_session(exception=None):
    app.session.close()

# A route with a CORS header, to enable your javascript client to access 
# JSON created from a database query.
@app.route('/whatever-data/', methods=['GET', 'OPTIONS'])
@cross_origin(headers=['Content-Type'])
def json_data():
    whatever_list = []
    results_json  = None
    try:
        # Use SQL Alchemy to select all Whatevers, WHERE size > 0.
        whatevers = app.session.query(Whatever).filter(Whatever.size > 0).all()
        if whatevers and len(whatevers) > 0:
            for whatever in whatevers:
                # Each whatever is able to return a serialized version of itself. 
                # Refer to your_model.
                whatever_list.append(whatever.serialize())
             # Convert a list to JSON. 
             results_json = json.dumps(whatever_list)
    except SQLAlchemyError as e:
        print 'Error {0}'.format(e)
        exit(0)

    if len(whatevers) < 1 or not results_json:
        exit(0)
    else:
        # Because we used json.dumps(), rather than jsonify(), 
        # we need to create a Flask Response object, here.
        return Response(response=str(results_json), mimetype='application/json')

if __name__ == '__main__':
    #@NOTE Not suitable for production. As configured, 
    #      your Flask service is in debug mode and publicly accessible.  
    app.run(debug=True, host='0.0.0.0', port=5001) # http://localhost:5001/

your_model contains the serialization method for your whatever, as well as the database connection manager (which could stand a little refactoring, but suffices to centralize the creation of database sessions, in bigger systems or Model/View/Control architectures). This happens to use postgreSQL, but could just as easily use any server side data store:

#!/usr/bin/env python 
# Filename: your_model.py
import time
import psycopg2
import psycopg2.pool
import psycopg2.extras
from   psycopg2.extensions        import adapt, register_adapter, AsIs
from   sqlalchemy                 import update
from   sqlalchemy.orm             import *
from   sqlalchemy.exc             import *
from   sqlalchemy.dialects        import postgresql
from   sqlalchemy                 import Table, Column, Integer, ForeignKey
from   sqlalchemy.ext.declarative import declarative_base

class SessionManager(object):
    @staticmethod
    def connect():
        engine = create_engine('postgresql://id:passwd@localhost/mydatabase', 
                                echo = True)
        Session = sessionmaker(bind = engine, 
                               autoflush = True, 
                               expire_on_commit = False, 
                               autocommit = False)
    session = Session()
    return session

  @staticmethod
  def declareBase():
      engine = create_engine('postgresql://id:passwd@localhost/mydatabase', echo=True)
      whatever_metadata = MetaData(engine, schema ='public')
      Base = declarative_base(metadata=whatever_metadata)
      return Base

Base = SessionManager.declareBase()

class Whatever(Base):
    """Create, supply information about, and manage the state of one or more whatever.
    """
    __tablename__         = 'whatever'
    id                    = Column(Integer, primary_key=True)
    whatever_digest       = Column(VARCHAR, unique=True)
    best_name             = Column(VARCHAR, nullable = True)
    whatever_timestamp    = Column(BigInteger, default = time.time())
    whatever_raw          = Column(Numeric(precision = 1000, scale = 0), default = 0.0)
    whatever_label        = Column(postgresql.VARCHAR, nullable = True)
    size                  = Column(BigInteger, default = 0)

    def __init__(self, 
                 whatever_digest = '', 
                 best_name = '', 
                 whatever_timestamp = 0, 
                 whatever_raw = 0, 
                 whatever_label = '', 
                 size = 0):
        self.whatever_digest         = whatever_digest
        self.best_name               = best_name
        self.whatever_timestamp      = whatever_timestamp
        self.whatever_raw            = whatever_raw
        self.whatever_label          = whatever_label

    # Serialize one way or another, just handle appropriately in the client.  
    def serialize(self):
        return {
            'best_name'     :self.best_name,
            'whatever_label':self.whatever_label,
            'size'          :self.size,
        }

In retrospect, I might have serialized the whatever objects as lists, rather than a Python dict, which might have simplified their processing in the Flask service, and I might have separated concerns better in the Flask implementation (The database call probably shouldn't be built-in the the route handler), but you can improve on this, once you have a working solution in your own development environment.

Also, I'm not suggesting people avoid JQuery. But, if JQuery's not in the picture, for one reason or another, this approach seems like a reasonable alternative.

It works, in any case.

Here's my implementation of DickFeynman's approach, in the the client:

<script type="text/javascript">
    var addr = "dev.yourserver.yourorg.tld"
    var port = "5001"

    function Get(whateverUrl){
        var Httpreq = new XMLHttpRequest(); // a new request
        Httpreq.open("GET",whateverUrl,false);
        Httpreq.send(null);
        return Httpreq.responseText;          
    }

    var whatever_list_obj = JSON.parse(Get("http://" + addr + ":" + port + "/whatever-data/"));
    whatever_qty = whatever_list_obj.length;
    for (var i = 0; i < whatever_qty; i++) {
        console.log(whatever_list_obj[i].best_name);
    }
</script>

I'm not going to list my console output, but I'm looking at a long list of whatever.best_name strings.

More to the point: The whatever_list_obj is available for use in my javascript namespace, for whatever I care to do with it, ...which might include generating graphics with D3.js, mapping with OpenLayers or CesiumJS, or calculating some intermediate values which have no particular need to live in my DOM.

A standard http GET request should do it. Then you can use JSON.parse() to make it into a json object.

function Get(yourUrl){
    var Httpreq = new XMLHttpRequest(); // a new request
    Httpreq.open("GET",yourUrl,false);
    Httpreq.send(null);
    return Httpreq.responseText;          
}

then

var json_obj = JSON.parse(Get(yourUrl));
console.log("this is the author name: "+json_obj.author_name);

that's basically it