Neo4J Rest Javascript wrapper

As a toy project, I started evaluating the Neo4J graph database and its Rest interface. I'm trying to write a simple graph visualization in JavaScript. In daily business I'm a Java developer and maybe that's the reason why I'm not that happy with my current approach. In Java I would have written a Neo4J class, with inner (data) classes for Node and Relation as their id is tightly couple to the server (http://localhost:7474/db/data/node/0). Maybe they even would get a common parent class.

In JavaScript I don't get away from my object oriented mind. I need some ideas in relation to JavaScript best practice.

```
function Neo4J(server) {
this.server = server;
}

function Node(connection, response) {
this.id = connection.sanitizeId(response['self']);
this.data = response['data'];
}

function Relation(connection, response) {
this.id = connection.sanitizeId(response['self']);
this.start = connection.sanitizeId(response['start']);
this.end = connection.sanitizeId(response['end']);
this.type = response['type'];
}

Neo4J.prototype = {
sanitizeId: function (id) {
return id.replace(this.server, "");
},

query: function (query, success) {
return post(this.server + "/cypher", {"query": query, "params": {} }, success);
},

getAllNodes: function (callback) {
var connection = this;
this.query("MATCH n RETURN n", function (nodes) {
$.each(nodes['data'], function (i, val) {
callback(new Node(connection, val[0]));
});
});
},

getAllRelations: function (callback) {
var connection = this;
this.query("START r=rel(*) RETURN r", function (nodes) {
$.each(nodes['data'], function (i, val) {
callback(new Relation(connection, val[0]));
});
});
},

getNode: function (id, callback) {
var connection = this;
get(this.server + id + "/properties", function (data) {

Interesting question,

The most striking thing is your repeated use of this pattern:

connection.sanitizeId(response['xxx']);

I dont believe sanitizeId should be a function of connection. Also, response is not a good name because it often is part of a response from the query and not the actual response. And, it is suggested that response.xxx makes more sense than response['xxx']. To that end, I feel that sanitizeId should be part of the response or data. Also, I would be tempted to call it simply sanitize.

function addSanitizer( data , server){
  data.sanitize = function( propertyName ){
    return this[propertyName].replace( server , "" );
  }
  return data;
}

Then you could do

function Node(response) {
    this.id   = response.sanitize( 'self' );
    this.data = response.data;
}

Or, if you agree that response might be misleading

function Node(node) {
    this.id   = node.sanitize( 'self' );
    this.data = node.data;  
}

Node with be called like this:

getNode: function (id, callback) {
    var connection = this;
    get(this.server + id + "/properties", function (data) {
        if (data == null) data = {data: {}};
        else data = {data: data};
        data['self'] = id;
        callback(new Node(this.addSanitizer( data, this.server ), data));
    });
},

From a style point:

• You are skipping new lines on your if statements: if (data == null) data = {data: {}};, dont do that

-
Your code stretches quite a bit horizontally, but not too bad, feel free to to split up some of your longer lines into something more readable, this :

createRelation: function (start, target, type, callback) {
    var connection = this;
    post(this.server + start + '/relationships', {to: this.server + target, type: type}, function (data) {
        callback(new Relation(connection, data));
    });
}

could be

createRelation: function (start, target, type, callback) {
    var connection = this,
        data = {to: this.server + target, type: type};
    post(this.server + start + '/relationships', data , function (data){
        callback(new Relation(connection, data));
    });
}

The code looks all in all fine to me. For a Java developer you did good :P

StackExchange Code Review Q#51202, answer score: 3