Mark Needham

Thoughts on Software Development

neo4j Spatial: Indexing football stadiums using the REST API

with one comment

Late last week my colleague Peter wrote up some documentation about creating spatial indexes in neo4j via HTTP, something I hadn’t realised was possible until then.

I previously wrote about indexing football stadiums using neo4j spatial but the annoying thing about the approach I described was that I was using neo4j in embedded mode which restricts you to using a JVM language.

The rest of my code is in Ruby so I thought I’d translate that code.

To recap, I’m parsing a CSV file of football stadiums that I downloaded from Chris Bell’s blog which looks like this:

Name,Team,Capacity,Latitude,Longitude
"Adams Park","Wycombe Wanderers",10284,51.6306,-0.800299
"Almondvale Stadium","Livingston",10122,55.8864,-3.52207
"Amex Stadium","Brighton and Hove Albion",22374,50.8609,-0.08014

The code to process the file and index the stadiums in neo4j is as follows (and is essentially a translation of the find_geometries_within_distance_using_cypher test):

require 'csv'
require 'httparty'
require 'json'
 
HTTParty.post("http://localhost:7474/db/data/ext/SpatialPlugin/graphdb/addSimplePointLayer", 
  :body => { :layer => 'geom', :lat => 'lat', :lon => 'lon' }.to_json,
  :headers => { 'Content-Type' => 'application/json' } )
 
HTTParty.post("http://localhost:7474/db/data/index/node", 		
  :body => { :name => 'geom', :config => { :provider => 'spatial', :geometry_type => 'point', :lat => 'lat', :lon => 'lon'  } }.to_json,
  :headers => { 'Content-Type' => 'application/json' } )
 
contents = CSV.read(File.join(File.dirname(__FILE__), 'data', 'stadiums.csv'))
contents.shift
contents.each do |row|
  name, team, capacity, lat, long = row
 
  node_id = HTTParty.post("http://localhost:7474/db/data/node", 		
    :body => { :lat => lat.to_f, :lon => long.to_f, :name => name, :team => team, :capacity => capacity }.to_json,
    :headers => { 'Content-Type' => 'application/json' } )['self'].split("/")[-1]
 
  HTTParty.post("http://localhost:7474/db/data/index/node/geom", 		
    :body => { :key => 'dummy', :value => 'dummy', :uri => "http://localhost:7474/db/data/node/#{node_id}"}.to_json,
    :headers => { 'Content-Type' => 'application/json' } )
end

One change from the previous version is that I’m not indexing the stadiums using point based geometry rather than wkt.

If we want to find the number of stadiums within 10 km of Centre Point in London we’d write the following query:

START node = node:geom('withinDistance:[51.521348,-0.128113, 10.0]') 
RETURN node.name, node.team;
==> +--------------------------------------------+
==> | node.name          | node.team             |
==> +--------------------------------------------+
==> | "Emirates Stadium" | "Arsenal"             |
==> | "Stamford Bridge"  | "Chelsea"             |
==> | "The Den"          | "Millwall"            |
==> | "Loftus Road"      | "Queens Park Rangers" |
==> | "Craven Cottage"   | "Fulham"              |
==> | "Brisbane Road"    | "Leyton Orient"       |
==> +--------------------------------------------+
==> 6 rows

I have put the code on github in case you’re interested in playing around with it.

Be Sociable, Share!

Written by Mark Needham

June 24th, 2013 at 7:17 am

Posted in neo4j

Tagged with