Adding all the train stations on the Sydney trains network into a knowledge graph

_create_station_relationships.py

import os

# Iterate through all .txt files in "data" folder
if not os.path.exists("output"):
    os.makedirs("output")

station_list = set()
for file in os.listdir("data"):
    if file.endswith(".txt"):
        text = open("data/{}".format(file), "r").read().split("\n")

        # Store all unseen stations
        station_list.update(text)

        # Store all relationships
        relationships = zip(text[:-1], text[1:])
        output_file_name = file.replace(".txt", ".csv")
        output = open("output/{}".format(output_file_name), "w")
        output.write("stop_id_from,stop_id_to")
        output.write("\n")
        output.write("\n".join("{},{}".format(parent, child) for parent, child in relationships))

        output.close()

# Write all stations to a file
station_list = sorted(list(station_list))
output = open("output/all_stations.csv", "w")
output.write("\n".join(station_list))
output.close()

# Print all outputted files
print(", ".join(["'{}'".format(i) for i in os.listdir("output")]))

_sydney_trains.cypher

LOAD CSV FROM 'file:///all_stations.csv' AS row
MERGE (s:Stop {name: row[0]});

UNWIND ['metro.csv', 't1.csv', 't1_western.csv', 't1_western_1.csv', 't2.csv', 't4.csv', 't4_1.csv', 't5.csv', 't6.csv', 't7.csv', 't8.csv', 't8_airport.csv', 't9.csv'] AS file
LOAD CSV WITH HEADERS FROM 'file:///' + file AS row
CALL(row) {
  MATCH (a:Stop {name: row.stop_id_from})
  MATCH (b:Stop {name: row.stop_id_to})
  OPTIONAL MATCH (a)-[r:NEXT_STOP]-(b)  // note the undirected pattern to check both directions
  WITH a, b, count(r) AS relCount
  WHERE relCount = 0
  MERGE (a)-[:NEXT_STOP]->(b)  
}

all_stations.csv

Allawah
Arncliffe
Artarmon
Ashfield
Asquith
Auburn
Banksia
Bankstown
Barangaroo
Bardwell Park
Beecroft
Bella Vista
Berala
Berowra
Beverly Hills
Bexley North
Birrong
Blacktown
Bondi Junction
Burwood
Cabramatta
Campbelltown
Canley Vale
Caringbah
Carlton
Carramar
Castle Hill
Casula
Central
Chatswood
Cheltenham
Cherrybrook
Chester Hill
Circular Quay
Clarendon
Clyde
Como
Concord West
Cronulla
Crows Nest
Croydon
Denistone
Domestic Airport
Doonside
East Hills
East Richmond
Eastwood
Edgecliff
Edmondson Park
Emu Plains
Engadine
Epping
Erskineville
Fairfield
Flemington
Gadigal
Glenfield
Gordon
Granville
Green Square
Guildford
Gymea
Harris Park
Heathcote
Helensburgh
Hills Showground
Holsworthy
Homebush
Hornsby
Hurstville
Ingleburn
International Airport
Jannali
Kellyville
Killara
Kings Cross
Kingsgrove
Kingswood
Kirrawee
Kogarah
Leightonfield
Leppington
Leumeah
Lewisham
Lidcombe
Lindfield
Liverpool
Loftus
Macarthur
Macdonaldtown
Macquarie Fields
Macquarie Park
Macquarie University
Marayong
Martin Place
Mascot
Meadowbank
Merrylands
Milsons Point
Minto
Miranda
Mortdale
Mount Colah
Mount Druitt
Mount Kuring-gai
Mulgrave
Museum
Narwee
Newtown
Normanhurst
North Ryde
North Strathfield
North Sydney
Norwest
Oatley
Olympic Park
Padstow
Panania
Parramatta
Pendle Hill
Pennant Hills
Penrith
Penshurst
Petersham
Pymble
Quakers Hill
Redfern
Regents Park
Revesby
Rhodes
Richmond
Riverstone
Riverwood
Rockdale
Rooty Hill
Roseville
Rouse Hill
Schofields
Sefton
Seven Hills
St James
St Leonards
St Marys
St Peters
Stanmore
Strathfield
Summer Hill
Sutherland
Sydenham
Tallawong
Tempe
Thornleigh
Toongabbie
Town Hall
Turramurra
Turrella
Victoria Cross
Villawood
Vineyard
Wahroonga
Waitara
Warrawee
Warwick Farm
Waterfall
Waterloo
Waverton
Wentworthville
Werrington
West Ryde
Westmead
Windsor
Wolli Creek
Wollstonecraft
Woolooware
Wynyard
Yagoona
Yennora

metro.csv

t1.csv

t1_western.csv

t1_western_1.csv

t2.csv

t4.csv

t4_1.csv

t5.csv

t6.csv

t7.csv

t8.csv

t8_airport.csv

t9.csv