- PostgreSQL - 9.5.13
- Postgis - 2.3.3
- Pgrouting 2.5.2
- Geoserver - 2.10.4
- Leaflet JS - 1.0.3
Create database - here named as routing
. Replace <user>
with your database user.
createdb routing -U <user>
Add postgis extension:
psql -U <user> -c "CREATE EXTENSION postgis;" routing
Add pgRouting extension:
psql -U <user> -c "CREATE EXTENSION pgrouting;" routing
Here we are using ogr2ogr to load roads.geojson (obtained from OSM) for a section of Nairobi. Again do not forget to replace <user>
with your database user.
ogr2ogr -select 'name,highway,oneway,surface' -lco GEOMETRY_NAME=the_geom -lco FID=id -f PostgreSQL PG:"dbname=routing user=<user>" -nln edges roads.geojson
A few things note on the above ogr2ogr command:
-select ‘name,highway,oneway,surface’
: Select the desired attributes/fields only from the data file. Other attributes in the data will not be imported-f PostgreSQL PG:”dbname=routing user=<user>
: Load the data into Postgres with<user>
and dbrouting
-lco GEOMETRY_NAME=the_geom
: Store the geometry in a field named the_geom-nlco FID=id
: Store the feature identifier in a field named id-nln edges
: Store the data in a table called edges
For more details on the possible options, please refer this ogr2ogr documentation
To accommodate pgr_createTopology
, we need to add source and target columns to our edges table and then execute the command. Note that we have to indicate the name of the table (‘edges’) and the tolerance for considering two vertices as the same in the network.
First fire up the psql
client with the correct user
and database
:
psql -U <user> -d routing
And then create the columns by typing the following:
ALTER TABLE edges ADD source INT4;
ALTER TABLE edges ADD target INT4;
SELECT pgr_nodeNetwork('edges', 0.00001);
NOTE We are using a tolerance of 0.00001 because our data is in EPSG:4326 (meter as projection unit - points have to be less than 0.00001 meters away from each other) Reference: https://gis.stackexchange.com/questions/229452/pgr-createtopology-how-tolerance
For details on pgr_nodeNetwork
function please refer from here
SELECT pgr_createTopology('edges_noded', 0.00001);
Details on pgr_createTopology
function here
Add Columns first:
ALTER TABLE edges_noded
ADD COLUMN name VARCHAR,
ADD COLUMN type VARCHAR;
Copy the data from the original table:
UPDATE edges_noded AS new
SET name=old.name,
type=old.highway
FROM edges as old
WHERE new.old_id=old.id;
We will simply use distance as the costing factor. Note you can also use other parameters like type of road, traffic etc..
Precalculate distance to save geoserver from calculating on each request:
Add Distance Column
ALTER TABLE edges_noded ADD distance FLOAT8;
Calculate distances in meters:
UPDATE edges_noded SET distance = ST_Length(ST_Transform(the_geom, 4326)::geography) / 1000;
SELECT * FROM pgr_dijkstra('SELECT id,source,target,distance as cost FROM edges_noded',1,2,false);
For details on pgr_dijkstra
please check here
Create 2 parameterized SQL Views to have the following code:
- Nearest Vertex SQL View
SELECT
v.id,
v.the_geom,
string_agg(distinct(e.name),',') AS name
FROM
edges_noded_vertices_pgr AS v,
edges_noded AS e
WHERE
v.id = (SELECT
id
FROM edges_noded_vertices_pgr
ORDER BY the_geom <-> ST_SetSRID(ST_MakePoint(%x%, %y%), 4326) LIMIT 1)
AND (e.source = v.id OR e.target = v.id)
GROUP BY v.id, v.the_geom
Validation for parametes To ensure that the sql view gets the correct parameters, add the below validation that checks the values as float type in the geoserver sql view under the parameters:
^[\d\.\+-eE]+$
- Shortest Path SQL View
SELECT
min(r.seq) AS seq,
e.old_id AS id,
e.name,
e.type,
sum(e.distance) AS distance,
ST_Collect(e.the_geom) AS geom
FROM pgr_dijkstra('SELECT id,source,target,distance AS cost
FROM edges_noded',%source%,%target%,false) AS r,edges_noded AS e
WHERE r.edge=e.id GROUP BY e.old_id,e.name,e.type
Validation Ensure parameters are integers
^[\d]+$
Please take a look at the code and it should be easy to follow along with the comments in the code