PostgreSQL complains about inexistent comparison function for element in primary key - postgresql

Question

I have a table in a PostgreSQL database in which I want to store the following columns:
STATION LOCATION SERVICE NORTH EAST
text point text real real
Each tuple(STATION, LOCATION, SERVICE) is unique, so I decided to make it a composite type and make it the primary key.
However, when I try to insert a new entry in the database I get the following error:
psycopg2.ProgrammingError: could not identify a comparison function for type point
I guess it is complaining that you cannot order two points in a 2D plane, but I cannot see how that is relevant. I have managed to use composite types that made use of points as primary keys in a test example, so I cannot see how this is different.
I want to know:
Why this is happening.
How it can be fixed, preferrably without changing the table schema.
Debugging information:
testdb=> \d ERROR_KEY
Composite type "public.error_key"
Column | Type | Modifiers
----------+-------+-----------
station | text |
location | point |
service | text |
testdb=> \d testtable
Table "public.testtable"
Column | Type | Modifiers
--------+-----------+-----------
key | error_key | not null
north | real |
east | real |
Indexes:
"testtable_pkey" PRIMARY KEY, btree (key)
For reference, this is the code I am using for the insertion:
from collections import namedtuple
import psycopg2
DB_NAME = 'testdb'
DB_USER = 'testuser'
DB_HOST = 'localhost'
DB_PASSWORD = '123456'
PVT_TABLE_NAME = 'testtable'
Coordinate = namedtuple('Coordinate', ['lat', 'lon'])
PVT_Error_Key = namedtuple('PVT_Error_Key',
['station', 'location', 'service'])
PVT_Error_Entry = namedtuple(
'PVT_Error_Entry', ['key', 'north', 'east'])
def _adapt_coordinate(coord):
"""
Adapter from Python class to Postgre geometric point
"""
lat = psycopg2.extensions.adapt(coord.lat)
lon = psycopg2.extensions.adapt(coord.lon)
return psycopg2.extensions.AsIs("'(%s, %s)'" % (lat, lon))
def _connect_to_db(db_name, db_user, db_host, db_password):
"""
Connects to a database and returns a cursor object to handle the connection
"""
connection_str = ('dbname=\'%s\' user=\'%s\' host=\'%s\' password=\'%s\''
% (db_name, db_user, db_host, db_password))
return psycopg2.connect(connection_str).cursor()
def main():
# Register the adapter for the location
psycopg2.extensions.register_adapter(Coordinate, _adapt_coordinate)
cursor = _connect_to_db(DB_NAME, DB_USER, DB_HOST, DB_PASSWORD)
# Create a dummy entry
entry = PVT_Error_Entry(
key=PVT_Error_Key(station='GKIR',
location=Coordinate(lat=12, lon=10),
service='E1'),
north=1, east=2)
# Insert the dummy entry in the database
cursor.execute(
'INSERT INTO %s '
'(KEY, NORTH, EAST) '
'VALUES((%%s, %%s, %%s), %%s, %%s)'
% PVT_TABLE_NAME,
(entry.key.station, entry.key.location, entry.key.service,
entry.north, entry.east))
# Retrieve and print all entries of the database
cursor.execute('SELECT * FROM %s', (PVT_TABLE_NAME))
rows = cursor.fetchall()
print(rows)
if __name__ == '__main__':
main()

Answer 1

You cannot use a column of type point in a primary key, e.g.:
create table my_table(location point primary key);
ERROR: data type point has no default operator class for access method "btree"
HINT: You must specify an operator class for the index or define a default operator class for the data type.
The error message is clear enough, you need to create a complete btree operator class for the type.
The full procedure is described in this answer: Creating custom “equality operator” for PostgreSQL type (point) for DISTINCT calls.
Update. With the workaround you mentioned in your comment
create table my_table(
x numeric,
y numeric,
primary key (x, y));
insert into my_table values
(1.1, 1.2);
you can always create a view, which can be queried just like a table:
create view my_view as
select point(x, y) as location
from my_table;
select *
from my_view;
location
-----------
(1.1,1.2)
(1 row)