DuckDB rewrite: add unique indexes for primary keys

Author: derhuerst

lib/agency.js

@@ -31,6 +31,9 @@ FROM read_csv(
 	-- We omit DATE/TIME/TIMESTAMP because GTFS formats them differently.
 	auto_type_candidates = ['NULL', 'BIGINT', 'DOUBLE', 'VARCHAR']
 );
+
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX agency_agency_id ON agency(agency_id);
 `)
 
 	workingState.nrOfRowsByName.set('agency', await queryNumberOfRows(db, 'agency', opt))

lib/levels.js

@@ -30,6 +30,8 @@ FROM read_csv(
 	}
 );
 
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX levels_level_id ON levels(level_id);
 `)
 
 	workingState.nrOfRowsByName.set('levels', await queryNumberOfRows(db, 'levels', opt))

lib/pathways.js

@@ -83,6 +83,9 @@ FROM read_csv(
 		${has_reversed_signposted_as ? `, reversed_signposted_as: 'TEXT'` : ``}
 	}
 );
+
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX pathways_pathway_id ON pathways(pathway_id);
 `)
 
 	workingState.nrOfRowsByName.set('pathways', await queryNumberOfRows(db, 'pathways', opt))

lib/routes.js

@@ -311,6 +311,9 @@ FROM read_csv(
 	}
 );
 
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX routes_route_id ON routes(route_id);
+
 CREATE INDEX routes_route_short_name ON routes (route_short_name);
 `)
 

lib/shapes.js

@@ -1,6 +1,7 @@
 'use strict'
 
 const GET = require('./get.js')
+const RUN = require('./run.js')
 const {queryIfColumnsExist} = require('./columns.js')
 
 // https://gtfs.org/documentation/schedule/reference/#shapestxt
@@ -66,6 +67,11 @@ FROM (
 	ORDER BY shape_id, shape_pt_sequence
 ) t
 GROUP BY shape_id;
+`)
+
+	await db[RUN](`\
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX shapes_shape_id ON shapes(shape_id);
 `)
 
 	// Note: This is not the number of shapes.txt rows!

lib/stop_times.js

@@ -90,6 +90,9 @@ FROM read_csv(
 	}
 );
 
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX stop_times_trip_id_stop_sequence ON stop_times(trip_id, stop_sequence);
+
 -- todo: are all of them beneficial/necessary?
 CREATE INDEX stop_times_trip_id ON stop_times (trip_id);
 CREATE INDEX stop_times_stop_id ON stop_times (stop_id);

lib/stops.js

@@ -179,6 +179,9 @@ ORDER BY root_id, recursion_level, stop_id;
 -- ADD CONSTRAINT stops_parent_station_fkey
 -- FOREIGN KEY (parent_station) REFERENCES stops;
 
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX stops_stop_id ON stops(stop_id);
+
 CREATE INDEX stops_parent_station ON stops (parent_station);
 ${opt.stopsLocationIndex ? `CREATE INDEX stops_stop_loc ON stops (stop_loc);` : ''}
 `)

lib/trips.js

@@ -82,6 +82,9 @@ FROM read_csv(
 		${has_bikes_allowed ? `bikes_allowed: 'INTEGER',` : ``}
 	}
 );
+
+-- For a primary key, DuckDB doesn't create an index automatically.
+CREATE UNIQUE INDEX trips_trip_id ON trips(trip_id);
 `)
 
 	workingState.nrOfRowsByName.set('trips', await queryNumberOfRows(db, 'trips', opt))