Tech doc for Hub Allocation

Problem: searching whether a given point is contained within a partition using complete poygen( columns which is of type geometry made by converting geoJson data to WKB)

Solution: Implementing indexing on a new Column bounds which will be of data type geometry by converting { lat_min , lat_max , lng_min , lng_max } for a partition.

Faster initial filtering: Can quickly eliminate partitions that don't intersect the query point's bounding box.
Reduced index size: Smaller than an index on the full polygon.

CREATE INDEX idx_partition_bounds ON partition USING gist(bounds);
//using GiST as it is better than GiN index for spatial data.

Implementation:

→ Adding one more column as hub_id to store to which hub a partition belongs

FindAllPartitions :

We will be fetching all partitions which contains given point(lat,lng) using ST_Contains which will check whether boundary of a partition contains given point geom or not.

Partition.findAllRequiredPartitions = async (organisationId, params = {}, options = {}) => {
		assert(organisationId, 'organisationId is required');
		const { lat, lng } = params;
		if (!lat || !lng) {
			throw helper.wrongInputError('Invalid location details');;
		}
		const query = `
			SELECT *
			FROM partition
			WHERE
			organisation_id = $1
			AND ST_Contains(bounds,ST_SetSRID (ST_MakePoint ($2, $3), 4326))
			AND is_active = $4
			`;
			//ST_Contains checks whether first geometry contains second geometry
			//ST_MakePoint : to create a point geometry. It takes coordinate values as input and returns a point object.
		const queryResult = await helper.executeQueryAsync(Partition, query, [organisationId, lat, lng, true], options);
		if (Array.isArray(queryResult) && queryResult.length) {
			return queryResult;
		}
		return [];
	};

Doubt: should we also include (ST_Intersects(bounds,ST_SetSRID (ST_MakePoint ($2, $3), 4326)) this in where condition

FindValidParition:

Fallback as of now is to throw err if no partition is found or if more than 1 is found.

	Partition.findValidPartition = async (organisationId, params = {}, options = {}) => {
		assert(organisationId, 'organisationId is required');
		const {lat, lng} = params;
		const currentTx = options.transaction;
		if (!lat || !lng) {
			return {};
		}
		const query = `
			SELECT *
			FROM partition
			WHERE
			organisation_id = $1
			AND ST_Contains(bounds,ST_SetSRID (ST_MakePoint ($2, $3), 4326))
			AND is_active = $4
			`;
		const queryResult = await helper.executeQueryAsync(Partition, query, [organisationId, lat, lng, true], {transaction: currentTx});
		if (Array.isArray(queryResult) && queryResult.length === 1) {
			return queryResult[0];
		} else {
			return helper.wrongInputError('Invalid location details');
		}
	};

UpdatePartition :

CreatePartition :

For hub Allocation using partition based method, a new IDB key will be used partition_based_hub_allocation while creating softadata and calling fn getConsignmentOriginHub

Similarly for destination hub allocation following logic will be added in getMultipleConsignmentDestinationHub fn:

Confluence Trial Project

Tech doc for Hub Allocation

Analytics

Related content