This project is an unofficial port of awslabs/dynamodb-geo, bringing creation and querying of geospatial data to Node JS developers using Amazon DynamoDB.
- Box Queries: Return all of the items that fall within a pair of geo points that define a rectangle as projected onto a sphere.
- Radius Queries: Return all of the items that are within a given radius of a geo point.
- Basic CRUD Operations: Create, retrieve, update, and delete geospatial data items.
- Customizable: Access to raw request and result objects from the AWS SDK for javascript.
- Fully Typed: This port is written in typescript and declaration files are bundled into releases.
Using npm or yarn:
npm install --save dynamodb-geo or yarn add dynamodb-geo.
First you'll need to import the AWS sdk and set up your DynamoDB connection:
const AWS = require('aws-sdk');
const ddb = new AWS.DynamoDB({ endpoint: new AWS.Endpoint('http://localhost:8000') }); // Local developmentNext you must create an instance of GeoDataManagerConfiguration for each geospatial table you wish to interact with. This is a container for various options (see API below), but you must always provide a DynamoDB instance and a table name.
const ddbGeo = require('dynamodb-geo');
const config = new ddbGeo.GeoDataManagerConfiguration(ddb, 'MyGeoTable');You may modify the config to change defaults.
config.longitudeFirst = true; // Use spec-compliant GeoJSON, incompatible with awslabs/dynamodb-geoFinally, you should instantiate a manager to query and write to the table using this config object.
const myGeoTableManager = new ddbGeo.GeoDataManager(config);The hashKeyLength is the number of most significant digits (in base 10) of the 64-bit geo hash to use as the hash key. Larger numbers will allow small geographical areas to be spread across DynamoDB partitions, but at the cost of performance as more queries need to be executed for box/radius searches that span hash keys. See these tests for an idea of how query performance scales with hashKeyLength for different search radii.
If your data is sparse, a large number will mean more RCUs since more empty queries will be executed and each has a minimum cost. However if your data is dense and hashKeyLength too short, more RCUs will be needed to read a hash key and a higher proportion will be discarded by server-side filtering.
From the AWS Query documentation
DynamoDB calculates the number of read capacity units consumed based on item size, not on the amount of data that is returned to an application. ... The number will also be the same whether or not you use a
FilterExpression
Optimally, you should pick the largest hashKeyLength your usage scenario allows. The wider your typical radius/box queries, the smaller it will need to be.
Note that the Java version uses a hashKeyLength of 6 by default. The same value will need to be used if you access the same data with both clients.
This is an important early choice, since changing your hashKeyLength will mean recreating your data.
GeoTableUtil has a static method getCreateTableRequest for helping you prepare a DynamoDB CreateTable request request, given a GeoDataManagerConfiguration.
You can modify this request as desired before executing it using AWS's DynamoDB SDK.
Example:
// Pick a hashKeyLength appropriate to your usage
config.hashKeyLength = 3;
// Use GeoTableUtil to help construct a CreateTableInput.
const createTableInput = ddbGeo.GeoTableUtil.getCreateTableRequest(config);
// Tweak the schema as desired
createTableInput.ProvisionedThroughput.ReadCapacityUnits = 2;
console.log('Creating table with schema:');
console.dir(createTableInput, { depth: null });
// Create the table
ddb.createTable(createTableInput).promise()
// Wait for it to become ready
.then(function () { return ddb.waitFor('tableExists', { TableName: config.tableName }).promise() })
.then(function () { console.log('Table created and ready!') });myGeoTableManager.putPoint({
RangeKeyValue: { S: '1234' }, // Use this to ensure uniqueness of the hash/range pairs.
GeoPoint: { // An object specifying latitutde and longitude as plain numbers. Used to build the geohash, the hashkey and geojson data
latitude: 51.51,
longitude: -0.13
},
PutItemInput: { // Passed through to the underlying DynamoDB.putItem request. TableName is filled in for you.
Item: { // The primary key, geohash and geojson data is filled in for you
country: { S: 'UK' }, // Specify attribute values using { type: value } objects, like the DynamoDB API.
capital: { S: 'London' }
},
// ... Anything else to pass through to `putItem`, eg ConditionExpression
}
}).promise()
.then(function() { console.log('Done!') });See also DynamoDB PutItem request
Note that you cannot update the hash key, range key, geohash or geoJson. If you want to change these, you'll need to recreate the record.
You must specify a RangeKeyValue, a GeoPoint, and an UpdateItemInput matching the DynamoDB UpdateItem request (TableName and Key are filled in for you).
myGeoTableManager.updatePoint({
RangeKeyValue: { S: '1234' },
GeoPoint: { // An object specifying latitutde and longitude as plain numbers.
latitude: 51.51,
longitude: -0.13
},
UpdateItemInput: { // TableName and Key are filled in for you
UpdateExpression: 'SET country = :newName',
ExpressionAttributeValues: {
':newName': { S: 'United Kingdom'}
}
}
}).promise()
.then(function() { console.log('Done!') });You must specify a RangeKeyValue and a GeoPoint. Optionally, you can pass DeleteItemInput matching DynamoDB DeleteItem request (TableName and Key are filled in for you).
myGeoTableManager.deletePoint({
RangeKeyValue: { S: '1234' },
GeoPoint: { // An object specifying latitutde and longitude as plain numbers.
latitude: 51.51,
longitude: -0.13
},
DeleteItemInput: { // Optional, any additional parameters to pass through.
// TableName and Key are filled in for you
// Example: Only delete if the point does not have a country name set
ConditionExpression: 'attribute_not_exists(country)'
}
}).promise()
.then(function() { console.log('Done!') });Query by rectangle by specifying a MinPoint and MaxPoint.
// Querying a rectangle
myGeoTableManager.queryRectangle({
MinPoint: {
latitude: 52.225730,
longitude: 0.149593
},
MaxPoint: {
latitude: 52.889499,
longitude: 0.848383
}
})
// Print the results, an array of DynamoDB.AttributeMaps
.then(console.log);Query by radius by specifying a CenterPoint and RadiusInMeter.
// Querying 100km from Cambridge, UK
myGeoTableManager.queryRadius({
RadiusInMeter: 100000,
CenterPoint: {
latitude: 52.225730,
longitude: 0.149593
}
})
// Print the results, an array of DynamoDB.AttributeMaps
.then(console.log);TODO: Docs (see the example for an example of a batch write)
These are public properties of a GeoDataManagerConfiguration instance. After creating the config object you may modify these properties.
Whether queries use the ConsistentRead option (for strongly consistent reads) or not (for eventually consistent reads, at half the cost).
This can also be overridden for individual queries as a query config option.
This library will automatically add GeoJSON-style position data to your stored items. The GeoJSON standard uses [lon,lat] ordering, but awslabs/dynamodb-geo uses [lat,lng].
This fork allows you to choose between awslabs/dynamodb-geo compatibility and GeoJSON standard compliance.
- Use
false([lat, lon]) for compatibility with awslabs/dynamodb-geo - Use
true([lon, lat]) for GeoJSON standard compliance. (default)
Note that this value should match the state of your existing data - if you change it you must update your database manually, or you'll end up with ambiguously mixed data.
The value of the type attribute in recorded GeoJSON points. Should normally be "Point", which is standards compliant.
Use "POINT" for compatibility with awslabs/dynamodb-geo.
This setting is only relevant for writes. This library doesn't inspect or set this value when reading/querying.
The name of the attribute storing the full 64-bit geohash. Its value is auto-generated based on item coordinates.
The name of the attribute storing the first hashKeyLength digits (default 2) of the geo hash, used as the hash (aka partition) part of a hash/range primary key pair. Its value is auto-generated based on item coordinates.
See above.
The name of the attribute storing the range key, used as the range (aka sort) part of a hash/range key primary key pair. Its value must be specified by you (hash-range pairs must be unique).
The name of the attribute which will contain the longitude/latitude pair in a GeoJSON-style point (see also longitudeFirst).
The name of the index to be created against the geohash. Only used for creating new tables.
See the example on Github
Currently, the library does not support composite keys. You may want to add tags such as restaurant, bar, and coffee shop, and search locations of a specific category; however, it is currently not possible. You need to create a table for each tag and store the items separately.
Although low level DynamoDB Query requests return paginated results, this library automatically pages through the entire result set. When querying a large area with many points, a lot of Read Capacity Units may be consumed.
The library retrieves candidate Geo points from the cells that intersect the requested bounds. The library then post-processes the candidate data, filtering out the specific points that are outside the requested bounds. Therefore, the consumed Read Capacity Units will be higher than the final results dataset. Typically 8 queries are exectued per radius or box search.
Because all paginated Query results are loaded into memory and processed, it may consume substantial amounts of memory for large datasets.
The Geohash used in this library is roughly centimeter precision. Therefore, the library is not suitable if your dataset has much higher density.