6.3.3 Custom providers

Implement ApiDataAccessLayer to connect an application storage service or wrap an existing provider with capabilities such as logging. Dataway uses this interface to query records and write batches.

Implementation contract

Implement ApiDataAccessLayer with these contracts:

• listObjects: match filters using case-sensitive equality combined with AND; each record includes ID and REVISION.
• write: apply each ordered batch atomically, undoing the whole batch on failure. Creation starts at revision 1; updates compare and increment the revision; deletes compare it before removal.
• Preserve omitted fields and clear explicit nulls. INFO routes are unique by (METHOD, PATH); route or revision conflicts throw DataConflictException.
• create() may return a custom DataMutation subtype. configureMapping applies name mappings; providers without mapping support reject non-empty overrides.

The store must ensure batch atomicity, for example through database transactions or Nacos snapshot CAS.

Example: log metadata writes

This wrapper logs the number of changes after a batch write returns successfully. It delegates queries, name mappings and mutation creation to the underlying provider, preserving its transactions, concurrency checks and extension points.

LoggingDataAccessLayer.java

import java.util.List;
import java.util.Map;
import net.hasor.dataway.dal.ApiDataAccessLayer;
import net.hasor.dataway.dal.DataMutation;
import net.hasor.dataway.dal.EntityType;
import net.hasor.dataway.dal.FieldDef;

public class LoggingDataAccessLayer implements ApiDataAccessLayer {
    private static final System.Logger LOGGER = System.getLogger(LoggingDataAccessLayer.class.getName());
    private final ApiDataAccessLayer delegate;

    public LoggingDataAccessLayer(ApiDataAccessLayer delegate) {
        this.delegate = delegate;
    }

    @Override
    public List<Map<FieldDef, String>> listObjects(EntityType type, Map<FieldDef, String> conditions) {
        return this.delegate.listObjects(type, conditions);
    }

    @Override
    public void write(List<DataMutation> mutations) {
        this.delegate.write(mutations);
        LOGGER.log(System.Logger.Level.INFO, "Applied metadata batch: {0} mutations", mutations.size());
    }

    @Override
    public DataMutation create() {
        return this.delegate.create();
    }

    @Override
    public void configureMapping(Map<EntityType, String> tables, Map<EntityType, Map<FieldDef, String>> fields) {
        this.delegate.configureMapping(tables, fields);
    }
}

Connect to Dataway

This example stores metadata through JDBC. source is a data source with the metadata tables already created:

DatawayConfig

import net.hasor.dataway.dal.ApiDataAccessLayer;
import net.hasor.dataway.dal.jdbc.JdbcDataAccessLayer;
import net.hasor.dataway.service.Dataway;
import net.hasor.dataway.service.DatawayConfig;

ApiDataAccessLayer storage = new JdbcDataAccessLayer(source);
ApiDataAccessLayer metadata = new LoggingDataAccessLayer(storage);
Dataway dataway = new DatawayConfig()
        .dataAccessLayer(metadata)
        .createDataway();

For Nacos, replace storage with NacosDataAccessLayer. See bean registration in Spring, Solon and Hasor.

This example reuses an existing provider's storage capabilities. A new storage backend must implement queries and atomic batch writes, and verify revision conflicts, duplicate routes, batch rollback and clearing fields with explicit nulls.