query/me.ahoo.wow.query.dataset

Package-level declarations

Types

How an access policy restricts the rows of a dataset bound to an aggregate (QueryDatasetAccess): by the current access attributes of the aggregate each row is of, never by what the row recorded when it was written. The restriction is one ELEMENT_MATCH on the reserved container FIELD, @access: the aggregate's current access record, none when the storage holds none of it, whose fields are its attributes by name as every record stores them (AccessValues: the built-in owner and space, and each declared one). It reads "the aggregate has a current access record that the predicate holds of":

DatasetQueryGateway

@ExperimentalWowApi

interface DatasetQueryGateway : QueryGateway

The query gateway of one dataset: the QueryGateway operations (single, list, paged, cursor, count, aggregate, describe), each with its explicit QueryCaller, admitted against the dataset's schema with the same filters, policies, budgets and masking as an aggregate's snapshots. Its aggregateType is the dataset's name.

DatasetQueryGateways

@ExperimentalWowApi

class DatasetQueryGateways(gateways: Collection<DatasetQueryGateway>)

The gateways of the datasets a node serves, by dataset name, following QueryDatasets: a dataset that joins or changes gets a new gateway, one that leaves loses its own.

DatasetQueryModelProfile

@ExperimentalWowApi

@InternalWowApi

class DatasetQueryModelProfile : QueryModelProfile

The record layout of one QueryDataset: the whole record is payload, its fields are the dataset's declaration, and the system fields are the columns the dataset names. A dataset has no aggregate id and no deletion flag, so a filter on either is an unknown field, and no default scope.

DatasetVisibility

@ExperimentalWowApi

fun interface DatasetVisibility

Which datasets a caller sees: those GET /wow/dataset lists, and those whose routes answer it; the routes of a dataset the caller does not see are a 404, as if it did not exist. It hides only what the caller could not read anyway, and every query is still restricted by the gateway's policies.

DefaultDatasetQueryGateway

@ExperimentalWowApi

class DefaultDatasetQueryGateway @InternalWowApi constructor(val dataset: QueryDataset, backend: QueryBackend, schemaProvider: QueryModelSchemaProvider, filters: List<QueryFilter> = emptyList(), policies: List<QueryPolicy> = emptyList(), observer: QueryObserver = QueryObserver.NONE, entryPolicy: QueryEntryPolicy = QueryEntryPolicy.DEFAULT, authorizer: ReadAuthorizer = ReadAuthorizer.ALLOW_ALL) : DatasetQueryGateway, QueryGateway

A DatasetQueryGateway over backend, admitting against the schema schemaProvider compiles.

QueryDataset

@ExperimentalWowApi

class QueryDataset @InternalWowApi constructor(val name: AggregateType, val scope: QueryDatasetScope, val declaration: QuerySchemaDeclaration, val identity: QueryField<*>? = null, val eventTime: QueryField<*>? = null, val title: String? = null, val storage: ObjectNode = WowJson.mapper.createObjectNode(), val unavailable: String? = null)

A dataset (the DATASET query model): a table-like record set, such as a BI wide table or the ClickHouse Sink's _state_flat view, declared at runtime or in a file rather than in the Manifest. Its fields are its declaration, at the record root; its record layout is its profile.

QueryDatasetAccess

@ExperimentalWowApi

class QueryDatasetAccess @InternalWowApi constructor(val aggregate: AggregateType, val id: QueryField<*>)

The aggregate whose access policy decides which of a dataset's rows a caller reads: each row is of one aggregate, the one whose id its id column holds. The rows are read as the aggregate's current access attributes let the caller read it, whatever the row recorded when it was written.

QueryDatasetBuilder

@ExperimentalWowApi

class QueryDatasetBuilder(name: AggregateType)

Declares a QueryDataset in code: its profile columns, its fields and their sensitivity.

QueryDatasetChange

@ExperimentalWowApi

sealed interface QueryDatasetChange

One change to the QueryDatasets: a dataset joined, was replaced by a new declaration, or left.

QueryDatasetConfig

@WowSpi

class QueryDatasetConfig @InternalWowApi constructor(val type: String, val options: Map<String, String> = emptyMap(), val resources: StoreResources = StoreResources.NONE)

What a QueryDatasetProvider opens from: its type, its settings (options, wow.<type>.* relative to it) and the application's resources. Fields may be added: the host builds it, so the constructor is internal to Wow; a provider's test builds one with of.

QueryDatasetFiles

@ExperimentalWowApi

class QueryDatasetFiles @JvmOverloads constructor(configDirectory: Path = Path.of("config"), classLoader: ClassLoader = Thread.currentThread().contextClassLoader ?: QueryDatasetFiles::class.java.classLoader)

Reads the dataset files (datasets are declared in files, not in the Manifest), one per dataset, named {context}.{dataset}.json in the QueryDatasetProperties format: from the working directory's <configDirectory>/wow/dataset/, which takes precedence, else from the classpath's META-INF/wow/dataset/.

QueryDatasetListener

@ExperimentalWowApi

interface QueryDatasetListener

Follows the changes of QueryDatasets: every listener validates a change before any applys it, so a listener that cannot follow a change (a gateway that cannot be built, a name an aggregate takes) refuses it for all of them.

QueryDatasetProvider

@WowSpi

interface QueryDatasetProvider

A dataset backend (the ClickHouse one, say): the QueryBindings of the datasets it serves, opened from its settings and the application's resources. The host finds it with ServiceLoader (META-INF/services) and opens the one that is available; with several it refuses the start, naming them, and an application's own QueryBindings<QueryDataset> bean takes the place of any.

QueryDatasets

@ExperimentalWowApi

class QueryDatasets(datasets: Collection<QueryDataset> = emptyList())

The runtime catalog of the datasets this node serves: declared in code, read from files (QueryDatasetFiles) or registered while the node runs, as dataset-registry does. The schema Catalog, the dataset gateways and the storages follow it through listen.

QueryDatasetScope

@ExperimentalWowApi

class QueryDatasetScope @InternalWowApi constructor(val tenant: QueryField<*>? = null, val owner: QueryField<*>? = null, val space: QueryField<*>? = null, val platform: Boolean = false, val access: QueryDatasetAccess? = null)

Which columns of a dataset hold its caller scope: the tenant (or, for a platform-level dataset, none), the owner and the space; and, when its rows are an aggregate's, the access of that aggregate. A dataset either names its tenant column or declares that it is platform-level, never both.