com.dooapp.gaedo.blueprints.AbstractBluePrintsBackedFinderService Maven / Gradle / Ivy
package com.dooapp.gaedo.blueprints;
import java.util.Arrays;
import java.util.Collection;
import java.util.Collections;
import java.util.HashMap;
import java.util.LinkedList;
import java.util.Map;
import java.util.SortedSet;
import java.util.TreeSet;
import java.util.logging.Level;
import java.util.logging.Logger;
import javax.persistence.CascadeType;
import com.dooapp.gaedo.blueprints.indexable.IndexableGraphBackedFinderService;
import com.dooapp.gaedo.blueprints.operations.Deleter;
import com.dooapp.gaedo.blueprints.operations.Loader;
import com.dooapp.gaedo.blueprints.operations.Updater;
import com.dooapp.gaedo.blueprints.strategies.GraphMappingStrategy;
import com.dooapp.gaedo.blueprints.strategies.StrategyType;
import com.dooapp.gaedo.blueprints.strategies.StrategyUtils;
import com.dooapp.gaedo.blueprints.transformers.Literals;
import com.dooapp.gaedo.blueprints.transformers.Tuples;
import com.dooapp.gaedo.extensions.migrable.Migrator;
import com.dooapp.gaedo.extensions.migrable.VersionMigratorFactory;
import com.dooapp.gaedo.extensions.views.InViewService;
import com.dooapp.gaedo.finders.FinderCrudService;
import com.dooapp.gaedo.finders.Informer;
import com.dooapp.gaedo.finders.QueryBuilder;
import com.dooapp.gaedo.finders.QueryExpression;
import com.dooapp.gaedo.finders.QueryStatement;
import com.dooapp.gaedo.finders.expressions.Expressions;
import com.dooapp.gaedo.finders.id.IdBasedService;
import com.dooapp.gaedo.finders.repository.ServiceRepository;
import com.dooapp.gaedo.finders.root.AbstractFinderService;
import com.dooapp.gaedo.finders.root.InformerFactory;
import com.dooapp.gaedo.properties.Property;
import com.dooapp.gaedo.properties.PropertyProvider;
import com.tinkerpop.blueprints.Direction;
import com.tinkerpop.blueprints.Edge;
import com.tinkerpop.blueprints.IndexableGraph;
import com.tinkerpop.blueprints.TransactionalGraph;
import com.tinkerpop.blueprints.Vertex;
/**
* Base class for all finder service using blueprints graphs as storage
*
* @author ndx
*
* @param
* @param
* @param
*/
public abstract class AbstractBluePrintsBackedFinderService>
extends AbstractFinderService
implements InViewService>,
IdBasedService {
/**
* Delegating driver delegates, like its name suggest, absolutly all calls to embedding service, as directly as possible.
* This implementation should be used only when no contextual information is available.
* @author ndx
*
*/
private class DelegatingDriver implements GraphDatabaseDriver {
@Override
public Vertex loadVertexFor(String objectVertexId, String className) {
return AbstractBluePrintsBackedFinderService.this.loadVertexFor(objectVertexId, className);
}
@Override
public Vertex createEmptyVertex(Class valueClass, String vertexId, Object value) {
return AbstractBluePrintsBackedFinderService.this.createEmptyVertex(vertexId, valueClass, value);
}
@Override
public String getIdOf(Vertex objectVertex) {
return getIdOfVertex(objectVertex);
}
@Override
public String getEffectiveType(Vertex vertex) {
return AbstractBluePrintsBackedFinderService.this.getEffectiveType(vertex);
}
@Override
public void setValue(Vertex vertex, Object value) {
AbstractBluePrintsBackedFinderService.this.setValue(vertex, value);
}
@Override
public Object getValue(Vertex vertex) {
return AbstractBluePrintsBackedFinderService.this.getValue(vertex);
}
@Override
public ServiceRepository getRepository() {
return AbstractBluePrintsBackedFinderService.this.getRepository();
}
@Override
public Edge createEdgeFor(Vertex fromVertex, Vertex toVertex, Property property) {
return AbstractBluePrintsBackedFinderService.this.createEdgeFor(fromVertex, toVertex, property);
}
@Override
public void removeSafely(Vertex notYetDeleted) {
GraphUtils.removeSafely(database, notYetDeleted);
}
}
private static final Logger logger = Logger.getLogger(IndexableGraphBackedFinderService.class.getName());
/**
* Graph used as database
*/
protected final GraphClass database;
/**
* Graph casted as transactional one if possible. It is used to offer
* support of transactionnal read operations (if graph is indeed a
* transactional one). This field may be NULL.
*/
protected final TransactionalGraph transactionSupport;
/**
* Accelerator cache linking classes objects to the collection of properties
* and cascade informations associated to persist those fields.
*/
protected Map, Map>> classes = new HashMap, Map>>();
/**
* Property provider indicating what, and how, saving infos from object
*/
protected final PropertyProvider propertyProvider;
/**
* Migrator for given contained class
*/
protected final Migrator migrator;
/**
* Get access to the service repository to handle links between objects
*/
protected final ServiceRepository repository;
private GraphMappingStrategy strategy;
/**
* Used lens is a list of named graphs uris
*/
protected SortedSet lens = new TreeSet(Arrays.asList(GraphUtils.GAEDO_CONTEXT));
public AbstractBluePrintsBackedFinderService(GraphClass graph, Class containedClass, Class informerClass, InformerFactory factory,
ServiceRepository repository, PropertyProvider provider) {
this(graph, containedClass, informerClass, factory, repository, provider, StrategyType.beanBased);
}
/**
* Constructor defining a service using a strategy type
* @param graph
* @param containedClass
* @param informerClass
* @param factory
* @param repository2
* @param provider
* @param beanbased
*/
public AbstractBluePrintsBackedFinderService(GraphClass graph, Class containedClass, Class informerClass, InformerFactory factory,
ServiceRepository repository, PropertyProvider provider, StrategyType strategy) {
this(graph, containedClass, informerClass, factory, repository, provider,
StrategyUtils.loadStrategyFor(strategy, containedClass, provider, VersionMigratorFactory.create(containedClass)));
}
public AbstractBluePrintsBackedFinderService(GraphClass graph, Class containedClass, Class informerClass, InformerFactory informerFactory,
ServiceRepository repository, PropertyProvider provider, GraphMappingStrategy strategy) {
super(containedClass, informerClass, informerFactory);
this.repository = repository;
this.propertyProvider = provider;
this.database = graph;
if (graph instanceof TransactionalGraph) {
transactionSupport = (TransactionalGraph) graph;
} else {
transactionSupport = null;
}
this.strategy = strategy;
strategy.reloadWith(this);
this.migrator = VersionMigratorFactory.create(containedClass);
// if there is a migrator, generate property from it
if (logger.isLoggable(Level.FINE)) {
logger.log(Level.FINE, "created graph service handling " + containedClass.getName());
}
}
@Override
public InformerFactory getInformerFactory() {
return super.getInformerFactory();
}
protected abstract Edge createEdgeFor(Vertex fromVertex, Vertex toVertex, Property property);
protected abstract Object getValue(Vertex vertex);
protected abstract void setValue(Vertex vertex, Object value);
protected abstract String getEffectiveType(Vertex vertex);
/**
* Get id of a given vertex, using any meanys required by implementation
* @param objectVertex
* @return
*/
protected abstract String getIdOfVertex(Vertex objectVertex);
/**
* Creates an empty vertex with given vertex id and vertex contained value class
* @param vertexId vertex id
* @param valueClass value class
* @param value TODO
* @return a vertex storing those informations
*/
protected abstract Vertex createEmptyVertex(String vertexId, Class valueClass, Object value);
/**
* To put object in graph, we have to find all its fields, then put them in
* graph elements. Notice this method directly calls
* {@link #doUpdate(Object, CascadeType, Map)}, just checking before that if
* an id must be generated. If an id must be generated, then it is (and so
* is associated vertex, to make sure no problem will arise later).
*
* @param toCreate
* @return
* @see com.dooapp.gaedo.AbstractCrudService#create(java.lang.Object)
*/
@Override
public DataType create(final DataType toCreate) {
return new TransactionalOperation(this) {
@Override
protected DataType doPerform() {
return doUpdate(toCreate, CascadeType.PERSIST, ObjectCache.create(CascadeType.PERSIST));
}
}.perform();
}
/**
* Delete id and all edges
*
* @param toDelete
* @see com.dooapp.gaedo.AbstractCrudService#delete(java.lang.Object)
*/
@Override
public void delete(final DataType toDelete) {
if (toDelete != null) {
new TransactionalOperation(this) {
@Override
protected Void doPerform() {
doDelete(toDelete, ObjectCache.create(CascadeType.REMOVE));
return null;
}
}.perform();
}
}
/**
* Local delete implementation
*
* @param toDelete
*/
private void doDelete(DataType toDelete, ObjectCache objectsBeingAccessed) {
String vertexId = getIdVertexId(toDelete, false /*
* no id
* generation
* on delete
*/);
Class toDeleteClass = toDelete.getClass();
Vertex objectVertex = loadVertexFor(vertexId, toDeleteClass.getName());
if (objectVertex != null) {
Map> containedProperties = strategy.getContainedProperties(toDelete, objectVertex, CascadeType.REMOVE);
new Deleter().performDelete(this, new VertexCachingDriver(new DelegatingDriver()), vertexId, objectVertex, toDeleteClass, containedProperties, toDelete, CascadeType.REMOVE, objectsBeingAccessed);
}
}
/**
* Delete an out edge vertex. Those are vertex corresponding to properties.
*
* @param objectVertex
* source object vertex, used for debugging purpose only
* @param valueVertex
* value vertex to remove
* @param value
* object value
*/
public void deleteOutEdgeVertex(Vertex objectVertex, Vertex valueVertex, Type value, ObjectCache objectsBeingUpdated) {
// Delete vertex and other associated ones, only if they have no
// other input links (elsewhere delete is silently ignored)
if (valueVertex.getEdges(Direction.IN).iterator().hasNext()) {
// There are incoming edges to that vertex. Do nothing but log
// it
if (logger.isLoggable(Level.FINE)) {
logger.log(Level.FINE,
"while deleting " + GraphUtils.toString(objectVertex) + "" + " we tried to delete " + GraphUtils.toString(valueVertex)
+ "" + " which has other incoming edges, so we didn't deleted it");
}
} else {
// OK, time to delete value vertex. Is it a managed node ?
if (repository.containsKey(value.getClass())) {
FinderCrudService finderCrudService = (FinderCrudService) repository.get(value.getClass());
if (finderCrudService instanceof AbstractBluePrintsBackedFinderService) {
((AbstractBluePrintsBackedFinderService) finderCrudService).doDelete(value, objectsBeingUpdated);
} else {
throw new IncompatibleServiceException(finderCrudService, value.getClass());
}
} else {
// Literal nodes can be deleted without any trouble
GraphUtils.removeSafely(database, valueVertex);
}
}
}
/**
* Gets the id vertex for the given object (if that object exists).
* Method is made public to allow some tests to call it
*
* @param object
* object to get id vertex for
* @param allowIdGeneration
* when set to true, an id may be created for that object
* @return first matching node if found, and null if not
*/
public Vertex getIdVertexFor(DataType object, boolean allowIdGeneration) {
String objectId = getIdVertexId(object, allowIdGeneration);
if(objectId==null)
return null;
return loadVertexFor(objectId, object.getClass().getName());
}
/**
* Notice it only works if id is a literal type.
* Method is made public to allow some tests to call it
*
* @param object
* object for which we want the id vertex id property
* @param requiresIdGeneration
* set to true when effective id generation is required. Allow to
* generate id only on create operations
* @return a composite id containing the service class, the data class and
* the instance value. This may be null if object has no id defined.
* @see GraphUtils#getIdVertexId(IndexableGraph, Class, Object, Property)
*/
public String getIdVertexId(DataType object, boolean requiresIdGeneration) {
if (requiresIdGeneration) {
strategy.generateValidIdFor(object);
}
return strategy.getIdString(object);
}
/**
* Get id of given object, provided of course it's an instance of this class
*
* @param data
* object to extract an id for
* @return id of that object
*/
public Object getIdOf(DataType data) {
return getIdVertexId(data, false);
}
@Override
public DataType update(final DataType toUpdate) {
return new TransactionalOperation(this) {
@Override
protected DataType doPerform() {
return doUpdate(toUpdate, CascadeType.MERGE, ObjectCache.create(CascadeType.MERGE));
}
}.perform();
}
/**
* here is a trick : we want id generation to happen only on first persist
* (that's to say on call to #create), but not on subsequent ones. So, as
* first call uses CascadeType.PERSIST and others uses CascadeType.MERGE, we
* can use that indication to separate them. It has the unfortunate
* inconvenient to force us to use only PERSIST during #create
*
* @param toUpdate
* object to update. if null, nothing will be done.
* @param cascade
* type. As mentionned upper, beware to value used !
* @param treeMap
* map of objects already used
*/
private DataType doUpdate(DataType toUpdate, CascadeType cascade, ObjectCache treeMap) {
if(toUpdate==null)
return toUpdate;
boolean generatesId = strategy.isIdGenerationRequired() ? (CascadeType.PERSIST == cascade) : false;
String objectVertexId = getIdVertexId(toUpdate, generatesId);
Class toUpdateClass = toUpdate.getClass();
Vertex objectVertex = loadVertexFor(objectVertexId, toUpdateClass.getName());
Map> containedProperties = strategy.getContainedProperties(toUpdate, objectVertex, cascade);
return (DataType) new Updater().performUpdate(this, new VertexCachingDriver(getDriver()), objectVertexId, objectVertex, toUpdateClass, containedProperties, toUpdate,
cascade, treeMap);
}
/**
* Get vertex associated to value. If object is managed by a service, we ask
* this service the value
*
* @param value
* value we want the vertex for
* @param cascade
* used cascade type, can be either {@link CascadeType#PERSIST}
* or {@link CascadeType#MERGE}
* @param objectsBeingUpdated
* map of objects currently being updated, it avoid some loops
* during update, but is absolutely NOT a persistent cache
* @return vertex for given value. May be null in some rare cases (typically obtaining a vertex that doesn't exist yet for deleting it)
*/
public Vertex getVertexFor(Object value, CascadeType cascade, ObjectCache objectsBeingUpdated) {
boolean allowIdGeneration = CascadeType.PERSIST.equals(cascade) || CascadeType.MERGE.equals(cascade);
// Here we suppose the service is the right one for the job (which may
// not be the case)
if (containedClass.isInstance(value)) {
return getVertexForInstanceOfDataType(value, cascade, objectsBeingUpdated, allowIdGeneration);
}
Class valueClass = value.getClass();
if (repository.containsKey(valueClass)) {
FinderCrudService service = repository.get(valueClass);
if (service instanceof AbstractBluePrintsBackedFinderService) {
return ((AbstractBluePrintsBackedFinderService) service).getVertexFor(value, cascade, objectsBeingUpdated);
} else {
throw new IncompatibleServiceException(service, valueClass);
}
} else if (Literals.containsKey(valueClass)) {
throw new CantCreateAVertexForALiteralException("impossible to create a vertex for "+value);
} else if (Tuples.containsKey(valueClass)) {
return getVertexForTuple(new VertexCachingDriver(getDriver()), value, cascade, objectsBeingUpdated);
} else {
/*
* // OK, we will persist this object by ourselves, which is really
* error-prone, but we do we have any other solution ? // But notice
* object is by design consderie Vertex objectVertex =
* objectVertex.setProperty(Properties.vertexId.name(),
* getIdVertexId(toUpdate));
* objectVertex.setProperty(Properties.kind.name(),
* Kind.managed.name());
* objectVertex.setProperty(Properties.type.name(),
* toUpdate.getClass().getName());
*/
throw new ObjectIsNotARealLiteralException(value, valueClass);
}
}
/**
* Get vertex for an instance of this service {@link #getContainedClass()}
* @param value value for which we search the vertex
* @param cascade current cascade type
* @param objectsBeingUpdated
* @param allowIdGeneration true if null id should be replaced by a new one
* @return a vertex for an instance of the given object. Under some rare circumstances, this vertex may be null.
*/
protected Vertex getVertexForInstanceOfDataType(Object value, CascadeType cascade, ObjectCache objectsBeingUpdated, boolean allowIdGeneration) {
// was there any vertex prior to that call ? (don't worry, it will be used later)
Vertex existing = getIdVertexFor(containedClass.cast(value), false);
Vertex returned = null;
if(existing==null) {
returned = getIdVertexFor(containedClass.cast(value), allowIdGeneration);
} else {
returned = existing;
}
if (returned == null) {
/*
* We als test cascade type here because we absolutely don't want to create vertices during search (as bug https://github.com/Riduidel/gaedo/issues/46 exposed)
*/
if(CascadeType.PERSIST==cascade || CascadeType.MERGE==cascade) {
doUpdate(containedClass.cast(value), cascade, objectsBeingUpdated);
}
returned = getIdVertexFor(containedClass.cast(value), allowIdGeneration);
} else {
/*
* vertex already exist, but maybe object needs an update.
* This can happen if MERGE has been set (directly or not), but also when cascading creations with @GeneratedValue set.
* Indeed, in such a case, previous call to #getIdVertexFor will create a vertex with a "good" id, which will put us in this very case.
* BUT cascade will be PERSIST, which is not supported. As a consequence, to avoid UPDATE on CREATE and support cascaded CREATE
* we check (when cascade is create) if existing vertex (that's to say the one priori to id generation) exist. If not, we cascade.
*/
if (CascadeType.MERGE == cascade ||(CascadeType.PERSIST==cascade && existing==null)) {
doUpdate(containedClass.cast(value), cascade, objectsBeingUpdated);
}
}
return returned;
}
protected Vertex getVertexForTuple(GraphDatabaseDriver driver, Object value, CascadeType cascade, ObjectCache objectsBeingUpdated) {
return GraphUtils.getVertexForTuple(this, driver, repository, value, cascade, objectsBeingUpdated);
}
/**
* Object query is done by simply looking up all objects of that class using
* a standard query
*
* @return an iterable over all objects of that class
* @see com.dooapp.gaedo.finders.FinderCrudService#findAll()
*/
@Override
public Iterable findAll() {
return find().matching(new QueryBuilder() {
/**
* An empty and starts with an initial match of true, but degrades
* it for each failure. So creating an empty and() is like creating
* a "true" statement, which in turn results into searching all
* objects of that class.
*
* @param informer
* @return an empty or matching all objects
* @see com.dooapp.gaedo.finders.QueryBuilder#createMatchingExpression(com.dooapp.gaedo.finders.Informer)
*/
@Override
public QueryExpression createMatchingExpression(InformerType informer) {
return Expressions.and();
}
}).getAll();
}
/**
* Load object starting with the given vertex root. Notice object is added
* to the accessed set with a weak key, this way, it should be faster to
* load it and to maintain instance unicity
*
* @param objectVertex
*
* @return loaded object
* @param objectsBeingAccessed
* map of objects currently being accessed, it avoid some loops
* during loading, but is absolutely NOT a persistent cache
* @see #loadObject(String, Vertex, Map)
*/
public DataType loadObject(String objectVertexId, ObjectCache objectsBeingAccessed) {
// If cast fails, well, that's some fuckin mess, no ?
Vertex objectVertex = loadVertexFor(objectVertexId, containedClass.getName());
return new Loader().loadObject(this, new VertexCachingDriver(getDriver()), objectVertexId, objectVertex, objectsBeingAccessed);
}
/**
* Load each vertex from the iterable
*
* @param verticesIterables an iterable over a list of vertices we will load
*
* @return loaded object
* @see #loadObject(Vertex,ObjectCache)
*/
public Iterable loadObjects(Iterable verticesIterable) {
return new VertexLoaderIterable(this, verticesIterable);
}
/**
* Load veretx associated to given object id.
*
* @param objectVertexId
* vertex id for which we want a vertex
* @param className class name used for value. This parameter is mainly useful to disambiguate values.
* @return loaded vertex if found, or an exception (I guess ?) if none found
*/
public abstract Vertex loadVertexFor(String objectVertexId, String className);
/**
* Load object from a vertex
*
* @param objectVertex
* @param objectsBeingAccessed
* map of objects currently being accessed, it avoid some loops
* during loading, but is absolutely NOT a persistent cache
* @return loaded object
* @see #loadObject(String, Vertex, Map)
*/
public DataType loadObject(Vertex objectVertex, ObjectCache objectsBeingAccessed) {
return new Loader().loadObject(this, objectVertex, objectsBeingAccessed);
}
/**
* we only consider first id element
*
* @param id
* collection of id
* @return object which has as vertexId the given property
* @see com.dooapp.gaedo.finders.id.IdBasedService#findById(java.lang.Object[])
*/
@Override
public DataType findById(final Object... id) {
// make sure entered type is a valid one
String vertexIdValue = strategy.getAsId(id[0]);
Vertex rootVertex = loadVertexFor(vertexIdValue, containedClass.getName());
if (rootVertex == null) {
// root vertex couldn't be found directly, mostly due to
// https://github.com/Riduidel/gaedo/issues/11
// So perform the longer (but always working) query
return find().matching(new QueryBuilder() {
@Override
public QueryExpression createMatchingExpression(InformerType informer) {
Collection ands = new LinkedList();
int index=0;
for(Property idProperty : strategy.getIdProperties()) {
ands.add(informer.get(idProperty.getName()).equalsTo(id[index++]));
}
return Expressions.and(ands.toArray(new QueryExpression[ands.size()]));
}
}).getFirst();
} else {
// root vertex can be directly found ! so load it immediatly
return loadObject(rootVertex, ObjectCache.create(CascadeType.REFRESH));
}
}
@Override
public Collection getIdProperties() {
return strategy.getIdProperties();
}
/**
* Get object associated to given key. Notice this method uses internal
* cache ({@link #objectsBeingAccessed}) before to resolve call on
* datastore.
*
* @param key
* @return
*/
public DataType getObjectFromKey(String key) {
return loadObject(key, ObjectCache.create(CascadeType.REFRESH));
}
/**
* @return the database
* @category getter
* @category database
*/
public GraphClass getDatabase() {
return database;
}
/**
* @return the repository
* @category getter
* @category repository
*/
public ServiceRepository getRepository() {
return repository;
}
/**
* Set id of object, and try to assign that object a vertex.
*
* @param value
* @param id
* @return
* @see com.dooapp.gaedo.finders.id.IdBasedService#assignId(java.lang.Object,
* java.lang.Object[])
*/
@Override
public boolean assignId(final DataType value, Object... id) {
/*
* We first make sure object is an instance of containedClass. This way,
* we can then use value class to create id vertex
*/
if (containedClass.isInstance(value)) {
strategy.assignId(value, id);
if (getIdVertexFor(value, false /*
* no id generation when assigning
* an id !
*/) == null) {
try {
TransactionalOperation operation = new TransactionalOperation(this) {
@Override
protected Boolean doPerform() {
String idVertexId = getIdVertexId(value, strategy.isIdGenerationRequired());
Vertex returned = getDriver().createEmptyVertex(value.getClass(), idVertexId, value);
getDriver().setValue(returned, idVertexId);
return true;
}
};
return operation.perform();
} catch (Exception e) {
return false;
}
} else {
return false;
}
} else {
return false;
}
}
/**
* @return the requiresIdGeneration
* @category getter
* @category requiresIdGeneration
*/
public boolean isRequiresIdGeneration() {
return strategy.isIdGenerationRequired();
}
/**
* Provides driver view to database. This driver is a way for us to expose
* low-level infos to graph without breaking huigh-level abstraction of a
* FinderService.
*
* @return a new driver each time, so please d'ont use that method !
* @deprecated driver should no more be obtained at each call from service, but rather loaded as soon as possible and kept all along query lifecycle
*/
@Deprecated
public GraphDatabaseDriver getDriver() {
return new DelegatingDriver();
}
@Override
protected QueryStatement createQueryStatement(QueryBuilder query) {
return new GraphQueryStatement(query,
this, repository);
}
/**
* @return the strategy
* @category getter
* @category strategy
*/
public GraphMappingStrategy getStrategy() {
return strategy;
}
@Override
public SortedSet getLens() {
return lens;
}
/**
* Set current lens. Notice when {@link #lens} is changed strategy is updated.
* @param lens
*/
public void setLens(SortedSet lens) {
SortedSet usedLens = new TreeSet(lens);
this.lens = Collections.unmodifiableSortedSet(usedLens);
this.strategy.reloadWith(this);
}
/**
* Informer factory has been made public to allow use of lens in finders
* @param informerFactory
* @see com.dooapp.gaedo.finders.root.AbstractFinderService#setInformerFactory(com.dooapp.gaedo.finders.root.InformerFactory)
*/
@Override
public void setInformerFactory(InformerFactory informerFactory) {
super.setInformerFactory(informerFactory);
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy