Java tutorial
/** * Copyright (C) 2013 Red Hat, Inc. (jdcasey@commonjava.org) * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.commonjava.maven.cartographer.agg; import java.util.ArrayList; import java.util.Collections; import java.util.HashMap; import java.util.HashSet; import java.util.List; import java.util.Map; import java.util.Map.Entry; import java.util.Set; import java.util.concurrent.CountDownLatch; import java.util.concurrent.ExecutorService; import javax.enterprise.context.ApplicationScoped; import javax.inject.Inject; import org.apache.commons.lang.exception.ExceptionUtils; import org.commonjava.cdi.util.weft.ExecutorConfig; import org.commonjava.maven.atlas.graph.RelationshipGraph; import org.commonjava.maven.atlas.graph.RelationshipGraphException; import org.commonjava.maven.atlas.graph.model.GraphPath; import org.commonjava.maven.atlas.graph.model.GraphPathInfo; import org.commonjava.maven.atlas.graph.rel.ParentRelationship; import org.commonjava.maven.atlas.graph.rel.ProjectRelationship; import org.commonjava.maven.atlas.graph.rel.RelationshipType; import org.commonjava.maven.atlas.ident.ref.ProjectVersionRef; import org.commonjava.maven.atlas.ident.util.JoinString; import org.commonjava.maven.cartographer.data.CartoDataException; import org.commonjava.maven.cartographer.discover.DiscoveryConfig; import org.commonjava.maven.cartographer.discover.DiscoveryResult; import org.commonjava.maven.cartographer.discover.ProjectRelationshipDiscoverer; import org.slf4j.Logger; import org.slf4j.LoggerFactory; @ApplicationScoped public class DefaultGraphAggregator implements GraphAggregator { private static final int MAX_BATCHSIZE = 8; private final Logger logger = LoggerFactory.getLogger(getClass()); @Inject private ProjectRelationshipDiscoverer discoverer; @Inject @ExecutorConfig(daemon = true, named = "carto-aggregator", priority = 9, threads = 8) private ExecutorService executor; protected DefaultGraphAggregator() { } public DefaultGraphAggregator(final ProjectRelationshipDiscoverer discoverer, final ExecutorService executor) { this.discoverer = discoverer; this.executor = executor; } @Override public void connectIncomplete(final RelationshipGraph graph, final AggregationOptions config) throws CartoDataException { if (graph != null && config.isDiscoveryEnabled()) { final Set<ProjectVersionRef> missing = new HashSet<ProjectVersionRef>(); logger.debug("Loading existing cycle participants..."); // final Set<ProjectVersionRef> cycleParticipants = loadExistingCycleParticipants( net ); final Set<ProjectVersionRef> seen = new HashSet<ProjectVersionRef>(); logger.debug("Loading initial set of GAVs to be resolved..."); final List<DiscoveryTodo> pending = loadInitialPending(graph, seen); final HashSet<DiscoveryTodo> done = new HashSet<DiscoveryTodo>(); int pass = 0; while (!pending.isEmpty()) { // final HashSet<DiscoveryTodo> current = new HashSet<DiscoveryTodo>( pending ); // pending.clear(); final HashSet<DiscoveryTodo> current = new HashSet<DiscoveryTodo>(MAX_BATCHSIZE); while (!pending.isEmpty() && current.size() < MAX_BATCHSIZE) { current.add(pending.remove(0)); } done.addAll(current); logger.debug("{}. {} in next batch of TODOs:\n {}", pass, current.size(), new JoinString("\n ", current)); final Set<DiscoveryTodo> newTodos = discover(current, config, /*cycleParticipants,*/missing, seen, pass); if (newTodos != null) { logger.debug("{}. Uncovered new batch of TODOs:\n {}", pass, new JoinString("\n ", newTodos)); for (final DiscoveryTodo todo : newTodos) { if (!done.contains(todo) && !pending.contains(todo)) { logger.debug("+= {}", todo); pending.add(todo); } } } pass++; } logger.info("Discovery complete. {} seen, {} missing in {} passes.", seen.size(), missing.size(), pass - 1); } } private Set<DiscoveryTodo> discover(final Set<DiscoveryTodo> todos, final AggregationOptions config, final Set<ProjectVersionRef> missing, final Set<ProjectVersionRef> seen, final int pass) throws CartoDataException { logger.info("Starting pass: {}", pass); logger.debug("{}. Performing discovery and cycle-detection on {} missing subgraphs:\n {}", pass, todos.size(), new JoinString("\n ", todos)); final Set<DiscoveryRunnable> runnables = executeTodoBatch(todos, config, missing, seen, /*cycleParticipants,*/pass); logger.debug("{}. Accounting for discovery results. Before discovery, these were missing:\n\n {}\n\n", pass, new JoinString("\n ", missing)); final Map<ProjectVersionRef, DiscoveryTodo> nextTodos = new HashMap<ProjectVersionRef, DiscoveryTodo>(); for (final DiscoveryRunnable r : runnables) { if (!processDiscoveryOutput(r, nextTodos, config.getDiscoveryConfig(), seen, pass)) { markMissing(r, missing, pass); } } logger.info("{}. After discovery, {} are missing", pass, missing.size()); logger.debug("Missing:\n\n {}\n\n", new JoinString("\n ", missing)); return new HashSet<DiscoveryTodo>(nextTodos.values()); } /** * Convert the current set of {@link DiscoveryTodo}'s into a set of * {@link DiscoveryRunnable}'s after first ensuring their corresponding GAVs * aren't already present, listed as missing, or listed as participants in a * relationship cycle. * * Then, execute all of these runnables and wait for processing to complete * before passing them back for output processing. * * @param todos The current set of {@link DiscoveryTodo}'s to process * @param config Configuration for how discovery should proceed * @param missing The accumulated list of confirmed-missing GAVs (NOT things * that have yet to be discovered) * @param cycleParticipants The accumulated list of GAVs that participate in * relationship cycles. These are NOT safe to continue processing, since they * will lead to infinite looping. * @param pass For diagnostic/logging purposes, the number of discovery passes * since discovery was initiated by the caller (part of the graph may have been * pre-existing) * @return The executed set of {@link DiscoveryRunnable} instances that contain * output to be processed and incorporated in the graph. */ private Set<DiscoveryRunnable> executeTodoBatch(final Set<DiscoveryTodo> todos, final AggregationOptions config, final Set<ProjectVersionRef> missing, final Set<ProjectVersionRef> seen, /*final Set<ProjectVersionRef> cycleParticipants,*/final int pass) { final Set<DiscoveryRunnable> runnables = new HashSet<DiscoveryRunnable>(todos.size()); final Set<ProjectVersionRef> roMissing = Collections.unmodifiableSet(missing); int idx = 0; for (final DiscoveryTodo todo : todos) { final ProjectVersionRef todoRef = todo.getRef(); if (missing.contains(todoRef)) { logger.info("{}.{}. Skipping missing reference: {}", pass, idx++, todoRef); continue; } // else if ( cycleParticipants.contains( todoRef ) ) // { // logger.info( "{}.{}. Skipping cycle-participant reference: {}", pass, idx++, todoRef ); // continue; // } // WAS: net.containsProject(todoRef) ...this is pretty expensive, since it requires traversal. Instead, we track as we go. else if (seen.contains(todoRef)) { logger.info("{}.{}. Skipping already-discovered reference: {}", pass, idx++, todoRef); continue; } // logger.info( "DISCOVER += {}", todo ); final DiscoveryRunnable runnable = new DiscoveryRunnable(todo, config, roMissing, discoverer, pass, idx); runnables.add(runnable); idx++; } final CountDownLatch latch = new CountDownLatch(runnables.size()); for (final DiscoveryRunnable runnable : runnables) { runnable.setLatch(latch); executor.execute(runnable); } try { latch.await(); } catch (final InterruptedException e) { logger.error("Interrupted on subgraph discovery."); return null; } return runnables; } /** * Process the output from a discovery runnable (discovery of relationships * related to a given GAV). This includes: * * <ul> * <li>Adding any accumulated metadata for the GAV</li> * <li>Determining which new relationships to store in the graph db related to the relationships in this result</li> * <li>Generating the next set of {@link DiscoveryTodo}'s related to the relationships in this result</li> * </ul> * * @param r The runnable containing discovery output to process for a specific * input GAV * @param nextTodos The accumulated next crop of {@link DiscoveryTodo}'s, which * MAY be augmented by output from this discovery runnable * @param net The top-level dependency graph for which discovery is taking place * @param config Configuration for how discovery should proceed * @param seen * @param pass For diagnostic/logging purposes, the number of discovery passes * since discovery was initiated by the caller (part of the graph may have been pre-existing) * @return true if output contained a valid result, or false to indicate the * GAV should be marked missing. * @throws CartoDataException */ private boolean processDiscoveryOutput(final DiscoveryRunnable r, final Map<ProjectVersionRef, DiscoveryTodo> nextTodos, final DiscoveryConfig config, final Set<ProjectVersionRef> seen, final int pass) throws CartoDataException { final DiscoveryTodo todo = r.getTodo(); final Throwable error = r.getError(); if (error != null) { try { todo.getGraph().storeProjectError(todo.getRef(), error); } catch (final RelationshipGraphException e) { logger.error(String.format( "Failed to store error for project: %s (%s). Error was:\n\n%s.\n\nStorage error:\n", todo.getRef(), e.getMessage(), ExceptionUtils.getFullStackTrace(error)), e); } return false; } final DiscoveryResult result = r.getResult(); if (result != null) { final RelationshipGraph graph = todo.getGraph(); final Map<String, String> metadata = result.getMetadata(); if (metadata != null) { try { graph.addMetadata(result.getSelectedRef(), metadata); } catch (final RelationshipGraphException e) { logger.error(String.format("Failed to store metadata for: %s in: %s. Reason: %s", result.getSelectedRef(), graph, e.getMessage()), e); } } final Set<ProjectRelationship<?>> discoveredRels = result.getAcceptedRelationships(); if (discoveredRels != null) { final Map<GraphPath<?>, GraphPathInfo> parentPathMap = todo.getParentPathMap(); seen.add(todo.getRef()); final int index = r.getIndex(); // De-selected relationships (not mutated) should be stored but NOT followed for discovery purposes. // Likewise, mutated (selected) relationships should be followed but NOT stored. logger.info("{}.{}. Processing {} new relationships for: {}", pass, index, discoveredRels.size(), result.getSelectedRef()); logger.debug("Relationships:\n {}", new JoinString("\n ", discoveredRels)); boolean contributedRels = false; int idx = 0; for (final ProjectRelationship<?> rel : discoveredRels) { final ProjectVersionRef relTarget = rel.getTarget().asProjectVersionRef(); if (!seen.contains(relTarget)) { for (final Entry<GraphPath<?>, GraphPathInfo> entry : parentPathMap.entrySet()) { GraphPath<?> path = entry.getKey(); GraphPathInfo pathInfo = entry.getValue(); final ProjectRelationship<?> selected = pathInfo.selectRelationship(rel, path); if (selected == null) { continue; } final ProjectVersionRef selectedTarget = selected.getTarget().asProjectVersionRef(); if (seen.contains(selectedTarget)) { continue; } contributedRels = true; path = graph.createPath(path, selected); if (path == null) { continue; } pathInfo = pathInfo.getChildPathInfo(selected); DiscoveryTodo nextTodo = nextTodos.get(selectedTarget); if (nextTodo == null) { nextTodo = new DiscoveryTodo(selectedTarget, path, pathInfo, graph); nextTodos.put(selectedTarget, nextTodo); logger.info("DISCOVER += {}", selectedTarget); } else { nextTodo.addParentPath(path, pathInfo); } } if (rel.isManaged()) { logger.debug( "{}.{}.{}. FORCE; NON-TRAVERSE: Adding managed relationship (for mutator use later): {}", pass, index, idx, rel); contributedRels = true; } else if (rel.getType() == RelationshipType.PARENT) { logger.debug("{}.{}.{}. FORCE; NON-TRAVERSE: Adding parent relationship: {}", pass, index, idx, rel); contributedRels = true; } else { logger.debug("{}.{}.{}. SKIP: {}", pass, index, idx, relTarget); } } else { logger.debug("{}.{}.{}. SKIP (already discovered): {}", pass, index, idx, relTarget); } idx++; } // if all relationships have been discarded by filter... if (!contributedRels && !discoveredRels.isEmpty()) { logger.debug( "{}.{}. INJECT: Adding terminal parent relationship to mark {} as resolved in the dependency graph.", pass, index, result.getSelectedRef()); try { graph.storeRelationships( new ParentRelationship(config.getDiscoverySource(), result.getSelectedRef())); } catch (final RelationshipGraphException e) { logger.error(String.format("Failed to store relationships for: %s in: %s. Reason: %s", result.getSelectedRef(), graph, e.getMessage()), e); } } } else { logger.debug("{}.{}. discovered relationships were NULL for: {}", pass, r.getIndex(), result.getSelectedRef()); } return true; } else { return false; } } // private void addToCycleParticipants( final Set<ProjectRelationship<?>> rejectedRelationships, final Set<ProjectVersionRef> cycleParticipants ) // { // for ( final ProjectRelationship<?> rejected : rejectedRelationships ) // { // cycleParticipants.add( rejected.getDeclaring() // .asProjectVersionRef() ); // cycleParticipants.add( rejected.getTarget() // .asProjectVersionRef() ); // } // } private void markMissing(final DiscoveryRunnable runnable, final Set<ProjectVersionRef> missing, final int pass) { final int index = runnable.getIndex(); final ProjectVersionRef originalRef = runnable.getTodo().getRef(); logger.debug("{}.{}. MISSING(1) += {}", pass, index, originalRef); missing.add(originalRef); final DiscoveryResult result = runnable.getResult(); if (result != null) { final ProjectVersionRef selectdRef = result.getSelectedRef(); if (!originalRef.equals(selectdRef)) { logger.debug("{}.{}. MISSING(2) += {}", pass, index, selectdRef); missing.add(selectdRef); } } } // private Set<ProjectVersionRef> loadExistingCycleParticipants( final EProjectNet net ) // { // final Set<ProjectVersionRef> participants = new HashSet<ProjectVersionRef>(); // final Set<EProjectCycle> cycles = net.getCycles(); // for ( final EProjectCycle cycle : cycles ) // { // participants.addAll( cycle.getAllParticipatingProjects() ); // } // // return participants; // } private List<DiscoveryTodo> loadInitialPending(final RelationshipGraph graph, final Set<ProjectVersionRef> seen) { logger.info("Using root-level mutator: {}", graph.getMutator()); final Set<ProjectVersionRef> initialIncomplete = graph.getIncompleteSubgraphs(); logger.info("Finding paths to:\n {} \n\nfrom:\n {}\n\n", new JoinString("\n ", initialIncomplete), new JoinString("\n ", graph.getRoots())); if (initialIncomplete == null || initialIncomplete.isEmpty()) { return new ArrayList<DiscoveryTodo>(); } final Map<GraphPath<?>, GraphPathInfo> pathMap = graph.getPathMapTargeting(initialIncomplete); if (pathMap == null || pathMap.isEmpty()) { return new ArrayList<DiscoveryTodo>(); } final Map<ProjectVersionRef, DiscoveryTodo> initialPending = new HashMap<ProjectVersionRef, DiscoveryTodo>(); for (final Entry<GraphPath<?>, GraphPathInfo> entry : pathMap.entrySet()) { final GraphPath<?> path = entry.getKey(); final GraphPathInfo pathInfo = entry.getValue(); final List<ProjectVersionRef> pathRefs = graph.getPathRefs(path); final ProjectVersionRef ref = pathRefs.remove(pathRefs.size() - 1); if (!pathRefs.isEmpty()) { logger.info("Already seen += {}", new JoinString(", ", pathRefs)); seen.addAll(pathRefs); } DiscoveryTodo todo = initialPending.get(ref); if (todo == null) { todo = new DiscoveryTodo(ref, path, pathInfo, graph); initialPending.put(ref, todo); logger.info("INIT-DISCOVER += {}", ref); } else { todo.addParentPath(path, pathInfo); } } logger.info("[INIT] {} subgraphs pending discovery", initialPending.size()); logger.debug("Initial pending:\n {}\n", new JoinString("\n ", initialPending.keySet())); return new ArrayList<DiscoveryTodo>(initialPending.values()); } }