Search in sources :

Example 1 with CrawlCommandEvent

use of won.bot.framework.eventbot.event.impl.crawl.CrawlCommandEvent in project webofneeds by researchstudio-sat.

the class CrawlAction method doRun.

@Override
protected void doRun(Event event, EventListener executingListener) throws Exception {
    if (!(event instanceof CrawlCommandEvent))
        return;
    CrawlCommandEvent crawlCommandEvent = (CrawlCommandEvent) event;
    EventListenerContext ctx = getEventListenerContext();
    logger.debug("starting crawl for {}", crawlCommandEvent.getStartURI());
    StopWatch stopWatch = new StopWatch();
    stopWatch.start();
    Dataset crawledData;
    try {
        crawledData = ctx.getLinkedDataSource().getDataForResourceWithPropertyPath(crawlCommandEvent.getStartURI(), crawlCommandEvent.getAtomURI(), crawlCommandEvent.getPropertyPaths(), crawlCommandEvent.getGetMaxRequest(), crawlCommandEvent.getMaxDepth());
    } catch (Exception e) {
        logger.debug("caught exception during crawl for {}", crawlCommandEvent.getStartURI(), e);
        ctx.getEventBus().publish(new CrawlCommandFailureEvent(crawlCommandEvent, "Could not crawl " + crawlCommandEvent.getStartURI() + " with WebID " + crawlCommandEvent.getAtomURI() + ": caught " + e));
        return;
    }
    stopWatch.stop();
    logger.debug("finished crawl for {} in {} millis", crawlCommandEvent.getStartURI(), stopWatch.getTotalTimeMillis());
    ctx.getEventBus().publish(new CrawlCommandSuccessEvent(crawlCommandEvent, crawledData, "Finished crawling " + crawlCommandEvent.getStartURI() + " in " + stopWatch.getTotalTimeMillis()));
}
Also used : EventListenerContext(won.bot.framework.eventbot.EventListenerContext) CrawlCommandEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandEvent) CrawlCommandFailureEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandFailureEvent) Dataset(org.apache.jena.query.Dataset) CrawlCommandSuccessEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandSuccessEvent) StopWatch(org.springframework.util.StopWatch)

Example 2 with CrawlCommandEvent

use of won.bot.framework.eventbot.event.impl.crawl.CrawlCommandEvent in project webofneeds by researchstudio-sat.

the class CrawlConnectionDataBehaviour method onActivate.

@Override
protected void onActivate(Optional<Object> message) {
    logger.debug("activating crawling connection data for connection {}", command.getConnectionURI());
    logger.debug("will deactivate automatically after " + abortTimeout);
    LinkedDataSource linkedDataSource = context.getLinkedDataSource();
    if (linkedDataSource instanceof CachingLinkedDataSource) {
        URI toInvalidate = WonRelativeUriHelper.createMessageContainerURIForConnection(command.getConnectionURI());
        ((CachingLinkedDataSource) linkedDataSource).invalidate(toInvalidate);
        ((CachingLinkedDataSource) linkedDataSource).invalidate(toInvalidate, command.getAtomURI());
    }
    context.getTaskScheduler().schedule(this::deactivate, new Date(System.currentTimeMillis() + abortTimeout.toMillis()));
    List<Path> propertyPaths = new ArrayList<>();
    PrefixMapping pmap = new PrefixMappingImpl();
    pmap.withDefaultMappings(PrefixMapping.Standard);
    pmap.setNsPrefix("won", WON.getURI());
    pmap.setNsPrefix("msg", WONMSG.getURI());
    propertyPaths.add(PathParser.parse("won:messageContainer", pmap));
    propertyPaths.add(PathParser.parse("won:messageContainer/rdfs:member", pmap));
    CrawlCommandEvent crawlAtomCommandEvent = new CrawlCommandEvent(command.getAtomURI(), command.getAtomURI(), propertyPaths, 10000, 5);
    propertyPaths = new ArrayList<>();
    propertyPaths.add(PathParser.parse("won:messageContainer", pmap));
    propertyPaths.add(PathParser.parse("won:messageContainer/rdfs:member", pmap));
    propertyPaths.add(PathParser.parse("won:messageContainer/rdfs:member/msg:correspondingRemoteMessage", pmap));
    propertyPaths.add(PathParser.parse("won:targetAtom", pmap));
    propertyPaths.add(PathParser.parse("won:targetConnection", pmap));
    CrawlCommandEvent crawlConnectionCommandEvent = new CrawlCommandEvent(command.getAtomURI(), command.getConnectionURI(), propertyPaths, 10000, 5);
    Dataset crawledData = DatasetFactory.createGeneral();
    // add crawlcommand listener
    this.subscribeWithAutoCleanup(CrawlCommandEvent.class, new ActionOnEventListener(context, new OrFilter(new SameEventFilter(crawlAtomCommandEvent), new SameEventFilter(crawlConnectionCommandEvent)), new CrawlAction(context)));
    // when the first crawl succeeds, start the second
    this.subscribeWithAutoCleanup(CrawlCommandSuccessEvent.class, new ActionOnEventListener(context, new CommandResultFilter(crawlAtomCommandEvent), new BaseEventBotAction(context) {

        @Override
        protected void doRun(Event event, EventListener executingListener) throws Exception {
            logger.debug("finished crawling atom data. ");
            Dataset dataset = ((CrawlCommandSuccessEvent) event).getCrawledData();
            RdfUtils.addDatasetToDataset(crawledData, dataset);
            // now crawl connection data
            context.getEventBus().publish(crawlConnectionCommandEvent);
        }
    }));
    // when we're done crawling, validate:
    this.subscribeWithAutoCleanup(CrawlCommandSuccessEvent.class, new ActionOnEventListener(context, new CommandResultFilter(crawlConnectionCommandEvent), new BaseEventBotAction(context) {

        @Override
        protected void doRun(Event event, EventListener executingListener) throws Exception {
            logger.debug("finished crawling atom data for connection {}", command.getConnectionURI());
            Dataset dataset = ((CrawlCommandSuccessEvent) event).getCrawledData();
            RdfUtils.addDatasetToDataset(crawledData, dataset);
            context.getEventBus().publish(new CrawlConnectionCommandSuccessEvent(command, crawledData));
            deactivate();
        }
    }));
    // when something goes wrong, abort
    this.subscribeWithAutoCleanup(CrawlCommandFailureEvent.class, new ActionOnFirstEventListener(context, new OrFilter(new CommandResultFilter(crawlConnectionCommandEvent), new CommandResultFilter(crawlAtomCommandEvent)), new BaseEventBotAction(context) {

        @Override
        protected void doRun(Event event, EventListener executingListener) throws Exception {
            CrawlCommandFailureEvent failureEvent = (CrawlCommandFailureEvent) event;
            logger.debug("crawling failed for connection {}, message: {}", command.getConnectionURI(), failureEvent.getMessage());
            context.getEventBus().publish(new CrawlConnectionCommandFailureEvent(failureEvent.getMessage(), command));
            deactivate();
        }
    }));
    // start crawling the atom - connection will be crawled when atom crawling is
    // done
    context.getEventBus().publish(crawlAtomCommandEvent);
}
Also used : SameEventFilter(won.bot.framework.eventbot.filter.impl.SameEventFilter) CrawlAction(won.bot.framework.eventbot.action.impl.crawl.CrawlAction) CrawlConnectionCommandSuccessEvent(won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandSuccessEvent) ArrayList(java.util.ArrayList) CrawlCommandSuccessEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandSuccessEvent) CommandResultFilter(won.bot.framework.eventbot.filter.impl.CommandResultFilter) LinkedDataSource(won.protocol.util.linkeddata.LinkedDataSource) CachingLinkedDataSource(won.protocol.util.linkeddata.CachingLinkedDataSource) URI(java.net.URI) CrawlCommandEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandEvent) ActionOnFirstEventListener(won.bot.framework.eventbot.listener.impl.ActionOnFirstEventListener) ActionOnEventListener(won.bot.framework.eventbot.listener.impl.ActionOnEventListener) EventListener(won.bot.framework.eventbot.listener.EventListener) Path(org.apache.jena.sparql.path.Path) CrawlCommandFailureEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandFailureEvent) Dataset(org.apache.jena.query.Dataset) OrFilter(won.bot.framework.eventbot.filter.impl.OrFilter) CrawlConnectionCommandFailureEvent(won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandFailureEvent) Date(java.util.Date) PrefixMapping(org.apache.jena.shared.PrefixMapping) CachingLinkedDataSource(won.protocol.util.linkeddata.CachingLinkedDataSource) BaseEventBotAction(won.bot.framework.eventbot.action.BaseEventBotAction) CrawlCommandEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandEvent) CrawlConnectionCommandFailureEvent(won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandFailureEvent) CrawlCommandFailureEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandFailureEvent) CommandResultEvent(won.bot.framework.eventbot.event.impl.cmd.CommandResultEvent) CrawlCommandSuccessEvent(won.bot.framework.eventbot.event.impl.crawl.CrawlCommandSuccessEvent) CrawlConnectionCommandSuccessEvent(won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandSuccessEvent) Event(won.bot.framework.eventbot.event.Event) CrawlConnectionCommandEvent(won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandEvent) ActionOnEventListener(won.bot.framework.eventbot.listener.impl.ActionOnEventListener) PrefixMappingImpl(org.apache.jena.shared.impl.PrefixMappingImpl) ActionOnFirstEventListener(won.bot.framework.eventbot.listener.impl.ActionOnFirstEventListener)

Aggregations

Dataset (org.apache.jena.query.Dataset)2 CrawlCommandEvent (won.bot.framework.eventbot.event.impl.crawl.CrawlCommandEvent)2 CrawlCommandFailureEvent (won.bot.framework.eventbot.event.impl.crawl.CrawlCommandFailureEvent)2 CrawlCommandSuccessEvent (won.bot.framework.eventbot.event.impl.crawl.CrawlCommandSuccessEvent)2 URI (java.net.URI)1 ArrayList (java.util.ArrayList)1 Date (java.util.Date)1 PrefixMapping (org.apache.jena.shared.PrefixMapping)1 PrefixMappingImpl (org.apache.jena.shared.impl.PrefixMappingImpl)1 Path (org.apache.jena.sparql.path.Path)1 StopWatch (org.springframework.util.StopWatch)1 EventListenerContext (won.bot.framework.eventbot.EventListenerContext)1 BaseEventBotAction (won.bot.framework.eventbot.action.BaseEventBotAction)1 CrawlAction (won.bot.framework.eventbot.action.impl.crawl.CrawlAction)1 Event (won.bot.framework.eventbot.event.Event)1 CommandResultEvent (won.bot.framework.eventbot.event.impl.cmd.CommandResultEvent)1 CrawlConnectionCommandEvent (won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandEvent)1 CrawlConnectionCommandFailureEvent (won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandFailureEvent)1 CrawlConnectionCommandSuccessEvent (won.bot.framework.eventbot.event.impl.crawlconnection.CrawlConnectionCommandSuccessEvent)1 CommandResultFilter (won.bot.framework.eventbot.filter.impl.CommandResultFilter)1