-
- All Implemented Interfaces:
-
ai.platon.pulsar.common.StartStopRunnable,ai.platon.pulsar.crawl.CrawlLoop
public class StreamingCrawlLoop extends AbstractCrawlLoop
-
-
Field Summary
Fields Modifier and Type Field Description private final GlobalCacheglobalCacheprivate final BooleanisRunningprivate DefaultCrawlEventHandlercrawlEventHandlerprivate final UrlFeederurlFeederprivate StreamingCrawler<UrlAware>crawlerprivate final GlobalCacheFactoryglobalCacheFactoryprivate final Integeridprivate LoadOptionsdefaultOptionsprivate final List<PriorityDataCollector<UrlAware>>collectorsprivate final Stringabstractprivate final Stringreportprivate final Stringnameprivate final ImmutableConfigconfig
-
Constructor Summary
Constructors Constructor Description StreamingCrawlLoop(GlobalCacheFactory globalCacheFactory, ImmutableConfig unmodifiedConfig, String name)
-
Method Summary
Modifier and Type Method Description final GlobalCachegetGlobalCache()final BooleangetIsRunning()final DefaultCrawlEventHandlergetCrawlEventHandler()final UnitsetCrawlEventHandler(DefaultCrawlEventHandler crawlEventHandler)UrlFeedergetUrlFeeder()The fetch iterable from which all fetch tasks are taken StreamingCrawler<UrlAware>getCrawler()UnitsetCrawler(StreamingCrawler<UrlAware> crawler)final GlobalCacheFactorygetGlobalCacheFactory()The global cache IntegergetId()LoadOptionsgetDefaultOptions()Data collector lower capacity UnitsetDefaultOptions(LoadOptions defaultOptions)Data collector lower capacity List<PriorityDataCollector<UrlAware>>getCollectors()StringgetAbstract()StringgetReport()StringgetName()ImmutableConfiggetConfig()Unitstart()Unitstop()Unitawait()-
-
Constructor Detail
-
StreamingCrawlLoop
StreamingCrawlLoop(GlobalCacheFactory globalCacheFactory, ImmutableConfig unmodifiedConfig, String name)
-
-
Method Detail
-
getGlobalCache
final GlobalCache getGlobalCache()
-
getIsRunning
final Boolean getIsRunning()
-
getCrawlEventHandler
final DefaultCrawlEventHandler getCrawlEventHandler()
-
setCrawlEventHandler
final Unit setCrawlEventHandler(DefaultCrawlEventHandler crawlEventHandler)
-
getUrlFeeder
UrlFeeder getUrlFeeder()
The fetch iterable from which all fetch tasks are taken
-
getCrawler
StreamingCrawler<UrlAware> getCrawler()
-
setCrawler
Unit setCrawler(StreamingCrawler<UrlAware> crawler)
-
getGlobalCacheFactory
final GlobalCacheFactory getGlobalCacheFactory()
The global cache
-
getDefaultOptions
LoadOptions getDefaultOptions()
Data collector lower capacity
-
setDefaultOptions
Unit setDefaultOptions(LoadOptions defaultOptions)
Data collector lower capacity
-
getCollectors
List<PriorityDataCollector<UrlAware>> getCollectors()
-
getAbstract
String getAbstract()
-
getConfig
ImmutableConfig getConfig()
-
start
@Synchronized() Unit start()
-
stop
@Synchronized() Unit stop()
-
-
-
-