mirror of
https://github.com/internetarchive/heritrix3.git
synced 2026-07-12 18:47:09 +00:00
164 lines
9.0 KiB
Plaintext
164 lines
9.0 KiB
Plaintext
root=object:java.util.HashMap
|
|
|
|
root.loggerModule=object:org.archive.crawler.framework.CrawlerLoggerModule
|
|
root.loggerModule.parent=reference:manager
|
|
root.loggerModule.dir=string:logs
|
|
|
|
root.seeds=object:org.archive.crawler.scope.SeedModule
|
|
root.seeds.directory=reference:manager
|
|
|
|
root.scope=object:org.archive.processors.deciderules.DecideRuleSequence
|
|
root.scope.rules=object:java.util.ArrayList
|
|
root.scope.rules.0=object:org.archive.processors.deciderules.RejectDecideRule
|
|
root.scope.rules.1=object:org.archive.crawler.deciderules.SurtPrefixedDecideRule
|
|
root.scope.rules.1.seeds=reference:root.seeds
|
|
root.scope.rules.1.surts-source-file=string:
|
|
root.scope.rules.1.seeds-as-surt-prefixes=boolean:true
|
|
root.scope.rules.1.surts-dump-file=string:
|
|
root.scope.rules.2=object:org.archive.processors.deciderules.TooManyHopsDecideRule
|
|
root.scope.rules.2.max-hops=int:20
|
|
root.scope.rules.3=object:org.archive.processors.deciderules.TransclusionDecideRule
|
|
root.scope.rules.3.max-trans-hops=int:3
|
|
root.scope.rules.4=object:org.archive.processors.deciderules.TooManyPathSegmentsDecideRule
|
|
root.scope.rules.4.max-path-depth=int:20
|
|
root.scope.rules.5=object:org.archive.processors.deciderules.PrerequisiteAcceptDecideRule
|
|
|
|
root.uriUniqFilter=object:org.archive.crawler.util.BdbUriUniqFilter
|
|
root.uriUniqFilter.bdb=reference:manager.bdb
|
|
|
|
root.uri-canonicalization-rules=object:java.util.ArrayList
|
|
root.uri-canonicalization-rules.0=object:org.archive.crawler.url.canonicalize.LowercaseRule
|
|
root.uri-canonicalization-rules.1=object:org.archive.crawler.url.canonicalize.StripUserinfoRule
|
|
root.uri-canonicalization-rules.2=object:org.archive.crawler.url.canonicalize.StripWWWNRule
|
|
root.uri-canonicalization-rules.3=object:org.archive.crawler.url.canonicalize.StripSessionIDs
|
|
root.uri-canonicalization-rules.4=object:org.archive.crawler.url.canonicalize.StripSessionCFIDs
|
|
root.uri-canonicalization-rules.5=object:org.archive.crawler.url.canonicalize.FixupQueryStr
|
|
|
|
root.order=object:org.archive.crawler.datamodel.CrawlOrder
|
|
root.order.disk-path=string:.
|
|
root.order.checkpoints-path=string:checkpoints
|
|
root.order.state-path=string:state
|
|
root.order.scratch-path=string:scratch
|
|
root.order.max-bytes-download=long:0
|
|
root.order.max-document-download=long:0
|
|
root.order.max-time-sec=long:0
|
|
root.order.max-toe-threads=int:1
|
|
root.order.recorder-out-buffer-bytes=int:4096
|
|
root.order.recorder-in-buffer-bytes=int:65536
|
|
root.order.http-headers=object:java.util.HashMap
|
|
root.order.http-headers.user-agent=string:Mozilla/5.0 (compatible; heritrix/11 +http://www.archive.org)
|
|
root.order.http-headers.from=string:info@archive.org
|
|
root.order.robots-honoring-policy=object:org.archive.processors.util.RobotsHonoringPolicy
|
|
root.order.robots-honoring-policy.masquerade=boolean:false
|
|
root.order.robots-honoring-policy.custom-robots=string:
|
|
root.order.robots-honoring-policy.user-agents=object:java.util.ArrayList
|
|
root.order.recover-path=string:
|
|
root.order.recover-retain-failures=boolean:false
|
|
|
|
root.queue-assignment-policy=object:org.archive.crawler.frontier.HostnameQueueAssignmentPolicy
|
|
|
|
root.controller=object:org.archive.crawler.framework.CrawlController
|
|
root.controller.sheet-manager=reference:manager
|
|
root.controller.logger-module=reference:root.loggerModule
|
|
root.controller.order=reference:root.order
|
|
|
|
root.controller.credential-store=object:org.archive.processors.credential.CredentialStore
|
|
|
|
root.controller.frontier=object:org.archive.crawler.frontier.BdbFrontier
|
|
root.controller.frontier.seeds=reference:root.seeds
|
|
root.controller.frontier.bdb=reference:bdb
|
|
root.controller.frontier.logger-module=reference:root.loggerModule
|
|
root.controller.frontier.controller=reference:root.controller
|
|
root.controller.frontier.queue-assignment-policy=reference:root.queue-assignment-policy
|
|
root.controller.frontier.uri-uniq-filter=reference:root.uriUniqFilter
|
|
root.controller.frontier.rules=reference:root.uri-canonicalization-rules
|
|
root.controller.frontier.manager=reference:manager
|
|
root.controller.frontier.delay-factor=float:0.0
|
|
root.controller.frontier.max-delay-ms=int:0
|
|
root.controller.frontier.min-delay-ms=int:0
|
|
root.controller.frontier.max-retries=int:10
|
|
root.controller.frontier.retry-delay-seconds=long:90
|
|
root.controller.frontier.preference-embed-hops=int:1
|
|
root.controller.frontier.total-bandwidth-usage-kb-sec=int:0
|
|
root.controller.frontier.max-per-host-bandwidth-usage-kb-sec=int:0
|
|
root.controller.frontier.force-queue-assignment=string:
|
|
root.controller.frontier.pause-at-finish=boolean:false
|
|
root.controller.frontier.hold-queues=boolean:true
|
|
root.controller.frontier.balance-replenish-amount=int:3000
|
|
root.controller.frontier.queue-total-budget=long:-1
|
|
root.controller.frontier.cost-policy=object:org.archive.crawler.frontier.ZeroCostAssignmentPolicy
|
|
|
|
root.controller.server-cache=object:org.archive.crawler.framework.CrawlerServerCache
|
|
root.controller.server-cache.bdb=reference:bdb
|
|
|
|
root.controller.processors=object:java.util.LinkedHashMap
|
|
|
|
root.controller.processors.Preselector=object:org.archive.crawler.prefetch.Preselector
|
|
root.controller.processors.Preselector.scope=reference:root.scope
|
|
root.controller.processors.Preselector.logger-module=reference:root.loggerModule
|
|
root.controller.processors.Preselector.recheck-scope=boolean:true
|
|
root.controller.processors.Preselector.block-all=boolean:false
|
|
root.controller.processors.Preselector.block-by-regexp=string:
|
|
|
|
root.controller.processors.Preprocessor=object:org.archive.crawler.prefetch.PreconditionEnforcer
|
|
root.controller.processors.Preprocessor.controller=reference:root.controller
|
|
root.controller.processors.Preprocessor.ip-validity-duration-seconds=int:21600
|
|
root.controller.processors.Preprocessor.robots-validity-duration-seconds=int:86400
|
|
|
|
root.controller.processors.DNS=object:org.archive.processors.fetcher.FetchDNS
|
|
root.controller.processors.DNS.server-cache=reference:root.controller.server-cache
|
|
root.controller.processors.DNS.accept-non-dns-resolves=boolean:true
|
|
|
|
root.controller.processors.HTTP=object:org.archive.processors.fetcher.FetchHTTP
|
|
root.controller.processors.HTTP.server-cache=reference:root.controller.server-cache
|
|
root.controller.processors.HTTP.credential-store=reference:root.controller.credential-store
|
|
root.controller.processors.HTTP.midfetch-rules=object:org.archive.processors.deciderules.DecideRuleSequence
|
|
root.controller.processors.HTTP.timeout-seconds=int:1200
|
|
root.controller.processors.HTTP.sotimeout-ms=int:20000
|
|
root.controller.processors.HTTP.max-length-bytes=long:0
|
|
root.controller.processors.HTTP.trust-level=string:open
|
|
root.controller.processors.HTTP.accept-headers=object:java.util.ArrayList
|
|
root.controller.processors.HTTP.http-proxy-host=string:
|
|
root.controller.processors.HTTP.http-proxy-port=int:0
|
|
root.controller.processors.HTTP.default-encoding=string:ISO-8859-1
|
|
root.controller.processors.HTTP.send-referer=boolean:true
|
|
root.controller.processors.HTTP.send-range=boolean:false
|
|
root.controller.processors.HTTP.send-connection-close=boolean:true
|
|
|
|
root.controller.processors.ExtractorHTTP=object:org.archive.processors.extractor.ExtractorHTTP
|
|
root.controller.processors.ExtractorHTML=object:org.archive.processors.extractor.ExtractorHTML
|
|
root.controller.processors.ExtractorCSS=object:org.archive.processors.extractor.ExtractorCSS
|
|
root.controller.processors.ExtractorJS=object:org.archive.processors.extractor.ExtractorJS
|
|
root.controller.processors.ExtractorSWF=object:org.archive.processors.extractor.ExtractorSWF
|
|
|
|
root.controller.processors.Archiver=object:org.archive.crawler.writer.ARCWriterProcessor
|
|
root.controller.processors.Archiver.server-cache=reference:root.controller.server-cache
|
|
root.controller.processors.Archiver.directory=reference:manager
|
|
root.controller.processors.Archiver.compress=boolean:true
|
|
root.controller.processors.Archiver.prefix=string:IAH
|
|
root.controller.processors.Archiver.suffix=string:archive.org
|
|
root.controller.processors.Archiver.max-size-bytes=long:100000000
|
|
root.controller.processors.Archiver.path=object:java.util.ArrayList
|
|
root.controller.processors.Archiver.path.0=string:arcs
|
|
root.controller.processors.Archiver.pool-max-active=int:5
|
|
root.controller.processors.Archiver.pool-max-wait=int:300000
|
|
root.controller.processors.Archiver.total-bytes-to-write=long:0
|
|
|
|
root.controller.processors.Updater=object:org.archive.crawler.postprocessor.CrawlStateUpdater
|
|
root.controller.processors.Updater.controller=reference:root.controller
|
|
|
|
root.controller.processors.LinksScoper=object:org.archive.crawler.postprocessor.LinksScoper
|
|
root.controller.processors.LinksScoper.scope=reference:root.scope
|
|
root.controller.processors.LinksScoper.logger-module=reference:root.loggerModule
|
|
root.controller.processors.LinksScoper.seed-redirects-new-seeds=boolean:true
|
|
root.controller.processors.LinksScoper.override-logger=boolean:false
|
|
|
|
root.controller.processors.Scheduler=object:org.archive.crawler.postprocessor.FrontierScheduler
|
|
root.controller.processors.Scheduler.frontier=reference:root.controller.frontier
|
|
|
|
root.controller.loggers=object:java.util.ArrayList
|
|
root.controller.loggers.0=object:org.archive.crawler.framework.StatisticsTracker
|
|
root.controller.loggers.0.seeds=reference:root.seeds
|
|
root.controller.loggers.0.bdb=reference:manager.bdb
|
|
root.controller.loggers.0.controller=reference:root.controller
|