Constant Field Values
Contents
com.digitalpebble.*
-
com.digitalpebble.stormcrawler.Constants Modifier and Type Constant Field Value public static final String
AllowRedirParamName
"redirections.allowed"
public static final String
defaultFetchIntervalParamName
"fetchInterval.default"
public static final String
DELETION_STREAM_NAME
"deletion"
public static final String
errorFetchIntervalParamName
"fetchInterval.error"
public static final String
fetchErrorCountParamName
"fetch.error.count"
public static final String
fetchErrorFetchIntervalParamName
"fetchInterval.fetch.error"
public static final int
MAX_ARRAY_SIZE
2147483639
public static final String
PARTITION_MODE_DOMAIN
"byDomain"
public static final String
PARTITION_MODE_HOST
"byHost"
public static final String
PARTITION_MODE_IP
"byIP"
public static final String
PARTITION_MODEParamName
"partition.url.mode"
public static final String
STATUS_ERROR_CAUSE
"error.cause"
public static final String
STATUS_ERROR_MESSAGE
"error.message"
public static final String
STATUS_ERROR_SOURCE
"error.source"
public static final String
StatusStreamName
"status"
-
com.digitalpebble.stormcrawler.bolt.FeedParserBolt Modifier and Type Constant Field Value public static final String
isFeedKey
"isFeed"
-
com.digitalpebble.stormcrawler.bolt.FetcherBolt Modifier and Type Constant Field Value public static final String
QUEUED_TIMEOUT_PARAM_KEY
"fetcher.timeout.queue"
-
com.digitalpebble.stormcrawler.bolt.JSoupParserBolt Modifier and Type Constant Field Value public static final String
ANCHORS_KEY_NAME
"anchors"
-
com.digitalpebble.stormcrawler.bolt.SimpleFetcherBolt Modifier and Type Constant Field Value public static final String
QUEUE_MODE_DOMAIN
"byDomain"
public static final String
QUEUE_MODE_HOST
"byHost"
public static final String
QUEUE_MODE_IP
"byIP"
public static final String
THROTTLE_STREAM
"throttle"
-
com.digitalpebble.stormcrawler.bolt.SiteMapParserBolt Modifier and Type Constant Field Value public static final String
foundSitemapKey
"foundSitemap"
public static final String
isSitemapKey
"isSitemap"
-
com.digitalpebble.stormcrawler.indexing.AbstractIndexerBolt Modifier and Type Constant Field Value public static final String
canonicalMetadataParamName
"indexer.canonical.name"
public static final String
ignoreEmptyFieldValueParamName
"indexer.ignore.empty.fields"
public static final String
metadata2fieldParamName
"indexer.md.mapping"
public static final String
metadataFilterParamName
"indexer.md.filter"
public static final String
textFieldParamName
"indexer.text.fieldname"
public static final String
textLengthParamName
"indexer.text.maxlength"
public static final String
urlFieldParamName
"indexer.url.fieldname"
-
com.digitalpebble.stormcrawler.parse.TextExtractor Modifier and Type Constant Field Value public static final String
EXCLUDE_PARAM_NAME
"textextractor.exclude.tags"
public static final String
INCLUDE_PARAM_NAME
"textextractor.include.pattern"
public static final String
NO_TEXT_PARAM_NAME
"textextractor.no.text"
public static final String
TEXT_MAX_TEXT_PARAM_NAME
"textextractor.skip.after"
-
com.digitalpebble.stormcrawler.persistence.AbstractQueryingSpout Modifier and Type Constant Field Value protected static final String
resetFetchDateParamName
"spout.reset.fetchdate.after"
protected static final String
StatusMaxDelayParamName
"spout.max.delay.queries"
protected static final String
StatusMinDelayParamName
"spout.min.delay.queries"
protected static final String
StatusTTLPurgatory
"spout.ttl.purgatory"
-
com.digitalpebble.stormcrawler.persistence.AbstractStatusUpdaterBolt Modifier and Type Constant Field Value public static final String
AS_IS_NEXTFETCHDATE_METADATA
"status.store.as.is.with.nextfetchdate"
-
com.digitalpebble.stormcrawler.persistence.AdaptiveScheduler Modifier and Type Constant Field Value public static final String
FETCH_INTERVAL_KEY
"fetchInterval"
public static final String
INTERVAL_DEC_RATE
"scheduler.adaptive.fetchInterval.rate.decr"
public static final String
INTERVAL_INC_RATE
"scheduler.adaptive.fetchInterval.rate.incr"
public static final String
INTERVAL_MAX
"scheduler.adaptive.fetchInterval.max"
public static final String
INTERVAL_MIN
"scheduler.adaptive.fetchInterval.min"
public static final String
SET_LAST_MODIFIED
"scheduler.adaptive.setLastModified"
public static final String
SIGNATURE_KEY
"signature"
public static final String
SIGNATURE_MODIFIED_KEY
"signatureChangeDate"
public static final String
SIGNATURE_OLD_KEY
"signatureOld"
-
com.digitalpebble.stormcrawler.persistence.DefaultScheduler Modifier and Type Constant Field Value public static final String
DELAY_METADATA
"scheduler.delay.mins"
-
com.digitalpebble.stormcrawler.persistence.Scheduler Modifier and Type Constant Field Value public static final String
schedulerClassParamName
"scheduler.class"
-
com.digitalpebble.stormcrawler.persistence.urlbuffer.SchedulingURLBuffer Modifier and Type Constant Field Value public static final String
MAXTIMEPARAM
"priority.buffer.max.time.msec"
-
com.digitalpebble.stormcrawler.persistence.urlbuffer.URLBuffer Modifier and Type Constant Field Value public static final String
bufferClassParamName
"urlbuffer.class"
-
com.digitalpebble.stormcrawler.protocol.AbstractHttpProtocol Modifier and Type Constant Field Value protected static final String
RESPONSE_COOKIES_HEADER
"set-cookie"
protected static final String
SET_HEADER_BY_REQUEST
"set-header"
-
com.digitalpebble.stormcrawler.protocol.HttpHeaders Modifier and Type Constant Field Value public static final String
CONTENT_DISPOSITION
"content-disposition"
public static final String
CONTENT_ENCODING
"content-encoding"
public static final String
CONTENT_LANGUAGE
"content-language"
public static final String
CONTENT_LENGTH
"content-length"
public static final String
CONTENT_LOCATION
"content-location"
public static final String
CONTENT_MD5
"content-md5"
public static final String
CONTENT_TYPE
"content-type"
public static final String
LAST_MODIFIED
"last-modified"
public static final String
LOCATION
"location"
public static final String
TRANSFER_ENCODING
"transfer-encoding"
-
com.digitalpebble.stormcrawler.protocol.ProtocolResponse Modifier and Type Constant Field Value public static final String
PROTOCOL_MD_PREFIX_PARAM
"protocol.md.prefix"
public static final String
PROTOCOL_VERSIONS_KEY
"_protocol_versions_"
public static final String
REQUEST_HEADERS_KEY
"_request.headers_"
public static final String
REQUEST_TIME_KEY
"_request.time_"
public static final String
RESPONSE_HEADERS_KEY
"_response.headers_"
public static final String
RESPONSE_IP_KEY
"_response.ip_"
public static final String
TRIMMED_RESPONSE_KEY
"http.trimmed"
public static final String
TRIMMED_RESPONSE_REASON_KEY
"http.trimmed.reason"
-
com.digitalpebble.stormcrawler.protocol.RobotRulesParser Modifier and Type Constant Field Value public static final String
cacheConfigParamName
"robots.cache.spec"
public static final String
errorcacheConfigParamName
"robots.error.cache.spec"
-
com.digitalpebble.stormcrawler.protocol.selenium.DelegatorRemoteDriverProtocol Modifier and Type Constant Field Value public static final String
PROTOCOL_IMPL_CONFIG
"selenium.delegated.protocol"
public static final String
USE_SELENIUM_KEY
"protocol.use.selenium"
-
com.digitalpebble.stormcrawler.spout.FileSpout Modifier and Type Constant Field Value public static final int
BATCH_SIZE
10000
-
com.digitalpebble.stormcrawler.util.MetadataTransfer Modifier and Type Constant Field Value public static final String
depthKeyName
"depth"
public static final String
maxDepthKeyName
"max.depth"
public static final String
metadataPersistParamName
"metadata.persist"
public static final String
metadataTransferClassParamName
"metadata.transfer.class"
public static final String
metadataTransferParamName
"metadata.transfer"
public static final String
trackDepthParamName
"metadata.track.depth"
public static final String
trackPathParamName
"metadata.track.path"
public static final String
urlPathKeyName
"url.path"
-
com.digitalpebble.stormcrawler.util.RobotsTags Modifier and Type Constant Field Value public static final String
ROBOTS_NO_CACHE
"robots.noCache"
public static final String
ROBOTS_NO_FOLLOW
"robots.noFollow"
public static final String
ROBOTS_NO_FOLLOW_STRICT
"robots.noFollow.strict"
public static final String
ROBOTS_NO_INDEX
"robots.noIndex"