Constant Field Values

Contents

• org.apache.*

org.apache.*

org.apache.nutch.analysis.NutchAnalysisConstants

public static final int	ACRONYM	2
public static final int	APOSTROPHE	14
public static final int	ATSIGN	13
public static final int	C_PLUS_PLUS	5
public static final int	C_SHARP	6
public static final int	CJK	18
public static final int	COLON	10
public static final int	DEFAULT	0
public static final int	DIGIT	19
public static final int	DOT	12
public static final int	EOF	0
public static final int	IRREGULAR_WORD	4
public static final int	LETTER	17
public static final int	MINUS	8
public static final int	PLUS	7
public static final int	QUOTE	9
public static final int	SIGRAM	3
public static final int	SLASH	11
public static final int	WHITE	15
public static final int	WORD	1
public static final int	WORD_PUNCT	16

org.apache.nutch.analysis.NutchDocumentAnalyzer

public static final int

INTER_ANCHOR_GAP

4

org.apache.nutch.clustering.carrot2.NutchInputComponent

public static final String	NUTCH_INPUT_HIT_DETAILS_ARRAY	"NUTCH_INPUT_HIT_DETAILS_ARRAY"
public static final String	NUTCH_INPUT_SUMMARIES_ARRAY	"NUTCH_INPUT_SUMMARIES_ARRAY"

org.apache.nutch.crawl.CrawlDatum

public static final String	FETCH_DIR_NAME	"crawl_fetch"
public static final String	GENERATE_DIR_NAME	"crawl_generate"
public static final String	PARSE_DIR_NAME	"crawl_parse"
public static final byte	STATUS_DB_FETCHED	2
public static final byte	STATUS_DB_GONE	3
public static final byte	STATUS_DB_MAX	31
public static final byte	STATUS_DB_NOTMODIFIED	6
public static final byte	STATUS_DB_REDIR_PERM	5
public static final byte	STATUS_DB_REDIR_TEMP	4
public static final byte	STATUS_DB_UNFETCHED	1
public static final byte	STATUS_FETCH_GONE	37
public static final byte	STATUS_FETCH_MAX	63
public static final byte	STATUS_FETCH_NOTMODIFIED	38
public static final byte	STATUS_FETCH_REDIR_PERM	36
public static final byte	STATUS_FETCH_REDIR_TEMP	35
public static final byte	STATUS_FETCH_RETRY	34
public static final byte	STATUS_FETCH_SUCCESS	33
public static final byte	STATUS_INJECTED	66
public static final byte	STATUS_LINKED	67
public static final byte	STATUS_PARSE_META	68
public static final byte	STATUS_SIGNATURE	65

org.apache.nutch.crawl.CrawlDb

public static final String	CRAWLDB_ADDITIONS_ALLOWED	"db.update.additions.allowed"
public static final String	CURRENT_NAME	"current"
public static final String	LOCK_NAME	".locked"

org.apache.nutch.crawl.CrawlDbFilter

public static final String	URL_FILTERING	"crawldb.url.filters"
public static final String	URL_NORMALIZING	"crawldb.url.normalizers"
public static final String	URL_NORMALIZING_SCOPE	"crawldb.url.normalizers.scope"

org.apache.nutch.crawl.CrawlDbReader

public static final int	CSV_FORMAT	1
public static final int	STD_FORMAT	0

org.apache.nutch.crawl.FetchSchedule

public static final int	SECONDS_PER_DAY	86400
public static final int	STATUS_MODIFIED	1
public static final int	STATUS_NOTMODIFIED	2
public static final int	STATUS_UNKNOWN	0

org.apache.nutch.crawl.Generator

public static final String	GENERATE_MAX_PER_HOST	"generate.max.per.host"
public static final String	GENERATE_MAX_PER_HOST_BY_IP	"generate.max.per.host.by.ip"
public static final String	GENERATE_UPDATE_CRAWLDB	"generate.update.crawldb"
public static final String	GENERATOR_COUNT_MODE	"generate.count.mode"
public static final String	GENERATOR_COUNT_VALUE_DOMAIN	"domain"
public static final String	GENERATOR_COUNT_VALUE_HOST	"host"
public static final String	GENERATOR_CUR_TIME	"generate.curTime"
public static final String	GENERATOR_DELAY	"crawl.gen.delay"
public static final String	GENERATOR_FILTER	"generate.filter"
public static final String	GENERATOR_MAX_COUNT	"generate.max.count"
public static final String	GENERATOR_MAX_NUM_SEGMENTS	"generate.max.num.segments"
public static final String	GENERATOR_MIN_SCORE	"generate.min.score"
public static final String	GENERATOR_NORMALISE	"generate.normalise"
public static final String	GENERATOR_TOP_N	"generate.topN"

org.apache.nutch.crawl.LinkDb

public static final String	CURRENT_NAME	"current"
public static final String	LOCK_NAME	".locked"

org.apache.nutch.crawl.LinkDbFilter

public static final String	URL_FILTERING	"linkdb.url.filters"
public static final String	URL_NORMALIZING	"linkdb.url.normalizer"
public static final String	URL_NORMALIZING_SCOPE	"linkdb.url.normalizer.scope"

org.apache.nutch.crawl.URLPartitioner

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_MODE_KEY	"partition.url.mode"

org.apache.nutch.fetcher.Fetcher

public static final String	CONTENT_REDIR	"content"
public static final int	PERM_REFRESH_TIME	5
public static final String	PROTOCOL_REDIR	"protocol"

org.apache.nutch.fetcher.OldFetcher

public static final String	CONTENT_REDIR	"content"
public static final int	PERM_REFRESH_TIME	5
public static final String	PROTOCOL_REDIR	"protocol"

org.apache.nutch.indexer.Indexer

public static final String

DONE_NAME

"index.done"

org.apache.nutch.indexer.IndexingFilters

public static final String

INDEXINGFILTER_ORDER

"indexingfilter.order"

org.apache.nutch.indexer.IndexMerger

public static final String

DONE_NAME

"merge.done"

org.apache.nutch.indexer.NutchDocument

public static final byte

VERSION

1

org.apache.nutch.indexer.field.FieldFilters

public static final String

FIELD_FILTER_ORDER

"field.filter.order"

org.apache.nutch.indexer.field.FieldIndexer

public static final String

DONE_NAME

"index.done"

org.apache.nutch.indexer.field.Fields

public static final String	ACTION	"action"
public static final String	ANCHOR	"anchor"
public static final String	BOOST	"boost"
public static final String	BOOSTFACTOR	"boostfactor"
public static final String	CACHE	"cache"
public static final String	COMPUTATION	"computation"
public static final String	CONTENT	"content"
public static final String	DIGEST	"digest"
public static final String	HOST	"host"
public static final String	ORIG_URL	"orig"
public static final String	SEG_URL	"segurl"
public static final String	SEGMENT	"segment"
public static final String	SITE	"site"
public static final String	TITLE	"title"
public static final String	TSTAMP	"tstamp"
public static final String	URL	"url"

org.apache.nutch.indexer.lucene.LuceneConstants

public static final String	FIELD_INDEX_PREFIX	"lucene.field.index."
public static final String	FIELD_PREFIX	"lucene.field."
public static final String	FIELD_STORE_PREFIX	"lucene.field.store."
public static final String	FIELD_VECTOR_PREFIX	"lucene.field.vector."
public static final String	INDEX_NO	"index.no"
public static final String	INDEX_NO_NORMS	"index.no_norms"
public static final String	INDEX_TOKENIZED	"index.tokenized"
public static final String	INDEX_UNTOKENIZED	"index.untokenized"
public static final String	LUCENE_PREFIX	"lucene."
public static final String	STORE_COMPRESS	"store.compress"
public static final String	STORE_NO	"store.no"
public static final String	STORE_YES	"store.yes"
public static final String	VECTOR_NO	"vector.no"
public static final String	VECTOR_OFFSET	"vector.offset"
public static final String	VECTOR_POS	"vector.pos"
public static final String	VECTOR_POS_OFFSET	"vector.pos_offset"
public static final String	VECTOR_YES	"vector.yes"

org.apache.nutch.indexer.solr.SolrConstants

public static final String	BOOST_FIELD	"boost"
public static final String	COMMIT_SIZE	"solr.commit.size"
public static final String	DIGEST_FIELD	"digest"
public static final String	ID_FIELD	"id"
public static final String	SERVER_URL	"solr.server.url"
public static final String	SOLR_PREFIX	"solr."
public static final String	TIMESTAMP_FIELD	"tstamp"
public static final String	URL_FIELD	"url"

org.apache.nutch.metadata.CreativeCommons

public static final String	LICENSE_LOCATION	"License-Location"
public static final String	LICENSE_URL	"License-Url"
public static final String	WORK_TYPE	"Work-Type"

org.apache.nutch.metadata.DublinCore

public static final String	CONTRIBUTOR	"contributor"
public static final String	COVERAGE	"coverage"
public static final String	CREATOR	"creator"
public static final String	DATE	"date"
public static final String	DESCRIPTION	"description"
public static final String	FORMAT	"format"
public static final String	IDENTIFIER	"identifier"
public static final String	LANGUAGE	"language"
public static final String	MODIFIED	"modified"
public static final String	PUBLISHER	"publisher"
public static final String	RELATION	"relation"
public static final String	RIGHTS	"rights"
public static final String	SOURCE	"source"
public static final String	SUBJECT	"subject"
public static final String	TITLE	"title"
public static final String	TYPE	"type"

org.apache.nutch.metadata.Feed

public static final String	FEED	"feed"
public static final String	FEED_AUTHOR	"author"
public static final String	FEED_PUBLISHED	"published"
public static final String	FEED_TAGS	"tag"
public static final String	FEED_UPDATED	"updated"

org.apache.nutch.metadata.HttpHeaders

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"

org.apache.nutch.metadata.Nutch

public static final String	CACHING_FORBIDDEN_ALL	"all"
public static final String	CACHING_FORBIDDEN_CONTENT	"content"
public static final String	CACHING_FORBIDDEN_KEY	"caching.forbidden"
public static final String	CACHING_FORBIDDEN_NONE	"none"
public static final String	CHAR_ENCODING_FOR_CONVERSION	"CharEncodingForConversion"
public static final String	FETCH_STATUS_KEY	"_fst_"
public static final String	FETCH_TIME_KEY	"_ftk_"
public static final String	GENERATE_TIME_KEY	"_ngt_"
public static final String	ORIGINAL_CHAR_ENCODING	"OriginalCharEncoding"
public static final String	PROTO_STATUS_KEY	"_pst_"
public static final String	REPR_URL_KEY	"_repr_"
public static final String	SCORE_KEY	"nutch.crawl.score"
public static final String	SEGMENT_NAME_KEY	"nutch.segment.name"
public static final String	SIGNATURE_KEY	"nutch.content.digest"

org.apache.nutch.metadata.Office

public static final String	APPLICATION_NAME	"Application-Name"
public static final String	AUTHOR	"Author"
public static final String	CHARACTER_COUNT	"Character Count"
public static final String	COMMENTS	"Comments"
public static final String	KEYWORDS	"Keywords"
public static final String	LAST_AUTHOR	"Last-Author"
public static final String	LAST_PRINTED	"Last-Printed"
public static final String	LAST_SAVED	"Last-Save-Date"
public static final String	PAGE_COUNT	"Page-Count"
public static final String	REVISION_NUMBER	"Revision-Number"
public static final String	TEMPLATE	"Template"
public static final String	WORD_COUNT	"Word-Count"

org.apache.nutch.microformats.reltag.RelTagParser

public static final String

REL_TAG

"Rel-Tag"

org.apache.nutch.net.URLFilters

public static final String

URLFILTER_ORDER

"urlfilter.order"

org.apache.nutch.net.URLNormalizers

public static final String	SCOPE_CRAWLDB	"crawldb"
public static final String	SCOPE_DEFAULT	"default"
public static final String	SCOPE_FETCHER	"fetcher"
public static final String	SCOPE_GENERATE_HOST_COUNT	"generate_host_count"
public static final String	SCOPE_INJECT	"inject"
public static final String	SCOPE_LINKDB	"linkdb"
public static final String	SCOPE_OUTLINK	"outlink"
public static final String	SCOPE_PARTITION	"partition"

org.apache.nutch.ontology.jena.OntologyImpl

public static final String

DELIMITER_SEARCHTERM

" "

org.apache.nutch.parse.HtmlParseFilters

public static final String

HTMLPARSEFILTER_ORDER

"htmlparsefilter.order"

org.apache.nutch.parse.ParseData

public static final String

DIR_NAME

"parse_data"

org.apache.nutch.parse.ParserFactory

public static final String

DEFAULT_PLUGIN

"*"

org.apache.nutch.parse.ParseStatus

public static final byte	FAILED	2
public static final short	FAILED_EXCEPTION	200
public static final short	FAILED_INVALID_FORMAT	203
public static final short	FAILED_MISSING_CONTENT	205
public static final short	FAILED_MISSING_PARTS	204
public static final short	FAILED_TRUNCATED	202
public static final byte	NOTPARSED	0
public static final byte	SUCCESS	1
public static final short	SUCCESS_REDIRECT	100

org.apache.nutch.parse.ParseText

public static final String

DIR_NAME

"parse_text"

org.apache.nutch.parse.msexcel.MSExcelParser

public static final String

MIME_TYPE

"application/vnd.ms-excel"

org.apache.nutch.parse.mspowerpoint.MSPowerPointParser

public static final String

MIME_TYPE

"application/vnd.ms-powerpoint"

org.apache.nutch.parse.msword.MSWordParser

public static final String

MIME_TYPE

"application/msword"

org.apache.nutch.protocol.Content

public static final String

DIR_NAME

"content"

org.apache.nutch.protocol.Protocol

public static final String	CHECK_BLOCKING	"protocol.plugin.check.blocking"
public static final String	CHECK_ROBOTS	"protocol.plugin.check.robots"

org.apache.nutch.protocol.ProtocolStatus

public static final int	ACCESS_DENIED	17
public static final int	BLOCKED	23
public static final int	EXCEPTION	16
public static final int	FAILED	2
public static final int	GONE	11
public static final int	MOVED	12
public static final int	NOTFETCHING	20
public static final int	NOTFOUND	14
public static final int	NOTMODIFIED	21
public static final int	PROTO_NOT_FOUND	10
public static final int	REDIR_EXCEEDED	19
public static final int	RETRY	15
public static final int	ROBOTS_DENIED	18
public static final int	SUCCESS	1
public static final int	TEMP_MOVED	13
public static final int	WOULDBLOCK	22

org.apache.nutch.protocol.ftp.Client

protected static final int	TERMINAL_TYPE	24
protected static final int	TERMINAL_TYPE_IS	0
protected static final int	TERMINAL_TYPE_SEND	1

org.apache.nutch.protocol.http.api.HttpBase

public static final int

BUFFER_SIZE

8192

org.apache.nutch.protocol.httpclient.HttpAuthenticationFactory

public static final String

WWW_AUTHENTICATE

"WWW-Authenticate"

org.apache.nutch.scoring.webgraph.LinkDatum

public static final byte	INLINK	1
public static final byte	OUTLINK	2

org.apache.nutch.scoring.webgraph.LinkDumper

public static final String

DUMP_DIR

"linkdump"

org.apache.nutch.scoring.webgraph.Loops

public static final String	LOOPS_DIR	"loops"
public static final String	ROUTES_DIR	"routes"

org.apache.nutch.scoring.webgraph.WebGraph

public static final String	INLINK_DIR	"inlinks"
public static final String	LOCK_NAME	".locked"
public static final String	NODE_DIR	"nodes"
public static final String	OUTLINK_DIR	"outlinks"

org.apache.nutch.searcher.FetchedSegments

public static final long

VERSION

1L

org.apache.nutch.searcher.LuceneSearchBean

public static final long

VERSION

1L

org.apache.nutch.searcher.NutchBean

public static final String

KEY

"nutchBean"

org.apache.nutch.searcher.Query.Clause

public static final String

DEFAULT_FIELD

"DEFAULT"

org.apache.nutch.searcher.QueryParams

public static final String	DEFAULT_DEDUP_FIELD	"site"
public static final int	DEFAULT_MAX_HITS_PER_DUP	2
public static final int	DEFAULT_NUM_HITS	10
public static final boolean	DEFAULT_REVERSE	false

org.apache.nutch.searcher.response.SearchServlet

public static final String	DEDUPE	"ddf"
public static final String	FIELDS	"field"
public static final String	LANG	"lang"
public static final String	NUM_DUPES	"dupes"
public static final String	QUERY	"query"
public static final String	RESPONSE_TYPE	"rt"
public static final String	REVERSE	"reverse"
public static final String	ROWS	"rows"
public static final String	SORT	"sort"
public static final String	START	"start"
public static final String	SUMMARY	"summary"

org.apache.nutch.tools.arc.ArcSegmentCreator

public static final String

URL_VERSION

"arc.url.version"

org.apache.nutch.util.EncodingDetector

public static final String	MIN_CONFIDENCE_KEY	"encodingdetector.charset.min.confidence"
public static final int	NO_THRESHOLD	-1

org.apache.nutch.util.domain.DomainSuffix

public static final float

DEFAULT_BOOST

1.0f