|
||||||||
PREV NEXT | FRAMES NO FRAMES |
org.apache.* |
---|
org.apache.manifoldcf.crawler.connectors.webcrawler.AuthenticationCredentials | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: AuthenticationCredentials.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.BasicParseState | ||
---|---|---|
protected static final int |
BASICPARSESTATE_IN_ATTR_LOOKING_FOR_VALUE |
12 |
protected static final int |
BASICPARSESTATE_IN_ATTR_NAME |
8 |
protected static final int |
BASICPARSESTATE_IN_ATTR_VALUE |
9 |
protected static final int |
BASICPARSESTATE_IN_COMMENT |
4 |
protected static final int |
BASICPARSESTATE_IN_DOUBLE_QUOTES_ATTR_VALUE |
14 |
protected static final int |
BASICPARSESTATE_IN_END_TAG_NAME |
11 |
protected static final int |
BASICPARSESTATE_IN_SINGLE_QUOTES_ATTR_VALUE |
13 |
protected static final int |
BASICPARSESTATE_IN_TAG_NAME |
7 |
protected static final int |
BASICPARSESTATE_IN_TAG_SAW_SLASH |
10 |
protected static final int |
BASICPARSESTATE_IN_UNQUOTED_ATTR_VALUE |
15 |
protected static final int |
BASICPARSESTATE_NORMAL |
0 |
protected static final int |
BASICPARSESTATE_SAWCOMMENTDASH |
5 |
protected static final int |
BASICPARSESTATE_SAWDASH |
3 |
protected static final int |
BASICPARSESTATE_SAWEXCLAMATION |
2 |
protected static final int |
BASICPARSESTATE_SAWLEFTBRACKET |
1 |
protected static final int |
BASICPARSESTATE_SAWSECONDCOMMENTDASH |
6 |
org.apache.manifoldcf.crawler.connectors.webcrawler.CookieManager | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: CookieManager.java 988245 2010-08-23 18:39:35Z kwright $" |
protected static final java.lang.String |
commentField |
"comment" |
protected static final java.lang.String |
commentURLField |
"commenturl" |
protected static final java.lang.String |
discardField |
"discard" |
protected static final java.lang.String |
domainField |
"domain" |
protected static final java.lang.String |
domainSpecifiedField |
"domainspecified" |
protected static final java.lang.String |
expirationDateField |
"expirationdate" |
protected static final java.lang.String |
keyField |
"sequencekey" |
protected static final java.lang.String |
nameField |
"name" |
protected static final java.lang.String |
ordinalField |
"ordinal" |
protected static final java.lang.String |
pathField |
"path" |
protected static final java.lang.String |
pathSpecifiedField |
"pathspecified" |
protected static final java.lang.String |
portBlankField |
"portblank" |
protected static final java.lang.String |
portField |
"ports" |
protected static final java.lang.String |
portSpecifiedField |
"portspecified" |
protected static final java.lang.String |
secureField |
"secure" |
protected static final java.lang.String |
valueField |
"value" |
protected static final java.lang.String |
versionField |
"version" |
protected static final java.lang.String |
versionSpecifiedField |
"versionspecified" |
org.apache.manifoldcf.crawler.connectors.webcrawler.CredentialsDescription | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: CredentialsDescription.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.DataCache | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: DataCache.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.DNSManager | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: DNSManager.java 988245 2010-08-23 18:39:35Z kwright $" |
protected static final java.lang.String |
expirationField |
"expirationtime" |
protected static final java.lang.String |
fqdnField |
"canonicalhostname" |
protected static final java.lang.String |
hostField |
"hostname" |
protected static final java.lang.String |
ipaddressField |
"ipaddress" |
org.apache.manifoldcf.crawler.connectors.webcrawler.FormData | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: FormData.java 988245 2010-08-23 18:39:35Z kwright $" |
public static final int |
SUBMITMETHOD_GET |
0 |
public static final int |
SUBMITMETHOD_POST |
1 |
org.apache.manifoldcf.crawler.connectors.webcrawler.FormDataAccumulator | ||
---|---|---|
protected static final int |
ELEMENTCATEGORY_FIXEDEXCLUSIVE |
1 |
protected static final int |
ELEMENTCATEGORY_FIXEDINCLUSIVE |
2 |
protected static final int |
ELEMENTCATEGORY_FREEFORM |
0 |
org.apache.manifoldcf.crawler.connectors.webcrawler.FormDataElement | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: FormDataElement.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.FormParseState | ||
---|---|---|
protected static final int |
FORMPARSESTATE_IN_FORM |
1 |
protected static final int |
FORMPARSESTATE_IN_OPTION |
4 |
protected static final int |
FORMPARSESTATE_IN_SELECT |
2 |
protected static final int |
FORMPARSESTATE_IN_TEXTAREA |
3 |
protected static final int |
FORMPARSESTATE_NORMAL |
0 |
org.apache.manifoldcf.crawler.connectors.webcrawler.IThrottledConnection | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: IThrottledConnection.java 988245 2010-08-23 18:39:35Z kwright $" |
public static final int |
FETCH_BAD_URI |
-101 |
public static final int |
FETCH_CIRCULAR_REDIRECT |
-100 |
public static final int |
FETCH_INTERRUPTED |
-104 |
public static final int |
FETCH_IO_ERROR |
-103 |
public static final int |
FETCH_NOT_TRIED |
-1 |
public static final int |
FETCH_SEQUENCE_ERROR |
-102 |
public static final int |
FETCH_UNKNOWN_ERROR |
-999 |
org.apache.manifoldcf.crawler.connectors.webcrawler.LoginCookies | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: LoginCookies.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.LoginParameters | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: LoginParameters.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.Messages | ||
---|---|---|
public static final java.lang.String |
DEFAULT_BUNDLE_NAME |
"org.apache.manifoldcf.crawler.connectors.webcrawler.common" |
public static final java.lang.String |
DEFAULT_PATH_NAME |
"org.apache.manifoldcf.crawler.connectors.webcrawler" |
org.apache.manifoldcf.crawler.connectors.webcrawler.PageCredentials | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: PageCredentials.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.RobotsManager | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: RobotsManager.java 988245 2010-08-23 18:39:35Z kwright $" |
protected static final java.lang.String |
expirationField |
"expirationtime" |
protected static final java.lang.String |
hostField |
"hostname" |
protected static final java.lang.String |
robotsField |
"robotsdata" |
org.apache.manifoldcf.crawler.connectors.webcrawler.ScriptParseState | ||
---|---|---|
protected static final int |
SCRIPTPARSESTATE_INSCRIPT |
1 |
protected static final int |
SCRIPTPARSESTATE_NORMAL |
0 |
org.apache.manifoldcf.crawler.connectors.webcrawler.SequenceCredentials | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: SequenceCredentials.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.ThrottleDescription | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: ThrottleDescription.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.ThrottledFetcher | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: ThrottledFetcher.java 989847 2010-08-26 17:52:30Z kwright $" |
protected static final java.lang.String |
dataFileFolder |
"/common/web/data/" |
protected static final int |
READ_CHUNK_LENGTH |
4096 |
protected static final boolean |
recordEverything |
false |
protected static final java.lang.String |
resultLogFile |
"/common/web/resultlog" |
protected static final long |
TIME_15MIN |
1500000L |
protected static final long |
TIME_1DAY |
86400000L |
protected static final long |
TIME_2HRS |
7200000L |
protected static final long |
TIME_5MIN |
300000L |
protected static final long |
TIME_6HRS |
21600000L |
org.apache.manifoldcf.crawler.connectors.webcrawler.TrustsDescription | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: TrustsDescription.java 988245 2010-08-23 18:39:35Z kwright $" |
org.apache.manifoldcf.crawler.connectors.webcrawler.WebcrawlerConfig | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: WebcrawlerConfig.java 995042 2010-09-08 13:10:06Z kwright $" |
public static final java.lang.String |
ATTR_BINREGEXP |
"binregexp" |
public static final java.lang.String |
ATTR_DOMAIN |
"domain" |
public static final java.lang.String |
ATTR_INSENSITIVE |
"caseinsensitive" |
public static final java.lang.String |
ATTR_MATCHREGEXP |
"match" |
public static final java.lang.String |
ATTR_NAMEREGEXP |
"name" |
public static final java.lang.String |
ATTR_PASSWORD |
"password" |
public static final java.lang.String |
ATTR_TRUSTEVERYTHING |
"trusteverything" |
public static final java.lang.String |
ATTR_TRUSTSTORE |
"truststore" |
public static final java.lang.String |
ATTR_TYPE |
"type" |
public static final java.lang.String |
ATTR_URLREGEXP |
"urlregexp" |
public static final java.lang.String |
ATTR_USERNAME |
"username" |
public static final java.lang.String |
ATTR_VALUE |
"value" |
public static final java.lang.String |
ATTRVALUE_BASIC |
"basic" |
public static final java.lang.String |
ATTRVALUE_FORM |
"form" |
public static final java.lang.String |
ATTRVALUE_LINK |
"link" |
public static final java.lang.String |
ATTRVALUE_NTLM |
"ntlm" |
public static final java.lang.String |
ATTRVALUE_REDIRECTION |
"redirection" |
public static final java.lang.String |
ATTRVALUE_SESSION |
"session" |
public static final java.lang.String |
NODE_ACCESSCREDENTIAL |
"accesscredential" |
public static final java.lang.String |
NODE_AUTHPAGE |
"authpage" |
public static final java.lang.String |
NODE_AUTHPARAMETER |
"authparameter" |
public static final java.lang.String |
NODE_BINDESC |
"bindesc" |
public static final java.lang.String |
NODE_EXCLUDES |
"excludes" |
public static final java.lang.String |
NODE_EXCLUDESINDEX |
"excludesindex" |
public static final java.lang.String |
NODE_INCLUDES |
"includes" |
public static final java.lang.String |
NODE_INCLUDESINDEX |
"includesindex" |
public static final java.lang.String |
NODE_LIMITTOSEEDS |
"limittoseeds" |
public static final java.lang.String |
NODE_MAXCONNECTIONS |
"maxconnections" |
public static final java.lang.String |
NODE_MAXFETCHESPERMINUTE |
"maxfetchesperminute" |
public static final java.lang.String |
NODE_MAXKBPERSECOND |
"maxkbpersecond" |
public static final java.lang.String |
NODE_SEEDS |
"seeds" |
public static final java.lang.String |
NODE_TRUST |
"trust" |
public static final java.lang.String |
PARAMETER_EMAIL |
"Email address" |
public static final java.lang.String |
PARAMETER_PROXYAUTHDOMAIN |
"Proxy authentication domain" |
public static final java.lang.String |
PARAMETER_PROXYAUTHPASSWORD |
"Proxy authentication password" |
public static final java.lang.String |
PARAMETER_PROXYAUTHUSERNAME |
"Proxy authentication user name" |
public static final java.lang.String |
PARAMETER_PROXYHOST |
"Proxy host" |
public static final java.lang.String |
PARAMETER_PROXYPORT |
"Proxy port" |
public static final java.lang.String |
PARAMETER_ROBOTSUSAGE |
"Robots usage" |
org.apache.manifoldcf.crawler.connectors.webcrawler.WebcrawlerConnector | ||
---|---|---|
public static final java.lang.String |
_rcsid |
"@(#)$Id: WebcrawlerConnector.java 995042 2010-09-08 13:10:06Z kwright $" |
public static final java.lang.String |
ACTIVITY_FETCH |
"fetch" |
public static final java.lang.String |
ACTIVITY_LOGON_END |
"end logon" |
public static final java.lang.String |
ACTIVITY_LOGON_START |
"begin logon" |
public static final java.lang.String |
ACTIVITY_ROBOTSPARSE |
"robots parse" |
protected static final java.lang.String |
FETCH_LOGIN |
"LOGIN" |
protected static final java.lang.String |
FETCH_ROBOTS |
"ROBOTS" |
protected static final java.lang.String |
FETCH_STANDARD |
"URL" |
public static final java.lang.String |
REL_LINK |
"link" |
public static final java.lang.String |
REL_REDIRECT |
"redirect" |
protected static final int |
RESULT_NO_DOCUMENT |
0 |
protected static final int |
RESULT_NO_VERSION |
1 |
protected static final int |
RESULT_RETRY_DOCUMENT |
3 |
protected static final int |
RESULT_VERSION_NEEDED |
2 |
protected static final int |
RESULTSTATUS_FALSE |
0 |
protected static final int |
RESULTSTATUS_NOTYETDETERMINED |
2 |
protected static final int |
RESULTSTATUS_TRUE |
1 |
protected static final int |
ROBOTS_ALL |
2 |
protected static final int |
ROBOTS_DATA |
1 |
protected static final int |
ROBOTS_NONE |
0 |
protected static final int |
SESSIONSTATE_LOGIN |
1 |
protected static final int |
SESSIONSTATE_NORMAL |
0 |
|
||||||||
PREV NEXT | FRAMES NO FRAMES |