org.wandora.application.tools.extractors.bookmark

Class BookmarkExtractor

java.lang.Object

org.wandora.application.tools.AbstractWandoraTool

org.wandora.application.tools.extractors.AbstractExtractor

org.wandora.application.tools.extractors.bookmark.BookmarkExtractor

All Implemented Interfaces:

java.io.Serializable, java.lang.Runnable, BrowserPluginExtractor, DropExtractor, WandoraTool, WandoraToolLogger, Handler, InterruptHandler, TopicMapLogger

public class BookmarkExtractor
extends AbstractExtractor

Author:

Eero

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
private java.lang.String	ADD_SI
static java.lang.String[]	contentTypes
private java.text.SimpleDateFormat	df
private java.lang.String	HREF_SI
private java.lang.String	ICON_SI
private java.lang.String[][]	itemProps
private java.lang.String[][]	itemTimeProps
private java.lang.String	LANG
private Topic	langTopic
private java.lang.String	MOD_SI
private java.lang.String	ROOT_SI

Fields inherited from class org.wandora.application.tools.extractors.AbstractExtractor

CUSTOM_EXTRACTOR, DONE_FAILED, DONE_MANY, DONE_ONE, EXACTLY_GIVEN_URLS, FILE_EXTRACTOR, FILE_PATTERN, GIVEN_URLS_AND_ALL_CRAWLED_DOCUMENTS, GIVEN_URLS_AND_CRAWLED_DOCUMENTS_IN_URL_DOMAIN, GIVEN_URLS_AND_LINKED_DOCUMENTS, GIVEN_URLS_AND_URL_BELOW, INFO_WAIT_WHILE_WORKING, LOG_TITLE, POINT_START_URL_TEXT, RAW_EXTRACTOR, SELECT_DIALOG_TITLE, STRING_EXTRACTOR_NOT_SUPPORTED_MESSAGE, URL_EXTRACTOR

Fields inherited from interface org.wandora.application.WandoraToolLogger

CLOSE, EXECUTE, INVISIBLE, VISIBLE, WAIT

Fields inherited from interface org.wandora.application.tools.browserextractors.BrowserPluginExtractor

RETURN_ERROR, RETURN_INFO

Constructor Summary

Constructors

Constructor and Description
BookmarkExtractor()

Method Summary

Modifier and Type	Method and Description
boolean	_extractTopicsFrom(java.io.File f, TopicMap t)
boolean	_extractTopicsFrom(java.lang.String str, TopicMap t)
boolean	_extractTopicsFrom(java.net.URL u, TopicMap t)
java.lang.String[]	getContentTypes() Returns an array of String containing the content-types this ContentHandler can process.
java.lang.String	getDescription() AdminToolManager views tool descriptions while user browses available tools and build user customizable GUI elements such as Tools menu.
int	getExtractorType()
javax.swing.Icon	getIcon() All tools may have identifying graphic icon used within tool GUI elements.
java.lang.String	getName() Tools name represent the tool in UI unless the tool has been given explicitly another GUI name.
private Topic	getOrCreateTopic(TopicMap tm, java.lang.String si)
private Topic	getOrCreateTopic(TopicMap tm, java.lang.String si, java.lang.String bn)
private Topic	getWandoraClass(TopicMap tm)
private void	makeSubclassOf(TopicMap tm, Topic t, Topic superclass)
private void	parse(org.jsoup.nodes.Document d, TopicMap t)
private void	parseCategory(org.jsoup.nodes.Element c, TopicMap t)
private void	parseCategory(org.jsoup.nodes.Element c, Topic parent, TopicMap t)
private void	parseItem(org.jsoup.nodes.Element i, Topic parent, TopicMap t)

Methods inherited from class org.wandora.application.tools.extractors.AbstractExtractor

acceptBrowserExtractRequest, addCrawlerUrl, browserExtractorConsumesPlainText, buildSI, buildSL, clearMasterSubject, createAssociation, createAssociation, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, croppedFilename, croppedFilename, croppedUrlString, croppedUrlString, doBrowserExtract, dropExtract, dropExtract, dropExtract, execute, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFromText, getBrowserExtractorName, getCrawlerMode, getForceContent, getForceFiles, getForceUrls, getGUIText, getGUIText, getInterruptsHandled, getMasterSubject, getType, getWandora, handle, handleContent, handleCustomType, handleFiles, handleForcedContent, handleInterrupt, handleStringContent, handleUrls, initializeCustomType, instantDropHandle, makeSubclassOfWandoraClass, runInOwnThread, setData, setDisplayName, setForceContent, setForceFiles, setForceUrls, setMasterSubject, setMasterSubject, setTopicMap, setupCrawler, setWandora, takeNap, urlEncode, useTempTopicMap, useURLCrawler

Methods inherited from class org.wandora.application.tools.AbstractWandoraTool

addUndoMarker, addUndoMarker, allowMultipleInvocations, clearAllThreads, clearThreads, clearThreads, clearToolLock, clearToolLock, clearToolLocks, configure, execute, execute, forceStop, forceStop, getContext, getCurrentLogger, getDefaultLogger, getHistory, getLastLogger, getState, getThreads, getThreads, getToolMenuItem, getToolMenuItem, getTopicName, hlog, initialize, interruptAllThreads, interruptThreads, interruptThreads, isConfigurable, isRunning, isRunning, lockLog, log, log, log, log, requiresRefresh, run, setContext, setDefaultLogger, setLogTitle, setProgress, setProgressMax, setState, setToolLogger, singleLog, singleLog, singleLog, solveContextTopicMap, solveNameForTopicMap, writeOptions

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface org.wandora.application.WandoraTool

configure, execute, execute, getContext, getToolMenuItem, hlog, initialize, isConfigurable, isRunning, log, log, log, log, requiresRefresh, setContext, setToolLogger, writeOptions

Methods inherited from interface org.wandora.application.WandoraToolLogger

forceStop, getHistory, getState, lockLog, setLogTitle, setProgress, setProgressMax, setState

Methods inherited from interface java.lang.Runnable

run

Field Detail

ROOT_SI

private java.lang.String ROOT_SI

HREF_SI

private java.lang.String HREF_SI

ADD_SI

private java.lang.String ADD_SI

MOD_SI

private java.lang.String MOD_SI

ICON_SI

private java.lang.String ICON_SI

LANG

private java.lang.String LANG

langTopic

private Topic langTopic

df

private java.text.SimpleDateFormat df

itemProps

private java.lang.String[][] itemProps

itemTimeProps

private java.lang.String[][] itemTimeProps

contentTypes

public static final java.lang.String[] contentTypes

Constructor Detail

BookmarkExtractor

public BookmarkExtractor()

Method Detail

getName

public java.lang.String getName()

Description copied from class: AbstractWandoraTool

Tools name represent the tool in UI unless the tool has been given explicitly another GUI name. All tools should override this method and return a valid string representing name of the tool.

Specified by:

getName in interface WandoraTool

Overrides:

getName in class AbstractExtractor

Returns:

String representing tool's name.

getDescription

public java.lang.String getDescription()

Description copied from class: AbstractWandoraTool

AdminToolManager views tool descriptions while user browses available tools and build user customizable GUI elements such as Tools menu. By default description equals the tool name. All tools should override this method.

Specified by:

getDescription in interface WandoraTool

Overrides:

getDescription in class AbstractExtractor

Returns:

String representing tool's description.

getExtractorType

public int getExtractorType()

Overrides:

getExtractorType in class AbstractExtractor

getContentTypes

public java.lang.String[] getContentTypes()

Description copied from interface: Handler

Returns an array of String containing the content-types this ContentHandler can process.

Specified by:

getContentTypes in interface Handler

Overrides:

getContentTypes in class AbstractExtractor

Returns:

String array representing content types.

getIcon

public javax.swing.Icon getIcon()

Description copied from class: AbstractWandoraTool

All tools may have identifying graphic icon used within tool GUI elements. getIcon should return Icon object of the tool.

Specified by:

getIcon in interface WandoraTool

Overrides:

getIcon in class AbstractExtractor

Returns:

icon for the tool.

_extractTopicsFrom

public boolean _extractTopicsFrom(java.io.File f,
                                  TopicMap t)
                           throws java.lang.Exception

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

_extractTopicsFrom

public boolean _extractTopicsFrom(java.net.URL u,
                                  TopicMap t)
                           throws java.lang.Exception

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

_extractTopicsFrom

public boolean _extractTopicsFrom(java.lang.String str,
                                  TopicMap t)
                           throws java.lang.Exception

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

parse

private void parse(org.jsoup.nodes.Document d,
                   TopicMap t)
            throws java.io.FileNotFoundException,
                   java.io.IOException,
                   TopicMapException,
                   java.text.ParseException

Throws:

java.io.FileNotFoundException

java.io.IOException

TopicMapException

java.text.ParseException

parseCategory

private void parseCategory(org.jsoup.nodes.Element c,
                           TopicMap t)
                    throws TopicMapException,
                           java.text.ParseException

Throws:

TopicMapException

java.text.ParseException

parseCategory

private void parseCategory(org.jsoup.nodes.Element c,
                           Topic parent,
                           TopicMap t)
                    throws TopicMapException,
                           java.text.ParseException

Throws:

TopicMapException

java.text.ParseException

parseItem

private void parseItem(org.jsoup.nodes.Element i,
                       Topic parent,
                       TopicMap t)
                throws TopicMapException,
                       java.text.ParseException

Throws:

TopicMapException

java.text.ParseException

getWandoraClass

private Topic getWandoraClass(TopicMap tm)
                       throws TopicMapException

Throws:

TopicMapException

getOrCreateTopic

private Topic getOrCreateTopic(TopicMap tm,
                               java.lang.String si)
                        throws TopicMapException

Throws:

TopicMapException

getOrCreateTopic

private Topic getOrCreateTopic(TopicMap tm,
                               java.lang.String si,
                               java.lang.String bn)
                        throws TopicMapException

Throws:

TopicMapException

makeSubclassOf

private void makeSubclassOf(TopicMap tm,
                            Topic t,
                            Topic superclass)
                     throws TopicMapException

Throws:

TopicMapException