org.wandora.application.tools.extractors.email

Class SimpleEmailExtractor

java.lang.Object

org.wandora.application.tools.AbstractWandoraTool

org.wandora.application.tools.extractors.AbstractExtractor

org.wandora.application.tools.extractors.email.SimpleEmailExtractor

All Implemented Interfaces:

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

public class SimpleEmailExtractor
extends AbstractExtractor
implements BrowserPluginExtractor

Author:

akivela

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
private java.lang.String	defaultLang
private java.util.regex.Pattern	emailPattern
private java.lang.String	emailPatternString
private int	extractedEmails
private SimpleEmailExtractorPanel	gui
private java.util.regex.Pattern	namePattern
private java.lang.String	namePatternString
private boolean	shouldExtractHeaders
private boolean	shouldExtractUnknownContentTypeAttachments
private java.util.ArrayList	visitedDirectories
private java.util.ArrayList	visitedEmailFolders

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.tools.browserextractors.BrowserPluginExtractor

RETURN_ERROR, RETURN_INFO

Fields inherited from interface org.wandora.application.WandoraToolLogger

CLOSE, EXECUTE, INVISIBLE, VISIBLE, WAIT

Constructor Summary

Constructors

Constructor and Description
SimpleEmailExtractor() Creates a new instance of SimpleEmailExtractor

Method Summary

Modifier and Type	Method and Description
boolean	_extractTopicsFrom(java.io.File file, TopicMap topicMap)
boolean	_extractTopicsFrom(java.io.File file, TopicMap topicMap, int type)
boolean	_extractTopicsFrom(java.lang.String str, TopicMap topicMap)
boolean	_extractTopicsFrom(java.net.URL url, TopicMap topicMap)
void	_extractTopicsFromStream(java.lang.String locator, java.io.InputStream inputStream, TopicMap map)
boolean	acceptBrowserExtractRequest(BrowserExtractRequest request, Wandora wandora)
Locator	buildSI(java.lang.String siend)
Topic	createAddressTypeTopic(TopicMap map)
Topic	createEmailAddressTopic(javax.mail.Address emailAddress, TopicMap map)
Topic	createEmailAddressTopic(javax.mail.Address emailAddress, Topic additionalType, TopicMap map)
Topic	createEmailAddressTopic(java.lang.String emailAddress, TopicMap map)
Topic	createEmailAddressTopic(java.lang.String emailAddress, Topic additionalType, TopicMap map)
Topic	createEmailsTypeTopic(TopicMap map)
Topic	createEmailTypeTopic(TopicMap map)
Topic	createFolderTypeTopic(TopicMap map)
Topic	createWandoraTypeTopic(TopicMap map)
java.lang.String	doBrowserExtract(BrowserExtractRequest request, Wandora wandora)
void	execute(Wandora wandora, Context context) Runs the tool.
void	extractContent(TopicMap map, Topic emailTopic, javax.mail.Part part)
void	extractDates(TopicMap map, Topic emailTopic, javax.mail.Message message)
void	extractHeaders(TopicMap map, Topic emailTopic, javax.mail.Part part)
void	extractMessageFlags(TopicMap map, Topic emailTopic, javax.mail.Message message)
void	extractRecipients(TopicMap map, Topic emailTopic, javax.mail.Message message)
void	extractReplyTo(TopicMap map, Topic emailTopic, javax.mail.Message message)
void	extractSender(TopicMap map, Topic emailTopic, javax.mail.Message message)
boolean	extractTopicsFromDBX(java.io.File file, TopicMap topicMap)
void	extractTopicsFromFolder(TopicMap map, javax.mail.Folder folder)
boolean	extractTopicsFromMBOX(java.io.File file, TopicMap topicMap)
Topic	extractTopicsFromMessage(TopicMap map, javax.mail.Message message)
java.lang.String	getDescription() AdminToolManager views tool descriptions while user browses available tools and build user customizable GUI elements such as Tools menu.
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.

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

addCrawlerUrl, browserExtractorConsumesPlainText, buildSL, clearMasterSubject, createAssociation, createAssociation, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, croppedFilename, croppedFilename, croppedUrlString, croppedUrlString, dropExtract, dropExtract, dropExtract, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFromText, getBrowserExtractorName, getContentTypes, getCrawlerMode, getExtractorType, 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.tools.browserextractors.BrowserPluginExtractor

getBrowserExtractorName

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

visitedEmailFolders

private java.util.ArrayList visitedEmailFolders

visitedDirectories

private java.util.ArrayList visitedDirectories

defaultLang

private java.lang.String defaultLang

shouldExtractHeaders

private boolean shouldExtractHeaders

shouldExtractUnknownContentTypeAttachments

private boolean shouldExtractUnknownContentTypeAttachments

gui

private SimpleEmailExtractorPanel gui

extractedEmails

private int extractedEmails

emailPatternString

private java.lang.String emailPatternString

emailPattern

private java.util.regex.Pattern emailPattern

namePatternString

private java.lang.String namePatternString

namePattern

private java.util.regex.Pattern namePattern

Constructor Detail

SimpleEmailExtractor

public SimpleEmailExtractor()

Creates a new instance of SimpleEmailExtractor

Method Detail

doBrowserExtract

public java.lang.String doBrowserExtract(BrowserExtractRequest request,
                                         Wandora wandora)
                                  throws TopicMapException

Specified by:

doBrowserExtract in interface BrowserPluginExtractor

Overrides:

doBrowserExtract in class AbstractExtractor

Throws:

TopicMapException

acceptBrowserExtractRequest

public boolean acceptBrowserExtractRequest(BrowserExtractRequest request,
                                           Wandora wandora)
                                    throws TopicMapException

Specified by:

acceptBrowserExtractRequest in interface BrowserPluginExtractor

Overrides:

acceptBrowserExtractRequest in class AbstractExtractor

Throws:

TopicMapException

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.

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.

execute

public void execute(Wandora wandora,
                    Context context)

Description copied from interface: WandoraTool

Runs the tool. This is the main entry to the tool. Here the context

Specified by:

execute in interface WandoraTool

Overrides:

execute in class AbstractExtractor

_extractTopicsFrom

public boolean _extractTopicsFrom(java.net.URL url,
                                  TopicMap topicMap)
                           throws java.lang.Exception

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

_extractTopicsFrom

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

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

_extractTopicsFrom

public boolean _extractTopicsFrom(java.io.File file,
                                  TopicMap topicMap)
                           throws java.lang.Exception

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

_extractTopicsFrom

public boolean _extractTopicsFrom(java.io.File file,
                                  TopicMap topicMap,
                                  int type)
                           throws java.lang.Exception

Throws:

java.lang.Exception

extractTopicsFromMBOX

public boolean extractTopicsFromMBOX(java.io.File file,
                                     TopicMap topicMap)
                              throws java.lang.Exception

Throws:

java.lang.Exception

extractTopicsFromDBX

public boolean extractTopicsFromDBX(java.io.File file,
                                    TopicMap topicMap)
                             throws java.lang.Exception

Throws:

java.lang.Exception

_extractTopicsFromStream

public void _extractTopicsFromStream(java.lang.String locator,
                                     java.io.InputStream inputStream,
                                     TopicMap map)

extractTopicsFromFolder

public void extractTopicsFromFolder(TopicMap map,
                                    javax.mail.Folder folder)

extractTopicsFromMessage

public Topic extractTopicsFromMessage(TopicMap map,
                                      javax.mail.Message message)

extractRecipients

public void extractRecipients(TopicMap map,
                              Topic emailTopic,
                              javax.mail.Message message)

extractSender

public void extractSender(TopicMap map,
                          Topic emailTopic,
                          javax.mail.Message message)

extractReplyTo

public void extractReplyTo(TopicMap map,
                           Topic emailTopic,
                           javax.mail.Message message)

extractMessageFlags

public void extractMessageFlags(TopicMap map,
                                Topic emailTopic,
                                javax.mail.Message message)

extractDates

public void extractDates(TopicMap map,
                         Topic emailTopic,
                         javax.mail.Message message)

extractContent

public void extractContent(TopicMap map,
                           Topic emailTopic,
                           javax.mail.Part part)

extractHeaders

public void extractHeaders(TopicMap map,
                           Topic emailTopic,
                           javax.mail.Part part)

createAddressTypeTopic

public Topic createAddressTypeTopic(TopicMap map)
                             throws TopicMapException

Throws:

TopicMapException

createEmailsTypeTopic

public Topic createEmailsTypeTopic(TopicMap map)
                            throws TopicMapException

Throws:

TopicMapException

createFolderTypeTopic

public Topic createFolderTypeTopic(TopicMap map)
                            throws TopicMapException

Throws:

TopicMapException

createEmailTypeTopic

public Topic createEmailTypeTopic(TopicMap map)
                           throws TopicMapException

Throws:

TopicMapException

createWandoraTypeTopic

public Topic createWandoraTypeTopic(TopicMap map)
                             throws TopicMapException

Throws:

TopicMapException

createEmailAddressTopic

public Topic createEmailAddressTopic(javax.mail.Address emailAddress,
                                     TopicMap map)
                              throws java.lang.Exception

Throws:

java.lang.Exception

createEmailAddressTopic

public Topic createEmailAddressTopic(java.lang.String emailAddress,
                                     TopicMap map)
                              throws java.lang.Exception

Throws:

java.lang.Exception

createEmailAddressTopic

public Topic createEmailAddressTopic(javax.mail.Address emailAddress,
                                     Topic additionalType,
                                     TopicMap map)
                              throws java.lang.Exception

Throws:

java.lang.Exception

createEmailAddressTopic

public Topic createEmailAddressTopic(java.lang.String emailAddress,
                                     Topic additionalType,
                                     TopicMap map)
                              throws java.lang.Exception

Throws:

java.lang.Exception

buildSI

public Locator buildSI(java.lang.String siend)

Overrides:

buildSI in class AbstractExtractor