org.wandora.application.tools.extractors.files

Class SimpleDocumentExtractor

java.lang.Object

org.wandora.application.tools.AbstractWandoraTool

org.wandora.application.tools.extractors.AbstractExtractor

org.wandora.application.tools.extractors.files.SimpleDocumentExtractor

All Implemented Interfaces:

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

public class SimpleDocumentExtractor
extends AbstractExtractor
implements WandoraTool, BrowserPluginExtractor

Author:

akivela

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
private Wandora	admin
static java.lang.String[]	contentTypes
protected static java.lang.String	DEFAULT_DATE_FORMAT
private java.lang.String	defaultLang
protected static java.lang.String	DOCUMENT_SI
protected static java.lang.String	SOURCE_SI
protected static java.lang.String	TOPIC_SI
private java.util.ArrayList<java.lang.String>	visitedDirectories

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
SimpleDocumentExtractor() Creates a new instance of SimpleDocumentExtractor

Method Summary

Modifier and Type	Method and Description
boolean	_extractTopicsFrom(java.io.File file, TopicMap topicMap)
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 topicMap, Topic topic)
boolean	browserExtractorConsumesPlainText()
java.lang.String	doBrowserExtract(BrowserExtractRequest request, Wandora wandora)
void	execute(Wandora wandora, Context context) Runs the tool.
void	fillDocumentTopic(Topic textTopic, TopicMap topicMap, java.lang.String content)
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.
Topic	getDocumentType(TopicMap tm)
java.lang.String	getGUIText(int textType)
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.
protected Topic	getOrCreateTopic(TopicMap tm, java.lang.String si)
protected Topic	getOrCreateTopic(TopicMap tm, java.lang.String si, java.lang.String bn)
Topic	getSourceType(TopicMap tm)
Topic	getTopicType(TopicMap tm)
Topic	getWandoraClass(TopicMap tm)
protected void	makeSubclassOf(TopicMap tm, Topic t, Topic superclass)
void	setBinaryEnrichment(Topic textTopic, TopicMap topicMap, byte[] content, java.lang.String mimeType)
void	setTextEnrichment(Topic topic, TopicMap topicMap, java.lang.String content, java.lang.String title)
java.lang.String	solveTitle(java.lang.String content)
boolean	useTempTopicMap()
boolean	useURLCrawler()

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

acceptBrowserExtractRequest, addCrawlerUrl, buildSI, buildSL, clearMasterSubject, createAssociation, createAssociation, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, croppedFilename, croppedFilename, croppedUrlString, croppedUrlString, dropExtract, dropExtract, dropExtract, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFromText, getBrowserExtractorName, getCrawlerMode, getExtractorType, getForceContent, getForceFiles, getForceUrls, 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

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, getType, 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 org.wandora.application.tools.browserextractors.BrowserPluginExtractor

acceptBrowserExtractRequest, getBrowserExtractorName

Methods inherited from interface java.lang.Runnable

run

Field Detail

TOPIC_SI

protected static java.lang.String TOPIC_SI

SOURCE_SI

protected static java.lang.String SOURCE_SI

DOCUMENT_SI

protected static java.lang.String DOCUMENT_SI

DEFAULT_DATE_FORMAT

protected static java.lang.String DEFAULT_DATE_FORMAT

defaultLang

private java.lang.String defaultLang

admin

private Wandora admin

visitedDirectories

private java.util.ArrayList<java.lang.String> visitedDirectories

contentTypes

public static final java.lang.String[] contentTypes

Constructor Detail

SimpleDocumentExtractor

public SimpleDocumentExtractor()

Creates a new instance of SimpleDocumentExtractor

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.

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.

useTempTopicMap

public boolean useTempTopicMap()

Overrides:

useTempTopicMap in class AbstractExtractor

useURLCrawler

public boolean useURLCrawler()

Overrides:

useURLCrawler in class AbstractExtractor

getGUIText

public java.lang.String getGUIText(int textType)

Overrides:

getGUIText in class AbstractExtractor

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

browserExtractorConsumesPlainText

public boolean browserExtractorConsumesPlainText()

Overrides:

browserExtractorConsumesPlainText in class AbstractExtractor

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.lang.String str,
                                  TopicMap topicMap)
                           throws java.lang.Exception

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

_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.io.File file,
                                  TopicMap topicMap)
                           throws java.lang.Exception

Specified by:

_extractTopicsFrom in class AbstractExtractor

Throws:

java.lang.Exception

_extractTopicsFromStream

public void _extractTopicsFromStream(java.lang.String locator,
                                     java.io.InputStream inputStream,
                                     TopicMap topicMap,
                                     Topic topic)

setBinaryEnrichment

public void setBinaryEnrichment(Topic textTopic,
                                TopicMap topicMap,
                                byte[] content,
                                java.lang.String mimeType)

setTextEnrichment

public void setTextEnrichment(Topic topic,
                              TopicMap topicMap,
                              java.lang.String content,
                              java.lang.String title)

solveTitle

public java.lang.String solveTitle(java.lang.String content)

fillDocumentTopic

public void fillDocumentTopic(Topic textTopic,
                              TopicMap topicMap,
                              java.lang.String content)

getWandoraClass

public Topic getWandoraClass(TopicMap tm)
                      throws TopicMapException

Throws:

TopicMapException

getTopicType

public Topic getTopicType(TopicMap tm)
                   throws TopicMapException

Throws:

TopicMapException

getSourceType

public Topic getSourceType(TopicMap tm)
                    throws TopicMapException

Throws:

TopicMapException

getDocumentType

public Topic getDocumentType(TopicMap tm)
                      throws TopicMapException

Throws:

TopicMapException

getOrCreateTopic

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

Throws:

TopicMapException

getOrCreateTopic

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

Throws:

TopicMapException

makeSubclassOf

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

Throws:

TopicMapException

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.