org.wandora.application.tools.extractors.ocr

Class OCRExtractor

java.lang.Object

org.wandora.application.tools.AbstractWandoraTool

org.wandora.application.tools.extractors.AbstractExtractor

org.wandora.application.tools.extractors.ocr.OCRExtractor

All Implemented Interfaces:

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

public class OCRExtractor
extends AbstractExtractor

Author:

Eero Lehtonen

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
private java.lang.String[]	contentTypes
protected java.lang.String	DATE_EXTRACTED_SI
protected java.lang.String	DATE_MODIFIED_SI
protected java.text.SimpleDateFormat	dateFormatter
protected java.lang.String	DOCUMENT_SI
protected java.lang.String	FILE_SIZE_SI
protected java.lang.String	SOURCE_SI
protected java.lang.String	TEMP_PATH
protected java.lang.String	TEXT_CONTENT_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
OCRExtractor()

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)
boolean	acceptBrowserExtractRequest(BrowserExtractRequest request, Wandora wandora)
Topic	createDocumentTypeTopic(TopicMap tm)
java.lang.String	doBrowserExtract(BrowserExtractRequest request, Wandora wandora)
java.lang.String	getBrowserExtractorName()
Topic	getContentType(TopicMap tm)
java.lang.String[]	getContentTypes() Returns an array of String containing the content-types this ContentHandler can process.
Topic	getDateModifiedType(TopicMap tm)
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.
protected Topic	getOrCreateLangTopic(TopicMap tm, java.lang.String lng6392)
protected Topic	getOrCreateTopic(TopicMap tm, java.lang.String si)
protected Topic	getOrCreateTopic(TopicMap tm, java.lang.String si, java.lang.String bn)
Topic	getSizeType(TopicMap tm)
Topic	getSourceType(TopicMap tm)
Topic	getTimeExtractedType(TopicMap tm)
Topic	getWandoraClass(TopicMap tm)
boolean	isConfigurable() Whether this tool is configurable.
protected void	makeSubclassOf(TopicMap tm, Topic t, Topic superclass)
private boolean	processFile(java.io.File f, TopicMap tm, Topic documentTopic)
boolean	useURLCrawler()

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

addCrawlerUrl, browserExtractorConsumesPlainText, buildSI, buildSL, clearMasterSubject, createAssociation, createAssociation, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, createTopic, croppedFilename, croppedFilename, croppedUrlString, croppedUrlString, dropExtract, dropExtract, dropExtract, execute, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFromText, 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

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, 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, 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

SOURCE_SI

protected java.lang.String SOURCE_SI

DOCUMENT_SI

protected java.lang.String DOCUMENT_SI

TEXT_CONTENT_SI

protected java.lang.String TEXT_CONTENT_SI

DATE_EXTRACTED_SI

protected java.lang.String DATE_EXTRACTED_SI

DATE_MODIFIED_SI

protected java.lang.String DATE_MODIFIED_SI

FILE_SIZE_SI

protected java.lang.String FILE_SIZE_SI

TEMP_PATH

protected java.lang.String TEMP_PATH

dateFormatter

protected java.text.SimpleDateFormat dateFormatter

contentTypes

private final java.lang.String[] contentTypes

Constructor Detail

OCRExtractor

public OCRExtractor()

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.

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.

useURLCrawler

public boolean useURLCrawler()

Overrides:

useURLCrawler in class AbstractExtractor

getExtractorType

public int getExtractorType()

Overrides:

getExtractorType in class AbstractExtractor

isConfigurable

public boolean isConfigurable()

Description copied from class: AbstractWandoraTool

Whether this tool is configurable. Configurable tools will provide an interface to configure themselves. To make a configurable tool you will need to:

• Implement isConfigurable method and return true there.
• Implement initialize method to initialize the tool with saved options.
• Implement configure method to provide the user interface for configuration and then save the new options.
• Implement writeOptions method to save the current options.

Specified by:

isConfigurable in interface WandoraTool

Overrides:

isConfigurable in class AbstractWandoraTool

Returns:

true if the tool is configurable.

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

getBrowserExtractorName

public java.lang.String getBrowserExtractorName()

Specified by:

getBrowserExtractorName in interface BrowserPluginExtractor

Overrides:

getBrowserExtractorName in class AbstractExtractor

_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

processFile

private boolean processFile(java.io.File f,
                            TopicMap tm,
                            Topic documentTopic)
                     throws TopicMapException

Throws:

TopicMapException

getContentType

public Topic getContentType(TopicMap tm)
                     throws TopicMapException

Throws:

TopicMapException

getDateModifiedType

public Topic getDateModifiedType(TopicMap tm)
                          throws TopicMapException

Throws:

TopicMapException

getTimeExtractedType

public Topic getTimeExtractedType(TopicMap tm)
                           throws TopicMapException

Throws:

TopicMapException

getSizeType

public Topic getSizeType(TopicMap tm)
                  throws TopicMapException

Throws:

TopicMapException

getSourceType

public Topic getSourceType(TopicMap tm)
                    throws TopicMapException

Throws:

TopicMapException

createDocumentTypeTopic

public Topic createDocumentTypeTopic(TopicMap tm)
                              throws TopicMapException

Throws:

TopicMapException

getWandoraClass

public Topic getWandoraClass(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

getOrCreateLangTopic

protected Topic getOrCreateLangTopic(TopicMap tm,
                                     java.lang.String lng6392)
                              throws TopicMapException

Throws:

TopicMapException