org.wandora.application.tools.extractors.bing

Class AbstractBingExtractor

java.lang.Object

org.wandora.application.tools.AbstractWandoraTool

org.wandora.application.tools.extractors.AbstractExtractor

org.wandora.application.tools.extractors.bing.AbstractBingExtractor

All Implemented Interfaces:

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

Direct Known Subclasses:

BingSearchResultExtractor

public abstract class AbstractBingExtractor
extends AbstractExtractor

Author:

akivela

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
private static java.util.HashMap<java.lang.String,java.lang.String>	auth
static java.lang.String	BING_CONTENTTYPE_SI
static java.lang.String	BING_DATETIME_SI
static java.lang.String	BING_DESCRIPTION_SI
static java.lang.String	BING_FILESIZE_SI
static java.lang.String	BING_HEIGHT_SI
static java.lang.String	BING_IMAGERESULT_SI
static java.lang.String	BING_ROOT
static java.lang.String	BING_TITLE_SI
static java.lang.String	BING_URL
static java.lang.String	BING_WEBRESULT_SI
static java.lang.String	BING_WIDTH_SI
private java.lang.String[]	contentTypes
protected java.lang.String	defaultEncoding
private static java.lang.String	dialogName
protected boolean	EXTRACT_RELEVANCE
static java.lang.String	LANG
static java.lang.String	QUERY_SI
static java.lang.String	SOURCE_SI
static java.lang.String	TOPIC_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
AbstractBingExtractor()

Method Summary

Modifier and Type	Method and Description
abstract boolean	_extractTopicsFrom(java.io.InputStream in, TopicMap topicMap)
void	configure(Wandora wandora, Options options, java.lang.String prefix) If the tool is configurable, shows an user interface to configure the tool.
void	fillQueryTopic(Topic queryTopic, TopicMap topicMap, java.lang.String query)
static void	forgetAuth()
Topic	getBingClass(TopicMap tm)
java.lang.String[]	getContentTypes() Returns an array of String containing the content-types this ContentHandler can process.
Topic	getContentTypeType(TopicMap tm)
Topic	getDateTimeTopic(java.lang.String datetime, TopicMap tm)
Topic	getDateTimeType(TopicMap tm)
Topic	getDescriptionTopic(java.lang.String description, TopicMap tm)
Topic	getDescriptionType(TopicMap tm)
protected java.lang.String	getFileContents(java.io.File file)
Topic	getFileSizeType(TopicMap tm)
Topic	getHeightType(TopicMap tm)
javax.swing.Icon	getIcon() All tools may have identifying graphic icon used within tool GUI elements.
Topic	getImageSearchResultTopic(TopicMap tm, java.lang.String title, java.lang.String url, java.lang.String width, java.lang.String height, java.lang.String contentType, java.lang.String fileSize)
Topic	getImageSearchResultType(TopicMap tm)
protected Topic	getOrCreateTopic(TopicMap tm, java.lang.String si)
protected Topic	getOrCreateTopic(TopicMap tm, java.lang.String si, java.lang.String bn)
Topic	getQueryType(TopicMap tm)
Topic	getSourceType(TopicMap tm)
protected java.lang.String	getStringFromDocument(org.w3c.dom.Document doc)
Topic	getTitleTopic(java.lang.String title, TopicMap tm)
Topic	getTitleType(TopicMap tm)
Topic	getTopicType(TopicMap tm)
WandoraToolType	getType() Tool type is used to categorize tools.
Topic	getWandoraClass(TopicMap tm)
Topic	getWebSearchResultTopic(TopicMap tm, java.lang.String title, java.lang.String description, java.lang.String url, java.lang.String datetime)
Topic	getWebSearchResultType(TopicMap tm)
Topic	getWidthType(TopicMap tm)
boolean	isConfigurable() Whether this tool is configurable.
protected void	makeSubclassOf(TopicMap tm, Topic t, Topic superclass)
java.util.HashMap<java.lang.String,java.lang.String>	solveAuth(Wandora wandora)
boolean	useURLCrawler()
void	writeOptions(Wandora wandora, Options options, java.lang.String prefix) If the tool is configurable, saves all current tool options.

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

_extractTopicsFrom, _extractTopicsFrom, _extractTopicsFrom, 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, getDescription, getExtractorType, getForceContent, getForceFiles, getForceUrls, getGUIText, getGUIText, getInterruptsHandled, getMasterSubject, getName, 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, 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

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

execute, execute, getContext, getToolMenuItem, hlog, initialize, isRunning, log, log, log, log, requiresRefresh, setContext, setToolLogger

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

EXTRACT_RELEVANCE

protected boolean EXTRACT_RELEVANCE

defaultEncoding

protected java.lang.String defaultEncoding

LANG

public static java.lang.String LANG

BING_ROOT

public static final java.lang.String BING_ROOT

See Also:

Constant Field Values

BING_URL

public static final java.lang.String BING_URL

See Also:

Constant Field Values

SOURCE_SI

public static final java.lang.String SOURCE_SI

See Also:

Constant Field Values

QUERY_SI

public static final java.lang.String QUERY_SI

See Also:

Constant Field Values

TOPIC_SI

public static final java.lang.String TOPIC_SI

See Also:

Constant Field Values

BING_TITLE_SI

public static final java.lang.String BING_TITLE_SI

See Also:

Constant Field Values

BING_DESCRIPTION_SI

public static final java.lang.String BING_DESCRIPTION_SI

See Also:

Constant Field Values

BING_DATETIME_SI

public static final java.lang.String BING_DATETIME_SI

See Also:

Constant Field Values

BING_WIDTH_SI

public static final java.lang.String BING_WIDTH_SI

See Also:

Constant Field Values

BING_HEIGHT_SI

public static final java.lang.String BING_HEIGHT_SI

See Also:

Constant Field Values

BING_CONTENTTYPE_SI

public static final java.lang.String BING_CONTENTTYPE_SI

See Also:

Constant Field Values

BING_FILESIZE_SI

public static final java.lang.String BING_FILESIZE_SI

See Also:

Constant Field Values

BING_WEBRESULT_SI

public static final java.lang.String BING_WEBRESULT_SI

See Also:

Constant Field Values

BING_IMAGERESULT_SI

public static final java.lang.String BING_IMAGERESULT_SI

See Also:

Constant Field Values

contentTypes

private final java.lang.String[] contentTypes

auth

private static java.util.HashMap<java.lang.String,java.lang.String> auth

dialogName

private static final java.lang.String dialogName

See Also:

Constant Field Values

Constructor Detail

AbstractBingExtractor

public AbstractBingExtractor()

Method Detail

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.

getType

public WandoraToolType getType()

Description copied from class: AbstractWandoraTool

Tool type is used to categorize tools. Tool type has no real effect today, it is merely an informative property of a tool.

Specified by:

getType in interface WandoraTool

Overrides:

getType in class AbstractExtractor

Returns:

Tool's type object.

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

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.

configure

public void configure(Wandora wandora,
                      Options options,
                      java.lang.String prefix)
               throws TopicMapException

Description copied from class: AbstractWandoraTool

If the tool is configurable, shows an user interface to configure the tool. The user may edit tool options which this method also need to save. Options specific to this tool should be saved with keys starting with the given prefix.

Specified by:

configure in interface WandoraTool

Overrides:

configure in class AbstractWandoraTool

Throws:

TopicMapException

writeOptions

public void writeOptions(Wandora wandora,
                         Options options,
                         java.lang.String prefix)

Description copied from class: AbstractWandoraTool

If the tool is configurable, saves all current tool options. Options specific to this tool should be saved with keys starting with the given prefix.

Specified by:

writeOptions in interface WandoraTool

Overrides:

writeOptions in class AbstractWandoraTool

_extractTopicsFrom

public abstract boolean _extractTopicsFrom(java.io.InputStream in,
                                           TopicMap topicMap)
                                    throws java.lang.Exception

Throws:

java.lang.Exception

fillQueryTopic

public void fillQueryTopic(Topic queryTopic,
                           TopicMap topicMap,
                           java.lang.String query)

getWebSearchResultTopic

public Topic getWebSearchResultTopic(TopicMap tm,
                                     java.lang.String title,
                                     java.lang.String description,
                                     java.lang.String url,
                                     java.lang.String datetime)
                              throws TopicMapException

Throws:

TopicMapException

getImageSearchResultTopic

public Topic getImageSearchResultTopic(TopicMap tm,
                                       java.lang.String title,
                                       java.lang.String url,
                                       java.lang.String width,
                                       java.lang.String height,
                                       java.lang.String contentType,
                                       java.lang.String fileSize)
                                throws TopicMapException

Throws:

TopicMapException

getTitleTopic

public Topic getTitleTopic(java.lang.String title,
                           TopicMap tm)
                    throws TopicMapException

Throws:

TopicMapException

getDescriptionTopic

public Topic getDescriptionTopic(java.lang.String description,
                                 TopicMap tm)
                          throws TopicMapException

Throws:

TopicMapException

getDateTimeTopic

public Topic getDateTimeTopic(java.lang.String datetime,
                              TopicMap tm)
                       throws TopicMapException

Throws:

TopicMapException

getTitleType

public Topic getTitleType(TopicMap tm)
                   throws TopicMapException

Throws:

TopicMapException

getDescriptionType

public Topic getDescriptionType(TopicMap tm)
                         throws TopicMapException

Throws:

TopicMapException

getDateTimeType

public Topic getDateTimeType(TopicMap tm)
                      throws TopicMapException

Throws:

TopicMapException

getWidthType

public Topic getWidthType(TopicMap tm)
                   throws TopicMapException

Throws:

TopicMapException

getHeightType

public Topic getHeightType(TopicMap tm)
                    throws TopicMapException

Throws:

TopicMapException

getContentTypeType

public Topic getContentTypeType(TopicMap tm)
                         throws TopicMapException

Throws:

TopicMapException

getFileSizeType

public Topic getFileSizeType(TopicMap tm)
                      throws TopicMapException

Throws:

TopicMapException

getBingClass

public Topic getBingClass(TopicMap tm)
                   throws TopicMapException

Throws:

TopicMapException

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

getQueryType

public Topic getQueryType(TopicMap tm)
                   throws TopicMapException

Throws:

TopicMapException

getWebSearchResultType

public Topic getWebSearchResultType(TopicMap tm)
                             throws TopicMapException

Throws:

TopicMapException

getImageSearchResultType

public Topic getImageSearchResultType(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

solveAuth

public java.util.HashMap<java.lang.String,java.lang.String> solveAuth(Wandora wandora)

forgetAuth

public static void forgetAuth()

getStringFromDocument

protected java.lang.String getStringFromDocument(org.w3c.dom.Document doc)

getFileContents

protected java.lang.String getFileContents(java.io.File file)
                                    throws java.io.IOException,
                                           java.io.FileNotFoundException

Throws:

java.io.IOException

java.io.FileNotFoundException