org.wandora.application.tools.extractors

Class IMDBExtractor

java.lang.Object

org.wandora.application.tools.AbstractWandoraTool

org.wandora.application.tools.extractors.AbstractExtractor

org.wandora.application.tools.extractors.IMDBExtractor

All Implemented Interfaces:

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

public class IMDBExtractor
extends AbstractExtractor
implements WandoraTool

Author:

olli

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static java.lang.String	ACTOR_SI
static java.lang.String	BIOGRAPHY_SI
static java.lang.String	COUNTRY_SI
static java.lang.String	DATE_SI
static java.lang.String	DATEOFBIRTH_SI
static java.lang.String	DATEOFDEATH_SI
static java.lang.String	DIRECTOR_SI
static java.lang.String	EPISODE_SI
static java.lang.String	GENRE_SI
static java.lang.String	KEYWORD_SI
static java.lang.String	LANGUAGE_SI
static java.lang.String	LOCATION_SI
static java.lang.String	MOVIE_SI
static java.lang.String	PERSON_SI
static java.lang.String	PLACE_SI
static java.lang.String	PLOT_SI
static java.lang.String	PRODUCER_SI
static java.lang.String	REALNAME_SI
static java.lang.String	RELEASEDATE_SI
static java.lang.String	RELEASEDATEINFO_SI
static java.lang.String	ROLE_SI
static java.lang.String	RUNTIME_SI
static java.lang.String	RUNTIMEINFO_SI
static java.lang.String	SHOW_SI
static java.lang.String	TVMINI_SI
static java.lang.String	TVMOVIE_SI
static java.lang.String	TVSHOW_SI
private java.util.HashSet<java.lang.String>	typeFilter
private static java.util.HashMap<java.lang.String,java.lang.String>	typeNames
static java.lang.String	VIDEOGAME_SI
static java.lang.String	VIDEOMOVIE_SI
static java.lang.String	YEAR_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
IMDBExtractor() Creates a new instance of IMDBExtractor

Method Summary

Modifier and Type	Method and Description
boolean	_extractTopicsFrom(java.io.File f, TopicMap t)
boolean	_extractTopicsFrom(java.lang.String str, TopicMap topicMap)
boolean	_extractTopicsFrom(java.net.URL u, TopicMap t)
Association	addAppearance(TopicMap tm, Topic actor, Topic show, Topic episode, Topic role, java.lang.String type)
void	addBiography(TopicMap tm, Topic person, java.lang.String text)
void	addDateOfBirth(TopicMap tm, Topic person, java.lang.String text)
void	addDateOfDeath(TopicMap tm, Topic person, java.lang.String text)
Association	addKeyword(TopicMap tm, Topic show, Topic keyword, java.lang.String keywordType)
void	addPlot(TopicMap tm, Topic show, java.lang.String text)
void	addRealName(TopicMap tm, Topic person, java.lang.String text)
void	addReleaseDate(TopicMap tm, Topic show, Topic date, Topic info)
void	addRuntime(TopicMap tm, Topic show, Topic runtime, Topic runtimeinfo)
boolean	browserExtractorConsumesPlainText()
private static java.lang.String	cleanLocator(java.lang.String s)
void	createSchemaTopics(TopicMap tm)
boolean	extractActorList(java.io.BufferedReader in, TopicMap tm)
boolean	extractBiographyList(java.io.BufferedReader in, TopicMap tm)
boolean	extractCountryList(java.io.BufferedReader in, TopicMap tm)
boolean	extractDirectorList(java.io.BufferedReader in, TopicMap tm)
boolean	extractFromReader(java.io.BufferedReader in, TopicMap tm)
boolean	extractGenreList(java.io.BufferedReader in, TopicMap tm)
boolean	extractKeywordList(java.io.BufferedReader in, TopicMap tm)
boolean	extractKeywords(java.io.BufferedReader in, TopicMap tm, java.lang.String keywordType)
boolean	extractLanguageList(java.io.BufferedReader in, TopicMap tm)
boolean	extractLocationList(java.io.BufferedReader in, TopicMap tm)
boolean	extractMovieList(java.io.BufferedReader in, TopicMap tm)
boolean	extractPersonList(java.io.BufferedReader in, TopicMap tm, java.lang.String type)
boolean	extractPlotList(java.io.BufferedReader in, TopicMap tm)
boolean	extractProducerList(java.io.BufferedReader in, TopicMap tm)
boolean	extractReleaseDateList(java.io.BufferedReader in, TopicMap tm)
boolean	extractRunningTimeList(java.io.BufferedReader in, TopicMap tm)
Topic	getActorTopic(TopicMap tm, java.lang.String name)
Topic	getDateTopic(TopicMap tm, java.lang.String text)
java.lang.String	getDescription() AdminToolManager views tool descriptions while user browses available tools and build user customizable GUI elements such as Tools menu.
Topic	getEpisodeTopic(TopicMap tm, Topic show, java.lang.String name)
java.lang.String	getGUIText(int textType)
Topic	getKeywordTopic(TopicMap tm, java.lang.String name, java.lang.String keywordType)
java.lang.String	getName() Tools name represent the tool in UI unless the tool has been given explicitly another GUI name.
Topic	getOrCreateTopic(TopicMap tm, java.lang.String si)
Topic	getPersonTopic(TopicMap tm, java.lang.String name)
Topic	getPersonTopic(TopicMap tm, java.lang.String name, java.lang.String type)
Topic	getPlaceTopic(TopicMap tm, java.lang.String text)
Topic	getRoleTopic(TopicMap tm, java.lang.String name)
Topic	getShowTopic(TopicMap tm, java.lang.String name)
boolean	useTempTopicMap()

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, doBrowserExtract, dropExtract, dropExtract, dropExtract, execute, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFrom, extractTopicsFromText, getBrowserExtractorName, getContentTypes, getCrawlerMode, getExtractorType, getForceContent, getForceFiles, getForceUrls, getGUIText, getIcon, 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, 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, execute, getContext, getIcon, 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 java.lang.Runnable

run

Field Detail

ACTOR_SI

public static final java.lang.String ACTOR_SI

See Also:

Constant Field Values

DIRECTOR_SI

public static final java.lang.String DIRECTOR_SI

See Also:

Constant Field Values

PRODUCER_SI

public static final java.lang.String PRODUCER_SI

See Also:

Constant Field Values

ROLE_SI

public static final java.lang.String ROLE_SI

See Also:

Constant Field Values

EPISODE_SI

public static final java.lang.String EPISODE_SI

See Also:

Constant Field Values

SHOW_SI

public static final java.lang.String SHOW_SI

See Also:

Constant Field Values

TVSHOW_SI

public static final java.lang.String TVSHOW_SI

See Also:

Constant Field Values

TVMINI_SI

public static final java.lang.String TVMINI_SI

See Also:

Constant Field Values

MOVIE_SI

public static final java.lang.String MOVIE_SI

See Also:

Constant Field Values

TVMOVIE_SI

public static final java.lang.String TVMOVIE_SI

See Also:

Constant Field Values

VIDEOMOVIE_SI

public static final java.lang.String VIDEOMOVIE_SI

See Also:

Constant Field Values

VIDEOGAME_SI

public static final java.lang.String VIDEOGAME_SI

See Also:

Constant Field Values

KEYWORD_SI

public static final java.lang.String KEYWORD_SI

See Also:

Constant Field Values

LANGUAGE_SI

public static final java.lang.String LANGUAGE_SI

See Also:

Constant Field Values

COUNTRY_SI

public static final java.lang.String COUNTRY_SI

See Also:

Constant Field Values

YEAR_SI

public static final java.lang.String YEAR_SI

See Also:

Constant Field Values

GENRE_SI

public static final java.lang.String GENRE_SI

See Also:

Constant Field Values

LOCATION_SI

public static final java.lang.String LOCATION_SI

See Also:

Constant Field Values

RUNTIME_SI

public static final java.lang.String RUNTIME_SI

See Also:

Constant Field Values

RUNTIMEINFO_SI

public static final java.lang.String RUNTIMEINFO_SI

See Also:

Constant Field Values

RELEASEDATE_SI

public static final java.lang.String RELEASEDATE_SI

See Also:

Constant Field Values

RELEASEDATEINFO_SI

public static final java.lang.String RELEASEDATEINFO_SI

See Also:

Constant Field Values

PLOT_SI

public static final java.lang.String PLOT_SI

See Also:

Constant Field Values

PERSON_SI

public static final java.lang.String PERSON_SI

See Also:

Constant Field Values

DATE_SI

public static final java.lang.String DATE_SI

See Also:

Constant Field Values

DATEOFBIRTH_SI

public static final java.lang.String DATEOFBIRTH_SI

See Also:

Constant Field Values

DATEOFDEATH_SI

public static final java.lang.String DATEOFDEATH_SI

See Also:

Constant Field Values

BIOGRAPHY_SI

public static final java.lang.String BIOGRAPHY_SI

See Also:

Constant Field Values

REALNAME_SI

public static final java.lang.String REALNAME_SI

See Also:

Constant Field Values

PLACE_SI

public static final java.lang.String PLACE_SI

See Also:

Constant Field Values

typeFilter

private java.util.HashSet<java.lang.String> typeFilter

typeNames

private static final java.util.HashMap<java.lang.String,java.lang.String> typeNames

Constructor Detail

IMDBExtractor

public IMDBExtractor()

Creates a new instance of IMDBExtractor

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.

useTempTopicMap

public boolean useTempTopicMap()

Overrides:

useTempTopicMap in class AbstractExtractor

getGUIText

public java.lang.String getGUIText(int textType)

Overrides:

getGUIText in class AbstractExtractor

browserExtractorConsumesPlainText

public boolean browserExtractorConsumesPlainText()

Overrides:

browserExtractorConsumesPlainText 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.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

extractFromReader

public boolean extractFromReader(java.io.BufferedReader in,
                                 TopicMap tm)
                          throws java.lang.Exception

Throws:

java.lang.Exception

createSchemaTopics

public void createSchemaTopics(TopicMap tm)
                        throws TopicMapException

Throws:

TopicMapException

getOrCreateTopic

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

Throws:

TopicMapException

cleanLocator

private static java.lang.String cleanLocator(java.lang.String s)

getPersonTopic

public Topic getPersonTopic(TopicMap tm,
                            java.lang.String name)
                     throws TopicMapException

Throws:

TopicMapException

getPersonTopic

public Topic getPersonTopic(TopicMap tm,
                            java.lang.String name,
                            java.lang.String type)
                     throws TopicMapException

Throws:

TopicMapException

getActorTopic

public Topic getActorTopic(TopicMap tm,
                           java.lang.String name)
                    throws TopicMapException

Throws:

TopicMapException

getShowTopic

public Topic getShowTopic(TopicMap tm,
                          java.lang.String name)
                   throws TopicMapException

Throws:

TopicMapException

getRoleTopic

public Topic getRoleTopic(TopicMap tm,
                          java.lang.String name)
                   throws TopicMapException

Throws:

TopicMapException

getEpisodeTopic

public Topic getEpisodeTopic(TopicMap tm,
                             Topic show,
                             java.lang.String name)
                      throws TopicMapException

Throws:

TopicMapException

getKeywordTopic

public Topic getKeywordTopic(TopicMap tm,
                             java.lang.String name,
                             java.lang.String keywordType)
                      throws TopicMapException

Throws:

TopicMapException

getDateTopic

public Topic getDateTopic(TopicMap tm,
                          java.lang.String text)
                   throws TopicMapException

Throws:

TopicMapException

getPlaceTopic

public Topic getPlaceTopic(TopicMap tm,
                           java.lang.String text)
                    throws TopicMapException

Throws:

TopicMapException

addAppearance

public Association addAppearance(TopicMap tm,
                                 Topic actor,
                                 Topic show,
                                 Topic episode,
                                 Topic role,
                                 java.lang.String type)
                          throws TopicMapException

Throws:

TopicMapException

addKeyword

public Association addKeyword(TopicMap tm,
                              Topic show,
                              Topic keyword,
                              java.lang.String keywordType)
                       throws TopicMapException

Throws:

TopicMapException

addDateOfBirth

public void addDateOfBirth(TopicMap tm,
                           Topic person,
                           java.lang.String text)
                    throws TopicMapException

Throws:

TopicMapException

addDateOfDeath

public void addDateOfDeath(TopicMap tm,
                           Topic person,
                           java.lang.String text)
                    throws TopicMapException

Throws:

TopicMapException

addRuntime

public void addRuntime(TopicMap tm,
                       Topic show,
                       Topic runtime,
                       Topic runtimeinfo)
                throws TopicMapException

Throws:

TopicMapException

addReleaseDate

public void addReleaseDate(TopicMap tm,
                           Topic show,
                           Topic date,
                           Topic info)
                    throws TopicMapException

Throws:

TopicMapException

addBiography

public void addBiography(TopicMap tm,
                         Topic person,
                         java.lang.String text)
                  throws TopicMapException

Throws:

TopicMapException

addRealName

public void addRealName(TopicMap tm,
                        Topic person,
                        java.lang.String text)
                 throws TopicMapException

Throws:

TopicMapException

addPlot

public void addPlot(TopicMap tm,
                    Topic show,
                    java.lang.String text)
             throws TopicMapException

Throws:

TopicMapException

extractDirectorList

public boolean extractDirectorList(java.io.BufferedReader in,
                                   TopicMap tm)
                            throws java.lang.Exception

Throws:

java.lang.Exception

extractProducerList

public boolean extractProducerList(java.io.BufferedReader in,
                                   TopicMap tm)
                            throws java.lang.Exception

Throws:

java.lang.Exception

extractActorList

public boolean extractActorList(java.io.BufferedReader in,
                                TopicMap tm)
                         throws java.lang.Exception

Throws:

java.lang.Exception

extractPersonList

public boolean extractPersonList(java.io.BufferedReader in,
                                 TopicMap tm,
                                 java.lang.String type)
                          throws java.lang.Exception

Throws:

java.lang.Exception

extractKeywordList

public boolean extractKeywordList(java.io.BufferedReader in,
                                  TopicMap tm)
                           throws java.lang.Exception

Throws:

java.lang.Exception

extractLanguageList

public boolean extractLanguageList(java.io.BufferedReader in,
                                   TopicMap tm)
                            throws java.lang.Exception

Throws:

java.lang.Exception

extractCountryList

public boolean extractCountryList(java.io.BufferedReader in,
                                  TopicMap tm)
                           throws java.lang.Exception

Throws:

java.lang.Exception

extractLocationList

public boolean extractLocationList(java.io.BufferedReader in,
                                   TopicMap tm)
                            throws java.lang.Exception

Throws:

java.lang.Exception

extractGenreList

public boolean extractGenreList(java.io.BufferedReader in,
                                TopicMap tm)
                         throws java.lang.Exception

Throws:

java.lang.Exception

extractMovieList

public boolean extractMovieList(java.io.BufferedReader in,
                                TopicMap tm)
                         throws java.lang.Exception

Throws:

java.lang.Exception

extractKeywords

public boolean extractKeywords(java.io.BufferedReader in,
                               TopicMap tm,
                               java.lang.String keywordType)
                        throws java.lang.Exception

Throws:

java.lang.Exception

extractBiographyList

public boolean extractBiographyList(java.io.BufferedReader in,
                                    TopicMap tm)
                             throws java.lang.Exception

Throws:

java.lang.Exception

extractPlotList

public boolean extractPlotList(java.io.BufferedReader in,
                               TopicMap tm)
                        throws java.lang.Exception

Throws:

java.lang.Exception

extractRunningTimeList

public boolean extractRunningTimeList(java.io.BufferedReader in,
                                      TopicMap tm)
                               throws java.lang.Exception

Throws:

java.lang.Exception

extractReleaseDateList

public boolean extractReleaseDateList(java.io.BufferedReader in,
                                      TopicMap tm)
                               throws java.lang.Exception

Throws:

java.lang.Exception