org.dice_research.topicmodeling.wikipedia.markup

Class StackBasedMarkupDeletingMachine

java.lang.Object

org.dice_research.topicmodeling.wikipedia.markup.StackBasedMarkupDeletingMachine

All Implemented Interfaces:

MarkupDetectingAutomatonObserver

public class StackBasedMarkupDeletingMachine
extends Object
implements MarkupDetectingAutomatonObserver

Field Summary

Fields

Modifier and Type	Field and Description
private static long	AFTER_MAKRO_MARKUP_PATTERN
private static long	AFTER_NOWIKI_TAG_PATTERN
private static long	AFTER_TABLE_ROW_PATTERN
private MarkupDetectingAutomaton	automaton
private static long	INSIDE_WIKI_TABLE_PATTERN
private static long	INSIDE_XML_COMMENT_PATTERN
static boolean	KEEP_TABLE_CONTENTS_DEFAULT
private boolean	keepTableContents
private static org.slf4j.Logger	LOGGER
private static long	NORMAL_TEXT_PATTERN
static boolean	REMOVE_CATEGORY_LINKS_DEFAULT
private boolean	removeCategoryLinks

Constructor Summary

Constructors

Constructor and Description
StackBasedMarkupDeletingMachine()
StackBasedMarkupDeletingMachine(boolean keepTableContents)
StackBasedMarkupDeletingMachine(boolean keepTableContents, boolean removeCategoryLinks)
StackBasedMarkupDeletingMachine(MarkupDetectingAutomaton automaton, boolean keepTableContents, boolean removeCategoryLinks)

Method Summary

Modifier and Type	Method and Description
private void	appendText(CharSequence text, StackBasedMarkupDeletingMachineState state)
private void	appendText(StackBasedMarkupDeletingMachineState state, int startPos)
private void	cleanupStack(StackBasedMarkupDeletingMachineState state)
private void	closeNextClosableMarkup(StackBasedMarkupDeletingMachineState state)
private void	closeNextClosableMarkup(StackBasedMarkupDeletingMachineState state, StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup nextClosableMarkup)
void	foundPattern(int patternId, int startPos, int endPos, MarkupDetectingAutomatonObserverState observerState, MarkupDetectingAutomatonState automatonState)
private String	getCleanString(String text)
String	getCleanText(String text)
private StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup	getNextClosableMarkup(SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack)
private boolean	handleClosingMarkup(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type, StackBasedMarkupDeletingMachineState state)
private void	handleClosingMarkup(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type, StackBasedMarkupDeletingMachineState state, MarkupDetectingAutomatonState automatonState)
private void	handleTag(StackBasedMarkupDeletingMachineState state, MarkupDetectingAutomatonState automatonState, int startPos, int endPos)
private boolean	isOpeningMarkupOnStack(SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack, StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type)
static void	main(String[] args)
private void	removeFromStack(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type, SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack)
private void	removeIntermediateMarkupsFromStack(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType intermediateType, StackBasedMarkupDeletingMachineState state, boolean appendTheirText)
private void	removeIntermediateMarkupsFromStack(StackBasedMarkupDeletingMachineState state, boolean appendTheirText, List<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType> intermediateType)
void	setKeepTableContents(boolean keepTableContents)
private void	setPatternBitSet(com.carrotsearch.hppc.BitSet patternBitSet, SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack)
void	setRemoveCategoryLinks(boolean removeCategoryLinks)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

LOGGER

private static final org.slf4j.Logger LOGGER

REMOVE_CATEGORY_LINKS_DEFAULT

public static final boolean REMOVE_CATEGORY_LINKS_DEFAULT

See Also:

Constant Field Values

KEEP_TABLE_CONTENTS_DEFAULT

public static final boolean KEEP_TABLE_CONTENTS_DEFAULT

See Also:

Constant Field Values

NORMAL_TEXT_PATTERN

private static final long NORMAL_TEXT_PATTERN

See Also:

Constant Field Values

AFTER_MAKRO_MARKUP_PATTERN

private static final long AFTER_MAKRO_MARKUP_PATTERN

See Also:

Constant Field Values

AFTER_NOWIKI_TAG_PATTERN

private static final long AFTER_NOWIKI_TAG_PATTERN

See Also:

Constant Field Values

INSIDE_XML_COMMENT_PATTERN

private static final long INSIDE_XML_COMMENT_PATTERN

See Also:

Constant Field Values

INSIDE_WIKI_TABLE_PATTERN

private static final long INSIDE_WIKI_TABLE_PATTERN

See Also:

Constant Field Values

AFTER_TABLE_ROW_PATTERN

private static final long AFTER_TABLE_ROW_PATTERN

See Also:

Constant Field Values

automaton

private MarkupDetectingAutomaton automaton

removeCategoryLinks

private boolean removeCategoryLinks

keepTableContents

private boolean keepTableContents

Constructor Detail

StackBasedMarkupDeletingMachine

public StackBasedMarkupDeletingMachine()

StackBasedMarkupDeletingMachine

public StackBasedMarkupDeletingMachine(boolean keepTableContents)

StackBasedMarkupDeletingMachine

public StackBasedMarkupDeletingMachine(boolean keepTableContents,
                                       boolean removeCategoryLinks)

StackBasedMarkupDeletingMachine

public StackBasedMarkupDeletingMachine(MarkupDetectingAutomaton automaton,
                                       boolean keepTableContents,
                                       boolean removeCategoryLinks)

Method Detail

foundPattern

public void foundPattern(int patternId,
                         int startPos,
                         int endPos,
                         MarkupDetectingAutomatonObserverState observerState,
                         MarkupDetectingAutomatonState automatonState)

Specified by:

foundPattern in interface MarkupDetectingAutomatonObserver

isOpeningMarkupOnStack

private boolean isOpeningMarkupOnStack(SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack,
                                       StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type)

appendText

private void appendText(StackBasedMarkupDeletingMachineState state,
                        int startPos)

appendText

private void appendText(CharSequence text,
                        StackBasedMarkupDeletingMachineState state)

handleClosingMarkup

private void handleClosingMarkup(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type,
                                 StackBasedMarkupDeletingMachineState state,
                                 MarkupDetectingAutomatonState automatonState)

handleClosingMarkup

private boolean handleClosingMarkup(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type,
                                    StackBasedMarkupDeletingMachineState state)

getNextClosableMarkup

private StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup getNextClosableMarkup(SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack)

removeFromStack

private void removeFromStack(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType type,
                             SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack)

removeIntermediateMarkupsFromStack

private void removeIntermediateMarkupsFromStack(StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType intermediateType,
                                                StackBasedMarkupDeletingMachineState state,
                                                boolean appendTheirText)

removeIntermediateMarkupsFromStack

private void removeIntermediateMarkupsFromStack(StackBasedMarkupDeletingMachineState state,
                                                boolean appendTheirText,
                                                List<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkupType> intermediateType)

setPatternBitSet

private void setPatternBitSet(com.carrotsearch.hppc.BitSet patternBitSet,
                              SimpleStack<StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup> stack)

getCleanText

public String getCleanText(String text)

getCleanString

private String getCleanString(String text)

cleanupStack

private void cleanupStack(StackBasedMarkupDeletingMachineState state)

closeNextClosableMarkup

private void closeNextClosableMarkup(StackBasedMarkupDeletingMachineState state)

closeNextClosableMarkup

private void closeNextClosableMarkup(StackBasedMarkupDeletingMachineState state,
                                     StackBasedMarkupDeletingMachineState.WikipediaOpeningMarkup nextClosableMarkup)

handleTag

private void handleTag(StackBasedMarkupDeletingMachineState state,
                       MarkupDetectingAutomatonState automatonState,
                       int startPos,
                       int endPos)

main

public static void main(String[] args)

setRemoveCategoryLinks

public void setRemoveCategoryLinks(boolean removeCategoryLinks)

setKeepTableContents

public void setKeepTableContents(boolean keepTableContents)