Marmelator/aho-corasick
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Updated source base to leverage JDK 1.7 syntax. Added more final modifiers. Eliminated parameter modification inside method. Some formatting. Changed TrieBuilder to offer CharSequence instead of String; revised Trie accordingly. Removed some duplication. NetBeans automatically translated the code to use static imports (as per JDK 1.7 syntax).

Browse Source
This commit is contained in:
djarvis 2016-11-29 18:38:00 -08:00
parent 2f1ec8d041
commit 2b5c2d654d
15 changed files with 180 additions and 172 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

60
src/main/java/org/ahocorasick/interval/IntervalNode.java
View File

@ -8,16 +8,16 @@ public class IntervalNode {
private enum Direction { LEFT, RIGHT }
private IntervalNode left = null;
private IntervalNode right = null;
private IntervalNode left;
private IntervalNode right;
private int point;
private List<Intervalable> intervals = new ArrayList<Intervalable>();
private List<Intervalable> intervals = new ArrayList<>();
public IntervalNode(List<Intervalable> intervals) {
public IntervalNode(final List<Intervalable> intervals) {
this.point = determineMedian(intervals);
List<Intervalable> toLeft = new ArrayList<Intervalable>();
List<Intervalable> toRight = new ArrayList<Intervalable>();
final List<Intervalable> toLeft = new ArrayList<>();
final List<Intervalable> toRight = new ArrayList<>();
for (Intervalable interval : intervals) {
if (interval.getEnd() < this.point) {
@ -37,7 +37,7 @@ public class IntervalNode {
}
}
public int determineMedian(List<Intervalable> intervals) {
public int determineMedian(final List<Intervalable> intervals) {
int start = -1;
int end = -1;
for (Intervalable interval : intervals) {
@ -53,17 +53,19 @@ public class IntervalNode {
return (start + end) / 2;
}
public List<Intervalable> findOverlaps(Intervalable interval) {
public List<Intervalable> findOverlaps(final Intervalable interval) {
final List<Intervalable> overlaps = new ArrayList<>();
List<Intervalable> overlaps = new ArrayList<Intervalable>();
if (this.point < interval.getStart()) { // Tends to the right
if (this.point < interval.getStart()) {
// Tends to the right
addToOverlaps(interval, overlaps, findOverlappingRanges(this.right, interval));
addToOverlaps(interval, overlaps, checkForOverlapsToTheRight(interval));
} else if (this.point > interval.getEnd()) { // Tends to the left
} else if (this.point > interval.getEnd()) {
// Tends to the left
addToOverlaps(interval, overlaps, findOverlappingRanges(this.left, interval));
addToOverlaps(interval, overlaps, checkForOverlapsToTheLeft(interval));
} else { // Somewhere in the middle
} else {
// Somewhere in the middle
addToOverlaps(interval, overlaps, this.intervals);
addToOverlaps(interval, overlaps, findOverlappingRanges(this.left, interval));
addToOverlaps(interval, overlaps, findOverlappingRanges(this.right, interval));
@ -72,26 +74,30 @@ public class IntervalNode {
return overlaps;
}
protected void addToOverlaps(Intervalable interval, List<Intervalable> overlaps, List<Intervalable> newOverlaps) {
for (Intervalable currentInterval : newOverlaps) {
protected void addToOverlaps(
final Intervalable interval,
final List<Intervalable> overlaps,
final List<Intervalable> newOverlaps) {
for (final Intervalable currentInterval : newOverlaps) {
if (!currentInterval.equals(interval)) {
overlaps.add(currentInterval);
}
}
}
protected List<Intervalable> checkForOverlapsToTheLeft(Intervalable interval) {
protected List<Intervalable> checkForOverlapsToTheLeft(final Intervalable interval) {
return checkForOverlaps(interval, Direction.LEFT);
}
protected List<Intervalable> checkForOverlapsToTheRight(Intervalable interval) {
protected List<Intervalable> checkForOverlapsToTheRight(final Intervalable interval) {
return checkForOverlaps(interval, Direction.RIGHT);
}
protected List<Intervalable> checkForOverlaps(Intervalable interval, Direction direction) {
List<Intervalable> overlaps = new ArrayList<Intervalable>();
for (Intervalable currentInterval : this.intervals) {
protected List<Intervalable> checkForOverlaps(
final Intervalable interval, final Direction direction) {
final List<Intervalable> overlaps = new ArrayList<>();
for (final Intervalable currentInterval : this.intervals) {
switch (direction) {
case LEFT :
if (currentInterval.getStart() <= interval.getEnd()) {
@ -105,15 +111,13 @@ public class IntervalNode {
break;
}
}
return overlaps;
}
protected List<Intervalable> findOverlappingRanges(IntervalNode node, Intervalable interval) {
if (node != null) {
return node.findOverlaps(interval);
}
return Collections.emptyList();
return node == null
? Collections.<Intervalable>emptyList()
: node.findOverlaps( interval );
}
}

18
src/main/java/org/ahocorasick/interval/IntervalTree.java
View File

@ -1,26 +1,26 @@
package org.ahocorasick.interval;
import java.util.Collections;
import static java.util.Collections.sort;
import java.util.List;
import java.util.Set;
import java.util.TreeSet;
public class IntervalTree {
private IntervalNode rootNode = null;
private final IntervalNode rootNode;
public IntervalTree(List<Intervalable> intervals) {
this.rootNode = new IntervalNode(intervals);
}
public List<Intervalable> removeOverlaps(List<Intervalable> intervals) {
public List<Intervalable> removeOverlaps(final List<Intervalable> intervals) {
// Sort the intervals on size, then left-most position
Collections.sort(intervals, new IntervalableComparatorBySize());
sort(intervals, new IntervalableComparatorBySize());
Set<Intervalable> removeIntervals = new TreeSet<Intervalable>();
final Set<Intervalable> removeIntervals = new TreeSet<>();
for (Intervalable interval : intervals) {
for (final Intervalable interval : intervals) {
// If the interval was already removed, ignore it
if (removeIntervals.contains(interval)) {
continue;
@ -31,17 +31,17 @@ public class IntervalTree {
}
// Remove all intervals that were overlapping
for (Intervalable removeInterval : removeIntervals) {
for (final Intervalable removeInterval : removeIntervals) {
intervals.remove(removeInterval);
}
// Sort the intervals, now on left-most position only
Collections.sort(intervals, new IntervalableComparatorByPosition());
sort(intervals, new IntervalableComparatorByPosition());
return intervals;
}
public List<Intervalable> findOverlaps(Intervalable interval) {
public List<Intervalable> findOverlaps(final Intervalable interval) {
return rootNode.findOverlaps(interval);
}

1
src/main/java/org/ahocorasick/interval/Intervalable.java
View File

@ -5,5 +5,4 @@ public interface Intervalable extends Comparable {
public int getStart();
public int getEnd();
public int size();
}

2
src/main/java/org/ahocorasick/interval/IntervalableComparatorByPosition.java
View File

@ -5,7 +5,7 @@ import java.util.Comparator;
public class IntervalableComparatorByPosition implements Comparator<Intervalable> {
@Override
public int compare(Intervalable intervalable, Intervalable intervalable2) {
public int compare(final Intervalable intervalable, final Intervalable intervalable2) {
return intervalable.getStart() - intervalable2.getStart();
}

4
src/main/java/org/ahocorasick/interval/IntervalableComparatorBySize.java
View File

@ -5,11 +5,13 @@ import java.util.Comparator;
public class IntervalableComparatorBySize implements Comparator<Intervalable> {
@Override
public int compare(Intervalable intervalable, Intervalable intervalable2) {
public int compare(final Intervalable intervalable, final Intervalable intervalable2) {
int comparison = intervalable2.size() - intervalable.size();
if (comparison == 0) {
comparison = intervalable.getStart() - intervalable2.getStart();
}
return comparison;
}

1
src/main/java/org/ahocorasick/trie/Emit.java
View File

@ -20,5 +20,4 @@ public class Emit extends Interval implements Intervalable {
public String toString() {
return super.toString() + "=" + this.keyword;
}
}

5
src/main/java/org/ahocorasick/trie/MatchToken.java
View File

@ -2,9 +2,9 @@ package org.ahocorasick.trie;
public class MatchToken extends Token {
private Emit emit;
private final Emit emit;
public MatchToken(String fragment, Emit emit) {
public MatchToken(final String fragment, final Emit emit) {
super(fragment);
this.emit = emit;
}
@ -18,5 +18,4 @@ public class MatchToken extends Token {
public Emit getEmit() {
return this.emit;
}
}

10
src/main/java/org/ahocorasick/trie/State.java
View File

@ -66,11 +66,11 @@ public class State {
return nextState(character, false);
}
public State nextStateIgnoreRootState(Character character) {
public State nextStateIgnoreRootState(final Character character) {
return nextState(character, true);
}
public State addState( String keyword ) {
public State addState(final String keyword ) {
State state = this;
for (final Character character : keyword.toCharArray()) {
@ -80,7 +80,7 @@ public class State {
return state;
}
public State addState(Character character) {
public State addState(final Character character) {
State nextState = nextStateIgnoreRootState(character);
if (nextState == null) {
nextState = new State(this.depth+1);
@ -93,14 +93,14 @@ public class State {
return this.depth;
}
public void addEmit(String keyword) {
public void addEmit(final String keyword) {
if (this.emits == null) {
this.emits = new TreeSet<>();
}
this.emits.add(keyword);
}
public void addEmit(Collection<String> emits) {
public void addEmit(final Collection<String> emits) {
for (String emit : emits) {
addEmit(emit);
}

148
src/main/java/org/ahocorasick/trie/Trie.java
View File

@ -1,6 +1,8 @@
package org.ahocorasick.trie;
import static java.lang.Character.isAlphabetic;
import static java.lang.Character.isWhitespace;
import static java.lang.Character.toLowerCase;
import java.util.ArrayList;
import java.util.Collection;
import java.util.List;
@ -35,42 +37,18 @@ public class Trie {
*
* @throws NullPointerException if the keyword is null.
*/
private void addKeyword(String keyword) {
if( keyword.isEmpty() ) {
return;
private void addKeyword( String keyword ) {
if( keyword.length() > 0 ) {
if( isCaseInsensitive() ) {
keyword = keyword.toLowerCase();
}
addState( keyword ).addEmit( keyword );
}
if( isCaseInsensitive() ) {
keyword = keyword.toLowerCase();
}
addState(keyword).addEmit(keyword);
}
/**
* Delegates to addKeyword.
*
* @param keywords List of search term to add to the list of search terms.
*/
private void addKeywords( final String[] keywords ) {
for( final String keyword : keywords ) {
addKeyword( keyword );
}
}
/**
* Delegates to addKeyword.
*
* @param keywords List of search term to add to the list of search terms.
*/
private void addKeywords( final Collection<String> keywords ) {
for( final String keyword : keywords ) {
addKeyword( keyword );
}
}
private State addState(final String keyword) {
return getRootState().addState(keyword);
private State addState( final String keyword ) {
return getRootState().addState( keyword );
}
public Collection<Token> tokenize(final String text) {
@ -94,11 +72,14 @@ public class Trie {
return tokens;
}
private Token createFragment(final Emit emit, final String text, final int lastCollectedPosition) {
private Token createFragment(
final Emit emit,
final String text,
final int lastCollectedPosition) {
return new FragmentToken(text.substring(lastCollectedPosition+1, emit == null ? text.length() : emit.getStart()));
}
private Token createMatch(Emit emit, String text) {
private Token createMatch(final Emit emit, final String text) {
return new MatchToken(text.substring(emit.getStart(), emit.getEnd()+1), emit);
}
@ -137,7 +118,7 @@ public class Trie {
// TODO: Maybe lowercase the entire string at once?
if (trieConfig.isCaseInsensitive()) {
character = Character.toLowerCase(character);
character = toLowerCase(character);
}
currentState = getState(currentState, character);
@ -163,14 +144,14 @@ public class Trie {
// TODO: Lowercase the entire string at once?
if (trieConfig.isCaseInsensitive()) {
character = Character.toLowerCase(character);
character = toLowerCase(character);
}
currentState = getState(currentState, character);
Collection<String> emitStrs = currentState.emit();
if (emitStrs != null && !emitStrs.isEmpty()) {
for (String emitStr : emitStrs) {
for (final String emitStr : emitStrs) {
final Emit emit = new Emit(position - emitStr.length() + 1, position, emitStr);
if (trieConfig.isOnlyWholeWords()) {
if (!isPartialMatch(text, emit)) {
@ -189,9 +170,9 @@ public class Trie {
private boolean isPartialMatch(final CharSequence searchText, final Emit emit) {
return (emit.getStart() != 0 &&
Character.isAlphabetic(searchText.charAt(emit.getStart() - 1))) ||
isAlphabetic(searchText.charAt(emit.getStart() - 1))) ||
(emit.getEnd() + 1 != searchText.length() &&
Character.isAlphabetic(searchText.charAt(emit.getEnd() + 1)));
isAlphabetic(searchText.charAt(emit.getEnd() + 1)));
}
private void removePartialMatches(final CharSequence searchText, final List<Emit> collectedEmits) {
@ -225,15 +206,16 @@ public class Trie {
}
}
private State getState(State currentState, final Character character) {
State newCurrentState = currentState.nextState(character);
private State getState(final State initialState, final Character character) {
State currentState = initialState;
State updatedState = currentState.nextState(character);
while (newCurrentState == null) {
while (updatedState == null) {
currentState = currentState.failure();
newCurrentState = currentState.nextState(character);
updatedState = currentState.nextState(character);
}
return newCurrentState;
return updatedState;
}
private void constructFailureStates() {
@ -266,7 +248,10 @@ public class Trie {
}
}
private boolean storeEmits(final int position, final State currentState, final EmitHandler emitHandler) {
private boolean storeEmits(
final int position,
final State currentState,
final EmitHandler emitHandler) {
boolean emitted = false;
final Collection<String> emits = currentState.emit();
@ -290,7 +275,8 @@ public class Trie {
}
/**
* Provides a fluent interface for constructing Trie instances.
* Constructs a TrieBuilder instance for configuring the Trie using a fluent
* interface.
*
* @return The builder used to configure its Trie.
*/
@ -298,6 +284,9 @@ public class Trie {
return new TrieBuilder();
}
/**
* Provides a fluent interface for constructing Trie instances.
*/
public static class TrieBuilder {
private final TrieConfig trieConfig = new TrieConfig();
@ -317,11 +306,11 @@ public class Trie {
* @return This builder.
* @throws NullPointerException if the keyword is null.
*/
public TrieBuilder addKeyword(final String keyword) {
this.trie.addKeyword(keyword);
public TrieBuilder addKeyword(final CharSequence keyword) {
getTrie().addKeyword( keyword.toString() );
return this;
}
/**
* Adds a list of keywords to the Trie's list of text search keywords.
*
@ -329,9 +318,12 @@ public class Trie {
*
* @return This builder.
*/
public TrieBuilder addKeywords(final String... keywords) {
this.trie.addKeywords(keywords);
return this;
public TrieBuilder addKeywords(final CharSequence... keywords) {
for( final CharSequence keyword : keywords ) {
addKeyword( keyword );
}
return this;
}
/**
@ -341,19 +333,18 @@ public class Trie {
*
* @return This builder.
*/
public TrieBuilder addKeywords(final Collection<String> keywords) {
this.trie.addKeywords(keywords);
return this;
public TrieBuilder addKeywords(final Collection<CharSequence> keywords) {
return addKeywords( keywords.toArray( new CharSequence[ keywords.size() ] ) );
}
/**
* Configure the Trie to ignore case when searching for keywords in
* the text.
* Configure the Trie to ignore case when searching for keywords in the
* text.
*
* @return This builder.
*/
public TrieBuilder ignoreCase() {
this.trieConfig.setCaseInsensitive(true);
getTrieConfig().setCaseInsensitive(true);
return this;
}
@ -363,7 +354,7 @@ public class Trie {
* @return This builder.
*/
public TrieBuilder ignoreOverlaps() {
this.trieConfig.setAllowOverlaps(false);
getTrieConfig().setAllowOverlaps(false);
return this;
}
@ -373,7 +364,7 @@ public class Trie {
* @return This builder.
*/
public TrieBuilder onlyWholeWords() {
this.trieConfig.setOnlyWholeWords(true);
getTrieConfig().setOnlyWholeWords(true);
return this;
}
@ -385,35 +376,48 @@ public class Trie {
* @return This builder.
*/
public TrieBuilder onlyWholeWordsWhiteSpaceSeparated() {
this.trieConfig.setOnlyWholeWordsWhiteSpaceSeparated(true);
getTrieConfig().setOnlyWholeWordsWhiteSpaceSeparated(true);
return this;
}
/**
* Configure the Trie to stop after the first keyword is found in the
* text.
* Configure the Trie to stop searching for matches after the first
* keyword is found in the text.
*
* @return This builder.
*/
public TrieBuilder stopOnHit() {
trie.trieConfig.setStopOnHit(true);
public TrieBuilder onlyFirstMatch() {
getTrieConfig().setStopOnHit(true);
return this;
}
/**
* Configure the Trie based on the builder settings.
* Construct the Trie using the builder settings.
*
* @return The configured Trie.
*/
public Trie build() {
this.trie.constructFailureStates();
getTrie().constructFailureStates();
return getTrie();
}
private Trie getTrie() {
return this.trie;
}
private TrieConfig getTrieConfig() {
return this.trieConfig;
}
/**
* @deprecated Use onlyFirstMatch()
*/
public TrieBuilder stopOnHit() {
return onlyFirstMatch();
}
/**
* @deprecated Use ignoreCase()
*
* @return This builder.
*/
public TrieBuilder caseInsensitive() {
return ignoreCase();
@ -421,8 +425,6 @@ public class Trie {
/**
* @deprecated Use ignoreOverlaps()
*
* @return This builder.
*/
public TrieBuilder removeOverlaps() {
return ignoreOverlaps();

8
src/main/java/org/ahocorasick/trie/handler/DefaultEmitHandler.java
View File

@ -1,21 +1,19 @@
package org.ahocorasick.trie.handler;
import org.ahocorasick.trie.Emit;
import java.util.ArrayList;
import java.util.List;
import org.ahocorasick.trie.Emit;
public class DefaultEmitHandler implements EmitHandler {
private List<Emit> emits = new ArrayList<>();
private final List<Emit> emits = new ArrayList<>();
@Override
public void emit(Emit emit) {
public void emit(final Emit emit) {
this.emits.add(emit);
}
public List<Emit> getEmits() {
return this.emits;
}
}

2
src/test/java/org/ahocorasick/interval/IntervalTest.java
View File

@ -44,7 +44,7 @@ public class IntervalTest {
@Test
public void comparable() {
Set<Interval> intervals = new TreeSet<Interval>();
Set<Interval> intervals = new TreeSet<>();
intervals.add(new Interval(4, 6));
intervals.add(new Interval(2, 7));
intervals.add(new Interval(3, 4));

4
src/test/java/org/ahocorasick/interval/IntervalTreeTest.java
View File

@ -12,7 +12,7 @@ public class IntervalTreeTest {
@Test
public void findOverlaps() {
List<Intervalable> intervals = new ArrayList<Intervalable>();
List<Intervalable> intervals = new ArrayList<>();
intervals.add(new Interval(0, 2));
intervals.add(new Interval(1, 3));
intervals.add(new Interval(2, 4));
@ -30,7 +30,7 @@ public class IntervalTreeTest {
@Test
public void removeOverlaps() {
List<Intervalable> intervals = new ArrayList<Intervalable>();
List<Intervalable> intervals = new ArrayList<>();
intervals.add(new Interval(0, 2));
intervals.add(new Interval(4, 5));
intervals.add(new Interval(2, 10));

5
src/test/java/org/ahocorasick/interval/IntervalableComparatorByPositionTest.java
View File

@ -4,6 +4,7 @@ import org.junit.Test;
import java.util.ArrayList;
import java.util.Collections;
import static java.util.Collections.sort;
import java.util.List;
import static junit.framework.Assert.assertEquals;
@ -12,11 +13,11 @@ public class IntervalableComparatorByPositionTest {
@Test
public void sortOnPosition() {
List<Intervalable> intervals = new ArrayList<Intervalable>();
List<Intervalable> intervals = new ArrayList<>();
intervals.add(new Interval(4,5));
intervals.add(new Interval(1,4));
intervals.add(new Interval(3,8));
Collections.sort(intervals, new IntervalableComparatorByPosition());
sort(intervals, new IntervalableComparatorByPosition());
assertEquals(4, intervals.get(0).size());
assertEquals(6, intervals.get(1).size());
assertEquals(2, intervals.get(2).size());

10
src/test/java/org/ahocorasick/interval/IntervalableComparatorBySizeTest.java
View File

@ -4,6 +4,8 @@ import org.junit.Test;
import java.util.ArrayList;
import java.util.Collections;
import static java.util.Collections.sort;
import static java.util.Collections.sort;
import java.util.List;
import static junit.framework.Assert.assertEquals;
@ -12,11 +14,11 @@ public class IntervalableComparatorBySizeTest {
@Test
public void sortOnSize() {
List<Intervalable> intervals = new ArrayList<Intervalable>();
List<Intervalable> intervals = new ArrayList<>();
intervals.add(new Interval(4,5));
intervals.add(new Interval(1,4));
intervals.add(new Interval(3,8));
Collections.sort(intervals, new IntervalableComparatorBySize());
sort(intervals, new IntervalableComparatorBySize());
assertEquals(6, intervals.get(0).size());
assertEquals(4, intervals.get(1).size());
assertEquals(2, intervals.get(2).size());
@ -24,10 +26,10 @@ public class IntervalableComparatorBySizeTest {
@Test
public void sortOnSizeThenPosition() {
List<Intervalable> intervals = new ArrayList<Intervalable>();
List<Intervalable> intervals = new ArrayList<>();
intervals.add(new Interval(4,7));
intervals.add(new Interval(2,5));
Collections.sort(intervals, new IntervalableComparatorBySize());
sort(intervals, new IntervalableComparatorBySize());
assertEquals(2, intervals.get(0).getStart());
assertEquals(4, intervals.get(1).getStart());
}

74
src/test/java/org/ahocorasick/trie/TrieTest.java
View File

@ -5,7 +5,9 @@ import java.util.Collection;
import java.util.Iterator;
import java.util.List;
import java.util.concurrent.ThreadLocalRandom;
import static java.util.concurrent.ThreadLocalRandom.current;
import static junit.framework.Assert.assertEquals;
import static org.ahocorasick.trie.Trie.builder;
import org.ahocorasick.trie.handler.EmitHandler;
import static org.junit.Assert.assertTrue;
import org.junit.Test;
@ -33,7 +35,7 @@ public class TrieTest {
@Test
public void keywordAndTextAreTheSame() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeyword(ALPHABET[0])
.build();
Collection<Emit> emits = trie.parseText(ALPHABET[0]);
@ -43,7 +45,7 @@ public class TrieTest {
@Test
public void keywordAndTextAreTheSameFirstMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeyword(ALPHABET[0])
.build();
Emit firstMatch = trie.firstMatch(ALPHABET[0]);
@ -52,7 +54,7 @@ public class TrieTest {
@Test
public void textIsLongerThanKeyword() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeyword(ALPHABET[0])
.build();
Collection<Emit> emits = trie.parseText(" " + ALPHABET[0]);
@ -62,7 +64,7 @@ public class TrieTest {
@Test
public void textIsLongerThanKeywordFirstMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeyword(ALPHABET[0])
.build();
Emit firstMatch = trie.firstMatch(" " + ALPHABET[0]);
@ -71,7 +73,7 @@ public class TrieTest {
@Test
public void variousKeywordsOneMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(ALPHABET)
.build();
Collection<Emit> emits = trie.parseText("bcd");
@ -81,7 +83,7 @@ public class TrieTest {
@Test
public void variousKeywordsFirstMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(ALPHABET)
.build();
Emit firstMatch = trie.firstMatch("bcd");
@ -90,7 +92,7 @@ public class TrieTest {
@Test
public void ushersTestAndStopOnHit() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(PRONOUNS)
.stopOnHit()
.build();
@ -103,7 +105,7 @@ public class TrieTest {
@Test
public void ushersTest() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(PRONOUNS)
.build();
Collection<Emit> emits = trie.parseText("ushers");
@ -116,7 +118,7 @@ public class TrieTest {
@Test
public void ushersTestWithCapitalKeywords() {
Trie trie = Trie.builder()
Trie trie = builder()
.ignoreCase()
.addKeyword("HERS")
.addKeyword("HIS")
@ -133,7 +135,7 @@ public class TrieTest {
@Test
public void ushersTestFirstMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(PRONOUNS)
.build();
Emit firstMatch = trie.firstMatch("ushers");
@ -142,7 +144,7 @@ public class TrieTest {
@Test
public void ushersTestByCallback() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(PRONOUNS)
.build();
@ -164,7 +166,7 @@ public class TrieTest {
@Test
public void misleadingTest() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeyword("hers")
.build();
Collection<Emit> emits = trie.parseText("h he her hers");
@ -174,7 +176,7 @@ public class TrieTest {
@Test
public void misleadingTestFirstMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeyword("hers")
.build();
Emit firstMatch = trie.firstMatch("h he her hers");
@ -183,7 +185,7 @@ public class TrieTest {
@Test
public void recipes() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(FOOD)
.build();
Collection<Emit> emits = trie.parseText("2 cauliflowers, 3 tomatoes, 4 slices of veal, 100g broccoli");
@ -196,7 +198,7 @@ public class TrieTest {
@Test
public void recipesFirstMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(FOOD)
.build();
Emit firstMatch = trie.firstMatch("2 cauliflowers, 3 tomatoes, 4 slices of veal, 100g broccoli");
@ -206,7 +208,7 @@ public class TrieTest {
@Test
public void longAndShortOverlappingMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeyword("he")
.addKeyword("hehehehe")
.build();
@ -223,7 +225,7 @@ public class TrieTest {
@Test
public void nonOverlapping() {
Trie trie = Trie.builder().removeOverlaps()
Trie trie = builder().removeOverlaps()
.addKeyword("ab")
.addKeyword("cba")
.addKeyword("ababc")
@ -238,7 +240,7 @@ public class TrieTest {
@Test
public void nonOverlappingFirstMatch() {
Trie trie = Trie.builder().removeOverlaps()
Trie trie = builder().removeOverlaps()
.addKeyword("ab")
.addKeyword("cba")
.addKeyword("ababc")
@ -250,7 +252,7 @@ public class TrieTest {
@Test
public void containsMatch() {
Trie trie = Trie.builder().removeOverlaps()
Trie trie = builder().removeOverlaps()
.addKeyword("ab")
.addKeyword("cba")
.addKeyword("ababc")
@ -260,7 +262,7 @@ public class TrieTest {
@Test
public void startOfChurchillSpeech() {
Trie trie = Trie.builder().removeOverlaps()
Trie trie = builder().removeOverlaps()
.addKeyword("T")
.addKeyword("u")
.addKeyword("ur")
@ -278,7 +280,7 @@ public class TrieTest {
@Test
public void partialMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.onlyWholeWords()
.addKeyword("sugar")
.build();
@ -289,7 +291,7 @@ public class TrieTest {
@Test
public void partialMatchFirstMatch() {
Trie trie = Trie.builder()
Trie trie = builder()
.onlyWholeWords()
.addKeyword("sugar")
.build();
@ -300,7 +302,7 @@ public class TrieTest {
@Test
public void tokenizeFullSentence() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(GREEK_LETTERS)
.build();
Collection<Token> tokens = trie.tokenize("Hear: Alpha team first, Beta from the rear, Gamma in reserve");
@ -318,7 +320,7 @@ public class TrieTest {
// @see https://github.com/robert-bor/aho-corasick/issues/5
@Test
public void testStringIndexOutOfBoundsException() {
Trie trie = Trie.builder().ignoreCase().onlyWholeWords()
Trie trie = builder().ignoreCase().onlyWholeWords()
.addKeywords(UNICODE)
.build();
Collection<Emit> emits = trie.parseText("TurninG OnCe AgAiN BÖRKÜ");
@ -332,7 +334,7 @@ public class TrieTest {
@Test
public void testIgnoreCase() {
Trie trie = Trie.builder().ignoreCase()
Trie trie = builder().ignoreCase()
.addKeywords(UNICODE)
.build();
Collection<Emit> emits = trie.parseText("TurninG OnCe AgAiN BÖRKÜ");
@ -346,7 +348,7 @@ public class TrieTest {
@Test
public void testIgnoreCaseFirstMatch() {
Trie trie = Trie.builder().ignoreCase()
Trie trie = builder().ignoreCase()
.addKeywords(UNICODE)
.build();
Emit firstMatch = trie.firstMatch("TurninG OnCe AgAiN BÖRKÜ");
@ -356,7 +358,7 @@ public class TrieTest {
@Test
public void tokenizeTokensInSequence() {
Trie trie = Trie.builder()
Trie trie = builder()
.addKeywords(GREEK_LETTERS)
.build();
Collection<Token> tokens = trie.tokenize("Alpha Beta Gamma");
@ -366,7 +368,7 @@ public class TrieTest {
// @see https://github.com/robert-bor/aho-corasick/issues/7
@Test
public void testZeroLength() {
Trie trie = Trie.builder().ignoreOverlaps().onlyWholeWords().ignoreCase()
Trie trie = builder().ignoreOverlaps().onlyWholeWords().ignoreCase()
.addKeyword("")
.build();
trie.tokenize("Try a natural lip and subtle bronzer to keep all the focus on those big bright eyes with NARS Eyeshadow Duo in Rated R And the winner is... Boots No7 Advanced Renewal Anti-ageing Glycolic Peel Kit ($25 amazon.com) won most-appealing peel.");
@ -377,7 +379,7 @@ public class TrieTest {
public void testUnicode1() {
String target = "LİKE THIS"; // The second character ('İ') is Unicode, which was read by AC as a 2-byte char
assertEquals("THIS", target.substring(5, 9)); // Java does it the right way
Trie trie = Trie.builder().ignoreCase().onlyWholeWords()
Trie trie = builder().ignoreCase().onlyWholeWords()
.addKeyword("this")
.build();
Collection<Emit> emits = trie.parseText(target);
@ -390,7 +392,7 @@ public class TrieTest {
@Test
public void testUnicode2() {
String target = "LİKE THIS"; // The second character ('İ') is Unicode, which was read by AC as a 2-byte char
Trie trie = Trie.builder()
Trie trie = builder()
.ignoreCase()
.onlyWholeWords()
.addKeyword("this")
@ -402,7 +404,7 @@ public class TrieTest {
@Test
public void testPartialMatchWhiteSpaces() {
Trie trie = Trie.builder()
Trie trie = builder()
.onlyWholeWordsWhiteSpaceSeparated()
.addKeyword("#sugar-123")
.build();
@ -420,7 +422,7 @@ public class TrieTest {
injectKeyword( text, keyword, interval );
Trie trie = Trie.builder()
Trie trie = builder()
.onlyWholeWords()
.addKeyword( keyword )
.build();
@ -436,10 +438,10 @@ public class TrieTest {
* @param count The number of numbers to generate.
* @return A character sequence filled with random digits.
*/
private StringBuilder randomNumbers( int count ) {
private StringBuilder randomNumbers( final int count ) {
final StringBuilder sb = new StringBuilder( count );
while( --count > 0 ) {
for( int i = count - 1; i >= 0; i-- ) {
sb.append( randomInt( 0, 10 ) );
}
@ -465,7 +467,7 @@ public class TrieTest {
}
private int randomInt( final int min, final int max ) {
return ThreadLocalRandom.current().nextInt( min, max );
return current().nextInt( min, max );
}
private void checkEmit(Emit next, int expectedStart, int expectedEnd, String expectedKeyword) {