enhanced identificator: using AtomicInteger for counter

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7785 6c8d7289-2bf4-0310-a012-ef5d649a1542
pull/1/head
orbiter 14 years ago
parent 16327d1cbe
commit f667b9c289

@ -29,6 +29,7 @@ import java.util.Iterator;
import java.util.List; import java.util.List;
import java.util.Map; import java.util.Map;
import java.util.Vector; import java.util.Vector;
import java.util.concurrent.atomic.AtomicInteger;
/** /**
* This class can try to identify the language a text is written in. * This class can try to identify the language a text is written in.
@ -37,14 +38,14 @@ public final class Identificator {
private static final LanguageStatisticsHolder languages = LanguageStatisticsHolder.getInstance(); private static final LanguageStatisticsHolder languages = LanguageStatisticsHolder.getInstance();
private final Map<Character, Integer> letter; private final Map<Character, AtomicInteger> letter;
private int letters; private int letters;
private String language; private String language;
public Identificator() { public Identificator() {
letter = new HashMap<Character, Integer>(); this.letter = new HashMap<Character, AtomicInteger>();
letters = 0; this.letters = 0;
language = null; this.language = null;
} }
/** /**
@ -73,7 +74,7 @@ public final class Identificator {
upperLimit = limit; upperLimit = limit;
} }
Identificator id = new Identificator(); final Identificator id = new Identificator();
// count number of characters in text // count number of characters in text
for (int i = 0; i < upperLimit; i++) id.inc(text.charAt(i)); for (int i = 0; i < upperLimit; i++) id.inc(text.charAt(i));
@ -83,25 +84,25 @@ public final class Identificator {
public void inc(final char c) { public void inc(final char c) {
if (!Character.isLetter(c)) return; if (!Character.isLetter(c)) return;
Character cc = Character.toLowerCase(c); final Character cc = Character.toLowerCase(c);
Integer i = letter.get(cc); final AtomicInteger i = this.letter.get(cc);
if (i == null) { if (i == null) {
letter.put(cc, 1); this.letter.put(cc, new AtomicInteger(1));
} else { } else {
letter.put(cc, i.intValue() + 1); i.incrementAndGet();
} }
letters++; this.letters++;
} }
public void add(String word) { public void add(final String word) {
if (word == null) return; if (word == null) return;
for (int i = 0; i < word.length(); i++) inc(word.charAt(i)); for (int i = 0; i < word.length(); i++) inc(word.charAt(i));
} }
public String getLanguage() { public String getLanguage() {
if (language != null) return language; // don't compute that twice if (this.language != null) return this.language; // don't compute that twice
if (letters == 0) return null; // not enough information available if (this.letters == 0) return null; // not enough information available
final LanguageStatistics testStat = new LanguageStatistics("text"); final LanguageStatistics testStat = new LanguageStatistics("text");
@ -110,14 +111,14 @@ public final class Identificator {
Character maxChar = null; Character maxChar = null;
int count = 0; int count = 0;
int max = 0; int max = 0;
for (Map.Entry<Character, Integer> e: letter.entrySet()) { for (final Map.Entry<Character, AtomicInteger> e: this.letter.entrySet()) {
character = e.getKey(); character = e.getKey();
count = e.getValue().intValue(); count = e.getValue().intValue();
if (count > max) { if (count > max) {
maxChar = character; maxChar = character;
max = count; max = count;
} }
testStat.put(character, ((float) 100) * ((float) count) / ((float) letters)); testStat.put(character, ((float) 100) * ((float) count) / (this.letters));
} }
// create list with relevant languages // create list with relevant languages
@ -136,7 +137,7 @@ public final class Identificator {
final float[] offsetList = new float[relevantLanguages.size()]; final float[] offsetList = new float[relevantLanguages.size()];
final int[] votesList = new int[relevantLanguages.size()]; final int[] votesList = new int[relevantLanguages.size()];
Iterator<Character> iter = testStat.keySet().iterator(); final Iterator<Character> iter = testStat.keySet().iterator();
float minimum; float minimum;
float offset = 0; float offset = 0;
float valueCharacter; float valueCharacter;
@ -175,8 +176,8 @@ public final class Identificator {
// prevents a language beeing reported that has won the voting, but // prevents a language beeing reported that has won the voting, but
// is still not the right language. // is still not the right language.
if (offset < 20) { if (offset < 20) {
language = languages.get(relevantLanguages.get(bestLanguage)).getName(); this.language = languages.get(relevantLanguages.get(bestLanguage)).getName();
return language; return this.language;
} }
return null; return null;

Loading…
Cancel
Save