edkuksa/russianmorphology
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fixing problem with - in word the english morphology not work correctly because word forms contains it

Browse Source 
git-svn-id: https://russianmorphology.googlecode.com/svn/trunk@97 d817d54c-26ab-11de-abc9-2f7d1455ff7a
This commit is contained in:
alexander.a.kuznetsov
2010-05-29 15:54:44 +00:00
parent 3ca9eb2cb9
commit 7bf8ef7d6f
12 changed files with 6393 additions and 6381 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
morph/src/main/java/org/apache/lucene/morphology/Heuristic.java
View File

@ -19,33 +19,33 @@ import java.io.Serializable;
public class Heuristic implements Serializable {
byte actualSuffixLengh;
byte actualSuffixLength;
String actualNormalSuffix;
short formMorphInfo;
short normalFormMorphInfo;
public Heuristic(String s) {
String[] strings = s.split("\\|");
actualSuffixLengh = Byte.valueOf(strings[0]);
actualSuffixLength = Byte.valueOf(strings[0]);
actualNormalSuffix = strings[1];
formMorphInfo = Short.valueOf(strings[2]);
normalFormMorphInfo = Short.valueOf(strings[3]);
}
public Heuristic(byte actualSuffixLengh, String actualNormalSuffix, short formMorphInfo, short normalFormMorphInfo) {
this.actualSuffixLengh = actualSuffixLengh;
public Heuristic(byte actualSuffixLength, String actualNormalSuffix, short formMorphInfo, short normalFormMorphInfo) {
this.actualSuffixLength = actualSuffixLength;
this.actualNormalSuffix = actualNormalSuffix;
this.formMorphInfo = formMorphInfo;
this.normalFormMorphInfo = normalFormMorphInfo;
}
public String transofrmWord(String w) {
if (w.length() - actualSuffixLengh < 0) return w;
return w.substring(0, w.length() - actualSuffixLengh) + actualNormalSuffix;
public String transformWord(String w) {
if (w.length() - actualSuffixLength < 0) return w;
return w.substring(0, w.length() - actualSuffixLength) + actualNormalSuffix;
}
public byte getActualSuffixLengh() {
return actualSuffixLengh;
public byte getActualSuffixLength() {
return actualSuffixLength;
}
public String getActualNormalSuffix() {
@ -67,7 +67,7 @@ public class Heuristic implements Serializable {
Heuristic heuristic = (Heuristic) o;
if (actualSuffixLengh != heuristic.actualSuffixLengh) return false;
if (actualSuffixLength != heuristic.actualSuffixLength) return false;
if (formMorphInfo != heuristic.formMorphInfo) return false;
if (normalFormMorphInfo != heuristic.normalFormMorphInfo) return false;
if (actualNormalSuffix != null ? !actualNormalSuffix.equals(heuristic.actualNormalSuffix) : heuristic.actualNormalSuffix != null)
@ -78,7 +78,7 @@ public class Heuristic implements Serializable {
@Override
public int hashCode() {
int result = (int) actualSuffixLengh;
int result = (int) actualSuffixLength;
result = 31 * result + (actualNormalSuffix != null ? actualNormalSuffix.hashCode() : 0);
result = 31 * result + (int) formMorphInfo;
result = 31 * result + (int) normalFormMorphInfo;
@ -87,6 +87,6 @@ public class Heuristic implements Serializable {
@Override
public String toString() {
return "" + actualSuffixLengh + "|" + actualNormalSuffix + "|" + formMorphInfo + "|" + normalFormMorphInfo;
return "" + actualSuffixLength + "|" + actualNormalSuffix + "|" + formMorphInfo + "|" + normalFormMorphInfo;
}
}

2
morph/src/main/java/org/apache/lucene/morphology/LuceneMorphology.java
View File

@ -55,7 +55,7 @@ public class LuceneMorphology extends MorphologyImpl {
for (Heuristic heuristic : heuristics) {
boolean isAdded = true;
for (Heuristic ch : result) {
isAdded = isAdded && !(ch.getActualNormalSuffix().equals(heuristic.getActualNormalSuffix()) && (ch.getActualSuffixLengh() == heuristic.getActualSuffixLengh()));
isAdded = isAdded && !(ch.getActualNormalSuffix().equals(heuristic.getActualNormalSuffix()) && (ch.getActualSuffixLength() == heuristic.getActualSuffixLength()));
}
if (isAdded) {
result.add(heuristic);

12
morph/src/main/java/org/apache/lucene/morphology/MorphologyImpl.java
View File

@ -68,7 +68,7 @@ public class MorphologyImpl implements Morphology {
int[] ints = decoderEncoder.encodeToArray(revertWord(s));
int ruleId = findRuleId(ints);
for (Heuristic h : rules[rulesId[ruleId]]) {
result.add(h.transofrmWord(s));
result.add(h.transformWord(s));
}
return result;
}
@ -78,7 +78,7 @@ public class MorphologyImpl implements Morphology {
int[] ints = decoderEncoder.encodeToArray(revertWord(s));
int ruleId = findRuleId(ints);
for (Heuristic h : rules[rulesId[ruleId]]) {
result.add(h.transofrmWord(s) + "|" + grammaInfo[h.getFormMorphInfo()]);
result.add(h.transformWord(s) + "|" + grammaInfo[h.getFormMorphInfo()]);
}
return result;
}
@ -180,9 +180,9 @@ public class MorphologyImpl implements Morphology {
rules = new Heuristic[amount][];
for (int i = 0; i < amount; i++) {
String s1 = bufferedReader.readLine();
Integer ruleLenght = Integer.valueOf(s1);
rules[i] = new Heuristic[ruleLenght];
for (int j = 0; j < ruleLenght; j++) {
Integer ruleLength = Integer.valueOf(s1);
rules[i] = new Heuristic[ruleLength];
for (int j = 0; j < ruleLength; j++) {
rules[i][j] = new Heuristic(bufferedReader.readLine());
}
}
@ -197,7 +197,6 @@ public class MorphologyImpl implements Morphology {
}
private void readSeparators(BufferedReader bufferedReader, Integer amount) throws IOException {
HashSet intetger = new HashSet<Integer>();
separators = new int[amount][];
for (int i = 0; i < amount; i++) {
String s1 = bufferedReader.readLine();
@ -206,7 +205,6 @@ public class MorphologyImpl implements Morphology {
for (int j = 0; j < wordLenght; j++) {
separators[i][j] = Integer.valueOf(bufferedReader.readLine());
}
intetger.add(separators[i][0]);
}
}