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

update to jdk11, fix migration issues, some typos, deprecated apis, bump up Lucene to 8.11

Browse Source
This commit is contained in:
Konstantin Perikov
2021-12-10 18:15:47 +00:00
parent b2fec940d0
commit 96900e334e
32 changed files with 184 additions and 198 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
morph/pom.xml
View File

@ -6,7 +6,6 @@
<version>1.5</version>
</parent>
<modelVersion>4.0.0</modelVersion>
<groupId>org.apache.lucene.morphology</groupId>
<artifactId>morph</artifactId>
<name>morph</name>
<version>1.5</version>

8
morph/src/main/java/org/apache/lucene/morphology/BaseLetterDecoderEncoder.java
View File

@ -21,7 +21,7 @@ import java.util.ArrayList;
public abstract class BaseLetterDecoderEncoder implements LetterDecoderEncoder {
public int[] encodeToArray(String s) {
ArrayList<Integer> integers = new ArrayList<Integer>();
ArrayList<Integer> integers = new ArrayList<>();
while (s.length() > 6) {
integers.add(encode(s.substring(0, 6)));
s = s.substring(6);
@ -37,11 +37,11 @@ public abstract class BaseLetterDecoderEncoder implements LetterDecoderEncoder {
}
public String decodeArray(int[] array) {
String result = "";
StringBuilder result = new StringBuilder();
for (int i : array) {
result += decode(i);
result.append(decode(i));
}
return result;
return result.toString();
}
public boolean checkString(String word) {

14
morph/src/main/java/org/apache/lucene/morphology/Heuristic.java
View File

@ -16,6 +16,7 @@
package org.apache.lucene.morphology;
import java.io.Serializable;
import java.util.Objects;
public class Heuristic implements Serializable {
@ -26,10 +27,10 @@ public class Heuristic implements Serializable {
public Heuristic(String s) {
String[] strings = s.split("\\|");
actualSuffixLength = Byte.valueOf(strings[0]);
actualSuffixLength = Byte.parseByte(strings[0]);
actualNormalSuffix = strings[1];
formMorphInfo = Short.valueOf(strings[2]);
normalFormMorphInfo = Short.valueOf(strings[3]);
formMorphInfo = Short.parseShort(strings[2]);
normalFormMorphInfo = Short.parseShort(strings[3]);
}
public Heuristic(byte actualSuffixLength, String actualNormalSuffix, short formMorphInfo, short normalFormMorphInfo) {
@ -70,15 +71,12 @@ public class Heuristic implements Serializable {
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)
return false;
return true;
return Objects.equals(actualNormalSuffix, heuristic.actualNormalSuffix);
}
@Override
public int hashCode() {
int result = (int) actualSuffixLength;
int result = actualSuffixLength;
result = 31 * result + (actualNormalSuffix != null ? actualNormalSuffix.hashCode() : 0);
result = 31 * result + (int) formMorphInfo;
result = 31 * result + (int) normalFormMorphInfo;

14
morph/src/main/java/org/apache/lucene/morphology/LetterDecoderEncoder.java
View File

@ -17,17 +17,17 @@ package org.apache.lucene.morphology;
public interface LetterDecoderEncoder {
public Integer encode(String string);
Integer encode(String string);
public int[] encodeToArray(String s);
int[] encodeToArray(String s);
public String decodeArray(int[] array);
String decodeArray(int[] array);
public String decode(Integer suffixN);
String decode(Integer suffixN);
public boolean checkCharacter(char c);
boolean checkCharacter(char c);
public boolean checkString(String word);
boolean checkString(String word);
public String cleanString(String s);
String cleanString(String s);
}

10
morph/src/main/java/org/apache/lucene/morphology/LuceneMorphology.java
View File

@ -34,13 +34,13 @@ public class LuceneMorphology extends MorphologyImpl {
protected void readRules(BufferedReader bufferedReader) throws IOException {
String s;
Integer amount;
int amount;
s = bufferedReader.readLine();
amount = Integer.valueOf(s);
amount = Integer.parseInt(s);
rules = new Heuristic[amount][];
for (int i = 0; i < amount; i++) {
String s1 = bufferedReader.readLine();
Integer ruleLenght = Integer.valueOf(s1);
int ruleLenght = Integer.parseInt(s1);
Heuristic[] heuristics = new Heuristic[ruleLenght];
for (int j = 0; j < ruleLenght; j++) {
heuristics[j] = new Heuristic(bufferedReader.readLine());
@ -51,7 +51,7 @@ public class LuceneMorphology extends MorphologyImpl {
private Heuristic[] modeifyHeuristic(Heuristic[] heuristics) {
ArrayList<Heuristic> result = new ArrayList<Heuristic>();
ArrayList<Heuristic> result = new ArrayList<>();
for (Heuristic heuristic : heuristics) {
boolean isAdded = true;
for (Heuristic ch : result) {
@ -61,7 +61,7 @@ public class LuceneMorphology extends MorphologyImpl {
result.add(heuristic);
}
}
return result.toArray(new Heuristic[result.size()]);
return result.toArray(new Heuristic[0]);
}
public boolean checkString(String s) {

27
morph/src/main/java/org/apache/lucene/morphology/MorphologyImpl.java
View File

@ -17,6 +17,7 @@ package org.apache.lucene.morphology;
import java.io.*;
import java.nio.charset.StandardCharsets;
import java.util.ArrayList;
import java.util.List;
@ -47,7 +48,7 @@ public class MorphologyImpl implements Morphology {
}
public List<String> getNormalForms(String s) {
ArrayList<String> result = new ArrayList<String>();
ArrayList<String> result = new ArrayList<>();
int[] ints = decoderEncoder.encodeToArray(revertWord(s));
int ruleId = findRuleId(ints);
boolean notSeenEmptyString = true;
@ -64,7 +65,7 @@ public class MorphologyImpl implements Morphology {
}
public List<String> getMorphInfo(String s) {
ArrayList<String> result = new ArrayList<String>();
ArrayList<String> result = new ArrayList<>();
int[] ints = decoderEncoder.encodeToArray(revertWord(s));
int ruleId = findRuleId(ints);
for (Heuristic h : rules[rulesId[ruleId]]) {
@ -100,14 +101,14 @@ public class MorphologyImpl implements Morphology {
private int compareToInts(int[] i1, int[] i2) {
int minLength = Math.min(i1.length, i2.length);
for (int i = 0; i < minLength; i++) {
int i3 = i1[i] < i2[i] ? -1 : (i1[i] == i2[i] ? 0 : 1);
int i3 = Integer.compare(i1[i], i2[i]);
if (i3 != 0) return i3;
}
return i1.length - i2.length;
}
public void writeToFile(String fileName) throws IOException {
OutputStreamWriter writer = new OutputStreamWriter(new FileOutputStream(fileName), "UTF-8");
OutputStreamWriter writer = new OutputStreamWriter(new FileOutputStream(fileName), StandardCharsets.UTF_8);
writer.write(separators.length + "\n");
for (int[] i : separators) {
writer.write(i.length + "\n");
@ -138,7 +139,7 @@ public class MorphologyImpl implements Morphology {
}
private void readFromInputStream(InputStream inputStream) throws IOException {
BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(inputStream, "UTF-8"));
BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(inputStream, StandardCharsets.UTF_8));
String s = bufferedReader.readLine();
Integer amount = Integer.valueOf(s);
@ -153,9 +154,9 @@ public class MorphologyImpl implements Morphology {
private void readGrammaInfo(BufferedReader bufferedReader) throws IOException {
String s;
Integer amount;
int amount;
s = bufferedReader.readLine();
amount = Integer.valueOf(s);
amount = Integer.parseInt(s);
grammarInfo = new String[amount];
for (int i = 0; i < amount; i++) {
grammarInfo[i] = bufferedReader.readLine();
@ -164,13 +165,13 @@ public class MorphologyImpl implements Morphology {
protected void readRules(BufferedReader bufferedReader) throws IOException {
String s;
Integer amount;
int amount;
s = bufferedReader.readLine();
amount = Integer.valueOf(s);
amount = Integer.parseInt(s);
rules = new Heuristic[amount][];
for (int i = 0; i < amount; i++) {
String s1 = bufferedReader.readLine();
Integer ruleLength = Integer.valueOf(s1);
int ruleLength = Integer.parseInt(s1);
rules[i] = new Heuristic[ruleLength];
for (int j = 0; j < ruleLength; j++) {
rules[i][j] = new Heuristic(bufferedReader.readLine());
@ -182,7 +183,7 @@ public class MorphologyImpl implements Morphology {
rulesId = new short[amount];
for (int i = 0; i < amount; i++) {
String s1 = bufferedReader.readLine();
rulesId[i] = Short.valueOf(s1);
rulesId[i] = Short.parseShort(s1);
}
}
@ -190,10 +191,10 @@ public class MorphologyImpl implements Morphology {
separators = new int[amount][];
for (int i = 0; i < amount; i++) {
String s1 = bufferedReader.readLine();
Integer wordLenght = Integer.valueOf(s1);
int wordLenght = Integer.parseInt(s1);
separators[i] = new int[wordLenght];
for (int j = 0; j < wordLenght; j++) {
separators[i][j] = Integer.valueOf(bufferedReader.readLine());
separators[i][j] = Integer.parseInt(bufferedReader.readLine());
}
}
}

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

@ -73,6 +73,6 @@ public class MorphologyAnalyzer extends Analyzer {
TokenFilter filter = new LowerCaseFilter(src);
filter = new MorphologyFilter(filter, luceneMorph);
return new TokenStreamComponents(r -> src.setReader(r), filter);
return new TokenStreamComponents(src::setReader, filter);
}
}