/**
* @version $Id: SyntaxStructure.java 1839 2014-04-16 02:33:51Z yukihiro-kinjyo $
*
* 2011/09/02 20:34:11
* @author imai
*
* Copyright 2011-2014 TIDAコンソーシアム All Rights Reserved.
*/
package com.tida_okinawa.corona.correction.morphem;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import org.eclipse.ui.views.properties.IPropertyDescriptor;
import org.eclipse.ui.views.properties.IPropertySource;
import org.eclipse.ui.views.properties.TextPropertyDescriptor;
import com.tida_okinawa.corona.correction.common.StringUtil;
import com.tida_okinawa.corona.io.model.MorphemeElement;
import com.tida_okinawa.corona.io.model.dic.TermClass;
/**
* 形態素・係り受け解析結果
*
* @author imai
*
*/
public class SyntaxStructure extends ArrayList<ISyntaxStructureElement> {
/**
* オブジェクトのシリアライズ用ID
*/
private static final long serialVersionUID = -1797704075541229412L;
/**
* Juman-KNP の出力のまま(デバッグ用)
*/
final List<String> original;
/**
* 係り先
*/
private final Map<Integer, Relation> destsMap = new HashMap<Integer, Relation>();
/**
* 係り元
*/
private final Map<Integer, List<Relation>> sourcesMap = new HashMap<Integer, List<Relation>>();
/**
* １つのテキストを、句読点で区切ってknpにかけるのに対応
*/
int index0 = 0;
int indexSentence = 0;
/**
* 句/文節を作る
*
* @param text
* KNPの結果のテキスト ex. *1 <文頭> ...
* @return 文節/句
*/
private ISyntaxStructureElement createElement(String text) {
int index = size();
ISyntaxStructureElement newElement = new SyntaxStructureElement(index, text);
super.add(newElement);
return newElement;
}
/**
* コンストラクタ
*
* @param text
* KNPの解析結果
*/
public SyntaxStructure(String text) {
this(StringUtil.splitFast(text));
// note: DBには 解析結果(List<String>)を連結したものを入れている
}
/**
* コンストラクタ
*
* @param texts
* KNPの解析結果
*/
public SyntaxStructure(List<String> texts) {
this(texts, "+"); //$NON-NLS-1$
}
/**
* 形態素のリストを取得
*
* @return 形態素
*/
public List<MorphemeElement> getMorphemeElemsnts() {
List<MorphemeElement> list = new ArrayList<MorphemeElement>();
for (ISyntaxStructureElement element : this) {
list.addAll(element.getMorphemes());
}
return list;
}
/**
* 形態素→文節/句
*
* @param me
* @return 文節/句
*/
public ISyntaxStructureElement getSyntaxStructureElement(MorphemeElement me) {
for (ISyntaxStructureElement element : this) {
if (element.getMorphemes().contains(me)) {
return element;
}
}
// never
return null;
}
/**
* コンストラクタ
*
* @param texts
* KNPの解析結果
* @param type
* 対象 "*" or "+"
*/
private SyntaxStructure(List<String> texts, String type) {
original = texts;
/**
* 処理中の句/文節
*/
ISyntaxStructureElement lastElement = null;
for (String text : texts) {
if (text.startsWith("#")) { //$NON-NLS-1$
// コメント行
} else if (text.startsWith("@")) { //$NON-NLS-1$
// Juman 次点候補 （無視する）
} else if (text.startsWith(type)) {
lastElement = createElement(text);
Relation relation = createRelation(lastElement, text);
if (relation.getDest() >= 0) { // 係り先がなければ -1
destsMap.put(lastElement.getIndex(), relation);
List<Relation> srcs = sourcesMap.get(relation.getDest());
if (srcs == null) {
srcs = new ArrayList<Relation>();
sourcesMap.put(relation.getDest(), srcs);
}
srcs.add(relation);
}
} else if (text.startsWith("*") || text.startsWith("+") || text.trim().equals("")) { //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
// 対象外の行, 空白行
// type ("*" or "+") は先に処理される
} else if (text.equals("EOS")) { //$NON-NLS-1$
// 次の文の文節の開始番号
index0 = size();
} else {
// 形態素
if (lastElement == null) {
lastElement = createElement(""); //$NON-NLS-1$
// 形態素だけを処理した場合
}
MorphemeElement morpheme = new MorphemeElement(text);
lastElement.getMorphemes().add(morpheme);
morpheme.indexSentence = indexSentence;
if (morpheme.getHinshiSaibunrui().equals(TermClass.PERIOD.getName())) {
indexSentence++; // 句点で分割
}
}
}
}
/**
* {@link Relation}を作る
*
* @param srcElement
* @param text
* @return 係り受けリレーション
*/
private Relation createRelation(ISyntaxStructureElement srcElement, String text) {
String s = text.substring(1).trim(); // #, * を除く
// 10D の D を除く
for (int pos = 0; pos < s.length(); pos++) {
char c = s.charAt(pos);
if ((c != '-') && (c < '0' || '9' < c)) {
int dst = Integer.parseInt(s.substring(0, pos));
if (dst > 0) { // dst==-1 : 係り先なし
dst += index0; // 前の文の番号だけずらす
}
String type = s.substring(pos, pos + 1);
return new Relation(srcElement.getIndex(), dst, type);
}
}
throw new IllegalArgumentException(text);
}
/**
* 表記をつなげて原文を取得
*
* @return 原文
*/
public String getText() {
StringBuilder text = new StringBuilder(3000);
for (ISyntaxStructureElement element : this) {
text.append(element.getHyouki());
}
return text.toString();
}
@Override
public String toString() {
// DB格納用の文字列を作る
// EOS はなくなるが、構文パターンに影響ないのでそのまま
StringBuilder sb = new StringBuilder(1000000); // 100KB
for (ISyntaxStructureElement element : this) {
sb.append(element.toString());
}
return sb.toString();
}
/**
* 係り受け先/元
*
*/
static public class Relation {
int src;
int dest;
String type;
Relation(int src, int dest, String type) {
this.src = src;
this.dest = dest;
this.type = type;
}
/**
* 係先の文節/句のID
*
* @return 係り元ID
*/
public int getSrc() {
return src;
}
/**
* 係り元の文節/句のID
*
* @return 係り先ID
*/
public int getDest() {
return dest;
}
/**
* 係先のタイプ
* "D", "P"
*
* @return 係り先のタイプ
*/
public String getType() {
return type;
}
@Override
public String toString() {
return src + "-" + type + "->" + dest; //$NON-NLS-1$ //$NON-NLS-2$
}
}
/**
* 句/文節
*
*/
class SyntaxStructureElement implements ISyntaxStructureElement {
/**
* この文節/句の番号(デバッグ用)
*/
final int index;
/**
* KNPの結果のテキスト
* ex. *1 <文頭> ...
*/
final String source;
/**
* この文節/句を構成する形態素
*/
final List<MorphemeElement> morphemes = new ArrayList<MorphemeElement>();
/**
*
* @param index
* @param text
* KNPの結果のテキスト ex. *1 <文頭> ...
*/
SyntaxStructureElement(int index, String text) {
this.index = index;
this.source = text;
}
@Override
public int getIndex() {
return index;
}
@Override
public List<MorphemeElement> getMorphemes() {
return morphemes;
}
@Override
public ISyntaxStructureElement getDependDestination() {
return getDependDestination(null);
}
@Override
public List<ISyntaxStructureElement> getDependSources() {
return getDependSources(null);
}
@Override
public String getHyouki() {
return getText(MorphemeElement.HYOUKI);
}
@Override
public String getYomi() {
return getText(MorphemeElement.YOMI);
}
/**
* 文節/句を構成する形態素をつなげる
*
* @param index
* @return 形態素
*/
private String getText(int index) {
StringBuilder s = new StringBuilder(2000);
for (MorphemeElement morpheme : morphemes) {
s.append(morpheme.get(index));
}
return s.toString();
}
@Override
public ISyntaxStructureElement getDependDestination(String type) {
Relation relation = destsMap.get(index);
if (relation != null) {
if (type == null || type.equals(relation.getType())) {
return SyntaxStructure.this.get(relation.getDest());
}
}
return null;
}
@Override
public List<ISyntaxStructureElement> getDependSources(String type) {
List<Relation> relations = sourcesMap.get(index);
List<ISyntaxStructureElement> srcs = new ArrayList<ISyntaxStructureElement>();
if (relations != null) {
for (Relation relation : relations) {
if (type == null || type.equals(relation.getType())) {
srcs.add(SyntaxStructure.this.get(relation.getSrc()));
}
}
}
return srcs;
}
@Override
public String toString() {
// DB格納用のテキストを作る
StringBuilder sb = new StringBuilder(10000); // 10KB
if (!source.isEmpty()) { // KNPをやっていない場合 source=""
// インデックスの差し替え
// TODO:係り受けタイプがこれ以外にあるかも？
String str = ""; //$NON-NLS-1$
int index = -1;
if (this.getDependDestination("D") != null) { //$NON-NLS-1$
index = this.getDependDestination("D").getIndex(); //$NON-NLS-1$
str = "+ " + index + source.substring(source.indexOf("D"));// 未格 //$NON-NLS-1$ //$NON-NLS-2$
} else if (this.getDependDestination("P") != null) { //$NON-NLS-1$
index = this.getDependDestination("P").getIndex(); //$NON-NLS-1$
str = "+ " + index + source.substring(source.indexOf("P"));// 連体 //$NON-NLS-1$ //$NON-NLS-2$
} else if (this.getDependDestination("A") != null) { //$NON-NLS-1$
index = this.getDependDestination("A").getIndex(); //$NON-NLS-1$
str = "+ " + index + source.substring(source.indexOf("A"));// 同格連体 //$NON-NLS-1$ //$NON-NLS-2$
} else {
str = source;
}
sb.append(str);
sb.append("\n"); //$NON-NLS-1$
}
for (MorphemeElement element : getMorphemes()) {
sb.append(element.toString());
sb.append("\n"); //$NON-NLS-1$
}
return sb.toString();
}
@Override
public Object getAdapter(@SuppressWarnings("rawtypes") Class adapter) {
if (adapter.equals(IPropertySource.class)) {
return source1;
}
return null;
}
private IPropertySource source1 = new IPropertySource() {
@Override
public IPropertyDescriptor[] getPropertyDescriptors() {
IPropertyDescriptor[] descriptor = new IPropertyDescriptor[] {
new TextPropertyDescriptor("id", "ID"), //$NON-NLS-1$ //$NON-NLS-2$
new TextPropertyDescriptor("text", Messages.SyntaxStructure_NOTATION), new TextPropertyDescriptor("src", Messages.MODIFICATION_DESTINATION), new TextPropertyDescriptor("dist", Messages.MODIFICATION_SOURCE), }; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
return descriptor;
}
@Override
public Object getPropertyValue(Object id) {
if (id.equals("id")) { //$NON-NLS-1$
return String.valueOf(getIndex());
}
if (id.equals("text")) { //$NON-NLS-1$
return getHyouki();
}
if (id.equals("src")) { //$NON-NLS-1$
if (getDependSources() != null && !getDependSources().isEmpty()) {
StringBuilder buf = new StringBuilder();
for (ISyntaxStructureElement src : getDependSources()) {
buf.append(" , " + src.getIndex() + ":" + src.getHyouki()); //$NON-NLS-1$ //$NON-NLS-2$
}
return buf.toString().substring(3);
}
return ""; //$NON-NLS-1$
}
if (id.equals("dist")) { //$NON-NLS-1$
if (getDependDestination() != null) {
// return getDst();
return String.valueOf(getDependDestination().getIndex()) + ":" + getDependDestination().getHyouki(); //$NON-NLS-1$
}
return ""; //$NON-NLS-1$
}
return null;
}
@Override
public boolean isPropertySet(Object id) {
return false;
}
@Override
public void resetPropertyValue(Object id) {
}
@Override
public void setPropertyValue(Object id, Object value) {
}
@Override
public Object getEditableValue() {
return null;
}
};
}
}