使用Lucene-Spatial實現整合地理位置的全文檢索
阿新 • • 發佈：2019-01-23
Lucene通過Spatial包提供了對基於地理位置的全文檢索的支援，最典型的應用場景就是：“搜尋中關村附近1公里內的火鍋店，並按遠近排序”。使用Lucene-Spatial新增對地理位置的支援，和之前普通文字搜尋主要有兩點區別：
1. 將座標資訊轉化為笛卡爾層，建立索引
private void indexLocation(Document document, JSONObject jo) throws Exception { double longitude = jo.getDouble("longitude"); double latitude = jo.getDouble("latitude"); document.add(new Field("lat", NumericUtils .doubleToPrefixCoded(latitude), Field.Store.YES, Field.Index.NOT_ANALYZED)); document.add(new Field("lng", NumericUtils .doubleToPrefixCoded(longitude), Field.Store.YES, Field.Index.NOT_ANALYZED)); for (int tier = startTier; tier <= endTier; tier++) { ctp = new CartesianTierPlotter(tier, projector, CartesianTierPlotter.DEFALT_FIELD_PREFIX); final double boxId = ctp.getTierBoxId(latitude, longitude); document.add(new Field(ctp.getTierFieldName(), NumericUtils .doubleToPrefixCoded(boxId), Field.Store.YES, Field.Index.NOT_ANALYZED_NO_NORMS)); } }
2. 搜尋時，指定使用DistanceQueryFilter
DistanceQueryBuilder dq = new DistanceQueryBuilder(latitude, longitude, miles, "lat", "lng", CartesianTierPlotter.DEFALT_FIELD_PREFIX, true, startTier, endTier); DistanceFieldComparatorSource dsort = new DistanceFieldComparatorSource( dq.getDistanceFilter()); Sort sort = new Sort(new SortField("geo_distance", dsort));
下面是基於Lucene3.2.0和JUnit4.8.2的完整程式碼。
首先準備測試用的資料：<dependencies> <dependency> <groupId>junit</groupId> <artifactId>junit</artifactId> <version>4.8.2</version> <type>jar</type> <scope>test</scope> </dependency> <dependency> <groupId>org.apache.lucene</groupId> <artifactId>lucene-core</artifactId> <version>3.2.0</version> <type>jar</type> <scope>compile</scope> </dependency> <dependency> <groupId>org.apache.lucene</groupId> <artifactId>lucene-spatial</artifactId> <version>3.2.0</version> <type>jar</type> <scope>compile</scope> </dependency> <dependency> <groupId>org.json</groupId> <artifactId>json</artifactId> <version>20100903</version> <type>jar</type> <scope>compile</scope> </dependency> </dependencies>
{"id":12,"title":"時尚碼頭美容美髮熱燙特價","longitude":116.3838183,"latitude":39.9629015}
{"id":17,"title":"審美個人美容美髮套餐","longitude":116.386564,"latitude":39.966102}
{"id":23,"title":"海底撈吃300送300","longitude":116.38629,"latitude":39.9629573}
{"id":26,"title":"僅98元！享原價335元李老爹","longitude":116.3846175,"latitude":39.9629125}
{"id":29,"title":"都美造型燙染美髮護理套餐","longitude":116.38629,"latitude":39.9629573}
{"id":30,"title":"僅售55元！原價80元的老舍茶館相聲下午場","longitude":116.0799914,"latitude":39.9655391}
{"id":33,"title":"僅售55元！原價80元的新笑聲客棧早場","longitude":116.0799914,"latitude":39.9655391}
{"id":34,"title":"僅售39元（紅色禮盒）！原價80元的平谷桃","longitude":116.0799914,"latitude":39.9655391}
{"id":46,"title":"僅售38元！原價180元地質禮堂白雪公主","longitude":116.0799914,"latitude":39.9655391}
{"id":49,"title":"僅99元！享原價342.7元自助餐","longitude":116.0799914,"latitude":39.9655391}
{"id":58,"title":"桑海教育暑期學生報名培訓九折優惠券","longitude":116.0799914,"latitude":39.9655391}
{"id":59,"title":"全國發貨：僅29元！貝玲妃超模粉紅高光光","longitude":116.0799914,"latitude":39.9655391}
{"id":65,"title":"海之嶼生態水族用品店抵用券","longitude":116.0799914,"latitude":39.9655391}
{"id":67,"title":"小區東門時尚燙染個人護理美髮套餐","longitude":116.3799914,"latitude":39.9655391}
{"id":74,"title":"《郭德綱相聲專輯》CD套裝","longitude":116.0799914,"latitude":39.9655391}根據上面的測試資料，編寫測試用例，分別搜尋座標（116.3838183, 39.9629015）3千米以內的“美髮”和全部內容，分別得到的結果應該是4條和6條。
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.fail;
import java.util.List;
import org.junit.Test;
public class LuceneSpatialTest {
private static LuceneSpatial spatialSearcher = new LuceneSpatial();
@Test
public void testSearch() {
try {
long start = System.currentTimeMillis();
List<String> results = spatialSearcher.search("美髮", 116.3838183, 39.9629015, 3.0);
System.out.println(results.size()
+ "個匹配結果，共耗時 "
+ (System.currentTimeMillis() - start) + "毫秒。\n");
assertEquals(4, results.size());
} catch (Exception e) {
fail("Exception occurs...");
e.printStackTrace();
}
}
@Test
public void testSearchWithoutKeyword() {
try {
long start = System.currentTimeMillis();
List<String> results = spatialSearcher.search(null, 116.3838183, 39.9629015, 3.0);
System.out.println( results.size()
+ "個匹配結果，共耗時 "
+ (System.currentTimeMillis() - start) + "毫秒.\n");
assertEquals(6, results.size());
} catch (Exception e) {
fail("Exception occurs...");
e.printStackTrace();
}
}
}下面是LuceneSpatial類，在建構函式中初始化變數和建立索引：
public class LuceneSpatial {
private Analyzer analyzer;
private IndexWriter writer;
private FSDirectory indexDirectory;
private IndexSearcher indexSearcher;
private IndexReader indexReader;
private String indexPath = "c:/lucene-spatial";
// Spatial
private IProjector projector;
private CartesianTierPlotter ctp;
public static final double RATE_MILE_TO_KM = 1.609344; //英里和公里的比率
public static final String LAT_FIELD = "lat";
public static final String LON_FIELD = "lng";
private static final double MAX_RANGE = 15.0; // 索引支援的最大範圍，單位是千米
private static final double MIN_RANGE = 3.0; // 索引支援的最小範圍，單位是千米
private int startTier;
private int endTier;
public LuceneSpatial() {
try {
init();
} catch (Exception e) {
e.printStackTrace();
}
}
private void init() throws Exception {
initializeSpatialOptions();
analyzer = new StandardAnalyzer(Version.LUCENE_32);
File path = new File(indexPath);
boolean isNeedCreateIndex = false;
if (path.exists() && !path.isDirectory())
throw new Exception("Specified path is not a directory");
if (!path.exists()) {
path.mkdirs();
isNeedCreateIndex = true;
}
indexDirectory = FSDirectory.open(new File(indexPath));
//建立索引
if (isNeedCreateIndex) {
IndexWriterConfig indexWriterConfig = new IndexWriterConfig(
Version.LUCENE_32, analyzer);
indexWriterConfig.setOpenMode(OpenMode.CREATE_OR_APPEND);
writer = new IndexWriter(indexDirectory, indexWriterConfig);
buildIndex();
}
indexReader = IndexReader.open(indexDirectory, true);
indexSearcher = new IndexSearcher(indexReader);
}
@SuppressWarnings("deprecation")
private void initializeSpatialOptions() {
projector = new SinusoidalProjector();
ctp = new CartesianTierPlotter(0, projector,
CartesianTierPlotter.DEFALT_FIELD_PREFIX);
startTier = ctp.bestFit(MAX_RANGE / RATE_MILE_TO_KM);
endTier = ctp.bestFit(MIN_RANGE / RATE_MILE_TO_KM);
}
private int mile2Meter(double miles) {
double dMeter = miles * RATE_MILE_TO_KM * 1000;
return (int) dMeter;
}
private double km2Mile(double km) {
return km / RATE_MILE_TO_KM;
}建立索引的具體實現：
private void buildIndex() {
BufferedReader br = null;
try {
//逐行新增測試資料到索引中，測試資料檔案和原始檔在同一個目錄下
br = new BufferedReader(new InputStreamReader(
LuceneSpatial.class.getResourceAsStream("data")));
String line = null;
while ((line = br.readLine()) != null) {
index(new JSONObject(line));
}
writer.commit();
} catch (Exception e) {
e.printStackTrace();
} finally {
if (br != null) {
try {
br.close();
} catch (IOException e) {
e.printStackTrace();
}
}
}
}
private void index(JSONObject jo) throws Exception {
Document doc = new Document();
doc.add(new Field("id", jo.getString("id"), Field.Store.YES,
Field.Index.ANALYZED));
doc.add(new Field("title", jo.getString("title"), Field.Store.YES,
Field.Index.ANALYZED));
//將位置資訊新增到索引中
indexLocation(doc, jo);
writer.addDocument(doc);
}
private void indexLocation(Document document, JSONObject jo)
throws Exception {
double longitude = jo.getDouble("longitude");
double latitude = jo.getDouble("latitude");
document.add(new Field("lat", NumericUtils
.doubleToPrefixCoded(latitude), Field.Store.YES,
Field.Index.NOT_ANALYZED));
document.add(new Field("lng", NumericUtils
.doubleToPrefixCoded(longitude), Field.Store.YES,
Field.Index.NOT_ANALYZED));
for (int tier = startTier; tier <= endTier; tier++) {
ctp = new CartesianTierPlotter(tier, projector,
CartesianTierPlotter.DEFALT_FIELD_PREFIX);
final double boxId = ctp.getTierBoxId(latitude, longitude);
document.add(new Field(ctp.getTierFieldName(), NumericUtils
.doubleToPrefixCoded(boxId), Field.Store.YES,
Field.Index.NOT_ANALYZED_NO_NORMS));
}
}搜尋的具體實現：
public List<String> search(String keyword, double longitude,
double latitude, double range) throws Exception {
List<String> result = new ArrayList<String>();
double miles = km2Mile(range);
DistanceQueryBuilder dq = new DistanceQueryBuilder(latitude,
longitude, miles, "lat", "lng",
CartesianTierPlotter.DEFALT_FIELD_PREFIX, true, startTier,
endTier);
//按照距離排序
DistanceFieldComparatorSource dsort = new DistanceFieldComparatorSource(
dq.getDistanceFilter());
Sort sort = new Sort(new SortField("geo_distance", dsort));
Query query = buildQuery(keyword);
//搜尋結果
TopDocs hits = indexSearcher.search(query, dq.getFilter(),
Integer.MAX_VALUE, sort);
//獲得各條結果相對應的距離
Map<Integer, Double> distances = dq.getDistanceFilter()
.getDistances();
for (int i = 0; i < hits.totalHits; i++) {
final int docID = hits.scoreDocs[i].doc;
final Document doc = indexSearcher.doc(docID);
final StringBuilder builder = new StringBuilder();
builder.append("找到了: ")
.append(doc.get("title"))
.append("， 距離: ")
.append(mile2Meter(distances.get(docID)))
.append("米。");
System.out.println(builder.toString());
result.add(builder.toString());
}
return result;
}
private Query buildQuery(String keyword) throws Exception {
//如果沒有指定關鍵字，則返回範圍內的所有結果
if (keyword == null || keyword.isEmpty()) {
return new MatchAllDocsQuery();
}
QueryParser parser = new QueryParser(Version.LUCENE_32, "title",
analyzer);
parser.setDefaultOperator(Operator.AND);
return parser.parse(keyword.toString());
}執行測試用例，可以得到下面的結果：
找到了: 時尚碼頭美容美髮熱燙特價， 距離: 0米。
找到了: 都美造型燙染美髮護理套餐， 距離: 210米。
找到了: 審美個人美容美髮套餐， 距離: 426米。
找到了: 小區東門時尚燙染個人護理美髮套餐， 距離: 439米。
4個匹配結果，共耗時 119毫秒。
找到了: 時尚碼頭美容美髮熱燙特價， 距離: 0米。
找到了: 僅98元！享原價335元李老爹， 距離: 68米。
找到了: 海底撈吃300送300， 距離: 210米。
找到了: 都美造型燙染美髮護理套餐， 距離: 210米。
找到了: 審美個人美容美髮套餐， 距離: 426米。
找到了: 小區東門時尚燙染個人護理美髮套餐， 距離: 439米。
6個匹配結果，共耗時 3毫秒.
參考文獻：
兩篇示例（其中大部分程式碼就來自於這裡）：