Lucene
OpenSource 검색 라이브러리로서 하둡 개발자로 잘 알려진 Doug Cutting이 개발했다. Lucene이라는 이름은 그의 아내 middle name 을 따서 지었다고.
Lucene은 Levenshtein distance에 기반한 fuzzy search 기능까지 있는 검색 라이브러리로 검색 능력이 뛰어나다고 한다.
Lucene-demo
위 경로에서 Lucene 최신 라이브러리를 다운로드 받으면 demo/lucene-demo-7.5.0.jar 데모 클래스가 있다.
IndexFiles를 이용하면 파일 시스템의 특정 경로 내의 모든 문서(text,txt,html 등)의 내용을 인덱싱할 수 있고 SearchFiles를 이용하면 문서 내의 키워드들을 이용해서 파일 위치를 찾아낼 수 있다.
Lucene 라이브러리 경로의 문서 색인하기
아래와 같이 IndexFiles를 이용해서 Lucene 경로의 모든 문서를 index 라는 인덱스로 생성한다.
java -cp Lucene/7.5.0/demo/lucene-demo-7.5.0.jar:Lucene/7.5.0/core/lucene-core-7.5.0.jar org.apache.lucene.demo.IndexFiles -index index -docs Lucene
다음과 같이 모든 문서들을 읽어들이고 색인함.
Indexing to directory 'index'...
adding Lucene/lucene-7.5.0.tgz
adding Lucene/7.5.0/licenses/ant-1.8.2.jar.sha1
adding Lucene/7.5.0/licenses/ant-LICENSE-ASL.txt
..
.
adding Lucene/5.5.5/suggest/lucene-suggest-5.5.5.jar
35647 total milliseconds
색인된 문서에서 “Apache Lucene is a high-performance, full-featured text search engine library.” 문자열 찾기
다음과 같이 index 인덱스를 조회하여 "Apache Lucene is a high-performance, full-featured text search engine library." 문자열을 검색한다.
java -cp Lucene/7.5.0/queryparser/lucene-queryparser-7.5.0.jar:Lucene/7.5.0/demo/lucene-demo-7.5.0.jar:Lucene/7.5.0/core/lucene-core-7.5.0.jar org.apache.lucene.demo.SearchFiles -query "Apache Lucene is a high-performance, full-featured text search engine library."
검색 결과 12967 건의 일치하는 문서가 발견되는데 ..
Searching for: apache lucene high performance full featured text search engine library
12967 total matching documents
1. Lucene/5.5.5/docs/core/overview-summary.html
2. Lucene/7.5.0/docs/core/overview-summary.html
3. Lucene/7.5.0/docs/index.html
4. Lucene/5.5.5/docs/index.html
5. Lucene/7.5.0/README.txt
6. Lucene/5.5.5/README.txt
7. Lucene/7.5.0/docs/highlighter/org/apache/lucene/search/vectorhighlight/package-summary.html
8. Lucene/5.5.5/docs/highlighter/org/apache/lucene/search/vectorhighlight/package-summary.html
9. Lucene/5.5.5/CHANGES.txt
10. Lucene/5.5.5/licenses/javax.servlet-LICENSE-CDDL.txt
위와 같이 검색 조건에 부합하는 것들 중에 top 10 결과만 출력해준다.
아래와 같이 명령으로 1순위 문서에 해당 문자열을 찾아보면
grep "Apache Lucene is a high-performance, full-featured text search engine library." Lucene/5.5.5/docs/core/overview-summary.html
해당 문자열이 1순위 문서에서 정확히 발견된다.
<div class="block">Apache Lucene is a high-performance, full-featured text search engine library.</div>
<div class="block"><p>Apache Lucene is a high-performance, full-featured text search engine library.
다음 명령으로 3순위 문서에 해당 문자열을 찾아보면
grep "Apache Lucene is a high-performance, full-featured text search engine library." Lucene/7.5.0/docs/index.html
안 나온다.
대신 Apache Lucene이라는 문자열만 3순위 문서에서 찾으면
grep "Apache Lucene" Lucene/7.5.0/docs/index.html
다음과 같이 조회된다.
<title>Apache Lucene 7.5.0 Documentation</title>
<a href="http://lucene.apache.org/core/"><img src="lucene_green_300.gif" title="Apache Lucene Logo" alt="Lucene" border="0"></a>
<h1>Apache Lucene<span style="vertical-align: top; font-size: x-small">TM</span> 7.5.0 Documentation</h1>
This is the official documentation for <b>Apache Lucene 7.5.0</b>. Additional documentation is available in the
audiences: first-time users looking to install Apache Lucene in their
Exact-Match가 아닌, Full-Text Search 기법을 이용해서 문서 내용을 색인한 것 같다. 정확하게는 Lucene-demo 소스를 읽어보고, Lucene-core 클래스를 공부해봐야할 것 같다.
Exact-Match
- SELECT SQL ( WHERE COL LIKE ‘%A%’ )
- grep
위와 같이 Exact-Match는 공백을 포함하여 모든 단어와 그 단어의 배열 순서가 완전히 일치하는 것들을 찾아낼 수 있다. 정확하지만 모든 record들을 full-scan 해야한다.
Full-Text Search
- email 검색
- 검색엔진 ( 구글 | 네이버 )
문서 내의 어떤 단어들이 있는지 먼저 파악하고, 각각의 단어들이 어느 문서에 위치했었는지를 색인한다.
Lucene-demo Source code
IndexFiles.java
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.lucene.demo;
import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.nio.charset.StandardCharsets;
import java.nio.file.FileVisitResult;
import java.nio.file.Files;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.nio.file.SimpleFileVisitor;
import java.nio.file.attribute.BasicFileAttributes;
import java.util.Date;
import org.apache.lucene.analysis.Analyzer;
import org.apache.lucene.analysis.standard.StandardAnalyzer;
import org.apache.lucene.document.LongPoint;
import org.apache.lucene.document.Document;
import org.apache.lucene.document.Field;
import org.apache.lucene.document.StringField;
import org.apache.lucene.document.TextField;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.IndexWriterConfig.OpenMode;
import org.apache.lucene.index.IndexWriterConfig;
import org.apache.lucene.index.Term;
import org.apache.lucene.store.Directory;
import org.apache.lucene.store.FSDirectory;
/** Index all text files under a directory.
* <p>
* This is a command-line application demonstrating simple Lucene indexing.
* Run it with no command-line arguments for usage information.
*/
public class IndexFiles {
private IndexFiles() {}
/** Index all text files under a directory. */
public static void main(String[] args) {
String usage = "java org.apache.lucene.demo.IndexFiles"
+ " [-index INDEX_PATH] [-docs DOCS_PATH] [-update]\n\n"
+ "This indexes the documents in DOCS_PATH, creating a Lucene index"
+ "in INDEX_PATH that can be searched with SearchFiles";
String indexPath = "index";
String docsPath = null;
boolean create = true;
for(int i=0;i<args.length;i++) {
if ("-index".equals(args[i])) {
indexPath = args[i+1];
i++;
} else if ("-docs".equals(args[i])) {
docsPath = args[i+1];
i++;
} else if ("-update".equals(args[i])) {
create = false;
}
}
if (docsPath == null) {
System.err.println("Usage: " + usage);
System.exit(1);
}
final Path docDir = Paths.get(docsPath);
if (!Files.isReadable(docDir)) {
System.out.println("Document directory '" +docDir.toAbsolutePath()+ "' does not exist or is not readable, please check the path");
System.exit(1);
}
Date start = new Date();
try {
System.out.println("Indexing to directory '" + indexPath + "'...");
Directory dir = FSDirectory.open(Paths.get(indexPath));
Analyzer analyzer = new StandardAnalyzer();
IndexWriterConfig iwc = new IndexWriterConfig(analyzer);
if (create) {
// Create a new index in the directory, removing any
// previously indexed documents:
iwc.setOpenMode(OpenMode.CREATE);
} else {
// Add new documents to an existing index:
iwc.setOpenMode(OpenMode.CREATE_OR_APPEND);
}
// Optional: for better indexing performance, if you
// are indexing many documents, increase the RAM
// buffer. But if you do this, increase the max heap
// size to the JVM (eg add -Xmx512m or -Xmx1g):
//
// iwc.setRAMBufferSizeMB(256.0);
IndexWriter writer = new IndexWriter(dir, iwc);
indexDocs(writer, docDir);
// NOTE: if you want to maximize search performance,
// you can optionally call forceMerge here. This can be
// a terribly costly operation, so generally it's only
// worth it when your index is relatively static (ie
// you're done adding documents to it):
//
// writer.forceMerge(1);
writer.close();
Date end = new Date();
System.out.println(end.getTime() - start.getTime() + " total milliseconds");
} catch (IOException e) {
System.out.println(" caught a " + e.getClass() +
"\n with message: " + e.getMessage());
}
}
/**
* Indexes the given file using the given writer, or if a directory is given,
* recurses over files and directories found under the given directory.
*
* NOTE: This method indexes one document per input file. This is slow. For good
* throughput, put multiple documents into your input file(s). An example of this is
* in the benchmark module, which can create "line doc" files, one document per line,
* using the
* <a href="../../../../../contrib-benchmark/org/apache/lucene/benchmark/byTask/tasks/WriteLineDocTask.html"
* >WriteLineDocTask</a>.
*
* @param writer Writer to the index where the given file/dir info will be stored
* @param path The file to index, or the directory to recurse into to find files to index
* @throws IOException If there is a low-level I/O error
*/
static void indexDocs(final IndexWriter writer, Path path) throws IOException {
if (Files.isDirectory(path)) {
Files.walkFileTree(path, new SimpleFileVisitor<Path>() {
@Override
public FileVisitResult visitFile(Path file, BasicFileAttributes attrs) throws IOException {
try {
indexDoc(writer, file, attrs.lastModifiedTime().toMillis());
} catch (IOException ignore) {
// don't index files that can't be read.
}
return FileVisitResult.CONTINUE;
}
});
} else {
indexDoc(writer, path, Files.getLastModifiedTime(path).toMillis());
}
}
/** Indexes a single document */
static void indexDoc(IndexWriter writer, Path file, long lastModified) throws IOException {
try (InputStream stream = Files.newInputStream(file)) {
// make a new, empty document
Document doc = new Document();
// Add the path of the file as a field named "path". Use a
// field that is indexed (i.e. searchable), but don't tokenize
// the field into separate words and don't index term frequency
// or positional information:
Field pathField = new StringField("path", file.toString(), Field.Store.YES);
doc.add(pathField);
// Add the last modified date of the file a field named "modified".
// Use a LongPoint that is indexed (i.e. efficiently filterable with
// PointRangeQuery). This indexes to milli-second resolution, which
// is often too fine. You could instead create a number based on
// year/month/day/hour/minutes/seconds, down the resolution you require.
// For example the long value 2011021714 would mean
// February 17, 2011, 2-3 PM.
doc.add(new LongPoint("modified", lastModified));
// Add the contents of the file to a field named "contents". Specify a Reader,
// so that the text of the file is tokenized and indexed, but not stored.
// Note that FileReader expects the file to be in UTF-8 encoding.
// If that's not the case searching for special characters will fail.
doc.add(new TextField("contents", new BufferedReader(new InputStreamReader(stream, StandardCharsets.UTF_8))));
if (writer.getConfig().getOpenMode() == OpenMode.CREATE) {
// New index, so we just add the document (no old document can be there):
System.out.println("adding " + file);
writer.addDocument(doc);
} else {
// Existing index (an old copy of this document may have been indexed) so
// we use updateDocument instead to replace the old one matching the exact
// path, if present:
System.out.println("updating " + file);
writer.updateDocument(new Term("path", file.toString()), doc);
}
}
}
}
SearchFiles.java
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.lucene.demo;
import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.nio.charset.StandardCharsets;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.util.Date;
import org.apache.lucene.analysis.Analyzer;
import org.apache.lucene.analysis.standard.StandardAnalyzer;
import org.apache.lucene.document.Document;
import org.apache.lucene.index.DirectoryReader;
import org.apache.lucene.index.IndexReader;
import org.apache.lucene.queryparser.classic.QueryParser;
import org.apache.lucene.search.IndexSearcher;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.ScoreDoc;
import org.apache.lucene.search.TopDocs;
import org.apache.lucene.store.FSDirectory;
/** Simple command-line based search demo. */
public class SearchFiles {
private SearchFiles() {}
/** Simple command-line based search demo. */
public static void main(String[] args) throws Exception {
String usage =
"Usage:\tjava org.apache.lucene.demo.SearchFiles [-index dir] [-field f] [-repeat n] [-queries file] [-query string] [-raw] [-paging hitsPerPage]\n\nSee http://lucene.apache.org/core/4_1_0/demo/ for details.";
if (args.length > 0 && ("-h".equals(args[0]) || "-help".equals(args[0]))) {
System.out.println(usage);
System.exit(0);
}
String index = "index";
String field = "contents";
String queries = null;
int repeat = 0;
boolean raw = false;
String queryString = null;
int hitsPerPage = 10;
for(int i = 0;i < args.length;i++) {
if ("-index".equals(args[i])) {
index = args[i+1];
i++;
} else if ("-field".equals(args[i])) {
field = args[i+1];
i++;
} else if ("-queries".equals(args[i])) {
queries = args[i+1];
i++;
} else if ("-query".equals(args[i])) {
queryString = args[i+1];
i++;
} else if ("-repeat".equals(args[i])) {
repeat = Integer.parseInt(args[i+1]);
i++;
} else if ("-raw".equals(args[i])) {
raw = true;
} else if ("-paging".equals(args[i])) {
hitsPerPage = Integer.parseInt(args[i+1]);
if (hitsPerPage <= 0) {
System.err.println("There must be at least 1 hit per page.");
System.exit(1);
}
i++;
}
}
IndexReader reader = DirectoryReader.open(FSDirectory.open(Paths.get(index)));
IndexSearcher searcher = new IndexSearcher(reader);
Analyzer analyzer = new StandardAnalyzer();
BufferedReader in = null;
if (queries != null) {
in = Files.newBufferedReader(Paths.get(queries), StandardCharsets.UTF_8);
} else {
in = new BufferedReader(new InputStreamReader(System.in, StandardCharsets.UTF_8));
}
QueryParser parser = new QueryParser(field, analyzer);
while (true) {
if (queries == null && queryString == null) { // prompt the user
System.out.println("Enter query: ");
}
String line = queryString != null ? queryString : in.readLine();
if (line == null || line.length() == -1) {
break;
}
line = line.trim();
if (line.length() == 0) {
break;
}
Query query = parser.parse(line);
System.out.println("Searching for: " + query.toString(field));
if (repeat > 0) { // repeat & time as benchmark
Date start = new Date();
for (int i = 0; i < repeat; i++) {
searcher.search(query, 100);
}
Date end = new Date();
System.out.println("Time: "+(end.getTime()-start.getTime())+"ms");
}
doPagingSearch(in, searcher, query, hitsPerPage, raw, queries == null && queryString == null);
if (queryString != null) {
break;
}
}
reader.close();
}
/**
* This demonstrates a typical paging search scenario, where the search engine presents
* pages of size n to the user. The user can then go to the next page if interested in
* the next hits.
*
* When the query is executed for the first time, then only enough results are collected
* to fill 5 result pages. If the user wants to page beyond this limit, then the query
* is executed another time and all hits are collected.
*
*/
public static void doPagingSearch(BufferedReader in, IndexSearcher searcher, Query query,
int hitsPerPage, boolean raw, boolean interactive) throws IOException {
// Collect enough docs to show 5 pages
TopDocs results = searcher.search(query, 5 * hitsPerPage);
ScoreDoc[] hits = results.scoreDocs;
int numTotalHits = Math.toIntExact(results.totalHits.value);
System.out.println(numTotalHits + " total matching documents");
int start = 0;
int end = Math.min(numTotalHits, hitsPerPage);
while (true) {
if (end > hits.length) {
System.out.println("Only results 1 - " + hits.length +" of " + numTotalHits + " total matching documents collected.");
System.out.println("Collect more (y/n) ?");
String line = in.readLine();
if (line.length() == 0 || line.charAt(0) == 'n') {
break;
}
hits = searcher.search(query, numTotalHits).scoreDocs;
}
end = Math.min(hits.length, start + hitsPerPage);
for (int i = start; i < end; i++) {
if (raw) { // output raw format
System.out.println("doc="+hits[i].doc+" score="+hits[i].score);
continue;
}
Document doc = searcher.doc(hits[i].doc);
String path = doc.get("path");
if (path != null) {
System.out.println((i+1) + ". " + path);
String title = doc.get("title");
if (title != null) {
System.out.println(" Title: " + doc.get("title"));
}
} else {
System.out.println((i+1) + ". " + "No path for this document");
}
}
if (!interactive || end == 0) {
break;
}
if (numTotalHits >= end) {
boolean quit = false;
while (true) {
System.out.print("Press ");
if (start - hitsPerPage >= 0) {
System.out.print("(p)revious page, ");
}
if (start + hitsPerPage < numTotalHits) {
System.out.print("(n)ext page, ");
}
System.out.println("(q)uit or enter number to jump to a page.");
String line = in.readLine();
if (line.length() == 0 || line.charAt(0)=='q') {
quit = true;
break;
}
if (line.charAt(0) == 'p') {
start = Math.max(0, start - hitsPerPage);
break;
} else if (line.charAt(0) == 'n') {
if (start + hitsPerPage < numTotalHits) {
start+=hitsPerPage;
}
break;
} else {
int page = Integer.parseInt(line);
if ((page - 1) * hitsPerPage < numTotalHits) {
start = (page - 1) * hitsPerPage;
break;
} else {
System.out.println("No such page");
}
}
}
if (quit) break;
end = Math.min(numTotalHits, start + hitsPerPage);
}
}
}
}
