private void testRightOpenRange(int precisionStep) throws Exception {
String field = "field" + precisionStep;
int count = 3000;
long lower = (count - 1) * distance + (distance / 3) + startOffset;
NumericRangeQuery<Long> q =
NumericRangeQuery.newLongRange(field, precisionStep, lower, null, true, true);
TopDocs topDocs = searcher.search(q, null, noDocs, Sort.INDEXORDER);
if (VERBOSE)
System.out.println(
"Found "
+ q.getTotalNumberOfTerms()
+ " distinct terms in right open range for field '"
+ field
+ "'.");
ScoreDoc[] sd = topDocs.scoreDocs;
assertNotNull(sd);
assertEquals("Score doc count", noDocs - count, sd.length);
Document doc = searcher.doc(sd[0].doc);
assertEquals("First doc", count * distance + startOffset, Long.parseLong(doc.get(field)));
doc = searcher.doc(sd[sd.length - 1].doc);
assertEquals("Last doc", (noDocs - 1) * distance + startOffset, Long.parseLong(doc.get(field)));
q = NumericRangeQuery.newLongRange(field, precisionStep, lower, null, true, false);
topDocs = searcher.search(q, null, noDocs, Sort.INDEXORDER);
sd = topDocs.scoreDocs;
assertNotNull(sd);
assertEquals("Score doc count", noDocs - count, sd.length);
doc = searcher.doc(sd[0].doc);
assertEquals("First doc", count * distance + startOffset, Long.parseLong(doc.get(field)));
doc = searcher.doc(sd[sd.length - 1].doc);
assertEquals("Last doc", (noDocs - 1) * distance + startOffset, Long.parseLong(doc.get(field)));
}
@Test
public void testInfiniteValues() throws Exception {
Directory dir = newDirectory();
IndexWriter writer =
new IndexWriter(
dir,
newIndexWriterConfig(
TEST_VERSION_CURRENT, new WhitespaceAnalyzer(TEST_VERSION_CURRENT)));
Document doc = new Document();
doc.add(new NumericField("double").setDoubleValue(Double.NEGATIVE_INFINITY));
doc.add(new NumericField("long").setLongValue(Long.MIN_VALUE));
writer.addDocument(doc);
doc = new Document();
doc.add(new NumericField("double").setDoubleValue(Double.POSITIVE_INFINITY));
doc.add(new NumericField("long").setLongValue(Long.MAX_VALUE));
writer.addDocument(doc);
doc = new Document();
doc.add(new NumericField("double").setDoubleValue(0.0));
doc.add(new NumericField("long").setLongValue(0L));
writer.addDocument(doc);
writer.close();
IndexSearcher s = new IndexSearcher(dir);
Query q = NumericRangeQuery.newLongRange("long", null, null, true, true);
TopDocs topDocs = s.search(q, 10);
assertEquals("Score doc count", 3, topDocs.scoreDocs.length);
q = NumericRangeQuery.newLongRange("long", null, null, false, false);
topDocs = s.search(q, 10);
assertEquals("Score doc count", 3, topDocs.scoreDocs.length);
q = NumericRangeQuery.newLongRange("long", Long.MIN_VALUE, Long.MAX_VALUE, true, true);
topDocs = s.search(q, 10);
assertEquals("Score doc count", 3, topDocs.scoreDocs.length);
q = NumericRangeQuery.newLongRange("long", Long.MIN_VALUE, Long.MAX_VALUE, false, false);
topDocs = s.search(q, 10);
assertEquals("Score doc count", 1, topDocs.scoreDocs.length);
q = NumericRangeQuery.newDoubleRange("double", null, null, true, true);
topDocs = s.search(q, 10);
assertEquals("Score doc count", 3, topDocs.scoreDocs.length);
q = NumericRangeQuery.newDoubleRange("double", null, null, false, false);
topDocs = s.search(q, 10);
assertEquals("Score doc count", 3, topDocs.scoreDocs.length);
s.close();
dir.close();
}
private void testSorting(int precisionStep) throws Exception {
final Random rnd = newRandom();
String field = "field" + precisionStep;
// 10 random tests, the index order is ascending,
// so using a reverse sort field should retun descending documents
for (int i = 0; i < 10; i++) {
long lower = (long) (rnd.nextDouble() * noDocs * distance) + startOffset;
long upper = (long) (rnd.nextDouble() * noDocs * distance) + startOffset;
if (lower > upper) {
long a = lower;
lower = upper;
upper = a;
}
Query tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, true, true);
TopDocs topDocs =
searcher.search(tq, null, noDocs, new Sort(new SortField(field, SortField.LONG, true)));
if (topDocs.totalHits == 0) continue;
ScoreDoc[] sd = topDocs.scoreDocs;
assertNotNull(sd);
long last = Long.parseLong(searcher.doc(sd[0].doc).get(field));
for (int j = 1; j < sd.length; j++) {
long act = Long.parseLong(searcher.doc(sd[j].doc).get(field));
assertTrue("Docs should be sorted backwards", last > act);
last = act;
}
}
}
protected <N extends Number> NumericRangeQuery<?> numericRange(
Class<N> clazz,
String field,
@Nullable N min,
@Nullable N max,
boolean minInc,
boolean maxInc) {
if (clazz.equals(Integer.class)) {
return NumericRangeQuery.newIntRange(field, (Integer) min, (Integer) max, minInc, maxInc);
} else if (clazz.equals(Double.class)) {
return NumericRangeQuery.newDoubleRange(field, (Double) min, (Double) max, minInc, minInc);
} else if (clazz.equals(Float.class)) {
return NumericRangeQuery.newFloatRange(field, (Float) min, (Float) max, minInc, minInc);
} else if (clazz.equals(Long.class)) {
return NumericRangeQuery.newLongRange(field, (Long) min, (Long) max, minInc, minInc);
} else if (clazz.equals(Byte.class) || clazz.equals(Short.class)) {
return NumericRangeQuery.newIntRange(
field,
min != null ? min.intValue() : null,
max != null ? max.intValue() : null,
minInc,
maxInc);
} else {
throw new IllegalArgumentException("Unsupported numeric type " + clazz.getName());
}
}
/**
* Factory that creates a <code>NumericRangeFilter</code>, that queries a <code>long</code> range
* using the default <code>precisionStep</code> {@link NumericUtils#PRECISION_STEP_DEFAULT} (4).
* You can have half-open ranges (which are in fact </≤ or >/≥ queries) by setting the
* min or max value to <code>null</code>. By setting inclusive to false, it will match all
* documents excluding the bounds, with inclusive on, the boundaries are hits, too.
*/
public static NumericRangeFilter<Long> newLongRange(
final String field,
Long min,
Long max,
final boolean minInclusive,
final boolean maxInclusive) {
return new NumericRangeFilter<Long>(
NumericRangeQuery.newLongRange(field, min, max, minInclusive, maxInclusive));
}
public void testOneMatchQuery() throws Exception {
NumericRangeQuery<Long> q =
NumericRangeQuery.newLongRange("ascfield8", 8, 1000L, 1000L, true, true);
assertSame(MultiTermQuery.CONSTANT_SCORE_BOOLEAN_QUERY_REWRITE, q.getRewriteMethod());
TopDocs topDocs = searcher.search(q, noDocs);
ScoreDoc[] sd = topDocs.scoreDocs;
assertNotNull(sd);
assertEquals("Score doc count", 1, sd.length);
}
private void testRangeSplit(int precisionStep) throws Exception {
final Random rnd = newRandom();
String field = "ascfield" + precisionStep;
// 50 random tests
for (int i = 0; i < 50; i++) {
long lower = (long) (rnd.nextDouble() * noDocs - noDocs / 2);
long upper = (long) (rnd.nextDouble() * noDocs - noDocs / 2);
if (lower > upper) {
long a = lower;
lower = upper;
upper = a;
}
// test inclusive range
Query tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, true, true);
TopDocs tTopDocs = searcher.search(tq, 1);
assertEquals(
"Returned count of range query must be equal to inclusive range length",
upper - lower + 1,
tTopDocs.totalHits);
// test exclusive range
tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, false, false);
tTopDocs = searcher.search(tq, 1);
assertEquals(
"Returned count of range query must be equal to exclusive range length",
Math.max(upper - lower - 1, 0),
tTopDocs.totalHits);
// test left exclusive range
tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, false, true);
tTopDocs = searcher.search(tq, 1);
assertEquals(
"Returned count of range query must be equal to half exclusive range length",
upper - lower,
tTopDocs.totalHits);
// test right exclusive range
tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, true, false);
tTopDocs = searcher.search(tq, 1);
assertEquals(
"Returned count of range query must be equal to half exclusive range length",
upper - lower,
tTopDocs.totalHits);
}
}
/**
* test for constant score + boolean query + filter, the other tests only use the constant score
* mode
*/
private void testRange(int precisionStep) throws Exception {
String field = "field" + precisionStep;
int count = 3000;
long lower = (distance * 3 / 2) + startOffset,
upper = lower + count * distance + (distance / 3);
NumericRangeQuery<Long> q =
NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, true, true);
NumericRangeFilter<Long> f =
NumericRangeFilter.newLongRange(field, precisionStep, lower, upper, true, true);
int lastTerms = 0;
for (byte i = 0; i < 3; i++) {
TopDocs topDocs;
int terms;
String type;
q.clearTotalNumberOfTerms();
f.clearTotalNumberOfTerms();
switch (i) {
case 0:
type = " (constant score filter rewrite)";
q.setRewriteMethod(MultiTermQuery.CONSTANT_SCORE_FILTER_REWRITE);
topDocs = searcher.search(q, null, noDocs, Sort.INDEXORDER);
terms = q.getTotalNumberOfTerms();
break;
case 1:
type = " (constant score boolean rewrite)";
q.setRewriteMethod(MultiTermQuery.CONSTANT_SCORE_BOOLEAN_QUERY_REWRITE);
topDocs = searcher.search(q, null, noDocs, Sort.INDEXORDER);
terms = q.getTotalNumberOfTerms();
break;
case 2:
type = " (filter)";
topDocs = searcher.search(new MatchAllDocsQuery(), f, noDocs, Sort.INDEXORDER);
terms = f.getTotalNumberOfTerms();
break;
default:
return;
}
if (VERBOSE)
System.out.println(
"Found " + terms + " distinct terms in range for field '" + field + "'" + type + ".");
ScoreDoc[] sd = topDocs.scoreDocs;
assertNotNull(sd);
assertEquals("Score doc count" + type, count, sd.length);
Document doc = searcher.doc(sd[0].doc);
assertEquals("First doc" + type, 2 * distance + startOffset, Long.parseLong(doc.get(field)));
doc = searcher.doc(sd[sd.length - 1].doc);
assertEquals(
"Last doc" + type, (1 + count) * distance + startOffset, Long.parseLong(doc.get(field)));
if (i > 0 && searcher.getIndexReader().getSequentialSubReaders().length == 1) {
assertEquals("Distinct term number is equal for all query types", lastTerms, terms);
}
lastTerms = terms;
}
}
public void testEqualsAndHash() throws Exception {
QueryUtils.checkHashEquals(NumericRangeQuery.newLongRange("test1", 4, 10L, 20L, true, true));
QueryUtils.checkHashEquals(NumericRangeQuery.newLongRange("test2", 4, 10L, 20L, false, true));
QueryUtils.checkHashEquals(NumericRangeQuery.newLongRange("test3", 4, 10L, 20L, true, false));
QueryUtils.checkHashEquals(NumericRangeQuery.newLongRange("test4", 4, 10L, 20L, false, false));
QueryUtils.checkHashEquals(NumericRangeQuery.newLongRange("test5", 4, 10L, null, true, true));
QueryUtils.checkHashEquals(NumericRangeQuery.newLongRange("test6", 4, null, 20L, true, true));
QueryUtils.checkHashEquals(NumericRangeQuery.newLongRange("test7", 4, null, null, true, true));
QueryUtils.checkEqual(
NumericRangeQuery.newLongRange("test8", 4, 10L, 20L, true, true),
NumericRangeQuery.newLongRange("test8", 4, 10L, 20L, true, true));
QueryUtils.checkUnequal(
NumericRangeQuery.newLongRange("test9", 4, 10L, 20L, true, true),
NumericRangeQuery.newLongRange("test9", 8, 10L, 20L, true, true));
QueryUtils.checkUnequal(
NumericRangeQuery.newLongRange("test10a", 4, 10L, 20L, true, true),
NumericRangeQuery.newLongRange("test10b", 4, 10L, 20L, true, true));
QueryUtils.checkUnequal(
NumericRangeQuery.newLongRange("test11", 4, 10L, 20L, true, true),
NumericRangeQuery.newLongRange("test11", 4, 20L, 10L, true, true));
QueryUtils.checkUnequal(
NumericRangeQuery.newLongRange("test12", 4, 10L, 20L, true, true),
NumericRangeQuery.newLongRange("test12", 4, 10L, 20L, false, true));
QueryUtils.checkUnequal(
NumericRangeQuery.newLongRange("test13", 4, 10L, 20L, true, true),
NumericRangeQuery.newFloatRange("test13", 4, 10f, 20f, true, true));
// difference to int range is tested in TestNumericRangeQuery32
}
private void testRandomTrieAndClassicRangeQuery(int precisionStep) throws Exception {
final Random rnd = newRandom();
String field = "field" + precisionStep;
int termCountT = 0, termCountC = 0;
for (int i = 0; i < 50; i++) {
long lower = (long) (rnd.nextDouble() * noDocs * distance) + startOffset;
long upper = (long) (rnd.nextDouble() * noDocs * distance) + startOffset;
if (lower > upper) {
long a = lower;
lower = upper;
upper = a;
}
// test inclusive range
NumericRangeQuery<Long> tq =
NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, true, true);
TermRangeQuery cq =
new TermRangeQuery(
field,
NumericUtils.longToPrefixCoded(lower),
NumericUtils.longToPrefixCoded(upper),
true,
true);
TopDocs tTopDocs = searcher.search(tq, 1);
TopDocs cTopDocs = searcher.search(cq, 1);
assertEquals(
"Returned count for NumericRangeQuery and TermRangeQuery must be equal",
cTopDocs.totalHits,
tTopDocs.totalHits);
termCountT += tq.getTotalNumberOfTerms();
termCountC += cq.getTotalNumberOfTerms();
// test exclusive range
tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, false, false);
cq =
new TermRangeQuery(
field,
NumericUtils.longToPrefixCoded(lower),
NumericUtils.longToPrefixCoded(upper),
false,
false);
tTopDocs = searcher.search(tq, 1);
cTopDocs = searcher.search(cq, 1);
assertEquals(
"Returned count for NumericRangeQuery and TermRangeQuery must be equal",
cTopDocs.totalHits,
tTopDocs.totalHits);
termCountT += tq.getTotalNumberOfTerms();
termCountC += cq.getTotalNumberOfTerms();
// test left exclusive range
tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, false, true);
cq =
new TermRangeQuery(
field,
NumericUtils.longToPrefixCoded(lower),
NumericUtils.longToPrefixCoded(upper),
false,
true);
tTopDocs = searcher.search(tq, 1);
cTopDocs = searcher.search(cq, 1);
assertEquals(
"Returned count for NumericRangeQuery and TermRangeQuery must be equal",
cTopDocs.totalHits,
tTopDocs.totalHits);
termCountT += tq.getTotalNumberOfTerms();
termCountC += cq.getTotalNumberOfTerms();
// test right exclusive range
tq = NumericRangeQuery.newLongRange(field, precisionStep, lower, upper, true, false);
cq =
new TermRangeQuery(
field,
NumericUtils.longToPrefixCoded(lower),
NumericUtils.longToPrefixCoded(upper),
true,
false);
tTopDocs = searcher.search(tq, 1);
cTopDocs = searcher.search(cq, 1);
assertEquals(
"Returned count for NumericRangeQuery and TermRangeQuery must be equal",
cTopDocs.totalHits,
tTopDocs.totalHits);
termCountT += tq.getTotalNumberOfTerms();
termCountC += cq.getTotalNumberOfTerms();
}
if (precisionStep == Integer.MAX_VALUE) {
assertEquals(
"Total number of terms should be equal for unlimited precStep", termCountT, termCountC);
} else {
System.out.println("Average number of terms during random search on '" + field + "':");
System.out.println(" Trie query: " + (((double) termCountT) / (50 * 4)));
System.out.println(" Classical query: " + (((double) termCountC) / (50 * 4)));
}
}
/**
* 测试各种查询条件
*
* @throws IOException
* @throws Exception
*/
@Test
public void testAdvancedQuery() throws Exception {
Term term = new Term("content", "spring");
// 1、最基本的查询条件
TermQuery query1 = new TermQuery(term);
// 2、数字范围查询条件 参数1、搜索域 参数2、范围的最小值 参数3、范围的最大值 参数4：是否包含最小值 参数5：是否包含最大值
NumericRangeQuery<Long> query2 =
NumericRangeQuery.newLongRange("size", 100L, 5000L, true, false);
// 3、布尔查询——实质是多条件的组合查询
BooleanQuery booleanQuery = new BooleanQuery();
booleanQuery.add(query1, Occur.MUST);
booleanQuery.add(query2, Occur.MUST_NOT);
// 4、matchall查询
MatchAllDocsQuery query3 = new MatchAllDocsQuery();
/** 以上的各种查询条件其实都是lucene底层查询语法的封装 */
/* System.out.println(query1);
System.out.println(query2);
System.out.println(booleanQuery);
System.out.println(query3);*/
/**
* content:spring size:[100 TO 5000} +content:spring -size:[100 TO 5000} :*
*
* <p>如果我们要直接使用查询语法来搜索，就需要用到queryparser来对语法进行解析得到各种查询条件query
*/
// --------------------------------------------------------------------------
// 使用查询解析器来解析语法生成查询条件
// lucene支持的常用查询与法
/** content:spring size:[100 TO 5000} +content:spring -size:[100 TO 5000} :* */
// 1、构造一个查询条件解析器 参数1：默认搜索域 参数2：解析查询语句时所用的分词器
QueryParser queryParser = new QueryParser("title", new IKAnalyzer());
// 2、传递一个用lucene查询语法表示的字符串给parser去解析
Query queryYufa1 = queryParser.parse("content:spring.txt");
System.out.println(queryYufa1);
// title:spring.txt title:spring title:txt
Query queryYufa2 = queryParser.parse("size:[2000 TO 4000}");
System.out.println(queryYufa2);
/** 多个域的查询解析器 */
String[] fields = {"title", "content", "author"};
// 参数1： 默认搜索域（多个） 参数2：用来解析查询短语的分词器
MultiFieldQueryParser queryParser2 = new MultiFieldQueryParser(fields, new IKAnalyzer());
// 可以给一个精确的查询语法来解析成查询条件对象query
Query query4 = queryParser2.parse("title:spring +content:springmvc");
// 也可以只给一个查询短语，让它自己去对默认搜索域进行解析
Query query5 = queryParser2.parse("小屌丝");
System.out.println("query4---- " + query4);
System.out.println("query5---- " + query5);
/**
* query4---- title:spring +content:springmvc query5---- (title:小 title:屌 title:丝) (content:小
* content:屌 content:丝) (author:小 author:屌 author:丝)
*/
TopDocs topDocs = indexSearcher.search(queryYufa2, 10);
int totalHits = topDocs.totalHits;
ScoreDoc[] scoreDocs = topDocs.scoreDocs;
System.out.println("总命中数： " + totalHits);
for (ScoreDoc scoreDoc : scoreDocs) {
Document doc = indexSearcher.doc(scoreDoc.doc);
System.out.println(doc.get("title"));
System.out.println(doc.get("content"));
System.out.println(doc.get("path"));
System.out.println(doc.get("size"));
}
}
@Override
public Query getRangeQuery(
QParser parser,
SchemaField field,
String min,
String max,
boolean minInclusive,
boolean maxInclusive) {
int ps = precisionStep;
Query query = null;
switch (type) {
case INTEGER:
query =
NumericRangeQuery.newIntRange(
field.getName(),
ps,
min == null ? null : Integer.parseInt(min),
max == null ? null : Integer.parseInt(max),
minInclusive,
maxInclusive);
break;
case FLOAT:
query =
NumericRangeQuery.newFloatRange(
field.getName(),
ps,
min == null ? null : Float.parseFloat(min),
max == null ? null : Float.parseFloat(max),
minInclusive,
maxInclusive);
break;
case LONG:
query =
NumericRangeQuery.newLongRange(
field.getName(),
ps,
min == null ? null : Long.parseLong(min),
max == null ? null : Long.parseLong(max),
minInclusive,
maxInclusive);
break;
case DOUBLE:
query =
NumericRangeQuery.newDoubleRange(
field.getName(),
ps,
min == null ? null : Double.parseDouble(min),
max == null ? null : Double.parseDouble(max),
minInclusive,
maxInclusive);
break;
case DATE:
query =
NumericRangeQuery.newLongRange(
field.getName(),
ps,
min == null ? null : dateField.parseMath(null, min).getTime(),
max == null ? null : dateField.parseMath(null, max).getTime(),
minInclusive,
maxInclusive);
break;
default:
throw new SolrException(
SolrException.ErrorCode.SERVER_ERROR, "Unknown type for trie field");
}
return query;
}
