use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.
the class TestMatchIndex method indexExampleFocusWithSpan.
@Test
public void indexExampleFocusWithSpan() throws IOException {
KrillIndex ki = new KrillIndex();
// abcabcabac
FieldDocument fd = new FieldDocument();
fd.addTV("base", "abcabcabac", "[(0-1)s:a|i:a|_0$<i>0<i>1|-:t$<i>10]" + "[(1-2)s:b|i:b|_1$<i>1<i>2|<>:s$<b>64<i>1<i>5<i>5]" + "[(2-3)s:c|i:c|_2$<i>2<i>3|<>:s$<b>64<i>2<i>7<i>7]" + "[(3-4)s:a|i:a|_3$<i>3<i>4]" + "[(4-5)s:b|i:b|_4$<i>4<i>5]" + "[(5-6)s:c|i:c|_5$<i>5<i>6]" + "[(6-7)s:a|i:a|_6$<i>6<i>7]" + "[(7-8)s:b|i:b|_7$<i>7<i>8]" + "[(8-9)s:a|i:a|_8$<i>8<i>9]" + "[(9-10)s:c|i:c|_9$<i>9<i>10]");
ki.addDoc(fd);
ki.commit();
SpanQuery sq;
Result kr;
// sq = new SpanWithinQuery(new SpanClassQuery(new SpanElementQuery(
// "base", "s"), (byte) 2), new SpanClassQuery(new SpanTermQuery(
// new Term("base", "s:b")), (byte) 3));
//
// kr = ki.search(sq, (short) 10);
// assertEquals(kr.getSerialQuery(),
// "spanContain({2: <base:s />}, {3: base:s:b})");
// assertEquals(kr.getMatch(0).getSnippetBrackets(),
// "a[{2:{3:b}cab}]cabac");
sq = new SpanFocusQuery(new SpanWithinQuery(new SpanClassQuery(new SpanElementQuery("base", "s"), (byte) 2), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:b")), (byte) 3)), (byte) 3);
kr = ki.search(sq, (short) 10);
assertEquals(kr.getSerialQuery(), "focus(3: spanContain({2: <base:s />}, {3: base:s:b}))");
assertEquals(kr.getMatch(0).getSnippetBrackets(), "a[[{3:b}]]cabcab ...");
}
use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.
the class TestMatchIndex method indexExampleFocusWithSkip.
@Ignore
public void indexExampleFocusWithSkip() throws IOException {
KrillIndex ki = new KrillIndex();
// abcabcabac
FieldDocument fd = new FieldDocument();
fd.addTV("base", "abcabcabac", // The payload should be ignored
// |<>:p#0-10<i>9]" +
"[(0-1)s:a|i:a|_0$<i>0<i>1|-:t$<i>10]" + "[(1-2)s:b|i:b|_1$<i>1<i>2|<>:s$<b>64<i>1<i>5<i>5]" + "[(2-3)s:c|i:c|_2$<i>2<i>3|<>:s$<b>64<i>2<i>7<i>7]" + "[(3-4)s:a|i:a|_3$<i>3<i>4]" + "[(4-5)s:b|i:b|_4$<i>4<i>5]" + "[(5-6)s:c|i:c|_5$<i>5<i>6]" + "[(6-7)s:a|i:a|_6$<i>6<i>7]" + "[(7-8)s:b|i:b|_7$<i>7<i>8]" + "[(8-9)s:a|i:a|_8$<i>8<i>9]" + "[(9-10)s:c|i:c|_9$<i>9<i>10]");
ki.addDoc(fd);
fd = new FieldDocument();
fd.addTV("base", "gbcgbcgbgc", "[(0-1)s:g|i:g|_0$<i>0<i>1|-:t$<i>10|<>:p$<b>64<i>0<i>10<i>9]" + "[(1-2)s:b|i:b|_1$<i>1<i>2|<>:s$<b>64<i>1<i>5<i>5]" + "[(2-3)s:c|i:c|_2$<i>2<i>3|<>:s$<b>64<i>2<i>7<i>7]" + "[(3-4)s:g|i:g|_3$<i>3<i>4]" + "[(4-5)s:b|i:b|_4$<i>4<i>5]" + "[(5-6)s:c|i:c|_5$<i>5<i>6]" + "[(6-7)s:g|i:g|_6$<i>6<i>7]" + "[(7-8)s:b|i:b|_7$<i>7<i>8]" + "[(8-9)s:g|i:g|_8$<i>8<i>9]" + "[(9-10)s:c|i:c|_9$<i>9<i>10]");
ki.addDoc(fd);
fd = new FieldDocument();
fd.addTV("base", "gbcgbcgbgc", "[(0-1)s:g|i:g|_0$<i>0<i>1|-:t$<i>10]" + "[(1-2)s:b|i:b|_1$<i>1<i>2]" + "[(2-3)s:c|i:c|_2$<i>2<i>3]" + "[(3-4)s:g|i:g|_3$<i>3<i>4]" + "[(4-5)s:b|i:b|_4$<i>4<i>5]" + "[(5-6)s:c|i:c|_5$<i>5<i>6]" + "[(6-7)s:g|i:g|_6$<i>6<i>7]" + "[(7-8)s:b|i:b|_7$<i>7<i>8]" + "[(8-9)s:g|i:g|_8$<i>8<i>9]" + "[(9-10)s:c|i:c|_9$<i>9<i>10]");
ki.addDoc(fd);
fd = new FieldDocument();
// contains(<p>, focus(3: contains({2:<s>}, {3:a})))
fd.addTV("base", "acabcabac", "[(0-1)s:a|i:a|_0$<i>0<i>1|-:t$<i>10|<>:p$<b>64<i>0<i>9<i>8]" + "[(1-2)s:b|i:b|_1$<i>1<i>2|<>:s$<b>64<i>1<i>5<i>5]" + "[(2-3)s:a|i:a|_2$<i>2<i>3|<>:s$<b>64<i>2<i>7<i>7]" + "[(3-4)s:b|i:b|_3$<i>3<i>4]" + "[(4-5)s:c|i:c|_4$<i>4<i>5]" + "[(5-6)s:a|i:a|_5$<i>5<i>6]" + "[(6-7)s:b|i:b|_6$<i>6<i>7]" + "[(7-8)s:a|i:a|_7$<i>7<i>8]" + "[(8-9)s:c|i:c|_8$<i>8<i>9]");
ki.addDoc(fd);
ki.commit();
SpanQuery sq;
Result kr;
KrillCollection kc = new KrillCollection(ki);
assertEquals("Documents", 4, kc.numberOf("documents"));
// within(<p>, focus(3:within({2:<s>}, {3:a})))
sq = new SpanWithinQuery(new SpanElementQuery("base", "p"), new SpanFocusQuery(new SpanWithinQuery(new SpanClassQuery(new SpanElementQuery("base", "s"), (byte) 2), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:a")), (byte) 3)), (byte) 3));
// fail("Skipping may go horribly wrong! (Known issue)");
Krill ks = new Krill(sq);
ks.getMeta().setStartIndex(0).setCount((short) 20).setContext(new SearchContext(true, (short) 5, true, (short) 5));
kr = ks.apply(ki);
// kr = ki.search(kc, sq, 0, (short) 20, true, (short) 5, true, (short) 5);
assertEquals(kr.getSerialQuery(), "spanContain(<base:p />, focus(3: spanContain({2: <base:s />}, {3: base:s:a})))");
assertEquals(12, kr.getTotalResults());
assertEquals("[a{2:bc{3:a}b}cabac]", kr.getMatch(0).getSnippetBrackets());
assertEquals("[ab{2:c{3:a}bcab}ac]", kr.getMatch(1).getSnippetBrackets());
assertEquals("[ab{2:cabc{3:a}}bac]", kr.getMatch(2).getSnippetBrackets());
}
use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.
the class TestMatchIndex method testEmbeddedClassQuery.
@Test
public void testEmbeddedClassQuery() throws IOException {
KrillIndex ki = new KrillIndex();
// abcabcabac
FieldDocument fd = new FieldDocument();
fd.addTV("base", "abcabcabac", "[(0-1)s:a|i:a|_0$<i>0<i>1|-:t$<i>10]" + "[(1-2)s:b|i:b|_1$<i>1<i>2]" + "[(2-3)s:c|i:c|_2$<i>2<i>3]" + "[(3-4)s:a|i:a|_3$<i>3<i>4]" + "[(4-5)s:b|i:b|_4$<i>4<i>5]" + "[(5-6)s:c|i:c|_5$<i>5<i>6]" + "[(6-7)s:a|i:a|_6$<i>6<i>7]" + "[(7-8)s:b|i:b|_7$<i>7<i>8]" + "[(8-9)s:a|i:a|_8$<i>8<i>9]" + "[(9-10)s:c|i:c|_9$<i>9<i>10]");
ki.addDoc(fd);
ki.commit();
SpanQuery sq;
Result kr;
sq = new SpanFocusQuery(new SpanClassQuery(new SpanNextQuery(new SpanClassQuery(new SpanTermQuery(new Term("base", "s:b")), (byte) 1), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:c")), (byte) 2)), (byte) 3), (byte) 3);
kr = ki.search(sq, (short) 10);
assertEquals("totalResults", kr.getTotalResults(), 2);
assertEquals("StartPos (0)", 1, kr.getMatch(0).startPos);
assertEquals("EndPos (0)", 3, kr.getMatch(0).endPos);
assertEquals("SnippetBrackets (0)", "a[[{3:{1:b}{2:c}}]]abcaba ...", kr.getMatch(0).getSnippetBrackets());
assertEquals("StartPos (1)", 4, kr.getMatch(1).startPos);
assertEquals("EndPos (1)", 6, kr.getMatch(1).endPos);
assertEquals("SnippetBrackets (1)", "abca[[{3:{1:b}{2:c}}]]abac", kr.getMatch(1).getSnippetBrackets());
assertEquals("Document count", 1, ki.numberOf("base", "documents"));
assertEquals("Token count", 10, ki.numberOf("base", "t"));
}
use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.
the class SpanQueryWrapper method toQuery.
/**
* Serialize the wrapped query and return a SpanQuery.
* This will be the final query and may be rewritten.
*
* @return A {@link SpanQuery} object.
* @throws QueryException
*/
public SpanQuery toQuery() throws QueryException {
if (this.isNull() || this.isEmpty()) {
return null;
}
;
// Wrap the query in a <base/s=t>, if it's extended to the right
if (this.isExtendedToTheRight()) {
return new SpanFocusQuery(new SpanWithinQuery("base/s:t", new SpanClassQuery(this.toFragmentQuery(), (byte) 254)), (byte) 254);
}
;
SpanQuery sq = this.toFragmentQuery();
log.info(sq.toString());
return sq;
}
use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.
the class TestNextIndex method indexExample6.
/**
* Skip to NextSpan
*/
@Test
public void indexExample6() throws IOException {
KrillIndex ki = new KrillIndex();
ki.addDoc(createFieldDoc1());
ki.addDoc(createFieldDoc2());
ki.addDoc(createFieldDoc3());
ki.commit();
SpanQuery sq = new SpanNextQuery(new SpanTermQuery(new Term("base", "s:c")), new SpanNextQuery(new SpanTermQuery(new Term("base", "s:d")), new SpanTermQuery(new Term("base", "s:b"))));
Result kr = ki.search(sq, (short) 10);
assertEquals("totalResults", kr.getTotalResults(), 1);
assertEquals("doc-number", 2, kr.getMatch(0).getLocalDocID());
assertEquals("StartPos", 0, kr.getMatch(0).startPos);
assertEquals("EndPos", 3, kr.getMatch(0).endPos);
sq = new SpanNextQuery(new SpanTermQuery(new Term("base", "s:c")), new SpanNextQuery(new SpanFocusQuery(new SpanClassQuery(new SpanTermQuery(new Term("base", "s:d")), (byte) 1), (byte) 1), new SpanFocusQuery(new SpanClassQuery(new SpanTermQuery(new Term("base", "s:b")), (byte) 2), (byte) 2)));
kr = ki.search(sq, (short) 10);
assertEquals("doc-number", 2, kr.getMatch(0).getLocalDocID());
assertEquals("StartPos", 0, kr.getMatch(0).startPos);
assertEquals("EndPos", 3, kr.getMatch(0).endPos);
// for (Match km : kr.getMatches()) {
// System.out.println(km.getStartPos() + "," + km.getEndPos()
// + " "
// + km.getSnippetBrackets());
// }
}
Aggregations