use of de.ids_mannheim.korap.query.QueryBuilder in project Krill by KorAP.
the class TestHighlight method highlightEscapes.
@Test
public void highlightEscapes() throws IOException, QueryException {
KrillIndex ki = new KrillIndex();
FieldDocument fd = new FieldDocument();
fd.addString("ID", "doc-1");
fd.addString("UID", "1");
fd.addString("textSigle", "c1/d1/1");
// Make this clean for HTML and Brackets!
fd.addTV("base", "Mit \"Mann\" & {Ma\\us}", "[(0-3)s:Mit|i:mit|_0#0-3|-:t$<i>4|<>:base/t:t$<b>64<i>0<i>20<i>4<b>0]" + "[(4-10)s:\"Mann\"|i:\"mann\"|base/l:\"Mann\"|_1#4-10]" + "[(11-12)s:&|i:&|base/l:&|_2#11-12]" + "[(13-20)s:{Ma\\us}|i:{ma\\us}|_3#13-20]");
ki.addDoc(fd);
// Commit!
ki.commit();
QueryBuilder kq = new QueryBuilder("base");
SpanQuery q = (SpanQuery) kq.tag("base/t:t").toQuery();
Krill qs = new Krill(q);
qs.getMeta().getContext().left.setToken(true).setLength((short) 0);
qs.getMeta().getContext().right.setToken(true).setLength((short) 0);
Result kr = ki.search(qs);
assertEquals((long) 1, kr.getTotalResults());
assertEquals("[[Mit \"Mann\" & \\{Ma\\\\us\\}]]", kr.getMatch(0).getSnippetBrackets());
assertEquals("<span class=\"context-left\"></span><span class=\"match\"><mark>Mit "Mann" & {Ma\\us}</mark></span><span class=\"context-right\"></span>", kr.getMatch(0).getSnippetHTML());
assertEquals("match-c1/d1/1-p0-4", kr.getMatch(0).getID());
Match km = ki.getMatchInfo("match-c1/d1/1-p0-4", "base", true, (ArrayList) null, (ArrayList) null, true, true, false);
assertEquals(0, km.getStartPos());
assertEquals("<span class=\"context-left\"></span>" + "<span class=\"match\"><mark><span title=\"base/t:t\">" + "Mit " + "<span title=\"base/l:"Mann"\">" + ""Mann"" + "</span>" + " " + "<span title=\"base/l:&\">&</span>" + " " + "{Ma\\us}" + "</span>" + "</mark></span>" + "<span class=\"context-right\"></span>", km.getSnippetHTML());
}
use of de.ids_mannheim.korap.query.QueryBuilder in project Krill by KorAP.
the class TestHighlight method highlightMissingBug.
@Test
public void highlightMissingBug() throws IOException, QueryException {
KrillIndex ki = new KrillIndex();
FieldDocument fd = new FieldDocument();
fd.addString("ID", "doc-1");
fd.addString("UID", "1");
fd.addTV("base", "abab", "[(0-1)s:a|i:a|_0#0-1|-:t$<i>4]" + "[(1-2)s:b|i:b|_1#1-2]" + "[(2-3)s:a|i:c|_2#2-3]" + "[(3-4)s:b|i:a|_3#3-4]");
ki.addDoc(fd);
fd = new FieldDocument();
fd.addString("ID", "doc-2");
fd.addString("UID", "2");
fd.addTV("base", "aba", "[(0-1)s:a|i:a|_0#0-1|-:t$<i>3]" + "[(1-2)s:b|i:b|_1#1-2]" + "[(2-3)s:a|i:c|_2#2-3]");
ki.addDoc(fd);
// Commit!
ki.commit();
fd = new FieldDocument();
fd.addString("ID", "doc-3");
fd.addString("UID", "3");
fd.addTV("base", "abab", "[(0-1)s:a|i:a|_0#0-1|-:t$<i>4]" + "[(1-2)s:b|i:b|_1#1-2]" + "[(2-3)s:a|i:c|_2#2-3]" + "[(3-4)s:b|i:a|_3#3-4]");
ki.addDoc(fd);
// Commit!
ki.commit();
fd = new FieldDocument();
fd.addString("ID", "doc-4");
fd.addString("UID", "4");
fd.addTV("base", "aba", "[(0-1)s:a|i:a|_0#0-1|-:t$<i>3]" + "[(1-2)s:b|i:b|_1#1-2]" + "[(2-3)s:a|i:c|_2#2-3]");
ki.addDoc(fd);
// Commit!
ki.commit();
QueryBuilder kq = new QueryBuilder("base");
SpanQuery q = (SpanQuery) kq.or(kq.nr(1, kq.seg("s:a"))).or(kq.nr(2, kq.seg("s:b"))).toQuery();
Result kr = ki.search(q);
assertEquals((long) 14, kr.getTotalResults());
assertEquals("[[{1:a}]]bab", kr.getMatch(0).getSnippetBrackets());
assertEquals("a[[{2:b}]]ab", kr.getMatch(1).getSnippetBrackets());
assertEquals("ab[[{1:a}]]b", kr.getMatch(2).getSnippetBrackets());
assertEquals("aba[[{2:b}]]", kr.getMatch(3).getSnippetBrackets());
assertEquals("[[{1:a}]]ba", kr.getMatch(4).getSnippetBrackets());
assertEquals("a[[{2:b}]]a", kr.getMatch(5).getSnippetBrackets());
assertEquals("ab[[{1:a}]]", kr.getMatch(6).getSnippetBrackets());
assertEquals("[[{1:a}]]bab", kr.getMatch(7).getSnippetBrackets());
assertEquals("a[[{2:b}]]ab", kr.getMatch(8).getSnippetBrackets());
assertEquals("ab[[{1:a}]]b", kr.getMatch(9).getSnippetBrackets());
assertEquals("aba[[{2:b}]]", kr.getMatch(10).getSnippetBrackets());
assertEquals("[[{1:a}]]ba", kr.getMatch(11).getSnippetBrackets());
assertEquals("a[[{2:b}]]a", kr.getMatch(12).getSnippetBrackets());
assertEquals("ab[[{1:a}]]", kr.getMatch(13).getSnippetBrackets());
kq = new QueryBuilder("base");
q = (SpanQuery) kq.or(kq.nr(1, kq.seg("i:a"))).or(kq.nr(2, kq.seg("i:c"))).toQuery();
Krill qs = new Krill(q);
qs.getMeta().getContext().left.setToken(true).setLength((short) 1);
qs.getMeta().getContext().right.setToken(true).setLength((short) 1);
kr = ki.search(qs);
assertEquals((long) 10, kr.getTotalResults());
assertEquals("[[{1:a}]]b ...", kr.getMatch(0).getSnippetBrackets());
assertEquals("... b[[{2:a}]]b", kr.getMatch(1).getSnippetBrackets());
assertEquals("... a[[{1:b}]]", kr.getMatch(2).getSnippetBrackets());
assertEquals("[[{1:a}]]b ...", kr.getMatch(3).getSnippetBrackets());
assertEquals("... b[[{2:a}]]", kr.getMatch(4).getSnippetBrackets());
assertEquals("[[{1:a}]]b ...", kr.getMatch(5).getSnippetBrackets());
assertEquals("... b[[{2:a}]]b", kr.getMatch(6).getSnippetBrackets());
assertEquals("... a[[{1:b}]]", kr.getMatch(7).getSnippetBrackets());
assertEquals("[[{1:a}]]b ...", kr.getMatch(8).getSnippetBrackets());
assertEquals("... b[[{2:a}]]", kr.getMatch(9).getSnippetBrackets());
qs.getMeta().getContext().left.setToken(true).setLength((short) 0);
qs.getMeta().getContext().right.setToken(true).setLength((short) 0);
kr = ki.search(qs);
assertEquals((long) 10, kr.getTotalResults());
assertEquals("[[{1:a}]] ...", kr.getMatch(0).getSnippetBrackets());
assertEquals("... [[{2:a}]] ...", kr.getMatch(1).getSnippetBrackets());
assertEquals("... [[{1:b}]]", kr.getMatch(2).getSnippetBrackets());
assertEquals("[[{1:a}]] ...", kr.getMatch(3).getSnippetBrackets());
assertEquals("... [[{2:a}]]", kr.getMatch(4).getSnippetBrackets());
assertEquals("[[{1:a}]] ...", kr.getMatch(5).getSnippetBrackets());
assertEquals("... [[{2:a}]] ...", kr.getMatch(6).getSnippetBrackets());
assertEquals("... [[{1:b}]]", kr.getMatch(7).getSnippetBrackets());
assertEquals("[[{1:a}]] ...", kr.getMatch(8).getSnippetBrackets());
assertEquals("... [[{2:a}]]", kr.getMatch(9).getSnippetBrackets());
q = (SpanQuery) kq.nr(3, kq.or(kq.nr(1, kq.seg("i:a"))).or(kq.nr(2, kq.seg("i:c")))).toQuery();
qs = new Krill(q);
qs.getMeta().getContext().left.setToken(true).setLength((short) 0);
qs.getMeta().getContext().right.setToken(true).setLength((short) 0);
kr = ki.search(qs);
assertEquals((long) 10, kr.getTotalResults());
assertEquals("[[{1:{3:a}}]] ...", kr.getMatch(0).getSnippetBrackets());
assertEquals("... [[{2:{3:a}}]] ...", kr.getMatch(1).getSnippetBrackets());
assertEquals("... [[{1:{3:b}}]]", kr.getMatch(2).getSnippetBrackets());
assertEquals("[[{1:{3:a}}]] ...", kr.getMatch(3).getSnippetBrackets());
assertEquals("... [[{2:{3:a}}]]", kr.getMatch(4).getSnippetBrackets());
assertEquals("[[{1:{3:a}}]] ...", kr.getMatch(5).getSnippetBrackets());
assertEquals("... [[{2:{3:a}}]] ...", kr.getMatch(6).getSnippetBrackets());
assertEquals("... [[{1:{3:b}}]]", kr.getMatch(7).getSnippetBrackets());
assertEquals("[[{1:{3:a}}]] ...", kr.getMatch(8).getSnippetBrackets());
assertEquals("... [[{2:{3:a}}]]", kr.getMatch(9).getSnippetBrackets());
}
use of de.ids_mannheim.korap.query.QueryBuilder in project Krill by KorAP.
the class TestBenchmarkSamples method simpleSegmentQuery.
@Test
public void simpleSegmentQuery() throws Exception {
// Construct index
KrillIndex ki = new KrillIndex();
// Indexing test files
for (String i : new String[] { "00001", "00002", "00003", "00004", "00005", "00006", "02439" }) {
ki.addDoc(getClass().getResourceAsStream("/wiki/" + i + ".json.gz"), true);
}
;
ki.commit();
t1 = System.nanoTime();
for (int i = 1; i <= rounds; i++) {
final QueryBuilder qb = new QueryBuilder("tokens");
final Krill ks = new Krill(qb.seg("mate/m:gender:masc").toQuery());
final Result kr = ks.apply(ki);
assertEquals(kr.getTotalResults(), 497);
}
;
t2 = System.nanoTime();
double seconds = (double) (t2 - t1) / 1000000000.0;
System.err.println("Seconds: " + seconds);
// Seconds: 9.465514311
// Seconds: 9.302011468
// Seconds: 9.052496918
// Seconds: 9.0567007
// Seconds: 9.113724089
// Seconds: 8.700548842
// Seconds: 9.390980437
// Seconds: 8.817503952
}
use of de.ids_mannheim.korap.query.QueryBuilder in project Krill by KorAP.
the class TestKrillCollectionIndex method filterExampleAtomicLegacy.
@Test
public void filterExampleAtomicLegacy() throws Exception {
// That's exactly the same test class, but with multiple atomic indices
// Construct index
KrillIndex ki = new KrillIndex();
// Indexing test files
for (String i : new String[] { "00001", "00002", "00003", "00004", "00005", "00006", "02439" }) {
ki.addDoc(getClass().getResourceAsStream("/wiki/" + i + ".json.gz"), true);
ki.commit();
}
;
CollectionBuilder kf = new CollectionBuilder();
// Create Virtual collections:
KrillCollection kc = new KrillCollection(ki);
assertEquals("Documents", 7, kc.numberOf("documents"));
// If this is set - everything is fine automatically ...
kc.filter(kc.build().term("corpusID", "WPD"));
assertEquals("Documents", 7, kc.numberOf("documents"));
// The virtual collection consists of all documents that have the textClass "reisen" and "freizeit"
/*
kc.filter(kf.and("textClass", "reisen").and("textClass",
"freizeit-unterhaltung"));
*/
kc.filter(kc.build().andGroup().with(kc.build().term("textClass", "reisen")).with(kc.build().term("textClass", "freizeit-unterhaltung")));
assertEquals("Documents", 5, kc.numberOf("documents"));
assertEquals("Tokens", 1678, kc.numberOf("tokens"));
assertEquals("Sentences", 194, kc.numberOf("sentences"));
assertEquals("Paragraphs", 139, kc.numberOf("paragraphs"));
// Subset this to all documents that have also the text
// kc.filter(kf.and("textClass", "kultur"));
kc.filter(kc.build().term("textClass", "kultur"));
assertEquals("Documents", 1, kc.numberOf("documents"));
assertEquals("Tokens", 405, kc.numberOf("tokens"));
assertEquals("Sentences", 75, kc.numberOf("sentences"));
assertEquals("Paragraphs", 48, kc.numberOf("paragraphs"));
// This is already filtered though ...
// kc.filter(kf.and("corpusID", "WPD"));
kc.filter(kc.build().term("corpusID", "WPD"));
assertEquals("Documents", 1, kc.numberOf("documents"));
assertEquals("Tokens", 405, kc.numberOf("tokens"));
assertEquals("Sentences", 75, kc.numberOf("sentences"));
assertEquals("Paragraphs", 48, kc.numberOf("paragraphs"));
// Create a query
Krill ks = new Krill(new QueryBuilder("tokens").seg("opennlp/p:NN").with("tt/p:NN"));
ks.setCollection(kc).getMeta().setStartIndex(0).setCount((short) 20).setContext(new SearchContext(true, (short) 5, true, (short) 5));
Result kr = ks.apply(ki);
/*
Result kr = ki.search(kc, query, 0, (short) 20, true, (short) 5, true,
(short) 5);
*/
assertEquals(kr.getTotalResults(), 70);
// kc.extend(kf.and("textClass", "uninteresting"));
kc.extend(kc.build().term("textClass", "uninteresting"));
assertEquals("Documents", 1, kc.numberOf("documents"));
kc.extend(kc.build().term("textClass", "wissenschaft"));
assertEquals("Documents", 3, kc.numberOf("documents"));
assertEquals("Tokens", 1669, kc.numberOf("tokens"));
assertEquals("Sentences", 188, kc.numberOf("sentences"));
assertEquals("Paragraphs", 130, kc.numberOf("paragraphs"));
// System.err.println(kc.toString());
// Test collectionbuilder simplifier!
/*
OrGroup(
AndGroup(
corpusID:WPD
textClass:reisen
textClass:freizeit-unterhaltung
textClass:kultur
corpusID:WPD
)
textClass:uninteresting
textClass:wissenschaft
)
*/
assertTrue(ki.delDocs("textClass", "wissenschaft"));
ki.commit();
assertEquals("Documents", 1, kc.numberOf("documents"));
assertEquals("Tokens", 405, kc.numberOf("tokens"));
assertEquals("Sentences", 75, kc.numberOf("sentences"));
assertEquals("Paragraphs", 48, kc.numberOf("paragraphs"));
}
use of de.ids_mannheim.korap.query.QueryBuilder in project Krill by KorAP.
the class TestKrillCollectionIndex method filterExample2Legacy.
@Test
public void filterExample2Legacy() throws Exception {
// Construct index
KrillIndex ki = new KrillIndex();
// Indexing test files
for (String i : new String[] { "00001", "00002", "00003", "00004", "00005", "00006", "02439" }) {
ki.addDoc(getClass().getResourceAsStream("/wiki/" + i + ".json.gz"), true);
}
;
ki.commit();
ki.addDoc(getClass().getResourceAsStream("/wiki/00012-fakemeta.json.gz"), true);
ki.commit();
/*
CollectionBuilderLegacy kf = new CollectionBuilderLegacy();
// Create Virtual collections:
KrillCollectionLegacy kc = new KrillCollectionLegacy(ki);
kc.filter(kf.and("textClass", "reisen").and("textClass",
"freizeit-unterhaltung"));
*/
KrillCollection kc = new KrillCollection(ki);
CollectionBuilder cb = kc.build();
kc.filter(cb.andGroup().with(cb.term("textClass", "reisen")).with(cb.term("textClass", "freizeit-unterhaltung")));
assertEquals("Documents", 5, kc.numberOf("documents"));
assertEquals("Tokens", 1678, kc.numberOf("tokens"));
assertEquals("Sentences", 194, kc.numberOf("sentences"));
assertEquals("Paragraphs", 139, kc.numberOf("paragraphs"));
// Create a query
Krill ks = new Krill(new QueryBuilder("tokens").seg("opennlp/p:NN").with("tt/p:NN"));
ks.setCollection(kc).getMeta().setStartIndex(0).setCount((short) 20).setContext(new SearchContext(true, (short) 5, true, (short) 5));
Result kr = ks.apply(ki);
assertEquals(kr.getTotalResults(), 369);
// kc.filter(kf.and("corpusID", "QQQ"));
kc.filter(cb.term("corpusID", "QQQ"));
assertEquals("Documents", 0, kc.numberOf("documents"));
assertEquals("Tokens", 0, kc.numberOf("tokens"));
assertEquals("Sentences", 0, kc.numberOf("sentences"));
assertEquals("Paragraphs", 0, kc.numberOf("paragraphs"));
ks.setCollection(kc);
// Create a query
kr = ks.apply(ki);
/*
kr = ki.search(kc, query, 0, (short) 20, true, (short) 5, true,
(short) 5);
*/
assertEquals(kr.getTotalResults(), 0);
}
Aggregations