Merge branch 'master' of ssh://korap.ids-mannheim.de:29418/KorAP/Krill
diff --git a/.gitignore b/.gitignore
index d7c1747..13c3b07 100644
--- a/.gitignore
+++ b/.gitignore
@@ -13,5 +13,6 @@
# /src/main/resources/
/src/main/resources/server.properties
+/src/main/resources/krill.properties
/src/main/resources/korap.conf
/bin
diff --git a/Changes b/Changes
index a6f30b2..7410203 100644
--- a/Changes
+++ b/Changes
@@ -1,3 +1,6 @@
+0.52 2015-06-10
+ - [bugfix] Fixed payload filtering in FocusSpans (margaretha)
+
0.51 2015-03-17
- This is a major version (prepared for the GitHub release)
- [cleanup] Changed groupID to "de.ids_mannheim.korap",
diff --git a/misc/krill.png b/misc/krill.png
new file mode 100644
index 0000000..1adab86
--- /dev/null
+++ b/misc/krill.png
Binary files differ
diff --git a/pom.xml b/pom.xml
index 821b2f1..1757cd9 100644
--- a/pom.xml
+++ b/pom.xml
@@ -24,7 +24,7 @@
<groupId>de.ids_mannheim.korap</groupId>
<artifactId>Krill</artifactId>
- <version>0.51</version>
+ <version>0.52</version>
<packaging>jar</packaging>
<name>Krill</name>
diff --git a/src/main/java/de/ids_mannheim/korap/KrillQuery.java b/src/main/java/de/ids_mannheim/korap/KrillQuery.java
index c966cc0..5a05bd5 100644
--- a/src/main/java/de/ids_mannheim/korap/KrillQuery.java
+++ b/src/main/java/de/ids_mannheim/korap/KrillQuery.java
@@ -9,6 +9,8 @@
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
+import com.fasterxml.jackson.databind.node.ArrayNode;
+import com.fasterxml.jackson.databind.node.ObjectNode;
import de.ids_mannheim.korap.query.QueryBuilder;
import de.ids_mannheim.korap.query.SpanWithinQuery;
@@ -17,6 +19,7 @@
import de.ids_mannheim.korap.query.wrap.SpanClassQueryWrapper;
import de.ids_mannheim.korap.query.wrap.SpanFocusQueryWrapper;
import de.ids_mannheim.korap.query.wrap.SpanQueryWrapper;
+import de.ids_mannheim.korap.query.wrap.SpanReferenceQueryWrapper;
import de.ids_mannheim.korap.query.wrap.SpanRegexQueryWrapper;
import de.ids_mannheim.korap.query.wrap.SpanRelationWrapper;
import de.ids_mannheim.korap.query.wrap.SpanRepetitionQueryWrapper;
@@ -85,10 +88,6 @@
private static final int MAX_CLASS_NUM = 255; // 127;
- // Variables used for relation queries
- private String direction;
- private byte[] classNumbers;
-
// Private class for koral:boundary objects
private class Boundary {
public int min, max;
@@ -121,7 +120,6 @@
};
};
-
/**
* Constructs a new object for query deserialization
* and building. Expects the name of an index field
@@ -133,10 +131,6 @@
*/
public KrillQuery (String field) {
this.field = field;
- this.direction = ">:";
- this.classNumbers = new byte[2];
- this.classNumbers[0] = (byte) 1;
- this.classNumbers[1] = (byte) 2;
};
@@ -229,7 +223,6 @@
throw new QueryException(766,
"Peripheral references are currently not supported");
}
- ;
JsonNode operands = json.get("operands");
@@ -257,9 +250,6 @@
if (number > MAX_CLASS_NUM)
throw new QueryException(709,
"Valid class numbers exceeded");
-
- this.classNumbers = null;
-
}
// Reference based on spans
@@ -303,12 +293,6 @@
// Get wrapped token
return this._segFromJson(json.get("wrap"));
- case "koral:relation":
- if (!json.has("wrap")) {
- throw new QueryException(718, "Missing relation term");
- }
- return this._termFromJson(json.get("wrap"), direction);
-
case "koral:span":
return this._termFromJson(json);
};
@@ -396,6 +380,11 @@
if (DEBUG)
log.trace("Operands are {}", operands);
+ SpanQueryWrapper spanReferenceQueryWrapper = _operationReferenceFromJSON(json, operands);
+ if (spanReferenceQueryWrapper != null) {
+ return spanReferenceQueryWrapper;
+ }
+
// Branch on operation
switch (operation) {
case "operation:junction":
@@ -435,6 +424,87 @@
throw new QueryException(711, "Unknown group operation");
};
+ private SpanQueryWrapper _operationReferenceFromJSON(JsonNode node, JsonNode operands)
+ throws QueryException {
+ boolean isReference = false;
+ int classNum = -1;
+ int refOperandNum = -1;
+ JsonNode childNode;
+
+ for (int i = 0; i < operands.size(); i++) {
+ childNode = operands.get(i);
+ if (childNode.has("@type")
+ && childNode.get("@type").asText()
+ .equals("koral:reference")
+ && childNode.has("operation")
+ && childNode.get("operation").asText()
+ .equals("operation:focus")
+ && !childNode.has("operands")) {
+
+ if (childNode.has("classRef")) {
+ classNum = childNode.get("classRef").get(0).asInt();
+ refOperandNum = i;
+ isReference = true;
+ break;
+ }
+ }
+ }
+
+ if (isReference) {
+ JsonNode resolvedNode = _resolveReference(node, operands,
+ refOperandNum, classNum);
+ return new SpanReferenceQueryWrapper(fromJson(resolvedNode),
+ (byte) classNum);
+ }
+
+ return null;
+ }
+
+ private JsonNode _resolveReference(JsonNode node, JsonNode operands,
+ int refOperandNum, int classNum) throws QueryException {
+ JsonNode referent = null;
+ ObjectMapper m = new ObjectMapper();
+ ArrayNode newOperands = m.createArrayNode();
+ boolean isReferentFound = false;
+ for (int i = 0; i < operands.size(); i++) {
+ if (i != refOperandNum) {
+ if (!isReferentFound) {
+ referent = _extractReferentClass(operands.get(i), classNum);
+ if (referent != null) isReferentFound = true;
+ }
+ newOperands.insert(i, operands.get(i));
+ }
+ }
+
+ if (isReferentFound) {
+ newOperands.insert(refOperandNum, referent);
+ ((ObjectNode) node).set("operands", newOperands);
+ return node;
+ }
+ else
+ throw new QueryException("Referent node is not found");
+
+ }
+
+ private JsonNode _extractReferentClass(JsonNode node, int classNum) {
+ JsonNode referent;
+ if (node.has("classOut") && node.get("classOut").asInt() == classNum) {
+ // System.out.println("found: " + node.toString());
+ return node;
+ }
+ else {
+ if (node.has("operands") && node.get("operands").isArray()) {
+ for (JsonNode childOperand : node.get("operands")) {
+ referent = _extractReferentClass(childOperand, classNum);
+ if (referent != null) {
+ return referent;
+ }
+ }
+ }
+ }
+ return null;
+ }
+
private SpanQueryWrapper _operationRelationFromJson (JsonNode operands,
JsonNode relation) throws QueryException {
@@ -446,16 +516,28 @@
SpanQueryWrapper operand1 = fromJson(operands.get(0));
SpanQueryWrapper operand2 = fromJson(operands.get(1));
-
- if (operand1.isEmpty()) {
+
+ String direction = ">:";
+ if (operand1.isEmpty() && !operand2.isEmpty()) {
direction = "<:";
}
- SpanQueryWrapper relationWrapper = fromJson(relation);
-
- return new SpanRelationWrapper(relationWrapper, operand1, operand2,
- classNumbers);
-
+ if (!relation.has("@type"))
+ throw new QueryException(701,
+ "JSON-LD group has no @type attribute");
+
+ if (relation.get("@type").asText().equals("koral:relation")) {
+ if (!relation.has("wrap")) {
+ throw new QueryException(718, "Missing relation term");
+ }
+ SpanQueryWrapper relationWrapper = _termFromJson(
+ relation.get("wrap"),
+ direction);
+ return new SpanRelationWrapper(relationWrapper, operand1, operand2);
+ }
+ else {
+ throw new QueryException(713, "Query type is not supported");
+ }
}
diff --git a/src/main/java/de/ids_mannheim/korap/query/SimpleSpanQuery.java b/src/main/java/de/ids_mannheim/korap/query/SimpleSpanQuery.java
index ff47287..de1e4c8 100644
--- a/src/main/java/de/ids_mannheim/korap/query/SimpleSpanQuery.java
+++ b/src/main/java/de/ids_mannheim/korap/query/SimpleSpanQuery.java
@@ -72,9 +72,10 @@
protected SpanQuery firstClause = null, secondClause = null;
protected List<SpanQuery> clauseList = null;
- private String field;
+ protected String field;
protected boolean collectPayloads;
+ public SimpleSpanQuery () {}
/**
* Constructs a new SimpleSpanQuery using the specified
diff --git a/src/main/java/de/ids_mannheim/korap/query/SpanClassFilterQuery.java b/src/main/java/de/ids_mannheim/korap/query/SpanClassFilterQuery.java
new file mode 100644
index 0000000..48790bc
--- /dev/null
+++ b/src/main/java/de/ids_mannheim/korap/query/SpanClassFilterQuery.java
@@ -0,0 +1,86 @@
+package de.ids_mannheim.korap.query;
+
+import java.io.IOException;
+import java.util.Map;
+
+import org.apache.lucene.index.AtomicReaderContext;
+import org.apache.lucene.index.Term;
+import org.apache.lucene.index.TermContext;
+import org.apache.lucene.search.spans.SpanQuery;
+import org.apache.lucene.search.spans.Spans;
+import org.apache.lucene.util.Bits;
+
+import de.ids_mannheim.korap.query.spans.ClassFilteredSpans;
+
+public class SpanClassFilterQuery extends SimpleSpanQuery {
+
+ public enum ClassOperation {
+ DISJOINT, INTERSECT, INCLUDE, EQUAL, DIFFER
+ }
+
+ private ClassOperation operation;
+ private byte classNum1, classNum2;
+
+ public SpanClassFilterQuery (SpanQuery sq, ClassOperation type,
+ int classNum1, int classNum2,
+ boolean collectPayloads) {
+ super(sq, collectPayloads);
+ this.operation = type;
+ this.classNum1 = (byte) classNum1;
+ this.classNum2 = (byte) classNum2;
+ }
+
+ @Override
+ public SimpleSpanQuery clone() {
+ return new SpanClassFilterQuery((SpanQuery) firstClause.clone(),
+ operation,
+ classNum1, classNum2, collectPayloads);
+ }
+
+ @Override
+ public Spans getSpans(AtomicReaderContext context, Bits acceptDocs,
+ Map<Term, TermContext> termContexts) throws IOException {
+ return new ClassFilteredSpans(this, context, acceptDocs,
+ termContexts);
+ }
+
+ @Override
+ public String toString(String field) {
+ StringBuilder sb = new StringBuilder();
+ sb.append("spanClassFilter(");
+ sb.append(firstClause.toString());
+ sb.append(",");
+ sb.append(operation);
+ sb.append(",");
+ sb.append(classNum1);
+ sb.append(",");
+ sb.append(classNum2);
+ sb.append(")");
+ return sb.toString();
+ }
+
+ public ClassOperation getOperation() {
+ return operation;
+ }
+
+ public void setOperation(ClassOperation operation) {
+ this.operation = operation;
+ }
+
+ public byte getClassNum1() {
+ return classNum1;
+ }
+
+ public void setClassNum1(byte classNum1) {
+ this.classNum1 = classNum1;
+ }
+
+ public byte getClassNum2() {
+ return classNum2;
+ }
+
+ public void setClassNum2(byte classNum2) {
+ this.classNum2 = classNum2;
+ }
+
+}
diff --git a/src/main/java/de/ids_mannheim/korap/query/SpanFocusQuery.java b/src/main/java/de/ids_mannheim/korap/query/SpanFocusQuery.java
index 305e42c..c144a58 100644
--- a/src/main/java/de/ids_mannheim/korap/query/SpanFocusQuery.java
+++ b/src/main/java/de/ids_mannheim/korap/query/SpanFocusQuery.java
@@ -21,16 +21,13 @@
* Modify the span of a match to the boundaries of a certain class.
*
* In case multiple classes are found with the very same number, the
- * span is
- * maximized to start on the first occurrence from the left and end on
- * the last
- * occurrence on the right.
+ * span is maximized to start on the first occurrence from the left
+ * and end on the last occurrence on the right.
*
* In case the class to modify on is not found in the subquery, the
- * match is
- * ignored.
+ * match is ignored.
*
- * @author diewald
+ * @author diewald, margaretha
*
* @see FocusSpans
*/
@@ -38,7 +35,8 @@
private List<Byte> classNumbers = new ArrayList<Byte>();
private boolean isSorted = true;
-
+ private boolean matchTemporaryClass = false;
+ private boolean removeTemporaryClasses = false;
/**
* Construct a new SpanFocusQuery.
@@ -59,7 +57,6 @@
public SpanFocusQuery (SpanQuery sq, List<Byte> classNumbers) {
super(sq, true);
this.classNumbers = classNumbers;
- isSorted = false;
};
@@ -83,6 +80,9 @@
public String toString (String field) {
StringBuffer buffer = new StringBuffer();
buffer.append("focus(");
+ if (matchTemporaryClass){
+ buffer.append("#");
+ }
if (classNumbers.size() > 1) {
buffer.append("[");
for (int i = 0; i < classNumbers.size(); i++) {
@@ -133,6 +133,9 @@
SpanFocusQuery spanFocusQuery = new SpanFocusQuery(
(SpanQuery) this.firstClause.clone(), this.getClassNumbers());
spanFocusQuery.setBoost(getBoost());
+ spanFocusQuery.setMatchTemporaryClass(this.matchTemporaryClass);
+ spanFocusQuery.setSorted(this.isSorted);
+ spanFocusQuery.setRemoveTemporaryClasses(this.removeTemporaryClasses);
return spanFocusQuery;
};
@@ -185,4 +188,20 @@
this.isSorted = isSorted;
}
+ public boolean matchTemporaryClass() {
+ return matchTemporaryClass;
+ }
+
+ public void setMatchTemporaryClass(boolean matchTemporaryClass) {
+ this.matchTemporaryClass = matchTemporaryClass;
+ }
+
+ public boolean removeTemporaryClasses() {
+ return removeTemporaryClasses;
+ }
+
+ public void setRemoveTemporaryClasses(boolean rem) {
+ this.removeTemporaryClasses = rem;
+ }
+
};
diff --git a/src/main/java/de/ids_mannheim/korap/query/SpanReferenceQuery.java b/src/main/java/de/ids_mannheim/korap/query/SpanReferenceQuery.java
new file mode 100644
index 0000000..9c2484d
--- /dev/null
+++ b/src/main/java/de/ids_mannheim/korap/query/SpanReferenceQuery.java
@@ -0,0 +1,61 @@
+package de.ids_mannheim.korap.query;
+
+import java.io.IOException;
+import java.util.Map;
+
+import org.apache.lucene.index.AtomicReaderContext;
+import org.apache.lucene.index.Term;
+import org.apache.lucene.index.TermContext;
+import org.apache.lucene.search.spans.SpanQuery;
+import org.apache.lucene.search.spans.Spans;
+import org.apache.lucene.util.Bits;
+
+import de.ids_mannheim.korap.query.spans.ReferenceSpans;
+
+public class SpanReferenceQuery extends SimpleSpanQuery {
+
+ private byte classNum;
+
+ public SpanReferenceQuery (SpanQuery firstClause, byte classNum,
+ boolean collectPayloads) {
+ super(firstClause, collectPayloads);
+ this.classNum = classNum;
+ }
+
+
+ @Override
+ public SimpleSpanQuery clone () {
+ // TODO Auto-generated method stub
+ return null;
+ }
+
+
+ @Override
+ public Spans getSpans (AtomicReaderContext context, Bits acceptDocs,
+ Map<Term, TermContext> termContexts) throws IOException {
+ // TODO Auto-generated method stub
+ return new ReferenceSpans(this, context, acceptDocs, termContexts);
+ }
+
+
+ @Override
+ public String toString (String field) {
+ StringBuilder sb = new StringBuilder();
+ sb.append("spanReference(");
+ sb.append(firstClause.toString());
+ sb.append(", ");
+ sb.append(classNum);
+ sb.append(")");
+ return sb.toString();
+ }
+
+
+ public byte getClassNum() {
+ return classNum;
+ }
+
+ public void setClassNum(byte classNum) {
+ this.classNum = classNum;
+ }
+
+}
diff --git a/src/main/java/de/ids_mannheim/korap/query/SpanRelationMatchQuery.java b/src/main/java/de/ids_mannheim/korap/query/SpanRelationMatchQuery.java
new file mode 100644
index 0000000..5bc962a
--- /dev/null
+++ b/src/main/java/de/ids_mannheim/korap/query/SpanRelationMatchQuery.java
@@ -0,0 +1,125 @@
+package de.ids_mannheim.korap.query;
+
+import java.io.IOException;
+import java.util.Map;
+
+import org.apache.lucene.index.AtomicReaderContext;
+import org.apache.lucene.index.Term;
+import org.apache.lucene.index.TermContext;
+import org.apache.lucene.search.spans.SpanQuery;
+import org.apache.lucene.search.spans.Spans;
+import org.apache.lucene.util.Bits;
+
+import de.ids_mannheim.korap.query.spans.FocusSpans;
+
+public class SpanRelationMatchQuery extends SimpleSpanQuery{
+
+ private SpanQuery operandQuery;
+ private SpanQuery operand2Query;
+ private SpanRelationQuery relationQuery;
+
+ public SpanRelationMatchQuery (SpanRelationQuery relation, SpanQuery operand,
+ boolean collectPayloads) {
+
+ checkVariables(relation, operand);
+ SpanFocusQuery sq = new SpanFocusQuery(new SpanSegmentQuery(
+ relationQuery, operandQuery, true),
+ relation.getTempClassNumbers());
+ sq.setMatchTemporaryClass(true);
+ sq.setRemoveTemporaryClasses(true);
+ sq.setSorted(false); // which operand to focus might be
+ // different from that to match
+
+ this.setFirstClause(sq);
+ this.collectPayloads = collectPayloads;
+ }
+
+ public SpanRelationMatchQuery (SpanRelationQuery relation, SpanQuery source,
+ SpanQuery target, boolean collectPayloads) {
+
+ checkVariables(relation, source, target);
+ SpanFocusQuery sq = null;
+ // match source
+ if (relationQuery.getDirection() == 0) {
+ sq = new SpanFocusQuery(new SpanSegmentQuery(relationQuery,
+ operandQuery, true), relation.getTempTargetNum());
+ }
+ // match target
+ else {
+ sq = new SpanFocusQuery(new SpanSegmentQuery(relationQuery,
+ operandQuery, true), relation.getTempSourceNum());
+ }
+ sq.setSorted(false);
+ sq.setMatchTemporaryClass(true);
+
+ SpanFocusQuery sq2 = new SpanFocusQuery(new SpanSegmentQuery(sq,
+ operand2Query, true), relation.getTempClassNumbers());
+ sq2.setMatchTemporaryClass(true);
+ sq2.setRemoveTemporaryClasses(true);
+ sq2.setSorted(false);
+
+ this.setFirstClause(sq2);
+ this.collectPayloads = collectPayloads;
+
+ }
+
+ public void checkVariables(SpanRelationQuery relation, SpanQuery operand) {
+ if (relation == null) {
+ throw new IllegalArgumentException(
+ "The relation query cannot be null.");
+ }
+ if (operand == null) {
+ throw new IllegalArgumentException(
+ "The operand query cannot be null.");
+ }
+ this.field = relation.getField();
+ if (!operand.getField().equals(field)) {
+ throw new IllegalArgumentException(
+ "Clauses must have the same field.");
+ }
+ this.relationQuery = relation;
+ this.operandQuery = operand;
+ }
+
+ public void checkVariables(SpanRelationQuery relation, SpanQuery operand, SpanQuery target) {
+ checkVariables(relation, operand);
+ if (target == null) {
+ if (operand == null) {
+ throw new IllegalArgumentException(
+ "The target query cannot be null.");
+ }
+ }
+ if (!target.getField().equals(field)) {
+ throw new IllegalArgumentException(
+ "Clauses must have the same field.");
+ }
+ this.operand2Query = target;
+ }
+
+ @Override
+ public SimpleSpanQuery clone() {
+ if (operand2Query != null) {
+ return new SpanRelationMatchQuery(
+ (SpanRelationQuery) relationQuery.clone(),
+ (SpanQuery) operandQuery.clone(),
+ (SpanQuery) operand2Query.clone(), collectPayloads);
+ }
+
+ return new SpanRelationMatchQuery(
+ (SpanRelationQuery) relationQuery.clone(),
+ (SpanQuery) operandQuery.clone(), collectPayloads);
+ }
+
+ @Override
+ public Spans getSpans(AtomicReaderContext context, Bits acceptDocs,
+ Map<Term, TermContext> termContexts) throws IOException {
+
+ return new FocusSpans((SpanFocusQuery) firstClause, context,
+ acceptDocs, termContexts);
+ }
+
+ @Override
+ public String toString(String field) {
+ return getFirstClause().toString();
+ }
+}
diff --git a/src/main/java/de/ids_mannheim/korap/query/SpanRelationQuery.java b/src/main/java/de/ids_mannheim/korap/query/SpanRelationQuery.java
index 8df29b3..8d06a7e 100644
--- a/src/main/java/de/ids_mannheim/korap/query/SpanRelationQuery.java
+++ b/src/main/java/de/ids_mannheim/korap/query/SpanRelationQuery.java
@@ -1,6 +1,8 @@
package de.ids_mannheim.korap.query;
import java.io.IOException;
+import java.util.Arrays;
+import java.util.List;
import java.util.Map;
import org.apache.lucene.index.AtomicReaderContext;
@@ -55,7 +57,12 @@
public class SpanRelationQuery extends SimpleSpanQuery {
private int direction = 0;
+ private byte tempSourceNum = 1;
+ private byte tempTargetNum = 2;
+ private byte sourceClass;
+ private byte targetClass;
+ private List<Byte> tempClassNumbers = Arrays.asList(tempSourceNum, tempTargetNum);
/**
* Constructs a SpanRelationQuery based on the given span query.
*
@@ -76,6 +83,13 @@
}
}
+ public SpanRelationQuery (SpanQuery firstClause, List<Byte> classNumbers,
+ boolean collectPayloads) {
+ this(firstClause, collectPayloads);
+ this.tempClassNumbers = classNumbers;
+ this.tempSourceNum = classNumbers.get(0);
+ this.tempTargetNum = classNumbers.get(1);
+ }
@Override
public SimpleSpanQuery clone () {
@@ -95,9 +109,25 @@
@Override
public String toString (String field) {
StringBuilder sb = new StringBuilder();
+ if (sourceClass > 0) {
+ sb.append("{");
+ sb.append(sourceClass);
+ sb.append(": source:");
+ }
+ if (targetClass > 0) {
+ sb.append("{");
+ sb.append(targetClass);
+ sb.append(": target:");
+ }
sb.append("spanRelation(");
sb.append(firstClause.toString(field));
sb.append(")");
+ if (sourceClass > 0) {
+ sb.append("}");
+ }
+ if (targetClass > 0) {
+ sb.append("}");
+ }
sb.append(ToStringUtils.boost(getBoost()));
return sb.toString();
}
@@ -110,4 +140,54 @@
this.direction = direction;
}
+ public List<Byte> getTempClassNumbers() {
+ return tempClassNumbers;
+ }
+
+ public void setTempClassNumbers(List<Byte> classNumbers) {
+ this.tempClassNumbers = classNumbers;
+ }
+
+ public byte getTempSourceNum() {
+ return tempSourceNum;
+ }
+
+ public void setTempSourceNum(byte sourceNum) {
+ this.tempSourceNum = sourceNum;
+ }
+
+ public byte getTempTargetNum() {
+ return tempTargetNum;
+ }
+
+ public void setTempTargetNum(byte targetNum) {
+ this.tempTargetNum = targetNum;
+ }
+
+ public byte getSourceClass() {
+ return sourceClass;
+ }
+
+ public void setSourceClass(byte sourceClass)
+ throws IllegalArgumentException {
+ if (sourceClass < 1) {
+ throw new IllegalArgumentException(
+ "Class number must be bigger than 0.");
+ }
+
+ this.sourceClass = sourceClass;
+ }
+
+ public byte getTargetClass() {
+ return targetClass;
+ }
+
+ public void setTargetClass(byte targetClass)
+ throws IllegalArgumentException {
+ if (targetClass < 1) {
+ throw new IllegalArgumentException(
+ "Class number must be bigger than 0.");
+ }
+ this.targetClass = targetClass;
+ }
}
diff --git a/src/main/java/de/ids_mannheim/korap/query/SpanSegmentQuery.java b/src/main/java/de/ids_mannheim/korap/query/SpanSegmentQuery.java
index bb69d4c..7e75543 100644
--- a/src/main/java/de/ids_mannheim/korap/query/SpanSegmentQuery.java
+++ b/src/main/java/de/ids_mannheim/korap/query/SpanSegmentQuery.java
@@ -46,7 +46,6 @@
this(firstClause, secondClause, true);
}
-
/**
* Constructs a SpanSegmentQuery from the two given SpanQueries.
*
@@ -61,16 +60,11 @@
* <code>false</code>.
*/
public SpanSegmentQuery (SpanQuery firstClause, SpanQuery secondClause,
- boolean collectPayloads) {
+ boolean collectPayloads) {
super(firstClause, secondClause, collectPayloads);
- }
-
-
- public SpanSegmentQuery (SpanRelationQuery firstClause,
- SimpleSpanQuery secondClause,
- boolean collectPayloads) {
- super(firstClause, secondClause, true);
- isRelation = true;
+ if (firstClause instanceof SpanRelationQuery) {
+ isRelation = true;
+ }
}
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/ClassFilteredSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/ClassFilteredSpans.java
new file mode 100644
index 0000000..bfd5f11
--- /dev/null
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/ClassFilteredSpans.java
@@ -0,0 +1,132 @@
+package de.ids_mannheim.korap.query.spans;
+
+import static de.ids_mannheim.korap.util.KrillByte.byte2int;
+
+import java.io.IOException;
+import java.util.BitSet;
+import java.util.Map;
+
+import org.apache.lucene.index.AtomicReaderContext;
+import org.apache.lucene.index.Term;
+import org.apache.lucene.index.TermContext;
+import org.apache.lucene.util.Bits;
+
+import de.ids_mannheim.korap.query.SpanClassFilterQuery;
+import de.ids_mannheim.korap.query.SpanClassFilterQuery.ClassOperation;
+
+public class ClassFilteredSpans extends SimpleSpans {
+
+ private BitSet bitset1, bitset2;
+ private ClassOperation operation;
+ private byte classNum1, classNum2;
+
+ public ClassFilteredSpans (SpanClassFilterQuery query,
+ AtomicReaderContext context, Bits acceptDocs,
+ Map<Term, TermContext> termContexts) throws IOException {
+ super(query, context, acceptDocs, termContexts);
+ this.operation = query.getOperation();
+ this.classNum1 = query.getClassNum1();
+ this.classNum2 = query.getClassNum2();
+ hasMoreSpans = firstSpans.next();
+ }
+
+ @Override
+ public boolean next() throws IOException {
+ while (hasMoreSpans) {
+ matchPayload.clear();
+ bitset1 = null;
+ bitset2 = null;
+ if (isClassOperationValid()) {
+ this.matchStartPosition = firstSpans.start();
+ this.matchEndPosition = firstSpans.end();
+ this.matchDocNumber = firstSpans.doc();
+ this.matchPayload.addAll(firstSpans.getPayload());
+ hasMoreSpans = firstSpans.next();
+ return true;
+ }
+ hasMoreSpans = firstSpans.next();
+ }
+ return false;
+ }
+
+ private boolean isClassOperationValid() throws IOException {
+ setBitsets();
+ int cardinality = Math.max(bitset1.cardinality(), bitset2.cardinality());
+ bitset1.and(bitset2);
+ // System.out.println("cardinality:" + cardinality);
+ switch (operation) {
+ case DISJOINT:
+ if (bitset1.cardinality() == 0) return true;
+ break;
+ case EQUAL:
+ if (cardinality == bitset1.cardinality()) return true;
+ break;
+ case DIFFER:
+ if (cardinality == 0 || cardinality != bitset1.cardinality())
+ return true;
+ break;
+ case INCLUDE:
+ if (bitset1.cardinality() == bitset2.cardinality()) {
+ return true;
+ }
+ break;
+ case INTERSECT:
+ if (bitset1.cardinality() > 0) return true;
+ break;
+ }
+
+ return false;
+ }
+
+ private void setBitsets() throws IOException {
+ BitSet bs = new BitSet();
+ int start, end;
+ // System.out.println("------------------------");
+ for (byte[] payload : firstSpans.getPayload()) {
+ if (payload.length == 9){
+ start = byte2int(payload, 0) + 1;
+ end = byte2int(payload, 4) + 1;
+ if (payload[8] == classNum1) {
+ // System.out.println("bitset1 " + start + " " +
+ // end);
+ if (bitset1 == null) {
+ bitset1 = new BitSet();
+ bitset1.set(start, end);
+ }
+ else {
+ bs.set(start, end);
+ bitset1.or(bs);
+ }
+ // System.out.println(bitset1);
+ }
+ else if (payload[8] == classNum2) {
+ // System.out.println("#bitset2 " + start + " " +
+ // end);
+ if (bitset2 == null) {
+ bitset2 = new BitSet();
+ bitset2.set(start, end);
+ }
+ else {
+ bs.set(start, end);
+ bitset2.or(bs);
+ // System.out.println("OR #2");
+ }
+ // System.out.println(bitset2);
+ }
+ }
+ }
+
+ }
+
+ @Override
+ public boolean skipTo(int target) throws IOException {
+ // TODO Auto-generated method stub
+ return false;
+ }
+
+ @Override
+ public long cost() {
+ // TODO Auto-generated method stub
+ return 0;
+ }
+}
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/ClassSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/ClassSpans.java
index 4e97dbc..ac83c4d 100644
--- a/src/main/java/de/ids_mannheim/korap/query/spans/ClassSpans.java
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/ClassSpans.java
@@ -26,16 +26,16 @@
* @author diewald
*/
-public class ClassSpans extends Spans {
- private List<byte[]> classedPayload;
- private final Spans spans;
- private byte number;
- private SpanQuery operand;
- private Boolean hasmorespans = false;
+public class ClassSpans extends SimpleSpans {
+ protected List<byte[]> classedPayload;
+ protected Spans spans;
+ protected byte number;
+ protected SpanQuery operand;
+ protected Boolean hasmorespans = false;
- private ByteBuffer bb = ByteBuffer.allocate(9);
+ protected ByteBuffer bb;
- private final static Logger log = LoggerFactory.getLogger(ClassSpans.class);
+ private final Logger log = LoggerFactory.getLogger(ClassSpans.class);
// This advices the java compiler to ignore all loggings
public static final boolean DEBUG = false;
@@ -69,6 +69,8 @@
// The highlighted payload
this.classedPayload = new ArrayList<byte[]>(3);
+
+ this.bb = ByteBuffer.allocate(9);
};
@@ -126,7 +128,7 @@
};
- private boolean addClassPayload () throws IOException {
+ protected boolean addClassPayload() throws IOException {
hasmorespans = true;
classedPayload.clear();
@@ -149,6 +151,12 @@
// Add highlight information as byte array
classedPayload.add(bb.array());
+
+ if (spans instanceof SimpleSpans) {
+ SimpleSpans ss = (SimpleSpans) spans;
+ this.hasSpanId = ss.hasSpanId;
+ this.spanId = ss.spanId;
+ }
return true;
};
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/ElementDistanceSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/ElementDistanceSpans.java
index 5060b31..ac8fe24 100644
--- a/src/main/java/de/ids_mannheim/korap/query/spans/ElementDistanceSpans.java
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/ElementDistanceSpans.java
@@ -14,16 +14,12 @@
/**
* Span enumeration of element-based distance span matches. Each match
- * consists
- * of two child spans. The element-distance between the child spans is
- * the
- * difference between the element position numbers where the child
- * spans are.
- * The element-distance unit can be a sentence or a paragraph. All
- * other child
- * spans' occurrences which are not in a sentence or a paragraph (with
- * respect
- * to the element distance type currently used), are ignored.
+ * consists of two child spans. The element-distance between the child
+ * spans is the difference between the element position numbers where
+ * the child spans are. The element-distance unit can be a sentence or
+ * a paragraph. All other child spans' occurrences which are not in a
+ * sentence or a paragraph (with respect to the element distance type
+ * currently used), are ignored.
*
* Note: elements cannot overlap with each other.
*
@@ -105,8 +101,7 @@
/**
* Add new possible (candidate) firstspans. Candidate firstspans
- * must be in
- * an element and not too far from the secondspan.
+ * must be in an element and not too far from the secondspan.
*
* @throws IOException
*/
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/FocusSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/FocusSpans.java
index 2da1558..9598ac0 100644
--- a/src/main/java/de/ids_mannheim/korap/query/spans/FocusSpans.java
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/FocusSpans.java
@@ -52,12 +52,10 @@
// This advices the java compiler to ignore all loggings
public static final boolean DEBUG = false;
- // private SimpleSpans originalSpans;
- private boolean isSorted;
+ private boolean isSorted, matchTemporaryClass, removeTemporaryClasses;
private List<CandidateSpan> candidateSpans;
private int windowSize = 10;
private int currentDoc;
- private byte number;
/**
@@ -86,13 +84,16 @@
}
classNumbers = query.getClassNumbers();
isSorted = query.isSorted();
+ matchTemporaryClass = query.matchTemporaryClass();
+ removeTemporaryClasses = query.removeTemporaryClasses();
candidateSpans = new ArrayList<CandidateSpan>();
hasMoreSpans = firstSpans.next();
currentDoc = firstSpans.doc();
- // matchPayload = new ArrayList<byte[]>(6);
this.query = query;
- hasSpanId = true;
+ if (getSpanId() > 0) {
+ hasSpanId = true;
+ }
}
@@ -164,7 +165,8 @@
for (byte[] payload : firstSpans.getPayload()) {
// No class payload - ignore
// this may be problematic for other calculated payloads!
- if (payload.length == 9) {
+ if ((!matchTemporaryClass && payload.length == 9)
+ || (matchTemporaryClass && payload.length == 10)) {
if (classNumbers.contains(payload[8])) {
isClassFound = true;
classStart = byte2int(payload, 0);
@@ -178,9 +180,13 @@
maxPos = classEnd;
}
}
- candidateSpan.getPayloads().add(payload.clone());
}
+ if (payload.length == 8
+ || (removeTemporaryClasses && payload.length == 10)) {
+ continue;
+ }
+ candidateSpan.getPayloads().add(payload.clone());
}
if (isClassFound) {
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/ReferenceSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/ReferenceSpans.java
new file mode 100644
index 0000000..a860afd
--- /dev/null
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/ReferenceSpans.java
@@ -0,0 +1,93 @@
+package de.ids_mannheim.korap.query.spans;
+
+import static de.ids_mannheim.korap.util.KrillByte.byte2int;
+
+import java.io.IOException;
+import java.util.Map;
+
+import org.apache.lucene.index.AtomicReaderContext;
+import org.apache.lucene.index.Term;
+import org.apache.lucene.index.TermContext;
+import org.apache.lucene.util.Bits;
+
+import de.ids_mannheim.korap.query.SpanReferenceQuery;
+
+public class ReferenceSpans extends SimpleSpans {
+
+ private byte classNum;
+
+ public ReferenceSpans (SpanReferenceQuery query,
+ AtomicReaderContext context, Bits acceptDocs,
+ Map<Term, TermContext> termContexts)
+ throws IOException {
+ super(query, context, acceptDocs, termContexts);
+ this.classNum = query.getClassNum();
+ hasMoreSpans = firstSpans.next();
+ }
+
+
+ @Override
+ public boolean next () throws IOException {
+ while (hasMoreSpans) {
+ if (hasSameClassPosition()) {
+ matchStartPosition = firstSpans.start();
+ matchEndPosition = firstSpans.end();
+ matchDocNumber = firstSpans.doc();
+ hasMoreSpans = firstSpans.next();
+ return true;
+ }
+ hasMoreSpans = firstSpans.next();
+ }
+ return false;
+ }
+
+
+ private boolean hasSameClassPosition () throws IOException {
+ int start=0, end=0;
+ boolean isFound = false;
+ boolean match = false;
+
+ matchPayload.clear();
+
+ for (byte[] payload : firstSpans.getPayload()) {
+ if (payload.length == 9 && payload[8] == classNum) {
+ if (isFound) {
+ if (start == byte2int(payload, 0) && end == byte2int(payload, 4)){
+ match = true;
+ continue;
+ }
+ match = false;
+ break;
+ }
+
+ start = byte2int(payload, 0);
+ end = byte2int(payload, 4);
+ isFound = true;
+ matchPayload.add(payload);
+ }
+ else {
+ matchPayload.add(payload);
+ }
+ }
+ return match;
+ }
+
+
+ @Override
+ public boolean skipTo (int target) throws IOException {
+ if (hasMoreSpans && (firstSpans.doc() < target)) {
+ if (!firstSpans.skipTo(target)) {
+ hasMoreSpans = false;
+ return false;
+ }
+ }
+ return next();
+ }
+
+
+ @Override
+ public long cost() {
+ return firstSpans.cost();
+ }
+
+}
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/RelationSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/RelationSpans.java
index f17043d..66afdbe 100644
--- a/src/main/java/de/ids_mannheim/korap/query/spans/RelationSpans.java
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/RelationSpans.java
@@ -58,7 +58,8 @@
protected Logger logger = LoggerFactory.getLogger(RelationSpans.class);
private List<CandidateRelationSpan> candidateList;
-
+ private byte tempSourceNum, tempTargetNum;
+ private byte sourceClass, targetClass;
/**
* Constructs RelationSpans from the given
@@ -77,6 +78,11 @@
throws IOException {
super(relationSpanQuery, context, acceptDocs, termContexts);
direction = relationSpanQuery.getDirection();
+ tempSourceNum = relationSpanQuery.getTempSourceNum();
+ tempTargetNum = relationSpanQuery.getTempTargetNum();
+ sourceClass = relationSpanQuery.getSourceClass();
+ targetClass = relationSpanQuery.getTargetClass();
+
candidateList = new ArrayList<>();
relationTermSpan = (TermSpans) firstSpans;
hasMoreSpans = relationTermSpan.next();
@@ -213,22 +219,47 @@
}
if (direction == 0) {
payload.add(createClassPayload(cs.getLeftStart(), cs.getLeftEnd(),
- (byte) 1));
+ tempSourceNum, false));
payload.add(createClassPayload(cs.getRightStart(),
- cs.getRightEnd(), (byte) 2));
+ cs.getRightEnd(), tempTargetNum, false));
+
+ if (sourceClass > 0) {
+ payload.add(createClassPayload(cs.getLeftStart(),
+ cs.getLeftEnd(), sourceClass, true));
+ }
+ if (targetClass > 0) {
+ payload.add(createClassPayload(cs.getRightStart(),
+ cs.getRightEnd(), targetClass, true));
+ }
+
}
else {
payload.add(createClassPayload(cs.getRightStart(),
- cs.getRightEnd(), (byte) 1));
+ cs.getRightEnd(), tempSourceNum, false));
payload.add(createClassPayload(cs.getLeftStart(), cs.getLeftEnd(),
- (byte) 2));
+ tempTargetNum, false));
+
+ if (sourceClass > 0) {
+ payload.add(createClassPayload(cs.getRightStart(),
+ cs.getRightEnd(), sourceClass, true));
+ }
+ if (targetClass > 0) {
+ payload.add(createClassPayload(cs.getLeftStart(),
+ cs.getLeftEnd(), targetClass, true));
+ }
}
cs.setPayloads(payload);
}
-
- private byte[] createClassPayload (int start, int end, byte classNumber) {
- ByteBuffer buffer = ByteBuffer.allocate(9);
+ private byte[] createClassPayload(int start, int end, byte classNumber,
+ boolean keep) {
+ ByteBuffer buffer = null;
+ if (keep) {
+ buffer = ByteBuffer.allocate(9);
+ }
+ else {
+ buffer = ByteBuffer.allocate(10);
+ }
buffer.putInt(start);
buffer.putInt(end);
buffer.put(classNumber);
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/SegmentSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/SegmentSpans.java
index 8ac53b8..9d40d31 100644
--- a/src/main/java/de/ids_mannheim/korap/query/spans/SegmentSpans.java
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/SegmentSpans.java
@@ -94,6 +94,7 @@
protected int findMatch () {
RelationSpans s1;
SimpleSpans s2;
+
if (firstSpans.start() == secondSpans.start()
&& firstSpans.end() == secondSpans.end()) {
@@ -101,13 +102,18 @@
s1 = (RelationSpans) firstSpans;
s2 = (SimpleSpans) secondSpans;
- //System.out.println("segment: " + s1.getRightStart() + " "
- // + s1.getRightEnd());
- if (s1.getLeftId() == s2.getSpanId()) {
+ if (s2.hasSpanId) {
+ if (s1.getLeftId() == s2.getSpanId()) {
+ setSpanId(s2.getSpanId());
+ setMatch();
+ return 0;
+ }
+ }
+ else {
setMatch();
- setSpanId(s2.getSpanId());
return 0;
}
+
}
else {
setMatch();
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedDistanceSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedDistanceSpans.java
index 2b93ece..60e2a8a 100644
--- a/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedDistanceSpans.java
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedDistanceSpans.java
@@ -100,12 +100,10 @@
/**
* Sets the list of matches for the span having the smallest
- * position (i.e.
- * between the first and the second spans), and its candidates
- * (i.e. its
- * counterparts). The candidates also must have smaller positions.
- * Simply
- * remove the span if it does not have any candidates.
+ * position (i.e. between the first and the second spans), and its
+ * candidates (i.e. its counterparts). The candidates also must
+ * have smaller positions. Simply remove the span if it does not
+ * have any candidates.
*
* @throws IOException
*/
@@ -117,13 +115,13 @@
hasMoreSecondSpans, firstSpanList);
// System.out.println("--------------------");
// System.out.println("firstSpanList:");
- // for (CandidateSpan cs: firstSpanList) {
- // System.out.println(cs.getStart() +" "+ cs.getEnd());
+ // for (CandidateSpan cs : firstSpanList) {
+ // System.out.println(cs.getStart() + " " + cs.getEnd());
// }
//
// System.out.println("secondSpanList:");
- // for (CandidateSpan cs: secondSpanList) {
- // System.out.println(cs.getStart() +" "+ cs.getEnd());
+ // for (CandidateSpan cs : secondSpanList) {
+ // System.out.println(cs.getStart() + " " + cs.getEnd());
// }
CandidateSpan currentFirstSpan, currentSecondSpan;
@@ -132,7 +130,7 @@
currentFirstSpan = firstSpanList.get(0);
currentSecondSpan = secondSpanList.get(0);
- if (currentFirstSpan.getEnd() < currentSecondSpan.getEnd()
+ if (currentFirstSpan.getStart() < currentSecondSpan.getStart()
|| isLastCandidateSmaller(currentFirstSpan,
currentSecondSpan)) {
// log.trace("current target: "
@@ -180,20 +178,16 @@
/**
* Tells if the last candidate from the secondSpanList has a
- * smaller end
- * position than the end position of the the last candidate from
- * the
- * firstSpanList.
+ * smaller end position than the end position of the the last
+ * candidate from the firstSpanList.
*
* @param currentFirstSpan
* the current firstspan
* @param currentSecondSpan
* the current secondspan
* @return <code>true</code> if the end position of the last
- * candidate from
- * the secondSpanList is smaller than that from the
- * firstSpanList,
- * <code>false</code> otherwise.
+ * candidate from the secondSpanList is smaller than that
+ * from the firstSpanList, <code>false</code> otherwise.
*/
private boolean isLastCandidateSmaller (CandidateSpan currentFirstSpan,
CandidateSpan currentSecondSpan) {
@@ -285,7 +279,6 @@
payloads.addAll(cs.getPayloads());
}
}
-
CandidateSpan match = new CandidateSpan(start, end, doc, cost, payloads);
match.setChildSpan(cs);
return match;
diff --git a/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedTokenDistanceSpans.java b/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedTokenDistanceSpans.java
index 3e50e76..0b8b055 100644
--- a/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedTokenDistanceSpans.java
+++ b/src/main/java/de/ids_mannheim/korap/query/spans/UnorderedTokenDistanceSpans.java
@@ -137,14 +137,16 @@
}
// left candidate
- if (cs.getEnd() < target.getStart())
+ if (cs.getEnd() < target.getStart()) {
actualDistance = target.getStart() - cs.getEnd() + 1;
- else
+ }
+ else {
// right candidate
actualDistance = cs.getStart() - target.getEnd() + 1;
-
- if (minDistance <= actualDistance && actualDistance <= maxDistance)
+ }
+ if (minDistance <= actualDistance && actualDistance <= maxDistance) {
matches.add(createMatchCandidate(target, cs, false));
+ }
}
return matches;
}
diff --git a/src/main/java/de/ids_mannheim/korap/query/wrap/SpanReferenceQueryWrapper.java b/src/main/java/de/ids_mannheim/korap/query/wrap/SpanReferenceQueryWrapper.java
new file mode 100644
index 0000000..02a734f
--- /dev/null
+++ b/src/main/java/de/ids_mannheim/korap/query/wrap/SpanReferenceQueryWrapper.java
@@ -0,0 +1,47 @@
+package de.ids_mannheim.korap.query.wrap;
+
+import org.apache.lucene.search.spans.SpanQuery;
+
+import de.ids_mannheim.korap.query.SpanReferenceQuery;
+import de.ids_mannheim.korap.util.QueryException;
+
+public class SpanReferenceQueryWrapper extends SpanQueryWrapper {
+
+ private SpanQueryWrapper subQuery;
+ private byte classNum;
+
+ public SpanReferenceQueryWrapper (SpanQueryWrapper subQueryWrapper,
+ byte classNum) {
+ this.subQuery = subQueryWrapper;
+ if (subQuery != null) {
+ this.isNull = false;
+ }
+ else
+ return;
+
+ if (subQuery.isEmpty) {
+ this.isEmpty = true;
+ return;
+ }
+
+ if (classNum < 0) {
+ throw new IllegalArgumentException(
+ "Class number must be bigger than 0.");
+ }
+ this.classNum = classNum;
+ }
+
+ @Override
+ public SpanQuery toQuery() throws QueryException {
+
+ if (this.isNull() || this.isEmpty()) {
+ return null;
+ }
+
+ SpanQuery sq = subQuery.retrieveNode(this.retrieveNode).toQuery();
+ if (sq == null) return null;
+
+ return new SpanReferenceQuery(sq, classNum, true);
+ }
+
+}
diff --git a/src/main/java/de/ids_mannheim/korap/query/wrap/SpanRelationWrapper.java b/src/main/java/de/ids_mannheim/korap/query/wrap/SpanRelationWrapper.java
index 17afa22..2535ce9 100644
--- a/src/main/java/de/ids_mannheim/korap/query/wrap/SpanRelationWrapper.java
+++ b/src/main/java/de/ids_mannheim/korap/query/wrap/SpanRelationWrapper.java
@@ -1,11 +1,11 @@
package de.ids_mannheim.korap.query.wrap;
-import java.util.ArrayList;
-
import org.apache.lucene.search.spans.SpanQuery;
+import org.apache.lucene.search.spans.SpanTermQuery;
import de.ids_mannheim.korap.query.SpanFocusQuery;
-import de.ids_mannheim.korap.query.SpanSegmentQuery;
+import de.ids_mannheim.korap.query.SpanRelationMatchQuery;
+import de.ids_mannheim.korap.query.SpanRelationQuery;
import de.ids_mannheim.korap.util.QueryException;
public class SpanRelationWrapper extends SpanQueryWrapper {
@@ -13,12 +13,9 @@
private SpanQueryWrapper relationQuery;
private SpanQueryWrapper subQuery1;
private SpanQueryWrapper subQuery2;
- private byte[] classNumbers;
-
public SpanRelationWrapper (SpanQueryWrapper relationWrapper,
- SpanQueryWrapper operand1,
- SpanQueryWrapper operand2, byte[] classNumbers) {
+ SpanQueryWrapper operand1, SpanQueryWrapper operand2) {
this.relationQuery = relationWrapper;
if (relationQuery != null) {
@@ -34,70 +31,63 @@
this.subQuery1 = operand1;
this.subQuery2 = operand2;
- this.classNumbers = classNumbers;
}
@Override
- public SpanQuery toQuery () throws QueryException {
+ public SpanQuery toQuery() throws QueryException {
if (this.isNull() || this.isEmpty()) {
return null;
}
- SpanQuery sq = relationQuery.retrieveNode(this.retrieveNode).toQuery();
- if (sq == null)
+ SpanTermQuery relationTermQuery = (SpanTermQuery) relationQuery.retrieveNode(
+ this.retrieveNode).toQuery();
+ if (relationTermQuery == null)
return null;
-
+
+ SpanRelationQuery rq = new SpanRelationQuery(relationTermQuery, true);
SpanQuery subq1, subq2;
+
if (subQuery1.isEmpty) {
if (!subQuery2.isEmpty) {
// match target
subq2 = subQuery2.retrieveNode(this.retrieveNode).toQuery();
- if (subq2 != null) {
- return createQuery(new SpanSegmentQuery(sq, subq2, true));
+ if (subQuery1.hasClass) {
+ rq.setSourceClass(subQuery1.getClassNumber());
}
+
+ return new SpanRelationMatchQuery(rq, subq2, true);
}
}
else if (subQuery2.isEmpty) {
if (!subQuery1.isEmpty) {
// match source
subq1 = subQuery1.retrieveNode(this.retrieveNode).toQuery();
- if (subq1 != null) {
- return createQuery(new SpanSegmentQuery(sq, subq1, true));
+ if (subQuery2.hasClass) {
+ rq.setTargetClass(subQuery2.getClassNumber());
}
+ return new SpanRelationMatchQuery(rq, subq1, true);
}
}
else {
// match both
subq1 = subQuery1.retrieveNode(this.retrieveNode).toQuery();
- if (subq1 != null) {
- SpanFocusQuery fq = new SpanFocusQuery(new SpanSegmentQuery(sq,
- subq1, true), (byte) 2);
- fq.setSorted(false);
- sq = fq;
- }
-
subq2 = subQuery2.retrieveNode(this.retrieveNode).toQuery();
- if (subq2 != null) {
- return createQuery(new SpanSegmentQuery(sq, subq2, true));
- }
+ return new SpanRelationMatchQuery(rq, subq1, subq2, true);
}
- return createQuery(sq);
- }
-
- private SpanQuery createQuery(SpanQuery sq) {
- ArrayList<Byte> classNumbers = new ArrayList<Byte>();
- if (this.classNumbers != null) {
- for (byte c : this.classNumbers) {
- if (c > 0) {
- classNumbers.add(c);
- }
- }
- return new SpanFocusQuery(sq, classNumbers);
+ // both empty
+ if (subQuery1.hasClass) {
+ rq.setSourceClass(subQuery1.getClassNumber());
}
- return sq;
-
+ if (subQuery2.hasClass) {
+ rq.setTargetClass(subQuery2.getClassNumber());
+ }
+ SpanFocusQuery fq = new SpanFocusQuery(rq, rq.getTempClassNumbers());
+ fq.setMatchTemporaryClass(true);
+ fq.setRemoveTemporaryClasses(true);
+ fq.setSorted(false);
+ return fq;
}
}
diff --git a/src/test/java/de/ids_mannheim/korap/index/TestClassFilterIndex.java b/src/test/java/de/ids_mannheim/korap/index/TestClassFilterIndex.java
new file mode 100644
index 0000000..77ef45d
--- /dev/null
+++ b/src/test/java/de/ids_mannheim/korap/index/TestClassFilterIndex.java
@@ -0,0 +1,297 @@
+package de.ids_mannheim.korap.index;
+
+import static org.junit.Assert.assertEquals;
+
+import java.io.IOException;
+
+import org.junit.Test;
+
+import de.ids_mannheim.korap.KrillIndex;
+import de.ids_mannheim.korap.query.DistanceConstraint;
+import de.ids_mannheim.korap.query.SpanClassFilterQuery;
+import de.ids_mannheim.korap.query.SpanClassFilterQuery.ClassOperation;
+import de.ids_mannheim.korap.query.SpanClassQuery;
+import de.ids_mannheim.korap.query.SpanDistanceQuery;
+import de.ids_mannheim.korap.query.SpanElementQuery;
+import de.ids_mannheim.korap.response.Result;
+
+public class TestClassFilterIndex {
+
+ private KrillIndex ki;
+ private Result kr;
+
+ @Test
+ public void testInclude() throws IOException {
+ ki = new KrillIndex();
+ ki.addDoc(TestReferenceIndex.createFieldDoc0());
+ ki.commit();
+
+ SpanElementQuery seq1 = new SpanElementQuery("tokens", "np");
+ SpanElementQuery seq2 = new SpanElementQuery("tokens", "vp");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+ SpanDistanceQuery sdq = new SpanDistanceQuery(scq1, scq2,
+ new DistanceConstraint(0, 1, false, false), true);
+
+ SpanClassFilterQuery sq = new SpanClassFilterQuery(sdq,
+ ClassOperation.INCLUDE, 2, 1, true);
+
+ assertEquals(
+ sq.toString(),
+ "spanClassFilter(spanDistance({1: <tokens:np />}, {2: <tokens:vp />}, "
+ + "[(w[0:1], notOrdered, notExcluded)]),INCLUDE,2,1)");
+
+ kr = ki.search(sq, (short) 10);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+ assertEquals(7, kr.getTotalResults());
+ assertEquals(1, kr.getMatch(0).getStartPos());
+ assertEquals(5, kr.getMatch(0).getEndPos());
+ assertEquals(
+ "Frankenstein, [{2:treat {1:my daughter} well}]. She is the one that saved ...",
+ kr.getMatch(0).getSnippetBrackets());
+ assertEquals(6, kr.getMatch(1).getStartPos());
+ assertEquals(18, kr.getMatch(1).getEndPos());
+ assertEquals(
+ "Frankenstein, treat my daughter well. She [{2:is {1:the one} that saved "
+ + "your master who you hold so dear}].", kr.getMatch(1)
+ .getSnippetBrackets());
+ assertEquals(
+ "Frankenstein, treat my daughter well. She [{2:is {1:the one that "
+ + "saved your master who you hold so dear}}].", kr
+ .getMatch(2).getSnippetBrackets());
+ assertEquals("Frankenstein, treat my daughter well. She [{2:is the one that "
+ + "saved {1:your master} who you hold so dear}].", kr
+ .getMatch(3).getSnippetBrackets());
+ assertEquals(
+ "Frankenstein, treat my daughter well. She [{2:is the one that saved your master who {1:you} hold so dear}].",
+ kr.getMatch(4).getSnippetBrackets());
+
+ }
+
+ @Test
+ public void testDisjoint() throws IOException {
+ ki = new KrillIndex();
+ ki.addDoc(TestReferenceIndex.createFieldDoc0());
+ ki.commit();
+
+ SpanElementQuery seq1 = new SpanElementQuery("tokens", "np");
+ SpanElementQuery seq2 = new SpanElementQuery("tokens", "vp");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+ SpanDistanceQuery sdq = new SpanDistanceQuery(scq1, scq2,
+ new DistanceConstraint(0, 1, false, false), true);
+
+ // kr = ki.search(sdq, (short) 10);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+
+ SpanClassFilterQuery sq = new SpanClassFilterQuery(sdq,
+ ClassOperation.DISJOINT, 2, 1, true);
+
+ kr = ki.search(sq, (short) 10);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+ assertEquals(0, kr.getMatch(0).getStartPos());
+ assertEquals(5, kr.getMatch(0).getEndPos());
+ assertEquals(
+ "[{1:Frankenstein}, {2:treat my daughter well}]. She is the one that saved ...",
+ kr.getMatch(0).getSnippetBrackets());
+ assertEquals(1, kr.getMatch(1).getStartPos());
+ assertEquals(6, kr.getMatch(1).getEndPos());
+ assertEquals(
+ "Frankenstein, [{2:treat my daughter well}. {1:She}] is the one that saved your ...",
+ kr.getMatch(1).getSnippetBrackets());
+
+ assertEquals(5, kr.getMatch(2).getStartPos());
+ assertEquals(18, kr.getMatch(2).getEndPos());
+ assertEquals(
+ "Frankenstein, treat my daughter well. [{1:She} {2:is the one that saved your master who you hold so dear}].",
+ kr.getMatch(2).getSnippetBrackets());
+ }
+
+ // Problem with SpanDistanceQuery - unordered distance spans,
+ // -> unsorted
+ @Test
+ public void testEqual() throws IOException {
+ ki = new KrillIndex();
+ ki.addDoc(TestReferenceIndex.createFieldDoc0());
+ ki.commit();
+
+ SpanElementQuery seq1 = new SpanElementQuery("tokens", "np");
+ SpanElementQuery seq2 = new SpanElementQuery("tokens", "prp");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+ SpanDistanceQuery sdq = new SpanDistanceQuery(scq1, scq2,
+ new DistanceConstraint(0, 1, false, false), true);
+
+ kr = ki.search(sdq, (short) 10);
+ assertEquals(6, kr.getTotalResults());
+
+ kr = ki.search(scq2, (short) 10);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+
+ SpanClassFilterQuery sq = new SpanClassFilterQuery(sdq,
+ ClassOperation.EQUAL, 2, 1, true);
+
+ kr = ki.search(sq, (short) 10);
+ assertEquals(5, kr.getMatch(0).getStartPos());
+ assertEquals(6, kr.getMatch(0).getEndPos());
+ assertEquals(14, kr.getMatch(1).getStartPos());
+ assertEquals(15, kr.getMatch(1).getEndPos());
+
+ }
+
+ @Test
+ public void testDiffer() throws IOException {
+ ki = new KrillIndex();
+ ki.addDoc(TestReferenceIndex.createFieldDoc0());
+ ki.commit();
+
+ SpanElementQuery seq1 = new SpanElementQuery("tokens", "np");
+ SpanElementQuery seq2 = new SpanElementQuery("tokens", "prp");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+ SpanDistanceQuery sdq = new SpanDistanceQuery(scq1, scq2,
+ new DistanceConstraint(0, 2, false, false), true);
+
+ SpanClassFilterQuery sq = new SpanClassFilterQuery(sdq,
+ ClassOperation.DIFFER, 1, 2, true);
+ kr = ki.search(sq, (short) 20);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+
+ assertEquals(9, kr.getTotalResults());
+ assertEquals(0, kr.getMatch(0).getStartPos());
+ assertEquals(3, kr.getMatch(0).getEndPos());
+ assertEquals(
+ "[{1:Frankenstein}, treat {2:my}] daughter well. She is the one ...",
+ kr.getMatch(0).getSnippetBrackets());
+
+ assertEquals(5, kr.getMatch(3).getStartPos());
+ assertEquals(9, kr.getMatch(3).getEndPos());
+ assertEquals(
+ "Frankenstein, treat my daughter well. [{2:She} is {1:the one}] that saved your master who you ...",
+ kr.getMatch(3).getSnippetBrackets());
+ // she is both prp and np
+ }
+
+
+ @Test
+ public void testIntersect() throws IOException {
+ ki = new KrillIndex();
+ ki.addDoc(TestReferenceIndex.createFieldDoc0());
+ ki.commit();
+
+ SpanElementQuery seq1 = new SpanElementQuery("tokens", "np");
+ SpanElementQuery seq2 = new SpanElementQuery("tokens", "vb");
+ SpanClassQuery scq = new SpanClassQuery(seq2, (byte) 3);
+ SpanDistanceQuery sdq = new SpanDistanceQuery(seq1, scq,
+ new DistanceConstraint(0, 1, false, false), true);
+ SpanClassQuery scq1 = new SpanClassQuery(sdq, (byte) 1);
+
+ SpanElementQuery seq3 = new SpanElementQuery("tokens", "prp");
+ SpanDistanceQuery sdq2 = new SpanDistanceQuery(seq3, seq2,
+ new DistanceConstraint(0, 1, false, false), true);
+ SpanClassQuery scq2 = new SpanClassQuery(sdq2, (byte) 2);
+
+ SpanDistanceQuery sdq3 = new SpanDistanceQuery(scq1, scq2,
+ new DistanceConstraint(0, 1, false, false), true);
+
+ SpanClassFilterQuery sq = new SpanClassFilterQuery(sdq3,
+ ClassOperation.INTERSECT, 1, 2, true);
+
+ assertEquals("spanClassFilter(spanDistance({1: spanDistance(<tokens:np />, " +
+ "{3: <tokens:vb />}, [(w[0:1], notOrdered, notExcluded)])}, " +
+ "{2: spanDistance(<tokens:prp />, <tokens:vb />, [(w[0:1], " +
+ "notOrdered, notExcluded)])}, [(w[0:1], notOrdered, notExcluded)]),INTERSECT,1,2)",
+ sq.toString());
+
+
+ kr = ki.search(sq, (short) 20);
+
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+
+ assertEquals(13, kr.getTotalResults());
+ assertEquals(0, kr.getMatch(0).getStartPos());
+ assertEquals(3, kr.getMatch(0).getEndPos());
+ assertEquals(
+ "[{1:Frankenstein, {2:{3:treat}}}{2: my}] daughter well. She is the one ...",
+ kr.getMatch(0).getSnippetBrackets());
+ assertEquals(1, kr.getMatch(1).getStartPos());
+ assertEquals(4, kr.getMatch(1).getEndPos());
+ assertEquals(
+ "Frankenstein, [{1:{2:{3:treat} my} daughter}] well. She is the one that ...",
+ kr.getMatch(1).getSnippetBrackets());
+ }
+
+
+ @Test
+ public void testMultipleSameClasses() throws IOException {
+
+ ki = new KrillIndex();
+ ki.addDoc(TestReferenceIndex.createFieldDoc0());
+ ki.commit();
+
+ SpanElementQuery seq1 = new SpanElementQuery("tokens", "nn");
+ SpanElementQuery seq = new SpanElementQuery("tokens", "prp");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq = new SpanClassQuery(seq, (byte) 1);
+
+ SpanDistanceQuery sdq = new SpanDistanceQuery(scq, scq1,
+ new DistanceConstraint(3, 5, false, false), true);
+
+ SpanElementQuery seq2 = new SpanElementQuery("tokens", "vp");
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+
+ SpanDistanceQuery sdq2 = new SpanDistanceQuery(sdq, scq2,
+ new DistanceConstraint(0, 1, false, false), true);
+
+ SpanClassFilterQuery sq = new SpanClassFilterQuery(sdq2,
+ ClassOperation.INCLUDE, 2, 1, true);
+
+ kr = ki.search(sdq2, (short) 20);
+ assertEquals(6, kr.getTotalResults());
+
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+
+ kr = ki.search(sq, (short) 20);
+
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+
+ assertEquals(6, kr.getMatch(0).getStartPos());
+ assertEquals(18, kr.getMatch(0).getEndPos());
+ assertEquals(
+ "Frankenstein, treat my daughter well. She [{2:is the {1:one} that saved {1:your} master who you hold so dear}].",
+ kr.getMatch(0).getSnippetBrackets());
+ }
+
+}
\ No newline at end of file
diff --git a/src/test/java/de/ids_mannheim/korap/index/TestReferenceIndex.java b/src/test/java/de/ids_mannheim/korap/index/TestReferenceIndex.java
new file mode 100644
index 0000000..4a56a04
--- /dev/null
+++ b/src/test/java/de/ids_mannheim/korap/index/TestReferenceIndex.java
@@ -0,0 +1,324 @@
+package de.ids_mannheim.korap.index;
+
+import static de.ids_mannheim.korap.TestSimple.getJSONQuery;
+import static org.junit.Assert.assertEquals;
+
+import java.io.IOException;
+
+import org.apache.lucene.index.Term;
+import org.apache.lucene.search.spans.SpanQuery;
+import org.apache.lucene.search.spans.SpanTermQuery;
+import org.junit.Test;
+
+import de.ids_mannheim.korap.KrillIndex;
+import de.ids_mannheim.korap.query.DistanceConstraint;
+import de.ids_mannheim.korap.query.SpanClassQuery;
+import de.ids_mannheim.korap.query.SpanDistanceQuery;
+import de.ids_mannheim.korap.query.SpanElementQuery;
+import de.ids_mannheim.korap.query.SpanFocusQuery;
+import de.ids_mannheim.korap.query.SpanNextQuery;
+import de.ids_mannheim.korap.query.SpanReferenceQuery;
+import de.ids_mannheim.korap.query.SpanRelationMatchQuery;
+import de.ids_mannheim.korap.query.SpanRelationQuery;
+import de.ids_mannheim.korap.query.wrap.SpanQueryWrapper;
+import de.ids_mannheim.korap.response.Result;
+import de.ids_mannheim.korap.util.QueryException;
+
+public class TestReferenceIndex {
+ private KrillIndex ki;
+ private Result kr;
+
+ @Test
+ public void testCase1 () throws IOException {
+ ki = new KrillIndex();
+ ki.addDoc(TestRelationIndex.createFieldDoc2());
+ ki.commit();
+
+ SpanTermQuery seq1 = new SpanTermQuery(new Term("base", "pos:V"));
+ SpanElementQuery seq2 = new SpanElementQuery("base", "np");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+ SpanNextQuery snq1 = new SpanNextQuery(scq1, scq2);
+
+ SpanFocusQuery sfq1 = new SpanFocusQuery(snq1, (byte) 2);
+
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
+ new Term("base", "<:child-of")), true);
+ // SpanSegmentQuery ssq = new SpanSegmentQuery(srq, sfq1,
+ // true);
+ // SpanFocusQuery sfq2 = new SpanFocusQuery(ssq, (byte) 1);
+ // sfq2.setSorted(false);
+ // sfq2.setMatchTemporaryClass(false);
+
+ SpanElementQuery seq3 = new SpanElementQuery("base", "pp");
+ SpanClassQuery scq3 = new SpanClassQuery(seq3, (byte) 3);
+ // SpanSegmentQuery ssq2 = new SpanSegmentQuery(sfq2, scq3,
+ // true);
+
+ SpanRelationMatchQuery rq = new SpanRelationMatchQuery(srq, sfq1, scq3, true);
+
+ // System.out.println(rq.toString());
+ SpanFocusQuery sfq3 = new SpanFocusQuery(rq, (byte) 1);
+
+ DistanceConstraint constraint = new DistanceConstraint(3, 3, true,
+ false);
+ SpanDistanceQuery sdq = new SpanDistanceQuery(sfq3, scq3, constraint,
+ true);
+
+ SpanReferenceQuery ref = new SpanReferenceQuery(sdq, (byte) 3, true);
+ // System.out.println(ref.toString());
+
+ kr = ki.search(ref, (short) 10);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+ assertEquals(
+ "spanReference(spanDistance(focus(1: focus(#[1,2]spanSegment("
+ + "focus(#1: spanSegment(spanRelation(base:<:child-of), focus(2: spanNext("
+ + "{1: base:pos:V}, {2: <base:np />})))), {3: <base:pp />}))), "
+ + "{3: <base:pp />}, [(w[3:3], ordered, notExcluded)]), 3)",
+ ref.toString());
+
+ assertEquals(1, kr.getMatch(0).getStartPos());
+ assertEquals(7, kr.getMatch(0).getEndPos());
+ }
+
+ @Test
+ public void testCase2() throws IOException, QueryException {
+
+ String filepath = getClass().getResource(
+ "/queries/reference/distance-reference.jsonld").getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+
+ // cat="vb" & cat="prp" & cat="nn" & #1 .notordered #2 & #1
+ // .{0,2} #3 & #3 -> #2
+
+ assertEquals(
+ "spanReference(focus(#[1,2]spanSegment(focus(#2: "
+ + "spanSegment(spanRelation(tokens:>:stanford/d:tag), "
+ + "focus(3: spanDistance(focus(1: spanDistance({1: <tokens:vb />}, "
+ + "{2: <tokens:prp />}, [(w[0:1], notOrdered, notExcluded)])), "
+ + "{3: <tokens:nn />}, [(w[0:2], notOrdered, notExcluded)])))), "
+ + "{2: <tokens:prp />})), 2)",
+ sq.toString());
+
+ SpanElementQuery seq1 = new SpanElementQuery("tokens", "vb");
+ // new SpanTermQuery(new Term("tokens", "c:vb"));
+ SpanElementQuery seq2 = new SpanElementQuery("tokens", "prp");
+ // new SpanTermQuery(new Term("tokens", "c:prp"));
+ SpanElementQuery seq3 = new SpanElementQuery("tokens", "nn");
+ // new SpanTermQuery(new Term("tokens", "c:nn"));
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+ SpanClassQuery scq3 = new SpanClassQuery(seq3, (byte) 3);
+
+ // vb .{0,1} prp
+ SpanDistanceQuery sdq1 = new SpanDistanceQuery(scq1, scq2,
+ new DistanceConstraint(0, 1, false, false), true);
+ SpanFocusQuery sfq1 = new SpanFocusQuery(sdq1, (byte) 1);
+
+ // vb .{0,2} nn
+ SpanDistanceQuery sdq2 = new SpanDistanceQuery(sfq1, scq3,
+ new DistanceConstraint(0, 2, false, false), true);
+ SpanFocusQuery sfq2 = new SpanFocusQuery(sdq2, (byte) 3);
+
+ // nn -> prp
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
+ new Term("tokens", ">:stanford/d:tag")), true);
+ SpanRelationMatchQuery rq = new SpanRelationMatchQuery(srq, sfq2, scq2,
+ true);
+
+ SpanReferenceQuery ref = new SpanReferenceQuery(rq, (byte) 2, true);
+
+ ki = new KrillIndex();
+ ki.addDoc(createFieldDoc0());
+ ki.commit();
+
+ kr = ki.search(ref, (short) 10);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+
+ assertEquals(sq.toString(), ref.toString());
+ assertEquals(2, kr.getMatch(0).getStartPos());
+ assertEquals(4, kr.getMatch(0).getEndPos());
+ assertEquals(5, kr.getMatch(1).getStartPos());
+ assertEquals(9, kr.getMatch(1).getEndPos());
+ assertEquals(11, kr.getMatch(2).getStartPos());
+ assertEquals(13, kr.getMatch(2).getEndPos());
+
+ // multiple references
+
+ SpanFocusQuery sfq3 = new SpanFocusQuery(ref, (byte) 1);
+ // vp -> nn
+ SpanRelationMatchQuery rq2 = new SpanRelationMatchQuery(srq, sfq3,
+ scq3, true);
+
+ SpanReferenceQuery ref2 = new SpanReferenceQuery(rq2, (byte) 3, true);
+
+ kr = ki.search(ref2, (short) 10);
+ assertEquals(1, kr.getMatch(0).getStartPos());
+ assertEquals(4, kr.getMatch(0).getEndPos());
+ assertEquals(10, kr.getMatch(1).getStartPos());
+ assertEquals(13, kr.getMatch(1).getEndPos());
+
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+ }
+
+ // multiple references
+ @Test
+ public void testCase3() throws IOException, QueryException {
+ ki = new KrillIndex();
+ ki.addDoc(createFieldDoc0());
+ ki.commit();
+
+ String filepath = getClass().getResource(
+ "/queries/reference/distance-multiple-references.jsonld")
+ .getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+
+ // 'cat="vb" & cat="prp" & cat="nn" & #1 .{0,1} #2 & #1 .{0,2}
+ // #3 & #3 ->stanford/d #2 & #1 ->stanford #3' annis
+ // without layer=c and + relation key
+ assertEquals(
+ "spanReference(focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(tokens:>:stanford/d:tag), "
+ + "focus(1: spanReference(focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(tokens:>:stanford/d:tag), "
+ + "focus(3: spanDistance(focus(1: spanDistance({1: <tokens:vb />}, {2: <tokens:prp />}, "
+ + "[(w[0:1], notOrdered, notExcluded)])), {3: <tokens:nn />}, [(w[0:2], notOrdered, notExcluded)])))), "
+ + "{2: <tokens:prp />})), 2)))), {3: <tokens:nn />})), 3)",
+ sq.toString());
+ kr = ki.search(sq, (short) 10);
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+ assertEquals(1, kr.getMatch(0).getStartPos());
+ assertEquals(4, kr.getMatch(0).getEndPos());
+ assertEquals(10, kr.getMatch(1).getStartPos());
+ assertEquals(13, kr.getMatch(1).getEndPos());
+ }
+
+ // multiple document
+ @Test
+ public void testCase4() throws Exception {
+ ki = new KrillIndex();
+ ki.addDoc(createFieldDoc0());
+ ki.addDoc(createFieldDoc1());
+ ki.commit();
+
+ String filepath = getClass().getResource(
+ "/queries/reference/distance-reference.jsonld")
+ .getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+
+ kr = ki.search(sq, (short) 10);
+
+ assertEquals(4, kr.getTotalResults());
+ assertEquals("doc-1", kr.getMatch(3).getDocID());
+ assertEquals(2, kr.getMatch(3).getStartPos());
+ assertEquals(4, kr.getMatch(3).getEndPos());
+ }
+
+ public static FieldDocument createFieldDoc1() {
+ FieldDocument fd = new FieldDocument();
+ fd.addString("ID", "doc-1");
+ fd.addTV(
+ "tokens",
+ "Frankenstein, treat my daughter well. She is the one that saved your master who you hold so dear.",
+
+ "[(0-12)s:Frankenstein|_0#0-12|<>:nn#0-12$<i>1<s>18|<>:s#0-37$<i>4<s>1|"
+ + "<>:np#0-13$<i>1<s>2|"
+ + "<:stanford/d:tag$<i>1<s>1<s>18<s>19]"
+
+ + "[(14-19)s:treat|_1#14-19|<>:vb#14-19$<i>2<s>19|<>:vp#14-36$<i>4<s>3|"
+ + ">:stanford/d:tag$<i>0<s>2<s>19<s>18|"
+ + ">:stanford/d:tag$<i>3<s>3<s>19<s>21|"
+ + ">:stanford/d:tag$<i>4<s>4<s>19<s>22]"
+
+ + "[(20-22)s:my|_2#20-22|<>:prp#20-22$<i>3<s>20|<>:np#20-31$<i>3<s>4]"
+
+ + "[(23-31)s:daughter|_3#23-31|<>:nn#23-31$<i>4<s>21|"
+ + ">:stanford/d:tag$<i>2<s>5<s>21<s>20]"
+
+ + "[(32-36)s:well|_4#32-36|<>:rb#32-36$<i>5<s>22|<>:advp#32-36$<i>4<s>5]");
+ return fd;
+ }
+
+ public static FieldDocument createFieldDoc0 () {
+ FieldDocument fd = new FieldDocument();
+ fd.addString("ID", "doc-0");
+ fd.addTV(
+ "tokens",
+ "Frankenstein, treat my daughter well. She is the one that saved your master who you hold so dear.",
+
+ "[(0-12)s:Frankenstein|_0#0-12|<>:nn#0-12$<i>1<s>18|<>:s#0-37$<i>5<s>1|"
+ + "<>:np#0-13$<i>1<s>2|"
+ + "<:stanford/d:tag$<i>1<s>1<s>18<s>19]"
+
+ + "[(14-19)s:treat|_1#14-19|<>:vb#14-19$<i>2<s>19|<>:vp#14-36$<i>5<s>3|"
+ + ">:stanford/d:tag$<i>0<s>2<s>19<s>18|"
+ + ">:stanford/d:tag$<i>3<s>3<s>19<s>21|"
+ + ">:stanford/d:tag$<i>4<s>4<s>19<s>22]"
+
+ + "[(20-22)s:my|_2#20-22|<>:prp#20-22$<i>3<s>20|<>:np#20-31$<i>4<s>4]"
+
+ + "[(23-31)s:daughter|_3#23-31|<>:nn#23-31$<i>4<s>21|"
+ + ">:stanford/d:tag$<i>2<s>5<s>21<s>20]"
+
+ + "[(32-36)s:well|_4#32-36|<>:rb#32-36$<i>5<s>22|<>:advp#32-36$<i>5<s>5]"
+
+ + "[(38-41)s:She|_5#38-41|<>:prp#38-41$<i>6<s>23|<>:np#38-41$<i>6<s>36|<>:s#38-97$<i>18<s>6]"
+
+ + "[(42-44)s:is|_6#42-44|<>:vb#42-44$<i>7<s>24|<>:vp#42-96$<i>18<s>7]"
+
+ + "[(45-48)s:the|_7#45-48|<>:dt#45-48$<i>8<s>25|<>:np#45-52$<i>9<s>8|<>:np#45-96$<i>18<s>9]"
+
+ + "[(49-52)s:one|_8#49-52|<>:nn#49-52$<i>9<s>26|"
+ + ">:stanford/d:tag$<i>5<s>6<s>26<s>23|"
+ + ">:stanford/d:tag$<i>6<s>7<s>26<s>24|"
+ + ">:stanford/d:tag$<i>7<s>8<s>26<s>25|"
+ + ">:stanford/d:tag$<i>10<s>9<s>26<s>28]"
+
+ + "[(53-57)s:that|_9#53-57|<>:rp#53-57$<i>10<s>27|<>:sb#53-96$<i>18<s>10]"
+
+ + "[(58-63)s:saved|_10#58-63|<>:vb#58-63$<i>11<s>28|<>:s#58-96$<i>18<s>11|"
+ + "<>:vp#58-96$<i>18<s>12|"
+ + ">:stanford/d:tag$<i>9<s>10<s>28<s>27|"
+ + ">:stanford/d:tag$<i>12<s>11<s>28<s>30|"
+ + ">:stanford/d:tag$<i>15<s>12<s>28<s>33]"
+
+ + "[(64-68)s:your|_11#64-68|<>:prp#64-68$<i>12<s>29|<>:np#64-75$<i>13<s>13]"
+
+ + "[(69-75)s:master|_12#69-75|<>:nn#69-75$<i>13<s>30|"
+ + ">:stanford/d:tag$<i>11<s>13<s>30<s>29]"
+
+ + "[(76-79)s:who|_13#76-79|<>:rp#76-79$<i>14<s>31|<>:sb#76-96$<i>18<s>14]"
+
+ + "[(80-83)s:you|_14#80-83|<>:prp#80-83$<i>15<s>32|<>:np#80-83$<i>15<s>37|<>:s#80-96$<i>18<s>15]"
+
+ + "[(84-88)s:hold|_15#84-88|<>:vb#84-88$<i>16<s>33|<>:vp#84-96$<i>18<s>16|"
+ + ">:stanford/d:tag$<i>13<s>14<s>33<s>31|"
+ + ">:stanford/d:tag$<i>14<s>15<s>33<s>32|"
+ + ">:stanford/d:tag$<i>17<s>16<s>33<s>35]"
+
+ + "[(89-91)s:so|_16#89-91|<>:rb#89-91$<i>17<s>341|<>:adjp#89-96$<i>18<s>17]"
+
+ + "[(92-96)s:dear|_17#92-96|<>:jj#92-96$<i>18<s>35|"
+ + ">:stanford/d:tag$<i>16<s>17<s>35<s>34]"
+ );
+
+ return fd;
+ }
+}
diff --git a/src/test/java/de/ids_mannheim/korap/index/TestRelationIndex.java b/src/test/java/de/ids_mannheim/korap/index/TestRelationIndex.java
index e0d6598..63d2a64 100644
--- a/src/test/java/de/ids_mannheim/korap/index/TestRelationIndex.java
+++ b/src/test/java/de/ids_mannheim/korap/index/TestRelationIndex.java
@@ -3,7 +3,6 @@
import static org.junit.Assert.assertEquals;
import java.io.IOException;
-import java.util.ArrayList;
import org.apache.lucene.index.Term;
import org.apache.lucene.search.spans.SpanQuery;
@@ -12,10 +11,11 @@
import de.ids_mannheim.korap.KrillIndex;
import de.ids_mannheim.korap.query.SpanAttributeQuery;
+import de.ids_mannheim.korap.query.SpanClassQuery;
import de.ids_mannheim.korap.query.SpanElementQuery;
import de.ids_mannheim.korap.query.SpanFocusQuery;
+import de.ids_mannheim.korap.query.SpanRelationMatchQuery;
import de.ids_mannheim.korap.query.SpanRelationQuery;
-import de.ids_mannheim.korap.query.SpanSegmentQuery;
import de.ids_mannheim.korap.query.SpanTermWithIdQuery;
import de.ids_mannheim.korap.query.SpanWithAttributeQuery;
import de.ids_mannheim.korap.response.Result;
@@ -128,7 +128,7 @@
+ "<:child-of$<i>0<s>3<s>3<s>1|"
+ "<:child-of$<i>7<i>0<i>1<s>4<s>2<s>3|"
+ "<:child-of$<i>7<i>1<i>7<s>5<s>2<s>2|"
- + "<:dep$<i>1<s>2<s>1<s>1|"
+ + "<:dep$<i>0<s>2<s>1<s>1|"
+ "r@:func=sbj$<i>0<i>7<s>1]"
+
@@ -149,14 +149,15 @@
+ "<:child-of$<i>4<b>0<i>3<s>14<s>3<s>1|"
+ "<:child-of$<i>7<i>2<i>4<s>15<s>4<s>3|"
+ "<:child-of$<i>7<i>4<i>7<s>16<s>4<s>2|"
- + "<:dep$<i>3<s>2<s>1<s>1]" +
+ + ">:parent-of$<i>7<i>4<i>7<s>17<s>4<s>2|"
+ + "<:dep$<i>3<s>3<s>1<s>1]" +
"[(3-4)s:Blümen|_3#14-20|pos:NN$<s>1|"
+ ">:child-of$<i>2<i>4<s>17<s>1<s>3|"
+ "<:dep$<i>1<s>2<s>1<s>1|" + ">:dep$<i>2<s>3<s>1<s>1|"
+ ">:dep$<i>4<s>4<s>1<s>1|"
- + "r@:func=head$<i>2<i>4<s>2|"
- + "r@:func=obj$<i>1<i>4<s>2]" +
+ + "r@:func=head$<i>2<i>4<s>3|"
+ + "r@:func=obj$<i>2<i>4<s>3]" +
"[(4-5)s:für|_4#21-24|pos:PREP$<s>1|<>:pp#21-38$<i>7<s>2|"
+ ">:child-of$<i>4<i>7<s>18<s>1<s>2|"
@@ -261,9 +262,7 @@
/**
- * Relations with attributes
- * need focusMulti on span relation query before
- * SpanWithAttributeQuery
+ * Relations only
* */
@Test
public void testCase3 () throws IOException {
@@ -288,32 +287,11 @@
assertEquals(3, kr.getMatch(4).getEndPos());
assertEquals(2, kr.getMatch(5).getStartPos());
assertEquals(4, kr.getMatch(5).getEndPos());
-
- ArrayList<Byte> classNumbers = new ArrayList<Byte>();
- classNumbers.add((byte) 1);
- classNumbers.add((byte) 2);
-
- SpanFocusQuery fq = new SpanFocusQuery(srq, classNumbers);
- kr = ki.search(fq, (short) 20);
- /*
- * for (Match km : kr.getMatches()) {
- * System.out.println(km.getStartPos() + "," + km.getEndPos()
- * + " " + km.getSnippetBrackets()); }
- */
- assertEquals((long) 13, kr.getTotalResults());
- assertEquals(0, kr.getMatch(0).getStartPos());
- assertEquals(1, kr.getMatch(0).getEndPos());
- assertEquals(0, kr.getMatch(1).getStartPos());
- assertEquals(7, kr.getMatch(1).getEndPos());
- assertEquals(0, kr.getMatch(2).getStartPos());
- assertEquals(7, kr.getMatch(2).getEndPos());
- assertEquals(1, kr.getMatch(3).getStartPos());
- assertEquals(7, kr.getMatch(3).getEndPos());
- assertEquals(1, kr.getMatch(4).getStartPos());
- assertEquals(7, kr.getMatch(4).getEndPos());
}
-
+ /**
+ * Relations only with/out attribute
+ * */
@Test
public void testCase4 () throws IOException {
ki.addDoc(createFieldDoc2());
@@ -322,12 +300,14 @@
SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
new Term("base", ">:child-of")), true);
- ArrayList<Byte> classNumbers = new ArrayList<Byte>();
- classNumbers.add((byte) 1);
- classNumbers.add((byte) 2);
+ SpanFocusQuery fq = new SpanFocusQuery(srq, srq.getTempClassNumbers());
+ fq.setMatchTemporaryClass(true);
+ fq.setRemoveTemporaryClasses(true);
+ fq.setSorted(false);
- SpanWithAttributeQuery wq = new SpanWithAttributeQuery(
- new SpanFocusQuery(srq, classNumbers), new SpanAttributeQuery(
+ // child-of with attr func=sbj
+ SpanWithAttributeQuery wq = new SpanWithAttributeQuery(fq,
+ new SpanAttributeQuery(
new SpanTermQuery(new Term("base", "r@:func=sbj")),
true), true);
@@ -337,7 +317,7 @@
assertEquals(7, kr.getMatch(0).getEndPos());
// child-of without attr func=sbj
- wq = new SpanWithAttributeQuery(new SpanFocusQuery(srq, classNumbers),
+ wq = new SpanWithAttributeQuery(fq,
new SpanAttributeQuery(new SpanTermQuery(new Term("base",
"r@:func=sbj")), true, true), true);
kr = ki.search(wq, (short) 20);
@@ -345,87 +325,9 @@
}
- @Test
- public void testCase5 () throws IOException {
-
- ki.addDoc(createFieldDoc2());
- ki.commit();
-
- SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
- new Term("base", "<:dep")), true);
- kr = ki.search(srq, (short) 10);
-
- ArrayList<Byte> classNumbers = new ArrayList<Byte>();
- classNumbers.add((byte) 1);
- classNumbers.add((byte) 2);
-
- SpanFocusQuery fq = new SpanFocusQuery(srq, classNumbers);
- kr = ki.search(fq, (short) 10);
- // for (Match km : kr.getMatches()) {
- // System.out.println(km.getStartPos() + "," + km.getEndPos()
- // + " "
- // + km.getSnippetBrackets());
- // }
-
- SpanAttributeQuery saq = new SpanAttributeQuery(new SpanTermQuery(
- new Term("base", "r@:func=obj")), true);
- kr = ki.search(saq, (short) 10);
-
- // child-of with attr func-obj
- SpanWithAttributeQuery wq = new SpanWithAttributeQuery(
- new SpanFocusQuery(srq, classNumbers), new SpanAttributeQuery(
- new SpanTermQuery(new Term("base", "r@:func=obj")),
- true), true);
-
- kr = ki.search(wq, (short) 10);
- assertEquals((long) 1, kr.getTotalResults());
- assertEquals(1, kr.getMatch(0).getStartPos()); // element
- assertEquals(4, kr.getMatch(0).getEndPos());
- }
-
-
- @Test
- public void testCase10 () throws IOException {
- ki.addDoc(createFieldDoc2());
- ki.commit();
- // target of a dependency relation
- SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
- new Term("base", "<:dep")), true);
- kr = ki.search(srq, (short) 10);
- assertEquals((long) 6, kr.getTotalResults());
-
- ArrayList<Byte> classNumbers = new ArrayList<Byte>();
- classNumbers.add((byte) 1);
- classNumbers.add((byte) 2);
-
- SpanFocusQuery fq = new SpanFocusQuery(srq, classNumbers);
- kr = ki.search(fq, (short) 10);
- assertEquals((long) 6, kr.getTotalResults());
-
- SpanAttributeQuery aq = new SpanAttributeQuery(new SpanTermQuery(
- new Term("base", "r@:func=head")), true);
- kr = ki.search(aq, (short) 10);
-
- // dependency relation, which is also a head
- SpanWithAttributeQuery wq = new SpanWithAttributeQuery(
- new SpanFocusQuery(srq, classNumbers), new SpanAttributeQuery(
- new SpanTermQuery(new Term("base", "r@:func=head")),
- true), true);
-
- kr = ki.search(wq, (short) 20);
-
- assertEquals((long) 2, kr.getTotalResults());
- assertEquals(2, kr.getMatch(0).getStartPos());
- assertEquals(4, kr.getMatch(0).getEndPos());
- assertEquals(5, kr.getMatch(1).getStartPos());
- assertEquals(7, kr.getMatch(1).getEndPos());
-
- }
-
-
/**
- * Match left return left
- * Match right return right
+ * Relation directions <br/>
+ * Relation with specific sources, return the sources
* */
@Test
public void testCase6 () throws IOException {
@@ -433,9 +335,19 @@
ki.commit();
// return all children that are NP
- SpanQuery rv = new SpanSegmentQuery(new SpanRelationQuery(
- new SpanTermQuery(new Term("base", ">:child-of")), true),
- new SpanElementQuery("base", "np"), true);
+ SpanElementQuery seq1 = new SpanElementQuery("base", "np");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
+ new Term("base", ">:child-of")), true);
+
+ SpanRelationMatchQuery rm = new SpanRelationMatchQuery(srq, scq1, true);
+ SpanFocusQuery rv = new SpanFocusQuery(rm, (byte) 1);
+ rv.setSorted(false);
+
+ assertEquals(
+ "focus(1: focus(#[1,2]spanSegment(spanRelation(base:>:child-of), {1: <base:np />})))",
+ rv.toString());
kr = ki.search(rv, (short) 10);
@@ -450,10 +362,10 @@
assertEquals(7, kr.getMatch(3).getEndPos());
// return all parents that are NP
- rv = new SpanSegmentQuery(new SpanRelationQuery(new SpanTermQuery(
- new Term("base", "<:child-of")), true), new SpanElementQuery(
- "base", "np"), true);
-
+ srq = new SpanRelationQuery(new SpanTermQuery(new Term("base",
+ "<:child-of")), true);
+ rm = new SpanRelationMatchQuery(srq, scq1, true);
+ rv = new SpanFocusQuery(rm, (byte) 1);
kr = ki.search(rv, (short) 10);
assertEquals(7, kr.getTotalResults());
@@ -473,24 +385,78 @@
assertEquals(7, kr.getMatch(6).getEndPos());
}
+ /**
+ * Dependency relations with attribute
+ * */
+ @Test
+ public void testCase5() throws IOException {
+ ki.addDoc(createFieldDoc2());
+ ki.commit();
+
+ // target of a dependency relation
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
+ new Term("base", "<:dep")), true);
+ kr = ki.search(srq, (short) 10);
+ assertEquals((long) 6, kr.getTotalResults());
+
+ SpanFocusQuery fq = new SpanFocusQuery(srq, srq.getTempClassNumbers());
+ fq.setMatchTemporaryClass(true);
+ fq.setRemoveTemporaryClasses(true);
+ // fq.setSorted(false);
+
+ kr = ki.search(fq, (short) 10);
+ assertEquals((long) 6, kr.getTotalResults());
+ // for (Match km : kr.getMatches()) {
+ // System.out.println(km.getStartPos() + "," + km.getEndPos()
+ // + " "
+ // + km.getSnippetBrackets());
+ // }
+ SpanAttributeQuery aq = new SpanAttributeQuery(new SpanTermQuery(
+ new Term("base", "r@:func=head")), true);
+ kr = ki.search(aq, (short) 10);
+
+ // dependency relation, which is also a head
+ SpanWithAttributeQuery wq = new SpanWithAttributeQuery(fq,
+ new SpanAttributeQuery(new SpanTermQuery(new Term("base",
+ "r@:func=head")), true), true);
+
+ kr = ki.search(wq, (short) 20);
+
+ assertEquals((long) 2, kr.getTotalResults());
+ assertEquals(2, kr.getMatch(0).getStartPos());
+ assertEquals(4, kr.getMatch(0).getEndPos());
+ assertEquals(5, kr.getMatch(1).getStartPos());
+ assertEquals(7, kr.getMatch(1).getEndPos());
+
+ }
/**
- * Match left, return right
- * sort by left, then sort by right
+ * Relation with specific sources, return any targets
* */
@Test
public void testCase7 () throws IOException {
ki.addDoc(createFieldDoc2());
ki.commit();
- // return all children that are NP
- SpanQuery rv = new SpanSegmentQuery(new SpanRelationQuery(
- new SpanTermQuery(new Term("base", ">:child-of")), true),
- new SpanElementQuery("base", "np"), true);
+ // match all children that are NP
+ SpanElementQuery seq1 = new SpanElementQuery("base", "np");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
+ new Term("base", ">:child-of")), true);
+ srq.setTargetClass((byte) 2);
+
+ SpanRelationMatchQuery rm = new SpanRelationMatchQuery(srq, scq1, true);
+ // SpanQuery rv = new SpanFocusQuery(rm, (byte) 1);
//return all parents of np
- SpanFocusQuery rv2 = new SpanFocusQuery(rv, (byte) 2);
+ SpanFocusQuery rv2 = new SpanFocusQuery(rm, (byte) 2);
rv2.setSorted(false);
+
+ assertEquals(
+ "focus(2: focus(#[1,2]spanSegment({2: target:spanRelation(base:>:child-of)}, {1: <base:np />})))",
+ rv2.toString());
+
kr = ki.search(rv2, (short) 10);
assertEquals((long) 4, kr.getTotalResults());
@@ -519,26 +485,32 @@
ki.commit();
//return source of dep relations to pos:NN
- SpanQuery rv = new SpanFocusQuery(new SpanSegmentQuery(
- new SpanRelationQuery(new SpanTermQuery(new Term("base",
- "<:dep")), true), new SpanTermWithIdQuery(new Term(
- "base", "pos:NN"), true), true), (byte) 1);
+
+ SpanTermWithIdQuery tq = new SpanTermWithIdQuery(new Term("base",
+ "pos:NN"), true);
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(new Term("base",
+ "<:dep")), true);
+ srq.setSourceClass((byte) 1);
+ SpanRelationMatchQuery rm = new SpanRelationMatchQuery(srq, tq, true);
+ SpanQuery rv = new SpanFocusQuery(rm, (byte) 1);
kr = ki.search(rv, (short) 10);
assertEquals((long) 3, kr.getTotalResults());
- assertEquals(1, kr.getMatch(0).getStartPos());
- assertEquals(2, kr.getMatch(0).getEndPos());
+ assertEquals(0, kr.getMatch(0).getStartPos());
+ assertEquals(1, kr.getMatch(0).getEndPos());
assertEquals(1, kr.getMatch(1).getStartPos());
assertEquals(2, kr.getMatch(1).getEndPos());
assertEquals(4, kr.getMatch(2).getStartPos());
assertEquals(5, kr.getMatch(2).getEndPos());
- //return target of dep relations from pos:NN
- rv = new SpanFocusQuery(
- new SpanSegmentQuery(new SpanRelationQuery(new SpanTermQuery(
- new Term("base", ">:dep")), true),
- new SpanTermWithIdQuery(new Term("base", "pos:NN"),
- true), true), (byte) 2);
+ // return target of dep relations from pos:NN
+ srq = new SpanRelationQuery(
+ new SpanTermQuery(
+ new Term("base", ">:dep")), true);
+ srq.setTargetClass((byte) 1);
+ rm = new SpanRelationMatchQuery(srq, tq, true);
+ rv = new SpanFocusQuery(rm, (byte) 1);
+
kr = ki.search(rv, (short) 10);
assertEquals((long) 3, kr.getTotalResults());
assertEquals(2, kr.getMatch(0).getStartPos());
@@ -552,8 +524,9 @@
/**
- * Relation with variable match right, return left sort by right,
- * then sort by left
+ * Relation with specific sources and return any targets <br/>
+ * Relation with specific sources and targets, return the targets <br/>
+ * Relation with specific sources and targets, return the sources
*
* @throws IOException
* */
@@ -563,12 +536,19 @@
ki.commit();
// return all children of np
- SpanFocusQuery rv = new SpanFocusQuery(new SpanSegmentQuery(
- new SpanRelationQuery(new SpanTermQuery(new Term("base",
- "<:child-of")), true), new SpanElementQuery("base",
- "np"), true), (byte) 1);
+ SpanElementQuery seq1 = new SpanElementQuery("base", "np");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
+ new Term("base", "<:child-of")), true);
+ srq.setSourceClass((byte) 2);
+ SpanRelationMatchQuery rm = new SpanRelationMatchQuery(srq, scq1, true);
+ SpanFocusQuery rv = new SpanFocusQuery(rm, (byte) 2);
rv.setSorted(false);
+ assertEquals(
+ "focus(2: focus(#[1,2]spanSegment({2: source:spanRelation(base:<:child-of)}, {1: <base:np />})))",
+ rv.toString());
+
kr = ki.search(rv, (short) 10);
assertEquals((long) 7, kr.getTotalResults());
@@ -586,12 +566,23 @@
assertEquals(6, kr.getMatch(5).getEndPos());
assertEquals(6, kr.getMatch(6).getStartPos());
assertEquals(7, kr.getMatch(6).getEndPos());
- // sorting left problem (solved)
// return all children of np that are articles
- SpanSegmentQuery rv2 = new SpanSegmentQuery(rv, new SpanTermQuery(
- new Term("base", "pos:ART")));
- kr = ki.search(rv2, (short) 10);
+
+ SpanTermWithIdQuery tiq = new SpanTermWithIdQuery(new Term("base",
+ "pos:ART"), true);
+ SpanClassQuery scq2 = new SpanClassQuery(tiq, (byte) 2);
+
+ srq = new SpanRelationQuery(new SpanTermQuery(new Term("base",
+ ">:child-of")), true);
+ rm = new SpanRelationMatchQuery(srq, scq2, scq1, true);
+ rv = new SpanFocusQuery(rm, (byte) 2);
+
+ assertEquals(
+ "focus(2: focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(base:>:child-of), {2: spanTermWithId(base:pos:ART)})), {1: <base:np />})))",
+ rv.toString());
+
+ kr = ki.search(rv, (short) 10);
assertEquals((long) 2, kr.getTotalResults());
assertEquals(2, kr.getMatch(0).getStartPos());
@@ -600,20 +591,45 @@
assertEquals(6, kr.getMatch(1).getEndPos());
// return all nps whose children are articles
- SpanSegmentQuery rv3 = new SpanSegmentQuery(rv,
- new SpanTermWithIdQuery(new Term("base", "pos:ART"), true));
-
-
- SpanFocusQuery sf = new SpanFocusQuery(rv3, (byte) 1);
+ SpanFocusQuery sf = new SpanFocusQuery(rm, (byte) 1);
+ assertEquals(
+ "focus(1: focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(base:>:child-of), {2: spanTermWithId(base:pos:ART)})), {1: <base:np />})))",
+ sf.toString());
+
kr = ki.search(sf, (short) 10);
assertEquals((long) 2, kr.getTotalResults());
-
assertEquals(2, kr.getMatch(0).getStartPos());
- assertEquals(3, kr.getMatch(0).getEndPos());
+ assertEquals(4, kr.getMatch(0).getEndPos());
assertEquals(5, kr.getMatch(1).getStartPos());
- assertEquals(6, kr.getMatch(1).getEndPos());
+ assertEquals(7, kr.getMatch(1).getEndPos());
}
+ @Test
+ public void testCase10() throws IOException {
+ ki.addDoc(createFieldDoc2());
+ ki.commit();
+ SpanElementQuery seq1 = new SpanElementQuery("base", "np");
+ SpanElementQuery seq2 = new SpanElementQuery("base", "np");
+ SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
+ SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
+
+ SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(
+ new Term("base", ">:child-of")), true);
+ SpanRelationMatchQuery rq = new SpanRelationMatchQuery(srq, scq2, true);
+
+ kr = ki.search(rq, (short) 10);
+
+ assertEquals((long) 4, kr.getTotalResults());
+ assertEquals(0, kr.getMatch(0).getStartPos());
+ assertEquals(7, kr.getMatch(0).getEndPos());
+ assertEquals(1, kr.getMatch(1).getStartPos());
+ assertEquals(7, kr.getMatch(1).getEndPos());
+ assertEquals(2, kr.getMatch(2).getStartPos());
+ assertEquals(7, kr.getMatch(2).getEndPos());
+ assertEquals(4, kr.getMatch(3).getStartPos());
+ assertEquals(7, kr.getMatch(3).getEndPos());
+ }
+
}
\ No newline at end of file
diff --git a/src/test/java/de/ids_mannheim/korap/query/TestSpanReferenceQueryJSON.java b/src/test/java/de/ids_mannheim/korap/query/TestSpanReferenceQueryJSON.java
new file mode 100644
index 0000000..b8a21b8
--- /dev/null
+++ b/src/test/java/de/ids_mannheim/korap/query/TestSpanReferenceQueryJSON.java
@@ -0,0 +1,84 @@
+package de.ids_mannheim.korap.query;
+
+import static de.ids_mannheim.korap.TestSimple.getJSONQuery;
+import static org.junit.Assert.assertEquals;
+
+import java.io.IOException;
+
+import org.apache.lucene.search.spans.SpanQuery;
+import org.junit.Test;
+
+import de.ids_mannheim.korap.query.wrap.SpanQueryWrapper;
+import de.ids_mannheim.korap.util.QueryException;
+
+public class TestSpanReferenceQueryJSON {
+
+ @Test
+ public void testFirstOperandRef() throws IOException, QueryException {
+
+ String filepath = getClass().getResource(
+ "/queries/reference/first-operand-reference.jsonld").getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+
+ // 'cat="V" & cat="NP" & cat="PP" & #2 . #1 & #1 ->dep #3 &
+ // #3 . #2
+ assertEquals(
+ "spanReference(spanNext({3: tokens:p:P}, focus(2: focus(#[1,2]spanSegment(focus(#2: "
+ + "spanSegment(spanRelation(tokens:>:mate/d:HEAD), focus(1: spanNext("
+ + "{2: tokens:p:V}, {1: <tokens:c:NP />})))), {3: tokens:p:P})))), 3)",
+ sq.toString());
+ }
+
+ @Test
+ public void testSecondOperandRef() throws QueryException {
+
+ String filepath = getClass().getResource(
+ "/queries/reference/second-operand-reference.jsonld").getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+
+ // 'cat="V" & cat="NP" & cat="PP" & #2 . #1 & #1 ->dep #3 &
+ // #2 . #3
+ assertEquals(
+ "spanReference(spanNext(focus(2: focus(#[1,2]spanSegment(focus(#2: "
+ + "spanSegment(spanRelation(tokens:>:mate/d:HEAD), focus(1: spanNext("
+ + "{2: tokens:p:V}, {1: <tokens:c:NP />})))), {3: tokens:p:P}))), "
+ + "{3: tokens:p:P}), 3)", sq.toString());
+ }
+
+ @Test
+ public void testMultipleReferences() throws QueryException {
+ String filepath = getClass().getResource(
+ "/queries/reference/multiple-references.jsonld").getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+
+ // 'cat="VP" & cat="NP" & cat="PP" & #1 . #2 & #2 . #3 & #1 .
+ // #3 & #2 ->dep #1'
+ assertEquals(
+ "spanReference(focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(tokens:>:mate/d:HEAD), "
+ + "focus(2: spanReference(spanNext(focus(1: spanNext(focus(2: spanNext({1: <tokens:c:VP />}, "
+ + "{2: <tokens:c:NP />})), {3: <tokens:c:PP />})), {3: <tokens:c:PP />}), 3)))), "
+ + "{1: <tokens:c:VP />})), 1)",
+ sq.toString());
+ }
+
+ @Test
+ public void testDistanceReferences() throws QueryException {
+ String filepath = getClass().getResource(
+ // "/queries/reference/distance-reference.jsonld").getFile();
+ "/queries/reference/bug-multiple-distance.jsonld").getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+
+ // 'cat="VP" & cat="NP" & cat="PP" & #1 . #2 & #2 . #3 & #1 .
+ // #3 & #2 ->dep #1'
+ assertEquals(
+ "spanReference(focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(tokens:>:stanford/d:tag), "
+ + "focus(2: spanDistance(focus(1: spanDistance(<tokens:c:vb />, {1: <tokens:c:prp />}, "
+ + "[(w[1:1], notOrdered, notExcluded)])), {2: <tokens:c:nn />}, [(w[0:2], ordered, notExcluded)])))), "
+ + "{1: <tokens:c:prp />})), 1)",
+ sq.toString());
+ }
+}
diff --git a/src/test/java/de/ids_mannheim/korap/query/TestSpanRelationQueryJSON.java b/src/test/java/de/ids_mannheim/korap/query/TestSpanRelationQueryJSON.java
index 2512d1d..fccd34a 100644
--- a/src/test/java/de/ids_mannheim/korap/query/TestSpanRelationQueryJSON.java
+++ b/src/test/java/de/ids_mannheim/korap/query/TestSpanRelationQueryJSON.java
@@ -19,7 +19,7 @@
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
assertEquals(
- "focus([1,2]spanSegment(tokens:>:mate/d:HEAD, <tokens:c:s />))",
+ "focus(#[1,2]spanSegment(spanRelation(tokens:>:mate/d:HEAD), <tokens:c:s />))",
sq.toString());
}
@@ -32,7 +32,7 @@
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
assertEquals(
- "focus([1,2]spanSegment(tokens:<:mate/d:HEAD, <tokens:c:vp />))",
+ "focus(#[1,2]spanSegment(spanRelation(tokens:<:mate/d:HEAD), <tokens:c:vp />))",
sq.toString());
}
@@ -45,7 +45,7 @@
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
assertEquals(
- "focus([1,2]spanSegment(focus(2: spanSegment(tokens:>:mate/d:HEAD, <tokens:c:s />)), <tokens:c:vp />))",
+ "focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(tokens:>:mate/d:HEAD), <tokens:c:s />)), <tokens:c:vp />))",
sq.toString());
}
@@ -58,7 +58,7 @@
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
assertEquals(
- "focus([1,2]spanSegment(focus(2: spanSegment(tokens:>:mate/d:HEAD, "
+ "focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(tokens:>:mate/d:HEAD), "
+ "spanElementWithAttribute(<tokens:c:s />, spanAttribute(tokens:@root)))), <tokens:c:vp />))",
sq.toString());
}
@@ -72,7 +72,7 @@
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
assertEquals(
- "focus([1,2]spanSegment(focus(2: spanSegment(tokens:>:mate/d:HEAD, "
+ "focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation(tokens:>:mate/d:HEAD), "
+ "spanElementWithAttribute(<tokens:c:s />, spanAttribute(tokens:type:top)))), <tokens:c:vp />))",
sq.toString());
}
@@ -85,7 +85,8 @@
"/queries/relation/relation-only.json").getFile();
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
- assertEquals("focus([1,2]tokens:<:mate/d:HEAD)", sq.toString());
+ assertEquals("focus(#[1,2]spanRelation(tokens:>:mate/d:HEAD))",
+ sq.toString());
}
@Test
@@ -96,18 +97,41 @@
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
assertEquals(
- "focus(1: spanSegment(tokens:<:mate/d:HEAD, <tokens:np />))",
+ "focus(1: focus(#[1,2]spanSegment(spanRelation(tokens:<:mate/d:HEAD), {1: <tokens:c:np />})))",
sq.toString());
}
@Test
- public void testFocusTarget () throws QueryException {
+ public void testFocusTarget() throws QueryException {
String filepath = getClass().getResource(
"/queries/relation/focus-target.json").getFile();
SpanQueryWrapper sqwi = getJSONQuery(filepath);
SpanQuery sq = sqwi.toQuery();
assertEquals(
- "focus(2: spanSegment(tokens:>:mate/d:HEAD, <tokens:s />))",
+ "focus(2: focus(#[1,2]spanSegment(focus(#2: spanSegment(spanRelation("
+ + "tokens:>:mate/d:HEAD), {1: <tokens:c:s />})), {2: <tokens:c:np />})))",
+ sq.toString());
+ }
+
+ @Test
+ public void testFocusEmptyTarget() throws QueryException {
+ String filepath = getClass().getResource(
+ "/queries/relation/focus-empty-target.json").getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+ assertEquals(
+ "focus(2: focus(#[1,2]spanSegment({2: target:spanRelation(tokens:>:mate/d:HEAD)}, {1: <tokens:c:s />})))",
+ sq.toString());
+ }
+
+ @Test
+ public void testFocusEmptyBoth() throws QueryException {
+ String filepath = getClass().getResource(
+ "/queries/relation/focus-empty-both.json").getFile();
+ SpanQueryWrapper sqwi = getJSONQuery(filepath);
+ SpanQuery sq = sqwi.toQuery();
+ assertEquals(
+ "focus(2: focus(#[1,2]{1: source:{2: target:spanRelation(tokens:>:mate/d:HEAD)}}))",
sq.toString());
}
}
diff --git a/src/test/java/de/ids_mannheim/korap/search/TestKrill.java b/src/test/java/de/ids_mannheim/korap/search/TestKrill.java
index 622a80b..973516e 100644
--- a/src/test/java/de/ids_mannheim/korap/search/TestKrill.java
+++ b/src/test/java/de/ids_mannheim/korap/search/TestKrill.java
@@ -1,35 +1,30 @@
package de.ids_mannheim.korap.search;
-import java.util.*;
-import java.io.*;
+import static de.ids_mannheim.korap.TestSimple.getString;
+import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertFalse;
+import static org.junit.Assert.assertNull;
-import static de.ids_mannheim.korap.TestSimple.*;
+import java.io.IOException;
+import java.util.HashMap;
-import de.ids_mannheim.korap.Krill;
-import de.ids_mannheim.korap.KrillMeta;
-import de.ids_mannheim.korap.KrillCollection;
-import de.ids_mannheim.korap.KrillQuery;
-import de.ids_mannheim.korap.KrillIndex;
-import de.ids_mannheim.korap.query.QueryBuilder;
-import de.ids_mannheim.korap.index.FieldDocument;
-import de.ids_mannheim.korap.collection.CollectionBuilder;
-import de.ids_mannheim.korap.response.SearchContext;
-import de.ids_mannheim.korap.response.Result;
-import java.nio.file.Files;
-import java.nio.file.FileSystem;
-import java.nio.file.Path;
-import java.nio.charset.StandardCharsets;
-import java.nio.ByteBuffer;
-
-import com.fasterxml.jackson.databind.ObjectMapper;
-import com.fasterxml.jackson.databind.JsonNode;
-
-import static org.junit.Assert.*;
import org.junit.Test;
-import org.junit.Ignore;
import org.junit.runner.RunWith;
import org.junit.runners.JUnit4;
+import com.fasterxml.jackson.databind.JsonNode;
+import com.fasterxml.jackson.databind.ObjectMapper;
+
+import de.ids_mannheim.korap.Krill;
+import de.ids_mannheim.korap.KrillCollection;
+import de.ids_mannheim.korap.KrillIndex;
+import de.ids_mannheim.korap.KrillMeta;
+import de.ids_mannheim.korap.collection.CollectionBuilder;
+import de.ids_mannheim.korap.index.FieldDocument;
+import de.ids_mannheim.korap.query.QueryBuilder;
+import de.ids_mannheim.korap.response.Result;
+import de.ids_mannheim.korap.response.SearchContext;
+
@RunWith(JUnit4.class)
public class TestKrill {
@Test
@@ -635,6 +630,9 @@
Result kr = ks.apply(ki);
assertEquals(kr.getSerialQuery(),
"focus(1: spanContain(<tokens:base/s:s />, {1: tokens:s:Leben}))");
+ assertEquals(40, kr.getMatch(0).getStartPos());
+ assertEquals(41, kr.getMatch(0).getEndPos());
+
assertEquals(
kr.getMatch(0).getSnippetBrackets(),
"... Initiative\" eine neue politische Gruppierung ins "
diff --git a/src/test/resources/queries/reference/bug-multiple-distance.jsonld b/src/test/resources/queries/reference/bug-multiple-distance.jsonld
new file mode 100644
index 0000000..4d59d1f
--- /dev/null
+++ b/src/test/resources/queries/reference/bug-multiple-distance.jsonld
@@ -0,0 +1,97 @@
+{
+ "query": {
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "vb",
+ "match": "match:eq"
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "prp",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "inOrder": false,
+ "distances": [{
+ "@type": "koral:distance",
+ "key": "w",
+ "boundary": {
+ "@type": "koral:boundary",
+ "min": 1,
+ "max": 1
+ },
+ "min": 1,
+ "max": 1
+ }]
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "nn",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "distances": [{
+ "@type": "koral:distance",
+ "key": "w",
+ "boundary": {
+ "@type": "koral:boundary",
+ "min": 0,
+ "max": 2
+ },
+ "min": 0,
+ "max": 2
+ }],
+ "inOrder": true
+ }]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "stanford",
+ "layer": "d",
+ "key":"tag"
+ }
+ }
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/reference/distance-multiple-references.jsonld b/src/test/resources/queries/reference/distance-multiple-references.jsonld
new file mode 100644
index 0000000..31ef098
--- /dev/null
+++ b/src/test/resources/queries/reference/distance-multiple-references.jsonld
@@ -0,0 +1,124 @@
+{
+ "query": {
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [3],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "key": "vb",
+ "match": "match:eq"
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:span",
+ "key": "prp",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "distances": [{
+ "@type": "koral:distance",
+ "key": "w",
+ "boundary": {
+ "@type": "koral:boundary",
+ "min": 0,
+ "max": 1
+ },
+ "min": 0,
+ "max": 1
+ }],
+ "inOrder": false
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 3,
+ "operands": [{
+ "@type": "koral:span",
+ "key": "nn",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "distances": [{
+ "@type": "koral:distance",
+ "key": "w",
+ "boundary": {
+ "@type": "koral:boundary",
+ "min": 0,
+ "max": 2
+ },
+ "min": 0,
+ "max": 2
+ }],
+ "inOrder": false
+ }]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "stanford",
+ "layer": "d",
+ "key":"tag"
+ }
+ }
+ }]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [3]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "stanford",
+ "layer": "d",
+ "key":"tag"
+ }
+ }
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/reference/distance-reference.jsonld b/src/test/resources/queries/reference/distance-reference.jsonld
new file mode 100644
index 0000000..cf23596
--- /dev/null
+++ b/src/test/resources/queries/reference/distance-reference.jsonld
@@ -0,0 +1,99 @@
+{
+ "query": {
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [3],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "key": "vb",
+ "match": "match:eq"
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:span",
+ "key": "prp",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "distances": [{
+ "@type": "koral:distance",
+ "key": "w",
+ "boundary": {
+ "@type": "koral:boundary",
+ "min": 0,
+ "max": 1
+ },
+ "min": 0,
+ "max": 1
+ }],
+ "inOrder": false
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 3,
+ "operands": [{
+ "@type": "koral:span",
+ "key": "nn",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "distances": [{
+ "@type": "koral:distance",
+ "key": "w",
+ "boundary": {
+ "@type": "koral:boundary",
+ "min": 0,
+ "max": 2
+ },
+ "min": 0,
+ "max": 2
+ }],
+ "inOrder": false
+ }]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "stanford",
+ "layer": "d",
+ "key":"tag"
+ }
+ }
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/reference/first-operand-reference.jsonld b/src/test/resources/queries/reference/first-operand-reference.jsonld
new file mode 100644
index 0000000..44792db
--- /dev/null
+++ b/src/test/resources/queries/reference/first-operand-reference.jsonld
@@ -0,0 +1,86 @@
+{
+ "query": {
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [3]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:token",
+ "wrap": {
+ "@type": "koral:term",
+ "layer": "p",
+ "key": "V",
+ "match": "match:eq"
+ }
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "NP",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "inOrder": true
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 3,
+ "operands": [{
+ "@type": "koral:token",
+ "wrap": {
+ "@type": "koral:term",
+ "layer": "p",
+ "key": "P",
+ "match": "match:eq"
+ }
+ }]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "mate",
+ "layer": "d",
+ "key" : "HEAD"
+ }
+ }
+ }]
+ }
+ ],
+ "inOrder": true
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/reference/multiple-references.jsonld b/src/test/resources/queries/reference/multiple-references.jsonld
new file mode 100644
index 0000000..f94cfb5
--- /dev/null
+++ b/src/test/resources/queries/reference/multiple-references.jsonld
@@ -0,0 +1,97 @@
+{
+ "query": {
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "VP",
+ "match": "match:eq"
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "NP",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "inOrder": true
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 3,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "PP",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "inOrder": true
+ }]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [3]
+ }
+ ],
+ "inOrder": true
+ }]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry" : "mate",
+ "layer": "d",
+ "key" : "HEAD"
+ }
+ }
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/reference/second-operand-reference.jsonld b/src/test/resources/queries/reference/second-operand-reference.jsonld
new file mode 100644
index 0000000..47388cc
--- /dev/null
+++ b/src/test/resources/queries/reference/second-operand-reference.jsonld
@@ -0,0 +1,86 @@
+{
+ "query": {
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [1],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:sequence",
+ "operands": [
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:token",
+ "wrap": {
+ "@type": "koral:term",
+ "layer": "p",
+ "key": "V",
+ "match": "match:eq"
+ }
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "NP",
+ "match": "match:eq"
+ }]
+ }
+ ],
+ "inOrder": true
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 3,
+ "operands": [{
+ "@type": "koral:token",
+ "wrap": {
+ "@type": "koral:term",
+ "layer": "p",
+ "key": "P",
+ "match": "match:eq"
+ }
+ }]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "mate",
+ "layer": "d",
+ "key" : "HEAD"
+ }
+ }
+ }]
+ },
+ {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [3]
+ }
+ ],
+ "inOrder": true
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/relation/focus-empty-both.json b/src/test/resources/queries/relation/focus-empty-both.json
new file mode 100644
index 0000000..1e9fa16
--- /dev/null
+++ b/src/test/resources/queries/relation/focus-empty-both.json
@@ -0,0 +1,39 @@
+{
+ "query": {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:token"
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:token"
+ }]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "mate",
+ "layer": "d",
+ "key": "HEAD"
+ }
+ }
+ }]
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/relation/focus-empty-target.json b/src/test/resources/queries/relation/focus-empty-target.json
new file mode 100644
index 0000000..78ac125
--- /dev/null
+++ b/src/test/resources/queries/relation/focus-empty-target.json
@@ -0,0 +1,42 @@
+{
+ "query": {
+ "@type": "koral:reference",
+ "operation": "operation:focus",
+ "classRef": [2],
+ "operands": [{
+ "@type": "koral:group",
+ "operation": "operation:relation",
+ "operands": [
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "s",
+ "match": "match:eq"
+ }]
+ },
+ {
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:token"
+ }]
+ }
+ ],
+ "relation": {
+ "@type": "koral:relation",
+ "wrap": {
+ "@type": "koral:term",
+ "foundry": "mate",
+ "layer": "d",
+ "key": "HEAD"
+ }
+ }
+ }]
+ },
+ "meta": {}
+}
diff --git a/src/test/resources/queries/relation/focus-source.json b/src/test/resources/queries/relation/focus-source.json
index d934a3b..68d5ce4 100644
--- a/src/test/resources/queries/relation/focus-source.json
+++ b/src/test/resources/queries/relation/focus-source.json
@@ -11,8 +11,15 @@
"@type": "koral:token"
},
{
- "@type": "koral:span",
- "key": "np"
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "np",
+ "match": "match:eq"
+ }]
}
],
"relation": {
diff --git a/src/test/resources/queries/relation/focus-target.json b/src/test/resources/queries/relation/focus-target.json
index daec972..46282a8 100644
--- a/src/test/resources/queries/relation/focus-target.json
+++ b/src/test/resources/queries/relation/focus-target.json
@@ -8,11 +8,26 @@
"operation": "operation:relation",
"operands": [
{
- "@type": "koral:span",
- "key": "s"
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 1,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "s",
+ "match": "match:eq"
+ }]
},
{
- "@type": "koral:token"
+ "@type": "koral:group",
+ "operation": "operation:class",
+ "classOut": 2,
+ "operands": [{
+ "@type": "koral:span",
+ "layer": "c",
+ "key": "np",
+ "match": "match:eq"
+ }]
}
],
"relation": {
