Minor changes after Poznan (with a broken test)
diff --git a/src/main/java/de/ids_mannheim/korap/KorapIndex.java b/src/main/java/de/ids_mannheim/korap/KorapIndex.java
index 4ccc4d0..5082be9 100644
--- a/src/main/java/de/ids_mannheim/korap/KorapIndex.java
+++ b/src/main/java/de/ids_mannheim/korap/KorapIndex.java
@@ -61,7 +61,7 @@
/*
Todo: Use FieldCache!
-
+ TODO: Reuse the indexreader everywhere - it should be threadsafe!
http://invertedindex.blogspot.co.il/2009/04/lucene-dociduid-mapping-and-payload.html
see korap/search.java -> retrieveTokens
@@ -704,12 +704,10 @@
kr.setBenchmarkSearchResults(t1, t2);
};
- /*
-TEMPORARY:
+ // Can be disabled TEMPORARILY
while (spans.next()) {
i++;
};
- */
for (KorapMatch km : atomicMatches) {
km.processHighlight(pto);
diff --git a/src/main/java/de/ids_mannheim/korap/KorapQuery.java b/src/main/java/de/ids_mannheim/korap/KorapQuery.java
index 0ee3895..12b34d3 100644
--- a/src/main/java/de/ids_mannheim/korap/KorapQuery.java
+++ b/src/main/java/de/ids_mannheim/korap/KorapQuery.java
@@ -56,7 +56,9 @@
// http://fasterxml.github.io/jackson-databind/javadoc/2.2.0/com/fasterxml/jackson/databind/JsonNode.html
// TODO: Exception messages are horrible!
- // Todo: Use the shortcuts implemented in this class instead of the wrapper constructors
+ // TODO: Use the shortcuts implemented in this class instead of the wrapper constructors
+ // TODO: Check for isArray()
+ // TODO: Check for the number of operands before getting them
public SpanQueryWrapperInterface fromJSON (JsonNode json) throws QueryException {
if (!json.has("@type")) {
@@ -140,6 +142,10 @@
if (!json.has("operands"))
throw new QueryException("SpanSequenceQuery needs operands");
+ JsonNode operands = json.get("operands");
+ if (!operands.isArray() || operands.size() < 2)
+ throw new QueryException("SpanSequenceQuery needs operands");
+
SpanSequenceQueryWrapper sseqqw = new SpanSequenceQueryWrapper(this.field);
for (JsonNode operand : json.get("operands")) {
sseqqw.append(this.fromJSON(operand));
diff --git a/src/test/java/de/ids_mannheim/korap/filter/TestKorapCollection.java b/src/test/java/de/ids_mannheim/korap/filter/TestKorapCollection.java
index 54864aa..8b880f7 100644
--- a/src/test/java/de/ids_mannheim/korap/filter/TestKorapCollection.java
+++ b/src/test/java/de/ids_mannheim/korap/filter/TestKorapCollection.java
@@ -26,7 +26,7 @@
// Indexing test files
for (String i : new String[] {"00001", "00002", "00003", "00004", "00005", "00006", "02439"}) {
FieldDocument fd = ki.addDocFile(
- getClass().getResource("/wiki/" + i + ".json.gz").getFile(), true
+ getClass().getResource("/wiki/" + i + ".json.gz").getFile(), true
);
};
ki.commit();
@@ -78,7 +78,7 @@
// System.err.println(kr.toJSON());
};
- @Ignore
+ @Test
public void filterExampleAtomic () throws IOException {
// That's exactly the same test class, but with multiple atomic indices
diff --git a/src/test/java/de/ids_mannheim/korap/query/TestKorapQueryJSON.java b/src/test/java/de/ids_mannheim/korap/query/TestKorapQueryJSON.java
index c62a71a..f8196a0 100644
--- a/src/test/java/de/ids_mannheim/korap/query/TestKorapQueryJSON.java
+++ b/src/test/java/de/ids_mannheim/korap/query/TestKorapQueryJSON.java
@@ -179,6 +179,14 @@
assertEquals(sqwi.toQuery().toString(), "tokens:base/p:foo");
};
+ @Test
+ public void queryJSONBspClass () {
+ SpanQueryWrapperInterface sqwi = jsonQuery(getClass().getResource("/queries/bsp-class.json").getFile());
+
+ // within(<np>,[base=Mann])
+ assertEquals(sqwi.toQuery().toString(), "{0: spanNext(tokens:tt/p:ADJA, tokens:mate/p:NN)}");
+ };
+
public static String getString (String path) {
StringBuilder contentBuilder = new StringBuilder();
diff --git a/src/test/resources/queries/bsp-class.json b/src/test/resources/queries/bsp-class.json
new file mode 100644
index 0000000..8125e6a
--- /dev/null
+++ b/src/test/resources/queries/bsp-class.json
@@ -0,0 +1 @@
+{"@context":{"korap":"http://korap.ids-mannheim.de/ns/query","@language":"de","operands":{"@id":"korap:operands","@container":"@list"},"relation":{"@id":"korap:relation","@type":"korap:relation#types"},"class":{"@id":"korap:class","@type":"xsd:integer"},"query":"korap:query","filter":"korap:filter","meta":"korap:meta"},"query":{"@type":"korap:group","class":"0","operands":[{"@type":"korap:sequence","operands":[{"@type":"korap:token","@value":{"@type":"korap:term","@value":"tt/p:ADJA","relation":"="}},{"@type":"korap:token","@value":{"@type":"korap:term","@value":"mate/p:NN","relation":"="}}]}]},"meta":[{"@type":"korap:meta-filter","@value":{"@type":"korap:term","@field":"korap:field#corpusID","@value":"WPD"}}],"startPage":1,"count":50,"context":{"left":["token",6],"right":["token",6]}}
diff --git a/src/test/resources/queries/bsp-startswith.json b/src/test/resources/queries/bsp-startswith.json
new file mode 100644
index 0000000..c986a5a
--- /dev/null
+++ b/src/test/resources/queries/bsp-startswith.json
@@ -0,0 +1 @@
+{"@context":{"korap":"http://korap.ids-mannheim.de/ns/query","@language":"de","operands":{"@id":"korap:operands","@container":"@list"},"relation":{"@id":"korap:relation","@type":"korap:relation#types"},"class":{"@id":"korap:class","@type":"xsd:integer"},"query":"korap:query","filter":"korap:filter","meta":"korap:meta"},"query":{"@type":"korap:group","relation":"position","position":"startswith","operands":[{"@type":"korap:element","@value":"s"},{"@type":"korap:sequence","operands":[{"@type":"korap:group","class":"0","operands":[{"@type":"korap:token","@value":{"@type":"korap:term","@value":"tt_p:ADJA","relation":"="}}]},{"@type":"korap:token","@value":{"@type":"korap:term","@value":"mate_p:NN","relation":"="}}]}]},"meta":[{"@type":"korap:meta-filter","@value":{"@type":"korap:term","@field":"korap:field#corpusID","@value":"A00"}},{"@type":"korap:meta-extend","@value":{"@type":"korap:term","@field":"korap:field#corpusID","@value":"A01"}}],"startPage":1,"count":50,"context":{"left":["token",6],"right":["token",6]}}
diff --git a/src/test/resources/queries/readme.txt b/src/test/resources/queries/readme.txt
index 5fa4709..f6f2584 100644
--- a/src/test/resources/queries/readme.txt
+++ b/src/test/resources/queries/readme.txt
@@ -1,4 +1,5 @@
bsp1.json: [base=foo]|([base=foo][base=bar])* meta author=Goethe&year=1815
+bsp1b.json: [base=foo]|([base=foo][base=bar]) meta author=Goethe&year=1815
bsp2.json: ([base=foo]|[base=bar])[base=foobar]
bsp3.json: shrink({[base=Mann]})
bsp4.json: shrink({[base=foo]}[orth=bar])