Make match expansion configurable (#150)

Change-Id: Ie4eb9098f6e7352918e1fd0d3bf74615d3508e7e

Changes

@@ -1,3 +1,6 @@
+0.62.6 2024-06-13
+    - [feature] Make match expansion configurable (close #150, margaretha)
+
 0.62.5 2024-06-11
     - [bugfix] cut primary data according to max values (margaretha, #143)
     - [enhancement] restrict match expansion by max token and context 

pom.xml

@@ -35,7 +35,7 @@
 
   <groupId>de.ids-mannheim.korap.krill</groupId>
   <artifactId>Krill</artifactId>
-  <version>0.62.5</version>
+  <version>0.62.6</version>
   <packaging>jar</packaging>
 
   <name>Krill</name>

src/main/java/de/ids_mannheim/korap/response/Match.java

@@ -1117,9 +1117,11 @@ public void expandContextToSpan (String element) {
 
         if (spanContext[0] >= 0
                 && spanContext[0] < spanContext[1]) {
-
-            int maxExpansionSize = KrillProperties.maxTokenMatchSize
-                    + KrillProperties.maxTokenContextSize;
+
+            int maxExpansionSize = KrillProperties.maxTokenMatchSize;
+            if (KrillProperties.matchExpansionIncludeContextSize) {
+                maxExpansionSize += KrillProperties.maxTokenContextSize;
+            }
 
             // Match needs to be cutted!
             boolean cutExpansion = false;

src/main/java/de/ids_mannheim/korap/util/KrillProperties.java

@@ -23,6 +23,8 @@ public class KrillProperties {
     public static int maxTokenContextSize = 60;
     public static int maxCharContextSize = 500;
 
+    public static boolean matchExpansionIncludeContextSize = false;
+
     public static String namedVCPath = "";
     public static boolean isTest = false;
 
@@ -100,6 +102,10 @@ public static void updateConfigurations (Properties  prop) {
         isTest = Boolean.parseBoolean(p);
 
         namedVCPath = prop.getProperty("krill.namedVC", "");
+
+        String matchExpansion = prop.getProperty(
+                "krill.match." + "expansion.includeContextSize", "false");
+        matchExpansionIncludeContextSize = Boolean.parseBoolean(matchExpansion);
     }
 
 

src/test/java/de/ids_mannheim/korap/index/TestMatchIdentifier.java

@@ -30,6 +30,7 @@
 import de.ids_mannheim.korap.response.Result;
 import de.ids_mannheim.korap.response.match.MatchIdentifier;
 import de.ids_mannheim.korap.response.match.PosIdentifier;
+import de.ids_mannheim.korap.util.KrillProperties;
 import de.ids_mannheim.korap.util.QueryException;
 
 @RunWith(JUnit4.class)
@@ -1219,6 +1220,8 @@ public void indexWithFieldInfo () throws IOException, QueryException {
 
     @Test
     public void indexCorolaTokensBugReplicated () throws IOException, QueryException {
+        KrillProperties.matchExpansionIncludeContextSize=false;
+
         KrillIndex ki = new KrillIndex();
 
         ki.addDoc(getClass().getResourceAsStream("/others/corola-bug.json"), false);
@@ -1239,14 +1242,16 @@ public void indexCorolaTokensBugReplicated () throws IOException, QueryException
 
         String str = km.getSnippetBrackets();
         assertTrue(str.contains("[<!>{drukola/l:au:a}"));
-        assertFalse(str.contains("<!>]"));
+        assertTrue(str.contains("<!>]"));
 
         km = ki.getMatchInfo("match-Corola-blog/BlogPost/370281_a_371610-p50-51", "tokens", null, null,false, false, true);
 
         // The match needs to be cutted on both sides!
         str = km.getSnippetBrackets();
         assertTrue(str.contains("[<!>{d"));
         assertTrue(str.contains("a}<!>]"));
+
+        KrillProperties.matchExpansionIncludeContextSize=true;
     };
 
 

src/test/resources/krill.properties

@@ -6,5 +6,6 @@ krill.index.commit.count = 15
 krill.namedVC = queries/collections/named-vcs/
 krill.test = true
 
+krill.match.expansion.includeContextSize = true
 krill.match.max.token=50
 krill.context.max.token=25