From f933fd33e081745259d12ab2b89d181483f25462 Mon Sep 17 00:00:00 2001
From: miconis <m.debonis991@gmail.com>
Date: Thu, 2 Jul 2020 17:04:17 +0200
Subject: [PATCH] implemented new function for clustering

---
 .../pace/clustering/WordsSuffixPrefix.java    | 42 +++++++++++++++++++
 .../clustering/ClusteringFunctionTest.java    | 13 ++++++
 2 files changed, 55 insertions(+)
 create mode 100644 dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsSuffixPrefix.java

diff --git a/dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsSuffixPrefix.java b/dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsSuffixPrefix.java
new file mode 100644
index 000000000..6086ac0a8
--- /dev/null
+++ b/dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsSuffixPrefix.java
@@ -0,0 +1,42 @@
+package eu.dnetlib.pace.clustering;
+
+import java.util.Collection;
+import java.util.Map;
+import java.util.Set;
+
+import com.google.common.collect.Sets;
+import eu.dnetlib.pace.config.Config;
+
+@ClusteringClass("wordssuffixprefix")
+public class WordsSuffixPrefix extends AbstractClusteringFunction {
+
+    public WordsSuffixPrefix(Map<String, Integer> params) {
+        super(params);
+    }
+
+    @Override
+    protected Collection<String> doApply(Config conf, String s) {
+        return suffixPrefix(s, param("len"), param("max"));
+    }
+
+    private Collection<String> suffixPrefix(String s, int len, int max) {
+
+        final int words = s.split(" ").length;
+        final Set<String> bigrams = Sets.newLinkedHashSet();
+        int i = 0;
+        while (++i < s.length() && bigrams.size() < max) {
+            int j = s.indexOf(" ", i);
+
+            int offset = j + len + 1 < s.length() ? j + len + 1 : s.length();
+
+            if (j - len > 0) {
+                String bigram = s.substring(j - len, offset).replaceAll(" ", "").trim();
+                if (bigram.length() >= 4) {
+                    bigrams.add(words+bigram);
+                }
+            }
+        }
+        return bigrams;
+    }
+
+}
diff --git a/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java b/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java
index 2bccdd5cb..832ba4bce 100644
--- a/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java
+++ b/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java
@@ -98,6 +98,19 @@ public class ClusteringFunctionTest extends AbstractPaceTest {
 		System.out.println(sp.apply(conf, Lists.newArrayList(title(s))));
 	}
 
+	@Test
+	public void testWordsSuffixPrefix() {
+
+		params.put("len", 3);
+		params.put("max", 4);
+
+		final ClusteringFunction sp = new WordsSuffixPrefix(params);
+
+		final String s = "Search for the Standard Model Higgs Boson";
+		System.out.println(s);
+		System.out.println(sp.apply(conf, Lists.newArrayList(title(s))));
+	}
+
 	@Test
 	public void testFieldValue() {