Merge branch 'master' into term-positions

code/index/test/nu/marginalia/ranking/domains/RankingAlgorithmWithRealDataTest.java

@@ -8,9 +8,14 @@ import java.util.List;
 
 // Test the ranking algorithm with prod data.  Will not run if the data is not available.
 // It's not feasible to include the data in the git repo, as it's ~6 GB of data.
+//
+// The data is available at
+// https://downloads.marginalia.nu/link-test-data.tar.gz
+//
 @Disabled
 class RankingAlgorithmWithRealDataTest {
 
+    /** Test the regular PageRank algorithm. */
     @Test
     public void testRegularPR() {
         if (!TestGraphSourceForLinkData.isAvailable()) {
@@ -26,6 +31,25 @@ class RankingAlgorithmWithRealDataTest {
         }
     }
 
+    /** Test the personalized PageRank algorithm. */
+    @Test
+    public void testRegularPPR() {
+        if (!TestGraphSourceForLinkData.isAvailable()) {
+            return;
+        }
+
+        var graphSource = new TestGraphSourceForLinkData();
+        var results = new PageRankDomainRanker(graphSource,
+                List.of(1476552) // wiby.me
+        )
+                .calculate(10, RankingResultListAccumulator::new);
+
+        for (int i = 0; i < results.size(); i++) {
+            System.out.println(i + " " + graphSource.getName(results.get(i)));
+        }
+    }
+
+    /** Test the inverted link graph pagerank algorithm */
     @Test
     public void testInvertedLinkGraph() {
         if (!TestGraphSourceForInvertedLinkData.isAvailable()) {
@@ -41,6 +65,7 @@ class RankingAlgorithmWithRealDataTest {
         }
     }
 
+    /** Test pagerank with similarity data */
     @Test
     public void testSimilarityPR() {
         if (!TestGraphSourceForSimilarityData.isAvailable()) {
@@ -56,6 +81,7 @@ class RankingAlgorithmWithRealDataTest {
         }
     }
 
+    /** Test personalized pagerank with similarity data */
     @Test
     public void testSimilarityPPR() {
         if (!TestGraphSourceForSimilarityData.isAvailable()) {

code/index/test/nu/marginalia/ranking/domains/TestGraphSourceForInvertedLinkData.java

@@ -16,6 +16,8 @@ import java.util.List;
 import java.util.Map;
 
 public class TestGraphSourceForInvertedLinkData implements GraphSource {
+    // The data is available at
+    // https://downloads.marginalia.nu/link-test-data.tar.gz
     private static Path domainDataPath = Paths.get("/home/vlofgren/Exports/Links/domains.export.tsv");
     private static Path[] linksDataPaths = new Path[] {
             Paths.get("/home/vlofgren/Exports/Links/domain-links-1.dat"),

code/index/test/nu/marginalia/ranking/domains/TestGraphSourceForLinkData.java

@@ -16,6 +16,9 @@ import java.util.List;
 import java.util.Map;
 
 public class TestGraphSourceForLinkData implements GraphSource {
+    // The data is available at
+    // https://downloads.marginalia.nu/link-test-data.tar.gz
+
     private static Path domainDataPath = Paths.get("/home/vlofgren/Exports/Links/domains.export.tsv");
     private static Path[] linksDataPaths = new Path[] {
             Paths.get("/home/vlofgren/Exports/Links/domain-links-1.dat"),

code/index/test/nu/marginalia/ranking/domains/TestGraphSourceForSimilarityData.java

@@ -15,6 +15,9 @@ import java.util.List;
 import java.util.Map;
 
 public class TestGraphSourceForSimilarityData implements GraphSource {
+    // The data is available at
+    // https://downloads.marginalia.nu/link-test-data.tar.gz
+
     private static Path domainDataPath = Paths.get("/home/vlofgren/Exports/Links/domains.export.tsv");
     private static Path similarityDataPath = Paths.get("/home/vlofgren/Exports/Links/neighbors.tsv");