(search) Soften domain limit constraints in URL deduplication

code/common/model/java/nu/marginalia/model/EdgeDomain.java

@@ -112,14 +112,6 @@ public class EdgeDomain implements Serializable {
         return topDomain;
     }
 
-    public String getDomainKey() {
-        int cutPoint = topDomain.indexOf('.');
-        if (cutPoint < 0) {
-            return topDomain;
-        }
-        return topDomain.substring(0, cutPoint).toLowerCase();
-    }
-
     /** If possible, try to provide an alias domain,
      * i.e. a domain name that is very likely to link to this one
      * */

code/common/model/test/nu/marginalia/model/EdgeDomainTest.java

@@ -8,14 +8,6 @@ import static org.junit.jupiter.api.Assertions.assertEquals;
 
 class EdgeDomainTest {
 
-    @Test
-    public void testSkepdic() throws URISyntaxException {
-        var domain = new EdgeUrl("http://www.skepdic.com/astrology.html");
-        assertEquals("skepdic", domain.getDomain().getDomainKey());
-        var domain2 = new EdgeUrl("http://skepdic.com/astrology.html");
-        assertEquals("skepdic", domain2.getDomain().getDomainKey());
-    }
-
     @Test
     public void testHkDomain() throws URISyntaxException {
         var domain = new EdgeUrl("http://l7072i3.l7c.net");

code/services-application/search-service/java/nu/marginalia/search/SearchOperator.java

@@ -112,13 +112,6 @@ public class SearchOperator {
                 .selectStrategy(queryResponse)
                 .clusterResults(queryResults, 25);
 
-        if (queryParams.humanQuery().equals("slackware linux")) {
-            logger.info("Query response: {}", queryResponse.results().subList(0, 5));
-            logger.info("Query results: {}", queryResults.subList(0, 5));
-            logger.info("Clustered results: {}", clusteredResults.subList(0, 5));
-        }
-
-
         // Log the query and results
 
         logger.info(queryMarker, "Human terms: {}", Strings.join(queryResponse.searchTermsHuman(), ','));

code/services-application/search-service/java/nu/marginalia/search/results/UrlDeduplicator.java

@@ -25,28 +25,14 @@ public class UrlDeduplicator {
     }
 
     public boolean shouldRemove(DecoratedSearchResultItem details) {
-        if (details.url.domain.topDomain.equals("slackware.com")) {
-            if (!deduplicateOnSuperficialHash(details)) {
-                logger.info("Rejecting on superficial hash " + details.url);
-                return true;
-            }
-            if (!deduplicateOnLSH(details)) {
-                logger.info("Rejecting on LSH for " + details.url);
-                return true;
-            }
-            if (!limitResultsPerDomain(details)) {
-                logger.info("Rejecting on limitResultsPerDomain for " + details.url);
-                return true;
-            }
-        }
-        else {
+
         if (!deduplicateOnSuperficialHash(details))
             return true;
         if (!deduplicateOnLSH(details))
             return true;
         if (!limitResultsPerDomain(details))
             return true;
-        }
+
         return false;
     }
 
@@ -76,7 +62,7 @@ public class UrlDeduplicator {
 
     private boolean limitResultsPerDomain(DecoratedSearchResultItem details) {
         final var domain = details.getUrl().getDomain();
-        final String key = domain.getDomainKey();
+        final String key = domain.toString();
 
         return keyCount.adjustOrPutValue(key, 1, 1) <= resultsPerKey;
     }