(search) Add debug logging for specific query

This reverts commit 39ce40bfeb.

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

@@ -112,6 +112,13 @@ 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/SearchResultClusterer.java

@@ -23,7 +23,7 @@ public class SearchResultClusterer {
     }
 
     /** No clustering, just return the results as is */
-    private static List<ClusteredUrlDetails> noOp(List<UrlDetails> results, int total) {
+    public static List<ClusteredUrlDetails> noOp(List<UrlDetails> results, int total) {
         if (results.isEmpty())
             return List.of();
 

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

@@ -85,7 +85,6 @@ public class SearchService extends JoobyService {
         String emptySvg = "<svg xmlns=\"http://www.w3.org/2000/svg\"></svg>";
         jooby.get("/site/{domain}/favicon", ctx -> {
             String domain = ctx.path("domain").value();
-            logger.info("Finding icon for domain {}", domain);
             try {
                 DbDomainQueries.DomainIdWithNode domainIdWithNode = domainQueries.getDomainIdWithNode(new EdgeDomain(domain));
                 var faviconMaybe = faviconClient.getFavicon(domain, domainIdWithNode.nodeAffinity());

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

@@ -25,13 +25,28 @@ public class UrlDeduplicator {
     }
 
     public boolean shouldRemove(DecoratedSearchResultItem details) {
-//        if (!deduplicateOnSuperficialHash(details))
-//            return true;
-//        if (!deduplicateOnLSH(details))
-//            return true;
-        if (!limitResultsPerDomain(details))
-            return true;
-
+        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;
     }