Adding todo for shortcircuit in PPTSS.

Signed-off-by: jzonthemtn <[email protected]>

opensearch-search-quality-evaluation-plugin/scripts/create-query-set-no-sampling.sh

@@ -1,12 +1,5 @@
 #!/bin/bash -e
 
-#QUERY_SET=`curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/queryset?name=test&description=fake&sampling=pptss" | jq .query_set | tr -d '"'`
-curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/queryset?name=test&description=fake&sampling=none&query_set_size=500"
-
-#echo ${QUERY_SET}
-
-#curl -s http://localhost:9200/search_quality_eval_query_sets/_search | jq
-
-# Run the query set now.
-#curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/run?id=${QUERY_SET}" | jq
+curl -s -X DELETE "http://localhost:9200/search_quality_eval_query_sets"
 
+curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/queryset?name=test&description=fake&sampling=none&query_set_size=10"

opensearch-search-quality-evaluation-plugin/scripts/create-query-set-using-pptss-sampling.sh

@@ -2,16 +2,4 @@
 
 curl -s -X DELETE "http://localhost:9200/search_quality_eval_query_sets"
 
-curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/queryset?name=test&description=fake&sampling=pptss&query_set_size=100"
-
-
-#QUERY_SET=`curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/queryset?name=test&description=fake&sampling=pptss" | jq .query_set | tr -d '"'`
-#curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/queryset?name=test&description=fake&sampling=pptss&query_set_size=100"
-
-#echo ${QUERY_SET}
-
-#curl -s -X GET http://localhost:9200/search_quality_eval_query_sets/_doc/${QUERY_SET} | jq
-
-# Run the query set now.
-#curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/run?id=${QUERY_SET}" | jq
-
+curl -s -X POST "http://localhost:9200/_plugins/search_quality_eval/queryset?name=test&description=fake&sampling=pptss&query_set_size=20"

opensearch-search-quality-evaluation-plugin/src/main/java/org/opensearch/eval/samplers/AllQueriesQuerySampler.java

@@ -56,6 +56,7 @@ public String sample() throws Exception {
         final SearchRequest searchRequest = new SearchRequest(SearchQualityEvaluationPlugin.UBI_QUERIES_INDEX_NAME);
         searchRequest.source(searchSourceBuilder);
 
+        // TODO: Don't use .get()
         final SearchResponse searchResponse = client.search(searchRequest).get();
 
         final Map<String, Long> queries = new HashMap<>();

opensearch-search-quality-evaluation-plugin/src/main/java/org/opensearch/eval/samplers/ProbabilityProportionalToSizeAbstractQuerySampler.java

@@ -83,8 +83,6 @@ public String sample() throws Exception {
 
         while (searchHits != null && searchHits.length > 0) {
 
-            LOGGER.info("search hits size = {}", searchHits.length);
-
             for(final SearchHit hit : searchHits) {
                 final Map<String, Object> fields = hit.getSourceAsMap();
                 userQueries.add(fields.get("user_query").toString());
@@ -137,7 +135,10 @@ public String sample() throws Exception {
         // This may require generating more random numbers than what was requested
         // because removing duplicate user queries will require randomly picking more queries.
         int count = 1;
-        while(querySet.size() < parameters.getQuerySetSize() && count < userQueries.size()) {
+
+        // TODO: How to short-circuit this such that if the same query gets picked over and over, the loop will never end.
+        final int max = 5000;
+        while(querySet.size() < parameters.getQuerySetSize() && count < max) {
 
             // Make a random number not yet used.
             double random;
@@ -160,7 +161,7 @@ public String sample() throws Exception {
             querySet.put(closestQuery, weights.get(closestQuery));
             count++;
 
-            // LOGGER.info("Generated random value: {}; Smallest delta = {}; Closest query = {}", random, smallestDelta, closestQuery);
+            //LOGGER.info("Generated random value: {}; Smallest delta = {}; Closest query = {}", random, smallestDelta, closestQuery);
 
         }