Add result aggregation for query templates (#283)

* Add QueryData class * Add test * Check for update queries * Move responsibility of QueryData to QueryHandler * Remove unused methods * Add tests * Fix authentication * Cleanup * Fix StringListQueryList * Modify QueryHandler and QueryData * Add executable query count and representative query count to QueryHandler * Update the saving template instances * Fix individual template instances results * Add some comments * Update schema * Change default behavior of query templates * Update tests * Fix configuration * Update documentation * Add some comments (to trigger GitHub actions) * Fix minor bug that caused an infinite loop * Change sparql endpoint for testing * Add javadocs * Refactor attribute name * Refactor method name * Add more javadocs * Update src/main/java/org/aksw/iguana/cc/query/QueryData.java Co-authored-by: Alexander Bigerl <[email protected]> * Revert "Update src/main/java/org/aksw/iguana/cc/query/QueryData.java" This reverts commit 8489f6e. * Delegate handling of query templates to an extra class * Trying to clarify comments * Add more comments * Change behavior of noOfQueries property in results * Remove unused import * Fix broken tests * Rename TemplateHandler to QueryTemplateHandler * Add javadoc string --------- Co-authored-by: Alexander Bigerl <[email protected]>
dice-group · Jan 24, 2025 · 20e1bee · 20e1bee
1 parent c1dc6da
commit 20e1bee
Show file tree

Hide file tree

Showing 23 changed files with 555 additions and 194 deletions.
diff --git a/docs/configuration/queries.md b/docs/configuration/queries.md
@@ -130,16 +130,21 @@ The results may look like the following:
 ### Configuration
 The `template` attribute has the following properties:
 
-| property | required | default | description                                                         | example                     |
-|----------|----------|---------|---------------------------------------------------------------------|-----------------------------|
-| endpoint | yes      |         | The endpoint to query.                                              | `http://dbpedia.org/sparql` |
-| limit    | no       | `2000`  | The maximum number of instances per query template.                 | `100`                       |
-| save     | no       | `true`  | If set to `true`, query instances will be saved in a separate file. | `false`                     |
+| property          | required | default | description                                                                                                                                                             | example                     |
+|-------------------|----------|---------|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------|
+| endpoint          | yes      |         | The endpoint to query.                                                                                                                                                  | `http://dbpedia.org/sparql` |
+| limit             | no       | `2000`  | The maximum number of instances per query template.                                                                                                                     | `100`                       |
+| save              | no       | `true`  | If set to `true`, query instances will be saved in a separate file.                                                                                                     | `false`                     |
+| individualResults | no       | `false` | If set to `true`, the results of each individual template instance will be reported, otherwise if set to `false` their results will be subsumed for the query template. | `true`                      |
 
 If the `save` attribute is set to `true`,
 the instances will be saved in a separate file in the same directory as the query templates.
 If the query templates are stored in a folder, the instances will be saved in the parent directory.
 
+If the `individualResults` attribute is set to `false`,
+the results of the instances will be subsumed for the query template.
+The query template will then be considered as an actual query in the results.
+
 Example of query configuration with query templates:
 ```yaml
 queries:
@@ -149,4 +154,5 @@ queries:
     endpoint: "http://dbpedia.org/sparql"
     limit: 100
     save: true
+    individualResults: true
 ```
diff --git a/example-suite.yml b/example-suite.yml
@@ -75,10 +75,11 @@ tasks:
         requestType: post query
         queries:
           path: "./example/query_pattern.txt"
-          pattern:
+          template:
             endpoint: "https://dbpedia.org/sparql"
             limit: 1000
             save: false
+            individualResults: false
         timeout: 180s
         completionTarget:
           duration: 1000s

diff --git a/graalvm/suite.yml b/graalvm/suite.yml
@@ -6,7 +6,7 @@ connections:
   - name: "Blazegraph"
     version: "1.1.1"
     dataset: "DatasetName"
-    endpoint: "http://localhost:9999/blazegraph/sparql"
+    endpoint: "https://query.wikidata.org/"
     authentication:
       user: "user"
       password: "test"
@@ -60,13 +60,14 @@ tasks:
           seed: 123
           lang: "SPARQL"
           template:
-            endpoint: "http://dbpedia.org/sparql"
+            endpoint: "https://dbpedia.org/sparql"
             limit: 1
             save: false
+            individualResults: false
         timeout: 2s
         connection: Blazegraph
         completionTarget:
-          duration: 1s
+          duration: 0.5s
         acceptHeader: "application/sparql-results+json"
         requestType: get query
         parseResults: true
@@ -78,7 +79,7 @@ tasks:
         timeout: 3m
         connection: Blazegraph
         completionTarget:
-          duration: 1s
+          duration: 0.5s
         requestType: get query
         acceptHeader: "application/sparql-results+json"
       - number: 1

diff --git a/schema/iguana-schema.json b/schema/iguana-schema.json
@@ -351,6 +351,9 @@
         },
         "save": {
           "type": "boolean"
+        },
+        "individualResults": {
+          "type": "boolean"
         }
       },
       "required": [

diff --git a/src/main/java/org/aksw/iguana/cc/metrics/impl/AggregatedExecutionStatistics.java b/src/main/java/org/aksw/iguana/cc/metrics/impl/AggregatedExecutionStatistics.java
@@ -29,7 +29,7 @@ public AggregatedExecutionStatistics() {
     public Model createMetricModel(List<HttpWorker> workers, List<HttpWorker.ExecutionStats>[][] data, IRES.Factory iresFactory) {
         Model m = ModelFactory.createDefaultModel();
         for (var worker : workers) {
-            for (int i = 0; i < worker.config().queries().getQueryCount(); i++) {
+            for (int i = 0; i < worker.config().queries().getRepresentedQueryCount(); i++) {
                 Resource queryRes = iresFactory.getWorkerQueryResource(worker, i);
                 m.add(createAggregatedModel(data[(int) worker.getWorkerID()][i], queryRes));
             }

diff --git a/src/main/java/org/aksw/iguana/cc/metrics/impl/EachExecutionStatistic.java b/src/main/java/org/aksw/iguana/cc/metrics/impl/EachExecutionStatistic.java
@@ -26,7 +26,7 @@ public EachExecutionStatistic() {
     public Model createMetricModel(List<HttpWorker> workers, List<HttpWorker.ExecutionStats>[][] data, IRES.Factory iresFactory) {
         Model m = ModelFactory.createDefaultModel();
         for (var worker : workers) {
-            for (int i = 0; i < worker.config().queries().getQueryCount(); i++) {
+            for (int i = 0; i < worker.config().queries().getRepresentedQueryCount(); i++) {
                 Resource workerQueryResource = iresFactory.getWorkerQueryResource(worker, i);
                 Resource queryRes = IRES.getResource(worker.config().queries().getQueryId(i));
                 BigInteger run = BigInteger.ONE;

diff --git a/src/main/java/org/aksw/iguana/cc/metrics/impl/QMPH.java b/src/main/java/org/aksw/iguana/cc/metrics/impl/QMPH.java
@@ -28,7 +28,7 @@ public Number calculateTaskMetric(List<HttpWorker> workers, List<HttpWorker.Exec
     @Override
     public Number calculateWorkerMetric(HttpWorker.Config worker, List<HttpWorker.ExecutionStats>[] data) {
         BigDecimal successes = BigDecimal.ZERO;
-        BigDecimal noq = BigDecimal.valueOf(worker.queries().getQueryCount());
+        BigDecimal noq = BigDecimal.valueOf(worker.queries().getExecutableQueryCount());
         Duration totalTime = Duration.ZERO;
         for (List<HttpWorker.ExecutionStats> datum : data) {
             for (HttpWorker.ExecutionStats exec : datum) {

diff --git a/src/main/java/org/aksw/iguana/cc/query/QueryData.java b/src/main/java/org/aksw/iguana/cc/query/QueryData.java
@@ -2,6 +2,7 @@
 
 import org.apache.jena.update.UpdateFactory;
 
+import java.io.IOException;
 import java.io.InputStream;
 import java.util.ArrayList;
 import java.util.Collection;
@@ -12,9 +13,23 @@
  * At the moment, it only stores if the query is an update query or not.
  *
  * @param queryId The id of the query
- * @param update  If the query is an update query
  */
-public record QueryData(int queryId, boolean update) {
+public record QueryData(int queryId, QueryType type, Integer templateId) {
+    public enum QueryType {
+        DEFAULT,
+        UPDATE,
+        TEMPLATE,
+        TEMPLATE_INSTANCE
+    }
+
+    /**
+     * Generates a list of QueryData objects for a collection of queries.
+     * The method uses the Jena library to check if the query is an update query.
+     * It only checks if the query is an update query or not and sets their index in the order they were given.
+     *
+     * @param queries collection of input streams of queries
+     * @return list of QueryData objects
+     */
     public static List<QueryData> generate(Collection<InputStream> queries) {
         final var queryData = new ArrayList<QueryData>();
         int i = 0;
@@ -25,8 +40,31 @@ public static List<QueryData> generate(Collection<InputStream> queries) {
             } catch (Exception e) {
                 update = false;
             }
-            queryData.add(new QueryData(i++, update));
+            queryData.add(new QueryData(i++, update ? QueryType.UPDATE : QueryType.DEFAULT, null));
+            try {
+                query.close();
+            } catch (IOException ignored) {}
         }
         return queryData;
     }
+
+    /**
+     * Checks if the given query is an update query.
+     * The method uses the Jena library to check if the query is an update query.
+     *
+     * @param query input stream of the query
+     * @return true if the query is an update query, false otherwise
+     */
+    public static boolean checkIfUpdate(InputStream query) {
+        try {
+            UpdateFactory.read(query); // Throws an exception if the query is not an update query
+            return true;
+        } catch (Exception e) {
+            return false;
+        }
+    }
+
+    public boolean update() {
+        return type == QueryType.UPDATE;
+    }
 }