Search in sources :

Example 6 with RdfCloudTripleStoreSelectivityEvaluationStatistics

use of org.apache.rya.rdftriplestore.evaluation.RdfCloudTripleStoreSelectivityEvaluationStatistics in project incubator-rya by apache.

the class QueryJoinSelectOptimizerTest method testOptimizeQ5.

@Test
public void testOptimizeQ5() throws Exception {
    RdfEvalStatsDAO<RdfCloudTripleStoreConfiguration> res = new ProspectorServiceEvalStatsDAO(conn, arc);
    AccumuloSelectivityEvalDAO accc = new AccumuloSelectivityEvalDAO();
    accc.setConf(arc);
    accc.setConnector(conn);
    accc.setRdfEvalDAO(res);
    accc.init();
    BatchWriter bw1 = conn.createBatchWriter("rya_prospects", config);
    BatchWriter bw2 = conn.createBatchWriter("rya_selectivity", config);
    String s1 = "predicateobject" + DELIM + "http://www.w3.org/2000/01/rdf-schema#label" + DELIM + "uri:dog";
    String s2 = "predicateobject" + DELIM + "uri:barksAt" + DELIM + "uri:cat";
    String s3 = "predicateobject" + DELIM + "uri:peesOn" + DELIM + "uri:hydrant";
    String s5 = "predicateobject" + DELIM + "uri:watches" + DELIM + "uri:television";
    String s4 = "predicateobject" + DELIM + "uri:eats" + DELIM + "uri:chickens";
    String s6 = "predicateobject" + DELIM + "uri:eats" + DELIM + "uri:kibble";
    String s7 = "predicateobject" + DELIM + "uri:rollsIn" + DELIM + "uri:mud";
    String s8 = "predicateobject" + DELIM + "uri:runsIn" + DELIM + "uri:field";
    String s9 = "predicateobject" + DELIM + "uri:smells" + DELIM + "uri:butt";
    String s10 = "predicateobject" + DELIM + "uri:eats" + DELIM + "uri:sticks";
    List<Mutation> mList = new ArrayList<Mutation>();
    List<Mutation> mList2 = new ArrayList<Mutation>();
    List<String> sList = Arrays.asList("subjectobject", "subjectpredicate", "subjectsubject", "predicateobject", "predicatepredicate", "predicatesubject");
    Mutation m1, m2, m3, m4, m5, m6, m7, m8, m9, m10, m11;
    m1 = new Mutation(s1 + DELIM + "3");
    m1.put(new Text("count"), new Text(""), new Value("5".getBytes()));
    m2 = new Mutation(s2 + DELIM + "2");
    m2.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m3 = new Mutation(s3 + DELIM + "1");
    m3.put(new Text("count"), new Text(""), new Value("2".getBytes()));
    m4 = new Mutation(s4 + DELIM + "1");
    m4.put(new Text("count"), new Text(""), new Value("0".getBytes()));
    m5 = new Mutation(s5 + DELIM + "1");
    m5.put(new Text("count"), new Text(""), new Value("1".getBytes()));
    m6 = new Mutation(s6 + DELIM + "1");
    m6.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m7 = new Mutation(s7 + DELIM + "1");
    m7.put(new Text("count"), new Text(""), new Value("2".getBytes()));
    m8 = new Mutation(s8 + DELIM + "1");
    m8.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m9 = new Mutation(s9 + DELIM + "1");
    m9.put(new Text("count"), new Text(""), new Value("1".getBytes()));
    m10 = new Mutation(s10 + DELIM + "1");
    m10.put(new Text("count"), new Text(""), new Value("1".getBytes()));
    mList.add(m1);
    mList.add(m2);
    mList.add(m3);
    mList.add(m4);
    mList.add(m5);
    mList.add(m6);
    mList.add(m7);
    mList.add(m8);
    mList.add(m9);
    mList.add(m10);
    bw1.addMutations(mList);
    bw1.close();
    Scanner scan = conn.createScanner("rya_prospects", new Authorizations());
    scan.setRange(new Range());
    for (Map.Entry<Key, Value> entry : scan) {
        System.out.println("Key row string is " + entry.getKey().getRow().toString());
        System.out.println("Key is " + entry.getKey());
        System.out.println("Value is " + (new String(entry.getValue().get())));
    }
    m1 = new Mutation(s1);
    m2 = new Mutation(s2);
    m3 = new Mutation(s3);
    m4 = new Mutation(s4);
    m5 = new Mutation(s5);
    m6 = new Mutation(s6);
    m7 = new Mutation(s7);
    m8 = new Mutation(s8);
    m9 = new Mutation(s9);
    m10 = new Mutation(s10);
    m11 = new Mutation(new Text("subjectpredicateobject" + DELIM + "FullTableCardinality"));
    m11.put(new Text("FullTableCardinality"), new Text("100"), EMPTY_VAL);
    int i = 2;
    int j = 3;
    int k = 4;
    int l = 5;
    Long count1;
    Long count2;
    Long count3;
    Long count4;
    for (String s : sList) {
        count1 = (long) i;
        count2 = (long) j;
        count3 = (long) k;
        count4 = (long) l;
        m1.put(new Text(s), new Text(count4.toString()), EMPTY_VAL);
        m2.put(new Text(s), new Text(count2.toString()), EMPTY_VAL);
        m3.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m4.put(new Text(s), new Text(count3.toString()), EMPTY_VAL);
        m5.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m6.put(new Text(s), new Text(count2.toString()), EMPTY_VAL);
        m7.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m8.put(new Text(s), new Text(count4.toString()), EMPTY_VAL);
        m9.put(new Text(s), new Text(count3.toString()), EMPTY_VAL);
        m10.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        i = 2 * i;
        j = 2 * j;
        k = 2 * k;
        l = 2 * l;
    }
    mList2.add(m1);
    mList2.add(m2);
    mList2.add(m3);
    mList2.add(m5);
    mList2.add(m4);
    mList2.add(m6);
    mList2.add(m7);
    mList2.add(m8);
    mList2.add(m9);
    mList2.add(m10);
    mList2.add(m11);
    bw2.addMutations(mList2);
    bw2.close();
    scan = conn.createScanner("rya_selectivity", new Authorizations());
    scan.setRange(new Range());
    for (Map.Entry<Key, Value> entry : scan) {
        System.out.println("Key row string is " + entry.getKey().getRow().toString());
        System.out.println("Key is " + entry.getKey());
        System.out.println("Value is " + (new String(entry.getKey().getColumnQualifier().toString())));
    }
    TupleExpr te = getTupleExpr(q5);
    System.out.println("Bindings are " + te.getBindingNames());
    RdfCloudTripleStoreSelectivityEvaluationStatistics ars = new RdfCloudTripleStoreSelectivityEvaluationStatistics(arc, res, accc);
    QueryJoinSelectOptimizer qjs = new QueryJoinSelectOptimizer(ars, accc);
    System.out.println("Originial query is " + te);
    qjs.optimize(te, null, null);
    System.out.println("Bindings are " + te.getBindingNames());
    System.out.print("Optimized query is " + te);
}
Also used : Scanner(org.apache.accumulo.core.client.Scanner) Authorizations(org.apache.accumulo.core.security.Authorizations) ProspectorServiceEvalStatsDAO(org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO) ArrayList(java.util.ArrayList) AccumuloSelectivityEvalDAO(org.apache.rya.joinselect.AccumuloSelectivityEvalDAO) QueryJoinSelectOptimizer(org.apache.rya.rdftriplestore.evaluation.QueryJoinSelectOptimizer) Text(org.apache.hadoop.io.Text) RdfCloudTripleStoreSelectivityEvaluationStatistics(org.apache.rya.rdftriplestore.evaluation.RdfCloudTripleStoreSelectivityEvaluationStatistics) Range(org.apache.accumulo.core.data.Range) TupleExpr(org.openrdf.query.algebra.TupleExpr) Value(org.apache.accumulo.core.data.Value) BatchWriter(org.apache.accumulo.core.client.BatchWriter) Mutation(org.apache.accumulo.core.data.Mutation) RdfCloudTripleStoreConfiguration(org.apache.rya.api.RdfCloudTripleStoreConfiguration) Map(java.util.Map) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Example 7 with RdfCloudTripleStoreSelectivityEvaluationStatistics

use of org.apache.rya.rdftriplestore.evaluation.RdfCloudTripleStoreSelectivityEvaluationStatistics in project incubator-rya by apache.

the class QueryJoinSelectOptimizerTest method testOptimizeQ2.

@Test
public void testOptimizeQ2() throws Exception {
    System.out.println("*********************QUERY2********************");
    RdfEvalStatsDAO<RdfCloudTripleStoreConfiguration> res = new ProspectorServiceEvalStatsDAO(conn, arc);
    AccumuloSelectivityEvalDAO accc = new AccumuloSelectivityEvalDAO();
    accc.setConf(arc);
    accc.setConnector(conn);
    accc.setRdfEvalDAO(res);
    accc.init();
    BatchWriter bw1 = conn.createBatchWriter("rya_prospects", config);
    BatchWriter bw2 = conn.createBatchWriter("rya_selectivity", config);
    String s1 = "predicateobject" + DELIM + "http://www.w3.org/2000/01/rdf-schema#label" + DELIM + "uri:dog";
    String s2 = "predicateobject" + DELIM + "uri:barksAt" + DELIM + "uri:cat";
    String s3 = "predicateobject" + DELIM + "uri:peesOn" + DELIM + "uri:hydrant";
    String s5 = "predicateobject" + DELIM + "uri:scratches" + DELIM + "uri:ears";
    String s4 = "predicateobject" + DELIM + "uri:eats" + DELIM + "uri:chickens";
    List<Mutation> mList = new ArrayList<Mutation>();
    List<Mutation> mList2 = new ArrayList<Mutation>();
    List<String> sList = Arrays.asList("subjectobject", "subjectpredicate", "subjectsubject", "predicateobject", "predicatepredicate", "predicatesubject");
    Mutation m1, m2, m3, m4, m5, m6;
    m1 = new Mutation(s1 + DELIM + "3");
    m1.put(new Text("count"), new Text(""), new Value("4".getBytes()));
    m2 = new Mutation(s2 + DELIM + "2");
    m2.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m3 = new Mutation(s3 + DELIM + "1");
    m3.put(new Text("count"), new Text(""), new Value("2".getBytes()));
    m4 = new Mutation(s4 + DELIM + "1");
    m4.put(new Text("count"), new Text(""), new Value("3".getBytes()));
    m5 = new Mutation(s5 + DELIM + "1");
    m5.put(new Text("count"), new Text(""), new Value("5".getBytes()));
    mList.add(m1);
    mList.add(m2);
    mList.add(m3);
    mList.add(m4);
    mList.add(m5);
    bw1.addMutations(mList);
    bw1.close();
    Scanner scan = conn.createScanner("rya_prospects", new Authorizations());
    scan.setRange(new Range());
    for (Map.Entry<Key, Value> entry : scan) {
        System.out.println("Key row string is " + entry.getKey().getRow().toString());
        System.out.println("Key is " + entry.getKey());
        System.out.println("Value is " + (new String(entry.getValue().get())));
    }
    m1 = new Mutation(s1);
    m2 = new Mutation(s2);
    m3 = new Mutation(s3);
    m4 = new Mutation(s4);
    m5 = new Mutation(s5);
    m6 = new Mutation(new Text("subjectpredicateobject" + DELIM + "FullTableCardinality"));
    m6.put(new Text("FullTableCardinality"), new Text("100"), EMPTY_VAL);
    int i = 2;
    int j = 3;
    int k = 4;
    Long count1;
    Long count2;
    Long count3;
    for (String s : sList) {
        count1 = (long) i;
        count2 = (long) j;
        count3 = (long) k;
        m1.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m2.put(new Text(s), new Text(count2.toString()), EMPTY_VAL);
        m3.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        m4.put(new Text(s), new Text(count3.toString()), EMPTY_VAL);
        m5.put(new Text(s), new Text(count1.toString()), EMPTY_VAL);
        i = 2 * i;
        j = 2 * j;
        k = 2 * k;
    }
    mList2.add(m1);
    mList2.add(m2);
    mList2.add(m3);
    mList2.add(m5);
    mList2.add(m4);
    mList2.add(m6);
    bw2.addMutations(mList2);
    bw2.close();
    // scan = conn.createScanner("rya_selectivity" , new Authorizations());
    // scan.setRange(new Range());
    // 
    // for (Map.Entry<Key, Value> entry : scan) {
    // System.out.println("Key row string is " + entry.getKey().getRow().toString());
    // System.out.println("Key is " + entry.getKey());
    // System.out.println("Value is " + (new String(entry.getKey().getColumnQualifier().toString())));
    // 
    // }
    TupleExpr te = getTupleExpr(q2);
    System.out.println("Bindings are " + te.getBindingNames());
    RdfCloudTripleStoreSelectivityEvaluationStatistics ars = new RdfCloudTripleStoreSelectivityEvaluationStatistics(arc, res, accc);
    QueryJoinSelectOptimizer qjs = new QueryJoinSelectOptimizer(ars, accc);
    System.out.println("Originial query is " + te);
    qjs.optimize(te, null, null);
    System.out.println("Optimized query is " + te);
    // System.out.println("Bindings are " + te.getBindingNames());
    Assert.assertTrue(te.equals(getTupleExpr(Q2)));
}
Also used : Scanner(org.apache.accumulo.core.client.Scanner) Authorizations(org.apache.accumulo.core.security.Authorizations) ProspectorServiceEvalStatsDAO(org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO) ArrayList(java.util.ArrayList) AccumuloSelectivityEvalDAO(org.apache.rya.joinselect.AccumuloSelectivityEvalDAO) QueryJoinSelectOptimizer(org.apache.rya.rdftriplestore.evaluation.QueryJoinSelectOptimizer) Text(org.apache.hadoop.io.Text) RdfCloudTripleStoreSelectivityEvaluationStatistics(org.apache.rya.rdftriplestore.evaluation.RdfCloudTripleStoreSelectivityEvaluationStatistics) Range(org.apache.accumulo.core.data.Range) TupleExpr(org.openrdf.query.algebra.TupleExpr) Value(org.apache.accumulo.core.data.Value) BatchWriter(org.apache.accumulo.core.client.BatchWriter) Mutation(org.apache.accumulo.core.data.Mutation) RdfCloudTripleStoreConfiguration(org.apache.rya.api.RdfCloudTripleStoreConfiguration) Map(java.util.Map) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Aggregations

QueryJoinSelectOptimizer (org.apache.rya.rdftriplestore.evaluation.QueryJoinSelectOptimizer)7 RdfCloudTripleStoreSelectivityEvaluationStatistics (org.apache.rya.rdftriplestore.evaluation.RdfCloudTripleStoreSelectivityEvaluationStatistics)7 ArrayList (java.util.ArrayList)6 Map (java.util.Map)6 BatchWriter (org.apache.accumulo.core.client.BatchWriter)6 Scanner (org.apache.accumulo.core.client.Scanner)6 Key (org.apache.accumulo.core.data.Key)6 Mutation (org.apache.accumulo.core.data.Mutation)6 Range (org.apache.accumulo.core.data.Range)6 Value (org.apache.accumulo.core.data.Value)6 Authorizations (org.apache.accumulo.core.security.Authorizations)6 Text (org.apache.hadoop.io.Text)6 RdfCloudTripleStoreConfiguration (org.apache.rya.api.RdfCloudTripleStoreConfiguration)6 AccumuloSelectivityEvalDAO (org.apache.rya.joinselect.AccumuloSelectivityEvalDAO)6 ProspectorServiceEvalStatsDAO (org.apache.rya.prospector.service.ProspectorServiceEvalStatsDAO)6 Test (org.junit.Test)6 TupleExpr (org.openrdf.query.algebra.TupleExpr)6 CloseableIteration (info.aduna.iteration.CloseableIteration)1 NoSuchElementException (java.util.NoSuchElementException)1 Configurable (org.apache.hadoop.conf.Configurable)1