Search in sources :

Example 26 with StreamExpression

use of org.apache.solr.client.solrj.io.stream.expr.StreamExpression in project lucene-solr by apache.

the class StreamExpressionTest method testCommitStream.

////////////////////////////////////////////
@Test
public void testCommitStream() throws Exception {
    CollectionAdminRequest.createCollection("destinationCollection", "conf", 2, 1).process(cluster.getSolrClient());
    AbstractDistribZkTestBase.waitForRecoveriesToFinish("destinationCollection", cluster.getSolrClient().getZkStateReader(), false, true, TIMEOUT);
    new UpdateRequest().add(id, "0", "a_s", "hello0", "a_i", "0", "a_f", "0", "s_multi", "aaaa", "s_multi", "bbbb", "i_multi", "4", "i_multi", "7").add(id, "2", "a_s", "hello2", "a_i", "2", "a_f", "0", "s_multi", "aaaa1", "s_multi", "bbbb1", "i_multi", "44", "i_multi", "77").add(id, "3", "a_s", "hello3", "a_i", "3", "a_f", "3", "s_multi", "aaaa2", "s_multi", "bbbb2", "i_multi", "444", "i_multi", "777").add(id, "4", "a_s", "hello4", "a_i", "4", "a_f", "4", "s_multi", "aaaa3", "s_multi", "bbbb3", "i_multi", "4444", "i_multi", "7777").add(id, "1", "a_s", "hello1", "a_i", "1", "a_f", "1", "s_multi", "aaaa4", "s_multi", "bbbb4", "i_multi", "44444", "i_multi", "77777").commit(cluster.getSolrClient(), "collection1");
    StreamExpression expression;
    TupleStream stream;
    Tuple t;
    StreamContext streamContext = new StreamContext();
    SolrClientCache solrClientCache = new SolrClientCache();
    streamContext.setSolrClientCache(solrClientCache);
    StreamFactory factory = new StreamFactory().withCollectionZkHost("collection1", cluster.getZkServer().getZkAddress()).withCollectionZkHost("destinationCollection", cluster.getZkServer().getZkAddress()).withFunctionName("search", CloudSolrStream.class).withFunctionName("update", UpdateStream.class).withFunctionName("commit", CommitStream.class);
    try {
        //Copy all docs to destinationCollection
        expression = StreamExpressionParser.parse("commit(destinationCollection, batchSize=2, update(destinationCollection, batchSize=5, search(collection1, q=*:*, fl=\"id,a_s,a_i,a_f,s_multi,i_multi\", sort=\"a_f asc, a_i asc\")))");
        stream = factory.constructStream(expression);
        stream.setStreamContext(streamContext);
        List<Tuple> tuples = getTuples(stream);
        //Ensure that all CommitStream tuples indicate the correct number of copied/indexed docs
        assert (tuples.size() == 1);
        t = tuples.get(0);
        assert (t.EOF == false);
        assertEquals(5, t.get("batchIndexed"));
        //Ensure that destinationCollection actually has the new docs.
        expression = StreamExpressionParser.parse("search(destinationCollection, q=*:*, fl=\"id,a_s,a_i,a_f,s_multi,i_multi\", sort=\"a_i asc\")");
        stream = new CloudSolrStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assertEquals(5, tuples.size());
        Tuple tuple = tuples.get(0);
        assert (tuple.getLong("id") == 0);
        assert (tuple.get("a_s").equals("hello0"));
        assert (tuple.getLong("a_i") == 0);
        assert (tuple.getDouble("a_f") == 0.0);
        assertList(tuple.getStrings("s_multi"), "aaaa", "bbbb");
        assertList(tuple.getLongs("i_multi"), Long.parseLong("4"), Long.parseLong("7"));
        tuple = tuples.get(1);
        assert (tuple.getLong("id") == 1);
        assert (tuple.get("a_s").equals("hello1"));
        assert (tuple.getLong("a_i") == 1);
        assert (tuple.getDouble("a_f") == 1.0);
        assertList(tuple.getStrings("s_multi"), "aaaa4", "bbbb4");
        assertList(tuple.getLongs("i_multi"), Long.parseLong("44444"), Long.parseLong("77777"));
        tuple = tuples.get(2);
        assert (tuple.getLong("id") == 2);
        assert (tuple.get("a_s").equals("hello2"));
        assert (tuple.getLong("a_i") == 2);
        assert (tuple.getDouble("a_f") == 0.0);
        assertList(tuple.getStrings("s_multi"), "aaaa1", "bbbb1");
        assertList(tuple.getLongs("i_multi"), Long.parseLong("44"), Long.parseLong("77"));
        tuple = tuples.get(3);
        assert (tuple.getLong("id") == 3);
        assert (tuple.get("a_s").equals("hello3"));
        assert (tuple.getLong("a_i") == 3);
        assert (tuple.getDouble("a_f") == 3.0);
        assertList(tuple.getStrings("s_multi"), "aaaa2", "bbbb2");
        assertList(tuple.getLongs("i_multi"), Long.parseLong("444"), Long.parseLong("777"));
        tuple = tuples.get(4);
        assert (tuple.getLong("id") == 4);
        assert (tuple.get("a_s").equals("hello4"));
        assert (tuple.getLong("a_i") == 4);
        assert (tuple.getDouble("a_f") == 4.0);
        assertList(tuple.getStrings("s_multi"), "aaaa3", "bbbb3");
        assertList(tuple.getLongs("i_multi"), Long.parseLong("4444"), Long.parseLong("7777"));
    } finally {
        CollectionAdminRequest.deleteCollection("destinationCollection").process(cluster.getSolrClient());
        solrClientCache.close();
    }
}
Also used : StreamExpression(org.apache.solr.client.solrj.io.stream.expr.StreamExpression) UpdateRequest(org.apache.solr.client.solrj.request.UpdateRequest) StreamFactory(org.apache.solr.client.solrj.io.stream.expr.StreamFactory) SolrClientCache(org.apache.solr.client.solrj.io.SolrClientCache) Tuple(org.apache.solr.client.solrj.io.Tuple) Test(org.junit.Test)

Example 27 with StreamExpression

use of org.apache.solr.client.solrj.io.stream.expr.StreamExpression in project lucene-solr by apache.

the class StreamExpressionTest method testUniqueStream.

@Test
public void testUniqueStream() throws Exception {
    new UpdateRequest().add(id, "0", "a_s", "hello0", "a_i", "0", "a_f", "0").add(id, "2", "a_s", "hello2", "a_i", "2", "a_f", "0").add(id, "3", "a_s", "hello3", "a_i", "3", "a_f", "3").add(id, "4", "a_s", "hello4", "a_i", "4", "a_f", "4").add(id, "1", "a_s", "hello1", "a_i", "1", "a_f", "1").commit(cluster.getSolrClient(), COLLECTIONORALIAS);
    StreamExpression expression;
    TupleStream stream;
    List<Tuple> tuples;
    StreamContext streamContext = new StreamContext();
    SolrClientCache solrClientCache = new SolrClientCache();
    streamContext.setSolrClientCache(solrClientCache);
    StreamFactory factory = new StreamFactory().withCollectionZkHost(COLLECTIONORALIAS, cluster.getZkServer().getZkAddress()).withFunctionName("search", CloudSolrStream.class).withFunctionName("unique", UniqueStream.class);
    try {
        // Basic test
        expression = StreamExpressionParser.parse("unique(search(" + COLLECTIONORALIAS + ", q=*:*, fl=\"id,a_s,a_i,a_f\", sort=\"a_f asc, a_i asc\"), over=\"a_f\")");
        stream = new UniqueStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 4);
        assertOrder(tuples, 0, 1, 3, 4);
        // Basic test desc
        expression = StreamExpressionParser.parse("unique(search(" + COLLECTIONORALIAS + ", q=*:*, fl=\"id,a_s,a_i,a_f\", sort=\"a_f desc, a_i desc\"), over=\"a_f\")");
        stream = new UniqueStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 4);
        assertOrder(tuples, 4, 3, 1, 2);
        // Basic w/multi comp
        expression = StreamExpressionParser.parse("unique(search(" + COLLECTIONORALIAS + ", q=*:*, fl=\"id,a_s,a_i,a_f\", sort=\"a_f asc, a_i asc\"), over=\"a_f, a_i\")");
        stream = new UniqueStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 5);
        assertOrder(tuples, 0, 2, 1, 3, 4);
        // full factory w/multi comp
        stream = factory.constructStream("unique(search(" + COLLECTIONORALIAS + ", q=*:*, fl=\"id,a_s,a_i,a_f\", sort=\"a_f asc, a_i asc\"), over=\"a_f, a_i\")");
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 5);
        assertOrder(tuples, 0, 2, 1, 3, 4);
    } finally {
        solrClientCache.close();
    }
}
Also used : StreamExpression(org.apache.solr.client.solrj.io.stream.expr.StreamExpression) UpdateRequest(org.apache.solr.client.solrj.request.UpdateRequest) StreamFactory(org.apache.solr.client.solrj.io.stream.expr.StreamFactory) SolrClientCache(org.apache.solr.client.solrj.io.SolrClientCache) Tuple(org.apache.solr.client.solrj.io.Tuple) Test(org.junit.Test)

Example 28 with StreamExpression

use of org.apache.solr.client.solrj.io.stream.expr.StreamExpression in project lucene-solr by apache.

the class StreamExpressionTest method testHashJoinStream.

@Test
public void testHashJoinStream() throws Exception {
    new UpdateRequest().add(id, "1", "side_s", "left", "join1_i", "0", "join2_s", "a", "ident_s", // 8, 9
    "left_1").add(id, "15", "side_s", "left", "join1_i", "0", "join2_s", "a", "ident_s", // 8, 9
    "left_1").add(id, "2", "side_s", "left", "join1_i", "0", "join2_s", "b", "ident_s", "left_2").add(id, "3", "side_s", "left", "join1_i", "1", "join2_s", "a", "ident_s", // 10
    "left_3").add(id, "4", "side_s", "left", "join1_i", "1", "join2_s", "b", "ident_s", // 11
    "left_4").add(id, "5", "side_s", "left", "join1_i", "1", "join2_s", "c", "ident_s", // 12
    "left_5").add(id, "6", "side_s", "left", "join1_i", "2", "join2_s", "d", "ident_s", "left_6").add(id, "7", "side_s", "left", "join1_i", "3", "join2_s", "e", "ident_s", // 14
    "left_7").add(id, "8", "side_s", "right", "join1_i", "0", "join2_s", "a", "ident_s", "right_1", "join3_i", // 1,15
    "0").add(id, "9", "side_s", "right", "join1_i", "0", "join2_s", "a", "ident_s", "right_2", "join3_i", // 1,15
    "0").add(id, "10", "side_s", "right", "join1_i", "1", "join2_s", "a", "ident_s", "right_3", "join3_i", // 3
    "1").add(id, "11", "side_s", "right", "join1_i", "1", "join2_s", "b", "ident_s", "right_4", "join3_i", // 4
    "1").add(id, "12", "side_s", "right", "join1_i", "1", "join2_s", "c", "ident_s", "right_5", "join3_i", // 5
    "1").add(id, "13", "side_s", "right", "join1_i", "2", "join2_s", "dad", "ident_s", "right_6", "join3_i", "2").add(id, "14", "side_s", "right", "join1_i", "3", "join2_s", "e", "ident_s", "right_7", "join3_i", // 7
    "3").commit(cluster.getSolrClient(), COLLECTIONORALIAS);
    StreamExpression expression;
    TupleStream stream;
    List<Tuple> tuples;
    StreamContext streamContext = new StreamContext();
    SolrClientCache solrClientCache = new SolrClientCache();
    streamContext.setSolrClientCache(solrClientCache);
    StreamFactory factory = new StreamFactory().withCollectionZkHost(COLLECTIONORALIAS, cluster.getZkServer().getZkAddress()).withFunctionName("search", CloudSolrStream.class).withFunctionName("hashJoin", HashJoinStream.class);
    try {
        // Basic test
        expression = StreamExpressionParser.parse("hashJoin(" + "search(collection1, q=\"side_s:left\", fl=\"id,join1_i,join2_s,ident_s\", sort=\"join1_i asc, join2_s asc, id asc\")," + "hashed=search(collection1, q=\"side_s:right\", fl=\"join1_i,join2_s,ident_s\", sort=\"join1_i asc, join2_s asc\")," + "on=\"join1_i, join2_s\")");
        stream = new HashJoinStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 8);
        assertOrder(tuples, 1, 1, 15, 15, 3, 4, 5, 7);
        // Basic desc
        expression = StreamExpressionParser.parse("hashJoin(" + "search(collection1, q=\"side_s:left\", fl=\"id,join1_i,join2_s,ident_s\", sort=\"join1_i desc, join2_s asc\")," + "hashed=search(collection1, q=\"side_s:right\", fl=\"join1_i,join2_s,ident_s\", sort=\"join1_i desc, join2_s asc\")," + "on=\"join1_i, join2_s\")");
        stream = new HashJoinStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 8);
        assertOrder(tuples, 7, 3, 4, 5, 1, 1, 15, 15);
        // Results in both searches, no join matches
        expression = StreamExpressionParser.parse("hashJoin(" + "search(collection1, q=\"side_s:left\", fl=\"id,join1_i,join2_s,ident_s\", sort=\"ident_s asc\")," + "hashed=search(collection1, q=\"side_s:right\", fl=\"id,join1_i,join2_s,ident_s\", sort=\"ident_s asc\")," + "on=\"ident_s\")");
        stream = new HashJoinStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 0);
        // Basic test with "on" mapping
        expression = StreamExpressionParser.parse("hashJoin(" + "search(collection1, q=\"side_s:left\", fl=\"id,join1_i,join3_i,ident_s\", sort=\"join1_i asc, join3_i asc, id asc\")," + "hashed=search(collection1, q=\"side_s:right\", fl=\"join1_i,join3_i,ident_s\", sort=\"join1_i asc, join3_i asc\")," + "on=\"join1_i=join3_i\")");
        stream = new HashJoinStream(expression, factory);
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assertEquals(17, tuples.size());
        //Does a lexical sort
        assertOrder(tuples, 1, 1, 15, 15, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 7);
    } finally {
        solrClientCache.close();
    }
}
Also used : StreamExpression(org.apache.solr.client.solrj.io.stream.expr.StreamExpression) UpdateRequest(org.apache.solr.client.solrj.request.UpdateRequest) StreamFactory(org.apache.solr.client.solrj.io.stream.expr.StreamFactory) SolrClientCache(org.apache.solr.client.solrj.io.SolrClientCache) Tuple(org.apache.solr.client.solrj.io.Tuple) Test(org.junit.Test)

Example 29 with StreamExpression

use of org.apache.solr.client.solrj.io.stream.expr.StreamExpression in project lucene-solr by apache.

the class StreamExpressionTest method testSortStream.

@Test
public void testSortStream() throws Exception {
    new UpdateRequest().add(id, "0", "a_s", "hello0", "a_i", "0", "a_f", "0").add(id, "2", "a_s", "hello2", "a_i", "2", "a_f", "0").add(id, "3", "a_s", "hello3", "a_i", "3", "a_f", "3").add(id, "4", "a_s", "hello4", "a_i", "4", "a_f", "4").add(id, "1", "a_s", "hello1", "a_i", "1", "a_f", "1").add(id, "5", "a_s", "hello1", "a_i", "1", "a_f", "2").commit(cluster.getSolrClient(), COLLECTIONORALIAS);
    StreamExpression expression;
    TupleStream stream;
    List<Tuple> tuples;
    StreamContext streamContext = new StreamContext();
    SolrClientCache solrClientCache = new SolrClientCache();
    streamContext.setSolrClientCache(solrClientCache);
    try {
        StreamFactory factory = new StreamFactory().withCollectionZkHost(COLLECTIONORALIAS, cluster.getZkServer().getZkAddress()).withFunctionName("search", CloudSolrStream.class).withFunctionName("sort", SortStream.class);
        // Basic test
        stream = factory.constructStream("sort(search(" + COLLECTIONORALIAS + ", q=*:*, fl=\"id,a_s,a_i,a_f\", sort=\"a_f asc\"), by=\"a_i asc\")");
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 6);
        assertOrder(tuples, 0, 1, 5, 2, 3, 4);
        // Basic test desc
        stream = factory.constructStream("sort(search(" + COLLECTIONORALIAS + ", q=*:*, fl=\"id,a_s,a_i,a_f\", sort=\"a_f asc\"), by=\"a_i desc\")");
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 6);
        assertOrder(tuples, 4, 3, 2, 1, 5, 0);
        // Basic w/multi comp
        stream = factory.constructStream("sort(search(" + COLLECTIONORALIAS + ", q=*:*, fl=\"id,a_s,a_i,a_f\", sort=\"a_f asc\"), by=\"a_i asc, a_f desc\")");
        stream.setStreamContext(streamContext);
        tuples = getTuples(stream);
        assert (tuples.size() == 6);
        assertOrder(tuples, 0, 5, 1, 2, 3, 4);
    } finally {
        solrClientCache.close();
    }
}
Also used : StreamExpression(org.apache.solr.client.solrj.io.stream.expr.StreamExpression) UpdateRequest(org.apache.solr.client.solrj.request.UpdateRequest) StreamFactory(org.apache.solr.client.solrj.io.stream.expr.StreamFactory) SolrClientCache(org.apache.solr.client.solrj.io.SolrClientCache) Tuple(org.apache.solr.client.solrj.io.Tuple) Test(org.junit.Test)

Example 30 with StreamExpression

use of org.apache.solr.client.solrj.io.stream.expr.StreamExpression in project lucene-solr by apache.

the class GatherNodesStream method toExpression.

private StreamExpression toExpression(StreamFactory factory, boolean includeStreams) throws IOException {
    StreamExpression expression = new StreamExpression(factory.getFunctionName(this.getClass()));
    // collection
    expression.addParameter(collection);
    if (includeStreams && !(tupleStream instanceof NodeStream)) {
        if (tupleStream instanceof Expressible) {
            expression.addParameter(((Expressible) tupleStream).toExpression(factory));
        } else {
            throw new IOException("This GatherNodesStream contains a non-expressible TupleStream - it cannot be converted to an expression");
        }
    } else {
        expression.addParameter("<stream>");
    }
    Set<Map.Entry<String, String>> entries = queryParams.entrySet();
    // parameters
    for (Map.Entry param : entries) {
        String value = param.getValue().toString();
        // SOLR-8409: This is a special case where the params contain a " character
        // Do note that in any other BASE streams with parameters where a " might come into play
        // that this same replacement needs to take place.
        value = value.replace("\"", "\\\"");
        expression.addParameter(new StreamExpressionNamedParameter(param.getKey().toString(), value));
    }
    if (metrics != null) {
        for (Metric metric : metrics) {
            expression.addParameter(metric.toExpression(factory));
        }
    }
    expression.addParameter(new StreamExpressionNamedParameter("zkHost", zkHost));
    expression.addParameter(new StreamExpressionNamedParameter("gather", zkHost));
    if (maxDocFreq > -1) {
        expression.addParameter(new StreamExpressionNamedParameter("maxDocFreq", Integer.toString(maxDocFreq)));
    }
    if (tupleStream instanceof NodeStream) {
        NodeStream nodeStream = (NodeStream) tupleStream;
        expression.addParameter(new StreamExpressionNamedParameter("walk", nodeStream.toString() + "->" + traverseTo));
    } else {
        expression.addParameter(new StreamExpressionNamedParameter("walk", traverseFrom + "->" + traverseTo));
    }
    expression.addParameter(new StreamExpressionNamedParameter("trackTraversal", Boolean.toString(trackTraversal)));
    StringBuilder buf = new StringBuilder();
    for (Traversal.Scatter sc : scatter) {
        if (buf.length() > 0) {
            buf.append(",");
        }
        buf.append(sc.toString());
    }
    expression.addParameter(new StreamExpressionNamedParameter("scatter", buf.toString()));
    return expression;
}
Also used : Expressible(org.apache.solr.client.solrj.io.stream.expr.Expressible) StreamExpressionNamedParameter(org.apache.solr.client.solrj.io.stream.expr.StreamExpressionNamedParameter) IOException(java.io.IOException) StreamExpression(org.apache.solr.client.solrj.io.stream.expr.StreamExpression) HashMap(java.util.HashMap) Map(java.util.Map)

Aggregations

StreamExpression (org.apache.solr.client.solrj.io.stream.expr.StreamExpression)82 Test (org.junit.Test)40 StreamExpressionNamedParameter (org.apache.solr.client.solrj.io.stream.expr.StreamExpressionNamedParameter)37 SolrClientCache (org.apache.solr.client.solrj.io.SolrClientCache)36 Tuple (org.apache.solr.client.solrj.io.Tuple)36 StreamFactory (org.apache.solr.client.solrj.io.stream.expr.StreamFactory)36 UpdateRequest (org.apache.solr.client.solrj.request.UpdateRequest)36 IOException (java.io.IOException)22 Expressible (org.apache.solr.client.solrj.io.stream.expr.Expressible)15 ModifiableSolrParams (org.apache.solr.common.params.ModifiableSolrParams)14 HashMap (java.util.HashMap)7 Map (java.util.Map)5 JettySolrRunner (org.apache.solr.client.solrj.embedded.JettySolrRunner)4 FieldComparator (org.apache.solr.client.solrj.io.comp.FieldComparator)4 Metric (org.apache.solr.client.solrj.io.stream.metrics.Metric)4 MeanMetric (org.apache.solr.client.solrj.io.stream.metrics.MeanMetric)3 MinMetric (org.apache.solr.client.solrj.io.stream.metrics.MinMetric)3 ArrayList (java.util.ArrayList)2 List (java.util.List)2 Explanation (org.apache.solr.client.solrj.io.stream.expr.Explanation)2