Added 'maxClauses' to boolean search (#338)

Eduardo Alonso · web-flow · commit e92acf92fe25 · 2017-07-04T09:02:34.000+02:00
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,12 +1,13 @@
 # Changelog
 
-## 3.0.14.0 (July 03, 2017)
+## 3.0.14.1 (upcoming)
 
-* Upgrade to Apache Cassandra 3.0.14
+* Add 'max_clauses' to boolean query
 
-## 3.0.13.1 (upcoming)
+## 3.0.14.0 (June 27, 2017)
 
-*Upgrade to Apache Lucene 5.5.4
+* Upgrade to Apache Lucene 5.5.4
+* Upgrade to Apache Cassandra 3.0.14
 
 ## 3.0.13.0 (April 17, 2017)
 
diff --git a/builder/src/main/java/com/stratio/cassandra/lucene/builder/search/condition/BooleanCondition.java b/builder/src/main/java/com/stratio/cassandra/lucene/builder/search/condition/BooleanCondition.java
@@ -39,6 +39,10 @@ public class BooleanCondition extends Condition<BooleanCondition> {
     @JsonProperty("not")
     private List<Condition> not;
 
+    /** The max boolean query clauses. */
+    @JsonProperty("max_clauses")
+    private Integer maxClauses = null;
+
     /**
      * Returns this with the specified mandatory conditions.
      *
@@ -71,4 +75,16 @@ public BooleanCondition not(Condition... conditions) {
         not = add(not, conditions);
         return this;
     }
+
+
+    /**
+     * Returns this builder with the specified max booleqna query clauses
+     *
+     * @param maxClauses teh booleanQuery allowed max clauses
+     * @return this builder with the specified conditions
+     */
+    public BooleanCondition maxClauses(Integer maxClauses) {
+        this.maxClauses = maxClauses;
+        return this;
+    }
 }
diff --git a/builder/src/test/java/com/stratio/cassandra/lucene/builder/BuilderTest.java b/builder/src/test/java/com/stratio/cassandra/lucene/builder/BuilderTest.java
@@ -484,6 +484,7 @@ public void testBooleanConditionFull() {
                               .should(match("f3", 3), match("f4", 4))
                               .not(match("f5", 5), match("f6", 6))
                               .boost(2)
+                              .maxClauses(10)
                               .build();
         String expected = "{\"type\":\"boolean\",\"boost\":2.0," +
                           "\"must\":[" +
@@ -495,7 +496,7 @@ public void testBooleanConditionFull() {
                           "],\"not\":[" +
                           "{\"type\":\"match\",\"field\":\"f5\",\"value\":5}," +
                           "{\"type\":\"match\",\"field\":\"f6\",\"value\":6}" +
-                          "]}";
+                          "],\"max_clauses\":10}";
         assertEquals("boolean is wrong", expected, actual);
     }
 
diff --git a/doc/documentation.rst b/doc/documentation.rst
@@ -2446,6 +2446,7 @@ Searches for rows matching boolean combinations of other searches.
          (, must: [(search,)?] )?
          (, should: [(search,)?] )?
          (, not: [(search,)?] )?
+         (, max_clauses: <max_clauses>)?
        }
     }');
 
@@ -2457,6 +2458,7 @@ where:
    OR … OR search_n
 -  **not**: represents the negation of the disjunction of searches:
    NOT(search_1 OR search_2 OR … OR search_n)
+-  **max_clauses**: this is the clauses limit for this query
 
 
 **Example 1:** search for rows where name ends with “a” AND food starts
diff --git a/plugin/src/main/java/com/stratio/cassandra/lucene/search/condition/BooleanCondition.java b/plugin/src/main/java/com/stratio/cassandra/lucene/search/condition/BooleanCondition.java
@@ -30,14 +30,14 @@
 import static org.apache.lucene.search.BooleanClause.Occur.*;
 
 /**
- * A {@link Condition} that matches documents matching boolean combinations of other queries, e.g. {@link
- * MatchCondition}s, {@link RangeCondition}s or other {@link BooleanCondition}s.
+ * A {@link Condition} that matches documents matching boolean combinations of other queries, e.g. {@link MatchCondition}s, {@link RangeCondition}s or other {@link BooleanCondition}s.
  *
  * @author Andres de la Pena {@literal <adelapena@stratio.com>}
  */
 public class BooleanCondition extends Condition {
 
     protected static final Logger logger = LoggerFactory.getLogger(BooleanCondition.class);
+    static final Integer DEFAULT_MAX_CLAUSES = 1024;
 
     /** The mandatory conditions. */
     public final List<Condition> must;
@@ -48,28 +48,36 @@ public class BooleanCondition extends Condition {
     /** The mandatory not conditions. */
     public final List<Condition> not;
 
+    /** The max boolean query clauses. */
+    final Integer maxClauses;
+
     /**
      * Returns a new {@link BooleanCondition} compound by the specified {@link Condition}s.
      *
-     * @param boost The boost for this query clause. Documents matching this clause will (in addition to the normal
-     * weightings) have their score multiplied by {@code boost}.
+     * @param boost The boost for this query clause. Documents matching this clause will (in addition to the normal weightings) have their score multiplied by {@code boost}.
      * @param must the mandatory conditions
      * @param should the optional conditions
      * @param not the mandatory not conditions
+     * @param maxClauses teh booleanQuery allowed max clauses
      */
     public BooleanCondition(Float boost,
                             List<Condition> must,
                             List<Condition> should,
-                            List<Condition> not) {
+                            List<Condition> not,
+                            Integer maxClauses) {
         super(boost);
         this.must = must == null ? Collections.EMPTY_LIST : must;
         this.should = should == null ? Collections.EMPTY_LIST : should;
         this.not = not == null ? Collections.EMPTY_LIST : not;
+        this.maxClauses = maxClauses == null ? DEFAULT_MAX_CLAUSES : maxClauses;
     }
 
     /** {@inheritDoc} */
     @Override
-    public BooleanQuery doQuery(Schema schema) {
+    public synchronized BooleanQuery doQuery(Schema schema) {
+        int oldMaxClauses= BooleanQuery.getMaxClauseCount();
+        BooleanQuery.setMaxClauseCount(maxClauses);
+
         BooleanQuery.Builder builder = new BooleanQuery.Builder();
         must.forEach(condition -> builder.add(condition.query(schema), MUST));
         should.forEach(condition -> builder.add(condition.query(schema), SHOULD));
@@ -78,7 +86,9 @@ public BooleanQuery doQuery(Schema schema) {
             logger.warn("Performing resource-intensive pure negation query {}", this);
             builder.add(new MatchAllDocsQuery(), FILTER);
         }
-        return builder.build();
+        BooleanQuery out=builder.build();
+        BooleanQuery.setMaxClauseCount(oldMaxClauses);
+        return out;
     }
 
     /** {@inheritDoc} */
diff --git a/plugin/src/main/java/com/stratio/cassandra/lucene/search/condition/builder/BooleanConditionBuilder.java b/plugin/src/main/java/com/stratio/cassandra/lucene/search/condition/builder/BooleanConditionBuilder.java
@@ -43,6 +43,10 @@ public class BooleanConditionBuilder extends ConditionBuilder<BooleanCondition,
     @JsonProperty("not")
     protected final List<ConditionBuilder<?, ?>> not = new LinkedList<>();
 
+    /** The max boolean query clauses. */
+    @JsonProperty("max_clauses")
+    protected Integer maxClauses = null;
+
     /**
      * Returns this builder with the specified mandatory conditions.
      *
@@ -76,6 +80,17 @@ public BooleanConditionBuilder not(ConditionBuilder<?, ?>... builders) {
         return this;
     }
 
+    /**
+     * Returns this builder with the specified max booleqna query clauses
+     *
+     * @param maxClauses teh booleanQuery allowed max clauses
+     * @return this builder with the specified conditions
+     */
+    public BooleanConditionBuilder maxClauses(Integer maxClauses) {
+        this.maxClauses = maxClauses;
+        return this;
+    }
+
     /**
      * Returns the {@link BooleanCondition} represented by this builder.
      *
@@ -86,6 +101,7 @@ public BooleanCondition build() {
         return new BooleanCondition(boost,
                                     must.stream().map(ConditionBuilder::build).collect(toList()),
                                     should.stream().map(ConditionBuilder::build).collect(toList()),
-                                    not.stream().map(ConditionBuilder::build).collect(toList()));
+                                    not.stream().map(ConditionBuilder::build).collect(toList()),
+                                    maxClauses);
     }
 }
diff --git a/plugin/src/test/java/com/stratio/cassandra/lucene/search/condition/BooleanConditionTest.java b/plugin/src/test/java/com/stratio/cassandra/lucene/search/condition/BooleanConditionTest.java
@@ -22,6 +22,7 @@
 
 import static com.stratio.cassandra.lucene.schema.SchemaBuilders.*;
 import static com.stratio.cassandra.lucene.search.SearchBuilders.*;
+import static com.stratio.cassandra.lucene.search.condition.BooleanCondition.DEFAULT_MAX_CLAUSES;
 import static org.junit.Assert.*;
 
 /**
@@ -55,6 +56,7 @@ public void testBuildDefaults() {
         assertEquals("Must is not set", 0, condition.must.size());
         assertEquals("Should is not set", 0, condition.should.size());
         assertEquals("Not is not set", 0, condition.not.size());
+        assertEquals("Max Clauses is not set", DEFAULT_MAX_CLAUSES, condition.maxClauses);
     }
 
     @Test
@@ -118,4 +120,46 @@ public void testToString() {
                      condition.toString());
     }
 
+    @Test
+    public void testOverMaxClauses() {
+        Schema schema = schema().mapper("name1", stringMapper())
+                                .mapper("name2", stringMapper())
+                                .mapper("name3", stringMapper())
+                                .mapper("name4", stringMapper())
+                                .mapper("name5", stringMapper())
+                                .mapper("name6", stringMapper()).build();
+        BooleanCondition condition = bool().maxClauses(5)
+                                           .must(match("name1","value1"))
+                                           .must(match("name2","value2"))
+                                           .must(match("name3","value3"))
+                                           .must(match("name4","value4"))
+                                           .must(match("name5","value5"))
+                                           .must(match("name6","value6")).build();
+
+        try {
+            condition.doQuery(schema);
+        } catch (org.apache.lucene.search.BooleanQuery.TooManyClauses e) {
+            assertTrue(true);
+            return;
+        }
+        fail("Creating a booleanQuery with more clauses than limited should throw an Exception");
+    }
+
+    @Test
+    public void testInMaxClauses() {
+        Schema schema = schema().mapper("name1", stringMapper())
+                                .mapper("name2", stringMapper())
+                                .mapper("name3", stringMapper())
+                                .mapper("name4", stringMapper())
+                                .mapper("name5", stringMapper())
+                                .mapper("name6", stringMapper()).build();
+        BooleanCondition condition = bool().maxClauses(10)
+                                           .must(match("name1","value1"))
+                                           .must(match("name2","value2"))
+                                           .must(match("name3","value3"))
+                                           .must(match("name4","value4"))
+                                           .must(match("name5","value5"))
+                                           .must(match("name6","value6")).build();
+        assertEquals("Query count clauses is wrong", 6, condition.doQuery(schema).clauses().size());
+    }
 }
diff --git a/testsAT/src/test/java/com/stratio/cassandra/lucene/testsAT/varia/TestBooleanExtremeClauses.java b/testsAT/src/test/java/com/stratio/cassandra/lucene/testsAT/varia/TestBooleanExtremeClauses.java
@@ -0,0 +1,61 @@
+package com.stratio.cassandra.lucene.testsAT.varia;
+
+import com.stratio.cassandra.lucene.builder.search.condition.BooleanCondition;
+import com.stratio.cassandra.lucene.testsAT.BaseIT;
+import com.stratio.cassandra.lucene.testsAT.util.CassandraUtils;
+import org.junit.AfterClass;
+import org.junit.BeforeClass;
+import org.junit.Test;
+import org.junit.runner.RunWith;
+import org.junit.runners.JUnit4;
+
+import java.util.LinkedHashMap;
+import java.util.Map;
+
+import static com.stratio.cassandra.lucene.builder.Builder.all;
+import static com.stratio.cassandra.lucene.builder.Builder.bool;
+import static com.stratio.cassandra.lucene.builder.Builder.range;
+
+/**
+ * @author Eduardo Alonso {@literal <eduardoalonso@stratio.com>}
+ */
+@RunWith(JUnit4.class)
+public class TestBooleanExtremeClauses extends BaseIT {
+
+    private static final int NUM_PARTITIONS = 100;
+    private static final int NUM_MAX_CLAUSES = 1000;
+    private static CassandraUtils utils;
+
+    @BeforeClass
+    public static void before() {
+        utils = CassandraUtils.builder("stateless_search_skinny")
+                              .withPartitionKey("pk")
+                              .withColumn("pk", "int")
+                              .withColumn("rc", "int")
+                              .build()
+                              .createKeyspace()
+                              .createTable()
+                              .createIndex();
+        for (Integer i = 0; i < NUM_PARTITIONS; i++) {
+            Map<String, String> data = new LinkedHashMap<>();
+            data.put("pk", i.toString());
+            data.put("rc", i.toString());
+            utils.insert(data);
+        }
+        utils.refresh();
+    }
+
+    @AfterClass
+    public static void after() {
+        CassandraUtils.dropKeyspaceIfNotNull(utils);
+    }
+
+    @Test
+    public void testQuery() throws Exception {
+        BooleanCondition bool= bool().maxClauses(NUM_MAX_CLAUSES + 1);
+        for (int i=0;i< NUM_MAX_CLAUSES ; i++ ) {
+            bool.must(range("rc").lower(0).upper(i).includeLower(true).includeUpper(true));
+        }
+        utils.query(bool).check(1);
+    }
+}
