Search in sources :

Example 1 with Or

use of org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.Or in project hive by apache.

the class TestParquetRecordReaderWrapper method testBuilder.

@Test
public void testBuilder() throws Exception {
    SearchArgument sarg = SearchArgumentFactory.newBuilder().startNot().startOr().isNull("x", PredicateLeaf.Type.LONG).between("y", PredicateLeaf.Type.LONG, 10L, 20L).in("z", PredicateLeaf.Type.LONG, 1L, 2L, 3L).nullSafeEquals("a", PredicateLeaf.Type.STRING, "stinger").end().end().build();
    MessageType schema = MessageTypeParser.parseMessageType("message test {" + " optional int32 x; required int32 y; required int32 z;" + " optional binary a;}");
    FilterPredicate p = ParquetFilterPredicateConverter.toFilterPredicate(sarg, schema);
    String expected = "and(and(and(not(eq(x, null)), not(and(lteq(y, 20), not(lt(y, 10))))), not(or(or(eq(z, 1), " + "eq(z, 2)), eq(z, 3)))), not(eq(a, Binary{\"stinger\"})))";
    assertEquals(expected, p.toString());
}
Also used : SearchArgument(org.apache.hadoop.hive.ql.io.sarg.SearchArgument) FilterPredicate(org.apache.parquet.filter2.predicate.FilterPredicate) MessageType(org.apache.parquet.schema.MessageType) Test(org.junit.Test)

Example 2 with Or

use of org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.Or in project hive by apache.

the class TestConvertAstToSearchArg method testExpression5.

@Test
public void testExpression5() throws Exception {
    /* (first_name < 'owen' or 'foobar' = substr(last_name, 4)) and
    first_name between 'david' and 'greg' */
    String exprStr = "<?xml version=\"1.0\" encoding=\"UTF-8\"?> \n" + "<java version=\"1.6.0_31\" class=\"java.beans.XMLDecoder\"> \n" + " <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "  <void property=\"children\"> \n" + "   <object class=\"java.util.ArrayList\"> \n" + "    <void method=\"add\"> \n" + "     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "      <void property=\"children\"> \n" + "       <object class=\"java.util.ArrayList\"> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "          <void property=\"children\"> \n" + "           <object class=\"java.util.ArrayList\"> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "              <void property=\"column\"> \n" + "               <string>first_name</string> \n" + "              </void> \n" + "              <void property=\"tabAlias\"> \n" + "               <string>orc_people</string> \n" + "              </void> \n" + "              <void property=\"typeInfo\"> \n" + "               <object id=\"PrimitiveTypeInfo0\" class=\"org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo\"> \n" + "                <void property=\"typeName\"> \n" + "                 <string>string</string> \n" + "                </void> \n" + "               </object> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "              <void property=\"typeInfo\"> \n" + "               <object idref=\"PrimitiveTypeInfo0\"/> \n" + "              </void> \n" + "              <void property=\"value\"> \n" + "               <string>owen</string> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "           </object> \n" + "          </void> \n" + "          <void property=\"genericUDF\"> \n" + "           <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "          </void> \n" + "          <void property=\"typeInfo\"> \n" + "           <object id=\"PrimitiveTypeInfo1\" class=\"org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo\"> \n" + "            <void property=\"typeName\"> \n" + "             <string>boolean</string> \n" + "            </void> \n" + "           </object> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "          <void property=\"children\"> \n" + "           <object class=\"java.util.ArrayList\"> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "              <void property=\"typeInfo\"> \n" + "               <object idref=\"PrimitiveTypeInfo0\"/> \n" + "              </void> \n" + "              <void property=\"value\"> \n" + "               <string>foobar</string> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "              <void property=\"children\"> \n" + "               <object class=\"java.util.ArrayList\"> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                  <void property=\"column\"> \n" + "                   <string>last_name</string> \n" + "                  </void> \n" + "                  <void property=\"tabAlias\"> \n" + "                   <string>orc_people</string> \n" + "                  </void> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object class=\"org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo\"> \n" + "                    <void property=\"typeName\"> \n" + "                     <string>int</string> \n" + "                    </void> \n" + "                   </object> \n" + "                  </void> \n" + "                  <void property=\"value\"> \n" + "                   <int>4</int> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "               </object> \n" + "              </void> \n" + "              <void property=\"genericUDF\"> \n" + "               <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFBridge\"> \n" + "                <void property=\"udfClassName\"> \n" + "                 <string>org.apache.hadoop.hive.ql.udf.UDFSubstr</string> \n" + "                </void> \n" + "                <void property=\"udfName\"> \n" + "                 <string>substr</string> \n" + "                </void> \n" + "               </object> \n" + "              </void> \n" + "              <void property=\"typeInfo\"> \n" + "               <object idref=\"PrimitiveTypeInfo0\"/> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "           </object> \n" + "          </void> \n" + "          <void property=\"genericUDF\"> \n" + "           <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPEqual\"/> \n" + "          </void> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo1\"/> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "       </object> \n" + "      </void> \n" + "      <void property=\"genericUDF\"> \n" + "       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPOr\"/> \n" + "      </void> \n" + "      <void property=\"typeInfo\"> \n" + "       <object idref=\"PrimitiveTypeInfo1\"/> \n" + "      </void> \n" + "     </object> \n" + "    </void> \n" + "    <void method=\"add\"> \n" + "     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "      <void property=\"children\"> \n" + "       <object class=\"java.util.ArrayList\"> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo1\"/> \n" + "          </void> \n" + "          <void property=\"value\"> \n" + "           <boolean>false</boolean> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "          <void property=\"column\"> \n" + "           <string>first_name</string> \n" + "          </void> \n" + "          <void property=\"tabAlias\"> \n" + "           <string>orc_people</string> \n" + "          </void> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "          </void> \n" + "          <void property=\"value\"> \n" + "           <string>david</string> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "          </void> \n" + "          <void property=\"value\"> \n" + "           <string>greg</string> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "       </object> \n" + "      </void> \n" + "      <void property=\"genericUDF\"> \n" + "       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFBetween\"/> \n" + "      </void> \n" + "      <void property=\"typeInfo\"> \n" + "       <object idref=\"PrimitiveTypeInfo1\"/> \n" + "      </void> \n" + "     </object> \n" + "    </void> \n" + "   </object> \n" + "  </void> \n" + "  <void property=\"genericUDF\"> \n" + "   <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPAnd\"/> \n" + "  </void> \n" + "  <void property=\"typeInfo\"> \n" + "   <object idref=\"PrimitiveTypeInfo1\"/> \n" + "  </void> \n" + " </object> \n" + "</java> \n";
    SearchArgumentImpl sarg = (SearchArgumentImpl) ConvertAstToSearchArg.create(conf, getFuncDesc(exprStr));
    List<PredicateLeaf> leaves = sarg.getLeaves();
    assertEquals(1, leaves.size());
    MessageType schema = MessageTypeParser.parseMessageType("message test { required int32 id;" + " required binary first_name; }");
    FilterPredicate p = ParquetFilterPredicateConverter.toFilterPredicate(sarg, schema);
    String expected = "and(lteq(first_name, Binary{\"greg\"}), not(lt(first_name, Binary{\"david\"})))";
    assertEquals(p.toString(), expected);
    assertEquals(PredicateLeaf.Type.STRING, leaves.get(0).getType());
    assertEquals(PredicateLeaf.Operator.BETWEEN, leaves.get(0).getOperator());
    assertEquals("first_name", leaves.get(0).getColumnName());
    assertEquals("leaf-0", sarg.getExpression().toString());
    assertNoSharedNodes(sarg.getExpression(), Sets.<ExpressionTree>newIdentityHashSet());
}
Also used : FilterPredicate(org.apache.parquet.filter2.predicate.FilterPredicate) MessageType(org.apache.parquet.schema.MessageType) Test(org.junit.Test)

Example 3 with Or

use of org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.Or in project hive by apache.

the class TestConvertAstToSearchArg method testExpression7.

@Test
public void testExpression7() throws Exception {
    /* (id < 10 and id < 11 and id < 12) or (id < 13 and id < 14 and id < 15) or
       (id < 16 and id < 17) or id < 18 */
    String exprStr = "<?xml version=\"1.0\" encoding=\"UTF-8\"?> \n" + "<java version=\"1.6.0_31\" class=\"java.beans.XMLDecoder\"> \n" + " <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "  <void property=\"children\"> \n" + "   <object class=\"java.util.ArrayList\"> \n" + "    <void method=\"add\"> \n" + "     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "      <void property=\"children\"> \n" + "       <object class=\"java.util.ArrayList\"> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "          <void property=\"children\"> \n" + "           <object class=\"java.util.ArrayList\"> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "              <void property=\"children\"> \n" + "               <object class=\"java.util.ArrayList\"> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                  <void property=\"children\"> \n" + "                   <object class=\"java.util.ArrayList\"> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                      <void property=\"children\"> \n" + "                       <object class=\"java.util.ArrayList\"> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                          <void property=\"column\"> \n" + "                           <string>id</string> \n" + "                          </void> \n" + "                          <void property=\"tabAlias\"> \n" + "                           <string>orc_people</string> \n" + "                          </void> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object id=\"PrimitiveTypeInfo0\" class=\"org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo\"> \n" + "                            <void property=\"typeName\"> \n" + "                             <string>int</string> \n" + "                            </void> \n" + "                           </object> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                          </void> \n" + "                          <void property=\"value\"> \n" + "                           <int>10</int> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                       </object> \n" + "                      </void> \n" + "                      <void property=\"genericUDF\"> \n" + "                       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "                      </void> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object id=\"PrimitiveTypeInfo1\" class=\"org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo\"> \n" + "                        <void property=\"typeName\"> \n" + "                         <string>boolean</string> \n" + "                        </void> \n" + "                       </object> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                      <void property=\"children\"> \n" + "                       <object class=\"java.util.ArrayList\"> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                          <void property=\"column\"> \n" + "                           <string>id</string> \n" + "                          </void> \n" + "                          <void property=\"tabAlias\"> \n" + "                           <string>orc_people</string> \n" + "                          </void> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                          </void> \n" + "                          <void property=\"value\"> \n" + "                           <int>11</int> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                       </object> \n" + "                      </void> \n" + "                      <void property=\"genericUDF\"> \n" + "                       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "                      </void> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object idref=\"PrimitiveTypeInfo1\"/> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                   </object> \n" + "                  </void> \n" + "                  <void property=\"genericUDF\"> \n" + "                   <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPAnd\"/> \n" + "                  </void> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo1\"/> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                  <void property=\"children\"> \n" + "                   <object class=\"java.util.ArrayList\"> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                      <void property=\"column\"> \n" + "                       <string>id</string> \n" + "                      </void> \n" + "                      <void property=\"tabAlias\"> \n" + "                       <string>orc_people</string> \n" + "                      </void> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                      </void> \n" + "                      <void property=\"value\"> \n" + "                       <int>12</int> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                   </object> \n" + "                  </void> \n" + "                  <void property=\"genericUDF\"> \n" + "                   <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "                  </void> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo1\"/> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "               </object> \n" + "              </void> \n" + "              <void property=\"genericUDF\"> \n" + "               <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPAnd\"/> \n" + "              </void> \n" + "              <void property=\"typeInfo\"> \n" + "               <object idref=\"PrimitiveTypeInfo1\"/> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "              <void property=\"children\"> \n" + "               <object class=\"java.util.ArrayList\"> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                  <void property=\"children\"> \n" + "                   <object class=\"java.util.ArrayList\"> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                      <void property=\"children\"> \n" + "                       <object class=\"java.util.ArrayList\"> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                          <void property=\"column\"> \n" + "                           <string>id</string> \n" + "                          </void> \n" + "                          <void property=\"tabAlias\"> \n" + "                           <string>orc_people</string> \n" + "                          </void> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                          </void> \n" + "                          <void property=\"value\"> \n" + "                           <int>13</int> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                       </object> \n" + "                      </void> \n" + "                      <void property=\"genericUDF\"> \n" + "                       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "                      </void> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object idref=\"PrimitiveTypeInfo1\"/> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                      <void property=\"children\"> \n" + "                       <object class=\"java.util.ArrayList\"> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                          <void property=\"column\"> \n" + "                           <string>id</string> \n" + "                          </void> \n" + "                          <void property=\"tabAlias\"> \n" + "                           <string>orc_people</string> \n" + "                          </void> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                        <void method=\"add\"> \n" + "                         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                          <void property=\"typeInfo\"> \n" + "                           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                          </void> \n" + "                          <void property=\"value\"> \n" + "                           <int>14</int> \n" + "                          </void> \n" + "                         </object> \n" + "                        </void> \n" + "                       </object> \n" + "                      </void> \n" + "                      <void property=\"genericUDF\"> \n" + "                       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "                      </void> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object idref=\"PrimitiveTypeInfo1\"/> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                   </object> \n" + "                  </void> \n" + "                  <void property=\"genericUDF\"> \n" + "                   <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPAnd\"/> \n" + "                  </void> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo1\"/> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "                  <void property=\"children\"> \n" + "                   <object class=\"java.util.ArrayList\"> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                      <void property=\"column\"> \n" + "                       <string>id</string> \n" + "                      </void> \n" + "                      <void property=\"tabAlias\"> \n" + "                       <string>orc_people</string> \n" + "                      </void> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                    <void method=\"add\"> \n" + "                     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                      <void property=\"typeInfo\"> \n" + "                       <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                      </void> \n" + "                      <void property=\"value\"> \n" + "                       <int>15</int> \n" + "                      </void> \n" + "                     </object> \n" + "                    </void> \n" + "                   </object> \n" + "                  </void> \n" + "                  <void property=\"genericUDF\"> \n" + "                   <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "                  </void> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo1\"/> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "               </object> \n" + "              </void> \n" + "              <void property=\"genericUDF\"> \n" + "               <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPAnd\"/> \n" + "              </void> \n" + "              <void property=\"typeInfo\"> \n" + "               <object idref=\"PrimitiveTypeInfo1\"/> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "           </object> \n" + "          </void> \n" + "          <void property=\"genericUDF\"> \n" + "           <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPOr\"/> \n" + "          </void> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo1\"/> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "          <void property=\"children\"> \n" + "           <object class=\"java.util.ArrayList\"> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "              <void property=\"children\"> \n" + "               <object class=\"java.util.ArrayList\"> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                  <void property=\"column\"> \n" + "                   <string>id</string> \n" + "                  </void> \n" + "                  <void property=\"tabAlias\"> \n" + "                   <string>orc_people</string> \n" + "                  </void> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                  </void> \n" + "                  <void property=\"value\"> \n" + "                   <int>16</int> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "               </object> \n" + "              </void> \n" + "              <void property=\"genericUDF\"> \n" + "               <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "              </void> \n" + "              <void property=\"typeInfo\"> \n" + "               <object idref=\"PrimitiveTypeInfo1\"/> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "            <void method=\"add\"> \n" + "             <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "              <void property=\"children\"> \n" + "               <object class=\"java.util.ArrayList\"> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "                  <void property=\"column\"> \n" + "                   <string>id</string> \n" + "                  </void> \n" + "                  <void property=\"tabAlias\"> \n" + "                   <string>orc_people</string> \n" + "                  </void> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "                <void method=\"add\"> \n" + "                 <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "                  <void property=\"typeInfo\"> \n" + "                   <object idref=\"PrimitiveTypeInfo0\"/> \n" + "                  </void> \n" + "                  <void property=\"value\"> \n" + "                   <int>17</int> \n" + "                  </void> \n" + "                 </object> \n" + "                </void> \n" + "               </object> \n" + "              </void> \n" + "              <void property=\"genericUDF\"> \n" + "               <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "              </void> \n" + "              <void property=\"typeInfo\"> \n" + "               <object idref=\"PrimitiveTypeInfo1\"/> \n" + "              </void> \n" + "             </object> \n" + "            </void> \n" + "           </object> \n" + "          </void> \n" + "          <void property=\"genericUDF\"> \n" + "           <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPAnd\"/> \n" + "          </void> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo1\"/> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "       </object> \n" + "      </void> \n" + "      <void property=\"genericUDF\"> \n" + "       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPOr\"/> \n" + "      </void> \n" + "      <void property=\"typeInfo\"> \n" + "       <object idref=\"PrimitiveTypeInfo1\"/> \n" + "      </void> \n" + "     </object> \n" + "    </void> \n" + "    <void method=\"add\"> \n" + "     <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc\"> \n" + "      <void property=\"children\"> \n" + "       <object class=\"java.util.ArrayList\"> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc\"> \n" + "          <void property=\"column\"> \n" + "           <string>id</string> \n" + "          </void> \n" + "          <void property=\"tabAlias\"> \n" + "           <string>orc_people</string> \n" + "          </void> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "        <void method=\"add\"> \n" + "         <object class=\"org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc\"> \n" + "          <void property=\"typeInfo\"> \n" + "           <object idref=\"PrimitiveTypeInfo0\"/> \n" + "          </void> \n" + "          <void property=\"value\"> \n" + "           <int>18</int> \n" + "          </void> \n" + "         </object> \n" + "        </void> \n" + "       </object> \n" + "      </void> \n" + "      <void property=\"genericUDF\"> \n" + "       <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan\"/> \n" + "      </void> \n" + "      <void property=\"typeInfo\"> \n" + "       <object idref=\"PrimitiveTypeInfo1\"/> \n" + "      </void> \n" + "     </object> \n" + "    </void> \n" + "   </object> \n" + "  </void> \n" + "  <void property=\"genericUDF\"> \n" + "   <object class=\"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPOr\"/> \n" + "  </void> \n" + "  <void property=\"typeInfo\"> \n" + "   <object idref=\"PrimitiveTypeInfo1\"/> \n" + "  </void> \n" + " </object>\n" + "</java>";
    SearchArgumentImpl sarg = (SearchArgumentImpl) ConvertAstToSearchArg.create(conf, getFuncDesc(exprStr));
    List<PredicateLeaf> leaves = sarg.getLeaves();
    assertEquals(9, leaves.size());
    MessageType schema = MessageTypeParser.parseMessageType("message test { required int32 id;" + " required binary first_name; }");
    FilterPredicate p = ParquetFilterPredicateConverter.toFilterPredicate(sarg, schema);
    String expected = "and(and(and(and(and(and(and(and(and(and(and(and(and(and(and(and(and(" + "or(or(or(lt(id, 18), lt(id, 10)), lt(id, 13)), lt(id, 16)), " + "or(or(or(lt(id, 18), lt(id, 11)), lt(id, 13)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 12)), lt(id, 13)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 10)), lt(id, 14)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 11)), lt(id, 14)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 12)), lt(id, 14)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 10)), lt(id, 15)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 11)), lt(id, 15)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 12)), lt(id, 15)), lt(id, 16))), " + "or(or(or(lt(id, 18), lt(id, 10)), lt(id, 13)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 11)), lt(id, 13)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 12)), lt(id, 13)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 10)), lt(id, 14)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 11)), lt(id, 14)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 12)), lt(id, 14)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 10)), lt(id, 15)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 11)), lt(id, 15)), lt(id, 17))), " + "or(or(or(lt(id, 18), lt(id, 12)), lt(id, 15)), lt(id, 17)))";
    assertEquals(p.toString(), expected);
    PredicateLeaf leaf = leaves.get(0);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(18L, leaf.getLiteral());
    leaf = leaves.get(1);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(10L, leaf.getLiteral());
    leaf = leaves.get(2);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(13L, leaf.getLiteral());
    leaf = leaves.get(3);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(16L, leaf.getLiteral());
    leaf = leaves.get(4);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(11L, leaf.getLiteral());
    leaf = leaves.get(5);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(12L, leaf.getLiteral());
    leaf = leaves.get(6);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(14L, leaf.getLiteral());
    leaf = leaves.get(7);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(15L, leaf.getLiteral());
    leaf = leaves.get(8);
    assertEquals(PredicateLeaf.Type.LONG, leaf.getType());
    assertEquals(PredicateLeaf.Operator.LESS_THAN, leaf.getOperator());
    assertEquals("id", leaf.getColumnName());
    assertEquals(17L, leaf.getLiteral());
    assertEquals("(and" + " (or leaf-0 leaf-1 leaf-2 leaf-3)" + " (or leaf-0 leaf-4 leaf-2 leaf-3)" + " (or leaf-0 leaf-5 leaf-2 leaf-3)" + " (or leaf-0 leaf-1 leaf-6 leaf-3)" + " (or leaf-0 leaf-4 leaf-6 leaf-3)" + " (or leaf-0 leaf-5 leaf-6 leaf-3)" + " (or leaf-0 leaf-1 leaf-7 leaf-3)" + " (or leaf-0 leaf-4 leaf-7 leaf-3)" + " (or leaf-0 leaf-5 leaf-7 leaf-3)" + " (or leaf-0 leaf-1 leaf-2 leaf-8)" + " (or leaf-0 leaf-4 leaf-2 leaf-8)" + " (or leaf-0 leaf-5 leaf-2 leaf-8)" + " (or leaf-0 leaf-1 leaf-6 leaf-8)" + " (or leaf-0 leaf-4 leaf-6 leaf-8)" + " (or leaf-0 leaf-5 leaf-6 leaf-8)" + " (or leaf-0 leaf-1 leaf-7 leaf-8)" + " (or leaf-0 leaf-4 leaf-7 leaf-8)" + " (or leaf-0 leaf-5 leaf-7 leaf-8))", sarg.getExpression().toString());
}
Also used : FilterPredicate(org.apache.parquet.filter2.predicate.FilterPredicate) MessageType(org.apache.parquet.schema.MessageType) Test(org.junit.Test)

Example 4 with Or

use of org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.Or in project hive by apache.

the class TestParquetFilterPredicate method testFilterColumnsThatDoNoExistOnSchemaHighOrder2.

@Test
public void testFilterColumnsThatDoNoExistOnSchemaHighOrder2() {
    MessageType schema = MessageTypeParser.parseMessageType("message test { required int32 a; required int32 b; }");
    SearchArgument sarg = SearchArgumentFactory.newBuilder().startAnd().startOr().equals("a", PredicateLeaf.Type.LONG, 1L).equals("b", PredicateLeaf.Type.LONG, 1L).end().startOr().equals("a", PredicateLeaf.Type.LONG, 999L).equals("none", PredicateLeaf.Type.LONG, 999L).end().end().build();
    FilterPredicate p = ParquetFilterPredicateConverter.toFilterPredicate(sarg, schema);
    String expected = "or(eq(a, 1), eq(b, 1))";
    assertEquals(expected, p.toString());
}
Also used : SearchArgument(org.apache.hadoop.hive.ql.io.sarg.SearchArgument) FilterPredicate(org.apache.parquet.filter2.predicate.FilterPredicate) MessageType(org.apache.parquet.schema.MessageType) Test(org.junit.Test)

Example 5 with Or

use of org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.Or in project hive by apache.

the class TestParquetFilterPredicate method testFilterColumnsThatDoNoExistOnSchemaHighOrder1.

@Test
public void testFilterColumnsThatDoNoExistOnSchemaHighOrder1() {
    MessageType schema = MessageTypeParser.parseMessageType("message test { required int32 a; required int32 b; }");
    SearchArgument sarg = SearchArgumentFactory.newBuilder().startOr().startAnd().equals("a", PredicateLeaf.Type.LONG, 1L).equals("none", PredicateLeaf.Type.LONG, 1L).end().startAnd().equals("a", PredicateLeaf.Type.LONG, 999L).equals("none", PredicateLeaf.Type.LONG, 999L).end().end().build();
    FilterPredicate p = ParquetFilterPredicateConverter.toFilterPredicate(sarg, schema);
    String expected = "or(eq(a, 1), eq(a, 999))";
    assertEquals(expected, p.toString());
}
Also used : SearchArgument(org.apache.hadoop.hive.ql.io.sarg.SearchArgument) FilterPredicate(org.apache.parquet.filter2.predicate.FilterPredicate) MessageType(org.apache.parquet.schema.MessageType) Test(org.junit.Test)

Aggregations

FilterPredicate (org.apache.parquet.filter2.predicate.FilterPredicate)17 Test (org.junit.Test)17 MessageType (org.apache.parquet.schema.MessageType)9 SearchArgument (org.apache.hadoop.hive.ql.io.sarg.SearchArgument)4 BinaryColumn (org.apache.parquet.filter2.predicate.Operators.BinaryColumn)3 Configuration (org.apache.hadoop.conf.Configuration)2 IntColumn (org.apache.parquet.filter2.predicate.Operators.IntColumn)2 And (org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.And)2 Or (org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.Or)2 ValueInspector (org.apache.parquet.filter2.recordlevel.IncrementallyUpdatedFilterPredicate.ValueInspector)2 ByteArrayInputStream (java.io.ByteArrayInputStream)1 ByteArrayOutputStream (java.io.ByteArrayOutputStream)1 ObjectInputStream (java.io.ObjectInputStream)1 ObjectOutputStream (java.io.ObjectOutputStream)1 ArrayList (java.util.ArrayList)1 Path (org.apache.hadoop.fs.Path)1 DataWritableReadSupport (org.apache.hadoop.hive.ql.io.parquet.read.DataWritableReadSupport)1 FileSplit (org.apache.hadoop.mapred.FileSplit)1 Job (org.apache.hadoop.mapreduce.Job)1 Group (org.apache.parquet.example.data.Group)1