Fix a bug of filter side in HashJoinOperator

apache · Jun 5, 2018 · 97bef17 · 97bef17
1 parent dfefe62
commit 97bef17
Show file tree

Hide file tree

Showing 4 changed files with 77 additions and 18 deletions.
diff --git a/autofix_cmakelists.py b/autofix_cmakelists.py
@@ -19,6 +19,7 @@
 
 import os
 import re
+import sys
 
 from collections import defaultdict
 
@@ -1356,7 +1357,7 @@ def __init__(self, path):
     self.path = path
 
     if "CMakeLists.txt" in os.listdir(path):
-      with open(self.path + "/CMakeLists.txt", "rb") as file:
+      with open(self.path + "/CMakeLists.txt", "r") as file:
         self.root = CMakeGroup(LineStream(file.readlines()))
         self.newfile = False
     else:
@@ -1524,8 +1525,8 @@ def addMissingSubdirectories(self, global_ctx):
           continue
 
       if name in subdirectories:
-        print "Warning: Possibly duplicated add_subdirectory(" + \
-              name + ") in", self.path
+        sys.stdout.write("Warning: Possibly duplicated add_subdirectory(" +
+                         name + ") in" + self.path)
 
       subdirectories[name] = node
 
@@ -1690,7 +1691,7 @@ def writeToFile(self):
     if len(output) == 0:
       return
 
-    with open(self.path + "/CMakeLists.txt", "wb") as file:
+    with open(self.path + "/CMakeLists.txt", "w") as file:
       file.write("".join(output))
       file.close()
 
@@ -1754,7 +1755,7 @@ def __init__(self, path, name):
     self.includes = []
 
     # Parse include items
-    with open(self.path + "/" + self.name, "rb") as file:
+    with open(self.path + "/" + self.name, "r") as file:
       dependencies = []
 
       for line in file:

diff --git a/query_optimizer/tests/execution_generator/Join.test b/query_optimizer/tests/execution_generator/Join.test
@@ -231,3 +231,62 @@ WHERE b.x > 10
 +-----------+--------------------+------------------------+----------------+
 |          6|                  61|                     601|              C6|
 +-----------+--------------------+------------------------+----------------+
+==
+
+# Semi-join w/o residual predicates
+SELECT SUM(x)
+FROM a
+WHERE w IN (SELECT w FROM b);
+--
++--------------------+
+|SUM(x)              |
++--------------------+
+|                 900|
++--------------------+
+==
+
+# Semi-join w/ residual predicates
+SELECT SUM(x)
+FROM a
+WHERE EXISTS (
+  SELECT *
+  FROM b
+  WHERE a.w = b.w
+    AND a.x < b.x
+);
+--
++--------------------+
+|SUM(x)              |
++--------------------+
+|                 500|
++--------------------+
+==
+
+# Anti-join w/o residual predicates
+SELECT SUM(x)
+FROM a
+WHERE w NOT IN (SELECT w FROM b);
+--
++--------------------+
+|SUM(x)              |
++--------------------+
+|                1000|
++--------------------+
+==
+
+# Anti-join w residual predicates
+SELECT SUM(x)
+FROM a
+WHERE NOT EXISTS (
+  SELECT *
+  FROM b
+  WHERE a.w = b.w
+    AND a.x < b.x
+);
+--
++--------------------+
+|SUM(x)              |
++--------------------+
+|                1400|
++--------------------+
+==
diff --git a/relational_operators/HashJoinOperator.cpp b/relational_operators/HashJoinOperator.cpp
@@ -733,9 +733,9 @@ void HashSemiJoinWorkOrder::executeWithResidualPredicate() {
         build_store.createValueAccessor());
     for (const std::pair<tuple_id, tuple_id> &hash_match
          : build_block_entry.second) {
-      // For each pair, 1st element is a tuple ID from the build relation in the
-      // given build block, 2nd element is a tuple ID from the probe relation.
-      if (filter.get(hash_match.second)) {
+      // For each pair, 1st element is a tuple ID from the probe relation, 2nd
+      // element is a tuple ID from the build relation in the given build block.
+      if (filter.get(hash_match.first)) {
         // We have already found matches for this tuple that belongs to the
         // probe side, skip it.
         continue;
@@ -744,7 +744,7 @@ void HashSemiJoinWorkOrder::executeWithResidualPredicate() {
                                                       hash_match.first,
                                                       *build_accessor,
                                                       hash_match.second)) {
-        filter.set(hash_match.second);
+        filter.set(hash_match.first);
       }
     }
   }
@@ -949,7 +949,7 @@ void HashAntiJoinWorkOrder::executeWithResidualPredicate() {
     std::unique_ptr<ValueAccessor> build_accessor(build_store.createValueAccessor());
     for (const std::pair<tuple_id, tuple_id> &hash_match
          : build_block_entry.second) {
-      if (!existence_map->get(hash_match.second)) {
+      if (!existence_map->get(hash_match.first)) {
         // We have already seen this tuple, skip it.
         continue;
       }
@@ -959,7 +959,7 @@ void HashAntiJoinWorkOrder::executeWithResidualPredicate() {
                                                       hash_match.second)) {
         // Note that the existence map marks a match as false, as needed by the
         // anti join definition.
-        existence_map->set(hash_match.second, false);
+        existence_map->set(hash_match.first, false);
       }
     }
   }

diff --git a/storage/CMakeLists.txt b/storage/CMakeLists.txt
@@ -360,10 +360,6 @@ target_link_libraries(quickstep_storage_BasicColumnStoreValueAccessor
                       quickstep_utility_BitVector
                       quickstep_utility_Macros
                       quickstep_utility_PtrVector)
-if (ENABLE_NETWORK_CLI)
-  target_link_libraries(quickstep_storage_BlockWire_proto
-                        ${PROTOBUF_LIBRARY})
-endif()
 target_link_libraries(quickstep_storage_BloomFilterIndexSubBlock
                       glog
                       quickstep_catalog_CatalogAttribute
@@ -1102,7 +1098,10 @@ if (ENABLE_DISTRIBUTED)
 endif()
 
 if (ENABLE_NETWORK_CLI)
+  target_link_libraries(quickstep_storage_BlockWire_proto
+                        ${PROTOBUF_LIBRARY})
   target_link_libraries(quickstep_storage_DataProviderThread
+                        ${GRPCPLUSPLUS_LIBRARIES}
                         glog
                         quickstep_catalog_CatalogDatabase
                         quickstep_catalog_CatalogRelation
@@ -1118,12 +1117,12 @@ if (ENABLE_NETWORK_CLI)
                         quickstep_threading_Thread
                         quickstep_threading_ThreadUtil
                         quickstep_utility_Macros
-                        tmb
-                        ${GRPCPLUSPLUS_LIBRARIES})
+                        tmb)
   target_link_libraries(quickstep_storage_StorageManager
                         ${GRPCPLUSPLUS_LIBRARIES}
                         quickstep_storage_BlockWire_proto)
 endif()
+
 if (QUICKSTEP_HAVE_FILE_MANAGER_HDFS)
   target_link_libraries(quickstep_storage_FileManagerHdfs
                         ${LIBHDFS3_LIBRARIES}
@@ -1581,9 +1580,9 @@ if (ENABLE_NETWORK_CLI)
                         quickstep_queryexecution_WorkerDirectory
                         quickstep_queryoptimizer_QueryProcessor
                         quickstep_queryoptimizer_tests_TestDatabaseLoader
+                        quickstep_storage_BlockWire_proto
                         quickstep_storage_DataProviderThread
                         quickstep_storage_StorageConstants
-                        quickstep_storage_BlockWire_proto 
                         tmb)
   add_test(DataProviderThread_unittest DataProviderThread_unittest)
   file(MAKE_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/data_provider_thread_test_data)