[DYOD] UCC Discovery Plugin (#2485)

The Unique Column Combination Discovery Plugin reads the LQP cache, determines candidate columns that can be used by optimization rules, and validates whether these candidate columns are unique.

.gitignore

@@ -5,7 +5,7 @@ build/*
 build-*/*
 *.make
 Makefile*
-coverage
+coverage*
 .vscode
 benchmark.json
 *.sublime-*

README.md

@@ -159,6 +159,7 @@ Contact: firstname.lastname@hpi.de
 ## Contributors
 -   Yannick   Bäumer
 -   Lawrence  Benson
+-   Lukas     Budach
 -   Timo      Djürken
 -   Alexander Dubrawski
 -   Fabian    Dumke
@@ -185,6 +186,7 @@ Contact: firstname.lastname@hpi.de
 -   Marvin    Keller
 -   Mirko     Krause
 -   Eva       Krebs
+-   Henok     Lachmann
 -   Sven      Lehmann
 -   Till      Lehmann
 -   Tom       Lichtenstein

resources/test_data/tbl/uniqueness_test_A.tbl

@@ -0,0 +1,6 @@
+A_a|b|c|
+int|int|string|
+4|10|this
+1|3|is
+13|2|unique
+6|10|column

resources/test_data/tbl/uniqueness_test_B.tbl

@@ -0,0 +1,6 @@
+B_a|d
+int|string|
+4|not
+5|unique
+10|unique
+2|column

src/lib/storage/pos_lists/entire_chunk_pos_list.hpp

@@ -21,7 +21,7 @@ class EntireChunkPosList : public AbstractPosList {
 
   // Implemented in hpp for performance reasons (to allow inlining)
   RowID operator[](const size_t index) const final {
-    return RowID{_common_chunk_id, static_cast<ChunkOffset>(index)};
+    return RowID{_common_chunk_id, ChunkOffset{static_cast<ChunkOffset::base_type>(index)}};
   }
 
   bool empty() const final;

src/plugins/CMakeLists.txt

@@ -35,6 +35,7 @@ add_plugin(NAME hyriseMvccDeletePlugin SRCS mvcc_delete_plugin.cpp mvcc_delete_p
 add_plugin(NAME hyriseSecondTestPlugin SRCS second_test_plugin.cpp second_test_plugin.hpp DEPS sqlparser)
 add_plugin(NAME hyriseTestPlugin SRCS test_plugin.cpp test_plugin.hpp DEPS sqlparser)
 add_plugin(NAME hyriseTestNonInstantiablePlugin SRCS non_instantiable_plugin.cpp)
+add_plugin(NAME hyriseUccDiscoveryPlugin SRCS ucc_discovery_plugin.cpp ucc_discovery_plugin.hpp DEPS sqlparser magic_enum compact_vector)
 
 # We define TEST_PLUGIN_DIR to always load plugins from the correct directory for testing purposes
 add_definitions(-DTEST_PLUGIN_DIR="${CMAKE_BINARY_DIR}/lib/")