Sprint 3 (hyrise#5)

src/lib/CMakeLists.txt

@@ -3,6 +3,14 @@ set(
     SOURCES
     all_type_variant.hpp
     null_value.hpp
+    operators/abstract_operator.cpp
+    operators/abstract_operator.hpp
+    operators/get_table.hpp
+    operators/print.cpp
+    operators/print.hpp
+    operators/table_scan.hpp
+    operators/table_wrapper.cpp
+    operators/table_wrapper.hpp
     resolve_type.hpp
     storage/abstract_attribute_vector.hpp
     storage/fixed_width_integer_vector.cpp
@@ -12,6 +20,8 @@ set(
     storage/chunk.hpp
     storage/dictionary_segment.cpp
     storage/dictionary_segment.hpp
+    storage/reference_segment.cpp
+    storage/reference_segment.hpp
     storage/storage_manager.cpp
     storage/storage_manager.hpp
     storage/table.cpp

src/lib/operators/abstract_operator.cpp

@@ -0,0 +1,27 @@
+#include "abstract_operator.hpp"
+
+namespace opossum {
+
+AbstractOperator::AbstractOperator(const std::shared_ptr<const AbstractOperator> left,
+                                   const std::shared_ptr<const AbstractOperator> right)
+    : _left_input(left), _right_input(right) {}
+
+void AbstractOperator::execute() {
+  _output = _on_execute();
+}
+
+std::shared_ptr<const Table> AbstractOperator::get_output() const {
+  // TODO(student): You should place some meaningful checks here
+
+  return _output;
+}
+
+std::shared_ptr<const Table> AbstractOperator::_left_input_table() const {
+  return _left_input->get_output();
+}
+
+std::shared_ptr<const Table> AbstractOperator::_right_input_table() const {
+  return _right_input->get_output();
+}
+
+}  // namespace opossum

src/lib/operators/abstract_operator.hpp

@@ -0,0 +1,59 @@
+#pragma once
+
+#include <memory>
+
+#include "types.hpp"
+
+namespace opossum {
+
+class Table;
+
+// AbstractOperator is the abstract super class for all operators. All operators have up to two input tables and one
+// output table. Their lifecycle has three phases:
+// 1. The operator is constructed. Previous operators are not guaranteed to have already executed, so operators must not
+// call get_output in their execute method
+// 2. The execute method is called from the outside (usually by the scheduler). This is where the heavy lifting is done.
+// By now, the input operators have already executed.
+// 3. The consumer (usually another operator) calls get_output. This should be very cheap. It is only guaranteed to
+// succeed if execute was called before. Otherwise, a nullptr or an empty table could be returned.
+//
+// Operators shall not be executed twice.
+
+class AbstractOperator : private Noncopyable {
+ public:
+  AbstractOperator(const std::shared_ptr<const AbstractOperator> left = nullptr,
+                   const std::shared_ptr<const AbstractOperator> right = nullptr);
+
+  virtual ~AbstractOperator() = default;
+
+  // We need to explicitly set the move constructor to default when we overwrite the copy constructor.
+  AbstractOperator(AbstractOperator&&) = default;
+  AbstractOperator& operator=(AbstractOperator&&) = default;
+
+  void execute();
+
+  // Returns the result of the operator.
+  std::shared_ptr<const Table> get_output() const;
+
+  // Get the input operators.
+  std::shared_ptr<const AbstractOperator> left_input() const;
+  std::shared_ptr<const AbstractOperator> right_input() const;
+
+ protected:
+  // Abstract method to actually execute the operator execute and get_output are split into two methods to allow for
+  // easier asynchronous execution.
+  virtual std::shared_ptr<const Table> _on_execute() = 0;
+
+  std::shared_ptr<const Table> _left_input_table() const;
+  std::shared_ptr<const Table> _right_input_table() const;
+
+  // Shared pointers to input operators. Can be nullptr, for example, if an operator is the leaf operator in the query
+  // plan or if the operator has only one input operator.
+  std::shared_ptr<const AbstractOperator> _left_input;
+  std::shared_ptr<const AbstractOperator> _right_input;
+
+  // Is nullptr until the operator is executed.
+  std::shared_ptr<const Table> _output;
+};
+
+}  // namespace opossum

src/lib/operators/get_table.hpp

@@ -0,0 +1,27 @@
+#pragma once
+
+#include "abstract_operator.hpp"
+#include "utils/assert.hpp"
+
+namespace opossum {
+
+// Operator to retrieve a table from the StorageManager by specifying its name.
+class GetTable : public AbstractOperator {
+ public:
+  explicit GetTable(const std::string& name) {
+    // TODO(student) implement it in a source file and change this to a declaration.
+  }
+
+  const std::string& table_name() const {
+    // TODO(student) implement it in a source file and change this to a declaration.
+    Fail("Implementation missing.");
+  }
+
+ protected:
+  std::shared_ptr<const Table> _on_execute() override {
+    // TODO(student) implement it in a source file and change this to a declaration.
+    Fail("Implementation missing.");
+  }
+};
+
+}  // namespace opossum

src/lib/operators/print.cpp

@@ -0,0 +1,110 @@
+#include "print.hpp"
+
+#include <iomanip>
+
+#include "operators/table_wrapper.hpp"
+#include "storage/abstract_segment.hpp"
+#include "storage/table.hpp"
+#include "type_cast.hpp"
+
+namespace {
+
+using namespace opossum;  // NOLINT(build/namespaces)
+
+std::string print_column_type(const std::shared_ptr<const Table>& table, const ColumnID column_id) {
+  auto stream = std::stringstream{};
+  stream << table->column_type(column_id);
+  if (table->column_nullable(column_id)) {
+    stream << "_null";
+  }
+
+  return stream.str();
+}
+
+}  // namespace
+
+namespace opossum {
+
+Print::Print(const std::shared_ptr<const AbstractOperator> in, std::ostream& out) : AbstractOperator(in), _out(out) {}
+
+void Print::print(std::shared_ptr<const Table>& table, std::ostream& out) {
+  auto table_wrapper = std::make_shared<TableWrapper>(table);
+  table_wrapper->execute();
+  Print(table_wrapper, out).execute();
+}
+
+std::shared_ptr<const Table> Print::_on_execute() {
+  auto widths = _column_string_widths(8, 20, _left_input_table());
+
+  // Print column headers.
+  _out << "=== Columns" << std::endl;
+  const auto left_column_count = _left_input_table()->column_count();
+  for (auto column_id = ColumnID{0}; column_id < left_column_count; ++column_id) {
+    _out << "|" << std::setw(widths[column_id]) << _left_input_table()->column_name(column_id) << std::setw(0);
+  }
+  _out << "|" << std::endl;
+  for (auto column_id = ColumnID{0}; column_id < left_column_count; ++column_id) {
+    _out << "|" << std::setw(widths[column_id]) << print_column_type(_left_input_table(), column_id) << std::setw(0);
+  }
+  _out << "|" << std::endl;
+
+  // print each chunk
+  const auto left_chunk_count = _left_input_table()->chunk_count();
+  for (auto chunk_id = ChunkID{0}; chunk_id < left_chunk_count; ++chunk_id) {
+    const auto chunk = _left_input_table()->get_chunk(chunk_id);
+
+    _out << "=== Chunk " << chunk_id << " === " << std::endl;
+
+    if (chunk->size() == 0) {
+      _out << "Empty chunk." << std::endl;
+      continue;
+    }
+
+    // Print the rows in the chunk.
+    const auto chunk_size = chunk->size();
+    for (size_t row = 0; row < chunk_size; ++row) {
+      _out << "|";
+      const auto column_count = chunk->column_count();
+      for (auto column_id = ColumnID{0}; column_id < column_count; ++column_id) {
+        // Yes, we use AbstractSegment::operator[] here, but since Print is not an operation that should be part of a
+        // regular query plan, let's keep things simple here.
+        _out << std::setw(widths[column_id]) << (*chunk->get_segment(column_id))[row] << "|" << std::setw(0);
+      }
+
+      _out << std::endl;
+    }
+  }
+
+  return _left_input_table();
+}
+
+// In order to print the table as an actual table, with columns being aligned, we need to calculate the number of
+// characters in the printed representation of each column `min` and `max` can be used to limit the width of the
+// columns - however, every column fits at least the column's name.
+std::vector<uint16_t> Print::_column_string_widths(uint16_t min, uint16_t max,
+                                                   const std::shared_ptr<const Table>& table) const {
+  auto widths = std::vector<uint16_t>(table->column_count());
+  // Calculate the length of the column name.
+  const auto column_count = table->column_count();
+  for (auto column_id = ColumnID{0}; column_id < column_count; ++column_id) {
+    widths[column_id] = std::max(min, static_cast<uint16_t>(table->column_name(column_id).size()));
+  }
+
+  // Go over all rows and find the maximum length of the printed representation of a value, up to max.
+  const auto left_chunk_count = _left_input_table()->chunk_count();
+  for (auto chunk_id = ChunkID{0}; chunk_id < left_chunk_count; ++chunk_id) {
+    auto chunk = _left_input_table()->get_chunk(chunk_id);
+
+    const auto column_count = chunk->column_count();
+    for (auto column_id = ColumnID{0}; column_id < column_count; ++column_id) {
+      for (auto row = size_t{0}; row < chunk->size(); ++row) {
+        auto cell_length =
+            static_cast<uint16_t>(boost::lexical_cast<std::string>((*chunk->get_segment(column_id))[row]).size());
+        widths[column_id] = std::max({min, widths[column_id], std::min(max, cell_length)});
+      }
+    }
+  }
+  return widths;
+}
+
+}  // namespace opossum

src/lib/operators/print.hpp

@@ -0,0 +1,24 @@
+#pragma once
+
+#include "abstract_operator.hpp"
+
+namespace opossum {
+
+/**
+ * operator to print the table with its data
+ */
+class Print : public AbstractOperator {
+ public:
+  explicit Print(const std::shared_ptr<const AbstractOperator> in, std::ostream& out = std::cout);
+
+  static void print(std::shared_ptr<const Table>& table, std::ostream& out = std::cout);
+
+ protected:
+  std::vector<uint16_t> _column_string_widths(uint16_t min, uint16_t max,
+                                              const std::shared_ptr<const Table>& table) const;
+  std::shared_ptr<const Table> _on_execute() override;
+
+  // stream to print the result
+  std::ostream& _out;
+};
+}  // namespace opossum

src/lib/operators/table_scan.hpp

@@ -0,0 +1,37 @@
+#pragma once
+
+#include "abstract_operator.hpp"
+#include "utils/assert.hpp"
+
+namespace opossum {
+
+class TableScan : public AbstractOperator {
+ public:
+  TableScan(const std::shared_ptr<const AbstractOperator>& in, const ColumnID column_id, const ScanType scan_type,
+            const AllTypeVariant search_value) {
+    // TODO(student) implement it in a source file and change this to a declaration.
+  }
+
+  ColumnID column_id() const {
+    // TODO(student) implement it in a source file and change this to a declaration.
+    Fail("Implementation missing.");
+  }
+
+  ScanType scan_type() const {
+    // TODO(student) implement it in a source file and change this to a declaration.
+    Fail("Implementation missing.");
+  }
+
+  const AllTypeVariant& search_value() const {
+    // TODO(student) implement it in a source file and change this to a declaration.
+    Fail("Implementation missing.");
+  }
+
+ protected:
+  std::shared_ptr<const Table> _on_execute() override {
+    // TODO(student) implement it in a source file and change this to a declaration.
+    Fail("Implementation missing.");
+  }
+};
+
+}  // namespace opossum

src/lib/operators/table_wrapper.cpp

@@ -0,0 +1,10 @@
+#include "table_wrapper.hpp"
+
+namespace opossum {
+
+TableWrapper::TableWrapper(const std::shared_ptr<const Table>& table) : _table(table) {}
+
+std::shared_ptr<const Table> TableWrapper::_on_execute() {
+  return _table;
+}
+}  // namespace opossum

src/lib/operators/table_wrapper.hpp

@@ -0,0 +1,21 @@
+#pragma once
+
+#include "abstract_operator.hpp"
+#include "utils/assert.hpp"
+
+namespace opossum {
+
+/**
+ * Operator that wraps a table.
+ */
+class TableWrapper : public AbstractOperator {
+ public:
+  explicit TableWrapper(const std::shared_ptr<const Table>& table);
+
+ protected:
+  std::shared_ptr<const Table> _on_execute() override;
+
+  // Table to retrieve
+  const std::shared_ptr<const Table> _table;
+};
+}  // namespace opossum

src/lib/storage/reference_segment.cpp

@@ -0,0 +1,43 @@
+#include "reference_segment.hpp"
+
+#include "storage/table.hpp"
+#include "utils/assert.hpp"
+
+namespace opossum {
+
+ReferenceSegment::ReferenceSegment(const std::shared_ptr<const Table>& referenced_table,
+                                   const ColumnID referenced_column_id, const std::shared_ptr<const PosList>& pos) {
+  // Implementation goes here
+}
+
+AllTypeVariant ReferenceSegment::operator[](const ChunkOffset chunk_offset) const {
+  // Implementation goes here
+  Fail("Implementation is missing.");
+}
+
+ChunkOffset ReferenceSegment::size() const {
+  // Implementation goes here
+  Fail("Implementation is missing.");
+}
+
+const std::shared_ptr<const PosList>& ReferenceSegment::pos_list() const {
+  // Implementation goes here
+  Fail("Implementation is missing.");
+}
+
+const std::shared_ptr<const Table>& ReferenceSegment::referenced_table() const {
+  // Implementation goes here
+  Fail("Implementation is missing.");
+}
+
+ColumnID ReferenceSegment::referenced_column_id() const {
+  // Implementation goes here
+  Fail("Implementation is missing.");
+}
+
+size_t ReferenceSegment::estimate_memory_usage() const {
+  // Implementation goes here
+  Fail("Implementation is missing.");
+}
+
+}  // namespace opossum