susumu.yata
null+****@clear*****
Tue Nov 18 16:39:59 JST 2014
susumu.yata 2014-11-18 16:39:59 +0900 (Tue, 18 Nov 2014) New Revision: 2f46fd714757e6e27e47645d5ddeea53abcbf6f2 https://github.com/groonga/grnxx/commit/2f46fd714757e6e27e47645d5ddeea53abcbf6f2 Message: Add Column<Vector<Text>>. (#110) Modified files: lib/grnxx/impl/column/base.cpp lib/grnxx/impl/column/vector.hpp lib/grnxx/impl/column/vector/Makefile.am lib/grnxx/impl/column/vector/text.cpp lib/grnxx/impl/column/vector/text.hpp Modified: lib/grnxx/impl/column/base.cpp (+4 -4) =================================================================== --- lib/grnxx/impl/column/base.cpp 2014-11-18 16:39:13 +0900 (422c17e) +++ lib/grnxx/impl/column/base.cpp 2014-11-18 16:39:59 +0900 (574d4c3) @@ -215,10 +215,10 @@ std::unique_ptr<ColumnBase> ColumnBase::create( column.reset(new impl::Column<Vector<GeoPoint>>(table, name, options)); break; } -// case TEXT_VECTOR_DATA: { -// column.reset(new impl::Column<Vector<Text>>(table, name, options)); -// break; -// } + case TEXT_VECTOR_DATA: { + column.reset(new impl::Column<Vector<Text>>(table, name, options)); + break; + } default: { throw "Not supported"; // TODO } Modified: lib/grnxx/impl/column/vector.hpp (+1 -1) =================================================================== --- lib/grnxx/impl/column/vector.hpp 2014-11-18 16:39:13 +0900 (9adc35b) +++ lib/grnxx/impl/column/vector.hpp 2014-11-18 16:39:59 +0900 (aed3214) @@ -5,6 +5,6 @@ #include "grnxx/impl/column/vector/float.hpp" #include "grnxx/impl/column/vector/geo_point.hpp" #include "grnxx/impl/column/vector/int.hpp" -//#include "grnxx/impl/column/vector/text.hpp" +#include "grnxx/impl/column/vector/text.hpp" #endif // GRNXX_IMPL_COLUMN_VECTOR_HPP Modified: lib/grnxx/impl/column/vector/Makefile.am (+2 -3) =================================================================== --- lib/grnxx/impl/column/vector/Makefile.am 2014-11-18 16:39:13 +0900 (980fe39) +++ lib/grnxx/impl/column/vector/Makefile.am 2014-11-18 16:39:59 +0900 (eeb7737) @@ -12,9 +12,8 @@ libgrnxx_impl_column_vector_la_SOURCES = \ bool.cpp \ float.cpp \ geo_point.cpp \ - int.cpp - -# text.cpp + int.cpp \ + text.cpp libgrnxx_impl_column_vector_includedir = ${includedir}/grnxx/impl/column/vector libgrnxx_impl_column_vector_include_HEADERS = \ Modified: lib/grnxx/impl/column/vector/text.cpp (+130 -67) =================================================================== --- lib/grnxx/impl/column/vector/text.cpp 2014-11-18 16:39:13 +0900 (8bae356) +++ lib/grnxx/impl/column/vector/text.cpp 2014-11-18 16:39:59 +0900 (7db175c) @@ -1,100 +1,163 @@ -#include "grnxx/impl/column/column_vector_text.hpp" +#include "grnxx/impl/column/vector/text.hpp" + +#include <cstring> -#include "grnxx/cursor.hpp" #include "grnxx/impl/db.hpp" #include "grnxx/impl/table.hpp" +//#include "grnxx/impl/index.hpp" namespace grnxx { namespace impl { -bool Column<Vector<Text>>::set(Error *error, Int row_id, - const Datum &datum) { - if (datum.type() != TEXT_VECTOR_DATA) { - GRNXX_ERROR_SET(error, INVALID_ARGUMENT, "Wrong data type"); - return false; +Column<Vector<Text>>::Column(Table *table, + const String &name, + const ColumnOptions &) + : ColumnBase(table, name, TEXT_VECTOR_DATA), + headers_(), + bodies_() {} + +Column<Vector<Text>>::~Column() {} + +void Column<Vector<Text>>::set(Int row_id, const Datum &datum) { + Vector<Text> new_value = parse_datum(datum); + if (!table_->test_row(row_id)) { + throw "Invalid row ID"; // TODO } - if (!table_->test_row(error, row_id)) { - return false; + if (new_value.is_na()) { + unset(row_id); + return; } - Vector<Text> value = datum.force_text_vector(); - if (value.size() == 0) { - headers_[row_id] = Header{ 0, 0 }; - return true; + Vector<Text> old_value = get(row_id); + if ((old_value == new_value).is_true()) { + return; } - Int text_headers_offset = text_headers_.size(); - if (!text_headers_.resize(error, text_headers_offset + value.size())) { - return false; + if (!old_value.is_na()) { + // TODO: Remove the old value from indexes. +// for (size_t i = 0; i < num_indexes(); ++i) { +// indexes_[i]->remove(row_id, old_value); +// } } - Int total_size = 0; - for (Int i = 0; i < value.size(); ++i) { - total_size += value[i].size(); + size_t value_id = row_id.value(); + if (value_id >= headers_.size()) { + headers_.resize(value_id + 1, na_header()); } - Int bodies_offset = bodies_.size(); - if (!bodies_.resize(error, bodies_offset + total_size)) { - return false; + // TODO: Insert the new value into indexes. +// for (size_t i = 0; i < num_indexes(); ++i) try { +// indexes_[i]->insert(row_id, datum)) { +// } catch (...) { +// for (size_t j = 0; j < i; ++i) { +// indexes_[j]->remove(row_id, datum); +// } +// throw; +// } + // TODO: Error handling. + size_t new_value_size = new_value.size().value(); + size_t text_headers_offset = text_headers_.size(); + text_headers_.resize(text_headers_offset + new_value_size); + size_t total_size = 0; + for (size_t i = 0; i < new_value_size; ++i) { + if (!new_value[i].is_na()) { + total_size += new_value[i].size().value(); + } } - headers_[row_id] = Header{ text_headers_offset, value.size() }; - for (Int i = 0; i < value.size(); ++i) { + size_t bodies_offset = bodies_.size(); + bodies_.resize(bodies_offset + total_size); + headers_[value_id] = Header{ text_headers_offset, new_value.size() }; + for (size_t i = 0; i < new_value_size; ++i) { text_headers_[text_headers_offset + i].offset = bodies_offset; - text_headers_[text_headers_offset + i].size = value[i].size(); - std::memcpy(&bodies_[bodies_offset], value[i].data(), value[i].size()); - bodies_offset += value[i].size(); + text_headers_[text_headers_offset + i].size = new_value[i].size(); + if (!new_value[i].is_na()) { + std::memcpy(&bodies_[bodies_offset], + new_value[i].data(), new_value[i].size().value()); + bodies_offset += new_value[i].size().value(); + } } - return true; } -bool Column<Vector<Text>>::get(Error *error, Int row_id, - Datum *datum) const { - if (!table_->test_row(error, row_id)) { - return false; +void Column<Vector<Text>>::get(Int row_id, Datum *datum) const { + size_t value_id = row_id.value(); + if (value_id >= headers_.size()) { + *datum = Vector<Text>::na(); + } else { + // TODO + *datum = get(row_id); } - *datum = get(row_id); - return true; } -unique_ptr<Column<Vector<Text>>> Column<Vector<Text>>::create( - Error *error, - Table *table, - const StringCRef &name, - const ColumnOptions &options) { - unique_ptr<Column> column(new (nothrow) Column); - if (!column) { - GRNXX_ERROR_SET(error, NO_MEMORY, "Memory allocation failed"); - return nullptr; - } - if (!column->initialize_base(error, table, name, - TEXT_VECTOR_DATA, options)) { - return nullptr; - } - if (!column->headers_.resize(error, table->max_row_id() + 1, - Header{ 0, 0 })) { - return nullptr; +bool Column<Vector<Text>>::contains(const Datum &datum) const { + // TODO: Use an index if exists. + Vector<Text> value = parse_datum(datum); + if (value.is_na()) { + for (size_t i = 0; i < headers_.size(); ++i) { + if (headers_[i].size.is_na()) { + return true; + } + } + } else { + for (size_t i = 0; i < headers_.size(); ++i) { + // TODO: Improve this. + if ((get(Int(i)) == value).is_true()) { + return true; + } + } } - return column; + return false; } -Column<Vector<Text>>::~Column() {} - -bool Column<Vector<Text>>::set_default_value(Error *error, - Int row_id) { - if (row_id >= headers_.size()) { - if (!headers_.resize(error, row_id + 1)) { - return false; +Int Column<Vector<Text>>::find_one(const Datum &datum) const { + // TODO: Use an index if exists. + Vector<Text> value = parse_datum(datum); + if (value.is_na()) { + for (size_t i = 0; i < headers_.size(); ++i) { + if (headers_[i].size.is_na()) { + return Int(i); + } + } + } else { + for (size_t i = 0; i < headers_.size(); ++i) { + // TODO: Improve this. + if ((get(Int(i)) == value).is_true()) { + return Int(i); + } } } - headers_[row_id] = Header{ 0, 0 }; - return true; + return Int::na(); } void Column<Vector<Text>>::unset(Int row_id) { - headers_[row_id] = Header{ 0, 0 }; + Vector<Text> value = get(row_id); + if (!value.is_na()) { + // TODO: Update indexes if exist. +// for (size_t i = 0; i < num_indexes(); ++i) { +// indexes_[i]->remove(row_id, value); +// } + headers_[row_id.value()] = na_header(); + } } -Column<Vector<Text>>::Column() - : ColumnBase(), - headers_(), - text_headers_(), - bodies_() {} +void Column<Vector<Text>>::read(ArrayCRef<Record> records, + ArrayRef<Vector<Text>> values) const { + if (records.size() != values.size()) { + throw "Data size conflict"; // TODO + } + for (size_t i = 0; i < records.size(); ++i) { + values.set(i, get(records[i].row_id)); + } +} + +Vector<Text> Column<Vector<Text>>::parse_datum(const Datum &datum) { + switch (datum.type()) { + case NA_DATA: { + return Vector<Text>::na(); + } + case TEXT_VECTOR_DATA: { + return datum.as_text_vector(); + } + default: { + throw "Wrong data type"; // TODO + } + } +} } // namespace impl } // namespace grnxx Modified: lib/grnxx/impl/column/vector/text.hpp (+36 -28) =================================================================== --- lib/grnxx/impl/column/vector/text.hpp 2014-11-18 16:39:13 +0900 (e97df22) +++ lib/grnxx/impl/column/vector/text.hpp 2014-11-18 16:39:59 +0900 (1c7c8a1) @@ -1,61 +1,69 @@ #ifndef GRNXX_IMPL_COLUMN_VECTOR_TEXT_HPP #define GRNXX_IMPL_COLUMN_VECTOR_TEXT_HPP -#include "grnxx/impl/column/column.hpp" +#include "grnxx/impl/column/base.hpp" namespace grnxx { namespace impl { +template <typename T> class Column; + template <> class Column<Vector<Text>> : public ColumnBase { public: - // -- Public API -- + // -- Public API (grnxx/column.hpp) -- - bool set(Error *error, Int row_id, const Datum &datum); - bool get(Error *error, Int row_id, Datum *datum) const; + Column(Table *table, const String &name, const ColumnOptions &options); + ~Column(); - // -- Internal API -- + void set(Int row_id, const Datum &datum); + void get(Int row_id, Datum *datum) const; - // Create a new column. - // - // Returns a pointer to the column on success. - // On failure, returns nullptr and stores error information into "*error" if - // "error" != nullptr. - static unique_ptr<Column> create(Error *error, - Table *table, - const StringCRef &name, - const ColumnOptions &options); + bool contains(const Datum &datum) const; + Int find_one(const Datum &datum) const; - ~Column(); + // -- Internal API (grnxx/impl/column/base.hpp) -- - bool set_default_value(Error *error, Int row_id); void unset(Int row_id); - // Return a value identified by "row_id". + // -- Internal API -- + + // Return a value. + // + // If "row_id" is valid, returns the stored value. + // If "row_id" is invalid, returns N/A. // - // Assumes that "row_id" is valid. Otherwise, the result is undefined. + // TODO: Vector cannot reuse allocated memory because of this interface. Vector<Text> get(Int row_id) const { - return Vector<Text>(&text_headers_[headers_[row_id].offset], - bodies_.data(), headers_[row_id].size); - } - - // Read values. - void read(ArrayCRef<Record> records, ArrayRef<Vector<Text>> values) const { - for (Int i = 0; i < records.size(); ++i) { - values.set(i, get(records.get_row_id(i))); + size_t value_id = row_id.value(); + if (value_id >= headers_.size()) { + return Vector<Text>::na(); + } + if (headers_[value_id].size.is_na()) { + return Vector<Text>::na(); } + return Vector<Text>(&text_headers_[headers_[value_id].offset], + bodies_.data(), headers_[value_id].size); } + // Read values. + // + // On failure, throws an exception. + void read(ArrayCRef<Record> records, ArrayRef<Vector<Text>> values) const; private: struct Header { - Int offset; + size_t offset; Int size; }; Array<Header> headers_; Array<Header> text_headers_; Array<char> bodies_; - Column(); + static constexpr Header na_header() { + return Header{ 0, Int::na() }; + } + + static Vector<Text> parse_datum(const Datum &datum); }; } // namespace impl -------------- next part -------------- HTML����������������������������...Download
