0
0
Fork 0
mirror of https://github.com/matrix-construct/construct synced 2024-11-05 05:18:54 +01:00
construct/include/ircd/db/row.h

274 lines
7.9 KiB
C++

/*
* Copyright (C) 2016 Charybdis Development Team
* Copyright (C) 2016 Jason Volk <jason@zemos.net>
*
* Permission to use, copy, modify, and/or distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice is present in all copies.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
* INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
*/
#pragma once
#define HAVE_IRCD_DB_ROW_H
namespace ircd::db
{
struct row;
// [SET] Delete row from DB (convenience to an op::DELETE delta)
void del(row &, const sopts & = {});
}
// A `row` is a collection of cells from different columns which all share the same
// key. This is an interface for dealing with those cells in the aggregate.
//
// Note that in a `row` each `cell` comes from a different `column`, but `cell::key()`
// will all return the same index value across the whole `row`. To get the names
// of the columns themselves to build ex. the key name of a JSON key-value pair,
// use `cell::col()`, which will be different for each `cell` across the `row`.
//
// The db::row::iterator iterates over the cells in a row; to iterate over
// multiple rows use the db::cursor
struct ircd::db::row
{
struct delta;
struct iterator;
struct const_iterator;
using value_type = cell &;
using reference = cell &;
using pointer = cell *;
using difference_type = size_t;
private: public:
std::vector<cell> its;
public:
auto empty() const { return its.empty(); }
auto size() const { return its.size(); }
bool valid() const; // true on any cell valid; false on empty
// [GET] Iterations
const_iterator begin() const;
const_iterator end() const;
iterator begin();
iterator end();
// [GET] Get iterator to cell
const_iterator find(const string_view &column) const;
iterator find(const string_view &column);
// [GET] Get cell
const cell &operator[](const string_view &column) const;
cell &operator[](const string_view &column);
// [SET] Perform operation
void operator()(const op &, const string_view &col, const string_view &val = {}, const sopts & = {});
// All cells
void load(const gopts & = {}); // !DANGER! not atomic
row(std::vector<cell> cells = {})
:its{std::move(cells)}
{}
row(database &,
const string_view &key = {},
const vector_view<string_view> &columns = {},
gopts opts = {});
template<class... T>
row(database &,
const string_view &key = {},
const json::tuple<T...> & = {},
gopts opts = {});
template<class pos> friend bool seeka(row &, const pos &);
template<class pos> friend bool seek(row &, const pos &);
friend size_t trim(row &, const std::function<bool (cell &)> &);
friend size_t trim(row &, const string_view &key); // remove invalid or not equal
friend size_t trim(row &); // remove invalid
};
namespace ircd::db
{
// [SET] Perform operations in a sequence as a single transaction. No template
// iterators supported yet, just a ptr range good for contiguous sequences.
void write(const row::delta *const &begin, const row::delta *const &end, const sopts & = {});
void write(const std::initializer_list<row::delta> &, const sopts & = {});
void write(const sopts &, const std::initializer_list<row::delta> &);
void write(const row::delta &, const sopts & = {});
}
struct ircd::db::row::const_iterator
{
using value_type = const cell &;
using reference = const cell &;
using pointer = const cell *;
using iterator_category = std::bidirectional_iterator_tag;
private:
friend class row;
decltype(row::its)::const_iterator it;
const_iterator(decltype(row::its)::const_iterator it)
:it{std::move(it)}
{}
public:
reference operator*() const { return it.operator*(); }
pointer operator->() const { return it.operator->(); }
const_iterator &operator++() { ++it; return *this; }
const_iterator &operator--() { --it; return *this; }
const_iterator() = default;
friend bool operator==(const const_iterator &, const const_iterator &);
friend bool operator!=(const const_iterator &, const const_iterator &);
};
struct ircd::db::row::iterator
{
using value_type = cell &;
using reference = cell &;
using pointer = cell *;
using iterator_category = std::bidirectional_iterator_tag;
private:
friend class row;
decltype(row::its)::iterator it;
iterator(decltype(row::its)::iterator it)
:it{std::move(it)}
{}
public:
reference operator*() const { return it.operator*(); }
pointer operator->() const { return it.operator->(); }
iterator &operator++() { ++it; return *this; }
iterator &operator--() { --it; return *this; }
iterator() = default;
friend bool operator==(const iterator &, const iterator &);
friend bool operator!=(const iterator &, const iterator &);
};
//
// A delta is an element of a database transaction. You can use this to make
// an all-succeed-or-all-fail commitment to multiple rows at once. It is also
// useful to make a commitment on a single row as a convenient way to compose
// all of a row's cells together.
//
struct ircd::db::row::delta
:std::tuple<op, row *>
{
delta(row &r, const enum op &op = op::SET)
:std::tuple<enum op, row *>{op, &r}
{}
delta(const enum op &op, row &r)
:std::tuple<enum op, row *>{op, &r}
{}
};
template<class... T>
ircd::db::row::row(database &d,
const string_view &key,
const json::tuple<T...> &t,
gopts opts)
:row{[&d, &key, &t, &opts]
() -> row
{
std::array<string_view, t.size()> cols;
json::_key_transform(t, std::begin(cols), std::end(cols));
return { d, key, cols, opts };
}()}
{
}
inline ircd::db::cell &
ircd::db::row::operator[](const string_view &column)
{
const auto it(find(column));
if(unlikely(it == end()))
throw schema_error("column '%s' not specified in the descriptor schema", column);
return *it;
}
inline const ircd::db::cell &
ircd::db::row::operator[](const string_view &column)
const
{
const auto it(find(column));
if(unlikely(it == end()))
throw schema_error("column '%s' not specified in the descriptor schema", column);
return *it;
}
inline ircd::db::row::iterator
ircd::db::row::end()
{
return { std::end(its) };
}
inline ircd::db::row::iterator
ircd::db::row::begin()
{
return { std::begin(its) };
}
inline ircd::db::row::const_iterator
ircd::db::row::end()
const
{
return { std::end(its) };
}
inline ircd::db::row::const_iterator
ircd::db::row::begin()
const
{
return { std::begin(its) };
}
inline bool
ircd::db::operator!=(const row::iterator &a, const row::iterator &b)
{
return a.it != b.it;
}
inline bool
ircd::db::operator==(const row::iterator &a, const row::iterator &b)
{
return a.it == b.it;
}
inline bool
ircd::db::operator!=(const row::const_iterator &a, const row::const_iterator &b)
{
return a.it != b.it;
}
inline bool
ircd::db::operator==(const row::const_iterator &a, const row::const_iterator &b)
{
return a.it == b.it;
}