diff options
Diffstat (limited to '')
-rw-r--r-- | src/lmdb/value_stream.h | 287 |
1 files changed, 287 insertions, 0 deletions
diff --git a/src/lmdb/value_stream.h b/src/lmdb/value_stream.h new file mode 100644 index 000000000..c9977221f --- /dev/null +++ b/src/lmdb/value_stream.h @@ -0,0 +1,287 @@ +// Copyright (c) 2018, The Monero Project +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without modification, are +// permitted provided that the following conditions are met: +// +// 1. Redistributions of source code must retain the above copyright notice, this list of +// conditions and the following disclaimer. +// +// 2. Redistributions in binary form must reproduce the above copyright notice, this list +// of conditions and the following disclaimer in the documentation and/or other +// materials provided with the distribution. +// +// 3. Neither the name of the copyright holder nor the names of its contributors may be +// used to endorse or promote products derived from this software without specific +// prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "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 COPYRIGHT HOLDER OR CONTRIBUTORS 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 + +#include <boost/range/iterator_range.hpp> +#include <cstdint> +#include <cstring> +#include <iterator> +#include <lmdb.h> +#include <utility> + +#include "span.h" + +namespace lmdb +{ + namespace stream + { + /* + \throw std::system_error if unexpected LMDB error. + \return 0 if `cur == nullptr`, otherwise count of values at current key. + */ + std::size_t count(MDB_cursor* cur); + + /*! + Calls `mdb_cursor_get` and does some error checking. + + \param cur is given to `mdb_cursor_get` without modification. + \param op is passed to `mdb_cursor_get` without modification. + \param key expected key size or 0 to skip key size check. + \param value expected value size or 0 to skip value size check. + + \throw std::system_error if `key != 0` and `key_.mv_size != key`. + \throw std::system_error if `value != 0` and `value_.mv_size != value`. + \throw std::system_error if `mdb_cursor_get` returns any error + other than `MDB_NOTFOUND`. + + \return {key bytes, value bytes} or two empty spans if `MDB_NOTFOUND`. + */ + std::pair<epee::span<const std::uint8_t>, epee::span<const std::uint8_t>> + get(MDB_cursor& cur, MDB_cursor_op op, std::size_t key, std::size_t value); + } + + /*! + An InputIterator for a fixed-sized LMDB value at a specific key. + + \tparam T The value type at the specific key. + \tparam F The value type being returned when dereferenced. + \tparam offset to `F` within `T`. + + \note This meets requirements for an InputIterator only. The iterator + can only be incremented and dereferenced. All copies of an iterator + share the same LMDB cursor, and therefore incrementing any copy will + change the cursor state for all (incrementing an iterator will + invalidate all prior copies of the iterator). Usage is identical + to `std::istream_iterator`. + */ + template<typename T, typename F = T, std::size_t offset = 0> + class value_iterator + { + MDB_cursor* cur; + epee::span<const std::uint8_t> values; + + void increment() + { + values.remove_prefix(sizeof(T)); + if (values.empty() && cur) + values = lmdb::stream::get(*cur, MDB_NEXT_DUP, 0, sizeof(T)).second; + } + + public: + using value_type = F; + using reference = value_type; + using pointer = void; + using difference_type = std::size_t; + using iterator_category = std::input_iterator_tag; + + //! Construct an "end" iterator. + value_iterator() noexcept + : cur(nullptr), values() + {} + + /*! + \param cur Iterate over values starting at this cursor position. + \throw std::system_error if unexpected LMDB error. This can happen + if `cur` is invalid. + */ + value_iterator(MDB_cursor* cur) + : cur(cur), values() + { + if (cur) + values = lmdb::stream::get(*cur, MDB_GET_CURRENT, 0, sizeof(T)).second; + } + + value_iterator(value_iterator const&) = default; + ~value_iterator() = default; + value_iterator& operator=(value_iterator const&) = default; + + //! \return True if `this` is one-past the last value. + bool is_end() const noexcept { return values.empty(); } + + //! \return True iff `rhs` is referencing `this` value. + bool equal(value_iterator const& rhs) const noexcept + { + return + (values.empty() && rhs.values.empty()) || + values.data() == rhs.values.data(); + } + + //! Invalidates all prior copies of the iterator. + value_iterator& operator++() + { + increment(); + return *this; + } + + //! \return A copy that is already invalidated, ignore + value_iterator operator++(int) + { + value_iterator out{*this}; + increment(); + return out; + } + + /*! + Get a specific field within `F`. Default behavior is to return + the entirety of `U`, despite the filtering logic of `operator*`. + + \pre `!is_end()` + + \tparam U must match `T`, used for `MONERO_FIELD` sanity checking. + \tparam G field type to extract from the value + \tparam uoffset to `G` type, or `0` when `std::is_same<U, G>()`. + + \return The field `G`, at `uoffset` within `U`. + */ + template<typename U, typename G = U, std::size_t uoffset = 0> + G get_value() const noexcept + { + static_assert(std::is_same<U, T>(), "bad MONERO_FIELD usage?"); + static_assert(std::is_pod<U>(), "value type must be pod"); + static_assert(std::is_pod<G>(), "field type must be pod"); + static_assert(sizeof(G) + uoffset <= sizeof(U), "bad field and/or offset"); + assert(sizeof(G) + uoffset <= values.size()); + assert(!is_end()); + + G value; + std::memcpy(std::addressof(value), values.data() + uoffset, sizeof(value)); + return value; + } + + //! \pre `!is_end()` \return The field `F`, at `offset`, within `T`. + value_type operator*() const noexcept { return get_value<T, F, offset>(); } + }; + + /*! + C++ wrapper for a LMDB read-only cursor on a fixed-sized value `T`. + + \tparam T value type being stored by each record. + \tparam D cleanup functor for the cursor; usually unique per db/table. + */ + template<typename T, typename D> + class value_stream + { + std::unique_ptr<MDB_cursor, D> cur; + public: + + //! Take ownership of `cur` without changing position. `nullptr` valid. + explicit value_stream(std::unique_ptr<MDB_cursor, D> cur) + : cur(std::move(cur)) + {} + + value_stream(value_stream&&) = default; + value_stream(value_stream const&) = delete; + ~value_stream() = default; + value_stream& operator=(value_stream&&) = default; + value_stream& operator=(value_stream const&) = delete; + + /*! + Give up ownership of the cursor. `count()`, `make_iterator()` and + `make_range()` can still be invoked, but return the empty set. + + \return Currently owned LMDB cursor. + */ + std::unique_ptr<MDB_cursor, D> give_cursor() noexcept + { + return {std::move(cur)}; + } + + /*! + Place the stream back at the first value. Newly created iterators + will start at the first value again. + + \note Invalidates all current iterators from `this`, including + those created with `make_iterator` or `make_range`. + */ + void reset() + { + if (cur) + lmdb::stream::get(*cur, MDB_FIRST_DUP, 0, 0); + } + + /*! + \throw std::system_error if LMDB has unexpected errors. + \return Number of values at this key. + */ + std::size_t count() const + { + return lmdb::stream::count(cur.get()); + } + + /*! + Return a C++ iterator over database values from current cursor + position that will reach `.is_end()` after the last duplicate key + record. Calling `make_iterator()` will return an iterator whose + `operator*` will return entire value (`T`). + `make_iterator<MONERO_FIELD(account, id)>()` will return an + iterator whose `operator*` will return a `decltype(account.id)` + object - the other fields in the struct `account` are never copied + from the database. + + \throw std::system_error if LMDB has unexpected errors. + \return C++ iterator starting at current cursor position. + */ + template<typename U = T, typename F = U, std::size_t offset = 0> + value_iterator<U, F, offset> make_iterator() const + { + static_assert(std::is_same<U, T>(), "was MONERO_FIELD used with wrong type?"); + return {cur.get()}; + } + + /*! + Return a range from current cursor position until last duplicate + key record. Useful in for-each range loops or in templated code + expecting a range of elements. Calling `make_range()` will return + a range of `T` objects. `make_range<MONERO_FIELD(account, id)>()` + will return a range of `decltype(account.id)` objects - the other + fields in the struct `account` are never copied from the database. + + \throw std::system_error if LMDB has unexpected errors. + \return An InputIterator range over values at cursor position. + */ + template<typename U = T, typename F = U, std::size_t offset = 0> + boost::iterator_range<value_iterator<U, F, offset>> make_range() const + { + return {make_iterator<U, F, offset>(), value_iterator<U, F, offset>{}}; + } + }; + + template<typename T, typename F, std::size_t offset> + inline + bool operator==(value_iterator<T, F, offset> const& lhs, value_iterator<T, F, offset> const& rhs) noexcept + { + return lhs.equal(rhs); + } + + template<typename T, typename F, std::size_t offset> + inline + bool operator!=(value_iterator<T, F, offset> const& lhs, value_iterator<T, F, offset> const& rhs) noexcept + { + return !lhs.equal(rhs); + } +} // lmdb + |