46ea8c9b8b
Fixes #7435 Adds an "eor" (end-of-record) column to cdc log. This is non-null only on last-in-timestamp group rows, i.e. end of a singular source "event". A client can use this as a shortcut to knowing whether or not he has a full cdc "record" for a given source mutation (single row change). Closes #7436
93 lines
4.3 KiB
C++
93 lines
4.3 KiB
C++
/*
|
|
* Copyright (C) 2020 ScyllaDB
|
|
*/
|
|
|
|
/*
|
|
* This file is part of Scylla.
|
|
*
|
|
* Scylla is free software: you can redistribute it and/or modify
|
|
* it under the terms of the GNU Affero General Public License as published by
|
|
* the Free Software Foundation, either version 3 of the License, or
|
|
* (at your option) any later version.
|
|
*
|
|
* Scylla is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with Scylla. If not, see <http://www.gnu.org/licenses/>.
|
|
*/
|
|
|
|
#pragma once
|
|
|
|
#include <vector>
|
|
#include <boost/dynamic_bitset.hpp>
|
|
#include "schema_fwd.hh"
|
|
#include "timestamp.hh"
|
|
#include "bytes.hh"
|
|
#include <seastar/util/noncopyable_function.hh>
|
|
|
|
class mutation;
|
|
|
|
namespace cdc {
|
|
|
|
// Represents a set of column ids of one kind (partition key, clustering key, regular row or static row).
|
|
// There already exists a column_set type, but it keeps ordinal_column_ids, not column_ids (ordinal column ids
|
|
// are unique across whole table, while kind-specific ids are unique only within one column kind).
|
|
// To avoid converting back and forth between ordinal and kind-specific ids, one_kind_column_set is used instead.
|
|
using one_kind_column_set = boost::dynamic_bitset<uint64_t>;
|
|
|
|
// An object that processes changes from a single, big mutation.
|
|
// It is intended to be used with process_changes_xxx_splitting. Those functions define the order and layout in which
|
|
// changes should appear in CDC log, and change_processor is responsible for producing CDC log rows from changes given
|
|
// by those two functions.
|
|
//
|
|
// The flow of calling its methods should go as follows:
|
|
// -> begin_timestamp #1
|
|
// -> produce_preimage (one call for each preimage row to be generated)
|
|
// -> process_change (one call for each part generated by the splitting function)
|
|
// -> produce_postimage (one call for each postimage row to be generated)
|
|
// -> begin_timestamp #2
|
|
// ...
|
|
class change_processor {
|
|
protected:
|
|
~change_processor() {};
|
|
public:
|
|
// Tells the processor that changes that follow from now on will be of given timestamp.
|
|
// This method must be called in increasing timestamp order.
|
|
// begin_timestamp can be called only once for a given timestamp and change_processor object.
|
|
// ts - timestamp of mutation parts
|
|
// is_last - determines if this will be the last timestamp to be processed by this change_processor instance.
|
|
virtual void begin_timestamp(api::timestamp_type ts, bool is_last) = 0;
|
|
|
|
// Tells the processor to produce a preimage for a given clustering/static row.
|
|
// ck - clustering key of the row for which to produce a preimage; if nullptr, static row preimage is requested
|
|
// columns_to_include - include information about the current state of those columns only, leave others as null
|
|
virtual void produce_preimage(const clustering_key* ck, const one_kind_column_set& columns_to_include) = 0;
|
|
|
|
// Tells the processor to produce a postimage for a given clustering/static row.
|
|
// Contrary to preimage, this requires data from all columns to be present.
|
|
// ck - clustering key of the row for which to produce a postimage; if nullptr, static row postimage is requested
|
|
virtual void produce_postimage(const clustering_key* ck) = 0;
|
|
|
|
// Processes a smaller mutation which is a subset of the big mutation.
|
|
// The mutation provided to process_change should be simple enough for it to be possible to convert it
|
|
// into CDC log rows - for example, it cannot represent a write to two columns of the same row, where
|
|
// both columns have different timestamp or TTL set.
|
|
// m - the small mutation to be converted into CDC log rows.
|
|
virtual void process_change(const mutation& m) = 0;
|
|
|
|
// Tells processor we have reached end of record - last part
|
|
// of a given timestamp batch
|
|
virtual void end_record() = 0;
|
|
};
|
|
|
|
bool should_split(const mutation& base_mutation);
|
|
void process_changes_with_splitting(const mutation& base_mutation, change_processor& processor,
|
|
bool enable_preimage, bool enable_postimage);
|
|
void process_changes_without_splitting(const mutation& base_mutation, change_processor& processor,
|
|
bool enable_preimage, bool enable_postimage);
|
|
|
|
}
|