Files
scylladb/alternator/executor.hh
Nadav Har'El 104ef5242b alternator: support attribute paths in ProjectionExpression
This patch fully implements support for attribute paths (e.g. a.b.c, a.d[3])
for the ProjectionExpression in the various operations where this parameter
is supported - GetItem, BatchGetItem, Query and Scan. After this patch, all
xfailing tests in test_projection_expression.py now pass.

In the previous patch we remembered in the "attrs_to_get" object not only
the top-level attributes to read from the table, but also how to filter
from it only the desired pieces of the nested document. In this patch we
add a filter() function to do this filtering, and call it in the right
places to post-process the JSON objects we read from the table.

We also had to fix reference resolution in paths to resolve all the
components of the path (e.g., #name1.#name2) and not just the top-level
attribute.

This is not the end of attribute path support, there are still other
expressions (ConditionExpression, UpdateExpression, FilterExpression,
ReturnValues) where they are not yet supported. This will come in following
patches.

Refs #5024

Signed-off-by: Nadav Har'El <nyh@scylladb.com>
2021-02-08 14:16:40 +02:00

193 lines
9.1 KiB
C++

/*
* Copyright 2019 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 Affero General Public License
* along with Scylla. If not, see <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <seastar/core/future.hh>
#include <seastar/http/httpd.hh>
#include "seastarx.hh"
#include <seastar/json/json_elements.hh>
#include <seastar/core/sharded.hh>
#include "service/storage_proxy.hh"
#include "service/migration_manager.hh"
#include "service/client_state.hh"
#include "db/timeout_clock.hh"
#include "alternator/error.hh"
#include "stats.hh"
#include "utils/rjson.hh"
namespace db {
class system_distributed_keyspace;
}
namespace query {
class partition_slice;
class result;
}
namespace cql3::selection {
class selection;
}
namespace service {
class storage_service;
}
namespace alternator {
class rmw_operation;
struct make_jsonable : public json::jsonable {
rjson::value _value;
public:
explicit make_jsonable(rjson::value&& value);
std::string to_json() const override;
};
struct json_string : public json::jsonable {
std::string _value;
public:
explicit json_string(std::string&& value);
std::string to_json() const override;
};
namespace parsed {
class path;
};
// A hierarchy_filter is used to specify only a part of the hierarchy that we
// wish to extract from a JSON object. For example when a ProjectionExpression
// requires that we only return certain pieces of a top-level attribute.
struct hierarchy_filter {
// Either "members" or "indexes" may be non-empty, but not both. This is
// because DynamoDB does not allow both a.b and a[1] to be requested at
// the same time. Those are considered to be "conflicting" document paths.
// We could allow both, if we wanted to diverge from DynamoDB.
// We need the extra unique_ptr<> here because libstdc++ unordered_map
// doesn't work with incomplete types :-(
std::unordered_map<std::string, std::unique_ptr<hierarchy_filter>> members;
std::unordered_map<unsigned, std::unique_ptr<hierarchy_filter>> indexes;
// Add a path to the filter. Throws a validation error if the path
// "overlaps" with one already in the filter (one is a sub-path
// of the other) or "conflicts" with it (both a member and index is
// requested).
void add(const parsed::path&);
// Run the filter on a given JSON object, dropping its parts which
// weren't asked to be kept. Modifies the given JSON object, or
// returns false to signify the entire object should be dropped.
// Important note: The JSON object is assumed to be using the DynamoDB
// conventions, i.e., it is really a map whose key has a type string,
// and the value is the real object.
bool filter(rjson::value&);
};
// We use lw_shared_ptr<hierarchy_filter> allows us to efficiently represent
// with a nullptr the common case where there is no hierarchy filter, i.e.,
// a top-level attribute is requested. We use lw_shared_ptr and not unique_ptr
// to allow copying attrs_to_get.
using attrs_to_get = std::unordered_map<std::string, lw_shared_ptr<hierarchy_filter>>;
class executor : public peering_sharded_service<executor> {
service::storage_proxy& _proxy;
service::migration_manager& _mm;
db::system_distributed_keyspace& _sdks;
service::storage_service& _ss;
// An smp_service_group to be used for limiting the concurrency when
// forwarding Alternator request between shards - if necessary for LWT.
smp_service_group _ssg;
public:
using client_state = service::client_state;
using request_return_type = std::variant<json::json_return_type, api_error>;
stats _stats;
static constexpr auto ATTRS_COLUMN_NAME = ":attrs";
static constexpr auto KEYSPACE_NAME_PREFIX = "alternator_";
static constexpr std::string_view INTERNAL_TABLE_PREFIX = ".scylla.alternator.";
executor(service::storage_proxy& proxy, service::migration_manager& mm, db::system_distributed_keyspace& sdks, service::storage_service& ss, smp_service_group ssg)
: _proxy(proxy), _mm(mm), _sdks(sdks), _ss(ss), _ssg(ssg) {}
future<request_return_type> create_table(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> describe_table(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> delete_table(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> update_table(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> put_item(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> get_item(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> delete_item(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> update_item(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> list_tables(client_state& client_state, service_permit permit, rjson::value request);
future<request_return_type> scan(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> describe_endpoints(client_state& client_state, service_permit permit, rjson::value request, std::string host_header);
future<request_return_type> batch_write_item(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> batch_get_item(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> query(client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit, rjson::value request);
future<request_return_type> tag_resource(client_state& client_state, service_permit permit, rjson::value request);
future<request_return_type> untag_resource(client_state& client_state, service_permit permit, rjson::value request);
future<request_return_type> list_tags_of_resource(client_state& client_state, service_permit permit, rjson::value request);
future<request_return_type> list_streams(client_state& client_state, service_permit permit, rjson::value request);
future<request_return_type> describe_stream(client_state& client_state, service_permit permit, rjson::value request);
future<request_return_type> get_shard_iterator(client_state& client_state, service_permit permit, rjson::value request);
future<request_return_type> get_records(client_state& client_state, tracing::trace_state_ptr, service_permit permit, rjson::value request);
future<> start();
future<> stop() { return make_ready_future<>(); }
future<> create_keyspace(std::string_view keyspace_name);
static tracing::trace_state_ptr maybe_trace_query(client_state& client_state, sstring_view op, sstring_view query);
static sstring table_name(const schema&);
static db::timeout_clock::time_point default_timeout();
static void set_default_timeout(db::timeout_clock::duration timeout);
private:
static db::timeout_clock::duration s_default_timeout;
public:
static schema_ptr find_table(service::storage_proxy&, const rjson::value& request);
private:
friend class rmw_operation;
static bool is_alternator_keyspace(const sstring& ks_name);
static sstring make_keyspace_name(const sstring& table_name);
static void describe_key_schema(rjson::value& parent, const schema&, std::unordered_map<std::string,std::string> * = nullptr);
static void describe_key_schema(rjson::value& parent, const schema& schema, std::unordered_map<std::string,std::string>&);
public:
static std::optional<rjson::value> describe_single_item(schema_ptr,
const query::partition_slice&,
const cql3::selection::selection&,
const query::result&,
const attrs_to_get&);
static void describe_single_item(const cql3::selection::selection&,
const std::vector<bytes_opt>&,
const attrs_to_get&,
rjson::value&,
bool = false);
void add_stream_options(const rjson::value& stream_spec, schema_builder&) const;
void supplement_table_info(rjson::value& descr, const schema& schema) const;
void supplement_table_stream_info(rjson::value& descr, const schema& schema) const;
};
}