2019-12-06 14:37:21 +00:00
|
|
|
#include <Common/config.h>
|
|
|
|
|
|
|
|
#if USE_AWS_S3
|
|
|
|
|
2019-12-03 16:23:24 +00:00
|
|
|
#include <IO/S3Common.h>
|
2019-05-23 09:03:39 +00:00
|
|
|
#include <Storages/StorageFactory.h>
|
2019-05-31 07:27:14 +00:00
|
|
|
#include <Storages/StorageS3.h>
|
2020-06-01 17:16:09 +00:00
|
|
|
#include <Storages/StorageS3Settings.h>
|
2019-05-23 09:03:39 +00:00
|
|
|
|
|
|
|
#include <Interpreters/Context.h>
|
|
|
|
#include <Interpreters/evaluateConstantExpression.h>
|
|
|
|
#include <Parsers/ASTLiteral.h>
|
|
|
|
|
2019-06-01 21:18:20 +00:00
|
|
|
#include <IO/ReadBufferFromS3.h>
|
2019-11-20 14:48:01 +00:00
|
|
|
#include <IO/ReadHelpers.h>
|
2019-06-01 21:18:20 +00:00
|
|
|
#include <IO/WriteBufferFromS3.h>
|
2019-11-20 14:48:01 +00:00
|
|
|
#include <IO/WriteHelpers.h>
|
2019-05-23 09:03:39 +00:00
|
|
|
|
|
|
|
#include <Formats/FormatFactory.h>
|
2019-05-31 07:27:14 +00:00
|
|
|
|
2019-05-23 09:03:39 +00:00
|
|
|
#include <DataStreams/IBlockOutputStream.h>
|
|
|
|
#include <DataStreams/AddingDefaultsBlockInputStream.h>
|
2020-01-26 13:12:15 +00:00
|
|
|
#include <DataStreams/narrowBlockInputStreams.h>
|
2019-05-23 09:03:39 +00:00
|
|
|
|
2020-01-26 14:05:51 +00:00
|
|
|
#include <DataTypes/DataTypeString.h>
|
|
|
|
|
2020-06-01 17:16:09 +00:00
|
|
|
#include <aws/core/auth/AWSCredentials.h>
|
2019-12-03 16:23:24 +00:00
|
|
|
#include <aws/s3/S3Client.h>
|
2020-05-25 09:26:50 +00:00
|
|
|
#include <aws/s3/model/ListObjectsV2Request.h>
|
2020-01-26 13:03:47 +00:00
|
|
|
|
|
|
|
#include <Common/parseGlobs.h>
|
2020-02-17 19:28:25 +00:00
|
|
|
#include <Common/quoteString.h>
|
2020-01-26 13:03:47 +00:00
|
|
|
#include <re2/re2.h>
|
2019-05-23 09:03:39 +00:00
|
|
|
|
2020-02-03 18:01:41 +00:00
|
|
|
#include <Processors/Sources/SourceWithProgress.h>
|
2020-05-18 10:00:22 +00:00
|
|
|
#include <Processors/Formats/InputStreamFromInputFormat.h>
|
2020-02-03 18:01:41 +00:00
|
|
|
#include <Processors/Pipe.h>
|
2021-05-07 07:07:44 +00:00
|
|
|
#include <Poco/Util/AbstractConfiguration.h>
|
2021-05-08 10:59:55 +00:00
|
|
|
#include <filesystem>
|
2019-05-23 09:03:39 +00:00
|
|
|
|
2021-05-08 10:59:55 +00:00
|
|
|
namespace fs = std::filesystem;
|
2019-05-31 07:27:14 +00:00
|
|
|
|
2019-05-23 09:03:39 +00:00
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
namespace ErrorCodes
|
|
|
|
{
|
|
|
|
extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH;
|
2020-01-30 06:17:55 +00:00
|
|
|
extern const int UNEXPECTED_EXPRESSION;
|
|
|
|
extern const int S3_ERROR;
|
2019-05-23 09:03:39 +00:00
|
|
|
}
|
2021-04-06 19:18:45 +00:00
|
|
|
class StorageS3Source::DisclosedGlobIterator::Impl
|
|
|
|
{
|
|
|
|
|
|
|
|
public:
|
|
|
|
Impl(Aws::S3::S3Client & client_, const S3::URI & globbed_uri_)
|
2021-04-12 17:07:01 +00:00
|
|
|
: client(client_), globbed_uri(globbed_uri_)
|
|
|
|
{
|
2021-04-10 02:21:18 +00:00
|
|
|
std::lock_guard lock(mutex);
|
|
|
|
|
2021-04-06 19:18:45 +00:00
|
|
|
if (globbed_uri.bucket.find_first_of("*?{") != globbed_uri.bucket.npos)
|
|
|
|
throw Exception("Expression can not have wildcards inside bucket name", ErrorCodes::UNEXPECTED_EXPRESSION);
|
|
|
|
|
|
|
|
const String key_prefix = globbed_uri.key.substr(0, globbed_uri.key.find_first_of("*?{"));
|
|
|
|
|
2021-04-12 16:51:52 +00:00
|
|
|
/// We don't have to list bucket, because there is no asterics.
|
2021-04-06 19:18:45 +00:00
|
|
|
if (key_prefix.size() == globbed_uri.key.size())
|
2021-04-12 16:51:52 +00:00
|
|
|
{
|
2021-04-06 19:18:45 +00:00
|
|
|
buffer.emplace_back(globbed_uri.key);
|
2021-04-12 16:51:52 +00:00
|
|
|
buffer_iter = buffer.begin();
|
|
|
|
is_finished = true;
|
|
|
|
return;
|
|
|
|
}
|
2021-04-06 19:18:45 +00:00
|
|
|
|
|
|
|
request.SetBucket(globbed_uri.bucket);
|
|
|
|
request.SetPrefix(key_prefix);
|
|
|
|
matcher = std::make_unique<re2::RE2>(makeRegexpPatternFromGlobs(globbed_uri.key));
|
2021-04-10 14:58:29 +00:00
|
|
|
fillInternalBufferAssumeLocked();
|
2021-04-06 19:18:45 +00:00
|
|
|
}
|
|
|
|
|
2021-04-13 10:59:02 +00:00
|
|
|
String next()
|
2021-04-10 02:21:18 +00:00
|
|
|
{
|
|
|
|
std::lock_guard lock(mutex);
|
|
|
|
return nextAssumeLocked();
|
|
|
|
}
|
|
|
|
|
|
|
|
private:
|
|
|
|
|
2021-04-13 10:59:02 +00:00
|
|
|
String nextAssumeLocked()
|
2021-04-06 19:18:45 +00:00
|
|
|
{
|
|
|
|
if (buffer_iter != buffer.end())
|
|
|
|
{
|
|
|
|
auto answer = *buffer_iter;
|
|
|
|
++buffer_iter;
|
|
|
|
return answer;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (is_finished)
|
2021-04-13 10:59:02 +00:00
|
|
|
return {};
|
2021-04-06 19:18:45 +00:00
|
|
|
|
2021-04-10 02:21:18 +00:00
|
|
|
fillInternalBufferAssumeLocked();
|
2021-04-06 19:18:45 +00:00
|
|
|
|
2021-04-10 02:21:18 +00:00
|
|
|
return nextAssumeLocked();
|
2021-04-06 19:18:45 +00:00
|
|
|
}
|
|
|
|
|
2021-04-10 02:21:18 +00:00
|
|
|
void fillInternalBufferAssumeLocked()
|
2021-04-06 19:18:45 +00:00
|
|
|
{
|
|
|
|
buffer.clear();
|
|
|
|
|
|
|
|
outcome = client.ListObjectsV2(request);
|
|
|
|
if (!outcome.IsSuccess())
|
|
|
|
throw Exception(ErrorCodes::S3_ERROR, "Could not list objects in bucket {} with prefix {}, S3 exception: {}, message: {}",
|
|
|
|
quoteString(request.GetBucket()), quoteString(request.GetPrefix()),
|
|
|
|
backQuote(outcome.GetError().GetExceptionName()), quoteString(outcome.GetError().GetMessage()));
|
|
|
|
|
|
|
|
const auto & result_batch = outcome.GetResult().GetContents();
|
|
|
|
|
|
|
|
buffer.reserve(result_batch.size());
|
|
|
|
for (const auto & row : result_batch)
|
|
|
|
{
|
|
|
|
String key = row.GetKey();
|
|
|
|
if (re2::RE2::FullMatch(key, *matcher))
|
|
|
|
buffer.emplace_back(std::move(key));
|
|
|
|
}
|
|
|
|
/// Set iterator only after the whole batch is processed
|
|
|
|
buffer_iter = buffer.begin();
|
|
|
|
|
|
|
|
request.SetContinuationToken(outcome.GetResult().GetNextContinuationToken());
|
2021-04-12 17:07:01 +00:00
|
|
|
|
2021-04-06 19:18:45 +00:00
|
|
|
/// It returns false when all objects were returned
|
|
|
|
is_finished = !outcome.GetResult().GetIsTruncated();
|
|
|
|
}
|
|
|
|
|
2021-04-10 02:21:18 +00:00
|
|
|
std::mutex mutex;
|
2021-04-06 19:18:45 +00:00
|
|
|
Strings buffer;
|
|
|
|
Strings::iterator buffer_iter;
|
|
|
|
Aws::S3::S3Client client;
|
|
|
|
S3::URI globbed_uri;
|
|
|
|
Aws::S3::Model::ListObjectsV2Request request;
|
|
|
|
Aws::S3::Model::ListObjectsV2Outcome outcome;
|
|
|
|
std::unique_ptr<re2::RE2> matcher;
|
|
|
|
bool is_finished{false};
|
|
|
|
};
|
|
|
|
|
|
|
|
StorageS3Source::DisclosedGlobIterator::DisclosedGlobIterator(Aws::S3::S3Client & client_, const S3::URI & globbed_uri_)
|
2021-04-10 02:21:18 +00:00
|
|
|
: pimpl(std::make_shared<StorageS3Source::DisclosedGlobIterator::Impl>(client_, globbed_uri_)) {}
|
2021-04-06 19:18:45 +00:00
|
|
|
|
2021-04-13 10:59:02 +00:00
|
|
|
String StorageS3Source::DisclosedGlobIterator::next()
|
2021-04-06 19:18:45 +00:00
|
|
|
{
|
|
|
|
return pimpl->next();
|
|
|
|
}
|
|
|
|
|
2019-12-03 16:23:24 +00:00
|
|
|
|
2021-03-22 17:12:31 +00:00
|
|
|
Block StorageS3Source::getHeader(Block sample_block, bool with_path_column, bool with_file_column)
|
2019-05-23 09:03:39 +00:00
|
|
|
{
|
2021-03-22 17:12:31 +00:00
|
|
|
if (with_path_column)
|
|
|
|
sample_block.insert({DataTypeString().createColumn(), std::make_shared<DataTypeString>(), "_path"});
|
|
|
|
if (with_file_column)
|
|
|
|
sample_block.insert({DataTypeString().createColumn(), std::make_shared<DataTypeString>(), "_file"});
|
2020-02-14 17:47:39 +00:00
|
|
|
|
2021-03-22 17:12:31 +00:00
|
|
|
return sample_block;
|
|
|
|
}
|
2020-02-14 17:47:39 +00:00
|
|
|
|
2021-03-22 17:12:31 +00:00
|
|
|
StorageS3Source::StorageS3Source(
|
|
|
|
bool need_path,
|
|
|
|
bool need_file,
|
2021-04-08 00:09:15 +00:00
|
|
|
const String & format_,
|
2021-03-22 17:12:31 +00:00
|
|
|
String name_,
|
2021-04-08 00:09:15 +00:00
|
|
|
const Block & sample_block_,
|
2021-04-12 19:35:26 +00:00
|
|
|
ContextPtr context_,
|
2021-04-08 00:09:15 +00:00
|
|
|
const ColumnsDescription & columns_,
|
|
|
|
UInt64 max_block_size_,
|
2021-05-19 21:42:25 +00:00
|
|
|
UInt64 max_single_read_retries_,
|
|
|
|
const String & compression_hint_,
|
2021-04-08 00:09:15 +00:00
|
|
|
const std::shared_ptr<Aws::S3::S3Client> & client_,
|
|
|
|
const String & bucket_,
|
2021-04-10 02:21:18 +00:00
|
|
|
std::shared_ptr<IteratorWrapper> file_iterator_)
|
2021-04-08 00:09:15 +00:00
|
|
|
: SourceWithProgress(getHeader(sample_block_, need_path, need_file))
|
2021-04-12 19:35:26 +00:00
|
|
|
, WithContext(context_)
|
2021-03-22 17:12:31 +00:00
|
|
|
, name(std::move(name_))
|
2021-04-08 00:09:15 +00:00
|
|
|
, bucket(bucket_)
|
|
|
|
, format(format_)
|
|
|
|
, columns_desc(columns_)
|
|
|
|
, max_block_size(max_block_size_)
|
2021-05-19 21:42:25 +00:00
|
|
|
, max_single_read_retries(max_single_read_retries_)
|
2021-04-08 00:09:15 +00:00
|
|
|
, compression_hint(compression_hint_)
|
|
|
|
, client(client_)
|
|
|
|
, sample_block(sample_block_)
|
2021-03-22 17:12:31 +00:00
|
|
|
, with_file_column(need_file)
|
|
|
|
, with_path_column(need_path)
|
2021-04-08 00:09:15 +00:00
|
|
|
, file_iterator(file_iterator_)
|
|
|
|
{
|
|
|
|
initialize();
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
bool StorageS3Source::initialize()
|
2021-03-22 17:12:31 +00:00
|
|
|
{
|
2021-04-13 10:59:02 +00:00
|
|
|
String current_key = (*file_iterator)();
|
|
|
|
if (current_key.empty())
|
2021-04-08 00:09:15 +00:00
|
|
|
return false;
|
2021-04-13 10:59:02 +00:00
|
|
|
|
2021-05-08 10:59:55 +00:00
|
|
|
file_path = fs::path(bucket) / current_key;
|
2021-04-08 00:09:15 +00:00
|
|
|
|
|
|
|
read_buf = wrapReadBufferWithCompressionMethod(
|
2021-05-19 21:42:25 +00:00
|
|
|
std::make_unique<ReadBufferFromS3>(client, bucket, current_key, max_single_read_retries), chooseCompressionMethod(current_key, compression_hint));
|
2021-04-12 19:35:26 +00:00
|
|
|
auto input_format = FormatFactory::instance().getInput(format, *read_buf, sample_block, getContext(), max_block_size);
|
2021-03-22 17:12:31 +00:00
|
|
|
reader = std::make_shared<InputStreamFromInputFormat>(input_format);
|
2020-02-14 17:47:39 +00:00
|
|
|
|
2021-04-08 00:09:15 +00:00
|
|
|
if (columns_desc.hasDefaults())
|
2021-04-12 19:35:26 +00:00
|
|
|
reader = std::make_shared<AddingDefaultsBlockInputStream>(reader, columns_desc, getContext());
|
2021-04-08 00:09:15 +00:00
|
|
|
|
|
|
|
initialized = false;
|
|
|
|
return true;
|
2021-03-22 17:12:31 +00:00
|
|
|
}
|
2020-02-03 18:01:41 +00:00
|
|
|
|
2021-03-22 17:12:31 +00:00
|
|
|
String StorageS3Source::getName() const
|
|
|
|
{
|
|
|
|
return name;
|
|
|
|
}
|
2019-05-23 09:03:39 +00:00
|
|
|
|
2021-03-22 17:12:31 +00:00
|
|
|
Chunk StorageS3Source::generate()
|
|
|
|
{
|
|
|
|
if (!reader)
|
|
|
|
return {};
|
|
|
|
|
|
|
|
if (!initialized)
|
|
|
|
{
|
2021-03-26 15:33:14 +00:00
|
|
|
reader->readPrefix();
|
2021-03-22 17:12:31 +00:00
|
|
|
initialized = true;
|
|
|
|
}
|
2019-05-23 09:03:39 +00:00
|
|
|
|
2021-03-22 17:12:31 +00:00
|
|
|
if (auto block = reader->read())
|
|
|
|
{
|
|
|
|
auto columns = block.getColumns();
|
|
|
|
UInt64 num_rows = block.rows();
|
|
|
|
|
|
|
|
if (with_path_column)
|
|
|
|
columns.push_back(DataTypeString().createColumnConst(num_rows, file_path)->convertToFullColumnIfConst());
|
|
|
|
if (with_file_column)
|
2019-05-31 07:27:14 +00:00
|
|
|
{
|
2021-03-22 17:12:31 +00:00
|
|
|
size_t last_slash_pos = file_path.find_last_of('/');
|
|
|
|
columns.push_back(DataTypeString().createColumnConst(num_rows, file_path.substr(
|
|
|
|
last_slash_pos + 1))->convertToFullColumnIfConst());
|
2019-05-31 07:27:14 +00:00
|
|
|
}
|
|
|
|
|
2021-03-22 17:12:31 +00:00
|
|
|
return Chunk(std::move(columns), num_rows);
|
|
|
|
}
|
|
|
|
|
2021-04-08 00:09:15 +00:00
|
|
|
reader->readSuffix();
|
2021-03-22 17:12:31 +00:00
|
|
|
reader.reset();
|
2021-04-08 00:09:15 +00:00
|
|
|
read_buf.reset();
|
2021-03-22 17:12:31 +00:00
|
|
|
|
2021-04-08 00:09:15 +00:00
|
|
|
if (!initialize())
|
|
|
|
return {};
|
|
|
|
|
|
|
|
return generate();
|
2021-03-22 17:12:31 +00:00
|
|
|
}
|
|
|
|
|
2021-04-12 19:35:26 +00:00
|
|
|
|
2021-04-12 17:07:01 +00:00
|
|
|
class StorageS3BlockOutputStream : public IBlockOutputStream
|
2021-03-22 17:12:31 +00:00
|
|
|
{
|
2021-04-12 17:07:01 +00:00
|
|
|
public:
|
|
|
|
StorageS3BlockOutputStream(
|
|
|
|
const String & format,
|
|
|
|
const Block & sample_block_,
|
|
|
|
ContextPtr context,
|
|
|
|
const CompressionMethod compression_method,
|
|
|
|
const std::shared_ptr<Aws::S3::S3Client> & client,
|
|
|
|
const String & bucket,
|
|
|
|
const String & key,
|
|
|
|
size_t min_upload_part_size,
|
|
|
|
size_t max_single_part_upload_size)
|
|
|
|
: sample_block(sample_block_)
|
2019-05-23 09:03:39 +00:00
|
|
|
{
|
2021-04-12 17:07:01 +00:00
|
|
|
write_buf = wrapWriteBufferWithCompressionMethod(
|
|
|
|
std::make_unique<WriteBufferFromS3>(client, bucket, key, min_upload_part_size, max_single_part_upload_size), compression_method, 3);
|
|
|
|
writer = FormatFactory::instance().getOutputStreamParallelIfPossible(format, *write_buf, sample_block, context);
|
|
|
|
}
|
2019-05-31 07:27:14 +00:00
|
|
|
|
2021-04-12 17:07:01 +00:00
|
|
|
Block getHeader() const override
|
|
|
|
{
|
|
|
|
return sample_block;
|
|
|
|
}
|
2019-05-31 07:27:14 +00:00
|
|
|
|
2021-04-12 17:07:01 +00:00
|
|
|
void write(const Block & block) override
|
|
|
|
{
|
|
|
|
writer->write(block);
|
|
|
|
}
|
2019-05-31 07:27:14 +00:00
|
|
|
|
2021-04-12 17:07:01 +00:00
|
|
|
void writePrefix() override
|
|
|
|
{
|
|
|
|
writer->writePrefix();
|
|
|
|
}
|
2019-05-31 07:27:14 +00:00
|
|
|
|
2021-04-12 17:07:01 +00:00
|
|
|
void flush() override
|
|
|
|
{
|
|
|
|
writer->flush();
|
|
|
|
}
|
2021-03-11 12:16:52 +00:00
|
|
|
|
2021-04-12 17:07:01 +00:00
|
|
|
void writeSuffix() override
|
|
|
|
{
|
2021-04-25 09:45:22 +00:00
|
|
|
try
|
|
|
|
{
|
|
|
|
writer->writeSuffix();
|
|
|
|
writer->flush();
|
|
|
|
write_buf->finalize();
|
|
|
|
}
|
|
|
|
catch (...)
|
|
|
|
{
|
|
|
|
/// Stop ParallelFormattingOutputFormat correctly.
|
|
|
|
writer.reset();
|
|
|
|
throw;
|
|
|
|
}
|
2021-04-12 17:07:01 +00:00
|
|
|
}
|
2019-05-31 07:27:14 +00:00
|
|
|
|
2021-04-12 17:07:01 +00:00
|
|
|
private:
|
|
|
|
Block sample_block;
|
|
|
|
std::unique_ptr<WriteBuffer> write_buf;
|
|
|
|
BlockOutputStreamPtr writer;
|
|
|
|
};
|
2019-05-23 09:03:39 +00:00
|
|
|
|
|
|
|
|
2019-12-04 16:06:55 +00:00
|
|
|
StorageS3::StorageS3(
|
|
|
|
const S3::URI & uri_,
|
2019-11-04 19:57:03 +00:00
|
|
|
const String & access_key_id_,
|
|
|
|
const String & secret_access_key_,
|
2019-12-04 16:06:55 +00:00
|
|
|
const StorageID & table_id_,
|
2019-09-22 22:13:42 +00:00
|
|
|
const String & format_name_,
|
2021-05-19 21:42:25 +00:00
|
|
|
UInt64 max_single_read_retries_,
|
2019-09-23 12:41:59 +00:00
|
|
|
UInt64 min_upload_part_size_,
|
2020-12-09 14:09:04 +00:00
|
|
|
UInt64 max_single_part_upload_size_,
|
2020-12-03 04:18:12 +00:00
|
|
|
UInt64 max_connections_,
|
2019-09-22 22:13:42 +00:00
|
|
|
const ColumnsDescription & columns_,
|
|
|
|
const ConstraintsDescription & constraints_,
|
2021-04-23 12:18:23 +00:00
|
|
|
const String & comment,
|
2021-04-10 23:33:54 +00:00
|
|
|
ContextPtr context_,
|
2021-04-13 20:17:25 +00:00
|
|
|
const String & compression_method_,
|
|
|
|
bool distributed_processing_)
|
2020-04-27 13:55:30 +00:00
|
|
|
: IStorage(table_id_)
|
2021-03-16 18:41:29 +00:00
|
|
|
, client_auth{uri_, access_key_id_, secret_access_key_, max_connections_, {}, {}} /// Client and settings will be updated later
|
2019-09-22 22:13:42 +00:00
|
|
|
, format_name(format_name_)
|
2021-05-19 21:42:25 +00:00
|
|
|
, max_single_read_retries(max_single_read_retries_)
|
2019-09-23 12:41:59 +00:00
|
|
|
, min_upload_part_size(min_upload_part_size_)
|
2020-12-09 14:09:04 +00:00
|
|
|
, max_single_part_upload_size(max_single_part_upload_size_)
|
2019-11-19 12:46:07 +00:00
|
|
|
, compression_method(compression_method_)
|
2020-07-13 14:13:30 +00:00
|
|
|
, name(uri_.storage_name)
|
2021-04-13 20:17:25 +00:00
|
|
|
, distributed_processing(distributed_processing_)
|
2019-09-22 22:13:42 +00:00
|
|
|
{
|
2021-03-16 18:41:29 +00:00
|
|
|
context_->getGlobalContext()->getRemoteHostFilter().checkURL(uri_.uri);
|
2020-06-19 15:39:41 +00:00
|
|
|
StorageInMemoryMetadata storage_metadata;
|
|
|
|
storage_metadata.setColumns(columns_);
|
|
|
|
storage_metadata.setConstraints(constraints_);
|
2021-04-23 12:18:23 +00:00
|
|
|
storage_metadata.setComment(comment);
|
2020-06-19 15:39:41 +00:00
|
|
|
setInMemoryMetadata(storage_metadata);
|
2021-03-16 18:41:29 +00:00
|
|
|
updateClientAndAuthSettings(context_, client_auth);
|
2019-09-22 22:13:42 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2020-08-03 13:54:14 +00:00
|
|
|
Pipe StorageS3::read(
|
2019-09-22 22:13:42 +00:00
|
|
|
const Names & column_names,
|
2020-06-16 14:25:08 +00:00
|
|
|
const StorageMetadataPtr & metadata_snapshot,
|
2020-09-20 17:52:17 +00:00
|
|
|
SelectQueryInfo & /*query_info*/,
|
2021-04-10 23:33:54 +00:00
|
|
|
ContextPtr local_context,
|
2019-06-01 21:18:20 +00:00
|
|
|
QueryProcessingStage::Enum /*processed_stage*/,
|
2019-05-23 09:03:39 +00:00
|
|
|
size_t max_block_size,
|
2020-01-26 13:12:15 +00:00
|
|
|
unsigned num_streams)
|
2019-05-23 09:03:39 +00:00
|
|
|
{
|
2021-03-16 18:41:29 +00:00
|
|
|
updateClientAndAuthSettings(local_context, client_auth);
|
2021-03-04 15:56:55 +00:00
|
|
|
|
2020-02-03 18:01:41 +00:00
|
|
|
Pipes pipes;
|
2020-01-26 14:05:51 +00:00
|
|
|
bool need_path_column = false;
|
|
|
|
bool need_file_column = false;
|
|
|
|
for (const auto & column : column_names)
|
|
|
|
{
|
|
|
|
if (column == "_path")
|
|
|
|
need_path_column = true;
|
|
|
|
if (column == "_file")
|
|
|
|
need_file_column = true;
|
|
|
|
}
|
2020-01-26 13:12:15 +00:00
|
|
|
|
2021-04-13 20:17:25 +00:00
|
|
|
std::shared_ptr<StorageS3Source::IteratorWrapper> iterator_wrapper{nullptr};
|
|
|
|
if (distributed_processing)
|
2021-04-10 02:21:18 +00:00
|
|
|
{
|
2021-04-13 20:17:25 +00:00
|
|
|
iterator_wrapper = std::make_shared<StorageS3Source::IteratorWrapper>(
|
|
|
|
[callback = local_context->getReadTaskCallback()]() -> String {
|
|
|
|
return callback();
|
|
|
|
});
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/// Iterate through disclosed globs and make a source for each file
|
|
|
|
auto glob_iterator = std::make_shared<StorageS3Source::DisclosedGlobIterator>(*client_auth.client, client_auth.uri);
|
|
|
|
iterator_wrapper = std::make_shared<StorageS3Source::IteratorWrapper>([glob_iterator]()
|
|
|
|
{
|
|
|
|
return glob_iterator->next();
|
|
|
|
});
|
|
|
|
}
|
2021-04-13 20:19:04 +00:00
|
|
|
|
2021-04-08 00:09:15 +00:00
|
|
|
for (size_t i = 0; i < num_streams; ++i)
|
2021-04-06 19:18:45 +00:00
|
|
|
{
|
2020-02-14 17:47:39 +00:00
|
|
|
pipes.emplace_back(std::make_shared<StorageS3Source>(
|
2020-01-26 14:05:51 +00:00
|
|
|
need_path_column,
|
|
|
|
need_file_column,
|
2020-01-26 13:12:15 +00:00
|
|
|
format_name,
|
|
|
|
getName(),
|
2020-06-16 14:25:08 +00:00
|
|
|
metadata_snapshot->getSampleBlock(),
|
2021-04-10 23:33:54 +00:00
|
|
|
local_context,
|
2020-10-02 12:38:50 +00:00
|
|
|
metadata_snapshot->getColumns(),
|
2020-01-26 13:12:15 +00:00
|
|
|
max_block_size,
|
2021-05-19 21:42:25 +00:00
|
|
|
max_single_read_retries,
|
2021-04-08 00:09:15 +00:00
|
|
|
compression_method,
|
2021-03-16 18:41:29 +00:00
|
|
|
client_auth.client,
|
|
|
|
client_auth.uri.bucket,
|
2021-04-10 02:21:18 +00:00
|
|
|
iterator_wrapper));
|
2021-04-06 19:18:45 +00:00
|
|
|
}
|
2021-04-10 14:58:29 +00:00
|
|
|
auto pipe = Pipe::unitePipes(std::move(pipes));
|
|
|
|
|
|
|
|
narrowPipe(pipe, num_streams);
|
|
|
|
return pipe;
|
2019-05-23 09:03:39 +00:00
|
|
|
}
|
|
|
|
|
2021-04-10 23:33:54 +00:00
|
|
|
BlockOutputStreamPtr StorageS3::write(const ASTPtr & /*query*/, const StorageMetadataPtr & metadata_snapshot, ContextPtr local_context)
|
2019-05-23 09:03:39 +00:00
|
|
|
{
|
2021-03-16 18:41:29 +00:00
|
|
|
updateClientAndAuthSettings(local_context, client_auth);
|
2019-05-29 12:54:31 +00:00
|
|
|
return std::make_shared<StorageS3BlockOutputStream>(
|
2020-12-09 14:09:04 +00:00
|
|
|
format_name,
|
|
|
|
metadata_snapshot->getSampleBlock(),
|
2021-04-12 19:35:26 +00:00
|
|
|
local_context,
|
2021-03-16 18:41:29 +00:00
|
|
|
chooseCompressionMethod(client_auth.uri.key, compression_method),
|
|
|
|
client_auth.client,
|
|
|
|
client_auth.uri.bucket,
|
|
|
|
client_auth.uri.key,
|
2020-12-09 14:09:04 +00:00
|
|
|
min_upload_part_size,
|
|
|
|
max_single_part_upload_size);
|
2019-05-23 09:03:39 +00:00
|
|
|
}
|
|
|
|
|
2021-03-16 18:41:29 +00:00
|
|
|
void StorageS3::updateClientAndAuthSettings(ContextPtr ctx, StorageS3::ClientAuthentificaiton & upd)
|
2021-03-04 15:56:55 +00:00
|
|
|
{
|
2021-03-16 18:41:29 +00:00
|
|
|
auto settings = ctx->getStorageS3Settings().getSettings(upd.uri.uri.toString());
|
|
|
|
if (upd.client && (!upd.access_key_id.empty() || settings == upd.auth_settings))
|
2021-03-04 15:56:55 +00:00
|
|
|
return;
|
|
|
|
|
2021-03-16 18:41:29 +00:00
|
|
|
Aws::Auth::AWSCredentials credentials(upd.access_key_id, upd.secret_access_key);
|
2021-03-04 15:56:55 +00:00
|
|
|
HeaderCollection headers;
|
2021-03-16 18:41:29 +00:00
|
|
|
if (upd.access_key_id.empty())
|
2021-03-04 15:56:55 +00:00
|
|
|
{
|
|
|
|
credentials = Aws::Auth::AWSCredentials(settings.access_key_id, settings.secret_access_key);
|
|
|
|
headers = settings.headers;
|
|
|
|
}
|
|
|
|
|
|
|
|
S3::PocoHTTPClientConfiguration client_configuration = S3::ClientFactory::instance().createClientConfiguration(
|
2021-05-01 17:41:16 +00:00
|
|
|
settings.region,
|
2021-03-16 18:41:29 +00:00
|
|
|
ctx->getRemoteHostFilter(), ctx->getGlobalContext()->getSettingsRef().s3_max_redirects);
|
2021-03-04 15:56:55 +00:00
|
|
|
|
2021-03-16 18:41:29 +00:00
|
|
|
client_configuration.endpointOverride = upd.uri.endpoint;
|
|
|
|
client_configuration.maxConnections = upd.max_connections;
|
2021-03-04 15:56:55 +00:00
|
|
|
|
2021-03-16 18:41:29 +00:00
|
|
|
upd.client = S3::ClientFactory::instance().create(
|
2021-03-04 15:56:55 +00:00
|
|
|
client_configuration,
|
2021-03-16 18:41:29 +00:00
|
|
|
upd.uri.is_virtual_hosted_style,
|
2021-03-04 15:56:55 +00:00
|
|
|
credentials.GetAWSAccessKeyId(),
|
|
|
|
credentials.GetAWSSecretKey(),
|
|
|
|
settings.server_side_encryption_customer_key_base64,
|
|
|
|
std::move(headers),
|
2021-03-18 07:36:11 +00:00
|
|
|
settings.use_environment_credentials.value_or(ctx->getConfigRef().getBool("s3.use_environment_credentials", false)),
|
2021-03-30 00:32:26 +00:00
|
|
|
settings.use_insecure_imds_request.value_or(ctx->getConfigRef().getBool("s3.use_insecure_imds_request", false)));
|
2021-03-04 15:56:55 +00:00
|
|
|
|
2021-03-16 18:41:29 +00:00
|
|
|
upd.auth_settings = std::move(settings);
|
2021-03-04 15:56:55 +00:00
|
|
|
}
|
|
|
|
|
2020-07-13 14:13:30 +00:00
|
|
|
void registerStorageS3Impl(const String & name, StorageFactory & factory)
|
2019-05-23 09:03:39 +00:00
|
|
|
{
|
2020-07-13 14:13:30 +00:00
|
|
|
factory.registerStorage(name, [](const StorageFactory::Arguments & args)
|
2019-06-17 00:06:14 +00:00
|
|
|
{
|
|
|
|
ASTs & engine_args = args.engine_args;
|
|
|
|
|
2019-11-04 19:57:03 +00:00
|
|
|
if (engine_args.size() < 2 || engine_args.size() > 5)
|
2019-06-17 00:06:14 +00:00
|
|
|
throw Exception(
|
2021-04-08 00:09:15 +00:00
|
|
|
"Storage S3 requires 2 to 5 arguments: url, [access_key_id, secret_access_key], name of used format and [compression_method].",
|
|
|
|
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
|
2019-06-17 00:06:14 +00:00
|
|
|
|
2020-03-09 01:22:33 +00:00
|
|
|
for (auto & engine_arg : engine_args)
|
2021-04-10 23:33:54 +00:00
|
|
|
engine_arg = evaluateConstantExpressionOrIdentifierAsLiteral(engine_arg, args.getLocalContext());
|
2019-06-17 00:06:14 +00:00
|
|
|
|
|
|
|
String url = engine_args[0]->as<ASTLiteral &>().value.safeGet<String>();
|
2019-12-06 14:37:21 +00:00
|
|
|
Poco::URI uri (url);
|
|
|
|
S3::URI s3_uri (uri);
|
2019-06-17 00:06:14 +00:00
|
|
|
|
2019-11-04 19:57:03 +00:00
|
|
|
String access_key_id;
|
|
|
|
String secret_access_key;
|
|
|
|
if (engine_args.size() >= 4)
|
|
|
|
{
|
|
|
|
access_key_id = engine_args[1]->as<ASTLiteral &>().value.safeGet<String>();
|
|
|
|
secret_access_key = engine_args[2]->as<ASTLiteral &>().value.safeGet<String>();
|
|
|
|
}
|
2019-06-17 00:06:14 +00:00
|
|
|
|
2021-05-19 21:42:25 +00:00
|
|
|
UInt64 max_single_read_retries = args.getLocalContext()->getSettingsRef().s3_max_single_read_retries;
|
2021-04-10 23:33:54 +00:00
|
|
|
UInt64 min_upload_part_size = args.getLocalContext()->getSettingsRef().s3_min_upload_part_size;
|
|
|
|
UInt64 max_single_part_upload_size = args.getLocalContext()->getSettingsRef().s3_max_single_part_upload_size;
|
|
|
|
UInt64 max_connections = args.getLocalContext()->getSettingsRef().s3_max_connections;
|
2019-09-23 12:41:59 +00:00
|
|
|
|
2019-11-19 12:46:07 +00:00
|
|
|
String compression_method;
|
2020-09-28 07:37:24 +00:00
|
|
|
String format_name;
|
2019-11-04 19:57:03 +00:00
|
|
|
if (engine_args.size() == 3 || engine_args.size() == 5)
|
2020-09-28 07:37:24 +00:00
|
|
|
{
|
2019-11-04 19:57:03 +00:00
|
|
|
compression_method = engine_args.back()->as<ASTLiteral &>().value.safeGet<String>();
|
2020-09-28 07:37:24 +00:00
|
|
|
format_name = engine_args[engine_args.size() - 2]->as<ASTLiteral &>().value.safeGet<String>();
|
|
|
|
}
|
2019-11-04 19:57:03 +00:00
|
|
|
else
|
2020-09-28 07:37:24 +00:00
|
|
|
{
|
2019-11-04 19:57:03 +00:00
|
|
|
compression_method = "auto";
|
2020-09-28 07:37:24 +00:00
|
|
|
format_name = engine_args.back()->as<ASTLiteral &>().value.safeGet<String>();
|
|
|
|
}
|
2019-11-19 12:46:07 +00:00
|
|
|
|
2020-09-30 12:04:21 +00:00
|
|
|
return StorageS3::create(
|
|
|
|
s3_uri,
|
|
|
|
access_key_id,
|
|
|
|
secret_access_key,
|
|
|
|
args.table_id,
|
|
|
|
format_name,
|
2021-05-19 21:42:25 +00:00
|
|
|
max_single_read_retries,
|
2020-09-30 12:04:21 +00:00
|
|
|
min_upload_part_size,
|
2020-12-09 14:09:04 +00:00
|
|
|
max_single_part_upload_size,
|
2020-12-03 04:18:12 +00:00
|
|
|
max_connections,
|
2020-09-30 12:04:21 +00:00
|
|
|
args.columns,
|
|
|
|
args.constraints,
|
2021-04-23 12:18:23 +00:00
|
|
|
args.comment,
|
2021-04-10 23:33:54 +00:00
|
|
|
args.getContext(),
|
2021-04-23 12:18:23 +00:00
|
|
|
compression_method);
|
2020-04-06 05:19:40 +00:00
|
|
|
},
|
|
|
|
{
|
|
|
|
.source_access_type = AccessType::S3,
|
2019-06-17 00:06:14 +00:00
|
|
|
});
|
2019-05-23 09:03:39 +00:00
|
|
|
}
|
2019-12-11 14:21:48 +00:00
|
|
|
|
2020-07-10 07:26:55 +00:00
|
|
|
void registerStorageS3(StorageFactory & factory)
|
|
|
|
{
|
2020-07-13 14:13:30 +00:00
|
|
|
return registerStorageS3Impl("S3", factory);
|
2020-07-10 07:26:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void registerStorageCOS(StorageFactory & factory)
|
|
|
|
{
|
2020-07-13 14:13:30 +00:00
|
|
|
return registerStorageS3Impl("COSN", factory);
|
2020-07-10 07:26:55 +00:00
|
|
|
}
|
|
|
|
|
2020-04-28 10:38:57 +00:00
|
|
|
NamesAndTypesList StorageS3::getVirtuals() const
|
2020-04-27 13:55:30 +00:00
|
|
|
{
|
2020-04-28 10:38:57 +00:00
|
|
|
return NamesAndTypesList{
|
2020-04-27 13:55:30 +00:00
|
|
|
{"_path", std::make_shared<DataTypeString>()},
|
|
|
|
{"_file", std::make_shared<DataTypeString>()}
|
|
|
|
};
|
|
|
|
}
|
|
|
|
|
2019-05-23 09:03:39 +00:00
|
|
|
}
|
2019-12-11 14:21:48 +00:00
|
|
|
|
2019-12-09 12:36:06 +00:00
|
|
|
#endif
|