2018-11-28 11:37:12 +00:00
|
|
|
#include "HTTPDictionarySource.h"
|
2016-11-15 19:51:06 +00:00
|
|
|
|
2018-12-10 15:25:45 +00:00
|
|
|
#include <DataStreams/IBlockOutputStream.h>
|
2017-05-25 19:26:17 +00:00
|
|
|
#include <DataStreams/OwningBlockInputStream.h>
|
2018-12-10 15:25:45 +00:00
|
|
|
#include <IO/ConnectionTimeouts.h>
|
2017-04-01 09:19:00 +00:00
|
|
|
#include <IO/ReadWriteBufferFromHTTP.h>
|
|
|
|
#include <IO/WriteBufferFromOStream.h>
|
2018-12-10 15:25:45 +00:00
|
|
|
#include <Interpreters/Context.h>
|
|
|
|
#include <Poco/Net/HTTPRequest.h>
|
2017-01-21 04:24:28 +00:00
|
|
|
#include <common/logger_useful.h>
|
2018-11-28 11:37:12 +00:00
|
|
|
#include "DictionarySourceFactory.h"
|
2018-12-10 15:25:45 +00:00
|
|
|
#include "DictionarySourceHelpers.h"
|
2018-11-28 11:37:12 +00:00
|
|
|
#include "DictionaryStructure.h"
|
|
|
|
|
2017-01-21 04:24:28 +00:00
|
|
|
|
2016-11-15 19:51:06 +00:00
|
|
|
namespace DB
|
|
|
|
{
|
2016-12-08 02:49:04 +00:00
|
|
|
static const size_t max_block_size = 8192;
|
|
|
|
|
|
|
|
|
2018-12-10 15:25:45 +00:00
|
|
|
HTTPDictionarySource::HTTPDictionarySource(
|
|
|
|
const DictionaryStructure & dict_struct_,
|
|
|
|
const Poco::Util::AbstractConfiguration & config,
|
|
|
|
const std::string & config_prefix,
|
|
|
|
Block & sample_block,
|
|
|
|
const Context & context)
|
|
|
|
: log(&Logger::get("HTTPDictionarySource"))
|
|
|
|
, update_time{std::chrono::system_clock::from_time_t(0)}
|
|
|
|
, dict_struct{dict_struct_}
|
|
|
|
, url{config.getString(config_prefix + ".url", "")}
|
|
|
|
, update_field{config.getString(config_prefix + ".update_field", "")}
|
|
|
|
, format{config.getString(config_prefix + ".format")}
|
|
|
|
, sample_block{sample_block}
|
|
|
|
, context(context)
|
|
|
|
, timeouts(ConnectionTimeouts::getHTTPTimeouts(context.getSettingsRef()))
|
2016-11-15 19:51:06 +00:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
2017-01-21 04:24:28 +00:00
|
|
|
HTTPDictionarySource::HTTPDictionarySource(const HTTPDictionarySource & other)
|
2018-12-10 15:25:45 +00:00
|
|
|
: log(&Logger::get("HTTPDictionarySource"))
|
|
|
|
, update_time{other.update_time}
|
|
|
|
, dict_struct{other.dict_struct}
|
|
|
|
, url{other.url}
|
|
|
|
, update_field{other.update_field}
|
|
|
|
, format{other.format}
|
|
|
|
, sample_block{other.sample_block}
|
|
|
|
, context(other.context)
|
|
|
|
, timeouts(ConnectionTimeouts::getHTTPTimeouts(context.getSettingsRef()))
|
2016-11-15 19:51:06 +00:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
2018-05-16 13:22:27 +00:00
|
|
|
void HTTPDictionarySource::getUpdateFieldAndDate(Poco::URI & uri)
|
2018-01-15 12:44:39 +00:00
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
if (update_time != std::chrono::system_clock::from_time_t(0))
|
2018-01-15 12:44:39 +00:00
|
|
|
{
|
|
|
|
auto tmp_time = update_time;
|
|
|
|
update_time = std::chrono::system_clock::now();
|
2018-02-15 13:08:23 +00:00
|
|
|
time_t hr_time = std::chrono::system_clock::to_time_t(tmp_time) - 1;
|
2018-12-10 15:25:45 +00:00
|
|
|
char buffer[80];
|
2018-01-15 12:44:39 +00:00
|
|
|
struct tm * timeinfo;
|
2018-12-10 15:25:45 +00:00
|
|
|
timeinfo = localtime(&hr_time);
|
2018-05-16 13:22:27 +00:00
|
|
|
strftime(buffer, 80, "%Y-%m-%d %H:%M:%S", timeinfo);
|
2018-01-15 12:44:39 +00:00
|
|
|
std::string str_time(buffer);
|
2018-05-16 13:22:27 +00:00
|
|
|
uri.addQueryParameter(update_field, str_time);
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
update_time = std::chrono::system_clock::now();
|
2018-05-16 13:22:27 +00:00
|
|
|
uri.addQueryParameter(update_field, "0000-00-00 00:00:00");
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-11-15 19:51:06 +00:00
|
|
|
BlockInputStreamPtr HTTPDictionarySource::loadAll()
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
LOG_TRACE(log, "loadAll " + toString());
|
2018-02-15 13:08:23 +00:00
|
|
|
Poco::URI uri(url);
|
2018-12-10 15:25:45 +00:00
|
|
|
auto in_ptr = std::make_unique<ReadWriteBufferFromHTTP>(
|
|
|
|
uri, Poco::Net::HTTPRequest::HTTP_GET, ReadWriteBufferFromHTTP::OutStreamCallback(), timeouts);
|
2017-05-25 19:21:57 +00:00
|
|
|
auto input_stream = context.getInputFormat(format, *in_ptr, sample_block, max_block_size);
|
2018-02-15 13:08:23 +00:00
|
|
|
return std::make_shared<OwningBlockInputStream<ReadWriteBufferFromHTTP>>(input_stream, std::move(in_ptr));
|
|
|
|
}
|
2018-01-15 12:44:39 +00:00
|
|
|
|
2018-02-15 13:08:23 +00:00
|
|
|
BlockInputStreamPtr HTTPDictionarySource::loadUpdatedAll()
|
|
|
|
{
|
2018-05-16 13:22:27 +00:00
|
|
|
Poco::URI uri(url);
|
|
|
|
getUpdateFieldAndDate(uri);
|
|
|
|
LOG_TRACE(log, "loadUpdatedAll " + uri.toString());
|
2018-12-10 15:25:45 +00:00
|
|
|
auto in_ptr = std::make_unique<ReadWriteBufferFromHTTP>(
|
|
|
|
uri, Poco::Net::HTTPRequest::HTTP_GET, ReadWriteBufferFromHTTP::OutStreamCallback(), timeouts);
|
2018-02-15 13:08:23 +00:00
|
|
|
auto input_stream = context.getInputFormat(format, *in_ptr, sample_block, max_block_size);
|
2017-05-25 19:21:57 +00:00
|
|
|
return std::make_shared<OwningBlockInputStream<ReadWriteBufferFromHTTP>>(input_stream, std::move(in_ptr));
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
BlockInputStreamPtr HTTPDictionarySource::loadIds(const std::vector<UInt64> & ids)
|
|
|
|
{
|
2017-05-25 19:27:23 +00:00
|
|
|
LOG_TRACE(log, "loadIds " << toString() << " size = " << ids.size());
|
2016-11-19 00:07:58 +00:00
|
|
|
|
2017-05-25 19:21:57 +00:00
|
|
|
ReadWriteBufferFromHTTP::OutStreamCallback out_stream_callback = [&](std::ostream & ostr)
|
|
|
|
{
|
|
|
|
WriteBufferFromOStream out_buffer(ostr);
|
|
|
|
auto output_stream = context.getOutputFormat(format, out_buffer, sample_block);
|
|
|
|
formatIDs(output_stream, ids);
|
2017-04-01 07:20:54 +00:00
|
|
|
};
|
2016-11-24 19:57:24 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
Poco::URI uri(url);
|
2018-12-10 15:25:45 +00:00
|
|
|
auto in_ptr = std::make_unique<ReadWriteBufferFromHTTP>(uri, Poco::Net::HTTPRequest::HTTP_POST, out_stream_callback, timeouts);
|
2017-05-25 19:21:57 +00:00
|
|
|
auto input_stream = context.getInputFormat(format, *in_ptr, sample_block, max_block_size);
|
|
|
|
return std::make_shared<OwningBlockInputStream<ReadWriteBufferFromHTTP>>(input_stream, std::move(in_ptr));
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|
|
|
|
|
2018-12-10 15:25:45 +00:00
|
|
|
BlockInputStreamPtr HTTPDictionarySource::loadKeys(const Columns & key_columns, const std::vector<size_t> & requested_rows)
|
2016-11-15 19:51:06 +00:00
|
|
|
{
|
2017-05-25 19:27:23 +00:00
|
|
|
LOG_TRACE(log, "loadKeys " << toString() << " size = " << requested_rows.size());
|
2016-11-22 15:03:54 +00:00
|
|
|
|
2017-05-25 19:21:57 +00:00
|
|
|
ReadWriteBufferFromHTTP::OutStreamCallback out_stream_callback = [&](std::ostream & ostr)
|
|
|
|
{
|
|
|
|
WriteBufferFromOStream out_buffer(ostr);
|
|
|
|
auto output_stream = context.getOutputFormat(format, out_buffer, sample_block);
|
2017-05-25 20:17:23 +00:00
|
|
|
formatKeys(dict_struct, output_stream, key_columns, requested_rows);
|
2017-04-01 07:20:54 +00:00
|
|
|
};
|
2016-11-22 15:03:54 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
Poco::URI uri(url);
|
2018-12-10 15:25:45 +00:00
|
|
|
auto in_ptr = std::make_unique<ReadWriteBufferFromHTTP>(uri, Poco::Net::HTTPRequest::HTTP_POST, out_stream_callback, timeouts);
|
2017-05-25 19:21:57 +00:00
|
|
|
auto input_stream = context.getInputFormat(format, *in_ptr, sample_block, max_block_size);
|
|
|
|
return std::make_shared<OwningBlockInputStream<ReadWriteBufferFromHTTP>>(input_stream, std::move(in_ptr));
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
bool HTTPDictionarySource::isModified() const
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
return true;
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
bool HTTPDictionarySource::supportsSelectiveLoad() const
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
return true;
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|
|
|
|
|
2018-01-15 12:44:39 +00:00
|
|
|
bool HTTPDictionarySource::hasUpdateField() const
|
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
return !update_field.empty();
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
|
|
|
|
2016-11-15 19:51:06 +00:00
|
|
|
DictionarySourcePtr HTTPDictionarySource::clone() const
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
return std::make_unique<HTTPDictionarySource>(*this);
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
std::string HTTPDictionarySource::toString() const
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
Poco::URI uri(url);
|
|
|
|
return uri.toString();
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|
|
|
|
|
2018-11-28 11:37:12 +00:00
|
|
|
void registerDictionarySourceHTTP(DictionarySourceFactory & factory)
|
|
|
|
{
|
|
|
|
auto createTableSource = [=](const DictionaryStructure & dict_struct,
|
|
|
|
const Poco::Util::AbstractConfiguration & config,
|
|
|
|
const std::string & config_prefix,
|
|
|
|
Block & sample_block,
|
|
|
|
const Context & context) -> DictionarySourcePtr {
|
|
|
|
if (dict_struct.has_expressions)
|
2018-12-10 15:25:45 +00:00
|
|
|
throw Exception{"Dictionary source of type `http` does not support attribute expressions", ErrorCodes::LOGICAL_ERROR};
|
2018-11-28 11:37:12 +00:00
|
|
|
|
|
|
|
return std::make_unique<HTTPDictionarySource>(dict_struct, config, config_prefix + ".http", sample_block, context);
|
|
|
|
};
|
|
|
|
factory.registerSource("http", createTableSource);
|
|
|
|
}
|
|
|
|
|
2016-11-15 19:51:06 +00:00
|
|
|
}
|