diff options
| author | max42 <[email protected]> | 2023-07-29 00:02:16 +0300 |
|---|---|---|
| committer | max42 <[email protected]> | 2023-07-29 00:02:16 +0300 |
| commit | 73b89de71748a21e102d27b9f3ed1bf658766cb5 (patch) | |
| tree | 188bbd2d622fa91cdcbb1b6d6d77fbc84a0646f5 /yt/cpp/mapreduce/client/client_reader.cpp | |
| parent | 528e321bcc2a2b67b53aeba58c3bd88305a141ee (diff) | |
YT-19210: expose YQL shared library for YT.
After this, a new target libyqlplugin.so appears. in open-source cmake build.
Diff in open-source YDB repo looks like the following: https://paste.yandex-team.ru/f302bdb4-7ef2-4362-91c7-6ca45f329264
Diffstat (limited to 'yt/cpp/mapreduce/client/client_reader.cpp')
| -rw-r--r-- | yt/cpp/mapreduce/client/client_reader.cpp | 232 |
1 files changed, 232 insertions, 0 deletions
diff --git a/yt/cpp/mapreduce/client/client_reader.cpp b/yt/cpp/mapreduce/client/client_reader.cpp new file mode 100644 index 00000000000..80759b12dcc --- /dev/null +++ b/yt/cpp/mapreduce/client/client_reader.cpp @@ -0,0 +1,232 @@ +#include "client_reader.h" + +#include "structured_table_formats.h" +#include "transaction.h" +#include "transaction_pinger.h" + +#include <yt/cpp/mapreduce/common/helpers.h> +#include <yt/cpp/mapreduce/common/retry_lib.h> +#include <yt/cpp/mapreduce/common/wait_proxy.h> + +#include <yt/cpp/mapreduce/interface/config.h> +#include <yt/cpp/mapreduce/interface/tvm.h> + +#include <yt/cpp/mapreduce/interface/logging/yt_log.h> + +#include <yt/cpp/mapreduce/io/helpers.h> +#include <yt/cpp/mapreduce/io/yamr_table_reader.h> + +#include <yt/cpp/mapreduce/http/helpers.h> +#include <yt/cpp/mapreduce/http/requests.h> +#include <yt/cpp/mapreduce/http/retry_request.h> + +#include <yt/cpp/mapreduce/raw_client/raw_requests.h> + +#include <library/cpp/yson/node/serialize.h> + +#include <util/random/random.h> +#include <util/stream/file.h> +#include <util/stream/str.h> +#include <util/string/builder.h> +#include <util/string/cast.h> + +namespace NYT { + +using ::ToString; + +//////////////////////////////////////////////////////////////////////////////// + +TClientReader::TClientReader( + const TRichYPath& path, + IClientRetryPolicyPtr clientRetryPolicy, + ITransactionPingerPtr transactionPinger, + const TClientContext& context, + const TTransactionId& transactionId, + const TFormat& format, + const TTableReaderOptions& options, + bool useFormatFromTableAttributes) + : Path_(path) + , ClientRetryPolicy_(std::move(clientRetryPolicy)) + , Context_(context) + , ParentTransactionId_(transactionId) + , Format_(format) + , Options_(options) + , ReadTransaction_(nullptr) +{ + if (options.CreateTransaction_) { + Y_VERIFY(transactionPinger, "Internal error: transactionPinger is null"); + ReadTransaction_ = MakeHolder<TPingableTransaction>( + ClientRetryPolicy_, + Context_, + transactionId, + transactionPinger->GetChildTxPinger(), + TStartTransactionOptions()); + Path_.Path(Snapshot( + ClientRetryPolicy_, + Context_, + ReadTransaction_->GetId(), + path.Path_)); + } + + if (useFormatFromTableAttributes) { + auto transactionId2 = ReadTransaction_ ? ReadTransaction_->GetId() : ParentTransactionId_; + auto newFormat = GetTableFormat(ClientRetryPolicy_, Context_, transactionId2, Path_); + if (newFormat) { + Format_->Config = *newFormat; + } + } + + TransformYPath(); + CreateRequest(); +} + +bool TClientReader::Retry( + const TMaybe<ui32>& rangeIndex, + const TMaybe<ui64>& rowIndex) +{ + if (CurrentRequestRetryPolicy_) { + // TODO we should pass actual exception in Retry function + yexception genericError; + auto backoff = CurrentRequestRetryPolicy_->OnGenericError(genericError); + if (!backoff) { + return false; + } + } + + try { + CreateRequest(rangeIndex, rowIndex); + return true; + } catch (const std::exception& ex) { + YT_LOG_ERROR("Client reader retry failed: %v", + ex.what()); + + return false; + } +} + +void TClientReader::ResetRetries() +{ + CurrentRequestRetryPolicy_ = nullptr; +} + +size_t TClientReader::DoRead(void* buf, size_t len) +{ + return Input_->Read(buf, len); +} + +void TClientReader::TransformYPath() +{ + for (auto& range : Path_.MutableRangesView()) { + auto& exact = range.Exact_; + if (IsTrivial(exact)) { + continue; + } + + if (exact.RowIndex_) { + range.LowerLimit(TReadLimit().RowIndex(*exact.RowIndex_)); + range.UpperLimit(TReadLimit().RowIndex(*exact.RowIndex_ + 1)); + exact.RowIndex_.Clear(); + + } else if (exact.Key_) { + range.LowerLimit(TReadLimit().Key(*exact.Key_)); + + auto lastPart = TNode::CreateEntity(); + lastPart.Attributes() = TNode()("type", "max"); + exact.Key_->Parts_.push_back(lastPart); + + range.UpperLimit(TReadLimit().Key(*exact.Key_)); + exact.Key_.Clear(); + } + } +} + +void TClientReader::CreateRequest(const TMaybe<ui32>& rangeIndex, const TMaybe<ui64>& rowIndex) +{ + if (!CurrentRequestRetryPolicy_) { + CurrentRequestRetryPolicy_ = ClientRetryPolicy_->CreatePolicyForGenericRequest(); + } + while (true) { + CurrentRequestRetryPolicy_->NotifyNewAttempt(); + + THttpHeader header("GET", GetReadTableCommand(Context_.Config->ApiVersion)); + if (Context_.ServiceTicketAuth) { + header.SetServiceTicket(Context_.ServiceTicketAuth->Ptr->IssueServiceTicket()); + } else { + header.SetToken(Context_.Token); + } + auto transactionId = (ReadTransaction_ ? ReadTransaction_->GetId() : ParentTransactionId_); + header.AddTransactionId(transactionId); + + const auto& controlAttributes = Options_.ControlAttributes_; + header.AddParameter("control_attributes", TNode() + ("enable_row_index", controlAttributes.EnableRowIndex_) + ("enable_range_index", controlAttributes.EnableRangeIndex_)); + header.SetOutputFormat(Format_); + + header.SetResponseCompression(ToString(Context_.Config->AcceptEncoding)); + + if (rowIndex.Defined()) { + auto& ranges = Path_.MutableRanges(); + if (ranges.Empty()) { + ranges.ConstructInPlace(TVector{TReadRange()}); + } else { + if (rangeIndex.GetOrElse(0) >= ranges->size()) { + ythrow yexception() + << "range index " << rangeIndex.GetOrElse(0) + << " is out of range, input range count is " << ranges->size(); + } + ranges->erase(ranges->begin(), ranges->begin() + rangeIndex.GetOrElse(0)); + } + ranges->begin()->LowerLimit(TReadLimit().RowIndex(*rowIndex)); + } + + header.MergeParameters(FormIORequestParameters(Path_, Options_)); + + auto requestId = CreateGuidAsString(); + + try { + const auto proxyName = GetProxyForHeavyRequest(Context_); + Response_ = Context_.HttpClient->Request(GetFullUrl(proxyName, Context_, header), requestId, header); + + Input_ = Response_->GetResponseStream(); + + YT_LOG_DEBUG("RSP %v - table stream", requestId); + + return; + } catch (const TErrorResponse& e) { + LogRequestError( + requestId, + header, + e.what(), + CurrentRequestRetryPolicy_->GetAttemptDescription()); + + if (!IsRetriable(e)) { + throw; + } + auto backoff = CurrentRequestRetryPolicy_->OnRetriableError(e); + if (!backoff) { + throw; + } + NDetail::TWaitProxy::Get()->Sleep(*backoff); + } catch (const std::exception& e) { + LogRequestError( + requestId, + header, + e.what(), + CurrentRequestRetryPolicy_->GetAttemptDescription()); + + Response_.reset(); + Input_ = nullptr; + + auto backoff = CurrentRequestRetryPolicy_->OnGenericError(e); + if (!backoff) { + throw; + } + NDetail::TWaitProxy::Get()->Sleep(*backoff); + } + } +} + +//////////////////////////////////////////////////////////////////////////////// + +} // namespace NYT |
