yql_facade.cpp 70 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024
  1. #include "yql_facade.h"
  2. #include <yql/essentials/core/yql_execution.h>
  3. #include <yql/essentials/core/yql_expr_csee.h>
  4. #include <yql/essentials/core/yql_expr_optimize.h>
  5. #include <yql/essentials/core/yql_expr_type_annotation.h>
  6. #include <yql/essentials/core/yql_opt_rewrite_io.h>
  7. #include <yql/essentials/core/yql_opt_proposed_by_data.h>
  8. #include <yql/essentials/core/yql_gc_transformer.h>
  9. #include <yql/essentials/core/type_ann/type_ann_expr.h>
  10. #include <yql/essentials/core/services/yql_plan.h>
  11. #include <yql/essentials/core/services/yql_eval_params.h>
  12. #include <yql/essentials/utils/log/context.h>
  13. #include <yql/essentials/utils/log/profile.h>
  14. #include <yql/essentials/utils/limiting_allocator.h>
  15. #include <yql/essentials/core/services/yql_out_transformers.h>
  16. #include <yql/essentials/core/extract_predicate/extract_predicate_dbg.h>
  17. #include <yql/essentials/providers/common/provider/yql_provider_names.h>
  18. #include <yql/essentials/providers/common/provider/yql_provider.h>
  19. #include <yql/essentials/providers/common/udf_resolve/yql_simple_udf_resolver.h>
  20. #include <yql/essentials/providers/common/udf_resolve/yql_outproc_udf_resolver.h>
  21. #include <yql/essentials/providers/common/udf_resolve/yql_udf_resolver_with_index.h>
  22. #include <yql/essentials/providers/common/arrow_resolve/yql_simple_arrow_resolver.h>
  23. #include <yql/essentials/providers/common/config/yql_setting.h>
  24. #include <yql/essentials/core/qplayer/udf_resolver/yql_qplayer_udf_resolver.h>
  25. #include <library/cpp/yson/node/node_io.h>
  26. #include <library/cpp/deprecated/split/split_iterator.h>
  27. #include <library/cpp/yson/writer.h>
  28. #include <library/cpp/string_utils/base64/base64.h>
  29. #include <util/stream/file.h>
  30. #include <util/stream/null.h>
  31. #include <util/string/join.h>
  32. #include <util/string/split.h>
  33. #include <util/generic/guid.h>
  34. #include <util/system/rusage.h>
  35. #include <util/generic/yexception.h>
  36. using namespace NThreading;
  37. namespace NYql {
  38. namespace {
  39. const size_t DEFAULT_AST_BUF_SIZE = 1024;
  40. const size_t DEFAULT_PLAN_BUF_SIZE = 1024;
  41. const TString FacadeComponent = "Facade";
  42. const TString SourceCodeLabel = "SourceCode";
  43. const TString GatewaysLabel = "Gateways";
  44. const TString ParametersLabel = "Parameters";
  45. const TString TranslationLabel = "Translation";
  46. const TString StaticUserFilesLabel = "UserFiles";
  47. const TString DynamicUserFilesLabel = "DynamicUserFiles";
  48. const TString StaticCredentialsLabel = "Credentials";
  49. const TString DynamicCredentialsLabel = "DynamicCredentials";
  50. class TUrlLoader : public IUrlLoader {
  51. public:
  52. TUrlLoader(TFileStoragePtr storage)
  53. : Storage_(storage)
  54. {}
  55. TString Load(const TString& url, const TString& token) override {
  56. auto file = Storage_->PutUrl(url, token);
  57. return TFileInput(file->GetPath()).ReadAll();
  58. }
  59. private:
  60. TFileStoragePtr Storage_;
  61. };
  62. template <typename... Params1, typename... Params2>
  63. TProgram::TStatus SyncExecution(
  64. TProgram* program,
  65. TProgram::TFutureStatus (TProgram::*method)(Params1...),
  66. Params2&&... params) {
  67. TProgram::TFutureStatus future =
  68. (program->*method)(std::forward<Params2>(params)...);
  69. YQL_ENSURE(future.Initialized());
  70. future.Wait();
  71. HandleFutureException(future);
  72. TProgram::TStatus status = future.GetValue();
  73. while (status == TProgram::TStatus::Async) {
  74. auto continueFuture = program->ContinueAsync();
  75. continueFuture.Wait();
  76. HandleFutureException(continueFuture);
  77. status = continueFuture.GetValue();
  78. }
  79. if (status == TProgram::TStatus::Error) {
  80. program->Print(program->ExprStream(), program->PlanStream());
  81. }
  82. return status;
  83. }
  84. std::function<TString(const TString&, const TString&)> BuildDefaultTokenResolver(TCredentials::TPtr credentials) {
  85. return [credentials](const TString& /*url*/, const TString& alias) -> TString {
  86. if (alias) {
  87. if (auto cred = credentials->FindCredential(TString("default_").append(alias))) {
  88. return cred->Content;
  89. }
  90. }
  91. return {};
  92. };
  93. }
  94. std::function<TString(const TString&, const TString&)> BuildCompositeTokenResolver(TVector<std::function<TString(const TString&, const TString&)>>&& children) {
  95. if (children.empty()) {
  96. return {};
  97. }
  98. if (children.size() == 1) {
  99. return std::move(children[0]);
  100. }
  101. return [children = std::move(children)](const TString& url, const TString& alias) -> TString {
  102. for (auto& c : children) {
  103. if (auto r = c(url, alias)) {
  104. return r;
  105. }
  106. }
  107. return {};
  108. };
  109. }
  110. } // namspace
  111. ///////////////////////////////////////////////////////////////////////////////
  112. // TProgramFactory
  113. ///////////////////////////////////////////////////////////////////////////////
  114. TProgramFactory::TProgramFactory(
  115. bool useRepeatableRandomAndTimeProviders,
  116. const NKikimr::NMiniKQL::IFunctionRegistry* functionRegistry,
  117. ui64 nextUniqueId,
  118. const TVector<TDataProviderInitializer>& dataProvidersInit,
  119. const TString& runner)
  120. : UseRepeatableRandomAndTimeProviders_(useRepeatableRandomAndTimeProviders)
  121. , FunctionRegistry_(functionRegistry)
  122. , NextUniqueId_(nextUniqueId)
  123. , DataProvidersInit_(dataProvidersInit)
  124. , Credentials_(MakeIntrusive<TCredentials>())
  125. , GatewaysConfig_(nullptr)
  126. , Runner_(runner)
  127. , ArrowResolver_(MakeSimpleArrowResolver(*functionRegistry))
  128. {
  129. }
  130. void TProgramFactory::UnrepeatableRandom() {
  131. UseUnrepeatableRandom = true;
  132. }
  133. void TProgramFactory::EnableRangeComputeFor() {
  134. EnableRangeComputeFor_ = true;
  135. }
  136. void TProgramFactory::AddUserDataTable(const TUserDataTable& userDataTable) {
  137. for (const auto& p : userDataTable) {
  138. if (!UserDataTable_.emplace(p).second) {
  139. ythrow yexception() << "UserDataTable already has user data block with key " << p.first;
  140. }
  141. }
  142. }
  143. void TProgramFactory::SetCredentials(TCredentials::TPtr credentials) {
  144. Credentials_ = std::move(credentials);
  145. }
  146. void TProgramFactory::SetGatewaysConfig(const TGatewaysConfig* gatewaysConfig) {
  147. GatewaysConfig_ = gatewaysConfig;
  148. }
  149. void TProgramFactory::SetModules(IModuleResolver::TPtr modules) {
  150. Modules_ = modules;
  151. }
  152. void TProgramFactory::SetUdfResolver(IUdfResolver::TPtr udfResolver) {
  153. UdfResolver_ = udfResolver;
  154. }
  155. void TProgramFactory::SetUdfIndex(TUdfIndex::TPtr udfIndex, TUdfIndexPackageSet::TPtr udfIndexPackageSet) {
  156. UdfIndex_ = std::move(udfIndex);
  157. UdfIndexPackageSet_ = std::move(udfIndexPackageSet);
  158. }
  159. void TProgramFactory::SetFileStorage(TFileStoragePtr fileStorage) {
  160. FileStorage_ = std::move(fileStorage);
  161. }
  162. void TProgramFactory::SetUrlPreprocessing(IUrlPreprocessing::TPtr urlPreprocessing) {
  163. UrlPreprocessing_ = std::move(urlPreprocessing);
  164. }
  165. void TProgramFactory::SetArrowResolver(IArrowResolver::TPtr arrowResolver) {
  166. ArrowResolver_ = arrowResolver;
  167. }
  168. void TProgramFactory::SetUrlListerManager(IUrlListerManagerPtr urlListerManager) {
  169. UrlListerManager_ = std::move(urlListerManager);
  170. }
  171. TProgramPtr TProgramFactory::Create(
  172. const TFile& file,
  173. const TString& sessionId,
  174. const TQContext& qContext,
  175. TMaybe<TString> gatewaysForMerge)
  176. {
  177. TString sourceCode = TFileInput(file).ReadAll();
  178. return Create(file.GetName(), sourceCode, sessionId, EHiddenMode::Disable, qContext, gatewaysForMerge);
  179. }
  180. TProgramPtr TProgramFactory::Create(
  181. const TString& filename,
  182. const TString& sourceCode,
  183. const TString& sessionId,
  184. EHiddenMode hiddenMode,
  185. const TQContext& qContext,
  186. TMaybe<TString> gatewaysForMerge)
  187. {
  188. auto randomProvider = UseRepeatableRandomAndTimeProviders_ && !UseUnrepeatableRandom && hiddenMode == EHiddenMode::Disable ?
  189. CreateDeterministicRandomProvider(1) : CreateDefaultRandomProvider();
  190. auto timeProvider = UseRepeatableRandomAndTimeProviders_ ?
  191. CreateDeterministicTimeProvider(10000000) : CreateDefaultTimeProvider();
  192. TUdfIndex::TPtr udfIndex = UdfIndex_ ? UdfIndex_->Clone() : nullptr;
  193. TUdfIndexPackageSet::TPtr udfIndexPackageSet = (UdfIndexPackageSet_ && hiddenMode == EHiddenMode::Disable) ? UdfIndexPackageSet_->Clone() : nullptr;
  194. IModuleResolver::TPtr moduleResolver = Modules_ ? Modules_->CreateMutableChild() : nullptr;
  195. IUrlListerManagerPtr urlListerManager = UrlListerManager_ ? UrlListerManager_->Clone() : nullptr;
  196. auto udfResolver = udfIndex ? NCommon::CreateUdfResolverWithIndex(udfIndex, UdfResolver_, FileStorage_) : UdfResolver_;
  197. // make UserDataTable_ copy here
  198. return new TProgram(FunctionRegistry_, randomProvider, timeProvider, NextUniqueId_, DataProvidersInit_,
  199. UserDataTable_, Credentials_, moduleResolver, urlListerManager,
  200. udfResolver, udfIndex, udfIndexPackageSet, FileStorage_, UrlPreprocessing_,
  201. GatewaysConfig_, filename, sourceCode, sessionId, Runner_, EnableRangeComputeFor_, ArrowResolver_, hiddenMode,
  202. qContext, gatewaysForMerge);
  203. }
  204. ///////////////////////////////////////////////////////////////////////////////
  205. // TProgram
  206. ///////////////////////////////////////////////////////////////////////////////
  207. TProgram::TProgram(
  208. const NKikimr::NMiniKQL::IFunctionRegistry* functionRegistry,
  209. const TIntrusivePtr<IRandomProvider> randomProvider,
  210. const TIntrusivePtr<ITimeProvider> timeProvider,
  211. ui64 nextUniqueId,
  212. const TVector<TDataProviderInitializer>& dataProvidersInit,
  213. const TUserDataTable& userDataTable,
  214. const TCredentials::TPtr& credentials,
  215. const IModuleResolver::TPtr& modules,
  216. const IUrlListerManagerPtr& urlListerManager,
  217. const IUdfResolver::TPtr& udfResolver,
  218. const TUdfIndex::TPtr& udfIndex,
  219. const TUdfIndexPackageSet::TPtr& udfIndexPackageSet,
  220. const TFileStoragePtr& fileStorage,
  221. const IUrlPreprocessing::TPtr& urlPreprocessing,
  222. const TGatewaysConfig* gatewaysConfig,
  223. const TString& filename,
  224. const TString& sourceCode,
  225. const TString& sessionId,
  226. const TString& runner,
  227. bool enableRangeComputeFor,
  228. const IArrowResolver::TPtr& arrowResolver,
  229. EHiddenMode hiddenMode,
  230. const TQContext& qContext,
  231. TMaybe<TString> gatewaysForMerge
  232. )
  233. : FunctionRegistry_(functionRegistry)
  234. , RandomProvider_(randomProvider)
  235. , TimeProvider_(timeProvider)
  236. , NextUniqueId_(nextUniqueId)
  237. , AstRoot_(nullptr)
  238. , Modules_(modules)
  239. , DataProvidersInit_(dataProvidersInit)
  240. , Credentials_(MakeIntrusive<NYql::TCredentials>(*credentials))
  241. , UrlListerManager_(urlListerManager)
  242. , UdfResolver_(udfResolver)
  243. , UdfIndex_(udfIndex)
  244. , UdfIndexPackageSet_(udfIndexPackageSet)
  245. , FileStorage_(fileStorage)
  246. , SavedUserDataTable_(userDataTable)
  247. , GatewaysConfig_(gatewaysConfig)
  248. , Filename_(filename)
  249. , SourceCode_(sourceCode)
  250. , SourceSyntax_(ESourceSyntax::Unknown)
  251. , SyntaxVersion_(0)
  252. , ExprRoot_(nullptr)
  253. , SessionId_(sessionId)
  254. , ResultType_(IDataProvider::EResultFormat::Yson)
  255. , ResultFormat_(NYson::EYsonFormat::Binary)
  256. , OutputFormat_(NYson::EYsonFormat::Pretty)
  257. , EnableRangeComputeFor_(enableRangeComputeFor)
  258. , ArrowResolver_(arrowResolver)
  259. , HiddenMode_(hiddenMode)
  260. , QContext_(qContext)
  261. , GatewaysForMerge_(gatewaysForMerge)
  262. {
  263. if (SessionId_.empty()) {
  264. SessionId_ = CreateGuidAsString();
  265. }
  266. if (QContext_.CanWrite()) {
  267. NYT::TNode credListNode = NYT::TNode::CreateList();
  268. Credentials_->ForEach([&](const TString name, const TCredential& cred) {
  269. credListNode.Add(NYT::TNode()
  270. ("Name", name)
  271. ("Category", cred.Category)
  272. ("Subcategory", cred.Subcategory));
  273. });
  274. auto credList = NYT::NodeToYsonString(credListNode, NYT::NYson::EYsonFormat::Binary);
  275. QContext_.GetWriter()->Put({FacadeComponent, StaticCredentialsLabel}, credList).GetValueSync();
  276. } else if (QContext_.CanRead()) {
  277. Credentials_ = MakeIntrusive<TCredentials>();
  278. Credentials_->SetUserCredentials({
  279. .OauthToken = "REPLAY_OAUTH",
  280. .BlackboxSessionIdCookie = "REPLAY_SESSIONID"
  281. });
  282. for (const auto& label : {StaticCredentialsLabel, DynamicCredentialsLabel}) {
  283. auto item = QContext_.GetReader()->Get({FacadeComponent, label}).GetValueSync();
  284. if (item) {
  285. auto node = NYT::NodeFromYsonString(item->Value);
  286. for (const auto& c : node.AsList()) {
  287. Credentials_->AddCredential(c["Name"].AsString(), TCredential(
  288. c["Category"].AsString(),c["Subcategory"].AsString(),"REPLAY"));
  289. }
  290. }
  291. }
  292. }
  293. if (QContext_.CanWrite() && !SavedUserDataTable_.empty()) {
  294. NYT::TNode userFilesNode = NYT::TNode::CreateList();
  295. for (const auto& p : SavedUserDataTable_) {
  296. userFilesNode.Add(p.first.Alias());
  297. }
  298. auto userFiles = NYT::NodeToYsonString(userFilesNode, NYT::NYson::EYsonFormat::Binary);
  299. QContext_.GetWriter()->Put({FacadeComponent, StaticUserFilesLabel}, userFiles).GetValueSync();
  300. } else if (QContext_.CanRead()) {
  301. SavedUserDataTable_.clear();
  302. for (const auto& label : {StaticUserFilesLabel, DynamicUserFilesLabel}) {
  303. auto item = QContext_.GetReader()->Get({FacadeComponent, label}).GetValueSync();
  304. if (item) {
  305. auto node = NYT::NodeFromYsonString(item->Value);
  306. for (const auto& alias : node.AsList()) {
  307. TUserDataBlock block;
  308. block.Type = EUserDataType::RAW_INLINE_DATA;
  309. YQL_ENSURE(SavedUserDataTable_.emplace(TUserDataKey::File(alias.AsString()), block).second);
  310. }
  311. }
  312. }
  313. }
  314. UserDataStorage_ = MakeIntrusive<TUserDataStorage>(fileStorage, SavedUserDataTable_, udfResolver, udfIndex);
  315. if (auto modules = dynamic_cast<TModuleResolver*>(Modules_.get())) {
  316. modules->AttachUserData(UserDataStorage_);
  317. modules->SetUrlLoader(new TUrlLoader(FileStorage_));
  318. modules->SetCredentials(Credentials_);
  319. if (QContext_) {
  320. modules->SetQContext(QContext_);
  321. }
  322. }
  323. if (UrlListerManager_) {
  324. UrlListerManager_->SetCredentials(Credentials_);
  325. UrlListerManager_->SetUrlPreprocessing(urlPreprocessing);
  326. }
  327. OperationOptions_.Runner = runner;
  328. UserDataStorage_->SetUrlPreprocessor(urlPreprocessing);
  329. if (QContext_) {
  330. UdfResolver_ = NCommon::WrapUdfResolverWithQContext(UdfResolver_, QContext_);
  331. if (QContext_.CanRead()) {
  332. auto item = QContext_.GetReader()->Get({FacadeComponent, GatewaysLabel}).GetValueSync();
  333. if (item) {
  334. YQL_ENSURE(LoadedGatewaysConfig_.ParseFromString(item->Value));
  335. if (GatewaysForMerge_) {
  336. YQL_ENSURE(LoadedGatewaysConfig_.MergeFromString(*GatewaysForMerge_));
  337. }
  338. GatewaysConfig_ = &LoadedGatewaysConfig_;
  339. }
  340. } else if (QContext_.CanWrite() && GatewaysConfig_) {
  341. TGatewaysConfig cleaned;
  342. if (GatewaysConfig_->HasYt()) {
  343. cleaned.MutableYt()->CopyFrom(GatewaysConfig_->GetYt());
  344. }
  345. if (GatewaysConfig_->HasFs()) {
  346. cleaned.MutableFs()->CopyFrom(GatewaysConfig_->GetFs());
  347. }
  348. if (GatewaysConfig_->HasYqlCore()) {
  349. cleaned.MutableYqlCore()->CopyFrom(GatewaysConfig_->GetYqlCore());
  350. }
  351. if (GatewaysConfig_->HasSqlCore()) {
  352. cleaned.MutableSqlCore()->CopyFrom(GatewaysConfig_->GetSqlCore());
  353. }
  354. if (GatewaysConfig_->HasDq()) {
  355. cleaned.MutableDq()->CopyFrom(GatewaysConfig_->GetDq());
  356. }
  357. auto data = cleaned.SerializeAsString();
  358. QContext_.GetWriter()->Put({FacadeComponent, GatewaysLabel}, data).GetValueSync();
  359. }
  360. if (QContext_.CanRead()) {
  361. auto item = QContext_.GetReader()->Get({FacadeComponent, ParametersLabel}).GetValueSync();
  362. if (item) {
  363. SetParametersYson(item->Value);
  364. }
  365. }
  366. }
  367. }
  368. TProgram::~TProgram() {
  369. try {
  370. CloseLastSession().GetValueSync();
  371. // stop all non complete execution before deleting TExprCtx
  372. with_lock (DataProvidersLock_) {
  373. DataProviders_.clear();
  374. }
  375. } catch (...) {
  376. Cerr << CurrentExceptionMessage() << Endl;
  377. }
  378. }
  379. void TProgram::ConfigureYsonResultFormat(NYson::EYsonFormat format) {
  380. ResultFormat_ = format;
  381. OutputFormat_ = format;
  382. }
  383. void TProgram::SetValidateOptions(NUdf::EValidateMode validateMode) {
  384. Y_ENSURE(!TypeCtx_, "TypeCtx_ already created");
  385. ValidateMode_ = validateMode;
  386. }
  387. void TProgram::SetDisableNativeUdfSupport(bool disable) {
  388. Y_ENSURE(!TypeCtx_, "TypeCtx_ already created");
  389. DisableNativeUdfSupport_ = disable;
  390. }
  391. void TProgram::SetUseTableMetaFromGraph(bool use) {
  392. Y_ENSURE(!TypeCtx_, "TypeCtx_ already created");
  393. UseTableMetaFromGraph_ = use;
  394. }
  395. TTypeAnnotationContextPtr TProgram::GetAnnotationContext() const {
  396. Y_ENSURE(TypeCtx_, "TypeCtx_ is not created");
  397. return TypeCtx_;
  398. }
  399. TTypeAnnotationContextPtr TProgram::ProvideAnnotationContext(const TString& username) {
  400. if (!TypeCtx_) {
  401. TypeCtx_ = BuildTypeAnnotationContext(username);
  402. TypeCtx_->OperationOptions = OperationOptions_;
  403. TypeCtx_->ValidateMode = ValidateMode_;
  404. TypeCtx_->DisableNativeUdfSupport = DisableNativeUdfSupport_;
  405. TypeCtx_->UseTableMetaFromGraph = UseTableMetaFromGraph_;
  406. }
  407. return TypeCtx_;
  408. }
  409. IPlanBuilder& TProgram::GetPlanBuilder() {
  410. if (!PlanBuilder_) {
  411. PlanBuilder_ = CreatePlanBuilder(*GetAnnotationContext());
  412. }
  413. return *PlanBuilder_;
  414. }
  415. void TProgram::SetParametersYson(const TString& parameters) {
  416. Y_ENSURE(!TypeCtx_, "TypeCtx_ already created");
  417. auto node = NYT::NodeFromYsonString(parameters);
  418. YQL_ENSURE(node.IsMap());
  419. for (const auto& x : node.AsMap()) {
  420. YQL_ENSURE(x.second.IsMap());
  421. YQL_ENSURE(x.second.HasKey("Data"));
  422. YQL_ENSURE(x.second.Size() == 1);
  423. }
  424. OperationOptions_.ParametersYson = node;
  425. if (auto modules = dynamic_cast<TModuleResolver*>(Modules_.get())) {
  426. modules->SetParameters(node);
  427. }
  428. if (UrlListerManager_) {
  429. UrlListerManager_->SetParameters(node);
  430. }
  431. if (QContext_.CanWrite()) {
  432. QContext_.GetWriter()->Put({FacadeComponent, ParametersLabel}, parameters).GetValueSync();
  433. }
  434. }
  435. bool TProgram::ExtractQueryParametersMetadata() {
  436. auto& types = *GetAnnotationContext();
  437. NYT::TNode params = NYT::TNode::CreateMap();
  438. Y_ENSURE(ExprCtx_);
  439. if (!ExtractParametersMetaAsYson(ExprRoot_, types, *ExprCtx_, params)) {
  440. return false;
  441. }
  442. ExtractedQueryParametersMetadataYson_ = NYT::NodeToYsonString(params, ResultFormat_);
  443. return true;
  444. }
  445. bool TProgram::FillParseResult(NYql::TAstParseResult&& astRes, NYql::TWarningRules* warningRules) {
  446. if (!astRes.Issues.Empty()) {
  447. if (!ExprCtx_) {
  448. ExprCtx_.Reset(new TExprContext(NextUniqueId_));
  449. }
  450. auto& iManager = ExprCtx_->IssueManager;
  451. if (warningRules) {
  452. for (auto warningRule: *warningRules) {
  453. iManager.AddWarningRule(warningRule);
  454. }
  455. }
  456. iManager.AddScope([this]() {
  457. TIssuePtr issueHolder = new TIssue();
  458. issueHolder->SetMessage(TStringBuilder() << "Parse " << SourceSyntax_);
  459. issueHolder->Severity = TSeverityIds::S_INFO;
  460. return issueHolder;
  461. });
  462. for (auto issue: astRes.Issues) {
  463. iManager.RaiseWarning(issue);
  464. }
  465. iManager.LeaveScope();
  466. }
  467. if (!astRes.IsOk()) {
  468. return false;
  469. }
  470. AstRoot_ = astRes.Root;
  471. AstPool_ = std::move(astRes.Pool);
  472. return true;
  473. }
  474. TString TProgram::GetSessionId() const {
  475. with_lock(SessionIdLock_) {
  476. return SessionId_;
  477. }
  478. }
  479. void TProgram::AddCredentials(const TVector<std::pair<TString, TCredential>>& credentials) {
  480. Y_ENSURE(!TypeCtx_, "TypeCtx_ already created");
  481. if (QContext_.CanWrite()) {
  482. NYT::TNode credListNode = NYT::TNode::CreateList();
  483. for (const auto& c : credentials) {
  484. credListNode.Add(NYT::TNode()
  485. ("Name", c.first)
  486. ("Category", c.second.Category)
  487. ("Subcategory", c.second.Subcategory));
  488. }
  489. auto credList = NYT::NodeToYsonString(credListNode, NYT::NYson::EYsonFormat::Binary);
  490. QContext_.GetWriter()->Put({FacadeComponent, DynamicCredentialsLabel}, credList).GetValueSync();
  491. }
  492. for (const auto& credential : credentials) {
  493. Credentials_->AddCredential(credential.first, credential.second);
  494. }
  495. if (auto modules = dynamic_cast<TModuleResolver*>(Modules_.get())) {
  496. modules->SetCredentials(Credentials_);
  497. }
  498. if (UrlListerManager_) {
  499. UrlListerManager_->SetCredentials(Credentials_);
  500. }
  501. }
  502. void TProgram::ClearCredentials() {
  503. Y_ENSURE(!TypeCtx_, "TypeCtx_ already created");
  504. Credentials_ = MakeIntrusive<TCredentials>();
  505. if (auto modules = dynamic_cast<TModuleResolver*>(Modules_.get())) {
  506. modules->SetCredentials(Credentials_);
  507. }
  508. if (UrlListerManager_) {
  509. UrlListerManager_->SetCredentials(Credentials_);
  510. }
  511. }
  512. void TProgram::AddUserDataTable(const TUserDataTable& userDataTable) {
  513. for (const auto& p : userDataTable) {
  514. if (!SavedUserDataTable_.emplace(p).second) {
  515. ythrow yexception() << "UserDataTable already has user data block with key " << p.first;
  516. }
  517. UserDataStorage_->AddUserDataBlock(p.first, p.second);
  518. }
  519. if (QContext_.CanWrite()) {
  520. NYT::TNode userFilesNode;
  521. for (const auto& p : userDataTable) {
  522. userFilesNode.Add(p.first.Alias());
  523. }
  524. auto userFiles = NYT::NodeToYsonString(userFilesNode, NYT::NYson::EYsonFormat::Binary);
  525. QContext_.GetWriter()->Put({FacadeComponent, DynamicUserFilesLabel}, userFiles).GetValueSync();
  526. }
  527. }
  528. void TProgram::HandleSourceCode(TString& sourceCode) {
  529. if (QContext_.CanWrite()) {
  530. QContext_.GetWriter()->Put({FacadeComponent, SourceCodeLabel}, sourceCode).GetValueSync();
  531. } else if (QContext_.CanRead()) {
  532. auto loaded = QContext_.GetReader()->Get({FacadeComponent, SourceCodeLabel}).GetValueSync();
  533. Y_ENSURE(loaded.Defined(), "No source code");
  534. sourceCode = loaded->Value;
  535. }
  536. }
  537. namespace {
  538. THashSet<TString> ExtractSqlFlags(const NYT::TNode& dataNode) {
  539. THashSet<TString> result;
  540. for (const auto& f : dataNode["SqlFlags"].AsList()) {
  541. result.insert(f.AsString());
  542. }
  543. return result;
  544. }
  545. } // namespace
  546. void UpdateSqlFlagsFromQContext(const TQContext& qContext, THashSet<TString>& flags) {
  547. if (qContext.CanRead()) {
  548. auto loaded = qContext.GetReader()->Get({FacadeComponent, TranslationLabel}).GetValueSync();
  549. if (!loaded) {
  550. return;
  551. }
  552. auto dataNode = NYT::NodeFromYsonString(loaded->Value);
  553. flags = ExtractSqlFlags(dataNode);
  554. }
  555. }
  556. void TProgram::HandleTranslationSettings(NSQLTranslation::TTranslationSettings& loadedSettings,
  557. NSQLTranslation::TTranslationSettings*& currentSettings)
  558. {
  559. if (QContext_.CanWrite()) {
  560. auto clusterMappingsNode = NYT::TNode::CreateMap();
  561. for (const auto& c : currentSettings->ClusterMapping) {
  562. clusterMappingsNode(c.first, c.second);
  563. }
  564. auto sqlFlagsNode = NYT::TNode::CreateList();
  565. for (const auto& f : currentSettings->Flags) {
  566. sqlFlagsNode.Add(f);
  567. }
  568. auto dataNode = NYT::TNode()
  569. ("ClusterMapping", clusterMappingsNode)
  570. ("V0Behavior", ui64(currentSettings->V0Behavior))
  571. ("V0WarnAsError", currentSettings->V0WarnAsError->Allow())
  572. ("DqDefaultAuto", currentSettings->DqDefaultAuto->Allow())
  573. ("BlockDefaultAuto", currentSettings->BlockDefaultAuto->Allow())
  574. ("SqlFlags", sqlFlagsNode);
  575. auto data = NYT::NodeToYsonString(dataNode, NYT::NYson::EYsonFormat::Binary);
  576. QContext_.GetWriter()->Put({FacadeComponent, TranslationLabel}, data).GetValueSync();
  577. } else if (QContext_.CanRead()) {
  578. auto loaded = QContext_.GetReader()->Get({FacadeComponent, TranslationLabel}).GetValueSync();
  579. if (!loaded) {
  580. return;
  581. }
  582. auto dataNode = NYT::NodeFromYsonString(loaded->Value);
  583. loadedSettings.ClusterMapping.clear();
  584. for (const auto& c : dataNode["ClusterMapping"].AsMap()) {
  585. loadedSettings.ClusterMapping[c.first] = c.second.AsString();
  586. }
  587. loadedSettings.Flags = ExtractSqlFlags(dataNode);
  588. loadedSettings.V0Behavior = (NSQLTranslation::EV0Behavior)dataNode["V0Behavior"].AsUint64();
  589. loadedSettings.V0WarnAsError = NSQLTranslation::ISqlFeaturePolicy::Make(dataNode["V0WarnAsError"].AsBool());
  590. loadedSettings.DqDefaultAuto = NSQLTranslation::ISqlFeaturePolicy::Make(dataNode["DqDefaultAuto"].AsBool());
  591. loadedSettings.BlockDefaultAuto = NSQLTranslation::ISqlFeaturePolicy::Make(dataNode["BlockDefaultAuto"].AsBool());
  592. currentSettings = &loadedSettings;
  593. }
  594. }
  595. bool TProgram::ParseYql() {
  596. YQL_PROFILE_FUNC(TRACE);
  597. YQL_ENSURE(SourceSyntax_ == ESourceSyntax::Unknown);
  598. SourceSyntax_ = ESourceSyntax::Yql;
  599. SyntaxVersion_ = 1;
  600. auto sourceCode = SourceCode_;
  601. HandleSourceCode(sourceCode);
  602. return FillParseResult(ParseAst(sourceCode));
  603. }
  604. bool TProgram::ParseSql() {
  605. YQL_PROFILE_FUNC(TRACE);
  606. static const THashMap<TString, TString> clusters = {
  607. { "plato", TString(YtProviderName) }
  608. };
  609. NSQLTranslation::TTranslationSettings settings;
  610. settings.ClusterMapping = clusters;
  611. return ParseSql(settings);
  612. }
  613. bool TProgram::ParseSql(const NSQLTranslation::TTranslationSettings& settings)
  614. {
  615. YQL_PROFILE_FUNC(TRACE);
  616. YQL_ENSURE(SourceSyntax_ == ESourceSyntax::Unknown);
  617. SourceSyntax_ = ESourceSyntax::Sql;
  618. SyntaxVersion_ = settings.SyntaxVersion;
  619. NYql::TWarningRules warningRules;
  620. auto sourceCode = SourceCode_;
  621. HandleSourceCode(sourceCode);
  622. NSQLTranslation::TTranslationSettings outerSettings = settings;
  623. NSQLTranslation::TTranslationSettings* currentSettings = &outerSettings;
  624. NSQLTranslation::TTranslationSettings loadedSettings;
  625. loadedSettings.PgParser = settings.PgParser;
  626. if (QContext_) {
  627. HandleTranslationSettings(loadedSettings, currentSettings);
  628. }
  629. currentSettings->EmitReadsForExists = true;
  630. return FillParseResult(SqlToYql(sourceCode, *currentSettings, &warningRules), &warningRules);
  631. }
  632. bool TProgram::Compile(const TString& username, bool skipLibraries) {
  633. YQL_PROFILE_FUNC(TRACE);
  634. Y_ENSURE(AstRoot_, "Program not parsed yet");
  635. if (!ExprCtx_) {
  636. ExprCtx_.Reset(new TExprContext(NextUniqueId_));
  637. }
  638. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_)) {
  639. return false;
  640. }
  641. TypeCtx_->IsReadOnly = true;
  642. if (!skipLibraries && Modules_.get()) {
  643. auto libs = UserDataStorage_->GetLibraries();
  644. for (auto lib : libs) {
  645. if (!Modules_->AddFromFile(lib, *ExprCtx_, SyntaxVersion_, 0)) {
  646. return false;
  647. }
  648. }
  649. }
  650. if (!CompileExpr(
  651. *AstRoot_, ExprRoot_, *ExprCtx_,
  652. skipLibraries ? nullptr : Modules_.get(),
  653. skipLibraries ? nullptr : UrlListerManager_.Get(), 0, SyntaxVersion_
  654. )) {
  655. return false;
  656. }
  657. return true;
  658. }
  659. bool TProgram::CollectUsedClusters() {
  660. using namespace NNodes;
  661. if (!UsedClusters_) {
  662. UsedClusters_.ConstructInPlace();
  663. UsedProviders_.ConstructInPlace();
  664. auto& typesCtx = *GetAnnotationContext();
  665. auto& ctx = *ExprCtx_;
  666. auto& usedClusters = *UsedClusters_;
  667. auto& usedProviders = *UsedProviders_;
  668. bool hasErrors = false;
  669. VisitExpr(ExprRoot_, [&typesCtx, &ctx, &usedClusters, &usedProviders, &hasErrors](const TExprNode::TPtr& node) {
  670. if (auto dsNode = TMaybeNode<TCoDataSource>(node)) {
  671. auto datasource = typesCtx.DataSourceMap.FindPtr(dsNode.Cast().Category());
  672. YQL_ENSURE(datasource, "Unknown DataSource: " << dsNode.Cast().Category().Value());
  673. TMaybe<TString> cluster;
  674. if (!(*datasource)->ValidateParameters(*node, ctx, cluster)) {
  675. hasErrors = true;
  676. return false;
  677. }
  678. usedProviders.insert(TString(dsNode.Cast().Category().Value()));
  679. if (cluster && *cluster != NCommon::ALL_CLUSTERS) {
  680. usedClusters.insert(*cluster);
  681. }
  682. }
  683. if (auto dsNode = TMaybeNode<TCoDataSink>(node)) {
  684. auto datasink = typesCtx.DataSinkMap.FindPtr(dsNode.Cast().Category());
  685. YQL_ENSURE(datasink, "Unknown DataSink: " << dsNode.Cast().Category().Value());
  686. TMaybe<TString> cluster;
  687. if (!(*datasink)->ValidateParameters(*node, ctx, cluster)) {
  688. hasErrors = true;
  689. return false;
  690. }
  691. usedProviders.insert(TString(dsNode.Cast().Category().Value()));
  692. if (cluster) {
  693. usedClusters.insert(*cluster);
  694. }
  695. }
  696. return true;
  697. });
  698. if (hasErrors) {
  699. UsedClusters_ = Nothing();
  700. UsedProviders_ = Nothing();
  701. return false;
  702. }
  703. }
  704. return true;
  705. }
  706. TProgram::TStatus TProgram::Discover(const TString& username) {
  707. YQL_PROFILE_FUNC(TRACE);
  708. auto m = &TProgram::DiscoverAsync;
  709. return SyncExecution(this, m, username);
  710. }
  711. TProgram::TFutureStatus TProgram::DiscoverAsync(const TString& username) {
  712. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  713. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  714. }
  715. TypeCtx_->DiscoveryMode = true;
  716. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  717. Transformer_ = TTransformationPipeline(TypeCtx_)
  718. .AddServiceTransformers()
  719. .AddParametersEvaluation(*FunctionRegistry_)
  720. .AddPreTypeAnnotation()
  721. .AddExpressionEvaluation(*FunctionRegistry_)
  722. .AddPreIOAnnotation()
  723. .Build();
  724. TFuture<void> openSession = OpenSession(username);
  725. if (!openSession.Initialized()) {
  726. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  727. }
  728. return openSession.Apply([this](const TFuture<void>& f) {
  729. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  730. try {
  731. f.GetValue();
  732. } catch (const std::exception& e) {
  733. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  734. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  735. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  736. }
  737. return AsyncTransform(*Transformer_, ExprRoot_, *ExprCtx_, false);
  738. });
  739. }
  740. TProgram::TStatus TProgram::Lineage(const TString& username, IOutputStream* traceOut, IOutputStream* exprOut, bool withTypes) {
  741. YQL_PROFILE_FUNC(TRACE);
  742. auto m = &TProgram::LineageAsync;
  743. return SyncExecution(this, m, username, traceOut, exprOut, withTypes);
  744. }
  745. TProgram::TFutureStatus TProgram::LineageAsync(const TString& username, IOutputStream* traceOut, IOutputStream* exprOut, bool withTypes) {
  746. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  747. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  748. }
  749. TypeCtx_->IsReadOnly = true;
  750. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  751. ExprStream_ = exprOut;
  752. Transformer_ = TTransformationPipeline(TypeCtx_)
  753. .AddServiceTransformers()
  754. .AddParametersEvaluation(*FunctionRegistry_)
  755. .AddPreTypeAnnotation()
  756. .AddExpressionEvaluation(*FunctionRegistry_)
  757. .AddIOAnnotation()
  758. .AddTypeAnnotation(TIssuesIds::CORE_TYPE_ANN, true)
  759. .AddPostTypeAnnotation()
  760. .Add(TExprOutputTransformer::Sync(ExprRoot_, traceOut), "ExprOutput")
  761. .AddCheckExecution(false)
  762. .AddLineageOptimization(LineageStr_)
  763. .Add(TExprOutputTransformer::Sync(ExprRoot_, exprOut, withTypes), "AstOutput")
  764. .Build();
  765. TFuture<void> openSession = OpenSession(username);
  766. if (!openSession.Initialized())
  767. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  768. SaveExprRoot();
  769. return openSession.Apply([this](const TFuture<void>& f) {
  770. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  771. try {
  772. f.GetValue();
  773. } catch (const std::exception& e) {
  774. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  775. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  776. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  777. }
  778. return AsyncTransformWithFallback(false);
  779. });
  780. }
  781. TProgram::TStatus TProgram::Validate(const TString& username, IOutputStream* exprOut, bool withTypes) {
  782. YQL_PROFILE_FUNC(TRACE);
  783. auto m = &TProgram::ValidateAsync;
  784. return SyncExecution(this, m, username, exprOut, withTypes);
  785. }
  786. TProgram::TFutureStatus TProgram::ValidateAsync(const TString& username, IOutputStream* exprOut, bool withTypes) {
  787. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  788. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  789. }
  790. TypeCtx_->IsReadOnly = true;
  791. TVector<TDataProviderInfo> dataProviders;
  792. with_lock (DataProvidersLock_) {
  793. dataProviders = DataProviders_;
  794. }
  795. for (const auto& dp : dataProviders) {
  796. if (!dp.RemoteClusterProvider || !dp.RemoteValidate) {
  797. continue;
  798. }
  799. if (auto cluster = dp.RemoteClusterProvider(UsedClusters_, UsedProviders_, SourceSyntax_)) {
  800. return dp.RemoteValidate(*cluster, SourceSyntax_, SourceCode_, *ExprCtx_);
  801. }
  802. }
  803. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  804. ExprStream_ = exprOut;
  805. Transformer_ = TTransformationPipeline(TypeCtx_)
  806. .AddServiceTransformers()
  807. .AddParametersEvaluation(*FunctionRegistry_)
  808. .AddPreTypeAnnotation()
  809. .AddExpressionEvaluation(*FunctionRegistry_)
  810. .AddIOAnnotation()
  811. .AddTypeAnnotation(TIssuesIds::CORE_TYPE_ANN, true)
  812. .Add(TExprOutputTransformer::Sync(ExprRoot_, exprOut, withTypes), "AstOutput")
  813. .Build();
  814. TFuture<void> openSession = OpenSession(username);
  815. if (!openSession.Initialized()) {
  816. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  817. }
  818. SaveExprRoot();
  819. return openSession.Apply([this](const TFuture<void>& f) {
  820. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  821. try {
  822. f.GetValue();
  823. } catch (const std::exception& e) {
  824. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  825. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  826. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  827. }
  828. return AsyncTransformWithFallback(false);
  829. });
  830. }
  831. TProgram::TStatus TProgram::Optimize(
  832. const TString& username,
  833. IOutputStream* traceOut,
  834. IOutputStream* tracePlan,
  835. IOutputStream* exprOut,
  836. bool withTypes)
  837. {
  838. YQL_PROFILE_FUNC(TRACE);
  839. auto m = &TProgram::OptimizeAsync;
  840. return SyncExecution(this, m, username, traceOut, tracePlan, exprOut, withTypes);
  841. }
  842. TProgram::TFutureStatus TProgram::OptimizeAsync(
  843. const TString& username,
  844. IOutputStream* traceOut,
  845. IOutputStream* tracePlan,
  846. IOutputStream* exprOut,
  847. bool withTypes)
  848. {
  849. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  850. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  851. }
  852. TypeCtx_->IsReadOnly = true;
  853. TVector<TDataProviderInfo> dataProviders;
  854. with_lock (DataProvidersLock_) {
  855. dataProviders = DataProviders_;
  856. }
  857. for (const auto& dp : dataProviders) {
  858. if (!dp.RemoteClusterProvider || !dp.RemoteOptimize) {
  859. continue;
  860. }
  861. if (auto cluster = dp.RemoteClusterProvider(UsedClusters_, UsedProviders_, SourceSyntax_)) {
  862. return dp.RemoteOptimize(*cluster,
  863. SourceSyntax_, SourceCode_, nullptr,
  864. TypeCtx_, ExprRoot_, *ExprCtx_, ExternalQueryAst_, ExternalQueryPlan_);
  865. }
  866. }
  867. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  868. ExprStream_ = exprOut;
  869. PlanStream_ = tracePlan;
  870. Transformer_ = TTransformationPipeline(TypeCtx_)
  871. .AddServiceTransformers()
  872. .AddParametersEvaluation(*FunctionRegistry_)
  873. .AddPreTypeAnnotation()
  874. .AddExpressionEvaluation(*FunctionRegistry_)
  875. .AddIOAnnotation()
  876. .AddTypeAnnotation(TIssuesIds::CORE_TYPE_ANN, true)
  877. .AddPostTypeAnnotation()
  878. .Add(TExprOutputTransformer::Sync(ExprRoot_, traceOut), "ExprOutput")
  879. .AddOptimization()
  880. .Add(TExprOutputTransformer::Sync(ExprRoot_, exprOut, withTypes), "AstOutput")
  881. .Build();
  882. TFuture<void> openSession = OpenSession(username);
  883. if (!openSession.Initialized())
  884. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  885. SaveExprRoot();
  886. return openSession.Apply([this](const TFuture<void>& f) {
  887. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  888. try {
  889. f.GetValue();
  890. } catch (const std::exception& e) {
  891. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  892. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  893. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  894. }
  895. return AsyncTransformWithFallback(false);
  896. });
  897. }
  898. TProgram::TStatus TProgram::OptimizeWithConfig(
  899. const TString& username, const IPipelineConfigurator& pipelineConf)
  900. {
  901. YQL_PROFILE_FUNC(TRACE);
  902. auto m = &TProgram::OptimizeAsyncWithConfig;
  903. return SyncExecution(this, m, username, pipelineConf);
  904. }
  905. TProgram::TFutureStatus TProgram::OptimizeAsyncWithConfig(
  906. const TString& username, const IPipelineConfigurator& pipelineConf)
  907. {
  908. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  909. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  910. }
  911. TypeCtx_->IsReadOnly = true;
  912. TVector<TDataProviderInfo> dataProviders;
  913. with_lock (DataProvidersLock_) {
  914. dataProviders = DataProviders_;
  915. }
  916. for (const auto& dp : DataProviders_) {
  917. if (!dp.RemoteClusterProvider || !dp.RemoteOptimize) {
  918. continue;
  919. }
  920. if (auto cluster = dp.RemoteClusterProvider(UsedClusters_, UsedProviders_, SourceSyntax_)) {
  921. return dp.RemoteOptimize(*cluster,
  922. SourceSyntax_, SourceCode_, &pipelineConf,
  923. TypeCtx_, ExprRoot_, *ExprCtx_, ExternalQueryAst_, ExternalQueryPlan_);
  924. }
  925. }
  926. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  927. TTransformationPipeline pipeline(TypeCtx_);
  928. pipelineConf.AfterCreate(&pipeline);
  929. pipeline.AddServiceTransformers();
  930. pipeline.AddParametersEvaluation(*FunctionRegistry_);
  931. pipeline.AddPreTypeAnnotation();
  932. pipeline.AddExpressionEvaluation(*FunctionRegistry_);
  933. pipeline.AddIOAnnotation();
  934. pipeline.AddTypeAnnotation(TIssuesIds::CORE_TYPE_ANN, true);
  935. pipeline.AddPostTypeAnnotation();
  936. pipelineConf.AfterTypeAnnotation(&pipeline);
  937. pipeline.AddOptimization();
  938. if (EnableRangeComputeFor_) {
  939. pipeline.Add(MakeExpandRangeComputeForTransformer(pipeline.GetTypeAnnotationContext()),
  940. "ExpandRangeComputeFor", TIssuesIds::CORE_EXEC);
  941. }
  942. pipeline.Add(CreatePlanInfoTransformer(*TypeCtx_), "PlanInfo");
  943. pipelineConf.AfterOptimize(&pipeline);
  944. Transformer_ = pipeline.Build();
  945. TFuture<void> openSession = OpenSession(username);
  946. if (!openSession.Initialized())
  947. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  948. SaveExprRoot();
  949. return openSession.Apply([this](const TFuture<void>& f) {
  950. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  951. try {
  952. f.GetValue();
  953. } catch (const std::exception& e) {
  954. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  955. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  956. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  957. }
  958. return AsyncTransformWithFallback(false);
  959. });
  960. }
  961. TProgram::TStatus TProgram::LineageWithConfig(
  962. const TString& username, const IPipelineConfigurator& pipelineConf)
  963. {
  964. YQL_PROFILE_FUNC(TRACE);
  965. auto m = &TProgram::LineageAsyncWithConfig;
  966. return SyncExecution(this, m, username, pipelineConf);
  967. }
  968. TProgram::TFutureStatus TProgram::LineageAsyncWithConfig(
  969. const TString& username, const IPipelineConfigurator& pipelineConf)
  970. {
  971. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  972. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  973. }
  974. TypeCtx_->IsReadOnly = true;
  975. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  976. TTransformationPipeline pipeline(TypeCtx_);
  977. pipelineConf.AfterCreate(&pipeline);
  978. pipeline.AddServiceTransformers();
  979. pipeline.AddParametersEvaluation(*FunctionRegistry_);
  980. pipeline.AddPreTypeAnnotation();
  981. pipeline.AddExpressionEvaluation(*FunctionRegistry_);
  982. pipeline.AddIOAnnotation();
  983. pipeline.AddTypeAnnotation(TIssuesIds::CORE_TYPE_ANN, true);
  984. pipeline.AddPostTypeAnnotation();
  985. pipelineConf.AfterTypeAnnotation(&pipeline);
  986. pipeline.AddCheckExecution(false);
  987. pipeline.AddLineageOptimization(LineageStr_);
  988. Transformer_ = pipeline.Build();
  989. TFuture<void> openSession = OpenSession(username);
  990. if (!openSession.Initialized())
  991. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  992. SaveExprRoot();
  993. return openSession.Apply([this](const TFuture<void>& f) {
  994. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  995. try {
  996. f.GetValue();
  997. } catch (const std::exception& e) {
  998. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  999. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  1000. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  1001. }
  1002. return AsyncTransformWithFallback(false);
  1003. });
  1004. }
  1005. TProgram::TStatus TProgram::Run(
  1006. const TString& username,
  1007. IOutputStream* traceOut,
  1008. IOutputStream* tracePlan,
  1009. IOutputStream* exprOut,
  1010. bool withTypes)
  1011. {
  1012. YQL_PROFILE_FUNC(TRACE);
  1013. auto m = &TProgram::RunAsync;
  1014. return SyncExecution(this, m, username, traceOut, tracePlan, exprOut, withTypes);
  1015. }
  1016. TProgram::TFutureStatus TProgram::RunAsync(
  1017. const TString& username,
  1018. IOutputStream* traceOut,
  1019. IOutputStream* tracePlan,
  1020. IOutputStream* exprOut,
  1021. bool withTypes)
  1022. {
  1023. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  1024. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  1025. }
  1026. TypeCtx_->IsReadOnly = (HiddenMode_ != EHiddenMode::Disable);
  1027. TVector<TDataProviderInfo> dataProviders;
  1028. with_lock (DataProvidersLock_) {
  1029. dataProviders = DataProviders_;
  1030. }
  1031. for (const auto& dp : DataProviders_) {
  1032. if (!dp.RemoteClusterProvider || !dp.RemoteRun) {
  1033. continue;
  1034. }
  1035. if (auto cluster = dp.RemoteClusterProvider(UsedClusters_, UsedProviders_, SourceSyntax_)) {
  1036. return dp.RemoteRun(*cluster, SourceSyntax_, SourceCode_,
  1037. OutputFormat_, ResultFormat_, nullptr,
  1038. TypeCtx_, ExprRoot_, *ExprCtx_, ExternalQueryAst_, ExternalQueryPlan_, ExternalDiagnostics_,
  1039. ResultProviderConfig_);
  1040. }
  1041. }
  1042. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  1043. ExprStream_ = exprOut;
  1044. PlanStream_ = tracePlan;
  1045. TTransformationPipeline pipeline(TypeCtx_);
  1046. pipeline.AddServiceTransformers();
  1047. pipeline.AddParametersEvaluation(*FunctionRegistry_);
  1048. pipeline.AddPreTypeAnnotation();
  1049. pipeline.AddExpressionEvaluation(*FunctionRegistry_);
  1050. pipeline.AddIOAnnotation();
  1051. pipeline.AddTypeAnnotation(TIssuesIds::CORE_TYPE_ANN, true);
  1052. pipeline.AddPostTypeAnnotation();
  1053. pipeline.Add(TExprOutputTransformer::Sync(ExprRoot_, traceOut), "ExprOutput");
  1054. pipeline.AddOptimization();
  1055. if (EnableRangeComputeFor_) {
  1056. pipeline.Add(MakeExpandRangeComputeForTransformer(pipeline.GetTypeAnnotationContext()),
  1057. "ExpandRangeComputeFor", TIssuesIds::CORE_EXEC);
  1058. }
  1059. pipeline.Add(TExprOutputTransformer::Sync(ExprRoot_, exprOut, withTypes), "AstOutput");
  1060. pipeline.Add(TPlanOutputTransformer::Sync(tracePlan, GetPlanBuilder(), OutputFormat_), "PlanOutput");
  1061. pipeline.AddRun(ProgressWriter_);
  1062. Transformer_ = pipeline.Build();
  1063. TFuture<void> openSession = OpenSession(username);
  1064. if (!openSession.Initialized()) {
  1065. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  1066. }
  1067. SaveExprRoot();
  1068. return openSession.Apply([this](const TFuture<void>& f) {
  1069. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  1070. try {
  1071. f.GetValue();
  1072. } catch (const std::exception& e) {
  1073. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  1074. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  1075. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  1076. }
  1077. return AsyncTransformWithFallback(false);
  1078. });
  1079. }
  1080. TProgram::TStatus TProgram::RunWithConfig(
  1081. const TString& username, const IPipelineConfigurator& pipelineConf)
  1082. {
  1083. YQL_PROFILE_FUNC(TRACE);
  1084. auto m = &TProgram::RunAsyncWithConfig;
  1085. return SyncExecution(this, m, username, pipelineConf);
  1086. }
  1087. TProgram::TFutureStatus TProgram::RunAsyncWithConfig(
  1088. const TString& username, const IPipelineConfigurator& pipelineConf)
  1089. {
  1090. if (!ProvideAnnotationContext(username)->Initialize(*ExprCtx_) || !CollectUsedClusters()) {
  1091. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  1092. }
  1093. TypeCtx_->IsReadOnly = (HiddenMode_ != EHiddenMode::Disable);
  1094. TVector<TDataProviderInfo> dataProviders;
  1095. with_lock (DataProvidersLock_) {
  1096. dataProviders = DataProviders_;
  1097. }
  1098. for (const auto& dp : DataProviders_) {
  1099. if (!dp.RemoteClusterProvider || !dp.RemoteRun) {
  1100. continue;
  1101. }
  1102. if (auto cluster = dp.RemoteClusterProvider(UsedClusters_, UsedProviders_, SourceSyntax_)) {
  1103. return dp.RemoteRun(*cluster, SourceSyntax_, SourceCode_,
  1104. OutputFormat_, ResultFormat_, &pipelineConf,
  1105. TypeCtx_, ExprRoot_, *ExprCtx_, ExternalQueryAst_, ExternalQueryPlan_, ExternalDiagnostics_,
  1106. ResultProviderConfig_);
  1107. }
  1108. }
  1109. Y_ENSURE(ExprRoot_, "Program not compiled yet");
  1110. TTransformationPipeline pipeline(TypeCtx_);
  1111. pipelineConf.AfterCreate(&pipeline);
  1112. pipeline.AddServiceTransformers();
  1113. pipeline.AddParametersEvaluation(*FunctionRegistry_);
  1114. pipeline.AddPreTypeAnnotation();
  1115. pipeline.AddExpressionEvaluation(*FunctionRegistry_);
  1116. pipeline.AddIOAnnotation();
  1117. pipeline.AddTypeAnnotation(TIssuesIds::CORE_TYPE_ANN, true);
  1118. pipeline.AddPostTypeAnnotation();
  1119. pipelineConf.AfterTypeAnnotation(&pipeline);
  1120. pipeline.AddOptimization();
  1121. if (EnableRangeComputeFor_) {
  1122. pipeline.Add(MakeExpandRangeComputeForTransformer(pipeline.GetTypeAnnotationContext()),
  1123. "ExpandRangeComputeFor", TIssuesIds::CORE_EXEC);
  1124. }
  1125. pipelineConf.AfterOptimize(&pipeline);
  1126. pipeline.AddRun(ProgressWriter_);
  1127. Transformer_ = pipeline.Build();
  1128. TFuture<void> openSession = OpenSession(username);
  1129. if (!openSession.Initialized()) {
  1130. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  1131. }
  1132. SaveExprRoot();
  1133. return openSession.Apply([this](const TFuture<void>& f) {
  1134. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  1135. try {
  1136. f.GetValue();
  1137. } catch (const std::exception& e) {
  1138. YQL_LOG(ERROR) << "OpenSession error: " << e.what();
  1139. ExprCtx_->IssueManager.RaiseIssue(ExceptionToIssue(e));
  1140. return NThreading::MakeFuture<TStatus>(IGraphTransformer::TStatus::Error);
  1141. }
  1142. return AsyncTransformWithFallback(false);
  1143. });
  1144. }
  1145. void TProgram::SaveExprRoot() {
  1146. TNodeOnNodeOwnedMap deepClones;
  1147. SavedExprRoot_ = ExprCtx_->DeepCopy(*ExprRoot_, *ExprCtx_, deepClones, /*internStrings*/false, /*copyTypes*/true, /*copyResult*/false, {});
  1148. }
  1149. std::optional<bool> TProgram::CheckFallbackIssues(const TIssues& issues) {
  1150. auto isFallback = std::optional<bool>();
  1151. auto checkIssue = [&](const TIssue& issue) {
  1152. if (issue.GetCode() == TIssuesIds::DQ_GATEWAY_ERROR) {
  1153. YQL_LOG(DEBUG) << "Gateway Error " << issue;
  1154. isFallback = false;
  1155. } else if (issue.GetCode() == TIssuesIds::DQ_GATEWAY_NEED_FALLBACK_ERROR) {
  1156. YQL_LOG(DEBUG) << "Gateway Fallback Error " << issue;
  1157. isFallback = true;
  1158. } else if (issue.GetCode() == TIssuesIds::DQ_OPTIMIZE_ERROR) {
  1159. YQL_LOG(DEBUG) << "Optimize Error " << issue;
  1160. isFallback = true;
  1161. } else if (issue.GetCode() >= TIssuesIds::YT_ACCESS_DENIED &&
  1162. issue.GetCode() <= TIssuesIds::YT_FOLDER_INPUT_IS_NOT_A_FOLDER &&
  1163. (issue.GetSeverity() == TSeverityIds::S_ERROR ||
  1164. issue.GetSeverity() == TSeverityIds::S_FATAL)) {
  1165. YQL_LOG(DEBUG) << "Yt Error " << issue;
  1166. isFallback = false;
  1167. }
  1168. };
  1169. std::function<void(const TIssuePtr& issue)> recursiveCheck = [&](const TIssuePtr& issue) {
  1170. checkIssue(*issue);
  1171. for (const auto& subissue : issue->GetSubIssues()) {
  1172. recursiveCheck(subissue);
  1173. }
  1174. };
  1175. for (const auto& issue : issues) {
  1176. checkIssue(issue);
  1177. // check subissues
  1178. for (const auto& subissue : issue.GetSubIssues()) {
  1179. recursiveCheck(subissue);
  1180. }
  1181. }
  1182. return isFallback;
  1183. }
  1184. TFuture<IGraphTransformer::TStatus> TProgram::AsyncTransformWithFallback(bool applyAsyncChanges)
  1185. {
  1186. return AsyncTransform(*Transformer_, ExprRoot_, *ExprCtx_, applyAsyncChanges).Apply([this](const TFuture<IGraphTransformer::TStatus>& res) {
  1187. auto status = res.GetValueSync();
  1188. if (status == IGraphTransformer::TStatus::Error
  1189. && !TypeCtx_->ForceDq
  1190. && SavedExprRoot_
  1191. && TypeCtx_->DqCaptured
  1192. && TypeCtx_->DqFallbackPolicy != EFallbackPolicy::Never)
  1193. {
  1194. auto issues = ExprCtx_->IssueManager.GetIssues();
  1195. bool isFallback = CheckFallbackIssues(issues).value_or(true);
  1196. if (!isFallback && TypeCtx_->DqFallbackPolicy != EFallbackPolicy::Always) {
  1197. // unrecoverable error
  1198. return res;
  1199. }
  1200. ExprRoot_ = SavedExprRoot_;
  1201. SavedExprRoot_ = nullptr;
  1202. UserDataStorage_->SetUserDataTable(std::move(SavedUserDataTable_));
  1203. ExprCtx_->IssueManager.Reset();
  1204. YQL_LOG(DEBUG) << "Fallback, Issues: " << issues.ToString();
  1205. ExprCtx_->Reset();
  1206. Transformer_->Rewind();
  1207. for (auto sink : TypeCtx_->DataSinks) {
  1208. sink->Reset();
  1209. }
  1210. for (auto source : TypeCtx_->DataSources) {
  1211. source->Reset();
  1212. }
  1213. TypeCtx_->Reset();
  1214. try {
  1215. CleanupLastSession().GetValueSync();
  1216. } catch (...) {
  1217. ExprCtx_->IssueManager.RaiseIssue(TIssue({}, "Failed to cleanup session: " + CurrentExceptionMessage()));
  1218. return NThreading::MakeFuture<IGraphTransformer::TStatus>(IGraphTransformer::TStatus::Error);
  1219. }
  1220. std::function<void(const TIssuePtr& issue)> toInfo = [&](const TIssuePtr& issue) {
  1221. if (issue->Severity == TSeverityIds::S_ERROR
  1222. || issue->Severity == TSeverityIds::S_FATAL
  1223. || issue->Severity == TSeverityIds::S_WARNING)
  1224. {
  1225. issue->Severity = TSeverityIds::S_INFO;
  1226. }
  1227. for (const auto& subissue : issue->GetSubIssues()) {
  1228. toInfo(subissue);
  1229. }
  1230. };
  1231. TIssue info("DQ cannot execute the query");
  1232. info.Severity = TSeverityIds::S_INFO;
  1233. for (auto& issue : issues) {
  1234. TIssuePtr newIssue = new TIssue(issue);
  1235. if (newIssue->Severity == TSeverityIds::S_ERROR
  1236. || issue.Severity == TSeverityIds::S_FATAL
  1237. || issue.Severity == TSeverityIds::S_WARNING)
  1238. {
  1239. newIssue->Severity = TSeverityIds::S_INFO;
  1240. }
  1241. for (auto& subissue : newIssue->GetSubIssues()) {
  1242. toInfo(subissue);
  1243. }
  1244. info.AddSubIssue(newIssue);
  1245. }
  1246. ExprCtx_->IssueManager.AddIssues({info});
  1247. ++FallbackCounter_;
  1248. // don't execute recapture again
  1249. ExprCtx_->Step.Done(TExprStep::Recapture);
  1250. AbortHidden_();
  1251. return AsyncTransformWithFallback(false);
  1252. }
  1253. if (status == IGraphTransformer::TStatus::Error && (TypeCtx_->DqFallbackPolicy == EFallbackPolicy::Never || TypeCtx_->ForceDq)) {
  1254. YQL_LOG(INFO) << "Fallback skipped due to per query policy";
  1255. }
  1256. return res;
  1257. });
  1258. }
  1259. TMaybe<TString> TProgram::GetQueryAst(TMaybe<size_t> memoryLimit) {
  1260. if (ExternalQueryAst_) {
  1261. return ExternalQueryAst_;
  1262. }
  1263. TStringStream astStream;
  1264. astStream.Reserve(DEFAULT_AST_BUF_SIZE);
  1265. if (ExprRoot_) {
  1266. std::unique_ptr<IAllocator> limitingAllocator;
  1267. TConvertToAstSettings settings;
  1268. settings.AnnotationFlags = TExprAnnotationFlags::None;
  1269. settings.RefAtoms = true;
  1270. settings.Allocator = TDefaultAllocator::Instance();
  1271. if (memoryLimit) {
  1272. limitingAllocator = MakeLimitingAllocator(*memoryLimit, TDefaultAllocator::Instance());
  1273. settings.Allocator = limitingAllocator.get();
  1274. }
  1275. auto ast = ConvertToAst(*ExprRoot_, *ExprCtx_, settings);
  1276. ast.Root->PrettyPrintTo(astStream, TAstPrintFlags::ShortQuote | TAstPrintFlags::PerLine);
  1277. return astStream.Str();
  1278. } else if (AstRoot_) {
  1279. AstRoot_->PrettyPrintTo(astStream, TAstPrintFlags::ShortQuote | TAstPrintFlags::PerLine);
  1280. return astStream.Str();
  1281. }
  1282. return Nothing();
  1283. }
  1284. TMaybe<TString> TProgram::GetQueryPlan(const TPlanSettings& settings) {
  1285. if (ExternalQueryPlan_) {
  1286. return ExternalQueryPlan_;
  1287. }
  1288. if (ExprRoot_) {
  1289. TStringStream planStream;
  1290. planStream.Reserve(DEFAULT_PLAN_BUF_SIZE);
  1291. NYson::TYsonWriter writer(&planStream, OutputFormat_);
  1292. PlanBuilder_->WritePlan(writer, ExprRoot_, settings);
  1293. return planStream.Str();
  1294. }
  1295. return Nothing();
  1296. }
  1297. TMaybe<TString> TProgram::GetDiagnostics() {
  1298. if (ExternalDiagnostics_) {
  1299. return ExternalDiagnostics_;
  1300. }
  1301. if (!TypeCtx_ || !TypeCtx_->Diagnostics) {
  1302. return Nothing();
  1303. }
  1304. if (!Transformer_) {
  1305. return Nothing();
  1306. }
  1307. TStringStream out;
  1308. NYson::TYsonWriter writer(&out, DiagnosticFormat_.GetOrElse(ResultFormat_));
  1309. writer.OnBeginMap();
  1310. writer.OnKeyedItem("Write");
  1311. writer.OnBeginList();
  1312. writer.OnListItem();
  1313. writer.OnBeginMap();
  1314. writer.OnKeyedItem("Data");
  1315. writer.OnBeginMap();
  1316. writer.OnKeyedItem("Diagnostics");
  1317. writer.OnBeginMap();
  1318. writer.OnKeyedItem("TransformStats");
  1319. auto transformStats = Transformer_->GetStatistics();
  1320. NCommon::TransformerStatsToYson("", transformStats, writer);
  1321. for (auto& datasink : TypeCtx_->DataSinks) {
  1322. writer.OnKeyedItem(datasink->GetName());
  1323. if (!datasink->CollectDiagnostics(writer)) {
  1324. writer.OnEntity();
  1325. }
  1326. }
  1327. writer.OnEndMap();
  1328. writer.OnEndMap();
  1329. writer.OnEndMap();
  1330. writer.OnEndList();
  1331. writer.OnEndMap();
  1332. return out.Str();
  1333. }
  1334. IGraphTransformer::TStatistics TProgram::GetRawDiagnostics() {
  1335. return Transformer_ ? Transformer_->GetStatistics() : IGraphTransformer::TStatistics::NotPresent();
  1336. }
  1337. TMaybe<TString> TProgram::GetTasksInfo() {
  1338. if (!TypeCtx_) {
  1339. return Nothing();
  1340. }
  1341. bool hasTasks = false;
  1342. TStringStream out;
  1343. NYson::TYsonWriter writer(&out, ResultFormat_);
  1344. writer.OnBeginMap();
  1345. writer.OnKeyedItem("Write");
  1346. writer.OnBeginList();
  1347. writer.OnListItem();
  1348. writer.OnBeginMap();
  1349. writer.OnKeyedItem("Tasks");
  1350. writer.OnBeginList();
  1351. for (auto& datasink : TypeCtx_->DataSinks) {
  1352. hasTasks = hasTasks || datasink->GetTasksInfo(writer);
  1353. }
  1354. writer.OnEndList();
  1355. writer.OnEndMap();
  1356. writer.OnEndList();
  1357. writer.OnEndMap();
  1358. if (hasTasks) {
  1359. return out.Str();
  1360. } else {
  1361. return Nothing();
  1362. }
  1363. }
  1364. TMaybe<TString> TProgram::GetStatistics(bool totalOnly, THashMap<TString, TStringBuf> extraYsons) {
  1365. if (!TypeCtx_) {
  1366. return Nothing();
  1367. }
  1368. TStringStream out;
  1369. NYson::TYsonWriter writer(&out, OutputFormat_);
  1370. // Header
  1371. writer.OnBeginMap();
  1372. writer.OnKeyedItem("ExecutionStatistics");
  1373. writer.OnBeginMap();
  1374. // Providers
  1375. THashSet<TStringBuf> processed;
  1376. for (auto& datasink : TypeCtx_->DataSinks) {
  1377. TStringStream providerOut;
  1378. NYson::TYsonWriter providerWriter(&providerOut);
  1379. if (datasink->CollectStatistics(providerWriter, totalOnly)) {
  1380. writer.OnKeyedItem(datasink->GetName());
  1381. writer.OnRaw(providerOut.Str());
  1382. processed.insert(datasink->GetName());
  1383. }
  1384. }
  1385. for (auto& datasource : TypeCtx_->DataSources) {
  1386. if (processed.insert(datasource->GetName()).second) {
  1387. TStringStream providerOut;
  1388. NYson::TYsonWriter providerWriter(&providerOut);
  1389. if (datasource->CollectStatistics(providerWriter, totalOnly)) {
  1390. writer.OnKeyedItem(datasource->GetName());
  1391. writer.OnRaw(providerOut.Str());
  1392. }
  1393. }
  1394. }
  1395. auto rusage = TRusage::Get();
  1396. // System stats
  1397. writer.OnKeyedItem("system");
  1398. writer.OnBeginMap();
  1399. writer.OnKeyedItem("MaxRSS");
  1400. writer.OnBeginMap();
  1401. writer.OnKeyedItem("max");
  1402. writer.OnInt64Scalar(rusage.MaxRss);
  1403. writer.OnEndMap();
  1404. writer.OnKeyedItem("MajorPageFaults");
  1405. writer.OnBeginMap();
  1406. writer.OnKeyedItem("count");
  1407. writer.OnInt64Scalar(rusage.MajorPageFaults);
  1408. writer.OnEndMap();
  1409. if (FallbackCounter_) {
  1410. writer.OnKeyedItem("Fallback");
  1411. writer.OnBeginMap();
  1412. writer.OnKeyedItem("count");
  1413. writer.OnInt64Scalar(FallbackCounter_);
  1414. writer.OnEndMap();
  1415. }
  1416. writer.OnEndMap(); // system
  1417. if (TypeCtx_->Modules) {
  1418. writer.OnKeyedItem("moduleResolver");
  1419. writer.OnBeginMap();
  1420. TypeCtx_->Modules->WriteStatistics(writer);
  1421. writer.OnEndMap();
  1422. }
  1423. // extra
  1424. for (const auto &[k, extraYson] : extraYsons) {
  1425. writer.OnKeyedItem(k);
  1426. writer.OnRaw(extraYson);
  1427. }
  1428. // Footer
  1429. writer.OnEndMap();
  1430. writer.OnEndMap();
  1431. return out.Str();
  1432. }
  1433. TMaybe<TString> TProgram::GetDiscoveredData() {
  1434. if (!TypeCtx_) {
  1435. return Nothing();
  1436. }
  1437. TStringStream out;
  1438. NYson::TYsonWriter writer(&out, OutputFormat_);
  1439. writer.OnBeginMap();
  1440. for (auto& datasource: TypeCtx_->DataSources) {
  1441. TStringStream providerOut;
  1442. NYson::TYsonWriter providerWriter(&providerOut);
  1443. if (datasource->CollectDiscoveredData(providerWriter)) {
  1444. writer.OnKeyedItem(datasource->GetName());
  1445. writer.OnRaw(providerOut.Str());
  1446. }
  1447. }
  1448. writer.OnEndMap();
  1449. return out.Str();
  1450. }
  1451. TMaybe<TString> TProgram::GetLineage() {
  1452. return LineageStr_;
  1453. }
  1454. TProgram::TFutureStatus TProgram::ContinueAsync() {
  1455. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  1456. return AsyncTransformWithFallback(true);
  1457. }
  1458. NThreading::TFuture<void> TProgram::Abort()
  1459. {
  1460. return CloseLastSession();
  1461. }
  1462. TIssues TProgram::Issues() const {
  1463. TIssues result;
  1464. if (ExprCtx_) {
  1465. result.AddIssues(ExprCtx_->IssueManager.GetIssues());
  1466. }
  1467. result.AddIssues(FinalIssues_);
  1468. CheckFatalIssues(result);
  1469. return result;
  1470. }
  1471. TIssues TProgram::CompletedIssues() const {
  1472. TIssues result;
  1473. if (ExprCtx_) {
  1474. result.AddIssues(ExprCtx_->IssueManager.GetCompletedIssues());
  1475. }
  1476. result.AddIssues(FinalIssues_);
  1477. CheckFatalIssues(result);
  1478. return result;
  1479. }
  1480. TIssue MakeNoBlocksInfoIssue(const TVector<TString>& names, bool isTypes) {
  1481. TIssue result;
  1482. TString msg = TStringBuilder() << "Most frequent " << (isTypes ? "types " : "callables ")
  1483. << "which do not support block mode: " << JoinRange(", ", names.begin(), names.end());
  1484. result.SetMessage(msg);
  1485. result.SetCode(isTypes ? TIssuesIds::CORE_TOP_UNSUPPORTED_BLOCK_TYPES : TIssuesIds::CORE_TOP_UNSUPPORTED_BLOCK_CALLABLES, TSeverityIds::S_INFO);
  1486. return result;
  1487. }
  1488. void TProgram::FinalizeIssues() {
  1489. FinalIssues_.Clear();
  1490. if (TypeCtx_) {
  1491. static const size_t topCount = 10;
  1492. auto noBlockTypes = TypeCtx_->GetTopNoBlocksTypes(topCount);
  1493. if (!noBlockTypes.empty()) {
  1494. FinalIssues_.AddIssue(MakeNoBlocksInfoIssue(noBlockTypes, true));
  1495. }
  1496. auto noBlockCallables = TypeCtx_->GetTopNoBlocksCallables(topCount);
  1497. if (!noBlockCallables.empty()) {
  1498. FinalIssues_.AddIssue(MakeNoBlocksInfoIssue(noBlockCallables, false));
  1499. }
  1500. }
  1501. }
  1502. NThreading::TFuture<void> TProgram::CleanupLastSession() {
  1503. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  1504. TString sessionId = GetSessionId();
  1505. if (sessionId.empty()) {
  1506. return MakeFuture();
  1507. }
  1508. TVector<TDataProviderInfo> dataProviders;
  1509. with_lock (DataProvidersLock_) {
  1510. dataProviders = DataProviders_;
  1511. }
  1512. TVector<NThreading::TFuture<void>> cleanupFutures;
  1513. cleanupFutures.reserve(dataProviders.size());
  1514. for (const auto& dp : dataProviders) {
  1515. if (dp.CleanupSession) {
  1516. dp.CleanupSession(sessionId);
  1517. }
  1518. if (dp.CleanupSessionAsync) {
  1519. cleanupFutures.push_back(dp.CleanupSessionAsync(sessionId));
  1520. }
  1521. }
  1522. return NThreading::WaitExceptionOrAll(cleanupFutures);
  1523. }
  1524. NThreading::TFuture<void> TProgram::CloseLastSession() {
  1525. YQL_LOG_CTX_ROOT_SESSION_SCOPE(GetSessionId());
  1526. TVector<TDataProviderInfo> dataProviders;
  1527. with_lock (DataProvidersLock_) {
  1528. dataProviders = DataProviders_;
  1529. }
  1530. auto promise = NThreading::NewPromise<void>();
  1531. TString sessionId;
  1532. with_lock(SessionIdLock_) {
  1533. // post-condition: SessionId_ will be empty
  1534. sessionId = std::move(SessionId_);
  1535. if (sessionId.empty()) {
  1536. return CloseLastSessionFuture_;
  1537. }
  1538. CloseLastSessionFuture_ = promise.GetFuture();
  1539. }
  1540. TVector<NThreading::TFuture<void>> closeFutures;
  1541. closeFutures.reserve(dataProviders.size());
  1542. for (const auto& dp : dataProviders) {
  1543. if (dp.CloseSession) {
  1544. dp.CloseSession(sessionId);
  1545. }
  1546. if (dp.CloseSessionAsync) {
  1547. closeFutures.push_back(dp.CloseSessionAsync(sessionId));
  1548. }
  1549. }
  1550. return NThreading::WaitExceptionOrAll(closeFutures)
  1551. .Apply([promise = std::move(promise)](const NThreading::TFuture<void>&) mutable {
  1552. promise.SetValue();
  1553. });
  1554. }
  1555. TString TProgram::ResultsAsString() const {
  1556. if (!ResultProviderConfig_)
  1557. return "";
  1558. TStringStream resultOut;
  1559. NYson::TYsonWriter yson(&resultOut, OutputFormat_);
  1560. yson.OnBeginList();
  1561. for (const auto& result: Results()) {
  1562. yson.OnListItem();
  1563. yson.OnRaw(result);
  1564. }
  1565. yson.OnEndList();
  1566. return resultOut.Str();
  1567. }
  1568. TTypeAnnotationContextPtr TProgram::BuildTypeAnnotationContext(const TString& username) {
  1569. auto typeAnnotationContext = MakeIntrusive<TTypeAnnotationContext>();
  1570. typeAnnotationContext->UserDataStorage = UserDataStorage_;
  1571. typeAnnotationContext->Credentials = Credentials_;
  1572. typeAnnotationContext->Modules = Modules_;
  1573. typeAnnotationContext->UrlListerManager = UrlListerManager_;
  1574. typeAnnotationContext->UdfResolver = UdfResolver_;
  1575. typeAnnotationContext->UdfIndex = UdfIndex_;
  1576. typeAnnotationContext->UdfIndexPackageSet = UdfIndexPackageSet_;
  1577. typeAnnotationContext->RandomProvider = RandomProvider_;
  1578. typeAnnotationContext->TimeProvider = TimeProvider_;
  1579. if (DiagnosticFormat_) {
  1580. typeAnnotationContext->Diagnostics = true;
  1581. }
  1582. typeAnnotationContext->ArrowResolver = ArrowResolver_;
  1583. typeAnnotationContext->FileStorage = FileStorage_;
  1584. typeAnnotationContext->QContext = QContext_;
  1585. typeAnnotationContext->HiddenMode = HiddenMode_;
  1586. if (UdfIndex_ && UdfIndexPackageSet_) {
  1587. // setup default versions at the beginning
  1588. // could be overridden by pragma later
  1589. UdfIndexPackageSet_->AddResourcesTo(UdfIndex_);
  1590. }
  1591. PlanBuilder_ = CreatePlanBuilder(*typeAnnotationContext);
  1592. THashSet<TString> providerNames;
  1593. TVector<TString> fullResultDataSinks;
  1594. TVector<std::function<TString(const TString&, const TString&)>> tokenResolvers;
  1595. for (const auto& dpi : DataProvidersInit_) {
  1596. auto dp = dpi(
  1597. username,
  1598. SessionId_,
  1599. GatewaysConfig_,
  1600. FunctionRegistry_,
  1601. RandomProvider_,
  1602. typeAnnotationContext,
  1603. ProgressWriter_,
  1604. OperationOptions_,
  1605. AbortHidden_,
  1606. QContext_
  1607. );
  1608. if (HiddenMode_ != EHiddenMode::Disable && !dp.SupportsHidden) {
  1609. continue;
  1610. }
  1611. providerNames.insert(dp.Names.begin(), dp.Names.end());
  1612. with_lock (DataProvidersLock_) {
  1613. DataProviders_.emplace_back(dp);
  1614. }
  1615. if (dp.Source) {
  1616. typeAnnotationContext->AddDataSource(dp.Names, dp.Source);
  1617. }
  1618. if (dp.Sink) {
  1619. typeAnnotationContext->AddDataSink(dp.Names, dp.Sink);
  1620. }
  1621. if (dp.TokenResolver) {
  1622. tokenResolvers.push_back(dp.TokenResolver);
  1623. }
  1624. if (dp.SupportFullResultDataSink) {
  1625. fullResultDataSinks.insert(fullResultDataSinks.end(), dp.Names.begin(), dp.Names.end());
  1626. }
  1627. }
  1628. TVector<TString> resultProviderDataSources;
  1629. if (providerNames.contains(YtProviderName)) {
  1630. resultProviderDataSources.push_back(TString(YtProviderName));
  1631. }
  1632. if (providerNames.contains(KikimrProviderName)) {
  1633. resultProviderDataSources.push_back(TString(KikimrProviderName));
  1634. }
  1635. if (providerNames.contains(RtmrProviderName)) {
  1636. resultProviderDataSources.push_back(TString(RtmrProviderName));
  1637. }
  1638. if (providerNames.contains(PqProviderName)) {
  1639. resultProviderDataSources.push_back(TString(PqProviderName));
  1640. }
  1641. if (providerNames.contains(DqProviderName)) {
  1642. resultProviderDataSources.push_back(TString(DqProviderName));
  1643. }
  1644. if (providerNames.contains(PureProviderName)) {
  1645. resultProviderDataSources.push_back(TString(PureProviderName));
  1646. }
  1647. if (!resultProviderDataSources.empty())
  1648. {
  1649. auto resultFormat = ResultFormat_;
  1650. auto writerFactory = [resultFormat] () { return CreateYsonResultWriter(resultFormat); };
  1651. ResultProviderConfig_ = MakeIntrusive<TResultProviderConfig>(*typeAnnotationContext,
  1652. *FunctionRegistry_, ResultType_, ToString((ui32)resultFormat), writerFactory);
  1653. ResultProviderConfig_->SupportsResultPosition = SupportsResultPosition_;
  1654. auto resultProvider = CreateResultProvider(ResultProviderConfig_);
  1655. typeAnnotationContext->AddDataSink(ResultProviderName, resultProvider);
  1656. typeAnnotationContext->AvailablePureResultDataSources = resultProviderDataSources;
  1657. }
  1658. if (!fullResultDataSinks.empty()) {
  1659. typeAnnotationContext->FullResultDataSink = fullResultDataSinks.front();
  1660. }
  1661. {
  1662. auto configProvider = CreateConfigProvider(*typeAnnotationContext, GatewaysConfig_, username);
  1663. typeAnnotationContext->AddDataSource(ConfigProviderName, configProvider);
  1664. }
  1665. tokenResolvers.push_back(BuildDefaultTokenResolver(typeAnnotationContext->Credentials));
  1666. typeAnnotationContext->UserDataStorage->SetTokenResolver(BuildCompositeTokenResolver(std::move(tokenResolvers)));
  1667. return typeAnnotationContext;
  1668. }
  1669. TFuture<void> TProgram::OpenSession(const TString& username)
  1670. {
  1671. TVector<TFuture<void>> openFutures;
  1672. with_lock (DataProvidersLock_) {
  1673. for (const auto& dp : DataProviders_) {
  1674. if (dp.OpenSession) {
  1675. auto future = dp.OpenSession(SessionId_, username, ProgressWriter_, OperationOptions_,
  1676. RandomProvider_, TimeProvider_);
  1677. openFutures.push_back(future);
  1678. }
  1679. }
  1680. }
  1681. return WaitExceptionOrAll(openFutures);
  1682. }
  1683. void TProgram::Print(IOutputStream* exprOut, IOutputStream* planOut, bool cleanPlan) {
  1684. TVector<TTransformStage> printTransformers;
  1685. const auto issueCode = TIssuesIds::DEFAULT_ERROR;
  1686. if (exprOut) {
  1687. printTransformers.push_back(TTransformStage(
  1688. TExprOutputTransformer::Sync(ExprRoot_, exprOut),
  1689. "ExprOutput",
  1690. issueCode));
  1691. }
  1692. if (planOut) {
  1693. if (cleanPlan) {
  1694. GetPlanBuilder().Clear();
  1695. }
  1696. printTransformers.push_back(TTransformStage(
  1697. TPlanOutputTransformer::Sync(planOut, GetPlanBuilder(), OutputFormat_),
  1698. "PlanOutput",
  1699. issueCode));
  1700. }
  1701. auto compositeTransformer = CreateCompositeGraphTransformer(printTransformers, false);
  1702. InstantTransform(*compositeTransformer, ExprRoot_, *ExprCtx_);
  1703. }
  1704. bool TProgram::HasActiveProcesses() {
  1705. with_lock (DataProvidersLock_) {
  1706. for (const auto& dp : DataProviders_) {
  1707. if (dp.HasActiveProcesses && dp.HasActiveProcesses()) {
  1708. return true;
  1709. }
  1710. }
  1711. }
  1712. return false;
  1713. }
  1714. bool TProgram::NeedWaitForActiveProcesses() {
  1715. with_lock (DataProvidersLock_) {
  1716. for (const auto& dp : DataProviders_) {
  1717. if (dp.HasActiveProcesses && dp.HasActiveProcesses() && dp.WaitForActiveProcesses) {
  1718. return true;
  1719. }
  1720. }
  1721. }
  1722. return false;
  1723. }
  1724. } // namespace NYql