builtin.cpp 153 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689269026912692269326942695269626972698269927002701270227032704270527062707270827092710271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737273827392740274127422743274427452746274727482749275027512752275327542755275627572758275927602761276227632764276527662767276827692770277127722773277427752776277727782779278027812782278327842785278627872788278927902791279227932794279527962797279827992800280128022803280428052806280728082809281028112812281328142815281628172818281928202821282228232824282528262827282828292830283128322833283428352836283728382839284028412842284328442845284628472848284928502851285228532854285528562857285828592860286128622863286428652866286728682869287028712872287328742875287628772878287928802881288228832884288528862887288828892890289128922893289428952896289728982899290029012902290329042905290629072908290929102911291229132914291529162917291829192920292129222923292429252926292729282929293029312932293329342935293629372938293929402941294229432944294529462947294829492950295129522953295429552956295729582959296029612962296329642965296629672968296929702971297229732974297529762977297829792980298129822983298429852986298729882989299029912992299329942995299629972998299930003001300230033004300530063007300830093010301130123013301430153016301730183019302030213022302330243025302630273028302930303031303230333034303530363037303830393040304130423043304430453046304730483049305030513052305330543055305630573058305930603061306230633064306530663067306830693070307130723073307430753076307730783079308030813082308330843085308630873088308930903091309230933094309530963097309830993100310131023103310431053106310731083109311031113112311331143115311631173118311931203121312231233124312531263127312831293130313131323133313431353136313731383139314031413142314331443145314631473148314931503151315231533154315531563157315831593160316131623163316431653166316731683169317031713172317331743175317631773178317931803181318231833184318531863187318831893190319131923193319431953196319731983199320032013202320332043205320632073208320932103211321232133214321532163217321832193220322132223223322432253226322732283229323032313232323332343235323632373238323932403241324232433244324532463247324832493250325132523253325432553256325732583259326032613262326332643265326632673268326932703271327232733274327532763277327832793280328132823283328432853286328732883289329032913292329332943295329632973298329933003301330233033304330533063307330833093310331133123313331433153316331733183319332033213322332333243325332633273328332933303331333233333334333533363337333833393340334133423343334433453346334733483349335033513352335333543355335633573358335933603361336233633364336533663367336833693370337133723373337433753376337733783379338033813382338333843385338633873388338933903391339233933394339533963397339833993400340134023403340434053406340734083409341034113412341334143415341634173418341934203421342234233424342534263427342834293430343134323433343434353436343734383439344034413442344334443445344634473448344934503451345234533454345534563457345834593460346134623463346434653466346734683469347034713472347334743475347634773478347934803481348234833484348534863487348834893490349134923493349434953496349734983499350035013502350335043505350635073508350935103511351235133514351535163517351835193520352135223523352435253526352735283529353035313532353335343535353635373538353935403541354235433544354535463547354835493550355135523553355435553556355735583559356035613562356335643565356635673568356935703571357235733574357535763577357835793580358135823583358435853586358735883589359035913592359335943595359635973598359936003601360236033604360536063607360836093610361136123613361436153616361736183619362036213622362336243625362636273628362936303631363236333634363536363637363836393640364136423643364436453646364736483649365036513652365336543655365636573658365936603661366236633664366536663667366836693670367136723673367436753676367736783679368036813682368336843685368636873688368936903691369236933694369536963697369836993700370137023703370437053706370737083709371037113712371337143715371637173718371937203721372237233724372537263727372837293730373137323733373437353736373737383739374037413742374337443745374637473748374937503751375237533754375537563757375837593760376137623763376437653766376737683769377037713772377337743775377637773778377937803781378237833784378537863787378837893790379137923793379437953796379737983799380038013802380338043805380638073808380938103811381238133814381538163817381838193820382138223823382438253826382738283829383038313832383338343835383638373838383938403841384238433844384538463847384838493850385138523853385438553856385738583859386038613862386338643865386638673868386938703871387238733874387538763877
  1. #include "node.h"
  2. #include "context.h"
  3. #include "list_builtin.h"
  4. #include "match_recognize.h"
  5. #include <yql/essentials/ast/yql_type_string.h>
  6. #include <yql/essentials/public/udf/udf_data_type.h>
  7. #include <yql/essentials/core/sql_types/simple_types.h>
  8. #include <yql/essentials/minikql/mkql_program_builder.h>
  9. #include <yql/essentials/minikql/mkql_type_ops.h>
  10. #include <yql/essentials/public/issue/yql_issue_id.h>
  11. #include <yql/essentials/parser/pg_catalog/catalog.h>
  12. #include <library/cpp/charset/ci_string.h>
  13. #include <library/cpp/yson/node/node_io.h>
  14. #include <util/string/builder.h>
  15. #include <util/string/cast.h>
  16. #include <util/string/util.h>
  17. #include <util/string/join.h>
  18. #include <util/system/env.h>
  19. #include <unordered_map>
  20. using namespace NYql;
  21. namespace NSQLTranslationV1 {
  22. extern const char SubqueryExtendFor[] = "SubqueryExtendFor";
  23. extern const char SubqueryUnionAllFor[] = "SubqueryUnionAllFor";
  24. extern const char SubqueryMergeFor[] = "SubqueryMergeFor";
  25. extern const char SubqueryUnionMergeFor[] = "SubqueryUnionMergeFor";
  26. extern const char SubqueryOrderBy[] = "SubqueryOrderBy";
  27. extern const char SubqueryAssumeOrderBy[] = "SubqueryAssumeOrderBy";
  28. TNodePtr MakeTypeConfig(const TPosition& pos, const TString& ns, const TVector<TNodePtr>& udfArgs) {
  29. if (ns == "clickhouse") {
  30. auto settings = NYT::TNode::CreateMap();
  31. auto args = NYT::TNode::CreateMap();
  32. for (ui32 i = 0; i < udfArgs.size(); ++i) {
  33. if (!udfArgs[i]->IsNull() && udfArgs[i]->IsLiteral()) {
  34. args[ToString(i)] = NYT::TNode()
  35. ("type", udfArgs[i]->GetLiteralType())
  36. ("value", udfArgs[i]->GetLiteralValue());
  37. }
  38. }
  39. settings["args"] = args;
  40. return (TDeferredAtom(pos, NYT::NodeToYsonString(settings))).Build();
  41. }
  42. return nullptr;
  43. }
  44. void AdjustCheckedAggFuncName(TString& aggNormalizedName, TContext& ctx) {
  45. if (!ctx.Scoped->PragmaCheckedOps) {
  46. return;
  47. }
  48. if (aggNormalizedName == "sum") {
  49. aggNormalizedName = "checked_sum";
  50. } else if (aggNormalizedName == "sumif") {
  51. aggNormalizedName = "checked_sumif";
  52. }
  53. }
  54. class TGroupingNode final: public TAstListNode {
  55. public:
  56. TGroupingNode(TPosition pos, const TVector<TNodePtr>& args)
  57. : TAstListNode(pos)
  58. , Args(args)
  59. {}
  60. bool DoInit(TContext& ctx, ISource* src) final {
  61. if (!src) {
  62. ctx.Error(Pos) << "Grouping function should have source";
  63. return false;
  64. }
  65. TVector<TString> columns;
  66. columns.reserve(Args.size());
  67. const bool isJoin = src->GetJoin();
  68. ISource* composite = src->GetCompositeSource();
  69. for (const auto& node: Args) {
  70. auto namePtr = node->GetColumnName();
  71. if (!namePtr || !*namePtr) {
  72. ctx.Error(Pos) << "GROUPING function should use columns as arguments";
  73. return false;
  74. }
  75. TString column = *namePtr;
  76. if (isJoin) {
  77. auto sourceNamePtr = node->GetSourceName();
  78. if (sourceNamePtr && !sourceNamePtr->empty()) {
  79. column = DotJoin(*sourceNamePtr, column);
  80. }
  81. }
  82. if (!src->IsGroupByColumn(column) && !src->IsAlias(EExprSeat::GroupBy, *namePtr) && (!composite || !composite->IsGroupByColumn(column))) {
  83. ctx.Error(node->GetPos()) << "Column '" << column << "' is not a grouping column";
  84. return false;
  85. }
  86. columns.emplace_back(column);
  87. }
  88. TString groupingColumn;
  89. if (!src->AddGrouping(ctx, columns, groupingColumn)) {
  90. return false;
  91. }
  92. Nodes.push_back(BuildAtom(Pos, "Member"));
  93. Nodes.push_back(BuildAtom(Pos, "row"));
  94. Nodes.push_back(BuildQuotedAtom(Pos, groupingColumn));
  95. return TAstListNode::DoInit(ctx, src);
  96. }
  97. TNodePtr DoClone() const final {
  98. return new TGroupingNode(Pos, CloneContainer(Args));
  99. }
  100. private:
  101. const TVector<TNodePtr> Args;
  102. };
  103. class TBasicAggrFunc final: public TAstListNode {
  104. public:
  105. TBasicAggrFunc(TPosition pos, const TString& name, TAggregationPtr aggr, const TVector<TNodePtr>& args)
  106. : TAstListNode(pos)
  107. , Name(name)
  108. , Aggr(aggr)
  109. , Args(args)
  110. {}
  111. TCiString GetName() const {
  112. return Name;
  113. }
  114. bool DoInit(TContext& ctx, ISource* src) final {
  115. if (!src) {
  116. ctx.Error(Pos) << "Unable to use aggregation function '" << Name << "' without data source";
  117. return false;
  118. }
  119. if (!DoInitAggregation(ctx, src)) {
  120. return false;
  121. }
  122. return TAstListNode::DoInit(ctx, src);
  123. }
  124. void CollectPreaggregateExprs(TContext& ctx, ISource& src, TVector<INode::TPtr>& exprs) override {
  125. if (Args.empty() || (Aggr->GetAggregationMode() != EAggregateMode::Distinct && Aggr->GetAggregationMode() != EAggregateMode::OverWindowDistinct)) {
  126. return;
  127. }
  128. auto& expr = Args.front();
  129. // need to initialize expr before checking whether it is a column
  130. auto clone = expr->Clone();
  131. if (!clone->Init(ctx, &src)) {
  132. return;
  133. }
  134. const auto column = clone->GetColumnName();
  135. if (column) {
  136. return;
  137. }
  138. auto tmpColumn = src.MakeLocalName("_yql_preagg_" + Name);
  139. YQL_ENSURE(!expr->GetLabel());
  140. expr->SetLabel(tmpColumn);
  141. PreaggregateExpr = expr;
  142. exprs.push_back(PreaggregateExpr);
  143. expr = BuildColumn(expr->GetPos(), tmpColumn);
  144. Aggr->MarkKeyColumnAsGenerated();
  145. }
  146. TNodePtr DoClone() const final {
  147. TAggregationPtr aggrClone = static_cast<IAggregation*>(Aggr->Clone().Release());
  148. return new TBasicAggrFunc(Pos, Name, aggrClone, CloneContainer(Args));
  149. }
  150. TAggregationPtr GetAggregation() const override {
  151. return Aggr;
  152. }
  153. private:
  154. bool DoInitAggregation(TContext& ctx, ISource* src) {
  155. if (PreaggregateExpr) {
  156. YQL_ENSURE(PreaggregateExpr->HasState(ENodeState::Initialized));
  157. if (PreaggregateExpr->IsAggregated() && !PreaggregateExpr->IsAggregationKey() && !Aggr->IsOverWindow()) {
  158. ctx.Error(Aggr->GetPos()) << "Aggregation of aggregated values is forbidden";
  159. return false;
  160. }
  161. }
  162. if (!Aggr->InitAggr(ctx, false, src, *this, Args)) {
  163. return false;
  164. }
  165. return src->AddAggregation(ctx, Aggr);
  166. }
  167. void DoUpdateState() const final {
  168. State.Set(ENodeState::Const, !Args.empty() && AllOf(Args, [](const auto& arg){ return arg->IsConstant(); }));
  169. State.Set(ENodeState::Aggregated);
  170. }
  171. TNodePtr PreaggregateExpr;
  172. protected:
  173. const TString Name;
  174. TAggregationPtr Aggr;
  175. TVector<TNodePtr> Args;
  176. };
  177. class TBasicAggrFactory final : public TAstListNode {
  178. public:
  179. TBasicAggrFactory(TPosition pos, const TString& name, TAggregationPtr aggr, const TVector<TNodePtr>& args)
  180. : TAstListNode(pos)
  181. , Name(name)
  182. , Aggr(aggr)
  183. , Args(args)
  184. {}
  185. TCiString GetName() const {
  186. return Name;
  187. }
  188. bool DoInit(TContext& ctx, ISource* src) final {
  189. if (!DoInitAggregation(ctx)) {
  190. return false;
  191. }
  192. auto factory = Aggr->AggregationTraitsFactory();
  193. auto apply = Y("Apply", factory, Y("ListType", "type"));
  194. auto columnIndices = Aggr->GetFactoryColumnIndices();
  195. if (columnIndices.size() == 1) {
  196. apply = L(apply, "extractor");
  197. } else {
  198. // make several extractors from main that returns a tuple
  199. for (ui32 arg = 0; arg < columnIndices.size(); ++arg) {
  200. auto partial = BuildLambda(Pos, Y("row"), Y("Nth", Y("Apply", "extractor", "row"), Q(ToString(columnIndices[arg]))));
  201. apply = L(apply, partial);
  202. }
  203. }
  204. Aggr->AddFactoryArguments(apply);
  205. Lambda = BuildLambda(Pos, Y("type", "extractor"), apply);
  206. return TAstListNode::DoInit(ctx, src);
  207. }
  208. TAstNode* Translate(TContext& ctx) const override {
  209. return Lambda->Translate(ctx);
  210. }
  211. TNodePtr DoClone() const final {
  212. TAggregationPtr aggrClone = static_cast<IAggregation*>(Aggr->Clone().Release());
  213. return new TBasicAggrFactory(Pos, Name, aggrClone, CloneContainer(Args));
  214. }
  215. TAggregationPtr GetAggregation() const override {
  216. return Aggr;
  217. }
  218. private:
  219. bool DoInitAggregation(TContext& ctx) {
  220. return Aggr->InitAggr(ctx, true, nullptr, *this, Args);
  221. }
  222. protected:
  223. const TString Name;
  224. TAggregationPtr Aggr;
  225. TVector<TNodePtr> Args;
  226. TNodePtr Lambda;
  227. };
  228. typedef THolder<TBasicAggrFunc> TAggrFuncPtr;
  229. class TLiteralStringAtom: public INode {
  230. public:
  231. TLiteralStringAtom(TPosition pos, TNodePtr node, const TString& info, const TString& prefix = {})
  232. : INode(pos)
  233. , Node(node)
  234. , Info(info)
  235. , Prefix(prefix)
  236. {
  237. }
  238. bool DoInit(TContext& ctx, ISource* src) override {
  239. Y_UNUSED(src);
  240. if (!Node) {
  241. ctx.Error(Pos) << Info;
  242. return false;
  243. }
  244. if (!Node->Init(ctx, src)) {
  245. return false;
  246. }
  247. Atom = MakeAtomFromExpression(Pos, ctx, Node, Prefix).Build();
  248. return true;
  249. }
  250. bool IsLiteral() const override {
  251. return Atom ? Atom->IsLiteral() : false;
  252. }
  253. TString GetLiteralType() const override {
  254. return Atom ? Atom->GetLiteralType() : "";
  255. }
  256. TString GetLiteralValue() const override {
  257. return Atom ? Atom->GetLiteralValue() : "";
  258. }
  259. TAstNode* Translate(TContext& ctx) const override {
  260. return Atom->Translate(ctx);
  261. }
  262. TPtr DoClone() const final {
  263. return new TLiteralStringAtom(GetPos(), SafeClone(Node), Info, Prefix);
  264. }
  265. void DoUpdateState() const override {
  266. YQL_ENSURE(Atom);
  267. State.Set(ENodeState::Const, Atom->IsConstant());
  268. State.Set(ENodeState::Aggregated, Atom->IsAggregated());
  269. State.Set(ENodeState::OverWindow, Atom->IsOverWindow());
  270. }
  271. private:
  272. TNodePtr Node;
  273. TNodePtr Atom;
  274. TString Info;
  275. TString Prefix;
  276. };
  277. class TYqlAsAtom: public TLiteralStringAtom {
  278. public:
  279. TYqlAsAtom(TPosition pos, const TVector<TNodePtr>& args)
  280. : TLiteralStringAtom(pos, args.size() == 1 ? args[0] : nullptr, "Literal string is required as argument")
  281. {
  282. }
  283. };
  284. class TYqlData: public TCallNode {
  285. public:
  286. TYqlData(TPosition pos, const TString& type, const TVector<TNodePtr>& args)
  287. : TCallNode(pos, type, 1, 1, args)
  288. {
  289. }
  290. bool DoInit(TContext& ctx, ISource* src) override {
  291. auto slot = NUdf::FindDataSlot(GetOpName());
  292. if (!slot) {
  293. ctx.Error(Pos) << "Unexpected type " << GetOpName();
  294. return false;
  295. }
  296. if (*slot == NUdf::EDataSlot::Decimal) {
  297. MinArgs = MaxArgs = 3;
  298. }
  299. if (!ValidateArguments(ctx)) {
  300. return false;
  301. }
  302. auto stringNode = Args[0];
  303. auto atom = stringNode->GetLiteral("String");
  304. if (!atom) {
  305. ctx.Error(Pos) << "Expected literal string as argument in " << GetOpName() << " function";
  306. return false;
  307. }
  308. TString value;
  309. if (*slot == NUdf::EDataSlot::Decimal) {
  310. const auto precision = Args[1]->GetLiteral("Int32");
  311. const auto scale = Args[2]->GetLiteral("Int32");
  312. if (!NKikimr::NMiniKQL::IsValidDecimal(*atom)) {
  313. ctx.Error(Pos) << "Invalid value " << atom->Quote() << " for type " << GetOpName();
  314. return false;
  315. }
  316. ui8 stub;
  317. if (!(precision && TryFromString<ui8>(*precision, stub))) {
  318. ctx.Error(Pos) << "Invalid precision " << (precision ? precision->Quote() : "") << " for type " << GetOpName();
  319. return false;
  320. }
  321. if (!(scale && TryFromString<ui8>(*scale, stub))) {
  322. ctx.Error(Pos) << "Invalid scale " << (scale ? scale->Quote() : "") << " for type " << GetOpName();
  323. return false;
  324. }
  325. Args[0] = BuildQuotedAtom(GetPos(), *atom);
  326. Args[1] = BuildQuotedAtom(GetPos(), *precision);
  327. Args[2] = BuildQuotedAtom(GetPos(), *scale);
  328. return TCallNode::DoInit(ctx, src);
  329. } else if (NUdf::GetDataTypeInfo(*slot).Features & (NUdf::DateType | NUdf::TzDateType | NUdf::TimeIntervalType)) {
  330. const auto out = NKikimr::NMiniKQL::ValueFromString(*slot, *atom);
  331. if (!out) {
  332. ctx.Error(Pos) << "Invalid value " << atom->Quote() << " for type " << GetOpName();
  333. return false;
  334. }
  335. switch (*slot) {
  336. case NUdf::EDataSlot::Date:
  337. case NUdf::EDataSlot::TzDate:
  338. value = ToString(out.Get<ui16>());
  339. break;
  340. case NUdf::EDataSlot::Date32:
  341. case NUdf::EDataSlot::TzDate32:
  342. value = ToString(out.Get<i32>());
  343. break;
  344. case NUdf::EDataSlot::Datetime:
  345. case NUdf::EDataSlot::TzDatetime:
  346. value = ToString(out.Get<ui32>());
  347. break;
  348. case NUdf::EDataSlot::Timestamp:
  349. case NUdf::EDataSlot::TzTimestamp:
  350. value = ToString(out.Get<ui64>());
  351. break;
  352. case NUdf::EDataSlot::Datetime64:
  353. case NUdf::EDataSlot::Timestamp64:
  354. case NUdf::EDataSlot::TzDatetime64:
  355. case NUdf::EDataSlot::TzTimestamp64:
  356. value = ToString(out.Get<i64>());
  357. break;
  358. case NUdf::EDataSlot::Interval:
  359. case NUdf::EDataSlot::Interval64:
  360. value = ToString(out.Get<i64>());
  361. if ('T' == atom->back()) {
  362. ctx.Error(Pos) << "Time prefix 'T' at end of interval constant. The designator 'T' shall be absent if all of the time components are absent.";
  363. return false;
  364. }
  365. break;
  366. default:
  367. Y_ABORT("Unexpected data slot");
  368. }
  369. if (NUdf::GetDataTypeInfo(*slot).Features & NUdf::TzDateType) {
  370. value += ",";
  371. value += NKikimr::NMiniKQL::GetTimezoneIANAName(out.GetTimezoneId());
  372. }
  373. } else if (NUdf::EDataSlot::Uuid == *slot) {
  374. char out[0x10];
  375. if (!NKikimr::NMiniKQL::ParseUuid(*atom, out)) {
  376. ctx.Error(Pos) << "Invalid value " << atom->Quote() << " for type " << GetOpName();
  377. return false;
  378. }
  379. value.assign(out, sizeof(out));
  380. } else {
  381. if (!NKikimr::NMiniKQL::IsValidStringValue(*slot, *atom)) {
  382. ctx.Error(Pos) << "Invalid value " << atom->Quote() << " for type " << GetOpName();
  383. return false;
  384. }
  385. value = *atom;
  386. }
  387. Args[0] = BuildQuotedAtom(GetPos(), value);
  388. return TCallNode::DoInit(ctx, src);
  389. }
  390. TPtr DoClone() const final {
  391. return new TYqlData(GetPos(), OpName, CloneContainer(Args));
  392. }
  393. };
  394. class TTableName : public TCallNode {
  395. public:
  396. TTableName(TPosition pos, const TVector<TNodePtr>& args, const TString& service)
  397. : TCallNode(pos, "TableName", 0, 2, args)
  398. , Service(service)
  399. , EmptyArgs(args.empty())
  400. {
  401. }
  402. bool DoInit(TContext& ctx, ISource* src) override {
  403. if (!ValidateArguments(ctx)) {
  404. return false;
  405. }
  406. if (Args.empty()) {
  407. if (!src) {
  408. ctx.Error(Pos) << "Unable to use TableName() without source";
  409. return false;
  410. }
  411. // TODO: TablePath() and TableRecordIndex() have more strict limitations
  412. if (src->GetJoin()) {
  413. ctx.Warning(Pos,
  414. TIssuesIds::YQL_EMPTY_TABLENAME_RESULT) << "TableName() may produce empty result when used in ambiguous context (with JOIN)";
  415. }
  416. if (src->HasAggregations()) {
  417. ctx.Warning(Pos,
  418. TIssuesIds::YQL_EMPTY_TABLENAME_RESULT) << "TableName() will produce empty result when used with aggregation.\n"
  419. "Please consult documentation for possible workaround";
  420. }
  421. Args.push_back(Y("TablePath", Y("DependsOn", "row")));
  422. }
  423. if (Args.size() == 2) {
  424. auto literal = Args[1]->GetLiteral("String");
  425. if (!literal) {
  426. ctx.Error(Args[1]->GetPos()) << "Expected literal string as second argument in TableName function";
  427. return false;
  428. }
  429. Args[1] = BuildQuotedAtom(Args[1]->GetPos(), *literal);
  430. } else {
  431. if (Service.empty()) {
  432. ctx.Error(GetPos()) << GetOpName() << " requires either service name as second argument or current cluster name";
  433. return false;
  434. }
  435. Args.push_back(BuildQuotedAtom(GetPos(), Service));
  436. }
  437. return TCallNode::DoInit(ctx, src);
  438. }
  439. TPtr DoClone() const final {
  440. return new TTableName(GetPos(), CloneContainer(Args), Service);
  441. }
  442. void DoUpdateState() const override {
  443. if (EmptyArgs) {
  444. State.Set(ENodeState::Const, false);
  445. } else {
  446. TCallNode::DoUpdateState();
  447. }
  448. }
  449. private:
  450. TString Service;
  451. const bool EmptyArgs;
  452. };
  453. class TYqlParseType final : public INode {
  454. public:
  455. TYqlParseType(TPosition pos, const TVector<TNodePtr>& args)
  456. : INode(pos)
  457. , Args(args)
  458. {}
  459. TAstNode* Translate(TContext& ctx) const override {
  460. if (Args.size() != 1) {
  461. ctx.Error(Pos) << "Expected 1 argument in ParseType function";
  462. return nullptr;
  463. }
  464. auto literal = Args[0]->GetLiteral("String");
  465. if (!literal) {
  466. ctx.Error(Args[0]->GetPos()) << "Expected literal string as argument in ParseType function";
  467. return nullptr;
  468. }
  469. auto parsed = ParseType(*literal, *ctx.Pool, ctx.Issues, Args[0]->GetPos());
  470. if (!parsed) {
  471. ctx.Error(Args[0]->GetPos()) << "Failed to parse type";
  472. return nullptr;
  473. }
  474. return parsed;
  475. }
  476. TNodePtr DoClone() const final {
  477. return new TYqlParseType(Pos, CloneContainer(Args));
  478. }
  479. void DoUpdateState() const final {
  480. State.Set(ENodeState::Const);
  481. }
  482. private:
  483. TVector<TNodePtr> Args;
  484. };
  485. class TYqlAddTimezone: public TCallNode {
  486. public:
  487. TYqlAddTimezone(TPosition pos, const TVector<TNodePtr>& args)
  488. : TCallNode(pos, "AddTimezone", 2, 2, args)
  489. {
  490. }
  491. bool DoInit(TContext& ctx, ISource* src) override {
  492. if (!ValidateArguments(ctx)) {
  493. return false;
  494. }
  495. Args[1] = Y("TimezoneId", Args[1]);
  496. return TCallNode::DoInit(ctx, src);
  497. }
  498. TNodePtr DoClone() const final {
  499. return new TYqlAddTimezone(Pos, CloneContainer(Args));
  500. }
  501. };
  502. class TYqlPgType: public TCallNode {
  503. public:
  504. TYqlPgType(TPosition pos, const TVector<TNodePtr>& args)
  505. : TCallNode(pos, "PgType", 1, 1, args)
  506. {
  507. }
  508. bool DoInit(TContext& ctx, ISource* src) final {
  509. if (!ValidateArguments(ctx)) {
  510. return false;
  511. }
  512. ui32 oid;
  513. if (Args[0]->IsIntegerLiteral() && TryFromString<ui32>(Args[0]->GetLiteralValue(), oid)) {
  514. if (!NPg::HasType(oid)) {
  515. ctx.Error(Args[0]->GetPos()) << "Unknown pg type oid: " << oid;
  516. return false;
  517. } else {
  518. Args[0] = BuildQuotedAtom(Args[0]->GetPos(), NPg::LookupType(oid).Name);
  519. }
  520. } else if (Args[0]->IsLiteral() && Args[0]->GetLiteralType() == "String") {
  521. if (!NPg::HasType(Args[0]->GetLiteralValue())) {
  522. ctx.Error(Args[0]->GetPos()) << "Unknown pg type: " << Args[0]->GetLiteralValue();
  523. return false;
  524. } else {
  525. Args[0] = BuildQuotedAtom(Args[0]->GetPos(), Args[0]->GetLiteralValue());
  526. }
  527. } else {
  528. ctx.Error(Args[0]->GetPos()) << "Expecting string literal with pg type name or integer literal with pg type oid";
  529. return false;
  530. }
  531. return TCallNode::DoInit(ctx, src);
  532. }
  533. TNodePtr DoClone() const final {
  534. return new TYqlPgType(Pos, CloneContainer(Args));
  535. }
  536. };
  537. class TYqlPgConst : public TCallNode {
  538. public:
  539. TYqlPgConst(TPosition pos, const TVector<TNodePtr>& args)
  540. : TCallNode(pos, "PgConst", 2, -1, args)
  541. {
  542. }
  543. bool DoInit(TContext& ctx, ISource* src) final {
  544. if (!ValidateArguments(ctx)) {
  545. return false;
  546. }
  547. if (!Args[0]->Init(ctx, src)) {
  548. return false;
  549. }
  550. if (Args[0]->IsLiteral()) {
  551. Args[0] = BuildQuotedAtom(Args[0]->GetPos(), Args[0]->GetLiteralValue());
  552. } else {
  553. auto value = MakeAtomFromExpression(Pos, ctx, Args[0]).Build();
  554. Args[0] = value;
  555. }
  556. if (Args.size() > 2) {
  557. TVector<TNodePtr> typeModArgs;
  558. typeModArgs.push_back(Args[1]);
  559. for (ui32 i = 2; i < Args.size(); ++i) {
  560. if (!Args[i]->IsLiteral()) {
  561. ctx.Error(Args[i]->GetPos()) << "Expecting literal";
  562. return false;
  563. }
  564. typeModArgs.push_back(BuildQuotedAtom(Args[i]->GetPos(), Args[i]->GetLiteralValue()));
  565. }
  566. Args.erase(Args.begin() + 2, Args.end());
  567. Args.push_back(new TCallNodeImpl(Pos, "PgTypeMod", typeModArgs));
  568. }
  569. return TCallNode::DoInit(ctx, src);
  570. }
  571. TNodePtr DoClone() const final {
  572. return new TYqlPgConst(Pos, CloneContainer(Args));
  573. }
  574. };
  575. class TYqlPgCast : public TCallNode {
  576. public:
  577. TYqlPgCast(TPosition pos, const TVector<TNodePtr>& args)
  578. : TCallNode(pos, "PgCast", 2, -1, args)
  579. {
  580. }
  581. bool DoInit(TContext& ctx, ISource* src) final {
  582. if (!ValidateArguments(ctx)) {
  583. return false;
  584. }
  585. if (Args.size() > 2) {
  586. TVector<TNodePtr> typeModArgs;
  587. typeModArgs.push_back(Args[1]);
  588. for (ui32 i = 2; i < Args.size(); ++i) {
  589. if (!Args[i]->IsLiteral()) {
  590. ctx.Error(Args[i]->GetPos()) << "Expecting literal";
  591. return false;
  592. }
  593. typeModArgs.push_back(BuildQuotedAtom(Args[i]->GetPos(), Args[i]->GetLiteralValue()));
  594. }
  595. Args.erase(Args.begin() + 2, Args.end());
  596. Args.push_back(new TCallNodeImpl(Pos, "PgTypeMod", typeModArgs));
  597. }
  598. return TCallNode::DoInit(ctx, src);
  599. }
  600. TNodePtr DoClone() const final {
  601. return new TYqlPgCast(Pos, CloneContainer(Args));
  602. }
  603. };
  604. class TYqlPgOp : public TCallNode {
  605. public:
  606. TYqlPgOp(TPosition pos, const TVector<TNodePtr>& args)
  607. : TCallNode(pos, "PgOp", 2, 3, args)
  608. {
  609. }
  610. bool DoInit(TContext& ctx, ISource* src) final {
  611. if (!ValidateArguments(ctx)) {
  612. return false;
  613. }
  614. if (!Args[0]->Init(ctx, src)) {
  615. return false;
  616. }
  617. if (!Args[0]->IsLiteral() || Args[0]->GetLiteralType() != "String") {
  618. ctx.Error(Args[0]->GetPos()) << "Expecting string literal as first argument";
  619. return false;
  620. }
  621. Args[0] = BuildQuotedAtom(Args[0]->GetPos(), Args[0]->GetLiteralValue());
  622. return TCallNode::DoInit(ctx, src);
  623. }
  624. TNodePtr DoClone() const final {
  625. return new TYqlPgOp(Pos, CloneContainer(Args));
  626. }
  627. };
  628. template <bool RangeFunction>
  629. class TYqlPgCall : public TCallNode {
  630. public:
  631. TYqlPgCall(TPosition pos, const TVector<TNodePtr>& args)
  632. : TCallNode(pos, "PgCall", 1, -1, args)
  633. {
  634. }
  635. bool DoInit(TContext& ctx, ISource* src) final {
  636. if (!ValidateArguments(ctx)) {
  637. return false;
  638. }
  639. if (!Args[0]->Init(ctx, src)) {
  640. return false;
  641. }
  642. if (!Args[0]->IsLiteral() || Args[0]->GetLiteralType() != "String") {
  643. ctx.Error(Args[0]->GetPos()) << "Expecting string literal as first argument";
  644. return false;
  645. }
  646. Args[0] = BuildQuotedAtom(Args[0]->GetPos(), Args[0]->GetLiteralValue());
  647. Args.insert(Args.begin() + 1, RangeFunction ? Q(Y(Q(Y(Q("range"))))) : Q(Y()));
  648. return TCallNode::DoInit(ctx, src);
  649. }
  650. TNodePtr DoClone() const final {
  651. return new TYqlPgCall<RangeFunction>(Pos, CloneContainer(Args));
  652. }
  653. };
  654. template <const char* Name>
  655. class TYqlSubqueryFor : public TCallNode {
  656. public:
  657. TYqlSubqueryFor(TPosition pos, const TVector<TNodePtr>& args)
  658. : TCallNode(pos, Name, 2, 2, args)
  659. {
  660. }
  661. bool DoInit(TContext& ctx, ISource* src) override {
  662. if (!ValidateArguments(ctx)) {
  663. return false;
  664. }
  665. Args[0] = Y("EvaluateExpr", Args[0]);
  666. return TCallNode::DoInit(ctx, src);
  667. }
  668. TNodePtr DoClone() const final {
  669. return new TYqlSubqueryFor<Name>(Pos, CloneContainer(Args));
  670. }
  671. };
  672. template <const char* Name>
  673. class TYqlSubqueryOrderBy : public TCallNode {
  674. public:
  675. TYqlSubqueryOrderBy(TPosition pos, const TVector<TNodePtr>& args)
  676. : TCallNode(pos, Name, 2, 2, args)
  677. {
  678. }
  679. bool DoInit(TContext& ctx, ISource* src) override {
  680. if (!ValidateArguments(ctx)) {
  681. return false;
  682. }
  683. Args[1] = Y("EvaluateExpr", Args[1]);
  684. return TCallNode::DoInit(ctx, src);
  685. }
  686. TNodePtr DoClone() const final {
  687. return new TYqlSubqueryOrderBy<Name>(Pos, CloneContainer(Args));
  688. }
  689. };
  690. template <bool Strict>
  691. class TYqlTypeAssert : public TCallNode {
  692. public:
  693. TYqlTypeAssert(TPosition pos, const TVector<TNodePtr>& args)
  694. : TCallNode(pos, Strict ? "EnsureType" : "EnsureConvertibleTo", 2, 3, args)
  695. {
  696. }
  697. bool DoInit(TContext& ctx, ISource* src) override {
  698. if (!ValidateArguments(ctx)) {
  699. return false;
  700. }
  701. if (!Args[1]->Init(ctx, src)) {
  702. return false;
  703. }
  704. if (Args.size() == 3) {
  705. if (!Args[2]->Init(ctx, src)) {
  706. return false;
  707. }
  708. auto message = MakeAtomFromExpression(Pos, ctx, Args[2]).Build();
  709. Args[2] = message;
  710. }
  711. return TCallNode::DoInit(ctx, src);
  712. }
  713. TNodePtr DoClone() const final {
  714. return new TYqlTypeAssert<Strict>(Pos, CloneContainer(Args));
  715. }
  716. };
  717. class TFromBytes final : public TCallNode {
  718. public:
  719. TFromBytes(TPosition pos, const TVector<TNodePtr>& args)
  720. : TCallNode(pos, "FromBytes", 2, 2, args)
  721. {}
  722. bool DoInit(TContext& ctx, ISource* src) override {
  723. if (!ValidateArguments(ctx)) {
  724. return false;
  725. }
  726. if (!Args[1]->Init(ctx, src)) {
  727. return false;
  728. }
  729. Args[1] = MakeAtomFromExpression(Pos, ctx, Y("FormatType", Args[1])).Build();
  730. return TCallNode::DoInit(ctx, src);
  731. }
  732. TNodePtr DoClone() const final {
  733. return new TFromBytes(Pos, CloneContainer(Args));
  734. }
  735. };
  736. class TYqlTaggedBase : public TCallNode {
  737. public:
  738. TYqlTaggedBase(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  739. : TCallNode(pos, opName, 2, 2, args)
  740. {}
  741. bool DoInit(TContext& ctx, ISource* src) override {
  742. if (!ValidateArguments(ctx)) {
  743. return false;
  744. }
  745. if (!Args[1]->Init(ctx, src)) {
  746. return false;
  747. }
  748. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  749. return TCallNode::DoInit(ctx, src);
  750. }
  751. };
  752. class TYqlAsTagged final : public TYqlTaggedBase {
  753. public:
  754. TYqlAsTagged(TPosition pos, const TVector<TNodePtr>& args)
  755. : TYqlTaggedBase(pos, "AsTagged", args)
  756. {}
  757. TNodePtr DoClone() const final {
  758. return new TYqlAsTagged(Pos, CloneContainer(Args));
  759. }
  760. };
  761. class TYqlUntag final : public TYqlTaggedBase {
  762. public:
  763. TYqlUntag(TPosition pos, const TVector<TNodePtr>& args)
  764. : TYqlTaggedBase(pos, "Untag", args)
  765. {}
  766. TNodePtr DoClone() const final {
  767. return new TYqlUntag(Pos, CloneContainer(Args));
  768. }
  769. };
  770. class TYqlVariant final : public TCallNode {
  771. public:
  772. TYqlVariant(TPosition pos, const TVector<TNodePtr>& args)
  773. : TCallNode(pos, "Variant", 3, 3, args)
  774. {}
  775. bool DoInit(TContext& ctx, ISource* src) override {
  776. if (!ValidateArguments(ctx)) {
  777. return false;
  778. }
  779. if (!Args[1]->Init(ctx, src)) {
  780. return false;
  781. }
  782. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  783. return TCallNode::DoInit(ctx, src);
  784. }
  785. TNodePtr DoClone() const final {
  786. return new TYqlVariant(Pos, CloneContainer(Args));
  787. }
  788. };
  789. class TYqlEnum final : public TCallNode {
  790. public:
  791. TYqlEnum(TPosition pos, const TVector<TNodePtr>& args)
  792. : TCallNode(pos, "Enum", 2, 2, args)
  793. {}
  794. bool DoInit(TContext& ctx, ISource* src) override {
  795. if (!ValidateArguments(ctx)) {
  796. return false;
  797. }
  798. if (!Args[0]->Init(ctx, src)) {
  799. return false;
  800. }
  801. Args[0] = MakeAtomFromExpression(Pos, ctx, Args[0]).Build();
  802. return TCallNode::DoInit(ctx, src);
  803. }
  804. TNodePtr DoClone() const final {
  805. return new TYqlEnum(Pos, CloneContainer(Args));
  806. }
  807. };
  808. class TYqlAsVariant final : public TCallNode {
  809. public:
  810. TYqlAsVariant(TPosition pos, const TVector<TNodePtr>& args)
  811. : TCallNode(pos, "AsVariant", 2, 2, args)
  812. {}
  813. bool DoInit(TContext& ctx, ISource* src) override {
  814. if (!ValidateArguments(ctx)) {
  815. return false;
  816. }
  817. if (!Args[1]->Init(ctx, src)) {
  818. return false;
  819. }
  820. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  821. return TCallNode::DoInit(ctx, src);
  822. }
  823. TNodePtr DoClone() const final {
  824. return new TYqlAsVariant(Pos, CloneContainer(Args));
  825. }
  826. };
  827. class TYqlAsEnum final : public TCallNode {
  828. public:
  829. TYqlAsEnum(TPosition pos, const TVector<TNodePtr>& args)
  830. : TCallNode(pos, "AsEnum", 1, 1, args)
  831. {}
  832. bool DoInit(TContext& ctx, ISource* src) override {
  833. if (!ValidateArguments(ctx)) {
  834. return false;
  835. }
  836. if (!Args[0]->Init(ctx, src)) {
  837. return false;
  838. }
  839. Args[0] = MakeAtomFromExpression(Pos, ctx, Args[0]).Build();
  840. return TCallNode::DoInit(ctx, src);
  841. }
  842. TNodePtr DoClone() const final {
  843. return new TYqlAsEnum(Pos, CloneContainer(Args));
  844. }
  845. };
  846. TNodePtr BuildFileNameArgument(TPosition pos, const TNodePtr& argument, const TString& prefix) {
  847. return new TLiteralStringAtom(pos, argument, "FilePath requires string literal as parameter", prefix);
  848. }
  849. template <typename TDerived, bool IsFile>
  850. class TYqlAtomBase: public TCallNode {
  851. public:
  852. TYqlAtomBase(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  853. : TCallNode(pos, opName, 1, 1, args)
  854. {}
  855. bool DoInit(TContext& ctx, ISource* src) override {
  856. if (!Args.empty()) {
  857. Args[0] = BuildFileNameArgument(Pos, Args[0], IsFile ? ctx.Settings.FileAliasPrefix : TString());
  858. }
  859. return TCallNode::DoInit(ctx, src);
  860. }
  861. TNodePtr DoClone() const final {
  862. return new TDerived(Pos, OpName, CloneContainer(Args));
  863. }
  864. bool IsLiteral() const override {
  865. return !Args.empty() ? Args[0]->IsLiteral() : false;
  866. }
  867. TString GetLiteralType() const override {
  868. return !Args.empty() ? Args[0]->GetLiteralType() : "";
  869. }
  870. TString GetLiteralValue() const override {
  871. return !Args.empty() ? Args[0]->GetLiteralValue() : "";
  872. }
  873. };
  874. class TYqlAtom final : public TYqlAtomBase<TYqlAtom, false>
  875. {
  876. using TBase = TYqlAtomBase<TYqlAtom, false>;
  877. using TBase::TBase;
  878. };
  879. class TFileYqlAtom final : public TYqlAtomBase<TFileYqlAtom, true>
  880. {
  881. using TBase = TYqlAtomBase<TFileYqlAtom, true>;
  882. using TBase::TBase;
  883. };
  884. class TTryMember final: public TCallNode {
  885. public:
  886. TTryMember(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  887. : TCallNode(pos, opName, 3, 3, args)
  888. {}
  889. bool DoInit(TContext& ctx, ISource* src) override {
  890. if (Args.size() != 3) {
  891. ctx.Error(Pos) << OpName << " requires exactly three arguments";
  892. return false;
  893. }
  894. for (const auto& arg : Args) {
  895. if (!arg->Init(ctx, src)) {
  896. return false;
  897. }
  898. }
  899. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  900. return TCallNode::DoInit(ctx, src);
  901. }
  902. TNodePtr DoClone() const final {
  903. return new TTryMember(Pos, OpName, CloneContainer(Args));
  904. }
  905. };
  906. template<bool Pretty>
  907. class TFormatTypeDiff final: public TCallNode {
  908. public:
  909. TFormatTypeDiff(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  910. : TCallNode(pos, opName, 3, 3, args)
  911. {}
  912. bool DoInit(TContext& ctx, ISource* src) override {
  913. if (Args.size() != 2) {
  914. ctx.Error(Pos) << OpName << " requires exactly 2 arguments";
  915. return false;
  916. }
  917. for (const auto& arg : Args) {
  918. if (!arg->Init(ctx, src)) {
  919. return false;
  920. }
  921. }
  922. Args.push_back(Q(Pretty ? "true" : "false"));
  923. OpName = "FormatTypeDiff";
  924. return TCallNode::DoInit(ctx, src);
  925. }
  926. TNodePtr DoClone() const final {
  927. return new TFormatTypeDiff<Pretty>(GetPos(), OpName, CloneContainer(Args));
  928. }
  929. };
  930. class TAddMember final: public TCallNode {
  931. public:
  932. TAddMember(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  933. : TCallNode(pos, opName, 3, 3, args)
  934. {}
  935. bool DoInit(TContext& ctx, ISource* src) override {
  936. if (Args.size() != 3) {
  937. ctx.Error(Pos) << OpName << " requires exactly three arguments";
  938. return false;
  939. }
  940. for (const auto& arg : Args) {
  941. if (!arg->Init(ctx, src)) {
  942. return false;
  943. }
  944. }
  945. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  946. return TCallNode::DoInit(ctx, src);
  947. }
  948. TNodePtr DoClone() const final {
  949. return new TAddMember(Pos, OpName, CloneContainer(Args));
  950. }
  951. };
  952. class TRemoveMember final: public TCallNode {
  953. public:
  954. TRemoveMember(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  955. : TCallNode(pos, opName, 2, 2, args)
  956. {}
  957. bool DoInit(TContext& ctx, ISource* src) override {
  958. if (Args.size() != 2) {
  959. ctx.Error(Pos) << OpName << " requires exactly two arguments";
  960. return false;
  961. }
  962. for (const auto& arg : Args) {
  963. if (!arg->Init(ctx, src)) {
  964. return false;
  965. }
  966. }
  967. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  968. return TCallNode::DoInit(ctx, src);
  969. }
  970. TNodePtr DoClone() const final {
  971. return new TRemoveMember(Pos, OpName, CloneContainer(Args));
  972. }
  973. };
  974. class TCombineMembers final: public TCallNode {
  975. public:
  976. TCombineMembers(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  977. : TCallNode(pos, opName, 1, -1, args)
  978. {}
  979. bool DoInit(TContext& ctx, ISource* src) override {
  980. if (Args.empty()) {
  981. ctx.Error(Pos) << "CombineMembers requires at least one argument";
  982. return false;
  983. }
  984. for (size_t i = 0; i < Args.size(); ++i) {
  985. Args[i] = Q(Y(Q(""), Args[i])); // flatten without prefix
  986. }
  987. return TCallNode::DoInit(ctx, src);
  988. }
  989. TNodePtr DoClone() const final {
  990. return new TCombineMembers(Pos, OpName, CloneContainer(Args));
  991. }
  992. };
  993. class TFlattenMembers final: public TCallNode {
  994. public:
  995. TFlattenMembers(TPosition pos, const TString& opName, const TVector<TNodePtr>& args)
  996. : TCallNode(pos, opName, 1, -1, args)
  997. {}
  998. bool DoInit(TContext& ctx, ISource* src) override {
  999. if (Args.empty()) {
  1000. ctx.Error(Pos) << OpName << " requires at least one argument";
  1001. return false;
  1002. }
  1003. for (size_t i = 0; i < Args.size(); ++i) {
  1004. if (!Args[i]->Init(ctx, src)) {
  1005. return false;
  1006. }
  1007. if (Args[i]->GetTupleSize() == 2) {
  1008. // flatten with prefix
  1009. Args[i] = Q(Y(
  1010. MakeAtomFromExpression(Pos, ctx, Args[i]->GetTupleElement(0)).Build(),
  1011. Args[i]->GetTupleElement(1)
  1012. ));
  1013. } else {
  1014. ctx.Error(Pos) << OpName << " requires arguments to be tuples of size 2: prefix and struct";
  1015. return false;
  1016. }
  1017. }
  1018. return TCallNode::DoInit(ctx, src);
  1019. }
  1020. TNodePtr DoClone() const final {
  1021. return new TFlattenMembers(Pos, OpName, CloneContainer(Args));
  1022. }
  1023. };
  1024. TString NormalizeTypeString(const TString& str) {
  1025. auto ret = to_title(str);
  1026. if (ret.StartsWith("Tz")) {
  1027. ret = "Tz" + to_title(ret.substr(2));
  1028. }
  1029. if (ret.StartsWith("Json")) {
  1030. ret = "Json" + to_title(ret.substr(4));
  1031. }
  1032. if (ret.StartsWith("Dy")) {
  1033. ret = "Dy" + to_title(ret.substr(2));
  1034. }
  1035. return ret;
  1036. }
  1037. static const TSet<TString> AvailableDataTypes = {"Bool", "String", "Uint32", "Uint64", "Int32", "Int64", "Float", "Double", "Utf8", "Yson", "Json", "JsonDocument",
  1038. "Date", "Datetime", "Timestamp", "Interval", "Uint8", "Int8", "Uint16", "Int16", "TzDate", "TzDatetime", "TzTimestamp", "Uuid", "Decimal", "DyNumber",
  1039. "Date32", "Datetime64", "Timestamp64", "Interval64", "TzDate32", "TzDatetime64", "TzTimestamp64"};
  1040. TNodePtr GetDataTypeStringNode(TContext& ctx, TCallNode& node, unsigned argNum, TString* outTypeStrPtr = nullptr) {
  1041. auto errMsgFunc = [&node, argNum]() {
  1042. static std::array<TString, 2> numToName = {{"first", "second"}};
  1043. TStringBuilder sb;
  1044. sb << "At " << numToName.at(argNum) << " argument of " << node.GetOpName() << " expected type string, available one of: "
  1045. << JoinRange(", ", AvailableDataTypes.begin(), AvailableDataTypes.end()) << ";";
  1046. return TString(sb);
  1047. };
  1048. auto typeStringNode = node.GetArgs().at(argNum);
  1049. auto typeStringPtr = typeStringNode->GetLiteral("String");
  1050. TNodePtr dataTypeNode;
  1051. if (typeStringPtr) {
  1052. TString typeString = NormalizeTypeString(*typeStringPtr);
  1053. if (!AvailableDataTypes.contains(typeString)) {
  1054. ctx.Error(typeStringNode->GetPos()) << "Bad type string: '" << typeString << "'. " << errMsgFunc();
  1055. return {};
  1056. }
  1057. if (outTypeStrPtr) {
  1058. *outTypeStrPtr = typeString;
  1059. }
  1060. dataTypeNode = typeStringNode->Q(typeString);
  1061. } else {
  1062. ctx.Error(typeStringNode->GetPos()) << errMsgFunc();
  1063. return {};
  1064. }
  1065. return dataTypeNode;
  1066. }
  1067. class TYqlParseFileOp final: public TCallNode {
  1068. public:
  1069. TYqlParseFileOp(TPosition pos, const TVector<TNodePtr>& args)
  1070. : TCallNode(pos, "ParseFile", 2, 2, args)
  1071. {}
  1072. bool DoInit(TContext& ctx, ISource* src) override {
  1073. if (!ValidateArguments(ctx)) {
  1074. return false;
  1075. }
  1076. auto dataTypeStringNode = GetDataTypeStringNode(ctx, *this, 0);
  1077. if (!dataTypeStringNode) {
  1078. return false;
  1079. }
  1080. auto aliasNode = BuildFileNameArgument(Args[1]->GetPos(), Args[1], ctx.Settings.FileAliasPrefix);
  1081. OpName = "Apply";
  1082. Args[0] = Y("Udf", Q("File.ByLines"), Y("Void"),
  1083. Y("TupleType",
  1084. Y("TupleType", Y("DataType", dataTypeStringNode)),
  1085. Y("StructType"),
  1086. Y("TupleType")));
  1087. Args[1] = Y("FilePath", aliasNode);
  1088. return TCallNode::DoInit(ctx, src);
  1089. }
  1090. TString GetOpName() const override {
  1091. return "ParseFile";
  1092. }
  1093. TNodePtr DoClone() const final {
  1094. return new TYqlParseFileOp(Pos, CloneContainer(Args));
  1095. }
  1096. };
  1097. class TYqlDataType final : public TCallNode {
  1098. public:
  1099. TYqlDataType(TPosition pos, const TVector<TNodePtr>& args)
  1100. : TCallNode(pos, "DataType", 1, 3, args)
  1101. {
  1102. FakeSource = BuildFakeSource(pos);
  1103. }
  1104. bool DoInit(TContext& ctx, ISource* src) override {
  1105. if (!ValidateArguments(ctx)) {
  1106. return false;
  1107. }
  1108. for (ui32 i = 0; i < Args.size(); ++i) {
  1109. if (!Args[i]->Init(ctx, FakeSource.Get())) {
  1110. return false;
  1111. }
  1112. Args[i] = MakeAtomFromExpression(Pos, ctx, Args[i]).Build();
  1113. }
  1114. return TCallNode::DoInit(ctx, src);
  1115. }
  1116. TNodePtr DoClone() const final {
  1117. return new TYqlDataType(Pos, CloneContainer(Args));
  1118. }
  1119. private:
  1120. TSourcePtr FakeSource;
  1121. };
  1122. class TYqlResourceType final : public TCallNode {
  1123. public:
  1124. TYqlResourceType(TPosition pos, const TVector<TNodePtr>& args)
  1125. : TCallNode(pos, "ResourceType", 1, 1, args)
  1126. {}
  1127. bool DoInit(TContext& ctx, ISource* src) override {
  1128. if (!ValidateArguments(ctx)) {
  1129. return false;
  1130. }
  1131. if (!Args[0]->Init(ctx, src)) {
  1132. return false;
  1133. }
  1134. Args[0] = MakeAtomFromExpression(Pos, ctx, Args[0]).Build();
  1135. return TCallNode::DoInit(ctx, src);
  1136. }
  1137. TNodePtr DoClone() const final {
  1138. return new TYqlResourceType(Pos, CloneContainer(Args));
  1139. }
  1140. };
  1141. class TYqlTaggedType final : public TCallNode {
  1142. public:
  1143. TYqlTaggedType(TPosition pos, const TVector<TNodePtr>& args)
  1144. : TCallNode(pos, "TaggedType", 2, 2, args)
  1145. {}
  1146. bool DoInit(TContext& ctx, ISource* src) override {
  1147. if (!ValidateArguments(ctx)) {
  1148. return false;
  1149. }
  1150. if (!Args[1]->Init(ctx, src)) {
  1151. return false;
  1152. }
  1153. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  1154. return TCallNode::DoInit(ctx, src);
  1155. }
  1156. TNodePtr DoClone() const final {
  1157. return new TYqlTaggedType(Pos, CloneContainer(Args));
  1158. }
  1159. };
  1160. class TYqlCallableType final : public TCallNode {
  1161. public:
  1162. TYqlCallableType(TPosition pos, const TVector<TNodePtr>& args)
  1163. : TCallNode(pos, "CallableType", 2, -1, args)
  1164. {}
  1165. bool DoInit(TContext& ctx, ISource* src) override {
  1166. if (!ValidateArguments(ctx)) {
  1167. return false;
  1168. }
  1169. if (!Args[0]->GetTupleNode()) {
  1170. ui32 numOptArgs;
  1171. if (!Parseui32(Args[0], numOptArgs)) {
  1172. ctx.Error(Args[0]->GetPos()) << "Expected either tuple or number of optional arguments";
  1173. return false;
  1174. }
  1175. Args[0] = Q(Y(BuildQuotedAtom(Args[0]->GetPos(), ToString(numOptArgs))));
  1176. }
  1177. if (!Args[1]->GetTupleNode()) {
  1178. Args[1] = Q(Y(Args[1]));
  1179. }
  1180. for (ui32 index = 2; index < Args.size(); ++index) {
  1181. if (!Args[index]->GetTupleNode()) {
  1182. Args[index] = Q(Y(Args[index]));
  1183. }
  1184. }
  1185. return TCallNode::DoInit(ctx, src);
  1186. }
  1187. TNodePtr DoClone() const final {
  1188. return new TYqlCallableType(Pos, CloneContainer(Args));
  1189. }
  1190. };
  1191. class TYqlTupleElementType final : public TCallNode {
  1192. public:
  1193. TYqlTupleElementType(TPosition pos, const TVector<TNodePtr>& args)
  1194. : TCallNode(pos, "TupleElementType", 2, 2, args)
  1195. {}
  1196. bool DoInit(TContext& ctx, ISource* src) override {
  1197. if (!ValidateArguments(ctx)) {
  1198. return false;
  1199. }
  1200. if (!Args[1]->Init(ctx, src)) {
  1201. return false;
  1202. }
  1203. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  1204. return TCallNode::DoInit(ctx, src);
  1205. }
  1206. TNodePtr DoClone() const final {
  1207. return new TYqlTupleElementType(Pos, CloneContainer(Args));
  1208. }
  1209. };
  1210. class TYqlStructMemberType final : public TCallNode {
  1211. public:
  1212. TYqlStructMemberType(TPosition pos, const TVector<TNodePtr>& args)
  1213. : TCallNode(pos, "StructMemberType", 2, 2, args)
  1214. {}
  1215. bool DoInit(TContext& ctx, ISource* src) override {
  1216. if (!ValidateArguments(ctx)) {
  1217. return false;
  1218. }
  1219. if (!Args[1]->Init(ctx, src)) {
  1220. return false;
  1221. }
  1222. Args[1] = MakeAtomFromExpression(Pos, ctx, Args[1]).Build();
  1223. return TCallNode::DoInit(ctx, src);
  1224. }
  1225. TNodePtr DoClone() const final {
  1226. return new TYqlStructMemberType(Pos, CloneContainer(Args));
  1227. }
  1228. };
  1229. class TYqlCallableArgumentType final : public TCallNode {
  1230. public:
  1231. TYqlCallableArgumentType(TPosition pos, const TVector<TNodePtr>& args)
  1232. : TCallNode(pos, "CallableArgumentType", 2, 2, args)
  1233. {}
  1234. bool DoInit(TContext& ctx, ISource* src) override {
  1235. if (!ValidateArguments(ctx)) {
  1236. return false;
  1237. }
  1238. ui32 index;
  1239. if (!Parseui32(Args[1], index)) {
  1240. ctx.Error(Args[1]->GetPos()) << "Expected index of the callable argument";
  1241. return false;
  1242. }
  1243. Args[1] = BuildQuotedAtom(Args[1]->GetPos(), ToString(index));
  1244. return TCallNode::DoInit(ctx, src);
  1245. }
  1246. TNodePtr DoClone() const final {
  1247. return new TYqlCallableArgumentType(Pos, CloneContainer(Args));
  1248. }
  1249. };
  1250. class TStructTypeNode : public TAstListNode {
  1251. public:
  1252. TStructTypeNode(TPosition pos, const TVector<TNodePtr>& exprs)
  1253. : TAstListNode(pos)
  1254. , Exprs(exprs)
  1255. {}
  1256. bool DoInit(TContext& ctx, ISource* src) override {
  1257. Nodes.push_back(BuildAtom(Pos, "StructType", TNodeFlags::Default));
  1258. for (const auto& expr : Exprs) {
  1259. const auto& label = expr->GetLabel();
  1260. if (!label) {
  1261. ctx.Error(expr->GetPos()) << "Structure does not allow anonymous members";
  1262. return false;
  1263. }
  1264. Nodes.push_back(Q(Y(Q(label), expr)));
  1265. }
  1266. return TAstListNode::DoInit(ctx, src);
  1267. }
  1268. TNodePtr DoClone() const final {
  1269. return new TStructTypeNode(Pos, CloneContainer(Exprs));
  1270. }
  1271. private:
  1272. const TVector<TNodePtr> Exprs;
  1273. };
  1274. template <bool IsStrict>
  1275. class TYqlIf final: public TCallNode {
  1276. public:
  1277. TYqlIf(TPosition pos, const TVector<TNodePtr>& args)
  1278. : TCallNode(pos, IsStrict ? "IfStrict" : "If", 2, 3, args)
  1279. {}
  1280. private:
  1281. TCallNode::TPtr DoClone() const override {
  1282. return new TYqlIf(GetPos(), CloneContainer(Args));
  1283. }
  1284. bool DoInit(TContext& ctx, ISource* src) override {
  1285. if (!ValidateArguments(ctx)) {
  1286. return false;
  1287. }
  1288. Args[0] = Y("Coalesce", Args[0], Y("Bool", Q("false")));
  1289. if (Args.size() == 2) {
  1290. Args.push_back(Y("Null"));
  1291. }
  1292. return TCallNode::DoInit(ctx, src);
  1293. }
  1294. };
  1295. class TYqlSubstring final: public TCallNode {
  1296. public:
  1297. TYqlSubstring(TPosition pos, const TString& name, const TVector<TNodePtr>& args)
  1298. : TCallNode(pos, name, 2, 3, args)
  1299. {}
  1300. private:
  1301. TCallNode::TPtr DoClone() const override {
  1302. return new TYqlSubstring(GetPos(), OpName, CloneContainer(Args));
  1303. }
  1304. bool DoInit(TContext& ctx, ISource* src) override {
  1305. if (Args.size() == 2) {
  1306. Args.push_back(Y("Null"));
  1307. }
  1308. return TCallNode::DoInit(ctx, src);
  1309. }
  1310. };
  1311. class TYqlIn final: public TCallNode {
  1312. public:
  1313. TYqlIn(TPosition pos, const TVector<TNodePtr>& args)
  1314. : TCallNode(pos, "IN", 3, 3, args)
  1315. {}
  1316. private:
  1317. TNodePtr DoClone() const final {
  1318. return new TYqlIn(Pos, CloneContainer(Args));
  1319. }
  1320. bool DoInit(TContext& ctx, ISource* src) override {
  1321. if (!ValidateArguments(ctx)) {
  1322. return false;
  1323. }
  1324. auto key = Args[0];
  1325. auto inNode = Args[1];
  1326. auto hints = Args[2];
  1327. const auto pos = inNode->GetPos();
  1328. if (!key->Init(ctx, src)) {
  1329. return false;
  1330. }
  1331. if (!inNode->Init(ctx, inNode->GetSource() ? nullptr : src)) {
  1332. return false;
  1333. }
  1334. if (inNode->GetLiteral("String")) {
  1335. ctx.Error(pos) << "Unable to use IN predicate with string argument, it won't search substring - "
  1336. "expecting tuple, list, dict or single column table source";
  1337. return false;
  1338. }
  1339. if (inNode->GetTupleSize() == 1) {
  1340. auto singleElement = inNode->GetTupleElement(0);
  1341. // TODO: 'IN ((select ...))' is parsed exactly like 'IN (select ...)' instead of a single element tuple
  1342. if (singleElement->GetSource() || singleElement->IsSelect()) {
  1343. TStringBuf parenKind = singleElement->GetSource() ? "" : "external ";
  1344. ctx.Warning(pos,
  1345. TIssuesIds::YQL_CONST_SUBREQUEST_IN_LIST) << "Using subrequest in scalar context after IN, "
  1346. << "perhaps you should remove "
  1347. << parenKind << "parenthesis here";
  1348. }
  1349. }
  1350. TVector<TNodePtr> hintElements;
  1351. for (size_t i = 0; i < hints->GetTupleSize(); ++i) {
  1352. hintElements.push_back(hints->GetTupleElement(i));
  1353. }
  1354. if (inNode->GetSource() || inNode->IsSelect()) {
  1355. hintElements.push_back(BuildHint(pos, "tableSource"));
  1356. }
  1357. if (!ctx.AnsiInForEmptyOrNullableItemsCollections.Defined()) {
  1358. hintElements.push_back(BuildHint(pos, "warnNoAnsi"));
  1359. } else if (*ctx.AnsiInForEmptyOrNullableItemsCollections) {
  1360. hintElements.push_back(BuildHint(pos, "ansi"));
  1361. }
  1362. OpName = "SqlIn";
  1363. MinArgs = MaxArgs = 3;
  1364. Args = {
  1365. inNode->GetSource() ? inNode->GetSource() : inNode,
  1366. key,
  1367. BuildTuple(pos, hintElements)
  1368. };
  1369. return TCallNode::DoInit(ctx, src);
  1370. }
  1371. static TNodePtr BuildHint(TPosition pos, const TString& name) {
  1372. return BuildTuple(pos, { BuildQuotedAtom(pos, name, NYql::TNodeFlags::Default) });
  1373. }
  1374. TString GetOpName() const override {
  1375. return "IN predicate";
  1376. }
  1377. };
  1378. class TYqlUdfBase : public TCallNode {
  1379. public:
  1380. TYqlUdfBase(TPosition pos, const TString& name)
  1381. : TCallNode(pos, "Udf", 1, 1, UdfArgs(pos, name))
  1382. {}
  1383. TYqlUdfBase(TPosition pos, const TString& name, const TVector<TNodePtr>& args, ui32 argsCount = 2)
  1384. : TCallNode(pos, "Udf", argsCount, argsCount, UdfArgs(pos, name, &args))
  1385. {}
  1386. protected:
  1387. TYqlUdfBase(TPosition pos, const TString& opName, ui32 minArgs, ui32 maxArgs, const TVector<TNodePtr>& args)
  1388. : TCallNode(pos, opName, minArgs, maxArgs, args)
  1389. {}
  1390. private:
  1391. static TVector<TNodePtr> UdfArgs(TPosition pos, const TString& name, const TVector<TNodePtr>* args = nullptr) {
  1392. TVector<TNodePtr> res = { BuildQuotedAtom(pos, name) };
  1393. if (args) {
  1394. res.insert(res.end(), args->begin(), args->end());
  1395. }
  1396. return res;
  1397. }
  1398. void DoUpdateState() const override {
  1399. TCallNode::DoUpdateState();
  1400. State.Set(ENodeState::Aggregated, false/*!RunConfig || RunConfig->IsAggregated()*/);
  1401. State.Set(ENodeState::Const, true /* FIXME: To avoid CheckAggregationLevel issue for non-const TypeOf. */);
  1402. }
  1403. private:
  1404. TNodePtr RunConfig;
  1405. };
  1406. class TYqlUdf final : public TYqlUdfBase {
  1407. public:
  1408. TYqlUdf(TPosition pos, const TString& name)
  1409. : TYqlUdfBase(pos, name)
  1410. {}
  1411. TYqlUdf(TPosition pos, const TString& name, const TVector<TNodePtr>& args, ui32 argsCount = 2)
  1412. : TYqlUdfBase(pos, name, args, argsCount)
  1413. {}
  1414. private:
  1415. TYqlUdf(const TYqlUdf& other)
  1416. : TYqlUdfBase(other.GetPos(), "Udf", other.MinArgs, other.MaxArgs, CloneContainer(other.Args))
  1417. {}
  1418. TNodePtr DoClone() const final {
  1419. return new TYqlUdf(*this);
  1420. }
  1421. };
  1422. class TYqlTypeConfigUdf final : public TYqlUdfBase {
  1423. public:
  1424. TYqlTypeConfigUdf(TPosition pos, const TString& name)
  1425. : TYqlUdfBase(pos, name)
  1426. {}
  1427. TYqlTypeConfigUdf(TPosition pos, const TString& name, const TVector<TNodePtr>& args, ui32 argsCount = 2)
  1428. : TYqlUdfBase(pos, name, args, argsCount)
  1429. {}
  1430. private:
  1431. TYqlTypeConfigUdf(const TYqlTypeConfigUdf& other)
  1432. : TYqlUdfBase(other.GetPos(), "Udf", other.MinArgs, other.MaxArgs, CloneContainer(other.Args))
  1433. {}
  1434. bool DoInit(TContext& ctx, ISource* src) override {
  1435. if (!ValidateArguments(ctx)) {
  1436. return false;
  1437. }
  1438. if (!Args[3]->Init(ctx, src)) {
  1439. return false;
  1440. }
  1441. Args[3] = MakeAtomFromExpression(Pos, ctx, Args[3]).Build();
  1442. return TYqlUdfBase::DoInit(ctx, src);
  1443. }
  1444. TNodePtr DoClone() const final {
  1445. return new TYqlTypeConfigUdf(*this);
  1446. }
  1447. };
  1448. class TWeakFieldOp final: public TCallNode {
  1449. public:
  1450. TWeakFieldOp(TPosition pos, const TVector<TNodePtr>& args)
  1451. : TCallNode(pos, "WeakField", 2, 3, args)
  1452. {}
  1453. bool DoInit(TContext& ctx, ISource* src) override {
  1454. if (!src) {
  1455. ctx.Error(Pos) << GetCallExplain() << " unable use without source";
  1456. return false;
  1457. }
  1458. src->AllColumns();
  1459. if (!ValidateArguments(ctx)) {
  1460. return false;
  1461. }
  1462. bool hasError = false;
  1463. for (auto& arg: Args) {
  1464. if (!arg->Init(ctx, src)) {
  1465. hasError = true;
  1466. continue;
  1467. }
  1468. }
  1469. if (hasError) {
  1470. return false;
  1471. }
  1472. PrecacheState();
  1473. const auto memberPos = Args[0]->GetPos();
  1474. TVector<TNodePtr> repackArgs = {BuildAtom(memberPos, "row", NYql::TNodeFlags::Default)};
  1475. if (auto literal = Args[1]->GetLiteral("String")) {
  1476. TString targetType;
  1477. if (!GetDataTypeStringNode(ctx, *this, 1, &targetType)) {
  1478. return false;
  1479. }
  1480. repackArgs.push_back(Args[1]->Q(targetType));
  1481. } else {
  1482. repackArgs.push_back(Args[1]);
  1483. }
  1484. TVector<TNodePtr> column;
  1485. auto namePtr = Args[0]->GetColumnName();
  1486. if (!namePtr || !*namePtr) {
  1487. ctx.Error(Pos) << GetCallExplain() << " expects column name as first argument";
  1488. return false;
  1489. }
  1490. auto memberName = *namePtr;
  1491. column.push_back(Args[0]->Q(*namePtr));
  1492. if (src->GetJoin() && !src->IsJoinKeysInitializing()) {
  1493. const auto sourcePtr = Args[0]->GetSourceName();
  1494. if (!sourcePtr || !*sourcePtr) {
  1495. ctx.Error(Pos) << GetOpName() << " required to have correlation name in case of JOIN for column at first parameter";
  1496. return false;
  1497. }
  1498. column.push_back(Args[0]->Q(*sourcePtr));
  1499. memberName = DotJoin(*sourcePtr, memberName);
  1500. }
  1501. if (!GetLabel()) {
  1502. SetLabel(memberName);
  1503. }
  1504. repackArgs.push_back(BuildTuple(memberPos, column));
  1505. if (Args.size() == 3) {
  1506. repackArgs.push_back(Args[2]);
  1507. }
  1508. ++MinArgs;
  1509. ++MaxArgs;
  1510. Args.swap(repackArgs);
  1511. return TCallNode::DoInit(ctx, src);
  1512. }
  1513. TNodePtr DoClone() const final {
  1514. return new TWeakFieldOp(Pos, CloneContainer(Args));
  1515. }
  1516. };
  1517. template <bool Join>
  1518. class TTableRow final : public INode {
  1519. public:
  1520. TTableRow(TPosition pos, const TVector<TNodePtr>& args)
  1521. : TTableRow(pos, args.size())
  1522. {}
  1523. TTableRow(TPosition pos, ui32 argsCount)
  1524. : INode(pos)
  1525. , ArgsCount(argsCount)
  1526. {}
  1527. bool DoInit(TContext& ctx, ISource* src) override {
  1528. if (!src || src->IsFake()) {
  1529. ctx.Error(Pos) << TStringBuilder() << (Join ? "Join" : "") << "TableRow requires data source";
  1530. return false;
  1531. }
  1532. if (ArgsCount > 0) {
  1533. ctx.Error(Pos) << "TableRow requires exactly 0 arguments";
  1534. return false;
  1535. }
  1536. src->AllColumns();
  1537. const bool isJoin = src->GetJoin();
  1538. if (!Join && ctx.SimpleColumns && isJoin) {
  1539. TNodePtr block = Y();
  1540. const auto& sameKeyMap = src->GetJoin()->GetSameKeysMap();
  1541. if (sameKeyMap) {
  1542. block = L(block, Y("let", "flatSameKeys", "row"));
  1543. for (const auto& sameKeysPair: sameKeyMap) {
  1544. const auto& column = sameKeysPair.first;
  1545. auto keys = Y("Coalesce");
  1546. auto sameSourceIter = sameKeysPair.second.begin();
  1547. for (auto end = sameKeysPair.second.end(); sameSourceIter != end; ++sameSourceIter) {
  1548. auto addKeyNode = Q(DotJoin(*sameSourceIter, column));
  1549. keys = L(keys, Y("TryMember", "row", addKeyNode, Y("Null")));
  1550. }
  1551. block = L(block, Y("let", "flatSameKeys", Y("AddMember", "flatSameKeys", Q(column), keys)));
  1552. sameSourceIter = sameKeysPair.second.begin();
  1553. for (auto end = sameKeysPair.second.end(); sameSourceIter != end; ++sameSourceIter) {
  1554. auto removeKeyNode = Q(DotJoin(*sameSourceIter, column));
  1555. block = L(block, Y("let", "flatSameKeys", Y("ForceRemoveMember", "flatSameKeys", removeKeyNode)));
  1556. }
  1557. }
  1558. block = L(block, Y("let", "row", "flatSameKeys"));
  1559. }
  1560. auto members = Y();
  1561. for (auto& joinLabel: src->GetJoin()->GetJoinLabels()) {
  1562. members = L(members, BuildQuotedAtom(Pos, joinLabel + "."));
  1563. }
  1564. block = L(block, Y("let", "res", Y("DivePrefixMembers", "row", Q(members))));
  1565. for (const auto& sameKeysPair: src->GetJoin()->GetSameKeysMap()) {
  1566. const auto& column = sameKeysPair.first;
  1567. auto addMemberKeyNode = Y("Member", "row", Q(column));
  1568. block = L(block, Y("let", "res", Y("AddMember", "res", Q(column), addMemberKeyNode)));
  1569. }
  1570. Node = Y("block", Q(L(block, Y("return", "res"))));
  1571. } else {
  1572. Node = ctx.EnableSystemColumns ? Y("RemoveSystemMembers", "row") : BuildAtom(Pos, "row", 0);
  1573. }
  1574. return true;
  1575. }
  1576. TAstNode* Translate(TContext& ctx) const override {
  1577. Y_DEBUG_ABORT_UNLESS(Node);
  1578. return Node->Translate(ctx);
  1579. }
  1580. void DoUpdateState() const override {
  1581. State.Set(ENodeState::Const, false);
  1582. }
  1583. TNodePtr DoClone() const final {
  1584. return new TTableRow<Join>(Pos, ArgsCount);
  1585. }
  1586. bool IsTableRow() const final {
  1587. return true;
  1588. }
  1589. private:
  1590. const size_t ArgsCount;
  1591. TNodePtr Node;
  1592. };
  1593. TTableRows::TTableRows(TPosition pos, const TVector<TNodePtr>& args)
  1594. : TTableRows(pos, args.size())
  1595. {}
  1596. TTableRows::TTableRows(TPosition pos, ui32 argsCount)
  1597. : INode(pos)
  1598. , ArgsCount(argsCount)
  1599. {}
  1600. bool TTableRows::DoInit(TContext& ctx, ISource* /*src*/) {
  1601. if (ArgsCount > 0) {
  1602. ctx.Error(Pos) << "TableRows requires exactly 0 arguments";
  1603. return false;
  1604. }
  1605. Node = ctx.EnableSystemColumns ? Y("RemoveSystemMembers", "inputRowsList") : BuildAtom(Pos, "inputRowsList", 0);
  1606. return true;
  1607. }
  1608. TAstNode* TTableRows::Translate(TContext& ctx) const {
  1609. Y_DEBUG_ABORT_UNLESS(Node);
  1610. return Node->Translate(ctx);
  1611. }
  1612. void TTableRows::DoUpdateState() const {
  1613. State.Set(ENodeState::Const, false);
  1614. }
  1615. TNodePtr TTableRows::DoClone() const {
  1616. return MakeIntrusive<TTableRows>(Pos, ArgsCount);
  1617. }
  1618. TSessionWindow::TSessionWindow(TPosition pos, const TVector<TNodePtr>& args)
  1619. : INode(pos)
  1620. , Args(args)
  1621. , FakeSource(BuildFakeSource(pos))
  1622. , Valid(false)
  1623. {}
  1624. void TSessionWindow::MarkValid() {
  1625. YQL_ENSURE(!HasState(ENodeState::Initialized));
  1626. Valid = true;
  1627. }
  1628. TNodePtr TSessionWindow::BuildTraits(const TString& label) const {
  1629. YQL_ENSURE(HasState(ENodeState::Initialized));
  1630. auto trueNode = Y("Bool", Q("true"));
  1631. if (Args.size() == 2) {
  1632. auto timeExpr = Args[0];
  1633. auto timeoutExpr = Args[1];
  1634. auto coalesceLess = [&](auto first, auto second) {
  1635. // first < second ?? true
  1636. return Y("Coalesce", Y("<", first, second), trueNode);
  1637. };
  1638. auto absDelta = Y("If",
  1639. coalesceLess("prev", "curr"),
  1640. Y("-", "curr", "prev"),
  1641. Y("-", "prev", "curr"));
  1642. auto newSessionPred = Y("And", Y("AggrNotEquals", "curr", "prev"), coalesceLess(timeoutExpr, absDelta));
  1643. auto timeoutLambda = BuildLambda(timeoutExpr->GetPos(), Y("prev", "curr"), newSessionPred);
  1644. auto sortSpec = Y("SortTraits", Y("TypeOf", label), trueNode, BuildLambda(Pos, Y("row"), Y("PersistableRepr", timeExpr)));
  1645. return Y("SessionWindowTraits",
  1646. Y("TypeOf", label),
  1647. sortSpec,
  1648. BuildLambda(Pos, Y("row"), timeExpr),
  1649. timeoutLambda);
  1650. }
  1651. auto orderExpr = Args[0];
  1652. auto initLambda = Args[1];
  1653. auto updateLambda = Args[2];
  1654. auto calculateLambda = Args[3];
  1655. auto sortSpec = Y("SortTraits", Y("TypeOf", label), trueNode, BuildLambda(Pos, Y("row"), Y("PersistableRepr", orderExpr)));
  1656. return Y("SessionWindowTraits",
  1657. Y("TypeOf", label),
  1658. sortSpec,
  1659. initLambda,
  1660. updateLambda,
  1661. calculateLambda);
  1662. }
  1663. bool TSessionWindow::DoInit(TContext& ctx, ISource* src) {
  1664. if (!src || src->IsFake()) {
  1665. ctx.Error(Pos) << "SessionWindow requires data source";
  1666. return false;
  1667. }
  1668. if (!(Args.size() == 2 || Args.size() == 4)) {
  1669. ctx.Error(Pos) << "SessionWindow requires either two or four arguments";
  1670. return false;
  1671. }
  1672. if (!Valid) {
  1673. ctx.Error(Pos) << "SessionWindow can only be used as a top-level GROUP BY / PARTITION BY expression";
  1674. return false;
  1675. }
  1676. if (Args.size() == 2) {
  1677. auto timeExpr = Args[0];
  1678. auto timeoutExpr = Args[1];
  1679. return timeExpr->Init(ctx, src) && timeoutExpr->Init(ctx, FakeSource.Get());
  1680. }
  1681. auto orderExpr = Args[0];
  1682. auto initLambda = Args[1];
  1683. auto updateLambda = Args[2];
  1684. auto calculateLambda = Args[3];
  1685. src->AllColumns();
  1686. return orderExpr->Init(ctx, src) && initLambda->Init(ctx, FakeSource.Get()) &&
  1687. updateLambda->Init(ctx, FakeSource.Get()) && calculateLambda->Init(ctx, FakeSource.Get());
  1688. }
  1689. TAstNode* TSessionWindow::Translate(TContext&) const {
  1690. YQL_ENSURE(false, "Translate is called for SessionWindow");
  1691. return nullptr;
  1692. }
  1693. void TSessionWindow::DoUpdateState() const {
  1694. State.Set(ENodeState::Const, false);
  1695. }
  1696. TNodePtr TSessionWindow::DoClone() const {
  1697. return new TSessionWindow(Pos, CloneContainer(Args));
  1698. }
  1699. TString TSessionWindow::GetOpName() const {
  1700. return "SessionWindow";
  1701. }
  1702. template<bool IsStart>
  1703. class TSessionStart final : public INode {
  1704. public:
  1705. TSessionStart(TPosition pos, const TVector<TNodePtr>& args)
  1706. : INode(pos)
  1707. , ArgsCount(args.size())
  1708. {
  1709. }
  1710. private:
  1711. TSessionStart(TPosition pos, size_t argsCount)
  1712. : INode(pos)
  1713. , ArgsCount(argsCount)
  1714. {}
  1715. bool DoInit(TContext& ctx, ISource* src) override {
  1716. if (!src || src->IsFake()) {
  1717. ctx.Error(Pos) << GetOpName() << " requires data source";
  1718. return false;
  1719. }
  1720. if (ArgsCount > 0) {
  1721. ctx.Error(Pos) << GetOpName() << " requires exactly 0 arguments";
  1722. return false;
  1723. }
  1724. auto windowName = src->GetWindowName();
  1725. OverWindow = windowName != nullptr;
  1726. TNodePtr sessionWindow;
  1727. if (windowName) {
  1728. auto spec = src->FindWindowSpecification(ctx, *windowName);
  1729. if (!spec) {
  1730. return false;
  1731. }
  1732. sessionWindow = spec->Session;
  1733. if (!sessionWindow) {
  1734. ctx.Error(Pos) << GetOpName() << " can not be used with window " << *windowName << ": SessionWindow specification is missing in PARTITION BY";
  1735. return false;
  1736. }
  1737. } else {
  1738. sessionWindow = src->GetSessionWindowSpec();
  1739. if (!sessionWindow) {
  1740. TString extra;
  1741. if (src->IsOverWindowSource()) {
  1742. extra = ". Maybe you forgot to add OVER `window_name`?";
  1743. }
  1744. if (src->HasAggregations()) {
  1745. ctx.Error(Pos) << GetOpName() << " can not be used here: SessionWindow specification is missing in GROUP BY" << extra;
  1746. } else {
  1747. ctx.Error(Pos) << GetOpName() << " can not be used without aggregation by SessionWindow" << extra;
  1748. }
  1749. return false;
  1750. }
  1751. if (!IsStart) {
  1752. ctx.Error(Pos) << GetOpName() << " with GROUP BY is not supported yet";
  1753. return false;
  1754. }
  1755. }
  1756. if (sessionWindow->HasState(ENodeState::Failed)) {
  1757. return false;
  1758. }
  1759. YQL_ENSURE(sessionWindow->HasState(ENodeState::Initialized));
  1760. YQL_ENSURE(sessionWindow->GetLabel());
  1761. Node = Y("Member", "row", BuildQuotedAtom(Pos, sessionWindow->GetLabel()));
  1762. if (OverWindow) {
  1763. Node = Y("Member", Node, BuildQuotedAtom(Pos, IsStart ? "start" : "state"));
  1764. }
  1765. return true;
  1766. }
  1767. TAstNode* Translate(TContext& ctx) const override {
  1768. Y_DEBUG_ABORT_UNLESS(Node);
  1769. return Node->Translate(ctx);
  1770. }
  1771. void DoUpdateState() const override {
  1772. State.Set(ENodeState::Const, false);
  1773. if (OverWindow) {
  1774. State.Set(ENodeState::OverWindow, true);
  1775. } else if (IsStart) {
  1776. State.Set(ENodeState::Aggregated, true);
  1777. }
  1778. }
  1779. TNodePtr DoClone() const override {
  1780. return new TSessionStart<IsStart>(Pos, ArgsCount);
  1781. }
  1782. TString GetOpName() const override {
  1783. return IsStart ? "SessionStart" : "SessionState";
  1784. }
  1785. const size_t ArgsCount;
  1786. bool OverWindow = false;
  1787. TNodePtr Node;
  1788. };
  1789. THoppingWindow::THoppingWindow(TPosition pos, const TVector<TNodePtr>& args)
  1790. : INode(pos)
  1791. , Args(args)
  1792. , FakeSource(BuildFakeSource(pos))
  1793. , Valid(false)
  1794. {}
  1795. void THoppingWindow::MarkValid() {
  1796. YQL_ENSURE(!HasState(ENodeState::Initialized));
  1797. Valid = true;
  1798. }
  1799. TNodePtr THoppingWindow::BuildTraits(const TString& label) const {
  1800. YQL_ENSURE(HasState(ENodeState::Initialized));
  1801. return Y(
  1802. "HoppingTraits",
  1803. Y("ListItemType", Y("TypeOf", label)),
  1804. BuildLambda(Pos, Y("row"), Y("Just", Y("SystemMetadata", Y("String", Q("write_time")), Y("DependsOn", "row")))),
  1805. Hop,
  1806. Interval,
  1807. Interval,
  1808. Q("true"),
  1809. Q("v2"));
  1810. }
  1811. bool THoppingWindow::DoInit(TContext& ctx, ISource* src) {
  1812. if (!src || src->IsFake()) {
  1813. ctx.Error(Pos) << "HoppingWindow requires data source";
  1814. return false;
  1815. }
  1816. if (!(Args.size() == 2)) {
  1817. ctx.Error(Pos) << "HoppingWindow requires two arguments";
  1818. return false;
  1819. }
  1820. if (!Valid) {
  1821. ctx.Error(Pos) << "HoppingWindow can only be used as a top-level GROUP BY expression";
  1822. return false;
  1823. }
  1824. auto hopExpr = Args[0];
  1825. auto intervalExpr = Args[1];
  1826. if (!(hopExpr->Init(ctx, FakeSource.Get()) && intervalExpr->Init(ctx, FakeSource.Get()))) {
  1827. return false;
  1828. }
  1829. Hop = ProcessIntervalParam(hopExpr);
  1830. Interval = ProcessIntervalParam(intervalExpr);
  1831. return true;
  1832. }
  1833. TAstNode* THoppingWindow::Translate(TContext&) const {
  1834. YQL_ENSURE(false, "Translate is called for HoppingWindow");
  1835. return nullptr;
  1836. }
  1837. void THoppingWindow::DoUpdateState() const {
  1838. State.Set(ENodeState::Const, false);
  1839. }
  1840. TNodePtr THoppingWindow::DoClone() const {
  1841. return new THoppingWindow(Pos, CloneContainer(Args));
  1842. }
  1843. TString THoppingWindow::GetOpName() const {
  1844. return "HoppingWindow";
  1845. }
  1846. TNodePtr THoppingWindow::ProcessIntervalParam(const TNodePtr& node) const {
  1847. auto literal = node->GetLiteral("String");
  1848. if (!literal) {
  1849. return Y("EvaluateExpr", node);
  1850. }
  1851. return new TYqlData(node->GetPos(), "Interval", {node});
  1852. }
  1853. TNodePtr BuildUdfUserTypeArg(TPosition pos, const TVector<TNodePtr>& args, TNodePtr customUserType) {
  1854. TVector<TNodePtr> argsTypeItems;
  1855. for (auto& arg : args) {
  1856. argsTypeItems.push_back(new TCallNodeImpl(pos, "TypeOf", TVector<TNodePtr>(1, arg)));
  1857. }
  1858. TVector<TNodePtr> userTypeItems;
  1859. userTypeItems.push_back(new TCallNodeImpl(pos, "TupleType", argsTypeItems));
  1860. userTypeItems.push_back(new TCallNodeImpl(pos, "StructType", {}));
  1861. if (customUserType) {
  1862. userTypeItems.push_back(customUserType);
  1863. } else {
  1864. userTypeItems.push_back(new TCallNodeImpl(pos, "TupleType", {}));
  1865. }
  1866. return new TCallNodeImpl(pos, "TupleType", userTypeItems);
  1867. }
  1868. TNodePtr BuildUdfUserTypeArg(TPosition pos, TNodePtr positionalArgs, TNodePtr namedArgs, TNodePtr customUserType) {
  1869. TVector<TNodePtr> userTypeItems;
  1870. userTypeItems.reserve(3);
  1871. userTypeItems.push_back(positionalArgs->Y("TypeOf", positionalArgs));
  1872. userTypeItems.push_back(positionalArgs->Y("TypeOf", namedArgs));
  1873. if (customUserType) {
  1874. userTypeItems.push_back(customUserType);
  1875. } else {
  1876. userTypeItems.push_back(new TCallNodeImpl(pos, "TupleType", {}));
  1877. }
  1878. return new TCallNodeImpl(pos, "TupleType", userTypeItems);
  1879. }
  1880. TVector<TNodePtr> BuildUdfArgs(const TContext& ctx, TPosition pos, const TVector<TNodePtr>& args,
  1881. TNodePtr positionalArgs, TNodePtr namedArgs, TNodePtr customUserType, TNodePtr typeConfig) {
  1882. if (!ctx.Settings.EnableGenericUdfs) {
  1883. return {};
  1884. }
  1885. TVector<TNodePtr> udfArgs;
  1886. udfArgs.push_back(new TAstListNodeImpl(pos));
  1887. udfArgs[0]->Add(new TAstAtomNodeImpl(pos, "Void", 0));
  1888. if (namedArgs) {
  1889. udfArgs.push_back(BuildUdfUserTypeArg(pos, positionalArgs, namedArgs, customUserType));
  1890. } else {
  1891. udfArgs.push_back(BuildUdfUserTypeArg(pos, args, customUserType));
  1892. }
  1893. if (typeConfig) {
  1894. udfArgs.push_back(typeConfig);
  1895. }
  1896. return udfArgs;
  1897. }
  1898. TNodePtr BuildSqlCall(TContext& ctx, TPosition pos, const TString& module, const TString& name, const TVector<TNodePtr>& args,
  1899. TNodePtr positionalArgs, TNodePtr namedArgs, TNodePtr customUserType, const TDeferredAtom& typeConfig, TNodePtr runConfig)
  1900. {
  1901. const TString fullName = module + "." + name;
  1902. TNodePtr callable;
  1903. if (to_lower(module) == "@yql") {
  1904. callable = BuildCallable(pos, module, name, {});
  1905. } else if (!ctx.Settings.EnableGenericUdfs) {
  1906. auto varName = ctx.AddSimpleUdf(fullName);
  1907. callable = new TAstAtomNodeImpl(pos, varName, TNodeFlags::ArbitraryContent);
  1908. }
  1909. if (callable) {
  1910. TVector<TNodePtr> applyArgs = { callable };
  1911. applyArgs.insert(applyArgs.end(), args.begin(), args.end());
  1912. return new TCallNodeImpl(pos, namedArgs ? "NamedApply" : "Apply", applyArgs);
  1913. }
  1914. TVector<TNodePtr> sqlCallArgs;
  1915. sqlCallArgs.push_back(BuildQuotedAtom(pos, fullName));
  1916. if (namedArgs) {
  1917. auto tupleNodePtr = positionalArgs->GetTupleNode();
  1918. YQL_ENSURE(tupleNodePtr);
  1919. TNodePtr positionalArgsNode = new TCallNodeImpl(pos, "PositionalArgs", tupleNodePtr->Elements());
  1920. sqlCallArgs.push_back(BuildTuple(pos, { positionalArgsNode, namedArgs }));
  1921. } else {
  1922. TNodePtr positionalArgsNode = new TCallNodeImpl(pos, "PositionalArgs", args);
  1923. sqlCallArgs.push_back(BuildTuple(pos, { positionalArgsNode }));
  1924. }
  1925. // optional arguments
  1926. if (customUserType) {
  1927. sqlCallArgs.push_back(customUserType);
  1928. } else if (!typeConfig.Empty()) {
  1929. sqlCallArgs.push_back(new TCallNodeImpl(pos, "TupleType", {}));
  1930. }
  1931. if (!typeConfig.Empty()) {
  1932. sqlCallArgs.push_back(typeConfig.Build());
  1933. } else if (runConfig) {
  1934. sqlCallArgs.push_back(BuildQuotedAtom(pos, ""));
  1935. }
  1936. if (runConfig) {
  1937. sqlCallArgs.push_back(runConfig);
  1938. }
  1939. return new TCallNodeImpl(pos, "SqlCall", sqlCallArgs);
  1940. }
  1941. class TCallableNode final: public INode {
  1942. public:
  1943. TCallableNode(TPosition pos, const TString& module, const TString& name, const TVector<TNodePtr>& args, bool forReduce)
  1944. : INode(pos)
  1945. , Module(module)
  1946. , Name(name)
  1947. , Args(args)
  1948. , ForReduce(forReduce)
  1949. {}
  1950. bool DoInit(TContext& ctx, ISource* src) override {
  1951. if (Module == "yql") {
  1952. Node = new TFuncNodeImpl(Pos, Name);
  1953. } else if (Module == "@yql") {
  1954. auto parsedName = StringContent(ctx, Pos, Name);
  1955. if (!parsedName) {
  1956. return false;
  1957. }
  1958. const TString yql("(" + parsedName->Content + ")");
  1959. TAstParseResult ast = ParseAst(yql, ctx.Pool.get());
  1960. /// TODO: do not drop warnings
  1961. if (ast.IsOk()) {
  1962. const auto rootCount = ast.Root->GetChildrenCount();
  1963. if (rootCount != 1) {
  1964. ctx.Error(Pos) << "Failed to parse YQL: expecting AST root node with single child, but got " << rootCount;
  1965. return false;
  1966. }
  1967. Node = AstNode(ast.Root->GetChild(0));
  1968. } else {
  1969. ctx.Error(Pos) << "Failed to parse YQL: " << ast.Issues.ToString();
  1970. return false;
  1971. }
  1972. if (src) {
  1973. src->AllColumns();
  1974. }
  1975. } else if (ctx.Settings.ModuleMapping.contains(Module)) {
  1976. Node = Y("bind", Module + "_module", Q(Name));
  1977. if (src) {
  1978. src->AllColumns();
  1979. }
  1980. } else {
  1981. TNodePtr customUserType = nullptr;
  1982. if (Module == "Tensorflow" && Name == "RunBatch") {
  1983. if (Args.size() > 2) {
  1984. auto passThroughAtom = Q("PassThrough");
  1985. auto passThroughType = Y("StructMemberType", Y("ListItemType", Y("TypeOf", Args[1])), passThroughAtom);
  1986. customUserType = Y("AddMemberType", Args[2], passThroughAtom, passThroughType);
  1987. Args.erase(Args.begin() + 2);
  1988. }
  1989. }
  1990. if ("Datetime" == Module || ("Yson" == Module && ctx.PragmaYsonFast))
  1991. Module.append('2');
  1992. TNodePtr typeConfig = MakeTypeConfig(Pos, to_lower(Module), Args);
  1993. if (ForReduce) {
  1994. TVector<TNodePtr> udfArgs;
  1995. udfArgs.push_back(BuildQuotedAtom(Pos, TString(Module) + "." + Name));
  1996. udfArgs.push_back(customUserType ? customUserType : new TCallNodeImpl(Pos, "TupleType", {}));
  1997. if (typeConfig) {
  1998. udfArgs.push_back(typeConfig);
  1999. }
  2000. Node = new TCallNodeImpl(Pos, "SqlReduceUdf", udfArgs);
  2001. } else {
  2002. auto udfArgs = BuildUdfArgs(ctx, Pos, Args, nullptr, nullptr, customUserType, typeConfig);
  2003. Node = BuildUdf(ctx, Pos, Module, Name, udfArgs);
  2004. }
  2005. }
  2006. return Node->Init(ctx, src);
  2007. }
  2008. TAstNode* Translate(TContext& ctx) const override {
  2009. Y_DEBUG_ABORT_UNLESS(Node);
  2010. return Node->Translate(ctx);
  2011. }
  2012. const TString* FuncName() const override {
  2013. return &Name;
  2014. }
  2015. const TString* ModuleName() const override {
  2016. return &Module;
  2017. }
  2018. void DoUpdateState() const override {
  2019. State.Set(ENodeState::Const, Node->IsConstant());
  2020. State.Set(ENodeState::Aggregated, Node->IsAggregated());
  2021. }
  2022. TNodePtr DoClone() const override {
  2023. return new TCallableNode(Pos, Module, Name, CloneContainer(Args), ForReduce);
  2024. }
  2025. void DoVisitChildren(const TVisitFunc& func, TVisitNodeSet& visited) const final {
  2026. Y_DEBUG_ABORT_UNLESS(Node);
  2027. Node->VisitTree(func, visited);
  2028. }
  2029. private:
  2030. TCiString Module;
  2031. TString Name;
  2032. TVector<TNodePtr> Args;
  2033. TNodePtr Node;
  2034. const bool ForReduce;
  2035. };
  2036. TNodePtr BuildCallable(TPosition pos, const TString& module, const TString& name, const TVector<TNodePtr>& args, bool forReduce) {
  2037. return new TCallableNode(pos, module, name, args, forReduce);
  2038. }
  2039. TNodePtr BuildUdf(TContext& ctx, TPosition pos, const TString& module, const TString& name, const TVector<TNodePtr>& args) {
  2040. if (to_lower(module) == "@yql") {
  2041. return BuildCallable(pos, module, name, args);
  2042. }
  2043. auto fullName = module + "." + name;
  2044. if (!args.empty()) {
  2045. return new TYqlUdf(pos, fullName, args, args.size() + 1);
  2046. } else {
  2047. auto varName = ctx.AddSimpleUdf(fullName);
  2048. return new TAstAtomNodeImpl(pos, varName, TNodeFlags::ArbitraryContent);
  2049. }
  2050. }
  2051. class TScriptUdf final: public INode {
  2052. public:
  2053. TScriptUdf(TPosition pos, const TString& moduleName, const TString& funcName, const TVector<TNodePtr>& args)
  2054. : INode(pos)
  2055. , ModuleName(moduleName)
  2056. , FuncName(funcName)
  2057. , Args(args)
  2058. {}
  2059. bool DoInit(TContext& ctx, ISource* src) override {
  2060. const bool isPython = ModuleName.find(TStringBuf("Python")) != TString::npos;
  2061. if (!isPython) {
  2062. if (Args.size() != 2) {
  2063. ctx.Error(Pos) << ModuleName << " script declaration requires exactly two parameters";
  2064. return false;
  2065. }
  2066. } else {
  2067. if (Args.size() < 1 || Args.size() > 2) {
  2068. ctx.Error(Pos) << ModuleName << " script declaration requires one or two parameters";
  2069. return false;
  2070. }
  2071. }
  2072. auto nameAtom = BuildQuotedAtom(Pos, FuncName);
  2073. auto scriptNode = Args.back();
  2074. if (!scriptNode->Init(ctx, src)) {
  2075. return false;
  2076. }
  2077. auto scriptStrPtr = Args.back()->GetLiteral("String");
  2078. if (!ctx.CompactNamedExprs && scriptStrPtr && scriptStrPtr->size() > SQL_MAX_INLINE_SCRIPT_LEN) {
  2079. scriptNode = ctx.UniversalAlias("scriptudf", std::move(scriptNode));
  2080. }
  2081. INode::TPtr type;
  2082. if (Args.size() == 2) {
  2083. type = Args[0];
  2084. } else {
  2085. // Python supports getting functions signatures right from docstrings
  2086. type = Y("EvaluateType", Y("ParseTypeHandle", Y("Apply",
  2087. Y("bind", "core_module", Q("PythonFuncSignature")),
  2088. Q(ModuleName),
  2089. scriptNode,
  2090. Y("String", nameAtom)
  2091. )));
  2092. }
  2093. if (!type->Init(ctx, src)) {
  2094. return false;
  2095. }
  2096. Node = Y("ScriptUdf", Q(ModuleName), nameAtom, type, scriptNode);
  2097. return true;
  2098. }
  2099. TAstNode* Translate(TContext& ctx) const override {
  2100. Y_UNUSED(ctx);
  2101. Y_DEBUG_ABORT_UNLESS(Node);
  2102. return Node->Translate(ctx);
  2103. }
  2104. void DoUpdateState() const override {
  2105. State.Set(ENodeState::Const, true);
  2106. }
  2107. TNodePtr DoClone() const final {
  2108. return new TScriptUdf(GetPos(), ModuleName, FuncName, CloneContainer(Args));
  2109. }
  2110. void DoVisitChildren(const TVisitFunc& func, TVisitNodeSet& visited) const final {
  2111. Y_DEBUG_ABORT_UNLESS(Node);
  2112. Node->VisitTree(func, visited);
  2113. }
  2114. private:
  2115. TString ModuleName;
  2116. TString FuncName;
  2117. TVector<TNodePtr> Args;
  2118. TNodePtr Node;
  2119. };
  2120. template <bool Sorted, bool Hashed>
  2121. class TYqlToDict final: public TCallNode {
  2122. public:
  2123. TYqlToDict(TPosition pos, const TString& mode, const TVector<TNodePtr>& args)
  2124. : TCallNode(pos, "ToDict", 4, 4, args)
  2125. , Mode(mode)
  2126. {}
  2127. private:
  2128. TCallNode::TPtr DoClone() const override {
  2129. return new TYqlToDict<Sorted, Hashed>(GetPos(), Mode, CloneContainer(Args));
  2130. }
  2131. bool DoInit(TContext& ctx, ISource* src) override {
  2132. if (Args.size() != 1) {
  2133. ctx.Error(Pos) << "ToDict required exactly one argument";
  2134. return false;
  2135. }
  2136. Args.push_back(BuildLambda(Pos, Y("val"), Y("Nth", "val", Q("0"))));
  2137. Args.push_back(BuildLambda(Pos, Y("val"), Y("Nth", "val", Q("1"))));
  2138. Args.push_back(Q(Y(Q(Sorted ? "Sorted" : Hashed ? "Hashed" : "Auto"), Q(Mode))));
  2139. return TCallNode::DoInit(ctx, src);
  2140. }
  2141. private:
  2142. TString Mode;
  2143. };
  2144. template <bool IsStart>
  2145. class THoppingTime final: public TAstListNode {
  2146. public:
  2147. THoppingTime(TPosition pos, const TVector<TNodePtr>& args = {})
  2148. : TAstListNode(pos)
  2149. {
  2150. Y_UNUSED(args);
  2151. }
  2152. private:
  2153. TNodePtr DoClone() const override {
  2154. return new THoppingTime(GetPos());
  2155. }
  2156. bool DoInit(TContext& ctx, ISource* src) override {
  2157. Y_UNUSED(ctx);
  2158. auto legacySpec = src->GetLegacyHoppingWindowSpec();
  2159. auto spec = src->GetHoppingWindowSpec();
  2160. if (!legacySpec && !spec) {
  2161. ctx.Error(Pos) << "No hopping window parameters in aggregation";
  2162. return false;
  2163. }
  2164. Nodes.clear();
  2165. const auto fieldName = legacySpec
  2166. ? "_yql_time"
  2167. : spec->GetLabel();
  2168. const auto interval = legacySpec
  2169. ? legacySpec->Interval
  2170. : dynamic_cast<THoppingWindow*>(spec.Get())->Interval;
  2171. if (!IsStart) {
  2172. Add("Member", "row", Q(fieldName));
  2173. return true;
  2174. }
  2175. Add("Sub",
  2176. Y("Member", "row", Q(fieldName)),
  2177. interval);
  2178. return true;
  2179. }
  2180. void DoUpdateState() const override {
  2181. State.Set(ENodeState::Aggregated, true);
  2182. }
  2183. };
  2184. class TInvalidBuiltin final: public INode {
  2185. public:
  2186. TInvalidBuiltin(TPosition pos, const TString& info)
  2187. : INode(pos)
  2188. , Info(info)
  2189. {
  2190. }
  2191. bool DoInit(TContext& ctx, ISource*) override {
  2192. ctx.Error(Pos) << Info;
  2193. return false;
  2194. }
  2195. TAstNode* Translate(TContext&) const override {
  2196. return nullptr;
  2197. }
  2198. TPtr DoClone() const override {
  2199. return new TInvalidBuiltin(GetPos(), Info);
  2200. }
  2201. private:
  2202. TString Info;
  2203. };
  2204. enum EAggrFuncTypeCallback {
  2205. NORMAL,
  2206. KEY_PAYLOAD,
  2207. PAYLOAD_PREDICATE,
  2208. TWO_ARGS,
  2209. COUNT,
  2210. HISTOGRAM,
  2211. LINEAR_HISTOGRAM,
  2212. PERCENTILE,
  2213. TOPFREQ,
  2214. TOP,
  2215. TOP_BY,
  2216. COUNT_DISTINCT_ESTIMATE,
  2217. LIST,
  2218. UDAF,
  2219. PG,
  2220. NTH_VALUE
  2221. };
  2222. struct TCoreFuncInfo {
  2223. TString Name;
  2224. ui32 MinArgs;
  2225. ui32 MaxArgs;
  2226. };
  2227. using TAggrFuncFactoryCallback = std::function<INode::TPtr(TPosition pos, const TVector<TNodePtr>& args, EAggregateMode aggMode, bool isFactory)>;
  2228. using TAggrFuncFactoryCallbackMap = std::unordered_map<TString, TAggrFuncFactoryCallback, THash<TString>>;
  2229. using TBuiltinFactoryCallback = std::function<TNodePtr(TPosition pos, const TVector<TNodePtr>& args)>;
  2230. using TBuiltinFactoryCallbackMap = std::unordered_map<TString, TBuiltinFactoryCallback, THash<TString>>;
  2231. using TCoreFuncMap = std::unordered_map<TString, TCoreFuncInfo, THash<TString>>;
  2232. TAggrFuncFactoryCallback BuildAggrFuncFactoryCallback(
  2233. const TString& functionName,
  2234. const TString& factoryName,
  2235. EAggrFuncTypeCallback type = NORMAL,
  2236. const TString& functionNameOverride = TString(),
  2237. const TVector<EAggregateMode>& validModes = {}) {
  2238. const TString realFunctionName = functionNameOverride.empty() ? functionName : functionNameOverride;
  2239. return [functionName, realFunctionName, factoryName, type, validModes] (TPosition pos, const TVector<TNodePtr>& args, EAggregateMode aggMode, bool isFactory) -> INode::TPtr {
  2240. if (!validModes.empty()) {
  2241. if (!IsIn(validModes, aggMode)) {
  2242. TString errorText;
  2243. if (TVector{EAggregateMode::OverWindow} == validModes) {
  2244. errorText = TStringBuilder()
  2245. << "Can't use window function " << functionName << " without window specification (OVER keyword is missing)";
  2246. } else {
  2247. errorText = TStringBuilder()
  2248. << "Can't use " << functionName << " in " << ToString(aggMode) << " aggregation mode";
  2249. }
  2250. return INode::TPtr(new TInvalidBuiltin(pos, errorText));
  2251. }
  2252. }
  2253. TAggregationPtr factory = nullptr;
  2254. switch (type) {
  2255. case NORMAL:
  2256. factory = BuildFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2257. break;
  2258. case KEY_PAYLOAD:
  2259. factory = BuildKeyPayloadFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2260. break;
  2261. case PAYLOAD_PREDICATE:
  2262. factory = BuildPayloadPredicateFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2263. break;
  2264. case TWO_ARGS:
  2265. factory = BuildTwoArgsFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2266. break;
  2267. case COUNT:
  2268. factory = BuildCountAggregation(pos, realFunctionName, factoryName, aggMode);
  2269. break;
  2270. case HISTOGRAM:
  2271. factory = BuildHistogramFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2272. break;
  2273. case LINEAR_HISTOGRAM:
  2274. factory = BuildLinearHistogramFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2275. break;
  2276. case PERCENTILE:
  2277. factory = BuildPercentileFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2278. break;
  2279. case TOPFREQ:
  2280. factory = BuildTopFreqFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2281. break;
  2282. case TOP:
  2283. factory = BuildTopFactoryAggregation<false>(pos, realFunctionName, factoryName, aggMode);
  2284. break;
  2285. case TOP_BY:
  2286. factory = BuildTopFactoryAggregation<true>(pos, realFunctionName, factoryName, aggMode);
  2287. break;
  2288. case COUNT_DISTINCT_ESTIMATE:
  2289. factory = BuildCountDistinctEstimateFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2290. break;
  2291. case LIST:
  2292. factory = BuildListFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2293. break;
  2294. case UDAF:
  2295. factory = BuildUserDefinedFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2296. break;
  2297. case PG:
  2298. factory = BuildPGFactoryAggregation(pos, realFunctionName, aggMode);
  2299. break;
  2300. case NTH_VALUE:
  2301. factory = BuildNthFactoryAggregation(pos, realFunctionName, factoryName, aggMode);
  2302. break;
  2303. }
  2304. if (isFactory) {
  2305. auto realArgs = args;
  2306. realArgs.erase(realArgs.begin()); // skip function name
  2307. return new TBasicAggrFactory(pos, functionName, factory, realArgs);
  2308. } else {
  2309. return new TBasicAggrFunc(pos, functionName, factory, args);
  2310. }
  2311. };
  2312. }
  2313. TAggrFuncFactoryCallback BuildAggrFuncFactoryCallback(
  2314. const TString& functionName,
  2315. const TString& factoryName,
  2316. const TVector<EAggregateMode>& validModes,
  2317. EAggrFuncTypeCallback type = NORMAL,
  2318. const TString& functionNameOverride = TString()) {
  2319. return BuildAggrFuncFactoryCallback(functionName, factoryName, type, functionNameOverride, validModes);
  2320. }
  2321. template<typename TType>
  2322. TBuiltinFactoryCallback BuildSimpleBuiltinFactoryCallback() {
  2323. return [] (TPosition pos, const TVector<TNodePtr>& args) -> TNodePtr {
  2324. return new TType(pos, args);
  2325. };
  2326. }
  2327. template<typename TType>
  2328. TBuiltinFactoryCallback BuildNamedBuiltinFactoryCallback(const TString& name) {
  2329. return [name] (TPosition pos, const TVector<TNodePtr>& args) -> TNodePtr {
  2330. return new TType(pos, name, args);
  2331. };
  2332. }
  2333. template<typename TType>
  2334. TBuiltinFactoryCallback BuildArgcBuiltinFactoryCallback(i32 minArgs, i32 maxArgs) {
  2335. return [minArgs, maxArgs] (TPosition pos, const TVector<TNodePtr>& args) -> TNodePtr {
  2336. return new TType(pos, minArgs, maxArgs, args);
  2337. };
  2338. }
  2339. template<typename TType>
  2340. TBuiltinFactoryCallback BuildNamedArgcBuiltinFactoryCallback(const TString& name, i32 minArgs, i32 maxArgs) {
  2341. return [name, minArgs, maxArgs] (TPosition pos, const TVector<TNodePtr>& args) -> TNodePtr {
  2342. return new TType(pos, name, minArgs, maxArgs, args);
  2343. };
  2344. }
  2345. template<typename TType>
  2346. TBuiltinFactoryCallback BuildNamedDepsArgcBuiltinFactoryCallback(ui32 reqArgsCount, const TString& name, i32 minArgs, i32 maxArgs) {
  2347. return [reqArgsCount, name, minArgs, maxArgs](TPosition pos, const TVector<TNodePtr>& args) -> TNodePtr {
  2348. return new TType(reqArgsCount, pos, name, minArgs, maxArgs, args);
  2349. };
  2350. }
  2351. template<typename TType>
  2352. TBuiltinFactoryCallback BuildBoolBuiltinFactoryCallback(bool arg) {
  2353. return [arg] (TPosition pos, const TVector<TNodePtr>& args) -> TNodePtr {
  2354. return new TType(pos, args, arg);
  2355. };
  2356. }
  2357. template<typename TType>
  2358. TBuiltinFactoryCallback BuildFoldBuiltinFactoryCallback(const TString& name, const TString& defaultValue) {
  2359. return [name, defaultValue] (TPosition pos, const TVector<TNodePtr>& args) -> TNodePtr {
  2360. return new TType(pos, name, "Bool", defaultValue, 1, args);
  2361. };
  2362. }
  2363. TNodePtr MakePair(TPosition pos, const TVector<TNodePtr>& args) {
  2364. TNodePtr list = new TAstListNodeImpl(pos, {
  2365. args[0],
  2366. args.size() > 1 ? args[1] : new TAstListNodeImpl(pos,{ new TAstAtomNodeImpl(pos, "Null", TNodeFlags::Default) })
  2367. });
  2368. return new TAstListNodeImpl(pos, {
  2369. new TAstAtomNodeImpl(pos, "quote", TNodeFlags::Default),
  2370. list
  2371. });
  2372. }
  2373. struct TBuiltinFuncData {
  2374. const TBuiltinFactoryCallbackMap BuiltinFuncs;
  2375. const TAggrFuncFactoryCallbackMap AggrFuncs;
  2376. const TCoreFuncMap CoreFuncs;
  2377. TBuiltinFuncData():
  2378. BuiltinFuncs(MakeBuiltinFuncs()),
  2379. AggrFuncs(MakeAggrFuncs()),
  2380. CoreFuncs(MakeCoreFuncs())
  2381. {
  2382. }
  2383. TBuiltinFactoryCallbackMap MakeBuiltinFuncs() {
  2384. TBuiltinFactoryCallbackMap builtinFuncs = {
  2385. // Branching
  2386. {"if", BuildSimpleBuiltinFactoryCallback<TYqlIf<false>>()},
  2387. {"ifstrict", BuildSimpleBuiltinFactoryCallback<TYqlIf<true>>() },
  2388. // String builtins
  2389. {"len", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Size", 1, 1)},
  2390. {"length", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Size", 1, 1)},
  2391. {"charlength", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Size", 1, 1)},
  2392. {"characterlength", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Size", 1, 1)},
  2393. {"substring", BuildNamedBuiltinFactoryCallback<TYqlSubstring>("Substring")},
  2394. {"find", BuildNamedBuiltinFactoryCallback<TYqlSubstring>("Find")},
  2395. {"rfind", BuildNamedBuiltinFactoryCallback<TYqlSubstring>("RFind")},
  2396. {"byteat", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ByteAt", 2, 2) },
  2397. {"startswith", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StartsWith", 2, 2)},
  2398. {"endswith", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EndsWith", 2, 2)},
  2399. // Numeric builtins
  2400. {"abs", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Abs", 1, 1) },
  2401. {"tobytes", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ToBytes", 1, 1) },
  2402. {"frombytes", BuildSimpleBuiltinFactoryCallback<TFromBytes>() },
  2403. // Compare builtins
  2404. {"minof", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Min", 1, -1)},
  2405. {"maxof", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Max", 1, -1)},
  2406. {"greatest", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Max", 1, -1)},
  2407. {"least", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Min", 1, -1)},
  2408. {"in", BuildSimpleBuiltinFactoryCallback<TYqlIn>()},
  2409. // List builtins
  2410. {"aslist", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AsListMayWarn", 0, -1)},
  2411. {"asliststrict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AsListStrict", 0, -1) },
  2412. {"listlength", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Length", 1, 1)},
  2413. {"listhasitems", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("HasItems", 1, 1)},
  2414. {"listextend", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListExtend", 0, -1)},
  2415. {"listextendstrict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListExtendStrict", 0, -1)},
  2416. {"listunionall", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListUnionAll", 0, -1) },
  2417. {"listzip", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListZip", -1, -1)},
  2418. {"listzipall", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListZipAll", -1, -1)},
  2419. {"listenumerate", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListEnumerate", 1, 3)},
  2420. {"listreverse", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListReverse", 1, 1)},
  2421. {"listskip", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListSkip", 2, 2)},
  2422. {"listtake", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTake", 2, 2)},
  2423. {"listhead", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListHead", 1, 1)},
  2424. {"listlast", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListLast", 1, 1)},
  2425. {"listsort", BuildBoolBuiltinFactoryCallback<TListSortBuiltin>(true)},
  2426. {"listsortasc", BuildBoolBuiltinFactoryCallback<TListSortBuiltin>(true)},
  2427. {"listsortdesc", BuildBoolBuiltinFactoryCallback<TListSortBuiltin>(false)},
  2428. {"listmap", BuildBoolBuiltinFactoryCallback<TListMapBuiltin>(false)},
  2429. {"listflatmap", BuildBoolBuiltinFactoryCallback<TListMapBuiltin>(true)},
  2430. {"listfilter", BuildNamedBuiltinFactoryCallback<TListFilterBuiltin>("ListFilter")},
  2431. {"listany", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListAny", 1, 1)},
  2432. {"listall", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListAll", 1, 1)},
  2433. {"listhas", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListHas", 2, 2)},
  2434. {"listmax", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListMax", 1, 1)},
  2435. {"listmin", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListMin", 1, 1)},
  2436. {"listsum", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListSum", 1, 1)},
  2437. {"listfold", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListFold", 3, 3)},
  2438. {"listfold1", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListFold1", 3, 3)},
  2439. {"listfoldmap", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListFoldMap", 3, 3)},
  2440. {"listfold1map", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListFold1Map", 3, 3)},
  2441. {"listavg", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListAvg", 1, 1)},
  2442. {"listconcat", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListConcat", 1, 2)},
  2443. {"listextract", BuildSimpleBuiltinFactoryCallback<TListExtractBuiltin>()},
  2444. {"listuniq", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListUniq", 1, 1)},
  2445. {"listuniqstable", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListUniqStable", 1, 1)},
  2446. {"listcreate", BuildSimpleBuiltinFactoryCallback<TListCreateBuiltin>()},
  2447. {"listfromrange", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListFromRange", 2, 3) },
  2448. {"listreplicate", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Replicate", 2, 2) },
  2449. {"listtakewhile", BuildNamedBuiltinFactoryCallback<TListFilterBuiltin>("ListTakeWhile") },
  2450. {"listskipwhile", BuildNamedBuiltinFactoryCallback<TListFilterBuiltin>("ListSkipWhile") },
  2451. {"listtakewhileinclusive", BuildNamedBuiltinFactoryCallback<TListFilterBuiltin>("ListTakeWhileInclusive") },
  2452. {"listskipwhileinclusive", BuildNamedBuiltinFactoryCallback<TListFilterBuiltin>("ListSkipWhileInclusive") },
  2453. {"listcollect", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListCollect", 1, 1) },
  2454. {"listnotnull", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListNotNull", 1, 1)},
  2455. {"listflatten", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListFlatten", 1, 1)},
  2456. {"listtop", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTop", 2, 3)},
  2457. {"listtopasc", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTopAsc", 2, 3)},
  2458. {"listtopdesc", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTopDesc", 2, 3)},
  2459. {"listtopsort", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTopSort", 2, 3)},
  2460. {"listtopsortasc", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTopSortAsc", 2, 3)},
  2461. {"listtopsortdesc", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTopSortDesc", 2, 3)},
  2462. {"listsample", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListSample", 2, 3)},
  2463. {"listsamplen", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListSampleN", 2, 3)},
  2464. {"listshuffle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListShuffle", 1, 2)},
  2465. // Dict builtins
  2466. {"dictlength", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Length", 1, 1)},
  2467. {"dicthasitems", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("HasItems", 1, 1)},
  2468. {"dictcreate", BuildSimpleBuiltinFactoryCallback<TDictCreateBuiltin>()},
  2469. {"setcreate", BuildSimpleBuiltinFactoryCallback<TSetCreateBuiltin>()},
  2470. {"asdict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AsDictMayWarn", 0, -1)},
  2471. {"asdictstrict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AsDictStrict", 0, -1)},
  2472. {"asset", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AsSetMayWarn", 0, -1)},
  2473. {"assetstrict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AsSetStrict", 0, -1)},
  2474. {"todict", BuildNamedBuiltinFactoryCallback<TYqlToDict<false, false>>("One")},
  2475. {"tomultidict", BuildNamedBuiltinFactoryCallback<TYqlToDict<false, false>>("Many")},
  2476. {"tosorteddict", BuildNamedBuiltinFactoryCallback<TYqlToDict<true, false>>("One")},
  2477. {"tosortedmultidict", BuildNamedBuiltinFactoryCallback<TYqlToDict<true, false>>("Many")},
  2478. {"tohasheddict", BuildNamedBuiltinFactoryCallback<TYqlToDict<false, true>>("One")},
  2479. {"tohashedmultidict", BuildNamedBuiltinFactoryCallback<TYqlToDict<false, true>>("Many")},
  2480. {"dictkeys", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictKeys", 1, 1) },
  2481. {"dictpayloads", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictPayloads", 1, 1) },
  2482. {"dictitems", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictItems", 1, 1) },
  2483. {"dictlookup", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Lookup", 2, 2) },
  2484. {"dictcontains", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Contains", 2, 2) },
  2485. // Atom builtins
  2486. {"asatom", BuildSimpleBuiltinFactoryCallback<TYqlAsAtom>()},
  2487. {"secureparam", BuildNamedBuiltinFactoryCallback<TYqlAtom>("SecureParam")},
  2488. {"void", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Void", 0, 0)},
  2489. {"emptylist", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EmptyList", 0, 0)},
  2490. {"emptydict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EmptyDict", 0, 0)},
  2491. {"callable", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Callable", 2, 2)},
  2492. {"way", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Way", 1, 1) },
  2493. {"dynamicvariant", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DynamicVariant", 3, 3) },
  2494. {"variant", BuildSimpleBuiltinFactoryCallback<TYqlVariant>() },
  2495. {"enum", BuildSimpleBuiltinFactoryCallback<TYqlEnum>() },
  2496. {"asvariant", BuildSimpleBuiltinFactoryCallback<TYqlAsVariant>() },
  2497. {"asenum", BuildSimpleBuiltinFactoryCallback<TYqlAsEnum>() },
  2498. {"astagged", BuildSimpleBuiltinFactoryCallback<TYqlAsTagged>() },
  2499. {"untag", BuildSimpleBuiltinFactoryCallback<TYqlUntag>() },
  2500. {"parsetype", BuildSimpleBuiltinFactoryCallback<TYqlParseType>() },
  2501. {"ensuretype", BuildSimpleBuiltinFactoryCallback<TYqlTypeAssert<true>>() },
  2502. {"ensureconvertibleto", BuildSimpleBuiltinFactoryCallback<TYqlTypeAssert<false>>() },
  2503. {"ensure", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Ensure", 2, 3) },
  2504. {"evaluateexpr", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EvaluateExpr", 1, 1) },
  2505. {"evaluateatom", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EvaluateAtom", 1, 1) },
  2506. {"evaluatetype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EvaluateType", 1, 1) },
  2507. {"unwrap", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Unwrap", 1, 2) },
  2508. {"just", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Just", 1, 1) },
  2509. {"nothing", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Nothing", 1, 1) },
  2510. {"formattype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("FormatType", 1, 1) },
  2511. {"formattypediff", BuildNamedBuiltinFactoryCallback<TFormatTypeDiff<false>>("FormatTypeDiff") },
  2512. {"formattypediffpretty", BuildNamedBuiltinFactoryCallback<TFormatTypeDiff<true>>("FormatTypeDiffPretty") },
  2513. {"pgtype", BuildSimpleBuiltinFactoryCallback<TYqlPgType>() },
  2514. {"pgconst", BuildSimpleBuiltinFactoryCallback<TYqlPgConst>() },
  2515. {"pgop", BuildSimpleBuiltinFactoryCallback<TYqlPgOp>() },
  2516. {"pgcall", BuildSimpleBuiltinFactoryCallback<TYqlPgCall<false>>() },
  2517. {"pgrangecall", BuildSimpleBuiltinFactoryCallback<TYqlPgCall<true>>() },
  2518. {"pgcast", BuildSimpleBuiltinFactoryCallback<TYqlPgCast>() },
  2519. {"frompg", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("FromPg", 1, 1) },
  2520. {"topg", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ToPg", 1, 1) },
  2521. {"pgor", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("PgOr", 2, 2) },
  2522. {"pgand", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("PgAnd", 2, 2) },
  2523. {"pgnot", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("PgNot", 1, 1) },
  2524. {"pgarray", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("PgArray", 1, -1) },
  2525. {"typeof", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TypeOf", 1, 1) },
  2526. {"instanceof", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("InstanceOf", 1, 1) },
  2527. {"datatype", BuildSimpleBuiltinFactoryCallback<TYqlDataType>() },
  2528. {"optionaltype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("OptionalType", 1, 1) },
  2529. {"listtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListType", 1, 1) },
  2530. {"streamtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StreamType", 1, 1) },
  2531. {"dicttype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictType", 2, 2) },
  2532. {"tupletype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TupleType", 0, -1) },
  2533. {"generictype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("GenericType", 0, 0) },
  2534. {"unittype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("UnitType", 0, 0) },
  2535. {"voidtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("VoidType", 0, 0) },
  2536. {"resourcetype", BuildSimpleBuiltinFactoryCallback<TYqlResourceType>() },
  2537. {"taggedtype", BuildSimpleBuiltinFactoryCallback<TYqlTaggedType>() },
  2538. {"varianttype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("VariantType", 1, 1) },
  2539. {"callabletype", BuildSimpleBuiltinFactoryCallback<TYqlCallableType>() },
  2540. {"optionalitemtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("OptionalItemType", 1, 1) },
  2541. {"listitemtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListItemType", 1, 1) },
  2542. {"streamitemtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StreamItemType", 1, 1) },
  2543. {"dictkeytype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictKeyType", 1, 1) },
  2544. {"dictpayloadtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictPayloadType", 1, 1) },
  2545. {"tupleelementtype", BuildSimpleBuiltinFactoryCallback<TYqlTupleElementType>() },
  2546. {"structmembertype", BuildSimpleBuiltinFactoryCallback<TYqlStructMemberType>() },
  2547. {"callableresulttype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("CallableResultType", 1, 1) },
  2548. {"callableargumenttype", BuildSimpleBuiltinFactoryCallback<TYqlCallableArgumentType>() },
  2549. {"variantunderlyingtype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("VariantUnderlyingType", 1, 1) },
  2550. {"variantitem", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("SqlVariantItem", 1, 1) },
  2551. {"fromysonsimpletype", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("FromYsonSimpleType", 2, 2) },
  2552. {"currentutcdate", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(0, "CurrentUtcDate", 0, -1) },
  2553. {"currentutcdatetime", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(0, "CurrentUtcDatetime", 0, -1) },
  2554. {"currentutctimestamp", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(0, "CurrentUtcTimestamp", 0, -1) },
  2555. { "currenttzdate", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(1, "CurrentTzDate", 1, -1) },
  2556. { "currenttzdatetime", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(1, "CurrentTzDatetime", 1, -1) },
  2557. { "currenttztimestamp", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(1, "CurrentTzTimestamp", 1, -1) },
  2558. {"currentoperationid", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("CurrentOperationId", 0, 0) },
  2559. {"currentoperationsharedid", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("CurrentOperationSharedId", 0, 0) },
  2560. {"currentauthenticateduser", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("CurrentAuthenticatedUser", 0, 0) },
  2561. {"addtimezone", BuildSimpleBuiltinFactoryCallback<TYqlAddTimezone>() },
  2562. {"removetimezone", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("RemoveTimezone", 1, 1) },
  2563. {"pickle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Pickle", 1, 1) },
  2564. {"stablepickle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StablePickle", 1, 1) },
  2565. {"unpickle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Unpickle", 2, 2) },
  2566. {"typehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TypeHandle", 1, 1) },
  2567. {"parsetypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ParseTypeHandle", 1, 1) },
  2568. {"typekind", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TypeKind", 1, 1) },
  2569. {"datatypecomponents", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DataTypeComponents", 1, 1) },
  2570. {"datatypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DataTypeHandle", 1, 1) },
  2571. {"optionaltypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("OptionalTypeHandle", 1, 1) },
  2572. {"listtypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListTypeHandle", 1, 1) },
  2573. {"streamtypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StreamTypeHandle", 1, 1) },
  2574. {"tupletypecomponents", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TupleTypeComponents", 1, 1) },
  2575. {"tupletypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TupleTypeHandle", 1, 1) },
  2576. {"structtypecomponents", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StructTypeComponents", 1, 1) },
  2577. {"structtypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StructTypeHandle", 1, 1) },
  2578. {"dicttypecomponents", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictTypeComponents", 1, 1) },
  2579. {"dicttypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("DictTypeHandle", 2, 2) },
  2580. {"resourcetypetag", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ResourceTypeTag", 1, 1) },
  2581. {"resourcetypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ResourceTypeHandle", 1, 1) },
  2582. {"taggedtypecomponents", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TaggedTypeComponents", 1, 1) },
  2583. {"taggedtypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("TaggedTypeHandle", 2, 2) },
  2584. {"varianttypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("VariantTypeHandle", 1, 1) },
  2585. {"voidtypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("VoidTypeHandle", 0, 0) },
  2586. {"nulltypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("NullTypeHandle", 0, 0) },
  2587. {"emptylisttypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EmptyListTypeHandle", 0, 0) },
  2588. {"emptydicttypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EmptyDictTypeHandle", 0, 0) },
  2589. {"callabletypecomponents", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("CallableTypeComponents", 1, 1) },
  2590. {"callableargument", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("CallableArgument", 1, 3) },
  2591. {"callabletypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("CallableTypeHandle", 2, 4) },
  2592. {"pgtypename", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("PgTypeName", 1, 1) },
  2593. {"pgtypehandle", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("PgTypeHandle", 1, 1) },
  2594. {"formatcode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("FormatCode", 1, 1) },
  2595. {"worldcode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("WorldCode", 0, 0) },
  2596. {"atomcode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AtomCode", 1, 1) },
  2597. {"listcode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ListCode", 0, -1) },
  2598. {"funccode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("FuncCode", 1, -1) },
  2599. {"lambdacode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("LambdaCode", 1, 2) },
  2600. {"evaluatecode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("EvaluateCode", 1, 1) },
  2601. {"reprcode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("ReprCode", 1, 1) },
  2602. {"quotecode", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("QuoteCode", 1, 1) },
  2603. {"lambdaargumentscount", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("LambdaArgumentsCount", 1, 1) },
  2604. {"lambdaoptionalargumentscount", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("LambdaOptionalArgumentsCount", 1, 1) },
  2605. {"subqueryextend", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("SubqueryExtend", 1, -1) },
  2606. {"subqueryunionall", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("SubqueryUnionAll", 1, -1) },
  2607. {"subquerymerge", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("SubqueryMerge", 1, -1) },
  2608. {"subqueryunionmerge", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("SubqueryUnionMerge", 1, -1) },
  2609. {"subqueryextendfor", BuildSimpleBuiltinFactoryCallback<TYqlSubqueryFor<SubqueryExtendFor>>() },
  2610. {"subqueryunionallfor", BuildSimpleBuiltinFactoryCallback<TYqlSubqueryFor<SubqueryUnionAllFor>>() },
  2611. {"subquerymergefor", BuildSimpleBuiltinFactoryCallback<TYqlSubqueryFor<SubqueryMergeFor>>() },
  2612. {"subqueryunionmergefor", BuildSimpleBuiltinFactoryCallback<TYqlSubqueryFor<SubqueryUnionMergeFor>>() },
  2613. {"subqueryorderby", BuildSimpleBuiltinFactoryCallback<TYqlSubqueryOrderBy<SubqueryOrderBy>>() },
  2614. {"subqueryassumeorderby", BuildSimpleBuiltinFactoryCallback<TYqlSubqueryOrderBy<SubqueryAssumeOrderBy>>() },
  2615. // Tuple builtins
  2616. {"astuple", BuildSimpleBuiltinFactoryCallback<TTupleNode>()},
  2617. // Struct builtins
  2618. {"trymember", BuildNamedBuiltinFactoryCallback<TTryMember>("TryMember")},
  2619. {"addmember", BuildNamedBuiltinFactoryCallback<TAddMember>("AddMember")},
  2620. {"replacemember", BuildNamedBuiltinFactoryCallback<TAddMember>("ReplaceMember")},
  2621. {"removemember", BuildNamedBuiltinFactoryCallback<TRemoveMember>("RemoveMember")},
  2622. {"forceremovemember", BuildNamedBuiltinFactoryCallback<TRemoveMember>("ForceRemoveMember")},
  2623. {"combinemembers", BuildNamedBuiltinFactoryCallback<TCombineMembers>("FlattenMembers")},
  2624. {"flattenmembers", BuildNamedBuiltinFactoryCallback<TFlattenMembers>("FlattenMembers")},
  2625. {"staticmap", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StaticMap", 2, 2) },
  2626. {"staticzip", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StaticZip", 1, -1) },
  2627. {"structunion", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StructUnion", 2, 3)},
  2628. {"structintersection", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StructIntersection", 2, 3)},
  2629. {"structdifference", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StructDifference", 2, 2)},
  2630. {"structsymmetricdifference", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StructSymmetricDifference", 2, 2)},
  2631. {"staticfold", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StaticFold", 3, 3)},
  2632. {"staticfold1", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("StaticFold1", 3, 3)},
  2633. // File builtins
  2634. {"filepath", BuildNamedBuiltinFactoryCallback<TFileYqlAtom>("FilePath")},
  2635. {"filecontent", BuildNamedBuiltinFactoryCallback<TFileYqlAtom>("FileContent")},
  2636. {"folderpath", BuildNamedBuiltinFactoryCallback<TFileYqlAtom>("FolderPath") },
  2637. {"files", BuildNamedBuiltinFactoryCallback<TFileYqlAtom>("Files")},
  2638. {"parsefile", BuildSimpleBuiltinFactoryCallback<TYqlParseFileOp>()},
  2639. // Misc builtins
  2640. {"coalesce", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Coalesce", 1, -1)},
  2641. {"nvl", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Coalesce", 1, -1) },
  2642. {"nanvl", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Nanvl", 2, 2) },
  2643. {"likely", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Likely", 1, -1)},
  2644. {"assumestrict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AssumeStrict", 1, 1)},
  2645. {"assumenonstrict", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("AssumeNonStrict", 1, 1)},
  2646. {"random", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(0, "Random", 1, -1)},
  2647. {"randomnumber", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(0, "RandomNumber", 1, -1)},
  2648. {"randomuuid", BuildNamedDepsArgcBuiltinFactoryCallback<TCallNodeDepArgs>(0, "RandomUuid", 1, -1) },
  2649. {"tablepath", BuildNamedBuiltinFactoryCallback<TCallDirectRow>("TablePath") },
  2650. {"tablerecordindex", BuildNamedBuiltinFactoryCallback<TCallDirectRow>("TableRecord") },
  2651. {"tablerow", BuildSimpleBuiltinFactoryCallback<TTableRow<false>>() },
  2652. {"jointablerow", BuildSimpleBuiltinFactoryCallback<TTableRow<true>>() },
  2653. {"tablerows", BuildSimpleBuiltinFactoryCallback<TTableRows>() },
  2654. {"weakfield", BuildSimpleBuiltinFactoryCallback<TWeakFieldOp>()},
  2655. {"version", BuildNamedArgcBuiltinFactoryCallback<TCallNodeImpl>("Version", 0, 0)},
  2656. {"systemmetadata", BuildNamedArgcBuiltinFactoryCallback<TCallDirectRow>("SystemMetadata", 1, -1)},
  2657. // Hint builtins
  2658. {"grouping", BuildSimpleBuiltinFactoryCallback<TGroupingNode>()},
  2659. // Window funcitons
  2660. {"rownumber", BuildNamedArgcBuiltinFactoryCallback<TWinRowNumber>("RowNumber", 0, 0)},
  2661. {"rank", BuildNamedArgcBuiltinFactoryCallback<TWinRank>("Rank", 0, 1)},
  2662. {"denserank", BuildNamedArgcBuiltinFactoryCallback<TWinRank>("DenseRank", 0, 1)},
  2663. {"lead", BuildNamedArgcBuiltinFactoryCallback<TWinLeadLag>("Lead", 1, 2)},
  2664. {"lag", BuildNamedArgcBuiltinFactoryCallback<TWinLeadLag>("Lag", 1, 2)},
  2665. {"percentrank", BuildNamedArgcBuiltinFactoryCallback<TWinRank>("PercentRank", 0, 1)},
  2666. {"cumedist", BuildNamedArgcBuiltinFactoryCallback<TWinCumeDist>("CumeDist", 0, 0)},
  2667. {"ntile", BuildNamedArgcBuiltinFactoryCallback<TWinNTile>("NTile", 1, 1)},
  2668. // Session window
  2669. {"sessionwindow", BuildSimpleBuiltinFactoryCallback<TSessionWindow>()},
  2670. {"sessionstart", BuildSimpleBuiltinFactoryCallback<TSessionStart<true>>()},
  2671. {"sessionstate", BuildSimpleBuiltinFactoryCallback<TSessionStart<false>>()},
  2672. // New hopping
  2673. {"hoppingwindow", BuildSimpleBuiltinFactoryCallback<THoppingWindow>()},
  2674. // Hopping intervals time functions
  2675. {"hopstart", BuildSimpleBuiltinFactoryCallback<THoppingTime<true>>()},
  2676. {"hopend", BuildSimpleBuiltinFactoryCallback<THoppingTime<false>>()},
  2677. };
  2678. return builtinFuncs;
  2679. }
  2680. TAggrFuncFactoryCallbackMap MakeAggrFuncs() {
  2681. constexpr auto OverWindow = EAggregateMode::OverWindow;
  2682. TAggrFuncFactoryCallbackMap aggrFuncs = {
  2683. {"min", BuildAggrFuncFactoryCallback("Min", "min_traits_factory")},
  2684. {"max", BuildAggrFuncFactoryCallback("Max", "max_traits_factory")},
  2685. {"minby", BuildAggrFuncFactoryCallback("MinBy", "min_by_traits_factory", KEY_PAYLOAD)},
  2686. {"maxby", BuildAggrFuncFactoryCallback("MaxBy", "max_by_traits_factory", KEY_PAYLOAD)},
  2687. {"sum", BuildAggrFuncFactoryCallback("Sum", "sum_traits_factory")},
  2688. {"sumif", BuildAggrFuncFactoryCallback("SumIf", "sum_if_traits_factory", PAYLOAD_PREDICATE) },
  2689. {"checked_sum", BuildAggrFuncFactoryCallback("CheckedSum", "checked_sum_traits_factory")},
  2690. {"checked_sumif", BuildAggrFuncFactoryCallback("CheckedSumIf", "checked_sum_if_traits_factory", PAYLOAD_PREDICATE) },
  2691. {"some", BuildAggrFuncFactoryCallback("Some", "some_traits_factory")},
  2692. {"somevalue", BuildAggrFuncFactoryCallback("SomeValue", "some_traits_factory")},
  2693. {"count", BuildAggrFuncFactoryCallback("Count", "count_traits_factory", COUNT)},
  2694. {"countif", BuildAggrFuncFactoryCallback("CountIf", "count_if_traits_factory")},
  2695. {"every", BuildAggrFuncFactoryCallback("Every", "and_traits_factory")},
  2696. {"booland", BuildAggrFuncFactoryCallback("BoolAnd", "and_traits_factory")},
  2697. {"boolor", BuildAggrFuncFactoryCallback("BoolOr", "or_traits_factory")},
  2698. {"boolxor", BuildAggrFuncFactoryCallback("BoolXor", "xor_traits_factory")},
  2699. {"bitand", BuildAggrFuncFactoryCallback("BitAnd", "bit_and_traits_factory")},
  2700. {"bitor", BuildAggrFuncFactoryCallback("BitOr", "bit_or_traits_factory")},
  2701. {"bitxor", BuildAggrFuncFactoryCallback("BitXor", "bit_xor_traits_factory")},
  2702. {"avg", BuildAggrFuncFactoryCallback("Avg", "avg_traits_factory")},
  2703. {"avgif", BuildAggrFuncFactoryCallback("AvgIf", "avg_if_traits_factory", PAYLOAD_PREDICATE) },
  2704. {"agglist", BuildAggrFuncFactoryCallback("AggregateList", "list2_traits_factory", LIST)},
  2705. {"aggrlist", BuildAggrFuncFactoryCallback("AggregateList", "list2_traits_factory", LIST)},
  2706. {"aggregatelist", BuildAggrFuncFactoryCallback("AggregateList", "list2_traits_factory", LIST)},
  2707. {"agglistdistinct", BuildAggrFuncFactoryCallback("AggregateListDistinct", "set_traits_factory", LIST)},
  2708. {"aggrlistdistinct", BuildAggrFuncFactoryCallback("AggregateListDistinct", "set_traits_factory", LIST)},
  2709. {"aggregatelistdistinct", BuildAggrFuncFactoryCallback("AggregateListDistinct", "set_traits_factory", LIST)},
  2710. {"median", BuildAggrFuncFactoryCallback("Median", "percentile_traits_factory", PERCENTILE)},
  2711. {"percentile", BuildAggrFuncFactoryCallback("Percentile", "percentile_traits_factory", PERCENTILE)},
  2712. {"mode", BuildAggrFuncFactoryCallback("Mode", "topfreq_traits_factory", TOPFREQ) },
  2713. {"topfreq", BuildAggrFuncFactoryCallback("TopFreq", "topfreq_traits_factory", TOPFREQ) },
  2714. {"top", BuildAggrFuncFactoryCallback("Top", "top_traits_factory", TOP)},
  2715. {"bottom", BuildAggrFuncFactoryCallback("Bottom", "bottom_traits_factory", TOP)},
  2716. {"topby", BuildAggrFuncFactoryCallback("TopBy", "top_by_traits_factory", TOP_BY)},
  2717. {"bottomby", BuildAggrFuncFactoryCallback("BottomBy", "bottom_by_traits_factory", TOP_BY)},
  2718. {"histogram", BuildAggrFuncFactoryCallback("AdaptiveWardHistogram", "histogram_adaptive_ward_traits_factory", HISTOGRAM, "Histogram")},
  2719. {"histogramcdf", BuildAggrFuncFactoryCallback("AdaptiveWardHistogramCDF", "histogram_cdf_adaptive_ward_traits_factory", HISTOGRAM, "HistogramCDF")},
  2720. {"adaptivewardhistogram", BuildAggrFuncFactoryCallback("AdaptiveWardHistogram", "histogram_adaptive_ward_traits_factory", HISTOGRAM)},
  2721. {"adaptivewardhistogramcdf", BuildAggrFuncFactoryCallback("AdaptiveWardHistogramCDF", "histogram_cdf_adaptive_ward_traits_factory", HISTOGRAM)},
  2722. {"adaptiveweighthistogram", BuildAggrFuncFactoryCallback("AdaptiveWeightHistogram", "histogram_adaptive_weight_traits_factory", HISTOGRAM)},
  2723. {"adaptiveweighthistogramcdf", BuildAggrFuncFactoryCallback("AdaptiveWeightHistogramCDF", "histogram_cdf_adaptive_weight_traits_factory", HISTOGRAM)},
  2724. {"adaptivedistancehistogram", BuildAggrFuncFactoryCallback("AdaptiveDistanceHistogram", "histogram_adaptive_distance_traits_factory", HISTOGRAM)},
  2725. {"adaptivedistancehistogramcdf", BuildAggrFuncFactoryCallback("AdaptiveDistanceHistogramCDF", "histogram_cdf_adaptive_distance_traits_factory", HISTOGRAM)},
  2726. {"blockwardhistogram", BuildAggrFuncFactoryCallback("BlockWardHistogram", "histogram_block_ward_traits_factory", HISTOGRAM)},
  2727. {"blockwardhistogramcdf", BuildAggrFuncFactoryCallback("BlockWardHistogramCDF", "histogram_cdf_block_ward_traits_factory", HISTOGRAM)},
  2728. {"blockweighthistogram", BuildAggrFuncFactoryCallback("BlockWeightHistogram", "histogram_block_weight_traits_factory", HISTOGRAM)},
  2729. {"blockweighthistogramcdf", BuildAggrFuncFactoryCallback("BlockWeightHistogramCDF", "histogram_cdf_block_weight_traits_factory", HISTOGRAM)},
  2730. {"linearhistogram", BuildAggrFuncFactoryCallback("LinearHistogram", "histogram_linear_traits_factory", LINEAR_HISTOGRAM)},
  2731. {"linearhistogramcdf", BuildAggrFuncFactoryCallback("LinearHistogramCDF", "histogram_cdf_linear_traits_factory", LINEAR_HISTOGRAM)},
  2732. {"logarithmichistogram", BuildAggrFuncFactoryCallback("LogarithmicHistogram", "histogram_logarithmic_traits_factory", LINEAR_HISTOGRAM)},
  2733. {"logarithmichistogramcdf", BuildAggrFuncFactoryCallback("LogarithmicHistogramCDF", "histogram_cdf_logarithmic_traits_factory", LINEAR_HISTOGRAM)},
  2734. {"loghistogram", BuildAggrFuncFactoryCallback("LogarithmicHistogram", "histogram_logarithmic_traits_factory", LINEAR_HISTOGRAM, "LogHistogram")},
  2735. {"loghistogramcdf", BuildAggrFuncFactoryCallback("LogarithmicHistogramCDF", "histogram_cdf_logarithmic_traits_factory", LINEAR_HISTOGRAM, "LogHistogramCDF")},
  2736. {"hyperloglog", BuildAggrFuncFactoryCallback("HyperLogLog", "hyperloglog_traits_factory", COUNT_DISTINCT_ESTIMATE)},
  2737. {"hll", BuildAggrFuncFactoryCallback("HyperLogLog", "hyperloglog_traits_factory", COUNT_DISTINCT_ESTIMATE, "HLL")},
  2738. {"countdistinctestimate", BuildAggrFuncFactoryCallback("HyperLogLog", "hyperloglog_traits_factory", COUNT_DISTINCT_ESTIMATE, "CountDistinctEstimate")},
  2739. {"variance", BuildAggrFuncFactoryCallback("Variance", "variance_0_1_traits_factory")},
  2740. {"stddev", BuildAggrFuncFactoryCallback("StdDev", "variance_1_1_traits_factory")},
  2741. {"populationvariance", BuildAggrFuncFactoryCallback("VariancePopulation", "variance_0_0_traits_factory")},
  2742. {"variancepopulation", BuildAggrFuncFactoryCallback("VariancePopulation", "variance_0_0_traits_factory")},
  2743. {"populationstddev", BuildAggrFuncFactoryCallback("StdDevPopulation", "variance_1_0_traits_factory")},
  2744. {"stddevpopulation", BuildAggrFuncFactoryCallback("StdDevPopulation", "variance_1_0_traits_factory")},
  2745. {"varpop", BuildAggrFuncFactoryCallback("VariancePopulation", "variance_0_0_traits_factory")},
  2746. {"stddevpop", BuildAggrFuncFactoryCallback("StdDevPopulation", "variance_1_0_traits_factory")},
  2747. {"varp", BuildAggrFuncFactoryCallback("VariancePopulation", "variance_0_0_traits_factory")},
  2748. {"stddevp", BuildAggrFuncFactoryCallback("StdDevPopulation", "variance_1_0_traits_factory")},
  2749. {"variancesample", BuildAggrFuncFactoryCallback("VarianceSample", "variance_0_1_traits_factory")},
  2750. {"stddevsample", BuildAggrFuncFactoryCallback("StdDevSample", "variance_1_1_traits_factory")},
  2751. {"varsamp", BuildAggrFuncFactoryCallback("VarianceSample", "variance_0_1_traits_factory")},
  2752. {"stddevsamp", BuildAggrFuncFactoryCallback("StdDevSample", "variance_1_1_traits_factory")},
  2753. {"vars", BuildAggrFuncFactoryCallback("VarianceSample", "variance_0_1_traits_factory")},
  2754. {"stddevs", BuildAggrFuncFactoryCallback("StdDevSample", "variance_1_1_traits_factory")},
  2755. {"correlation", BuildAggrFuncFactoryCallback("Correlation", "correlation_traits_factory", TWO_ARGS)},
  2756. {"corr", BuildAggrFuncFactoryCallback("Correlation", "correlation_traits_factory", TWO_ARGS, "Corr")},
  2757. {"covariance", BuildAggrFuncFactoryCallback("CovarianceSample", "covariance_sample_traits_factory", TWO_ARGS, "Covariance")},
  2758. {"covariancesample", BuildAggrFuncFactoryCallback("CovarianceSample", "covariance_sample_traits_factory", TWO_ARGS)},
  2759. {"covarsamp", BuildAggrFuncFactoryCallback("CovarianceSample", "covariance_sample_traits_factory", TWO_ARGS, "CovarSamp")},
  2760. {"covar", BuildAggrFuncFactoryCallback("CovarianceSample", "covariance_sample_traits_factory", TWO_ARGS, "Covar")},
  2761. {"covars", BuildAggrFuncFactoryCallback("CovarianceSample", "covariance_sample_traits_factory", TWO_ARGS, "CovarS")},
  2762. {"covariancepopulation", BuildAggrFuncFactoryCallback("CovariancePopulation", "covariance_population_traits_factory", TWO_ARGS)},
  2763. {"covarpop", BuildAggrFuncFactoryCallback("CovariancePopulation", "covariance_population_traits_factory", TWO_ARGS, "CovarPop")},
  2764. {"covarp", BuildAggrFuncFactoryCallback("CovariancePopulation", "covariance_population_traits_factory", TWO_ARGS, "CovarP")},
  2765. {"udaf", BuildAggrFuncFactoryCallback("UDAF", "udaf_traits_factory", UDAF)},
  2766. // Window functions
  2767. {"firstvalue", BuildAggrFuncFactoryCallback("FirstValue", "first_value_traits_factory", {OverWindow})},
  2768. {"lastvalue", BuildAggrFuncFactoryCallback("LastValue", "last_value_traits_factory", {OverWindow})},
  2769. {"nthvalue", BuildAggrFuncFactoryCallback("NthValue", "nth_value_traits_factory", {OverWindow}, NTH_VALUE)},
  2770. {"firstvalueignorenulls", BuildAggrFuncFactoryCallback("FirstValueIgnoreNulls", "first_value_ignore_nulls_traits_factory", {OverWindow})},
  2771. {"lastvalueignorenulls", BuildAggrFuncFactoryCallback("LastValueIgnoreNulls", "last_value_ignore_nulls_traits_factory", {OverWindow})},
  2772. {"nthvalueignorenulls", BuildAggrFuncFactoryCallback("NthValueIgnoreNulls", "nth_value_ignore_nulls_traits_factory", {OverWindow}, NTH_VALUE)},
  2773. // MatchRecognize navigation functions
  2774. {"first", BuildAggrFuncFactoryCallback("First", "first_traits_factory")},
  2775. {"last", BuildAggrFuncFactoryCallback("Last", "last_traits_factory")},
  2776. };
  2777. return aggrFuncs;
  2778. }
  2779. TCoreFuncMap MakeCoreFuncs() {
  2780. TCoreFuncMap coreFuncs = {
  2781. {"listindexof", { "IndexOf", 2, 2}},
  2782. {"testbit", { "TestBit", 2, 2}},
  2783. {"setbit", { "SetBit", 2, 2}},
  2784. {"clearbit", { "ClearBit", 2, 2}},
  2785. {"flipbit", { "FlipBit", 2, 2 }},
  2786. {"toset", { "ToSet", 1, 1 }},
  2787. {"setisdisjoint", { "SetIsDisjoint", 2, 2}},
  2788. {"setintersection", { "SetIntersection", 2, 3}},
  2789. {"setincludes", { "SetIncludes", 2, 2}},
  2790. {"setunion", { "SetUnion", 2, 3}},
  2791. {"setdifference", { "SetDifference", 2, 2}},
  2792. {"setsymmetricdifference", { "SetSymmetricDifference", 2, 3}},
  2793. {"listaggregate", { "ListAggregate", 2, 2}},
  2794. {"dictaggregate", { "DictAggregate", 2, 2}},
  2795. {"aggregatetransforminput", { "AggregateTransformInput", 2, 2}},
  2796. {"aggregatetransformoutput", { "AggregateTransformOutput", 2, 2}},
  2797. {"aggregateflatten", { "AggregateFlatten", 1, 1}},
  2798. {"choosemembers", { "ChooseMembers", 2, 2}},
  2799. {"removemembers", { "RemoveMembers", 2, 2}},
  2800. {"forceremovemembers", { "ForceRemoveMembers", 2, 2}},
  2801. {"structmembers", { "StructMembers", 1, 1}},
  2802. {"gathermembers", { "GatherMembers", 1, 1}},
  2803. {"renamemembers", { "RenameMembers", 2, 2}},
  2804. {"forcerenamemembers", { "ForceRenameMembers", 2, 2}},
  2805. {"spreadmembers", { "SpreadMembers", 2, 2}},
  2806. {"forcespreadmembers", { "ForceSpreadMembers", 2, 2}},
  2807. {"listfromtuple", { "ListFromTuple", 1, 1}},
  2808. {"listtotuple", { "ListToTuple", 2, 2}},
  2809. {"opaque", { "Opaque", 1, 1}},
  2810. };
  2811. return coreFuncs;
  2812. }
  2813. };
  2814. TNodePtr BuildBuiltinFunc(TContext& ctx, TPosition pos, TString name, const TVector<TNodePtr>& args,
  2815. const TString& originalNameSpace, EAggregateMode aggMode, bool* mustUseNamed, bool warnOnYqlNameSpace) {
  2816. const TBuiltinFuncData* funcData = Singleton<TBuiltinFuncData>();
  2817. const TBuiltinFactoryCallbackMap& builtinFuncs = funcData->BuiltinFuncs;
  2818. const TAggrFuncFactoryCallbackMap& aggrFuncs = funcData->AggrFuncs;
  2819. const TCoreFuncMap& coreFuncs = funcData->CoreFuncs;
  2820. for (auto& arg: args) {
  2821. if (!arg) {
  2822. return nullptr;
  2823. }
  2824. }
  2825. TString normalizedName(name);
  2826. TString nameSpace(originalNameSpace);
  2827. TString ns = to_lower(nameSpace);
  2828. if (ns.empty()) {
  2829. TMaybe<TIssue> error = NormalizeName(pos, normalizedName);
  2830. if (!error.Empty()) {
  2831. return new TInvalidBuiltin(pos, error->GetMessage());
  2832. }
  2833. auto coreFunc = coreFuncs.find(normalizedName);
  2834. if (coreFunc != coreFuncs.end()) {
  2835. ns = "core";
  2836. name = coreFunc->second.Name;
  2837. if (args.size() < coreFunc->second.MinArgs || args.size() > coreFunc->second.MaxArgs) {
  2838. return new TInvalidBuiltin(pos, TStringBuilder() << name << " expected from "
  2839. << coreFunc->second.MinArgs << " to " << coreFunc->second.MaxArgs << " arguments, but got: " << args.size());
  2840. }
  2841. if (coreFunc->second.MinArgs != coreFunc->second.MaxArgs) {
  2842. name += ToString(args.size());
  2843. }
  2844. }
  2845. }
  2846. TString moduleResource;
  2847. if (ctx.Settings.ModuleMapping.contains(ns)) {
  2848. moduleResource = ctx.Settings.ModuleMapping.at(ns);
  2849. }
  2850. if (ns == "js") {
  2851. ns = "javascript";
  2852. nameSpace = "JavaScript";
  2853. }
  2854. if (ns == "datetime2") {
  2855. ctx.Warning(pos, TIssuesIds::YQL_DEPRECATED_DATETIME2) << "DateTime2:: is a temporary alias for DateTime:: which will be removed in the future, use DateTime:: instead";
  2856. }
  2857. if (ns == "datetime") {
  2858. ns = "datetime2";
  2859. nameSpace = "DateTime2";
  2860. }
  2861. auto scriptType = NKikimr::NMiniKQL::ScriptTypeFromStr(ns);
  2862. switch (scriptType) {
  2863. case NKikimr::NMiniKQL::EScriptType::Python:
  2864. case NKikimr::NMiniKQL::EScriptType::Python3:
  2865. case NKikimr::NMiniKQL::EScriptType::ArcPython3:
  2866. scriptType = NKikimr::NMiniKQL::EScriptType::Python3;
  2867. break;
  2868. case NKikimr::NMiniKQL::EScriptType::Python2:
  2869. scriptType = NKikimr::NMiniKQL::EScriptType::ArcPython2;
  2870. break;
  2871. case NKikimr::NMiniKQL::EScriptType::SystemPython2:
  2872. scriptType = NKikimr::NMiniKQL::EScriptType::Python2;
  2873. break;
  2874. default:
  2875. break;
  2876. }
  2877. if (ns == "yql" || ns == "@yql") {
  2878. if (warnOnYqlNameSpace && GetEnv("YQL_DETERMINISTIC_MODE").empty()) {
  2879. ctx.Warning(pos, TIssuesIds::YQL_S_EXPRESSIONS_CALL)
  2880. << "It is not recommended to directly access s-expressions functions via YQL::" << Endl
  2881. << "This mechanism is mostly intended for temporary workarounds or internal testing purposes";
  2882. }
  2883. if (ns == "yql") {
  2884. return new TCallNodeImpl(pos, name, -1, -1, args);
  2885. }
  2886. } else if (moduleResource) {
  2887. auto exportName = ns == "core" ? name : "$" + name;
  2888. TVector<TNodePtr> applyArgs = {
  2889. new TCallNodeImpl(pos, "bind", {
  2890. BuildAtom(pos, ns + "_module", 0), BuildQuotedAtom(pos, exportName)
  2891. })
  2892. };
  2893. applyArgs.insert(applyArgs.end(), args.begin(), args.end());
  2894. return new TCallNodeImpl(pos, "Apply", applyArgs);
  2895. } else if (ns == "hyperscan" || ns == "pcre" || ns == "pire" || ns.StartsWith("re2")) {
  2896. TString moduleName(nameSpace);
  2897. moduleName.to_title();
  2898. if ((args.size() == 1 || args.size() == 2) && (name.StartsWith("Multi") || (ns.StartsWith("re2") && name == "Capture"))) {
  2899. TVector<TNodePtr> multiArgs{
  2900. ns.StartsWith("re2") && name == "Capture" ? MakePair(pos, args) : args[0],
  2901. new TCallNodeImpl(pos, "Void", 0, 0, {}),
  2902. args[0]
  2903. };
  2904. auto fullName = moduleName + "." + name;
  2905. return new TYqlTypeConfigUdf(pos, fullName, multiArgs, multiArgs.size() + 1);
  2906. } else if (!(ns.StartsWith("re2") && name == "Options")) {
  2907. auto newArgs = args;
  2908. if (ns.StartsWith("re2")) {
  2909. // convert run config is tuple of string and optional options
  2910. if (args.size() == 1 || args.size() == 2) {
  2911. newArgs[0] = MakePair(pos, args);
  2912. if (args.size() == 2) {
  2913. newArgs.pop_back();
  2914. }
  2915. } else {
  2916. return new TInvalidBuiltin(pos, TStringBuilder() << ns << "." << name << " expected one or two arguments.");
  2917. }
  2918. }
  2919. return BuildUdf(ctx, pos, moduleName, name, newArgs);
  2920. }
  2921. } else if (ns == "datetime2" && (name == "Parse")) {
  2922. return BuildUdf(ctx, pos, nameSpace, name, args);
  2923. } else if (ns == "pg" || ns == "pgagg" || ns == "pgproc") {
  2924. bool isAggregateFunc = NYql::NPg::HasAggregation(name, NYql::NPg::EAggKind::Normal);
  2925. bool isNormalFunc = NYql::NPg::HasProc(name, NYql::NPg::EProcKind::Function);
  2926. if (!isAggregateFunc && !isNormalFunc) {
  2927. return new TInvalidBuiltin(pos, TStringBuilder() << "Unknown function: " << name);
  2928. }
  2929. if (isAggregateFunc && isNormalFunc) {
  2930. if (ns == "pg") {
  2931. return new TInvalidBuiltin(pos, TStringBuilder() << "Ambigious function: " << name << ", use either PgAgg:: or PgProc:: namespace");
  2932. } else if (ns == "pgagg") {
  2933. isNormalFunc = false;
  2934. } else {
  2935. isAggregateFunc = false;
  2936. }
  2937. }
  2938. if (isAggregateFunc && ns == "pgproc") {
  2939. return new TInvalidBuiltin(pos, TStringBuilder() << "Invalid namespace for aggregation function: " << name << ", use either Pg:: or PgAgg:: namespace");
  2940. }
  2941. if (isNormalFunc && ns == "pgagg") {
  2942. return new TInvalidBuiltin(pos, TStringBuilder() << "Invalid namespace for normal function: " << name << ", use either Pg:: or PgProc:: namespace");
  2943. }
  2944. if (isAggregateFunc) {
  2945. if (aggMode == EAggregateMode::Distinct) {
  2946. return new TInvalidBuiltin(pos, "Distinct is not supported yet for PG aggregation ");
  2947. }
  2948. return BuildAggrFuncFactoryCallback(name, "", EAggrFuncTypeCallback::PG)(pos, args, aggMode, false);
  2949. } else {
  2950. YQL_ENSURE(isNormalFunc);
  2951. TVector<TNodePtr> pgCallArgs;
  2952. pgCallArgs.push_back(BuildLiteralRawString(pos, name));
  2953. pgCallArgs.insert(pgCallArgs.end(), args.begin(), args.end());
  2954. return new TYqlPgCall<false>(pos, pgCallArgs);
  2955. }
  2956. } else if (name == "MakeLibraPreprocessor") {
  2957. if (args.size() != 1) {
  2958. return new TInvalidBuiltin(pos, TStringBuilder() << name << " requires exactly one argument");
  2959. }
  2960. auto settings = NYT::TNode::CreateMap();
  2961. auto makeUdfArgs = [&args, &pos, &settings]() {
  2962. return TVector<TNodePtr> {
  2963. args[0],
  2964. new TCallNodeImpl(pos, "Void", {}),
  2965. BuildQuotedAtom(pos, NYT::NodeToYsonString(settings))
  2966. };
  2967. };
  2968. auto structNode = args[0]->GetStructNode();
  2969. if (!structNode) {
  2970. if (auto callNode = args[0]->GetCallNode()) {
  2971. if (callNode->GetOpName() == "AsStruct") {
  2972. return BuildUdf(ctx, pos, nameSpace, name, makeUdfArgs());
  2973. }
  2974. }
  2975. return new TInvalidBuiltin(pos, TStringBuilder() << name << " requires struct as argument");
  2976. }
  2977. for (const auto& item : structNode->GetExprs()) {
  2978. const auto& label = item->GetLabel();
  2979. if (label == "Entities") {
  2980. auto callNode = item->GetCallNode();
  2981. if (!callNode || callNode->GetOpName() != "AsListMayWarn") {
  2982. return new TInvalidBuiltin(pos, TStringBuilder() << name << " entities must be list of strings");
  2983. }
  2984. auto entities = NYT::TNode::CreateList();
  2985. for (const auto& entity : callNode->GetArgs()) {
  2986. if (!entity->IsLiteral() || entity->GetLiteralType() != "String") {
  2987. return new TInvalidBuiltin(pos, TStringBuilder() << name << " entity must be string literal");
  2988. }
  2989. entities.Add(entity->GetLiteralValue());
  2990. }
  2991. settings(label, std::move(entities));
  2992. } else if (label == "EntitiesStrategy") {
  2993. if (!item->IsLiteral() || item->GetLiteralType() != "String") {
  2994. return new TInvalidBuiltin(
  2995. pos, TStringBuilder() << name << " entities strategy must be string literal"
  2996. );
  2997. }
  2998. if (!EqualToOneOf(item->GetLiteralValue(), "whitelist", "blacklist")) {
  2999. return new TInvalidBuiltin(
  3000. pos,
  3001. TStringBuilder() << name << " got invalid entities strategy: expected 'whitelist' or 'blacklist'"
  3002. );
  3003. }
  3004. settings(label, item->GetLiteralValue());
  3005. } else if (label == "Mode") {
  3006. if (!item->IsLiteral() || item->GetLiteralType() != "String") {
  3007. return new TInvalidBuiltin(
  3008. pos, TStringBuilder() << name << " mode must be string literal"
  3009. );
  3010. }
  3011. settings(label, item->GetLiteralValue());
  3012. } else if (EqualToOneOf(label, "BlockstatDict", "ParseWithFat")) {
  3013. continue;
  3014. } else {
  3015. return new TInvalidBuiltin(
  3016. pos,
  3017. TStringBuilder()
  3018. << name << " got unsupported setting: " << label
  3019. << "; supported: Entities, EntitiesStrategy, BlockstatDict, ParseWithFat" );
  3020. }
  3021. }
  3022. return BuildUdf(ctx, pos, nameSpace, name, makeUdfArgs());
  3023. } else if (scriptType != NKikimr::NMiniKQL::EScriptType::Unknown) {
  3024. auto scriptName = NKikimr::NMiniKQL::IsCustomPython(scriptType) ? nameSpace : TString(NKikimr::NMiniKQL::ScriptTypeAsStr(scriptType));
  3025. return new TScriptUdf(pos, scriptName, name, args);
  3026. } else if (ns.empty()) {
  3027. if (auto simpleType = LookupSimpleType(normalizedName, ctx.FlexibleTypes, /* isPgType = */ false)) {
  3028. const auto type = *simpleType;
  3029. if (NUdf::FindDataSlot(type)) {
  3030. YQL_ENSURE(type != "Decimal");
  3031. return new TYqlData(pos, type, args);
  3032. }
  3033. if (type.StartsWith("pg") || type.StartsWith("_pg")) {
  3034. TVector<TNodePtr> pgConstArgs;
  3035. if (!args.empty()) {
  3036. pgConstArgs.push_back(args.front());
  3037. pgConstArgs.push_back(new TCallNodeImpl(pos, "PgType", { BuildQuotedAtom(pos,
  3038. TString(type.StartsWith("pg") ? "" : "_") + type.substr(type.StartsWith("pg") ? 2 : 3), TNodeFlags::Default) }));
  3039. pgConstArgs.insert(pgConstArgs.end(), args.begin() + 1, args.end());
  3040. }
  3041. return new TYqlPgConst(pos, pgConstArgs);
  3042. } else if (type == "Void" || type == "EmptyList" || type == "EmptyDict") {
  3043. return new TCallNodeImpl(pos, type, 0, 0, args);
  3044. } else {
  3045. return new TInvalidBuiltin(pos, TStringBuilder() << "Can not create objects of type " << type);
  3046. }
  3047. }
  3048. if (normalizedName == "decimal") {
  3049. if (args.size() == 2) {
  3050. TVector<TNodePtr> dataTypeArgs = { BuildQuotedAtom(pos, "Decimal", TNodeFlags::Default) };
  3051. for (auto& arg : args) {
  3052. if (auto literal = arg->GetLiteral("Int32")) {
  3053. dataTypeArgs.push_back(BuildQuotedAtom(pos, *literal, TNodeFlags::Default));
  3054. } else {
  3055. dataTypeArgs.push_back(MakeAtomFromExpression(ctx.Pos(), ctx, arg).Build());
  3056. }
  3057. }
  3058. return new TCallNodeImpl(pos, "DataType", dataTypeArgs);
  3059. }
  3060. return new TYqlData(pos, "Decimal", args);
  3061. }
  3062. if (normalizedName == "tablename") {
  3063. return new TTableName(pos, args, ctx.Scoped->CurrService);
  3064. }
  3065. if (normalizedName == "aggregationfactory") {
  3066. if (args.size() < 1 || !args[0]->GetLiteral("String")) {
  3067. return new TInvalidBuiltin(pos, "AGGREGATION_FACTORY requries a function name");
  3068. }
  3069. auto aggNormalizedName = *args[0]->GetLiteral("String");
  3070. auto error = NormalizeName(pos, aggNormalizedName);
  3071. if (!error.Empty()) {
  3072. return new TInvalidBuiltin(pos, error->GetMessage());
  3073. }
  3074. if (aggNormalizedName == "aggregateby") {
  3075. return new TInvalidBuiltin(pos, "AGGREGATE_BY is not allowed to use with AGGREGATION_FACTORY");
  3076. }
  3077. if (aggNormalizedName == "multiaggregateby") {
  3078. return new TInvalidBuiltin(pos, "MULTI_AGGREGATE_BY is not allowed to use with AGGREGATION_FACTORY");
  3079. }
  3080. if (aggMode == EAggregateMode::Distinct || aggMode == EAggregateMode::OverWindowDistinct) {
  3081. return new TInvalidBuiltin(pos, "DISTINCT can only be used in aggregation functions");
  3082. }
  3083. if (to_lower(*args[0]->GetLiteral("String")).StartsWith("pg::")) {
  3084. auto name = args[0]->GetLiteral("String")->substr(4);
  3085. const bool isAggregateFunc = NYql::NPg::HasAggregation(name, NYql::NPg::EAggKind::Normal);
  3086. if (!isAggregateFunc) {
  3087. return new TInvalidBuiltin(pos, TStringBuilder() << "Unknown aggregation function: " << *args[0]->GetLiteral("String"));
  3088. }
  3089. return BuildAggrFuncFactoryCallback(name, "", EAggrFuncTypeCallback::PG)(pos, args, aggMode, true);
  3090. }
  3091. AdjustCheckedAggFuncName(aggNormalizedName, ctx);
  3092. auto aggrCallback = aggrFuncs.find(aggNormalizedName);
  3093. if (aggrCallback == aggrFuncs.end()) {
  3094. return new TInvalidBuiltin(pos, TStringBuilder() << "Unknown aggregation function: " << *args[0]->GetLiteral("String"));
  3095. }
  3096. switch (ctx.GetColumnReferenceState()) {
  3097. case EColumnRefState::MatchRecognizeMeasures:
  3098. [[fallthrough]];
  3099. case EColumnRefState::MatchRecognizeDefine:
  3100. return new TInvalidBuiltin(pos, "Cannot use aggregation factory inside the MATCH_RECOGNIZE context");
  3101. default:
  3102. if ("first" == aggNormalizedName || "last" == aggNormalizedName) {
  3103. return new TInvalidBuiltin(pos, "Cannot use FIRST and LAST outside the MATCH_RECOGNIZE context");
  3104. }
  3105. return (*aggrCallback).second(pos, args, aggMode, true);
  3106. }
  3107. }
  3108. if (normalizedName == "aggregateby" || normalizedName == "multiaggregateby") {
  3109. const bool multi = (normalizedName == "multiaggregateby");
  3110. if (args.size() != 2) {
  3111. return new TInvalidBuiltin(pos, TStringBuilder() << (multi ? "MULTI_AGGREGATE_BY" : "AGGREGATE_BY") << " requries two arguments");
  3112. }
  3113. auto name = multi ? "MultiAggregateBy" : "AggregateBy";
  3114. auto aggr = BuildFactoryAggregation(pos, name, "", aggMode, multi);
  3115. return new TBasicAggrFunc(pos, name, aggr, args);
  3116. }
  3117. AdjustCheckedAggFuncName(normalizedName, ctx);
  3118. auto aggrCallback = aggrFuncs.find(normalizedName);
  3119. if (aggrCallback != aggrFuncs.end()) {
  3120. switch (ctx.GetColumnReferenceState()) {
  3121. case EColumnRefState::MatchRecognizeMeasures: {
  3122. auto result = (*aggrCallback).second(pos, args, aggMode, false);
  3123. return BuildMatchRecognizeVarAccess(pos, std::move(result));
  3124. }
  3125. case EColumnRefState::MatchRecognizeDefine:
  3126. return BuildMatchRecognizeDefineAggregate(ctx.Pos(), normalizedName, args);
  3127. default:
  3128. if ("first" == normalizedName || "last" == normalizedName) {
  3129. return new TInvalidBuiltin(pos, "Cannot use FIRST and LAST outside the MATCH_RECOGNIZE context");
  3130. }
  3131. return (*aggrCallback).second(pos, args, aggMode, false);
  3132. }
  3133. }
  3134. if (aggMode == EAggregateMode::Distinct || aggMode == EAggregateMode::OverWindowDistinct) {
  3135. return new TInvalidBuiltin(pos, "DISTINCT can only be used in aggregation functions");
  3136. }
  3137. auto builtinCallback = builtinFuncs.find(normalizedName);
  3138. if (builtinCallback != builtinFuncs.end()) {
  3139. return (*builtinCallback).second(pos, args);
  3140. } else if (normalizedName == "udf") {
  3141. if (mustUseNamed && *mustUseNamed) {
  3142. *mustUseNamed = false;
  3143. }
  3144. return new TUdfNode(pos, args);
  3145. } else if (normalizedName == "asstruct" || normalizedName == "structtype") {
  3146. if (args.empty()) {
  3147. return new TCallNodeImpl(pos, normalizedName == "asstruct" ? "AsStruct" : "StructType", 0, 0, args);
  3148. }
  3149. if (mustUseNamed && *mustUseNamed) {
  3150. *mustUseNamed = false;
  3151. YQL_ENSURE(args.size() == 2);
  3152. Y_DEBUG_ABORT_UNLESS(args[0]->GetTupleNode());
  3153. auto posArgs = args[0]->GetTupleNode();
  3154. if (posArgs->IsEmpty()) {
  3155. if (normalizedName == "asstruct") {
  3156. return args[1];
  3157. } else {
  3158. Y_DEBUG_ABORT_UNLESS(args[1]->GetStructNode());
  3159. auto namedArgs = args[1]->GetStructNode();
  3160. return new TStructTypeNode(pos, namedArgs->GetExprs());
  3161. }
  3162. }
  3163. }
  3164. return new TInvalidBuiltin(pos, TStringBuilder() <<
  3165. (normalizedName == "asstruct" ? "AsStruct" : "StructType") <<
  3166. " requires all argument to be named");
  3167. } else if (normalizedName == "expandstruct") {
  3168. if (mustUseNamed) {
  3169. if (!*mustUseNamed) {
  3170. return new TInvalidBuiltin(pos, TStringBuilder() << "ExpandStruct requires at least one named argument");
  3171. }
  3172. *mustUseNamed = false;
  3173. }
  3174. YQL_ENSURE(args.size() == 2);
  3175. Y_DEBUG_ABORT_UNLESS(args[0]->GetTupleNode());
  3176. Y_DEBUG_ABORT_UNLESS(args[1]->GetStructNode());
  3177. auto posArgs = args[0]->GetTupleNode();
  3178. if (posArgs->GetTupleSize() != 1) {
  3179. return new TInvalidBuiltin(pos, TStringBuilder() << "ExpandStruct requires all arguments except first to be named");
  3180. }
  3181. TVector<TNodePtr> flattenMembersArgs = {
  3182. BuildTuple(pos, {BuildQuotedAtom(pos, ""), posArgs->GetTupleElement(0)}),
  3183. BuildTuple(pos, {BuildQuotedAtom(pos, ""), args[1]}),
  3184. };
  3185. return new TCallNodeImpl(pos, "FlattenMembers", 2, 2, flattenMembersArgs);
  3186. } else if (normalizedName == "visit" || normalizedName == "visitordefault") {
  3187. bool withDefault = normalizedName == "visitordefault";
  3188. TNodePtr variant;
  3189. TVector<TNodePtr> labels, handlers;
  3190. TMaybe<TNodePtr> dflt;
  3191. if (mustUseNamed && *mustUseNamed) {
  3192. *mustUseNamed = false;
  3193. auto &positional = *args[0]->GetTupleNode();
  3194. if (positional.GetTupleSize() != (withDefault ? 2 : 1)) {
  3195. return new TInvalidBuiltin(pos, TStringBuilder() << name
  3196. << " requires exactly " << (withDefault ? 2 : 1) << " positional arguments when named args are used");
  3197. }
  3198. auto &named = *args[1]->GetStructNode();
  3199. variant = positional.GetTupleElement(0);
  3200. auto &namedExprs = named.GetExprs();
  3201. labels.reserve(namedExprs.size());
  3202. handlers.reserve(namedExprs.size());
  3203. for (size_t idx = 0; idx < namedExprs.size(); idx++) {
  3204. labels.push_back(BuildQuotedAtom(pos, namedExprs[idx]->GetLabel()));
  3205. handlers.push_back(namedExprs[idx]);
  3206. }
  3207. if (withDefault) {
  3208. dflt = positional.GetTupleElement(positional.GetTupleSize() - 1);
  3209. }
  3210. } else {
  3211. size_t minArgs = withDefault ? 2 : 1;
  3212. if (args.size() < minArgs) {
  3213. return new TInvalidBuiltin(pos, TStringBuilder() << name
  3214. << " requires at least " << minArgs << " positional arguments");
  3215. }
  3216. variant = args[0];
  3217. labels.reserve(args.size() - minArgs);
  3218. handlers.reserve(args.size() - minArgs);
  3219. for (size_t idx = 0; idx < args.size() - minArgs; idx++) {
  3220. labels.push_back(BuildQuotedAtom(pos, ToString(idx)));
  3221. handlers.push_back(args[minArgs + idx]);
  3222. }
  3223. if (withDefault) {
  3224. dflt = args[1];
  3225. }
  3226. }
  3227. TVector<TNodePtr> resultArgs;
  3228. resultArgs.reserve(1 + labels.size() + handlers.size());
  3229. resultArgs.emplace_back(std::move(variant));
  3230. for (size_t idx = 0; idx < labels.size(); idx++) {
  3231. resultArgs.emplace_back(std::move(labels[idx]));
  3232. resultArgs.emplace_back(std::move(handlers[idx]));
  3233. }
  3234. if (dflt.Defined()) {
  3235. resultArgs.emplace_back(std::move(dflt->Get()));
  3236. }
  3237. return new TCallNodeImpl(pos, "SqlVisit", 1, -1, resultArgs);
  3238. } else if (normalizedName == "sqlexternalfunction") {
  3239. return new TCallNodeImpl(pos, "SqlExternalFunction", args);
  3240. } else {
  3241. return new TInvalidBuiltin(pos, TStringBuilder() << "Unknown builtin: " << name);
  3242. }
  3243. }
  3244. TNodePtr positionalArgs;
  3245. TNodePtr namedArgs;
  3246. if (mustUseNamed && *mustUseNamed) {
  3247. YQL_ENSURE(args.size() == 2);
  3248. positionalArgs = args[0];
  3249. namedArgs = args[1];
  3250. *mustUseNamed = false;
  3251. }
  3252. TVector<TNodePtr> usedArgs = args;
  3253. TNodePtr customUserType = nullptr;
  3254. if (ns == "json") {
  3255. ctx.Warning(pos, TIssuesIds::YQL_DEPRECATED_JSON_UDF) << "Json UDF is deprecated. Please use JSON API instead";
  3256. ns = "yson";
  3257. nameSpace = "Yson";
  3258. if (name == "Serialize") {
  3259. name = "SerializeJson";
  3260. }
  3261. else if (name == "Parse") {
  3262. name = "ParseJson";
  3263. }
  3264. }
  3265. if (ctx.PragmaYsonFast && ns == "yson") {
  3266. ns.append('2');
  3267. nameSpace.append('2');
  3268. }
  3269. if (ns.StartsWith("yson")) {
  3270. if (name == "ConvertTo" && usedArgs.size() > 1) {
  3271. customUserType = usedArgs[1];
  3272. usedArgs.erase(usedArgs.begin() + 1);
  3273. }
  3274. if (name == "Serialize") {
  3275. if (usedArgs) {
  3276. usedArgs.resize(1U);
  3277. }
  3278. } else if (ctx.PragmaYsonFast && name == "SerializeJsonEncodeUtf8") {
  3279. name = "SerializeJson";
  3280. if (usedArgs.size() < 2U) {
  3281. usedArgs.emplace_back(BuildYsonOptionsNode(pos, ctx.PragmaYsonAutoConvert, ctx.PragmaYsonStrict, ctx.PragmaYsonFast));
  3282. }
  3283. positionalArgs = BuildTuple(pos, usedArgs);
  3284. auto encodeUtf8 = BuildLiteralBool(pos, true);
  3285. encodeUtf8->SetLabel("EncodeUtf8");
  3286. namedArgs = BuildStructure(pos, {encodeUtf8});
  3287. usedArgs = {positionalArgs, namedArgs};
  3288. } else if (name.StartsWith("From")) {
  3289. name = "From";
  3290. } else if (name == "GetLength" || name.StartsWith("ConvertTo") || name.StartsWith("Parse") || name.StartsWith("SerializeJson")) {
  3291. if (usedArgs.size() < 2U) {
  3292. usedArgs.emplace_back(BuildYsonOptionsNode(pos, ctx.PragmaYsonAutoConvert, ctx.PragmaYsonStrict, ctx.PragmaYsonFast));
  3293. }
  3294. } else if (name == "Contains" || name.StartsWith("Lookup") || name.StartsWith("YPath")) {
  3295. if (usedArgs.size() < 3U) {
  3296. usedArgs.push_back(BuildYsonOptionsNode(pos, ctx.PragmaYsonAutoConvert, ctx.PragmaYsonStrict, ctx.PragmaYsonFast));
  3297. }
  3298. }
  3299. }
  3300. if (ns == "datetime2" && name == "Update") {
  3301. if (namedArgs) {
  3302. TStructNode* castedNamedArgs = namedArgs->GetStructNode();
  3303. Y_DEBUG_ABORT_UNLESS(castedNamedArgs);
  3304. auto exprs = castedNamedArgs->GetExprs();
  3305. for (auto& arg : exprs) {
  3306. if (arg->GetLabel() == "Timezone") {
  3307. arg = new TCallNodeImpl(pos, "TimezoneId", 1, 1, { arg });
  3308. arg->SetLabel("TimezoneId");
  3309. }
  3310. }
  3311. namedArgs = BuildStructure(pos, exprs);
  3312. usedArgs.pop_back();
  3313. usedArgs.push_back(namedArgs);
  3314. };
  3315. }
  3316. TNodePtr typeConfig = MakeTypeConfig(pos, ns, usedArgs);
  3317. return BuildSqlCall(ctx, pos, nameSpace, name, usedArgs, positionalArgs, namedArgs, customUserType, TDeferredAtom(typeConfig, ctx), nullptr);
  3318. }
  3319. } // namespace NSQLTranslationV1