1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150 |
- #include "mkql_computation_node_ut.h"
- #include <yql/essentials/minikql/mkql_runtime_version.h>
- namespace NKikimr {
- namespace NMiniKQL {
- Y_UNIT_TEST_SUITE(TMiniKQLMapJoinCoreTest) {
- Y_UNIT_TEST_LLVM(TestInnerOnTuple) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto optionalUi64Type = pb.NewDataType(NUdf::TDataType<ui64>::Id, true);
- const auto tupleType = pb.NewTupleType({optionalUi64Type, optionalUi64Type});
- const auto emptyOptionalUi64 = pb.NewEmptyOptional(optionalUi64Type);
- const auto key1 = pb.NewTuple(tupleType, {
- pb.NewOptional(pb.NewDataLiteral<ui64>(1)),
- pb.NewOptional(pb.NewDataLiteral<ui64>(1)),
- });
- const auto key2 = pb.NewTuple(tupleType, {
- pb.NewOptional(pb.NewDataLiteral<ui64>(2)),
- pb.NewOptional(pb.NewDataLiteral<ui64>(2)),
- });
- const auto key3 = pb.NewTuple(tupleType, {
- pb.NewOptional(pb.NewDataLiteral<ui64>(3)),
- emptyOptionalUi64,
- });
- const auto key4 = pb.NewTuple(tupleType, {
- pb.NewOptional(pb.NewDataLiteral<ui64>(4)),
- pb.NewOptional(pb.NewDataLiteral<ui64>(4)),
- });
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", tupleType},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToSortedDict(list2, false,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Right", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::Inner, {0U}, {1U, 0U}, {0U, 1U}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- Y_UNIT_TEST_LLVM(TestInner) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(3);
- const auto key4 = pb.NewDataLiteral<ui32>(4);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, false,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Right", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::Inner, {0U}, {1U, 0U}, {0U, 1U}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestInnerMulti) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Right", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::Inner, {0U}, {1U, 0U}, {0U, 1U}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeft) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(3);
- const auto key4 = pb.NewDataLiteral<ui32>(4);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, false,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Right", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::Left, {0U}, {1U, 0U}, {0U, 1U}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "A");
- UNIT_ASSERT(!tuple.GetElement(1));
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftMulti) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Right", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::Left, {0U}, {1U, 0U}, {0U, 1U}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "A");
- UNIT_ASSERT(!tuple.GetElement(1));
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftSemi) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::LeftSemi, {0U}, {1U, 1U, 0U, 0U}, {}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(0).Get<ui32>(), 2);
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "B");
- UNIT_ASSERT(iterator.Next(tuple));
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(0).Get<ui32>(), 2);
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "C");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftOnly) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::LeftOnly, {0U}, {1U, 1U, 0U, 0U}, {}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(0).Get<ui32>(), 1);
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "A");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftSemiWithNullKey) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key0 = pb.NewEmptyOptional(pb.NewDataType(NUdf::TDataType<ui32>::Id, true));
- const auto key1 = pb.NewOptional(pb.NewDataLiteral<ui32>(1));
- const auto key2 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key3 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key4 = pb.NewOptional(pb.NewDataLiteral<ui32>(3));
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id, true)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key0), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key0), "Payload", payload3),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::LeftSemi, {0U}, {1U, 1U, 0U, 0U}, {}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(0).Get<ui32>(), 2);
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "B");
- UNIT_ASSERT(iterator.Next(tuple));
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(0).Get<ui32>(), 2);
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "C");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftOnlyWithNullKey) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key0 = pb.NewEmptyOptional(pb.NewDataType(NUdf::TDataType<ui32>::Id, true));
- const auto key1 = pb.NewOptional(pb.NewDataLiteral<ui32>(1));
- const auto key2 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key3 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key4 = pb.NewOptional(pb.NewDataLiteral<ui32>(3));
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto structType = pb.NewStructType({
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id, true)},
- {"Payload", pb.NewDataType(NUdf::TDataType<char*>::Id)}
- });
- const auto list1 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key0), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key1), "Payload", payload1),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload2),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload3)
- });
- const auto list2 = pb.NewList(structType, {
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key0), "Payload", payload3),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key2), "Payload", payload4),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key3), "Payload", payload5),
- pb.AddMember(pb.AddMember(pb.NewEmptyStruct(), "Key", key4), "Payload", payload6)
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Member(item, "Key");
- },
- [&](TRuntimeNode item) {
- return pb.AddMember(pb.NewEmptyStruct(), "Payload", pb.Member(item, "Payload"));
- });
- const auto resultType = pb.NewFlowType(pb.NewStructType({
- {"Left", pb.NewDataType(NUdf::TDataType<char*>::Id)},
- {"Key", pb.NewDataType(NUdf::TDataType<ui32>::Id)},
- }));
- const auto pgmReturn = pb.Collect(pb.MapJoinCore(pb.ToFlow(list1), dict2, EJoinKind::LeftOnly, {0U}, {1U, 1U, 0U, 0U}, {}, resultType));
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNIT_ASSERT(!tuple.GetElement(0));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(0).Get<ui32>(), 1);
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "A");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- }
- #if !defined(MKQL_RUNTIME_VERSION) || MKQL_RUNTIME_VERSION >= 18u
- Y_UNIT_TEST_SUITE(TMiniKQLWideMapJoinCoreTest) {
- Y_UNIT_TEST_LLVM(TestInner) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(4);
- const auto key4 = pb.NewDataLiteral<ui32>(4);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, false,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::Inner, {0U}, {1U, 0U}, {0U, 1U}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestInnerMulti) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::Inner, {0U}, {1U, 0U}, {0U, 1U}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeft) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(3);
- const auto key4 = pb.NewDataLiteral<ui32>(4);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, false,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::Left, {0U}, {1U, 0U}, {0U, 1U}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "A");
- UNIT_ASSERT(!tuple.GetElement(1));
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftMulti) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::Left, {0U}, {1U, 0U}, {0U, 1U}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "A");
- UNIT_ASSERT(!tuple.GetElement(1));
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "X");
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(1), "Y");
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftSemi) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<ui32>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::LeftSemi, {0U}, {1U, 0U, 0U, 1U}, {}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(1).Get<ui32>(), 2);
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(1).Get<ui32>(), 2);
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftOnly) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key1 = pb.NewDataLiteral<ui32>(1);
- const auto key2 = pb.NewDataLiteral<ui32>(2);
- const auto key3 = pb.NewDataLiteral<ui32>(2);
- const auto key4 = pb.NewDataLiteral<ui32>(3);
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<ui32>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::LeftOnly, {0U}, {1U, 0U, 0U, 1U}, {}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "A");
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(1).Get<ui32>(), 1);
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftSemiWithNullKey) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key0 = pb.NewEmptyOptional(pb.NewDataType(NUdf::TDataType<ui32>::Id, true));
- const auto key1 = pb.NewOptional(pb.NewDataLiteral<ui32>(1));
- const auto key2 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key3 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key4 = pb.NewOptional(pb.NewDataLiteral<ui32>(3));
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id, true),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key0, payload4}),
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key0, payload3}),
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<ui32>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::LeftSemi, {0U}, {1U, 0U, 0U, 1U}, {}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "B");
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(1).Get<ui32>(), 2);
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "C");
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(1).Get<ui32>(), 2);
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- Y_UNIT_TEST_LLVM(TestLeftOnlyWithNullKey) {
- for (ui32 pass = 0; pass < 1; ++pass) {
- TSetup<LLVM> setup;
- TProgramBuilder& pb = *setup.PgmBuilder;
- const auto key0 = pb.NewEmptyOptional(pb.NewDataType(NUdf::TDataType<ui32>::Id, true));
- const auto key1 = pb.NewOptional(pb.NewDataLiteral<ui32>(1));
- const auto key2 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key3 = pb.NewOptional(pb.NewDataLiteral<ui32>(2));
- const auto key4 = pb.NewOptional(pb.NewDataLiteral<ui32>(3));
- const auto payload1 = pb.NewDataLiteral<NUdf::EDataSlot::String>("A");
- const auto payload2 = pb.NewDataLiteral<NUdf::EDataSlot::String>("B");
- const auto payload3 = pb.NewDataLiteral<NUdf::EDataSlot::String>("C");
- const auto payload4 = pb.NewDataLiteral<NUdf::EDataSlot::String>("X");
- const auto payload5 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Y");
- const auto payload6 = pb.NewDataLiteral<NUdf::EDataSlot::String>("Z");
- const auto tupleType = pb.NewTupleType({
- pb.NewDataType(NUdf::TDataType<ui32>::Id, true),
- pb.NewDataType(NUdf::TDataType<char*>::Id)
- });
- const auto list1 = pb.NewList(tupleType, {
- pb.NewTuple({key0, payload4}),
- pb.NewTuple({key1, payload1}),
- pb.NewTuple({key2, payload2}),
- pb.NewTuple({key3, payload3})
- });
- const auto list2 = pb.NewList(tupleType, {
- pb.NewTuple({key0, payload3}),
- pb.NewTuple({key2, payload4}),
- pb.NewTuple({key3, payload5}),
- pb.NewTuple({key4, payload6})
- });
- const auto dict2 = pb.ToHashedDict(list2, true,
- [&](TRuntimeNode item) {
- return pb.Nth(item, 0U);
- },
- [&](TRuntimeNode item) {
- return pb.NewTuple({pb.Nth(item, 1U)});
- });
- const auto resultType = pb.NewFlowType(pb.NewMultiType({
- pb.NewDataType(NUdf::TDataType<char*>::Id),
- pb.NewDataType(NUdf::TDataType<ui32>::Id)
- }));
- const auto pgmReturn = pb.Collect(pb.NarrowMap(pb.MapJoinCore(pb.ExpandMap(pb.ToFlow(list1),
- [&](TRuntimeNode item) -> TRuntimeNode::TList { return {pb.Nth(item, 0U), pb.Nth(item, 1U)}; }),
- dict2, EJoinKind::LeftOnly, {0U}, {1U, 0U, 0U, 1U}, {}, resultType),
- [&](TRuntimeNode::TList items) -> TRuntimeNode { return pb.NewTuple(items); })
- );
- const auto graph = setup.BuildGraph(pgmReturn);
- const auto iterator = graph->GetValue().GetListIterator();
- NUdf::TUnboxedValue tuple;
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "X");
- UNIT_ASSERT(!tuple.GetElement(1));
- UNIT_ASSERT(iterator.Next(tuple));
- UNBOXED_VALUE_STR_EQUAL(tuple.GetElement(0), "A");
- UNIT_ASSERT_VALUES_EQUAL(tuple.GetElement(1).Get<ui32>(), 1);
- UNIT_ASSERT(!iterator.Next(tuple));
- UNIT_ASSERT(!iterator.Next(tuple));
- }
- }
- }
- #endif
- }
- }
|