// Tests load and save from text file. TEST(TTable, LoadSave) { TTableContext Context; // Create schema. Schema GradeS; GradeS.Add(TPair<TStr,TAttrType>("Class", atStr)); GradeS.Add(TPair<TStr,TAttrType>("Area", atStr)); GradeS.Add(TPair<TStr,TAttrType>("Quarter", atStr)); GradeS.Add(TPair<TStr,TAttrType>("Grade 2011", atInt)); GradeS.Add(TPair<TStr,TAttrType>("Grade 2012", atInt)); GradeS.Add(TPair<TStr,TAttrType>("Grade 2013", atInt)); TIntV RelevantCols; RelevantCols.Add(0); RelevantCols.Add(1); RelevantCols.Add(2); RelevantCols.Add(3); RelevantCols.Add(4); RelevantCols.Add(5); PTable P = TTable::LoadSS(GradeS, "table/grades.txt", &Context, RelevantCols); EXPECT_EQ(5, P->GetNumRows().Val); EXPECT_EQ(5, P->GetNumValidRows().Val); EXPECT_EQ(7, P->GetIntVal("Grade 2011", 0).Val); EXPECT_EQ(9, P->GetIntVal("Grade 2013", 4).Val); EXPECT_STREQ("Compilers", P->GetStrVal("Class", 3).CStr()); P->SaveSS("table/p1.txt"); // Test SaveSS by loading the saved table and testing values again. GradeS.Add(TPair<TStr,TAttrType>("_id", atInt)); P = TTable::LoadSS(GradeS, "table/p1.txt", &Context, RelevantCols); EXPECT_EQ(5, P->GetNumRows().Val); EXPECT_EQ(5, P->GetNumValidRows().Val); EXPECT_EQ(7, P->GetIntVal("Grade 2011", 0).Val); EXPECT_EQ(9, P->GetIntVal("Grade 2013", 4).Val); EXPECT_STREQ("Compilers", P->GetStrVal("Class", 3).CStr()); }
// Tests parallel join function. TEST(TTable, ParallelJoin) { TTableContext Context; Schema LJS; LJS.Add(TPair<TStr,TAttrType>("Src", atInt)); LJS.Add(TPair<TStr,TAttrType>("Dst", atInt)); TIntV RelevantCols; RelevantCols.Add(0); RelevantCols.Add(1); PTable T1 = TTable::LoadSS(LJS, "table/soc-LiveJournal1_small.txt", &Context, RelevantCols); EXPECT_EQ(499, T1->GetNumRows().Val); EXPECT_EQ(499, T1->GetNumValidRows().Val); PTable T2 = TTable::LoadSS(LJS, "table/soc-LiveJournal1_small.txt", &Context, RelevantCols); EXPECT_EQ(499, T2->GetNumRows().Val); EXPECT_EQ(499, T2->GetNumValidRows().Val); PTable P = T1->Join("Src", T2, "Dst"); EXPECT_EQ(24, P->GetNumRows().Val); EXPECT_EQ(24, P->GetNumValidRows().Val); }
int main(int argc, char* argv[]) { TEnv Env(argc, argv); TStr PrefixPath = Env.GetArgs() > 1 ? Env.GetArg(1) : TStr(""); double ts1 = Tick(); TTableContext Context; TVec<PTable> NodeTblV = TVec<PTable>(); TVec<PTable> EdgeTblV = TVec<PTable>(); Schema NodeSchema = Schema(); Schema EdgeSchema = Schema(); LoadFlickrTables(PrefixPath, Context, NodeTblV, NodeSchema, EdgeTblV, EdgeSchema); double ts2 = Tick(); int ExpectedSz = 0; for (TVec<PTable>::TIter it = NodeTblV.BegI(); it < NodeTblV.EndI(); it++) { PTable Table = *it; ExpectedSz += Table->GetNumRows(); } THash<TStr, TInt> Hash(ExpectedSz); TStrV OriNIdV(ExpectedSz); MergeNodeTables(NodeTblV, NodeSchema, Hash, OriNIdV); PTable EdgeTable = MergeEdgeTables(EdgeTblV, EdgeSchema, Hash, Context); double ts3 = Tick(); TStrV V; TStrV VE; VE.Add(EdgeSchema.GetVal(2).GetVal1()); PNEANet Graph = TSnap::ToNetwork<PNEANet>(EdgeTable, EdgeSchema.GetVal(0).GetVal1(), EdgeSchema.GetVal(1).GetVal1(), V, V, VE, aaLast); double ts4 = Tick(); //int nExps = 1; int nExps = 40; TIntFltH PageRankResults; for (int i = 0; i < nExps; i++) { PageRankResults = TIntFltH(ExpectedSz); #ifdef USE_OPENMP TSnap::GetWeightedPageRankMP2(Graph, PageRankResults, EdgeSchema.GetVal(2).GetVal1(), 0.849999999999998, 0.0001, 10); #else TSnap::GetWeightedPageRank(Graph, PageRankResults, EdgeSchema.GetVal(2).GetVal1(), 0.849999999999998, 0.0001, 10); #endif } double ts5 = Tick(); PSOut ResultOut = TFOut::New(PrefixPath + TStr("page-rank-results.tsv")); for (TIntFltH::TIter it = PageRankResults.BegI(); it < PageRankResults.EndI(); it++) { ResultOut->PutStrFmtLn("%s\t%f9", OriNIdV[it.GetKey()].CStr(), it.GetDat().Val); } double ts6 = Tick(); bool isPar = false; #ifdef USE_OPENMP isPar = true; #endif // PSOut FeaturesOut = TFOut::New(PrefixPath + "features.txt"); // FeaturesOut->PutStrFmtLn("Photo %d", PPhotoTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Users %d", PUserTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Tags %d", PTagTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Comments %d", PCommentTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Locations %d", PLocationTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Photo - Owner %d", PPhotoOwnerTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Photo - Comment %d", PPhotoCommentTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Photo - Location %d", PPhotoLocationTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Comment - User %d", PCommentUserTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Comment - User %d", PCommentUserTbl->GetNumRows().Val); //// FeaturesOut->PutStrFmtLn("Photo - Tagger %d", PPhotoTaggerTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Tagger - Tag %d", PTaggerTagTbl->GetNumRows().Val); // FeaturesOut->PutStrFmtLn("Total number of nodes = %d", Graph->GetNodes()); // FeaturesOut->PutStrFmtLn("Total number of edges = %d", Graph->GetEdges()); PSOut TimeOut = TFOut::New(PrefixPath + TStr("time.txt"), true); TimeOut->PutStrFmtLn("Experiment Weighted - %s - %s", PrefixPath.CStr(), (isPar ? "Parallel" : "Sequential")); TimeOut->PutStrFmtLn("Input Time = %f", GetCPUTimeUsage(ts1, ts2)); TimeOut->PutStrFmtLn("Preprocessing Time = %f", GetCPUTimeUsage(ts2, ts3)); TimeOut->PutStrFmtLn("Conversion Time = %f", GetCPUTimeUsage(ts3, ts4)); TimeOut->PutStrFmtLn("Computing Time = %f", GetCPUTimeUsage(ts4, ts5)/nExps); TimeOut->PutStrFmtLn("Output Time = %f", GetCPUTimeUsage(ts5, ts6)); return 0; }