C++ THash::GetDat方法代码示例

void TTable::Count(TStr CountColName, TStr Col){
  if(!ColTypeMap.IsKey(Col)){TExcept::Throw("no such column " + Col);}
  TIntV CntCol(NumRows);
  switch(GetColType(Col)){
    case INT:{
      THash<TInt,TIntV> T;  // can't really estimate the size of T for constructor hinting
      TIntV& Column = IntCols[GetColIdx(Col)];
      GroupByIntCol(Col, T, TIntV(0), true);
      for(TRowIterator it = BegRI(); it < EndRI(); it++){
        CntCol[it.GetRowIdx()] = T.GetDat(Column[it.GetRowIdx()]).Len();
      }
      break;
    }
    case FLT:{
      THash<TFlt,TIntV> T;
      TFltV& Column = FltCols[GetColIdx(Col)];
      GroupByFltCol(Col, T, TIntV(0), true);
      for(TRowIterator it = BegRI(); it < EndRI(); it++){
         CntCol[it.GetRowIdx()] = T.GetDat(Column[it.GetRowIdx()]).Len();
      }
      break;
    }
    case STR:{
      THash<TStr,TIntV> T;
      GroupByStrCol(Col, T, TIntV(0), true);
      for(TRowIterator it = BegRI(); it < EndRI(); it++){
        CntCol[it.GetRowIdx()] = T.GetDat(GetStrVal(Col, it.GetRowIdx())).Len();
      }
    }
  }
  // add count column
  IntCols.Add(CntCol);
  AddSchemaCol(CountColName, INT);
  ColTypeMap.AddDat(CountColName, TPair<TYPE,TInt>(INT, IntCols.Len()-1));
}

// IN-OUT edges are swapped (so that the prog runs faster)
// Send message via IN edge proportional to the OUT edge weight
void TWgtNet::ReinforceEdges(const int& NIters) {
  THash<TInt, TFlt> OutWgtSumH;
  for (TNodeI NI = BegNI(); NI < EndNI(); NI++) {
    double wgt = 0;
    for (int e = 0; e < NI.GetOutDeg(); e++) { 
      wgt += NI.GetOutEDat(e); }
    OutWgtSumH.AddDat(NI.GetId(), wgt);
  }
  printf("Reinforcing edges for %d iterations\n", NIters);
  // iterate
  TExeTm ExeTm;
  for (int iter = 0; iter < NIters; iter++) {
    for (TNodeI NI = BegNI(); NI < EndNI(); NI++) {
      const double X = TInt::Rnd.GetUniDev() * OutWgtSumH.GetDat(NI.GetId());
      double x = 0;  int e = 0;
      for ( ; x + NI.GetOutEDat(e) < X; e++) {
        x += NI.GetOutEDat(e); }
      IAssert(IsEdge(NI.GetOutNId(e), NI.GetId()));
      GetEDat(NI.GetOutNId(e), NI.GetId()) += 1; // reinforce the edge
      OutWgtSumH.GetDat(NI.GetOutNId(e)) += 1; 
    }
    if (iter % (NIters/100) == 0) {
      printf("\r%d [%s]", iter, ExeTm.GetStr()); 
    }
  }
  printf(" done.\n");
}

/// save bipartite community affiliation into gexf file
void TAGMUtil::SaveBipartiteGephi(const TStr& OutFNm, const TIntV& NIDV, const TVec<TIntV>& CmtyVV, const double MaxSz, const double MinSz, const TIntStrH& NIDNameH, const THash<TInt, TIntTr>& NIDColorH, const THash<TInt, TIntTr>& CIDColorH ) {
    /// Plot bipartite graph
    if (CmtyVV.Len() == 0) {
        return;
    }
    double NXMin = 0.1, YMin = 0.1, NXMax = 250.00, YMax = 30.0;
    double CXMin = 0.3 * NXMax, CXMax = 0.7 * NXMax;
    double CStep = (CXMax - CXMin) / (double) CmtyVV.Len(), NStep = (NXMax - NXMin) / (double) NIDV.Len();
    THash<TInt,TIntV> NIDComVH;
    TAGMUtil::GetNodeMembership(NIDComVH, CmtyVV);

    FILE* F = fopen(OutFNm.CStr(), "wt");
    fprintf(F, "<?xml version='1.0' encoding='UTF-8'?>\n");
    fprintf(F, "<gexf xmlns='http://www.gexf.net/1.2draft' xmlns:viz='http://www.gexf.net/1.1draft/viz' xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' xsi:schemaLocation='http://www.gexf.net/1.2draft http://www.gexf.net/1.2draft/gexf.xsd' version='1.2'>\n");
    fprintf(F, "\t<graph mode='static' defaultedgetype='directed'>\n");
    fprintf(F, "\t\t<nodes>\n");
    for (int c = 0; c < CmtyVV.Len(); c++) {
        int CID = c;
        double XPos = c * CStep + CXMin;
        TIntTr Color = CIDColorH.IsKey(CID)? CIDColorH.GetDat(CID) : TIntTr(120, 120, 120);
        fprintf(F, "\t\t\t<node id='C%d' label='C%d'>\n", CID, CID);
        fprintf(F, "\t\t\t\t<viz:color r='%d' g='%d' b='%d'/>\n", Color.Val1.Val, Color.Val2.Val, Color.Val3.Val);
        fprintf(F, "\t\t\t\t<viz:size value='%.3f'/>\n", MaxSz);
        fprintf(F, "\t\t\t\t<viz:shape value='square'/>\n");
        fprintf(F, "\t\t\t\t<viz:position x='%f' y='%f' z='0.0'/>\n", XPos, YMax);
        fprintf(F, "\t\t\t</node>\n");
    }

    for (int u = 0; u < NIDV.Len(); u++) {
        int NID = NIDV[u];
        TStr Label = NIDNameH.IsKey(NID)? NIDNameH.GetDat(NID): "";
        double Size = MinSz;
        double XPos = NXMin + u * NStep;
        TIntTr Color = NIDColorH.IsKey(NID)? NIDColorH.GetDat(NID) : TIntTr(120, 120, 120);
        double Alpha = 1.0;
        fprintf(F, "\t\t\t<node id='%d' label='%s'>\n", NID, Label.CStr());
        fprintf(F, "\t\t\t\t<viz:color r='%d' g='%d' b='%d' a='%.1f'/>\n", Color.Val1.Val, Color.Val2.Val, Color.Val3.Val, Alpha);
        fprintf(F, "\t\t\t\t<viz:size value='%.3f'/>\n", Size);
        fprintf(F, "\t\t\t\t<viz:shape value='square'/>\n");
        fprintf(F, "\t\t\t\t<viz:position x='%f' y='%f' z='0.0'/>\n", XPos, YMin);
        fprintf(F, "\t\t\t</node>\n");
    }
    fprintf(F, "\t\t</nodes>\n");
    fprintf(F, "\t\t<edges>\n");
    int EID = 0;
    for (int u = 0; u < NIDV.Len(); u++) {
        int NID = NIDV[u];
        if (NIDComVH.IsKey(NID)) {
            for (int c = 0; c < NIDComVH.GetDat(NID).Len(); c++) {
                int CID = NIDComVH.GetDat(NID)[c];
                fprintf(F, "\t\t\t<edge id='%d' source='C%d' target='%d'/>\n", EID++, CID, NID);
            }
        }
    }
    fprintf(F, "\t\t</edges>\n");
    fprintf(F, "\t</graph>\n");
    fprintf(F, "</gexf>\n");
}

 TFltIntIntTr FindMxQEdge() {
   while (true) {
     if (MxQHeap.Empty()) { break; }
     const TFltIntIntTr TopQ = MxQHeap.PopHeap();
     if (! CmtyQH.IsKey(TopQ.Val2) || ! CmtyQH.IsKey(TopQ.Val3)) { continue; }
     if (TopQ.Val1!=CmtyQH.GetDat(TopQ.Val2).GetMxQ() && TopQ.Val1!=CmtyQH.GetDat(TopQ.Val3).GetMxQ()) { continue; }
     return TopQ;
   }
   return TFltIntIntTr(-1, -1, -1);
 }

// YES I COPIED AND PASTED CODE my section leader would be so ashamed :D
void LSH::MinHash(THash<TMd5Sig, TIntSet>& ShingleToQuoteIds,
    TVec<THash<TIntV, TIntSet> >& SignatureBandBuckets) {
  TRnd RandomGenerator; // TODO: make this "more random" by incorporating time
  for (int i = 0; i < NumBands; ++i) {
    THash < TInt, TIntV > Inverted; // (QuoteID, QuoteSignatureForBand)
    THash < TIntV, TIntSet > BandBuckets; // (BandSignature, QuoteIDs)
    for (int j = 0; j < BandSize; ++j) {
      // Create new signature
      TVec < TMd5Sig > Signature;
      ShingleToQuoteIds.GetKeyV(Signature);
      Signature.Shuffle(RandomGenerator);

      // Place in bucket - not very efficient
      int SigLen = Signature.Len();
      for (int k = 0; k < SigLen; ++k) {
        TIntSet CurSet = ShingleToQuoteIds.GetDat(Signature[k]);
        for (TIntSet::TIter l = CurSet.BegI(); l < CurSet.EndI(); l++) {
          TInt Key = l.GetKey();
          if (Inverted.IsKey(Key)) {
            TIntV CurSignature = Inverted.GetDat(Key);
            if (CurSignature.Len() <= j) {
              CurSignature.Add(k);
              Inverted.AddDat(Key, CurSignature);
            }
          } else {
            TIntV NewSignature;
            NewSignature.Add(k);
            Inverted.AddDat(Key, NewSignature);
          }
        }
      }
    }

    TIntV InvertedKeys;
    Inverted.GetKeyV(InvertedKeys);
    TInt InvertedLen = InvertedKeys.Len();
    for (int k = 0; k < InvertedLen; ++k) {
      TIntSet Bucket;
      TIntV Signature = Inverted.GetDat(InvertedKeys[k]);
      if (BandBuckets.IsKey(Signature)) {
        Bucket = BandBuckets.GetDat(Signature);
      }
      Bucket.AddKey(InvertedKeys[k]);
      BandBuckets.AddDat(Signature, Bucket);
    }

    SignatureBandBuckets.Add(BandBuckets);
    Err("%d out of %d band signatures computed\n", i + 1, NumBands);
  }
  Err("Minhash step complete!\n");
}

int main(int argc, char *argv[]) {
  TStr BaseString = "/lfs/1/tmp/curis/week/QBDB.bin";
  TFIn BaseFile(BaseString);
  TQuoteBase *QB = new TQuoteBase;
  TDocBase *DB = new TDocBase;
  QB->Load(BaseFile);
  DB->Load(BaseFile);

  TIntV QuoteIds;
  QB->GetAllQuoteIds(QuoteIds);

  int NumQuotes = QuoteIds.Len();
  THash<TInt, TStrSet> PeakCounts;
  for (int i = 0; i < NumQuotes; i++) {
    TQuote CurQuote;
    if (QB->GetQuote(QuoteIds[i], CurQuote)) {
      TVec<TSecTm> Peaks;
      CurQuote.GetPeaks(DB, Peaks);
      TStr QuoteString;
      CurQuote.GetParsedContentString(QuoteString);
      TStrSet StringSet;
      if (PeakCounts.IsKey(Peaks.Len())) {
        StringSet = PeakCounts.GetDat(Peaks.Len());
      }
      StringSet.AddKey(QuoteString);
      PeakCounts.AddDat(Peaks.Len(), StringSet);
    }
  }

  TIntV PeakCountKeys;
  PeakCounts.GetKeyV(PeakCountKeys);
  PeakCountKeys.Sort(true);
  for (int i = 0; i < PeakCountKeys.Len(); i++) {
    TStrSet CurSet = PeakCounts.GetDat(PeakCountKeys[i]);
    if (CurSet.Len() > 0) {
      printf("QUOTES WITH %d PEAKS\n", PeakCountKeys[i].Val);
      printf("#########################################\n");
      THashSet<TStr> StringSet = PeakCounts.GetDat(PeakCountKeys[i]);
      for (THashSet<TStr>::TIter l = StringSet.BegI(); l < StringSet.EndI(); l++) {
        printf("%s\n", l.GetKey().CStr());
      }
      printf("\n");
    }
  }
  delete QB;
  delete DB;
  return 0;
}

void LSH::ElCheapoHashing(TQuoteBase *QuoteBase, TInt ShingleLen,
    THash<TMd5Sig, TIntSet>& ShingleToQuoteIds) {
  fprintf(stderr, "Hashing shingles the el cheapo way...\n");
  TIntV QuoteIds;
  QuoteBase->GetAllQuoteIds(QuoteIds);
  for (int qt = 0; qt < QuoteIds.Len(); qt++) {
    if (qt % 1000 == 0) {
      fprintf(stderr, "%d out of %d completed\n", qt, QuoteIds.Len());
    }
    TQuote Q;
    QuoteBase->GetQuote(QuoteIds[qt], Q);

    // Put x-character (or x-word) shingles into hash table; x is specified by ShingleLen parameter
    TStr QContentStr;
    Q.GetParsedContentString(QContentStr);
    TChA QContentChA = TChA(QContentStr);

    for (int i = 0; i < QContentChA.Len() - ShingleLen + 1; i++) {
      TChA ShingleChA = TChA();
      for (int j = 0; j < ShingleLen; j++) {
        ShingleChA.AddCh(QContentChA.GetCh(i + j));
      }
      TStr Shingle = TStr(ShingleChA);
      const TMd5Sig ShingleMd5(Shingle);
      TIntSet ShingleQuoteIds;
      if (ShingleToQuoteIds.IsKey(ShingleMd5)) {
        ShingleQuoteIds = ShingleToQuoteIds.GetDat(ShingleMd5);
      }

      ShingleQuoteIds.AddKey(QuoteIds[qt]);
      ShingleToQuoteIds.AddDat(ShingleMd5, ShingleQuoteIds);
    }
  }
  Err("Done with el cheapo hashing!\n");
}

/// Shingles by words
void LSH::HashShinglesOfClusters(TQuoteBase *QuoteBase,
    TClusterBase *ClusterBase, TIntV& ClusterIds, TInt ShingleLen,
    THash<TMd5Sig, TIntV>& ShingleToClusterIds) {
  Err("Hashing shingles of clusters...\n");
  for (int i = 0; i < ClusterIds.Len(); i++) {
    if (i % 1000 == 0) {
      fprintf(stderr, "%d out of %d completed\n", i, ClusterIds.Len());
    }
    TCluster C;
    ClusterBase->GetCluster(ClusterIds[i], C);
    //fprintf(stderr, "%d vs. %d\n", ClusterIds[i].Val, C.GetId().Val);

    // Put x-word shingles into hash table; x is specified by ShingleLen parameter
    THashSet < TMd5Sig > CHashedShingles;
    GetHashedShinglesOfCluster(QuoteBase, C, ShingleLen, CHashedShingles);
    for (THashSet<TMd5Sig>::TIter Hash = CHashedShingles.BegI();
        Hash < CHashedShingles.EndI(); Hash++) {
      TIntV ShingleClusterIds;
      if (ShingleToClusterIds.IsKey(*Hash)) {
        ShingleClusterIds = ShingleToClusterIds.GetDat(*Hash);
      }
      ShingleClusterIds.Add(ClusterIds[i]);
      ShingleToClusterIds.AddDat(*Hash, ShingleClusterIds);
    }
  }
  Err("Done hashing!\n");
}

void TSockSys::Connect(const uint64& SockId, const PSockHost& SockHost, const int& PortN) {
	// make sure it's a valid socket
	IAssert(IsSock(SockId));
	uv_tcp_t* SockHnd = SockIdToHndH.GetDat(SockId);
	// make sure we got a valid socket host
	IAssert(SockHost->IsOk());
	// get connection handle
	uv_connect_t* ConnectHnd = (uv_connect_t*)malloc(sizeof(uv_connect_t));
	// special handling for v4 and v6
	int ResCd = 0;
	if (SockHost->IsIpv4()) {
		// get address
		struct sockaddr_in Addr = uv_ip4_addr(SockHost->GetIpNum().CStr(), PortN);
		// establish connection
		ResCd = uv_tcp_connect(ConnectHnd, SockHnd, Addr, TSockSys::OnConnect);
	} else if (SockHost->IsIpv6()) {
		// get address
		struct sockaddr_in6 Addr = uv_ip6_addr(SockHost->GetIpNum().CStr(), PortN);
		// establish connection
		ResCd = uv_tcp_connect6(ConnectHnd, SockHnd, Addr, TSockSys::OnConnect);
	}
	// check for errors
	if (ResCd != 0) {
		// cleanup first
		free(SockHnd);
		// and throw exception
		throw TExcept::New("SockSys.Connect: Error establishing socket connection: " + SockSys.GetLastErr());
	}
}

int ComputeKCore(const PUNGraph& G) {
  int cnt = 0;
  for(TUNGraph::TNodeI NI = G->BegNI(); NI < G->EndNI(); NI++)
    cnt = max(cnt, NI.GetOutDeg());
  THashSet <TInt> D[cnt+1];
  THash <TInt, TInt> deg;
  for(TUNGraph::TNodeI NI = G->BegNI(); NI < G->EndNI(); NI++) { 
    TInt tmp = NI.GetOutDeg() - G->IsEdge(NI.GetId(), NI.GetId() );
    D[tmp.Val].AddKey(NI.GetId());
    deg.AddDat(NI.GetId()) = tmp;
  }
  int max_k = 0;
  for(int num_iters = 0;num_iters < G->GetNodes(); num_iters++)
    for(int i = 0; i < cnt; i++)
      if(D[i].Empty() == 0) {
        max_k = max(max_k, i);
        TInt a = *(D[i].BegI());
        D[i].DelKey(a);
        deg.AddDat(a.Val) = -1; // Hope overwriting works
        TUNGraph::TNodeI NI = G->GetNI(a.Val);
        for(int e = 0; e < NI.GetOutDeg(); e++) {
          TInt b = NI.GetOutNId(e);
          if(deg.GetDat(b) >= 0) {
            int Id = deg.GetKeyId(b); 
            D[deg[Id].Val].DelKey(b);
            deg[Id] = deg[Id] - 1;  //Hope the overwriting works
            D[deg[Id]].AddKey(b);
          }
        }
        break;
      }
  return max_k;
}

// wrong reading of string attributes
void TTable::BuildGraphTopology(PNEAGraph& Graph, THash<TFlt, TInt>& FSrNodeMap, THash<TFlt, TInt>& FDsNodeMap) {
  TYPE SrCT = GetColType(SrcCol);
  TInt SrIdx = GetColIdx(SrcCol);
  TYPE DsCT = GetColType(DstCol);
  TInt DsIdx = GetColIdx(DstCol);
  TInt SrcCnt = 0;
  TInt DstCnt = 0;
  
  for(TRowIterator RowI = BegRI(); RowI < EndRI(); RowI++) {
    if (SrCT == INT && DsCT == INT) {
      Graph->AddNode(IntCols[SrIdx][RowI.GetRowIdx()]);
      Graph->AddNode(IntCols[DsIdx][RowI.GetRowIdx()]);
      Graph->AddEdge(IntCols[SrIdx][RowI.GetRowIdx()], IntCols[DsIdx][RowI.GetRowIdx()], RowI.GetRowIdx());
    } else if (SrCT == INT && DsCT == FLT) {
      Graph->AddNode(IntCols[SrIdx][RowI.GetRowIdx()]);
      TFlt val = FltCols[DsIdx][RowI.GetRowIdx()];
      if (!FDsNodeMap.IsKey(val)) {
	      FDsNodeMap.AddDat(val, DstCnt++);
      }
      Graph->AddNode(FDsNodeMap.GetDat(val));
      Graph->AddEdge(IntCols[SrIdx][RowI.GetRowIdx()], FDsNodeMap.GetDat(val));
    } else if (SrCT == INT && DsCT == STR) {
      Graph->AddNode(IntCols[SrIdx][RowI.GetRowIdx()]);
      Graph->AddNode(StrColMaps[DsIdx][RowI.GetRowIdx()]);
      Graph->AddEdge(IntCols[SrIdx][RowI.GetRowIdx()], StrColMaps[DsIdx][RowI.GetRowIdx()], RowI.GetRowIdx());
    } else if (SrCT == FLT && DsCT == INT) {
      Graph->AddNode(IntCols[DsIdx][RowI.GetRowIdx()]);
      TFlt val = FltCols[SrIdx][RowI.GetRowIdx()];
      if (!FSrNodeMap.IsKey(val)) {
	      FSrNodeMap.AddDat(val, SrcCnt++);
      }
      Graph->AddNode(FSrNodeMap.GetDat(val));
      Graph->AddEdge(FSrNodeMap.GetDat(val), IntCols[SrIdx][RowI.GetRowIdx()], RowI.GetRowIdx());
    } else if (SrCT == FLT && DsCT == STR) {
      Graph->AddNode(StrColMaps[DsIdx][RowI.GetRowIdx()]);
      TFlt val = FltCols[SrIdx][RowI.GetRowIdx()];
      if (!FSrNodeMap.IsKey(val)) {
	      FSrNodeMap.AddDat(val, SrcCnt++);
      }
      Graph->AddNode(FSrNodeMap.GetDat(val));
      Graph->AddEdge(FSrNodeMap.GetDat(val), IntCols[SrIdx][RowI.GetRowIdx()], RowI.GetRowIdx());
    } else if (SrCT == FLT && DsCT == FLT) {
      TFlt val = FltCols[SrIdx][RowI.GetRowIdx()];
      if (!FSrNodeMap.IsKey(val)) {
	      FSrNodeMap.AddDat(val, SrcCnt++);
      }
      Graph->AddNode(FSrNodeMap.GetDat(val));
      val = FltCols[DsIdx][RowI.GetRowIdx()];
      if (!FDsNodeMap.IsKey(val)) {
	      FDsNodeMap.AddDat(val, DstCnt++);
      }
      Graph->AddNode(FDsNodeMap.GetDat(val));
      Graph->AddEdge(FSrNodeMap.GetDat(val), FDsNodeMap.GetDat(val), RowI.GetRowIdx());
    }
  }
}

TInt TTable::GetNId(TStr Col, TInt RowIdx, THash<TFlt, TInt>& FSrNodeMap, THash<TFlt, TInt>& FDsNodeMap) {
  TYPE CT = GetColType(Col);
  TInt Idx = GetColIdx(Col);
  if (CT == INT) {
    return IntCols[RowIdx][Idx];
  } else if (CT == FLT) {
    if (Col == SrcCol) {
      return FSrNodeMap.GetDat(FltCols[Idx][RowIdx]);
    } else if (Col == DstCol) {
      return FDsNodeMap.GetDat(FltCols[Idx][RowIdx]);
    } else {
      TExcept::Throw("Column " + Col + " is not source node or destination column");
    }
  } else {
    return StrColMaps[RowIdx][Idx]();
  }
  return 0;
}

void TSockSys::DelSock(const uint64& SockId) {
	if (Active) {
		uv_tcp_t* SockHnd = SockIdToHndH.GetDat(SockId);
		// close the handle
		uv_close((uv_handle_t*)SockHnd, TSockSys::OnClose);
		// mark it's under closing so we don't handle it again in the destructor
		ClosedSockIdSet.AddKey(SockId);
	}
}

void TSockSys::DelIfSockTimer(const uint64& SockId) {
	if (SockIdToTimerHndH.IsKey(SockId)) {
		// get timer handle
		uv_timer_t* TimerHnd = SockIdToTimerHndH.GetDat(SockId);
		// stop the timer
		uv_timer_stop(TimerHnd);
		// remove shortcuts
		SockIdToTimerHndH.DelKey(SockId);
		TimerHndToSockIdH.DelKey((uint64)TimerHnd);
		// remove shortcuts
	}
}

TStr GetFlagStr(const TGraphFlag& GraphFlag) {
  static THash<TInt, TStr> GraphFlagToStrH;
  if (GraphFlagToStrH.Empty()) {
    GraphFlagToStrH.AddDat((int) gfUndef, "Undef");
    GraphFlagToStrH.AddDat((int) gfDirected, "Directed");
    GraphFlagToStrH.AddDat((int) gfMultiGraph, "MultiGraph");
    GraphFlagToStrH.AddDat((int) gfNodeDat, "NodeDat");
    GraphFlagToStrH.AddDat((int) gfEdgeDat, "EdgeDat");
    GraphFlagToStrH.AddDat((int) gfSources, "Sources");
  }
  return GraphFlagToStrH.GetDat((int) GraphFlag);
}