void LogOutput::ComputeOldRankString(THash<TInt, TInt>& OldRankings, TInt& ClusterId, TInt CurRank, TStr& OldRankStr) {
TInt OldRanking;
if (OldRankings.IsKeyGetDat(ClusterId, OldRanking)) {
TInt Difference = OldRanking - CurRank;
if (Difference < 0) {
OldRankStr = "<b><center><font color=\\\"red\\\">" + Difference.GetStr() + "</font></center></b>";
} else if (Difference > 0) {
OldRankStr = "<b><center><font color=\\\"green\\\">+" + Difference.GetStr() + "</font></center></b>";
} else {
OldRankStr = "<b><center>0</center></b>";
}
} else {
OldRankStr = "<center>new!</center>";
}
}
int Intersect1(TUNGraph::TNodeI Node, TStr NNodes){
int br = 0;
for (int i = 0; i<Node.GetDeg(); i++)
{
TInt digi = Node.GetNbrNId(i);
TStr buf = "";
buf = digi.GetStr();
if (NNodes.SearchStr(buf.CStr()) != -1)
br++;
}
TInt digi = Node.GetId();
TStr buf = digi.GetStr();
if (NNodes.SearchStr(buf.CStr()) != -1)
br++;
return br;
}
int Intersect(TUNGraph::TNodeI Node, TStr NNodes){
int br = 0;
TInt digi = -1;
TStr buf = "";
for (int i = 0; i<Node.GetDeg(); i++)
{
digi = Node.GetNbrNId(i);
TStr buf = digi.GetStr();
if (NNodes.IsStrIn(buf.CStr()))
br++;
}
digi = Node.GetId();
buf = digi.GetStr();
if (NNodes.IsStrIn(buf.CStr()))
br++;
return br;
}
//Function to read a table of nodes
PTable AddNodeTable(TTableContext& Context) {
Schema NodeScm;
NodeScm.Add(TPair<TStr, TAttrType>("NodeID", atStr));
char FileName[50];
int ColCnt = 0;
printf("Adding Node Table\n");
printf("Enter filename and number of columns (>= 1) \n");
scanf("%s %d", FileName, &ColCnt);
for (TInt i = 1; i < ColCnt; i++) {
TStr ColName = "Attribute" + i.GetStr();
NodeScm.Add(TPair<TStr, TAttrType>(ColName, atStr));
}
TStr FName(FileName);
PTable T = TTable::LoadSS(NodeScm, FName, Context);
return T;
}
int main(){
TTableContext Context;
// create scheme
Schema AnimalS;
AnimalS.Add(TPair<TStr,TAttrType>("Animal", atStr));
AnimalS.Add(TPair<TStr,TAttrType>("Size", atStr));
AnimalS.Add(TPair<TStr,TAttrType>("Location", atStr));
AnimalS.Add(TPair<TStr,TAttrType>("Number", atInt));
TIntV RelevantCols;
RelevantCols.Add(0);
RelevantCols.Add(1);
RelevantCols.Add(2);
RelevantCols.Add(3);
PTable P = TTable::LoadSS("Animals", AnimalS, "tests/animals.txt", Context, RelevantCols);
P->SaveSS("tests/p1.txt");
TStrV cols;
cols.Add("Size");
cols.Add("Number");
TVec<PTable> R = P->SpliceByGroup(cols);
for (TInt i = 0; i < R.Len(); i++) {
TStr fn = i.GetStr();
R[i]->SaveSS("tests/sznumber" + fn + ".txt");
}
P->Unique(cols, true);
P->SaveSS("tests/p2.txt");
TStrV group1;
group1.Add("Location");
P->Group(group1, "LocationGroup");
P->SaveSS("tests/p3.txt");
return 0;
}
TStr DCluster::GetClusterString(TQuoteBase *QB, TDocBase *DB, TCluster& C, TFreqTripleV &FreqV, TInt NumPeaks, TStr &End) {
TStr Response;
TSecTm Start;
TStr RepStr, RepURL;
C.GetBirthDate(Start);
C.GetRepresentativeQuoteString(RepStr, QB);
C.GetRepresentativeQuoteURL(QB, DB, RepURL);
TStrV QuoteV;
RepStr.SplitOnStr(" ", QuoteV);
TSecTm First, Last, Peak;
DCluster::GetFMP(FreqV, First, Last, Peak, RepStr);
Response = C.GetId().GetStr() + "\t" + Start.GetDtYmdStr() + "\t" + End + "\t";
Response += C.GetNumUniqueQuotes().GetStr() + "\t" + C.GetNumQuotes().GetStr() + "\t";
Response += NumPeaks.GetStr() + "\t" + TInt(QuoteV.Len()).GetStr() + "\t";
Response += RepStr + "\t" + RepURL + "\t";
Response += First.GetYmdTmStr() + "\t" + Last.GetYmdTmStr() + "\t" + Peak.GetYmdTmStr() + "\t";
Response += TBool::GetStr(TBool(C.IsArchived())) + "\t" + C.GetDiscardState().GetStr();
return Response;
}
// Function to read in a table of edges
PTable AddEdgeTable(TTableContext& Context) {
char FileName[200];
int ColCnt;
int Reverse;
printf("Adding Edge Table\n");
printf("Enter filename, number of columns (>= 2), and whether reverse? (reverse = 1, not reverse = 0\n");
scanf("%s %d %d", FileName, &ColCnt, &Reverse);
Schema EdgeScm;
if (Reverse == 1) {
EdgeScm.Add(TPair<TStr, TAttrType>("DstID", atStr));
EdgeScm.Add(TPair<TStr, TAttrType>("SrcID", atStr));
}
else {
EdgeScm.Add(TPair<TStr, TAttrType>("SrcID", atStr));
EdgeScm.Add(TPair<TStr, TAttrType>("DstID", atStr));
}
for (TInt i = 1; i < ColCnt-1; i++) {
TStr ColName = "Attribute" + i.GetStr();
EdgeScm.Add(TPair<TStr, TAttrType>(ColName, atStr));
}
TStr FName(FileName);
PTable T = TTable::LoadSS(EdgeScm, FName, Context);
return T;
}
void LogOutput::LogValue(const TStr Key, TInt Value) {
OutputValues.AddDat(Key, Value.GetStr());
}
TStr to_string () {
if (previous == NULL) return id.GetStr() + ", NULL, " + depth.GetStr ();
else return id.GetStr() + ", " + previous->id.GetStr() + ", "
+ depth.GetStr ();
}
void LSH::MinHash(TQuoteBase *QB, THashSet<TMd5Sig>& Shingles,
TVec<THash<TMd5Sig, TIntSet> >& SignatureBandBuckets) {
Err("Creating buckets...\n");
THash < TMd5Sig, TIntV > Signatures;
ComputeSignatures(Shingles, Signatures, NumBands * BandSize);
// bucket creation
for (int i = 0; i < NumBands; ++i) {
SignatureBandBuckets.Add(THash<TMd5Sig, TIntSet>());
}
// bucket filling
int NumShingles = Shingles.Len();
THash<TInt, TQuote> Quotes;
QB->GetIdToTQuotes(Quotes);
THash<TInt, TQuote>::TIter CurI = Quotes.BegI();
THash<TInt, TQuote>::TIter EndI = Quotes.EndI();
TQuote Q; // SKYFALL
for (; CurI < EndI; CurI++) {
Q = CurI.GetDat();
TStrV Content;
Q.GetParsedContent(Content);
TInt Id = Q.GetId();
// signature for quote
int ContentLen = Content.Len();
TVec < TIntV > Signature;
for (int i = 0; i < ContentLen; i++) {
const TMd5Sig ShingleMd5(Content[i]);
Signature.Add(Signatures.GetDat(ShingleMd5));
}
// place in bucket
if (ContentLen < WordWindow) {
for (int i = 0; i < NumBands; ++i) {
TStr Sig;
for (int j = 0; j < BandSize; ++j) {
int CurSig = i * BandSize + j;
TInt min = NumShingles;
for (int k = 0; k < ContentLen; k++) {
if (Signature[k][CurSig] < min) {
min = Signature[k][CurSig];
}
}
Sig += min.GetStr() + "-";
}
//Err(Sig.CStr());
const TMd5Sig SigMd5(Sig);
TIntSet Bucket;
SignatureBandBuckets[i].IsKeyGetDat(SigMd5, Bucket);
Bucket.AddKey(Id);
SignatureBandBuckets[i].AddDat(SigMd5, Bucket);
}
} else {
}
}
Err("Minhash step complete!\n");
}