void TRSet::SaveXml(const TStr& FNm){
TFOut FOut(FNm); FILE* fOut=FOut.GetFileId();
fprintf(fOut, "<RSet>\n");
// fixed fields
if (!GetUrlStr().Empty()){
fprintf(fOut, " <Url>%s</Url>\n", TXmlLx::GetXmlStrFromPlainStr(GetUrlStr()).CStr());}
if (!GetNextUrlStr().Empty()){
fprintf(fOut, " <NextUrl>%s</NextUrl>\n", TXmlLx::GetXmlStrFromPlainStr(GetNextUrlStr()).CStr());}
if (!GetQueryStr().Empty()){
fprintf(fOut, " <Query>%s</Query>\n", TXmlLx::GetXmlStrFromPlainStr(GetQueryStr()).CStr());}
if (GetAllHits()!=-1){
fprintf(fOut, " <AllHits>%d</AllHits>\n", GetAllHits());}
// hits
fprintf(fOut, " <Hits Size=\"%d\">\n", GetHits());
for (int HitN=0; HitN<GetHits(); HitN++){
TStr HitUrlStr; TStr HitTitleStr; TStr HitSrcNm; TStr HitCtxStr;
GetHit(HitN, HitUrlStr, HitTitleStr, HitSrcNm, HitCtxStr);
fprintf(fOut, " <Hit Num=\"%d\">\n", 1+HitN);
fprintf(fOut, " <Url>%s</Url>\n", TXmlLx::GetXmlStrFromPlainStr(HitUrlStr).CStr());
fprintf(fOut, " <Title>%s</Title>\n", TXmlLx::GetXmlStrFromPlainStr(HitTitleStr).CStr());
if (!HitSrcNm.Empty()){
fprintf(fOut, " <Source>%s</Source>\n", TXmlLx::GetXmlStrFromPlainStr(HitSrcNm).CStr());}
fprintf(fOut, " <Snippet>%s</Snippet>\n", TXmlLx::GetXmlStrFromPlainStr(HitCtxStr).CStr());
fprintf(fOut, " </Hit>\n");
}
fprintf(fOut, " </Hits>\n");
fprintf(fOut, "</RSet>");
}
PBowDocBs TRSet::GetBowDocBs(
const TStr& SwSetTypeNm, const TStr& StemmerTypeNm,
const int& MxNGramLen, const int& MnNGramFq) const {
// prepare stop-words
PSwSet SwSet=TSwSet::GetSwSet(SwSetTypeNm);
// prepare stemmer
PStemmer Stemmer=TStemmer::GetStemmer(StemmerTypeNm);
// prepare n-grams
TStrV HtmlStrV(GetHits(), 0);
for (int HitN=0; HitN<GetHits(); HitN++){
TStr TitleStr=GetHitTitleStr(HitN);
TStr CtxStr=GetHitCtxStr(HitN);
TStr HtmlStr=TitleStr+". "+CtxStr;
HtmlStrV.Add(HtmlStr);
}
PNGramBs NGramBs=TNGramBs::GetNGramBsFromHtmlStrV(
HtmlStrV, MxNGramLen, MnNGramFq, SwSet, Stemmer);
// create document-base
printf("Create Bag-Of-Words Base ... ");
PBowDocBs BowDocBs=TBowDocBs::New();
BowDocBs->PutNGramBs(NGramBs);
for (int HitN=0; HitN<GetHits(); HitN++){
BowDocBs->AddHtmlDoc(GetHitTitleStr(HitN), TStrV(), HtmlStrV[HitN], true);
}
BowDocBs->AssertOk();
printf("Done.\n");
// return bag-of-words
return BowDocBs;
}
void
CMSHit::CountHitsByType(int& Independent,
int& Dependent,
double Threshold,
int MaxI) const
{
int i;
Independent = Dependent = 0;
int LastIon(-1), LastCharge(-1), LastNumber(-1);
for (i = 0; i < GetHits(); i++) {
if(GetHitInfo(i).GetIntensity() > MaxI*Threshold) {
if(GetHitInfo(i).GetIonSeries() == LastIon && GetHitInfo(i).GetCharge() == LastCharge) {
if(LastNumber + 1 == GetHitInfo(i).GetNumber())
Dependent++;
else
Independent++;
}
else
Independent++;
LastIon = GetHitInfo(i).GetIonSeries();
LastCharge = GetHitInfo(i).GetCharge();
LastNumber = GetHitInfo(i).GetNumber();
}
}
}
// return number of hits above threshold
int CMSHit::CountHits(double Threshold, int MaxI)
{
int i, retval(0);
for(i = 0; i < GetHits(); i++)
if(SetHitInfo(i).GetIntensity() > MaxI*Threshold)
retval++;
return retval;
}
void
PoorManWindow::UpdateHitsLabel()
{
if (Lock()) {
sprintf(fHitsLabel, B_TRANSLATE("Hits: %lu"), GetHits());
fHitsView->SetText(fHitsLabel);
Unlock();
}
}
void TGgSchRSet::Merge(const PGgSchRSet& RSet){
if (RSet.Empty()){return;}
// create hash table of existing urls
TStrH TitleStrH(GetHits());
for (int HitN=0; HitN<GetHits(); HitN++){
TitleStrH.AddKey(GetHit(HitN)->TitleStr);
}
// merge hits
for (int HitN=0; HitN<RSet->GetHits(); HitN++){
PGgSchRef Ref=RSet->GetHit(HitN);
if (!TitleStrH.IsKey(Ref->TitleStr)){
AddHit(Ref);
TitleStrH.AddKey(Ref->TitleStr);
}
}
// reset fixed fields
PutNextUrlStr("");
PutAllHits(-1);
}
void TRSet::Merge(const PRSet& RSet){
if (RSet.Empty()){return;}
// create hash table of existing urls
TStrH UrlStrH(GetHits());
for (int HitN=0; HitN<GetHits(); HitN++){
UrlStrH.AddKey(GetHitUrlStr(HitN));
}
// merge hits
for (int HitN=0; HitN<RSet->GetHits(); HitN++){
TStr HitUrlStr; TStr HitTitleStr; TStr HitSrcNm; TStr HitCtxStr;
RSet->GetHit(HitN, HitUrlStr, HitTitleStr, HitSrcNm, HitCtxStr);
if (!UrlStrH.IsKey(HitUrlStr)){
AddHit(HitUrlStr, HitTitleStr, HitSrcNm, HitCtxStr);
UrlStrH.AddKey(HitUrlStr);
}
}
// reset fixed fields
PutNextUrlStr("");
PutAllHits(-1);
}
int TRSet::GetHitN(const TStr& UrlStr, const bool& LcP) const {
int Hits=GetHits();
TStr LcUrlStr=UrlStr.GetLc();
for (int HitN=0; HitN<Hits; HitN++){
if (LcP){
if (GetHitUrlStr(HitN).GetLc()==LcUrlStr){return HitN;}
} else {
if (GetHitUrlStr(HitN)==UrlStr){return HitN;}
}
}
return -1;
}
void TGgSchRSet::SaveXml(const TStr& FNm){
TFOut FOut(FNm); FILE* fOut=FOut.GetFileId();
fprintf(fOut, "<RSet>\n");
// fixed fields
if (!GetUrlStr().Empty()){
fprintf(fOut, " <Url>%s</Url>\n", TXmlLx::GetXmlStrFromPlainStr(GetUrlStr()).CStr());}
if (!GetNextUrlStr().Empty()){
fprintf(fOut, " <NextUrl>%s</NextUrl>\n", TXmlLx::GetXmlStrFromPlainStr(GetNextUrlStr()).CStr());}
if (!GetQueryStr().Empty()){
fprintf(fOut, " <Query>%s</Query>\n", TXmlLx::GetXmlStrFromPlainStr(GetQueryStr()).CStr());}
if (GetAllHits()!=-1){
fprintf(fOut, " <AllHits>%d</AllHits>\n", GetAllHits());}
// hits
fprintf(fOut, " <Hits Size=\"%d\">\n", GetHits());
for (int HitN=0; HitN<GetHits(); HitN++){
PGgSchRef Ref=GetHit(HitN);
Ref->SaveXml(fOut, 1+HitN);
}
fprintf(fOut, " </Hits>\n");
fprintf(fOut, "</RSet>");
}
// for poisson test
// return number of hits above threshold scaled by m/z positions
int CMSHit::CountHits(double Threshold, int MaxI, int High)
{
int i ;
float retval(0);
for(i = 0; i < GetHits(); i++)
if(SetHitInfo(i).GetIntensity() > MaxI*Threshold) {
if (SetHitInfo(i).GetMZ() > High/2)
retval += 0.5 + 2.0*(High - SetHitInfo(i).GetMZ())/(float)High;
else
retval += 1.5 - 2.0*SetHitInfo(i).GetMZ()/(float)High;
}
return (int)(retval+0.5);
}
PBowDocBs TGgSchRSet::GetBowDocBs() const {
// prepare stop-words, stemming
PSwSet SwSet=TSwSet::GetSwSet(swstEn523);
PStemmer Stemmer=TStemmer::New(stmtPorter, true);
// prepare n-grams
TStrV HtmlStrV(GetHits(), 0);
for (int HitN=0; HitN<GetHits(); HitN++){
TStr HtmlStr=GetHit(HitN)->TitleStr;
HtmlStrV.Add(HtmlStr);
}
PNGramBs NGramBs=TNGramBs::GetNGramBsFromHtmlStrV(
HtmlStrV, 3, 3, SwSet, Stemmer);
// create document-base
printf("Create Bag-Of-Words Base ... ");
PBowDocBs BowDocBs=TBowDocBs::New();
BowDocBs->PutNGramBs(NGramBs);
for (int HitN=0; HitN<GetHits(); HitN++){
BowDocBs->AddHtmlDoc(TInt::GetStr(HitN), TStrV(), HtmlStrV[HitN], true);
}
BowDocBs->AssertOk();
printf("Done.\n");
// return bag-of-words
return BowDocBs;
}
int URLData::TopTen()
{
//add to a list, all the website visited and their hits
for( auto it = dataMap.begin(); it != dataMap.end(); ++it )
{
addTopTen(it->first,(it->second)->GetHits());
}
//sort this list according to their hits number (less first)
listTopTen.sort();
//displays the last 10 elements of this list
auto itend = listTopTen.end();
for (int i = 0; i < TOP_DISP; i++)
{
itend--;
//displays each ten first website with this syntax : "Website : X" (X, number of hits)
std::cout << itend->GetName() << " : " << itend->GetHits() << std::endl;
}
return 0;
}
