void TFtrGenSparseNumeric::Split(
const TStr& Str, int& Id, TStr& Val) const {
if (!Str.IsChIn(':')) { TExcept::Throw("Wrong sparse numeric '" + Str + "'!"); }
//split
TStr IdStr; Str.SplitOnCh(IdStr, ':', Val);
// parse id
if (!IdStr.IsInt(Id)) { TExcept::Throw("Wrong sparse numeric '" + Str + "'!"); }
}
int TSAppSrvFun::GetFldInt(const TStrKdV& FldNmValPrV, const TStr& FldNm, const int& DefInt) {
if (!IsFldNm(FldNmValPrV, FldNm)) { return DefInt; }
TStr IntStr = GetFldVal(FldNmValPrV, FldNm, "");
EAssertR(IntStr.IsInt(), "Parameter '" + FldNm + "' not a number");
return IntStr.GetInt();
}
int TSAppSrvFun::GetFldInt(const TStrKdV& FldNmValPrV, const TStr& FldNm) {
EAssertR(IsFldNm(FldNmValPrV, FldNm), "Missing parameter '" + FldNm + "'");
TStr IntStr = GetFldVal(FldNmValPrV, FldNm, "");
EAssertR(IntStr.IsInt(), "Parameter '" + FldNm + "' not a number");
return IntStr.GetInt();
}
void TWebTxtBsSrv::OnHttpRq(const int& SockId, const PHttpRq& HttpRq){
// request parameters
TStr RqContTypeStr=THttp::TextHtmlFldVal;
PUrlEnv UrlEnv;
TStr QueryStr; TStr EQueryStr;
TStr HitSetStr; TStr AcceptStr;
// prepare & extract search-environment
if (HttpRq->IsOk()){
// prepare search-environment
PUrl Url=HttpRq->GetUrl();
UrlEnv=HttpRq->GetUrlEnv();
// if empty search-environment and url-path is not empty
if (UrlEnv->Empty()&&
(Url->GetPathSegs()>0)&&(!Url->GetPathSeg(0).Empty())){
// get document name
TStr DocNm=Url->GetPathSeg(Url->GetPathSegs()-1);
if (WebTxtBs->GetTxtBs()->IsDoc(DocNm)){
// document exists in text-base
TStr DocStr=WebTxtBs->GetTxtBs()->GetDocStr(DocNm);
PSIn HttpBodySIn=TMIn::New(DocStr);
PHttpResp HttpResp=
THttpResp::New(THttp::OkStatusCd, RqContTypeStr, false, HttpBodySIn);
SendHttpResp(SockId, HttpResp);
} else {
// ordinary http request
TWebSrv::OnHttpRq(SockId, HttpRq);
}
// end if no search request
return;
}
// extract fields from search-environment
QueryStr=UrlEnv->GetVal(QueryUrlFldNm).GetTrunc();
EQueryStr=THtmlLx::GetEscapedStr(QueryStr);
HitSetStr=UrlEnv->GetVal(HitSetUrlFldNm).GetTrunc();
AcceptStr=UrlEnv->GetVal(AcceptUrlFldNm).GetTrunc();
if (AcceptStr.Empty()){RqContTypeStr=GetRqContType(HttpRq);}
else {RqContTypeStr=AcceptStr;}
}
// hit-set
int HitSetN=1;
HitSetStr.IsInt(true, 1, TInt::Mx, HitSetN);
int HitSetDocs=GetVarVal(RqContTypeStr, "HitSetDocs").GetInt();
int StrHitSets=GetVarVal(RqContTypeStr, "StrHitSets").GetInt();
// output buffer
TChA OutChA(10000);
// header
TStr HdTpl=GetTplVal(RqContTypeStr, "Header");
HdTpl.ChangeStrAll(QueryMacro, EQueryStr);
OutChA+=HdTpl;
// html body
if (HttpRq->IsOk()){
if (!QueryStr.Empty()){
// execute query
PTxtBsRes TxtBsRes=WebTxtBs->Search(QueryStr);
TStr EWixExpStr=THtmlLx::GetEscapedStr(TxtBsRes->GetWixExpStr());
// log string
TChA QueryInfoChA;
QueryInfoChA+="Query: "+QueryStr;
//QueryInfoChA+=" ["+GetPeerNm(SockId)+"]";
QueryInfoChA+=" ["+TSecTm::GetCurTm().GetStr()+"]";
TNotify::OnNotify(Notify, ntInfo, QueryInfoChA);
SLog->PutStr(QueryInfoChA); SLog->PutLn(); SLog->Flush();
// query-results processing
if (TxtBsRes->IsOk()){
// result header
TStr ResultHdTpl=GetTplVal(RqContTypeStr, "ResultHd");
ResultHdTpl.ChangeStrAll(QueryMacro, EWixExpStr);
ResultHdTpl.ChangeStrAll(HitsMacro, TInt::GetStr(TxtBsRes->GetDocs()));
OutChA+=ResultHdTpl;
// result records
int MnDocN; int MxDocN;
TxtBsRes->GetHitSetMnMxDocN(HitSetN, HitSetDocs, MnDocN, MxDocN);
for (int DocN=MnDocN; DocN<=MxDocN; DocN++){
// get result document data
int MxDocTitleLen=GetVarVal(RqContTypeStr, "MxDocTitleLen").GetInt();
int MxDocCtxLen=GetVarVal(RqContTypeStr, "MxDocCtxLen").GetInt();
TStr DocNm; TStr DocTitleStr; TStr DocStr; TStr DocCtxStr;
TxtBsRes->GetDocInfo(DocN, MxDocTitleLen, MxDocCtxLen,
DocNm, DocTitleStr, DocStr, DocCtxStr);
if (DocTitleStr.Empty()){DocTitleStr=DocNm;}
// result record
TStr ResultRecTpl=GetTplVal(RqContTypeStr, "ResultRec");
ResultRecTpl.ChangeStrAll(HitNumMacro, TInt::GetStr(DocN+1));
ResultRecTpl.ChangeStrAll(DocAddrMacro, DocNm);
ResultRecTpl.ChangeStrAll(DocTitleMacro, DocTitleStr);
ResultRecTpl.ChangeStrAll(DocCtxMacro, DocCtxStr);
OutChA+=ResultRecTpl;
}
// result footer
TStr ResultFtTpl=GetTplVal(RqContTypeStr, "ResultFt");
OutChA+=ResultFtTpl;
// hit-set
AddHitSetChA(TxtBsRes, RqContTypeStr, HitSetN, HitSetDocs, StrHitSets,
UrlEnv, OutChA);
} else {
// bad query
TStr BadQueryTpl=GetTplVal(RqContTypeStr, "BadQuery");
BadQueryTpl.ChangeStrAll(QueryMacro, EWixExpStr);
OutChA+=BadQueryTpl;
}
}
} else {
// bad http-request
TStr BadHttpRqTpl=GetTplVal(RqContTypeStr, "BadHttpRq");
OutChA+=BadHttpRqTpl;
}
// footer
TStr FtTpl=GetTplVal(RqContTypeStr, "Footer");
FtTpl.ChangeStrAll(QueryMacro, EQueryStr);
OutChA+=FtTpl;
// construct & send response
PSIn HttpBodySIn=TMIn::New(OutChA);
PHttpResp HttpResp=
THttpResp::New(THttp::OkStatusCd, RqContTypeStr, false, HttpBodySIn);
SendHttpResp(SockId, HttpResp);
}
// load from allactors.zip that was prepared by Brad Malin in 2005
PImdbNet TImdbNet::LoadTxt(const TStr& ActorFNm) {
PImdbNet Net = TImdbNet::New();
TStrV ColV;
char line [2024];
int NLines=0, DupEdge=0, Year, Position, ActorNId, MovieNId;
TIntH ActorNIdH;
THash<TIntPr, TInt> MovieNIdH;
FILE *F = fopen(ActorFNm.CStr(), "rt"); fgets(line, 2024, F);
while (! feof(F)) {
memset(line, 0, 2024);
fgets(line, 2024, F);
if (strlen(line) == 0) break;
TStr(line).SplitOnAllCh('|', ColV, false); IAssert(ColV.Len() == 7);
const int NameStrId = Net->AddStr(ColV[0].GetTrunc().GetLc()+" "+ColV[1].GetTrunc().GetLc());
const int MovieStrId = Net->AddStr(ColV[2].GetTrunc().GetLc());
TStr YearStr = ColV[3].GetTrunc();
if (YearStr.Len() > 4) YearStr = YearStr.GetSubStr(0, 3);
Year = 1; YearStr.IsInt(Year);
const TMovieTy MovieTy = TImdbNet::GetMovieTy(ColV[4]);
Position = TInt::Mx; ColV[5].GetTrunc().IsInt(Position);
IAssert(ColV[6].GetTrunc()[0] == 'M' || ColV[6].GetTrunc()[0]=='F');
const bool IsMale = ColV[6].GetTrunc()[0] == 'M';
// create nodes
if (ActorNIdH.IsKey(NameStrId)) {
ActorNId = ActorNIdH.GetDat(NameStrId); }
else {
ActorNId = Net->AddNode(-1, TImdbNode(NameStrId, Year, Position, IsMale));
ActorNIdH.AddDat(NameStrId, ActorNId);
}
if (MovieNIdH.IsKey(TIntPr(MovieStrId, Year))) {
MovieNId = MovieNIdH.GetDat(TIntPr(MovieStrId, Year)); }
else {
MovieNId = Net->AddNode(-1, TImdbNode(NameStrId, Year, MovieTy));
MovieNIdH.AddDat(TIntPr(MovieStrId, Year), MovieNId);
}
if (! Net->IsEdge(ActorNId, MovieNId)) {
Net->AddEdge(ActorNId, MovieNId); }
else { DupEdge++; }
if (++NLines % 100000 == 0) printf("\r %dk ", NLines/1000);
}
fclose(F);
printf("duplicate edges: %d\n", DupEdge);
printf("nodes: %d\n", Net->GetNodes());
printf("edges: %d\n", Net->GetEdges());
printf("actors: %d\n", ActorNIdH.Len());
printf("movies: %d\n", MovieNIdH.Len());
// set the actor year to the year of his first movie
int NUpdates=0;
for (TNet::TNodeI NI = Net->BegNI(); NI < Net->EndNI(); NI++) {
if (NI().IsActor()) {
int MinYear = NI().GetYear();
for (int e = 0; e < NI.GetOutDeg(); e++) {
const TImdbNode& NodeDat = Net->GetNDat(NI.GetOutNId(e));
if (NodeDat.IsMovie()) MinYear = TMath::Mn(MinYear, NodeDat.GetYear());
}
if (NI().Year != MinYear) NUpdates++;
NI().Year = MinYear;
}
}
printf("updated actor times: %d\n", NUpdates);
return Net;
}
