// space seprated sequence of words (includes all non-blank characters, i.e., punctuations)
TChA TStrUtil::GetCleanStr(const TChA& ChA) {
char *b = (char *) ChA.CStr();
while (*b && ! TCh::IsAlNum(*b)) {
b++;
}
if (*b == 0) {
return TChA();
}
TChA OutChA(ChA.Len());
char *e = b;
bool ws=false;
while (*e) {
while (*e && TCh::IsWs(*e)) {
e++;
ws=true;
}
if (! *e) {
break;
}
if (ws) {
OutChA.AddCh(' ');
ws=false;
}
OutChA.AddCh(*e);
e++;
}
//OutChA.ToLc();
return OutChA;
}
// space separated sequence of words, remove all punctuations, etc.
TChA TStrUtil::GetCleanWrdStr(const TChA& ChA) {
char *b = (char *) ChA.CStr();
while (*b && ! TCh::IsAlNum(*b)) {
b++;
}
if (*b == 0) {
return TChA();
}
TChA OutChA(ChA.Len());
char *e = b, tmp;
while (*e) {
b = e;
while (*e && (TCh::IsAlNum(*e) || ((*e=='\'' || *e=='-') && TCh::IsAlNum(*(e+1))))) {
e++;
}
if (b < e) {
tmp = *e;
*e=0;
OutChA += b;
OutChA.AddCh(' ');
*e = tmp;
}
while (*e && ! TCh::IsAlNum(*e)) {
e++;
}
if (! *e) {
break;
}
}
OutChA.DelLastCh();
OutChA.ToLc();
return OutChA;
}
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);
}