void TTransCorpus::SaveSgm(const TStr& OutOrgFNm,
const TStr& OutTransFNm, const TStr& OutRefTransFNm) {
// open files
PSOut OrgSOut = !OutOrgFNm.Empty() ? TFOut::New(OutOrgFNm) : NULL;
PSOut TransSOut = !OutTransFNm.Empty() ? TFOut::New(OutTransFNm) : NULL;
PSOut RefTransSOut = !OutRefTransFNm.Empty() ? TFOut::New(OutRefTransFNm) : NULL;
// check which are given
const bool IsOrgP = !OrgSOut.Empty() && IsOrg();
const bool IsTransP = !TransSOut.Empty() && IsTrans();
const bool IsRefTransP = !RefTransSOut.Empty() && IsRefTrans();
// prepare headers
if (IsOrgP) {
OrgSOut->PutStrLn("<srcset setid=\"tmp\" srclang=\"source\" trglang=\"target\">");
OrgSOut->PutStrLn("<DOC docid=\"tmpdoc\">");
}
if (IsTransP) {
TransSOut->PutStrLn("<tstset setid=\"tmp\" srclang=\"source\" trglang=\"target\">");
TransSOut->PutStrLn("<DOC docid=\"tmpdoc\" sysid=\"trans\">");
}
if (IsRefTransP) {
RefTransSOut->PutStrLn("<refset setid=\"tmp\" srclang=\"source\" trglang=\"target\">");
RefTransSOut->PutStrLn("<DOC docid=\"tmpdoc\" sysid=\"ref\">");
}
// output sentences
TIntV SentIdV; GetSentIdV(SentIdV);
for (int SentIdN = 0; SentIdN < SentIdV.Len(); SentIdN++) {
const int SentId = SentIdV[SentIdN];
if (IsOrgP) {
OrgSOut->PutStrLn(TStr::Fmt(
"<p><seg id=\"%d\">%s</seg></p>",
SentId, GetOrgStr(SentId).CStr()));
}
if (IsTransP) {
TransSOut->PutStrLn(TStr::Fmt(
"<p><seg id=\"%d\">%s</seg></p>",
SentId, GetTransStr(SentId).CStr()));
}
if (IsRefTransP) {
RefTransSOut->PutStrLn(TStr::Fmt(
"<p><seg id=\"%d\">%s</seg></p>",
SentId, GetRefTransStrV(SentId)[0].CStr()));
}
}
// prepare footers
if (IsOrgP) {
OrgSOut->PutStrLn("</DOC>");
OrgSOut->PutStrLn("</srcset>");
}
if (IsTransP) {
TransSOut->PutStrLn("</DOC>");
TransSOut->PutStrLn("</tstset>");
}
if (IsRefTransP) {
RefTransSOut->PutStrLn("</DOC>");
RefTransSOut->PutStrLn("</refset>");
}
}
void TTransCorpus::SaveTxt(const TStr& OutFBase, const TStr& OutOrgFNm,
const TStr& OutTransFNm, const TStr& OutRefTransFNm, TStrV& OrgFNmV,
TStrV& TransFNmV, TStrV& RefTransFNmV, const int& LinesPerFile) {
// prepare filenames
OrgFNmV.Clr(); TransFNmV.Clr(); RefTransFNmV.Clr();
if (!OutOrgFNm.Empty()) { OrgFNmV.Add(GetOutFNm(OutFBase, 0, LinesPerFile, OutOrgFNm)); }
if (!OutTransFNm.Empty()) { TransFNmV.Add(GetOutFNm(OutFBase, 0, LinesPerFile, OutTransFNm)); }
if (!OutRefTransFNm.Empty()) { RefTransFNmV.Add(GetOutFNm(OutFBase, 0, LinesPerFile, OutRefTransFNm)); }
// open files
PSOut OrgSOut = !OutOrgFNm.Empty() ? TFOut::New(OrgFNmV.Last()) : PSOut();
PSOut TransSOut = !OutTransFNm.Empty() ? TFOut::New(TransFNmV.Last()) : PSOut();
PSOut RefTransSOut = !OutRefTransFNm.Empty() ? TFOut::New(RefTransFNmV.Last()) : PSOut();
// check which are given
const bool IsOrgP = !OrgSOut.Empty() && IsOrg();
const bool IsTransP = !TransSOut.Empty() && IsTrans();
const bool IsRefTransP = !RefTransSOut.Empty() && IsRefTrans();
// print warnings
if (!IsOrgP) { printf("No original sentences!\n"); }
if (!IsTransP) { printf("No machine translation sentences!\n"); }
if (!IsRefTransP) { printf("No reference translation sentences!\n"); }
// go over all the sentences and store them in the file
TIntV SentIdV; GetSentIdV(SentIdV);
for (int SentIdN = 0; SentIdN < SentIdV.Len(); SentIdN++) {
const int SentId = SentIdV[SentIdN];
if (IsOrgP) { OrgSOut->PutStrLn(GetOrgStr(SentId)); }
if (IsTransP) { TransSOut->PutStrLn(GetTransStr(SentId)); }
if (IsRefTransP) { RefTransSOut->PutStrLn(GetRefTransStrV(SentId)[0]); }
// should we break and go to next file?
if ((LinesPerFile!=-1) && (SentIdN>0) && (SentIdN % LinesPerFile == 0)) {
// prepare filenames
if (!OutOrgFNm.Empty()) { OrgFNmV.Add(GetOutFNm(OutFBase, SentIdN, LinesPerFile, OutOrgFNm)); }
if (!OutTransFNm.Empty()) { TransFNmV.Add(GetOutFNm(OutFBase, SentIdN, LinesPerFile, OutTransFNm)); }
if (!OutRefTransFNm.Empty()) { RefTransFNmV.Add(GetOutFNm(OutFBase, SentIdN, LinesPerFile, OutRefTransFNm)); }
// open next files files
OrgSOut = !OutOrgFNm.Empty() ? TFOut::New(OrgFNmV.Last()) : PSOut();
TransSOut = !OutTransFNm.Empty() ? TFOut::New(TransFNmV.Last()) : PSOut();
RefTransSOut = !OutRefTransFNm.Empty() ? TFOut::New(RefTransFNmV.Last()) : PSOut();
}
}
}
