TEST(TStr, ToLc) {
TStr Mixedcase = "AbCd";
TStr Lowercase = "abcd";
TStr Empty = "";
TStr Empty2;
Empty.ToLc();
Mixedcase.ToLc();
EXPECT_EQ(Mixedcase, Lowercase);
EXPECT_EQ(Empty, Empty2);
}
void TStrUtil::GetStdNameV(TStr AuthorNames, TStrV& StdNameV) {
AuthorNames.ChangeChAll('\n', ' ');
AuthorNames.ToLc();
// split into author names
TStrV AuthV, TmpV, Tmp2V;
// split on 'and'
AuthorNames.SplitOnStr(" and ", TmpV);
int i;
for (i = 0; i < TmpV.Len(); i++) {
TmpV[i].SplitOnAllCh(',', Tmp2V); AuthV.AddV(Tmp2V); }
// split on '&'
TmpV = AuthV; AuthV.Clr();
for (i = 0; i < TmpV.Len(); i++) {
TmpV[i].SplitOnAllCh('&', Tmp2V); AuthV.AddV(Tmp2V); }
// split on ','
TmpV = AuthV; AuthV.Clr();
for (i = 0; i < TmpV.Len(); i++) {
TmpV[i].SplitOnAllCh(',', Tmp2V); AuthV.AddV(Tmp2V); }
// split on ';'
TmpV = AuthV; AuthV.Clr();
for (i = 0; i < TmpV.Len(); i++) {
TmpV[i].SplitOnAllCh(';', Tmp2V); AuthV.AddV(Tmp2V); }
// standardize names
StdNameV.Clr();
//printf("\n*** %s\n", AuthorNames.CStr());
for (i = 0; i < AuthV.Len(); i++) {
TStr StdName = GetStdName(AuthV[i]);
if (! StdName.Empty()) {
//printf("\t%s ==> %s\n", AuthV[i].CStr(), StdName.CStr());
StdNameV.Add(StdName);
}
}
}
// <last_name>_<first name innitial>
TStr TStrUtil::GetStdName(TStr AuthorName) {
TStr StdName;
AuthorName.ToLc();
AuthorName.ChangeChAll('\n', ' ');
AuthorName.ChangeChAll('.', ' ');
// if there is a number in the name, remove it and everything after it
int i, pos = 0;
while (pos<AuthorName.Len() && (AuthorName[pos]!='#' && !TCh::IsNum(AuthorName[pos]))) {
pos++;
}
if (pos < AuthorName.Len()) {
AuthorName = AuthorName.GetSubStr(0, pos-1).ToTrunc();
}
if (AuthorName.Empty()) {
return TStr::GetNullStr();
}
// replace everything after '('
int b = AuthorName.SearchCh('(');
if (b != -1) {
AuthorName = AuthorName.GetSubStr(0, b-1).ToTrunc();
}
// skip if contains ')'
if (AuthorName .SearchCh(')')!=-1) {
return TStr::GetNullStr();
}
// skip if it is not a name
if (AuthorName .SearchStr("figures")!=-1 || AuthorName .SearchStr("macros")!=-1
|| AuthorName .SearchStr("univ")!=-1 || AuthorName .SearchStr("institute")!=-1) {
return TStr::GetNullStr();
}
// remove all non-letters (latex tags, ...)
TChA NewName;
for (i = 0; i < AuthorName.Len(); i++) {
const char Ch = AuthorName[i];
if (TCh::IsAlpha(Ch) || TCh::IsWs(Ch) || Ch=='-') {
NewName += Ch;
}
}
StdName = NewName;
StdName.ToTrunc();
TStrV AuthNmV;
StdName.SplitOnWs(AuthNmV);
// too short -- not a name
if (! AuthNmV.Empty() && AuthNmV.Last() == "jr") AuthNmV.DelLast();
if (AuthNmV.Len() < 2) return TStr::GetNullStr();
const TStr LastNm = AuthNmV.Last();
if (! TCh::IsAlpha(LastNm[0]) || LastNm.Len() == 1) return TStr::GetNullStr();
IAssert(isalpha(AuthNmV[0][0]));
return TStr::Fmt("%s_%c", LastNm.CStr(), AuthNmV[0][0]);
}