void LSH::GetHashedShinglesOfCluster(TQuoteBase *QuoteBase, TCluster& C,
TInt ShingleLen, THashSet<TMd5Sig>& HashedShingles) {
TIntV QuoteIds;
C.GetQuoteIds(QuoteIds);
for (int qt = 0; qt < QuoteIds.Len(); qt++) {
TQuote Q;
QuoteBase->GetQuote(QuoteIds[qt], Q);
TStr QContentStr;
Q.GetContentString(QContentStr);
TStr QContentStrNoPunc;
TStringUtil::RemovePunctuation(QContentStr, QContentStrNoPunc);
TStrV QContentV;
QContentStrNoPunc.SplitOnWs(QContentV);
for (int i = 0; i < QContentV.Len() - ShingleLen + 1; i++) {
TStr Shingle;
for (int j = 0; j < ShingleLen; j++) {
if (j > 0) {
Shingle.InsStr(Shingle.Len(), " ");
}
Shingle.InsStr(Shingle.Len(), QContentV[i + j]);
}
TMd5Sig ShingleMd5(Shingle);
HashedShingles.AddKey(ShingleMd5);
}
}
}
TEST(TStr, InsStr) {
TStr Str = "abcda";
TStr Empty = "";
Str.InsStr(2, "xk");
EXPECT_EQ(Str, "abxkcda");
Str.InsStr(2, "");
EXPECT_EQ(Str, "abxkcda");
Str.InsStr(0, "f");
EXPECT_EQ(Str, "fabxkcda");
Str.InsStr(8, "f");
EXPECT_EQ(Str, "fabxkcdaf");
dup2(2, 1); // redirect stdout to stderr (Assert emits a printf to stdout)
EXPECT_ANY_THROW(Str.InsStr(100, "kek"));
EXPECT_ANY_THROW(Str.InsStr(-100, "kek"));
}