private void CheckTermsOrder(IndexReader r, ISet <string> allTerms, bool isTop)
{
TermsEnum terms = MultiFields.GetFields(r).GetTerms("f").GetEnumerator();
BytesRef last = new BytesRef();
ISet <string> seenTerms = new JCG.HashSet <string>();
while (terms.MoveNext())
{
BytesRef term = terms.Term;
Assert.IsTrue(last.CompareTo(term) < 0);
last.CopyBytes(term);
string s = term.Utf8ToString();
Assert.IsTrue(allTerms.Contains(s), "term " + TermDesc(s) + " was not added to index (count=" + allTerms.Count + ")");
seenTerms.Add(s);
}
if (isTop)
{
Assert.IsTrue(allTerms.SetEquals(seenTerms));
}
// Test seeking:
IEnumerator <string> it = seenTerms.GetEnumerator();
while (it.MoveNext())
{
BytesRef tr = new BytesRef(it.Current);
Assert.AreEqual(TermsEnum.SeekStatus.FOUND, terms.SeekCeil(tr), "seek failed for term=" + TermDesc(tr.Utf8ToString()));
}
}
// we have a segment, in NFD. Find all the strings that are canonically equivalent to it.
private string[] GetEquivalents(string segment)
{
ISet <string> result = new JCG.HashSet <string>();
ISet <string> basic = GetEquivalents2(segment);
ISet <string> permutations = new JCG.HashSet <string>();
// now get all the permutations
// add only the ones that are canonically equivalent
// TODO: optimize by not permuting any class zero.
using (IEnumerator <string> it = basic.GetEnumerator())
{
while (it.MoveNext())
{
string item = it.Current;
permutations.Clear();
#pragma warning disable 612, 618
Permute(item, SKIP_ZEROS, permutations);
#pragma warning restore 612, 618
using (IEnumerator <string> it2 = permutations.GetEnumerator())
{
while (it2.MoveNext())
{
string possible = it2.Current;
/*
* String attempt = Normalizer.normalize(possible, Normalizer.DECOMP, 0);
* if (attempt.equals(segment)) {
*/
if (Normalizer.Compare(possible, segment, 0) == 0)
{
if (PROGRESS)
{
Console.Out.WriteLine("Adding Permutation: " + Utility.Hex(possible));
}
result.Add(possible);
}
else
{
if (PROGRESS)
{
Console.Out.WriteLine("-Skipping Permutation: " + Utility.Hex(possible));
}
}
}
}
}
}
// convert into a String[] to clean up storage
string[] finalResult = new string[result.Count];
result.CopyTo(finalResult, 0);
return(finalResult);
}
private static ISet <string> DifFiles(string[] files1, string[] files2)
{
ISet <string> set1 = new JCG.HashSet <string>();
ISet <string> set2 = new JCG.HashSet <string>();
ISet <string> extra = new JCG.HashSet <string>();
for (int x = 0; x < files1.Length; x++)
{
set1.Add(files1[x]);
}
for (int x = 0; x < files2.Length; x++)
{
set2.Add(files2[x]);
}
IEnumerator <string> i1 = set1.GetEnumerator();
while (i1.MoveNext())
{
string o = i1.Current;
if (!set2.Contains(o))
{
extra.Add(o);
}
}
IEnumerator <string> i2 = set2.GetEnumerator();
while (i2.MoveNext())
{
string o = i2.Current;
if (!set1.Contains(o))
{
extra.Add(o);
}
}
return(extra);
}
public IEnumerator <Type> GetEnumerator()
{
return(types.GetEnumerator());
}