// two strings that are canonically equivalent must test
// equal under a canonical caseless match
// see UAX #21 Case Mappings and Jitterbug 2021 and
// Unicode Technical Committee meeting consensus 92-C31
private void compare(String s1, String s2)
{
if (s1.Length == 1 && s2.Length == 1)
{
if (Normalizer.Compare(UTF16.CharAt(s1, 0), UTF16.CharAt(s2, 0), Normalizer.COMPARE_IGNORE_CASE) != 0)
{
Errln("Normalizer.compare(int,int) failed for s1: "
+ Utility.Hex(s1) + " s2: " + Utility.Hex(s2));
}
}
if (s1.Length == 1 && s2.Length > 1)
{
if (Normalizer.Compare(UTF16.CharAt(s1, 0), s2, Normalizer.COMPARE_IGNORE_CASE) != 0)
{
Errln("Normalizer.compare(int,String) failed for s1: "
+ Utility.Hex(s1) + " s2: " + Utility.Hex(s2));
}
}
if (s1.Length > 1 && s2.Length > 1)
{
// TODO: Re-enable this tests after UTC fixes UAX 21
if (Normalizer.Compare(s1.ToCharArray(), s2.ToCharArray(), Normalizer.COMPARE_IGNORE_CASE) != 0)
{
Errln("Normalizer.compare(char[],char[]) failed for s1: "
+ Utility.Hex(s1) + " s2: " + Utility.Hex(s2));
}
}
}
// 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 HashSet <string>();
ISet <string> basic = GetEquivalents2(segment);
ISet <string> permutations = new 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);
}
/// <summary>
/// See if the decomposition of cp2 is at segment starting at <paramref name="segmentPos"/>
/// (with canonical rearrangment!).
/// If so, take the remainder, and return the equivalents.
/// </summary>
/// <param name="comp"></param>
/// <param name="segment"></param>
/// <param name="segmentPos"></param>
/// <param name="buf"></param>
/// <returns></returns>
private ISet <string> Extract(int comp, string segment, int segmentPos, StringBuffer buf)
{
if (PROGRESS)
{
Console.Out.WriteLine(" extract: " + Utility.Hex(UTF16.ValueOf(comp))
+ ", " + Utility.Hex(segment.Substring(segmentPos)));
}
string decomp = nfcImpl.GetDecomposition(comp);
if (decomp == null)
{
decomp = UTF16.ValueOf(comp);
}
// See if it matches the start of segment (at segmentPos)
bool ok = false;
int cp;
int decompPos = 0;
int decompCp = UTF16.CharAt(decomp, 0);
decompPos += UTF16.GetCharCount(decompCp); // adjust position to skip first char
//int decompClass = getClass(decompCp);
buf.Length = 0; // initialize working buffer, shared among callees
for (int i = segmentPos; i < segment.Length; i += UTF16.GetCharCount(cp))
{
cp = UTF16.CharAt(segment, i);
if (cp == decompCp)
{ // if equal, eat another cp from decomp
if (PROGRESS)
{
Console.Out.WriteLine(" matches: " + Utility.Hex(UTF16.ValueOf(cp)));
}
if (decompPos == decomp.Length)
{ // done, have all decomp characters!
buf.Append(segment.Substring(i + UTF16.GetCharCount(cp))); // add remaining segment chars
ok = true;
break;
}
decompCp = UTF16.CharAt(decomp, decompPos);
decompPos += UTF16.GetCharCount(decompCp);
//decompClass = getClass(decompCp);
}
else
{
if (PROGRESS)
{
Console.Out.WriteLine(" buffer: " + Utility.Hex(UTF16.ValueOf(cp)));
}
// brute force approach
UTF16.Append(buf, cp);
/* TODO: optimize
* // since we know that the classes are monotonically increasing, after zero
* // e.g. 0 5 7 9 0 3
* // we can do an optimization
* // there are only a few cases that work: zero, less, same, greater
* // if both classes are the same, we fail
* // if the decomp class < the segment class, we fail
*
* segClass = getClass(cp);
* if (decompClass <= segClass) return null;
*/
}
}
if (!ok)
{
return(null); // we failed, characters left over
}
if (PROGRESS)
{
Console.Out.WriteLine("Matches");
}
if (buf.Length == 0)
{
return(SET_WITH_NULL_STRING); // succeed, but no remainder
}
string remainder = buf.ToString();
// brute force approach
// to check to make sure result is canonically equivalent
/*
* String trial = Normalizer.normalize(UTF16.valueOf(comp) + remainder, Normalizer.DECOMP, 0);
* if (!segment.regionMatches(segmentPos, trial, 0, segment.length() - segmentPos)) return null;
*/
if (0 != Normalizer.Compare(UTF16.ValueOf(comp) + remainder, segment.Substring(segmentPos), 0))
{
return(null);
}
// get the remaining combinations
return(GetEquivalents2(remainder));
}
/**
* Verify the conformance of the given line of the Unicode
* normalization (UTR 15) test suite file. For each line,
* there are five columns, corresponding to field[0]..field[4].
*
* The following invariants must be true for all conformant implementations
* c2 == NFC(c1) == NFC(c2) == NFC(c3)
* c3 == NFD(c1) == NFD(c2) == NFD(c3)
* c4 == NFKC(c1) == NFKC(c2) == NFKC(c3) == NFKC(c4) == NFKC(c5)
* c5 == NFKD(c1) == NFKD(c2) == NFKD(c3) == NFKD(c4) == NFKD(c5)
*
* @param field the 5 columns
* @param line the source line from the test suite file
* @return true if the test passes
*/
private bool checkConformance(String[] field, String line, int options)
{
bool pass = true;
StringBuffer buf = new StringBuffer(); // scratch
String @out, fcd;
int i = 0;
for (i = 0; i < 5; ++i)
{
int fieldNum = i + 1;
if (i < 3)
{
pass &= checkNorm(Normalizer.NFC, options, field[i], field[1], fieldNum);
pass &= checkNorm(Normalizer.NFD, options, field[i], field[2], fieldNum);
}
pass &= checkNorm(Normalizer.NFKC, options, field[i], field[3], fieldNum);
pass &= checkNorm(Normalizer.NFKD, options, field[i], field[4], fieldNum);
cross(field[4] /*NFKD String*/, field[3] /*NFKC String*/, Normalizer.NFKC);
cross(field[3] /*NFKC String*/, field[4] /*NFKD String*/, Normalizer.NFKD);
}
compare(field[1], field[2]);
compare(field[0], field[1]);
compare(field[0], field[2]);
// test quick checks
if (NormalizerQuickCheckResult.No == Normalizer.QuickCheck(field[1], Normalizer.NFC, options))
{
Errln("Normalizer error: quickCheck(NFC(s), Normalizer.NFC) is Normalizer.NO");
pass = false;
}
if (Normalizer.NO == Normalizer.QuickCheck(field[2], Normalizer.NFD, options))
{
Errln("Normalizer error: quickCheck(NFD(s), Normalizer.NFD) is Normalizer.NO");
pass = false;
}
if (Normalizer.NO == Normalizer.QuickCheck(field[3], Normalizer.NFKC, options))
{
Errln("Normalizer error: quickCheck(NFKC(s), Normalizer.NFKC) is Normalizer.NO");
pass = false;
}
if (Normalizer.NO == Normalizer.QuickCheck(field[4], Normalizer.NFKD, options))
{
Errln("Normalizer error: quickCheck(NFKD(s), Normalizer.NFKD) is Normalizer.NO");
pass = false;
}
if (!Normalizer.IsNormalized(field[1], Normalizer.NFC, options))
{
Errln("Normalizer error: isNormalized(NFC(s), Normalizer.NFC) is false");
pass = false;
}
if (!field[0].Equals(field[1]) && Normalizer.IsNormalized(field[0], Normalizer.NFC, options))
{
Errln("Normalizer error: isNormalized(s, Normalizer.NFC) is TRUE");
pass = false;
}
if (!Normalizer.IsNormalized(field[3], Normalizer.NFKC, options))
{
Errln("Normalizer error: isNormalized(NFKC(s), Normalizer.NFKC) is false");
pass = false;
}
if (!field[0].Equals(field[3]) && Normalizer.IsNormalized(field[0], Normalizer.NFKC, options))
{
Errln("Normalizer error: isNormalized(s, Normalizer.NFKC) is TRUE");
pass = false;
}
// test api that takes a char[]
if (!Normalizer.IsNormalized(field[1].ToCharArray(), 0, field[1].Length, Normalizer.NFC, options))
{
Errln("Normalizer error: isNormalized(NFC(s), Normalizer.NFC) is false");
pass = false;
}
// test api that takes a codepoint
if (!Normalizer.IsNormalized(UTF16.CharAt(field[1], 0), Normalizer.NFC, options))
{
Errln("Normalizer error: isNormalized(NFC(s), Normalizer.NFC) is false");
pass = false;
}
// test FCD quick check and "makeFCD"
fcd = Normalizer.Normalize(field[0], Normalizer.FCD);
if (Normalizer.NO == Normalizer.QuickCheck(fcd, Normalizer.FCD, options))
{
Errln("Normalizer error: quickCheck(FCD(s), Normalizer.FCD) is Normalizer.NO");
pass = false;
}
// check FCD return length
{
char[] fcd2 = new char[fcd.Length * 2];
char[] src = field[0].ToCharArray();
int fcdLen = Normalizer.Normalize(src, 0, src.Length, fcd2, fcd.Length, fcd2.Length, Normalizer.FCD, 0);
if (fcdLen != fcd.Length)
{
Errln("makeFCD did not return the correct length");
}
}
if (Normalizer.NO == Normalizer.QuickCheck(fcd, Normalizer.FCD, options))
{
Errln("Normalizer error: quickCheck(FCD(s), Normalizer.FCD) is Normalizer.NO");
pass = false;
}
if (Normalizer.NO == Normalizer.QuickCheck(field[2], Normalizer.FCD, options))
{
Errln("Normalizer error: quickCheck(NFD(s), Normalizer.FCD) is Normalizer.NO");
pass = false;
}
if (Normalizer.NO == Normalizer.QuickCheck(field[4], Normalizer.FCD, options))
{
Errln("Normalizer error: quickCheck(NFKD(s), Normalizer.FCD) is Normalizer.NO");
pass = false;
}
@out = iterativeNorm(new StringCharacterIterator(field[0]), Normalizer.FCD, buf, +1, options);
@out = iterativeNorm(new StringCharacterIterator(field[0]), Normalizer.FCD, buf, -1, options);
@out = iterativeNorm(new StringCharacterIterator(field[2]), Normalizer.FCD, buf, +1, options);
@out = iterativeNorm(new StringCharacterIterator(field[2]), Normalizer.FCD, buf, -1, options);
@out = iterativeNorm(new StringCharacterIterator(field[4]), Normalizer.FCD, buf, +1, options);
@out = iterativeNorm(new StringCharacterIterator(field[4]), Normalizer.FCD, buf, -1, options);
@out = Normalizer.Normalize(fcd, Normalizer.NFD);
if ([email protected](field[2]))
{
Errln("Normalizer error: NFD(FCD(s))!=NFD(s)");
pass = false;
}
if (!pass)
{
Errln("FAIL: " + line);
}
if (field[0] != field[2])
{
// two strings that are canonically equivalent must test
// equal under a canonical caseless match
// see UAX #21 Case Mappings and Jitterbug 2021 and
// Unicode Technical Committee meeting consensus 92-C31
int rc;
if ((rc = Normalizer.Compare(field[0], field[2], (options << Normalizer.COMPARE_NORM_OPTIONS_SHIFT) | Normalizer.COMPARE_IGNORE_CASE)) != 0)
{
Errln("Normalizer.compare(original, NFD, case-insensitive) returned " + rc + " instead of 0 for equal");
pass = false;
}
}
return(pass);
}