public static Bimachine CreateForEnglish()
{
var alphabet = Enumerable.Range(32, 95).Select(x => (char)x)
.Concat(new[] { '\t', '\n', '\v', '\f', '\r' })
.ToHashSet();
var whitespaces = new[] { ' ', '\t', '\n' };
var upperCaseLetters = Enumerable.Range(65, 27).Select(x => (char)x);
var lowerCaseLetters = Enumerable.Range(97, 27).Select(x => (char)x);
var digits = new[] { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9' };
var letters = upperCaseLetters.Concat(lowerCaseLetters);
var riseCase = alphabet
.Select(symbol =>
FstBuilder.FromWordPair(
symbol.ToString(),
char.IsLower(symbol)
? symbol.ToString().ToUpper()
: symbol.ToString()))
.Aggregate((aggr, fst) => aggr.Union(fst))
.Star();
var multiWordExprList = new[] { "AT LEAST", "IN SPITE OF", "HEAD OVER HEELS" };
var multiWordExpr =
multiWordExprList
.Select(exp => FsaBuilder.FromWord(exp))
.Aggregate((aggr, fsa) => aggr.Union(fsa));
var token =
FsaBuilder.FromSymbolSet(letters)
.Plus()
.Union(
FsaBuilder.FromSymbolSet(digits).Plus(),
riseCase.Compose(multiWordExpr.Identity()).Domain(),
FsaBuilder.FromSymbolSet(alphabet.Except(whitespaces)));
var insertLeadingNewLine =
FstBuilder.FromWordPair(string.Empty, "\n")
.Concat(FsaBuilder.FromSymbolSet(alphabet).Star().Identity());
var clearSpaces =
FsaBuilder.FromSymbolSet(whitespaces)
.Plus()
.Product(FsaBuilder.FromWord(" "))
.ToLmlRewriter(alphabet);
var markTokens =
token.Identity()
.Concat(FstBuilder.FromWordPair(string.Empty, "\n"))
.ToLmlRewriter(alphabet);
var clearLeadingSpace =
insertLeadingNewLine.Compose(
FstBuilder.FromWordPair("\n ", "\n").ToRewriter(alphabet),
insertLeadingNewLine.Inverse());
return(clearSpaces.Compose(markTokens, clearLeadingSpace).ToBimachine(alphabet));
}
public void EpsilonFreeSimpleConstructionTest()
{
// a*
var fsa = FsaBuilder.FromWord("a").Star().EpsilonFree();
Assert.DoesNotContain(fsa.Transitions, t => string.IsNullOrEmpty(t.Label));
Assert.DoesNotContain(new[] { "ca", "aaba", "b", "cc" }, fsa.Recognize);
Assert.True(new[] { "aaaa", "a", "aa", string.Empty, "aaaaaaaa" }.All(fsa.Recognize));
}
public void WordFsaBuilderTest()
{
var fsa = FsaBuilder.FromWord("abc");
Assert.Equal(4, fsa.States.Count);
Assert.False(fsa.Recognize(string.Empty));
Assert.False(fsa.Recognize("a"));
Assert.False(fsa.Recognize("abca"));
Assert.True(fsa.Recognize("abc"));
}
public void StarFsaTest()
{
var fsa = FsaBuilder.FromWord("a").Star();
Assert.Equal(3, fsa.States.Count);
Assert.Single(fsa.Initial);
Assert.Equal(2, fsa.Final.Count);
Assert.False(fsa.Recognize("ab"));
Assert.True(new[] { "aaaa", "a", "aa", string.Empty, "aaaaaaaa" }.All(fsa.Recognize));
}
public void ConcatMultipleFsaTest()
{
var fsa1 = FsaBuilder.FromWord("ab");
var fsa2 = FsaBuilder.FromWord("cde");
var fsa3 = FsaBuilder.FromWord("f").Star();
var fsa = fsa1.Concat(fsa2, fsa3);
Assert.True(fsa.Recognize("abcdef"));
Assert.True(fsa.Recognize("abcdefffffff"));
Assert.False(fsa.Recognize("abcdff"));
}
public void ComplexFsaConstructionTest1()
{
// (a|b)*c
var fsa = FsaBuilder.FromWord("a")
.Union(FsaBuilder.FromWord("b"))
.Star()
.Concat(FsaBuilder.FromWord("c"));
Assert.DoesNotContain(new[] { "ca", "aaba", string.Empty, "cc" }, fsa.Recognize);
Assert.True(new[] { "abbac", "ac", "bc", "ababbbbac", "c" }.All(fsa.Recognize));
}
public void OptionFsaTest()
{
var fsa = FsaBuilder.FromWord("ab").Optional();
Assert.Equal(4, fsa.States.Count);
Assert.Equal(2, fsa.Initial.Count);
Assert.Equal(2, fsa.Final.Count);
Assert.False(fsa.Recognize("b"));
Assert.False(fsa.Recognize("a"));
Assert.True(new[] { "ab", string.Empty }.All(fsa.Recognize));
}
public void EpsilonFreeConstructionTest()
{
// (a|b)+c
var fsa = FsaBuilder.FromWord("a")
.Union(FsaBuilder.FromWord("b"))
.Plus()
.Concat(FsaBuilder.FromWord("c"))
.EpsilonFree();
Assert.DoesNotContain(fsa.Transitions, t => string.IsNullOrEmpty(t.Label));
Assert.True(new[] { "abbac", "ac", "bc", "ababbbbac", "aac" }.All(fsa.Recognize));
Assert.DoesNotContain(new[] { "ca", "aaba", string.Empty, "cc", "c" }, fsa.Recognize);
}
public void StarFsaTest1()
{
var fsa = FsaBuilder.FromWord("abc").Star();
Assert.Equal(5, fsa.States.Count);
Assert.Single(fsa.Initial);
Assert.Equal(2, fsa.Final.Count);
Assert.False(fsa.Recognize("abcabcabcb"));
Assert.False(fsa.Recognize("ab"));
Assert.True(fsa.Recognize(string.Empty));
Assert.True(fsa.Recognize("abc"));
Assert.True(fsa.Recognize("abcabcabc"));
}
public void UnionEpsilonFsaTest()
{
var fsa1 = FsaBuilder.FromWord("abc");
var fsa2 = FsaBuilder.FromEpsilon();
var fsa = fsa1.Union(fsa2);
Assert.Equal(5, fsa.States.Count);
Assert.Equal(2, fsa.Initial.Count);
Assert.Equal(2, fsa.Final.Count);
Assert.True(fsa.Recognize(string.Empty));
Assert.False(fsa.Recognize("a"));
Assert.True(fsa.Recognize("abc"));
Assert.False(fsa.Recognize("abca"));
}
public void ComplexFsaConstructionTest()
{
// ab*c
var fsa =
FsaBuilder.FromWord("a").Concat(
FsaBuilder.FromWord("b").Star(),
FsaBuilder.FromWord("c"));
Assert.False(fsa.Recognize(string.Empty));
Assert.False(fsa.Recognize("ab"));
Assert.True(fsa.Recognize("abc"));
Assert.True(fsa.Recognize("ac"));
Assert.True(fsa.Recognize("abbbbc"));
}
public void ConcatFsaTest()
{
var fsa1 = FsaBuilder.FromWord("abc");
var fsa2 = FsaBuilder.FromWord("de");
var fsa = fsa1.Concat(fsa2);
Assert.Equal(7, fsa.States.Count);
Assert.Single(fsa.Initial);
Assert.Single(fsa.Final);
Assert.False(fsa.Recognize(string.Empty));
Assert.False(fsa.Recognize("a"));
Assert.False(fsa.Recognize("abc"));
Assert.False(fsa.Recognize("de"));
Assert.True(fsa.Recognize("abcde"));
}
public void ComplexFsaConstructionTest2()
{
// .*@.*\.com
var all = FsaBuilder.All(
Enumerable.Range(97, 27).Select(Convert.ToChar).ToHashSet());
var fsa = all
.Concat(
FsaBuilder.FromWord("@"),
all,
FsaBuilder.FromWord(".com"))
.Determinize();
Assert.DoesNotContain(new[] { "*****@*****.**", "you@@gmail.com", "*****@*****.**", "*****@*****.**" }, fsa.Recognize);
Assert.True(new[] { "*****@*****.**", "*****@*****.**", "*****@*****.**" }.All(fsa.Recognize));
}