public IPatternUnit WithSuffix(string suffix, bool ignoreCase = false)
{
Type |= PatternUnitType.Suffix;
Suffix = suffix;
CaseSensitive = !ignoreCase;
return(this);
}
public IPatternUnit WithPrefix(string prefix, bool ignoreCase = false)
{
Type |= PatternUnitType.Prefix;
Prefix = prefix;
CaseSensitive = !ignoreCase;
return(this);
}
public IPatternUnit WithLength(int minLength, int maxLength)
{
Type |= PatternUnitType.Length;
MinLength = minLength;
MaxLength = maxLength;
return(this);
}
public IPatternUnit WithToken(string token, bool ignoreCase = false)
{
Type |= PatternUnitType.Token;
Token = token;
CaseSensitive = !ignoreCase;
TokenHash = ignoreCase ? IgnoreCaseHash64(token.AsSpan()) : Hash64(token.AsSpan());
return(this);
}
public IPatternUnit WithTokens(IEnumerable <string> tokens, bool ignoreCase = false)
{
if (tokens.Count() == 1)
{
return(WithToken(tokens.First(), ignoreCase));
}
Type |= PatternUnitType.Set;
Set = new HashSet <string>(tokens);
CaseSensitive = !ignoreCase;
SetHashes = new HashSet <ulong>(tokens.Select(tk => ignoreCase ? IgnoreCaseHash64(tk.AsSpan()) : Hash64(tk.AsSpan())));
return(this);
}
public PatternUnit(PatternMatchingMode mode, bool optional, bool caseSensitive, PatternUnitType type, PartOfSpeech[] pos, string suffix, string prefix, string shape, string token, HashSet <string> set, string entityType, HashSet <ulong> setHashes, ulong tokenHash, PatternUnit leftSide, PatternUnit rightSide)
{
Mode = mode;
Optional = optional;
CaseSensitive = caseSensitive;
Type = type;
POS = pos;
Suffix = suffix;
Prefix = prefix;
Shape = shape?.AsSpan().Shape(false);
Token = token;
Set = set;
EntityType = entityType;
SetHashes = setHashes ?? (set is null ? null : new HashSet <ulong>(set.Select(tk => CaseSensitive ? PatternUnitPrototype.Hash64(tk.AsSpan()) : PatternUnitPrototype.IgnoreCaseHash64(tk.AsSpan()))));
TokenHash = tokenHash;
LeftSide = leftSide;
RightSide = rightSide;
}
public PatternUnit(PatternMatchingMode mode, bool optional, bool caseSensitive, PatternUnitType type, PartOfSpeech[] pos, string suffix, string prefix, string shape, string token, HashSet <string> set, string entityType, HashSet <ulong> setHashes, ulong tokenHash, PatternUnit leftSide, PatternUnit rightSide)
{
Mode = mode;
Optional = optional;
CaseSensitive = caseSensitive;
Type = type;
POS = pos;
Suffix = suffix;
Prefix = prefix;
Shape = shape?.AsSpan().Shape(false);
Token = token;
Set = set;
EntityType = entityType;
SetHashes = setHashes ?? (set is null ? null : new HashSet <ulong>(set.Select(tk => CaseSensitive ? PatternUnitPrototype.Hash64(tk.AsSpan()) : PatternUnitPrototype.IgnoreCaseHash64(tk.AsSpan()))));
TokenHash = tokenHash;
LeftSide = leftSide;
RightSide = rightSide;
_splitSuffix = Suffix?.Split(splitCharWithWhitespaces, StringSplitOptions.RemoveEmptyEntries)?.Distinct()?.ToArray();
_splitPrefix = Prefix?.Split(splitCharWithWhitespaces, StringSplitOptions.RemoveEmptyEntries)?.Distinct()?.ToArray();
_splitEntityType = EntityType is object?new HashSet <string>(EntityType.Split(splitChar, StringSplitOptions.RemoveEmptyEntries)) : null;
_splitShape = Shape is object?new HashSet <string>(Shape.Split(splitCharWithWhitespaces, StringSplitOptions.RemoveEmptyEntries)) : null;
}
public PatternUnit(IPatternUnit prototype)
{
var p = (PatternUnitPrototype)prototype;
Mode = p.Mode;
Optional = p.Optional;
CaseSensitive = p.CaseSensitive;
Type = p.Type;
POS = p.POS;
Suffix = p.Suffix;
Prefix = p.Prefix;
Shape = p.Shape;
Token = p.Token;
Set = p.Set;
EntityType = p.EntityType;
SetHashes = p.SetHashes ?? (p.Set is null ? null : new HashSet <ulong>(p.Set.Select(token => p.CaseSensitive ? PatternUnitPrototype.Hash64(token.AsSpan()) : PatternUnitPrototype.IgnoreCaseHash64(token.AsSpan()))));
TokenHash = p.TokenHash;
LeftSide = p.LeftSide is object?new PatternUnit(p.LeftSide) : null;
RightSide = p.RightSide is object?new PatternUnit(p.RightSide) : null;
ValidChars = p.ValidChars;
MinLength = p.MinLength;
MaxLength = p.MaxLength;
}
public IPatternUnit LikeEmail()
{
Type |= PatternUnitType.LikeEmail; return(this);
}
public IPatternUnit IsClosingParenthesis()
{
Type |= PatternUnitType.IsClosingParenthesis; return(this);
}
public IPatternUnit IsTitleCase()
{
Type |= PatternUnitType.IsTitleCase; return(this);
}
public IPatternUnit LikeURL()
{
Type |= PatternUnitType.LikeURL; return(this);
}
public IPatternUnit IsPunctuation()
{
Type |= PatternUnitType.IsPunctuation; return(this);
}
public IPatternUnit IsUpperCase()
{
Type |= PatternUnitType.IsUpperCase; return(this);
}
public IPatternUnit IsLetterOrDigit()
{
Type |= PatternUnitType.IsLetterOrDigit; return(this);
}
public IPatternUnit IsEmoji()
{
Type |= PatternUnitType.IsEmoji; return(this);
}
public IPatternUnit IsAlpha()
{
Type |= PatternUnitType.IsAlpha; return(this);
}
public IPatternUnit HasNumeric()
{
Type |= PatternUnitType.HasNumeric; return(this);
}
public IPatternUnit WithPOS(params PartOfSpeech[] pos)
{
Type |= pos.Length > 1 ? PatternUnitType.MultiplePOS : PatternUnitType.POS;
POS = pos;
return(this);
}
public IPatternUnit WithShape(string shape)
{
Type |= PatternUnitType.Shape;
Shape = shape.AsSpan().Shape(false);
return(this);
}
public IPatternUnit WithoutEntityType(string entityType)
{
Type |= PatternUnitType.NotEntity;
EntityType = entityType;
return(this);
}
