public AbbreviationCapturer(Language language)
{
Language = language;
var commonWords = AbbreviationCapturerCommonWords.Get(language);
DiscardCommonWords = new PatternUnit(PatternUnitPrototype.Single().WithTokens(commonWords, ignoreCase: true));
Stopwords = new HashSet <ulong>(StopWords.Spacy.For(Language).Select(w => w.AsSpan().IgnoreCaseHash64()).ToArray());
}
public bool IsMatch(Span <Token> tokens, out int consumedTokens)
{
int largestMatch = -1;
for (int i = 0; i < Patterns.Count; i++)
{
var currentToken = 0;
for (int j = 0; j < Patterns[i].Length; j++)
{
PatternUnit currentPattern = Patterns[i][j];
int ct = currentToken;
bool hasMatched = false;
while (ct < tokens.Length && currentPattern.IsMatch(ref tokens[ct]))
{
ct++;
hasMatched = true;
if (currentPattern.Mode == PatternMatchingMode.Single)
{
break;
}
}
if (hasMatched)
{
currentToken = ct;
}
else
{
if (!currentPattern.Optional)
{
currentToken = int.MinValue; //Didn't match a mandatory token, so abort
break;
}
}
}
if (largestMatch < currentToken)
{
largestMatch = currentToken;
}
}
if (largestMatch > 0)
{
consumedTokens = largestMatch;
return(true);
}
else
{
consumedTokens = 0;
return(false);
}
}
private IEnumerable <string> GetTokens(PatternUnit pu)
{
if (pu.Token is object)
{
yield return(pu.Token);
}
if (pu.Set is object)
{
foreach (var t in pu.Set)
{
yield return(t);
}
}
}
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 bool IsSimpleForgetEntityType(PatternUnit pu, string entityType, bool ignoreCase)
{
return(IsSimpleForgetEntityType(pu, entityType) && pu.CaseSensitive == !ignoreCase);
}
public bool IsSimpleForgetEntityType(PatternUnit pu, string entityType)
{
return(pu.Mode == PatternMatchingMode.Single && ((pu.Type == (PatternUnitType.Token | PatternUnitType.Entity)) || (pu.Type == (PatternUnitType.Set | PatternUnitType.Entity))) && pu.EntityType == entityType);
}
