//public void RemoveForgetPattern(string entityType, string name)
//{
// if (!Data.AutoOptimizePatterns.Value) { return; }
// RWLock.EnterWriteLock();
// try
// {
// if (Data.RemovePatternsPerEntityType.TryGetValue(entityType, out var list))
// {
// list.RemoveAll(p => p.Name == name);
// if (list.Count == 0) { Data.RemovePatternsPerEntityType.Remove(entityType); }
// }
// }
// finally
// {
// RWLock.ExitWriteLock();
// }
//}
//public void RemoveAddPattern(string entityType, string name)
//{
// RWLock.EnterWriteLock();
// try
// {
// if (Data.AddPatternPerEntityType.TryGetValue(entityType, out var mp))
// {
// mp.Patterns.RemoveAll(p => p.Name == name);
// if (mp.Patterns.Count == 0) { Data.AddPatternPerEntityType.Remove(entityType); }
// RWLock.ExitWriteLock();
// }
// }
// finally
// {
// RWLock.ExitWriteLock();
// }
//}
public void OptimizeMatchingPattern(string entityType, MatchingPattern mp, bool isAdd)
{
var patterns = mp.Patterns.ToArray();
mp.Patterns.Clear();
var ignorePatterns = patterns.Where(p => p.All(pu => isAdd ? !IsSimpleAddEntityType(pu, entityType) : !IsSimpleForgetEntityType(pu, entityType))).ToList();
mp.Patterns.AddRange(ignorePatterns);
foreach (var ignoreCase in new bool[] { false, true })
{
var words = new Dictionary <int, List <List <string[]> > >();
foreach (var mergePatterns in patterns.Where(p => p.All(pu => isAdd ? IsSimpleAddEntityType(pu, entityType, ignoreCase) : IsSimpleForgetEntityType(pu, entityType, ignoreCase))))
{
if (!words.TryGetValue(mergePatterns.Length, out var wordList))
{
wordList = new List <List <string[]> >();
words[mergePatterns.Length] = wordList;
}
wordList.Add(mergePatterns.Select(pu => GetTokens(pu).ToArray()).ToList());
}
foreach (var kv in words)
{
var len = kv.Key; var wordList = kv.Value;
var newPatterns = new List <IPatternUnit>();
var hs = new HashSet <string> [len];
for (int i = 0; i < len; i++)
{
hs[i] = new HashSet <string>();
}
foreach (var wl in wordList)
{
for (int i = 0; i < len; i++)
{
foreach (var w in wl[i])
{
hs[i].Add(w);
}
}
}
newPatterns.AddRange(hs.Select(h => PatternUnitPrototype.Single().WithTokens(h, ignoreCase)));
foreach (var pu in newPatterns)
{
if (isAdd)
{
pu.WithoutEntityType(entityType);
}
else
{
pu.WithEntityType(entityType);
}
}
mp.Patterns.Add(newPatterns.Select(prot => new PatternUnit(prot)).ToArray());
}
}
}
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());
}
private bool MatchSet(ref Token token)
{
if (SetHashes is null)
{
//No need to lock here, as we would just replace one with another equal set if there is a colision
SetHashes = new HashSet <ulong>(Set.Select(tk => CaseSensitive ? PatternUnitPrototype.Hash64(tk.AsSpan()) : PatternUnitPrototype.IgnoreCaseHash64(tk.AsSpan())));
}
return(SetHashes.Contains(GetTokenHash(ref token)));
}
private ulong GetTokenHash(ref Token token)
{
return(CaseSensitive ? PatternUnitPrototype.Hash64(token.ValueAsSpan) : PatternUnitPrototype.IgnoreCaseHash64(token.ValueAsSpan));
}
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;
_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(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(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;
}
