internal void ParseDictionary(Dictionary dictionary)
{
AffixRules rules = null;
Encoding encoding = null;
string affixFile = dictionary.GetFile(DictionaryFileType.Affix);
if (null != affixFile)
{
encoding = Utils.DetectEncoding(affixFile);
if (null == encoding)
{
encoding = EncodingDetector.DetectEncoding(affixFile);
}
rules = this.affixParser.Parse(affixFile, encoding);
}
string fileName = dictionary.GetFile(DictionaryFileType.Dictionary);
DictionaryWithFlags rawDict = this.parser.Parse(fileName, encoding);
foreach (DictionaryItemWithFlags item in rawDict)
{
if (null == item.Flags)
{
dictionary.Add(item.Word);
}
else
{
dictionary.AddRange(rules.GetPossibleWords(item));
}
}
}
internal void ParseConfusionMatrixes(Dictionary dictionary)
{
DictionaryFileType[] matrixes = new DictionaryFileType[] {
DictionaryFileType.DeletetionsMatrix,
DictionaryFileType.InsertionsMatrix,
DictionaryFileType.SubstitutionsMatrix,
DictionaryFileType.TranspositionsMatrix
};
foreach (DictionaryFileType type in matrixes)
{
string file = dictionary.GetFile(type);
if (null != file)
{
ConfusionMatrix matrix = this.matrixParser.ParseMatrix(file);
dictionary.AddConfusionMatrix(ConvertFileTypeToEditOperation(type), matrix);
}
}
}
internal void ParseFrequences(Dictionary dictionary)
{
FrequencyVector<string> oneChrFrq = this.frequencyParser.ParseFrequency(dictionary.GetFile(DictionaryFileType.OneCharFrequences));
dictionary.AddFrequencyVector(FrequencyVectorType.OneChar, oneChrFrq);
FrequencyVector<string> twoChrFrq = this.frequencyParser.ParseFrequency(dictionary.GetFile(DictionaryFileType.TwoCharFrequences));
dictionary.AddFrequencyVector(FrequencyVectorType.TwoChar, twoChrFrq);
}
// todo move
internal void ParseSimpleDictionary(Dictionary dictionary)
{
string file = dictionary.GetFile(DictionaryFileType.LineDictionary);
if (null != file)
{
Encoding enc = EncodingDetector.DetectEncoding(file);
using (StreamReader reader = new StreamReader(file, enc))
{
while (!reader.EndOfStream)
{
dictionary.Add(reader.ReadLine());
}
}
}
}
internal void ParseNgrams(Dictionary dictionary)
{
DictionaryFileType[] files = new DictionaryFileType[]
{
DictionaryFileType.UnigramFrequences,
DictionaryFileType.DigramFrequences,
DictionaryFileType.TrigramFrequences
};
foreach (DictionaryFileType type in files)
{
string file = dictionary.GetFile(type);
if (null != file)
{
NgramCollection collection = this.ngramParser.ParseNgrams(file);
NgramType ngramType;
switch (type)
{
case DictionaryFileType.DigramFrequences:
ngramType = NgramType.Digram;
break;
case DictionaryFileType.TrigramFrequences:
ngramType = NgramType.Trigram;
break;
default:
ngramType = NgramType.Unigram;
break;
}
dictionary.AddNgramCollection(ngramType, collection);
}
}
}