public Sentence CreateSentence(BookId bookId, string validatedSentence, int indexInBook)
{
validatedSentence = string.Join(" ", _textProcessor.GetWordsEnumerable(validatedSentence));
var sentenceType = SentenceType.Normal;
var sentenceId = SentenceId.New();
var words = _textProcessor.GetWordsEnumerable(validatedSentence).ToArray();
var wordsList = new List <Word>(words.Length);
var wordsInSentence = words
.GroupBy(word => word)
.Select(group => new
{
Word = group.Key,
Count = group.Count()
}).ToDictionary(x => x.Word);
foreach (var word in words)
{
if (word.All(character => TextConstants.PunctuationCharacters.Contains(character)) &&
word.Length > 1)
{
sentenceType = SentenceType.Other;
}
if (word.EndsWith(".?", StringComparison.Ordinal) || word.EndsWith(".!", StringComparison.Ordinal))
{
sentenceType = SentenceType.Other;
}
}
var rawWordsInSentence = words
.Select(word => _textProcessor.NormalizeWord(word))
.GroupBy(word => word)
.Select(group => new
{
Word = group.Key,
Count = group.Count()
}).ToDictionary(x => x.Word);
var index = 0;
foreach (var word in words)
{
var keyPairs = CreateKeyPairs(validatedSentence, word);
var rawWord = _textProcessor.NormalizeWord(word);
wordsList.Add(new Word(
sentenceId, index,
word, rawWord,
wordsInSentence[word].Count, rawWord == string.Empty ? 0 : rawWordsInSentence[rawWord].Count,
keyPairs));
index++;
}
return(new Sentence(bookId, sentenceId, sentenceType, indexInBook, validatedSentence, wordsList));
}
public static IEnumerable <string> GetWordsEnumerable(
this ITextProcessor textProcessor,
string text,
LanguageInformation languageInformation)
{
return(textProcessor.GetWordsEnumerable(text)
.Where(word => languageInformation.IsAllLettersAllowed(word)));
}
public static IEnumerable <string> GetNormalizedWordsEnumerable(
this ITextProcessor textProcessor, string text)
{
return(textProcessor.GetWordsEnumerable(text)
.Select(word => textProcessor.NormalizeWord(word))
.Where(word => word.Length > 0)
.Distinct());
}
public static IEnumerable <string> GetNormalizedWordsEnumerable(
this ITextProcessor textProcessor,
string text,
LanguageInformation languageInformation)
{
return(textProcessor.GetWordsEnumerable(text)
.Select(word => textProcessor.NormalizeWord(word))
.Where(word => languageInformation.IsAllLettersAllowed(word))
.Where(word => word.Length > 0)
.Distinct());
}
private async ValueTask <IEnumerable <string> > FindRandomWordsAsync(int maxCount, bool rawWords)
{
var randomSentences = await FindRandomSentencesAsync(maxCount, 1)
.ConfigureAwait(false);
var wordsEnumerable = randomSentences.SelectMany(sentence => _textProcessor.GetWordsEnumerable(sentence.Value));
if (rawWords)
{
wordsEnumerable = wordsEnumerable.Select(word => _textProcessor.NormalizeWord(word));
}
var distinctWords = wordsEnumerable.Distinct().ToList();
return(Enumerable.Range(0, maxCount)
.Select(index => RandomNumberGenerator.GetInt32(0, distinctWords.Count))
.Select(index => distinctWords[index])
.ToList());
}