public static WordFeatureRow[] Convert(string[] words)
{
var result = new WordFeatureRow[words.Length];
for (var i = 0; i < words.Length; i++)
{
result[i] = new WordFeatureRow(words[i]);
}
return(result);
}
private static void ApostropheEnd_Func_Bloom(Func <WordFeatureRow, bool> isWord_Word, Func <WordFeatureRow, bool> isUnknown_Word, int bloomFilterSize)
{
var n = 0;
var filter = new BloomFilter(bloomFilterSize, 1);
//var filter = new ArrayBloomFilter(6000, 60000 * 8, 2);
foreach (var feature in wordFeatures)
{
filter.Add(feature.Word);
n++;
}
filter.ShowInfo();
solutionCreator.WriteData(filter);
var negativeStats = new WeightedSet <string>();
CalcErrors(word =>
{
var w = GetApostropheEndBase(word);
w = substCalc.SubstituteWordByCommands(w);
if (!filter.Contains(w))
{
negativeStats.Increment("!filter.Contains", 1); return(false);
}
var feature = new WordFeatureRow(w);
if (!isWord_Word(feature))
{
negativeStats.Increment("!isWord_Word", 1); return(false);
}
if (!isUnknown_Word(feature))
{
negativeStats.Increment("!isUnknown_Word", 1); return(false);
}
return(true);
});
Console.WriteLine("Negative stats:");
foreach (var kvp in negativeStats)
{
Console.WriteLine(new { reason = kvp.Key, count = kvp.Value });
}
Console.WriteLine();
}
// Сохранить тесты с сайта
// HolaWeb.SaveTests(9000, @"D:\Develop\holachallenge\js\testcases");
// Создать файл всех используемых в фильтре слов
// File.WriteAllLines(Path.Combine(dir, "words_out.txt"), wordFeatures.Select(f => f.Word).ToArray());
public static void Main()
{
try
{
jsonSerializer = new JsonSerializerMaster();
infixCalc = new InfixCalc(jsonSerializer, dir);
originalWords = File.ReadAllLines(Path.Combine(dir, "words_uniq.txt"));
originalNotWords = File.ReadAllLines(Path.Combine(dir, "notwords_uniq.txt"));
words = originalWords.Where(w => !w.EndsWith("'s")).ToArray();
substCalc = new SubstCalc(jsonSerializer, dir);
lettersCount = new LettersCount(jsonSerializer);
solutionCreator = new SolutionCreator(dir, substCalc);
//substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "suffix first.json", 6), false);
//substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "prefix first.json", 7), true);
//substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "suffix second.json", 6), false);
//substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "prefix second.json", 7), false);
//substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "suffix third.json", 6), false); // 4.8 по "ing"->"ed"
//words = substCalc.ReplaceByCommands(words);
substCalc.AddCommand(substCalc.GetFirstPopularSuffixReplaces(), false);
substCalc.AddCommand(substCalc.GetPrefixReplaceMap(words), true);
substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "trash\\suffix second 3 5 50.json", 5), false);
substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "trash\\prefix second 3 5 50.json", 6), true);
substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "trash\\suffix third.json", 5), false);
substCalc.AddCommand(substCalc.GetAffixReplaceMap(words, "trash\\prefix third.json", 6), true);
words = substCalc.ReplaceByCommands(words);
mainBloomLength = 61450;
Func <WordFeatureRow, bool> isWord =
w => w.Length <= 14 &&
w.VowInRow <= 3 &&
w.ConsInRow <= 4 &&
w.Word.Count(c => c == '\'') <= 1;
wordFeatures = WordFeatureRow.Convert(words).Where(isWord).ToArray();
ApostropheEnd_C2_LetterPositions_BloomPrefix_Func_Bloom(isWord);
}
catch (Exception ex)
{
Console.WriteLine(ex);
}
}
