public static void CalcWordsCount(JsonSerializerMaster jsonSerializer, int requestCount)
{
var words = 0;
var notWords = 0;
for (var i = 0; i < requestCount; i++)
{
using (var web = new WebClient())
{
web.BaseAddress = baseUrl;
var json = web.DownloadString("");
var data = jsonSerializer.Deserialize <Dictionary <string, bool> >(json);
var wordsHere = data.Count(kvp => kvp.Value);
var notWordsHere = data.Count - wordsHere;
words += wordsHere;
notWords += notWordsHere;
Console.WriteLine($"[{i}]: words {wordsHere}, not words {notWordsHere}");
}
}
Console.WriteLine(new { words, notWords, wordsMean = (double)words / (words + notWords) });
}
public WordsGeneratorOld(JsonSerializerMaster jsonSerializer, string dir, string[] words, string alphabet)
{
this.jsonSerializer = jsonSerializer;
this.alphabet = alphabet;
this.dir = dir;
this.words = words.Where(w => w.Length == wordLen).ToArray();
wordsHash = this.words.ToHashSet();
alphabetIndexes = alphabet.Select((c, i) => new { c, i }).ToDictionary(a => a.c, a => a.i);
}
// 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);
}
}
public LettersCount(JsonSerializerMaster jsonSerializer)
{
this.jsonSerializer = jsonSerializer;
var stats = new Dictionary <char, int[]>
{
{ '\'', new int[] { 234279, 226, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'a', new int[] { 94043, 96525, 36614, 6568, 717, 39, 1, 0, 0, 0, 0, 0, 0 } },
{ 'b', new int[] { 201739, 30462, 2187, 113, 6, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'c', new int[] { 158387, 64089, 11043, 917, 63, 7, 1, 0, 0, 0, 0, 0, 0 } },
{ 'd', new int[] { 178320, 50227, 5613, 338, 9, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'e', new int[] { 91271, 95304, 39346, 7818, 727, 41, 0, 0, 0, 0, 0, 0, 0 } },
{ 'f', new int[] { 216231, 16081, 2109, 77, 9, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'g', new int[] { 196931, 34803, 2657, 107, 9, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'h', new int[] { 179544, 49839, 4991, 131, 2, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'i', new int[] { 105872, 95363, 28753, 4196, 317, 6, 0, 0, 0, 0, 0, 0, 0 } },
{ 'j', new int[] { 229921, 4513, 72, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'k', new int[] { 216418, 16803, 1227, 52, 7, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'l', new int[] { 146909, 71332, 14708, 1425, 131, 1, 1, 0, 0, 0, 0, 0, 0 } },
{ 'm', new int[] { 176480, 51811, 5883, 320, 13, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'n', new int[] { 134824, 81728, 16514, 1391, 49, 1, 0, 0, 0, 0, 0, 0, 0 } },
{ 'o', new int[] { 123124, 82043, 25070, 3869, 386, 15, 0, 0, 0, 0, 0, 0, 0 } },
{ 'p', new int[] { 180038, 47779, 6379, 306, 5, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'q', new int[] { 230828, 3602, 77, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'r', new int[] { 119267, 95181, 18759, 1265, 35, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 's', new int[] { 147439, 69917, 15148, 1861, 127, 14, 1, 0, 0, 0, 0, 0, 0 } },
{ 't', new int[] { 133724, 80173, 18907, 1629, 74, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'u', new int[] { 175077, 53487, 5587, 332, 24, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'v', new int[] { 217794, 16096, 607, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'w', new int[] { 223053, 11001, 442, 11, 0, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'x', new int[] { 227921, 6534, 49, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'y', new int[] { 198793, 33684, 1986, 44, 0, 0, 0, 0, 0, 0, 0, 0, 0 } },
{ 'z', new int[] { 226260, 7576, 633, 23, 15, 0, 0, 0, 0, 0, 0, 0, 0 } },
};
thresholds = stats.ToDictionary(
kvp => kvp.Key,
kvp => kvp.Value.FirstIndexOf(i => i < 1500));
}
public SubstCalc(JsonSerializerMaster jsonSerializer, string dir)
{
this.jsonSerializer = jsonSerializer;
this.dir = dir;
replaceCommands = new List <SubstReplaceCommand>();
}
public InfixCalc(JsonSerializerMaster jsonSerializer, string dir)
{
this.jsonSerializer = jsonSerializer;
this.dir = dir;
}
public WordsGenerator(JsonSerializerMaster jsonSerializer, string dir)
{
this.jsonSerializer = jsonSerializer;
this.dir = dir;
}
public static void Save(JsonSerializerMaster jsonSerializer, string dir, string[] words)
{
var wordFeatures = Convert(words);
File.WriteAllText(Path.Combine(dir, "word_features.json"), jsonSerializer.Serialize(wordFeatures));
}
public static WordFeatureRow[] Read(JsonSerializerMaster jsonSerializer, string dir)
{
return(jsonSerializer.Deserialize <WordFeatureRow[]>(File.ReadAllText(Path.Combine(dir, "word_features.json"))));
}