public void Process()
{
var words = wordSource.GetWords();
var statistics = new OccurrenceStatistics(words);
var grammarInfo = grammarInfoParser.GetGrammarInfo(statistics.OccurrenceCount.Keys);
statistics = grammarFormJoiner.Join(statistics, grammarInfo);
IEnumerable<string> filteredWords = statistics.OccurrenceCount.Keys;
foreach (var filter in wordFilters)
filteredWords = filter.Filter(filteredWords, grammarInfo);
var filteredWordsSet = new HashSet<string>(filteredWords);
var orderedRatings = statistics.OccurrenceCount
.Select(pair => new WordRating(pair.Key, pair.Value))
.Where(item => filteredWordsSet.Contains(item.Word))
.OrderByDescending(item => item.OccurencesCount)
.Take(wordCount)
.ToArray();
var wordRectangles = fontManager.GenerateFonts(orderedRatings);
var cloudScheme = cloudGenerator.Generate(wordRectangles);
var coloredCloudScheme = colorManager.GenerateColors(cloudScheme);
cloudRenderer.Render(coloredCloudScheme);
}
public OccurrenceStatistics Join(OccurrenceStatistics statistics,
IReadOnlyDictionary<string, WordGrammarInfo> grammarInfo)
{
var wordsGroupedByInitialForm = statistics.OccurrenceCount.Keys
.Where(grammarInfo.ContainsKey)
.GroupBy(word => grammarInfo[word].InitialForm);
return new OccurrenceStatistics(wordsGroupedByInitialForm
.Select(wordForms => new
{
MostCommonForm = wordForms
.OrderByDescending(form => statistics.OccurrenceCount[form])
.First(),
InitialForm = wordForms.Key,
TotalCount = wordForms.Select(form => statistics.OccurrenceCount[form]).Sum()
})
.ToDictionary(item => item.MostCommonForm, item => item.TotalCount));
}
public void Join_selectsMostCommonForm()
{
var filter = new GrammarFormJoiner();
var statistics = new OccurrenceStatistics(new Dictionary<string, int>
{
{"активный", 10}, {"активное", 20}, {"команд", 40},
});
var grammarInfo = new Dictionary<string, WordGrammarInfo>
{
{"активный", new WordGrammarInfo("активный", PartOfSpeech.Adjective)},
{"активное", new WordGrammarInfo("активный", PartOfSpeech.Adjective)},
{"команд", new WordGrammarInfo("команда", PartOfSpeech.Noun)},
};
statistics = filter.Join(statistics, grammarInfo);
statistics.OccurrenceCount.Keys.Should().BeEquivalentTo("активное", "команд");
}
public void Join_mergesGrammarForms()
{
var filter = new GrammarFormJoiner();
var statistics = new OccurrenceStatistics(new Dictionary<string, int>
{
{"активный", 10}, {"активное", 20}, {"команд", 40},
});
var grammarInfo = new Dictionary<string, WordGrammarInfo>
{
{"активный", new WordGrammarInfo("активный", PartOfSpeech.Adjective)},
{"активное", new WordGrammarInfo("активный", PartOfSpeech.Adjective)},
{"команд", new WordGrammarInfo("команда", PartOfSpeech.Noun)},
};
statistics = filter.Join(statistics, grammarInfo);
statistics.OccurrenceCount.Keys.Should().HaveCount(2)
.And.Contain(key => grammarInfo[key].InitialForm == "активный" &&
statistics.OccurrenceCount[key] == 30)
.And.Contain(key => grammarInfo[key].InitialForm == "команда" &&
statistics.OccurrenceCount[key] == 40);
}
