Холстинин Егор

TagsCloudVisualization/BitmapExtensions.cs

@@ -0,0 +1,22 @@
+using System.Drawing;
+using System.Drawing.Imaging;
+
+namespace TagsCloudVisualization;
+
+public static class BitmapExtensions
+{
+    public static void SaveImage(this Bitmap bitmap, string outputFilePath, ImageFormat imageFormat)
+    {
+        outputFilePath = Path.GetFullPath(outputFilePath);
+        var outputFileName = Path.GetFileName(outputFilePath);
+        var outputFileDirectory = Path.GetDirectoryName(outputFilePath);
+
+        Directory.CreateDirectory(outputFileDirectory);
+
+        var savePath = Path.Combine(outputFileDirectory, $"{outputFileName}.{imageFormat.ToString().ToLower()}");
+
+        bitmap.Save(savePath, imageFormat);
+
+        Console.WriteLine($"Image is saved to {savePath}");
+    }
+}

TagsCloudVisualization/DullCheckers/IDullWordChecker.cs

@@ -0,0 +1,6 @@
+namespace TagsCloudVisualization;
+
+public interface IDullWordChecker
+{
+    public bool Check(WordAnalysis word);
+}

TagsCloudVisualization/DullCheckers/MystemDullWordChecker.cs

@@ -0,0 +1,33 @@
+using System.Text;
+
+namespace TagsCloudVisualization;
+
+public class MystemDullWordChecker : IDullWordChecker
+{
+    private HashSet<string> removedPartOfSpeech;
+    private HashSet<string> excludedWords = new();
+
+    public MystemDullWordChecker(HashSet<string> removedPartOfSpeech, string? excludedWordsFile)
+    {
+        this.removedPartOfSpeech = removedPartOfSpeech;
+        if (excludedWordsFile is null)
+            return;
+
+        try
+        {
+            excludedWords =
+                new HashSet<string>(File.ReadAllText(excludedWordsFile, Encoding.UTF8).Split(Environment.NewLine));
+        }
+        catch (FileNotFoundException e)
+        {
+            Console.WriteLine($"Could not find specified excluded words file {excludedWordsFile}. " +
+                              $"No words will be excluded.");
+        }
+    }
+
+    public bool Check(WordAnalysis wordAnalysis)
+    {
+        return removedPartOfSpeech.Any(dullPart => wordAnalysis.GrammarAnalysis.StartsWith(dullPart))
+               || excludedWords.Contains(wordAnalysis.Lexema.ToLower());
+    }
+}

TagsCloudVisualization/IRectangleLayouter.cs

@@ -0,0 +1,9 @@
+using System.Drawing;
+
+namespace TagsCloudVisualization;
+
+public interface IRectangleLayouter
+{
+    public Rectangle PutNextRectangle(Size rectangleSize);
+    public Rectangle PutNextRectangle(SizeF rectangleSize);
+}

TagsCloudVisualization/LayoutDrawer.cs

@@ -0,0 +1,57 @@
+using System.Drawing.Imaging;
+using System.Drawing;
+
+namespace TagsCloudVisualization;
+
+public class LayoutDrawer
+{
+    private IInterestingWordsParser interestingWordsParser;
+    private IRectangleLayouter rectangleLayouter;
+    private IPalette palette;
+    private Font font;
+
+    public LayoutDrawer(IInterestingWordsParser interestingWordsParser,
+        IRectangleLayouter rectangleLayouter,
+        IPalette palette,
+        Font font)
+    {
+        this.interestingWordsParser = interestingWordsParser;
+        this.rectangleLayouter = rectangleLayouter;
+        this.palette = palette;
+        this.font = font;
+    }
+
+    public Bitmap CreateLayoutImageFromFile(string inputFilePath,
+        Size imageSize,
+        int minimumFontSize)
+    {
+        var bitmap = new Bitmap(imageSize.Width, imageSize.Height);
+        using var graphics = Graphics.FromImage(bitmap);
+
+        inputFilePath = Path.GetFullPath(inputFilePath);
+
+        var sortedWordsCount = interestingWordsParser.GetInterestingWords(inputFilePath)
+            .GroupBy(s => s)
+            .Select(group => new { Word = group.Key, Count = group.Count() })
+            .OrderByDescending(wordCount => wordCount.Count);
+        var mostWordOccurrencies = sortedWordsCount.Max(arg => arg.Count);
+
+        graphics.Clear(palette.GetBackgroundColor());
+
+        foreach (var wordCount in sortedWordsCount)
+        {
+            var rectangleFont = new Font(font.FontFamily, 
+                Math.Max(font.Size * wordCount.Count / mostWordOccurrencies, minimumFontSize));
+            var rectangleSize = graphics.MeasureString(wordCount.Word, rectangleFont);
+
+            var textRectangle = rectangleLayouter.PutNextRectangle(rectangleSize);
+            var x = textRectangle.X + imageSize.Width / 2;
+            var y = textRectangle.Y + imageSize.Height / 2;
+
+            using var brush = new SolidBrush(palette.GetNextWordColor());
+            graphics.DrawString(wordCount.Word, rectangleFont, brush, x, y);
+        }
+
+        return bitmap;
+    }
+}

TagsCloudVisualization/Palletes/IPalette.cs

@@ -0,0 +1,10 @@
+using System.Drawing;
+
+namespace TagsCloudVisualization;
+
+public interface IPalette
+{
+    public Color GetNextWordColor();
+
+    public Color GetBackgroundColor();
+}

TagsCloudVisualization/Palletes/Palette.cs

@@ -0,0 +1,27 @@
+using System.Drawing;
+
+namespace TagsCloudVisualization;
+
+public class Palette : IPalette
+{
+    public Palette(Color[] textColor, Color backgroundColor)
+    {
+        TextColor = textColor;
+        BackgroundColor = backgroundColor;
+    }
+
+    private Color[] TextColor { get; set; }
+    private int currentColorId = 0;
+    private Color BackgroundColor { get; set; }
+
+    public Color GetNextWordColor()
+    {
+        if (currentColorId >= TextColor.Length) currentColorId = 0;
+        return TextColor[currentColorId++];
+    }
+
+    public Color GetBackgroundColor()
+    {
+        return BackgroundColor;
+    }
+}

TagsCloudVisualization/PointGenerators/IPointGenerator.cs

@@ -0,0 +1,8 @@
+using System.Drawing;
+
+namespace TagsCloudVisualization;
+
+public interface IPointGenerator
+{
+    Point GetNextPoint();
+}

TagsCloudVisualization/PointGenerators/LissajousCurvePointGenerator.cs

@@ -0,0 +1,29 @@
+using System.Drawing;
+
+namespace TagsCloudVisualization;
+
+public class LissajousCurvePointGenerator : IPointGenerator
+{
+    private int xAmplitude = 100;
+    private int yAmplitude = 100;
+    private int xConstant = 19;
+    private int yConstant = 20;
+    private double delta = Math.PI / 2;
+    private double parameter = 0;
+
+    public Point GetNextPoint()
+    {
+        parameter += 0.01;
+        var x = Math.Round(xAmplitude * Math.Sin(xConstant * parameter + delta));
+        var y = Math.Round(yAmplitude * Math.Sin(yConstant * parameter));
+
+        if (parameter > 20)
+        {
+            parameter = 0;
+            xAmplitude += 20;
+            yAmplitude += 20;
+        }
+
+        return new Point((int)x, (int)y);
+    }
+}

TagsCloudVisualization/PointGenerators/SpiralPointGenerator.cs

@@ -0,0 +1,26 @@
+using System.Drawing;
+
+namespace TagsCloudVisualization;
+
+public class SpiralPointGenerator : IPointGenerator
+{
+    public Point Center { get; } = new(0, 0);
+    public int Radius { get; private set; }
+    public double Angle { get; private set; }
+    public int RadiusDelta { get; private set; } = 1;
+    public double AngleDelta { get; private set; } = Math.PI / 60;
+
+    public Point GetNextPoint()
+    {
+        var x = (int)Math.Round(Center.X + Radius * Math.Cos(Angle));
+        var y = (int)Math.Round(Center.Y + Radius * Math.Sin(Angle));
+
+        var nextAngle = Angle + AngleDelta;
+        var angleMoreThan2Pi = Math.Abs(nextAngle) >= Math.PI * 2;
+
+        Radius = angleMoreThan2Pi ? Radius + RadiusDelta : Radius;
+        Angle = angleMoreThan2Pi ? 0 : nextAngle;
+
+        return new Point(x, y);
+    }
+}

TagsCloudVisualization/Program.cs

@@ -0,0 +1,99 @@
+using System.ComponentModel.DataAnnotations;
+using System.Drawing;
+using System.Drawing.Imaging;
+using McMaster.Extensions.CommandLineUtils;
+using Microsoft.Extensions.DependencyInjection;
+
+namespace TagsCloudVisualization;
+
+public class Program
+{
+    public static int Main(string[] args) => CommandLineApplication.Execute<Program>(args);
+
+    [Argument(0, Description = "Path to txt file with words")]
+    [Required(ErrorMessage = "Expected to get path to file with words as first positional argument." +
+                             "\nExample: C:\\PathTo\\File.txt\nOr relative to exe: PathTo\\File.txt")]
+    public string InputFilePath { get; set; }
+
+    [Argument(1, Description = "Path to output file")]
+    [Required(ErrorMessage = "Expected to get output file path as second positional argument." +
+                             "\nExample: C:\\PathTo\\File\nOr relative to exe: PathTo\\File")]
+    public string OutputFilePath { get; set; }
+
+    [Option("-w", Description = "Image width in pixels")]
+    private int ImageWidth { get; set; } = 1000;
+
+    [Option("-h", Description = "Image height in pixels")]
+    private int ImageHeight { get; set; } = 1000;
+
+    [Option("-bc", Description = "Image background color from KnownColor enum")]
+    private Color BackgroundColor { get; set; } = Color.Wheat;
+
+    [Option("-tc", Description = "Image words colors sequence array from KnownColor enum. " +
+                                 "Can be set multiple times for sequence. Example: -tc black -tc white")]
+    private Color[] TextColor { get; set; } = { Color.Black };
+
+    [Option("-ff", Description = "Font used for words")]
+    private string FontFamily { get; set; } = "Arial";
+
+    [Option("-fs", Description = "Max font size in em")]
+    private int FontSize { get; set; } = 50;
+
+    [Option("-mfs", Description = "Min font size in em")]
+    private int MinimalFontSize { get; set; } = 0;
+
+    [Option("-img", Description = "Output image format. Choosen from ImageFormat")]
+    private ImageFormat SaveImageFormat { get; set; } = ImageFormat.Png;
+
+    [Option("-ef", Description = "Txt file with words to exclude. 1 word in line. Words must be lexems.")]
+    private string ExcludedWordsFile { get; set; }
+
+    [Option("-rp", Description = "Parts of speech abbreviations that are excluded from parsed words. " +
+                                 "More info here https://yandex.ru/dev/mystem/doc/ru/grammemes-values")]
+    private HashSet<string> RemovedPartsOfSpeech { get; set; } = new()
+        { "ADVPRO", "APRO", "INTJ", "CONJ", "PART", "PR", "SPRO" };
+
+    [Option("-square", Description = "Will use another algorithm to generate square tag cloud instead of circular.")]
+    private bool SqareAlgorithm { get; set; }
+
+
+    private void OnExecute()
+    {
+        var services = new ServiceCollection();
+        services.AddTransient<Font>(x => new Font(FontFamily, FontSize));
+        services.AddTransient<IPalette>(x => new Palette(TextColor, BackgroundColor));
+        if (SqareAlgorithm)
+            services.AddTransient<IPointGenerator, LissajousCurvePointGenerator>();
+        else
+            services.AddTransient<IPointGenerator, SpiralPointGenerator>();
+        services.AddTransient<IDullWordChecker>(x =>
+            new MystemDullWordChecker(RemovedPartsOfSpeech, ExcludedWordsFile));
+        services.AddTransient<IInterestingWordsParser, MystemWordsParser>();
+        services.AddTransient<IRectangleLayouter, RectangleLayouter>();
+        services.AddTransient<LayoutDrawer>();
+
+        using var provider = services.BuildServiceProvider();
+
+        var layoutDrawer = provider.GetRequiredService<LayoutDrawer>();
+        try
+        {
+            layoutDrawer
+                .CreateLayoutImageFromFile(InputFilePath, new Size(ImageWidth, ImageHeight), MinimalFontSize)
+                .SaveImage(OutputFilePath, SaveImageFormat);
+        }
+        catch (Exception ex)
+        {
+            if (ex is FileNotFoundException or DirectoryNotFoundException)
+            {
+                Console.WriteLine(ex.Message);
+                if (!Path.IsPathRooted(InputFilePath))
+                    Console.WriteLine("Relative paths are searched realative to .exe file. " +
+                                      "Try giving an absolute path.");
+            }
+            else
+            {
+                throw;
+            }
+        }
+    }
+}