Address all warnings

Directory.Build.props

@@ -10,12 +10,12 @@
  <PackageReleaseNotes>https://github.com/russcam/lingua-dotnet/releases</PackageReleaseNotes>
  <PackageTags>lingua, language detection</PackageTags>
  <LangVersion>latest</LangVersion>
- <!-- TODO: fix all warnings and enable -->
- <TreatWarningsAsErrors>false</TreatWarningsAsErrors>
+ <TreatWarningsAsErrors>true</TreatWarningsAsErrors>
  <IsPackable>False</IsPackable>
  <Nullable>enable</Nullable>
  <ImplicitUsings>enable</ImplicitUsings>
  <MinVerMinimumMajorMinor>1.0</MinVerMinimumMajorMinor>
+ <Deterministic>true</Deterministic>
  </PropertyGroup>
 
  <PropertyGroup>

build/UnicodeScriptGenerator/Program.cs

@@ -52,7 +52,10 @@ public enum UnicodeScript
  Unknown,
 }
 
-public static class CharExtensions
+/// <summary>
+/// Extension method for <see cref=""char""/> to determine its <see cref=""UnicodeScript""/> property.
+/// </summary>
+public static class UnicodeScriptInfo
 {
  private static readonly int[] ScriptStarts =
  {");

src/Lingua/Api/CharExtensions.cs

@@ -2,7 +2,7 @@
 
 internal static class CharExtensions
 {
- private static readonly HashSet<Alphabet> AlphabetsWithLogograms = LanguageExtensions.LanguagesSupportingLogograms
+ private static readonly HashSet<Alphabet> AlphabetsWithLogograms = LanguageInfo.LanguagesSupportingLogograms
  .SelectMany(l => l.Alphabets())
  .ToHashSet();
 

src/Lingua/Api/IO/FilesWriter.cs

@@ -0,0 +1,25 @@
+namespace Lingua.Api.IO;
+
+internal abstract class PathValidation
+{
+ public static void CheckInputFilePath(string inputFilePath) {
+ if (!Path.IsPathRooted(inputFilePath))
+ throw new ArgumentException($"Input file path '{inputFilePath}' is not absolute");
+
+ if (!File.Exists(inputFilePath))
+ throw new DirectoryNotFoundException($"Input file '{inputFilePath}' does not exist");
+
+ var attributes = File.GetAttributes(inputFilePath);
+ if (!attributes.HasFlag(FileAttributes.Normal))
+ throw new ArgumentException($"Input file path '{inputFilePath}' does not represent a regular file");
+ }
+
+ public static void CheckOutputDirectoryPath(string outputDirectoryPath)
+ {
+ if (!Path.IsPathRooted(outputDirectoryPath))
+ throw new ArgumentException($"Output directory path '{outputDirectoryPath}' is not absolute");
+
+ if (!Directory.Exists(outputDirectoryPath))
+ throw new DirectoryNotFoundException($"Output directory '{outputDirectoryPath}' does not exist");
+ }
+}

src/Lingua/Api/IO/LanguageModelFilesWriter.cs → src/Lingua/Api/IO/LanguageModelWriter.cs

@@ -1,23 +1,25 @@
 using System.Text;
-using System.Text.Json;
 using Lingua.Internal;
-using Lingua.Internal.IO;
+using static Lingua.Api.IO.PathValidation;
 
 namespace Lingua.Api.IO;
 
-public class LanguageModelFilesWriter : FilesWriter
+/// <summary>
+/// Creates language model files and writes them to a directory.
+/// </summary>
+public class LanguageModelWriter
 {
  /// <summary>
  /// Creates language model files and writes them to a directory.
  /// </summary>
  /// <param name="inputFilePath">The path to a txt file used for language model creation.</param>
- /// <param name="inputFileCharset">The encoding of <see cref="inputFilePath"/>. Defaults to <see cref="Encoding.UTF8"/></param>
+ /// <param name="encoding">The encoding of <paramref name="inputFilePath"/>. Defaults to <see cref="Encoding.UTF8"/></param>
  /// <param name="outputDirectoryPath">The directory where the language model files are to be written.</param>
  /// <param name="language">The language for which to create language models.</param>
  /// <param name="charClass"> A regex character class as supported by <see cref="System.Text.RegularExpressions.Regex"/></param>
  public void CreateAndWriteLanguageModelFiles(
  string inputFilePath,
- Encoding? inputFileCharset,
+ Encoding? encoding,
  string outputDirectoryPath,
  Language language,
  string charClass = "\\p{L}"
@@ -26,42 +28,42 @@ public void CreateAndWriteLanguageModelFiles(
  CheckInputFilePath(inputFilePath);
  CheckOutputDirectoryPath(outputDirectoryPath);
 
- inputFileCharset ??= Encoding.UTF8;
+ encoding ??= Encoding.UTF8;
  var unigramModel = CreateLanguageModel(
  inputFilePath,
- inputFileCharset,
+ encoding,
  language,
  1,
  charClass,
  new Dictionary<Ngram, int>()
  );
  var bigramModel = CreateLanguageModel(
  inputFilePath,
- inputFileCharset,
+ encoding,
  language,
  2,
  charClass,
  unigramModel.AbsoluteFrequencies
  );
  var trigramModel = CreateLanguageModel(
  inputFilePath,
- inputFileCharset,
+ encoding,
  language,
  3,
  charClass,
  bigramModel.AbsoluteFrequencies
  );
  var quadrigramModel = CreateLanguageModel(
  inputFilePath,
- inputFileCharset,
+ encoding,
  language,
  4,
  charClass,
  trigramModel.AbsoluteFrequencies
  );
  var fivegramModel = CreateLanguageModel(
  inputFilePath,
- inputFileCharset,
+ encoding,
  language,
  5,
  charClass,