X10D/tools/SourceGenerator/EmojiRegexGenerator.cs

using System.Text;
using System.Text.RegularExpressions;
using Microsoft.CodeAnalysis;
using Microsoft.CodeAnalysis.Text;

namespace SourceGenerator;

[Generator]
internal sealed class EmojiRegexGenerator : ISourceGenerator
{
    // ReSharper disable once IdentifierTypo
    private const string TwemojiRegexUrl = "https://raw.githubusercontent.com/twitter/twemoji-parser/master/src/lib/regex.js";
    private static readonly HttpClient HttpClient = new();
    private string _emojiRegex = string.Empty;

    /// <inheritdoc />
    public void Initialize(GeneratorInitializationContext context)
    {
        string response = HttpClient.GetStringAsync(TwemojiRegexUrl).GetAwaiter().GetResult();
        var regex = new Regex(@"export default /(?<regex>.+)/g;", RegexOptions.Compiled, Regex.InfiniteMatchTimeout);
        using var reader = new StringReader(response);

        while (reader.ReadLine() is { } line)
        {
            if (!line.StartsWith("export default /"))
            {
                continue;
            }

            Match match = regex.Match(line);
            if (!match.Success)
            {
                continue;
            }

            _emojiRegex = $"^{match.Groups["regex"].Value}$";
            break;
        }
    }

    /// <inheritdoc />
    public void Execute(GeneratorExecutionContext context)
    {
        if (string.IsNullOrEmpty(_emojiRegex))
        {
            return;
        }

        var builder = new StringBuilder();
        builder.AppendLine("// This file was auto-generated by the X10D source generator");
        builder.AppendLine("// Do not edit this file manually!");
        builder.AppendLine();

        builder.AppendLine("using System.Text.RegularExpressions;");
        builder.AppendLine();
        builder.AppendLine("namespace X10D.Text;");
        builder.AppendLine();
        builder.AppendLine("internal static class EmojiRegex");
        builder.AppendLine("{");
        builder.AppendLine("    internal static readonly Regex Value = new Regex(");
        builder.AppendLine($"        @\"{_emojiRegex}\",");
        // ReSharper disable once StringLiteralTypo
        builder.AppendLine("        RegexOptions.Compiled | RegexOptions.CultureInvariant | RegexOptions.Singleline");
        builder.AppendLine("    );");
        builder.AppendLine("}");

        context.AddSource("EmojiRegex.g.cs", SourceText.From(builder.ToString(), Encoding.UTF8));
    }
}