FileReferenceTokenizer.cs

using System;
using System.Collections.Generic;
using System.IO;
using System.Threading.Tasks;
using Waher.Persistence.FullTextSearch.Files;
using Waher.Runtime.Inventory;
 
namespace Waher.Persistence.FullTextSearch.Tokenizers
{
    public class FileReferenceTokenizer : ITokenizer
    {
        public FileReferenceTokenizer()
        {
        }
 
        public Grade Supports(Type Object)
        {
            if (Object == typeof(FileReference))
                return Grade.Ok;
            else
                return Grade.NotAtAll;
        }
 
        public async Task Tokenize(object Value, TokenizationProcess Process)
        {
            if (Value is FileReference Ref)
                await Tokenize(Ref, Process);
        }
 
        public static async Task Tokenize(FileReference Reference, TokenizationProcess Process)
        {
            if (File.Exists(Reference.FileName) &&
                TryGetFileTokenizer(Reference.FileName, out IFileTokenizer Tokenizer))
            {
                await Tokenizer.Tokenize(Reference, Process);
            }
        }
 
        public static bool HasTokenizer(string FileName)
        {
            return TryGetFileTokenizer(FileName, out _);
        }
 
        public static bool TryGetFileTokenizer(string FileName, out IFileTokenizer Tokenizer)
        {
            string Extension = Path.GetExtension(FileName).ToLower();
 
            if (Extension.StartsWith("."))
                Extension = Extension.Substring(1);
 
            lock (tokenizers)
            {
                if (!tokenizers.TryGetValue(Extension, out Tokenizer))
                    Tokenizer = null;
                else if (Tokenizer is null)
                    return false;
            }
 
            Tokenizer = Types.FindBest<IFileTokenizer, string>(Extension);
 
            lock (tokenizers)
            {
                tokenizers[Extension] = Tokenizer;
            }
 
            return !(Tokenizer is null);
        }
 
        private static readonly Dictionary<string, IFileTokenizer> tokenizers = new Dictionary<string, IFileTokenizer>();
 
        static FileReferenceTokenizer()
        {
            Types.OnInvalidated += Types_OnInvalidated;
        }
 
        private static void Types_OnInvalidated(object Sender, EventArgs e)
        {
            lock (tokenizers)
            {
                tokenizers.Clear();
            }
        }
    }
}