BangumiRenamer/EpisodeParser.cs

using System.Collections.Concurrent;
using System.Text;
using System.Text.RegularExpressions;
using Newtonsoft.Json;

namespace ConsoleApp1;

public class EpisodeInfo
{
    public string path;
    public string name;
    public string session;
    public string episode;
    public string group;
    public string type; // others, episode, subtitle
    public string language;
}

public class EpisodeParseResult
{
    public bool success;
    public string originalQuestion;
    public EpisodeInfo parseResult;
}

public class EpisodeParser
{
    // todo: 添加解析年份
    private bool _running = false;
    
    private const string PromptPath = "Prompt.txt";

    private readonly string _prompt;
    private readonly OllamaHelper _ollama;

    private ConcurrentQueue<string> _questions;
    private ConcurrentQueue<EpisodeParseResult> _results;
    
    public bool Running => _running;
    public int TotalQuestions => _questions.Count + _results.Count;
    public int CompletedQuestions => _results.Count;
    
    public int RestQuestions => _questions.Count;

    public bool TryGetResult(out EpisodeParseResult result)
    {
        return _results.TryDequeue(out result);
    }
    
    public EpisodeParser()
    {
        _prompt = File.ReadAllText(PromptPath);
        _ollama = new OllamaHelper();
        _questions = new ConcurrentQueue<string>();
        _results = new ConcurrentQueue<EpisodeParseResult>();
    }

    public void Append(string question)
    {
        _questions.Enqueue(question);
    }

    public void Start()
    {
        if (_running) return ;
        _running = true;
        DoParse();
    }

    private string Preprocess(string respoonds)
    {
        return respoonds.Replace("```json\n", "").Replace("```", "");
    }

    private static string RemoveNonDigits(string s)
    {
        return Regex.Replace(s, @"[^\d\.]*", "");
    }

    private static string RemoveFrontZeros(string s)
    {
        var result = new StringBuilder();
        bool front = true;
        foreach (var c in s)
        {
            if (front && c == '0') continue;
            front = false;
            result.Append(c);
        }
        return result.ToString();
    }
    
    private static string ProcessSession(string session)
    {
        session = RemoveFrontZeros(RemoveNonDigits(session));
        if (session.Length > 2) session = ""; 
        return session == "" ? "1" : session;
    }

    private static string ProcessEpisode(string episode)
    {
        episode = RemoveFrontZeros(RemoveNonDigits(episode));
        return episode == "" ? "1" : episode;
    }
    
    private void FinalProcess(EpisodeInfo info)
    {
        if (info.type == "others")
        {
            info.episode = "";
            info.session = ProcessSession(info.session);
        }
        else
        {
            info.episode = ProcessEpisode(info.episode);
            info.session = ProcessSession(info.session);
        }
    }
    
    private async void DoParse()
    {
        while (_questions.TryDequeue(out string question))
        {
            var result = new EpisodeParseResult();
            result.originalQuestion = question;
            var responds = await _ollama.Ask(_prompt + $"\"{question}\"");
            try
            {
                responds = Preprocess(responds);
                result.parseResult = JsonConvert.DeserializeObject<EpisodeInfo>(responds);
                result.parseResult.path = question;
                result.success = result.parseResult != null;
                if (result.success)
                {
                    FinalProcess(result.parseResult);
                }
            }
            catch (Exception _)
            {
                result.success = false;    
            }
            _results.Enqueue(result);
        }
        _running = false;
    }
}