using System.IO;
using System.Text;
using System.Text.RegularExpressions;
using Microsoft.Data.Sqlite;

namespace AxCopilot.Services;

public partial class CodeIndexService
{
    // ── 검색 + TF-IDF + Dispose ──────────────────────────────────────────

    // ── 검색 ────────────────────────────────────────────────────────────

    /// <summary>시맨틱 검색: 질문과 가장 관련 있는 코드 청크를 반환합니다.</summary>
    public List<SearchResult> Search(string query, int maxResults = 5)
    {
        if (!_indexed || _db == null || _totalDocs == 0)
            return new();

        var queryTokens = Tokenize(query);
        if (queryTokens.Count == 0) return new();

        // 쿼리 토큰의 DF 조회
        var dfMap = new Dictionary<string, int>();
        foreach (var token in queryTokens.Keys)
        {
            using var cmd = _db.CreateCommand();
            cmd.CommandText = "SELECT df FROM doc_freq WHERE token = @t";
            cmd.Parameters.AddWithValue("@t", token);
            var result = cmd.ExecuteScalar();
            if (result != null) dfMap[token] = Convert.ToInt32(result);
        }

        // 후보 청크 검색: 쿼리 토큰이 하나라도 포함된 청크만
        var candidateChunks = new HashSet<int>();
        foreach (var token in queryTokens.Keys)
        {
            using var cmd = _db.CreateCommand();
            cmd.CommandText = "SELECT DISTINCT chunk_id FROM tokens WHERE token = @t";
            cmd.Parameters.AddWithValue("@t", token);
            using var reader = cmd.ExecuteReader();
            while (reader.Read()) candidateChunks.Add(reader.GetInt32(0));
        }

        if (candidateChunks.Count == 0) return new();

        // 각 후보 청크의 TF-IDF 유사도 계산
        var scored = new List<(int ChunkId, double Score)>();

        foreach (var chunkId in candidateChunks)
        {
            // 청크의 토큰 TF 로드
            var docTf = new Dictionary<string, int>(StringComparer.OrdinalIgnoreCase);
            using (var cmd = _db.CreateCommand())
            {
                cmd.CommandText = "SELECT token, tf FROM tokens WHERE chunk_id = @cid";
                cmd.Parameters.AddWithValue("@cid", chunkId);
                using var reader = cmd.ExecuteReader();
                while (reader.Read())
                    docTf[reader.GetString(0)] = reader.GetInt32(1);
            }

            var score = ComputeTfIdfSimilarity(queryTokens, docTf, dfMap);
            if (score > 0.01)
                scored.Add((chunkId, score));
        }

        // 상위 결과 추출
        var topChunks = scored
            .OrderByDescending(s => s.Score)
            .Take(maxResults)
            .ToList();

        var results = new List<SearchResult>();
        foreach (var (chunkId, score) in topChunks)
        {
            using var cmd = _db.CreateCommand();
            cmd.CommandText = """
                SELECT f.path, c.start_line, c.end_line, c.content
                FROM chunks c JOIN files f ON c.file_id = f.id
                WHERE c.id = @cid
            """;
            cmd.Parameters.AddWithValue("@cid", chunkId);
            using var reader = cmd.ExecuteReader();
            if (reader.Read())
            {
                results.Add(new SearchResult
                {
                    FilePath = reader.GetString(0),
                    StartLine = reader.GetInt32(1),
                    EndLine = reader.GetInt32(2),
                    Score = score,
                    Preview = reader.GetString(3) is { Length: > 200 } s ? s[..200] + "..." : reader.GetString(3),
                });
            }
        }

        return results;
    }

    /// <summary>기존 인덱스가 있으면 로드합니다 (앱 재시작 시).</summary>
    public void TryLoadExisting(string workFolder)
    {
        if (string.IsNullOrEmpty(workFolder) || !Directory.Exists(workFolder)) return;

        var dbPath = GetDbPath(workFolder);
        if (!File.Exists(dbPath)) return;

        EnsureDb(workFolder);
        _totalDocs = GetTotalChunkCount();
        _indexed = _totalDocs > 0;

        if (_indexed)
            LogService.Info($"기존 코드 인덱스 로드: {_totalDocs}개 청크 [{workFolder}]");
    }

    // ── TF-IDF 계산 ─────────────────────────────────────────────────────

    private double ComputeTfIdfSimilarity(
        Dictionary<string, int> queryTf,
        Dictionary<string, int> docTf,
        Dictionary<string, int> dfMap)
    {
        double dotProduct = 0, queryNorm = 0, docNorm = 0;

        foreach (var (token, qtf) in queryTf)
        {
            var df = dfMap.GetValueOrDefault(token, 0);
            var idf = Math.Log(1.0 + _totalDocs / (1.0 + df));
            var qWeight = qtf * idf;
            queryNorm += qWeight * qWeight;

            if (docTf.TryGetValue(token, out var dtf))
            {
                var dWeight = dtf * idf;
                dotProduct += qWeight * dWeight;
            }
        }

        foreach (var (token, dtf) in docTf)
        {
            var df = dfMap.GetValueOrDefault(token, 0);
            var idf = Math.Log(1.0 + _totalDocs / (1.0 + df));
            var dWeight = dtf * idf;
            docNorm += dWeight * dWeight;
        }

        if (queryNorm == 0 || docNorm == 0) return 0;
        return dotProduct / (Math.Sqrt(queryNorm) * Math.Sqrt(docNorm));
    }

    // ── 토큰화 ──────────────────────────────────────────────────────────

    /// <summary>텍스트를 토큰으로 분할하고 빈도를 계산합니다. 스톱워드 제거 포함.</summary>
    private static Dictionary<string, int> Tokenize(string text)
    {
        var tf = new Dictionary<string, int>(StringComparer.OrdinalIgnoreCase);
        var words = Regex.Split(text, @"[^a-zA-Z0-9가-힣_]+")
            .SelectMany(SplitCamelCase)
            .Where(w => w.Length >= 2 && !StopWords.Contains(w));

        foreach (var word in words)
        {
            var lower = word.ToLowerInvariant();
            tf.TryGetValue(lower, out var count);
            tf[lower] = count + 1;
        }

        // 바이그램 추가 (구문 검색 품질 향상)
        var wordList = words.Select(w => w.ToLowerInvariant()).ToList();
        for (int i = 0; i < wordList.Count - 1; i++)
        {
            var bigram = $"{wordList[i]}_{wordList[i + 1]}";
            tf.TryGetValue(bigram, out var bc);
            tf[bigram] = bc + 1;
        }

        return tf;
    }

    private static IEnumerable<string> SplitCamelCase(string word)
    {
        if (string.IsNullOrEmpty(word)) yield break;
        var sb = new StringBuilder();
        foreach (var ch in word)
        {
            if (char.IsUpper(ch) && sb.Length > 0)
            {
                yield return sb.ToString();
                sb.Clear();
            }
            sb.Append(ch);
        }
        if (sb.Length > 0) yield return sb.ToString();
    }

    // ── Dispose ─────────────────────────────────────────────────────────

    public void Dispose()
    {
        _db?.Dispose();
        _db = null;
    }
}