Files
Finlytic/FinlyticFundamentals/Services/YahooFinanceScraper.cs
T

313 lines
14 KiB
C#

using System;
using System.Collections.Generic;
using System.Linq;
using System.Net.Http;
using System.Threading;
using System.Threading.Tasks;
using FinlyticCore.Clients;
using FinlyticCore.Dtos.Fundamentals;
using FinlyticCore.Dtos.Yahoo;
using FinlyticCore.Services;
using FinlyticCore.Services.Yahoo;
using FinlyticFundamentals.Database;
using FinlyticFundamentals.Util;
using Microsoft.Extensions.Logging;
namespace FinlyticFundamentals.Services;
public interface IYahooFinanceScraper
{
/// <summary>
/// Ermittelt den primären Börsenticker zu einer ISIN anhand von Börsenplatz-Prioritäten.
/// </summary>
Task<TickerInfoDto?> ResolveTickerFromIsinAsync(string isin, CancellationToken cancellationToken = default);
/// <summary>
/// Ermittelt alle gefundenen Börsenticker zu einer ISIN, sortiert nach Priorität.
/// </summary>
Task<List<TickerInfoDto>> ResolveAllTickersFromIsinAsync(string isin, CancellationToken cancellationToken = default);
/// <summary>
/// Ruft Fundamental- und Unternehmensdaten primär über die Yahoo Finance API ab
/// und fällt automatisch auf den Playwright HTML Scraper zurück, falls keine Daten vorhanden sind.
/// </summary>
/// <param name="symbolOrIsin">Das Tickersymbol (z. B. "MSFT") oder die ISIN.</param>
/// <param name="forceHtmlScrape">Erzwingt sofortiges HTML-Scraping ohne API-Vorprüfung.</param>
/// <param name="cancellationToken">Abbruch-Token.</param>
/// <returns>Das aggregierte <see cref="YahooQuoteSummaryModulesDto"/> oder <c>null</c>.</returns>
Task<YahooQuoteSummaryModulesDto?> GetQuoteSummaryModulesAsync(
string symbolOrIsin,
bool forceHtmlScrape = false,
CancellationToken cancellationToken = default);
}
public class YahooFinanceScraper : IYahooFinanceScraper
{
private readonly YahooFinanceClient _yahooApiClient;
private readonly IYahooFinanceHtmlClient _htmlScraperClient;
private readonly Microsoft.Extensions.Configuration.IConfiguration _configuration;
private readonly IFinlyticLogger<YahooFinanceScraper, FundamentalsDbContext> _finlyticLogger;
public YahooFinanceScraper(
YahooFinanceClient yahooApiClient,
IYahooFinanceHtmlClient htmlScraperClient,
Microsoft.Extensions.Configuration.IConfiguration configuration,
IFinlyticLogger<YahooFinanceScraper, FundamentalsDbContext> finlyticLogger)
{
_yahooApiClient = yahooApiClient;
_htmlScraperClient = htmlScraperClient;
_configuration = configuration;
_finlyticLogger = finlyticLogger;
}
/// <inheritdoc />
public async Task<TickerInfoDto?> ResolveTickerFromIsinAsync(string isin, CancellationToken cancellationToken = default)
{
var tickers = await ResolveAllTickersFromIsinAsync(isin, cancellationToken);
return tickers.FirstOrDefault();
}
/// <inheritdoc />
public async Task<List<TickerInfoDto>> ResolveAllTickersFromIsinAsync(string isin, CancellationToken cancellationToken = default)
{
if (string.IsNullOrWhiteSpace(isin)) return new List<TickerInfoDto>();
var cleanIsin = isin.Trim().ToUpperInvariant();
var symbols = new List<(string symbol, string exchange, int priority)>();
// Crypto / Trade Republic interne ISINs (beginnend mit 'X', z. B. XF000BTC0017)
if (cleanIsin.StartsWith("X", StringComparison.OrdinalIgnoreCase))
{
var (cryptoSubtitle, cryptoName) = await FinlyticCore.Utils.CryptoSubtitleResolver.ResolveCryptoInfoAsync(
cleanIsin, _configuration.GetConnectionString("DefaultConnection"), cancellationToken);
if (!string.IsNullOrWhiteSpace(cryptoSubtitle))
{
var cryptoEur = $"{cryptoSubtitle}-EUR";
var cryptoUsd = $"{cryptoSubtitle}-USD";
symbols.Add((cryptoEur, "Crypto", 0));
symbols.Add((cryptoUsd, "Crypto", 1));
try
{
var searchRes = await _yahooApiClient.SearchAsync(cryptoSubtitle, quotesCount: 10, cancellationToken: cancellationToken);
if (searchRes?.Quotes != null)
{
foreach (var q in searchRes.Quotes.Where(q => !string.IsNullOrEmpty(q.Symbol)))
{
if (!symbols.Any(s => s.symbol.Equals(q.Symbol, StringComparison.OrdinalIgnoreCase)))
{
symbols.Add((q.Symbol, q.Exchange ?? "Crypto", 2));
}
}
}
}
catch { }
await _finlyticLogger.LogInfoAsync(SettingKeys.FundamentalsChannel,
"[YahooFinanceScraper] Resolved Crypto ISIN {Isin} to {Symbol} using Subtitle {Sub}",
cleanIsin, cryptoEur, cryptoSubtitle);
return symbols
.OrderBy(s => s.priority)
.Select(s => new TickerInfoDto { Ticker = s.symbol, Exchange = s.exchange })
.ToList();
}
}
try
{
// 1. Suche via ISIN - der allererste Ticker von Yahoo Finance ist der absolute Primary Ticker
var primary = await _yahooApiClient.SearchAsync(cleanIsin, quotesCount: 20, cancellationToken: cancellationToken);
var quotes = primary?.Quotes ?? new List<YahooSearchQuoteDto>();
var validQuotes = quotes.Where(q => !string.IsNullOrEmpty(q.Symbol)).ToList();
if (validQuotes.Count > 0)
{
var first = validQuotes[0];
symbols.Add((first.Symbol, first.Exchange ?? string.Empty, 0));
foreach (var q in validQuotes.Skip(1))
{
if (!symbols.Any(s => s.symbol.Equals(q.Symbol, StringComparison.OrdinalIgnoreCase)))
{
symbols.Add((q.Symbol, q.Exchange ?? string.Empty, Math.Max(1, GetExchangePriority(q.Symbol, cleanIsin))));
}
}
}
// 2. Falls Ticker gefunden, aber mit Unternehmensname noch mehr Exchangeticker auffindbar sind
if (validQuotes.Count > 0)
{
var companyName = validQuotes[0].LongName ?? validQuotes[0].ShortName;
if (!string.IsNullOrWhiteSpace(companyName))
{
var secondary = await _yahooApiClient.SearchAsync(companyName, quotesCount: 20, cancellationToken: cancellationToken);
foreach (var q in secondary?.Quotes ?? new List<YahooSearchQuoteDto>())
{
if (!string.IsNullOrEmpty(q.Symbol) &&
!symbols.Any(s => s.symbol.Equals(q.Symbol, StringComparison.OrdinalIgnoreCase)))
{
symbols.Add((q.Symbol, q.Exchange ?? string.Empty, Math.Max(1, GetExchangePriority(q.Symbol, cleanIsin))));
}
}
}
}
}
catch (Exception ex)
{
await _finlyticLogger.LogWarningAsync(SettingKeys.YahooClientChannel, ex,
"[YahooFinanceScraper] Fehler beim Auflösen des Tickers für ISIN '{Isin}'", cleanIsin);
}
return symbols
.OrderBy(s => s.priority)
.Select(s => new TickerInfoDto(){Ticker = s.symbol, Exchange = s.exchange})
.ToList();
}
/// <inheritdoc />
public async Task<YahooQuoteSummaryModulesDto?> GetQuoteSummaryModulesAsync(
string symbolOrIsin,
bool forceHtmlScrape = false,
CancellationToken cancellationToken = default)
{
if (string.IsNullOrWhiteSpace(symbolOrIsin)) return null;
var symbol = symbolOrIsin.Trim().ToUpperInvariant();
// Falls eine ISIN übergeben wurde, zuerst Ticker auflösen
if (IsIsin(symbol))
{
var resolvedTicker = await ResolveTickerFromIsinAsync(symbol, cancellationToken);
if (resolvedTicker != null)
{
symbol = resolvedTicker.Ticker;
}
}
YahooQuoteSummaryModulesDto? apiModules = null;
// -------------------------------------------------------------
// 1. PRIMÄRE DATENQUELLE: Yahoo Finance API (Cookie/Crumb)
// -------------------------------------------------------------
if (!forceHtmlScrape)
{
try
{
await _finlyticLogger.LogInfoAsync(SettingKeys.YahooClientChannel,
"[YahooFinanceScraper] Starte primären API-Abruf für '{Symbol}'...", symbol);
var apiResponse = await _yahooApiClient.GetFullQuoteSummaryAsync(symbol, cancellationToken);
apiModules = apiResponse?.QuoteSummary?.Result?.FirstOrDefault();
if (apiModules != null && HasSufficientData(apiModules))
{
await _finlyticLogger.LogInfoAsync(SettingKeys.YahooClientChannel,
"[YahooFinanceScraper] Erfolgreich Daten über API bezogen für '{Symbol}'.", symbol);
return apiModules;
}
await _finlyticLogger.LogWarningAsync(SettingKeys.YahooClientChannel,
"[YahooFinanceScraper] API lieferte unvollständige Daten für '{Symbol}'. Initiiere Fallback...", symbol);
}
catch (Exception ex)
{
await _finlyticLogger.LogWarningAsync(SettingKeys.YahooClientChannel, ex,
"[YahooFinanceScraper] API-Abruf fehlgeschlagen für '{Symbol}'. Wechsle zu Scraper...", symbol);
}
}
// -------------------------------------------------------------
// 2. FALLBACK DATENQUELLE: Playwright HTML Scraper
// -------------------------------------------------------------
YahooQuoteSummaryModulesDto? htmlModules = null;
try
{
await _finlyticLogger.LogInfoAsync(SettingKeys.YahooClientChannel,
"[YahooFinanceScraper] Starte HTML-Scraper Fallback für '{Symbol}'...", symbol);
htmlModules = await _htmlScraperClient.ScrapeQuoteSummaryModulesAsync(symbol, cancellationToken);
}
catch (Exception ex)
{
await _finlyticLogger.LogErrorAsync(SettingKeys.YahooClientChannel, ex,
"[YahooFinanceScraper] HTML-Scraper Fallback ebenfalls fehlgeschlagen für '{Symbol}'.", symbol);
}
// -------------------------------------------------------------
// 3. Zusammenführen (Merge API & HTML Fallback)
// -------------------------------------------------------------
if (apiModules == null) return htmlModules;
if (htmlModules == null) return apiModules;
return MergeModules(apiModules, htmlModules);
}
/// <summary>
/// Prüft, ob das Modul-DTO die wesentlichen Fundamentalblöcke enthält.
/// </summary>
private static bool HasSufficientData(YahooQuoteSummaryModulesDto modules)
{
return modules.SummaryDetail != null ||
modules.FinancialData != null ||
modules.DefaultKeyStatistics != null;
}
/// <summary>
/// Führt API- und Scraper-Daten zusammen, damit Lücken in API-Responses geschlossen werden.
/// </summary>
private static YahooQuoteSummaryModulesDto MergeModules(
YahooQuoteSummaryModulesDto primary,
YahooQuoteSummaryModulesDto secondary)
{
return new YahooQuoteSummaryModulesDto(
QuoteType: primary.QuoteType ?? secondary.QuoteType,
AssetProfile: primary.AssetProfile ?? secondary.AssetProfile,
FinancialData: primary.FinancialData ?? secondary.FinancialData,
DefaultKeyStatistics: primary.DefaultKeyStatistics ?? secondary.DefaultKeyStatistics,
SummaryDetail: primary.SummaryDetail ?? secondary.SummaryDetail,
IncomeStatementHistory: primary.IncomeStatementHistory ?? secondary.IncomeStatementHistory,
IncomeStatementHistoryQuarterly: primary.IncomeStatementHistoryQuarterly ?? secondary.IncomeStatementHistoryQuarterly,
BalanceSheetHistory: primary.BalanceSheetHistory ?? secondary.BalanceSheetHistory,
BalanceSheetHistoryQuarterly: primary.BalanceSheetHistoryQuarterly ?? secondary.BalanceSheetHistoryQuarterly,
CashflowStatementHistory: primary.CashflowStatementHistory ?? secondary.CashflowStatementHistory,
CashflowStatementHistoryQuarterly: primary.CashflowStatementHistoryQuarterly ?? secondary.CashflowStatementHistoryQuarterly,
CalendarEvents: primary.CalendarEvents ?? secondary.CalendarEvents
);
}
private static bool IsIsin(string value)
{
return value.Length == 12 &&
char.IsLetter(value[0]) &&
char.IsLetter(value[1]) &&
value.All(char.IsLetterOrDigit);
}
private static int GetExchangePriority(string symbol, string isin)
{
bool isGermanIsin = isin.StartsWith("DE", StringComparison.OrdinalIgnoreCase);
if (isGermanIsin)
{
if (symbol.EndsWith(".DE", StringComparison.OrdinalIgnoreCase)) return 1; // Xetra
if (symbol.EndsWith(".F", StringComparison.OrdinalIgnoreCase)) return 2; // Frankfurt
if (symbol.EndsWith(".STU", StringComparison.OrdinalIgnoreCase)) return 3; // Stuttgart
if (symbol.EndsWith(".SG", StringComparison.OrdinalIgnoreCase)) return 4; // Stuttgart (alt)
if (symbol.EndsWith(".HM", StringComparison.OrdinalIgnoreCase)) return 5; // Hamburg
if (!symbol.Contains('.')) return 6; // US Primary
}
else
{
if (!symbol.Contains('.')) return 1; // US Primary (NASDAQ, NYSE)
if (symbol.EndsWith(".DE", StringComparison.OrdinalIgnoreCase)) return 2; // Xetra
if (symbol.EndsWith(".F", StringComparison.OrdinalIgnoreCase)) return 3; // Frankfurt
if (symbol.EndsWith(".L", StringComparison.OrdinalIgnoreCase)) return 4; // London
if (symbol.EndsWith(".PA", StringComparison.OrdinalIgnoreCase)) return 5; // Paris
}
return 10;
}
}