public void DoesNotLeakMemory()
{
var symbol = Symbols.AAPL;
var config = new SubscriptionDataConfig(typeof(TradeBar), symbol, Resolution.Daily, TimeZones.NewYork, TimeZones.NewYork, false, false, false, false, TickType.Trade, false);
var security = new Security(
SecurityExchangeHours.AlwaysOpen(TimeZones.NewYork),
config,
new Cash(CashBook.AccountCurrency, 0, 1),
SymbolProperties.GetDefault(CashBook.AccountCurrency),
ErrorCurrencyConverter.Instance
);
var mapFileProvider = new LocalDiskMapFileProvider();
var factorFileProvider = new LocalDiskFactorFileProvider(mapFileProvider);
var mapFileResolver = mapFileProvider.Get(security.Symbol.ID.Market);
var fileProvider = new DefaultDataProvider();
var factory = new BaseDataSubscriptionEnumeratorFactory(false, mapFileResolver, factorFileProvider);
GC.Collect();
var ramUsageBeforeLoop = OS.TotalPhysicalMemoryUsed;
var date = new DateTime(1998, 1, 1);
const int iterations = 1000;
for (var i = 0; i < iterations; i++)
{
var request = new SubscriptionRequest(false, null, security, config, date, date);
using (var enumerator = factory.CreateEnumerator(request, fileProvider))
{
enumerator.MoveNext();
}
date = date.AddDays(1);
}
GC.Collect();
var ramUsageAfterLoop = OS.TotalPhysicalMemoryUsed;
Log.Trace($"RAM usage - before: {ramUsageBeforeLoop} MB, after: {ramUsageAfterLoop} MB");
Assert.IsTrue(ramUsageAfterLoop - ramUsageBeforeLoop < 10);
}
public void FindsFactorFilesWithErrors()
{
var provider = new LocalDiskFactorFileProvider();
var factorFileFolder = Path.Combine(Globals.DataFolder, "equity", QuantConnect.Market.USA, "factor_files");
foreach (var fileName in Directory.EnumerateFiles(factorFileFolder))
{
var ticker = Path.GetFileNameWithoutExtension(fileName).ToUpper(CultureInfo.InvariantCulture);
var symbol = Symbol.Create(ticker, SecurityType.Equity, QuantConnect.Market.USA);
try
{
provider.Get(symbol);
}
catch (Exception exception)
{
Console.WriteLine(ticker + ": " + exception.Message);
}
}
}
public void TestsFileSystemDataFeedSpeed()
{
var job = new BacktestNodePacket();
var resultHandler = new BacktestingResultHandler();
var mapFileProvider = new LocalDiskMapFileProvider();
var factorFileProvider = new LocalDiskFactorFileProvider(mapFileProvider);
var dataProvider = new DefaultDataProvider();
var algorithm = new BenchmarkTest();
var feed = new FileSystemDataFeed();
feed.Initialize(algorithm, job, resultHandler, mapFileProvider, factorFileProvider, dataProvider);
algorithm.Initialize();
var feedThreadStarted = new ManualResetEvent(false);
Task.Factory.StartNew(() =>
{
feedThreadStarted.Set();
feed.Run();
});
feedThreadStarted.WaitOne();
var stopwatch = Stopwatch.StartNew();
var lastMonth = -1;
var count = 0;
foreach (var timeSlice in feed)
{
if (timeSlice.Time.Month != lastMonth)
{
Console.WriteLine(DateTime.Now + " - Time: " + timeSlice.Time);
lastMonth = timeSlice.Time.Month;
}
count++;
}
Console.WriteLine("Count: " + count);
stopwatch.Stop();
Console.WriteLine("Elapsed time: " + stopwatch.Elapsed);
}
public override void Initialize()
{
SetStartDate(2014, 6, 5); //Set Start Date
SetEndDate(2014, 6, 5); //Set End Date
UniverseSettings.DataNormalizationMode = DataNormalizationMode.SplitAdjusted;
_aapl = AddEquity(Ticker, Resolution.Minute).Symbol;
var dataProvider =
Composer.Instance.GetExportedValueByTypeName <IDataProvider>(Config.Get("data-provider",
"DefaultDataProvider"));
var mapFileProvider = new LocalDiskMapFileProvider();
mapFileProvider.Initialize(dataProvider);
var factorFileProvider = new LocalDiskFactorFileProvider();
factorFileProvider.Initialize(mapFileProvider, dataProvider);
_factorFile = factorFileProvider.Get(_aapl) as CorporateFactorProvider;
}
public void TestsFileSystemDataFeedSpeed()
{
var job = new BacktestNodePacket();
var resultHandler = new BacktestingResultHandler();
var mapFileProvider = new LocalDiskMapFileProvider();
var factorFileProvider = new LocalDiskFactorFileProvider(mapFileProvider);
var dataProvider = new DefaultDataProvider();
var algorithm = PerformanceBenchmarkAlgorithms.SingleSecurity_Second;
var feed = new FileSystemDataFeed();
var dataManager = new DataManager(feed, new UniverseSelection(feed, algorithm), algorithm.Settings, algorithm.TimeKeeper);
algorithm.SubscriptionManager.SetDataManager(dataManager);
feed.Initialize(algorithm, job, resultHandler, mapFileProvider, factorFileProvider, dataProvider, dataManager);
algorithm.Initialize();
algorithm.PostInitialize();
var count = 0;
var stopwatch = Stopwatch.StartNew();
var lastMonth = algorithm.StartDate.Month;
foreach (var timeSlice in feed)
{
if (timeSlice.Time.Month != lastMonth)
{
var elapsed = stopwatch.Elapsed.TotalSeconds;
var thousands = count / 1000d;
Console.WriteLine($"{DateTime.Now} - Time: {timeSlice.Time}: KPS: {thousands/elapsed}");
lastMonth = timeSlice.Time.Month;
}
count++;
}
Console.WriteLine("Count: " + count);
stopwatch.Stop();
feed.Exit();
Console.WriteLine($"Elapsed time: {stopwatch.Elapsed} KPS: {count/1000d/stopwatch.Elapsed.TotalSeconds}");
}
/// <summary>
/// Iterates over each equity directory and aggregates the data into the coarse file
/// </summary>
/// <param name="dataDirectory">The Lean /Data directory</param>
/// <param name="ignoreMaplessSymbols">Ignore symbols without a QuantQuote map file.</param>
public static IEnumerable <string> ProcessEquityDirectories(string dataDirectory, bool ignoreMaplessSymbols)
{
var exclusions = ReadExclusionsFile(ExclusionsFile);
var equity = Path.Combine(dataDirectory, "equity");
foreach (var directory in Directory.EnumerateDirectories(equity))
{
var dailyFolder = Path.Combine(directory, "daily");
var mapFileFolder = Path.Combine(directory, "map_files");
var coarseFolder = Path.Combine(directory, "fundamental", "coarse");
if (!Directory.Exists(coarseFolder))
{
Directory.CreateDirectory(coarseFolder);
}
var factorFileProvider = new LocalDiskFactorFileProvider();
var files = ProcessDailyFolder(dailyFolder, coarseFolder, MapFileResolver.Create(mapFileFolder), factorFileProvider, exclusions, ignoreMaplessSymbols);
foreach (var file in files)
{
yield return(file);
}
}
}
/// <summary>
/// Iterates each daily file in the specified <paramref name="dailyFolder"/> and adds a line for each
/// day to the approriate coarse file
/// </summary>
/// <param name="dailyFolder">The folder with daily data</param>
/// <param name="coarseFolder">The coarse output folder</param>
/// <param name="mapFileResolver"></param>
/// <param name="exclusions">The symbols to be excluded from processing</param>
/// <param name="ignoreMapless">Ignore the symbols without a map file.</param>
/// <param name="startDate">The starting date for processing</param>
/// <param name="symbolResolver">Function used to provide symbol resolution. Default resolution uses the zip file name to resolve
/// the symbol, specify null for this behavior.</param>
/// <returns>A collection of the generated coarse files</returns>
public static ICollection <string> ProcessDailyFolder(string dailyFolder, string coarseFolder, MapFileResolver mapFileResolver, HashSet <string> exclusions, bool ignoreMapless, DateTime startDate, Func <string, string> symbolResolver = null)
{
const decimal scaleFactor = 10000m;
Log.Trace("Processing: {0}", dailyFolder);
var start = DateTime.UtcNow;
// load map files into memory
var symbols = 0;
var maplessCount = 0;
var dates = new HashSet <DateTime>();
// instead of opening/closing these constantly, open them once and dispose at the end (~3x speed improvement)
var writers = new Dictionary <string, StreamWriter>();
var dailyFolderDirectoryInfo = new DirectoryInfo(dailyFolder).Parent;
if (dailyFolderDirectoryInfo == null)
{
throw new Exception("Unable to resolve market for daily folder: " + dailyFolder);
}
var market = dailyFolderDirectoryInfo.Name.ToLower();
var fundamentalDirectoryInfo = new DirectoryInfo(coarseFolder).Parent;
if (fundamentalDirectoryInfo == null)
{
throw new Exception("Unable to resolve fundamental path for coarse folder: " + coarseFolder);
}
var fineFundamentalFolder = Path.Combine(fundamentalDirectoryInfo.FullName, "fine");
var mapFileProvider = new LocalDiskMapFileProvider();
var factorFileProvider = new LocalDiskFactorFileProvider(mapFileProvider);
// open up each daily file to get the values and append to the daily coarse files
foreach (var file in Directory.EnumerateFiles(dailyFolder))
{
try
{
var symbol = Path.GetFileNameWithoutExtension(file);
if (symbol == null)
{
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Unable to resolve symbol from file: {0}", file);
continue;
}
if (symbolResolver != null)
{
symbol = symbolResolver(symbol);
}
symbol = symbol.ToUpper();
if (exclusions.Contains(symbol))
{
Log.Trace("Excluded symbol: {0}", symbol);
continue;
}
// check if symbol has any fine fundamental data
var firstFineSymbolDate = DateTime.MaxValue;
if (Directory.Exists(fineFundamentalFolder))
{
var fineSymbolFolder = Path.Combine(fineFundamentalFolder, symbol.ToLower());
var firstFineSymbolFileName = Directory.Exists(fineSymbolFolder) ? Directory.GetFiles(fineSymbolFolder).OrderBy(x => x).FirstOrDefault() : string.Empty;
if (firstFineSymbolFileName.Length > 0)
{
firstFineSymbolDate = DateTime.ParseExact(Path.GetFileNameWithoutExtension(firstFineSymbolFileName), "yyyyMMdd", CultureInfo.InvariantCulture);
}
}
ZipFile zip;
using (var reader = Compression.Unzip(file, out zip))
{
var checkedForMapFile = false;
symbols++;
string line;
while ((line = reader.ReadLine()) != null)
{
//20150625.csv
var csv = line.Split(',');
var date = DateTime.ParseExact(csv[0], DateFormat.TwelveCharacter, CultureInfo.InvariantCulture);
// spin past old data
if (date < startDate)
{
continue;
}
if (ignoreMapless && !checkedForMapFile)
{
checkedForMapFile = true;
if (!mapFileResolver.ResolveMapFile(symbol, date).Any())
{
// if the resolved map file has zero entries then it's a mapless symbol
maplessCount++;
break;
}
}
var close = decimal.Parse(csv[4]) / scaleFactor;
var volume = long.Parse(csv[5]);
var dollarVolume = close * volume;
var coarseFile = Path.Combine(coarseFolder, date.ToString("yyyyMMdd") + ".csv");
dates.Add(date);
// try to resolve a map file and if found, regen the sid
var sid = SecurityIdentifier.GenerateEquity(SecurityIdentifier.DefaultDate, symbol, market);
var mapFile = mapFileResolver.ResolveMapFile(symbol, date);
if (!mapFile.IsNullOrEmpty())
{
// if available, us the permtick in the coarse files, because of this, we need
// to update the coarse files each time new map files are added/permticks change
sid = SecurityIdentifier.GenerateEquity(mapFile.FirstDate, mapFile.OrderBy(x => x.Date).First().MappedSymbol, market);
}
if (mapFile == null && ignoreMapless)
{
// if we're ignoring mapless files then we should always be able to resolve this
Log.Error(string.Format("CoarseGenerator.ProcessDailyFolder(): Unable to resolve map file for {0} as of {1}", symbol, date.ToShortDateString()));
continue;
}
// check if symbol has fine fundamental data for the current date
var hasFundamentalDataForDate = date >= firstFineSymbolDate;
// get price and split factors from factor files
var leanSymbol = new Symbol(sid, symbol);
var factorFile = factorFileProvider.Get(leanSymbol);
var factorFileRow = factorFile?.GetScalingFactors(date);
var priceFactor = factorFileRow?.PriceFactor ?? 1m;
var splitFactor = factorFileRow?.SplitFactor ?? 1m;
// sid,symbol,close,volume,dollar volume,has fundamental data,price factor,split factor
var coarseFileLine = $"{sid},{symbol},{close},{volume},{Math.Truncate(dollarVolume)},{hasFundamentalDataForDate},{priceFactor},{splitFactor}";
StreamWriter writer;
if (!writers.TryGetValue(coarseFile, out writer))
{
writer = new StreamWriter(new FileStream(coarseFile, FileMode.Create, FileAccess.Write, FileShare.Write));
writers[coarseFile] = writer;
}
writer.WriteLine(coarseFileLine);
}
}
if (symbols % 1000 == 0)
{
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Completed processing {0} symbols. Current elapsed: {1} seconds", symbols, (DateTime.UtcNow - start).TotalSeconds.ToString("0.00"));
}
}
catch (Exception err)
{
// log the error and continue with the process
Log.Error(err.ToString());
}
}
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Saving {0} coarse files to disk", dates.Count);
// dispose all the writers at the end of processing
foreach (var writer in writers)
{
writer.Value.Dispose();
}
var stop = DateTime.UtcNow;
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Processed {0} symbols into {1} coarse files in {2} seconds", symbols, dates.Count, (stop - start).TotalSeconds.ToString("0.00"));
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Excluded {0} mapless symbols.", maplessCount);
return(writers.Keys);
}
