private static FactorFile GetFactorFileToUse(
SubscriptionDataConfig config,
IFactorFileProvider factorFileProvider)
{
var factorFileToUse = new FactorFile(config.Symbol.Value, new List <FactorFileRow>());
if (!config.IsCustomData &&
config.SecurityType == SecurityType.Equity)
{
try
{
var factorFile = factorFileProvider.Get(config.Symbol);
if (factorFile != null)
{
factorFileToUse = factorFile;
}
}
catch (Exception err)
{
Log.Error(err, "CorporateEventEnumeratorFactory.GetFactorFileToUse(): Factors File: "
+ config.Symbol.ID + ": ");
}
}
return(factorFileToUse);
}
/// <summary>
/// Advances the enumerator to the next element of the collection.
/// </summary>
/// <returns>
/// True if the enumerator was successfully advanced to the next element;
/// False if the enumerator has passed the end of the collection.
/// </returns>
public bool MoveNext()
{
var underlyingReturnValue = _rawDataEnumerator.MoveNext();
Current = _rawDataEnumerator.Current;
if (underlyingReturnValue &&
Current != null &&
_factorFileProvider != null &&
_config.DataNormalizationMode != DataNormalizationMode.Raw)
{
if (Current.Time >= _nextTradableDate)
{
_factorFile = _factorFileProvider.Get(_config.Symbol);
_config.PriceScaleFactor = _factorFile.GetPriceScale(Current.Time.Date, _config.DataNormalizationMode, _config.ContractDepthOffset, _config.DataMappingMode);
// update factor files every day
_nextTradableDate = Current.Time.Date.AddDays(1);
if (_liveMode)
{
// in live trading we add a offset to make sure new factor files are available
_nextTradableDate = _nextTradableDate.Add(Time.LiveAuxiliaryDataOffset);
}
}
Current = Current.Normalize(_config.PriceScaleFactor, _config.DataNormalizationMode, _config.SumOfDividends);
}
return(underlyingReturnValue);
}
/// <summary>
/// Initializes this instance
/// </summary>
/// <param name="config">The <see cref="SubscriptionDataConfig"/></param>
/// <param name="factorFileProvider">The factor file provider to use</param>
/// <param name="mapFileProvider">The <see cref="Data.Auxiliary.MapFile"/> provider to use</param>
/// <param name="startTime">Start date for the data request</param>
public void Initialize(
SubscriptionDataConfig config,
IFactorFileProvider factorFileProvider,
IMapFileProvider mapFileProvider,
DateTime startTime)
{
_config = config;
_mapFile = mapFileProvider.ResolveMapFile(_config);
_factorFile = factorFileProvider.Get(_config.Symbol) as CorporateFactorProvider;
}
/// <summary>
/// Advances the enumerator to the next element of the collection.
/// </summary>
/// <returns>
/// True if the enumerator was successfully advanced to the next element;
/// False if the enumerator has passed the end of the collection.
/// </returns>
public bool MoveNext()
{
var underlyingReturnValue = _rawDataEnumerator.MoveNext();
Current = _rawDataEnumerator.Current;
if (underlyingReturnValue &&
Current != null &&
_factorFileProvider != null &&
_config.DataNormalizationMode != DataNormalizationMode.Raw)
{
if (Current.Time.Date > _lastTradableDate)
{
_factorFile = _factorFileProvider.Get(_config.Symbol);
_lastTradableDate = Current.Time.Date;
_config.PriceScaleFactor = _factorFile.GetPriceScale(_lastTradableDate, _config.DataNormalizationMode, _config.ContractDepthOffset, _config.DataMappingMode);
}
Current = Current.Normalize(_config.PriceScaleFactor, _config.DataNormalizationMode, _config.SumOfDividends);
}
return(underlyingReturnValue);
}
/// <summary>
/// Subscription data reader takes a subscription request, loads the type, accepts the data source and enumerate on the results.
/// </summary>
/// <param name="config">Subscription configuration object</param>
/// <param name="periodStart">Start date for the data request/backtest</param>
/// <param name="periodFinish">Finish date for the data request/backtest</param>
/// <param name="resultHandler">Result handler used to push error messages and perform sampling on skipped days</param>
/// <param name="mapFileResolver">Used for resolving the correct map files</param>
/// <param name="factorFileProvider">Used for getting factor files</param>
/// <param name="tradeableDates">Defines the dates for which we'll request data, in order, in the security's exchange time zone</param>
/// <param name="isLiveMode">True if we're in live mode, false otherwise</param>
/// <param name="includeAuxilliaryData">True if we want to emit aux data, false to only emit price data</param>
public SubscriptionDataReader(SubscriptionDataConfig config,
DateTime periodStart,
DateTime periodFinish,
IResultHandler resultHandler,
MapFileResolver mapFileResolver,
IFactorFileProvider factorFileProvider,
IEnumerable <DateTime> tradeableDates,
bool isLiveMode,
bool includeAuxilliaryData = true)
{
//Save configuration of data-subscription:
_config = config;
_auxiliaryData = new Queue <BaseData>();
//Save Start and End Dates:
_periodStart = periodStart;
_periodFinish = periodFinish;
//Save access to securities
_isLiveMode = isLiveMode;
_includeAuxilliaryData = includeAuxilliaryData;
//Save the type of data we'll be getting from the source.
//Create the dynamic type-activators:
var objectActivator = ObjectActivator.GetActivator(config.Type);
_resultHandler = resultHandler;
_tradeableDates = tradeableDates.GetEnumerator();
if (objectActivator == null)
{
_resultHandler.ErrorMessage("Custom data type '" + config.Type.Name + "' missing parameterless constructor E.g. public " + config.Type.Name + "() { }");
_endOfStream = true;
return;
}
//Create an instance of the "Type":
var userObj = objectActivator.Invoke(new object[] {});
_dataFactory = userObj as BaseData;
//If its quandl set the access token in data factory:
var quandl = _dataFactory as Quandl;
if (quandl != null)
{
if (!Quandl.IsAuthCodeSet)
{
Quandl.SetAuthCode(Config.Get("quandl-auth-token"));
}
}
_factorFile = new FactorFile(config.Symbol.Value, new List <FactorFileRow>());
_mapFile = new MapFile(config.Symbol.Value, new List <MapFileRow>());
// load up the map and factor files for equities
if (!config.IsCustomData && config.SecurityType == SecurityType.Equity)
{
try
{
var mapFile = mapFileResolver.ResolveMapFile(config.Symbol.ID.Symbol, config.Symbol.ID.Date);
// only take the resolved map file if it has data, otherwise we'll use the empty one we defined above
if (mapFile.Any())
{
_mapFile = mapFile;
}
var factorFile = factorFileProvider.Get(_config.Symbol);
_hasScaleFactors = factorFile != null;
if (_hasScaleFactors)
{
_factorFile = factorFile;
}
}
catch (Exception err)
{
Log.Error(err, "Fetching Price/Map Factors: " + config.Symbol.ID + ": ");
}
}
_subscriptionFactoryEnumerator = ResolveDataEnumerator(true);
}
/// <summary>
/// Initializes the <see cref="SubscriptionDataReader"/> instance
/// </summary>
/// <remarks>Should be called after all consumers of <see cref="NewTradableDate"/> event are set,
/// since it will produce events.</remarks>
public void Initialize()
{
if (_initialized)
{
return;
}
//Save the type of data we'll be getting from the source.
try
{
_dataFactory = _config.GetBaseDataInstance();
}
catch (ArgumentException exception)
{
OnInvalidConfigurationDetected(new InvalidConfigurationDetectedEventArgs(_config.Symbol, exception.Message));
_endOfStream = true;
return;
}
//If its quandl set the access token in data factory:
var quandl = _dataFactory as Quandl;
if (quandl != null)
{
if (!Quandl.IsAuthCodeSet)
{
Quandl.SetAuthCode(Config.Get("quandl-auth-token"));
}
}
// If Tiingo data, set the access token in data factory
var tiingo = _dataFactory as TiingoPrice;
if (tiingo != null)
{
if (!Tiingo.IsAuthCodeSet)
{
Tiingo.SetAuthCode(Config.Get("tiingo-auth-token"));
}
}
// If USEnergyAPI data, set the access token in data factory
var energyInformation = _dataFactory as USEnergyAPI;
if (energyInformation != null)
{
if (!USEnergyAPI.IsAuthCodeSet)
{
USEnergyAPI.SetAuthCode(Config.Get("us-energy-information-auth-token"));
}
}
// If Fred data, set the access token in data factory
var fred = _dataFactory as FredApi;
if (fred != null)
{
if (!FredApi.IsAuthCodeSet)
{
FredApi.SetAuthCode(Config.Get("fred-auth-token"));
}
}
_factorFile = new FactorFile(_config.Symbol.Value, new List <FactorFileRow>());
_mapFile = new MapFile(_config.Symbol.Value, new List <MapFileRow>());
// load up the map files for equities, options, and custom data if it supports it.
// Only load up factor files for equities
if (_dataFactory.RequiresMapping())
{
try
{
var mapFile = _mapFileResolver.ResolveMapFile(_config.Symbol, _config.Type);
// only take the resolved map file if it has data, otherwise we'll use the empty one we defined above
if (mapFile.Any())
{
_mapFile = mapFile;
}
if (!_config.IsCustomData && !_config.SecurityType.IsOption())
{
var factorFile = _factorFileProvider.Get(_config.Symbol);
_hasScaleFactors = factorFile != null;
if (_hasScaleFactors)
{
_factorFile = factorFile;
// if factor file has minimum date, update start period if before minimum date
if (!_isLiveMode && _factorFile != null && _factorFile.FactorFileMinimumDate.HasValue)
{
if (_periodStart < _factorFile.FactorFileMinimumDate.Value)
{
_periodStart = _factorFile.FactorFileMinimumDate.Value;
OnNumericalPrecisionLimited(
new NumericalPrecisionLimitedEventArgs(_config.Symbol,
$"[{_config.Symbol.Value}, {_factorFile.FactorFileMinimumDate.Value.ToShortDateString()}]"));
}
}
}
if (_periodStart < mapFile.FirstDate)
{
_periodStart = mapFile.FirstDate;
OnStartDateLimited(
new StartDateLimitedEventArgs(_config.Symbol,
$"[{_config.Symbol.Value}," +
$" {mapFile.FirstDate.ToString("yyyy-MM-dd", CultureInfo.InvariantCulture)}]"));
}
}
}
catch (Exception err)
{
Log.Error(err, "Fetching Price/Map Factors: " + _config.Symbol.ID + ": ");
}
}
_delistingDate = _config.Symbol.GetDelistingDate(_mapFile);
// adding a day so we stop at EOD
_delistingDate = _delistingDate.AddDays(1);
UpdateDataEnumerator(true);
_initialized = true;
}
/// <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="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, IFactorFileProvider factorFileProvider,
HashSet <string> exclusions, bool ignoreMapless, 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(Globals.DataFolder, SecurityType.Equity.ToLower(), Market.USA.ToLower(), "fundamental", "fine");
var mapFileProvider = new LocalDiskMapFileProvider();
// open up each daily file to get the values and append to the daily coarse files
foreach (var file in Directory.EnumerateFiles(dailyFolder, "*.zip"))
{
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);
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 = $"{sid},{symbol},{close},{volume},{Math.Truncate(dollarVolume)},{hasFundamentalDataForDate},{priceFactor},{splitFactor}";
StreamWriter writer;
if (!writers.TryGetValue(coarseFile, out writer))
{
writer = new StreamWriter(new FileStream(coarseFile, FileMode.Create, FileAccess.Write, FileShare.Write));
writers[coarseFile] = writer;
}
writer.WriteLine(coarseFileLine);
}
}
if (symbols % 1000 == 0)
{
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Completed processing {0} symbols. Current elapsed: {1} seconds", symbols, (DateTime.UtcNow - start).TotalSeconds.ToString("0.00"));
}
}
catch (Exception err)
{
// log the error and continue with the process
Log.Error(err.ToString());
}
}
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Saving {0} coarse files to disk", dates.Count);
// dispose all the writers at the end of processing
foreach (var writer in writers)
{
writer.Value.Dispose();
}
var stop = DateTime.UtcNow;
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Processed {0} symbols into {1} coarse files in {2} seconds", symbols, dates.Count, (stop - start).TotalSeconds.ToString("0.00"));
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Excluded {0} mapless symbols.", maplessCount);
return(writers.Keys);
}
/// <summary>
/// Runs this instance.
/// </summary>
/// <returns></returns>
public bool Run()
{
var startTime = DateTime.UtcNow;
var success = true;
Log.Trace($"CoarseUniverseGeneratorProgram.ProcessDailyFolder(): Processing: {_dailyDataFolder.FullName}");
var symbolsProcessed = 0;
var filesRead = 0;
var dailyFilesNotFound = 0;
var coarseFilesGenerated = 0;
var mapFileResolver = _mapFileProvider.Get(_market);
var blackListedTickers = new HashSet <string>();
if (_blackListedTickersFile.Exists)
{
blackListedTickers = File.ReadAllLines(_blackListedTickersFile.FullName).ToHashSet();
}
var marketFolder = _dailyDataFolder.Parent;
var fineFundamentalFolder = new DirectoryInfo(Path.Combine(marketFolder.FullName, "fundamental", "fine"));
if (!fineFundamentalFolder.Exists)
{
Log.Error($"CoarseUniverseGenerator.Run(): FAIL, Fine Fundamental folder not found at {fineFundamentalFolder}! ");
return(false);
}
var securityIdentifierContexts = PopulateSidContex(mapFileResolver, blackListedTickers);
var dailyPricesByTicker = new ConcurrentDictionary <string, List <TradeBar> >();
var outputCoarseContent = new ConcurrentDictionary <DateTime, List <string> >();
var parallelOptions = new ParallelOptions {
MaxDegreeOfParallelism = Environment.ProcessorCount / 2
};
try
{
Parallel.ForEach(securityIdentifierContexts, parallelOptions, sidContext =>
{
var symbol = new Symbol(sidContext.SID, sidContext.LastTicker);
var symbolCount = Interlocked.Increment(ref symbolsProcessed);
Log.Debug($"CoarseUniverseGeneratorProgram.Run(): Processing {symbol}");
var factorFile = _factorFileProvider.Get(symbol);
// Populate dailyPricesByTicker with all daily data by ticker for all tickers of this security.
foreach (var ticker in sidContext.Tickers)
{
var dailyFile = new FileInfo(Path.Combine(_dailyDataFolder.FullName, $"{ticker}.zip"));
if (!dailyFile.Exists)
{
Log.Error($"CoarseUniverseGeneratorProgram.Run(): {dailyFile} not found!");
Interlocked.Increment(ref dailyFilesNotFound);
continue;
}
if (!dailyPricesByTicker.ContainsKey(ticker))
{
dailyPricesByTicker.AddOrUpdate(ticker, ParseDailyFile(dailyFile));
Interlocked.Increment(ref filesRead);
}
}
// Look for daily data for each ticker of the actual security
for (int mapFileRowIndex = sidContext.MapFileRows.Length - 1; mapFileRowIndex >= 1; mapFileRowIndex--)
{
var ticker = sidContext.MapFileRows[mapFileRowIndex].Item2.ToLowerInvariant();
var endDate = sidContext.MapFileRows[mapFileRowIndex].Item1;
var startDate = sidContext.MapFileRows[mapFileRowIndex - 1].Item1;
List <TradeBar> tickerDailyData;
if (!dailyPricesByTicker.TryGetValue(ticker, out tickerDailyData))
{
Log.Error($"CoarseUniverseGeneratorProgram.Run(): Daily data for ticker {ticker.ToUpperInvariant()} not found!");
continue;
}
var tickerFineFundamentalFolder = Path.Combine(fineFundamentalFolder.FullName, ticker);
var fineAvailableDates = Enumerable.Empty <DateTime>();
if (Directory.Exists(tickerFineFundamentalFolder))
{
fineAvailableDates = Directory.GetFiles(tickerFineFundamentalFolder, "*.zip")
.Select(f => DateTime.ParseExact(Path.GetFileNameWithoutExtension(f), DateFormat.EightCharacter, CultureInfo.InvariantCulture))
.ToList();
}
// Get daily data only for the time the ticker was
foreach (var tradeBar in tickerDailyData.Where(tb => tb.Time >= startDate && tb.Time <= endDate))
{
var coarseRow = GenerateFactorFileRow(ticker, sidContext, factorFile, tradeBar, fineAvailableDates, fineFundamentalFolder);
outputCoarseContent.AddOrUpdate(tradeBar.Time,
new List <string> {
coarseRow
},
(time, list) =>
{
lock (list)
{
list.Add(coarseRow);
return(list);
}
});
}
}
if (symbolCount % 1000 == 0)
{
var elapsed = DateTime.UtcNow - startTime;
Log.Trace($"CoarseUniverseGeneratorProgram.Run(): Processed {symbolCount} in {elapsed:g} at {symbolCount / elapsed.TotalMinutes:F2} symbols/minute ");
}
});
_destinationFolder.Create();
var startWriting = DateTime.UtcNow;
Parallel.ForEach(outputCoarseContent, coarseByDate =>
{
var filename = $"{coarseByDate.Key.ToString(DateFormat.EightCharacter, CultureInfo.InvariantCulture)}.csv";
var filePath = Path.Combine(_destinationFolder.FullName, filename);
Log.Debug($"CoarseUniverseGeneratorProgram.Run(): Saving {filename} with {coarseByDate.Value.Count} entries.");
File.WriteAllLines(filePath, coarseByDate.Value.OrderBy(cr => cr));
var filesCount = Interlocked.Increment(ref coarseFilesGenerated);
if (filesCount % 1000 == 0)
{
var elapsed = DateTime.UtcNow - startWriting;
Log.Trace($"CoarseUniverseGeneratorProgram.Run(): Processed {filesCount} in {elapsed:g} at {filesCount / elapsed.TotalSeconds:F2} files/second ");
}
});
Log.Trace($"\n\nTotal of {coarseFilesGenerated} coarse files generated in {DateTime.UtcNow - startTime:g}:\n" +
$"\t => {filesRead} daily data files read.\n");
}
catch (Exception e)
{
Log.Error(e, $"CoarseUniverseGeneratorProgram.Run(): FAILED!");
success = false;
}
return(success);
}
/// <summary>
/// Iterates each daily file in the specified <paramref name="dailyFolder"/> and adds a line for each
/// day to the appropriate coarse file
/// </summary>
/// <param name="dailyFolder">The folder with daily data.</param>
/// <param name="coarseFolder">The coarse output folder.</param>
/// <param name="mapFileResolver">The map file resolver.</param>
/// <param name="factorFileProvider">The factor file provider.</param>
/// <param name="exclusions">The symbols to be excluded from processing.</param>
/// <param name="ignoreMapless">Ignore the symbols without a map file.</param>
/// <param name="symbolResolver">Function used to provide symbol resolution. Default resolution uses the zip file name to resolve
/// the symbol, specify null for this behavior.</param>
/// <returns>Collection with the names of the newly generated coarse files.</returns>
/// <exception cref="Exception">
/// Unable to resolve market for daily folder: " + dailyFolder
/// or
/// Unable to resolve fundamental path for coarse folder: " + coarseFolder
/// </exception>
public static ICollection <string> ProcessDailyFolder(string dailyFolder, string coarseFolder, MapFileResolver mapFileResolver, IFactorFileProvider factorFileProvider,
HashSet <string> exclusions, bool ignoreMapless, Func <string, string> symbolResolver = null)
{
const decimal scaleFactor = 10000m;
Log.Trace("Processing: {0}", dailyFolder);
var start = DateTime.UtcNow;
// load map files into memory
var symbols = 0;
var maplessCount = 0;
var dates = new HashSet <DateTime>();
// instead of opening/closing these constantly, open them once and dispose at the end (~3x speed improvement)
var writers = new Dictionary <string, StreamWriter>();
var marketDirectoryInfo = new DirectoryInfo(dailyFolder).Parent;
if (marketDirectoryInfo == null)
{
throw new Exception($"Unable to resolve market for daily folder: {dailyFolder}");
}
var market = marketDirectoryInfo.Name.ToLowerInvariant();
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(marketDirectoryInfo.FullName, "fundamental", "fine");
// open up each daily file to get the values and append to the daily coarse files
foreach (var file in Directory.EnumerateFiles(dailyFolder, "*.zip"))
{
try
{
var ticker = Path.GetFileNameWithoutExtension(file);
var fineAvailableDates = Enumerable.Empty <DateTime>();
var tickerFineFundamentalFolder = Path.Combine(fineFundamentalFolder, ticker);
if (Directory.Exists(tickerFineFundamentalFolder))
{
fineAvailableDates = Directory.GetFiles(tickerFineFundamentalFolder, "*.zip")
.Select(f => DateTime.ParseExact(Path.GetFileNameWithoutExtension(f), DateFormat.EightCharacter, CultureInfo.InvariantCulture))
.ToList();
}
if (ticker == null)
{
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Unable to resolve symbol from file: {0}", file);
continue;
}
if (symbolResolver != null)
{
ticker = symbolResolver(ticker);
}
ticker = ticker.ToUpperInvariant();
if (exclusions != null && exclusions.Contains(ticker))
{
Log.Trace("Excluded symbol: {0}", ticker);
continue;
}
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);
if (ignoreMapless && !checkedForMapFile)
{
checkedForMapFile = true;
if (!mapFileResolver.ResolveMapFile(ticker, date).Any())
{
// if the resolved map file has zero entries then it's a mapless symbol
maplessCount++;
break;
}
}
var close = Parse.Decimal(csv[4]) / scaleFactor;
var volume = Parse.Long(csv[5]);
var dollarVolume = close * volume;
var coarseFile = Path.Combine(coarseFolder, date.ToStringInvariant("yyyyMMdd") + ".csv");
dates.Add(date);
// try to resolve a map file and if found, regen the sid
var sid = SecurityIdentifier.GenerateEquity(SecurityIdentifier.DefaultDate, ticker, market);
var mapFile = mapFileResolver.ResolveMapFile(ticker, 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($"CoarseGenerator.ProcessDailyFolder(): Unable to resolve map file for {ticker} as of {date.ToStringInvariant("d")}");
continue;
}
// get price and split factors from factor files
var symbol = new Symbol(sid, ticker);
var factorFile = factorFileProvider.Get(symbol);
var factorFileRow = factorFile?.GetScalingFactors(date);
var priceFactor = factorFileRow?.PriceFactor ?? 1m;
var splitFactor = factorFileRow?.SplitFactor ?? 1m;
// Check if security has fine file within a trailing month for a date-ticker set.
// There are tricky cases where a folder named by a ticker can have data for multiple securities.
// e.g GOOG -> GOOGL (GOOG T1AZ164W5VTX) / GOOCV -> GOOG (GOOCV VP83T1ZUHROL) case.
// The fine data in the 'fundamental/fine/goog' folder will be for 'GOOG T1AZ164W5VTX' up to the 2014-04-02 and for 'GOOCV VP83T1ZUHROL' afterward.
// Therefore, date before checking if the security has fundamental data for a date, we need to filter the fine files the map's first date.
var firstDate = mapFile?.FirstDate ?? DateTime.MinValue;
var hasFundamentalDataForDate = fineAvailableDates.Where(d => d >= firstDate).Any(d => date.AddMonths(-1) <= d && d <= date);
// The following section handles mergers and acquisitions cases.
// e.g. YHOO -> AABA (YHOO R735QTJ8XC9X)
// The dates right after the acquisition, valid fine fundamental data for AABA are still under the former ticker folder.
// Therefore if no fine fundamental data is found in the 'fundamental/fine/aaba' folder, it searches into the 'yhoo' folder.
if (mapFile != null && mapFile.Count() > 2 && !hasFundamentalDataForDate)
{
var previousTicker = mapFile.LastOrDefault(m => m.Date < date)?.MappedSymbol;
if (previousTicker != null)
{
var previousTickerFineFundamentalFolder = Path.Combine(fineFundamentalFolder, previousTicker);
if (Directory.Exists(previousTickerFineFundamentalFolder))
{
var previousTickerFineAvailableDates = Directory.GetFiles(previousTickerFineFundamentalFolder, "*.zip")
.Select(f => DateTime.ParseExact(Path.GetFileNameWithoutExtension(f), DateFormat.EightCharacter, CultureInfo.InvariantCulture))
.ToList();
hasFundamentalDataForDate = previousTickerFineAvailableDates.Where(d => d >= firstDate).Any(d => date.AddMonths(-1) <= d && d <= date);
}
}
}
// sid,symbol,close,volume,dollar volume,has fundamental data,price factor,split factor
var coarseFileLine = string.Format(
CultureInfo.InvariantCulture,
"{0},{1},{2},{3},{4},{5},{6},{7}",
sid,
ticker,
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($"CoarseGenerator.ProcessDailyFolder(): Completed processing {symbols} symbols. Current elapsed: {(DateTime.UtcNow - start).TotalSeconds.ToStringInvariant("0.00")} seconds");
}
}
catch (Exception err)
{
// log the error and continue with the process
Log.Error(err.ToString());
}
}
Log.Trace("CoarseGenerator.ProcessDailyFolder(): Saving {0} coarse files to disk", dates.Count);
// dispose all the writers at the end of processing
foreach (var writer in writers)
{
writer.Value.Dispose();
}
var stop = DateTime.UtcNow;
Log.Trace($"CoarseGenerator.ProcessDailyFolder(): Processed {symbols} symbols into {dates.Count} coarse files in {(stop - start).TotalSeconds.ToStringInvariant("0.00")} seconds");
Log.Trace($"CoarseGenerator.ProcessDailyFolder(): Excluded {maplessCount} mapless symbols.");
return(writers.Keys);
}
/// <summary>
/// Initializes the <see cref="SubscriptionDataReader"/> instance
/// </summary>
/// <remarks>Should be called after all consumers of <see cref="NewTradableDate"/> event are set,
/// since it will produce events.</remarks>
public void Initialize()
{
if (_initialized)
{
return;
}
//Save the type of data we'll be getting from the source.
try
{
_dataFactory = _config.Type.GetBaseDataInstance();
}
catch (ArgumentException exception)
{
OnInvalidConfigurationDetected(new InvalidConfigurationDetectedEventArgs(exception.Message));
_endOfStream = true;
return;
}
//If its quandl set the access token in data factory:
var quandl = _dataFactory as Quandl;
if (quandl != null)
{
if (!Quandl.IsAuthCodeSet)
{
Quandl.SetAuthCode(Config.Get("quandl-auth-token"));
}
}
// If Tiingo data, set the access token in data factory
var tiingo = _dataFactory as TiingoDailyData;
if (tiingo != null)
{
if (!Tiingo.IsAuthCodeSet)
{
Tiingo.SetAuthCode(Config.Get("tiingo-auth-token"));
}
}
// If USEnergyInformation data, set the access token in data factory
var energyInformation = _dataFactory as USEnergyInformation;
if (energyInformation != null)
{
if (!USEnergyInformation.IsAuthCodeSet)
{
USEnergyInformation.SetAuthCode(Config.Get("us-energy-information-auth-token"));
}
}
_factorFile = new FactorFile(_config.Symbol.Value, new List <FactorFileRow>());
_mapFile = new MapFile(_config.Symbol.Value, new List <MapFileRow>());
// load up the map files for equities, options, and custom data if it supports it.
// Only load up factor files for equities
if (_config.TickerShouldBeMapped())
{
try
{
var mapFile = _mapFileResolver.ResolveMapFile(_config.Symbol, _config.Type);
// only take the resolved map file if it has data, otherwise we'll use the empty one we defined above
if (mapFile.Any())
{
_mapFile = mapFile;
}
if (!_config.IsCustomData && _config.SecurityType != SecurityType.Option)
{
var factorFile = _factorFileProvider.Get(_config.Symbol);
_hasScaleFactors = factorFile != null;
if (_hasScaleFactors)
{
_factorFile = factorFile;
// if factor file has minimum date, update start period if before minimum date
if (!_isLiveMode && _factorFile != null && _factorFile.FactorFileMinimumDate.HasValue)
{
if (_periodStart < _factorFile.FactorFileMinimumDate.Value)
{
_periodStart = _factorFile.FactorFileMinimumDate.Value;
OnNumericalPrecisionLimited(
new NumericalPrecisionLimitedEventArgs(
$"Data for symbol {_config.Symbol.Value} has been limited due to numerical precision issues in the factor file. " +
$"The starting date has been set to {_factorFile.FactorFileMinimumDate.Value.ToShortDateString()}."));
}
}
}
}
}
catch (Exception err)
{
Log.Error(err, "Fetching Price/Map Factors: " + _config.Symbol.ID + ": ");
}
}
// Estimate delisting date.
switch (_config.Symbol.ID.SecurityType)
{
case SecurityType.Future:
_delistingDate = _config.Symbol.ID.Date;
break;
case SecurityType.Option:
_delistingDate = OptionSymbol.GetLastDayOfTrading(_config.Symbol);
break;
default:
_delistingDate = _mapFile.DelistingDate;
break;
}
// adding a day so we stop at EOD
_delistingDate = _delistingDate.AddDays(1);
_subscriptionFactoryEnumerator = ResolveDataEnumerator(true);
_initialized = true;
}
/// <summary>
/// Subscription data reader takes a subscription request, loads the type, accepts the data source and enumerate on the results.
/// </summary>
/// <param name="config">Subscription configuration object</param>
/// <param name="periodStart">Start date for the data request/backtest</param>
/// <param name="periodFinish">Finish date for the data request/backtest</param>
/// <param name="resultHandler">Result handler used to push error messages and perform sampling on skipped days</param>
/// <param name="mapFileResolver">Used for resolving the correct map files</param>
/// <param name="factorFileProvider">Used for getting factor files</param>
/// <param name="dataProvider">Used for getting files not present on disk</param>
/// <param name="dataCacheProvider">Used for caching files</param>
/// <param name="tradeableDates">Defines the dates for which we'll request data, in order, in the security's exchange time zone</param>
/// <param name="isLiveMode">True if we're in live mode, false otherwise</param>
/// <param name="includeAuxilliaryData">True if we want to emit aux data, false to only emit price data</param>
public SubscriptionDataReader(SubscriptionDataConfig config,
DateTime periodStart,
DateTime periodFinish,
IResultHandler resultHandler,
MapFileResolver mapFileResolver,
IFactorFileProvider factorFileProvider,
IDataProvider dataProvider,
IEnumerable <DateTime> tradeableDates,
bool isLiveMode,
IDataCacheProvider dataCacheProvider,
bool includeAuxilliaryData = true)
{
//Save configuration of data-subscription:
_config = config;
_auxiliaryData = new Queue <BaseData>();
//Save Start and End Dates:
_periodStart = periodStart;
_periodFinish = periodFinish;
_dataProvider = dataProvider;
_dataCacheProvider = dataCacheProvider;
//Save access to securities
_isLiveMode = isLiveMode;
_includeAuxilliaryData = includeAuxilliaryData;
//Save the type of data we'll be getting from the source.
//Create the dynamic type-activators:
var objectActivator = ObjectActivator.GetActivator(config.Type);
_resultHandler = resultHandler;
_tradeableDates = tradeableDates.GetEnumerator();
if (objectActivator == null)
{
_resultHandler.ErrorMessage("Custom data type '" + config.Type.Name + "' missing parameterless constructor E.g. public " + config.Type.Name + "() { }");
_endOfStream = true;
return;
}
//Create an instance of the "Type":
var userObj = objectActivator.Invoke(new object[] { config.Type });
_dataFactory = userObj as BaseData;
//If its quandl set the access token in data factory:
var quandl = _dataFactory as Quandl;
if (quandl != null)
{
if (!Quandl.IsAuthCodeSet)
{
Quandl.SetAuthCode(Config.Get("quandl-auth-token"));
}
}
// If Tiingo data, set the access token in data factory
var tiingo = _dataFactory as TiingoDailyData;
if (tiingo != null)
{
if (!Tiingo.IsAuthCodeSet)
{
Tiingo.SetAuthCode(Config.Get("tiingo-auth-token"));
}
}
_factorFile = new FactorFile(config.Symbol.Value, new List <FactorFileRow>());
_mapFile = new MapFile(config.Symbol.Value, new List <MapFileRow>());
// load up the map and factor files for equities
if (!config.IsCustomData && config.SecurityType == SecurityType.Equity)
{
try
{
var mapFile = mapFileResolver.ResolveMapFile(config.Symbol.ID.Symbol, config.Symbol.ID.Date);
// only take the resolved map file if it has data, otherwise we'll use the empty one we defined above
if (mapFile.Any())
{
_mapFile = mapFile;
}
var factorFile = factorFileProvider.Get(_config.Symbol);
_hasScaleFactors = factorFile != null;
if (_hasScaleFactors)
{
_factorFile = factorFile;
// if factor file has minimum date, update start period if before minimum date
if (!_isLiveMode && _factorFile != null && _factorFile.FactorFileMinimumDate.HasValue)
{
if (_periodStart < _factorFile.FactorFileMinimumDate.Value)
{
_periodStart = _factorFile.FactorFileMinimumDate.Value;
_resultHandler.DebugMessage(
string.Format("Data for symbol {0} has been limited due to numerical precision issues in the factor file. The starting date has been set to {1}.",
config.Symbol.Value,
_factorFile.FactorFileMinimumDate.Value.ToShortDateString()));
}
}
}
}
catch (Exception err)
{
Log.Error(err, "Fetching Price/Map Factors: " + config.Symbol.ID + ": ");
}
}
// load up the map and factor files for underlying of equity option
if (!config.IsCustomData && config.SecurityType == SecurityType.Option)
{
try
{
var mapFile = mapFileResolver.ResolveMapFile(config.Symbol.Underlying.ID.Symbol, config.Symbol.Underlying.ID.Date);
// only take the resolved map file if it has data, otherwise we'll use the empty one we defined above
if (mapFile.Any())
{
_mapFile = mapFile;
}
}
catch (Exception err)
{
Log.Error(err, "Map Factors: " + config.Symbol.ID + ": ");
}
}
// Estimate delisting date.
switch (_config.Symbol.ID.SecurityType)
{
case SecurityType.Future:
_delistingDate = _config.Symbol.ID.Date;
break;
case SecurityType.Option:
_delistingDate = OptionSymbol.GetLastDayOfTrading(_config.Symbol);
break;
default:
_delistingDate = _mapFile.DelistingDate;
break;
}
_subscriptionFactoryEnumerator = ResolveDataEnumerator(true);
}
/// <summary>
/// Initializes the <see cref="SubscriptionDataReader"/> instance
/// </summary>
public void Initialize()
{
if (_initialized)
{
return;
}
//Save the type of data we'll be getting from the source.
//Create the dynamic type-activators:
var objectActivator = ObjectActivator.GetActivator(_config.Type);
if (objectActivator == null)
{
OnInvalidConfigurationDetected(
new InvalidConfigurationDetectedEventArgs(
$"Custom data type \'{_config.Type.Name}\' missing parameterless constructor " +
$"E.g. public {_config.Type.Name}() {{ }}"));
_endOfStream = true;
return;
}
//Create an instance of the "Type":
var userObj = objectActivator.Invoke(new object[] { _config.Type });
_dataFactory = userObj as BaseData;
//If its quandl set the access token in data factory:
var quandl = _dataFactory as Quandl;
if (quandl != null)
{
if (!Quandl.IsAuthCodeSet)
{
Quandl.SetAuthCode(Config.Get("quandl-auth-token"));
}
}
// If Tiingo data, set the access token in data factory
var tiingo = _dataFactory as TiingoDailyData;
if (tiingo != null)
{
if (!Tiingo.IsAuthCodeSet)
{
Tiingo.SetAuthCode(Config.Get("tiingo-auth-token"));
}
}
_factorFile = new FactorFile(_config.Symbol.Value, new List <FactorFileRow>());
_mapFile = new MapFile(_config.Symbol.Value, new List <MapFileRow>());
// load up the map and factor files for equities
if (!_config.IsCustomData && _config.SecurityType == SecurityType.Equity)
{
try
{
var mapFile = _mapFileResolver.ResolveMapFile(_config.Symbol.ID.Symbol, _config.Symbol.ID.Date);
// only take the resolved map file if it has data, otherwise we'll use the empty one we defined above
if (mapFile.Any())
{
_mapFile = mapFile;
}
var factorFile = _factorFileProvider.Get(_config.Symbol);
_hasScaleFactors = factorFile != null;
if (_hasScaleFactors)
{
_factorFile = factorFile;
// if factor file has minimum date, update start period if before minimum date
if (!_isLiveMode && _factorFile != null && _factorFile.FactorFileMinimumDate.HasValue)
{
if (_periodStart < _factorFile.FactorFileMinimumDate.Value)
{
_periodStart = _factorFile.FactorFileMinimumDate.Value;
OnNumericalPrecisionLimited(
new NumericalPrecisionLimitedEventArgs(
$"Data for symbol {_config.Symbol.Value} has been limited due to numerical precision issues in the factor file. " +
$"The starting date has been set to {_factorFile.FactorFileMinimumDate.Value.ToShortDateString()}."));
}
}
}
}
catch (Exception err)
{
Log.Error(err, "Fetching Price/Map Factors: " + _config.Symbol.ID + ": ");
}
}
// load up the map and factor files for underlying of equity option
if (!_config.IsCustomData && _config.SecurityType == SecurityType.Option)
{
try
{
var mapFile = _mapFileResolver.ResolveMapFile(_config.Symbol.Underlying.ID.Symbol, _config.Symbol.Underlying.ID.Date);
// only take the resolved map file if it has data, otherwise we'll use the empty one we defined above
if (mapFile.Any())
{
_mapFile = mapFile;
}
}
catch (Exception err)
{
Log.Error(err, "Map Factors: " + _config.Symbol.ID + ": ");
}
}
// Estimate delisting date.
switch (_config.Symbol.ID.SecurityType)
{
case SecurityType.Future:
_delistingDate = _config.Symbol.ID.Date;
break;
case SecurityType.Option:
_delistingDate = OptionSymbol.GetLastDayOfTrading(_config.Symbol);
break;
default:
_delistingDate = _mapFile.DelistingDate;
break;
}
_subscriptionFactoryEnumerator = ResolveDataEnumerator(true);
_initialized = true;
}
public void RetrievesFromDisk()
{
var factorFile = FactorFileProvider.Get(Symbols.SPY);
Assert.IsNotNull(factorFile);
}
/// <summary>
/// Setups a new <see cref="Subscription"/> which will consume a blocking <see cref="EnqueueableEnumerator{T}"/>
/// that will be feed by a worker task
/// </summary>
/// <param name="request">The subscription data request</param>
/// <param name="enumerator">The data enumerator stack</param>
/// <param name="factorFileProvider">The factor file provider</param>
/// <param name="enablePriceScale">Enables price factoring</param>
/// <returns>A new subscription instance ready to consume</returns>
public static Subscription CreateAndScheduleWorker(
SubscriptionRequest request,
IEnumerator <BaseData> enumerator,
IFactorFileProvider factorFileProvider,
bool enablePriceScale)
{
var exchangeHours = request.Security.Exchange.Hours;
var enqueueable = new EnqueueableEnumerator <SubscriptionData>(true);
var timeZoneOffsetProvider = new TimeZoneOffsetProvider(request.Configuration.ExchangeTimeZone, request.StartTimeUtc, request.EndTimeUtc);
var subscription = new Subscription(request, enqueueable, timeZoneOffsetProvider);
var config = subscription.Configuration;
enablePriceScale = enablePriceScale && config.PricesShouldBeScaled();
var lastTradableDate = DateTime.MinValue;
Func <int, bool> produce = (workBatchSize) =>
{
try
{
var count = 0;
while (enumerator.MoveNext())
{
// subscription has been removed, no need to continue enumerating
if (enqueueable.HasFinished)
{
enumerator.DisposeSafely();
return(false);
}
var data = enumerator.Current;
// Use our config filter to see if we should emit this
// This currently catches Auxiliary data that we don't want to emit
if (data != null && !config.ShouldEmitData(data, request.IsUniverseSubscription))
{
continue;
}
// In the event we have "Raw" configuration, we will force our subscription data
// to precalculate adjusted data. The data will still be emitted as raw, but
// if the config is changed at any point it can emit adjusted data as well
// See SubscriptionData.Create() and PrecalculatedSubscriptionData for more
var requestMode = config.DataNormalizationMode;
if (config.SecurityType == SecurityType.Equity)
{
requestMode = requestMode != DataNormalizationMode.Raw ? requestMode : DataNormalizationMode.Adjusted;
}
// We update our price scale factor when the date changes for non fill forward bars or if we haven't initialized yet.
// We don't take into account auxiliary data because we don't scale it and because the underlying price data could be fill forwarded
if (enablePriceScale && data?.Time.Date > lastTradableDate && data.DataType != MarketDataType.Auxiliary && (!data.IsFillForward || lastTradableDate == DateTime.MinValue))
{
var factorFile = factorFileProvider.Get(request.Configuration.Symbol);
lastTradableDate = data.Time.Date;
request.Configuration.PriceScaleFactor = factorFile.GetPriceScale(data.Time.Date, requestMode, config.ContractDepthOffset, config.DataMappingMode);
}
SubscriptionData subscriptionData = SubscriptionData.Create(
config,
exchangeHours,
subscription.OffsetProvider,
data,
requestMode,
enablePriceScale ? request.Configuration.PriceScaleFactor : null);
// drop the data into the back of the enqueueable
enqueueable.Enqueue(subscriptionData);
count++;
// stop executing if added more data than the work batch size, we don't want to fill the ram
if (count > workBatchSize)
{
return(true);
}
}
}
catch (Exception exception)
{
Log.Error(exception, $"Subscription worker task exception {request.Configuration}.");
}
// we made it here because MoveNext returned false or we exploded, stop the enqueueable
enqueueable.Stop();
// we have to dispose of the enumerator
enumerator.DisposeSafely();
return(false);
};
WeightedWorkScheduler.Instance.QueueWork(config.Symbol, produce,
// if the subscription finished we return 0, so the work is prioritized and gets removed
() =>
{
if (enqueueable.HasFinished)
{
return(0);
}
var count = enqueueable.Count;
return(count > WeightedWorkScheduler.MaxWorkWeight ? WeightedWorkScheduler.MaxWorkWeight : count);
}
);
return(subscription);
}
