/// <summary>
/// Records the error in the parsing report and determines if to continue with the parsing.
/// </summary>
private bool ShouldContinueAfterError(ParsingErrorReason error, FileParsingReport parsingReport, string usageFile, long lineNumber)
{
parsingReport.Errors.Add(new ParsingError(Path.GetFileName(usageFile), lineNumber, error));
if (parsingReport.Errors.Count > _maximumParsingErrorsCount)
{
parsingReport.IsCompletedSuccessfuly = false;
return(false);
}
return(true);
}
/// <summary>
/// Parse a usage events file into usage events items.
/// Expected format is userId,itemId,timestamp[,event-type[,event-weight]
/// </summary>
private IEnumerable <SarUsageEvent> ParseUsageEventsFile(string usageFile, FileParsingReport parsingReport, DateTime defaultEventTimestamp)
{
using (var reader = new TextFieldParser(usageFile)
{
Delimiters = new[] { "," }
})
{
while (!reader.EndOfData)
{
string[] fields;
parsingReport.TotalLinesCount++;
try
{
fields = reader.ReadFields();
}
catch (MalformedLineException ex)
{
if (ShouldContinueAfterError(ParsingErrorReason.MalformedLine,
parsingReport, usageFile, ex.LineNumber))
{
continue;
}
yield break;
}
ParsingErrorReason?parsingError;
ParsingErrorReason?parsingWarning;
SarUsageEvent usageEvent = ParseUsageEvent(fields, defaultEventTimestamp, out parsingError, out parsingWarning);
if (parsingError.HasValue)
{
if (ShouldContinueAfterError(parsingError.Value,
parsingReport, usageFile, reader.LineNumber - 1))
{
continue;
}
yield break;
}
if (parsingWarning.HasValue)
{
parsingReport.Warnings.Add(new ParsingError(Path.GetFileName(usageFile), reader.LineNumber - 1,
parsingWarning.Value));
continue;
}
parsingReport.SuccessfulLinesCount++;
yield return(usageEvent);
}
}
}
/// <summary>
/// Parse a catalog file to <see cref="SarCatalogItem"/> items.
/// </summary>
/// <param name="catalogFilePath">The file to parse</param>
/// <param name="cancellationToken">A cancellation token used to abort the operation</param>
/// <param name="catalogItems">The parsed catalog items</param>
/// <param name="featureNames">The parsed names of the catalog items features, in the same order as the feature values in the catalog</param>
/// <returns>The parsing report</returns>
public FileParsingReport ParseCatalogFile(
string catalogFilePath,
CancellationToken cancellationToken,
out IList <SarCatalogItem> catalogItems,
out string[] featureNames)
{
if (string.IsNullOrWhiteSpace(catalogFilePath))
{
throw new ArgumentNullException(nameof(catalogFilePath));
}
if (!File.Exists(catalogFilePath))
{
throw new ArgumentException($"Failed to find catalog file under '{catalogFilePath}'", nameof(catalogFilePath));
}
_tracer.TraceInformation("Starting catalog file parsing");
// parse the catalog file into catalog items
var featureNamesIndex = new ConcurrentDictionary <string, uint>();
var parsingReport = new FileParsingReport();
IList <SarCatalogItem> parsedCatalogItems = ParseCatalogFile(catalogFilePath, featureNamesIndex, parsingReport).ToList();
if (!parsingReport.IsCompletedSuccessfuly)
{
catalogItems = new List <SarCatalogItem>(0);
featureNames = new string[0];
_tracer.TraceError("Failed parsing catalog file");
return(parsingReport);
}
cancellationToken.ThrowIfCancellationRequested();
// clear the feature index as it is no longer needed
featureNames = featureNamesIndex.OrderBy(kvp => kvp.Value).Select(kvp => kvp.Key).ToArray();
featureNamesIndex.Clear();
// inflate feature vectors
int numberOfFeatures = featureNames.Length;
catalogItems = parsedCatalogItems.Select(item => InflateFeaturesVector(item, numberOfFeatures)).ToList();
_tracer.TraceInformation("Finished catalog file parsing");
return(parsingReport);
}
/// <summary>
/// Parses each file found in the input folder into usage events,
/// while indexing the found user ids and item ids.
/// </summary>
/// <param name="usageFolder">A folder containing the usage files to parse</param>
/// <param name="cancellationToken">A cancellation token used to abort the operation</param>
/// <param name="usageEvents">The parsed usage events</param>
/// <returns>The parsing report</returns>
public FileParsingReport ParseUsageEventFiles(string usageFolder, CancellationToken cancellationToken, out IList <SarUsageEvent> usageEvents)
{
if (string.IsNullOrWhiteSpace(usageFolder))
{
throw new ArgumentNullException(nameof(usageFolder));
}
if (!Directory.Exists(usageFolder))
{
throw new ArgumentException($"Failed to find usage files folder: '{usageFolder}'", nameof(usageFolder));
}
_tracer.TraceInformation("Starting usage files parsing");
DefaultEventTimestamp = DateTime.UtcNow;
MostRecentEventTimestamp = DefaultEventTimestamp;
var parsingReport = new FileParsingReport();
usageEvents = ParseUsageEventFilesInternal(usageFolder, parsingReport, cancellationToken).ToList();
_tracer.TraceInformation("Finished usage files parsing");
return(parsingReport);
}
/// <summary>
/// Parse the input catalog file into catalog items.
/// </summary>
private IEnumerable <SarCatalogItem> ParseCatalogFile(
string catalogFilePath,
ConcurrentDictionary <string, uint> featureNamesIndex,
FileParsingReport parsingReport)
{
using (var reader = new TextFieldParser(catalogFilePath)
{
Delimiters = new[] { "," }
})
{
while (!reader.EndOfData)
{
string[] fields;
try
{
parsingReport.TotalLinesCount++;
fields = reader.ReadFields();
}
catch (MalformedLineException)
{
parsingReport.Errors.Add(
new ParsingError(Path.GetFileName(catalogFilePath), reader.ErrorLineNumber,
ParsingErrorReason.MalformedLine));
if (parsingReport.Errors.Count > _maximumParsingErrorsCount)
{
parsingReport.IsCompletedSuccessfuly = false;
yield break;
}
continue;
}
ParsingErrorReason?parsingError;
ParsingErrorReason?parsingWarning;
SarCatalogItem catalogItem = ParseCatalogItem(fields, featureNamesIndex, out parsingError, out parsingWarning);
if (parsingError.HasValue)
{
parsingReport.Errors.Add(
new ParsingError(Path.GetFileName(catalogFilePath), reader.LineNumber - 1, parsingError.Value));
if (parsingReport.Errors.Count > _maximumParsingErrorsCount)
{
parsingReport.IsCompletedSuccessfuly = false;
yield break;
}
continue;
}
if (parsingWarning.HasValue)
{
parsingReport.Warnings.Add(
new ParsingError(Path.GetFileName(catalogFilePath), reader.LineNumber - 1, parsingWarning.Value));
continue;
}
parsingReport.SuccessfulLinesCount++;
yield return(catalogItem);
}
}
// no more lines to parse - mark the parsing as successful
parsingReport.IsCompletedSuccessfuly = true;
}
/// <summary>
/// Parses each file found in the input folder into usage events,
/// while indexing the found user ids and item ids.
/// </summary>
private IEnumerable <SarUsageEvent> ParseUsageEventFilesInternal(string usageFolder, FileParsingReport parsingReport, DateTime defaultEventTimestamp, CancellationToken cancellationToken)
{
foreach (string usageFile in Directory.GetFiles(usageFolder))
{
cancellationToken.ThrowIfCancellationRequested();
_tracer.TraceInformation($"Parsing file {Path.GetFileName(usageFile)} ({(double)new FileInfo(usageFile).Length/(1024*1024):F2} MB)");
foreach (SarUsageEvent sarUsageEvent in ParseUsageEventsFile(usageFile, parsingReport, defaultEventTimestamp))
{
yield return(sarUsageEvent);
}
if (parsingReport.Errors.Count > _maximumParsingErrorsCount)
{
parsingReport.IsCompletedSuccessfuly = false;
yield break;
}
}
// no more lines to parse - mark the parsing as successful
parsingReport.IsCompletedSuccessfuly = true;
}
