private void UpdateField(string prefix, string fieldToUpdate, EdgarTaskState state, int value)
{
string field = prefix + fieldToUpdate;
int currentValue = (int)state.Dataset.GetType().GetProperty(field).GetValue(state.Dataset);
state.Dataset.GetType().GetProperty(field).SetValue(state.Dataset, value);
state.DatasetSharedRepo.UpdateEdgarDataset(state.Dataset, field);
}
private EdgarTaskState[] LoadCalTxtNum(EdgarDataset ds, IEdgarDatasetsRepository repo, ConcurrentDictionary <string, int> subs, ConcurrentDictionary <string, int> tags, ConcurrentDictionary <string, int> dims)
{
List <EdgarTaskState> states = new List <EdgarTaskState>();
List <Task> tasks = new List <Task>();
//process calc file
EdgarTaskState stateCalc = new EdgarTaskState(EdgarDatasetCalculation.FILE_NAME, ds, repo);
states.Add(stateCalc);
calcParser.Submissions = subs;
calcParser.Tags = tags;
log.Info("Datasetid " + ds.Id.ToString() + " -- starting calcService.Process(...)");
tasks.Add(Task.Factory.StartNew(() =>
calcParser.Process(stateCalc, false, EdgarDatasetCalculation.FILE_NAME, "Calculations")) //false --> to avoid to have too many threads
);
//process text file
EdgarTaskState stateText = new EdgarTaskState(EdgarDatasetText.FILE_NAME, ds, repo);
states.Add(stateText);
textParser.Dimensions = dims;
textParser.Submissions = subs;
textParser.Tags = tags;
log.Info("Datasetid " + ds.Id.ToString() + " -- starting textService.Process(...)");
tasks.Add(Task.Factory.StartNew(() =>
{
textParser.Process(stateText, true, EdgarDatasetText.FILE_NAME, "Texts");
for (int i = 0; i < MAX_TRIALS; i++)
{
if (!string.IsNullOrEmpty(stateText.FileNameToReprocess))
{
string filee = stateText.FileNameToReprocess.Split('\\').Last();
textParser.Process(stateText, true, stateText.FileNameToReprocess, "Texts");
}
}
}));
//Process num file
EdgarTaskState stateNum = new EdgarTaskState(EdgarDatasetNumber.FILE_NAME, ds, repo);
states.Add(stateNum);
numParser.Dimensions = dims;
numParser.Submissions = subs;
numParser.Tags = tags;
log.Info("Datasetid " + ds.Id.ToString() + " -- starting numService.Process(...)");
tasks.Add(Task.Factory.StartNew(() =>
{
bool parallel = ConfigurationManager.AppSettings["run_num_in_parallel"] == "true";
numParser.Process(stateNum, parallel, EdgarDatasetNumber.FILE_NAME, "Numbers");
}));
Task.WaitAll(tasks.ToArray());
return(states.ToArray());
}
public virtual void Process(EdgarTaskState state, bool processInParallel, string fileToProcess, string fieldToUpdate)
{
try
{
Stopwatch watch = System.Diagnostics.Stopwatch.StartNew();
Log.Info("Datasetid " + state.Dataset.Id.ToString() + " -- " + fileToProcess + " -- BEGIN PROCESS");
int savedInDb;
if (!IsAlreadyProcessed(state, fieldToUpdate, out savedInDb))
{
string cacheFolder = ConfigurationManager.AppSettings["cache_folder"];
string tsvFileName = state.Dataset.RelativePath.Replace("/", "\\").Replace(".zip", "") + "\\" + fileToProcess;
string filepath = cacheFolder + tsvFileName;
string[] allLines = File.ReadAllLines(filepath);
string header = allLines[0];
UpdateTotalField(state, fieldToUpdate, allLines.Length - 1);
ConcurrentBag <int> missing;
if (savedInDb == 0)
{
missing = null;
}
else
{
missing = GetMissingLines(state.Dataset.Id, allLines.Length - 1);
}
ProcessFile(missing, fileToProcess, fieldToUpdate, state, allLines, header, cacheFolder, tsvFileName, processInParallel);
Log.Info("Datasetid " + state.Dataset.Id.ToString() + " -- " + fileToProcess + " -- Process finished, updating dataset status");
savedInDb = state.DatasetSharedRepo.GetCount <T>(state.Dataset.Id);
UpdateProcessedField(state, fieldToUpdate, savedInDb);
}
else
{
state.FileNameToReprocess = null;
Log.Info("Datasetid " + state.Dataset.Id.ToString() + " -- " + fileToProcess + " -- The complete file is already processed");
}
watch.Stop();
TimeSpan ts = watch.Elapsed;
string elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:00}", ts.Hours, ts.Minutes, ts.Seconds, ts.Milliseconds / 10);
Log.Info("Datasetid " + state.Dataset.Id.ToString() + " -- " + fileToProcess + " -- END PROCESS -- time: " + elapsedTime);
state.ResultOk = true;
}
catch (Exception ex)
{
state.ResultOk = false;
state.Exception = new EdgarDatasetException(fileToProcess, ex);
Log.Fatal("Datasetid " + state.Dataset.Id.ToString() + " -- " + fileToProcess + " -- Error: " + ex.Message, ex);
}
}
private bool IsAlreadyProcessed(EdgarTaskState state, string fieldToUpdate, out int savedInDb)
{
using (IEdgarDatasetsRepository repo = new EdgarRepository())
{
savedInDb = repo.GetCount <T>(state.Dataset.Id);
int processed = (int)state.Dataset.GetType().GetProperty("Processed" + fieldToUpdate).GetValue(state.Dataset);
if (savedInDb != processed)
{
UpdateProcessedField(state, fieldToUpdate, savedInDb);
}
int total = (int)state.Dataset.GetType().GetProperty("Total" + fieldToUpdate).GetValue(state.Dataset);
return(savedInDb == processed && processed == total && total != 0);
}
}
private EdgarTaskState[] LoadSubTagDim(EdgarDataset ds, IEdgarDatasetsRepository repo)
{
List <EdgarTaskState> states = new List <EdgarTaskState>();
EdgarTaskState stateSubs, stateTag, stateDim;
stateSubs = new EdgarTaskState(EdgarDatasetSubmission.FILE_NAME, ds, repo);
stateTag = new EdgarTaskState(EdgarDatasetTag.FILE_NAME, ds, repo);
stateDim = new EdgarTaskState(EdgarDatasetDimension.FILE_NAME, ds, repo);
states.Add(stateSubs);
states.Add(stateTag);
states.Add(stateDim);
IList <Task> tasks = new List <Task>();
log.Info("Datasetid " + ds.Id.ToString() + " -- starting submissionService.Process(...)");
tasks.Add(Task.Factory.StartNew(() =>
submissionParser.Process(stateSubs, false, EdgarDatasetSubmission.FILE_NAME, "Submissions")//false --> to avoid to have too many threads
));
log.Info("Datasetid " + ds.Id.ToString() + " -- starting tagService.Process(...)");
tasks.Add(Task.Factory.StartNew(() =>
{
if (ConfigurationManager.AppSettings["run_tag_in_parallel"] == "true")
{
tagParser.Process(stateTag, true, EdgarDatasetTag.FILE_NAME, "Tags");
}
else
{
tagParser.Process(stateTag, false, EdgarDatasetTag.FILE_NAME, "Tags");
}
}));
log.Info("Datasetid " + ds.Id.ToString() + " -- starting dimensionService.Process(...)");
tasks.Add(Task.Factory.StartNew(() =>
{
dimensionParser.Process(stateDim, false, EdgarDatasetDimension.FILE_NAME, "Dimensions");
}));
Task.WaitAll(tasks.ToArray());
return(states.ToArray());
}
private EdgarTaskState[] LoadRenPre(EdgarDataset ds, IEdgarDatasetsRepository repo, ConcurrentDictionary <string, int> subs, ConcurrentDictionary <string, int> tags, ConcurrentDictionary <string, int> nums, ConcurrentDictionary <string, int> texts)
{
List <EdgarTaskState> states = new List <EdgarTaskState>();
List <Task> tasks = new List <Task>();
EdgarTaskState stateRen = new EdgarTaskState(EdgarDatasetRender.FILE_NAME, ds, repo);
states.Add(stateRen);
EdgarTaskState statePre = new EdgarTaskState(EdgarDatasetPresentation.FILE_NAME, ds, repo);
states.Add(statePre);
tasks.Add(Task.Factory.StartNew(() =>
{
renderingParser.Subs = subs;
log.Info("Datasetid " + ds.Id.ToString() + " -- starting renderingService.Process(...)");
//Presentations has a relationship to renders
renderingParser.Process(stateRen, true, EdgarDatasetRender.FILE_NAME, "Renders");
presentationParser.Subs = subs;
presentationParser.Tags = tags;
log.Info("Datasetid " + ds.Id.ToString() + " -- loading all rens for presentationService.Process(...)");
presentationParser.Renders = renderingParser.GetAsConcurrent(ds.Id);
presentationParser.Nums = nums;
presentationParser.Texts = texts;
log.Info("Datasetid " + ds.Id.ToString() + " -- starting presentationService.Process(...)");
if (ConfigurationManager.AppSettings["run_pre_in_parallel"] == "true")
{
presentationParser.Process(statePre, true, EdgarDatasetPresentation.FILE_NAME, "Presentations"); //parallel execution
}
else
{
presentationParser.Process(statePre, false, EdgarDatasetPresentation.FILE_NAME, "Presentations");//sequential execution
}
}
));
Task.WaitAll(tasks.ToArray());
return(states.ToArray());
}
public abstract void ProcessFile(ConcurrentBag <int> missing, string fileToProcess, string fieldToUpdate, EdgarTaskState state, string[] allLines, string header, string cacheFolder, string tsvFileName, bool processInParallel);
private void UpdateTotalField(EdgarTaskState state, string fieldToUpdate, int value)
{
UpdateField("Total", fieldToUpdate, state, value);
}
private void UpdateProcessedField(EdgarTaskState state, string fieldToUpdate, int value)
{
UpdateField("Processed", fieldToUpdate, state, value);
}
