public override void InitializeRelatedTables(DataProcessor processor, TableDataBuilder table)
{
if (!InlineFields)
{
var tables = processor.TableMap.Tables;
var current = tables.FirstOrDefault(t => t.Name == TableName) as TableDataBuilder;
if (current != null)
{
if (!current.Schema.FieldsAreEqual(LookupTableBuilder.Schema) || current.GetType() != LookupTableBuilder.GetType())
{
throw new InvalidOperationException("Lookup tables with the same name must have the same schema and type to be shared");
}
LookupTableBuilder.Schema.ClearRelations();
LookupTableBuilder = current;
}
else
{
//Insert dimension tables in the start, since it is referenced by other tables
tables.Insert(0, LookupTableBuilder);
}
table.Schema.Associate(LookupTableBuilder.Schema,
table.Schema.Fields.Where(f => Fields.Contains(f)).ToArray(),
LookupTableBuilder.Schema.Keys.Select(fp => fp.Value), true);
}
//Initialize realted tables in field mappers
foreach (var fm in FieldMappers)
{
fm.InitializeRelatedTables(processor, InlineFields ? table : LookupTableBuilder);
}
base.InitializeRelatedTables(processor, table);
}
public void Initialize(DataProcessor processor, TableDataBuilder parentTable)
{
foreach (var mapper in TableMappers)
{
mapper.Initialize(processor, parentTable);
}
}
public override void InitializeRelatedTables(DataProcessor processor, TableDataBuilder table)
{
foreach (var fm in FieldMappers.SelectMany(fms => fms))
{
fm.InitializeRelatedTables(processor, table);
}
}
public override void Initialize(DataProcessor processor)
{
foreach (var fm in FieldMappers.SelectMany(fm=>fm))
{
fm.Initialize(processor);
}
base.Initialize(processor);
}
public override void Initialize(DataProcessor processor)
{
//Initialize realted tables in field mappers
foreach (var fm in FieldMappers)
{
fm.Initialize(processor);
}
base.Initialize(processor);
}
public virtual void Initialize(DataProcessor processor, TableDataBuilder parentTable)
{
Builders = new List<BuilderInfo>();
foreach (var def in TableDefinitions)
{
var fieldMappers = def.FieldMappers.ToArray();
foreach (var mapper in fieldMappers)
{
mapper.Initialize(processor);
}
var builder = fieldMappers.Length == 0
? null
: new TableDataBuilder(def.Name, fieldMappers);
var builderInfo = new BuilderInfo(builder);
if( builder != null )
{
if (parentTable != null)
{
builder.LinkParentTable(parentTable);
}
builder.EnsureKey(KeyFactory.Default);
processor.TableMap.Tables.Add(builderInfo.Builder);
}
foreach (var fieldMapper in fieldMappers)
{
fieldMapper.InitializeRelatedTables(processor, builderInfo.Builder);
}
if (def.TableMappers != null)
{
foreach (var childTable in def.TableMappers)
{
builderInfo.Children.Add(childTable);
childTable.Initialize(processor, builder ?? parentTable);
}
}
Builders.Add(builderInfo);
}
}
public override void Initialize(DataProcessor processor)
{
Key.Initialize(processor);
_labelLoaders = new List<LabelLoader>();
if (Labels != null)
{
foreach (var label in Labels)
{
var loader = new LabelLoader(label.Value);
loader.LabelProvider.Initialize(processor);
_labelLoaders.Add(loader);
}
}
base.Initialize(processor);
}
public void InitializeRelatedTables(DataProcessor processor, TableDataBuilder table)
{
}
public void Initialize(DataProcessor processor)
{
}
public virtual void Initialize(DataProcessor processor)
{
_fields = CreateFields().ToArray();
}
public void Initialize(DataProcessor processor)
{
Lookup = processor.FieldLookup;
}
public void Run()
{
if (Status != JobStatus.Pending) throw new InvalidOperationException("Job is not pending");
try
{
Directory.CreateDirectory(TempDirectory);
File.WriteAllText(Path.Combine(TempDirectory, "specification.json"), Specification.ToString());
SetStatus(JobStatus.Preparing);
Specification.Initialize(this);
var source = Specification.CreateDataSource();
PostProcessors = Specification.CreatePostProcessors().ToArray();
var jobDirectory = TempDirectory;
//The processors will consume data from this collection;
var items = new BlockingCollection<object>(ExecutionSettings.DataSourceBufferSize);
//Create the processors
var processors = Enumerable.Range(0, ExecutionSettings.ProcessingThreads).Select(i =>
{
IItemFieldLookup lookup = null;
try
{
lookup = new ItemDatabaseFieldLookup(Database.GetDatabase(ExecutionSettings.DatabaseName),
Specification.DefaultLanguage,
ExecutionSettings.FieldCacheSize);
}
catch (Exception ex)
{
Log.Error("Error initailizing item field lookup", ex, this);
}
var proc = new DataProcessor(Specification.CreateRootMapper())
{
BatchSize = ExecutionSettings.BatchSize,
FieldLookup = lookup
};
var exporter = Specification.CreateExporter(jobDirectory) as CsvExporter; //Move PartititionPrefix so this cast isn't necessary
if (exporter == null)
{
exporter = new CsvExporter(jobDirectory);
}
exporter.PartitionPrefix = "~" + i + "_";
exporter.KeepOutput = true; //Don't delete the job's main directory
proc.BatchWriter = new TableDataBatchWriter(exporter)
{
SyncLock = this,
MaximumSize = ExecutionSettings.SizeLimit
};
proc.Initialize();
return proc;
}).ToArray();
var hasUpdates = false;
//Allow post processors to validate their conditions (if any). This allows the job to fail before data is processed
//Allow post processors to filter data source for updates
foreach (var pp in PostProcessors)
{
pp.Validate(processors[0].Tables, Specification);
if (pp.UpdateDataSource(processors[0].Tables, source))
{
if (hasUpdates)
{
throw new InvalidOperationException("Only one post processor can update the data source");
}
hasUpdates = true;
}
}
EstimatedClientItemCount = source.Count;
//Start the processors
var processingThreads = processors.Select(p =>
{
var t = new Thread(() =>
{
try
{
p.Process(items.GetConsumingEnumerable());
}
catch (Exception ex)
{
LastException = ex;
SetStatus(JobStatus.Failed, ex.Message);
}
});
t.Start();
return t;
}).ToArray();
if (Status == JobStatus.Failed)
{
throw LastException;
}
SetStatus(JobStatus.Running);
source.ItemLoaded += (sender, args) =>
{
if (StatusUpdateFrequency <= 0 || ItemsProcessed % StatusUpdateFrequency == 0)
{
OnProgress();
}
ItemsProcessed = args;
RowsCreated = processors.Sum(p => p.RowsCreated);
};
//Add items to the collection that the processors consume
foreach (var item in source)
{
if (Status != JobStatus.Running)
{
break;
}
if (processors.Any(p => p.BatchWriter.End))
{
break;
}
items.Add(item);
}
items.CompleteAdding();
//Wait for processors to finish
foreach (var p in processingThreads)
{
p.Join();
}
RowsCreated = processors.Sum(p => p.RowsCreated);
if (Status == JobStatus.Running)
{
//Now we know how many items we got for sure. Update progress to 100%
EstimatedClientItemCount = ItemsProcessed;
SetStatus(JobStatus.Merging);
using (var csvWriter = Specification.CreateExporter(TempDirectory))
{
var tables = MergedTableData.FromTableSets(processors.Select(p => p.Tables)).ToArray();
var w = csvWriter as CsvExporter;
if (w == null || w.KeepOutput)
{
tables = csvWriter.Export(tables).ToArray();
}
File.WriteAllText(Path.Combine(jobDirectory, "schema.json"),
tables.Select(t => t.Schema).Serialize());
foreach (var postProcessor in PostProcessors)
{
CurrentPostProcessor = postProcessor;
SetStatus(JobStatus.PostProcessing, postProcessor.Name);
postProcessor.Process(jobDirectory, tables, Specification);
}
CurrentPostProcessor = null;
foreach (var proc in processors)
{
SizeLimitExceeded = SizeLimitExceeded || proc.BatchWriter.End;
proc.BatchWriter.Dispose();
}
SetStatus(JobStatus.Completing);
}
SetStatus(JobStatus.Completed);
}
}
catch (Exception ex)
{
Log.Error("Job failed", ex, this);
LastException = ex;
SetStatus(JobStatus.Failed, ex.ToString());
}
try
{
OnJobEnded();
}
catch (Exception ex)
{
Log.Error("Exception occured after job ended", ex, this);
LastException = ex;
}
EndDate = DateTime.Now;
try
{
if (Status == JobStatus.Canceled || Status == JobStatus.Failed)
{
Delete();
}
}
catch (Exception ex)
{
Log.Error("Exception occured while deleting job", ex, this);
}
}