public override void Execute(
Job job,
Action<string> updateStatus,
IRepository repositoryService,
IJobManager jobManager)
{
foreach(var file in job.Files.OfType<CreateFile>())
{
updateStatus(string.Format("Downloading file: \"{0}\"...", file.Name));
using(var stream = repositoryService.Download(file.Link.Id))
{
using(var fstream = new FileStream(
Path.Combine(Directory.GetCurrentDirectory(), string.Format("{0}.data", file.Name)),
FileMode.Create))
{
var buff = new byte[256];
while(true)
{
var isCompleted = stream.Read(buff, 0, buff.Length) < buff.Length;
fstream.Write(buff, 0, buff.Length);
if(isCompleted)
break;
}
}
}
}
}
public Shared.Japi.Job CreateJob()
{
var job = new Job { Name = "MapReduce" };
var assembly = new UploadAssembly(typeof(Map).Assembly);
// map
Console.WriteLine("Generating \"map\" tasks..");
var fstLayer = new List<Task>();
var n = 0;
foreach (var path in pathes)
{
fstLayer.Add(new Task
{
EntryPoint = new EntryPoint
{
Assembly = assembly,
ClassName = typeof(Map).FullName
},
InputFiles = new List<File>
{
new UploadFile(path)
},
Name = string.Format("MapReduce.Map({0})", n),
Parameters = new Dictionary<string, string>
{
{ "MapReduce.MapIndex", n.ToString() },
{ "MapReduce.ItemsCount", takeTop.ToString() }
},
OutputFiles = Sequence.Items(new CreateFile(string.Format("PartialResult_{0}", n)))
.Cast<File>()
.ToList()
});
n++;
}
// reduce
Console.WriteLine("Generating \"reduce\" tasks...");
var resultTask = new Task
{
EntryPoint = new EntryPoint
{
Assembly = assembly,
ClassName = typeof(Reduce).FullName
},
InputFiles = (from t in fstLayer
from f in t.OutputFiles
select f)
.ToList(),
Name = "MapReduce.Reduce()",
Parameters = new Dictionary<string, string> { { "MapReduce.ItemsCount", takeTop.ToString() } },
OutputFiles = new List<File> { new CreateFile("Result") },
Dependencies = fstLayer
};
job.Tasks.AddRange(fstLayer);
job.Tasks.Add(resultTask);
return job;
}
public Job CreateJob()
{
var job = new Job { Name = "DF.SDK.ComputeWordFrequencies" };
var assembly = new UploadAssembly(typeof(BuildWordsList).Assembly);
// 1st layer:
Console.WriteLine("Generating 1st layer tasks...");
var fstLayer = new List<Task>();
var n = 0;
foreach(var path in pathes)
{
fstLayer.Add(new Task
{
EntryPoint = new EntryPoint
{
Assembly = assembly,
ClassName = typeof(BuildWordsList).FullName
},
InputFiles = new List<File>
{
new UploadFile(path)
},
Name = string.Format(typeof(BuildWordsList).FullName + "({0})", n),
Parameters = new Dictionary<string, string> { { "Layer.NextSize", nodes.ToString() }, {"Node", n.ToString() } },
OutputFiles = (from target in Enumerable.Range(0, nodes)
let name = string.Format("HashDistribute_{0}_to_{1})", n, target)
select new CreateFile(name)
{
Metadata = new Dictionary<string, string>
{
{"Node",target.ToString()}
}
})
.Cast<File>()
.ToList()
});
n++;
}
// 2nd layer
Console.WriteLine("Generating 2nd layer tasks...");
var sndLayer = new List<Task>();
for(n = 0; n < nodes; n++)
{
sndLayer.Add(new Task
{
EntryPoint = new EntryPoint
{
Assembly = assembly,
ClassName = typeof(TakeTopList).FullName
},
InputFiles = (from t in fstLayer
from f in t.OutputFiles
where int.Parse(f.Metadata["Node"]) == n
select f)
.ToList(),
Name = string.Format(typeof(TakeTopList).FullName + "({0})", n),
Parameters = new Dictionary<string, string> { { "TakeTop", takeTop.ToString() } },
OutputFiles =
new List<File> { new CreateFile(string.Format("TakeTopList_{0}", n)) },
Dependencies = fstLayer
});
}
// 3rd layer
Console.WriteLine("Generating 3rd layer tasks...");
var resultTask = new Task
{
EntryPoint = new EntryPoint
{
Assembly = assembly,
ClassName = typeof(TakeTopList).FullName
},
InputFiles = (from t in sndLayer
from f in t.OutputFiles
select f)
.ToList(),
Name = typeof(TakeTopList).FullName + "(last)",
Parameters = new Dictionary<string, string> { { "TakeTop", takeTop.ToString() } },
OutputFiles = new List<File> { new CreateFile("Result") },
Dependencies = sndLayer
};
job.Tasks.AddRange(fstLayer);
job.Tasks.AddRange(sndLayer);
job.Tasks.Add(resultTask);
return job;
}