public void verify_id_is_opaque_and_not_contains_blob_id()
{
var info = new QueueInfo("test", "", "docx");
QueueHandler sut = new QueueHandler(info, _db);
StreamReadModel rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.docx"),
FormatInfo = new FormatInfo()
{
DocumentFormat = new DocumentFormat("thumb.small"),
BlobId = new BlobId("blob.1"),
PipelineId = new PipelineId("thumbnail"),
},
DocumentDescriptorId = new DocumentDescriptorId(1),
Handle = new DocumentHandle("Revision_2"),
};
sut.Handle(rm, new TenantId("test_tenant"));
var collection = _db.GetCollection<QueuedJob>("queue.test");
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(1));
var job = collection.AsQueryable().Single();
Assert.That(job.BlobId, Is.EqualTo(new BlobId("blob.1")));
Assert.That(job.TenantId, Is.EqualTo(new TenantId("test_tenant")));
Assert.That(job.DocumentDescriptorId, Is.EqualTo(new DocumentDescriptorId(1)));
Assert.That(job.Handle.ToString(), Is.EqualTo(rm.Handle));
Assert.That(job.Id.ToString(), Is.Not.Contains("blob.1"), "Id should not contains internal concempts like blob id");
Assert.That(job.Id.ToString(), Is.Not.Contains("tenant"), "Id should not contains internal concempts like tenant id");
Assert.That(job.Parameters.Keys, Is.Not.Contains(JobKeys.BlobId));
Assert.That(job.Parameters.Keys, Is.Not.Contains(JobKeys.DocumentId));
}
public void verify_should_create_job_mime_extension(String mimetypes, String extensions, String fileName, Boolean expected)
{
QueueInfo sut = new QueueInfo("TEST", extensions : extensions, mimeTypes: mimetypes);
StreamReadModel sr = new StreamReadModel();
sr.Filename = new Core.Model.FileNameWithExtension(fileName);
Assert.That(sut.ShouldCreateJob(sr), Is.EqualTo(expected));
}
public QueueHandler(QueueInfo info, IMongoDatabase database)
{
_collection = database.GetCollection<QueuedJob>("queue." + info.Name);
_collection.Indexes.CreateOne(
Builders<QueuedJob>.IndexKeys.Ascending(x => x.Status).Ascending(x => x.StreamId).Ascending(x => x.SchedulingTimestamp),
new CreateIndexOptions() { Name = "ForGetNextJobQuery" });
_collection.Indexes.CreateOne(
Builders<QueuedJob>.IndexKeys.Ascending(x => x.TenantId).Ascending(x => x.BlobId),
new CreateIndexOptions() { Name = "TenantAndBlob", Unique = false });
_collection.Indexes.CreateOne(
Builders<QueuedJob>.IndexKeys.Ascending(x => x.Handle).Ascending(x => x.Status),
new CreateIndexOptions() { Name = "HandleAndStatus", Unique = false });
_info = info;
Name = info.Name;
_statsAggregationQuery = BsonDocument.Parse(@"
{
_id : '$Status',
c : {$sum:1}
}");
//timeout of polling time is the maximum timeout allowed before a job is considered to be locked
//but give 10 seconds to each job to start
var millisecondTimeout = info.JobLockTimeout * 60 * 1000;
_healthCheck = MetricHeartBeatHealthCheck.Create(
"Job queue " + info.Name,
millisecondTimeout,
TimeSpan.FromMilliseconds(millisecondTimeout - 10000));
Logger = NullLogger.Instance;
}
public DocumentStoreTestConfigurationForPollQueue( QueueInfo[] queueInfo, String engineVersion = "v3")
{
IsQueueManager = true;
QueueJobsPollInterval = 50; //poll each 50 milliseconds.
QueueStreamPollInterval = 50;
EngineVersion = engineVersion;
this.QueueInfoList = queueInfo;
}
public void verify_file_extension_on_handler_filter_exact_extension()
{
var info = new QueueInfo("test", "", "pdf|doc");
QueueHandler sut = new QueueHandler(info, _db);
StreamReadModel rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.docx")
};
sut.Handle(rm, new TenantId("test"));
var collection = _db.GetCollection<QueuedJob>("queue.test");
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(0));
}
public void queue_with_only_manual_execution(
String mimetypes,
String extensions,
String pipeline,
String fileName)
{
QueueInfo sut = new QueueInfo("TEST",
extensions: extensions,
mimeTypes: mimetypes,
pipeline : pipeline);
StreamReadModel sr = new StreamReadModel();
sr.Filename = new Core.Model.FileNameWithExtension(fileName);
Assert.That(sut.ShouldCreateJob(sr), Is.EqualTo(false));
}
public QueueHandler(QueueInfo info, IMongoDatabase database)
{
_collection = database.GetCollection <QueuedJob>("queue." + info.Name);
_collection.Indexes.CreateOne(
Builders <QueuedJob> .IndexKeys.Ascending(x => x.Status).Ascending(x => x.StreamId).Ascending(x => x.SchedulingTimestamp),
new CreateIndexOptions()
{
Name = "ForGetNextJobQuery"
});
_collection.Indexes.CreateOne(
Builders <QueuedJob> .IndexKeys.Ascending(x => x.TenantId).Ascending(x => x.BlobId),
new CreateIndexOptions()
{
Name = "TenantAndBlob", Unique = false
});
_collection.Indexes.CreateOne(
Builders <QueuedJob> .IndexKeys.Ascending(x => x.Handle).Ascending(x => x.Status),
new CreateIndexOptions()
{
Name = "HandleAndStatus", Unique = false
});
_info = info;
Name = info.Name;
_statsAggregationQuery = BsonDocument.Parse(@"
{
_id : '$Status',
c : {$sum:1}
}");
//timeout of polling time is the maximum timeout allowed before a job is considered to be locked
//but give 10 seconds to each job to start
var millisecondTimeout = info.JobLockTimeout * 60 * 1000;
_healthCheck = MetricHeartBeatHealthCheck.Create(
"Job queue " + info.Name,
millisecondTimeout,
TimeSpan.FromMilliseconds(millisecondTimeout - 10000));
Logger = NullLogger.Instance;
}
private QueueHandler CreateAGenericJob(QueueInfo info, String tenant = "test", Dictionary<String, Object> customData = null)
{
QueueHandler sut = GetSut(info);
HandleStreamToCreateJob(sut, tenant, customData);
return sut;
}
private QueueHandler GetSut(QueueInfo info)
{
return new QueueHandler(info, _db);
}
public void verify_set_error_status()
{
var info = new QueueInfo("test", "tika", "");
info.MaxNumberOfFailure = 2;
QueueHandler sut = CreateAGenericJob(info);
var nextJob = sut.GetNextJob("", "handle", null, null);
sut.SetJobExecuted(nextJob.Id, "Error 42");
var collection = _db.GetCollection<QueuedJob>("queue.test");
var job = collection.Find(Builders<QueuedJob>.Filter.Eq(j => j.Id, nextJob.Id)).SingleOrDefault();
Assert.That(job.ExecutionError, Is.EqualTo("Error 42"));
Assert.That(job.ErrorCount, Is.EqualTo(1));
Assert.That(job.Status, Is.EqualTo(QueuedJobExecutionStatus.ReQueued));
}
public void verify_job_is_generated_with_custom_parameters()
{
var info = new QueueInfo("test", "tika", "");
info.Parameters = new Dictionary<string, string>() { { "Custom", "CustomValue" } };
QueueHandler sut = CreateAGenericJob(info);
var nextJob = sut.GetNextJob("", "handle", null, null);
Assert.That(nextJob.Parameters["Custom"], Is.EqualTo("CustomValue"));
}
public void verify_max_number_of_falure()
{
var info = new QueueInfo("test", "tika", "");
info.MaxNumberOfFailure = 2;
QueueHandler sut = CreateAGenericJob(info);
var nextJob = sut.GetNextJob("", "handle", null, null);
Assert.That(nextJob, Is.Not.Null);
var jobId = nextJob.Id;
sut.SetJobExecuted(nextJob.Id, "Error 42");
nextJob = sut.GetNextJob("", "handle", null, null);
Assert.That(nextJob, Is.Not.Null);
sut.SetJobExecuted(nextJob.Id, "Error 42");
nextJob = sut.GetNextJob("", "handle", null, null);
Assert.That(nextJob, Is.Null, "After two failure the job should not be returned anymore");
var collection = _db.GetCollection<QueuedJob>("queue.test");
var job = collection.Find(Builders<QueuedJob>.Filter.Eq(j => j.Id, jobId)).SingleOrDefault();
Assert.That(job.ExecutionError, Is.EqualTo("Error 42"));
Assert.That(job.ErrorCount, Is.EqualTo(2));
Assert.That(job.Status, Is.EqualTo(QueuedJobExecutionStatus.Failed));
}
public void verify_get_next_job_set_identity()
{
var info = new QueueInfo("test", "", "pdf|docx");
QueueHandler sut = new QueueHandler(info, _db);
StreamReadModel rm = new StreamReadModel()
{
Id = 1L,
Handle = "FirstHandle",
Filename = new FileNameWithExtension("test.docx"),
EventType = HandleStreamEventTypes.DocumentHasNewFormat,
FormatInfo = new FormatInfo()
{
PipelineId = new PipelineId("soffice"),
DocumentFormat = new DocumentFormat("office"),
BlobId = new BlobId("soffice.1")
},
DocumentDescriptorId = new DocumentDescriptorId(1),
};
sut.Handle(rm, new TenantId("test"));
rm.Handle = "SecondHandle";
rm.Id = 2L;
//This is typical situation when handle is de-duplicated, because
//and handle is assigned to another document, but the underling blob id is the same.
sut.Handle(rm, new TenantId("test"));
var collection = _db.GetCollection<QueuedJob>("queue.test");
//no need to schedule another job
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(1));
}
public void verify_pipeline_id_filter()
{
var info = new QueueInfo("test", "tika", "");
QueueHandler sut = new QueueHandler(info, _db);
StreamReadModel rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.docx"),
EventType = HandleStreamEventTypes.DocumentHasNewFormat,
FormatInfo = new FormatInfo()
{
PipelineId = new PipelineId("soffice")
}
};
sut.Handle(rm, new TenantId("test"));
var collection = _db.GetCollection<QueuedJob>("queue.test");
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(0), "pipeline filter is not filtering out unwanted pipeline");
rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.docx"),
FormatInfo = new FormatInfo()
{
PipelineId = new PipelineId("tika"),
DocumentFormat = new DocumentFormat("tika"),
BlobId = new BlobId("tika.1")
}
};
sut.Handle(rm, new TenantId("test"));
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(1), "pipeline filter is not filtering in admitted pipeline");
}
public void verify_filtering_on_mime_types()
{
var mimeTypeDocx = MimeTypes.GetMimeTypeByExtension("docx");
var info = new QueueInfo("test", mimeTypes: mimeTypeDocx);
QueueHandler sut = new QueueHandler(info, _db);
StreamReadModel rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.pdf"),
FormatInfo = new FormatInfo()
{
DocumentFormat = new DocumentFormat("thumb.small"),
BlobId = new BlobId("blob.1"),
PipelineId = new PipelineId("thumbnail")
}
};
sut.Handle(rm, new TenantId("test"));
var collection = _db.GetCollection<QueuedJob>("queue.test");
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(0));
rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.docx"),
FormatInfo = new FormatInfo()
{
DocumentFormat = new DocumentFormat("thumb.small"),
BlobId = new BlobId("blob.1"),
PipelineId = new PipelineId("thumbnail")
}
};
sut.Handle(rm, new TenantId("test"));
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(1));
}
public void verify_job_parameters_contains_mime_type()
{
var info = new QueueInfo("test", "", "docx");
QueueHandler sut = new QueueHandler(info, _db);
StreamReadModel rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.docx"),
FormatInfo = new FormatInfo()
{
DocumentFormat = new DocumentFormat("thumb.small"),
BlobId = new BlobId("blob.1"),
PipelineId = new PipelineId("thumbnail"),
},
DocumentDescriptorId = new DocumentDescriptorId(1),
};
sut.Handle(rm, new TenantId("test_tenant"));
var collection = _db.GetCollection<QueuedJob>("queue.test");
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(1));
var job = collection.AsQueryable().Single();
Assert.That(job.BlobId, Is.EqualTo(new BlobId("blob.1")));
Assert.That(job.Parameters[JobKeys.MimeType], Is.EqualTo(MimeTypes.GetMimeTypeByExtension("docx")));
}
public void verify_job_created_with_handle_metadata()
{
var info = new QueueInfo("test", "", "pdf|docx");
QueueHandler sut = new QueueHandler(info, _db);
var customData = new DocumentCustomData()
{
{"test" , "value"},
{"complex" , 42},
};
StreamReadModel rm = new StreamReadModel()
{
Id = 1L,
Handle = "FirstHandle",
Filename = new FileNameWithExtension("test.docx"),
EventType = HandleStreamEventTypes.DocumentHasNewFormat,
FormatInfo = new FormatInfo()
{
PipelineId = new PipelineId("soffice"),
DocumentFormat = new DocumentFormat("office"),
BlobId = new BlobId("soffice.1")
},
DocumentDescriptorId = new DocumentDescriptorId(1),
DocumentCustomData = customData,
};
sut.Handle(rm, new TenantId("test"));
var collection = _db.GetCollection<QueuedJob>("queue.test");
Assert.That(collection.AsQueryable().Single().HandleCustomData, Is.EquivalentTo(customData));
}
