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_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_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));
}
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_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));
}
private static void HandleStreamToCreateJob(
QueueHandler sut,
String tenant = "test",
Dictionary <String, Object> customData = null)
{
StreamReadModel rm = new StreamReadModel()
{
Filename = new FileNameWithExtension("test.docx"),
EventType = HandleStreamEventTypes.DocumentHasNewFormat,
FormatInfo = new FormatInfo()
{
PipelineId = new PipelineId("tika"),
DocumentFormat = new DocumentFormat("tika"),
BlobId = new BlobId("tika." + lastBlobId++)
},
DocumentCustomData = new DocumentCustomData(customData ?? new Dictionary <String, Object>()),
};
sut.Handle(rm, new TenantId(tenant));
}
public void Verify_file_extension_permitted()
{
var info = new QueueInfo("test", "", "pdf|docx");
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"),
DocumentFormat = new DocumentFormat("office"),
BlobId = new BlobId("soffice.1")
}
};
sut.Handle(rm, new TenantId("test"));
var collection = _db.GetCollection <QueuedJob>("queue.test");
Assert.That(collection.AsQueryable().Count(), Is.EqualTo(1));
}