public void query_gemini_records()
{
IDocumentStore store = new DocumentStore {
Url = RavenUrl
}.Initialize();
RavenUtility.WaitForIndexing(store);
string peakDistrictBbox = BoundingBoxUtility.ToWkt(new BoundingBox
{
North = 53.6m,
South = 53.0m,
East = -1.52m,
West = -2.14m
});
Stopwatch watch = Stopwatch.StartNew();
using (IDocumentSession db = store.OpenSession())
{
List <Record> results = db.Query <Record, RecordSpatialIndex>()
.Customize(x => x.RelatesToShape(FieldNames.Spatial, peakDistrictBbox, SpatialRelation.Intersects))
.Where(i => i.Gemini.Title.StartsWith("GA"))
//.Take(10)
.ToList();
//results.Count().Should().Be(10);
Console.WriteLine(results.Count);
}
Console.WriteLine(watch.ElapsedMilliseconds);
}
public void load_gemini_records()
{
string dir = @"C:\Users\PETMON\Downloads\nbn-gemini2";
XNamespace gmd = "http://www.isotc211.org/2005/gmd";
var q = (from f in Directory.GetFiles(dir, "*.xml")
where File.ReadLines(f).Any()
let xml = XDocument.Load(f)
let p = new
{
North = (decimal)xml.Descendants(gmd + "northBoundLatitude").Single(),
South = (decimal)xml.Descendants(gmd + "southBoundLatitude").Single(),
East = (decimal)xml.Descendants(gmd + "eastBoundLongitude").Single(),
West = (decimal)xml.Descendants(gmd + "westBoundLongitude").Single(),
}
where p.North < 61 && p.South > 50 && p.East <2 && p.West> -12
// ignore bad data
select new
{
File = Path.GetFileName(f),
Wkt = BoundingBoxUtility.ToWkt(new BoundingBox
{
North = p.North,
South = p.South,
East = p.East,
West = p.West,
})
})
.ToList();
Console.WriteLine(q.Count());
q.ForEach(Console.WriteLine);
IDocumentStore store = new DocumentStore {
Url = RavenUrl
}.Initialize();
using (IDocumentSession db = store.OpenSession())
{
foreach (var x in q)
{
var item = new Record
{
Gemini = new Metadata
{
Title = x.File,
},
Wkt = x.Wkt,
};
db.Store(item);
}
db.SaveChanges();
}
IndexCreation.CreateIndexes(typeof(Record).Assembly, store);
}
private static void PerformDenormalizations(Record record)
{
// we store the bounding box as wkt so we can index it
if (!BoundingBoxUtility.IsBlank(record.Gemini.BoundingBox))
{
record.Wkt = BoundingBoxUtility.ToWkt(record.Gemini.BoundingBox);
}
}
void ValidateBoundingBox(Record record, ValidationResult <Record> result)
{
if (!BoundingBoxUtility.IsBlank(record.Gemini.BoundingBox))
{
if (!BoundingBoxUtility.IsValid(record.Gemini.BoundingBox))
{
result.Errors.Add("Invalid bounding box", r => r.Gemini.BoundingBox);
}
}
}
public HttpResponseMessage AddOTBoundingBoxes()
{
var query = new RecordQueryInputModel
{
F = new FilterOptions {
Keywords = new[] { "vocab.jncc.gov.uk/jncc-category/Overseas Territories" }
},
P = 0,
N = 1024,
};
var records = _queryer.Query(query).ToList();
foreach (var record in records)
{
// note incorrect 's' in vocabs.jncc!
var ots = record.Gemini.Keywords.Where(k => k.Vocab == "http://vocabs.jncc.gov.uk/overseas-territories").ToList();
if (ots.Any())
{
var allBoxes = BoundingBoxes.Groups.Single(g => g.Name == "Overseas Territories").Boxes
.Concat(new[] // currently don't have these so just use 0,0,0,0
{
new KnownBoundingBox {
Name = "Sovereign Base Areas Cyprus", Box = new BoundingBox()
},
new KnownBoundingBox {
Name = "British Antarctic Territory", Box = new BoundingBox()
},
});
var boxes = ots.Select(ot => allBoxes.SingleOrDefault(b => b.Name == ot.Value)).ToList();
if (boxes.Any(b => b == null))
{
throw new Exception("Some box couldn't be looked up for record " + record.Id);
}
var box = BoundingBoxUtility.MinimumOf(boxes.Select(b => b.Box));
record.Gemini.BoundingBox = box;
}
}
db.SaveChanges();
return(new HttpResponseMessage());
}
public void should_store_bounding_box_as_wkt()
{
var database = Mock.Of <IDocumentSession>();
var service = new RecordService(database, ValidatorStub());
var e = Library.Example();
var record = new Record
{
Gemini = e,
Footer = new Footer()
};
service.Update(record, TestUser);
string expectedWkt = BoundingBoxUtility.ToWkt(e.BoundingBox);
Mock.Get(database).Verify(db => db.Store(It.Is((Record r) => r.Wkt == expectedWkt)));
}
void PerformGeminiValidation(Record record, ValidationResult <Record> result)
{
// structured to match the gemini doc
// 1 title is validated at basic level
// 2 alternative title not used as optional
// 3 Dataset language, conditional - data resource contains textual information
// lets assume all data resources contain text
// data_type is enum so can't be null, will default to eng
// 4 abstract is mandatory
if (record.Gemini.Abstract.IsBlank())
{
result.Errors.Add("Abstract must be provided" + GeminiSuffix, r => r.Gemini.Abstract);
}
// 5 topic_category_must_not_be_blank
if (record.Gemini.TopicCategory.IsBlank())
{
result.Errors.Add(String.Format("Topic Category must be provided" + GeminiSuffix),
r => r.Gemini.TopicCategory);
}
// 6 keywords mandatory
if (record.Gemini.Keywords.Count == 0)
{
result.Errors.Add("Keywords must be provided" + GeminiSuffix, r => r.Gemini.Keywords);
}
// 7 temporal extent is mandatory - at least Begin must be provided
if (record.Gemini.TemporalExtent.Begin.IsBlank())
{
result.Errors.Add("Temporal Extent must be provided" + GeminiSuffix,
r => r.Gemini.TemporalExtent.Begin);
}
// 8 DatasetReferenceDate mandatory
if (record.Gemini.DatasetReferenceDate.IsBlank())
{
result.Errors.Add("Dataset Reference Date must be provided" + GeminiSuffix,
r => r.Gemini.DatasetReferenceDate);
}
// 10 Lineage is mandatory
if (record.Gemini.Lineage.IsBlank())
{
result.Errors.Add("Lineage must be provided" + GeminiSuffix, r => r.Gemini.Lineage);
}
// 15 extent is optional and not used
// 16 Vertical extent information is optional and not used
// 17 Spatial reference system is optional
// 18 Spatial resolution, where it can be specified it should - so its optional
// 19 resource location, conditional
// when online access is availble, should be a valid url
// when do not yet perform a get request and get a 200 response, the only true way to validate a url
if (record.Gemini.ResourceLocator.IsNotBlank())
{
Uri url;
if (Uri.TryCreate(record.Gemini.ResourceLocator, UriKind.Absolute, out url))
{
if (url.Scheme != Uri.UriSchemeHttp)
{
result.Errors.Add("Resource locator must be an http url",
r => r.Gemini.ResourceLocator);
}
}
else
{
result.Errors.Add("Resource locator must be a valid url",
r => r.Gemini.ResourceLocator);
}
}
// 21 DataFormat optional
// 23 reponsible Organisation
if (record.Gemini.ResponsibleOrganisation.Email.IsBlank())
{
result.Errors.Add("Email address for responsible organisation must be provided" + GeminiSuffix,
r => r.Gemini.ResponsibleOrganisation.Email);
}
if (record.Gemini.ResponsibleOrganisation.Name.IsBlank())
{
result.Errors.Add("Name of responsible organisation must be provided" + GeminiSuffix,
r => r.Gemini.ResponsibleOrganisation.Name);
}
if (record.Gemini.ResponsibleOrganisation.Role.IsBlank())
{
result.Errors.Add("Role of responsible organisation must be provided" + GeminiSuffix,
r => r.Gemini.ResponsibleOrganisation.Role);
}
// 24 frequency of update is optional
// 25 limitations on publci access is mandatory
if (record.Gemini.LimitationsOnPublicAccess.IsBlank())
{
result.Errors.Add("Limitations On Public Access must be provided" + GeminiSuffix,
r => r.Gemini.LimitationsOnPublicAccess);
}
// 26 use constraints are mandatory
if (record.Gemini.UseConstraints.IsBlank())
{
result.Errors.Add("Use Constraints must be provided (if there are none, leave as 'no conditions apply')",
r => r.Gemini.UseConstraints);
}
// 27 Additional information source is optional
// 30 metadatadate is mandatory
if (record.Gemini.MetadataDate.Equals(DateTime.MinValue))
{
result.Errors.Add("A metadata reference date must be provided" + GeminiSuffix,
r => r.Gemini.MetadataDate);
}
// 33 Metadatalanguage
// 35 Point of contacts
// org name and email contact mandatory
if (record.Gemini.MetadataPointOfContact.Email.IsBlank())
{
result.Errors.Add("A metadata point of contact email address must be provided" + GeminiSuffix,
r => r.Gemini.MetadataPointOfContact.Email);
}
if (record.Gemini.MetadataPointOfContact.Name.IsBlank())
{
result.Errors.Add("A metadata point of contact organisation name must be provided" + GeminiSuffix,
r => r.Gemini.MetadataPointOfContact.Name);
}
if (record.Gemini.MetadataPointOfContact.Role != "pointOfContact")
{
result.Errors.Add("The metadata point of contact role must be 'pointOfContact'" + GeminiSuffix,
r => r.Gemini.MetadataPointOfContact.Name);
}
// 36 Unique resource identifier
// 39 resource type is mandatory
if (record.Gemini.ResourceType.IsBlank())
{
result.Errors.Add("A resource type must be provided" + GeminiSuffix,
r => r.Gemini.ResourceType);
}
// 40 Keywords from controlled vocabularys must be defined, they cannot be added.
//ValidateControlledKeywords(record, recordValidationResult<Record>);
// Conformity, required if claiming conformity to INSPIRE
// not yet implemented
// Equivalent scale, optional
// we're going to try to squash gemini and non-geographic iso metadata together in the same validation
if (record.Gemini.ResourceType != "nonGeographicDataset")
{
// BoundingBox
// mandatory
// valid todo
if (BoundingBoxUtility.IsBlank(record.Gemini.BoundingBox))
{
result.Errors.Add("A bounding box must be supplied", r => r.Gemini.BoundingBox);
}
}
}
