public JsonResult GetScientificName(string name)
{
WikipediaReader wikipediaRaeder = new WikipediaReader();
string scientificName = wikipediaRaeder.GetScientificName(name);
return(Json(scientificName));
}
public void WikipediaReaderTest()
{
var xmlDump = GetXmlDump();
using var xmlReader = new WikiDumpXmlReader(xmlDump);
var wikiReader = new WikipediaReader(
xmlReader,
WikipediaReader.DefaultFilter,
blockSize: 1);
var corpus = wikiReader.Read().ToArray();
Assert.Equal(2, corpus.Length);
Assert.Equal(1, corpus[0].Documents.Count);
Assert.Equal(1, corpus[0].Metadata.Count);
var doc1 = corpus[0].Documents[0];
Assert.Equal("Simple page", doc1.Metadata.Title);
Assert.Equal("=Simple Page=\nSome text", doc1.Data);
Assert.Equal("Simple page", corpus[0].Metadata[doc1.Metadata.Id].Title);
Assert.Equal(doc1.Metadata.Id, corpus[0].Metadata[doc1.Metadata.Id].Id);
Assert.Equal(1, corpus[1].Documents.Count);
Assert.Equal(1, corpus[1].Metadata.Count);
var doc2 = corpus[1].Documents[0];
Assert.Equal("Another page", doc2.Metadata.Title);
Assert.Equal("Hello world", doc2.Data);
Assert.Equal("Another page", corpus[1].Metadata[doc2.Metadata.Id].Title);
Assert.Equal(doc2.Metadata.Id, corpus[1].Metadata[doc2.Metadata.Id].Id);
}
public static Node CreateOrSetParents(string scientificName, Type type, SubjectManager subjectManager)
{
string[] nameArray = scientificName.Split(' ');
WikipediaReader wReader = new WikipediaReader();
/**
* Lactuca -> Genus
* Lactuca sativa -> Species
* Lactuca sativa var. capitata -> Variation - Subspecies
* Lactuca sativa var. capitata Larissa -> SubSpecies
* Lactuca sativa Larissa -> SubSpecies
* */
//Lactuca sativa - create genus return genus
Taxon genus = GetOrCreateGenus(nameArray[0], subjectManager);
if (nameArray.Count() == 2)
{
return(genus);
}
//Lactuca sativa Larissa -> SubSpecies
if (nameArray.Count() == 3 && !nameArray.Contains("var."))
{
string name = nameArray[0] + " " + nameArray[1];
return(GetOrCreateSpecies(name, type, subjectManager, genus));
}
//Lactuca sativa var. capitata -> Species
if (nameArray.Count() == 4 && nameArray.Contains("var."))
{
string name = nameArray[0] + " " + nameArray[1];
return(GetOrCreateSpecies(name, type, subjectManager, genus));
}
//Lactuca sativa var. capitata Larissa -> SubSpecies
if (nameArray.Count() == 5 && nameArray.Contains("var."))
{
string name = nameArray[0] + " " + nameArray[1];
return(GetOrCreateSpecies(name, type, subjectManager, genus));
}
if (nameArray.Count() > 5)
{
string name = nameArray[0] + " " + nameArray[1];
return(GetOrCreateSpecies(name, type, subjectManager, genus));
}
return(null);
}
static void TransformWikiDump()
{
string pathToSave = wikiPath;
PrepareOutputDirectory(pathToSave);
using var xmlReader = new WikiDumpXmlReader(wikiDumpFilePath);
ICorpusReader <string> reader = new WikipediaReader(
xmlReader,
WikipediaReader.DefaultFilter,
(ushort)BlockSize,
CorpusSize);
ICorpusWriter <string> writer = new CorpusZipWriter <string>(pathToSave, stringDataSerializer);
writer.Write(reader.Read());
}
private static Species GetOrCreateSpecies(string speciesName, Type type, SubjectManager subjectManager, Taxon genus)
{
WikipediaReader wReader = new WikipediaReader();
Species species = new Species();
if (type.Equals(typeof(Plant)))
{
species = new Plant();
if (subjectManager.GetAll <Species>().Any(s => s.ScientificName.Equals(speciesName)))
{
species = subjectManager.GetAll <Plant>().FirstOrDefault(s => s.ScientificName.Equals(speciesName));
}
else
{
species.ScientificName = speciesName;
species.Name = wReader.GetName(speciesName);
species.Parent = genus;
species.Rank = TaxonRank.Species;
subjectManager.Create(species);
}
}
else if (type.Equals(typeof(Animal)))
{
species = new Animal();
if (subjectManager.GetAll <Species>().Any(s => s.ScientificName.Equals(speciesName)))
{
species = subjectManager.GetAll <Animal>().FirstOrDefault(s => s.ScientificName.Equals(speciesName));
}
else
{
species.ScientificName = speciesName;
species.Name = wReader.GetName(speciesName);
species.Parent = genus;
species.Rank = TaxonRank.Species;
subjectManager.Create(species);
}
}
return(species);
}
private static Taxon GetOrCreateGenus(string genusName, SubjectManager subjectManager)
{
WikipediaReader wReader = new WikipediaReader();
Taxon genus = new Taxon();
if (subjectManager.GetAll <Taxon>().Any(s => s.ScientificName.Equals(genusName)))
{
genus = subjectManager.GetAll <Taxon>().FirstOrDefault(s => s.ScientificName.Equals(genusName));
if (string.IsNullOrEmpty(genus.Name) && !string.IsNullOrEmpty(genus.ScientificName))
{
genus.Name = wReader.GetName(genus.ScientificName);
}
if (string.IsNullOrEmpty(genus.ScientificName) && !string.IsNullOrEmpty(genus.Name))
{
genus.ScientificName = wReader.GetScientificName(genus.Name);
}
}
else
{
genus.ScientificName = genusName;
genus.Name = wReader.GetName(genusName);
if (String.IsNullOrEmpty(genus.Name))
{
genus.Name = genusName;
}
genus.Rank = TaxonRank.Genus;
subjectManager.Create(genus);
}
return(genus);
}