/// <summary>
/// Load named entities from output files
/// </summary>
private static void LoadNamedEntitiesFromFiles()
{
if (NamedEntityReader.FilesExist(PersonsFileName, OrganizationsFileName, LocationsFileName))
{
Persons = NamedEntityReader.ReadData(PersonsFileName);
Organizations = NamedEntityReader.ReadData(OrganizationsFileName);
Locations = NamedEntityReader.ReadData(LocationsFileName);
var finder = new Finder(Persons, Organizations, Locations);
finder.Find();
}
else
{
Console.WriteLine("Files not exists. Load from XML? (Y/N): ");
if (Console.ReadLine().ToLower().Equals("y"))
{
LoadNamedEntitiesFromXML();
}
}
}
/// <summary>
/// Load named entities from input XML
/// </summary>
private static void LoadNamedEntitiesFromXML()
{
//string filePath = GetDataPath(@"input_skwiki-latest-pages-articles.xml")
string filePath = GetDataPath();
var reader = new InputDataReader();
reader.SetPagesFromInputFile(filePath);
var pages = reader.Pages;
//init arrays
Infoboxes = new List<Infobox>();
Geoboxes = new List<Geobox>();
Citations = new List<Citation>();
//init arrays for output
Persons = new List<string>();
Organizations = new List<string>();
Locations = new List<string>();
CategorizeProperties(pages);
//get entities
Persons = Persons.Select(x => WordUtils.TrimNonLetterCharacters(x, true)).OrderBy(x => x).Distinct(StringComparer.CurrentCultureIgnoreCase).ToList();
Organizations = Organizations.Select(x => WordUtils.TrimNonLetterCharacters(x)).OrderBy(x => x).Distinct(StringComparer.CurrentCultureIgnoreCase).ToList();
Locations = Locations.Select(x => WordUtils.TrimNonLetterCharacters(x, false)).OrderBy(x => x).Distinct(StringComparer.CurrentCultureIgnoreCase).ToList();
//write data into output files
NamedEntityWriter.WriteData(Persons, PersonsFileName);
NamedEntityWriter.WriteData(Organizations, OrganizationsFileName);
NamedEntityWriter.WriteData(Locations, LocationsFileName);
//start finder
var finder = new Finder(Persons, Organizations, Locations);
finder.Find();
}