public static void Initialize()
{
if (GlobalOrgs != null)
{
return;
}
GlobalOrgs = new Pullenti.Ner.Core.IntOntologyCollection();
Pullenti.Ner.Org.OrganizationReferent org;
Pullenti.Ner.Core.IntOntologyItem oi;
using (Pullenti.Ner.Processor geoProc = Pullenti.Ner.ProcessorService.CreateEmptyProcessor())
{
geoProc.AddAnalyzer(new Pullenti.Ner.Geo.GeoAnalyzer());
Dictionary <string, Pullenti.Ner.Geo.GeoReferent> geos = new Dictionary <string, Pullenti.Ner.Geo.GeoReferent>();
for (int k = 0; k < 3; k++)
{
Pullenti.Morph.MorphLang lang = (k == 0 ? Pullenti.Morph.MorphLang.RU : (k == 1 ? Pullenti.Morph.MorphLang.EN : Pullenti.Morph.MorphLang.UA));
string name = (k == 0 ? "Orgs_ru.dat" : (k == 1 ? "Orgs_en.dat" : "Orgs_ua.dat"));
byte[] dat = ResourceHelper.GetBytes(name);
if (dat == null)
{
throw new Exception(string.Format("Can't file resource file {0} in Organization analyzer", name));
}
using (MemoryStream tmp = new MemoryStream(OrgItemTypeToken.Deflate(dat)))
{
tmp.Position = 0;
XmlDocument xml = new XmlDocument();
xml.Load(tmp);
foreach (XmlNode x in xml.DocumentElement.ChildNodes)
{
org = new Pullenti.Ner.Org.OrganizationReferent();
string abbr = null;
foreach (XmlNode xx in x.ChildNodes)
{
if (xx.LocalName == "typ")
{
org.AddSlot(Pullenti.Ner.Org.OrganizationReferent.ATTR_TYPE, xx.InnerText, false, 0);
}
else if (xx.LocalName == "nam")
{
org.AddSlot(Pullenti.Ner.Org.OrganizationReferent.ATTR_NAME, xx.InnerText, false, 0);
}
else if (xx.LocalName == "epo")
{
org.AddSlot(Pullenti.Ner.Org.OrganizationReferent.ATTR_EPONYM, xx.InnerText, false, 0);
}
else if (xx.LocalName == "prof")
{
org.AddSlot(Pullenti.Ner.Org.OrganizationReferent.ATTR_PROFILE, xx.InnerText, false, 0);
}
else if (xx.LocalName == "abbr")
{
abbr = xx.InnerText;
}
else if (xx.LocalName == "geo")
{
Pullenti.Ner.Geo.GeoReferent geo;
if (!geos.TryGetValue(xx.InnerText, out geo))
{
Pullenti.Ner.AnalysisResult ar = geoProc.Process(new Pullenti.Ner.SourceOfAnalysis(xx.InnerText), null, lang);
if (ar != null && ar.Entities.Count == 1 && (ar.Entities[0] is Pullenti.Ner.Geo.GeoReferent))
{
geo = ar.Entities[0] as Pullenti.Ner.Geo.GeoReferent;
geos.Add(xx.InnerText, geo);
}
else
{
}
}
if (geo != null)
{
org.AddSlot(Pullenti.Ner.Org.OrganizationReferent.ATTR_GEO, geo, false, 0);
}
}
}
oi = org.CreateOntologyItemEx(2, true, true);
if (oi == null)
{
continue;
}
if (abbr != null)
{
oi.Termins.Add(new Pullenti.Ner.Core.Termin(abbr, null, true));
}
if (k == 2)
{
GlobalOrgsUa.AddItem(oi);
}
else
{
GlobalOrgs.AddItem(oi);
}
}
}
}
}
return;
}
public static void Initialize()
{
m_StdTails = new Pullenti.Ner.Core.TerminCollection();
m_StdNames = new Pullenti.Ner.Core.TerminCollection();
m_VervotWords = new Pullenti.Ner.Core.TerminCollection();
Pullenti.Ner.Core.Termin t;
t = new Pullenti.Ner.Core.Termin("INCORPORATED");
t.AddAbridge("INC.");
m_StdTails.Add(t);
t = new Pullenti.Ner.Core.Termin("CORPORATION");
t.AddAbridge("CORP.");
m_StdTails.Add(t);
t = new Pullenti.Ner.Core.Termin("LIMITED");
t.AddAbridge("LTD.");
m_StdTails.Add(t);
t = new Pullenti.Ner.Core.Termin("AG");
m_StdTails.Add(t);
t = new Pullenti.Ner.Core.Termin("GMBH");
m_StdTails.Add(t);
foreach (string s in new string[] { "ЗАКАЗЧИК", "ИСПОЛНИТЕЛЬ", "РАЗРАБОТЧИК", "БЕНЕФИЦИАР", "ПОЛУЧАТЕЛЬ", "ОТПРАВИТЕЛЬ", "ИЗГОТОВИТЕЛЬ", "ПРОИЗВОДИТЕЛЬ", "ПОСТАВЩИК", "АБОНЕНТ", "КЛИЕНТ", "ВКЛАДЧИК", "СУБЪЕКТ", "ПРОДАВЕЦ", "ПОКУПАТЕЛЬ", "АРЕНДОДАТЕЛЬ", "АРЕНДАТОР", "СУБАРЕНДАТОР", "НАЙМОДАТЕЛЬ", "НАНИМАТЕЛЬ", "АГЕНТ", "ПРИНЦИПАЛ", "ПРОДАВЕЦ", "ПОСТАВЩИК", "ПОДРЯДЧИК", "СУБПОДРЯДЧИК" })
{
m_StdTails.Add(new Pullenti.Ner.Core.Termin(s)
{
Tag = s
});
}
foreach (string s in new string[] { "ЗАМОВНИК", "ВИКОНАВЕЦЬ", "РОЗРОБНИК", "БЕНЕФІЦІАР", "ОДЕРЖУВАЧ", "ВІДПРАВНИК", "ВИРОБНИК", "ВИРОБНИК", "ПОСТАЧАЛЬНИК", "АБОНЕНТ", "КЛІЄНТ", "ВКЛАДНИК", "СУБ'ЄКТ", "ПРОДАВЕЦЬ", "ПОКУПЕЦЬ", "ОРЕНДОДАВЕЦЬ", "ОРЕНДАР", "СУБОРЕНДАР", "НАЙМОДАВЕЦЬ", "НАЙМАЧ", "АГЕНТ", "ПРИНЦИПАЛ", "ПРОДАВЕЦЬ", "ПОСТАЧАЛЬНИК", "ПІДРЯДНИК", "СУБПІДРЯДНИК" })
{
m_StdTails.Add(new Pullenti.Ner.Core.Termin(s, Pullenti.Morph.MorphLang.UA)
{
Tag = s
});
}
t = new Pullenti.Ner.Core.Termin("РАЗРАБОТКА ПРОГРАММНОГО ОБЕСПЕЧЕНИЯ");
t.AddAbridge("РАЗРАБОТКИ ПО");
m_StdNames.Add(t);
foreach (string s in new string[] { "СПЕЦИАЛЬНОСТЬ", "ДИАГНОЗ" })
{
m_VervotWords.Add(new Pullenti.Ner.Core.Termin(s));
}
foreach (string s in new string[] { "СПЕЦІАЛЬНІСТЬ", "ДІАГНОЗ" })
{
m_VervotWords.Add(new Pullenti.Ner.Core.Termin(s, Pullenti.Morph.MorphLang.UA));
}
m_StdNouns = new Pullenti.Ner.Core.TerminCollection();
for (int k = 0; k < 2; k++)
{
string name = (k == 0 ? "NameNouns_ru.dat" : "NameNouns_ua.dat");
byte[] dat = ResourceHelper.GetBytes(name);
if (dat == null)
{
throw new Exception(string.Format("Can't file resource file {0} in Organization analyzer", name));
}
string str = Encoding.UTF8.GetString(OrgItemTypeToken.Deflate(dat));
foreach (string line0 in str.Split('\n'))
{
string line = line0.Trim();
if (string.IsNullOrEmpty(line))
{
continue;
}
if (k == 0)
{
m_StdNouns.Add(new Pullenti.Ner.Core.Termin(line));
}
else
{
m_StdNouns.Add(new Pullenti.Ner.Core.Termin(line)
{
Lang = Pullenti.Morph.MorphLang.UA
});
}
}
}
}