/// <summary>
/// Adds the specified sufix: $ is last letter of nominative, # is n, t or s
/// </summary>
/// <param name="sufix">The sufix.</param>
/// <param name="gram">The gram.</param>
/// <returns></returns>
public morphRule Add(string sufix, gramFlags gram)
{
gram.type = type;
morphRule output = new morphRule(sufix, gram);
rules.Add(output);
return(output);
}
public wordnetSymsetResults query_srb(List <String> srb_tokens, ILogBuilder response, Boolean buildModel = true)
{
wordnetSymsetResults output = new wordnetSymsetResults();
getReady();
List <DataRow> matches = new List <DataRow>();
foreach (String tkn in srb_tokens)
{
matches.AddRangeUnique(table.Select(SRB_COLUMN_TOKEN + " LIKE '" + tkn + "'"));
}
foreach (DataRow dr in matches)
{
String eng = dr[SRB_COLUMN_TOKEN].toStringSafe();
String code = dr[SRB_COLUMN_CODE].toStringSafe();
output.Add(code, eng);
}
if (buildModel)
{
foreach (var pair in output)
{
termExploreModel md = null;
String srb = pair.Key;
if (!output.models.ContainsKey(srb))
{
md = new termExploreModel(srb);
output.models.Add(srb, md);
}
else
{
md = output.models[srb];
}
String symc = pair.Value[0].ToString();
pos_type pt = posConverter.wordNetFirstNumToPosType.getValue(symc, pos_type.none);
gramFlags gr = new gramFlags();
gr.Set(pt);
md.gramSet.Add(gr);
md.wordnetPrimarySymsets.AddUnique(pair.Value);
}
}
return(output);
}
public bool discoverGram(termExploreItem item, ILogBuilder loger, bool debug = true)
{
//List<termExploreItem> inst = new List<termExploreItem>();
//exploreModel.instances.ForEach(x => inst.Add(x));
//inst.Add(exploreModel);
// instanceCountCollection<pos_type> pct = new instanceCountCollection<pos_type>();
bool failed = false;
//// <--------------- Trying to resolve alone
//foreach (termExploreItem item in inst)
//{
if (loger != null)
{
loger.AppendLine("Item:" + item.inputForm);
}
instanceCountCollection <object> res = termDiscoveryResolver.resolveQuery(item.inputForm);
res.reCalculate();
if (res.Count > 0)
{
List <object> sorted = res.getSorted();
if (item.gramSet.getPosType() != pos_type.none)
{
sorted.RemoveAll(x => x is pos_type);
}
gramFlags gf = new gramFlags();
if (sorted.Any(x => x is pos_type))
{
gf.Set((pos_type)sorted.First(x => x is pos_type));
}
//pct.AddInstance(gf.type, 1);
var tl = posConverter.posTypeVsPattern[gf.type];
sorted.RemoveAll(x => !tl.Contains(x.GetType()));
if (loger != null)
{
loger.AppendLine("Votes:");
for (int i = 0; i < Math.Max(sorted.Count(), 20); i++)
{
loger.Append(sorted[i].ToString() + "; ");
}
}
if (sorted.Any(x => x is pos_gender))
{
gf.Set((pos_gender)sorted.First(x => x is pos_gender));
}
if (sorted.Any(x => x is pos_gramaticalCase))
{
gf.Set((pos_gramaticalCase)sorted.First(x => x is pos_gramaticalCase));
}
if (sorted.Any(x => x is pos_verbform))
{
gf.Set((pos_verbform)sorted.First(x => x is pos_verbform));
}
if (sorted.Any(x => x is pos_number))
{
gf.Set((pos_number)sorted.First(x => x is pos_number));
}
if (sorted.Any(x => x is pos_degree))
{
gf.Set((pos_degree)sorted.First(x => x is pos_degree));
}
if (sorted.Any(x => x is pos_person))
{
gf.Set((pos_person)sorted.First(x => x is pos_person));
}
if (loger != null)
{
loger.AppendLine("Final gram:" + gf.ToString());
}
item.gramSet.Add(gf);
}
else
{
if (item.inputForm.Length < 4)
{
return(false);
}
//item.flags = termExploreItemEnumFlag.none;
failed = true;
}
return(failed);
}
/// <summary>
/// Explores definition on an unknown term
/// </summary>
/// <param name="term">The term.</param>
/// <param name="loger">The loger.</param>
/// <param name="shortExplore">if set to <c>true</c> [short explore].</param>
/// <param name="debug">if set to <c>true</c> [debug].</param>
/// <returns></returns>
public List <termExploreModel> explore(string term, ILogBuilder loger, bool shortExplore = true, bool debug = true, termExploreModel exploreModel = null)
{
term = term.Trim();
List <termExploreModel> output = new List <termExploreModel>();
if (modelRegistry.ContainsKey(term))
{
return(modelRegistry[term]);
}
if (missing.Contains(term))
{
return(GetModels(term));
}
if (term.isNumber())
{
termExploreModel tmp = makeTempModel(term, pos_type.NUMnumerical);
tmp.flags = termExploreItemEnumFlag.datapoint;
if (loger != null)
{
loger.AppendLine("Term [" + term + "] detected to be number.");
}
AddModel(tmp);
return(GetModels(term));
}
// <----- drugi test
exploreModel = termExploreProcedures.exploreWithHunspell(new termExploreItem(term), loger);
List <string> suggests = new List <string>();
exploreModel.instances.ForEach(x => suggests.Add(x.inputForm));
//languageManagerDBNamedEntities.manager.exploreEntities(exploreModel.rootWord, exploreModel);
suggests.Add(exploreModel.rootWord);
// s
apertiumDictionaryResult result = languageManagerApertium.manager.query(suggests, apertiumDictQueryScope.exact, apertiumDictNeedleSide.serbian);
if (result.Any())
{
List <termExploreItem> gramCheck = new List <termExploreItem>();
gramFlags gr = null;
if (result.termVsGramFlags.ContainsKey(exploreModel.inputForm))
{
exploreModel.gramSet.Add(new gramFlags(result.termVsGramFlags[exploreModel.inputForm]));
if (exploreModel.lemma == null)
{
exploreModel.lemma = exploreModel.instances[exploreModel.inputForm];
}
gramCheck.Add(exploreModel);
if (debug)
{
if (loger != null)
{
loger.AppendLine("Apertium discovered model [" + exploreModel.inputForm + "]");
}
}
}
else
{
//if (loger != null) loger.AppendLine("Apertium failed to discover [" + exploreModel.inputForm + "]");
}
foreach (termExploreItem item in exploreModel.instances)
{
if (result.termVsGramFlags.ContainsKey(item.inputForm))
{
item.gramSet.Add(new gramFlags(result.termVsGramFlags[item.inputForm]));
gramCheck.Add(exploreModel);
exploreModel.lemmaForm = item.inputForm;
if (exploreModel.lemma == null)
{
exploreModel.lemma = item;
}
if (debug)
{
if (loger != null)
{
loger.AppendLine("Apertium discovered model [" + item.inputForm + "]");
}
}
}
else
{
//if (loger != null) loger.AppendLine("Apertium failed to discover [" + item.inputForm + "]");
}
}
exploreModel.translations.AddRange(result.GetEnglish());
gramCheck.RemoveAll(x => posConverter.posTypeVsPattern[x.gramSet.getPosType()].Count() == 0);
int disc = 0;
foreach (var gram in gramCheck)
{
if (discoverGram(gram, loger, debug))
{
disc++;
}
}
if (loger != null)
{
loger.AppendLine("Gram [" + term + "] autodiscovered for [" + disc + "] / [" + gramCheck.Count() + "]");
}
if (debug)
{
if (loger != null)
{
exploreModel.ToString(loger, true, false);
manager.constructor.saveTermModel(exploreModel, "Apertium_");
}
}
AddModel(exploreModel);
exploreModel.flags = termExploreItemEnumFlag.aper;
if (shortExplore)
{
return(GetModels(term));
}
}
else
{
if (loger != null)
{
loger.AppendLine("Apertium failed to discover any information on [" + term + "]");
}
}
if (loger != null)
{
loger.consoleAltColorToggle();
}
// <------------------ APERTIUM ^^
foreach (string s in suggests)
{
languageManagerDBNamedEntities.manager.exploreEntities(s, exploreModel);
}
if (exploreModel.flags == termExploreItemEnumFlag.namedEntity)
{
AddModel(exploreModel);
if (debug)
{
if (loger != null)
{
exploreModel.ToString(loger, true, false);
manager.constructor.saveTermModel(exploreModel, "NamedEntity_");
loger.AppendLine("Named entities discovered model [" + exploreModel.inputForm + "]:" + exploreModel.gramSet.ToString());
}
}
if (shortExplore)
{
return(GetModels(term));
}
}
else
{
if (loger != null)
{
if (debug)
{
loger.AppendLine("Named entities found nothing for [" + exploreModel.inputForm + "]:" + exploreModel.gramSet.ToString());
}
}
}
if (loger != null)
{
loger.consoleAltColorToggle();
}
// <------------------ NAMED ENTITY ^^
// <----------------- Wordnet
wordnetSymsetResults resSrWordnet = languageManagerWordnet.manager.query_srb(suggests, loger);
bool found = false;
if (resSrWordnet.Any())
{
foreach (termExploreItem item in exploreModel.instances)
{
if (resSrWordnet.GetByKey(item.inputForm).Any())
{
exploreModel.lemma = item;
exploreModel.lemmaForm = item.inputForm;
exploreModel.translations.AddRange(resSrWordnet.GetValues());
exploreModel.synonyms.AddRange(resSrWordnet.GetKeys());
exploreModel.flags = termExploreItemEnumFlag.srWNet;
found = true;
item.gramSet.Add(new gramFlags(new Enum[] { resSrWordnet.models[item.inputForm].gramSet.getPosType() }));
}
}
foreach (termExploreItem item in exploreModel.instances)
{
discoverGram(item, loger, debug);
}
}
if (found)
{
if (loger != null)
{
loger.AppendLine("SerbianWordNet discovered model [" + term + "]:" + exploreModel.gramSet.ToString());
}
if (debug)
{
if (loger != null)
{
exploreModel.ToString(loger, true, false);
manager.constructor.saveTermModel(exploreModel, "SrWordNet_");;
}
}
AddModel(exploreModel);
exploreModel.flags = termExploreItemEnumFlag.srWNet;
if (shortExplore)
{
return(GetModels(term));
}
}
else
{
if (loger != null)
{
if (debug)
{
loger.AppendLine("Serbian wordnet found nothing for [" + term + "]");
}
}
}
// <------------------ SERBIAN WORD NET ^^
bool failed = discoverGram(exploreModel, loger, debug);
exploreModel.instances.ForEach(x => discoverGram(x, loger, debug));
int d = 0;
List <termExploreItem> lastCheck = new List <termExploreItem>();
foreach (var gram in lastCheck)
{
if (discoverGram(gram, loger, debug))
{
d++;
}
}
if (debug)
{
if (loger != null)
{
loger.AppendLine("The last check [" + term + "] autodiscovered for [" + d + "] / [" + lastCheck.Count() + "]");
}
}
if (d == 0)
{
failed = true;
}
if (loger != null)
{
loger.consoleAltColorToggle();
}
// <------------------ LAST CHECK ^^
if (!failed)
{
exploreModel.flags = termExploreItemEnumFlag.termExplorer;
AddModel(exploreModel);
return(GetModels(term));
}
else
{
if (debug)
{
if (loger != null)
{
loger.AppendLine("Exploration failed for [" + term + "] -- creating temporary term model");
}
}
output.Add(makeTempModel(term, pos_type.TEMP));
missing.Add(term);
return(output);
}
}
public morphRule(string __sufix, gramFlags __gram)
{
sufix = __sufix;
gramEntry = __gram;
}