protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSurfaceSlotToKnownValue(Name,
HelperDataIfConjunctionParetntId.Name
, Consts.SurfaceSlotDefinedValues.AdjunctConditionClause
, sourceSentence, data);
}
public static List<PostPropcessingActionInfo> ChangePartOfSpeech(string actionName, string columnName, string newPartOfSpeechValue, Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var words = new List<KeyValuePair<int, string>>();
foreach (DataRow row in data.Rows)
{
var sourceId = row.Field<int>(columnName);
if (sourceId > 0)
words.Add(new KeyValuePair<int, string>(sourceId, newPartOfSpeechValue));
}
foreach (var pair in words)
{
var wordId = pair.Key;
var newPartOfSpeech = pair.Value;
var modifiedWord = sourceSentence.Elements.Find(x => x.Id == wordId);
var oldModWordPartOfSpeech = modifiedWord.GrammarInfo.PartOfSpeech.Value;
modifiedWord.GrammarInfo.PartOfSpeech = PartOfSpeech.ReadFromString(newPartOfSpeech);
result.Add(new PostPropcessingActionInfo(actionName, wordId, nameof(modifiedWord.GrammarInfo.PartOfSpeech), oldModWordPartOfSpeech, modifiedWord.GrammarInfo.PartOfSpeech.Value));
}
return result;
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeTextValue(Name,
HelperDataFalseDirectObjectReconstructionSubstantivatorControllerId.Name
, HelperDataFalseDirectObjectReconstructionSubstantivatorControllerText.Name
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangePartOfSpeech(Name,
HelperDataFindFalseParticleIIds.Name
, PartOfSpeech.Conjunction.Value
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSyntacticParent(Name,
HelperDataFindFalseConjunctionSParetnForSId.Name
, HelperDataFindFalseConjunctionSNewGrandParentId.Name
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangePartOfSpeech(Name,
HelperDataColumnElementId.Name
, PartOfSpeech.Preposition.Value
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSyntacticParent(Name,
HelperDataConjunctedWordsForPSSWordId.Name
, HelperDataConjunctedWordsForPSSConjunctedId.Name
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSyntacticParent(Name,
HelperDataUPAdjectiveForElliptedNounPESTrue.Name
, HelperDataUPAdjectiveForElliptedNounSSPES.Name
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var wordPairs = new Dictionary<int, int>();
var firstGroupingColumnName = HelperDataQuasiUniformPartsAttachmentMainType2ConfirmedNr.Name;
var secondGroupingColumnName = HelperDataQuasiUniformPartsAttachmentMainType2ParentNr.Name;
var sourceColumnName = HelperDataQuasiUniformPartsAttachmentMainType2ParentId.Name;
var destColumnName = HelperDataQuasiUniformPartsAttachmentMainType2ConfirmedId.Name;
foreach (DataRow row in data.Rows)
{
if (row.Field<int>(firstGroupingColumnName) > -1)
row.SetField(secondGroupingColumnName, row.Field<int>(firstGroupingColumnName));
if (row.Field<int>(secondGroupingColumnName) > -1)
row.SetField(firstGroupingColumnName, row.Field<int>(secondGroupingColumnName));
}
foreach (var group in data.AsEnumerable().GroupBy(grp => grp.Field<int>(firstGroupingColumnName)))
{
var parentColValues = group.ToList().Find(x => x.Field<int>(sourceColumnName) > -1);
var childColValues = group.ToList().Find(x => x.Field<int>(destColumnName) > -1);
var parentId = parentColValues?.Field<int>(sourceColumnName) ?? -1;
var childId = childColValues?.Field<int>(destColumnName) ?? -1;
if (parentId > 0 && childId > 0)
{
wordPairs.Add(parentId, childId);
}
}
PostProcessingUtils.AddConjunctionLink(Name, wordPairs, sourceSentence, result);
return result;
}
public int ConjunctedWordId(Sentence sent)
{
// ищем соч связь между прилагательными через соч связь существительных
if (FalseParentId > -1)
{
var falseElement = sent.GetElement(FalseParentId);
var conjunctedWord = sent.Elements.Find(x => falseElement.ConjunctedWithId == x.Id);
var firstChild = sent.SyntacticChildren(conjunctedWord).First(x =>
(x.SyntacticRole.Value == SyntacticRole.AgreedAttribute.Value
|| x.SyntacticRole.Value == SyntacticRole.ConcordantAttr.Value)
&& x.GrammarInfo.IsSingular
&& x.ConjunctedWithId == -1);
return firstChild.Id;
}
else if (SuspectedParentId > -1)
{
var suspectedElement = sent.GetElement(SuspectedParentId);
var conjunctedWord = sent.Elements.Find(x => x.ConjunctedWithId == suspectedElement.Id);
var firstChild = sent.SyntacticChildren(conjunctedWord).First(x =>
(x.SyntacticRole.Value == SyntacticRole.AgreedAttribute.Value
|| x.SyntacticRole.Value == SyntacticRole.ConcordantAttr.Value)
&& x.GrammarInfo.IsSingular
&& x.ConjunctedWithId == -1);
return firstChild.Id;
}
return -1;
}
public static List<PredicatePartHelperObject> Analyze(int wordId, PredicatesBFinalResults predicates,
Sentence sentence)
{
Debug.Assert(sentence.Elements.Count > 0);
if (predicates.All.Count == 0)
return null;
Log.InfoFormat("Запущен алгоритм поиска управляющего сказуемого для слова с Id = {0}.", wordId);
var canContinue = true;
var parentElement = sentence.SyntacticParent(wordId);
while (canContinue)
{
var rootPredicates = predicates.All.FindAll(x => x.PartType == PredicatePartType.Root);
if (rootPredicates.Select(x => x.Id).Contains(parentElement.Id))
return
predicates.Items.Find(
x => x.Find(y => y.PartType == PredicatePartType.Root).Id == parentElement.Id);
parentElement = sentence.SyntacticParent(parentElement);
canContinue = parentElement != null;
}
return null;
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSyntacticParent(Name
, HelperDataElliptedNounLinkFeatureCoordinatingConjunctionId.Name
, HelperDataElliptedNounLinkFeatureSSPESId.Name
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSurfaceSlotToKnownValue(Name,
HelperDataThenConjunctionId.Name
, Consts.SurfaceSlotDefinedValues.CorrelateCondTemp
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSyntacticParent(Name
, HelperDataConcordanceResultlSPSSId.Name
, HelperDataLostConjunctionLinkConjunctedWithSPSSParentId.Name
, sourceSentence, data);
}
public void Process(Sentence sourceSentence, BaseHelperData data)
{
Results = DoProcess(sourceSentence, data);
if (Results.Count > 0)
Log.InfoFormat("Выполнены действия по модификации лога:{0}{1}", Environment.NewLine, _dumpPostProcessingResults(Results));
else
Log.InfoFormat("Действий по модификации лога не было выполнено. [{0}] ", Name);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
return PostProcessingUtils.ChangeSurfaceSlot(Name,
HelperDataQuasiUniformPartsAttachmentMainType2ConfirmedId.Name
, HelperDataQuasiUniformPartsAttachmentMainType2ParentId.Name
, HelperDataQuasiUniformPartsAttachmentMainType2ConfirmedNr.Name
, HelperDataQuasiUniformPartsAttachmentMainType2ParentNr.Name
, sourceSentence, data);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var predicateValues = data.ColumnIntPositiveValues(HelperDataNearestPredicateAfterThenPredicateId.Name);
var predicateValue = -1;
if (predicateValues.Any())
predicateValue = predicateValues.First();
return PostProcessingUtils.ChangeSyntacticParentForKnownValue(Name,
HelperDataIfConjunctionParetntId.Name,
predicateValue
, sourceSentence, data);
}
public void ApplyAlgorithms(string initialText,
Action<AlgorithmsStageType, BaseSentenceAlgorithm, BaseSentenceAlgorithmData> singleAlgorithmCompleteCallback,
Action<AlgorithmsStageType, BaseSentenceAlgorithm> singleAlgorithmStartCallback)
{
// сначала просто парсим, а потом применяем потоком все добавленные алгоритмы
Log.InfoFormat("Применяем базовый алгоритм: {0}", _baseAlgorithm.Info.InternalName);
(_baseAlgorithm as ComprenoSentenceParseAlgorithm).InitialText = initialText;
var algResult = _baseAlgorithm.Process(null, true);
InitialSentece = algResult.Sentence;
_doApplyAlgorithms(singleAlgorithmCompleteCallback, singleAlgorithmStartCallback);
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var wordPairs = new Dictionary<int, int>();
// добавляем сочинительную связь между ССПСС и ПСС
foreach (DataRow row in data.Rows)
if (row.Field<int>(HelperDataConjunctedWordsForPSSConjunctedId.Name) > 0)
wordPairs.Add(row.Field<int>(HelperDataConjunctedWordsForPSSConjunctedId.Name), row.Field<int>(HelperDataConjunctedWordsForPSSWordId.Name));
PostProcessingUtils.AddConjunctionLink(Name, wordPairs, sourceSentence, result);
return result;
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var words = new List<int>();
foreach (DataRow row in data.Rows)
{
var sspWordId = row.Field<int>(HelperDataFindElliptedNounIntermedaiteFalseUpId.Name);
if (sspWordId > 0)
words.Add(sspWordId);
}
PostProcessingUtils.DeleteConjunctionLink(Name, words, sourceSentence, result);
return result;
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var words = new List<int>();
foreach (DataRow row in data.Rows)
{
var suspectedId = row.Field<int>(HelperDataBetweenAndISuspectedId.Name);
if (suspectedId > 0)
words.Add(suspectedId);
}
PostProcessingUtils.DeleteConjunctionLink(Name, words, sourceSentence, result);
return result;
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var words = new List<KeyValuePair<int, string>>();
foreach (DataRow row in data.Rows)
{
var parentForS = row.Field<int>(HelperDataFindFalseConjunctionSParetnForSId.Name);
if (parentForS > 0)
words.Add(new KeyValuePair<int, string>(parentForS, SyntacticRole.IndirectObject.Value));
}
PostProcessingUtils.ChangeSyntacticRole(Name, words, sourceSentence, result);
return result;
}
public static Sentence Copy(Sentence sourceSentence)
{
var result = new Sentence
{
ParsingResult = ParsingResult.ReadFromString(sourceSentence.ParsingResult.Value),
Text = sourceSentence.Text,
XML = sourceSentence.XML,
};
foreach (var element in sourceSentence.Elements)
result.Elements.Add(SentenceElement.Copy(element));
result.Version++;
return result;
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var wordPairs = new Dictionary<int, int>();
foreach (DataRow row in data.Rows)
{
var pesId = row.Field<int>(HelperDataUPAdjectiveForElliptedNounPESTrue.Name);
var sspesId = row.Field<int>(HelperDataUPAdjectiveForElliptedNounSSPES.Name);
if (pesId > -1 && sspesId > -1)
wordPairs.Add(pesId, sspesId);
}
// добавляем сочинительную связь между ССПСС и ПСС
PostProcessingUtils.AddConjunctionLink(Name, wordPairs, sourceSentence, result);
return result;
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var wordPairs = new Dictionary<int, int>();
foreach (DataRow row in data.Rows)
{
var parentId = row.Field<int>(HelperDataLostConjunctionLinkConjunctedWithSPSSId.Name);
var childId = row.Field<int>(HelperDataConcordanceResultlSPSSId.Name);
if (parentId > 0 && childId > 0)
{
wordPairs.Add(parentId, childId);
}
}
PostProcessingUtils.AddConjunctionLink(Name, wordPairs, sourceSentence, result);
return result;
}
private object[] _processParticiple(SentenceElement element, Sentence sentence, UniformPartsFinalResults upRows)
{
var participleId = element.Id;
var hasChildrenNotServicePart = sentence.SyntacticChildren(participleId).Any(x =>
x.SyntacticRole.Value != SyntacticRole.ParticleNegative.Value
&& x.SyntacticRole.Value != SyntacticRole.CoordinatingConjunction.Value
&& x.SyntacticRole.Value != SyntacticRole.Particle.Value);
if (hasChildrenNotServicePart)
{
Log.InfoFormat(" Обнаружили потомка не СЧР. Причастие -- Причастие Финальное 1");
var finalParticiple1 = participleId;
return new object[] {participleId, finalParticiple1, -1, finalParticiple1, -1};
}
Log.InfoFormat("Причастие - Причастие - предварительное 2)");
if (!upRows.All.Any(x => x.Id == participleId))
{
Log.InfoFormat(" Причастие не принадлежит ни одному из рядов. Причастие -- Признак 1");
var participleFeature = participleId;
return new object[] {participleId, -1, -1, -1, participleFeature};
}
var rowNr = upRows.All.Find(x => x.Id == participleId).RowNr;
Log.InfoFormat(" Причастие принадлежит к ряду ОЧ {0}. Причастие - предваритиельное 3", rowNr);
if (sentence.Elements.FindAll(x => upRows.All.Select(y => y.Id).Contains(x.SyntacticParentWordId)).Any(x =>
x.SyntacticRole.Value != SyntacticRole.ParticleNegative.Value
&& x.SyntacticRole.Value != SyntacticRole.CoordinatingConjunction.Value
&& x.SyntacticRole.Value != SyntacticRole.Particle.Value))
{
Log.InfoFormat(" У сочиненных с рассматриваемым причастий нет потомков не СЧР. Причастие -- Признак 2");
var participleFeature = participleId;
return new object[] { participleId, -1, -1, -1, participleFeature };
}
else
{
Log.InfoFormat(" У сочиненных с рассматриваемым причастий есть хотя бы один потомок не СЧР. Причастие -- Финальное 2");
var participleFinal2 = participleId;
return new object[] { participleId, -1, participleFinal2, participleFinal2, -1 };
}
}
public static void AddConjunctionLink(string actionName, Dictionary<int, int> wordPairs, Sentence sourceSentence, List<PostPropcessingActionInfo> results)
{
foreach (var pair in wordPairs)
{
var leftWord = sourceSentence.Elements.Find(x => x.Id == pair.Key);
var rightWord = sourceSentence.Elements.Find(x => x.Id == pair.Value);
if (leftWord != null && rightWord != null)
{
var oldLeftWordConjuctedWithId = leftWord.ConjunctedWithId;
var oldLeftWordConjuctedWithGUID = leftWord.ConjunctedWithGUID;
leftWord.ConjunctedWithId = rightWord.Id;
leftWord.ConjunctedWithGUID = rightWord.GUID;
results.Add(new PostPropcessingActionInfo(actionName, pair.Key, nameof(leftWord.ConjunctedWithId), oldLeftWordConjuctedWithId.ToString(), leftWord.ConjunctedWithId.ToString()));
results.Add(new PostPropcessingActionInfo(actionName, pair.Key, nameof(leftWord.ConjunctedWithGUID), oldLeftWordConjuctedWithGUID, leftWord.ConjunctedWithGUID));
}
}
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var words = new List<int>();
foreach (DataRow row in data.Rows)
{
var sspWordId = row.Field<int>(HelperDataSubjectAndPredicateConjunctionSubjectIdSSP.Name);
if (sspWordId > 0)
words.Add(sspWordId);
var pssWordId = row.Field<int>(HelperDataSubjectAndPredicateConjunctionPredicateIdPSS.Name);
if (pssWordId > 0)
words.Add(pssWordId);
}
PostProcessingUtils.DeleteConjunctionLink(Name, words, sourceSentence, result);
return result;
}
private void _prepareData(DataRowCollection attachInfos, List<QuasiUPAttachmentHelperObject> attachments, Sentence sentence)
{
if (attachInfos.Count <= 0) return;
foreach (DataRow attachInfo in attachInfos)
{
// 3.6
var attachIdType1 =
attachInfo.Field<int>(HelperDataQuasiUniformPartsAttachmentPreliminaryIdType1.Name);
// 3.2
var attachIdType2 =
attachInfo.Field<int>(HelperDataQuasiUniformPartsAttachmentPreliminaryIdType2.Name);
// 102
var attachNr = attachInfo.Field<int>(HelperDataQuasiUniformPartsAttachmentPreliminaryNr.Name);
if (attachIdType1 > 0)
{
var newAttach = new QuasiUPAttachmentHelperObject(sentence.Elements.Find(x => x.Id == attachIdType1))
{
IsInitiallyType1 = true,
IsInitiallyType2 = false,
SourceNr = attachNr
};
Log.InfoFormat($"Создан КвазиОЧ Т1: Id: {newAttach.Id}, Текст: {newAttach.Text}");
attachments.Add(newAttach);
}
else if (attachIdType2 > 0)
{
var newAttach = new QuasiUPAttachmentHelperObject(sentence.Elements.Find(x => x.Id == attachIdType2))
{
IsInitiallyType2 = true,
IsInitiallyType1 = false,
SourceNr = attachNr
};
Log.InfoFormat($"Создан КвазиОЧ Т2: Id: {newAttach.Id}, Текст: {newAttach.Text}");
attachments.Add(newAttach);
}
}
}
protected override List<PostPropcessingActionInfo> DoProcess(Sentence sourceSentence, BaseHelperData data)
{
var result = new List<PostPropcessingActionInfo>();
var wordPairsgroups =
data.Rows.Cast<DataRow>().GroupBy(grp => grp.Field<int>(HelperDataParticleISourceId.Name));
var wordPairs = new Dictionary<int, int>();
foreach (var group in wordPairsgroups)
{
wordPairs.Add(
group.Select(x => x.Field<int>(HelperDataParticleIParentConjuctedWordId.Name)).First(x => x > -1),
group.Select(x => x.Field<int>(HelperDataParticleIParentWordId.Name)).First(x => x > -1)
);
}
// добавляем сочинительную связь между родитилем И, и текстовым элементом, имеющим такую синт роль
PostProcessingUtils.AddConjunctionLink(Name, wordPairs, sourceSentence, result);
return result;
}