Labojumi pārtaisītajās redukcijās.

LUMII-AILab · Apr 10, 2019 · 7128e3f · 7128e3f
1 parent b758124
commit 7128e3f
Show file tree

Hide file tree

Showing 5 changed files with 50 additions and 29 deletions.
diff --git a/LVTB2UD/src/lv/ailab/lvtb/universalizer/pml/xmldom/XmlDomANode.java b/LVTB2UD/src/lv/ailab/lvtb/universalizer/pml/xmldom/XmlDomANode.java
@@ -6,6 +6,7 @@
 import lv.ailab.lvtb.universalizer.pml.PmlANode;
 import lv.ailab.lvtb.universalizer.pml.utils.PmlANodeListUtils;
 import lv.ailab.lvtb.universalizer.transformator.morpho.AnalyzerWrapper;
+import org.w3c.dom.Document;
 import org.w3c.dom.Element;
 import org.w3c.dom.Node;
 import org.w3c.dom.NodeList;
@@ -739,17 +740,20 @@ public boolean splitMorphoEllipsis(String idPostfix)
 		if (reductionField == null || reductionField.isEmpty()) return false;
 		try
 		{
+			XmlDomANode parentToAppend = getParent();
+			if (!parentToAppend.isPhraseNode()) parentToAppend = this;
+			Document ownerDoc = domNode.getOwnerDocument();
 			// Children container
 			Node childenNode = (Node) XPathEngine.get().evaluate(
-					"./children", domNode, XPathConstants.NODE);
+					"./children", parentToAppend.domNode, XPathConstants.NODE);
 			if (childenNode == null)
 			{
-				childenNode = domNode.getOwnerDocument().createElement("children");
-				domNode.appendChild(childenNode);
+				childenNode = ownerDoc.createElement("children");
+				parentToAppend.domNode.appendChild(childenNode);
 			}
 
 			// Node itself
-			Element newTokenNode = domNode.getOwnerDocument().createElement("node");
+			Element newTokenNode = ownerDoc.createElement("node");
 			childenNode.appendChild(newTokenNode);
 
 			// id attribute
@@ -769,9 +773,9 @@ public boolean splitMorphoEllipsis(String idPostfix)
 			newTokenNode.appendChild(ord);
 
 			// Role.
-			Node roleNode = domNode.getOwnerDocument().createElement("role");
+			Node roleNode = ownerDoc.createElement("role");
 			newTokenNode.appendChild(roleNode);
-			roleNode.appendChild(domNode.getOwnerDocument().createTextNode(LvtbRoles.ELLIPSIS_TOKEN));
+			roleNode.appendChild(ownerDoc.createTextNode(LvtbRoles.ELLIPSIS_TOKEN));
 
 			return true;
 		}

diff --git a/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/Sentence.java b/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/Sentence.java
@@ -194,8 +194,12 @@ protected HashMap<String, HashSet<String>> getGov2subj(PmlANode aNode, HashMap<S
 						&& ((phrasePart.getM() != null && MorphoTransformator.isTrueAux(phrasePart.getM().getLemma()))
 							|| MorphoTransformator.isTrueAux(phrasePart.getReductionLemma())))
 					continue;
+				// Do not add coordination conjuctions and punctuation.
 				if (LvtbCoordTypes.CRDPARTS.equals(phraseType) && !LvtbRoles.CRDPART.equals(partRole))
 					continue;
+				// Do not add split ellipsis nodes.
+				if (LvtbRoles.ELLIPSIS_TOKEN.equals(partRole))
+					continue;
 
 				String partId = phrasePart.getId();
 				HashSet<String> collectedPartSubjs = resultAccumulator.get(partId);

diff --git a/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/syntax/NewSyntaxTransformator.java b/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/syntax/NewSyntaxTransformator.java
@@ -1,8 +1,6 @@
 package lv.ailab.lvtb.universalizer.transformator.syntax;
 
-import lv.ailab.lvtb.universalizer.pml.LvtbRoles;
 import lv.ailab.lvtb.universalizer.pml.PmlANode;
-import lv.ailab.lvtb.universalizer.pml.utils.PmlANodeListUtils;
 import lv.ailab.lvtb.universalizer.transformator.Sentence;
 import lv.ailab.lvtb.universalizer.transformator.TransformationParams;
 import lv.ailab.lvtb.universalizer.transformator.morpho.XPosLogic;
@@ -139,9 +137,11 @@ else if (aNode.isPureReductionNode())
 			if (redXPostag.matches("v..([^p].*|p[du].*)") || ! params.UD_STANDARD_NULLNODES)
 			{
 				String newIdStub = newBasicRoot.getId();
-				List<PmlANode> tokenNodes = aNode.getChildren(LvtbRoles.ELLIPSIS_TOKEN);
-				if (tokenNodes != null && !tokenNodes.isEmpty())
-					newIdStub = PmlANodeListUtils.getFirstByDescOrd(tokenNodes).getId();
+				//List<PmlANode> tokenNodes = aNode.getChildren(LvtbRoles.ELLIPSIS_TOKEN);
+				//if (tokenNodes != null && !tokenNodes.isEmpty())
+				//	newIdStub = PmlANodeListUtils.getFirstByDescOrd(tokenNodes).getId();
+				PmlANode tokenNode = s.pmlTree.getDescendant(aNode.getId() + Sentence.ID_POSTFIX);
+				if (tokenNode != null) newIdStub = tokenNode.getId();
 				s.createNewEnhEllipsisNode(aNode, newIdStub, params.ADD_NODE_IDS);
 			}
 

diff --git a/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/syntax/PhrasePartDepLogic.java b/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/syntax/PhrasePartDepLogic.java
@@ -266,6 +266,12 @@ else if (nominal && ultimateAux)
 				return Tuple.of(UDv2Relations.XCOMP, null);
 		}
 
+		if (lvtbRole.equals(LvtbRoles.ELLIPSIS_TOKEN))
+		{
+			String tag = aNode.getAnyTag();
+			if (tag.matches("z.*")) return Tuple.of(UDv2Relations.PUNCT, null);
+		}
+
 		StandardLogger.l.doInsentenceWarning(String.format(
 				"\"%s\" (%s) in \"%s\" has no UD label.",
 				lvtbRole, nodeId, phraseType));

diff --git a/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/syntax/PhraseTransformator.java b/LVTB2UD/src/lv/ailab/lvtb/universalizer/transformator/syntax/PhraseTransformator.java
@@ -675,36 +675,43 @@ protected PmlANode noModXPredToUD(PmlANode xNode)
 	{
 		// Get basElems and warn if there is none.
 		List<PmlANode> basElems = xNode.getChildren(LvtbRoles.BASELEM);
+		if (basElems != null && basElems.size() > 1)
+			throw new IllegalArgumentException(String.format(
+					"\"%s\" in sentence \"%s\" has multiple \"basElem\"",
+					xNode.getPhraseType(), s.id));
 		PmlANode basElem = PmlANodeListUtils.getLastByDeepOrd(basElems);
 		if (basElem == null)
 			throw new IllegalArgumentException(String.format(
 					"\"%s\" in sentence \"%s\" has no \"basElem\"",
 					xNode.getPhraseType(), s.id));
+		PmlANode newRoot = basElem;
+
 		List<PmlANode> auxes = xNode.getChildren(LvtbRoles.AUXVERB);
 		PmlANode lastAux = PmlANodeListUtils.getLastByDeepOrd(auxes);
-		if (lastAux == null)
+		/*if (lastAux == null)
 			throw new IllegalArgumentException(String.format(
 					"\"%s\" in sentence \"%s\" has neither \"auxVerb\" nor \"mod\"",
-					xNode.getPhraseType(), s.id));
-		if (auxes.size() > 1) for (int i = 0; i < auxes.size(); i++)
+					xNode.getPhraseType(), s.id));*/
+		if (lastAux != null)
 		{
-			String auxLemma = lastAux.getM().getLemma();
+			if (auxes.size() > 1) for (int i = 0; i < auxes.size(); i++)
+			{
+				String auxLemma = lastAux.getM().getLemma();
+				String auxRedLemma = lastAux.getReductionLemma();
+				if (!MorphoTransformator.isTrueAux(auxLemma) &&
+						!MorphoTransformator.isTrueAux(auxRedLemma))
+					StandardLogger.l.doInsentenceWarning(String.format(
+							"xPred \"%s\" has multiple auxVerb one of which has lemma \"%s\".",
+							xNode.getParent().getId(), auxLemma));
+			}
+
+			PmlMNode lastAuxM = lastAux.getM();
+			String auxLemma = lastAuxM == null ? null : lastAuxM.getLemma();
 			String auxRedLemma = lastAux.getReductionLemma();
-			if (!MorphoTransformator.isTrueAux(auxLemma) &&
-					!MorphoTransformator.isTrueAux(auxRedLemma))
-				StandardLogger.l.doInsentenceWarning(String.format(
-						"xPred \"%s\" has multiple auxVerb one of which has lemma \"%s\".",
-						xNode.getParent().getId(), auxLemma));
+			boolean ultimateAux = MorphoTransformator.isTrueAux(auxLemma)
+					|| MorphoTransformator.isTrueAux(auxRedLemma);
+			if (!ultimateAux) newRoot = lastAux;
 		}
-
-		PmlMNode lastAuxM = lastAux.getM();
-		String auxLemma = lastAuxM == null ? null : lastAuxM.getLemma();
-		String auxRedLemma = lastAux.getReductionLemma();
-		boolean ultimateAux = MorphoTransformator.isTrueAux(auxLemma)
-				|| MorphoTransformator.isTrueAux(auxRedLemma);
-
-		PmlANode newRoot = basElem;
-		if (!ultimateAux) newRoot = lastAux;
 		List<PmlANode> children = xNode.getChildren();
 		s.relinkAllConstituents(newRoot, children, xNode,
 				params.PROPAGATE_CONJUNCTS, params.NO_EDEP_DUPLICATES);