improvements for R-pronouns and relativisation (#10)

Co-authored-by: Jan Odijk <[email protected]>

mwe_query/canonicalform.py

@@ -386,7 +386,14 @@ def mknewnode(stree, mwetop, atts, annotations):
             newnode.attrib['maxnodecount'] = f'{len(stree)}'
     return newnode
 
-
+def expandnonheadwordnode(nonheadwordnode, phrasenodeproperties):
+    phraserel = gav(nonheadwordnode, 'rel')
+    newnonheadwordnode = copy.copy(nonheadwordnode)
+    newnonheadwordnode.attrib['rel'] = 'hd'
+    phrasenode = ET.Element('node', attrib=phrasenodeproperties)
+    phrasenode.attrib['rel'] = phraserel
+    phrasenode.append(newnonheadwordnode)
+    return phrasenode
 def zullenheadclause(stree: SynTree) -> bool:
     if stree.tag == 'node':
         cat = gav(stree, 'cat')
@@ -1016,9 +1023,10 @@ def newgenvariants(stree: SynTree) -> List[SynTree]:
             Rpronounobj1node = copy.copy(obj1node)
             Rpronounobj1node.attrib['lemma'] = 'er|hier|daar|waar|ergens|nergens|overal'
             Rpronounobj1node.attrib['pt'] = 'vnw'
+            newphrase = expandnonheadwordnode(Rpronounobj1node, {})
             for child in newppnode2:
                 newppnode2.remove(child)
-            newppnode2.append(Rpronounobj1node)
+            newppnode2.append(newphrase)
             newppnode2.append(newvz2)
 
             # pp with R-pronoun object which has been replaced by a full NO with a dummymod
@@ -1047,11 +1055,15 @@ def newgenvariants(stree: SynTree) -> List[SynTree]:
                 pppronadvvcnode.append(pronadvnode1)
                 pppronadvvcnode.append(newvcnode)
 
+            # pp's with a pronominal adverb. e.g. daarnaar
             pprel = gav(ppnode, 'rel')
             pronadvnode = getpronadv(vzlemma, pprel)
+            pronadvppnode = expandnonheadwordnode(pronadvnode, {'cat': 'pp', 'rel': pprel})
+            pronadvnode.attrib['rel'] = 'hd'
+            pronadvppnode.append(pronadvnode)
 
             alternativesnode = mkalternativesnode([[ppnode], [newppnode2], [newppnode3], [
-                                                  pppobj1vcnode], [pppronadvvcnode], [pronadvnode]])
+                                                  pppobj1vcnode], [pppronadvvcnode], [pronadvppnode]])
             parent.append(alternativesnode)
 
         vblgennpnodeids = newstree.xpath(
@@ -1404,8 +1416,9 @@ def relpronsubst(stree: SynTree) -> SynTree:
 def expandfull(stree: SynTree) -> SynTree:
     # possibly add getlcat
     stree1 = relpronsubst(stree)
-    stree2 = indextransform(stree1)
-    return stree2
+    stree2 = expandnonheadwords(stree1)
+    stree3 = indextransform(stree2)
+    return stree3
 
 
 def gettopnode(stree):

mwe_query/lcat.py

@@ -8,6 +8,8 @@
 import copy
 import lxml.etree as ET
 
+dummy = 'dummy'
+
 
 def expandnonheadwords(stree: SynTree) -> SynTree:
     # it is presupposed that the input stree is not None
@@ -47,7 +49,8 @@ def getlcatatt(node: SynTree) -> str:
 
 def mkphrase(child: SynTree) -> SynTree:
     newnode = ET.Element('node')
-    newnode.attrib['id'] = child.attrib['id'] + 'a'
+    if 'íd' in child.attrib:
+        newnode.attrib['id'] = child.attrib['id'] + 'a'
     lcat = getlcatatt(child)
     if lcat in validcats:
         newnode.attrib['cat'] = lcat
@@ -176,6 +179,8 @@ def getlcat(node: SynTree, prel=None) -> str:  # noqa: C901
         result = 'np'
     elif pt == 'spec':
         result = None
+    elif pt == dummy:
+        result = None
     else:
         print('Unknown att value (pt) encountered in:')
         ET.dump(node)

mwe_query/trymwes.py

@@ -0,0 +1,52 @@
+from sastadev.alpinoparsing import parse
+from lcat import expandnonheadwords
+from sastadev.treebankfunctions import indextransform
+from lxml import etree
+from canonicalform import generatequeries, expandfull
+
+debug = False
+
+geenhaankraaien = ('0geen *haan zal naar iets kraaien',
+                   ['Daar kraait geen haan naar', 'Hier heeft geen haan naar gekraaid',
+                    'geen haan kraaide daarnaar', 'geen haan kraaide ernaar dat hij niet kwam',
+                    'geen haan kraaide er naar dat hij niet kwam', 
+                    'er is geen haan die daar naar kraait', ]
+                   )
+
+def select(mweutts, utt=None):
+    if utt is None:
+        result = mweutts
+    else:
+        result = (mweutts[0], [mweutts[1][utt]])
+    return result
+
+def getparses(utterances):
+    uttparses = []
+    for utterance in utterances:
+        uttparse = parse(utterance)
+        uttparses.append(uttparse)
+    return uttparses
+
+def trysomemwes():
+    mwe, utterances = select(geenhaankraaien)
+    mwequeries = generatequeries(mwe)
+    labeledmwequeries = (('MWEQ', mwequeries[0]), ('NMQ', mwequeries[1]), ('MLQ', mwequeries[2]))
+    uttparses = getparses(utterances)
+    for utterance, uttparse in zip(utterances, uttparses):
+        print(f'{utterance}:')
+        expandeduttparse = expandfull(uttparse)
+        if debug:
+            etree.dump(expandeduttparse)
+        for label, mwequery in labeledmwequeries:
+            results = expandeduttparse.xpath(mwequery)
+            if debug:
+                print('Found hits:')
+                for result in results:
+                    etree.dump(result)
+            print(f'{label}: {len(results)}')
+
+
+
+
+if __name__ == '__main__':
+    trysomemwes()