def actual_process(cls,mygroup):
tagsOfForms=[ u"inf" , u"ger" , u"pp.m.sg" , u"pri.p1.sg" , u"pri.p2.sg", u"pri.p3.sg" , u"pri.p1.pl" , u"pri.p2.pl", u"pri.p3.pl", u"prs.p1.sg" , u"prs.p2.sg", u"prs.p3.sg" , u"prs.p1.pl" , u"prs.p2.pl", u"prs.p3.pl", u"pii.p1.sg" , u"pii.p2.sg", u"pii.p3.sg" , u"pii.p1.pl" , u"pii.p2.pl", u"pii.p3.pl", u"pis.p1.sg" , u"pis.p2.sg", u"pis.p3.sg" , u"pis.p1.pl" , u"pis.p2.pl", u"pis.p3.pl", u"pis.p1.sg" , u"pis.p2.sg", u"pis.p3.sg" , u"pis.p1.pl" , u"pis.p2.pl", u"pis.p3.pl", u"ifi.p1.sg" , u"ifi.p2.sg", u"ifi.p3.sg" , u"ifi.p1.pl" , u"ifi.p2.pl", u"ifi.p3.pl", u"fti.p1.sg" , u"fti.p2.sg", u"fti.p3.sg" , u"fti.p1.pl" , u"fti.p2.pl", u"fti.p3.pl", u"pis.p1.sg" , u"pis.p2.sg", u"pis.p3.sg" , u"pis.p1.pl" , u"pis.p2.pl", u"pis.p3.pl", u"cni.p1.sg" , u"cni.p2.sg", u"cni.p3.sg" , u"cni.p1.pl" , u"cni.p2.pl", u"cni.p3.pl", u"imp.p1.pl" , u"imp.p2.sg", u"imp.p3.sg" , u"imp.p1.pl" , u"imp.p2.pl", u"imp.p3.pl", u"pp.m.sg", u"pp.f.sg", u"pp.m.pl", u"pp.f.pl"]
foundEntries=[ [] for mylist in tagsOfForms ]
#search for tags in entries
for i in range(len(tagsOfForms)):
pattern=tagsOfForms[i]
for entry in mygroup:
tagsOfLexicalForm=u".".join(entry.lexicalForm.get_tags())
if tagsOfLexicalForm.count(pattern) > 0:
foundEntries[i].append(entry)
if all( len(pfoundEntries) > 0 for pfoundEntries in foundEntries ):
if not mygroup[0].lexicalForm.get_lemma() in SpaVblexProcessor.engVblexDict:
print >> sys.stderr, "Discarded group: not found in bilingual: "+str(mygroup)
else:
for token in SpaVblexProcessor.engVblexDict[mygroup[0].lexicalForm.get_lemma()]:
valencies=ValenciesProcessor.valencyDict.get(createGFTokenNoValency(token))
if not valencies or len(valencies) == 0:
SpaVblexProcessor.print_GF(token,[pfoundEntries[0].surfaceForm for pfoundEntries in foundEntries])
else:
for valency in valencies:
SpaVblexProcessor.print_GF(token,[pfoundEntries[0].surfaceForm for pfoundEntries in foundEntries],valency)
else:
print >> sys.stderr, "Discarded Group: "+str(mygroup)
def actual_process(cls,mygroup):
tagsOfForms=[[u"inf"] , [u"pri",u"p3",u"sg"] , [u"past"] , [u"pp"] , [u"ger"] ]
foundEntries=[ [] for mylist in tagsOfForms ]
groupNoSep=[ entry for entry in mygroup if len ( set([u"hasprpers",u"hasthat",u"hasthis"]) & set(entry.lexicalForm.get_tags()) ) == 0 ]
#search for tags in entries
for i in range(len(tagsOfForms)):
pattern=u".".join(tagsOfForms[i])
for entry in groupNoSep:
tagsOfLexicalForm=u".".join(entry.lexicalForm.get_tags())
if tagsOfLexicalForm.count(pattern) > 0:
foundEntries[i].append(entry)
if all( len(pfoundEntries) > 0 for pfoundEntries in foundEntries ):
valencies=ValenciesProcessor.valencyDict.get(createGFTokenNoValency(groupNoSep[0].lexicalForm.get_lemma()))
if not valencies or len(valencies) == 0:
EngVblexProcessor.print_GF(groupNoSep[0].lexicalForm.get_lemma(),[pfoundEntries[0].surfaceForm for pfoundEntries in foundEntries])
else:
for valency in valencies:
EngVblexProcessor.print_GF(groupNoSep[0].lexicalForm.get_lemma(),[pfoundEntries[0].surfaceForm for pfoundEntries in foundEntries],valency)
else:
print >> sys.stderr, "Discarded Group: "+str(mygroup)