def execComposePart(self, pp, cp):
parClustIdx = pp._clustIdx
chdClustIdx = cp._clustIdx
pcl = Clust.getClust(parClustIdx)
ccl = Clust.getClust(chdClustIdx)
dep = pp.getArguments()[cp._parArgIdx]._path.getDep()
pp._relTreeRoot.addChild(dep, cp._relTreeRoot)
nrti = RelType.getRelType(pp._relTreeRoot)
ncl = Clust.getClust(next(iter(Clust.getClustsWithRelType(nrti))))
nci = ncl.getId()
pp.removeArgument(cp._parArgIdx, clust_only=True)
for argIdx, arg in pp.getArguments().items():
pp.unsetArgClust(argIdx)
arg._argPart.unsetParent()
pp.changeClust(nci, nrti, clust_only=True)
for argIdx, arg in pp.getArguments().items():
ati = arg._path.getArgType()
aci = -1
if ati not in ncl._argTypeIdx_argClustIdxs:
aci = ncl.createArgClust(ati)
elif len(ncl._argTypeIdx_argClustIdxs[ati]) == 0:
aci = ncl.createArgClust(ati)
else:
aci = next(iter(ncl._argTypeIdx_argClustIdxs[ati]))
arg._argPart.setParent(pp, argIdx)
pp.setArgClustOnly(argIdx, aci)
pp.setRelTypeIdx(nrti)
for argIdx, arg in cp.getArguments():
ati = arg._path.getArgType()
aci = -1
if ati not in ncl._argTypeIdx_argClustIdxs:
aci = ncl.createArgClust(ati)
elif len(ncl._argTypeIdx_argClustIdxs[ati]) == 0:
aci = ncl.createArgClust(ati)
else:
aci = next(iter(ncl._argTypeIdx_argClustIdxs[ati]))
cp.unsetArgClustOnly(argIdx)
pp.setArgClust(pp.addArgument(arg), aci)
arg._argPart.setParent(pp, pp.addArgument(arg))
cp.destroy()
return None
def part_from_node(ai, sj, sent, k, tok):
if not Parse.isIgnore(sent, k):
tn = TreeNode(genTreeNodeID(ai, sj, k), tok)
part = Part(tn)
relTypeIdx = part.getRelTypeIdx()
clustIdxs = Clust.getClustsWithRelType(relTypeIdx)
if clustIdxs is not None:
clustIdx = next(iter(clustIdxs))
else:
clustIdx = Clust.createClust(relTypeIdx)
part.setClust(clustIdx)
return None
def scoreOpComposePart(self, pp, cp):
score = 0
rcl, acl = [Clust.getClust(x._clustIdx) for x in (pp, cp)]
ptn, ctn = [x._relTreeRoot for x in (pp, cp)]
ptn.addChild(dep, ctn)
nrti = RelType.getRelType(ptn)
if Clust.getClustsWithRelType(nrti) is None:
return score
pai = cp._parArgIdx
pcarg = pp.getArguments()[pai]
dep = pcarg._path.getDep()
orti = pp._relTypeIdx
ncl = Clust.getClust(Clust.getClustsWithRelType(nrti).next())
nci = ncl._clustIdx
if pp.getParPart() is not None:
ppp = pp.getParPart()
ppcl = Clust.getClust(ppp.getClustIdx())
ac = ppcl._argClusts[ppp.getArgClust(pp.getParArgIdx())]
oc = ac._chdClustIdx_cnt[rcl._clustIdx]
nc = ac._chdClustIdx_cnt[nci]
else:
oc = Clust.clustIdx_rootCnt[rcl]
nc = Clust.clustIdx_rootCnt[ncl]
score += log(nc) - log(oc)
for aci, ais in pp._argClustIdx_argIdxs.items():
ac = rcl._argClusts[aci]
score -= (log(ac._argNum_cnt[len(ais)]) - log(ac._ttlArgCnt))
for ai in ais:
arg = pp.getArgument(ai)
score -= (log(ac._chdClustIdx_cnt[arg._argPart._clustIdx]) \
- log(ac._ttlArgCnt))
score -= (log(ac._argTypeIdx_cnt[arg._path.getArgType()]) \
- log(ac._ttlArgCnt))
ai_newaci = dict()
for ai, arg in pp._args.items():
if ai == pai:
pass
else:
ati = arg._path.getArgType()
aci = ncl._argTypeIdx_argClustIdxs[ati].next()
ai_newaci[ai] = aci
newArgClustIdx_ais = dict()
for ai, aci in ai_newaci.items():
if aci not in newArgClustIdx_ais:
newArgClustIdx_ais[aci] = set()
newArgClustIdx_ais[aci].add(ais)
for aci, ais in newArgClustIdx_ais.items():
ac = ncl._argClusts[aci]
score += (log(ac._argNum_cnt[len(ais)]) - log(ac._ttlArgCnt))
for ai in ais:
arg = pp.getArgument(ai)
score -= (log(ac._chdClustIdx_cnt[arg._argPart._clustIdx]) \
- log(ac._ttlArgCnt))
score -= (log(ac._argTypeIdx_cnt[arg._path.getArgType()]) \
- log(ac._ttlArgCnt))
return score
def execCompose(self, op):
parClustIdx = op._parClustIdx
chdClustIdx = op._chdClustIdx
new_clust_id = -1
#
# If either cluster are None, return -1
#
if Clust.getClust(parClustIdx) is None or Clust.getClust(
chdClustIdx) is None:
return -1
new_clust = None
parent_child_pair = (parClustIdx, chdClustIdx)
part_ids = set()
part_ids.update(
Part.pairClustIdxs_pairPartRootNodeIds[parent_child_pair])
deleted_parts = []
for parent_id, child_id in part_ids:
if parent_id in deleted_parts or child_id in deleted_parts:
continue
parent_part = Part.getPartByRootNodeId(parent_id)
child_part = Part.getPartByRootNodeId(child_id)
dep = parent_part.getArguments()[
child_part._parArgIdx]._path.getDep()
parent_part._relTreeRoot.addChild(dep, child_part._relTreeRoot)
nrti = RelType.getRelType(parent_part._relTreeRoot)
if new_clust is None:
# on first loop
rel_clusts = Clust.getClustsWithRelType(nrti)
if rel_clusts is None:
new_clust = Clust.getClust(Clust.createClust(nrti))
elif len(rel_clusts) > 1:
raise Exception
else:
new_clust = Clust.getClust(next(iter(rel_clusts)))
new_clust_id = new_clust.getId()
parent_part.removeArgument(child_part._parArgIdx)
if parent_part.getClustIdx() != new_clust_id:
for argIdx in parent_part.getArguments():
parent_part.unsetArgClust(argIdx)
arg = parent_part.getArgument(argIdx)
arg._argPart.unsetParent()
parent_part.changeClust(new_clust_id, nrti)
for argIdx, arg in parent_part.getArguments().items():
arg_type = arg._path.getArgType()
arg_clust_id = -1
if arg_type not in new_clust._argTypeIdx_argClustIdxs:
arg_clust_id = new_clust.createArgClust(arg_type)
elif len(
new_clust._argTypeIdx_argClustIdxs[arg_type]) == 0:
arg_clust_id = new_clust.createArgClust(arg_type)
else:
arg_clust_id = next(
iter(new_clust._argTypeIdx_argClustIdxs[arg_type]))
arg._argPart.setParent(parent_part, argIdx)
parent_part.setArgClust(argIdx, arg_clust_id)
parent_part.setRelTypeIdx(nrti)
else:
parent_part.unsetRelTypeIdx()
parent_part.setRelTypeIdx(nrti)
#
# Connect the child part's arguments directly to the parent part now
#
for argIdx, arg in child_part.getArguments().items():
child_part.unsetArgClust(argIdx)
arg_type = arg._path.getArgType()
arg_clust_id = -1
if arg_type not in new_clust._argTypeIdx_argClustIdxs:
arg_clust_id = new_clust.createArgClust(arg_type)
elif len(new_clust._argTypeIdx_argClustIdxs[arg_type]) == 0:
arg_clust_id = new_clust.createArgClust(arg_type)
else:
arg_clust_id = next(
iter(new_clust._argTypeIdx_argClustIdxs[arg_type]))
newArgIdx = parent_part.addArgument(arg)
arg._argPart.setParent(parent_part, newArgIdx)
parent_part.setArgClust(newArgIdx, arg_clust_id)
#
# Remove the old child part
#
deleted_parts.append(child_part.getRelTreeRoot().getId())
child_part.destroy()
# Part.clustIdx_pairClustIdxs[parClustIdx].remove(pci)
# Part.clustIdx_pairClustIdxs[chdClustIdx].remove(pci)
del Part.pairClustIdxs_pairPartRootNodeIds[parent_child_pair]
return new_clust_id