def run(self):
with self.input()[0] as i:
graphIndex = i.query()
with self.input()[1] as i:
nodeIndex = i.query()
row = []
column = []
data = []
floatType = False
row_shape = 0
col_shape = 0
for ix in range(2, len(self.input())):
with self.input()[ix] as i:
D = i.query()
for ID, entry in D.items():
if ID not in graphIndex:
continue
gI = graphIndex[ID]
row_shape = max(row_shape, gI)
for n, c in entry['kernel_bag'].items():
if n not in nodeIndex:
continue
nI = nodeIndex[n]
col_shape = max(col_shape, nI)
floatType = floatType or isinstance(c, float)
row.append(gI)
column.append(nI)
data.append(c)
dtype = np.float64 if floatType else np.uint64
phi = coo_matrix((data, (row, column)),
shape=(row_shape + 1, col_shape + 1),
dtype=dtype).tocsr()
if self.tfidf:
phi = TfidfTransformer(sublinear_tf=True).fit_transform(phi)
NZ = phi.nonzero()
data = phi[NZ].A
shape = phi.get_shape()
out = {
'graphIndex': graphIndex,
'nodeIndex': nodeIndex,
'rows': NZ[0].tolist(),
'columns': NZ[1].tolist(),
'data': data.tolist()[0],
'row_shape': shape[0],
'column_shape': shape[1]
}
with self.output() as o:
o.emit(out)
