def wrapMatrixFile(matrixFile, protocol, dfsDir=None, maxCores=1):
"""
Factory method for wrapping a matrix stored in a file in coordinate format.
"""
if protocol == MatrixWrapper.RAW:
raws = [[] for i in range(0, maxCores)]
i = 0
infile = file(matrixFile, "r")
for line in infile:
for elem in line.strip().split(";"):
raws[i%maxCores].append(elem)
i += 1
infile.close()
return MatrixWrapper(["raw://"+";".join(elems) for elems in raws if len(elems) > 0])
elif protocol == MatrixWrapper.DFS:
if not os.path.exists(dfsDir):
os.mkdir(dfsDir)
# delete any previous gluster files in the directory
[os.remove(dfsDir+"/"+f) for f in os.listdir(dfsDir)]
splitFileToFiles(matrixFile, dfsDir, "", maxCores)
return MatrixWrapper(files(dfsDir))
else:
raise Exception('only dfs or raw protocol are supported')
def wrapMatrix(A, protocol, dfsDir=None, maxCores=1):
"""
Factory method for wrapping a numpy/scipy matrix object.
"""
if type(A) == coo_matrix:
Acoo = A.tocoo()
if protocol == MatrixWrapper.RAW:
raws = [[] for i in range(0, maxCores)]
for i in range(0, len(Acoo.row)):
raws[i%maxCores].append("%d,%d,%.14f" % (Acoo.row[i], Acoo.col[i], Acoo.data[i]))
return MatrixWrapper(["raw://"+";".join(elems) for elems in raws if len(elems) > 0])
elif protocol == MatrixWrapper.DFS:
import os.path
if not os.path.exists(dfsDir):
os.mkdir(dfsDir)
# delete any previous gluster files in the directory
[os.remove(dfsDir+"/"+f) for f in os.listdir(dfsDir)]
# save in coordinate format and split matrix to multiple files
splitMatrixToFiles(A, dfsDir, "", maxCores)
return MatrixWrapper(files(dfsDir))
else:
raise Exception('only dfs or raw protocol are supported')
else:
raise Exception('only coo_matrix is supported for now')
