def get_edges(self):
if not self.has_downloaded():
self.download()
graph_file = config.get_download_file_path(self.name,
self.name + '.graph')
self.logger.info("reading {}".format(graph_file))
cmd = "{} -server -cp {} it.unimi.dsi.webgraph.ArcListASCIIGraph {} {}".format(
utils.get_java_command(), config.WEBGRAPH_JAR_PATH,
self.name + '-hc-t', self.name)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, self.download_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
csvfile = os.path.join(self.download_dir, self.name)
edges = pd.read_csv(csvfile, header=None, sep='\t')
edges.columns = ['src', 'dest']
utils.remove_if_file_exit(csvfile)
return edges
def run(self, data, seed=None):
if seed is None:
seed = np.random.randint(999999)
params = {'seed': seed}
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_pajek):
data.to_pajek()
with utils.TempDir() as tmp_dir:
def link_file(path, destname=None):
if destname is None:
destname = path.split("/")[-1]
destpath = os.path.join(tmp_dir, destname)
utils.remove_if_file_exit(destpath)
os.symlink(path, destpath)
return destpath
pajek = link_file(data.file_pajek, 'pajek.txt')
cmd = "{} {} {} {}".format(
config.get_OSLOM_prog('infomap', data.is_directed()), seed,
pajek, 1)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfile = glob.glob(tmp_dir + "/pajek.tree")[0]
import pandas as pd
output = pd.read_csv(outputfile, sep=" ", skiprows=1, header=None)
output['cluster'] = output.loc[:, 0].map(
lambda u: int(u.split(':')[0]))
output['node'] = output.loc[:, 2].astype(np.int)
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, f=False, m=None,seed=None):
if False and (data.is_directed()):
raise Exception("only undirected is supported")
if seed is not None:self.logger.info("seed ignored")
params = {}
params['f'] = f
params['m'] = m
if not utils.file_exists(data.file_hig):
data.to_higformat()
cmd = ["./hirecs"]
cmd.append("-oje")
if f: cmd.append('-f')
if m is not None: cmd.append("-m{}".format(m))
cmd.append(data.file_hig)
cmd.append("> output")
cmd = " ".join(cmd)
with utils.TempDir() as tmp_dir:
with open(os.path.join(tmp_dir, "tmpcmd"), 'wt') as f: f.write(cmd)
self.logger.info("Running " + cmd)
cmd = "bash tmpcmd"
utils.link_file(os.path.join(config.HIRECS_PATH, 'hirecs'), tmp_dir)
utils.link_file(os.path.join(config.HIRECS_PATH, 'libhirecs.so'), tmp_dir)
utils.link_file(data.file_hig, tmp_dir)
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception("Run command with error status code {}".format(status))
with open (os.path.join(tmp_dir, "output"), "r") as f:
output = json.load(f)
mod = output['mod']
communities = output['communities']
clusters = {}
for c in communities:
clusters[int(c)] = [int(u) for u in communities[c].keys()]
self.logger.info("Made %d clusters in %f seconds with modularity %f" % (len(clusters), timecost, mod))
result = {}
result['runname'] = self.name
result['params'] = params
result['overlap'] = True
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self,
data,
seed=None,
lamb=1,
trials=5,
temp_step=0.999,
initial_temp=1e-6):
params = locals()
del params['data']
del params['self']
if seed is None:
seed = np.random.randint(999999)
params['seed'] = seed
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = os.path.join(tmp_dir, 'edges.txt')
utils.remove_if_file_exit(pajek)
os.symlink(data.file_edges, pajek)
cmd = "{} {} {} {} {} {} {}".format(
config.get_OSLOM_prog('modopt', data.is_directed()), pajek,
seed, lamb, trials, temp_step, initial_temp)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfile = pajek + "part"
with open(outputfile) as f:
lines = [u.strip() for u in f]
lines = [[int(v) for v in u.split("\t")] for u in lines]
clusters = dict(enumerate(lines))
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def to_higformat(self, filepath=None):
if (filepath == None):
filepath = self.file_hig
if utils.file_exists(filepath):
return filepath
if not utils.file_exists(self.file_pajek): self.to_pajek()
cmd = "python {} {}".format(config.HIG_CONVERT_PROG, self.file_pajek)
self.logger.info("running " + cmd)
status = utils.shell_run_and_wait(cmd)
if (status != 0):
raise Exception("run command failed: " + str(status))
return filepath
def run(self, data, seed=None):
self.logger.warning(
"dct assumes node starts with zero and is continuous")
if seed is None:
seed = np.random.randint(999999)
params = {'seed': seed, 'prog': self.progname}
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
#tmp_dir = "/tmp/abc"
pajek = utils.link_file(data.file_edges,
dest_dir=tmp_dir,
destname='edges.txt')
cmd = "{} {}".format(
config.get_dct_prog(self.progname, data.is_directed()), pajek)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(
cmd, tmp_dir, env={'SEED': str(seed)}))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfiles = glob.glob(tmp_dir + "/output*.cluster")
import pandas as pd
df_from_each_file = (pd.read_csv(f, sep=" ", header=None)
for f in outputfiles)
output = pd.concat(df_from_each_file, ignore_index=True)
output.columns = ['node', 'cluster']
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, seed=None):
self.logger.warning(
"dct::seq_louvain assumes node starts with zero and is continuous")
if seed is None:
seed = np.random.randint(999999)
params = {'seed': seed}
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = utils.link_file(data.file_edges,
dest_dir=tmp_dir,
destname='edges.txt')
cmd = "{} -f -s {} -o output {}".format(
config.get_dct_prog('infomap', data.is_directed()), seed,
pajek)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfile = tmp_dir + "/output"
import pandas as pd
output = pd.read_csv(outputfile, sep=" ", header=None)
output.columns = ['cluster']
output['node'] = range(len(output))
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, algorithm=4, minpts=4, epsilon=0.5, alpha=32768 , beta=32768 , thread=-1, seed=None):
params = locals()
del(params['self']);del(params['data'])
if seed is not None:self.logger.info("seed ignored")
thread = utils.get_num_thread(thread)
params['thread'] = thread
if False and (data.is_directed() or data.is_weighted()):
raise Exception("only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_anyscan):
data.to_anyscan()
cmd = "{} -c {} -i {} -m {} -e {} -o {} -a {} -b {} -t {}".format(
config.ANYSCAN_PROG, algorithm, data.file_anyscan, minpts, epsilon, 'output', alpha, beta, thread)
self.logger.info("Running " + cmd)
with utils.TempDir() as tmp_dir:
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 1: # anyscan always return 1
raise Exception("Run command with error status code {}".format(status))
with open (os.path.join(tmp_dir, "output"), "r") as output:
lines = [u.strip() for u in output.readlines()]
n_nodes = int(lines[0])
clusters_list = lines[1].split(" ")
clusters_list = [int(u) for u in clusters_list ]
if (n_nodes != len(clusters_list)):
raise Exception("#node is not equals #cluster")
from collections import defaultdict
clusters = defaultdict(list)
for n, c in enumerate(clusters_list):
clusters[c].append(n)
self.logger.info("Made %d clusters in %f seconds" % (len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, mu=3, epsilon=0.5, prog='pScan', seed=None):
assert prog in ['pScan', 'ppScan', 'ppScanSSE']
if seed is not None:self.logger.info("seed ignored")
params = {'mu':mu, 'epsilon':epsilon, 'prog':prog}
if False and (data.is_directed() or data.is_weighted()):
raise Exception("only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_scanbin):
data.to_scanbin()
with utils.TempDir() as tmp_dir:
scanbin = data.file_scanbin
os.symlink(os.path.join(scanbin, 'b_degree.bin'), os.path.join(tmp_dir, 'b_degree.bin'))
os.symlink(os.path.join(scanbin, 'b_adj.bin'), os.path.join(tmp_dir, 'b_adj.bin'))
if prog == 'pScan':
EXE = config.PSCAN_PROG
elif prog == 'ppScan':
EXE = config.PPSCAN_PROG
elif prog == 'ppScanSSE':
EXE = config.PPSCANSSE_PROG
cmd = "{} {} {} {} {}".format(EXE, tmp_dir, epsilon, mu, 'output')
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception("Run command with error status code {}".format(status))
outputfile = glob.glob(tmp_dir + "/result-*-*.txt")[0]
import pandas as pd
output = pd.read_csv(outputfile, sep=" ")
clusters = output[output['c/n'] == 'c'][['vertex_id', 'cluster_id']]
others = output[output['c/n'] != 'c'][['vertex_id', 'cluster_id']]
clusters = clusters.groupby('cluster_id').apply(lambda u: list(u['vertex_id'])).to_dict()
others = others.values.tolist()
self.logger.info("Made %d clusters in %f seconds" % (len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
result['noise_or_hub'] = others
save_result(result)
self.result = result
return self
def to_scanbin(self, filepath=None):
if (filepath == None):
filepath = self.file_scanbin
if utils.file_exists(os.path.join(
filepath, "b_degree.bin")) and utils.file_exists(
os.path.join(filepath, "b_adj.bin")):
return filepath
utils.create_dir_if_not_exists(filepath)
if not utils.file_exists(self.file_edges): self.to_edgelist()
cmd = "{} {} {} {}".format(config.SCAN_CONVERT_PROG, self.file_edges,
"b_degree.bin", "b_adj.bin")
self.logger.info("running " + cmd)
status = utils.shell_run_and_wait(cmd, filepath)
if (status != 0):
raise Exception("run command failed: " + str(status))
return filepath
def to_mcl_mci(self, filepath=None):
if (filepath == None):
filepath = self.file_mcl_mci
if utils.file_exists(filepath):
return filepath
if not utils.file_exists(self.file_edges): self.to_edgelist()
if not self.is_directed():
cmd = "{} --stream-mirror -123 {} -o {} --write-binary".format(
config.MCL_CONVERT_PROG, self.file_edges, filepath)
else:
cmd = "{} -123 {} -o {} --write-binary".format(
config.MCL_CONVERT_PROG, self.file_edges, filepath)
self.logger.info("running " + cmd)
status = utils.shell_run_and_wait(cmd)
if (status != 0):
raise Exception("run command failed: " + str(status))
return filepath
def run(self, data, **kwargs):
if False and (data.is_directed()):
raise Exception("only undirected is supported")
params = dict(kwargs)
params = {k:v for k, v in params.items() if v is not None }
if not data.is_directed():
params['Sym'] = 1
params['d'] = "output"
if "r" not in params: params['r'] = 0.1
if not utils.file_exists(data.file_edges):
data.to_edgelist()
txt_params = " ".join(["-{} {}".format(k, v) for k, v in params.items()])
cmd = "{} -jar {} -i {} {} ".format(utils.get_java_command(), config.GANXISW_PROG, "edges.txt", txt_params)
with utils.TempDir() as tmp_dir:
utils.remove_if_file_exit(os.path.join(tmp_dir, "output"), True)
utils.create_dir_if_not_exists(os.path.join(tmp_dir, "output"))
self.logger.info("Running " + cmd)
utils.link_file(data.file_edges, tmp_dir, "edges.txt")
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception("Run command with error status code {}".format(status))
outputfile = glob.glob(os.path.join(tmp_dir, "output/SLPAw*.icpm"))[0]
clusters = []
with open (os.path.join(tmp_dir, outputfile), "r") as f:
for line in f:
clusters.append([int(u) for u in line.strip().split(" ")])
clusters = dict(enumerate(clusters))
self.logger.info("Made %d clusters in %f seconds" % (len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def to_topgc_fromat(self, filepath=None):
if (filepath == None):
filepath = self.file_topgc
if utils.file_exists(filepath):
return filepath
if not utils.file_exists(self.file_edges): self.to_edgelist()
with TempDir() as tmp_dir:
cmd = "cat {}|sort -k1,1 -k2,2 -n | {} > {} || rm {}".format(
self.file_edges, config.get_cdc_prog('mkidx'), filepath,
filepath)
self.logger.info("Running " + cmd)
with open(os.path.join(tmp_dir, "tmpcmd"), 'wt') as f:
f.write(cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait("bash -x tmpcmd", tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
assert utils.file_exists(filepath)
return filepath
def run(self, data, cutoff_r=0.01, inflation_in=2, NBDisimilarity_q=0.3, seed=None):
if False and (data.is_directed()):
raise Exception("only undirected is supported")
if seed is not None:self.logger.info("seed ignored")
params = locals();del params['self'];del params['data']
if not utils.file_exists(data.file_edges):
data.to_edgelist()
cmd = "{} {} {} {} {}".format(config.LABLE_RANK_PROG, "edges.txt", cutoff_r, inflation_in, NBDisimilarity_q)
with utils.TempDir() as tmp_dir:
utils.remove_if_file_exit(os.path.join(tmp_dir, "output"), True)
utils.create_dir_if_not_exists(os.path.join(tmp_dir, "output"))
self.logger.info("Running " + cmd)
utils.link_file(data.file_edges, tmp_dir, "edges.txt")
timecost, status = utils.timeit(lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception("Run command with error status code {}".format(status))
outputfile = glob.glob(os.path.join(tmp_dir, "output/LabelRank*.icpm"))[0]
clusters = []
with open (os.path.join(tmp_dir, outputfile), "r") as f:
for line in f:
clusters.append([int(u) for u in line.strip().split(" ")])
clusters = dict(enumerate(clusters))
self.logger.info("Made %d clusters in %f seconds" % (len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def md5check(self):
cmd = 'md5sum --ignore-missing -c {}.md5sums'.format(self.name)
status = utils.shell_run_and_wait(
cmd, config.get_download_file_path(self.name))
return status == 0
def run(self, data, **kwargs):
if False and (data.is_directed() or data.is_weighted()):
raise Exception(
"only undirected and unweighted graph is supported")
params = dict(kwargs)
params = {u: v for u, v in params.items() if v is not None}
if "loss" not in params: params['loss'] = 'modularity'
argparams = dict(params)
if argparams['loss'] in ['pmod', 'mom']:
if argparams['loss'] not in argparams:
raise Exception(
"You have to specify pmod=<val> or mom=<val> for the loss function"
)
loss_args = "--loss {{{},{}}}".format(argparams['loss'],
argparams[argparams['loss']])
del argparams[argparams['loss']]
del argparams['loss']
else:
loss_args = "--loss {}".format(argparams["loss"])
del argparams['loss']
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = utils.link_file(data.file_edges,
dest_dir=tmp_dir,
destname='edges.txt')
cmdargs = ["--{} {}".format(u, v) for u, v in argparams.items()]
cmdargs.append(loss_args)
cmdargs.append("-o lsooutput")
cmdargs = " ".join(cmdargs)
cmd = "{} {} {}".format(config.LSO_CLUSTER_PROG, pajek, cmdargs)
with open(os.path.join(tmp_dir, "tmpcmd"), 'wt') as f:
f.write(cmd)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait("bash -x ./tmpcmd", tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfiles = tmp_dir + "/lsooutput"
import pandas as pd
output = pd.read_csv(outputfiles, sep="\t", header=None)
output.columns = ['node', 'cluster']
clusters = output
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self,
data,
vmax_start=None,
vmax_end=None,
c=None,
niter=None,
seed=None):
if False and (data.is_directed() or data.is_weighted()):
raise Exception(
"only undirected and unweighted graph is supported")
params = locals()
del (params['self'])
del (params['data'])
params['f'] = data.file_edges
params['o'] = 'output'
params = {
u.replace("_", "-"): v
for u, v in params.items() if v is not None
}
if not utils.file_exists(data.file_edges):
data.to_edgelist()
cmd = "{} {} {}".format(
config.STREAMCOM_PROG, " ".join([
'{}{} {}'.format('-' if len(u) == 1 else '--', u, v)
for u, v in params.items()
]), data.file_edges)
self.logger.info("Running " + cmd)
with utils.TempDir() as tmp_dir:
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
output = glob.glob(os.path.join(tmp_dir, "output*"))[0]
with open(os.path.join(tmp_dir, output), "r") as output:
lines = [u.strip() for u in output.readlines()]
from collections import defaultdict
clusters = defaultdict(list)
for c, line in enumerate(lines):
if line.startswith('#'): continue
for n in line.split(" "):
clusters[c].append(int(n))
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self, data, **kwargs):
if False and (data.is_directed()):
raise Exception("only undirected is supported")
params = dict(kwargs)
if "seed" in params:
if params['seed'] is not None:
self.logger.info("seed ignored")
del params['seed']
#params['abc'] = ''
params['o'] = 'output'
params = {u: v for u, v in params.items() if v is not None}
if not utils.file_exists(data.file_mcl_mci):
data.to_mcl_mci()
if not utils.file_exists(data.file_mcl_mci):
raise Exception("failed to crate mcl mci format file")
with utils.TempDir() as tmp_dir:
cmd1 = "{} {} {}".format(
config.MCL_PROG, data.file_mcl_mci, " ".join([
'{}{} {}'.format('-' if len(u) == 1 else '--', u,
v).strip() for u, v in params.items()
]))
cmd2 = "{} -imx {} -o cluster.output".format(
config.MCLDUMP_PROG, 'output')
cmdfile = os.path.join(tmp_dir, "tmpcmd.sh")
with open(cmdfile, 'wt') as f:
f.write(cmd1 + "\n")
f.write(cmd2 + "\n")
self.logger.info("Running " + cmd1)
self.logger.info("Running " + cmd2)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait("bash " + cmdfile, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
with open(os.path.join(tmp_dir, "cluster.output"), "r") as output:
lines = [u.strip() for u in output.readlines()]
from collections import defaultdict
clusters = defaultdict(list)
for line in lines:
cluster, node = line.split("\t")[:2]
clusters[int(cluster)].append(int(node))
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self,
data,
execution='async',
ncpus=None,
scheduler=None,
engine_opts=None,
graph_opts=None,
scheduler_opts=None,
seed=None):
if seed is not None: self.logger.info("seed ignored")
params = locals()
del params['self']
del params['data']
del params['seed']
params = {u: v for u, v in params.items() if v is not None}
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
ncpus = utils.get_num_thread(ncpus)
params['ncpus'] = ncpus
params['weighted'] = data.is_weighted() * 1
params['directed'] = data.is_directed() * 1
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = os.path.join(tmp_dir, 'edges.txt')
utils.remove_if_file_exit(pajek)
os.symlink(data.file_edges, pajek)
args = " ".join(
["--{} {}".format(u, v) for u, v in params.items()])
cmd = "{} --graph {} --saveprefix=output.cluster {}".format(
config.get_powergraph_prog('label_propagation',
data.is_directed()), pajek, args)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfiles = glob.glob(tmp_dir + "/output.cluster.*")
import pandas as pd
df_from_each_file = (pd.read_csv(f, sep="\t", header=None)
for f in outputfiles)
output = pd.concat(df_from_each_file, ignore_index=True)
output.columns = ['node', 'cluster']
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def generate(self, seed=None):
params = self.params
if params['name'] == 'Erdos-Renyi':
gtype = snap.PNGraph if params['directed'] else snap.PUNGraph
g = snap.GenRndGnm(gtype, params['n_node'], params['n_edge'])
lst = []
for EI in g.Edges():
lst.append([EI.GetSrcNId(), EI.GetDstNId()])
return pd.DataFrame(lst, columns=['src', 'dest']), None
elif params['name'] == 'ovp_LFR':
if seed is None:
seed = np.random.randint(999999)
with utils.TempDir() as tmpdir:
with open(os.path.join(tmpdir, "seed.txt"), 'wt') as seedf:
seedf.write(str(seed))
program = config.LFR_PROG
newparams = dict(self.params)
if 'directed' in newparams: del newparams['directed']
if 'weighted' in newparams: del newparams['weighted']
newparams['name'] = 'LFR'
cmd = program + " " + " ".join([
"-" + str(u[0]) + " " + str(u[1])
for u in newparams.items() if u[1] is not None
])
self.logger.info("Runing '{}' with seed {}".format(cmd, seed))
self.logger.info("working dir: " + tmpdir)
status = utils.shell_run_and_wait(cmd, working_dir=tmpdir)
if status != 0:
raise Exception(
"run command failed. status={}".format(status))
edgefile = "LFR.nsa" if self.params['directed'] else "LFR.nse"
edgefile = os.path.join(tmpdir, edgefile)
if self.params['weighted']:
edges = pd.read_csv(edgefile,
sep='\t',
header=None,
skiprows=1,
dtype={2: np.float})
edges.columns = ['src', 'dest', 'weight']
else:
edges = pd.read_csv(edgefile,
sep='\t',
header=None,
skiprows=1,
usecols=[0, 1])
edges.columns = ['src', 'dest']
gtfile = "LFR.nmc"
gtfile = os.path.join(tmpdir, gtfile)
gt = pd.read_csv(gtfile, sep='\t', header=None)
gt.columns = ['node', 'cluster']
return self.make_offset_0(edges, gt)
elif params['name'] == 'LFR':
if seed is None:
seed = np.random.randint(999999)
with utils.TempDir() as tmpdir:
with open(os.path.join(tmpdir, "time_seed.dat"),
'wt') as seedf:
seedf.write(str(seed))
program = config.get_LFR_prog(weighted=params['weighted'],
directed=params['directed'],
hier='hier' in params
and params['hier'])
newparams = dict(self.params)
if 'directed' in newparams: del newparams['directed']
if 'weighted' in newparams: del newparams['weighted']
if 'name' in newparams: del newparams['name']
if 'hier' in newparams: del newparams['hier']
cmd = program + " " + " ".join([
"-" + str(u[0]) + " " + str(u[1])
for u in newparams.items() if u[1] is not None
])
self.logger.info("Runing '{}' with seed {}".format(cmd, seed))
self.logger.info("working dir: " + tmpdir)
status = utils.shell_run_and_wait(cmd, working_dir=tmpdir)
if status != 0:
raise Exception(
"run command failed. status={} with seed {}".format(
status, seed))
edgefile = "network.dat"
edgefile = os.path.join(tmpdir, edgefile)
if self.params['weighted']:
edges = pd.read_csv(edgefile,
sep='\t',
header=None,
skiprows=1,
dtype={2: np.float})
edges.columns = ['src', 'dest', 'weight']
else:
edges = pd.read_csv(edgefile,
sep='\t',
header=None,
skiprows=1)
edges.columns = ['src', 'dest']
def read_comm(fname):
gtfile = fname
gtfile = os.path.join(tmpdir, gtfile)
gt = pd.read_csv(gtfile, sep='\t', header=None)
gt.columns = ['node', 'cluster']
return gt
if 'hier' in params and params['hier']:
gt = read_comm("community_first_level.dat"), read_comm(
"community_second_level.dat")
else:
gt = read_comm("community.dat")
return self.make_offset_0(edges, gt)
else:
raise Exception("unknown " + params['name'])
def run(self,
data,
v=5,
v1=None,
v2=None,
prop=None,
repeat=None,
mo=None,
nosplit=False,
extrasimplify=False,
q=False,
seed=None):
assert (v1 is None and v2 is None) or (v1 is not None
and v2 is not None)
params = locals()
del params['self']
del params['data']
if seed is not None: self.logger.info("seed ignored")
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = os.path.join(tmp_dir, 'edges.txt')
utils.remove_if_file_exit(pajek)
os.symlink(data.file_edges, pajek)
cmd = "{} -cp {} COPRA {} -w -v {}".format(
utils.get_java_command(),
config.get_OSLOM_prog('copra', data.is_directed()), pajek, v)
if (v1 is not None and v2 is not None):
cmd += "-vs {} {}".format(v1, v2)
if prop is not None: cmd += ' -prop {}'.format(prop)
if repeat is not None: cmd += ' -repeat {}'.format(repeat)
if mo is not None: cmd += ' -mo'
if nosplit is not None: cmd += ' -nosplit'
if extrasimplify is not None: cmd += ' -extrasimplify'
if q is not None: cmd += ' -q'
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfile = os.path.join(tmp_dir, 'clusters-edges.txt')
with open(outputfile) as f:
lines = [u.strip() for u in f]
lines = [[int(v) for v in u.split(" ")] for u in lines]
clusters = dict(enumerate(lines))
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self,
data,
startk=None,
finalk=None,
runs=None,
ensemblesize=None,
algorithm=None,
seed=None):
if False and (data.is_directed() or data.is_weighted()):
raise Exception(
"only undirected and unweighted graph is supported")
params = locals()
del (params['self'])
del (params['data'])
if seed is None:
seed = np.random.randint(999999)
params['seed'] = seed
params['inpfmt'] = 'e'
params['outfile'] = 'output'
params['outfmt'] = 'l'
params = {u: v for u, v in params.items() if v is not None}
if not utils.file_exists(data.file_edges):
data.to_edgelist()
cmd = "{} {} {}".format(
config.CGGC_PROG,
" ".join(['--{}={}'.format(u, v) for u, v in params.items()]),
data.file_edges)
self.logger.info("Running " + cmd)
with utils.TempDir() as tmp_dir:
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
with open(os.path.join(tmp_dir, "output"), "r") as output:
lines = [u.strip() for u in output.readlines()]
from collections import defaultdict
clusters = defaultdict(list)
for c, line in enumerate(lines):
if line.startswith('#'): continue
for n in line.split(" "):
clusters[c].append(int(n))
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self,
data,
seed=None,
thread=None,
threshold=1e-3,
vThresh=0.0,
maxIter=10,
prior=True):
if seed is None:
seed = np.random.randint(999999)
params = locals()
del params['self']
del params['data']
params = {u: v for u, v in params.items() if v is not None}
thread = utils.get_num_thread(thread)
params['thread'] = thread
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
pajek = utils.link_file(data.file_edges, tmp_dir, 'edges.txt')
cmd = "{prog} {seed} {graph} {thread} 1 {threshold} {vThresh} {maxIter} {output_dir} {prior}"\
.format(prog=config.get_powergraph_prog('RelaxMap', data.is_directed()), seed=seed,
graph=pajek, thread=thread, threshold=threshold, vThresh=vThresh, output_dir=tmp_dir,
maxIter=maxIter, prior='prior' if prior else 'normal').strip()
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfile = os.path.join(tmp_dir, "edges.tree")
import pandas as pd
output = pd.read_csv(outputfile, sep=" ", header=None, skiprows=1)
output['node'] = output.loc[:, 2].astype(np.int)
output['cluster'] = output.loc[:, 0].map(
lambda u: u.split(':')[0]).astype(np.int)
clusters = output[['cluster', 'node']]
clusters = clusters.groupby('cluster').apply(
lambda u: list(u['node'])).to_dict()
self.logger.info("Made %d clusters in %f seconds" %
(len(clusters), timecost))
result = {}
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
def run(self,
data,
seed=None,
r=10,
hr=50,
t=0.1,
cp=0.5,
fast=False,
singlet=False,
infomap=False,
copra=False,
louvain=False,
runs=1):
params = locals()
del params['data']
del params['self']
if seed is None:
seed = np.random.randint(999999)
params['seed'] = seed
if (data.is_directed() or data.is_weighted()) and False:
raise Exception(
"only undirected and unweighted graph is supported")
if int(infomap) + int(copra) + int(louvain) > 1:
raise Exception("only of infomap, corpra, louvain can be true")
if not utils.file_exists(data.file_edges):
data.to_edgelist()
with utils.TempDir() as tmp_dir:
# tmp_dir = "/tmp/abc"
def link_file(path, destname=None):
if destname is None:
destname = path.split("/")[-1]
destpath = os.path.join(tmp_dir, destname)
utils.remove_if_file_exit(destpath)
os.symlink(path, destpath)
return destpath
pajek = link_file(data.file_edges)
if copra:
_ = link_file(
config.get_OSLOM_prog('copra', data.is_directed()))
if infomap:
_ = link_file(config.get_OSLOM_prog('infomap_script', True))
_ = link_file(config.get_OSLOM_prog('infomap', True))
_ = link_file(config.get_OSLOM_prog('infomap_script', False))
_ = link_file(config.get_OSLOM_prog('infomap', False))
if louvain:
_ = link_file(
config.get_OSLOM_prog('louvain_script',
data.is_directed()))
_ = link_file(
config.get_OSLOM_prog('convert', data.is_directed()))
_ = link_file(
config.get_OSLOM_prog('community', data.is_directed()))
_ = link_file(
config.get_OSLOM_prog('hierarchy', data.is_directed()))
cmd = "{} -f {} -{} -r {} -hr {} -seed {} -t {} -cp {}".format(
config.get_OSLOM_prog('oslom', data.is_directed()), pajek,
'w' if data.is_weighted() else 'uw', r, hr, seed, t, cp)
if fast: cmd += " -fast"
if singlet: cmd += " -singlet"
if infomap: cmd += " -infomap {}".format(runs)
if copra: cmd += " -copra {}".format(runs)
if louvain: cmd += " -louvain {}".format(runs)
self.logger.info("Running " + cmd)
timecost, status = utils.timeit(
lambda: utils.shell_run_and_wait(cmd, tmp_dir))
if status != 0:
raise Exception(
"Run command with error status code {}".format(status))
outputfiles = glob.glob(
os.path.join(tmp_dir, "edges.txt_oslo_files", "tp*"))
clusters = {}
for tp in outputfiles:
fname = tp.split("/")[-1]
if fname == 'tp':
level = 0
else:
level = int(fname[2:])
with open(tp) as f:
lines = [u.strip() for u in f if not u.startswith('#')]
lines = [[int(v) for v in u.split(" ")] for u in lines]
clusters[level] = dict(enumerate(lines))
max_level = max(list(clusters.keys()))
self.logger.info(
"Made %d levels of clusters with #clusters %s in %f seconds" %
(len(clusters), str([len(u)
for u in clusters.values()]), timecost))
result = {}
result['multilevel'] = True
result['num_level'] = len(clusters)
result['max_level'] = max_level
result['runname'] = self.name
result['params'] = params
result['dataname'] = data.name
result['meta'] = self.get_meta()
result['timecost'] = timecost
result['clusters'] = clusters
save_result(result)
self.result = result
return self
