def generate_project_kernel(project, cluster_json=None):
""" run graph kernel computation """
out_dir = dot.dot_dirs(project)[0]
print("Generating kernel for {0} with dot files in {1}.".format(project, out_dir))
kernel_file_path = dot.get_kernel_path(project, out_dir)
if cluster_json:
print("Using clustering output for node relabeling:")
graph_kernel_cmd = ['python',
common.get_simprog('precompute_kernel.py'),
os.path.join(dot.dot_dir(project), out_dir),
kernel_file_path,
cluster_json
]
common.run_cmd(graph_kernel_cmd, True)
else:
graph_kernel_cmd = ['python',
common.get_simprog('precompute_kernel.py'),
os.path.join(dot.dot_dir(project), out_dir),
kernel_file_path
]
common.run_cmd(graph_kernel_cmd, True)
print("Generated kernel file for {0} in {1}.".format(project, kernel_file_path))
def generate_project_kernel(project, cluster_json=None):
""" run graph kernel computation """
project_dir = common.get_project_dir(project)
dot_dirs = dot.dot_dirs(project)
if not dot_dirs:
print("No graphs generated for {}".format(project))
return
out_dir = dot_dirs[0]
kernel_file_path = dot.get_kernel_path(project, out_dir)
if cluster_json:
graph_kernel_cmd = [
'python',
common.get_simprog('precompute_kernel.py'), project_dir,
kernel_file_path, cluster_json
]
common.run_cmd(graph_kernel_cmd, 'graphkernel')
else:
graph_kernel_cmd = [
'python',
common.get_simprog('precompute_kernel.py'), project_dir,
kernel_file_path
]
common.run_cmd(graph_kernel_cmd, 'graphkernel')
def get_method_map(project_list, include_all=True):
dot_to_method_map = {}
for project in project_list:
project_dot_dirs = []
if include_all:
project_dot_dirs = dot.dot_dirs(project)
else:
project_dot_dirs = [dot.dot_dirs(project)[0]]
for output_dir in project_dot_dirs:
method_file = dot.get_method_path(project, output_dir)
if not os.path.isfile(method_file):
print("Cannot find method file for project {0} at {1}".format(project, method_file))
sys.exit(0)
with open(method_file, "r") as mf:
content = mf.readlines()
for line in content:
line = line.rstrip()
items = line.split('\t')
method_name = items[0]
method_dot = items[1]
method_dot_path = dot.get_dot_path(project, output_dir, method_dot)
dot_to_method_map[method_dot_path] = method_name
return dot_to_method_map
def get_dot_method_map(proj_lst):
dot_method_map = {}
for proj in proj_lst:
output_dir_lst = dot.dot_dirs(proj)
for output_dir in output_dir_lst:
method_file = dot.get_method_path(proj, output_dir)
with open(method_file, "r") as mf:
for line in mf:
line = line.rstrip()
items = line.split("\t")
method_name = items[0]
method_dot = items[1]
method_dot_path = dot.get_dot_path(proj, output_dir,
method_dot)
dot_method_map[method_dot_path] = method_name
return dot_method_map
def gather_kernels(projects, corpus_kernel_file):
print("Gathering kernels from projects {0}".format(" and ".join(projects)))
with open(corpus_kernel_file, "w") as corpus_kernel_file_handle:
for project in projects:
project_dir = common.get_project_dir(project)
out_dir = dot.dot_dirs(project)[0] # only consider the first one
project_kernel_file_path = dot.get_kernel_path(project, out_dir)
if os.path.isfile(project_kernel_file_path):
with open(project_kernel_file_path, "r") as fi:
corpus_kernel_file_handle.write(fi.read())
else:
print ("No kernel file find for project {0}.\n {1} is not a file.".format(
project,
project_kernel_file_path
))