def start_coloring(filename, td, cfgfile, output):
m = ccalgorithm_factory(cfgfile, False, None)
p, fn = os.path.split(filename)
graphname, e = os.path.splitext(fn)
col = m.start(load_graph(filename), td)
# Store results in common folder and, if supplied, in output file
save_file(col, 'colorings/' + graphname + str(td), False)
if output:
save_file(col, output, True)
def start_coloring(filename, td, cfgfile, output):
m = ccalgorithm_factory(cfgfile, False)
p, fn = os.path.split(filename)
graphname, e = os.path.splitext(fn)
col = m.start(load_graph(filename), td)
# Store results in common folder and, if supplied, in output file
save_file(col, 'colorings/' + graphname + str(td), False)
if output:
save_file(col, output, True)
def get_pattern_from_file(filename):
"""
Load pattern from a file
:param filename: The name of the pattern file
:return: A graph object representing the pattern along with its treedepth
"""
# Argument is a filename
try:
# Try to load the graph from file
H = load_graph(filename)
# Return pattern along with lower bound on its treedepth
return H, treedepth(H)
except Exception:
# Invalid file extension
pattern_argument_error_msg(filename)
def get_pattern_from_file(filename):
"""
Load pattern from a file
:param filename: The name of the pattern file
:return: A graph object representing the pattern along with its treedepth
"""
# Argument is a filename
try:
# Try to load the graph from file
H = load_graph(filename)
# Return pattern along with lower bound on its treedepth
return H, treedepth(H)
except Exception:
# Invalid file extension
pattern_argument_error_msg(filename)
for v in cols:
f.write(str(v)+': '+str(cols[v])+'\n')
f.close()
if __name__=="__main__":
parser = argparse.ArgumentParser()
parser.add_argument("graph", help="A graph in any supported format "
"(see graphformats.py)")
parser.add_argument("colors", help="A vertex coloring file")
parser.add_argument("p", help="the number of color classes that yield "
"low-td components, i.e. the coloring is (p+1)-"
"centered coloring",
type=int)
args = parser.parse_args()
graph = load_graph(args.graph)
remove_loops(graph)
cols = load_coloring(args.colors)
p = args.p
print "Loaded {0} and coloring {1} ({2} colors)".format(args.graph,
args.colors,
len(cols))
res = cols.is_proper(graph) and check_tree_depth(graph, graph, cols, p)
print "{0}-centered coloring: {1}".format(p+1, res)
print ""
optcols = merge_colors(graph, cols, p)
size = len(optcols)
print "Best coloring found has size", size
def parse_pattern_argument(pattern):
"""
Parses the 'pattern' command line argument.
Checks to see if this argument is a filename or
a description of the pattern
:param pattern: Filename or description of pattern
:return: A tuple with the pattern graph and a lower
bound on its treedepth
"""
import os
# Get the name of the file and the file extension
name, ext = os.path.splitext(pattern)
# There is no extension, so argument is a description
# of the pattern
if ext == "":
import re
p = re.compile(r'(\d*)')
# Parse out the different parts of the argument
args = filter(lambda x: x != "" and x != ",", p.split(pattern))
# There are two parts
if len(args) == 2 and args[0] not in pattern_gen.bipartite_patterns:
try:
# Get the generator for the pattern type
generator = pattern_gen.get_generator(args[0])
# Get the number of vertices provided
pattern_num_vertices = int(args[1])
# Generate the pattern
H = generator(pattern_num_vertices)
# Return the pattern along with its treedepth
return H, treedepth(H, args[0], pattern_num_vertices)
except KeyError:
pattern_argument_error_msg()
# Bipartite pattern type provided
elif len(args) == 3 and args[0] in pattern_gen.bipartite_patterns:
# Make sure it is a valid bipartite pattern
try:
generator = pattern_gen.get_generator(args[0])
# Try to get the two set sizes
m = int(args[1])
n = int(args[2])
# Generate the pattern
H = generator(m, n)
# Return the pattern along with its treedepth
return H, treedepth(H, args[0], m, n)
except (KeyError, ValueError):
# Invalid sizes provided
pattern_argument_error_msg()
else:
# Number of vertices not provided in argument
pattern_argument_error_msg()
else:
# Argument is a filename
try:
# Try to load the graph from file
H = load_graph(pattern)
# Return pattern along with lower bound on its treedepth
return H, treedepth(H)
except Exception:
# Invalid file extension
pattern_argument_error_msg()
def runPipeline(graph, pattern, cfgFile, colorFile, color_no_verify, output,
verbose, profile):
"""Basic running of the pipeline"""
if profile: # time profiling
readProfile = cProfile.Profile()
readProfile.enable()
H, td_lower = parse_pattern_argument(pattern)
# Read graphs from file
G = load_graph(graph)
td = len(H)
G_path, G_local_name = os.path.split(graph)
G_name, G_extension = os.path.splitext(G_local_name)
if verbose:
print "G has {0} vertices and {1} edges".format(len(G), G.num_edges())
if profile: # time profiling
readProfile.disable()
printProfileStats("reading graphs", readProfile)
colorProfile = cProfile.Profile()
colorProfile.enable()
# Find p-centered coloring
if colorFile is None:
coloring = p_centered_coloring(G, td, cfgFile, verbose)
save_file(coloring, 'colorings/' + G_name + str(td), False, verbose)
else:
coloring = coloring_from_file(colorFile, G, td, cfgFile, verbose,
not color_no_verify)
if profile: # time profiling
colorProfile.disable()
printProfileStats("coloring", colorProfile)
patternProfile = cProfile.Profile()
patternProfile.enable()
# Get configuration settings for dynamic programming
cfgParser = parse_config_safe(cfgFile)
kpat_name = cfgParser.get('compute', 'k_pattern')
# tab_name = cfgParser.get('dp', 'tab')
table_hints = {
'forward': cfgParser.getboolean('compute', 'table_forward'),
'reuse': cfgParser.getboolean('compute', 'table_reuse')
}
count_name = cfgParser.get('combine', 'count')
sweep_name = cfgParser.get('decompose', 'sweep')
patternClass = import_modules("lib.pattern_counting.dp."+kpat_name)
count_class = import_modules('lib.pattern_counting.double_count.' +
count_name)
sweep_class = import_modules('lib.decomposition.' + sweep_name)
# Count patterns
pattern_counter = PatternCounter(G, H, td_lower, coloring,
pattern_class=patternClass,
table_hints=table_hints,
decomp_class=sweep_class,
combiner_class=count_class,
verbose=verbose)
patternCount = pattern_counter.count_patterns()
print "Number of occurrences of H in G: {0}".format(patternCount)
if profile: # time profiling
patternProfile.disable()
printProfileStats("pattern counting", patternProfile)
def runPipeline(graph, pattern, cfgFile, colorFile, color_no_verify, output,
verbose, profile, multifile, execution_data):
"""Basic running of the pipeline"""
# Check if execution_data flag is set
execdata = execution_data is not None
if execdata and pattern == "multi":
print "CONCUSS does not support outputting execution data while using the multi-motif flag"
sys.exit(1)
if execdata:
# Check if directory exists already
if os.path.isdir("./execdata"):
# delete it, if it does exist
shutil.rmtree("./execdata")
# make new directory called "execdata"
os.mkdir("./execdata")
if profile: # time profiling
readProfile = cProfile.Profile()
readProfile.enable()
# Parse the multifile if counting multiple patterns
if pattern == 'multi':
multi, td_list = parse_multifile(multifile)
# Parse pattern argument
else:
basic_pattern = is_basic_pattern(pattern)
if basic_pattern:
H, td_lower = get_pattern_from_generator(pattern)
else:
H, td_lower = get_pattern_from_file(pattern)
multi = [H]
td_list = [td_lower]
# Read graphs from file
G = load_graph(graph)
td = len(max(multi, key=len))
G_path, G_local_name = os.path.split(graph)
G_name, G_extension = os.path.splitext(G_local_name)
if verbose:
print "G has {0} vertices and {1} edges".format(len(G), G.num_edges())
if profile: # time profiling
readProfile.disable()
printProfileStats("reading graphs", readProfile)
colorProfile = cProfile.Profile()
colorProfile.enable()
# Find p-centered coloring
if colorFile is None:
coloring = p_centered_coloring(G, td, cfgFile, verbose, execdata)
save_file(coloring, 'colorings/' + G_name + str(td), False, verbose)
else:
coloring = coloring_from_file(colorFile, G, td, cfgFile, verbose,
not color_no_verify)
if profile: # time profiling
colorProfile.disable()
printProfileStats("coloring", colorProfile)
patternProfile = cProfile.Profile()
patternProfile.enable()
# Get configuration settings for dynamic programming
cfgParser = parse_config_safe(cfgFile)
kpat_name = cfgParser.get('compute', 'k_pattern')
# tab_name = cfgParser.get('dp', 'tab')
table_hints = {
'forward': cfgParser.getboolean('compute', 'table_forward'),
'reuse': cfgParser.getboolean('compute', 'table_reuse')
}
count_name = cfgParser.get('combine', 'count')
sweep_name = cfgParser.get('decompose', 'sweep')
patternClass = import_modules("lib.pattern_counting.dp." + kpat_name)
count_class = import_modules('lib.pattern_counting.double_count.' +
count_name)
sweep_class = import_modules('lib.decomposition.' + sweep_name)
# Output for Count stage
if execdata:
count_path = 'execdata/count/'
if not os.path.exists(count_path):
os.makedirs(count_path)
big_component_file = open(count_path+'big_component.txt', 'w')
tdd_file = open(count_path+'tdd.txt', 'w')
dp_table_file = open(count_path+'dp_table.txt', 'w')
else:
big_component_file = None
tdd_file = None
dp_table_file = None
# Output for Combine stage
if execdata:
combine_path = 'execdata/combine/'
if not os.path.exists(combine_path):
os.makedirs(combine_path)
# Open the file that needs to be passed to the count combiner
colset_count_file = open('execdata/combine/counts_per_colorset.txt',
'w')
else:
# If execution data is not requested, we don't need to open a file
colset_count_file = None
if count_name != "InclusionExclusion" and execdata:
print "CONCUSS can only output execution data using the",
print "InclusionExclusion combiner class."
# Check if there is incomplete execution data written
if os.path.isdir("./execdata"):
# delete it, if it does exist
shutil.rmtree("./execdata")
# Exit the program with an error code of 1
sys.exit(1)
pattern_counter = PatternCounter(G, multi, td_list, coloring,
pattern_class=patternClass,
table_hints=table_hints,
decomp_class=sweep_class,
combiner_class=count_class,
verbose=verbose,
big_component_file=big_component_file,
tdd_file=tdd_file,
dp_table_file=dp_table_file,
colset_count_file=colset_count_file)
pattern_count = pattern_counter.count_patterns()
# Patterns have been counted, print output
if pattern == "multi":
with open(multifile[0], 'r') as pattern_file:
pattern_names = [pat[:-1] for pat in pattern_file]
else:
pattern_names = [pattern]
for i in range(len(pattern_names)):
print "Number of occurrences of {0} in G: {1}".format(pattern_names[i], pattern_count[i])
if execdata:
# Close count stage files
big_component_file.close()
tdd_file.close()
dp_table_file.close()
# Close the color set file
colset_count_file.close()
# if execution data flag is set
# make and write to visinfo.cfg
with open('execdata/visinfo.cfg', 'w') as visinfo:
write_visinfo(visinfo, graph, pattern)
# write execution data to zip
with ZipFile(execution_data, 'w') as exec_zip:
# exec_zip.write("execdata/visinfo.cfg", "visinfo.cfg")
# Write data from 'execdata' directory to the zip:
rootDir = './execdata/'
for dir_name, _, file_list in os.walk(rootDir):
for f_name in file_list:
full_path = dir_name + '/' + f_name
exec_zip.write(
full_path,
'/'.join(full_path.split('/')[2:]))
exec_zip.write(cfgFile, os.path.split(cfgFile)[1])
exec_zip.write(graph, os.path.split(graph)[1])
# Check to see if the user specified a basic pattern
if basic_pattern:
from lib.graph.graphformats import write_edgelist
# Write the pattern graph object to a file as an edgelist
with open(pattern + '.txt', 'w') as pattern_file:
write_edgelist(H, pattern_file)
# Write the file to the zip
exec_zip.write(
pattern + '.txt',
os.path.split(pattern + '.txt')[1])
# File written to zip, delete it
os.remove(pattern + '.txt')
else:
exec_zip.write(pattern, os.path.split(pattern)[1])
# delete execution data stored in folder
shutil.rmtree('./execdata')
if profile: # time profiling
patternProfile.disable()
printProfileStats("pattern counting", patternProfile)
def parse_pattern_argument(pattern):
"""
Parses the 'pattern' command line argument.
Checks to see if this argument is a filename or
a description of the pattern
:param pattern: Filename or description of pattern
:return: A tuple with the pattern graph and a lower
bound on its treedepth
"""
# Get the name of the file and the file extension
name, ext = os.path.splitext(pattern)
# There is no extension, so argument is a description
# of the pattern
if ext == "":
import re
p = re.compile(r'(\d*)')
# Parse out the different parts of the argument
args = filter(lambda x: x != "" and x != ",", p.split(pattern))
# There are two parts
if len(args) == 2 and args[0] not in pattern_gen.bipartite_patterns:
try:
# Get the generator for the pattern type
generator = pattern_gen.get_generator(args[0])
# Get the number of vertices provided
pattern_num_vertices = int(args[1])
# Generate the pattern
H = generator(pattern_num_vertices)
# Return the pattern along with its treedepth
return H, treedepth(H, args[0], pattern_num_vertices)
except KeyError:
pattern_argument_error_msg(pattern)
# Bipartite pattern type provided
elif len(args) == 3 and args[0] in pattern_gen.bipartite_patterns:
# Make sure it is a valid bipartite pattern
try:
generator = pattern_gen.get_generator(args[0])
# Try to get the two set sizes
m = int(args[1])
n = int(args[2])
# Generate the pattern
H = generator(m, n)
# Return the pattern along with its treedepth
return H, treedepth(H, args[0], m, n)
except (KeyError, ValueError):
# Invalid sizes provided
pattern_argument_error_msg(pattern)
else:
# Number of vertices not provided in argument
pattern_argument_error_msg(pattern)
else:
# Argument is a filename
try:
# Try to load the graph from file
H = load_graph(pattern)
# Return pattern along with lower bound on its treedepth
return H, treedepth(H)
except Exception:
# Invalid file extension
pattern_argument_error_msg(pattern)
def cycle(num_vertices):
"""
Compute the lower bound on treedepth of a cycle.
:param num_vertices: Tuple containing number of elements
:return: the treedepth
"""
# return lower bound
return int(ceil(log(num_vertices[0], 2))) + 1
def biclique(num_vertices):
"""
Compute the lower bound on treedepth of a clique.
:param num_vertices: 2-Tuple containing number of elements in
first and second set
:return: the treedepth
"""
# The treedepth of a biclique is the number of nodes in
# the smaller set + 1
return min(num_vertices[0], num_vertices[1]) + 1
if __name__ == "__main__":
G = load_graph(sys.argv[1])
print treedepth(G)
def runPipeline(
graph, pattern, cfgFile, colorFile, color_no_verify, output, verbose, profile, multifile, execution_data
):
"""Basic running of the pipeline"""
# Check if execution_data flag is set
execdata = execution_data is not None
if execdata and pattern == "multi":
print "CONCUSS does not support outputting execution data while using the multi-motif flag"
sys.exit(1)
if execdata:
# Check if directory exists already
if os.path.isdir("./execdata"):
# delete it, if it does exist
shutil.rmtree("./execdata")
# make new directory called "execdata"
os.mkdir("./execdata")
if profile: # time profiling
readProfile = cProfile.Profile()
readProfile.enable()
# Parse the multifile if counting multiple patterns
if pattern == "multi":
multi, td_list = parse_multifile(multifile)
# Parse pattern argument
else:
basic_pattern = is_basic_pattern(pattern)
if basic_pattern:
H, td_lower = get_pattern_from_generator(pattern)
else:
H, td_lower = get_pattern_from_file(pattern)
multi = [H]
td_list = [td_lower]
# Read graphs from file
G = load_graph(graph)
td = len(max(multi, key=len))
G_path, G_local_name = os.path.split(graph)
G_name, G_extension = os.path.splitext(G_local_name)
if verbose:
print "G has {0} vertices and {1} edges".format(len(G), G.num_edges())
if profile: # time profiling
readProfile.disable()
printProfileStats("reading graphs", readProfile)
colorProfile = cProfile.Profile()
colorProfile.enable()
# Find p-centered coloring
if colorFile is None:
coloring = p_centered_coloring(G, td, cfgFile, verbose, execdata)
save_file(coloring, "colorings/" + G_name + str(td), False, verbose)
else:
coloring = coloring_from_file(colorFile, G, td, cfgFile, verbose, not color_no_verify)
if profile: # time profiling
colorProfile.disable()
printProfileStats("coloring", colorProfile)
patternProfile = cProfile.Profile()
patternProfile.enable()
# Get configuration settings for dynamic programming
cfgParser = parse_config_safe(cfgFile)
kpat_name = cfgParser.get("compute", "k_pattern")
# tab_name = cfgParser.get('dp', 'tab')
table_hints = {
"forward": cfgParser.getboolean("compute", "table_forward"),
"reuse": cfgParser.getboolean("compute", "table_reuse"),
}
count_name = cfgParser.get("combine", "count")
sweep_name = cfgParser.get("decompose", "sweep")
patternClass = import_modules("lib.pattern_counting.dp." + kpat_name)
count_class = import_modules("lib.pattern_counting.double_count." + count_name)
sweep_class = import_modules("lib.decomposition." + sweep_name)
# Output for Count stage
if execdata:
count_path = "execdata/count/"
if not os.path.exists(count_path):
os.makedirs(count_path)
big_component_file = open(count_path + "big_component.txt", "w")
tdd_file = open(count_path + "tdd.txt", "w")
dp_table_file = open(count_path + "dp_table.txt", "w")
else:
big_component_file = None
tdd_file = None
dp_table_file = None
# Output for Combine stage
if execdata:
combine_path = "execdata/combine/"
if not os.path.exists(combine_path):
os.makedirs(combine_path)
# Open the file that needs to be passed to the count combiner
colset_count_file = open("execdata/combine/counts_per_colorset.txt", "w")
else:
# If execution data is not requested, we don't need to open a file
colset_count_file = None
if count_name != "InclusionExclusion" and execdata:
print "CONCUSS can only output execution data using the",
print "InclusionExclusion combiner class."
# Check if there is incomplete execution data written
if os.path.isdir("./execdata"):
# delete it, if it does exist
shutil.rmtree("./execdata")
# Exit the program with an error code of 1
sys.exit(1)
pattern_counter = PatternCounter(
G,
multi,
td_list,
coloring,
pattern_class=patternClass,
table_hints=table_hints,
decomp_class=sweep_class,
combiner_class=count_class,
verbose=verbose,
big_component_file=big_component_file,
tdd_file=tdd_file,
dp_table_file=dp_table_file,
colset_count_file=colset_count_file,
)
pattern_count = pattern_counter.count_patterns()
# Patterns have been counted, print output
if pattern == "multi":
with open(multifile[0], "r") as pattern_file:
pattern_names = [pat[:-1] for pat in pattern_file]
else:
pattern_names = [pattern]
for i in range(len(pattern_names)):
print "Number of occurrences of {0} in G: {1}".format(pattern_names[i], pattern_count[i])
if execdata:
# Close count stage files
big_component_file.close()
tdd_file.close()
dp_table_file.close()
# Close the color set file
colset_count_file.close()
# if execution data flag is set
# make and write to visinfo.cfg
with open("execdata/visinfo.cfg", "w") as visinfo:
write_visinfo(visinfo, graph, pattern)
# write execution data to zip
with ZipFile(execution_data, "w") as exec_zip:
# exec_zip.write("execdata/visinfo.cfg", "visinfo.cfg")
# Write data from 'execdata' directory to the zip:
rootDir = "./execdata/"
for dir_name, _, file_list in os.walk(rootDir):
for f_name in file_list:
full_path = dir_name + "/" + f_name
exec_zip.write(full_path, "/".join(full_path.split("/")[2:]))
exec_zip.write(cfgFile, os.path.split(cfgFile)[1])
exec_zip.write(graph, os.path.split(graph)[1])
# Check to see if the user specified a basic pattern
if basic_pattern:
from lib.graph.graphformats import write_edgelist
# Write the pattern graph object to a file as an edgelist
with open(pattern + ".txt", "w") as pattern_file:
write_edgelist(H, pattern_file)
# Write the file to the zip
exec_zip.write(pattern + ".txt", os.path.split(pattern + ".txt")[1])
# File written to zip, delete it
os.remove(pattern + ".txt")
else:
exec_zip.write(pattern, os.path.split(pattern)[1])
# delete execution data stored in folder
shutil.rmtree("./execdata")
if profile: # time profiling
patternProfile.disable()
printProfileStats("pattern counting", patternProfile)
def cycle(num_vertices):
"""
Compute the lower bound on treedepth of a cycle.
:param num_vertices: Tuple containing number of elements
:return: the treedepth
"""
# return lower bound
return int(ceil(log(num_vertices[0], 2))) + 1
def biclique(num_vertices):
"""
Compute the lower bound on treedepth of a clique.
:param num_vertices: 2-Tuple containing number of elements in
first and second set
:return: the treedepth
"""
# The treedepth of a biclique is the number of nodes in
# the smaller set + 1
return min(num_vertices[0], num_vertices[1]) + 1
if __name__ == "__main__":
G = load_graph(sys.argv[1])
print treedepth(G)