def build_reference_graph():
""" Builds a reference graph which shall be the same
than the graph of ../test_files.
A PyGraphml graph is returned.
"""
g = pgml.Graph()
f1 = g.add_node("1")
f2 = g.add_node("2")
f3 = g.add_node("3")
f4 = g.add_node("4")
f5 = g.add_node("5")
f6 = g.add_node("6")
f1['key1'] = 'motorcycle_test.rb'
f2['key1'] = 'lib/motorcycle.rb'
f3['key1'] = 'motorcycle_2.rb'
f4['key1'] = 'motorcycle_3.rb'
f5['key1'] = 'motorcycle_4.rb'
f6['key1'] = 'lib/sidecar.rb'
g.add_edge(f1, f2)
g.add_edge(f1, f3)
g.add_edge(f1, f4)
g.add_edge(f1, f5)
g.add_edge(f4, f6)
return g
def main(args):
maven_modules = find_maven_modules(args.maven_module_paths)
graph = pygraphml.Graph()
nodes_store = NodesStore(graph)
DependencyGraphBuilder(graph, nodes_store).build_graph(maven_modules)
if (args.include_parent_edges):
ParentEdgeBuilder(graph, nodes_store).build_graph(maven_modules)
write_graph(args.graph_output_path, graph)
def build_reference_graph( self ):
""" Builds a reference graph which shall be the same
than the graph of ../test_files.
A PyGraphml graph is returned.
"""
g = pgml.Graph()
f_1_c = g.add_node( "1" )
f_2_c = g.add_node( "2" )
f_3_c = g.add_node( "3" )
existing_h = g.add_node("4")
iostream = g.add_node("5")
lib_f_1_h = g.add_node("6")
lib_f_2_h = g.add_node("7")
lib_f_3_h = g.add_node("8")
lib2_f_4_h = g.add_node("9")
lib2_f_1_h = g.add_node("10")
e_1_x = g.add_node("11")
e_2_y = g.add_node("12")
f_1_c ['key1'] = 'src/f_1.c'
f_2_c ['key1'] = 'src/f_2.c'
f_3_c ['key1'] = 'src/f_3.c'
existing_h ['key1'] = '../non/existing.h'
iostream ['key1'] = 'iostream'
lib_f_1_h ['key1'] = 'inc/lib/f_1.h'
lib_f_2_h ['key1'] = 'inc/lib/f_2.h'
lib_f_3_h ['key1'] = 'inc/lib/f_3.h'
lib2_f_4_h ['key1'] = 'inc/lib2/f_4.h'
lib2_f_1_h ['key1'] = 'inc/lib2/f_1.h'
e_1_x ['key1'] = 'inc/lib/e_1.x'
e_2_y ['key1'] = 'inc/lib/e_2.y'
g.add_edge( f_3_c, existing_h )
g.add_edge( f_3_c, iostream )
g.add_edge( f_2_c, lib2_f_1_h )
g.add_edge( f_1_c, iostream )
g.add_edge( f_1_c, lib_f_2_h )
g.add_edge( f_1_c, lib_f_1_h )
g.add_edge( lib_f_2_h, lib_f_1_h )
g.add_edge( lib_f_1_h, lib_f_3_h )
g.add_edge( lib_f_1_h, lib2_f_4_h )
g.add_edge( lib_f_3_h, lib_f_1_h )
return g
def build_reference_graph(self):
""" Builds a reference graph which shall be the same
than the graph of ../test_files.
A PyGraphml graph is returned.
"""
""" List of files by doing:
find -iname '*.py' -or -iname '*.pyw' -or -iname '*.py3'
Use as basis for constructing the graph
./py/pack3/__init__.py
./py/pack3/file1.py
./py/file3.pyw
./py/pack2/__init__.py
./py/pack2/file3.py
./py/pack2/file2.py
./py/pack2/file1.py
./py/file2.py3
./py/file1.py
./py/pack1/__init__.py
./py/pack1/file3.py
./py/pack1/file2.py
./py/pack1/file1.py
./py/pack1/subpack1/__init__.py
./py/pack1/subpack1/file3.py
./py/pack1/subpack1/file2.py
./py/pack1/subpack1/file1.py
"""
g = pgml.Graph()
f1 = g.add_node("1")
f2 = g.add_node("2")
f3 = g.add_node("3")
p1_f1 = g.add_node("4")
p1_f2 = g.add_node("5")
p1_f3 = g.add_node("6")
p1_init = g.add_node("7")
p1_s1_f1 = g.add_node("8")
p1_s1_f2 = g.add_node("9")
p1_s1_f3 = g.add_node("10")
p1_s1_init = g.add_node("11")
p2_f1 = g.add_node("12")
p2_f2 = g.add_node("13")
p2_f3 = g.add_node("14")
p2_init = g.add_node("15")
p3_f1 = g.add_node("16")
p3_init = g.add_node("17")
bogusfile = g.add_node("18")
filecmp = g.add_node("19")
os = g.add_node("20")
pickle = g.add_node("21")
sys = g.add_node("22")
f1['key1'] = 'file1.py'
f2['key1'] = 'file2.py3'
f3['key1'] = 'file3.pyw'
p1_f1['key1'] = 'pack1/file1.py'
p1_f2['key1'] = 'pack1/file2.py'
p1_f3['key1'] = 'pack1/file3.py'
p1_init['key1'] = 'pack1/__init__.py'
p1_s1_f1['key1'] = 'pack1/subpack1/file1.py'
p1_s1_f2['key1'] = 'pack1/subpack1/file2.py'
p1_s1_f3['key1'] = 'pack1/subpack1/file3.py'
p1_s1_init['key1'] = 'pack1/subpack1/__init__.py'
p2_f1['key1'] = 'pack2/file1.py'
p2_f2['key1'] = 'pack2/file2.py'
p2_f3['key1'] = 'pack2/file3.py'
p2_init['key1'] = 'pack2/__init__.py'
p3_f1['key1'] = 'pack3/file1.py'
p3_init['key1'] = 'pack3/__init__.py'
bogusfile['key1'] = 'bogusfilename'
filecmp['key1'] = 'filecmp'
os['key1'] = 'os'
pickle['key1'] = 'pickle'
sys['key1'] = 'sys'
# Root
g.add_edge(f1, p1_s1_f1)
g.add_edge(f1, p1_init)
g.add_edge(f1, p2_init)
g.add_edge(f2, sys)
g.add_edge(f2, p1_s1_init) # *-import
g.add_edge(f2, p1_f1)
g.add_edge(f2, os)
g.add_edge(f2, p2_init)
g.add_edge(f3, p2_init) # *-import
g.add_edge(f3, bogusfile)
# Pack 1
g.add_edge(p1_f1, p1_s1_f1)
g.add_edge(p1_f1, p2_f1)
g.add_edge(p1_f2, f1)
g.add_edge(p1_f2, f3)
g.add_edge(p1_f2, p1_s1_f1)
g.add_edge(p1_f3, sys)
g.add_edge(p1_init, pickle)
# Pack 1, Subpack 1
g.add_edge(p1_s1_f1, f3)
g.add_edge(p1_s1_f2, sys)
g.add_edge(p1_s1_f2, f1)
g.add_edge(p1_s1_f3, filecmp)
# Pack 2 (__init__ has __all__ import)
g.add_edge(p2_f2, filecmp)
g.add_edge(p2_f3, p1_f2)
g.add_edge(p2_init, p2_f1)
g.add_edge(p2_init, p2_f2)
g.add_edge(p2_init, p2_f3)
# Pack 3 (__init__ has __all__ import)
g.add_edge(p3_init, p3_f1)
return g
import pymongo
import pygraphml
# MongoDB
db = pymongo.MongoClient('127.0.0.1', 27017).superblogger
g = pygraphml.Graph()
nodes = dict()
for user in db.users.find({
'counts.followed_by': {
'$gte': 1000000
},
'follows': {
'$exists': True
}
}):
nodes[user['id']] = g.add_node(user['username'])
nodes[user['id']]['follows'] = user['counts']['follows']
nodes[user['id']]['followed_by'] = user['counts']['followed_by']
for user in db.users.find({
'counts.followed_by': {
'$gte': 1000000
},
'follows': {
'$exists': True
}
}):
for followed_user_id in user['follows']:
if followed_user_id in nodes: