def test_get_packages(self):
print("Testing singularity.analysis.utils.get_packages")
from singularity.package import get_packages
print("Case 1: Default returns Docker operating systems")
packages = get_packages()
self.assertEqual(len(packages),46)
print("Case 2: Family specified to Docker library")
packages = get_packages(family="docker-library")
self.assertEqual(len(packages),117)
def compare_packages(packages_set1=None, packages_set2=None, by=None):
'''compare_packages will compare one image or package to one image or package. If
the folder isn't specified, the default singularity packages (included with install)
will be used (os vs. docker library). Images will take preference over packages
:param packages_set1: a list of package files not defined uses docker-library
:param packages_set2: a list of package files, not defined uses docker-os
:by: metrics to compare by (files.txt and or folders.txt)
'''
if packages_set1 == None:
packages_set1 = get_packages('docker-library')
if packages_set2 == None:
packages_set2 = get_packages('docker-os')
if by == None:
by = ['files.txt']
if not isinstance(by, list):
by = [by]
if not isinstance(packages_set1, list):
packages_set1 = [packages_set1]
if not isinstance(packages_set2, list):
packages_set2 = [packages_set2]
comparisons = dict()
for b in by:
bot.debug("Starting comparisons for %s" % b)
df = pandas.DataFrame(index=packages_set1, columns=packages_set2)
for package1 in packages_set1:
for package2 in packages_set2:
if package1 != package2:
sim = calculate_similarity(image_package1=package1,
image_package2=package2,
by=b)[b]
else:
sim = 1.0
name1 = os.path.basename(package1).replace('.img.zip', '')
name2 = os.path.basename(package2).replace('.img.zip', '')
bot.debug("%s vs. %s: %s" % (name1, name2, sim))
df.loc[package1, package2] = sim
df.index = [
os.path.basename(x).replace('.img.zip', '')
for x in df.index.tolist()
]
df.columns = [
os.path.basename(x).replace('.img.zip', '')
for x in df.columns.tolist()
]
comparisons[b] = df
return comparisons
def test_container_similarity(self):
print(
"Testing singularity.analysis.compare.container_similarity_vector")
import pandas
from singularity.analysis.compare import container_similarity_vector
from singularity.package import get_packages
packages_set = get_packages('docker-os')[0:2]
vector = container_similarity_vector(container1=self.container,
custom_set=packages_set)
self.assertTrue('files.txt' in vector)
self.assertTrue(isinstance(vector['files.txt'], pandas.DataFrame))
print(
"Testing singularity.analysis.compare.compare_singularity_images")
from singularity.analysis.compare import compare_singularity_images
sim = compare_singularity_images(self.container, self.comparator)
self.assertTrue(isinstance(sim, pandas.DataFrame))
self.assertTrue(sim.loc[self.container, self.comparator] -
0.4803262269280298 < 0.01)
print("Testitng singularity.analysis.compare.compare_containers")
from singularity.analysis.compare import compare_containers
comparison = compare_containers(self.container, self.comparator)
self.assertTrue('files.txt' in comparison)
for key in ['total1', 'total2', 'intersect', 'unique2', 'unique1']:
self.assertTrue(key in comparison['files.txt'])
print("Testing singularity.analysis.compare.calculate_similarity")
from singularity.analysis.compare import calculate_similarity
sim = calculate_similarity(self.container, self.comparator)
self.assertTrue(sim['files.txt'] - 0.4921837537163134 < 0.01)
def container_similarity_vector(container1=None,
packages_set=None,
by=None,
custom_set=None):
'''container similarity_vector is similar to compare_packages, but intended
to compare a container object (singularity image or singularity hub container)
to a list of packages. If packages_set is not provided, the default used is
'docker-os'. This can be changed to 'docker-library', or if the user wants a custom
list, should define custom_set.
:param container1: singularity image or singularity hub container.
:param packages_set: a name of a package set, provided are docker-os and docker-library
:param custom_set: a list of package files, used first if provided.
:by: metrics to compare by (files.txt and or folders.txt)
'''
if custom_set == None:
if packages_set == None:
packages_set = get_packages('docker-os')
else:
packages_set = custom_set
if by == None:
by = ['files.txt']
if not isinstance(by, list):
by = [by]
if not isinstance(packages_set, list):
packages_set = [packages_set]
comparisons = dict()
for b in by:
bot.debug("Starting comparisons for %s" % b)
df = pandas.DataFrame(columns=packages_set)
for package2 in packages_set:
sim = calculate_similarity(container1=container1,
image_package2=package2,
by=b)[b]
name1 = os.path.basename(package2).replace('.img.zip', '')
bot.debug("container vs. %s: %s" % (name1, sim))
df.loc["container", package2] = sim
df.columns = [
os.path.basename(x).replace('.img.zip', '')
for x in df.columns.tolist()
]
comparisons[b] = df
return comparisons
# A quick example of making a package tree with data derived from calculate_similarity.py
# Set environment variable MESSAGELEVEL to CRITICAL to disable debug output
from singularity.views import make_package_tree
from singularity.analysis.compare import compare_packages
from singularity.package import get_packages
import pickle
# Compare your own data
data = pickle.load(open('comparisons.pkl','rb'))['files.txt']
plt = make_package_tree(matrix=data)
# Compare docker-os to docker-os
package_set1 = get_packages(family='docker-os')
package_set2 = get_packages(family='docker-os')
data = compare_packages(packages_set1=package_set1,
packages_set2=package_set2)['files.txt']
plt = make_package_tree(matrix=data)
# Show the plot
plt.show()
# or save to file
plt.savefig('examples/package_tree/docker-os.png')