def estimate_os(container=None,image_package=None,sudopw=None,return_top=True):
'''estimate os will compare a package to singularity python's database of
operating system images, and return the docker image most similar
:param return_top: return only the most similar (estimated os) default True
:param image_package: the package created from the image to estimate.
FIGURE OUT WHAT DATA WE NEED
'''
if image_package == None:
SINGULARITY_HUB = os.environ.get('SINGULARITY_HUB',"False")
# Visualization deployed local or elsewhere
if SINGULARITY_HUB == "False":
image_package = make_package(container,remove_image=True,sudopw=sudopw)
comparison = compare_packages(packages_set1=[image_package])['files.txt'].transpose()
else:
comparison = container_similarity_vector(container1=container)['files.txt'].transpose()
else:
comparison = compare_packages(packages_set1=[image_package])['files.txt'].transpose()
comparison.columns = ['SCORE']
most_similar = comparison['SCORE'].idxmax()
print("Most similar OS found to be ", most_similar)
if return_top == True:
return most_similar
return comparison
def test_compare_packages(self):
print("Testing singularity.analysis.compare.compare_packages")
from singularity.analysis.compare import compare_packages
pwd = get_installdir()
pkg1 = "%s/tests/data/busybox-2016-02-16.img.zip" %(pwd)
pkg2 = "%s/tests/data/cirros-2016-01-04.img.zip" %(pwd)
comparison = compare_packages(pkg1,pkg2)
self.assertTrue('files.txt' in comparison)
self.assertTrue(isinstance(comparison['files.txt'],pandas.DataFrame))
def make_package_tree(matrix=None,
labels=None,
width=25,
height=10,
title=None,
font_size=None):
'''make package tree will make a dendrogram comparing a matrix of packages
:param matrix: a pandas df of packages, with names in index and columns
:param labels: a list of labels corresponding to row names, will be
pulled from rows if not defined
:param title: a title for the plot, if not defined, will be left out.
:returns a plot that can be saved with savefig
'''
from matplotlib import pyplot as plt
from scipy.cluster.hierarchy import (dendrogram, linkage)
if font_size is None:
font_size = 8.
from scipy.cluster.hierarchy import cophenet
from scipy.spatial.distance import pdist
if not isinstance(matrix, pandas.DataFrame):
bot.info(
"No pandas DataFrame (matrix) of similarities defined, will use default."
)
matrix = compare_packages()['files.txt']
title = 'Docker Library Similarity to Base OS'
Z = linkage(matrix, 'ward')
c, coph_dists = cophenet(Z, pdist(matrix))
if labels == None:
labels = matrix.index.tolist()
plt.figure(figsize=(width, height))
if title != None:
plt.title(title)
plt.xlabel('image index')
plt.ylabel('distance')
dendrogram(
Z,
leaf_rotation=90., # rotates the x axis labels
leaf_font_size=font_size, # font size for the x axis labels
labels=labels)
return plt
package_directory = '%s/examples/package_image/packages' % (base)
package_set1 = glob("%s/*.zip" % (package_directory))
# Option 3: provide no input args, and default (os) for package_set1 is used
###############################################################################
# Choose another set of packages to compare to
###############################################################################
# Option 1: specify another (same or different) family of packages
package_set2 = get_packages(family='docker-os')
# Option 2: Same as above
# Option 3: Don't specify any packages, use defaults
###############################################################################
# Run the analysis
###############################################################################
# Use your own input arguments...
comparisons = compare_packages(packages_set1=package_set1,
packages_set2=package_set2,
by="folders.txt")
# Or use defaults
comparisons = compare_packages() # docker-library vs. docker-os,
# by files.txt
# Save to file
pickle.dump(result, open('comparisons.pkl', 'wb'))
# 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')