Profiling memory leaks

Out:

-------------- Performance analysis -------------
 Name   | Executions   | Mean time   | Total time
--------+--------------+-------------+--------------
-------------------------------------------------


------------------------------- Memory analysis --------------------------------
 Name                    |   Executions | Mean size change   | Total size change
-------------------------+--------------+--------------------+---------------------
 nothing                 |            1 | 6.25391e+00 kB     | 6.25391e+00 kB
 one list                |            1 | 6.74531e+01 kB     | 6.74531e+01 kB
 one list - and clear    |            1 | 4.13281e+00 kB     | 4.13281e+00 kB
 dynamic string creation |         1000 | 2.01815e+00 kB     | 2.01815e+03 kB
--------------------------------------------------------------------------------

import matplotlib.pyplot as plt

from sorts.profiling import Profiler

p = Profiler(track_memory=True)

#As the profiler data is also stored in Python tracked memory
# a diff of "nothing" will still result in more allocation of memory
# including initialization for structures and the like
p.snapshot('nothing')
p.memory_diff('nothing')

#this allocates memory
p.snapshot('one list')
lst = list(range(2000))
p.memory_diff('one list')

#and if we take care to delete the variable
#only profiling allocations are left over
del lst
p.memory_diff('one list', save='one list - and clear')

#this iteration changes allocation each iteration and does not clean up
lsts = []
for i in range(1000):
    p.snapshot('dynamic string creation')
    lsts.append('test'*i)
    p.memory_diff('dynamic string creation')


print(p)

#so it might be a good idea to plot the trend
fig, ax = plt.subplots(1,1)
ax.plot(p.memory_stats['dynamic string creation'])
ax.set_title('Dynamic string creation in loop')
ax.set_xlabel('Iteration')
ax.set_ylabel('Allocation [kB]')
plt.show()