Bug 1255472 - Sanity check Windows 10 version data

Bug 1255472 - Sanity check Windows 10 version data.ipynb

Bug 1255472 - Sanity check Windows 10 version data.py

# coding: utf-8

# ### Bug 1255472 - Sanity check Windows 10 version data

# In[76]:

import ujson as json
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import plotly.plotly as py
import operator
from distutils.version import StrictVersion

from moztelemetry import get_pings, get_pings_properties, get_one_ping_per_client, get_clients_history

get_ipython().magic(u'pylab inline')


# Let's fetch the main pings.

# In[66]:

submission_dates = ("20160419", "20160421")
build_ids = ("20160419000000", "20160499999999")
main_pings = get_pings(sc,
                       app="Firefox",
                       channel="nightly",
                       submission_date=submission_dates,
                       build_id=build_ids,
                       doc_type="main",
                       schema="v4",
                       fraction=1.0)


# ... and extract only the attributes we need from the Telemetry submissions:

# In[67]:

subset = get_pings_properties(main_pings, ["environment/system/os/name",
                                           "environment/system/os/version",
                                           "environment/system/os/windowsUBR",
                                           "environment/system/os/windowsBuildNumber",
                                          ])


# In[68]:

subset.count()


# Pick the pings coming from Windows.

# In[69]:

windows_pings = subset.filter(lambda p: p.get("environment/system/os/name") == "Windows_NT")
windows_pings.count()


# ### Check that pings from Windows ver < 10 have no UBR/build number fields

# Get the pings for Windows < 10.0

# In[70]:

windows_lessthan_10 = windows_pings.filter(lambda p: StrictVersion(p.get("environment/system/os/version")) < StrictVersion("10.0"))
windows_lessthan_10.count()


# For these pings, we don't expect to see neither windowsUBR or windowsBuildNumber. 

# In[71]:

windows_lessthan_10.filter(lambda p: (p.get("environment/system/os/windowsUBR", None) != None) or                                      (p.get("environment/system/os/windowsBuildNumber", None) != None)).count()


# ### Check that pings from Windows ver >= 10 have sane UBR/build number fields

# In[72]:

windows_gethan_10 = windows_pings.filter(lambda p: StrictVersion(p.get("environment/system/os/version")) >= StrictVersion("10.0"))
windows_gethan_10.count()


# If we're on Windows ver >= 10.0, we expect to see both the UBR and the windows build number.

# In[73]:

def validate_ubr_buildnumber(p):
    windows_ubr = p.get("environment/system/os/windowsUBR", None)
    windows_build_number = p.get("environment/system/os/windowsBuildNumber", None)
    
    if windows_ubr is None:
        return ("No UBR", 1)
    
    if type(windows_ubr) != int:
        return ("Unexpecteed UBR " + type(windows_ubr).__name__, 1)
    
    if windows_build_number is None:
        return ("No Build Number", 1)

    if type(windows_build_number) != int:
        return ("Unexpecteed Build Number " + type(windows_build_number).__name__, 1)
    
    return ("", 1)

windows_gethan_10.map(validate_ubr_buildnumber).countByKey()


# ### Check the distribution of the UBR/build number values

# In[82]:

build_values = windows_gethan_10.map(lambda p: (str(p.get("environment/system/os/windowsBuildNumber", "")) +                                                 ", " + str(p.get("environment/system/os/windowsUBR", "")), 1))
sorted(build_values.countByKey().items(), key=operator.itemgetter(1), reverse=True)