[FIXED] Sort dict by key when key is a version

Issue

I'm trying to print out a dictionary (actually a defaultdict) where the keys are version numbers (in the form 6.0 or 6.1.2) and the values are floats.

# Parse file contents
firmware_percents = defaultdict(float)
with open('test.csv', 'r') as file:
    reader = csv.DictReader(file)

    for row in reader:
        # Don't care about iphone vs ipad
        # Must be a string - StrictVersion is apparently unhashable
        rev = row["Firmware Version"].split(" ")[2]

        # Dump extra spaces, the less than (assume .1%), and % sign
        percent = float(row["% of Sessions"].strip().lstrip("<").strip("%"))
        firmware_percents[rev] += percent

def pretty_print(d):
    for k in sorted(d, key=d.get, reverse=True):
        print("{0}: {1:.1f}".format(k, d[k]))

print("All versions:")
pretty_print(firmware_percents)

However, when I do this some versions are printing out of order:

All versions:
7.0: 44.2
7.0.2: 25.7
6.1.3: 14.2
6.1.4: 5.5
7.0.1: 3.2
6.1: 2.7
# You get the point

Using this input file:

Firmware Version,Sessions,% of Sessions
" iPhone 5.0.1 ",20," <0.1% "
" iPhone 6.0 ",26," 0.1% "
" iPhone 5.1.1 ",69," 0.3% "
" iPhone 5.1 ",2," <0.1% "
" iPhone 7.0 ",7401," 31.5% "
" iPhone 6.1 ",337," 1.4% "
" iPhone 6.1.3 ",2193," 9.3% "
" iPhone 6.1.2 ",84," 0.4% "
" iPhone 7.0.1 ",747," 3.2% "
" iPhone 7.0.2 ",4619," 19.7% "
" iPhone 6.0.1 ",37," 0.2% "
" iPhone 6.0.2 ",1," <0.1% "
" iPhone 6.1.4 ",1281," 5.5% "
" iPad 5.0 ",4," <0.1% "
" iPad 5.1 ",100," 0.4% "
" iPad 5.1.1 ",545," 2.3% "
" iPad 6.0 ",16," <0.1% "
" iPad 6.1 ",305," 1.3% "
" iPhone 7.0.3 ",1," <0.1% "
" iPhone 6.1.1 ",1," <0.1% "
" iPad 7.0 ",2979," 12.7% "
" iPad 6.0.1 ",100," 0.4% "
" iPad 6.1.3 ",1139," 4.9% "
" iPad 6.0.2 ",5," <0.1% "
" iPad 6.1.2 ",65," 0.3% "
" iPad 7.0.2 ",1404," 6.0% "

I've tried pprint, and while it does sort in the correct order it doesn't format the floats (so I end up with numbers like 14.5999999996). It also occasionally prints strangeness like defaultdict(<class 'float'>, {'5': 3.3, '6': 24.2, '7': 73.2}) when I try to do major version only.

How can I ensure these print in version order with formatted percentages?

By order I mean it sorts by major then minor then build/super-minor (7.0.2 > 7.0.1 > 6.1.4 > 6.1 etc etc)?

Solution

I think @dornhege spotted the main issue first: key=d.get means that you're sorting by the values, not the keys.

I'd leverage the stdlib package distutils to get the version stuff handled automatically:

>>> sorted(firmware_percents, key=distutils.version.StrictVersion, reverse=True)
['7.0.3', '7.0.2', '7.0.1', '7.0', '6.1.4', '6.1.3', '6.1.2', '6.1.1', '6.1', '6.0.2', '6.0.1', '6.0', '5.1.1', '5.1', '5.0.1', '5.0']
>>> sorted(firmware_percents, key=distutils.version.LooseVersion, reverse=True)
['7.0.3', '7.0.2', '7.0.1', '7.0', '6.1.4', '6.1.3', '6.1.2', '6.1.1', '6.1', '6.0.2', '6.0.1', '6.0', '5.1.1', '5.1', '5.0.1', '5.0']

This has advantages over splitting by . because it'll be smarter about handling things like 7.0.2rc1.

Answered By - DSM