Generates charts about the split test data collected by the script collectsplitdata.py

gensplitcharts.py

import json
import pygal

# Parsing the json file
content = json.load(open('split.json'))

# App app names
app_names = list({x['what'] for x in content['app']})

# All file names
file_names = list({x['what'] for x in content['files']})

# Returns how many times a certain file appears in the file or in the
# app list
get_num_changes = \
    lambda k, f: len([x for x in content[k] if x['what'] == f])


def is_invalid(label):
    return ((label is None) or
            ('split' in label) or
            ('migrations' in label) or
            ('util/switches.py' in label))


# Time to gen some nice charts
def changes(kind, labels, opts=None):

    # Reading parameters
    opts = opts or {}
    limit = opts.get('limit', -1)
    sortfunc = opts.get('sortfunc')
    reverse = opts.get('reverse', False)

    # Gathering all data, cause we need to sort it
    data = [(item, get_num_changes(kind, item)) for item in labels
            if not is_invalid(item)]
    data.sort(key=sortfunc, reverse=reverse)

    # Limitting data according to the caller's request
    if limit > 0:
        data = data[:limit]

    # Translating the data to the format expected by matplotlib
    chart = pygal.HorizontalBar(truncate_legend=35)
    for label, num_changes in data:
        if kind == 'files':
            label = '/'.join(label.split('/')[-2:])

        chart.add(label, float(num_changes))

    chart.render_to_file('split-per-{}.svg'.format(kind))


def files_per_change():

    # Preparing the data
    commits = {}
    for file_ in content['files']:
        commits[file_['hash']] = commits.get(file_['hash'], -1) + 1
    commits = sorted(commits.items(), key=lambda x: x[1], reverse=True)
    commits = commits[:10]

    # Creating the chart
    chart = pygal.HorizontalBar()
    for commit, num_files in commits:
        chart.add(commit, num_files)
    chart.render_to_file('split-per-changes.svg')

if __name__ == '__main__':

    changes('app', app_names, {
        'sortfunc': lambda x: x[1],
        'limit': 5,
        'reverse': True,
    })

    changes('files', file_names, {
        'sortfunc': lambda x: x[1],
        'limit': 10,
        'reverse': True,
    })

    files_per_change()