energy

energy.py

def test_energy(countries):
    """
    Input: a series/ the Country column in Energy
    utf-8 encoded i.e. when reading Energy use
    encoding='utf-8'
    """
    encodedC = '11,7,7,14,7,6,8,19,9,7,5,9,7,10,7,7,10,8,7,7,6,5,7,6,7,32,22,8,6,22,17,8,12,7,10,8,8,6,14,24,4,5,5,9,42,8,7,5,12,10,13,7,4,7,6,14,37,32,7,8,8,18,7,5,11,17,7,7,8,14,16,4,7,6,13,16,5,6,7,7,5,9,6,9,7,10,4,9,8,6,13,6,5,8,7,7,5,9,4,4,7,11,6,5,7,5,6,6,10,5,8,6,10,32,6,7,7,7,5,13,9,10,10,6,8,8,4,5,16,10,10,9,6,10,8,10,10,7,10,7,7,5,5,11,13,11,9,5,7,4,24,6,4,8,5,6,16,8,4,11,6,8,11,5,11,19,7,7,18,6,12,21,11,25,32,5,21,12,7,6,10,12,9,12,8,8,15,7,12,11,5,9,18,5,8,9,6,11,20,10,8,41,11,4,5,19,7,6,12,24,6,6,7,20,14,27,13,28,7,10,7,9,8,25,5,6,8'
    outcome = ['Failed\n', 'Passed\n']
    
    
    
    energy = pd.DataFrame()
    energy['original'] = pd.read_excel('Energy Indicators.xls',
                                   usecols=[1],encoding='utf-8',
                                   index_col=0).loc['Afghanistan':'Zimbabwe'].index.tolist()
    energy['tested'] = countries.str.len()
    energy['actual'] = encodedC.split(',')
    energy['actual'] = energy['actual'].astype(int)
    try:
        energy['Country'] = countries
    except Exception as e:
        print('Failed, error: ',e)
    
    res = 'Test number of records: '
    res += outcome[len(countries)==len(energy)]
    
    res += 'Test the column name: '
    res += outcome [countries.name == 'Country']
    
    res += 'Equality Test: '
    res += outcome[energy['tested'].equals(energy['actual'])]
    
    if not energy['tested'].equals(energy['actual']):
        res += '\nMismatched countries:\n'
        mismatch = energy.loc[energy['tested'] != (energy['actual']), [
            'original', 'Country', 'tested', 'actual']].values.tolist()
        res += '\n'.join('"{:}" miss-cleaned as  "{:}"'.format(o, r)
                         for o, r, s, v in mismatch)
    return res
print(test_energy(get_energy().loc[:,'Country']))