summarize

#!/usr/bin/env python

# summarize file.csv

import os.path
import csv
import sys
import statistics

def read_file(fname, data):
    print("Loading data from %s" % fname)
    first = True
    with open(fname, 'r') as f:
        reader = csv.reader(f)
        for row in reader:
            if first:
                i = None
                for i_ in range(len(row)):
                    if row[i_] == '':
                        i = i_ 
                first = False
            val = row[i-1]
            ndx = tuple(row[1:i-1])
            date = row[0]
            if not ndx in data:
                data[ndx] = []
            if val != "":
                data[ndx].append( float(val) )
    return date

def compute_stats(fname, data, date, extend=None):
    #
    # Check if the file already has data with the same timestamp
    #
    if extend is True and os.path.exists(fname):
        with open(fname, 'r') as f:
            reader = csv.reader(f)
            for row in reader:
                if row[0] == date:
                    print("Date '%s' already exists in %s" % (date, fname))
                    return
    #
    # Compute stats
    #
    if extend is True:
        print("Appending data to %s" % fname)
        mode = 'a'
    else:
        print("Writing data to %s" % fname)
        mode = 'w'
    with open(fname, mode) as f:
        for ndx in sorted(data.keys()):
            values = data[ndx]
            if len(values) > 0:
                f.write( ",".join( map(str, [date] + list(ndx) + [min(values), statistics.mean(values), max(values), 0 if len(values) == 1 else statistics.stdev(values)]) ) )
                f.write("\n")


print("")
data = {}
fname = 'benchmarks/%s_releases.csv' % sys.argv[1]
if os.path.exists(fname):
    date = read_file(fname, data)
    fname = 'benchmarks/%s_releases_stats.csv' % sys.argv[1]
    compute_stats(fname, data, date, extend=False)

print("")
data = {}
fname = 'benchmarks/%s_branches.csv' % sys.argv[1]
if os.path.exists(fname):
    date = read_file(fname, data)
    fname = 'benchmarks/%s_branches_stats_history.csv' % sys.argv[1]
    compute_stats(fname, data, date, extend=True)
    fname = 'benchmarks/%s_branches_stats.csv' % sys.argv[1]
    compute_stats(fname, data, date, extend=False)