WCA avg letter count by country

# Alexandre Campos, 02/03/2019
# [email protected]

# program made for the WCA Statistics, requested by Marius Rombout van Riemsdijk

# place this program.py alongside with the tsv (unziped) export from
# https://www.worldcubeassociation.org/results/misc/WCA_export.tsv.zip
# and run it with
# python program.py

import csv

def show_results(labels, results_list, list_of_lists, limit = float("inf"), direction = "asc", string_format = "%s"):
    """'Labels is' intuitive. Results_list is the main order list and list_of_lists are other lists that are gonna be ordered as well (eg.: country, name, etc). You can set limit if you wish. Direction = 'asc' if you wanna the lower results first; direction = 'des' if you wanna bigger results first."""

    pos = 1
    prev = None

    out = "Pos"
    for x in labels:
        out += "\t"+x
    print out

    temp = []
    if direction == "asc":
        temp = sorted(zip(results_list, range(len(results_list))))
    elif direction == "des":
        temp = sorted(zip(results_list, range(len(results_list))))[::-1]
    else:
        raise ValueError, "Invalid direction: %s"%direction

    for a, b in temp:
        out = ""
        if a != prev:
            if pos > limit:
                break
            out+="%s)\t"%pos
        else:
            out += "---\t"
        out += format(a, string_format)+"\t"
        for x in list_of_lists:
            out += str(x[b])+"\t"
        print out.strip()
        prev = a
        pos += 1

def just_letters(s):
    out = ""
    for x in s:
        if x.isalpha(): out += x
    return out

def avg(l):
    if len(l) == 0: return 0.
    return 1.0*sum(l)/len(l)

def main():
    country_list = []
    count = []

    header = True

    with open("WCA_export_Persons.tsv") as tsvfile:

        tsvreader = csv.reader(tsvfile, delimiter="\t")
        for line in tsvreader:

            if header:
                header = False
                continue

            name = line[2].decode("utf-8")
            if "(" in name:
                name = name[:name.index("(")]

            # remove spaces . and -
            name = just_letters(name)

            name_count = sum(map(len, name.split()))
            country = line[3]

            i = -1
            try:
                i = country_list.index(country)
            except:
                country_list.append(country)
                count.append([])
            count[i].append(name_count)

    avg_list = map(avg, count)
    no_of_competitors = map(len, count)
    labels = ["Avg", "Country", "# of competitors"]
    list_of_lists = [country_list, no_of_competitors]

    assert len(labels) == len(list_of_lists)+1

    limit = 1000
    show_results(labels, avg_list, list_of_lists, limit, "des", ".2f")

main()