sort-txn-count.py

import os
import csv
import glob


# Script to parse csv, count txn of each address and sort and save to file.
# Place a csv file next to this program


# Config

# Use script file's directory
pyfiledir = os.path.dirname(os.path.realpath(__file__))

outputPath = pyfiledir + os.sep + 'output.csv'

targetCol = 'To'

# First n number of rows of a file, 0 for no limit
lines_limit = 100


# The input
files = glob.glob(pyfiledir+'/*.csv')


results = {}

print('\nProcess is starting...', '\n')

line_count = 0

for filepath in files:
    if(filepath == outputPath):
        continue  # ignore output
    current_file_line_count = 0
    with open(filepath, mode='r') as csv_file:
        csv_reader = csv.DictReader(csv_file)
        for row in csv_reader:  # looping through lines
            if(targetCol not in row):
                continue  # only the data lines
            if(lines_limit != 0 and current_file_line_count >= lines_limit):
                break
            current_file_line_count += 1
            line_count += 1
            key = row[targetCol]
            if(key not in results):
                results[key] = [filepath]
            else:
                results[key].append(filepath)


print(f'Total number of lines: {line_count}', '\n')

resultAsList = []

for key in results:
    resultAsList.append({'Address': key, 'Count': len(results[key])})


def sortFunc(e):
    return e['Count']


resultAsList.sort(reverse=True, key=sortFunc)


# Write output
with open(outputPath, mode='w', newline='') as csv_file:
    fieldnames = ['Address', 'Count']
    writer = csv.DictWriter(csv_file, fieldnames=fieldnames)

    writer.writeheader()

    for row in resultAsList:
        writer.writerow(row)


print('File written: ' + outputPath, '\n')
print('Completed!', '\n')