Das deutsche Python-Forum

Code: Alles auswählen

with open(eingabename, encoding="utf-8") as grunddaten, open(ausgabename, mode="w", encoding="utf-8") as ausschnitt:
    for zeile in grunddaten:
        alle_felder = zeile.split(";")
        manche_felder = [alle_felder[1], alle_felder[2], alle_felder[4]]
        ausschnitt.write(";".join(manche_felder) + "\n")

Code: Alles auswählen

import csv
import time
import pandas as pd
import numpy as np

start_time = time.perf_counter()

with open("convertcsv.csv", newline="") as read_file, open("new-csv.csv", "w", newline="") as write_file:
    reader = csv.DictReader(read_file, delimiter=";")
    writer = csv.writer(write_file, delimiter=";")
    writer.writerow(["first", "word", "age"])
    for row in reader:
        writer.writerow([row["first"], row["word"], row["age"]])
        
print(f"csv\t{round(time.perf_counter() - start_time, 3)} s")

start_time = time.perf_counter()

df = pd.read_csv("convertcsv.csv", delimiter=";", header=0, usecols=["first", "word", "age"])
df.to_csv("new-pd.csv", index=False, sep=";", header=True)

print(f"pandas\t{round(time.perf_counter() - start_time, 3)} s")

start_time = time.perf_counter()

array = np.loadtxt("convertcsv.csv", dtype=np.object, delimiter=";", usecols=[0, 2, 4])
np.savetxt("new-np.csv", array, delimiter=";", header="first;word;age", encoding="utf-8", fmt="%s")

print(f"numpy\t{round(time.perf_counter() - start_time, 3)} s")

>>>csv     16.759 s
>>>pandas  5.246 s
>>>numpy   12.669 s