CODEWITHSUNDEEP
pythonfilecsv
James Cooper
James Cooper

Reputation: 57

Python - sort csv data from multiple files based on columns

I have a folder with multiple files, each with a varying number of columns in each file. I want to go through the directory, open each file and loop through each line, writing the line to a new CSV file based on the number of columns in that line. I want to end up with a single big CSV for all lines with 14 columns, another big CSV for all lines with 18 columns, and the last CSV with all the other columns.

Here's what I have so far.

import pandas as pd
import glob
import os
import csv


path = r'C:\Users\Vladimir\Documents\projects\ETLassig\W3SVC2'
all_files = glob.glob(os.path.join(path, "*.log")) 

for file in all_files:
    for line in file:
        if len(line.split()) == 14:
            with open('c14.csv', 'wb') as csvfile:
                csvwriter = csv.writer(csvfile, delimiter=' ')
                csvwriter.writerow([line])
        elif len(line.split()) == 18:
            with open('c14.csv', 'wb') as csvfile:
                csvwriter = csv.writer(csvfile, delimiter=' ')
                csvwriter.writerow([line])          
            #open 18.csv
        else:
            with open('misc.csv', 'wb') as csvfile:
                csvwriter = csv.writer(csvfile, delimiter=' ')
                csvwriter.writerow([line])
print(c14.csv)

Can anyone offer any feedback on how to approach this?

Upvotes: 0

Views: 348

Answers (2)

gboffi
gboffi

Reputation: 25023

Beforehand, please note that you can copy the lines as is from the input files to the output ones, no need for the CSV machinery.

That said, I propose to use a dictionary of file objects and the get method of dictionaries, that permits to specify a default value.

files = {14:open('14.csv', 'wb'),
         18:open('18.csv', 'wb')}
other = open('other.csv', 'wb')

for file in all_files:
    for line in open(file):
        llen = len(line.split())
        target = files.get(llen, other)
        target.write(line)

If you have to process some million records then note that, because

In [20]: a = 'a '*20                                                                      

In [21]: %timeit len(a.split())                                                           
599 ns ± 1.59 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)

In [22]: %timeit a.count(' ')+1                                                           
328 ns ± 1.28 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)
 

you should substitute the for loops above with

for file in all_files:
    for line in open(file):
        fields_count = line.count(' ')+1
        target = files.get(fields_count, other)
        target.write(line)

Should because, even if we speak of nano seconds , the file system access is in the same ballpark

In [23]: f = open('dele000', 'w')                                                         

In [24]: %timeit f.write(a)                                                               
508 ns ± 154 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)

as splitting/counting.

Upvotes: 0

Mehrdad Pedramfar
Mehrdad Pedramfar

Reputation: 11073

You can add all of your columns as as a list in list:

l = []
for file in [your_files]:
    with open(file, 'r') as f:
        for line in f.readlines()
            l.appned(line.split(" "))

Now you have list of lists, so just sort them with length of sublists then put it in a new file:

l.sort(key=len)

with open(outputfile, 'w'):
     # Write  lines here as you want

Upvotes: 5

Related Questions