Overwriting a specific cell in a CSV file

def update_csv_cell(address, new_value): filepath = '/home/idle/so/some.csv' dialect_params = dict(delimiter=';') col_num, row_num = address with open(filepath, 'r+b') as csvfile: reader = csv.reader(csvfile, **dialect_params) lines = [] for current_line in reader: if reader.line_num == row_num: current_line[col_num-1] = new_value lines.append(current_line) csvfile.seek(0) csv.writer(csvfile, **dialect_params).writerows(lines) csvfile.truncate() update_csv_cell(address=(5, 6), new_value='replaced') 

The above is the base code, without error handling.

Bad news more than good.

Good: the csv module allows you to work with ordinary file objects, which means you have access to all methods of such objects (see seek, tell, etc.)

Bad 1: the csv module reads the entire file, so during the iteration over the file lines, the tell method will always return the end position of the file, so there is no other simple solution than to read all the data, change it and save it on the horizon.

Bad 2: The approach to accessing the file data by position in the document is unreliable, because if everything can be relatively stable with columns, then it is very possible not to guess with a number, thus spoiling the data.

When working with CSV (and indeed when working with data) it is very convenient and almost always more productive to use Pandas :

Example:

read / parse CSV in Pandas Data Frame with the read_csv () method:

 import pandas as pd In [115]: df = pd.read_csv('c:/temp/example.csv', sep=',') In [116]: df Out[116]: ABCDE 0 54 9 74 73 59 1 34 20 66 81 14 2 89 33 87 98 41 3 74 78 19 4 25 4 30 12 7 52 29 5 33 45 39 73 81 6 49 28 48 69 99 

change the 6th row and 5th column (note: in Pandas, as in Python, the numbering starts from zero):

 In [119]: df.ix[6-1, 5-1] += 1000 In [120]: df Out[120]: ABCDE 0 54 9 74 73 59 1 34 20 66 81 14 2 89 33 87 98 41 3 74 78 19 4 25 4 30 12 7 52 29 5 33 45 39 73 1081 6 49 28 48 69 99 

write back to CSV:

 df.to_csv('c:/temp/example.csv', index=False) 

Result:

 A,B,C,D,E 54,9,74,73,59 34,20,66,81,14 89,33,87,98,41 74,78,19,4,25 30,12,7,52,29 33,45,39,73,1081 49,28,48,69,99 

How to quickly and easily generate test data (this is how test CSV was created):

 import numpy as np df = pd.DataFrame(np.random.randint(0,100,size=(7, 5)), columns=list('ABCDE')) df.to_csv('c:/temp/example.csv', index=False) 

or in one line:

 pd.DataFrame(np.random.randint(0,100,size=(7, 5)), columns=list('ABCDE')).to_csv('c:/temp/example.csv', index=False) 

• read file to list
• change item in list
• write list to file.

Try without a module. If you can:

• Read from file;
• Break a string with a separator;
• Write to file

then you can solve your problem without using CSV.

• with open (.... as f
• str.split (here is your delimiter)
• f.write (...