Question

I’m trying to pick up Python. As part of the learning process I’m porting a project I wrote in Java to Python. I’m at a section now where I have a list of CSV headers of the form:

headers = [a, b, c, d, e, .....]

and separate lists of groups that these headers should be broken up into, e.g.:

headers_for_list_a = [b, c, e, ...]
headers_for_list_b = [a, d, k, ...]
. . .

I want to take the CSV data and turn it into dict’s based on these groups, e.g.:

list_a = [
          {b:val_1b, c:val_1c, e:val_1e, ... },
          {b:val_2b, c:val_2c, e:val_2e, ... },
          {b:val_3b, c:val_3c, e:val_3e, ... },
          . . . 
         ]

where for example, val_1b is the first row of the ‘b’ column, val_3c is the third row of the ‘c’ column, etc.

My first “Java instinct” is to do something like:

for row in data:
    for col_num, val in enumerate(row):
        col_name = headers[col_num]
        if col_name in group_a:
            dict_a[col_name] = val
        elif headers[col_cum] in group_b:
            dict_b[col_name] = val
        ...
    list_a.append(dict_a)
    list_b.append(dict_b)
    ...     

However, this method seems inefficient/unwieldy and doesn’t posses the elegance that Python programmers are constantly talking about. Is there a more “Zen-like” way I should try- keeping with the philosophy of Python?

Answer 1

csv.DictReader

import csv

groups = dict(a=headers_for_list_a, b=headers_for_list_b)
lists = dict((name, []) for name in groups)

for row in csv.DictReader(csvfile, fieldnames=headers):
    for name, grp_headers in groups.items():
        lists[name].append(dict((header, row[header]) for header in grp_headers))