Contents

1. Understanding CSVs
   1. Reading a CSV
   2. Writing a CSV
   3. DictReader/Writer
2. Too slow?
3. Exercises

Understanding CSVs

A comma-separated values (CSV) file is a convenient way to store data that can be easily read and written. Despite the name, many CSVs files aren’t even comma-delimited (tab is common, but people rarely call them TSV) and there is no standard way of doing it.

You might think you can trivially read and write CSVs using .split(',') to read them and ','.join(fields) to write them. However, this is risky; if some of the fields contain commas themselves, those fields will be enclosed by quotes and you’ll end up misparsing.

Reading a CSV

The simplest way to read a CSV is the csv.reader function. This will return a reader object that allows you to read each row as a list. An example adapted from the documentation:

>>> import csv
>>> with open('eggs.csv', 'U') as csvfile:
...     spamreader = csv.reader(csvfile)
...     for row in spamreader:
...         print(', '.join(row))
Spam, Spam, Spam, Spam, Spam, Baked Beans
Spam, Lovely Spam, Wonderful Spam

This demonstrates another useful Python construct, using with to mark of a block where you will use a file, which will automatically close it when you are done with it.

Writing a CSV

Similarly, csv.writer allows you to write a CSV by writing one row at a time. Note that when you write a CSV file, you want to set the mode to 'wb'.

DictReader/Writer

What if you don’t want to rely on hard-coding the order of the fields in each line? You can use the DictReader and DictWriter classes to help. These allow you to read and write each row as a dictionary, with keys being the field names and values being the value of that field in each row.

Too slow?

The standard Python CSV parser is designed to handle a lot of strange input well, including Excel files. If you care more about speed than broad features, take a look at read_csv in pandas.

Exercises

Here are some exercises to get you used to working with the CSV module. Try these out using a sample CSV file.

1. Write a function that reads a CSV and uses a defaultdict to store a list of the values for each item as it reads in the CSV. Use csv.reader and csv.writer.
2. Write a second function that takes the dictionary produced above and then computes the minimum, maximum, and mean values for each item. (You’ll need to write your own function to compute the mean.) Write these values to another CSV with four fields: item, min, mean, max.
3. When you’ve got everything working, replace the reader and writer with csv.DictReader and csv.DictWriter.