Important classes and functions of the CSV module: open(), reader(), writer(), writerows(), DictReader(), DictWriter()

📖 Theory

The comma that breaks split()

Armed with last lesson's open(), you parse marks.csv yourself: line.split(','). Works beautifully... until row 40:

104,"Patel, Riya",78

The name contains a comma. Your split() produces four pieces instead of three, and the score column now holds " Riya".

Quoted fields, embedded commas, stray newlines: the CSV format has edge cases, and hand-rolled parsers die on all of them. Python's csv module exists so you never write split(',') again.

📖 Theory

A trained postman for tabular mail

Reading CSV by split() is grabbing envelopes and tearing them at every comma, even the commas inside the letter.

The csv module is a trained postman: csv.reader opens each envelope properly (quotes respected) and hands you the contents as a neat list. csv.writer is the same postman sealing your lists back into valid envelopes. The Dict pair goes further: envelopes labelled by column name instead of position.

📖 Theory

reader and writer, formally

Both wrap an already-open file:

Reading: csv.reader(f) yields each row as a list of strings: ['101', 'DBMS', '78']. Note: every value is a string, even the numbers: convert with int()/float() before maths. The header line arrives as an ordinary first row: skip it with next(reader).

Writing: csv.writer(f) offers writerow(one_list) and writerows(list_of_lists).

One format-specific rule: open files for csv with `newline=''`, or Windows users get a blank line after every row.

⌨️ Practical

The four workers on marks.csv

import csv

# READER: rows as lists (strings!)
with open('marks.csv', 'r', newline='') as f:
    reader = csv.reader(f)
    header = next(reader)            # skip the header row
    for row in reader:
        print(row[0], int(row[2]))   # roll, score as a NUMBER

# WRITER: lists back to disk
with open('toppers.csv', 'w', newline='') as f:
    writer = csv.writer(f)
    writer.writerow(['roll', 'subject', 'score'])          # one row
    writer.writerows([[101, 'DBMS', 88], [103, 'DBMS', 91]])  # many

# DICTREADER: rows as dictionaries, keys from the header
with open('marks.csv', newline='') as f:
    for row in csv.DictReader(f):
        print(row['roll'], row['score'])   # by NAME, not position

# DICTWRITER: dictionaries to disk (fieldnames required)
with open('report.csv', 'w', newline='') as f:
    w = csv.DictWriter(f, fieldnames=['roll', 'score'])
    w.writeheader()                        # do not forget this line
    w.writerows([{'roll': 101, 'score': 88}])

This example runs in Gri-Learn on the web, where you can edit it and see the output.

📊 At a glance

The csv module at a glance

💭 Think first

Why did the maths crash?

A student computes a total: for row in csv.reader(f): total = total + row[2]. Python raises TypeError: unsupported operand type(s). The CSV is perfectly valid. Before tapping: what is row[2], actually, and what is the one-word fix?

🧩 Quiz

When is DictReader clearly better than plain reader?

1. When the CSV's column order might change: row['score'] still works, row[2] silently breaks
2. When the file is very large: DictReader is faster
3. When the values must arrive as numbers instead of strings
4. When the file has no header row

⚠️ Watch out

The three csv-module slips

newline='' forgotten when writing: blank lines between rows (the classic Windows symptom, and an exam favourite "what is wrong with this code").

writeheader() skipped with DictWriter: a headerless file that breaks every DictReader downstream.

Header row processed as data with plain reader: 'roll' fails int(): skip it with next(reader) first.

📖 Theory

One format, three tools now

You now hold three CSV instruments: the sqlite3 shell's .import/.export (Unit 2), Python's csv module (row-by-row control, this lesson), and next lesson pandas' read_csv (the whole file as one object, one line). Choosing among them is real engineering judgement: shell for one-off loads, csv module for streaming row logic, pandas for analysis. Exams love asking two ways to read a CSV in Python: you now have both.

✅ Summary

Key takeaways

• csv.reader yields rows as lists of STRINGS; convert numbers with int()/float(); skip the header with next().
• csv.writer: writerow(list) and writerows(list of lists).
• DictReader keys rows by header names: survives column reordering; DictWriter needs fieldnames + writeheader().
• Open csv files with newline='' (writing especially) to avoid blank-line rows.
• Reader vs DictReader = position-based vs name-based access.
• Memory hook: a trained postman, envelopes by position or by label.