Reading CSV and JSON with the standard library

Why these two formats

CSV means comma-separated values. It is a plain text file where each line is a row and commas separate the columns; the first line usually holds the column names (the header). Excel, Google Sheets, bank statements and government data portals can all export CSV. JSON (JavaScript Object Notation) is the format most web APIs use: it has objects in curly brackets and lists in square brackets, which map straight onto Python dicts and lists.

Text that behaves like a file

The csv module reads from a file. In the browser tasks there are no files, so we keep the CSV in a string and wrap it with io.StringIO, which makes a string behave like an open file. At work you would use open() on a real file instead; everything else stays the same.

csv.reader: rows as lists

import csv, io

text = """item,qty,price
milk,2,30
bread,1,45
rice,5,60
"""
for row in csv.reader(io.StringIO(text)):
    print(row)
# ['item', 'qty', 'price']
# ['milk', '2', '30']
# ['bread', '1', '45']
# ['rice', '5', '60']

csv.DictReader: rows as dicts

DictReader reads the header line for you and gives each row as a dict keyed by column name. That is exactly the list-of-dicts table from the previous topic, so it is usually the better choice.

rows = list(csv.DictReader(io.StringIO(text)))
print(rows[0])   # {'item': 'milk', 'qty': '2', 'price': '30'}

total = sum(int(r["qty"]) * int(r["price"]) for r in rows)
print(total)     # 405

Blanks and messy values

Real files have gaps. A blank cell arrives as an empty string "", and int("") raises a ValueError. Decide on purpose what a blank means: skip the row, count it as 0, or keep it as None (missing). For money and marks, None is usually the honest choice, because 0 is a real value. Module 4 goes deeper into handling missing data.

text = """name,city,salary
Asha,Pune,52000
Ravi,"Mumbai, MH",
Meena,Delhi,61000
"""
for r in csv.DictReader(io.StringIO(text)):
    salary = int(r["salary"]) if r["salary"].strip() else None
    print(r["name"], r["city"], salary)
# Asha Pune 52000
# Ravi Mumbai, MH None
# Meena Delhi 61000

JSON: loads and dumps

json.loads (load string) turns JSON text into Python values: objects become dicts, arrays become lists, true and false become True and False, and null becomes None. json.dumps (dump string) goes the other way, which is how you send data to an API or save it.

import json

text = '{"city": "Pune", "orders": [250, 120], "paid": true, "coupon": null}'
data = json.loads(text)
print(data["orders"], sum(data["orders"]))   # [250, 120] 370
print(data["paid"], data["coupon"])           # True None

print(json.dumps({"item": "chai", "price": 15}))           # {"item": "chai", "price": 15}
print(json.dumps({"price": "₹250"}))                      # {"price": "\u20b9250"}
print(json.dumps({"price": "₹250"}, ensure_ascii=False))  # {"price": "₹250"}

At work: real files and pandas

import csv

with open("sales.csv", newline="", encoding="utf-8") as f:
    rows = list(csv.DictReader(f))

import pandas as pd
df = pd.read_csv("sales.csv")   # pandas also guesses number types for you

Common mistakes

• Doing maths on text: "30" + "45" gives "3045". Fix: convert with int() or float() as soon as you read the value.
• Splitting lines with split(","). It breaks on quoted values like "Mumbai, MH". Fix: always use the csv module.
• Crashing on blanks with int(""). Fix: check value.strip() first and use None (or skip the row) on purpose.
• Mixing up loads and load. Fix: loads and dumps work with strings; load and dump work with open files.
• Writing JSON by hand with single quotes. JSON needs double quotes, so {'a': 1} is not valid JSON. Fix: build a dict and call json.dumps.