Standard Input/Output — stdin/stdout Redirection

After completing this topic, you will be able to:

Explain the roles of stdin, stdout, and stderr, and connect programs' input and output using redirection (>, <) and pipes (|) in the terminal.

The Three Channels

When a program runs, the operating system automatically opens three channels:

Name	Direction	Purpose	Python
stdin (standard input)	External → Program	Keyboard input	`input()`, `sys.stdin`
stdout (standard output)	Program → External	Normal results	`print()`, `sys.stdout`
stderr (standard error)	Program → External	Errors/warnings	`sys.stderr`

By default, stdin is connected to the keyboard, and stdout and stderr are connected to the screen (terminal).

`print()` Writes to stdout

python

# hello.py
print("안녕하세요")  # Outputs to stdout

bash

$ python hello.py
안녕하세요

print() internally calls sys.stdout.write(). You can also use it directly:

python

import sys

sys.stdout.write("Output to stdout\n")
sys.stderr.write("Output to stderr\n")

Both lines will appear on the screen, but through different channels. This difference becomes apparent with redirection.

Redirection — Directing Channels to Files

You can change the direction of input and output using > and < in the terminal.

Redirecting stdout to a File (`>`)

bash

$ python hello.py > output.txt

Nothing will appear on the screen. The result of print() will be written to the output.txt file.

bash

$ cat output.txt
안녕하세요

>> appends to the existing content:

bash

$ python hello.py >> output.txt  # Appends to existing content

Taking stdin from a File (`<`)

python

# count.py
import sys
lines = sys.stdin.readlines()
print(f"Total {len(lines)} lines")

bash

$ python count.py < data.txt
Total 42 lines

The content of data.txt is fed into stdin instead of keyboard input.

Separating stderr Only

python

# process.py
import sys
print("Processing result: Success")
sys.stderr.write("Warning: Some data is missing\n")

bash

$ python process.py > result.txt 2> error.txt

> redirects only stdout, and 2> redirects only stderr. This allows you to separate results and errors into different files.

Pipes — Connecting Programs

| (pipe) connects the stdout of one program to the stdin of the next program:

bash

$ cat data.csv | python process.py | python report.py > final.txt

text

cat → (stdout) → | → (stdin) → process.py → (stdout) → | → (stdin) → report.py → final.txt

Each program receives data from its input, processes it, and passes the result to the next. This is the core of the Unix philosophy of combining small programs to perform complex tasks.

Reading stdin Line by Line in Python

python

# upper.py — Converts input to uppercase
import sys

for line in sys.stdin:
    print(line.strip().upper())

bash

$ echo "hello world" | python upper.py
HELLO WORLD

$ cat names.txt | python upper.py
ALICE
BOB
CHARLIE

sys.stdin is iterable. Using a for loop to read line by line is memory-efficient.

Relationship between `input()` and stdin

python

name = input("Name: ")  # Prompt → stderr? No, stdout

input() internally:

Prints the prompt string to stdout
Reads a line from stdin and returns it

Therefore, caution is needed when using it with pipes:

bash

$ echo "철수" | python -c "name = input(); print(f'Hello, {name}')"
Hello, 철수

The prompt is meaningless in a pipe. When creating CLI tools, it is better to read sys.stdin directly instead of using input().

Practical Usage Examples

python

# csv_filter.py — Outputs only rows that meet specific conditions
import sys
import csv

reader = csv.reader(sys.stdin)
header = next(reader)
print(",".join(header))

for row in reader:
    if float(row[2]) > 100:  # 3rd column is greater than 100
        print(",".join(row))

bash

$ cat sales.csv | python csv_filter.py > filtered.csv

Using stdin/stdout eliminates the need to hardcode file names. Any file can be connected via a pipe, increasing reusability.

Pipe Chaining — UNIX Philosophy

bash

# Find only ERRORs in the log and sort them by frequency
$ cat server.log | grep "ERROR" | sort | uniq -c | sort -rn | head -5

Each program does only one thing well, and they are connected by pipes. Python scripts can also be part of this chain:

python

# word_count.py — Calculates word frequency
import sys
from collections import Counter

words = []
for line in sys.stdin:
    words.extend(line.strip().split())

for word, count in Counter(words).most_common(10):
    print(f"{count:>5} {word}")

bash

$ cat article.txt | python word_count.py
   42 the
   31 and
   28 to

Outputting Progress to stderr

python

import sys

total = 1000
for i in range(total):
    # Processing logic...
    if i % 100 == 0:
        print(f"Progress: {i}/{total}", file=sys.stderr)

    print(f"Result: {i * 2}")  # Actual output → stdout

bash

$ python process.py > results.txt
Progress: 0/1000      ← Appears on the screen (stderr)
Progress: 100/1000
...

stdout is redirected to the results.txt file, while stderr still appears on the screen. This is a practical pattern for separating progress messages and data output.

Key Summary

Syntax	Meaning
`>`	Redirect stdout to a file (overwrite)
`>>`	Redirect stdout to a file (append)
`<`	Take a file as stdin
`2>`	Redirect stderr to a file
`	`
`2>&1`	Merge stderr into stdout

/dev/null — Discarding Output

bash

# Discard stdout (only see errors)
$ python noisy_script.py > /dev/null

# Discard stderr (only see results)
$ python noisy_script.py 2> /dev/null

# Discard both (completely silent)
$ python noisy_script.py > /dev/null 2>&1

/dev/null is a "garbage can." It is used to suppress unnecessary output in automated scripts.

python

# The same pattern in Python
import os, sys

if os.environ.get("QUIET"):
    sys.stdout = open(os.devnull, "w")

subprocess — Building Pipes in Python

python

import subprocess

# Execute an external command + capture stdout
result = subprocess.run(
    ["ls", "-la"],
    capture_output=True,
    text=True
)
print(result.stdout)
print(result.stderr)

# Pipe chaining
p1 = subprocess.Popen(["cat", "data.txt"], stdout=subprocess.PIPE)
p2 = subprocess.Popen(["grep", "ERROR"], stdin=p1.stdout, stdout=subprocess.PIPE)
output = p2.communicate()[0].decode()

You can programmatically connect stdin/stdout of shell commands within Python.

print() writes to stdout, and error messages go to stderr. Understanding this separation makes redirection and pipes more natural.

BioPlayground

Standard Input/Output — stdin/stdout Redirection

Standard Input/Output — stdin/stdout Redirection

After completing this topic, you will be able to:

The Three Channels

`print()` Writes to stdout

Redirection — Directing Channels to Files

Redirecting stdout to a File (`>`)

Taking stdin from a File (`<`)

Separating stderr Only

Pipes — Connecting Programs

Reading stdin Line by Line in Python

Relationship between `input()` and stdin

Practical Usage Examples

Pipe Chaining — UNIX Philosophy

Outputting Progress to stderr

Key Summary

/dev/null — Discarding Output

subprocess — Building Pipes in Python

💬 Questions & Comments

Standard Input/Output — stdin/stdout Redirection

Standard Input/Output — stdin/stdout Redirection

After completing this topic, you will be able to:

The Three Channels

print() Writes to stdout

Redirection — Directing Channels to Files

Redirecting stdout to a File (>)

Taking stdin from a File (<)

Separating stderr Only

Pipes — Connecting Programs

Reading stdin Line by Line in Python

Relationship between input() and stdin

Practical Usage Examples

Pipe Chaining — UNIX Philosophy

Outputting Progress to stderr

Key Summary

/dev/null — Discarding Output

subprocess — Building Pipes in Python

💬 Questions & Comments

`print()` Writes to stdout

Redirecting stdout to a File (`>`)

Taking stdin from a File (`<`)

Relationship between `input()` and stdin