Globbing in Python, Part 1

What globbing is

Globbing is pattern matching for file paths. Instead of listing every file by name, you describe a shape and the system finds the matches.

*.py        → all Python files
data_?.csv  → data_1.csv, data_A.csv, but not data_10.csv
[abc].txt   → a.txt, b.txt, or c.txt

Python has a built-in glob module that does exactly this. No pip install needed.

The basics: * and ?

The two most common patterns:

• * matches any number of characters (including zero)
• ? matches exactly one character

import glob

# All Python files in the current directory
for f in glob.glob("*.py"):
    print(f)

# Files like data_1.csv, data_X.csv — but not data_10.csv
for f in glob.glob("data_?.csv"):
    print(f)

* is greedy. *.py matches setup.py, my_script.py, even .py (an empty name with an extension). If you want non-hidden files only, combine it:

# Hidden files (starting with .) are NOT matched by *
glob.glob("*.py")    # finds app.py, not .hidden.py

Character classes: [...]

Square brackets match one character from a set:

# Exactly one character from the set a, b, or c
glob.glob("[abc].txt")   # a.txt, b.txt, c.txt — not ab.txt

# Ranges work too
glob.glob("file[0-9].log")  # file0.log, file1.log, ... file9.log

# Negation with ^ or !
glob.glob("file[^0-9].log")  # fileX.log, fileA.log — not file1.log

Recursive glob with **

Python 3.5+ supports recursive globbing with **. This is the most useful pattern for real work:

import glob

# All Python files in current dir and every subdirectory
for f in glob.glob("**/*.py", recursive=True):
    print(f)

# All markdown files anywhere under content/
for f in glob.glob("content/**/*.md", recursive=True):
    print(f)

Without recursive=True, ** is treated like * — it does NOT descend into subdirectories.

glob vs os.scandir vs pathlib.Path.glob

Python gives you three ways to glob. Use each when:

from pathlib import Path

# Same as glob.glob("**/*.py", recursive=True)
for f in Path(".").glob("**/*.py"):
    print(f)

# Filter by something else
for f in Path(".").glob("**/*.py"):
    if f.stat().st_size > 1000:
        print(f"{f} ({f.stat().st_size} bytes)")

pathlib is the modern choice. It returns Path objects (not strings), so you can call .read_text(), .rename(), .stem, .suffix directly.

Common pitfalls

1. Hidden files are invisible to * — files starting with . are skipped by default. If you need them, use glob.glob(".*") explicitly.

2. ** without recursive=True is a no-op for recursion — it acts like * and only matches the current directory.

3. Glob returns strings, not sorted — the order is filesystem-dependent. If you need sorted results, wrap it: sorted(glob.glob("*.py")).

4. No match returns an empty list, not an error — always handle the empty case if the result drives further logic.

files = glob.glob("nonexistent/*.py")
if not files:
    print("No matches — check the directory and pattern.")

Try it yourself

Open a Python REPL and try:

import glob
glob.glob("**/*.md", recursive=True)

See how many markdown files exist under your current directory. Then try glob.glob("content/posts/**/*.md", recursive=True) to list every blog post.

In Part 2, we'll go further — fnmatch for string-only matching, brace expansion, writing a custom glob engine, and globbing across remote filesystems.