Apps Artificial Intelligence CSS DevOps Go JavaScript Laravel Linux MongoDB MySQL PHP Python Rust Svelte Vue

Extract Text from PDFs in Python with PyMuPDF

1 min read .
Updated on
Extract Text from PDFs in Python with PyMuPDF

Extracting text from PDF files is useful for search, indexing, analysis, migration, and accessibility workflows. PyMuPDF provides a fast Python API for reading PDF pages and extracting their text.

Install PyMuPDF

python -m pip install pymupdf

Current PyMuPDF versions support the pymupdf import name. Older examples often use import fitz, which is still seen in existing codebases.

Extract Each Page to a Text File

from pathlib import Path
import pymupdf


def extract_text_from_pdf(pdf_path, output_dir):
    output = Path(output_dir)
    output.mkdir(parents=True, exist_ok=True)

    with pymupdf.open(pdf_path) as document:
        for page_number, page in enumerate(document, start=1):
            text = page.get_text()
            file_path = output / f"page_{page_number}.txt"
            file_path.write_text(text, encoding="utf-8")
            print(f"Saved page {page_number} to {file_path}")


extract_text_from_pdf("data/file.pdf", "output_texts")

How It Works

  1. pymupdf.open() opens the PDF document.
  2. Path.mkdir(..., exist_ok=True) creates the output directory when needed.
  3. Iterating over the document yields each page.
  4. page.get_text() extracts the page’s text representation.
  5. Path.write_text() writes UTF-8 text to a separate file.
  6. The context manager closes the PDF automatically.

Extract Only Selected Pages

For example, extract pages 1 through 3:

with pymupdf.open("data/file.pdf") as document:
    for page_number in range(min(3, len(document))):
        text = document[page_number].get_text()
        print(text)

Remember that Python indexes pages from zero even though humans normally number PDF pages starting from one.

Important Limitation: Scanned PDFs

page.get_text() extracts embedded text. If a PDF page is only a scanned image, there may be little or no text to extract. In that case you need an OCR workflow rather than ordinary PDF text extraction.

Layout Considerations

PDFs store positioned text rather than semantic paragraphs. Multi-column documents, tables, headers, and unusual reading orders can therefore require extra processing. PyMuPDF supports other extraction formats such as blocks, words, dictionaries, and HTML when you need more layout information.

Conclusion

PyMuPDF makes embedded PDF text easy to extract page by page. Use a context manager to close documents reliably, write output with an explicit encoding, and distinguish normal text extraction from OCR when working with scanned documents.

Python

Related Posts

Arrays in Python: Lists, `array`, and NumPy
Updated on
Arrays in Python: Lists, `array`, and NumPy
Python has several ways to represent sequence data. The right choice depends on whether you need flexible general-purpose containers, compact typed storage, or high-performance numerical operations. 1. Lists: The Default General-Purpose Sequence Python lists are flexible and can contain objects of different types: Copy numbers = [1, 2, 3, 4, 5] mixed = [1, "Python", 3.14, True] print(numbers[0]) numbers[1] = 10 numbers.append(6) For most application code, a list is the correct default.
Python Array
Common Types of Functions in Python
Updated on
Common Types of Functions in Python
Python functions range from ordinary def functions to lambdas, generators, methods, and asynchronous functions. Understanding the differences helps you choose the clearest abstraction for each task. 1. Built-in Functions Python includes many functions that are available without imports: Copy print("Hello, world!") print(len([1, 2, 3])) print(sum([1, 2, 3])) Other examples include type(), range(), enumerate(), zip(), min(), and max().
Python
Python Dictionaries: A Comprehensive Guide
Updated on
Python Dictionaries: A Comprehensive Guide
Python dictionaries are mutable mappings that associate unique keys with values. They are a natural fit for configuration, structured records, caches, lookup tables, and JSON-like application data. Create a Dictionary Use a literal: Copy person = { "name": "Alice", "age": 30, "city": "New York", } Or use dict():
Python
Python Lists: From Basics to Nested Lists
Updated on
Python Lists: From Basics to Nested Lists
Python lists are mutable ordered collections and one of the most commonly used data structures in the language. They work well for sequences that need to grow, shrink, reorder, or contain arbitrary Python objects. Create Lists Copy empty_list = [] numbers = [1, 2, 3, 4, 5] mixed = [1, "Hello", 3.14, True] A list can contain values of different types, although homogeneous lists are often easier to process predictably.
Python
Sets in Python: A Practical Beginner Guide
Updated on
Sets in Python: A Practical Beginner Guide
A Python set is a mutable collection of unique, hashable elements. Sets are especially useful for membership tests, deduplication, and mathematical set operations such as union and intersection. Create a Set Use braces for a non-empty set: Copy numbers = {1, 2, 3, 4, 5} Duplicates are removed automatically:
Python
Understanding Tuples in Python: A Practical Guide
Updated on
Understanding Tuples in Python: A Practical Guide
A tuple is an ordered Python collection whose item references cannot be reassigned after creation. Tuples are useful for fixed groups of related values, function return values, dictionary keys when their contents are hashable, and lightweight records. Create Tuples Copy coordinates = (10, 20) colors = ("red", "green", "blue") The comma is what creates a tuple, not the parentheses alone. A one-item tuple therefore needs a trailing comma:
Python Tuple
chevron-up