Introduction to PyMuPDF: Features, Installation, and Usage Guide

1. PyMuPDF Overview

Before introducing PyMuPDF, it is useful to understand MuPDF, a lightweight PDF, XPS, and e‑book viewer library. MuPDF consists of a core library, command‑line tools, and platform‑specific viewers, offering high‑quality anti‑aliased rendering and precise layout metrics.

MuPDF supports many document formats such as PDF, XPS, OpenXPS, CBZ, EPUB, and FictionBook 2. It allows annotation and form filling on PDF documents, and its command‑line tools can convert documents to HTML, SVG, PDF, CBZ, etc., as well as run JavaScript scripts for document manipulation. PyMuPDF (current version 1.18.17) is the Python binding for MuPDF (current version 1.18.*). It provides access to file extensions ".pdf", ".xps", ".oxps", ".cbz", ".fb2", and ".epub", and can also handle about ten popular image formats such as ".png", ".jpg", ".bmp", and ".tiff" as if they were documents.

2. Features

Decrypt files

Access metadata, links, and bookmarks

Render pages as raster images (e.g., PNG) or vector formats (SVG)

Search text

Extract text and images

Convert to other formats: PDF, (X)HTML, XML, JSON, plain text, etc.

Create, merge, or split PDF pages; insert, delete, rearrange, or modify pages, annotations, and form fields

Extract or insert images and fonts

Full support for embedded files

Re‑format PDFs for duplex printing, spot colors, watermarks, or logos

Comprehensive password protection (encrypt, decrypt, set permissions, user/owner passwords)

Support optional content for images, text, and drawings

Access and modify low‑level PDF structure

Command‑line utility python -m fitz … with features such as encryption/decryption, sub‑document creation, document concatenation, image/font extraction, and layout‑preserving text extraction

New: layout‑preserving text extraction! The script fitzcliy.py with the sub‑command gettext can output text that closely matches the original physical layout, including surrounding images and multi‑column tables.

3. Installation

PyMuPDF

can be installed from source or via pre‑built wheels. Wheels are available on PyPI for Windows, Linux, and macOS, supporting Python 3.6‑3.9 (64‑bit) and, more recently, many‑linux2014_aarch64 for ARM platforms.

Optional dependencies for enhanced functionality include: Pillow – required for Pixmap.pil_save() and

Pixmap.pil_tobytes()

fontTools

– required for

Document.subset_fonts()

pymupdf-fonts

– provides a collection of fonts for text output

Install with: pip install PyMuPDF Import the library using the historic name fitz:

import fitz

4. Basic Usage

4.1 Import and Check Version

import fitz
print(fitz.__doc__)
# Example output:
# PyMuPDF 1.18.16: Python bindings for the MuPDF 1.18.0 library.
# Version date: 2021-08-05 00:00:01.
# Built for Python 3.8 on linux (64‑bit).

4.2 Open a Document

doc = fitz.open(filename)  # filename must be an existing file path

The call returns a Document object. Documents can also be opened from memory or created as empty PDFs, and can be used as context managers.

4.3 Document Methods and Properties

Method/Property

Description Document.page_count Number of pages (int) Document.metadata Metadata dictionary Document.get_toc() Retrieve table of contents (list) Document.load_page() Load a specific page

Example:

>> doc.page_count
1
>>> doc.metadata
{'format': 'PDF 1.7', 'title': '', 'author': '', 'subject': '', 'keywords': '', 'creator': '', 'producer': 'Foxit Reader PDF Printer 10.0.130.3456', 'creationDate': "D:20210810173328+08'00'", 'modDate': "D:20210810173328+08'00'", 'trapped': '', 'encryption': None}

4.4 Retrieve Metadata

Document.metadata

returns a Python dict with keys such as producer, format, author, creationDate, etc. Not all fields are guaranteed to contain data for every document.

4.5 Get Table of Contents

toc = doc.get_toc()

4.6 Working with Pages

Pages are the core of MuPDF functionality. You can render a page as a raster image:

pix = page.get_pixmap()

pix

is a Pixmap object containing the RGB (or RGBA) image data. Various options allow control over resolution, colorspace, alpha channel, rotation, cropping, etc.

Save the raster image: pix.save("page-%i.png" % page.number) Or obtain a vector SVG image:

svg = page.get_svg_image()

Extract Text and Images

text = page.get_text(opt)

The opt argument can be one of: "text" – plain text with line breaks "blocks" – list of text blocks (paragraphs) "words" – list of words without spaces "html" – full visual HTML representation "dict" / "json" – same information as HTML but as Python dict or JSON string "rawdict" / "rawjson" – superset including XML‑style character details "xhtml" – text version with embedded images "xml" – text with full character position and font information

Search Text

areas = page.search_for("mupdf")

This returns a list of rectangles where the string "mupdf" (case‑insensitive) occurs, useful for highlighting or cross‑referencing.

5. PDF‑Specific Operations

Only PDF documents can be modified with PyMuPDF. Other formats are read‑only, but any document can be converted to PDF using Document.convert_to_pdf().

Save changes with Document.save(). You can choose incremental saving (fast, appends changes to the original file) or create a new file.

5.1 Page Manipulation

Document.delete_page()

/ Document.delete_pages() – remove pages Document.copy_page() / Document.fullcopy_page() / Document.move_page() – copy or move pages within the same document Document.select() – keep only selected page numbers, effectively creating a new PDF Document.insert_page() / Document.new_page() – insert new pages

5.2 Concatenating and Splitting PDFs

Append another PDF:

# Append complete doc2 to the end of doc1
doc1.insert_pdf(doc2)

Split a document (first 10 pages and last 10 pages):

doc2 = fitz.open()  # new empty PDF
doc2.insert_pdf(doc1, to_page=9)               # first 10 pages
doc2.insert_pdf(doc1, from_page=len(doc1)-10)  # last 10 pages
doc2.save("first-and-last-10.pdf")

5.3 Closing Documents

When finished, close the document to release file handles and buffers:

doc.close()

6. Additional Resources

The article also includes promotional QR codes for a free Python public course and links to related Python tutorials, but the technical content above constitutes the core instructional material.